Public Choice Analyses of American Economic History

Studies in Public Choice

Joshua Hall · Marcus Witcher Editors

Public Choice Analyses of American Economic History Volume 2

Studies in Public Choice Volume 37

Series Editor Randall G. Holcombe, Tallahassee, FL, USA Founding Editor Gordon Tullock, Fairfax, VA, USA

More information about this series at http://www.springer.com/series/6550

Joshua Hall • Marcus Witcher Editors

Public Choice Analyses of American Economic History Volume 2

123

Editors Joshua Hall Department of Economics West Virginia University Morgantown, WV, USA

Marcus Witcher West Virginia University Morgantown, WV, USA

ISSN 0924-4700 Studies in Public Choice ISBN 978-3-319-95818-7 ISBN 978-3-319-95819-4 (eBook) https://doi.org/10.1007/978-3-319-95819-4 Library of Congress Control Number: 2018938802 © Springer International Publishing AG, part of Springer Nature 2018 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG. The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Acknowledgements

We would like to thank the Center for Free Enterprise at West Virginia University for support for this project. Morgantown, WV, USA Morgantown, WV, USA

Joshua C. Hall Marcus M. Witcher

v

Contents

1

An Economic Analysis of the 1st Nationalist Movement of 1783 . . . . . . . John Lovett and Grant Ferguson

2

After Johnny Came Marching Home: The Political Economy of Veterans’ Benefits in the Nineteenth Century . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sung Won Kang and Hugh Rockoff

1

27

3

Asian Exclusion in American Immigration Policy . . . . . . . . . . . . . . . . . . . . . . . Zachary Gochenour

57

4

The Political Economy of the Arbitration Act of 1888 . . . . . . . . . . . . . . . . . . . Joshua Gotkin

69

5

The Political Economy of Bank Entry Restrictions: A Theory of Unit Banking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Charles W. Calomiris and Carlos D. Ramírez

99

6

Partisanship and Electoral Reform: Change in Congressional Cohesion, 1877–1932. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 Rick K. Wilson

7

The Economics and Politics of Unit Banking: Evidence from the McFadden Banking Bill of 1927 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139 Marcus M. Witcher

8

Immigrant Ethnic Composition and the Adoption of Women’s Suffrage in the United States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 Ho-Po Crystal Wong, J. R. Clark, and Joshua C. Hall

vii

Contributors

Charles W. Calomiris Columbia Business School, New York, NY, USA NBER, Cambridge, MA, USA J. R. Clark The University of Tennessee at Chattanooga, Chattanooga, TN, USA Grant Ferguson Texas Christian University, Fort Worth, TX, USA Zachary Gochenour James Madison University, Harrisonburg, VA, USA Joshua Gotkin ERS Group, Tallahassee, FL, USA Joshua C. Hall West Virginia University, Morgantown, WV, USA Sung Won Kang Korea Environment Institute, Sejong, Republic of Korea John Lovett Texas Christian University, Fort Worth, TX, USA Carlos D. Ramírez Department of Economics, George Mason University, Fairfax, VA, USA Hugh Rockoff Rutgers University, New Brunswick, NJ, USA Rick K. Wilson Rice University, Houston, TX, USA Marcus M. Witcher West Virginia University, Morgantown, WV, USA Ho-Po Crystal Wong National Tsing Hua University, Hsinchu, Taiwan

ix

Chapter 1

An Economic Analysis of the 1st Nationalist Movement of 1783 John Lovett and Grant Ferguson

Abstract There is a strong tradition of both: (1) considering the Articles of Confederation a dead-end in U.S. constitutional development, and (2) examining the economic motivation behind the drafting and ratification of the U.S. Constitution. The constitutional movement, however, was not the first attempt at a stronger central government. This paper first overviews the 1783 drive to strengthen the Articles of Confederation that very nearly succeeded. Had this movement succeeded, the impetus for the drafting of the Constitution would have been significantly reduced. In short, the failure of the 1783 “1st nationalist movement” was not destined to happen, but is a critical turning point in U.S. constitutional history. After overviewing the importance of this turning point in history, this paper investigates the economic and geographic interests behind the “first nationalist movement.” A nationalist sentiment voting index is constructed for each delegate to the Articles of Confederation Congress. This index is regressed on state and delegate characteristics. Not surprisingly, nationalists tend to be from populous states and states less dependent on foreign trade. Being a military veteran is also associated with nationalist leanings. There are some surprises, however. State debts appear to play little role in nationalist sentiment and prior service in state government is positively correlated with nationalism. Finally, resulting measures of each delegate’s nationalist sentiment are examined.

J. Lovett () · G. Ferguson Texas Christian University, Fort Worth, TX, USA e-mail: [email protected]; [email protected] © Springer International Publishing AG, part of Springer Nature 2018 J. Hall, M. Witcher (eds.), Public Choice Analyses of American Economic History, Studies in Public Choice 37, https://doi.org/10.1007/978-3-319-95819-4_1

1

2

J. Lovett and G. Ferguson

1.1 Introduction The Articles of Confederation, the constitution governing the United States from 1781 until the ratification of the U.S. Constitution, are typically portrayed as a deadend in U.S. history.1 If anything positive is implied about the Articles, it is that they served as both: (1) a necessary step for keeping the states together during the war and, (2) a bad example to inspire those drafting and ratifying the Constitution. The relatively few works with a more positive, or at least nuanced, portrayal describe the Articles as an interim step in the development of the early American federal system (McDonald 2000; LaCroix 2010), namely moving part-way from a system of sovereign states to a federal system in which the national government is, overall, ascendant. Even among these works, the strong impression is that the Articles of Confederation were not only a sub-standard form of government, but also not a survivable form of government. The goals of this paper are three-fold. The first is to argue that the Articles of Confederation were not destined to be a dead-end. In fact, there were very real attempts to strengthen the Articles in 1781 and then again 1783. This “first nationalist movement” was not the work of those wanting to keep an anemic Confederation alive. Instead, it was the best hope for, and led by, those favoring a stronger national government (“nationalists” as termed in this paper). After the failure of this movement, nationalists eventually abandoned hope of strengthening the Articles of Confederation and took a new track starting in 1785. This second, more famous path, involved replacing the Articles rather than modifying them. It led to the Mount Vernon Conference in 1785, the Annapolis Convention in 1786, the Philadelphia Convention in 1787, and finally the ratification of the Constitution. Had this earlier nationalist movement succeeded, which it very nearly did, it is entirely probable that the young United States would have been left with a national government more functional than the original Articles of Confederation but much weaker than the national government under the later Constitution. The impetus for replacing the Articles, via a relatively radical and quasi-legal convention process, would have been greatly diminished. The history of the United States would indeed have been very different. In short the failed attempt to strengthen the Articles should be seen as a turning point in U.S. history. The second objective of this paper is to investigate the economic and geographic interests behind this first nationalist movement. Beginning with Beard (1913) and more recently with McGuire and Ohsfeldt (1984, 1986, 1989), McGuire (1988), and McGuire (2003), there is a strong tradition of examining the economic motivation behind the drafting and ratification of the U.S. Constitution. This paper conducts a preliminary analysis of voting to strengthen the Articles of Confederation in 1783. A “nationalist” voting index is first created for each delegate to the Congress. This

1 While

the drafting of the Articles of Confederation was completed and submitted to the states for ratification in late 1777, they were not finally ratified until March 1, 1781. They did, however, guide the Second Continental Congress in the interim.

1 An Economic Analysis of the 1st Nationalist Movement of 1783

3

gives us a quantitative measure of nationalist sentiment during this little studied period. Next, this index is regressed against delegate characteristics. The results shed light on which economic/geographic interests favored a stronger national government. Finally (the third goal), the nationalism voting index is examined in the light of common historical perceptions of some of the more famous delegates. The undeconstructed voting index offers a quantitative estimate of each delegate’s politics in practice. The part of the index not “explained” by state interests (the residuals plus a bit more) can be viewed as a more “pure” or independent measure of each delegate’s ideology.

1.2 The First “Nationalist Movement,” 1781 and 1783 The Articles of Confederation have long been recognized as a very weak system of national government. The national government, via the Congress, did have the power to pass legislation. It also established a court to hear disputes between states. It was, therefore, more than just an alliance of sovereign states. Nonetheless, the ability of the national government to assert power was very limited. It lacked specified powers such as the authority to raise armies and, probably most important of all, to tax. Voting was by state with the default being a state votes “Nay.” If a state delegation was absent, under-represented, or split, this was equivalent to an automatic “Nay.”2 In addition, a super-majority was required for passage: seven states voting “Yeah” for procedural issues, nine “Yeahs” for ordinary legislation, and 13 “Yeahs” for amendments. While this might have been partially offset by only having one house (versus two houses and an executive under the Constitution), the general consensus is that it was very hard for the Congress to pass much. Those desiring a stronger national government recognized this and made attempts to strengthen the government under the Articles. A relatively broad-based movement (with the debate and correspondence actually beginning in late 1780) started in 1781. The main goal was to give Congress the authority and mandate to directly tax the states (by amending the Articles, i.e. requiring the consent of all 13 states) via a 5% “impost” or import tax. All but the Rhode Islanders eventually approved this. A second, more directed attempt was made in 1783. The Congress’ Superintendent of Finance, Robert Morris, was the leader of this movement. Again, taxing authority was central. However, this second movement also dealt with other nationalist issues

2A

state had to be represented by at least two delegates (and not more than seven) for it to cast a vote. Therefore states with zero or one delegate (ex. Georgia during all of 1783) were automatically “Nay” votes. States with an adequate number of delegates, but no majority among the state delegates, were likewise “Nay” votes. For example, if North Carolina has two delegates in attendance, two being the norm for most states, and one delegate votes “Yeah” and the other “Nay,” North Carolina therefore votes “Nay.”

4

J. Lovett and G. Ferguson

such as the size, composition, and role of a peacetime army. It is this 1783 movement that this paper attempts to quantitatively analyze. There are several reasons for limiting the study to 1783. First, the 1783 movement encompasses more issues than the 1781 movement. Finally, in 1783, one does not have to worry about military events (ex. Are the British on your state’s soil?) dictating each delegate’s voting pattern. Still, the final nature of the peace is uncertain, giving nationalists enough of a foreign threat to find support for a stronger government. Although it may not have been evident at the time, the 1783 movement was the last hurrah for any nationalist attempts to reform the Articles of Confederation. While the impost is debated at the state level until 1787 (with New York being the main holdout this time, see Dougherty (2001, pp. 60–73)), Robert Morris and much of his proposals were gone by late 1784. This lack of a specific agenda makes the identification of votes for the voting index more difficult and results in fewer available votes after 1783. Once the Treaty of Paris is signed and the results become known, the spectre of a foreign threat greatly diminishes (although many aspects of the peace leaked out in the year proceeding the treaty’s signing), temporarily weakening the nationalist movement. Supporting the idea that the nationalist movement declined after 1783 is research from authors attempting to identify factions in the Congress. These speak of a situation in which groupings shifted from nationalist versus localist factions in the years through 1783, to regional (in particular north-south) factions in 1784 and 1785.3 In short, it is difficult to develop a set of well-defined nationalism votes in 1784 and even more difficult thereafter. Besides losing an identifiable nationalist agenda, alternative methods for strengthening the federal government, other than via actions on the floor of the Congress, began to appear viable starting around 1785. After these years prospects for expanding federal powers in the Congress were much more limited. It became obvious that the nationalists were not going to get significant, if any, extensions of federal power on the floor of the Congress. The prospects for legislation, emanating from the Congress itself, that significantly strengthened the national government are described as reaching a high point shortly before the Treaty of Paris (1783) and to have visibly declined thereafter. Much of this decline is attributed to the removal of wartime pressures for a unified and powerful national government (Ferguson 1961; Rakove 1979). Further, the possibility of a convention outside of the Congress which would strengthen or even replace the Articles was becoming very real.4 By 1785 some nationalists would come to believe that the best way to promote a strong

3 On

this point see Jillson and Wilson (1994, pp. 244–267) and Henderson (1974, pp. 281–378). In fact, 1783 is the only year that Henderson specifically identifies nationalist/anti-nationalist factions. 4 The first serious calls for an outside convention came in 1785 (Jensen 1979, p. 33). In this instance the Massachusetts legislature instructed its delegates in the Congress to propose such a convention. The Massachusetts delegates, however, refused. The first two convention outside the Congress, were between Maryland and Virginia, were also in 1785. This was limited to trade issues. In 1786 the Annapolis Convention would be held. This convention called for the Philadelphia Convention

1 An Economic Analysis of the 1st Nationalist Movement of 1783

5

national government was to vote against measures in the Congress which would strengthen the Articles. The logic was that any strengthening of the Articles would only be minor and would halt any impetus for action through outside conventions. James Madison, a Virginia delegate and supporter of the 1783 nationalist movement, stated in December of 1785, “I think it better to trust to further experience and even distress, . . . than to try a temporary measure which may stand in the way of a permanent one.”5 Other delegates were beginning to share Madison’s belief that a limited strengthening of the federal government would prevent more sweeping reforms (Rakove 1979, pp. 370–381). The question arises of whether or not this movement, like the constitutional movement a few years later, had strong economic and geographic bases. Did this push for a stronger central government take place before the nationalist/antinationalist divide congealed around economic and geographic interests? This movement occurred before political coalitions were well-established. Peace was newly at hand, but there was not yet a peace treaty. It is not impossible to imagine a period in which nationalism was based mostly on ideology and shared experience within the Congress. A second question is, if economic and geographic interests were aligned with this nationalist movement, what were these interests?

1.3 Overview of the Methodology Research on the causes of votes in legislatures typically identifies three key aspects: (1) party, (2) constituency effects, and (3) the personal ideology of the representative (Kuklinski 1977; Lawrence 2007; Rosenson 2003; Ramey 2015). These three factors, or some combination of them and other influences, are known as the “standard model” of legislator representation in Congress (Hill 2015). Interestingly, America’s Congress under the Articles of Confederation lacked parties (Aldrich 1993). While there were groups of legislators that tended to vote the same way on certain issues, such clustering was fleeting (Jillson and Wilson 1994); there was nothing resembling the party structures that would be seen under the Constitution. This is, in a way, fortunate, because there are numerous mechanisms by which parties can influence voting and thereby mask both constituency effects and personal ideology (Kiewiet and McCubbins 1991; Miller 2005; Rohde 2013; Cox and McCubbins 2005). As shown below, the reduced “standard model” of legislator voting under the Articles of Confederation is just a combination of constituency effects and personal issue preferences.

which drafted the Constitution, the document which would replace the Articles and provide for a much stronger national government. 5 Quoted from Rakove (1979, p. 369).

6

J. Lovett and G. Ferguson

Delegate i s Voting Pattern = Constituency Effects + i s Ideology/Issue Preferences

(1.1)

As is common practice, voting patterns are measured by a voting index. In this case an index, termed NAT, measuring the frequency in which a delegate votes in favor of a strong national government is constructed. Constituency effects are proxied by measurable characteristics of each delegate’s state and the local area in which he normally resides. A delegate’s ideology/issue preferences is assumed to be a function of both measurable characteristics of the delegate (ex. military service) and unmeasurable personal preferences. The voting index values (NAT) are then regressed on state level and individual characteristics of the delegates. The magnitude and confidence levels of the estimated coefficients give us insight into which interests supported a stronger national government and which opposed it. In addition, the estimated effects of individual delegate characteristics (ex. wartime military service) plus the residual can be interpreted as the delegate’s ideology/issue preference. In fact, it has long been standard practice in voting index studies to interpret the each individual’s residual as a measure of pure ideology (Kau and Rubin 1979; Kalt and Zupan 1984, 1990; Carson and Oppenheimer 1984; Levitt 1996; Voeten 2005). Indeed, Lopez and Ramirez (2008) note that variations of this technique appear “pervasively.” Further, unlike other common and otherwise valuable techniques for identifying the preferences of members of Congress, such as DW-NOMINATE (Poole and Rosenthal 1997), preferences obtained through the residual technique are theoretically free of the influence of parties and less formal vote trading groups. Finally, the small size of the Congress of Confederation should be noted. After discarding a few delegates who were present only for one or two votes, there are only 48 delegates in the voting index. This means the number of explanatory variables has to be rather limited. It is also natural to expect relatively low levels of confidence for the results. However, that is the nature of the data.

1.4 Choosing the Nationalism Votes The first step in choosing the votes for the index is to define nationalism. Nationalism is broadly defined as desiring a central government with much power and a large role relative to that of the state governments. Nationalism is defined, more specifically, according to the agendas put forth during this period by Robert Morris and Alexander Hamilton. Morris was the biggest figure in what is sometimes referred to as the “nationalist” period in the later days of the war and the early postwar years, roughly 1782–1784 (Ferguson 1961). During these years he served as Superintendent of Finance and led the initiative to give the Congress greater powers.

1 An Economic Analysis of the 1st Nationalist Movement of 1783

7

Alexander Hamilton was a friend and correspondent of Morris.6 Hamilton had already espoused many of his arguments for a stronger central government by 1782. These can be found in a series of writings termed The Continentalist (Hamilton 1904a). Hamilton would later author many of The Federalists Papers, lead the fight for the ratification of the Constitution in New York, and actively work to strengthen the federal government as Secretary of Treasury in Washington’s administration. The following five tenets, from Morris’ and Hamilton’s agendas, are used in choosing votes for the voting index.7 • Taxes: Nationalists favor taxes specifically for the use of the U.S. government. Preferably, these should be collected and administered by agents of the U.S. government.8 • Public Credit: Nationalists favor U.S. government, not state, assumption of most of the Revolutionary war debt. Such debt should be funded but not completely retired. • Military: A permanent and professional army was desired by nationalists. Issues of pay to the Army, including severance pay and pay non-active officers (the half-pay for life issue) also fall under this category.9 • Bureaucracy: A professional and relatively large bureaucracy for the U.S. government was sought by nationalists.10

6 See

“Letter of Hamilton to Robert Morris” (Hamilton 1904d) in The Works of Alexander Hamilton. 7 Three very prominent features of Hamilton’s political thought are not explicitly mentioned in these five tenants. The first of these is regulation of foreign trade. We haven’t included this as a separate category because it is implicitly a part of taxing authority and expanding the jurisdiction of U.S. government. The second missing feature is Hamilton’s desire for a strong executive. This is excluded because: (1) we do not have writings to indicate Morris advocated this (although his actions indicate he favored much authority and strong actions for the quasi-executive boards), (2) the desire to have a professional bureaucracy is quite similar, and (3) there aren’t any votes regarding a true executive in this period. Finally, Hamilton advocated the establishment of a National Bank. Morris, obviously after his role in establishing the Bank of North America, was also a proponent of this. This is not included simply because there are no votes regarding the bank during the years studied. 8 For Hamilton’s views on this point, see Letter to James Duane, 3 September, 1780 (Hamilton 1904b), The Continentalist No. IV (Hamilton 1904a), and Letter to George Washington of 08 April, 1783 (Hamilton 1904c). See also Ferguson (1961, pp. xv, 116, 142–143, 146–148, 160–161) and Main (1961, p. 15, Chap. 15). The prefatory notes to the Journals of the Continental Congress for the years 1783–1785 also refer to an independent income for the federal government as one of the principle issues of these years. These prefactory notes were written in the twentieth century by the editor(s) compiling the various papers and into the Journals of the Continental Congress. 9 For Hamilton’s views on this point see Hamilton (1904b), Jensen (1965), and Ferguson (1961, pp. 50, 115, 157–159, 164, 169–170). 10 For Hamilton’s views on this point see Hamilton (1904b), The Continentalist No. VI (Hamilton 1904a), Wood (1987, pp. 81–93), Main (1961, pp. 15–17), and Ferguson (1961, p. 116).

8

J. Lovett and G. Ferguson

• Jurisdiction: Nationalists favored expanding the jurisdiction of the U.S. government relative to that of the state governments.11 In addition, the vote must, in the context of the proceedings within the Continental Congress, have a clear nationalist position (either yeah or nay). For example, on 07 January of 1783, there were four recorded votes on the issue of how to honor (or fund) outstanding Continental dollars. This is clearly an issue with nationalist/localist ramifications. Nonetheless, it is difficult to discern which position would enhance the nationalist position for each vote. The four votes include a vote on whether to postpone consideration of the matter (for which there could be a nationalist reason), changes in wording, etc. Based on the context in the Journal of the Continental Congress (Ford et al. Various, pp. 30-042 (vol. 24)), none of these votes has a definitively nationalist or localist position. Descriptions of the votes used can be found in Tables 1.1 and 1.2. To identify each vote by number, we used New York State Historical Association’s Atlas of Congressional Roll Calls, Volume I (Lord 1943a). The Journals of the Congress of the Continental Congress are the primary source of the vote, and the immediate debate surrounding it.12

1.5 Calculating NAT, the Voting Index Measure In simple voting indices, each vote counts the same. This is appropriate if there is low absenteeism among the voters and there are a great number of votes. However, if some delegates are present for only few votes, they are likely to have inaccurate index values. High rates of absenteeism were chronic in the Congress of Confederation.13 Consider the case of a delegate, present for only a few votes, who is roughly in the middle of the Congress when it comes to nationalist sentiment. If the delegate’s few votes are ones in which only a few extreme localists vote in favor of strengthening the central government, he will vote “pro-national” on all the votes.

11 Hamilton

wrote in 1780 (Hamilton 1904b, p. 213) that “The fundamental defect is a want of power in Congress.” He referred to Congress’ need for more power other times in Letter to James Duane and the The Continentalist No. 1 such as on pp. 41–43, and p. 51 of The Continentalist No. III. Although he doesn’t specifically use the word jurisdiction or authority he is obviously talking about these. Furthermore, it seems difficult to eliminate a general expansion of congressional authority from the more specific points (a to d) given above. 12 Ford et al. (Various, pp. 5–6). This reference is from the prefactory (i.e. editor’s) notes. Beginning with the first Continental Congress in 1774, daily notes or journals were kept of the body’s proceedings. These journals were more succinct and haphazard than the House and Senate journals later seen under the Constitution. At times there were multiple journals in existence, none of which, by themselves, gave a complete description of events. In the early twentieth century these various journals were compiled into the Journals of the Congress of the Continental Congress. The same name is applied to the records of the Congress both before and after it adopted the Articles of Confederation. 13 On absenteeism, see Jillson and Wilson (1990, pp. 153–163) and Montross (1950, pp. 382, 396).

1 An Economic Analysis of the 1st Nationalist Movement of 1783

9

Table 1.1 Nationalism index votes, 1st half of 1783 Date Vote Description Natl Yea 06 Feb 1783 912A That the states be required to pass laws and appoint Yea 16 commissioners to procure accurate estimates of the value of all lands (including buildings and improvements) and to pass laws to collect taxes to meet Congress’ requisitions (rejected) 12 Feb 1783 914 To adopt proposition stating that Congress is of the Yea 24 opinion that the establishment of permanent and adequate funds on taxes or duties to operate in just proportions throughout the US are indespensably needed to do justice to public creditor, restore public credit, and meet the exigencies of war (adopted) 14 Feb 1783 916 To postpone consideration of the report calling for each Nay 9 state’s legislature to submit the name of a commissioner to oversee the estimation of the value of all land (including improvements) within that state (14 Feb 1783) 17 Feb 1783 917 To adopt resolution report calling each state’s legislature Yea 18 to submit the name of a commissioner to oversee the estimation of the value of all land (including improvements) within that state and to report these values to the Congress; said values are to serve as the basis for proportioning sums to be raised to support the public credit and contingent expenses (rejected) 17 Feb 1783 918 To adopt the above resolution with the following Yea 22 changes: the date to submit the commissioner’s names to Congress is delayed from 01 Jan ’84 to 01 Mar ’84 and to a grand committee be appointed by Congress rather than a committee of commissioners from each state (adopted) 26 Feb 1783 920 To amend resolution granting officers of the Continental Yea 20 Army full pay for 5 years (adopted) 26 Feb 1783 921 To amend resolution referred to above by granting Yea 21 officers full pay for 5 years (adopted) 26 Feb 1783 922 To adopt paragraph granting officers full pay for 5 years Yea 22 (adopted) 10–15 March: Newburgh Conspiracy/Letter 12-March: News reaches Philadelphia that a preliminary US-Britain peace was signed 19-March: British commander Guy Carlton confirms the preliminary treaty 18 Mar 1783 929 To adopt resolution granting officers full pay for 5 years Yea 27 (rejected) 21 Mar 1786 932 To take up for consideration and completion the part of Yea 9 the report on the public credit which relates to imposts on imported goods and merchandise 18 Apr 1783 947 To adopt the act recommending to the states that they Yea 25 invest in Congress the power to levy duties on imports for 25 years, that they appropriate substantial revenues to paying off the federal war debt, make acceptable cessions of western lands, and ratify the proposed revision of the articles so that states shall pay for expenses for the common defence and general welfare based on their population (free citizens + 3/5 slaves) (adopted)

Nay 9

5

20

9

5

8 7 8

8 22

4

10

J. Lovett and G. Ferguson

Table 1.2 Nationalism index votes, 2nd half of 1783 Date Vote 07 Aug 1783 974

Description To postpone motion that the Commander-in-Chief be requested to attend Congress, that a committee be appointed to confer with him on the peace arrangement and to report the proper manner of receiving him, in order to consider motion that a committee be appointed to report the proper measures to be adopted with respect to the reception of the Commander-in-Chief (rejected) 07 Aug 1783 975 To appoint a committee to confer with the Commander-in-Chief on the peace arrangement (carried) 13 Aug 1783 976 To substitute “union” for “government” in the letter responding to the inhabitants of New Jersey; said letter states “. . . Congress received with pleasure their congratulations on the success of the war, are obliged by the respect and affection for the federal government in their address, and highly approve of their patriotic disposition. . . ” (rejected) 27 Aug 1783 981 To amend the motion to take into consideration what powers exist in Congress by the Confederation, for the purpose of forming a military peace establishment “to consider the question of a peace establishment” (rejected) 16 Sept 1783 995 To let the words preserving the Agent of the Marine’s position stand in the act eliminating the Marine Department except for the agent of Marine (accepted, words stand) 17 Sept 1783 996 To retain the provision providing that officers who do not accept the proposal of their state shall nevertheless be granted the benefits granted by the Congress (accepted, words stand) 26 Sept 1783 1010 To adopt the paragraph proposing that a special committee be appointed to deliberate and report on a means of strengthening American commerce with Europe through obtaining additional support of the Union from the several states. . . ” (adopted) 04 Nov 1783 1063 That the Commander-in-Chief be authorized to and directed to, after the evacuation on New York by the British, to discharge the federal army except for 500 men and officers, or such as he feels necessary (rejected)

Natl Yea Nay Nay 13 12

Yea

20

6

Nay 4

20

Yea

14

10

Yea

15

6

Yea

20

4

Yea

16

6

Nay 12

3

Therefore, his NAT will equal 1 indicating he is an ardent nationalist. However, the nature of the votes in question, not the delegate’s preferences, were the main factor in determining this high ranking. Weighting each vote based on how strongly national or localist the outcome is the ideal solution for this problem. However, it would be preferable to avoid subjective weightings. Instead the rarity of the two positions is used as an indicator of how

1 An Economic Analysis of the 1st Nationalist Movement of 1783

11

strongly national or localist they are. For example, voting pro-national in a case in which only 20% of the delegates vote pro-national is more strongly indicative of nationalist sentiment than voting pro-national when 80% of the delegates do likewise. These weights used are equal to the ordinal ranking (from 0% to 100%) of the average delegate voting a particular position. In particular, the following weighting scheme, %ProNatj is the proportion of delegates voting the nationalist position on vote j , is used. Weightij =

%ProNat 1 + (1 − %ProNat) =1− 2 2

(1.2)

if delegate i votes a pro-national position on vote j . Otherwise, Weightij =

%ProNat 0 + (1 − %ProNat) =1− 2 2

(1.3)

if delegate i votes a localist position on vote j . For illustration, if there are 100 members in the legislature, and a vote is split 50–50, the average delegate voting in favor of the nationalist position has an ordinal ranking of 75%. The average delegate voting against the nationalist position has an ordinal ranking of 25%. On another vote in which 80% of the delegates voted pro-national, the ordinal position of the average “yeah” vote would be 60% (halfway between 20% and 100%), whereas the average anti-national voter’s percent rank is 10% (halfway between 0% and 20%). NAT is simply the average of a delegate’s weighted values for all votes for which he was present.14  Weightij NAT i = (1.4) n where n = the number of votes for which delegate i was present and where Weightij = 0 if the delegate was absent. In case one has doubts about this weighting scheme, solace can be taken in the fact that the resulting measures are not greatly changed from the unweighted values. The correlation coefficients for the weighted and unweighted values of NAT are

14 Since

NAT is bounded, there is the possibility of a clustering of measures at the extreme values. There is also the possibility of forecasted values of NAT falling outside of the feasible range (see Kau and Rubin 1979). For this reason, the natural log of the odds ratio, i.e. ln[(upperbound) − (NAT/(upperbound − NAT))], is often used. The upper bound for NAT is 1 if the unweighted values are used. It differs when weighted values are used. We do not use this weighting procedure for two reasons. First, with our data, there is not a clustering of values at the extreme ends of the spectrum. Instead they have a relatively normal distribution. Secondly, fitted values do not fall outside of the feasible range. Since the problems this transformation is designed to fix are not present, the simpler, untransformed values are used. Consistency was the criteria for choosing this weighting. In particular, consider a case in which delegate A was present on all votes for which delegate B was present. In addition, A was present for an additional set of votes. Further assume that on the set of votes common to A and B, A voted pro-national more often than B. A should have a higher value for NAT than B to be consistent. The weighting scheme used in this thesis improved this consistency.

12

J. Lovett and G. Ferguson

quite high.15 Finally, to be included, a delegate had to be present for at least 3 of the 20 votes in the index. The weighting procedure, discussed earlier, should address much of the problem caused by the high levels of absenteeism. Nonetheless, we are not comfortable including delegates with excessive absenteeism. This excludes three delegates thereby reducing the sample size from 51 to 48. Like all voting studies, this analysis assumes single peaked preferences; a preferred point over the range measured by the voting index, with the delegate’s utility continually decreasing as one moves either direction away from this point. Multi-peaked preferences are problematic in voting studies. An example of multipeaked preferences would be a delegate whose first choice is greatly strengthening the Articles, second choice is doing nothing to strengthen them, and the absolute last choice is moderately strengthening them. His dislike for proposals to moderately strengthen the Articles presumably stems from a fear that this would preclude any more radical strengthening in the future. It is therefore critical that this nationalist movement be seen as the only game in town for nationalists. Evidence seems to indicate this is the case. Robert Morris led the 1783 push and others who would later be labelled “Federalists” supported it. There is no well-known evidence of any serious discussion of scrapping the Articles at this point. Further, there was no history of a significant failure to strengthen the Articles of Confederation like there would be after the 1783 movement failed. For a nationalist, this movement was the best game in town in 1783. The one known exception is Alexander Hamilton, arguably the most nationalist delegate in 1783. He indicated a preference for scrapping the Articles via a convention as early as 1780. He is on record as opposing many of these measures, including being the only delegate to vote with the (famously anti-nationalist) Rhode Islanders on the famous impost vote. Hamilton clearly has multi-peaked preferences.16 His desired outcome is off the scale of the voting index, and his 15 The

correlation between the unweighted and weighted index is 0.9942 when calculated based on delegates who voted at least five times. 16 Alexander Hamilton is known to have opposed the 1783 impost measure (i.e. adopted the localist position) as it was worded because he thought it was too limited an extension of national authority. In particular, the central government was given fewer revenue sources than in previous plans. He believed passage of the measure would preclude further extensions of central government authority in this regard (Ferguson 1961, Chap. 7). Hamilton even advocated a general convention to strengthen the federal government, rather than working within the framework of the Articles, as early as 1780 (Jensen 1965, pp. 50–51). Hamilton, during these years, appeared to be alone in seriously viewing an outside convention as a possibility, as well as believing that significant extensions of federal power could be had if lesser, compromise extensions were avoided. According to Rakove (1979, p. 346) a 1784 proposal for a convention to suggest amendments to the Articles was never given serious consideration. In addition to Alexander Hamilton, General Phillip Schuyler by 1781 supported a national convention to produce a government superseding the Articles (Beard 1913, p. 55). That is, Phillip Schuyler is the only other national figure for whom we have evidence that he may have opposed strengthening the Articles for nationalist reasons. General Schuyler, however, was never a delegate to the Congress of Confederation. Further, beyond Alexander Hamilton and Phillip Schuyler, we have found no evidence indicated that any other nationalist figures desired to keep the Articles weak in attempt to further nationalist goals. Perhaps the best

1 An Economic Analysis of the 1st Nationalist Movement of 1783

13

second favorite outcome is on the anti-nationalist end of the index. While tempted to drop Hamilton from the data, we are uncomfortable with the idea of dropping individual observations. We therefore retain Hamilton’s data. If, however, he shows up as anti-nationalist, the caveat regarding the peakedness of his preferences applies.

1.6 The Deconstruction of NAT As stated previously, the decomposition of NAT, by regressing it on economic variables, has two goals. First, it gives us information on which economic interests were in favour of strengthening the federal government. Secondly, it will later allow us to calculate a nationalist value (index) for each delegate that is relatively “pure” or independent of what their constituent interests dictate. The right-hand side variables to be used in decomposing NAT are as follows.17

1.6.1 State-Level Delegate Variables • Trade: This is an index of the importance of trade, per capita, in the state. Three components, equally weighted, go into the index. The first of these is Export, the monetary value of exports per capita. A linear trend was fitted to trade data estimated by Shepard and Walton (1976) for the years just prior to the Revolution, and the early years under the Constitution.18 Duties is the second item. This is

evidence of this is Hamilton voting against the impost in April of 1783. He was only one of four to vote against this (versus 25 in favor). His “support” came from Higginson of Massachusetts and the Rhode Island delegation, recognized as the most localists by Alexander Hamilton as well historians of today. Hamilton stated that he viewed the Rhode Island delegation as opposing any sort of compromise and doing their utmost to sink the impost (Ford et al. Various, p. 902 (Vol. 25)). See also, Ferguson (1961, pp. 152–155). David Howell especially is credited with the defeat of the impost. For more on the localists, see Henderson (1974, p. 320). Henderson only classifies four delegates as localist. Two of these four are David Howell and William Ellery of Rhode Island. 17 Population, namely the natural log of estimated state population for that year was used in an earlier study. We will likely bring this back in when (and if) we do a fixed effects study (which gives us many more degrees of freedom). There are several arguments for using a population measure. First, representation under a stronger national government was likely to remain in part on the basis of states rather than population. This, of course, happened under the Constitution. Secondly, if representation was related to population, the level of requisitions would likely be also. More populous states already saw higher requisitions but their burden might rise with a move to representation by population. These imply that a doubling of a state’s population would lead to less than a doubling of a state’s interest in a stronger national government. Finally, there was discussion on changing the representation (and taxation) to the value of improved land and buildings these years. The population data is found in United States Bureau of the Census (1975). More information on the sources of pre-1790 data can be found in Sutherland (1936). 18 Correlation coefficients between these trade measures range between 0.67 and 0.92. Data on exports, duties, and tonnage is found in Gales and (by the authority of the U.S. Congress, pp. 140–

14

J. Lovett and G. Ferguson

duties per capita collected later, under the early Constitutional period (1790– 1791). 1790 through 1791 data is used because it is available (unlike data between the end of the colonial period and 1790). In addition, it reflects the actual amount of duties collected under a stronger national government. This should make Duties a good proxy for the amount of duties a delegate in 1783 should expect a stronger national government would collect. Tonnage is the final component of the Trade index measure. Tonnage is the measured tonnage handled by each state in 1790–1792 (collections did not begun in North Carolina and Rhode Island until 1791–1792). The variable Trade is simply the arithmetic average of a (0– 100) Export index, Duties index, and Tonnage index. Trade is used as an indicator of the expected tax burden under a strengthened Confederation. Taxes on imports, as well as for a while exports, were almost certain to be the main tax supporting a stronger U.S. government (Rakove 1979, p. 346; Jensen 1979, p. 32). While certain domestic industries might benefit from this, individuals with interests in foreign trade would not. This suggest a negative relationship between Trade and NAT. • Debt: This is estimated state debt per capita.19 Most of this was Revolutionary War debt assumed by the states and not yet retired. States with a large state debt per capita are expected to be more pro-national ceteris paribus. This is based on the expectation that a stronger central government would be more likely to assume these debts (Ferguson 1961; McGuire and Ohsfeldt 1984, 1986, 1989; Dougherty 2001). A positive coefficient is therefore expected. • Pop: This is the natural log of estimated state population for that year. 1790 population figures are from the 1790 census. The log is used in the expectation that the impact of this variable increases at a decreasing rate. Population estimates are available for all of the states for various years from 1780 to 1785.20

165, p. 250). We generated fitted values for exports per capita by fitting a linear time trend between pre-Revolutionary figures reported by Shepard and Walton (1976) and the 1790–1792 figures used RawEXP. We call this measure Export. Per capita gross import duties per capita (Duties) for 1790– 1791 as well as per capita tonnage (Tonnage) for 1790–1792 (after collections were finally begun in North Carolina and Rhode Island) was also calculated. The reason the coefficients change from year to year, even though most of the values are for 1790–1791, is that the mix of delegates changes.There are different numbers of delegates from some states each year. 19 Figures for 1790 come from “Report on the Assumption of State Debts” by Alexander Hamilton in Lowrie and Franklin (1832, pp. 28–29). Also, Ratchford (1941, pp. 50–51). Estimates for various years in the 1780s are available for some of the states Ratchford (1941, p. 45) We fitted a linear trend for states in which two data points were available and used the 1790 data otherwise. The states for which two data points were available did not exhibit radical changes over this period. Therefore we felt it appropriate to use 1790 data in all years for the other states. 20 This data is found in Government (1976, pp. 25–37, pp. 1168–1171). More information on the sources of pre 1790 data can be found in Sutherland (1936). An exponential growth curve was fitted to interpolate data for the missing years. NAT should be positively related to Pop. Under the Articles of Confederation states were officially sovereign and voting was by state. Under a more federal system voting would likely to be apportioned (in part) by population or a closely related measure. Large states would gain representation and small states would lose implying a positive coefficient for Pop (Jensen 1966, 1979). Further, even in modern times, U.S. Senators

1 An Economic Analysis of the 1st Nationalist Movement of 1783

15

• West: This is the natural log of non-ceded lands per capita plus 0.1. Non-ceded means lands that have not been ceded by treaty from Native American tribes. The log is used in the expectation that the impact of this variable increases at a decreasing rate. The 0.1 is added because some states have no claims to disputed lands. West is as indicated below.

West =

Square miles of unceded land claimed by state + 0.1 State population

(1.5)

Only land cessions that were eventually recognized by the federal government are counted as ceded. Virginia’s sessions are a potential wildcard in this study. Virginia did not formally turn over her last western land claims until later in 1783 and these were not accepted by the Congress until 1784. However, Virginia had agreed to do so prior to 1783. Therefore, Virginia’s land cessions of 1783 are pro-rated by being counted as half ceded (the arithmetic average of the preand post-1784 cession values). Alternate results, based on counting Virginia’s 1784 cessions are available upon request from the authors. The basic results are unchanged. Besides representing a source of jurisdictional conflict, these lands also represent a potential source of revenue for the government controlling them. States claiming these lands would be likely to want to keep them for themselves and prevent them from going over to the “common good” as would be likely with a stronger central government (Hamilton 1904a). In sum, a negative correlation is hypothesized between NAT and WEST. • Indian: RawIndian is the proportion (0–1) of a state’s non-Indian citizens living within 50 miles of Indian lands. This variable was generated as follows. First the 1780 and 1790 population maps in Friis (1968) were enlarged. These maps are derived from state level censuses, tax data, and the parts of the national 1790 census not burned by the British, etc. On these maps Friis (1968) placed a dot to mark the approximate location of every 200 or so rural citizens. More information on the sources of this population data can be found in Sutherland (1936). We then transferred boundary lines of non-ceded lands onto this map. These boundaries were obtained from the following sources: Kappler (1904, pp. 5–10), Cappon (1976, p. 61), and Prucha (1992). Cessions never recognised by the federal government, even many years after the fact, are excluded. Examples include those by the state of “Franklin” and several of Georgia’s cessions during this period. The boundaries were extended 50 miles and the number of dots within the extended boundaries was counted. The 1790 values and 1780 values were interpolated to produce an estimate for the year in question. This was then divided by the state’s estimated population for that year (see the next sub-section).

from larger states are estimated to be more supportive of broad national interests over smaller and more localized projects (Atlas et al. 1997).

16

J. Lovett and G. Ferguson

For years in which boundaries changed, the average of the before and after values was taken. Several counts were taken for each state for each year to insure consistency. Those living in modern-day West Virginia and Kentucky were considered to be Virginians. Those in Tennessee were counted as citizens of North Carolina. Those in Mississippi and Alabama were considered Georgia citizens. With the above exceptions, modern state boundaries were used in classifying citizens. Those living in the Northwest territory were not counted as citizens of any state. Indian is simply the log of RawIndian plus 0.1. States with a large population near Indian lands might be expected to clamor for federal assistance both for measures against the Indians as well as assistance in acquiring land for further settlement. This coefficient is expected to be positive.

1.6.2 Individual Delegate Variables • Frontier: Frontier is the natural log of the distance (in miles) a delegate lived from the nearest non-ceded lands.21 A priori, the relationship between distance from the frontier and nationalist sentiment is less than clear unless Indian and Frontier are separable measures. Populations near Indian lands (and delegates from those areas) might want federal assistance both for measures against the Indians as well as assistance in acquiring land for further settlement. There is also the stereotype of frontiersmen as being suspicious of any outside authority be it state or national. At the very least, it has been recognized that there were different frontier and coastal interests in the Congress.22 Fortunately, Frontier, West, and Indian are quite separable. • Vet: Vet is a dummy variable with a value of 1 if the delegate served in the military (continental or state), and a 0 otherwise. 41.7% of the delegates are veterans. The strong expectation is that service in the military will be associated with a preference for a strong national government. This view is pervasive in historical writing on the early national period. The rationale is based on the idea that military service exposes an individual to the shortcomings of a weak national government (the Congress was woefully negligent in fulfilling its promises to fund the army and pay soldiers). Further, military veterans, both then and now often prefer a strong central authority over the “bickering” of self-interested politicians. • State: State is a dummy variable with a value of 1 if the delegate previously served in the state government aside from the state legislature. The reason for excluding state legislative service is that it is almost universal among delegates. The delegates were chosen by their respective state legislatures. While having been a state legislature was not a necessary condition for service as a delegate to

21 The 22 See,

delegate’s stated home of residence was found in the Lord (1943b). for example, Morgan and Schmidt (1976, pp. 13–16) and Jensen (1965, Chap. 13).

1 An Economic Analysis of the 1st Nationalist Movement of 1783

17

the Continental Congress, it was far and away the norm. Ex ante, the authors can see the linkage between this variable going either way. Service during the war could have shown a delegate first-hand the need for greater coordination among the states thereby making him more nationalistic. Alternatively, state service could be associated with state patronage or a preference/respect for state-level authority. 56.3% of the delegates have state government service.23 The following equation is estimated using ordinary least squares regression: NAT = cˆ + Bˆ 1 Trade + Bˆ 2 Debt + Bˆ 3 Pop + Bˆ 4 West + Bˆ 5 Indian + Bˆ 6 Frontier + Bˆ 7 Vet + Bˆ 8 State + NatRes

(1.6)

Finally, the (lucky or unlucky) absence of any Georgia delegates in 1783 should also be noted. While Georgia’s absence does cost observations, it allows one to conduct and investigation in which Frontier, Indian, and West are separable. Georgia had the most western land claims, and 100% of U.S. citizens in Georgia were less than 50 miles from Indian lands. Even with the use of natural logs, the Georgia delegates are head and shoulders above the other states when it comes to the value of these three variables and they become largely inseparable. The Georgia delegates return in 1784, but by that time the movement to strengthen the Articles is largely dead. Several nationalists who supported the 1783 push now favor attempting to replace the Articles of Confederation with a new Constitution.

1.7 Results The results of estimating Eq. (1.6) using ordinary least squares are given in Table 1.3.24 Examining Table 1.3, one sees that the roughly half of the variation in voting that is “explained” by the variables and the F-statistic is quite significant. The coefficients are mostly of the expected sign with the exception of State. State was the variable for which the authors were least certain, but, ex ante, the general expectation was that the coefficient would be negative rather the positive. The estimated magnitude of a 1-standard-deviation increase in each explanatory variable is shown in the two rightmost columns. For example, if a delegate’s Trade value increases by 1 standard error, his value for NAT is predicted to fall by 0.331 standard errors. The far right column shows these effects with after normalizing the NAT values so that the least nationalist (by voting record) had a NAT value of 0, 23 The Vet

and State variables come from Congress (2005), the National Governors Association and Wikipedia. 24 We might switch to GLS at some time in the future. While the dependent variable is bounded by 0 and 1 (and the measured values fall between 0.183 and 0.686), the predicted value all fall within this range. This, and the desire to keep things simple (the sample is not extensive enough to deserve lots of manipulation) are why the authors stuck with OLS.

18

J. Lovett and G. Ferguson

Table 1.3 OLS decomposition of NAT RHS variable Intercept

Estimated coefficient −0.0223

Trade

−0.000792

Debt

0.0000302

Pop

0.0839

West

−0.0475

Indian

0.0706

Frontier

−0.0818

Vet

0.0473

State

0.0645

R2 Adj R 2 N

0.515 0.416 48

p-stat (sig) (t-stat) 97.0% (−0.0378) 1.82% (−2.47) 98.9% (0.0139) 0.995% (2.71) 16.3% (−1.42) 23.5% (1.21) 13.1% (1.54) 11.1% (1.63) 3.19% (2.23)

Estimated  in NAT from +1 s.d. s.d. change in NAT  NAT if range = 0–100 N.A. N.A. −0.331

−8.18

0.00210

+0.0520

0.474

+11.7

−0.289

−7.13

+0.220

+5.4497

−0.254

−6.26

+0.189

+4.69

+0.260

+6.43

F-stat Sig of F-Stat

5.18 0.00019%

Note: t-statistics in parentheses in column 3. Decomposition is for 3 or more votes, 1/2 VA session counted

and the most nationalist delegate a NAT value of 100. If a delegate’s Trade value increases by 1 standard error, his value for this normalized NAT is fall by 8.18 (ex. from 50 to 41.2). Given that the last two variables, Vet and State, are by nature binary, a 1-standard-deviation increase in Vet or State is rather nonsensical. This would mean going from (for instance) 0% military veteran to 50% (actually, the standard error of Vet is 0.498) military veteran. One either served in the military or not. Therefore, for Vet and State, the estimated effect of going from a 0 (no service) to 1 (service) is shown at the bottom of the cell. The very small sample size reduces one’s expectations of finding highly significant estimated coefficients. Nonetheless, Trade and Pop are significant at better than the 5% level. Trade s sign and significance and estimated magnitude (discussed below) support the hypothesis that much of the voting revolved around the expectation that a tax on trade (the impost) would be a central feature in any strengthening of the national government. Likewise, the results for Pop support the hypothesis that smaller states would lose relative influence in any strengthening of the national government.

1 An Economic Analysis of the 1st Nationalist Movement of 1783

19

State s estimated coefficient, especially in combination with its high level of confidence and magnitude of impact, is probably the largest surprise in the result. Perhaps the experience of serving in the public sector during the war, in any measure not just the military, gave one a sense of national purpose (and need) that outweighed any state versus national government jealousies. Vet and Frontier are almost significant at the 10% level, not a bad showing considering the very small number of observations (48). Military service indeed appears to lead one to favor a stronger national government as expected. The values (negative coefficient) for Frontier suggest that the stereotype of the individualist, anti-authoritarian, frontiersmen holds true. West and Indian do not “perform” particularly well but are at least of the right sign. The evidence weakly supports the idea that state claims to western lands made a state and its representatives fear a stronger national government might diminish any gains that state could hope to gain from its western lands. Likewise, having a large population near Native American lands does seem to be correlated with wanting a larger federal presence to provide aid and/or reduce externalities by helping coordinate Indian policy. Again, however, the evidence only provides weak support for this idea. After the negative (and highly significant) estimated coefficient for State, the biggest surprise for the authors is that Debt is markedly insignificant. Debt s lack of any estimated effect is actually quite an important conclusion given all the attention Hamilton’s later debt assumption plan receives. Perhaps without a stronger national government already in existence, and without a definitive plan for the federal assumption of state debts, this was not, in 1783, as much of an issue at it would be under the Constitution.

1.8 Quantitative Measures of Nationalist Sentiment One should not forget that this study generates quantitative measures of nationalist leanings during the first nationalist movement. While these measures are arguably quite fallible, they could be a significant addition to the study of the nationalist movement that followed the American Revolution and continued into the early national period. Most information on where a figure stood is both anecdotal and geared toward the later part of this period. The Federalist vs. Anti-Federalist debates during the drafting of the Constitution, the ratification of the Constitution, and the first three presidencies are the basis for most classifications. This study provides a snapshot for the earlier years of the Early National period. The first measure of nationalist sentiment is simply, NAT, the voting index. Figure 1.1 shows the results of this measure for each of the delegates. Making the comparison a bit easier is the lucky fact that the arithmetic mean of the voting index is almost exactly 50% (50.6%). Although the authors are certainly not experts on the delegates, and are in fact very naive regarding many of them, the results in Fig. 1.1 fit the general history well. As one would expect, the Rhode Islanders led by David Howell

20

Fig. 1.1 NAT

J. Lovett and G. Ferguson

1 An Economic Analysis of the 1st Nationalist Movement of 1783

21

and Virginia’s Arthur Lee (famous for his contrary rhetoric in the Continental Congress and brother of both Richard Henry Lee and Francis Lightfoot Lee) are the leading anti-nationalists. Even without the Pinkney’s, South Carolina’s delegation is the most nationalist. James Duane of New York is showing up as a strong nationalist as expected. Alexander Hamilton (NY) is, as stated earlier, already against strengthening the Articles if it means keeping them alive longer than necessary. He is not, however, against strengthening the national government in ways that do not run the risk of turning the Articles into a survivable long-term government. Hamilton, for example, votes in favor of all the measures designed to provide for a permanent officer corps. Accordingly, Hamilton shows up as an antinationalist but not as extreme a one as Arthur Lee or the Rhode Islanders. There could, however, be a few surprises. Elbridge Gerry of Massachusetts will later become famous being one of the few delegates to the Philadelphia constitutional convention to vote against the Constitution. At this point, however, he appears to be in support of a government much stronger than the Articles of Confederation. There was a large gap between a slight strengthening of the national government and going all the way to the U.S. Constitution. Madison is also interesting. Madison will go on to become one of the leading intellectual minds behind the Constitution but also a leader of the (anti-federalist) DemocraticRepublican Party. Madison’s 1783 voting index suggests that his attitudes towards government did not necessarily change. Instead, while he preferred something stronger than the Articles of Confederation, he did not prefer a stronger national government in every regard. With the caveat that every added quantitative step a researcher introduces creates more room for error, one can go a step further and interpret the residuals as a more pure measure of sentiment. This interpretation of voting index residuals, as discussed previously, is quite common. Regarding the Articles of Confederation there are occasional references to at least some of the delegates having such divided sentiments; sentiments based on their state’s interests versus their own personal sentiments. Morgan and Schmidt (1976), for example describes Thomas Burke of North Carolina (a delegate from 1777 to 1781) as having major qualms regarding North Carolina’s claims to western lands. On the one hand, Burke felt it was in his state’s interest to oppose the cessions of these lands. On the other hand, Burke strongly sympathized with the nationalist idea of ceding or otherwise using these lands to the benefit of the United States as a whole (Morgan and Schmidt 1976, pp. 74–76). A difficulty with this “residual as pure ideology” argument is that there is not always a sharp dividing line between a representative acting on behalf of a constituency and acting based on his or her personal beliefs. A congressional district in West Virginia today is more likely to elect someone who truly believes that coal mining is good for the nation than is an urban Los Angeles district. At best, the residual can be interpreted as a representative’s ideology that is not based on, or in sync with, the interests he or she represents. Therefore, the authors prefer the term “independent ideology” to “pure ideology.”

22

J. Lovett and G. Ferguson

Continuing with this idea of “independent ideology,” not all of the voting index explained by the right-hand-side variables needs to be taken out. If a right-handside variable measures a characteristic of the delegate that is not highly correlated with his constituency, the author believes the effect of that variable should count as part of the delegate’s independent ideology. For example, assume Congresswoman Elizabeth Bennet has a degree in the liberal arts, contributes to several organizations for the arts, and buys 300 fiction books a year. Her constituency, however, does not differ from the norm in any measurable way when it comes to supporting the arts. Just because we quantify the type of degree each congressperson has, their giving to the arts, and the number of books they buy, and put these into my regressions does not mean their estimated effects should be attributed to her constituency. In short, we prefer to put back in the estimated effects of the three individual level characteristics; Vet, State, and Frontier, before presenting our “independent ideology” measures. The derivation of this measure, termed NatInd, is shown below. NatInd = Bˆ 6 Frontier + Bˆ 7 Vet + Bˆ 8 State + NatRes

(1.7)

We are very confident including the estimated effects of Vet in NatInd.25 For the most part, a delegate’s military service was a thing of the past. They idea that a delegate would be favoring a stronger national government to gain some favors from the military seems very far-fetched. Including the effects of State and Frontier is a harder call since these can measure a constituent interest. Although the state legislature chose that state’s delegates to the Congress of Confederation, a frontier delegate likely had a career path that as some point meant representing the frontier. Likewise, a delegate with lots of connections in state government probably wished to keep many in the state establishment happy so he or she might have a job to return to when they are no longer a delegate. Still, the possible constituency effects of Frontier and State are less direct than the state-level variables Trade, Debt, Pop, West, and Indian. Again, it was each state’s legislature that chose the delegates to the Congress of Confederation. Adding the estimated effects of Vet, Frontier, and State results in NatInd having a mean that is different than zero. When presenting NatInd in Fig. 1.2, this mean has been subtracted. The range, however, has not been otherwise normalized. For example, the −16.2% value for David Howell of Rhode Island simply means that his NAT measure is 0.162 below where it would be based solely on his state’s characteristics. The actual NAT values range from a low of 0.183 to a high of 0.686 rather than 0–1. If we normalized the ideology measure NAT in the Congress to run from 0 to 1, Howell’s normalized NatInd would be −32.2%. The results are shown in Fig. 1.2. A couple of delegates are of interest. The Rhode Island delegates, while still anti-nationalists, are less rabidly anti-nationalists than their unadjusted voting record would imply. Rhode Island was more exposed to

25 Adding B ˆ 1 Vet + Bˆ 2 State + Bˆ 3 Frontier results in NatInd having a mean that is different than zero. When presenting NatInd in Fig. 1.2, this mean has been subtracted.

1 An Economic Analysis of the 1st Nationalist Movement of 1783

23

Fig. 1.2 NATInd, nationalist sentiment independent of state interests

foreign trade than any other state. One can argue that much of the Rhode Island delegations opposition to strengthening the national government was the rational fear of taxes levied on foreign trade. Arthur Lee of Virginia is the most antinationalist according to NatInd, something the authors find quite believable (he was the rhetorical leader of the anti-establishment faction in the Congress). Finally, the delegates from Delaware stand out even more so than they did in Fig. 1.1. While the authors are not overly familiar with either Gunning Bedford Jr. or Eleazer McComb of Delaware, perhaps they deserve more investigation.

24

J. Lovett and G. Ferguson

1.9 Conclusion One goals of this paper have been threefold. The first is to draw further attention to this critical, but often neglected turning point in U.S. history. The Articles of Confederation are often depicted as doomed to fail. In fact, they almost were strengthened in way that might have allowed them to exist as the United States’ national government for much longer than they did. This possibility of “success” does not mean the outcome under a strengthened Articles of Confederation would be good. Most economists (including the lead author) strongly prefer the U.S. Constitution to strengthened Articles of Confederation. Instead, the idea that the Articles were almost strengthened in a way that would have reduced the impetus for the Constitution should make one more appreciative of the narrow and often lucky path the United States took to get its Constitution. The second goal is to provide some rudimentary analysis of what interests supported (and opposed) this first nationalist movement. Populous states with a relatively low exposure to international trade tended to be the biggest supporters. Veterans and, surprisingly, those with service in state governments (outside of a state legislature) also supported a stronger national government. Finally, a set of (admittedly imperfect) quantitative measures of this set of forgotten founding fathers have been generated. Perhaps these, in conjunction with the more traditional narrative histories, can shed light on this formative period.

References Aldrich JH (1993) Rational choice and turnout. Am J Polit Sci 37(1):246–278 Atlas CM, Hendershott RJ, Zupan MA (1997) Optimal effort allocation by U.S. senators: the role of constituency size. Public Choice 92(3):221–229 Beard C (1913) An economic interpretation of the constitution. MacMillan, New York, NY Cappon LJ (1976) Atlas of early American history: the revolutionary era, 1760–1790. Princeton University Press, Princeton, NJ Carson RT, Oppenheimer JA (1984) A method of estimating the personal ideology of political representatives. Am Polit Sci Rev 78(1):163–178 Congress U (2005) Biographical directory of the United States Congress, 1774 –2005: the Continental Congress. United States Government Printing Office, Washington, DC Cox G, McCubbins MD (2005) Setting the agenda: responsible party government in the U.S. House of Representatives. Cambridge University Press, Cambridge Dougherty K (2001) Collective action under the articles of confederation. Cambridge University Press, Cambridge Ferguson JE (1961) The power of the purse: a history of American public finance; 1776–1790. University of North Carolina Press, Chapel Hill, NC Ford WC, Hunt G, Fitzpatrick JC, Hill RR (eds) (Various) Journals of the Continental Congress, 1774–1789. Government Printing Office, Washington, DC Friis HR (1968) A series of population maps of the colonies and the United States: 1625–1790. American Geographical Society, New York, NY Gales (by the authority of the US Congress) S (1832) American State Papers: commerce and navigation volume 1. U.S. Congress, Washington, DC

1 An Economic Analysis of the 1st Nationalist Movement of 1783

25

Government U (1976) Historical statistics of the United States from colonial times to present. United States Government Printing Office, Washington, DC Hamilton A (1904a) The continentalist. In: Lodge HC (ed) The works of Alexander Hamiton, vol 2. GP Putnam’s Sons, New York, pp 243–291 Hamilton A (1904b) The continentalist. In: Lodge HC (ed) The works of Alexander Hamiton, vol 1. GP Putnam’s Sons, New York, pp 213–239 Hamilton A (1904c) The continentalist. In: Lodge HC (ed) The works of Alexander Hamiton, vol 9. GP Putnam’s Sons, New York, pp 331–337 Hamilton A (1904d) Letter of Hamilton to Robert Morris. In: Lodge HC (ed) The works of Alexander Hamiton, vol 3. GP Putnam’s Sons, New York, pp 319–341 Henderson JH (1974) Party politics in the Continental Congress. McGraw Hill, New York, NY Hill KQ (2015) In search of general theory. J Polit 74(4):917–931 Jensen M (1965) The new nation: a history of the United States during the confederation; 1781– 1789. Vintage Books, New York, NY Jensen M (1966) The articles of confederation: an interpretation of the social-constitutional history of the American Revolution, 1774 –1781. The University of Wisconsin Press, Madison, WI Jensen M (1979) The making of the constitution. Robert E. Krieger Publishing Co., Malabar, FL Jillson CC, Wilson R (1990) Leadership and coordination in legislatures: the President of the First American Congress, 1774–1789. Congress and the Presidency 17(12):85–107 Jillson CC, Wilson R (1994) Congressional dynamics: structure, coordination, and choice in the First American Congress, 1774 –1789. Stanford University Press, Stanford, CA Kalt JP, Zupan MA (1984) Capture and ideology in the economic theory of politics. Am Econ Rev 74(3):279–300 Kalt JP, Zupan MA (1990) The apparent ideological behavior of legislators: testing for principalagent slack in political institutions. J Law Econ 33(1):103–131 Kappler CJ (1904) Indian affairs, volume II: laws and treaties. U.S. Government Printing Office, Washington, DC Kau JB, Rubin PH (1979) Self-interest, ideology, and logrolling in congressional voting. J Law Econ 22(2):365–384 Kiewiet DR, McCubbins MD (1991) The logic of delegation. University of Chicago Press, Chicago, IL Kuklinski JH (1977) District competitiveness and legislative roll call behavior: a reassessment of the marginality hypothesis. Am J Polit Sci 21(3):627–638 LaCroix A (2010) The ideological origins of American Federalism. Harvard University Press, Cambridge, MA Lawrence CN (2007) Of shirking, outlier, and statistical artifacts: Lame-Duck legislators and support for impeachment. Polit Res Q 60(1):159–162 Levitt SD (1996) How do senators vote? Disentangling the role of voter preferences, party affiliation, and senator ideology. Am Econ Rev 86(3):425–441 Lopez EJ, Ramirez CD (2008) Mr. Smith and the economy: the influence of economic conditions on individual legislator voting. Public Choice 136(1):1–117 Lord CL (ed) (1943a) The atlas of Congressional Roll Calls: volume I, the Continental Congresses and the Congresses of Confederation. New York State Historical Society, New York, NY Lord CL (1943b) The atlas of Congressional Roll Calls: volume I, the Continental Congresses and the Congresses of Confederation. New York State Historical Association, Cooperstown, NY Lowrie W, Franklin WS (1832) American State Papers: finance volume 1. Gales and Seaton, Washington, DC Main JT (1961) The antifederalists: critics of the constitution, 1781–1788. University of North Carolina Press, Chapel Hill, NC McDonald F (2000) State’s rights and the union: imperium in imperio, 1776–1876. University of Kansas Press, Lawrence, KS McGuire RA (1988) Constitution making: a rational choice model of the Federal Convention of 1787. Am J Polit Sci 32(2):483–522 McGuire RA (2003) To form a more perfect union. Oxford University Press, New York, NY

26

J. Lovett and G. Ferguson

McGuire RA, Ohsfeldt RL (1984) Economic interests and the American Constitution: a qualitative rehabilitation of Charles A. Beard. J Econ Hist 44(2):509–519 McGuire RA, Ohsfeldt RL (1986) An economic model of voting behavior over specific issues at the Constitutional Convention of 1787. J Econ Hist 46(1):79–111 McGuire RA, Ohsfeldt RL (1989) Self-interest, agency theory, and political voting behavior: the ratification of the United States Constitution. Am Econ Rev 79(1):219–234 Miller GJ (2005) The political evolution of principal-agent models. Ann Rev Polit Sci 8:203–225 Montross L (1950) The reluctant rebels: the story of the Continental Congresses, 1774–1789. Harper and Brothers, New York, NY Morgan D, Schmidt W (1976) North Carolinians in the Continental Congress. John Blair, WinstonSalem, NC Poole KT, Rosenthal H (1997) Congress: a political-economic history of Roll Call Voting. Oxford University Press, New York, NY Prucha FP (1992) Atlas of North American Indians. University of Nebraska Press, Lincoln, NE Rakove J (1979) The beginnings of national politics. Knopf, New York Ramey A (2015) Weighing the alternatives: preferences, parties, and constituency in roll-call voting. J Polit 77(2):421–432 Ratchford B (1941) American state debts. Duke University Press, Durham, NC Rohde DW (2013) Reflections on the practice of theorizing: conditional party government in the twenty-first century. J Polit 75(4):849–864 Rosenson BA (2003) Legislative voting on ethics reform in two states: the influence of economic self-interest, ideology, and institutional power. Public Integrity 5(3):205–222 Shepard JF, Walton GM (1976) Economic change after the American revolution. Explor Econ Hist 13(4):397–422 Sutherland SM (1936) Population distribution in Colonial America. Colombia University Press, New York United States Bureau of the Census (1975) Historical statistics of the United States, colonial times to 1970. US Department of Commerce, Bureau of the Census, Washington, DC, p 93 Voeten EW (2005) Legislator preferences, ideal points and the spatial model in the European Parliament Paper presented at the conference on legislative behavior in the U.S., Europe and beyond. Berkeley, CA, February 25 Wood GS (1987) Interests and disinterestedness in the making of the constitution. In: Beeman R, Botein S, Carter ECI (eds) Beyond confederation, origins of the constitution and American National Identity. University of North Carolina Press, Chapel Hill, NC

Chapter 2

After Johnny Came Marching Home: The Political Economy of Veterans’ Benefits in the Nineteenth Century Sung Won Kang and Hugh Rockoff

Abstract This paper explores new estimates of the number of veterans and the value of veterans’ benefits—both cash benefits and land grants—from the Revolution to 1900. Benefits, it turns out, varied substantially from war to war. The veterans of the War of 1812, in particular, received a smaller amount of benefits than did the veterans of the other nineteenth century wars. A number of factors appear to account for the differences across wars. Some are familiar from studies of other government programs: the previous history of veterans’ benefits, the wealth of the United States, the number of veterans relative to the population, and the lobbying efforts of lawyers and other agents employed by veterans. Some are less familiar. There were several occasions, for example, when public attitudes toward the war appeared to influence the amount of benefits. Perhaps the most important factor, however, was the state of the federal treasury. When the federal government ran a surplus, veterans were likely to receive additional benefits; when it ran a deficit, veterans’ hopes for additional benefits went unfilled. Veterans’ benefits were, to use the terms a bit freely, more like a luxury than a necessity.

let us strive on to finish the work we are in, to bind up the nation’s wounds, to care for him who shall have borne the battle and for his widow and his orphan,. . . 1 (Abraham Lincoln)

1 From

Lincoln’s second inaugural, this statement became the motto of the Veterans Department.

S. W. Kang Korea Environment Institute, Sejong, Republic of Korea e-mail: [email protected] H. Rockoff () Rutgers University, New Brunswick, NJ, USA e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2018 J. Hall, M. Witcher (eds.), Public Choice Analyses of American Economic History, Studies in Public Choice 37, https://doi.org/10.1007/978-3-319-95819-4_2

27

28

S. W. Kang and H. Rockoff

2.1 Veterans’ Benefits by War This paper compares veterans’ benefits across America’s nineteenth-century wars and explores the determinants of the level of benefits. These comparisons are based on new estimates of the number of pensions, the value of cash pensions, and the amount and value of land grants. The amount of benefits, it turns out, varied substantially from war to war and depended on several factors—the wealth of the United States, the previous history of veterans’ benefits, the number of veterans relative to the general population, lobbying by agents for the veterans, and so on. Perhaps the most important factor, however, was the state of the federal treasury. Veterans’ benefits were, to use the term somewhat freely, a luxury: When the federal treasury was in the black demands for additional benefits were approved; when it was in the red demands for additional benefits were denied. By veterans’ benefits we mean simply money or other assets paid to former members of the armed forces. In many cases benefits were simply cash paid to a veteran, often in the veteran’s old age, or to the veteran’s widow or children. Inkind benefits, however, were important in some cases. Land grants were important before the Civil War. Old soldier’s homes, as they were termed, have been important since the Civil War, although facilities for retired army and naval personnel were established earlier. Health care provided through the veterans system has been important since the First World War, although military hospitals were sometimes pressed into service even earlier. Educational benefits were first provided, as far as we are aware, for veterans of World War I and became famous with the GI bill for veterans of World War II. Veterans have also been the beneficiaries of veterans’ preference laws that give veterans a leg up in the pursuit of civil service jobs. Private firms could also introduce a preference for veterans. Here we will focus on the amounts spent by the federal government for cash payments and the amount and value of the land grants. Although it is hard to be certain, we believe that our totals include the great bulk of benefits paid to America’s veterans in the nineteenth century. Since benefits were paid after, sometimes long after, a war ended and the veteran returned to civilian life, we will use present values to compare veterans’ benefits of different wars. In effect, we are asking what the value would be of a bond given to each veteran at the end of the war that would have produced income equal to the benefits the veteran later received. As soon as one begins to calculate present values, of course, various questions arise: what discount rate should be used, how money paid to widows and children should be counted, and so on. We address these questions by computing variants, for example by using alternative discount rates. As it turns out, the broad differences to which we draw attention do not depend on the way present values are calculated.

2 After Johnny Came Marching Home: The Political Economy of Veterans’. . .

29

2.2 What Explains the Level of Benefits? In many cases veterans’ benefits were simply deferred pay. When soldiers volunteered or were drafted they were promised a reward after the war. The contract might have been explicit or implicit: Retirement benefits might have been part of the recruitment package; but even when there was no explicit promise soldiers might have assumed that they would be rewarded after the war simply because legislation rewarding veterans was passed after previous wars. Sometimes there is more to veterans’ benefits, however, than deferred pay. We also reward veterans to express our gratitude for the risks they have taken and the hardships they have suffered. The analogy with gratuities paid for more mundane services, although it may be unpleasant to think about it in these terms, is apt. We often give a gratuity because it is part of the pay for the services rendered, but sometimes we give a larger gratuity to reward extraordinary service. Sometimes we use military language to describe a civilian gratuity: we reward someone generously because they “went beyond the call of duty.” In the nineteenth century civilian language was used, on occasion, to describe veterans’ benefits: Veterans’ benefits were referred to as the “veterans’ gratuity.” George Washington referred to the distinction between deferred pay and gratuities in a circular letter to the governors defending the pension awarded his officers. The pension was part of their hire, “. . . it was the price of their blood, and of your independency; it is therefore more than a common debt, it is a debt of honor; it can never be considered as a pension or gratuity, nor be cancelled until it is fairly discharged” (Glasson 1918, p. 43). The decision to reward veterans was always a contentious political decision that was influenced by a number of factors. We will discuss these factors in this section in general terms and then apply them to the cases that follow. Most of these forces can be grouped under six headings: • The previous history of veterans’ benefits. Veterans’ benefits are a good example of a path-dependent process. The amount veterans received after the last war produces a strong gravitational force. If veterans of the last war received $100 per month starting at age 62, a strong presumption is established that veterans of the next war should get same pension. If they receive less, they have a strong basis for claiming that they have been treated unfairly. If they receive more, they can be criticized for being greedy, and the way is opened for veterans of previous wars who can demand more benefits based on equity. This does not mean that benefits inevitably ratchet upward. At times benefit systems have come under heavy criticism. Perhaps after the last war benefits were extended to individuals with dubious claims to service. Or perhaps benefits were extended through “pension marriages”—aged veterans marrying young women so the women and their children could receive a pension after the veterans died. If so, there may be a reaction in the form of stricter eligibility requirements when the next set of veterans ask for benefits. As with other path-dependent processes, the accidents of history may shape the course of events. The occupant of the White House might be important: A former military officer, for example, might be more sympathetic to veterans.

30

S. W. Kang and H. Rockoff

• Secular changes in national wealth. As the United States has grown richer it has been able to reward its veterans more generously. Indeed, for some purposes it makes more sense to measure benefits relative to national wealth when comparing benefits across wars. In other words, the right question might be whether veterans of one war received more or less relative to wealth in their era compared with veterans of another war. • The number of veterans. The number of veterans is an important determinant of their political influence. As usual, numbers cut two ways. A larger number of veterans means that veterans (and their families) have more votes to offer politicians who support higher benefits. On the other hand, a larger number of veterans means that taxes have to be raised in order to pay for higher benefits, or that the benefits provided other citizens have to be cut. Either decision involves heavy political costs. It may be possible to provide a small number of veterans with higher benefits without raising taxes to the point that the increase is visible to tax payers. • The effectiveness of individuals and organizations working on behalf of veterans. After every war veterans have joined new or existing veterans’ organizations that have lobbied on their behalf. The effectiveness of those organizations and their leaders has played a role in determining the level of benefits. At times lawyers, or other paid agents of veterans have carried out important lobbying efforts, hoping to profit if they win additional benefits for veterans. • Public attitudes towards the war. How much did the veterans actually suffer during the war? How much did they accomplish? How just was the war? Our first reaction is that the answers to these questions should not matter: all veterans should be well treated. But as we will show below, the answers to these questions did influence the level and timing of benefits. • The state of the federal treasury. If the treasury is running a surplus, and more surpluses are in the offing, the demands of veterans can be met easily; if the treasury is running a deficit, and more deficits are in the offing, veterans will find the road to higher pensions difficult. The crucial thing, of course, is how the public and the politicians view the state of the treasury. The same budget will look very different to politicians who hold a balanced budget sacrosanct than to politicians who believe a deficit is a good way of assuring full employment. The state of the treasury will be more important when benefits are paid from general revenues. In the early part of the nineteenth century some naval benefits were paid from a fund derived from the sale of captured foreign merchant ships and their cargoes. This method of finance changes the role of the taxpayer. Naval pensions and the role of the Naval Prize Fund are discussed in detail in Clark et al. (2003). Most nineteenth century veterans, however, were army veterans, so the state of the treasury was the key factor. In the antebellum period the federal government’s and the states’ “land budgets” were often as important as fiscal budgets.2 There were

2 Grubb

(2005) stresses the importance of land to the “net asset” position of the United States.

2 After Johnny Came Marching Home: The Political Economy of Veterans’. . .

31

times when generous grants of land could be made to veterans even when cash grants seemed difficult; in other words times when governments were land rich and cash poor. Veterans’ benefits, we will argue, can be characterized like consumer spending. Benefits to compensate veterans who were injured in the course of the war are analogous to necessities: The government will pay without regard to the current state of the budget. Old-age pensions, however, were like luxuries: The government would “buy” them provided its income was high enough relative to its expenditures. These six factors have often been cited in previous studies of veterans benefits in the nineteenth century such as Bodenger (1971), Oberly (1990), Skocpol (1992), Resch (1999), Teipe (2002), Clark et al. (2003), and the classic with which all other studies of nineteenth-century pensions begin, Glasson (1918). We cannot claim, therefore, that our analysis, including our emphasis the state of the federal treasury, is unprecedented. Indeed, any explanation of something as important as veterans’ benefits that was based on an entirely new set of explanatory factors would be highly suspect. Rather, our claim is that we have sharpened the existing portrait of the political economy veterans’ benefits and established more rigorously some of the empirical generalizations already in the literature by developing and exploring data on the number and value of cash pensions and land grants. We will begin by comparing cash pensions awarded to veterans of the antebellum wars.

2.3 Service Pensions for Veterans of the Antebellum Wars Figure 2.1 compares the present discounted value of the cash benefits received by veterans for service in the Revolution, the War of 1812, and the Mexican War. In each case we deflated the annual amount spent on a pension by a cost of living index (Carter et al. 2006), discounted the resulting series back to the end of the war at 3% or 6%,3 and divided by the number of soldiers and sailors who served in the war (Table 2.1). Evidently, the veterans of the War of 1812 received substantially less than did the veterans of the Revolution or the Mexican War. The difference between the benefits when discounted at 3% and when discounted at 6% points to the important role of timing: The veterans of the War of 1812 do relatively better when a low rate of discount is used. A natural if somewhat cynical conclusion that follows from Fig. 2.1 is that it is the difference in the success of American arms that explains the difference in benefits. America won the Revolutionary War and the Mexican War—indeed America won great empires in these wars—but lost, or at least accomplished very little, in the War of 1812. This interpretation, we argue below in more detail, may have an element of

3 Over

the years 1798–1860 the yield on U.S. long-term bonds averaged (in years when it is available) 5.71%; the yield on New England municipals averaged 5.11 (Carter et al. 2006, series Cj1192, Cj1194).

32

S. W. Kang and H. Rockoff

S100 S90 S80 S70 S60 S50 S40 S30 S20 S10 S0 Revolution

War of 1812 Six Percent

Mexican War

Three Percent

Fig. 2.1 Present value of service pensions. Source: Authors’ calculations using data in Kang and Rockoff (2007, Appendix, Table A2) Table 2.1 Basic war statistics War (1) Start (2) End (3) Number serving (4) Number serving as a % of the population (5) Battle deaths per 1000 (6) Casualties per 1000 (7) Start of a service pension (8) Service pensionsc

American revolution 1775 1783 217,000a 8.81

War of 1812 1812 1815 286,730 3.71

Mexican War 1846 1849 78,718 0.37

Civil War (Union) 1861 1865 2,213,363 6.84b

20.4

7.9

22.0

63.4

n.a. 1818

n.a. 1871

221.5 1887

292.0 1890

$12.59

$2.52

$24.35

$104.87

a The

exact figure is not known. This is the midpoint of figures frequently used by the Department of Defense b The figure is 10.09% if the 1,050,000 estimated Confederate soldiers is added c Per capita, $1860, discounted at 6%. Sources by row: (1–6) Carter et al. (2006, Table Ed1-5); (7) See narrative Sects. 2.3–2.6 in this chapter; (8) authors’ calculations based on Table A2 of Carter et al. (2006, Series Ed327)

truth to it, although another, perhaps more important factor, the state of the federal treasury, was also at work: Veterans of the Revolution benefited from periods of treasury surpluses, while veterans of the War of 1812 came due for their pension when the budget was in deficit.

2 After Johnny Came Marching Home: The Political Economy of Veterans’. . .

33

S250.00

S200.00

S150.00

S100.00

S50.00

S0.00 1818

1828

1838

Nominal

1848

1858

1868

Constant Prices

1878

1888

1898

1908

Constant Wages of Unskilled Labor

Fig. 2.2 Standard annual pension, non-commissioned officers and privates, 1818–1916. Source: Glasson (1918, passim) and Carter et al. (2006, series Cc2, Ba4218)

In terms of nominal amounts, the idea that veterans of a given war should be given what veterans of the proceeding war were given worked to an unusual extent in the nineteenth century. The standard rate of $8 per month did not change from 1818 to 1903! Some variation in the real value of pensions was produced, as shown in Fig. 2.2, by movements in prices and wages. It is evident, however, that price movements were not a major source of the differences revealed by Fig. 2.1. Thus, the main factor behind the differences in Fig. 2.1 was the time that veterans had to wait for their service pensions. One way to highlight the role of timing is to calculate the pensions of a hypothetical standardized veteran. Consider a private or noncommissioned officer, who (1) survived a war without a disability entitling him to a pension, who (2) was eligible for a service pension when it became available, who (3) lived for 70 years after the war, and who (4) left no dependents. Although hypothetical, these calculations provide a picture of the relative size of the pensions; both in terms of what the legislators had in mind, and in terms of what the veterans ultimately received. When we do this we find that our hypothetical long-lived veteran of the Revolution would have received a service pension worth $64 in 1860s, our hypothetical veteran of the War of 1812 would have received $27, and our hypothetical veteran of the Mexican War would have received $143. These figures are higher than the estimates in Fig. 2.1 that reflect the actual experience of the veterans, but the ordering is similar. Evidently, the Veterans of the War of 1812 received less in large part because they had to wait so long for their service pensions. We therefore need to look at the history of how and when these benefit programs

34

S. W. Kang and H. Rockoff

were established to explain why the veterans of the War of 1812 did less well than the veterans of the other antebellum wars.

2.4 Service Pensions: The Revolutionary War The Revolutionary War was no exception to the rule that debate over veterans’ benefits starts with what was awarded to veterans of the last war. English and colonial practice was to pay compensation to all soldiers who were injured and unable to work, but to pay service pensions only to officers. Legislation compensating Revolutionary soldiers for debilitating injuries was passed soon after the Revolution began. The rate for privates and noncommissioned officers was $5.00 per month.4 Pensions for widows and orphans of officers were also well accepted. One of the first awards, in April 1777, was to the youngest son of General Mercer who died in the Battle of Princeton. All told, General Mercer’s youngest son would receive $3876, perhaps $1.6 million in today’s money (2004) using the wages of unskilled labor as an inflator.5 But the officers’ demand for half-pay for life, although based on current British practice, proved highly controversial. At first, General Washington opposed the idea of half pay on the grounds that it would be too expensive; but when conditions in the army deteriorated in 1777–1778, and officers began returning home, Washington changed his mind and supported half-pay. In May 1778 Congress promised half pay, although only for 7 years, for officers who served until the end of the war, and $80 mustering out pay (1 year’s pay) for common soldiers, about $33,000 in today’s money (2004) using the wages of unskilled labor as an inflator (Williamson 2004). In October 1780 Congress, under pressure from the officers and from Washington, increased the offer to officers to half pay for life (rather than 7 years) for those who served until the end of the war. The story, however, was not to end happily for the officers. Early in March 1783 the officers with the American army at Newburgh, New York became increasingly anxious about whether arrears of pay and the commitment to half-pay for life would be fulfilled once a treaty of peace was signed. On March 10, 1783 an anonymous letter appeared urging the officers not to lay down their arms and disband until their demands for back pay and pensions were met: the famous ‘Newburgh Conspiracy.’

4A

crude estimate of daily wages of unskilled labor during the Revolution varies between $0.40 and $0.56. So $5.00 represents what an unskilled laborer could earn in 9–13 days: a small sum to live on, but an important supplement for veterans who had other sources of income. Wages during the Revolution were estimated by using an index money wages for unskilled labor available during the Revolution (Carter et al. 2006, series Ba4218) to backdate daily wages for unskilled labor in Philadelphia (Carter et al. 2006, series Ba4219), a series that first becomes available in 1785. 5 U.S. President’s Commission on Veterans’ Pensions.1877. “Statement of Annual Appropriations and Expenditures for Army and Navy Pensions from March 4, 1789 to June 30, 1876.” 45th Congress, 1st Session, S. Exec. Doc. 4. Williamson (2004).

2 After Johnny Came Marching Home: The Political Economy of Veterans’. . .

35

Washington met this challenge by calling a general meeting of the officers on March 15. At that meeting Washington addressed his men and assured them that Congress would fulfill its promises. The story goes that during the speech Washington began to read a letter from Congress and said ‘you will permit me to put on my spectacles, for I have not only grown gray but almost blind in the service of my country.’6 The story goes that Washington’s officers, some of whom did not know that he wore glasses, were so overcome with emotion that they began to cry. Washington’s emotional appeal, it is said, put an end to the threat. Perhaps the officers were not serious about their threats; it is hard to know. But if the officers had rebelled it could have driven the United States off its path toward democracy. Flexner, who takes the threats seriously, claims that the meeting on March 15 ‘was probably the most important single gathering ever held in the United States’ (Flexner 1974, p. 177). On March 22, with the threat of mutiny quashed, and the budget once more on the front burner, Congress passed the “Commutation Act,” which promised former officers cash or interest bearing securities amounting at face value to full pay for 5 years in lieu of half-pay for life.7 About 2500 officers received securities under the Act (Bodenger 1971, p. 21). The events in Newburgh and the Congressional response are the extreme example of the conflict between the claims of the veterans and the state of the federal treasury that would affect veterans’ benefits for the remainder of the century. As it turned out, the fear of the officers that Congress would fail to pay them once their services were no longer needed proved well founded. Under the Confederation the government failed to pay interest on the commutation bonds. Eventually, these debts were assumed by the federal government under Alexander Hamilton’s plan for reorganizing the debt. Indeed, they were the largest part of the debt taken over by the Federal government. In the meantime, however, many officers had sold their securities to investors at heavy discounts. In 1954 E. James Ferguson studied what happened in Maryland, the state with the best-preserved records, and found that about 52% of the securities issued to Maryland’s soldiers had been acquired by the State of Maryland because they could be used to satisfy state debts. Of the 48% that remained in private hands about 82% had been transferred to private investors at a price that might have been, according to Ferguson (1954, p. 41), 20–40 cents on the dollar (Ferguson 1954, p. 41). The officers continually petitioned Congress for additional compensation. And in 1818 officers in “reduced circumstances” received a pension of $20 per month, about $4200 per month in today’s money (2004) using the wages of unskilled labor as the inflator (Williamson 2004). As we will explain below, 1818 fell in a period of extraordinary federal budget surpluses. Finally, in

6 Flexner

provides a clear and moving account of the conspiracy (Flexner 1974, pp. 168–178). Historians are not agreed on how far the plotters were really prepared to go. In an interesting exchange in the William and Mary Quarterly, Kohn (1970) argued that it was a full blown conspiracy; a position strongly disputed by Skeen and Kohn (1974). 7 The par value of the commutation bonds equaled half pay for 16 years discounted at 6%.

36

S. W. Kang and H. Rockoff

1828 (during another run of surpluses) Congress granted the surviving officers full pay for the remainder of their lives. Service pensions for officers were an old story. There was no precedent, however, in the American experience, as far as we have been able to learn, for a service pension for privates or noncommissioned officers. In a new nation that took democracy seriously, however, it was natural for privates and noncommissioned officers to feel they also deserved a pension, and the public to agree. In the case of noncommissioned officers and privates the precedent, in effect, was the pension received by the officers. The enlisted men, moreover, could also point to failures on the part of Congress to live up to its promises. Congress had paid the enlisted soldiers in Continental dollars that had depreciated rapidly, and sometimes had failed to pay altogether. True, Congress had given the enlisted men interest-bearing securities for the arrears in pay and for the $80 mustering out bonus, but Congress had failed to pay interest on these securities, as it had on the securities given to the officers. And many of the enlisted men, like the officers, had sold their securities at steep discounts from their face value. In the years following the War of 1812 the state of the federal treasury was promising for the passage of a service pension for veterans of the Revolution.8 The United States ran large deficits during the War of 1812; in 1816 and 1817, however, the treasury ran large surpluses, as can be seen in Fig. 2.3. The surplus was 56% (measured as a percentage of expenditures) in 1816, and 52% in 1817 (Carter et al. 2006, series Ea584, Ea585). In retrospect we can see that part of the surpluses were temporary, the result of a postwar surge in imports that produced increased tariff revenues. One result, nevertheless, was an important piece of veterans’ legislation. The law of 1818, mentioned above in connection with the officers, also provided a pension of $8 per month for privates and noncommissioned officers who had served 9 months during the Revolution and were now in “reduced circumstances”—about $1680 per month in today’s money (2004) using wages of unskilled labor as the inflator (Williamson 2004). As can be seen in Fig. 2.3, spending on pensions ratcheted upward as a result. Spending on pensions under the law of 1818 significantly exceeded compensation for wounded veterans including those wounded in the War of 1812 and for the dependents of soldiers and sailors who died while in service (the lowest line in Fig. 2.3). The new law was soon in trouble. The War Department was overwhelmed with applications and it was clear that many veterans were misrepresenting their resources. The widely discussed case of Ebenezer Huntington, a wealthy former officer from Connecticut who had managed to get a pension, personified the corruption. The problem even reached into the War Department where bureaucrats enrolled veterans whose assets exceeded the Department’s standards. The treasury surpluses, moreover, disappeared soon after the law of 1818 took effect: By 1820 the 8 The

following two paragraphs depend heavily on Resch (1999).

2 After Johnny Came Marching Home: The Political Economy of Veterans’. . .

37

S20,000,000

S15,000,000

S10,000,000 Impact of the Law of 1818 S5,000,000

S0 1816

1817

1818

1819

1820

1821

1822

1823

1824

-S5,000,000 Total Pensions

Compensation

Surplus

Fig. 2.3 Service pensions for veterans of the revolution, 1816–1824. Source: Kang and Rockoff (2007, Appendix, Table A2) and Carter et al. (2006, Series Ea584, Ea585)

deficit was 2% of expenditures (Carter et al. 2006, series Ea584, Ea585). Congress responded to the public’s outrage over corruption in the pension system and the deterioration in the budget by striking all of the veterans from the rolls and requiring that they reapply. In the summer of 1820 Americans watched in fascination as veterans of the Revolution mustered prior to presenting their applications: Old soldiers shouldered their crutches as they had once shouldered their muskets. The effect of the re-enrollment amendment shows up clearly in Fig. 2.3 as a sharp contraction in net expenditures in 1821.9 The 1820 amendment, and its vigorous enforcement by Secretary of War John C. Calhoun and the head of the pension bureau James L. Edwards, put an end to the scandals and saved the program. Now public attention shifted from men unfairly added to the rolls to men who unfairly excluded. There were many possible inequities. A veteran who had gone into debt to buy property on the expectation of a pension might be denied a pension on the grounds that his gross assets (one of the criteria used by the War Department) were too high. A veteran who deeded his property to his children in exchange for long-term care, a common practice, would qualify for a pension; a veteran in similar circumstances who kept title to his assets 9 Our

estimates of net expenditure are the differences between warrants issued and repayments. Since repayments could be based on warrants issued in previous years, our estimate of zero net expenditure in fiscal 1821 is consistent with some veterans receiving payments from the government in that year.

38

S. W. Kang and H. Rockoff

would not. The budget, moreover, having recovered from the effects of the Panic of 1819, was now solidly in the black: the surplus was 35% of expenditures in 1822 and 40% in 1823 (Carter et al. 2006, series Ea584, Ea585). In 1823, therefore, Congress allowed veterans to submit amended applications. Veterans were now free to restate their assets in a way that met War Department standards. The 1823 amendment does not seem to have left a strong imprint on the pension rolls (Table A1, column 4), but it does illustrate the willingness of Congress to address inequities provided the budget was favorable. Pressure now began to build for a non-means-tested pension and a reduction in the required term of service. The legislation of 1828, mentioned above, that awarded officers in the continental army who served until the end of the war full pay for life, made a similar award to privates and noncommissioned officers. The decline in the number of veterans and the strict service requirement (until the end of the war), however, meant that few enlisted men would be pensioned under the law of 1828. With this precedent on the books, however, it is easy to see why pressure built to expand benefits further. Veterans who had served, but not until the end of the war, or who had served in state or local militias rather than the Continental Army, now had a basis for claiming a pension. In his first State of the Union Message in 1829 Andrew Jackson, a hero of the War of 1812 and the Indian Wars, advocated extending benefits to the militia who served in the Revolution, although only when a veteran “. . . is unable to maintain himself in comfort.”10 Legislation passed in the House, but stalled in the Senate. Among those opposed to increased benefits was Senator Robert Y. Hayne of South Carolina. Hayne pointed out that there was a close connection between the protective tariff—southerners such as Hayne famously had labeled the tariff of 1828 the “Tariff of Abominations”—and veterans’ pensions. To increase the pension rolls was to delay the day when tariffs could be reduced (Glasson 1918, pp. 77–78). In 1832, however, despite opposition from men like Hayne, legislation provided full pay for life to all veterans without regard to their economic or physical status who had served for 2 years in the regular army, the navy, or (significantly) the militias. Those who had served for shorter periods received proportionately less. As can be seen in Fig. 2.4, the Act of 1832 was passed during a run of years when the federal budget was in surplus. In 1831 the surplus was 87% of expenditures and in 1832, 84%. The surplus had been larger, measured in this way, in only two preceding years (Carter et al. 2006, series Ea584, Ea585). Spending on Revolutionary pensions increased as a result of the legislation. The amounts of additional spending, however, were relatively small, although the effect on the surplus was not negligible. Fifty years had passed since the end of the war, and the number of eligible veterans had been greatly reduced. Nevertheless, the law of 1832 set an important precedent. From this time onward veterans could claim that equity with veterans of the Revolution required that they be awarded a pension based on service alone when they reached an advanced age. Shortly after passing this bill,

10 Andrew

Jackson, State of the Union Message, December 8, 1929.

2 After Johnny Came Marching Home: The Political Economy of Veterans’. . .

39

S25,000,000

S20,000,000

S15,000,000

S10,000,000

S5,000,000

S0 1830

1831

1832

1833

1834

1835

1838

1837

1838

1839

-S5,000,000

-S10,000,000

-S15,000,000 Total Pensions

Law of 1832 Pensions

Surplus

Fig. 2.4 Service pensions for veterans of the revolution, 1830–1839. Source: Kang and Rockoff (2007, Appendix, Table 2) and Carter et al. (2006, Series Ea584, Ea585)

which in some measure increased the need for long-term taxes and showed how future surpluses could be spent, Jackson signed the tariff legislation that led to the Nullification Crisis. The Revolutionary War also set the precedent for pensioning widows. During the war the widows of soldiers or sailors killed in battle were pensioned, a colonial tradition. But under the Act of July 4, 1836 widows who were married before 1794 to revolutionary officers, soldiers, seamen, and marines became eligible for a pension. Thus, women who had married a non-commissioned officer or soldier who had survived the war became eligible for a pension, even if the marriage had taken place a few years after the end of the war. The effect of the pensioning of widows, orphans, and other dependents can be seen in Fig. 2.4 as the opening of the gap after 1836 between total expenditures and Act of 1832 expenditures. This law was a remarkable break with colonial tradition. At the peak over 9000 widows and other dependents would be on the rolls (Table A1, column 5). Pensioning the widows of ordinary soldiers and seamen set a powerful precedent: The widows of veterans of all future wars would be pensioned. Legislation in March 1837 directed the Navy Pension Fund to pay the widows and orphans who were now eligible for a pension from the time of the veteran’s death. This legislation, according to Clark, Craig, and Wilson, although passed when the Navy Pension Fund appeared to be in good financial shape, doomed it (Clark et al. 2003, pp. 58–60). In a few years the fund was gone. Clark, Craig, and Wilson are right that the increase in benefits for veterans was partly due to the attempt

40

S. W. Kang and H. Rockoff

to build up the navy because of the tensions with Britain over the U.S. boundary with Canada. And Theda Skocpol is right that the expansion of the franchise (by reducing property requirements) made it more important politically to pension the widows of ordinary soldiers and sailors than it had been earlier (Skocpol 1992, p. 92). However, the pensioning of the widows and orphans is also consistent with a simple fiscal theory of the pensions: Fiscal 1836 saw the last and largest in a series of surpluses.11 The federal budget would go into deficit after the Panic of 1837. Time soon took its toll. By 1848 little more than 3000 widows remained on the rolls (Table A1, column 5). An act of July 29, 1848 pensioned widows who were married before January 1800. In other words it pensioned women who had married veterans between 1794 (the cutoff under the law of 1836) and 1800. This amendment, a subsequent amendments along the same lines, produced small increases in the number of widows on the rolls. An attempt was made at the time the widows of the Revolutionary soldiers were being pensioned to pension the veterans of various engagements with Native Americans that had taken place in the years immediately following the Revolution.12 This effort ran into opposition based on the political morality of the Indian Wars as they were known. Opponents of pensioning these veterans contended that these wars were different from the Revolution: the Revolutionary War was carried out by the Congress with the patriotic goal of securing independence; the Indian Wars were carried out by private individuals with the goal of private gain. Veterans of the Indian Wars would have to wait for their pensions. As we have seen, Revolutionary War Veterans had to wait 35 years (1783–1818) for a means-tested service pension, and 45 years (1783–1828) for a non-meanstested service pension. Given that the precedent of an old-age pension had been established and that the United States had grown richer, one might have expected that the veterans of the War of 1812 would have received their old-age pension in fewer years. In 1845, 30 years after the end of the War of 1812, for example, real per capita income was 30% higher than it was in 1818 when the Revolutionary War Pension was created (Carter et al. 2006, series Ca11). However, this was not to be. Veterans of the War of 1812 would have to wait longer for their service pensions even though the country had grown richer. Before exploring the cash pensions received by the veterans of the War of 1812 and subsequent wars, however, we need to look at the land grants made to veterans of the Revolution and subsequent wars. As Farley Grubb has recently argued, it is a mistake to look solely at cash expenditures, because land was an important part of the government’s “net asset position” Grubb (2005). It is possible that including land grants would alter our conclusion that the veterans of the War of 1812 did badly compared with the veterans of other nineteenth-century wars. The land grant story, however, reinforces the cash pension story.

11 In

total dollars. The largest surplus measured as a percentage of expenditures occurred in 1835. “Pensions to persons engaged in Indian Wars.” 24th Congress, 1st Session, H. Rpt. 338.

12 U.S. House. 1836.

2 After Johnny Came Marching Home: The Political Economy of Veterans’. . .

41

2.5 Land Grants Land grants for military service were a long-standing colonial tradition.13 Land was awarded, for example, to soldiers who had participated in the Pequot War in 1637. Maryland awarded land in 1645 for help in suppressing the rebellion led by the pirate Richard Ingle. Land was also awarded to the veterans of King Philip’s War in 1675–76. The French and Indian War of 1754–1763 provided the immediate precedent for the Revolutionary land grants. The war began in the west with the campaigns in which George Washington took part. A proclamation by Governor Dinwiddie of Virginia issued in 1754 authorized land grants to soldiers who participated. At the conclusion of the War, George III’s Royal Proclamation of 1763 rewarded veterans 50–5000 acres depending on their rank.14 The reliance of colonial authorities on land grants is easily understood. As Paul Wallace Gates put it in his magisterial History of Public Land Law Development, in that era America was “long on land and short on money” (Gates and Swenson 1968, p. 249). It was natural, therefore, that America would reward veterans of the Revolution with land grants. The earliest offer came in 1776, but ironically it was an offer of 50 acres to British soldiers who deserted (Gates and Swenson 1968, p. 251). To make sure that the offer reached the Hessians, the German mercenaries who were one of the main targets of the offer, it was printed in German on the backs of tobacco wrappers (Hibbard and Gates 1924, p. 118). Both the Continental Congress and the states made grants. The grants made by the Continental Congress went to veterans of the Continental Army; the grants made by the states went to veterans of their militias and as supplemental grants to their citizens who served in the Continental Army. The southern states were especially generous. Virginia, the most populous state, and the one with the most western land, steadily increased its bounties during the course of the war as it became harder and harder to enlist men. Toward the end of the war ordinary soldiers and sailors were promised 300 acres and a slave to work the land. It appears to be North Carolina, however, that made the largest grants. Under a law of 1780 the North Carolina scale started with 640 acres for privates and 1000 acres for noncommissioned officers. The grants to officers, in many cases, were far more generous than the grants to enlisted men. The Continental Congress gave brigadier generals 850 acres and major generals 1100 acres. North Carolina added 12,000 for both brigadier and major generals, but rewarded Major General Nathaniel Greene, the hero of the Revolution in the South, with a “little dukedom,” as one Tennessee legislator put it, of 25,000

13 The

examples in this paragraph are drawn from the Introduction to Lloyd DeWitt Bockstruck’s forthcoming volume on bounty and donation land grants in British North America. We thank Dr. Bockstruck for making the Introduction available to us prior to publication. 14 The Proclamation can be read at http://www.ushistory.org/declaration/related/proc63.htm (Accessed Friday, May 25, 2007).

42

S. W. Kang and H. Rockoff

acres (Gates and Swenson 1968, pp. 252–53).15 Virginia added an additional 10,000 acres for brigadier generals and 15,000 for major generals. Virginia granted Major General Charles Lee, 16,875 acres, Brigadier General Daniel Morgan, 23,328 acres, and Major General Horatio Gates, 31,000 acres (Bockstruck 1996, pp. 196, 308, 376). Washington undoubtedly could have had an enormous amount of land. He refused, however, to accept any land for his service during the Revolution, even though his personal economy depended on the extensive lands awarded to him for his service in the French and Indian War (Flexner 1974, pp. 53–54, 194). The Northern states made smaller distinctions among ranks. Whereas Virginia gave privates 300 acres and major generals 15,000, a ratio of 1–50; New York gave privates 500 acres and major generals 5500, a ratio of 1–11 (Gates and Swenson 1968, p. 252). The states with large claims on western lands were, of course, in a position to reward their veterans more generously. The expectation that those claims might have to be surrendered to the central government simply added another incentive to grant land to their veterans before the opportunity was lost. Although the veterans of the Revolution were promised a great deal of land, its monetary value is difficult to estimate, and may not have been very great. Some preliminary efforts to distribute federal land were made in the mid-1780s, but it was not until 1796 that a military track in Ohio was made available where small holders could exercise their warrants (Gates and Swenson 1968, p. 259). In the meantime most veterans had sold their warrants to speculators. The largest of the speculators in federal warrants was the New Jersey politician Jonathan Dayton who acquired 90,936 acres by buying warrants (Gates and Swenson 1968, p. 260). The story of how land was distributed by the states varied from state to state, but the tendency for veterans to experience long delays and to sell their warrants to speculators and developers was the norm. The policy of rewarding soldiers with land grants continued after the Revolution. Texas, as might be expected based on its vast acreages, made generous grants. The Texas constitution of 1836 promised single men over the age of 17 who were loyal to the state a third of a league (1476 acres) and families “a League and Labor” (4.606 acres). An act of December 21, 1837, awarded 640 acres to veterans who had participated in various engagements during the Texas War of Independence such as the Goliad campaigns of 1835 and 1836, and the battle of the Alamo.16 Initially, veterans of the War of 1812 who served in the regular army received bounties of 160 acres; an amount that was raised to 320 acres toward the end of the war (Hickey 1989, pp. 243–44). About 29,000 of the 60,000 regulars took them up. There were, however, several problems with the grants from the point of view of the veterans. First, the grants were located in a few restricted western areas, in part because it was hoped that the veterans would be a buffer against Native

15 The

area of the island of Manhattan is about 13,000 acres; the borough, which includes some water and adjacent islands, is about 22,000 acres. 16 See Land Grants in The Handbook of Texas Online; http://www.tsha.utexas.edu/handbook/ online/articles/LL/mpl1.html, accessed 1/13/2018.

2 After Johnny Came Marching Home: The Political Economy of Veterans’. . .

43

Americans. This was also true of the Revolutionary warrants, but many of the Revolutionary grants were in highly productive farmland, such as those in South Central Ohio. Although some of the other areas reserved for the veterans of 1812 did contain good farmland, one area in Michigan reserved for veterans of the War of 1812 was replaced when it was found to be too swampy. The warrants, moreover, could not be sold to a broker, at least not without some legal maneuvering; they were not “assignable” in the language of the time. Although veterans and their agents found various ways around this limitation, it raised the cost of selling the warrants. The most important factor reducing the grants to the veterans of the War of 1812, however, was simply that the veterans who had served in state militias were not entitled to grants. State governments granted land to many veterans who had served in militias during the Revolution. However, the states were no longer in a position to make land grants because they had ceded their western lands to the federal government, and the Federal government could not be persuaded to step in a grant land to men who had served in state militias. It may have been the officers of the War of 1812 who felt most aggrieved. As we noted, the officers of the Revolutionary War received very handsome land grants. In 1811, however, when bounties were first offered to soldiers for the War of 1812, there was nothing extra for the officers. Further attempts to reward the officers were made during the war and in 1817, but nothing came of them.17 As Gates suggests, Congress was probably reacting to the perception that great estates had arisen as a result of the huge grants made at the close of the Revolution. Congress also may have been reacting to the related perception that the land grants had produced excessive speculation in western lands, always a political liability. There was also the perception that the fighting was not very intense. In a memorial to Congress in 1831, a group of War of 1812 officers pointed out the Revolutionary War precedents and asked for similar treatment. They also pointed out that U.S. citizens living in Canada in 1812 and who had volunteered to serve with the American forces had received land grants, and in that case the officers had been given larger grants than the enlisted men—up to 960 acres for a colonel. The officers conceded that they may not have suffered as much as the officers of the Revolutionary armies, but they insisted that they had accomplished as much. “The [officers in the War of 1812] . . . exhibited the same valor and love of liberty [as the officers in the Revolution], and, although they may not as a body have suffered as much, yet their zeal was not less, nor their exertions less meritorious or successful.”18 One wonders, of course, whether Congress found the last part of their claim persuasive. Unlike their brethren in the Revolution and the Mexican War which we will discuss below, the veterans of the War of 1812 had conquered little land for the United States.

17 Gates

and Swenson (1968, p. 262)—especially footnote 37. House. 1831. “Memorial of Sundry Officers of the United States’ Army in the Late War, Praying for a Grant of Land in Consideration of their Services.” 21st Congress, 2nd Session, H. Doc. 25.

18 U.S.

44

S. W. Kang and H. Rockoff

No legislation responding to these complaints by the veterans of the War of 1812 was forthcoming, however, until the land grants given to the veterans of the Mexican-American war (1847–1848) ignited a campaign to secure additional land for the veterans of 1812. The “Ten Regiments Act” of 1847 designed to fill the ranks for the Mexican War provided a warrant for 160 acres of land anywhere in the public domain for every regular and volunteer who joined the war against Mexico and served for 12 months or until the end of the war. The warrants, moreover, were assignable, they could be sold to a broker for cash if the veteran did not want to settle the land. According to Oberly “After the act’s passage and the president’s signature, army recruiters busily signed up men on the basis of the new bounty” (Oberly 1990, p. 11). The passage of the Ten Regiments Act reminded the surviving veterans of the War of 1812 of their old grievances. They demanded parity with the veterans of the Mexican War, and Congress obliged soon after the war with Mexico ended. This is not surprising. The veterans of 1812 had a strong claim based on equity, and granting it was easy given the immense booty, including California, taken from Mexico. The Bounty Land Act of 1850 provided veterans of the War of 1812, of various Indian Wars, and of various other fights with up to 160 acres depending on length of service (1 month of service for 40 acres and 9 months for the full 160 acres). The warrants awarded under the Bounty Land Act, however, could not be sold to brokers, a nod to the opponents of the “speculators.” It was implausible, however, that the veterans of the War of 1812, many now in their sixties, would move West to take up farming or even that they would go through the laborious process of taking possession of and selling their land. The veterans wanted assignable bounties. The brokers were also an important voice for liberalization of the Bounty Land Act because they were running out of warrants issued under the Ten Regiments Act. Congress responded with what was popularly known as the Assignment Bill of 1852, which allowed veterans of the War of 1812 to sell their warrants to brokers. The fight to secure bounties for the veterans of the War of 1812 had been led by the United Brethren of the War of 1812. Joel B. Sutherland, a physician in the War of 1812 and a successful politician in the postwar era, headed the organization, and proved to be a skillful champion for the veterans. In 1855 the veterans held a march on the White House holding aloft a banner bearing the slogan under which they had gone to war in 1812: “Free Trade and Sailor’s Rights.” President Pierce met with a delegation, but did not commit himself to their demands (Oberly 1990, pp. 35–36). Further pressure, however, produced the Old Soldiers Act of 1855, which gave 160 acres to any soldier, whether in the regular army or the militia, who had served at least 2 weeks in the War of 1812. This was the high point for the United Brethren, which never held another national convention. We have made some rough estimates, shown in Fig. 2.5, of the amount of land given to the veterans of the Revolution, the War of 1812, and the Mexican War. These estimates are subject to a wide margin of error. Nevertheless, we believe that the impression created by the chart is probably broadly correct. The details of how we made the estimates are in the appendix of Kang and Rockoff (2007). The bar for the War of 1812 is divided into two parts. The lower part shows the amount awarded

2 After Johnny Came Marching Home: The Political Economy of Veterans’. . .

45

200.0 175.0

Acres per Veteran

150.0 125.0 100.0

Grants made to Veterans of the War of 1812 after the Mexican War

75.0 50.0 25.0 0.0 Revolution

War of 1812

Mexican War

Fig. 2.5 Land bounties in three antebellum wars, acres per veteran. Source: Kang and Rockoff (2007, Appendix, Table A3)

during the war. The upper part shows the amount awarded to the veterans of the War of 1812 after the Mexican War. The veterans of the War of 1812, evidently, received less initially, but caught up many years later. Without the latter grants the veterans of the War of 1812 would have done less well in terms of total land received, than the veterans of the Revolution.19 We have also made some back-of-the envelope estimates of the real value to the veterans of the land grants. These estimates are subject to larger margins of error than the estimates of the total amount of land received because we lack data on exactly when veterans disposed of their land, how much they were paid for it, and what they were charged in transaction costs. Nevertheless, we made some estimates by assuming various schedules according to which veterans disposed of their land, and by using some available information on prices. The details of our calculation are in the appendix of Kang and Rockoff (2007). These estimates show that on average the veterans of the Revolution received the equivalent $28 in 1860 dollars in land warrants, veterans of the War of 1812 received $14, and veterans of the Mexican

19 Veterans

of subsequent wars also received help in buying land, but it appears that the Mexican War was the last in which land grants were a prominent component of benefits. Under the Homestead Act (1862) land could be acquired by paying $1.25 per acre or by farming the land for 5 years. Veterans could substitute military service for residence down to 1 year. Veterans of the Civil War also received the right to buy supplemental amounts of land. Many veterans probably made use of their military service in this way. However, the lack of transferability of the benefit limited its value to most soldiers.

46

S. W. Kang and H. Rockoff

War received $137. These estimates again suggest that the veterans of the War of 1812 did less well than either the veterans of the Revolution or the Mexican War.

2.6 Service Pensions: The War of 1812 and the Mexican War Sutherland, after his success with the land grants, began to lobby for a non-meanstested cash pension. He did not live, however, to see it awarded in 1871, 56 years after the end of the war. By this time the ranks of the veterans had been greatly thinned, and hence the potential cost of the program was relatively low. Nevertheless, there was some strong opposition to a non-means-tested pension based on the budget. Senator John Sherman, the Chair of the Committee on Finance, warned that a bill that pensioned all veterans of the War of 1812 would be too expensive; he wanted it restricted to indigent veterans. Sherman mentioned the precedent set by the law of 1832 but dismissed it on the grounds that the veterans of the Revolution had been paid initially with worthless paper. Sherman complained, moreover, that the Revolutionary War pension had produced pension marriages, and wanted the War of 1812 pensions limited to widows who had been married to soldiers at the time of the war. To add all of the widows would add, in Sherman’s view, $5 million to annual expenditures.20 When the law was finally enacted a pension was granted to all soldiers and sailors who had served 60 days—with no requirement that the veterans be in reduced circumstances—but it did limit benefits to widows who had been married to a veteran before the treaty of peace was signed and who had not remarried. In 1878 the service pension for veterans of the War of 1812 was liberalized: only 2 weeks of service in the war were required. This change produced a small increase of about 1200 in the number of survivors on the rolls (Table A1 continued, column 6). There was at the time, we should no longer be surprised, a surplus in the federal budget. Indeed, the federal budget had been in surplus in every year after the Civil War. This was consistent with the policy of deflating prices and returning to the gold standard at the prewar par. Running a surplus was a way of retiring the greenbacks and the federal debt. By the late 1870s the dollar was close to the prewar par, and return to the gold standard was in sight. The monetary case for maintaining a surplus was less compelling, and therefore the case for spending more on the veterans was more compelling. Why did the veterans of the War of 1812 and their heirs do less well than did the veterans of the Revolution? One factor was bad timing. The veterans of the Revolution in “reduced circumstances” won lifetime pensions, as we noted above, 35 years after the end of the Revolution, and a non-means-tested 10 years after that. With the same lags, the veterans of the War of 1812 would have been “due”

20 “Forty-First

Congress; Third Session,” New York Times, Feb 4, 1871, p. 2. Federal expenditures in 1871 were $292 million (Carter et al. 2006, seies Ea585).

2 After Johnny Came Marching Home: The Political Economy of Veterans’. . .

47

for their means-tested pension in 1850 and due for their non-means-tested pension in 1860. The timing was unfortunate. In 1850 the budget was emerging from 3 years in the red. The veterans of the War of 1812, moreover, were in the midst of their fight for land bounties on a par with the veterans of the Mexican War. Indeed, the land bounties that they won, since they were easily convertible into cash, were a substitute for a cash pension. The amounts they received from the land grants, however, were small compared with a cash pension at the Revolutionary rate. This would have been true, even if the public lands had maintained their market value, but the rapid distribution of the public lands depressed the value of federal lands. The second date, 1860, when the veterans of the War of 1812 were due for their non-means-tested pension, falls just before the Civil War. Congress did not put all other projects aside simply because the nation was at war. Nevertheless, pensioning the veterans of the War of 1812 during the Civil War would have appeared injudicious in light of the rapid increase in the number of wounded veterans, widows and orphans on the pension rolls and the inflation and large deficits that marked the Civil War years. The number of pensions compensating for death and disability rose from 7991 in 1860 to 154,477 in 1867 (Table A1, columns 2, 3); the deficit rose from $7 million in 1860 to close to a billion in 1865; and the price level doubled (Carter et al. 2006, series Ea586). So part of the explanation for why the veterans of the War of 1812 received less than the veterans of the Revolution was simply bad timing: Their pension obligations fell due when the federal treasury was stretched to the limit by the Civil War. There may have been something more at work: the perception that American arms had done badly in the War of 1812. Perhaps, as mentioned in our discussion of the land grants, this group of veterans did not deserve as large a “gratuity” as the veterans of more successful wars. It is hard to prove that the lack of success of American arms contributed to the delay in the awarding of the service pension. Few politicians would have admitted to this concern even if it had consciously or unconsciously influenced their thinking. However, it was from the start one of the most controversial wars in American history. The Federalists, the dominant party in New England, along with sizable groups of supporters in the South and West, opposed the war at every turn. Donald Hickey writes that in Congress the Federalists “unanimously opposed the declaration of war in June of 1812, and thereafter they voted against almost every proposal to raise men or money, to foster privateering, or to restrict trade with the enemy” (Hickey 1989, p. 255). To be sure, once the war was over, many Americans adopted the view that the war had been a great moral victory, even if the treaty had restored the status quo antebellum. The Federalists party disappeared as a result of its opposition to the war. Andrew Jackson and William Henry Harrison parlayed wartime victories into the Presidency. Nevertheless, it is possible that some remaining bitterness may have influenced the treatment of the veterans. One piece of suggestive evidence is that the successful fight for land grants in the 1850s was accompanied by an effort to burnish the image of the war. In 1847, just as the Mexican War began, Congressman Charles J. Ingersoll, who proved to be a strong ally of the veterans, published a two-volume revisionist history of the War of

48

S. W. Kang and H. Rockoff

1812. The Ingersoll volumes, soon followed by others along the same lines, argued that the army and navy had fought praiseworthy campaigns and accomplished a great deal for the country (Oberly 1990, p. 30). In the end we probably will never know for certain that the controversial nature of the war and its ambiguous conclusion slowed the awarding of service pensions to the veterans of the War of 1812. But we cannot rule out the conclusion reached by the U.S. President’s Commission on Veterans’ Benefits in 1956 that one factor working against benefits for the veterans of the War of 1812 was that “There was little land fighting during the war, not many men were engaged in any one battle, and it was felt that the army did not accomplish much.”21 The veterans of the Mexican War turned to the National Association of Mexican War Veterans to lobby for their pensions.22 Its founder and leader was Alexander Kenaday, an eccentric and sometimes radical populist labor leader from California. The principal demand was for the same pension, $8.00 per month, which veterans of the Revolution and the War of 1812 had gotten. After persistent lobbying, this pension was granted in 1887 for veterans aged 62 or more. From a fiscal point of view it was a favorable period for awarding a pension. The budget had been in surplus for several years; in 1886 the surplus was about to 39% of expenditures. Thus, veterans of the Mexican War had to wait a shorter period (39 years) for their pension than did the veterans of the War of 1812 (56 years). The wait for veterans of the Mexican War, however, was comparable to that experienced by the veterans of the Revolution: 35 years for a means-tested pension and 45 years for a non-meanstested pension. The awarding of the pension created a period of confusion for Kenaday, who was also the editor of the organization’s newspaper and, not incidentally, a prominent pension agent. He soon turned, however, to the next logical step: an increase in the pension. An increase to $12 per month was finally achieved in 1903. The pensioners got their raise; it should not surprise us, during a period of surplus. From 1894 through 1899 the budget was always in a deficit averaging −9.6% of expenditures per year; from 1900 through 1903 the budget was in a surplus averaging +11.4% per year (Carter et al. 2006, series Ea 584, Ea 585). Thus it appears that the major determinants of the old-age pension for the veterans of the Mexican War were the pensions received by veterans of earlier wars and the state of the federal budget. Public attitudes toward the war, however, did play a role in the timing. The Mexican War, although highly successful from a military point of view and highly profitable in terms of the wealth added to the United States, had aroused opposition from the start. Many Northerners saw it as an imperialist war undertaken by a southern president (James K. Polk) designed to add new slave territories to the United States. Ulysses Grant in a famous passage in his Memoirs 21 U.S. President’s Commission on Veterans’ Pensions. 1956. A Report on Veterans’ Benefits in the

United States, Staff Report No. I: The Historical Development of Veterans’ Benefits in the United States. Printed as U.S. House of Representatives, House Committee Print No. 244, 84th Cong., 2nd Sess, p. 66. 22 The next two paragraphs are based on Davies (1948).

2 After Johnny Came Marching Home: The Political Economy of Veterans’. . .

49

recalled his opposition to the annexation of Texas and to the Mexican War which he regarded as “one of the most unjust ever waged by a stronger against a weaker nation” (Grant 1885, p. 53). Many of the soldiers who had taken part, moreover, were southerners who then fought for the Confederacy during the Civil War. Robert E. Lee was the most prominent example. According to Wallace E. Davies, early efforts to secure a pension for the Mexican War veterans failed during Republican administrations “because of apprehensions that such a measure might benefit many ex-Confederates” (Davies 1948, pp. 226–228). It was not until 1887 when the Southfriendly Democrats were in power—they controlled the White House and the House of Representatives, although not the Senate—that the Mexican War veterans secured their service pension, and even then some former Confederates were disqualified.

2.7 Service Pensions: The Civil War23 The Civil War pension proved highly contentious. The reason was not that the individual pension was significantly larger than the earlier pensions. Controversy on that issue, as we have seen, was minimized by keeping the nominal pension constant. Economists looking for sticky nominal prices would have to look hard to find a better example. As can be seen in Fig. 2.2, the post-Civil War deflation raised the purchasing power of the standard $8 per month pension, measured with the consumer price index. But even at the end of the century it was no more than on a par with the level in the late antebellum period. In terms of wage units the standard pension, as shown by the lower line in Fig. 2.2, was worth less in the postbellum era. In 1818, to put it somewhat differently, the standard annual pension could hire a farm laborer for 10 months; but by 1899 it could hire a farm laborer for only 7 months.24 Deflating the pension by the wage of unskilled labor is relevant to the actual pattern of consumption of many veterans. As the veteran aged he spent more of his income, in many cases, on services such as nursing care and less on goods. The Civil War pension system was often criticized because it led to corrupt “pension marriages.” An elderly veteran would marry a young woman simply so that she would be eligible for a pension. One defense offered for the pension marriages was that they were really part of an exchange that benefited the veteran. The veteran received nursing care from a young woman for which the veteran was unable to pay out of his current income (because it was fixed in nominal terms and wages had risen), and in return the young woman entered into a pension marriage and received the pension after the veteran died. 23 We

discuss only the pensions for Union veterans. Confederate veterans did not receive federal pensions. (Except for the last surviving Confederate veteran, who was pardoned and given a lifetime pension—at age 105.) The Confederates, however, did receive pensions from southern states. These are described in Ratchford and Heise (1938). Short (2001) explores the (small) effect of these pensions on the retirement behavior of Confederate veterans. 24 The monthly earnings with board for farm laborers is from Carter et al. (2006, series Bc 4234– 4243).

50

S. W. Kang and H. Rockoff

Rather than the amount per veteran, the real sources of contention were the large number of veterans who received pensions, which meant that transfers to veterans materially affected taxes, and the North-south divide. As shown in Table 2.1 Union forces amounted to 6.84% of the 1861 population. This percentage meant that the cost of the pensions was visible to the average taxpayer. The federal government, moreover, was financed to a substantial extent by tariffs. Southerners, therefore, saw themselves as twice abused. First they paid high tariffs on imported goods, a tariff that protected northern industries from competition, but made it harder for southerners to export cotton and other cash crops. Then the proceeds of the tariff were distributed to union veterans and their families. In 1895, for example, tariffs accounted for 47% of total revenues; and pensions (mainly Civil War pensions) accounted for 40% of total federal expenditures.25 The legislation providing pensions for the Civil War veterans evolved along much the same lines as the legislation providing pensions of the veterans of earlier wars. A general law of 1862 provided a pension at the rate of $8–$30 per month (depending on rank) for totally disabled soldiers, or to the widows, orphans, or mothers of soldiers killed in battle. Subsequent legislation refined the table of benefits, creating specific benefits for a long list of disabilities. Two pieces of legislation were crucial to the postwar expansion of the Civil War pension: the Arrears Act of 1879, and the Dependent Pensions Act of 1890. The Arrears Act became law in January 1879. Many soldiers had been placed on the rolls and started receiving benefits from dates well after the end of the war. There was, as many veterans and their advocates pointed out, a potential inequity here. A soldier who had shown great courage in battle might be receiving less than another soldier with an inferior military record simply because the first soldier had been slow to apply for a pension out of pride or lack of understanding. The Arrears Act would make good this inequity by paying a soldier who qualified for a pension an additional amount to cover what he was due from the time of his discharge to the beginning of his pension. The Arrears Act, moreover, applied not only to those currently on the rolls but also to those who would be added in the future. There was some uncertainty about the cost of the Arrears Act, and the plausibility of a relatively low estimate may have aided passage of the act. At a cabinet meeting at which the bill, having already passed both houses, was discussed, the Secretary of the Treasury, John Sherman, offered a figure of $150 million as the cost of the act, while Carl Schurz, the Secretary of the Interior, offered a figure of $50 million based on an estimate by the Pension Office. President Hayes expressed some concern about the cost of the additional pensions, but signed the bill.26 The cost of the Arrears Act eventually reached $200 million.27

25 U.S.

Bureau of the Census. 1975. Historical Statistics of the United States, Colonial Times to 1970, Bicentennial Edition. Washington, DC: Government Printing Office. 26 New York Times, “Yesterday’s Cabinet Session,” January 22, 1879, p. 1 27 U.S. President’s Commission on Veterans’ Pensions. 1956. A Report on Veterans’ Benefits in the United States, Staff Report No. I: The Historical Development of Veterans’ Benefits in the United

2 After Johnny Came Marching Home: The Political Economy of Veterans’. . .

51

Skocpol has argued that the Arrears Act was not as closely associated with a bulging surplus in the federal treasury as was the law of 1890 (discussed below) (Skocpol 1993, pp. 98–99). Nevertheless, some account should be taken of the budget. The budget was in surplus, and had been for a number of years. The surplus, moreover, as we noted in our discussion of the War of 1812 pension, was no longer needed to deflate the price level because the United States had returned to gold at the prewar par. Although we should not make too much of the coincidence in timing, the return to the gold standard was accomplished on January 1, 1879 and the Arrears Act was passed by Congress on January 25. The Dependent Pensions Act of 1890, as most historians have recognized, clearly owed its existence to large budget surpluses. The law pensioned all veterans who had been honorably discharged and who now suffered from a disability that prevented them from performing manual labor.28 There was no requirement, as there had been under previous laws, that the disability be the result of wounds suffered during the war. The law was not technically a service pension. It would not be until 1907 that a pure service pension became law. The Pension Office, however, was instructed to grant a pension to all veterans who were 65 or older unless they were unusually vigorous, so in effect the law of 1890 was close to a service pension. The notion that the law of 1890 amounted to a service pension has been widely accepted (Costa 1995, pp. 300–301). Figure 2.6 shows that the spending that resulted from this law

S200,000,000

S150,000,000

S100,000,000

S50,000,000

S0 1885

1888

1887

1888

1889

1890

1891

1892

1893

1894

1895

-S50,000,000

-S100,000,000 Total Pensions

Law of 1890 Pensions

Surplus

Fig. 2.6 Service pensions for veterans of the civil war and the surplus, 1885–1895. Source: Kang and Rockoff (2007, Appendix, Table A2) and Carter et al. (2006, Series Ea584, Ea585) States. Printed as U.S. House of Representatives, House Committee Print No. 244, 84th Cong., 2nd Sess, p. 81. 28 The infirmity could not be the result of the veteran’s “own vicious habits.”

52

S. W. Kang and H. Rockoff

(and from previous liberalizations) was sufficient to erase the surpluses that had persisted in the second half of the 1880s. The close connection between the bulging surpluses in the federal treasury, which were produced by the tariff, and the liberalization of the pension in the early 1890s was no secret. Republican politicians made their support for generous pensions and high tariffs clear. As the Republican Party platform put it in 1888: The legislation of Congress [on pensions] should conform to the pledges made by a loyal people and be so enlarged and extended as to provide against the possibility that any man who honorably wore the Federal uniform shall become the inmate of an almshouse, or dependent upon private charity. In the presence of an overflowing treasury it would be a public scandal to do less for those whose valorous service preserved the government.

The next, and last, plank in the platform invited “. . . all workingmen, whose prosperity is seriously threatened by the free-trade policy of the present Administration,” to vote for the Republican candidate.29 After the election President Benjamin Harrison followed through on Republican promises. His first appointment to head the Pension Bureau was James Tanner a legless Union veteran and who soon became a political liability. In one of his bombastic speeches, Tanner famously promised that as Pension Commissioner he would raise the pensions of veterans receiving only a pittance—a veteran might receive less than a dollar a week for a small disability—even if his decisions raised from some lips the prayer: “God help the surplus” (McMurry 1926, pp. 345–347). Green B. Raum, a staunch but more discreet friend of the veteran replaced Tanner. The erosion of the surplus may have affected the next piece of legislation, which although minor in terms of fiscal effects, is nevertheless revealing. In 1892 Congress provided pensions for veterans of the Indian Wars. The youngest veterans who became eligible were veterans of the Seminole wars between 1832 and 1842. In other words, the minimum time that veterans of the Indian Wars had to wait was 50 years (1842–1892): Veterans of later engagements with Native Americans were not yet old enough for a pension.

2.8 Some Tentative Generalizations Veterans’ benefits varied substantially from war to war in the nineteenth century and were usually the subject of bitter controversy. Veterans of the War of 1812 received less than did veterans of either the Revolution or the Mexican War. Union veterans of the Civil War did well compared with veterans of earlier wars, but only after a contentious political debate.

29 Republican Platform 1888, American Presidency Project, http://www.presidency.ucsb.edu/index.

php. The Democratic platform noted that the Cleveland Administration had paid out more in pensions than any previous administration, but made it clear that the Democrats intended to cut the tariff and other Civil War taxes.

2 After Johnny Came Marching Home: The Political Economy of Veterans’. . .

53

Several factors account for the differences across wars. Veterans’ benefits evolved, first of all, along a path-dependent process. The precedent set by past policies shaped what was considered a fair reward for service. Veterans of any given war would point to how veterans of previous wars were treated and demand equity. Veterans who served in the militia would point to veterans who had served in the regular army and demand equity. Privates and noncommissioned officers could point to the awards made to commissioned officers and demand equity. Opponents of generous benefits, however, could point to what they regarded as abuses of previous systems and demanded greater economy. We would like to believe that feelings toward veterans were independent of the war they fought. There are some suggestions in the historical record, however, that the politics of a war influenced how and when its veterans were rewarded. This is an inherently difficult proposition to prove because the it is not the sort of thing that people involved in the political process are likely to talk about. Nevertheless, we can point to several possible examples: (1) In the years immediately following the Revolutionary War, veterans of the militia were considered the real heroes, and veterans of the Continental Army were sometimes perceived as lesser soldiers, as mere mercenaries. With the rise of the image of the suffering soldier, associated especially with Valley Forge, however, pressure built for cash benefits for veterans of the Continental Army. (2) Attempts were made to add veterans of campaigns carried out against Native Americans in the wake of the Revolution to the service pension rolls, but these efforts ran into opposition based partly on the contention that these were wars carried out for private gain. (3) The veterans of the War of 1812 received less in cash and land than either the veterans of the Revolution or the veterans of the Mexican War. There were a number of reasons, but one may have been that the war itself was politically divisive, and went badly for the United States. (4) The Mexican War was much more successful from a military point of view than the War of 1812. Nevertheless, in the 1880s the veterans of the Mexican War were frustrated on some occasions in their campaign for a service pension for political reasons. Republican Congresses were unsympathetic to what they regarded as a southern war undertaken by a southern president to expand slavery, and fought by men who later served in the Confederate military. (5) Confederate veterans, of course, received nothing from the federal government. Even pensioners from antebellum wars who lived in the South during the Civil War had to prove their loyalty during the Civil War to remain on the rolls.30 We would also like to believe that veterans’ benefits are independent of fiscal politics. After all, since its founding the United States has been a wealthy nation compared to most other nations, and has always been in a position to reward veterans generously whatever the current balance of taxes and spending. Nevertheless, in case after case, the state of the treasury was an important determinant of the level of benefits: When federal coffers were full, claims for additional benefits were agree to; when coffers were empty, claims for additional benefits were rejected.

30 U.S.

Senate, 1867. “Report of the Commissioner of Pensions.”

54

S. W. Kang and H. Rockoff

Here are some examples. (1) The officers of the Revolutionary Army asked for half-pay for life, the practice in the British army. The American officers, however, were forced to settle for bonds worth 5 year’s pay at face value. Under the Confederation, moreover, Congress failed to pay the interest on the bonds, and many former officers parted with their bonds at a fraction of their face value. (2) In 1818, however, after a 2-year run of treasury surpluses, The Continental Army’s officers and common soldiers in “reduced circumstances” received a lifetime pension. In 1828 and in 1832, after another run of surpluses that began in 1825, pensions were awarded to veterans without regard to their economic status, and in 1836, after the last and largest of these surpluses; widows of ordinary soldiers were pensioned, even when the marriage took place after the war. (3) The veterans of the War of 1812 were not successful in winning postwar rewards until the Mexican War created the precedent for a generous land bounty and, not incidentally, the means—the vast lands won from Mexico—for awarding it. The veterans of the War of 1812 eventually qualified for a cash service pension in 1871 in the middle of a run of double-digit surpluses. Even in 1871, moreover, 56 years after the end of the war, there was some sentiment for means-testing the pension to save expenses. (4) The Arrears Act of 1879, which increased the incentives for Civil War veterans to apply for pensions, also was passed during a period of surpluses, and when the surpluses could no longer be defended on the grounds that they were needed to deflate prices and restore the gold standard. (5) The Dependent Pensions Act of 1890 was passed during a period of large treasury surpluses, and the connection with the budget was explicit. The Republicans promised to support a high tariff and to expand pension benefits. The association between budget surpluses and veterans’ pensions that we have identified for the nineteenth century is less surprising than it would be in the twentieth century because surpluses were common. Indeed, between 1800 and 1899 the federal budget was in surplus in 69 out of 100 years, and in most of those years the surplus was fairly large. Nevertheless, one can show that there was a strong association between surpluses and pension legislation. Perhaps the simplest calculation that illustrates the association between budget surpluses and pension legislation is the following. If the chance of a surplus was 0.69 the probability that all nine major pieces of pension legislation that we have identified (Table 2.2) would have fallen by chance in surplus years was only about 3.5% (0.699). If we confine our attention to periods in which there was a run of years with surpluses an even clearer association between surpluses and pension laws emerges. The probability that by chance all nine pieces of legislation would have fallen into a year in which the budget was in surplus and had been in surplus in the two previous years was only 0.20% (0.509). A more formal method of making the same point is to estimate a regression in which the dependent variable is a binary variable that takes the value one in years when legislation expanding benefits for veterans was passed and the independent variables are current and lagged values of the surplus in the federal budget measured as a percentage of revenues. We ran several regressions of this sort. We started with both the surplus and the first lagged value of the surplus. We dropped the former because it was not significant, although it was signed correctly. The result was the following.

2 After Johnny Came Marching Home: The Political Economy of Veterans’. . .

55

Table 2.2 Pension legislation in the nineteenth century Year 1818 1828 1832 1836 1855 1871 1879 1887 1890

Purpose of legislation Service Pensions for veterans who served in the Continental Army and who were in “reduced circumstances” Full pay for life granted to certain officers and to noncommissioned officers and privates who served in the Continental Army until the end of the war Service Pensions for veterans of the Revolution who served in the Continental Army or in a militia, whether or not in reduced circumstances Pensions for widows of veterans of the Revolution Land grants for veterans of the War of 1812 Service Pensions for veterans of the War of 1812 The Arrears Act awarded pensioners a lump sum to cover the time between the end of the war and the time when they qualified for a pension Service pensions for veterans of the Mexican War Dependent Pension Act comes close to being a service pension for veterans of the Civil War

Source: The narrative in Sects. 2.3–2.7

Legislation = −2.21 + 0.38 ∗ Surplus(−1)

(2.1)

This was a probit regression with a MccFadden R2 of 0.20. The z-statistic on Surplus(-1) was 2.36, which indicates that Surplus(-1) was significant at the 0.02 level. Adding additional lags of the surplus did not yield significant variables or improve the fit. The results were similar when logit or extreme value methods were used to estimate the equation. Evidently, as both the narratives and the equations suggests there was a strong association between the condition of the federal treasury and pension legislation for veterans. One might hope that decisions on veterans’ benefits would be above politics. In the nineteenth-century, at least, the opposite was true. Affection for the veterans as individuals often took second place to the politics of war and to current budget realities. Acknowledgements We are very grateful for the comments on previous drafts that we received during presentations at the National Bureau of Economic Research Summer Institute (Cambridge, July 2006), the International Economic History Meetings (Helsinki, August 2006), and the Southern Economic Association Meetings (Charleston, November 2006). We owe a special debt to John James and Robert Whaples who were the discussants on the latter occasion. Lloyd DeWitt Bockstruck and Lee Craig graciously took time to answer questions about their areas of expertise. Daniel Fleisch and Adam Yaari provided thoughtful research assistance.The idea for this paper emerged from a conversation that one of us had many years ago with Lawrence Fisher of Rutgers University, Newark.

56

S. W. Kang and H. Rockoff

References Bockstruck LD (1996) Revolutionary war bounty land grants. Genealogical Publishing Company, Baltimore Bodenger RG (1971) Soldiers’ bonuses: a history of veterans’ benefits in the United States, 1776–1967. Pennsylvania State University, Philadelphia Carter SB, Gartner SS, Haines MR, Olmstead AL, Sutch R, Wright G (2006) Historical statistics of the United States: Millennial edition, vol 3. Cambridge University Press, Cambridge Clark RL, Craig LA, Wilson JW (2003) A history of public sector pensions in the United States. University of Pennsylvania Press, Philadelphia Costa DL (1995) Pensions and retirement: evidence from union army veterans. Q J Econ 110(2):297–319 Davies WE (1948) The Mexican War veterans as an organized group. Miss Val Hist Rev 35(2): 221–238 Ferguson EJ (1954) Speculation in the revolutionary debt: the ownership of public securities in Maryland, 1790. J Econ Hist 14(1):35–45 Flexner JT (1974) Washington: the indispensable man. New American Library, New York Gates PW, Swenson RW (1968) History of public land law development. US Government Printing Office, Washington, DC Glasson WH (1918) Federal military pensions in the United States. Oxford University Press, New York Grant US (1885) Personal Memoirs of US Grant. Charles Webster and Company, New York Grubb F (2005) The net asset position of the US national government, 1784–1802: Hamilton’s blessing or the spoils of war? National Bureau of Economic Research Working Paper No. 11868 Hibbard BH, Gates PW (1924) A history of the public land policies. The University of Wisconsin Press, Madison Hickey DR (1989) The war of 1812: a forgotten conflict. University of Illinois Press, Urbana Kang SW, Rockoff H (2007) After Johnny came marching home: the political economy of veterans’ benefits in the nineteenth century. National Bureau of Economic Research Working Paper No. 13223 Kohn RH (1970) The inside history of the Newburgh conspiracy: America and the coup d’etat. William Mary Q 27(2):188–220 McMurry DL (1926) The bureau of pensions during the administration of President Harrison. Miss Val Hist Rev 13(3):343–364 Oberly JW (1990) Sixty million acres: American veterans and the public lands before the civil war. Kent State University Press, Kent Ratchford BU, Heise K (1938) Confederate pensions. South Econ J 5(2):207–217 Resch JP (1999) Suffering soldiers: revolutionary war veterans, moral sentiment, and political culture in the early Republic. University of Massachusetts Press, Amherst Short JS (2001) The retirement of the rebels: georgia confederate pensions and retirement behavior in the New South. Indiana University, Dissertation Skeen CE, Kohn RH (1974) The Newburgh conspiracy reconsidered. William Mary Q 31(2): 273–298 Skocpol T (1992) Protecting soldiers and mothers: the political origins of social policy in the United States. Harvard University Press, Cambridge Skocpol T (1993) America’s first social security system: the expansion of benefits for Civil War veterans. Polit Sci Q 108(1):85–116 Teipe EJ (2002) America’s first veterans and the revolutionary war pensions. Edwin Mellen Press, Lewiston Williamson SH (2004) An index of the wage of unskilled labor from 1774 to the present. EH.net

Chapter 3

Asian Exclusion in American Immigration Policy Zachary Gochenour

Abstract The closing of the United States to immigrants is arguably the most economically and socially significant policy shift in American history. The U.S. had virtually open borders until 1875, when the first of a series of federal laws prohibiting or limiting immigration of particular groups was passed. The first such group was Asian immigrants, mostly Chinese, who were excluded by a series of bills in the late nineteenth century. Using data from the U.S. Congressional record, I attempt to explain the policy shift in public choice terms: identifying voting patterns that can be explained by shifts in public and elite opinion, the incentives of policymakers, and changing economic conditions. Explanations of the policy shift from previous scholarship are evaluated in light of roll-call voting data and NOMINATE scores.

3.1 Introduction How did the United States come to limit immigration? Almost all the original European immigration to the United States was from Great Britain and a few other Western European countries. After the colonial period there was relatively little immigration until 1830, but economic opportunities in rapidly growing American cities like New York and Chicago, as well as frontier opportunities in railroad work and mining brought immigrants by the millions. Immigration totaled only 8385 in 1820, but had grown to 1,713,000 in 1850, mostly consisting of individuals of Irish, German, British, and French origin (Dolan 2010). The invention of steam powered ships lowered fares for trans-oceanic travel, eventually bringing immigrants from Italy, Eastern Europe, and Asia. Until the 1850s, there was no major organized political resistance to immigration. Federal laws restricting immigration took effect in 1875, and by 1924 immigration from most of the world was strictly capped by quotas based on national origin. The policy change was drastic and effectively

Z. Gochenour () James Madison University, Harrisonburg, VA, USA © Springer International Publishing AG, part of Springer Nature 2018 J. Hall, M. Witcher (eds.), Public Choice Analyses of American Economic History, Studies in Public Choice 37, https://doi.org/10.1007/978-3-319-95819-4_3

57

58

Z. Gochenour

changed demographic composition of the country, representing the most economically and socially significant shift in American political history and sparking an intense debate that continues to this day. Historical scholarship on immigration emphasizes the role of racism in support for immigration restriction (Dolan 2010; Peffer 1986; Kennedy et al. 2002; Van Nuys 2002). This explanation has some intuitive appeal consistent with the legislative history: most immigration acts targeted specific ethnicities and countries of origin, starting with immigrants whose looks and customs would have been most foreign to Anglo-American citizens of the nineteenth century such as those from China and Japan.1 The immigrants during the American Gilded Age were also much more likely to be non-Protestant and even non-Christian. The narrative which emphasizes xenophobia, however, gives little insight into how these fears were translated into policy. Why did the restrictions take the form they did? What explains the timing of the restrictions—xenophobia was nothing new: nativist sentiment and political action predate the passage of the first federal immigration law by at least three decades. Restriction took almost three quarters of a century to fully implement, and despite the severity of the policy shift, immigration remained high during most of the period. Following Timmer and Williams (1998), Goldin (1994) and others, this article investigates what other factors might explain the timing and content of this policy shift. Particularly we will focus on the legislative push for Chinese Exclusion, the first major policy goal of the anti-immigration movement in the United States. Was there a role for economic conditions and public finance, as at least one economic historian has suggested (Kanazawa 2005)? Following the lead of researchers in public choice, recent research by Gilens and Page (2014) has emphasized the role of elite opinion and special interests in determining U.S. policy. Could elite opinion and special interests have been the real force behind the policy shift? If so, which interests and groups stood to gain the most from changing policy? The contribution of this paper is to fuse economic, demographic, and political explanations and judge the relative merits of each using roll-call voting data from the U.S. Congress. The paper is organized as follows: Sect. 3.2 describes the legislative history leading up to the Chinese Exclusion Act and its renewals. Section 3.3 describes the data used in my analysis: DW-NOMINATE (Poole and Rosenthal 1997, 1984, 2007) scores and the U.S. Congressional record for topics related to federal immigration policy. Section 3.4 continues the analysis and compares it to other explanations given by historians and economists. Section 3.5 concludes.

1 Some

historians emphasize that the racism was driven in part by economic competition. See, for example, Saxton (1975, pp. 72–75), Mann (1982, pp. 33–56), Daniels (2011, pp. 33–56), Marks (1994, pp. 300–301), and Takaki (1989, pp. 92–112).

3 Asian Exclusion in American Immigration Policy

59

3.2 Legislative History Despite the promise of the Burlingame Treaty,2 the first federal immigration law was the Page Act of 1875.3 The law classified as “undesirable” any individual from Asia who intended to become a contract laborer. It imposed a $2000 fine and maximum 2-year jail sentence for traffickers who brought people from Asia without their full and voluntary consent for the purposes of holding them to a term of service. The law was named after its sponsor, Representative Horace F. Page, a Republican who introduced it to “end the danger of cheap Chinese labor and immoral Chinese women.” The large number of Chinese laborers the Western United States, who usually came without their families intending to return to China, resulted in high demand for prostitutes, which infuriated natives (Peffer 1986). Other justifications given for the law include the American Medical Association’s belief that Chinese immigrants “carried distinct germs to which they were immune, but from which whites would die if exposed” (Luibhéid 2002). Historians have generally attributed the success of the law to American concern for maintaining social ideals of marriage and morality (Abrams 2005). The act sailed through Congress with little opposition. The ultimate effect of the law would be to severely limit the supply of Chineseborn women in the United States. In 1882 alone, during the few months before the enactment of the Chinese Exclusion Act of 1882 and the beginning of its enforcement, 39,579 Chinese entered the U.S., and only 136 of them were women (Abrams 2005). How did the U.S. get to the point where it would consider banning immigration from an entire country, and later most of the world? Congress considered immigration issues very sparsely between the Alien and Sedition Acts (1798) and the Civil War. There was briefly a policy of encouraging immigration; Lincoln submitted in his address to the 38th Congress: “establishing a system for the encouragement of immigration. . . there is still a great deficiency of laborers in every field of industry.” Lincoln later said that he regarded immigrants as a replenishing stream which were “appointed by Providence to repair the ravages of internal war, and its waste of national strength and wealth.”4 Opposition to these pro-immigrant sentiments were concerned primarily with alien criminality. The principal group of aliens attracting attention for criminality were the Chinese, called “coolies.”5 An 1867 bill resolving “this Government. . . and all its agencies” to prevent the coolie trade was adopted with no debate, possibly because

2 Signed

in 1868, this treaty established friendly relations, and open immigration, between China and the United States. In fact, Chinese immigration was specifically encouraged by the treaty. The policy was officially reversed when the treaty was renegotiated in 1880. 3 An Act Supplementary to the Acts in Relation to Immigration. Sect. 141, 18 Stat. 477, 1873March 1875. 4 38th Congress, 2nd session. 5 The word “coolie” has a contentious etymology, but is generally thought to mean “worker” or “peon,” usually used in a pejorative sense.

60

Z. Gochenour

of linking the coolie trade to the slavery issue (Hutchinson 1981). The problem for immigration restriction proponents is that the federal government had little authority or capacity to regulate immigration. In late 1867, Congressman Johnson of California offered a resolution of inquiry to the House, asking “whether the Congress can by legislation prevent the immigration and importation of Chinese and other inferior races to our country.” Objection was raised and the resolution was not adopted, but this began a series of Congressional argumentation about limiting immigration from “inferior” races that would continue through the twentieth century. Through the 1860s and 1870s, several bills and resolutions were considered for the exclusion of Chinese laborers from American society. One bill was even suggested to make the importation of all immigrants under any labor contract unlawful (Hutchinson 1981). Most bills restricting immigration were referred to committee and remained (died) there. By the time of the 42nd Congress (1871–1873), most explicitly pro-immigrant sentiment had evaporated in Congress. President Grant encouraged Congress to legislate for the health and safety of immigrants arriving by steamship, which most immigrants did.6 Immigration restriction language in Congress was almost always couched in language to make it seem good and necessary for the benefit of the immigrants themselves. There is little evidence, though, that most or even a large fraction of Chinese immigrants were brought against their will. Many were brought not on long-term service contracts but borrowed money from brokers to pay for their trip (Abrams 2005). Work was plentiful in the American West, particularly on railroads, especially compared to the wages available at the time in the Chinese Empire. This “place premium” has long been—and continues to be—a major factor influencing immigrant flows (Clemens 2011). The Chester A. Arthur administration saw a flood of immigration bills, mostly related to Chinese immigration. Arthur himself recommended to Congress that they should be “considerate of the interest and susceptibilities of the Chinese government in any modification of the immigration laws” (Hutchinson 1981). Among those groups petitioning Congress for further limitation of Chinese immigration were the San Francisco Board of Trade, the New York Union League Club, and the Milwaukee Trades Assembly. All these organizations cited economic pressure on native workers as a major reason to restrict Chinese immigration. In 1880, the Democratic party convention plank called for a complete restriction of Chinese immigration, exceptions on for travel, education, and foreign commerce. The Republican party only called for restrictions for Chinese criminals and those “likely to become a public charge.”

6 In a message to Congress on December 4, 1871, he wrote “The number of immigrants, ignorant of

our laws, habits, etc., coming into our country annually, has become so great, and the impositions practiced upon them so numerous and flagrant, that I suggest congressional action for their protection.”

3 Asian Exclusion in American Immigration Policy

61

The Chinese Exclusion Act, signed into law on May 8, 1882, was the second federal immigration law passed. Congress passed the bill over Arthur’s veto.7 It was originally written to limit immigration from China for a period of 10 years, though it was renewed in 1892 and made permanent in 1902.8 The act provided deportation instructions for illegal Chinese laborers, effectively creating the first illegal immigrants to the United States. Many Chinese had come to the United States to work in railroad construction and gold mining (Norton 1924), but the end of the gold rush reduced labor demand. Many of these laborers sought work in other industries, and the increased competition was the stated impetus for restrictive legislation. Labor leader Denis Kearney and his Workingman’s Party, as well as by California Governor John Bigler, blamed Chinese coolies for depressed wage levels and called for federal solutions.9 Kanazawa (2005) argues that Chinese immigrants were important taxpayers when both the state and localities were experiencing major fiscal difficulties starting in the 1850s, making it difficult to build support for immigration restriction until the fiscal situation improved.10 The most vocal critic of the act was the anti-slavery and anti-imperialist Republican Senator George Frisbie Hoar of Massachusetts, who described the Act as “nothing less than the legalization of racial discrimination.” The act received considerable support from the nascent labor unions in the United States, such as the Knights of Labor (Kennedy et al. 2002). The year 1882 also saw the passage of another piece of federal immigration legislation which established a head tax on immigrants, meant to pay for the expanding federal immigration bureaucracy. Prior to the 1880s, immigration was primarily handled by state governments and city port authorities. The 1882 act also created a new group of immigrants subject to restriction, those “likely to become a public charge,” which would leave a lasting legacy on the immigration debate. The bureaucracy was further expanded by the Immigration Act of 1891 which established a Bureau of Immigration within the Treasury Department whose role was to oversee and enforce federal immigration policies. In 1885, Canada passed a similar law called the Chinese Immigration Act, placing a head tax on Chinese immigrants and directing the proceeds toward the development of an immigration bureaucracy. The Scott Act (1888) expanded on the Chinese Exclusion Act by prohibiting re-entry of deported Chinese. The Chinese Exclusion Act was extended in 1892 with another law known as the Geary Act, sponsored by Democratic California Congressman Thomas Geary. The act was not only extended for another 10 years, but several new onerous

7 Arthur

claimed the bill was a repudiation of treaty obligations to the Chinese Empire. was finally repealed by the Magnuson Act of 1943. 9 The expressed goal of the California Workingman’s Party was to “rid the Country of cheap Chinese labor” according to Kearney (1878), and their primary enemies were Chinese immigrants and the Central Pacific Railroad. Their party slogan was “The Chinese must go!” 10 Kanazawa (2005) reports that in several years during the 1850 and 1860s, taxes on foreign miners represented over 15% of total state revenues. 8 It

62

Z. Gochenour

requirements were added for Chinese nationals living in the U.S., such as a requirement to carry a residency permit at all times or risk deportation. The law was extremely unpopular among Chinese workers, and a protest was led by the Chinese Consolidated Benevolent Association encouraging workers not to register, to refuse to pay for transportation costs if they were deported, and to issue several legal challenges in court.11 The first immigration restriction not to specifically target Chinese workers was the Immigration Act of 1903, also called the Anarchist Exclusion Act. It codified previous immigration law, and added four inadmissible classes: anarchists, people with epilepsy, beggars, and importers of prostitutes. Legislative focus returned to Asia with the passage of the Asiatic Barred Zone Act of 1917. Eventually, by 1924, immigration from most of the world would be restricted by strict quotas based on national origin. This quota system remained in effect until 1965.

3.3 Data A comprehensive analysis of any major program must include more than just the history of successful legislation. Most immigration acts that were ultimately passed faced little Congressional opposition in the final vote. However, final votes show only a small portion of the debate. Many acts went through multiple votes before reaching a consensus in Congress. To this end, I have used roll call voting information collected by Poole and Rosenthal for the NOMINATE (Nominal ThreeStep Estimation) and VoteView projects (Poole and Rosenthal 1997) which are taken directly from the U.S. Congressional record. This data tracks yea or nay responses for almost every Congressional roll-call vote.12 Later I will discuss the use of NOMINATE scores, which enable us to see how voting behavior on particular issues across both time and chambers of Congress are correlated with each other. The analysis begins with the 37th Congress and ends with the 51st (the interval 1862–1892). During this time, the House of Representatives and Senate each rollcall voted four times on the issue of immigration. Of course, several other bills and proposals were heard and most died in committee. Another important thing to mention is that not all votes in Congress are not roll-call votes, even though they are all matters of public record. There are voice votes (asking the chamber to say “aye” or “no”) and division or standing votes (where the presiding officer counts Members), and these types of votes do not indicate by name how a member voted. 11 The

Geary Act was ultimately upheld by the United States Supreme Court in an opinion by Justice Horace Gray, Fong Yue Ting v. United States, 149 U.S. 698, 13 S. Ct. 1016. 37 L.Ed. 905 (1893), Justices David Josiah Brewer, Stephen J. Field, and Chief Justice Melville Fuller dissenting. The majority opinion states that the U.S., as a sovereign nations, has the right to exclude any non-citizen for any reason. 12 A small number of Congressional roll-call votes, especially in the pre-Civil-War period, have been lost.

3 Asian Exclusion in American Immigration Policy

63

Roll call voting is usually used when the vote is expected to be close or is otherwise a key vote in Congress. For this analysis, since we are interested in how individual members of Congress from various regions voted, I will only use roll-call votes.

3.3.1 NOMINATE NOMINATE is a multidimensional scaling application developed by the political scientists (Poole and Rosenthal 1997) to analyze preferential and choice data, such as legislative roll-call voting behavior. Several forms of NOMINATE exist, but current applications use the Dynamic Weighted NOMINATE score which allows comparisons to be made between houses and over time using consistent scores. This is particularly useful in tracking the role of ideology and party coalitions in shaping policy. NOMINATE assumes that alternative choices can be projected on a basic, low-dimensional Euclidian space. Within that space, individual legislators have utility functions which are normally distributed and maximized at their ideal point. NOMINATE also assumes that individuals also have symmetric, single-peaked utility functions which center on their ideal point. Ideal points are determined from observing choices, with individuals exhibiting similar preferences placed more closely than those behaving dissimilarly. Legislators are given scores which place them closer to other legislators who have similar ideal points. NOMINATE is useful because it gives quantitative measures of Representatives’ and Senators’ ideology across chambers and across time. Errors in prediction can also be quite valuable because they are indicative of cases where individual voters went against party lines and voted in a manner inconsistent with the preferences they revealed through other votes. A key finding from NOMINATE and other literature in political science is low-dimensionality in Congress (Poole and Rosenthal 1997). Most votes can be predicted by a single, unidimensional score on a continuum of liberal-conservative economic policy, despite the many complexities of Congressional politics. Interestingly, there is often a second-dimension that represents some salient social issue of the day. During this period, the social dimension is related to racial issues such as slavery or civil rights, or regional issues related to the Civil War and Reconstruction (Poole and Rosenthal 1997). The second dimension is highly context dependent. Today, the most important second dimension is probably religiosity or social conservatism; in Europe, the second dimension is usually conceived as proor anti-integration with other EU member nations (Szczerbiak and Taggart 2008). An important part of my analysis will be to see how much strength of ideology in either dimension affected votes on immigration and naturalization related issues in the period. Previous work has not explored the role of ideology and political polarization in the development of immigration policy.

64

Z. Gochenour

3.4 Analysis Most previous scholarship in economics on this topic has focused on market factors (Timmer and Williams 1998; Goldin 1994; Shughart et al. 1986) in determining the timing and content of immigration legislation. Timmer and Williams (1998) conclude that there is no consistent evidence that macroeconomic conditions as measured by growth and unemployment had consistent influence on shifts in policy.13 Relative wages and income inequality appears to have been a more important, but weak, determinant. They conclude that over the long run, the United States and other countries try to maintain the relative economic position of unskilled labor compared to skilled labor, landowners, and industrialists. Foreman-Peck (1992) defends claims that the 1830s anti-immigrant movement was counteracted by the Democratic Party’s recruitment of immigrant voters, and immigration restriction was heavily supported by native voters but floundered in areas where immigrants could not or did not vote. Shughart et al. (1986) claim that immigration enforcement is lenient during times of wage and employment growth while it is more strictly enforced during times of economic contraction. Goldin (1994) echoes this sentiment, concluding that immigration restrictions were held at bay during the largest immigrant flows because the new immigrants were able to capture many Congressional districts. She also suspects that the South became anti-immigrant after 1900 when they “realized immigrants would never settle there.” I use a binomial logistic multivariate regression model of the form: VoteToRestrict = α + β1 Urbanization + β2 ForeignBorn +β3 ImmigrantGrowth + β4 Democrat + β5 LiberalConservative +β6 RacialIssues + 

(3.1)

Urbanization data is on a per-state basis and is included due to its inclusion in Goldin (1994) which suggested it may be an important variable. Foreign-born and immigrant growth are both taken from U.S. Census data. Democrat is a dummy variable indicating whether the voting member of Congress was part of the Democratic Party. The final two regressors, liberal-conservative and racial-issues, are DW-NOMINATE scores for each voting member of Congress at the time of the roll-call. Table 3.1 shows the results of this analysis for votes that passed, while Table 3.2 presents the results from bills that failed. Both DW-NOMINATE scores are shown to be statistically and economically significant predictors of voting in favor of

13 Macroeconomic

conditions considered are real wage growth, growth in real GDP, unemployment, real wage effects, relative wages of unskilled to per capita GDP, trade openness, human capital content of immigrants, rate of immigration—all with lagged indicators. The authors do find that changes in real wages are a significant explanatory variable in accounting for the Congressional vote to override Wilson’s veto in 1917.

3 Asian Exclusion in American Immigration Policy

65

Table 3.1 Explaining congressional voting on key immigration bills that passed, 1862–1892 Date Chamber Vote (Yea-Nay) Urbanization Foreign born Immigration growth Democrat Liberal-conservative Racial-issues Adj. R-Squared

1/28/1879 H 155-72 −0.45 (1.52) −1.030 (1.11) 0.328 (2.01) −0.422 (2.55) 0.125 (1.01) 0.279 (1.04) 0.23

3/9/1882 S 29-15 −0.031 (0.05) −0.311 (0.14) 0.209 (1.37) 0.087 (0.29) 0.910 (0.92) 0.166 (0.55) 0.12

3/23/1882 H 167-66 −0.035 (1.47) 0.044 (0.50) 0.187 (1.65) −0.146 (2.12) 0.589 (2.40) 0.045 (0.06) 0.19

1/10/1892 H 180-65 −0.188 (1.71) 0.029 (0.09) 0.254 (1.40) −0.109 (1.39) 0.399 (2.11) 0.044 (0.01) 0.22

1/15/1892 S 67-16 0.044 (0.03) 0.019 (0.09) 0.193 (1.10) −0.444 (1.92) 0.371 (3.04) 0.201 (1.48) 0.27

Note: Absolute t-statistics are in parentheses. Each column represents a separate logistic regression with the dependent variable equaling 1 if a member of Congress voted “Yes” on a bill that would restrict legislation, 0 if nay

Table 3.2 Explaining congressional voting on key immigration bills that Passed, 1862–1892

Date Chamber Vote (Yea-Nay) Urbanization Foreign born Immigration growth Democrat Liberal-conservative Racial-issues Adj. R-Squared

6/24/1878 H 80-148 0.104 (0.01) −1.410 (2.00) 0.282 (1.20) 0.091 (0.01) 0.399 (1.97) 0.181 (0.04) 0.31

3/1/1882 S 18-24 −0.702 (1.03) −0.490 (0.98) −0.108 (0.88) −0.145 (1.16) 0.460 (2.08) −0.226 (0.75) 0.28

10/5/1892 S 14-30 −0.088 (0.99) 0.116 (0.09) 0.066 (0.00) −0.322 (2.10) 0.291 (1.75) 0.004 (0.00) 0.21

Note: Absolute t-statistics are in parentheses. Each column represents a separate logistic regression with the dependent variable equaling 1 if a member of Congress voted “Yes” on a bill that would restrict legislation, 0 if nay

66

Z. Gochenour

restriction. Urbanization, the proportion of foreign born, and the rate of growth of the foreign-born to not seem to have a significant effect, in many cases even the sign of the effect is inconsistent. Both the West and the Northeast had large numbers of immigrants, but voted in opposite ways on the issue of Chinese exclusion. Further work should continue this analysis to other time periods of U.S. history. One potential explanation not considered by previous literature is that the ideology of decision makers, independent of factors that might describe their constituencies, is an important predictor of change. A key assumption behind the previous work is that policy is both sensitive to public opinion and that public opinion is determined primarily by narrowly economic self-interest. An alternative explanation that emphasizes the role of interest groups in directing policy and a passive public is consistent with both rational ignorance about policy (following Downs (1954) and the Virginia Public Choice school) and rational irrationality (following Caplan 2007), where interests groups can capitalize on the public’s biases. But since voters are largely unaware of the political decisions made by the Congressmen, there is considerable leeway for individual voting members to bring their own ideology to the voting table. These results should be considered a small contribution supporting the view that Congressional ideology is important in policy.

3.5 Concluding Remarks The history of immigration legislation in the United States seems carefully targeted at particular ethnic groups. Chinese exclusion was the first14 such program, but it would extend to various ethnicities from Asia, Africa, Eastern Europe and Southern Europe in future decades. Only in much more recent history has concern been extended to natives of the Americas, such as those immigrating from Central or South America. Based on the content of publicized discussions in Congress, some responsibility for immigration restriction belongs to America’s nascent labor movement. Work by Boudreaux and DiLorenzo (1993) shows a similar pattern for the development of antitrust policy in the United States, which was also strongly supported by organized labor. There can be little doubt that economic considerations played a major role in developing attitudes toward immigrants, but it can be difficult to capture this idea by looking at purely economic data. Potentially, these economic concerns are not driven by real economic variables like a drop in employment or productivity, but by socialpsychological factors that may take years to develop. What started as a concern over job availability in the American West seems to have rippled through the political system over decades, even including the boom years of the late nineteenth century.

14 Possibly

with the exception of French, which was a major target nationality for the Alien and Sedition Acts.

3 Asian Exclusion in American Immigration Policy

67

Previous scholarship has emphasized the roles of xenophobia and nativism, labor market pressures, and the increased influence of urbanites and established immigrant communities. Roll-call data strongly supports the role of market forces, following previous works by economists, particularly Kanazawa (2005), Timmer and Williams (1998), and Goldin (1994). Roll-call data also supports Goldin’s (1994) contention that urbanized areas, independent of party and ideology, remained pro-immigration even as the rural south switched to restrictionism. During the economic book of the 1890s, support for restriction was weaker across the board, which several economists have previously claimed or predicted (Kanazawa 2005; Shughart et al. 1986; Goldin 1994). NOMINATE data also shows a role for ideology: as progressive ideology took shape, it came to incorporate new ideas and became closely associated with organized labor. Perhaps most importantly, the legislative history of Chinese exclusion represents a sea-change in migration policy to the United States. Concerns stemming from Chinese immigration, especially immigration of criminal aliens, led the federal government to take control of the migration system. Since then, the federal immigration bureaucracy has expanded greatly. The relationship with today’s immigration debate is clear. While current laws do not target ethnicities specifically, they are engineered partly to maintain wages for unskilled laborers, even at the expense of other considerations. While most workers and capital owners should expect to gain from increased immigration, widespread “folk economic” beliefs about the negative effects of foreigners often leads people to support anti-immigrant policy (Caplan 2007). Future research should expand this analysis to other countries and time periods. Evidence suggests that the restriction of immigration to the United States increased immigration to the United States increased immigration to other countries such as the United Kingdom, Canada, Brazil, and Argentina (Foreman-Peck 1992). Following the work of Foreman-Peck (1992) and Shughart et al. (1986), attention should be given to the responses of other countries to the policy changes in the United States. Acknowledgements The author is grateful to Bryan Caplan, Keith Poole, Alex Nowrasteh, Samuel Wilson, Don Boudreaux, Peter Leeson, and Josh Hall for helpful comments and suggestions.

References Abrams K (2005) Polygamy, prostitution, and the federalization of immigration law. Columbia Law Rev 105(3):641–716 Boudreaux DJ, DiLorenzo TJ (1993) The protectionist roots of antitrust. Rev Aust Econ 6(2): 81–96 Caplan B (2007) The myth of the rational voter: why democracies choose bad policies. Princeton University Press, Princeton, NJ

68

Z. Gochenour

Clemens MA (2011) Economics and emigration: trillion-dollar bills on the sidewalk? J Econ Perspect 25(3):83–106 Daniels R (2011) Asian America: Chinese and Japanese in the United States since 1850. University of Washington Press, Seattle, WA Dolan JP (2010) The Irish Americans: a history. Bloomsbury Publishing, New York, NY Downs A (1954) An economic theory of democracy. Harper & Row, New York Foreman-Peck J (1992) A political economy of international migration, 1815–1914. Manch Sch 60(4):359–376 Gilens M, Page BI (2014) Testing theories of American politics: Elites, interest groups, and average citizens. Perspect Polit 12(3):564–581 Goldin C (1994) The political economy of immigration restriction in the United States 1890–1921. In: Goldin C, Libecap GD (eds) The regulated economy: a historical approach to political economy. University of Chicago Press, Chicago, IL, pp 223–258 Hutchinson EP (1981) Legislative history of American immigration policy. University of Pennsylvania Press, Philadelphia Kanazawa M (2005) Immigration, exclusion, and taxation: Anti-Chinese legislation in gold rush California. J Econ Hist 65(3):779–805 Kearney D (1878) The workingman’s party of California: An epitome of its rise and progress. Bacon, San Francisco Kennedy DM, Cohen L, Bailey TA (2002) The American Pageant. Houghton Mifflin, Lexington, MA Luibhéid E (2002) Entry denied: controlling sexuality at the border. University of Minnesota Press, Minneapolis, MN Mann R (1982) After the gold rush: society in grass valley and Nevada City, California, 1849–1870. Stanford University Press, Palo Alto, CA Marks PM (1994) Precious dust: the North American gold rush era, 1848–1900. William Morrow & Company, New York, NY Norton HK (1924) The story of California: from the earliest days to the present. AC McClurg & Company, Chicago, IL Peffer GA (1986) Forbidden families: Emigration experiences of Chinese women under the Page law, 1875–1882. J Am Ethnic Hist 6(1):28–46 Poole KT, Rosenthal H (1984) The polarization of american politics. J Polit 46(4):1061–1079 Poole KT, Rosenthal HL (1997) Congress: a political-economic history of roll call voting. Oxford University Press, New York, NY Poole KT, Rosenthal HL (2007) Ideology and congress. Transaction Publishers, New Brunswick, NJ Saxton A (1975) The indispensable enemy: labor and the anti-Chinese movement in California. Univ of California Press, Berkeley, CA Shughart WF, Tollison RD, Kimenyi MS (1986) The political economy of immigration restrictions. Yale J Regul 4(1):79–98 Szczerbiak A, Taggart P (2008) Opposing Europe? Oxford University Press, New York, NY Takaki R (1989) Strangers from a different shore: a history of Asian Americans. Little, Brown and Company, New York Timmer AS, Williams JG (1998) Immigration policy prior to the 1930s: Labor markets, policy interactions, and globalization backlash. Popul Dev Rev 24(4):739–771 Van Nuys F (2002) Americanizing the West: race, immigrants, and citizenship, 1890–1930. University Press of Kansas, Lawrence, KS

Chapter 4

The Political Economy of the Arbitration Act of 1888 Joshua Gotkin

Abstract The goal of this chapter was to identify who supported and who opposed the Arbitration Act of 1888. A probit model was estimated with explanatory variables to capture the role of constituent interest, and legislator private interest. The role of legislator private interests was ruled out as having any significant effect on the passage of the Arbitration Bill. None of the occupation variables or personal characteristic variables were statistically significant. The results support the belief that the external cost of strikes played a significant part in the passage of the legislation. Representatives in states with urban populations and high levels of strike activity were more likely to vote in favor of the Arbitration Act. This places an emphasis on the legislation’s potential benefits in reducing the external costs of strikes to society. How were the private or special interests served? Organized labor supported arbitration legislation, and the effect of union involvement in strikes was predicted to be positive. However, this variable had a negative effect on the passage of the legislation. The passage of the bill implies that one group was more successful in influencing the legislative outcome. The effect of manufacturing and railroad wealth was negative and statistically significant, however, the marginal effects of these variables were small. Did some other factor influence the passage of this legislation over the protests of both railroad and manufacturing capital? During this period the Federal government was developing a more acute awareness of the public welfare and its relationship to private business interests. There is no denying that the Arbitration Act was legislation passed to promote the public welfare by reducing railroad strike activity, but it was also part of the fundamental debate that revolved around the changing role of the Federal government in the late nineteenth century.

J. Gotkin () ERS Group, Tallahassee, FL, USA e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2018 J. Hall, M. Witcher (eds.), Public Choice Analyses of American Economic History, Studies in Public Choice 37, https://doi.org/10.1007/978-3-319-95819-4_4

69

70

J. Gotkin

4.1 Introduction Mr. REED, of Maine. Mr. Chairman, one of the greatest delusions in the world is the hope that the evils in this world are to be cured by legislation.1 Mr. GLOVER, of Missouri. Gentlemen say, “Let [capital and labor] fight it out.” But, gentlemen, while they are fighting it out the commerce of this country is injured and destroyed. How long are they to be allowed, by culpable lack of legislation, to continue a conflict that grows worse and worse as it goes on? 2

A central question in the analysis of arbitration and mediation legislation is, who supported it? Was this legislation in the “public interest”, or did it benefit a particular interest group or set of interest groups? By identifying who favored or opposed this type of labor legislation and why, a clear picture can be constructed of the reasons for its creation. The goal of this chapter is to discern which groups were involved and what their opinions were regarding the legislative adjustment of labor disputes, and how they were able to influence the formulation and passage of the legislation. A basic model of legislative choice is applied to the voting behavior of members of the House of Representatives on an early version of the Arbitration Bill. In analyzing the voting behavior, I assume that a representative votes in response to the demands of a his or her constituents, and possibly his or her own interests (Fiorina 1974; Welch and Peters 1983; Peltzman 1984; Kalt and Zupan 1984; McGuire and Ohsfeldt 1986). The stated intended effect of this type of legislation was to reduce strike activity. As such, if railroad management and labor felt that strike-imposed costs could be reduced without compromising their ability to maximize their shares of the economic rents in the railroad industry, then both groups would have supported arbitration legislation. The other parties interested in reducing strike activity were those strongly affected by the external costs of strikes. Businesses which depended on the railroads for supplies of intermediate goods, and the shipment of final goods to market, would favor any legislation that reduced railroad strike activity. Consumers of railroad passenger service, and individuals who depended on the railroads to provide consumption goods, would also favor legislation that reduced strike activity. Farmers who relied on rail transport for delivery of crops to market would favor this type of legislation as well. Legislative action aimed at solving the problem of railroad strikes coincided with railroad strike activity. Strikes evidently brought significant pressure to bear on Congress. Whether this pressure originated from railroad owners, railroad labor, shipper, or other interests is what this chapter seeks to determine.

1 Congressional 2 Congressional

Record, 49-1, p. 2966. Record, 49-1, p. 2970.

4 The Political Economy of the Arbitration Act of 1888

71

4.2 Legislative Activity The railroad strikes of 1877 awakened government and the public to the ability of labor disputes between railroad owners and labor to impose enormous costs on society. However, there was no general consensus of how to solve the railroad strike problem. Suggested remedies ranged from increasing the size of the army and passing legislation to prohibit strikes, to the use of compulsory arbitration for all labor disputes. The first concerted effort by Congress occurred during the Southwest Strikes of 1885 and 1886. The strikes in the Southwestern United States on the lines owned by Jay Gould were the result of the railroads cutting wages in response to the business cycle downturn that began in late 1884. It was at the height of these strikes in 1886 that the Congress began to consider actual labor legislation in earnest. The first two bills introduced never made it out of committee. The Glover Bill, House Resolution 7081, and the Andersen Bill, House Resolution 7020, contained provisions for compulsory enforcement of the award in a court of law. The bill introduced by Representative John J. O’Neill of Missouri, House Resolution 7031, provided for voluntary arbitration with court enforcement of the arbitrated award. The bill that emerged from the House’s Committee on Labor was weaker than HR7081 and HR7020, and HR7081. Table 4.1 provides a detailed description and listing of the legislative activity during this period. House Resolution 7479 provided for the voluntary arbitration of disputes by a temporary board of three arbitrators. The award of this board was non-binding, and trusted to the “power of public opinion for the enforcement of the award.”3 Representative John J. O’Neill explained that the bill was the “best we [the Committee on Labor] could do.”4 He acknowledged that many members of the House would find the bill disappointing, but that the bill provided for the only method of resolving disputes that was both consistent with the Constitution and that “did not conflict with State rights or State laws.”5 The debates in Congress over House Resolution 7479 centered upon three polar positions. The first saw the Bill as absolutely ineffectual; the second was a noninterventionist position which saw no legal role for government in regulating the relationship between labor and capital; and the third view asserted that the Bill, having high moral purpose and the welfare of the nation at its heart, was necessary to ensure peaceful settlement of labor disputes. The first position was articulated by Representative William D. Kelley of Pennsylvania who contended, “I will vote for any Bill for the relief of oppressed labor; but I will not vote for this mass of

3 Congressional

Record, 49-1, p. 2959. Record, 49-1, p. 2959. 5 Congressional Record, 49-1, p. 2959. 4 Congressional

72

J. Gotkin

Table 4.1 Congressional activity on railroad labor legislation 1886–1888 Date 03/22/1886

Bill number HR7020

03/22/1886

HR7081

03/29/1886

HR7313

03/31/1886 03/31/1886 03/31/1886

HR7081 HR7313 HR7479

04/3/1886 04/05/1886

HR7479 HR7479

04/06/1886 04/12/1886 02/28/1887 03/04/1887 03/16/1888

HR7479 HR7479 HR7479 S2396

03/19/1888 03/19/1888

HR8665

04/17/1888 04/18/1888 04/19/1888 07/28/1888

HR8665 HR8665 HR8665 HR8665

09/17/1888 10/01/1888

HR8665 HR8665

Description Bill introduced to create U.S. Commission of Arbitration by Rep. Andersen (KS, Rep); sent to Committee on Labor; dies in committee Bill introduced to provide for the settlement of labor disputes by arbitration, by Reps. Glover (MO, Dem.) and Bumes (MO, Dem.); sent to Committee on Labor Bill introduced to create boards of arbitration for the speedy settlement of controversies and differences between common carriers and their employees, by Rep. O’Neill (MO, Dem.); sent to Committee on Labor Bill withdrawn by Rep Glover Replaced by HR7479 Replacing HR7313; bill is favorably reported on from committee by Rep. O’Neill Passes House 199 to 30 Bill referred to Senate Committee on Education and Labor, by Sen. Blair (NH, Dem.) Bill favorably reported from committee, by Sen. Blair Curtin Committee appointed to investigate the Southwest strike Passes Senate without division The Bill is pocket vetoed by President Cleveland Sen. Blair reintroduces HR7479; referred to Committee on Education and Labor Special House Committee appointed to investigate the Atchison, Topeka, and Sante F6 Railroad strikes Rep. O’Neill reintroduces HR7479; sent to House Committee on Labor Bill favorably reported by House Committee on Labor Bill passes Housea Bill referred to Senate Committee on Education and Labor Bill reported without amendment by Senate Committee on Education and Labor, by Sen. Wilson (IA, Dem.) Passes Senatea Signed by President Cleveland

Source: Congressional Record, 49-1, and 50-1 a Denotes that the actual vote was not recorded

words unless my vote be accompanied with this notice that the bill means nothing.”6 Representative Julius Caesar Burrows of Michigan agreed that “this seems to be an entirely harmless bill.”7

6 Congressional 7 Congressional

Record, 49-1, p. 2962. Record, 49-1, p. 2962.

4 The Political Economy of the Arbitration Act of 1888

73

One of the strongest opponents of HR7479, Representative Martin A. Foran of Ohio, who served on the House Committee on Labor, was “opposed to all adjustments of labor disputes between labor and capital which is based upon legislative enactment.”8 He found the idea of forced conciliation amongst men as “repugnant to the human mind.”9 The thought of Congress actually proposing to enact legislation to provide for the arbitration of labor disputes was “nonsense run mad.”10 The opponents of the bill objected to government intervention in labor disputes on several grounds—sovereignty of State’s rights, the rights of capital, and the constitutionality of the legislation. Representative John J. O’Neill, who introduced the bill and also served on the House Committee on Labor, was a staunch supporter of labor and saw the bill as both necessary and effective. He felt that this type of legislation would “compel the parties on both sides to appeal to reason and to say that brute force shall not and need not be used to obtain human rights.”11 Proponents of the legislation considered the labor problem, the conflict between labor and capital, as a fundamental issue in human relations that Congress, that Government, was obligated to solve.12 Arbitration legislation was, if not the cure, at least a step in the direction of discovering a solution to the strike problem.13 Representative Gibson of West Virginia, a proponent of the legislation, felt it was necessary for the government to “take hold of all controversies arising between capital and labor, arising in public institutions and corporations, and settle them, not purely for labor alone, not purely in the interest of capital alone, but in the interest of the public and in behalf of the country.”14 One side of the debate felt that labor and capital had to be compelled to resolve their differences. The other side objected to the government enacting such legislation. Both sides recognized the severe consequences of railroad strikes, but still could not agree on how to solve the problem. The bill, HR7479, finally passed both houses of Congress in late February of 1887, but was subsequently vetoed by President Cleveland. The President claimed to favor arbitration, but felt the bill was not “sufficient.”15 In a meeting with Representative John J. O’Neill, the President explained that the lack of government involvement in the administration of the legislation’s provisions caused him to withhold his signature from the bill. The bill originally submitted by Representative O’Neill to the President had a provision for the awards of the

8 Rep.

Foran had acted as the prosecuting attorney for the City of Cleveland against the striking railroad workers in 1877. 9 Congressional Record, 49-1, p. 2962. 10 Congressional Record, 49-1, p. 2963. 11 Congressional Record, 49-1, p. 2960. 12 Congressional Record, 49-1, p. 2963. 13 Congressional Record, 49-1, p. 3014. 14 Congressional Record, 49-1, p. 3009. 15 Washington Post, March 10, 1887.

74

J. Gotkin

temporary boards of arbitration to be reported to the Clerk of the United States Court. This gave to the award, “effect and form of a judgment.”16 Representative O’Neill informed the President that a bill which contained any compulsion of the award would never pass Congress, but that he would attempt to prepare a “bill for the next Congress in harmony with the President’s view.”17 The next round of legislative activity did not occur until the Burlington Strike in February of 1888. By mid-March, when HR7479 was reintroduced in both the Senate and the House, strike breakers had been hired to replace striking workers on the Chicago, Burlington, and Quincy; and injunctions had been issued to block the sympathy boycott of Burlington cars. Once again, negotiations had broken down between railroad management and workers, and the nation was threatened by a strike “of gigantic proportions” (McMurry 1956, p. 70). The bill re-designated as HR8665 passed the House in April with little debate, and was amended to provide for an investigative board appointed by the President. The bill was sent on to the Senate, but did not pass until September. The Arbitration Act of 1888 was signed into law on October 1 by President Cleveland, too late to be any use in the Burlington Strike. The next step in this analysis is the specification of variables to capture constituent and legislator interests. Organized labor favored arbitration legislation and railroad owners opposed it. A great deal of evidence was entered into the record of the Congressional debates indicating that manufacturing interests also favored arbitration. By quantitatively modeling voting behavior, hypotheses can be constructed to test whether Representatives voted in line with their constituents’ interests, special interests, or in the public interest.

4.3 Determinants of Voting Behavior 4.3.1 Modeling Voting Behavior An elected representative’s maximization problem is one in which the representative seeks to maximize the probability of reelection by introducing and voting for legislation which benefits their constituencies. This suggests that how a representative votes on a particular piece of legislation is a function of his constituents’ interests. The probability of representative i voting in favor of bill j can be expressed in the following form Pij = fi (Xi ) (4.1) where Xi is a vector of representative i’s constituents’ interests.

16 Washington 17 Washington

Post, March 10, 1887. Post, March 10, 1887.

4 The Political Economy of the Arbitration Act of 1888

75

Kalt and Zupan (1984) and McGuire and Ohsfeldt (1986) feel that this model is too simple and that a representative’s personal economic interests and ideologies need to be included as determinants of voting behavior. They also argue that constituents’ interests should include both economic and ideological interests. The modified model can be expressed as Pij = fi (Xi ,Zi )

(4.2)

where Xi is a vector of representative i’s constituent interests including both economic and ideological interests, and Zi is a vector of the personal economic interests and ideology of representative i.

4.3.2 Constituent Interests The selection of a set of variables to measure constituent interests needs to be both general and specific in order to capture the general or public interests, and special interests. Population, employment, wealth, and strike data are collected from US Census Bureau (1882a,b), Poor (1886), and US Commissioner of Labor (1888). These data are not easily disaggregated down to the district level—a clear deficiency which cannot be easily overcome. This does potentially introduce some aggregation bias into the results. If a state’s demographics were homogenous across all districts, then the aggregation bias would be non-existent. However, for most states, especially those with urban areas, there may be significant differences across congressional districts. The aggregation bias may over—or under—emphasize the importance of an interest group in a Representative’s district. To the extent that arbitration legislation applied to railroad operations across an entire state, railroad labor and owners would have attempted to influence all of a state’s Representatives, minimizing the aggregation bias. In order to control for some constituent interests at the district level, an urban dummy variable and a district level party attachment variable have been used. A summary of the data, and descriptive statistics for the data by voting behavior, are provided in Table 4.2. The importance of railroads is likely to be highly correlated with the size of railroad operations in a member’s district. The number of miles of track operated, revenues, level of investment, and profitability should be good measures of the strength of railroad interests in a Representative’s district. These variables are available in Poor (1886) on an annual, by state basis. The effect of railroads on a state tended to be widespread; therefore, the use of state level data should not introduce too much aggregation bias. The total wealth of railroads per capita is used as a measure of the size of railroad operations in a state.18 Total investment and

18 The

total wealth and miles of track operated in a state were highly correlated and collinear.

76

J. Gotkin

Table 4.2 Description of data used in the voting analysis Variables Democrata Southern Democrat

Serving first-term Age College educated Agricultural Manufacturer Trade and transportation Professional and personal services Urban district

Southwest Strikeb

Percent Democratc Percent Blackd Percent Immigrantd Percent employed agricultured Percent employed manufacturingd Percent employed trade and non-railroad transportationd Percent employed railroad transportationd Farm wealth per capitad

Manufacturing wealth per capitad Railroad wealth per capitad, e Percent strikes union organizedf

Definition Member’s political party affiliation: Democrat = 1; Republican = 0 Dummy variable for Democratic Representatives from the deep South: Virginia, Alabama, Arkansas, Florida, Georgia, Louisiana, Mississippi, N. Carolina, S. Carolina, and Texas Dummy variable equal to 1 if the 49th Congress was the member’s first term Member’s age Dummy variable for completion of college education Previous occupation agriculture/farming dummy variable Previous occupation manufacturing dummy variable Previous occupation trade and transportation dummy variable Previous occupation professional and personal services: lawyer, banking, journalist/publisher, doctor, college professor, public servant, engineer, soldier, and clergyman Dummy variable equal to 1 if a member’s district was urban. These are cities with population greater than 19,000 and identified as being one of the 100 principal cities in the U.S as defined by the census, 1880 Dummy variable for states affected by the Southwest strike: Arkansas, Kansas, Louisiana, Missouri, Nebraska, Tennessee, and Texas The percent of voters in a district that voted Democratic in the House elections of 1884 Percent of state population, Black, 1880 Percent of state population foreign born, 1880 Percent of persons in a state employed in agriculture, 1880 Percent of persons in a state employed in manufacturing, 1880 Percent of persons in a state employed in trade and non-railroad transportation, 1880 Percent of persons in a state employed in railroad transportation, 1880 The amount invested in land, buildings, fences, and implements plus the value of agricultural products, divided by a state’s population, 1880 The amount invested in plant and equipment plus the value of manufactured goods, divided by a state’s population, 1880 The amount invested in railroads plus railroad revenues, divided by a state’s population, 1885 Number of union organized strikes, divided by the total number of strikes for 1885–1886, by state (continued)

4 The Political Economy of the Arbitration Act of 1888

77

Table 4.2 (continued) Variables Percent of days lost to railroad strikes Percent of days lost to non-railroad strikes

Definition The average length of railroad strikes multiplied by the number of employees involved in railroad strikes for 1885–1886, divided by railroad employment, multiplied by 624 days, by state The average length of non-railroad strikes multiplied by the number of employees involved in non-railroad strikes for 1885–1886, divided by non-agricultural, non-railroad employment, multiplied by 624 days, by state

Sources: a All characteristics of house members from ICPSR 7803 b Various issues of Washington Post and New York Times and Poor (1886) c Congressional Quarterly (1985) d US Census Bureau (1882a,b); Poor (1886); US Commissioner of Labor (1888)

revenues for 1885 are summed together and divided by a state’s population to arrive at railroad wealth per capita. The expected effect of railroad interests on a Representative’s voting behavior on the Arbitration Bill is negative, because the railroads opposed this type of legislation (Ko 1926, p. 56). Jay Gould, president of the Missouri Pacific Railway Company, favored arbitration, but only between the company and individual workers (Klein 1986, p. 360). During the Southwest strike of 1886, the Missouri Pacific refused the Knights of Labor’s request for arbitration.19 Gould felt that compulsory arbitration would find few friends, and that an employer’s right to dismiss and hire workers should “remain unimpaired under all circumstances”20 If a Representative’s state was characterized by significant railroad interests—high railroad wealth per capita—then a Representative would be less likely to vote in favor of arbitration legislation. Labor was divided on the issue. The railroad brotherhoods and the Knights of Labor favored arbitration, as evidenced by their appeals for arbitration of disputes. The Knights of Labor’s platform called for the “enactment of laws providing for the arbitration of differences between employers and employees”21 During the debates on the Arbitration Bill, letters received from labor organizations were presented as evidence in favor of the bill. In fact, many of the letters favored compulsory arbitration. A letter received from the Eureka Assembly 4180, Knights of Labor in Bonne Terre, Missouri called for legislation that would “compel the recognition of labor organizations, and compel corporations to arbitrate differences between their employees.”22 However, there is some evidence that the Arbitration Bill was not favorably viewed by certain Representatives of organized labor. The Workmen’s Ethical

19 Washington

Post, March, 17, and 30, 1886. Post, April 25, 1886. 21 Congressional Record, 49-1, p. 3036. 22 Congressional Record, 49-1, p. 2973. 20 Washington

78

J. Gotkin

Society of Washington opposed the Arbitration Bill.23 The Post interviewed several members of the Knights of Labor who were adamantly opposed to the Arbitration Bill. These Knights were worried that the lawyers of “rich railroad corporations” could prolong the process of arbitration “indefinitely” and turn the proceedings to the railroad’s favor.24 For the most part, however, labor opposition to the Arbitration Bill was very limited. The majority of workers and labor organizations favored arbitration and arbitration legislation. Overwhelmingly, organized labor wanted employers to recognize and negotiate with them as the representatives of workers. Employers were generally quite successful in ignoring the demands of workers. They employed lockouts, hired strike breakers, and used injunctions to resolve strikes, rather than enter into negotiations with striking workers (Eggert 1967, pp. 56–59). Organized labor felt that arbitration legislation would promote or even force employers to negotiate with unions. The data available for union membership on the state level, and even the national level for the 1880s is very poor. Garlock (1982), in a study of the Knights of Labor, can only account for membership in less than one third of the Local Assemblies of the Knights. There is no way to judge how representative the sample is for which he has membership numbers. Rather than use these membership numbers as a measure of union activity, the number of union organized strikes was divided by the total number of strikes in a state between 1885 and 1886. Given the position of organized labor, Representatives from states with heavy union activity would be more likely to vote for arbitration legislation. A positive effect is expected for the percent of strikes union organized. In addition, the percentage of a state’s labor force employed in agriculture, trade and non-railroad transportation, manufacturing, and railroad transportation, are included as a more general measure of labor’s importance in a member’s constituency. The evidence from the debates favors a positive effect for both manufacturing and railroad transportation employment. Given agriculture’s weaker ties to railroad interests, Representatives with significant agricultural interests would be less likely to vote in favor of the arbitration legislation, unless agriculture was highly dependent on the railroads for delivery of crops. Persons occupied in trade and non-railroad transportation should have supported arbitration legislation since they were affected by the external costs of railroad strikes. A positive effect for percent employed in trade and non-railroad transportation would be consistent with this scenario. However, it is possible that the effect could be negative. The arbitration legislation represented a form of regulation of the interstate commerce, and individuals engaged in trade and transportation, fearing government intervention and regulation, would not have favored the arbitration legislation.

23 Washington 24 Washington

Post, April 10, 1886. Post, March 31, 1886.

4 The Political Economy of the Arbitration Act of 1888

79

The other groups that stood to benefit from a reduction in strike activity were those affected by the external costs of strikes. The fact that there were more than two parties involved in a railway strike was clearly recognized by congressmen.25 Manufacturing operations were often impaired by strikes, and Representatives who served states with large manufacturing sectors should have voted in favor of the arbitration legislation. The manufacturing/corporate lobby in congress favored the passage of the Arbitration Bill.26 Representative Glover of St. Louis, Missouri, was concerned for his manufacturing constituents. He spoke out in favor of the Arbitration Bill during the debates, reporting that in March alone “260,000,000 pounds of freight, valued at $40,000,000, would have moved had there been no strike. Ten flouring mills and scores of manufacturing establishments have been forced to shut down for want of raw materials.”27 The manufacturing wealth per capita in a state is used to ascertain the position of manufacturing interests on the bill. The only evidence entered into the House debates from manufacturing interests suggests that they supported the Arbitration Bill.28 This is not definitive evidence of manufacturing interests’ disposition towards arbitration. If manufacturing interests favored arbitration, then high manufacturing wealth per capita would have a positive effect on the probability of a Representative voting in favor of the Arbitration Bill. However, if manufacturing interests harbored similar opinions to those espoused by Jay Gould, then an inverse relationship would be expected. No direct evidence surfaced regarding farming interests’ position on the arbitration issue. The farm wealth per capita variable captures the importance of agriculture in a state. The aggregation bias is likely to be significant for this variable. For example, the Midwestern States had both high farm wealth and manufacturing wealth per capita, but this wealth was not evenly distributed across congressional districts. However, even in the aggregate, Representatives from states with high farm wealth and dependence on railroad transportation would be more likely to vote in favor of arbitration legislation. Also, the agrarian protests of the 1880s, which were predominately aimed at the railroads and big business (Puth 1988, pp. 303–309), would have increased the likelihood of voting for arbitration legislation. Therefore, a positive effect would be consistent for the farm wealth per capita variable. The last few constituent variables are included to capture general state and district level characteristics. Blacks resided in predominately rural areas and were concentrated in the South. Any effect associated with the percent of the population that was Black is more likely to be a pure regional effect, than caused by the preferences of Blacks, who were becoming increasingly disenfranchised in the Southern States. The expected effect for the percentage of the population that was

25 Congressional

Record, 49-1, p. 3017. Record, 49-1, p. 3019. 27 Congressional Record, 49-1, p. 2970. 28 Congressional Record, 49-1, pp. 3019, 3022–24. 26 Congressional

80

J. Gotkin

Black is unclear. The percent of the population that were immigrants is included to further differentiate regional differences. The expected effect for this variable is also unclear. Urban areas were highly dependent on the railroads for the supply of consumption goods. Representative Glover feared for the well-being of his constituents, concerned that railroad strikes could potentially threaten the people of St. Louis with “famine in the midst of plenty.”29 The urban district dummy variable should help diminish some of the aggregation bias. This variable is equal to 1 if a Representative’s district was located in an urban area. The cities included in this variable had populations over 19,000 and were identified as being one of the 100 principal cities in the United States by the 1880 census (US Census Bureau 1882b, pp. 379–380). The expected effect for the urban district dummy variable is positive. The legislative activity in early April 1886 on arbitration legislation occurred against the backdrop of the Southwest Strike. A dummy variable is constructed and set equal to 1 for states that were involved in the strike. The Missouri Pacific and Gould System of railroads serviced Texas, Louisiana, Oklahoma, Arkansas, Tennessee, Colorado, Kansas, Missouri, and Nebraska. The cities of Ft. Worth, New Orleans, and St. Louis were particularly hard hit by the strike. The Representatives from these states were well aware of the injurious effects of strikes and should have voted in favor of arbitration. Thus, a positive effect is expected for the Southwest Strike dummy variable. The other strike activity variables—percent of days lost to railroad and nonrailroad strikes, should have a similar effect on how a Representative voted, and thus, should also have a positive effect on the probability of voting in favor of the arbitration legislation. The final constituent variable is included to capture constituent ideology. The percentage of Democratic votes cast for United States Representatives in a congressional district was used to identify constituent party attachment. Values in the 45–55% range are indicative of districts where Representatives might be more likely to cross party lines when voting, and have weaker party attachments. The effect for this variable should parallel that of the Representative’s party variable—Democrats tended not to favor the arbitration legislation, because of beliefs about the proper role of the Federal government, and the sovereignty of States’ rights. Therefore, the higher the percentage of Democrats in a Representative’s district, the less likely a Representative would be to vote in favor of arbitration legislation.

4.3.3 Representative Self-Interest Why might a legislator vote in a manner contrary to his or her constituent’s interests? One reason would be because a Representative had a personal interest that might

29 Congressional

Record, 49-1, p. 2970.

4 The Political Economy of the Arbitration Act of 1888

81

be affected by the passage of arbitration legislation. The railroads feigned being in favor of arbitration, but they clearly had little incentive to arbitrate disputes, especially with union representatives, given their success in dealing with strikes in other ways. If a member of the House had private railroad interests it is likely, that he would have voted against the Arbitration Bill. A legislator with manufacturing interests might have voted in favor of the bill, even if railroads represented a significant proportion of their constituency. A member’s last occupation before taking service in the House may reveal private economic interests. The occupations are collapsed into four categories—agricultural, manufacturer, trade and transportation, and professional and personal services. If private economic interests contribute to voting behavior, then the effects of these dummy variables would be statistically significant. A Representative would vote in favor of the legislation if he felt that his private economic interests would be served, and against it otherwise. The professional and personal services dummy variable includes, in descending proportions, lawyers, bankers, doctors, journalists/publishers, college professors, public servants, engineers, clergymen, and soldiers. Lawyers comprised 83% of the professional and personal services category, and 69% of overall House membership. There was only one member in the trade and transportation category with significant ties to the railroad industry. Representative William Scott of Pennsylvania had been the “president or director of various lines, aggregating over 22,000 miles of completed road, the greatest number of miles of railroad, probably, which any one individually was ever an officer or director of” (United States Congress 1886, p. 80). If private interest had influenced Representative Scott, then he would have voted against the Arbitration Bill. Another potential cause for voting opposite the interests of one’s constituency is personal “ideology.” This type of behavior can often be captured by how a member has voted on similar issues. For instance, if a legislator generally voted against big business, then it is likely they would have voted for the Arbitration Bill even when large numbers of their constituents favored the arbitration legislation. However, the problem of a Representative’s “ideology” is generally not worrisome, since a Representative’s “ideology” is highly correlated with his constituent’s “ideology” (Peltzman 1984). There is a positive correlation of 0.6375 between a Representative’s Democratic party affiliation, and the percentage voted Democratic in a congressional district. A Representative’s party affiliation is often considered as a good proxy for a legislator’s “ideology.” Even so, there is evidence that some Representatives did vote contrary to the best interests of their constituents. Martin Foran, an Ohio Democrat who represented constituents in Cleveland, felt that the government had no business regulating the relationship between labor and capital. In fact, he found the idea of forcing or

82

J. Gotkin

compelling men to be conciliatory as “repugnant to the human mind.”30 Foran represented an urban constituency with strong manufacturing interests, yet he took up a position in opposition to the bill. A Representative may have voted in a manner contrary to his constituents based on whether or not he felt that government had a legal obligation to solve the labor problem. The constitutionality of the arbitration legislation revolved around the issues of state sovereignty and the powers of the Federal government. The Democrats tended to be strong proponents of State’s rights, and thus would have been more likely to vote against the Arbitration Bill. Therefore, a negative effect is expected on the Democratic dummy variable. A dummy variable is also included to capture the voting behavior of Southern Democrats. They were strong proponents of States’ rights and tended to consistently vote together. Therefore, they would have been more likely to have voted against the Arbitration Bill. In addition to party affiliation and occupation, a Representative’s age, level of education, and congressional service are included as general measures of a Representative’s personal characteristics. The arbitration legislation was controversial; a vote in favor of the bill represented support of a fundamental shift in the role of the Federal government in the economy. Representatives serving their first terms in the House may have been reluctant to take a stand on this heated issue, thus a negative effect is expected for the serving first term dummy variable. The fundamental change this bill represented would be more likely embraced by younger members of the House; thus, age should have a negative effect on the probability of voting for the legislation. The level of education is also likely to have a negative effect on the probability of voting in favor of the legislation. Representatives with more education might have been more prone to challenge the legislation along constitutional grounds. The next section examines the actual voting behavior of House members with respect to the arbitration legislation.

4.4 The Voting Data The Arbitration Act of 1888 passed the House of Representatives on April 18, 1888.31 It passed the Senate on September 17, 1888.32 However, no record of how individual Representatives or Senators voted was reported in the Congressional Record, The New York Times, or the Washington Post. This fact makes it impossible to analyze voting behavior on the passage of the final bill. The earlier version of the Arbitration Act, passed during the previous legislative session (HR7479), which was vetoed by President Cleveland, has several recorded roll call votes related to the bill. There are three roll call votes in the House on April

30 Congressional

Record, 49-1, p. 2962. Record, 50-1, p. 3109. 32 Congressional Record, 50-1, p. 8609. 31 Congressional

4 The Political Economy of the Arbitration Act of 1888

83

Table 4.3 Description of roll call votes Votes Vote 1 Vote 2 Vote 3 Vote 4

Voting behavior: 0 = opposed to arbitration; 1 = in favor of arbitration; 2 = not voting Vote not to adjourn and take up debate on the Arbitration Bill, April 1, 1886. 193 in favor, 64 opposed, 66 not votinga Vote to change order of business and take up debate on the Arbitration Bill, April 2, 1886. 154 in favor, 71 opposed, 98 not votingb Vote to put aside debate on the Silver Bill and take up debate on the Arbitration Bill, April 3, 1886. 128 in favor, 87 opposed, 108 not votingc Vote for passage of the Arbitration Bill, HR7479, April 3, 1886. 199 in favor, 30 opposed, 94 not votingd

Sources: a Congressional Record, 49-1, p. 3027 b Congressional Record. 49-1, pp. 3032–33 c Congressional Record, 49-1, pp. 3059–60 d Congressional Record, 49-1, p. 3066

1, 2, and 3 1886, for consideration of taking up debate on the bill.33 The fourth roll call is for the House’s passage of the bill.34 A description of these roll call votes can be found in Table 4.3. The Senate passed the bill on February 28, 1887.35 The Congressional Record does not report individual voting behavior or the tally of votes, but the Washington Post reports that the bill passed the Senate “without division.”36 The available data for roll call votes pertaining to the arbitration of labor disputes restricts the analysis of the arbitration legislation to the House activity on the earlier version of the Arbitration Bill, HR7479. The debates over the earlier bill and the final bill are very similar, as are the bills themselves. The only major difference was a provision in the final version authorizing the President to create an investigative commission to study the causes of disputes between common carriers and their employees. The next section addresses the econometric modeling of voting behavior.

4.5 Econometric Modeling of Voting and Abstention Behavior The actual voting behavior of Representatives was restricted to three distinct choices. They could vote in favor or against a particular piece of legislation, but they could also choose not to vote. This is discrete information. A legislator cannot

33 Congressional

Record, 49-1, pp. 3027, 3032–33, and 3059–60. Record, 49-1, p. 3066. 35 Congressional Record, 49-2, p. 2376. 36 Washington Post, March 1, 1887. 34 Congressional

84

J. Gotkin

express his or her preferences over legislative outcomes by stating how much he or she favors a bill. This necessitates the use of discrete choice econometric models (Amemiya 1981; Maddala 1983). The underlying structure of these types of discrete choice models is that the observed behavior, the dependent variable, yi , is discrete. In the case of a binary discrete variable, the dependent variable can take on a value of zero or one. The task of these models is to predict the probability that the dependent variable will take on one of the discrete values of the observed variable. Specifically, what is the probability that the dependent variable, yi , will take on some value in the discrete distribution of yi , conditional on some set of deterministic characteristics, Xi . The actual probability of observing, yi is unobservable and can be defined as yi∗ = X β + ui It then follows that

(4.3)



E(yi∗ |Xi ) = Xi β

(4.4)

In the binary choice case, then the 

Pr(yi = 1) = Pr(ui > −Xi β) 

= 1 − F (−Xi β) 

= F (Xi β)

(4.5) (4.6) (4.7)

where F is the cumulative distribution function of u. The preferences for or against arbitration could be easily modeled using univariate dichotomous limited dependent variable models—a simple probit or logit specification, where the cumulative distribution function of u would be normal or logistic. However, there are three types of voting behavior to be explained. In cases where there are more than two choices, polychotomous limited dependent variable models’ should be used. There are several types of these models, and the choice of which one to use is dictated by the characteristics of the actual decision process being modeled. The model selection process for voting behavior hinges on understanding the reasons why some Representatives chose to abstain from voting. There may have been some type of strategic behavior involved in the choice of not voting. Some Representatives may have felt it was better not to vote than go on record as either opposing or favoring the arbitration of railroad labor disputes. If enough Representatives choose not to vote, legislative action can be blocked for lack of a quorum. However, the reasons for not voting were often not strategically motivated, but simply the result of circumstance. A House member might have been detained elsewhere—at home visiting constituents, ill, or on personal business. There is evidence that all of the above reasons contributed to not voting. When the Arbitration Bill went to a vote on April 3, 1886 Representative Pidcock was

4 The Political Economy of the Arbitration Act of 1888

85

Fig. 4.1 Voting decisions

“unavoidably absent, owing to the death of a near relative;” Representative Lehlbach was “detained in his committee-room;” and Representative Riggs was “absent on account of sickness.”37 The Washington Post reported on April 4, 1886 that “many who would have voted and who did vote to have the bill shut out from further consideration (Vote3), did not dare go on record as voting against it when the straight question was put.” Ninety-four members of the House did not vote on the arbitration question (Vote4). In addition to identifying the deterministic factors underlying particular actions, it is important to understand the relationships between the choices when specifying the proper model. If the multiple decisions, vote yes, vote no, or do not vote, are thought to be independent and mutually exclusive, then the multinomial logit model is an appropriate choice. This implies that the decision process can be illustrated like in Fig. 4.1. If a Representative chose not to cast a vote for strategic reasons that were related to being in favor of or against the Arbitration Bill, then not voting could not have been independent of the other two choices. Given this relationship between the choices, the underlying assumption of the independence of irrelevant alternatives (Hausman and McFadden 1984), which is required in the multinomial model specification, is violated. Therefore the use of a multinomial specification would be improper. An ordered probit model would be appropriate if the observed discrete variable systematically corresponded to the unobservable continuous variable in the following way:

yi =

⎧ ∗ ⎪ ⎪ ⎨0 if yi ≤ ν, then vote against arbitration

1 if ν < yi∗ < ν + γ , then do not vote ⎪ ⎪ ⎩2 if y ∗ ≥ ν + γ , then vote for arbitration i

(4.8)

If the voting behavior followed this type of decision process, those who abstained from voting would be ordered somewhere between voting in favor of arbitration and opposing arbitration as in Fig. 4.2. Representatives who abstained could be viewed as being indifferent to voting for or against the bill. The evidence presented above indicates that some Representatives who chose not to vote may have been opposed to arbitration. These Representatives would not have been indifferent to arbitration. The choices need only be differenti-

37 Congressional

Record, 49-1, p. 3066.

86

J. Gotkin

Fig. 4.2 Ordered voting behavior

ated from each other. If not voting Representatives all chose not to vote because they opposed the bill, but were afraid of the repercussions, then an ordered model would be appropriate. What this implies is that the individual groups must be homogeneous in their preferences over the three actions. If they are not, then it would be difficult to order them. The Representatives who chose not to vote were not a very homogeneous group with respect to their reasons. In fact, the mean characteristics of the group that did not vote were very similar to the mean characteristics of the group that voted (see Table 4.4. The two groups’ means differ statistically significantly for only eight of the twenty-three variables. The mean characteristics of the not voting group and those of the group that voted in favor of the bill differed statistically significantly for only nine of the twenty-three variables. The not voting group was more similar to the group that voted for the legislation than it was to the group opposed to the legislation. The relationships between the three groups do not seem to meet the underlying assumptions that would be imposed by an ordered model, which suggests that this type of model is also inappropriate. The last of the polychotomous choice models to evaluate is a sequential model. Sequential models view the decision process in stages. A legislator would first decide whether or not to cast a vote, and then in the next stage the decision to vote for or against the Arbitration Bill would be made. See Fig. 4.3 for a graphical representation. The simplest version of this model is estimated under the assumption that the decisions at each stage are independent (Maddala 1983). Under this assumption, each of the decisions is modeled as an independent binary choice model, and can be estimated with either a simple probit or logit. The probability of voting in favor of the Arbitration Bill in the simple sequential model is expressed as 



Pr(yi = 1) = F (Xi β1 )F (Xi β2 )

(4.9)

The parameter β1 is determined by dividing the entire sample into two groups— voting and not voting. The probability of choosing to vote in the first stage is expressed as 

Pr(y1i = 1) = F (Xi β1 )

(4.10)

4 The Political Economy of the Arbitration Act of 1888

87

Table 4.4 Mean characteristics of members of the 49th House of Representatives by voting behavior on Vote4—HR7479 Variable Democrat Southern Democrat Serving first-term Age College educated Agricultural Manufacturer Trade and transportation Professional and personal services Urban district Southwest Strike Percent Democrat Percent Black Percent immigrant Percent employed agriculture Percent employed manufacturing

All members 0.563 (0.496) 0.216 (0.413) 0.375 (0.485) 49.04 (8.56) 0.684 (0.465) 0.049 (0.217) 0.052 (0.223) 0.062 (0.241) 0.835 (0.371) 0.343 (0.475) 0.170 (0.376) 52.35 (15.35) 12.99 (18.23) 13.35 (9.68) 44.88 (21.09) 21.65 (13.06)

Voted 0.528 (0.500) 0.196 (0.398) 0.366 (0.483) 48.76 (8.76) 0.716 (0.451) 0.043 (0.204) 0.026 (0.160) 0.069 (0.255) 0.860 (0.347) 0.332 (0.472) 0.192 (0.395) 51.12 (14.15) 12.16 (17.50) 13.10 (9.37) 45.62 (20.48) 21.36 (12.82)

Members who voted Nay 1.000 (0.000) 0.733 (0.449) 0.433 (0.504) 50.06 (9.26) 0.866 (0.345) 0.033 (0.182) 0.000 (0.000) 0.000 (0.000) 0.966 (0.183) 0.200 (0.407) 0.200 (0.407) 66.13 (14.72) 35.16 (19.97) 3.90 (5.46) 63.65 (15.28) 10.32 (7.26)

Members who voted Yea 0.457 (0.499) 0.115 (0.321) 0.356 (0.480) 48.56 (8.69) 0.693 (0.462) 0.045 (0.208) 0.030 (0.171) 0.080 (0.272) 0.844 (0.363) 0.351 (0.478) 0.191 (0.394) 48.86 (12.63) 8.69 (14.21) 14.48 (9.059) 42.90 (19.81) 23.03 (12.66)

Not voting 0.649 (0.479) 0.266 (0.444) 0.394 (0.491) 49.72 (8.04) 0.606 (0.491) 0.063 (0.245) 0.117 (0.323) 0.042 (0.203) 0.776 (0.418) 0.372 (0.486) 0.117 (0.323) 55.35 (17.67) 15.01 (19.85) 13.96 (10.43) 43.08 (22.51) 22.35 (13.69) (continued)

88

J. Gotkin

Table 4.4 (continued) Members who voted Nay 5.27 (3.18)

Members who voted Yea Not voting 9.09 9.51 (3.76) (4.88)

1.50 (0.59)

0.94 (0.54)

1.58 (0.55)

1.33 (0.56)

280.25 (113.58) 165.22 (135.92) 179.64 (102.23) 50.68 (20.17) 0.137 (0.302)

287.92 (114.12) 159.35 (130.55) 185.00 (104.12) 48.88 (18.62) 0.148 (0.315)

194.13 (92.75) 61.29 (64.32) 112.13 (68.50) 48.81 (27.73) 0.157 (0.371)

302.07 (110.46) 174.13 (131.69) 195.99 (104.26) 48.89 (16.93) 0.147 (0.307)

261.55 (110.65) 179.53 (147.94) 166.56 (96.77) 55.07 (23.05) 0.111 (0.266)

Percent days lost to non-railroad strikes

0.253 (0.258)

0.244 (0.251)

0.062 (0.126)

0.2722 (0.254)

0.275 (0.273)

Observations

323

229

30

199

94

Variable All members Voted Percent employed 8.86 8.59 trade and non-rail (4.22) (3.90) road transportation Percent employed 8.86 railroad (4.22) transportation Farm wealth per capita Manufacturing wealth per capita Railroad wealth per capita Percent strikes union organized Percent of days lost to railroad strikes

Note: Standard deviations in parentheses

In the second stage, β2 is determined by taking the subsample of voting House members, segregating them by whether they voted for or against the bill, and then estimating a binary probit or logit. 

Pr(y2i = 1) = F (Xi β2 )

(4.11)

There are two problems in estimating this model with the voting data. The first and already discussed, is whether the decisions to vote, and voting in favor or against the legislation are independent, The other issue is whether the explanatory variables are properly specified for the choices in each stage (Hausman and McFadden 1984). Due to the heterogeneity in the reasons for choosing not to vote, it is likely that specifying a set of explanatory variables will be very difficult.

4 The Political Economy of the Arbitration Act of 1888

89

Fig. 4.3 Sequential voting behavior Fig. 4.4 Simple binary vote

The estimation of the first stage of the sequential model provides evidence in support of the mis-specification hypothesis. The stage 1 estimates only correctly predict 73% of the vote/not vote decision, and the McFadden R-square is only 0.1208. In fact, the sequential model yields a predicted probability of voting in favor of the bill of only 0.6620, which is less than 0.8690 the actual proportion that voted in favor of the legislation. Simply predicting a vote in favor of the Arbitration Bill 86.9% of the time would be a better estimator of voting behavior than the sequential model. Thus, another of the polychotomous limited dependent variable models is dismissed. Given the characteristics of the data, the use of any of the polychotomous limited dependent variable models is ruled out. The difficulty associated with clearly identifying the reasons behind the choice not to vote is common to all voting studies. The approach generally opted for is to simply ignore those Representatives that did not cast votes (Gilligan et al. 1989; Poole and Rosenthal 1991). Once the voting behavior has been reduced to a simple binary choice, excluding all not voting Representatives, a simple probit model can be estimated (Fig. 4.4). The probability of voting in favor of the arbitration legislation is expressed as 

Pr(yi = 1) = F (Xi β)

(4.12)

90

J. Gotkin

where F is the cumulative normal distribution function, and the predicted voting behavior, yˆi , is determined by the following method  yˆi =



ˆ < 0.5 0 if F (Xi β)  ˆ ≥ 0.5 1 if F (X β)

(4.13)

i

The results using this simple specification for Vote4 are presented in the next section.

4.6 Estimation and Results The probit analysis of House members’ voting behavior will focus on the final House vote on the Arbitration Bill HR7479 (Vote4) and be restricted to voting members only. However, an attempt was made to include the information contained in the other roll call votes. A voting index was created following Kalt and Zupan (1984). The number of votes in favor of arbitration legislation are divided by the number of times that a Representative voted. This creates a dependent variable bounded by zero and one. The voting index did not perform well even though 61% of the voting members did not alter their position on the Arbitration Bill. This is probably a result of the nature of the other roll call votes, which were not exclusive to the question of arbitration legislation. On Vote3, a vote in favor of taking up debate on the Arbitration Bill may have been a vote against the Silver Bill, rather than a vote in favor of arbitration. Probits were estimated for all of the roll call votes, but only the results for Vote4 had good predictive power. The discussion in the previous section offered justification for ignoring the behavior of non-voting Representatives. However, the other roll call votes and information contained in the Congressional Record can be used to examine why some members chose not to vote. The Washington Post claimed that some Representatives chose not to vote rather than go on record as opposed to the bill.38 The results of these additional investigations of voting behavior will be presented and discussed after the results for the probit analysis of Vote4. The purpose of the probit analysis is to identify who supported and who opposed the passage of this legislation (see Table 4.5 for the probit analysis results). The first group of variables controls for a Representative’s personal characteristics. A Representative’s personal economic interests may impel voting behavior which is contrary to the best interests of his or her constituents. The previous occupation dummies, which were included to capture personal economic interest, are not individually or jointly statistically significant; a likelihood ratio test of their joint significance yields a chi-squared statistic of 1.896 with three degrees of freedom. In fact, none

38 Washington

Post, April 4, 1886F.

4 The Political Economy of the Arbitration Act of 1888

91

Table 4.5 Results for probit estimation of voting behavior Variable Democrat Southern Democrat Serving first-term Age College educated Agricultural Manufacturer Trade and transportation Urban district Southwest Strike Percent Democrat Percent Black Percent immigrant Percent employed agriculture Percent employed manufacturing Percent employed trade and non-railroad transportation Percent employed railroad transportation Farm wealth per capita Manufacturing wealth per capita Railroad wealth per capita

Vote4 −4.683 (5.736) −2.873* (1.663) −0.368 (0.355) −0.016 (0.0185) −0.553 (0.466) 0.547 (0.944) 2.977 (26.651) 3.091 (15.662) 0.273 (0.4222) 1.107 (0.914) −0.008 (0.014) −0.062 (0.049) 0.594* (0.330) −0.137 (0.144) 0.370 (0.267) −0.782 (0.554) 2.207 (2.285) −0.021 (0.014) −0.073* (0.037) −0.038* (0.023)

Elasticity at means −1.13E-07

Weighted aggregate elasticity −0.406

−2.58E-08

−0.139

−6.18E-09

−0.013

−3.65E-08

−0.069

−1.81E-08

−0.039

1.09E-09

0.002

3.56E-09

1.38E-06

9.87E-09

4.32E-06

4.14E-09

0.005

9.72E-09

0.027

−1.89E-08

−0.044

−3.48E-08

−0.146

3.56E-07

0.312

−2.86E07

−0.721

3.61E-07

0.399

−3.07E-07

−0.426

1.51E-07

0.207

−2.74E-07

−0.408

−5.38E-07

−0.515

−3.22E-07

−0.440 (continued)

92

J. Gotkin

Table 4.5 (continued) Variable Percent strikes union organized Percent of days lost to railroad strikes Percent days lost to non-railroad strikes Constant

Vote4 −0.022 (0.016) −1.322 (1.765) 7.372** (3.574) 28.841 (19.722)

Elasticity at means −4.93E-08

Weighted aggregate elasticity −0.085

8.95E-09

−0.028

8.24E-08

0.063

1.32E-06

2.503

Notes: Standard errors in parentheses *Denotes statistical significance at the 10% level and ** at the 5% level, respectively N=229. Percentage of right predictions. Log-likelihood = −43.088. Log-likelihood with constant term only = −88.919. Readers interested in the full calculation of the elasticity at the means and the weighted aggregate elasticity should consult my dissertation, discussed in Gotkin (1997)

of the Representative personal characteristic variables, except for party affiliation and being a Southern Democrat, are individually or jointly statistically significant. I find no evidence that any personal economic interests drove voting behavior. The effect of being a member of the Democratic party alone is statistically insignificant, but negative.39 However, Southern Democrats were on average statistically significantly more likely to vote against the bill—73% of the Representatives who voted against the bill were Southern Democrats (see Table 4.4). The primary reason behind Democratic opposition to the bill was the belief that the bill infringed on State sovereignty, and Southern Democrats were among the most strident supporters of these rights.40 Representative John Reagan of Texas served constituents that were affected by the Southwest Strike, but voted the party line. Democratic ideology did not win out entirely, since the bill would never have passed without some Democratic support, because the House was Democratically controlled. Republicans were in the majority of those who voted in favor of the Arbitration Bill, indicating a small, but statistically insignificant, difference in voting behavior based on party ideology—the negative coefficient on party affiliation. However, it should be noted that not a single Republican voted against the Arbitration Bill, and that being Democratic reduced the overall probability of voting in favor of the bill by 41% on the margin (see Table 4.5, column 3). The second group of variables represents constituent and special interests. Legislators who represented urban constituents were more likely to vote in favor of arbitration. In states affected by the Southwest Strike, Representatives were also more likely to vote in favor of the Arbitration Bill. The extent to which a district 39 The

Democratic dummy variable is insignificant even when the Southern Democrat dummy variable is dropped from the estimation. 40 Congressional Record, 49-1, pp. 3019–21.

4 The Political Economy of the Arbitration Act of 1888

93

voted Democratic had a small, but statistically insignificant negative impact on the probability of voting in favor of the Arbitration Bill. The population percentage variables yielded the expected results. The percent of the population that was Black had a negative, but statistically insignificant effect. Since Blacks were disenfranchised in the South, it is unlikely that this effect was due to the wishes of Black constituents, but is rather capturing regional effects. High concentrations of immigrants in a state’s population had a positive and statistically significant effect on the probability of a Representative voting in favor of the bill. Immigrants tended to reside in urban areas and hold manufacturing jobs, which directly exposed them to the external costs of strikes. Therefore, they would have been highly likely to support the legislation. The employment variables were supposed to help control for the influence of labor on a Representative’s voting behavior. As expected, higher percentages of agricultural employment in a Representative’s state had a negative impact on the probability of voting for the arbitration legislation. Even with a one standard deviation increase in the percent employed in agriculture, the Arbitration Bill would still have passed. Table 4.6 presents the effects of a one standard deviation increase in selected variables on the passage of the Arbitration Bill. The evidence provided during the debates in the House showed strong support for arbitration by manufacturing labor. The Representatives from states with a higher proportion of manufacturing employment were more likely to vote in favor of arbitration. A one standard deviation increase in the percentage of manufacturing

Table 4.6 Effect of a one standard deviation change on the probability of voting in favor of arbitration legislation (selected variables) Variable Percent employed agriculture Percent employed manufacturing Percent employed trade and non-railroad transportation Percent employed railroad Farm wealth per capita Manufacturing wealth per capita Railroad wealth per capita Percent strikes union organized Percent of days lost to railroad strikes Percent of days lost to non-railroad strikes Actual

Probability of voting for legislation 0.590 0.999 0.579

Predicted number of votes in favor of the legislation 132 229 131

0.964 0.614 0.229 0.547 0.821 0.820 0.982 0.868

224 135 56 127 196 196 229 210

Notes: The predicted probability is computed by taking the parameter vector from the probit estimation for Vote4, adding one standard deviation to the selected variable for each voting member, computing individual probabilities, and then taking the average to get the new predicted probability of voting in favor of the Arbitration Bill. The predicted number of votes is the sum of the predicted “yes” votes, i.e., where yˆi = 1

94

J. Gotkin

employment in a Representative’s state would have resulted in unanimous passage of HR7479. Railroad employment also had a positive, but statistically insignificant effect on voting in favor of arbitration legislation. The negative effect of trade and non-railroad transportation employment supports the prediction that a more fundamental issue may have been at the core of the debate on this bill. Individuals engaged in trade and non-railroad transportation were worried about the implications of Federal regulation of the relationship between capital and labor. The wealth variables all have a negative impact on the probability of a Representative voting in favor of the bill. Apparently, manufacturing interests concurred with the railroad interests on the issue of government regulation of the relationship between labor and capital. In fact, a one standard deviation increase in a state’s manufacturing wealth per capita would have caused the defeat of HR7479. Railroad wealth per capita significantly decreased the probability of the bill’s passage, but even with a one standard deviation increase in railroad wealth per capita, the bill would have still passed. Manufacturing interests were very concerned about the implications of this legislation’s effects on the relationship between capital and labor. The strike activity and union involvement in strike activity variables yielded some unexpected results. Railroad strikes had a negative, but statistically insignificant impact on the probability of the Arbitration Bill’s passage. This implies that Representatives in states with relatively more railroad strike activity were less likely to vote in favor of the bill. The overall marginal effect is actually quite small, reducing the probability of passage by only 2.8%. The effect of non-railroad strikes was positive and statistically significant, which was expected. The marginal effect of non-railroad strikes is small, but a one standard deviation increase in a state’s strike activity predicts unanimous passage of the legislation. The measure of union involvement employed in the model refers to all strikes. The reason this measure was used rather than a disaggregated measure of railroad and non-railroad strikes was due to strong multicollinearity between the days lost to railroad strikes and the railroad strikes union involvement variables. The aggregated union involvement variable actually does a better job of predicting the voting behavior of the Representatives. The negative effect for this variable is unexpected. Representatives in states with heavy union involvement in strikes were more likely to vote against the legislation, but not statistically significantly. The legislation passed by a significant majority, and would have passed even if all House members who had abstained from voting had instead voted against the bill. Nevertheless, some interesting insights may be gained by a more careful examination of the non-voting members.

4 The Political Economy of the Arbitration Act of 1888

95

4.7 Examination of the Non-voting House Members In the modeling section, several statistical and modeling results were offered as justification for the exclusion of non-voting Representatives from the analysis. However, a further investigation is warranted given the Washington Post’s article of April 4, 1886, which reported that many House members abstained from voting because they opposed the bill. Poole and Rosenthal (1997) report that the average abstention rate in the fortyninth session of the House of Representatives was approximately 29%. Ninety-four House members, 29.1%, abstained from voting on the HR7479. The vote on HR7479 was lopsided, passing by a margin of 37% (199 to 30). Poole and Rosenthal (1997) actually predict a lower turnout on lopsided issues such as the Arbitration Bill. They claim that turnout should increase with the probability of being decisive. It should also be true that abstentions are more likely on the majority side of a vote. Their results support this “silent majority” hypothesis only after the 63rd House. The behavior of nonvoting members will be analyzed for conformity to Poole and Rosenthal’s (1997) findings, and to test the validity of the Washington Post’s claim. The Post’s article attributes strategic behavior to “many” House members’ voting behavior. There were only nine House members who switched from a no vote on Vote3 to an abstention on Vote4 (see Table 4.7). Of these, only four Representatives, all Southern Democrats, had gone on the record as officially opposed to the legislation. This seems to lend little support to the Post’s claim that “many who would have voted and who did vote to have the bill shut out from further consideration (Vote3), did not dare go on record as voting against it when the straight question was put.”41 The mean characteristics of the non-voting Representatives are more similar to the mean characteristics of those who voted for the bill than against it. As a more rigorous test of both the strategic behavior and the “silent majority” hypotheses, the coefficients estimated in the probit analysis can be employed in computing the probabilities of voting choices, given an individual’s characteristics. These results are reported in Table 4.8. This method of analysis predicts that not voting Representatives would have voted yes with a probability of 82.5%. The majority of Table 4.7 Cross tabulation of voting behavior on Vote3 and Vote4

41 Washington

Post, April 4, 1886.

Vote4 Vote 3

Frequencies 0 1 2 Total

0 28 1 1 30

1 50 123 26 199

2 9 4 81 94

Total 87 128 108 323

96

J. Gotkin

Table 4.8 Prediction of voting behavior for members not voting For all members not voting Predicted probability of voting in favor of arbitration Predicted percentage voting in favor of arbitration For members voting Predicted probability of voting in favor of arbitration Predicted percentage voting in favor of arbitration Percentage that actually voted in favor of arbitration

0.825 88.3% 0.868 90.8% 86.9%

Notes: The predicted probability is computed by taking the parameter vector from the probit estimation for Vote4 and computing individual probabilities, then the mean of these probabilities is computed to get a prediction of the overall probability of voting in favor of the Arbitration Bill for voting and non-voting House members. See Gotkin (1997) and the dissertation it describes for further details

those who abstained from voting would have voted in favor of the Arbitration Bill. These findings support the “silent majority” hypothesis, and reject the presence of any significant strategic behavior.

4.8 Conclusion The goal of this chapter was to identify who supported and who opposed the Arbitration Bill, and which of these groups were successful in influencing the legislative outcome. A probit model was estimated with explanatory variables to capture the role of constituent interest, and legislator private interest. The role of legislator private interests was ruled out as having any significant effect on the passage of the Arbitration Bill. None of the occupation variables or personal characteristic variables were statistically significant. A Representative’s economic interests did not play a significant role in this legislation. The results support the belief that the external cost of strikes played a significant part in the passage of the legislation. Representatives in states with urban populations and high levels of strike activity were more likely to vote in favor of the Arbitration Bill. This places an emphasis on the legislation’s potential benefits in reducing the external costs of strikes to society. How were the private or special interests served? Organized labor supported arbitration legislation, and the effect of union involvement in strikes was predicted to be positive. However, this variable had a negative effect on the passage of the legislation. Perhaps organized labor felt that the bill did not go far enough. Some Representatives, along with the Knights of Labor, favored compulsory arbitration and felt that voluntary arbitration would not be effective.42 Representative Glover of Missouri was outspoken in his support of compulsory 42 Congressional

Record, 49-1, p. 2970.

4 The Political Economy of the Arbitration Act of 1888

97

arbitration, and he voted against the Arbitration Bill. A vote against organized labor may also have been the result of anti-union sentiments. The strike problem was often attributed to the growth of unions. The passage of the bill implies that one group was more successful in influencing the legislative outcome. None of the employment variables were statistically significant, but the effect of manufacturing and railroad wealth was negative and statistically significant. However, the marginal effects of all these variables were small. Did some other factor influence the passage of this legislation over the protests of both railroad and manufacturing capital? Even though party affiliation alone was not statistically significant until it was interacted with the Southern states, it did play a role in the debates and passage of the bill. The Arbitration Bill was charged with politics, and the debates were polarized with respect to party affiliation—not a single Republican voted against the bill.43 The Democrats, especially Southern Democrats, were opposed to this type of legislation. One of the central issues of the debate on the Arbitration Bill was not whether the bill would be effective, but whether or not it was the responsibility and in the power of the Federal government to regulate businesses that operated in the public interest. The bill was attacked as being unconstitutional, because it infringed on State sovereignty. When the earlier version of the Arbitration Bill passed the House, Congress was deadlocked in its debate of bills to regulate the interstate commerce (Gilligan et al. 1989). President Cleveland vetoed this first bill, but by the Fall of 1888 when the Arbitration Act was signed into law, the role of the Federal government in the regulation of the interstate commerce had been established. The Federal government was developing a more acute awareness of the public welfare and its relationship to private business interests. The role of the Federal government in the regulation of the interstate commerce and big business was one of the fundamental issues of the late nineteenth century.44 Many in Congress believed that railroad companies had a duty to serve the public. Railroads as common carriers had an obligation to “run the trains at the hour named, to run them to the places named, to be ready to carry for all who may apply, to accept and to deliver freight in accordance with its advertised schedule. These are obligations which are imposed by the very nature of the work of a common carrier, and by the relation which this service holds to the public, for they are functions of which he can not divest himself.45 There is no denying that the Arbitration Act was legislation passed to promote the public welfare by reducing railroad strike activity, but it was also part of the fundamental debate that revolved around the changing role of the Federal

43 New

York Times, May 28, 1886, 5-1. Supreme Court in the Wabash v. Illinois, 1886, and Munn v. Illinois, 1877 cases set the legal precedence for federal government regulation of interstate commerce. The Interstate Commerce Act of 1887, and the Sherman Antitrust Act of 1890 were major pieces of legislation. 45 Congressional Record, 49-1, p. 3016. 44 The

98

J. Gotkin

government in the late nineteenth century. Whether or not the legislation was effective in reducing strike activity is a question to be answered elsewhere.

References Amemiya T (1981) Qualitative response models: a survey. J Econ Lit 19(4):1483–1536 Congressional Quarterly (1985) Guide to U.S. Elections. Congressional Quarterly, Washington DC Eggert GG (1967) Railroad labor disputes: the beginnings of federal strike policy. University of Michigan Press, Ann Arbor Fiorina MP (1974) Representatives, roll calls, and constituencies. Lexington Books, Lanham Garlock J (1982) Guide to the local assemblies of the knights of labor. Greenwood Publishing Group, Westport Gilligan TW, Marshall WJ, Weingast BR (1989) Regulation and the theory of legislative choice: the Interstate Commerce Act of 1887. J Law Econ 32(1):35–61 Gotkin J (1997) The legislated adjustment of labor disputes: an empirical analysis, 1880 to 1894. J Econ Hist 57(2):481–484 Hausman J, McFadden D (1984) Specification tests for the multinomial logit model. Econometrica 52(5):1219–1240 Kalt JP, Zupan MA (1984) Capture and ideology in the economic theory of politics. Am Econ Rev 74(3):279–300 Klein M (1986) Jay Gould: a revisionist interpretation. Bus Econ Hist 15(1):55–68 Ko TT (1926) Governmental methods of adjusting labor disputes in North America and Australasia. Columbia University, New York Maddala GS (1983) Limited-dependent and qualitative variables in econometrics. Cambridge University Press, Cambridge McGuire RA, Ohsfeldt RL (1986) An economic model of voting behaviour over specific issues at the Constitutional Convention of 1787. J Econ Hist 46(1):79–111 McMurry DLC (1956) The Great Burlington strike of 1888: a case history in labor relations. Harvard University Press, Cambridge Peltzman S (1984) Constituent interest and congressional voting. J Law Econ 27(1):181–210 Poole KT, Rosenthal H (1991) Patterns of congressional voting. Am J Polit Sci 35(1):228–278 Poole KT, Rosenthal H (1997) Congress: a political-economic history of roll call voting. Oxford University, New York Poor HV (1886) Manual of the railroads of the United States. HV & HW Poor, New York Puth R (1988) American economic history. Dryden Press, New York United States Congress (1886) Congressional Directory, 49-1. Government Printing Office, Washington DC US Census Bureau (1882a) Tenth Census of the United States: Population. Government Printing Office, Washington DC US Census Bureau (1882b) Tenth Census of the United States: Mortality. Government Printing Office, Washington DC US Commissioner of Labor (1888) Third Annual Report of the Commissioner of Labor: Strikes and Lockouts. Government Printing Office, Washington DC Welch S, Peters JG (1983) Private interests and public interests: an analysis of the impact of personal finance on Congressional voting on agriculture issues. J Polit 45(2):378–396

Chapter 5

The Political Economy of Bank Entry Restrictions: A Theory of Unit Banking Charles W. Calomiris and Carlos D. Ramírez

Abstract Conventional wisdom has it that entry barriers in banking (for example, historical branch banking restrictions in the United States) are motivated by special interest groups, with small, local banks playing a central role in lobbying for protection. In particular, it is thought that unit (single-office) banks in the United States favored branching restrictions because they wanted (and needed) protection from competition from large, multi-office banks. Historically, however, branch banking restrictions also had the support of some classes of borrowers. Borrower support for entry barriers varied across states, and varied over time within states. We show that entry barriers affect the terms on which borrowers access credit, and can sometimes be beneficial for some classes of borrowers. While it is true that branch banking tends to increase the overall supply of credit to borrowers, it is also true that in the presence of imperfect capital markets, borrowers may benefit from barriers to entry because such barriers limit the options of the banks in the loan market. We develop a model that shows how branching restrictions (or more generically, barriers to varying the inter-regional allocation of credit by banks) create strategic advantages for borrowers that hold their wealth in the form of immobile factors of production (e.g., land). These advantages tend to be present only when borrowers’ net worth levels are sufficiently high. Our results indicate that bank clients, not just unit bankers themselves, may have supported unit banking laws out of informed self interest. We argue that these results also have broader implications for explaining the economic circumstances under which entry barriers to global banking are erected or removed in emerging market economies today.

C. W. Calomiris Columbia Business School, New York, NY, USA NBER, Cambridge, MA, USA e-mail: [email protected] C. D. Ramírez () George Mason University, Department of Economics, Fairfax, VA, USA e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2018 J. Hall, M. Witcher (eds.), Public Choice Analyses of American Economic History, Studies in Public Choice 37, https://doi.org/10.1007/978-3-319-95819-4_5

99

100

C. W. Calomiris and C. D. Ramírez

5.1 Introduction Entry barriers in banking have been an important fact of life historically and currently in many countries. In emerging market economies today, one of the revolutionary changes taking hold in some countries is the entry of foreignowned banks on more or less equal footing with domestically owned institutions. For example, Argentina, Chile, Mexico, and Brazil saw their banking systems transformed into largely foreign-owned systems by the end of the 1990s. But as recently as the 1980s, foreign ownership in these countries was the exception rather than the rule. In many other developing countries, particularly in Asia, there also have been rigid barriers to foreign entry, which have been relaxed recently (World Bank 2018). Why, in general, is there so much resistance to competition from foreign-owned banks? And why is it that resistance is sometimes overcome, as it has been in many emerging market countries? One fact that many observers have noticed is that, historically, limits on foreign entry tend to be relaxed after severe adverse economic shocks. For example, in Mexico the financial crisis of 1994–1995 clearly set the stage for the liberalization of foreign entry after 1997. The same pattern is visible in the history of the relaxation of entry barriers in the United States. During the bank distress years of 1920–1939, 15 states relaxed their branching restrictions, while in the four decades that followed (1939–1979) only four states relaxed branching limits. When bank distress returned in the 1980s, once again 15 states relaxed their branching rules (Mengle 1990; Kroszner and Strahan 1999; Calomiris 2000, pp. 63–67; Calomiris and Haber 2014, Chapters 6–7). One explanation of barriers to entry revolves around the role of local bankers in lobbying for entry barriers. That perspective could also explain the link between economic distress and the relaxation of entry barriers, if economic distress weakens the political power of the local banks. This is certainly a plausible explanation, and in our view, captures an important part of the political struggle over entry barriers. But this is not the only possible explanation, and we will argue that there are reasons to believe that, by itself, it is an inadequate explanation. In this paper we develop an alternative theoretical approach to explaining entry barriers, which focuses on the gains certain classes of borrowers receive from those barriers under certain circumstances. We apply our model of borrower preference for entry barriers to the historical case of historical U.S. bank entry barriers laws limiting branching. We argue that a perspective that takes account of borrowers’ preferences is necessary to explain aspects of the political choice for limits on branching in the United States, and we present theoretical explanations for why borrowers sometimes supported entry barriers.

5 The Political Economy of Bank Entry Restrictions: A Theory of Unit Banking

101

5.1.1 Entry Barriers in U.S. History Branch banking restrictions have been among the longest-lasting financial regulations in the United States. State laws that had restricted or prohibited the establishment of commercial bank branches date back to the last century. Except for the First and Second Banks of the United States (1791–1811, and 1816–1836, respectively), antebellum state bank charters dictated the location and activities for each bank at the state level. In the landmark Supreme Court case, Bank of Augusta v. Earle (1939), the court affirmed that states could limit banking activities within their states by banks chartered in other states, effectively ending the possibility of interstate branching without federal government action. Before the Civil War banks chartered by Northern states were unit banks, while many states in the South permitted branch banking. In the postbellum period, branching restrictions continued to be a matter of state law. The creation of national banks under the National Banking Act of 1863 did not materially alter that fact. National banks were chartered to operate within individual states. Although there was no explicit prohibition of within-state branching in the National Banking Act, the Comptrollers of the Currency, who oversaw national banks, interpreted some of the Act’s clauses as implicitly prohibiting the establishment of branches. Although the McFadden Act of 1927 allowed national banks to establish branches, they were allowed these branches only if state law permitted it, and even in such cases, branching was restricted to the city limits of where the main branch was located. During the banking distress of the agricultural price declines of the 1920s and the Great Depression of the 1930s, many unit banks failed, and branching became more popular. That threat to unit banking was not sufficient to produce a change in law at that time, however, owing especially to the federal government decisions to provide deposit insurance (which primarily benefited small unit banks in the 1930s), and the government’s opposition to branch banking or other means of consolidation (Calomiris 2010; Calomiris and Haber 2014, Chapter 6). The one-town, one-bank structure that characterized the commercial banking industry throughout most of U.S. history has only recently given way to nationwide branch banking, beginning in the 1980s. Change occurred in the wake of a new wave of bank distress, as well as rising urbanization, which made rural interests less powerful than they had been in the early twentieth century. The relaxation of branching limits occurred first through changes in state law and regional interstate agreements that permitted branching. These initial changes culminated in the enactment of the Riegle-Neal Interstate Banking and Branching Efficiency Act (IBBEA) in 1994, which further promoted cross-state mergers and acquisitions in the banking industry.1 Despite an enormous literature on the economics and politics of branching limits, a convincing theoretical explanation for branching restrictions remains elusive.

1 For

reviews, see Berger et al. (1995) and Calomiris (2000).

102

C. W. Calomiris and C. D. Ramírez

Regulation can be welfare enhancing when competitive market forces produce monopoly or negative externalities. However, it is hard to justify the existence of branch banking restrictions on these grounds. Branching restrictions decrease the total number of banks that can compete within a local market. Fewer banks with many branches tend to produce greater entry and competition, especially in sparsely populated areas. Branching restrictions tend to limit the supply of credit, increase bank failure risk, and can promote monopoly power by local banks. In recognition of the many shortcomings of a unit banking system, many researchers maintain that vested interests within the industry, rather that the political preferences of consumers, best explain the existence and duration of branch banking restrictions. (See, for example, Economides et al. (1996), and Kroszner and Strahan (1999).) The most obvious interest group identified in this literature is unit bankers, who stood to gain a great deal from restrictions that prevented the branches of larger banks headquartered elsewhere from entering their local markets. This explanation, although compelling, seems incomplete by itself for several reasons. First, there is the unmistakable fact that unit banking laws were often quite popular. Limits on branching were a prominent part of William Jennings Bryan’s populist platform. In at least one case, in Illinois in 1924, the question of whether to permit branching was put to a referendum and was defeated (White 1985). Clearly, there was more to the support for unit banking than the political lobbying of unit bankers. Second, an explanation that focuses on the rent-seeking behavior of unit bankers neglects the fact that competition among unit bankers within a city or county can be just as effective as entry by branching banks in limiting the rents of unit banks. With the exception of the most rural locations, towns, cities, and counties typically contained many competing local banks, and the dissemination of the automobile by the 1920s increased the range of competition among nearby unit banks. That history suggests that it would be useful to construct a model of the political economy of the support for unit banking that does not rely entirely on unit bankers’ desire to preserve monopoly rents. Third, as we show more formally below, some bank borrowers likely stood to gain strategically from supporting limits on branch banking. Unit banking served as a commitment device to prevent local banks from moving funds out of the local economy. In our model, we consider circumstances under which borrowers might have been advantaged by these limits, despite their costs. Our model suggests that some borrowers of unit banks were willing to absorb the costs associated with the regulatory choice of unit banking (reflected in higher loan rates and lower deposit rates) because unit banking provided benefits to those borrowers that more than offset these costs. According to our model, the benefit borrowers receive stems from differences in loan pricing policies of unit banks that come about from imperfections in capital markets. The model predicts that, when the net worth of landowners (farm-owners and homeowners) is sufficiently high, this benefit accrued by borrowers more than offsets the disadvantages from limiting branching.

5 The Political Economy of Bank Entry Restrictions: A Theory of Unit Banking

103

Calomiris (2000, p. 65) shows that, around the turn of the twentieth century, states with relatively high per capita wealth in rural areas supported restrictions on branching. The recent empirical work of Rajan and Ramcharan also offers support for our model. Rajan and Ramcharan (2011) find that land holding concentration is associated with fewer banks per capita in U.S. counties during the 1920s. Controlling for a large array of factors, including state fixed effects, their results indicate that landowner wealth not just suppressed the number of banks at the county level, but also increased interest rates, and reduced local loan supply. In subsequent work, Rajan and Ramcharan (2016) provide more direct evidence of landowner interests influencing local bank legislation through congressional votes and other political channels. Their evidence relies on a detailed analysis of the McFadden Act of 1927, which permitted national banks (and state bank member of the Federal Reserve) to open branches in the cities where the parent institution was located, as long as state laws also permitted branch banking for state-chartered institutions. Their voting model shows that land concentration at the district level significantly reduced the probability of a “Yea” vote in the initial version (1926) of the act, even after controlling for a large set of factors.

5.2 Theory We develop a model of borrower preferences for unit or branch banking. The model depends on imperfect capital markets to generate a borrower demand for unit banking under some states of the world (when wealth is high). It emphasizes the benefits for borrowers from the loan pricing strategy that unit banking produces.2 That element of the model can explain why collapses in land prices (such as in the U.S. in 1920s and 1980s) were associated with erosion of political support for maintaining entry restrictions.

5.2.1 Branch Banking, Diversification, and Loan Pricing In a geographical place (which could be a state or a country), there are two regions (which could be considered counties or states). Each region has a continuum of borrowers of mass Nk , where k denotes the region (either 1 or 2). In each region, there are two types of borrowers and two types of projects: Type A borrowers can engage only in Type 1 projects, which have a certain gross return, Rc ; Type B borrowers can engage in either Type 1 projects or Type 2 projects, which pay a gross return of Rs if successful and 0 if unsuccessful. The probability of a successful outcome is p, so (1-p) is the probability of an unsuccessful outcome. We

2 The

model adapts some features of Calomiris and Hubbard (1990).

104

C. W. Calomiris and C. D. Ramírez

assume that the size of the project, X, is the same for both types. Also, we assume that the gross return of a Type 1 project exceeds the expected gross return of a Type 2 project, with the safe return in between. That is, Rc > (1+r)X > pRs , where r is the risk-free interest rate. Type A borrowers constitute α percent of total borrowers in a region and Type B borrowers constitute (1 − α) percent. Borrowers are aware of their own type, and are risk-neutral. There is only one lending period in the model, which we identify as Period 2. (We discuss what takes place in Period 1 below.) At the beginning of the lending period, borrowers demand and the bank supplies loan funds for the projects, given the amount of collateral owned by the borrowers. Banks are risk-averse. We model banking behavior using the standard portfolio selection framework. All banks face a choice as to how to invest their funds: either in loans (which will finance projects) or in riskless assets (“government bonds”). After choosing the proportion of assets that will be invested in loans, banks choose the interest rate to charge borrowers. Unit banks can make loans only to borrowers in the region in which the bank is located. Branch banks can make loans to borrowers in both regions. If θkb is the proportion of bank assets invested in loans in region k (region 1 or 2), under banking regime b (unit or branch), (1 − θkb ) is the proportion of assets invested in government bonds. Since branch banks can offer loans in both regions, they face a larger set of portfolio choices than unit banks do. They choose the proportion of assets to be invested in Region 1, θ1branch , the proportion of assets to be invested in Region 2, θ2branch , and the proportion of assets to be invested in government bonds, (1 − θ1branch − θ2branch ). To generate demand for loan diversification under branch banking, we assume, without loss of generality, that the outcomes of Type 2 projects are perfectly negatively correlated across the two regions. Hence, if a successful outcome takes place in Region 1, an unsuccessful one takes place in Region 2.3 We assume that the total quantity of funds available for lending in a given region is not sufficient to fund all borrowers in that region. Thus, if F k is the total quantity of funds available in Region k, Fk < XN k . For clarity and simplicity, we make the assumption that F1 = F2 = F¯ < min[XN 1,XN 2 ]. Since (F¯ /XN k ) < 1, there may be credit rationing in equilibrium if moral hazard limits interest rate increases to clear the market and if not enough funds can be imported from the other region (which only branch banks will be able to do).

5.2.1.1

Timeline of Events

Period 1 In period 1, voters (borrowers) in both regions choose the type of banking regime that will service the region (which can be conceived to be a particular state in

3 In

a model with a large number of regions, independence of outcomes, rather than negative correlation, would produce similar results.

5 The Political Economy of Bank Entry Restrictions: A Theory of Unit Banking

105

the United States). They do so by voting on a law that either allows or disallows branching. Their basis for preferring one regime to the other is their expected profits. For simplicity, in some of our discussion below we assume that initial collateral levels are the same across regions. In the appendix we show that this does not affect our results qualitatively. Period 2 The level of collateral in each region is exogenously given at the start of the lending period, as described above. Thus, all borrowers in region k start with the same amount of collateral, Ck , and demand the same loan amount, X − Ck , where X is the project size. We assume that X is the same in both regions. (a) First, the bank chooses θkb . Assuming it is positive, the bank then chooses an interest rate, ik , to charge its borrowers. The same rate is charged to all borrowers because borrowers are not observably different. The bank sets this interest rate without knowing: (i) the type of any given borrower, only that the probability of encountering a Type A borrower is α and the probability of encountering a Type B borrower is (1 − α); (ii) the type of shock that will occur during the period, only that the probability of a good outcome is p and the probability of a bad outcome is (1 − p). (b) Borrowers choose whether to accept the interest rate offer. Assuming that they do, Type A will do project 1, while Type B borrowers must choose the type of project they will undertake (either 1 or 2). (c) At the end of the period, after the shock occurs, the borrowers realize their returns from their projects they repay their loans, if possible, and consume the rest, if any.

5.2.1.2

Solution of the Model

In order to solve the model, we first determine what takes place in Period 2, the lending period. Once the expected payoffs are derived, we can analyze the voting decision that takes place in Period 1.

5.2.2 Period 2: Lending Period Demand for Loans The following analysis applies to a particular region only. In Period 2, all borrowers start with collateral Ck , and the bank is unable to distinguish between the two types of borrowers.

106

C. W. Calomiris and C. D. Ramírez

5.2.2.1

Type A Borrowers

A Type A borrower can only engage in a Type 1 project. His expected return from this investment is Rc −(1+ik )(X−Ck ). The (gross) opportunity cost of his collateral is the alternative (risk-free) investment with rate of return, 1 + r. Hence, the total return (net of the opportunity cost) is: π1,k (ik ,Ck ) ≡ Rc − (1 + ik )(X − Ck ) − (1 + r)Ck

(5.1)

Since, by assumption, funds are scarce, borrowers do not know with certainty whether they will be offered credit. However, they form expectations of bank credit offers based on a rational understanding of a bank’s optimal credit allocation decision. Let the probability of being offered funds in region k under banking regime b be λbk . Naturally, this probability will be related to θkb , and will be determined by Ck , and the choice of banking regime. After accounting for the probability of being offered credit, a Type A borrower’s expected return (before being granted credit) is: ΠkA,b ≡ λbk (θkb,Ck ) × π1,k (ik ,Ck ) 5.2.2.2

(5.2)

Type B Borrowers

A Type B borrower can engage in either a Type 1 or a Type 2 project. His expected return from investment in a Type 1 project is π1,k (ik ,Ck ). His expected return from investment in a Type 2 project for period 1 is: π2,k (ik ,Ck ) ≡ p{Rs − (1 + ik )(X − Ck )} − (1 + r)Ck

(5.3)

After accounting for the probability of being offered credit, a Type B borrower’s expected return (before being granted credit), if he chooses to do a Type j project is: ΠkB,b ≡ λbk (θkb,Ck ) × πj,k (ik ,Ck )

(5.4)

We now derive conditions that determine project choice and profitability for both Type A and Type B borrowers. Assuming he gets credit, a Type A borrower’s expected return is π1,k (ik ,Ck ). He will only wish to borrow as long as this return is positive. Hence, a Type A borrower will borrow (invest) only if ik ≤ ikmax,1 , where4 : π1,k (ikmax,1,Ck ) = 0

4 Note

that π1,k (ik ,Ck ) is decreasing in ik .

(5.5)

5 The Political Economy of Bank Entry Restrictions: A Theory of Unit Banking

Equivalently:

 1 + ikmax,1 ≡

Rc − (1 + r)Ck X − Ck

107

(5.6)

A Type B borrower can engage in either a Type 1 or a Type 2 project. His expected return from investment in a Type 1 project is π1,k (ik ,Ck ); his expected return from investment in a Type 2 project is π2,k (ik ,Ck ). A Type B borrower will prefer a Type 1 to a Type 2 project if, for a given collateral level, π1,k (ik ,Ck ) ≥ π2,k (ik ,Ck ). This will happen as long as ik ≤ ikswitch , where5 :  1 + ikswitch

≡

Rc − pRs (1 − p)(X − Ck )

(5.7)

Hence, a Type B borrower will borrow for a Type 1 project only if ik ≤ min[ikswitch,ikmax,1 ]; he will borrow for a Type 2 project only if ikswitch < ik ≤ ikmax,2 , where: π2,k (ikmax,2,Ck ) = 0 (5.8) Equivalently: 1 + ikmax,2

≡

 R − ( 1 )(1 + r)C s k p X − Ck

(5.9)

Case 1 (Ck < C ∗ ) We show in the appendix that if Ck is below the threshold level C ∗ , then ikswitch < ikmax,1 < ikmax,2 . The threshold level C ∗ is defined as follows: C∗ ≡



p 1−p



RS − RC 1+r

(5.10)

Case 2 (Ck ≥ C ∗ ) We show in the appendix that if Ck is above the threshold level C ∗ instead, then ikswitch ≥ ikmax,1 ≥ ikmax,2 . The choices of project type as a function of interest rates and collateral level are summarized in Figs. 5.1 and 5.2. Supply of Loans: Bank Behavior and Interest Rate Offers We model the bank’s risk-averse behavior using Tobin’s (1958) portfolio diversification approach, which incorporates a risk-free alternative to the efficient set of feasible investment portfolios.6 Since the bank’s problem depends on the banking

that ikswitch is simply the interest rate that satisfies the following: π1,k (ikswitch,Ck ) = π2,k (ikswitch,Ck ). 6 Portfolio allocation models in banking have a large history in the literature. Some of the wellknown papers include Pyle (1971, 1972). For a comprehensive survey see Santomero (1984). 5 Note

108

C. W. Calomiris and C. D. Ramírez

Both Type A and Type B borrowers doing Type 1 projects 0

Type A borrowers doing Type 1 projects; Type B borrowers doing Type 2 projects iswitch k

Only Type B borrowers doing Type 2 projects

No borrowing

ikmax,1

imax,2 k

Fig. 5.1 C < C ∗

Both Type A and Type B borrowers doing Type 1 projects 0

Both Type A and Type B borrowers doing Type 1 projects ikmax,2

No borrowing

imax,1 k

iswitch k

Fig. 5.2 C ≥ C ∗

regime in which it is operating (either unit or branching), we discuss each solution separately.

5.2.2.3

Unit Banks

Unit banks have several risk-return combinations available that span their portfolio choice set. These combinations depend on the level of collateral. We list them below in Table 5.1. Lemma 5.1 establishes the range of α values that are of interest in our model, in particular when Ck < C ∗ . Intuitively, it states that when Ck < C ∗ , if the fraction of Type A borrowers in the market is high enough, the bank will find it more profitable to lend at the interest rate that makes Type A borrowers indifferent as to whether to undertake the project, even though it makes Type B borrowers do Type 2 projects. Since α is exogenous, we assume it is greater than α ∗ throughout the rest of the model.7

α < α ∗ the model yields uninteresting or trivial results, especially if α is so low that ρr > ρi max,1 .

7 When

k

5 The Political Economy of Bank Entry Restrictions: A Theory of Unit Banking

109

Table 5.1 Risk-return alternatives under unit banking Interest rate

Ck < C ∗ (ikswitch < ikmax,1 < ikmax,2 ) Risk Expected return

1. r on bonds None ikswitch

ρr ≡ (1 + r)

on None. All ρi switch ≡ (1 + ikswitch ) k borrowers do Type 1

2. loans

3. ikmax,1 on loans 4. ikmax,2 on loans

Some. (1 − α) ρi max,1 ≡ (α + (1 − α)p) k fraction do (1 + ikmax,1 ) Type 2 ρi max,2 ≡ p(1 + ikmax,2 ) Significant. k All do Type 2

Ck ≥ C ∗ (ikswitch ≥ ikmax,1 ≥ ikmax,2 ) Risk Expected return None ρr ≡ (1 + r) ρi switch ≡ N.A. No borrower k participates None. All do Type 1

ρi max,1 ≡ (1 + ikmax,1 )

None. All do Type 1

ρi max,2 ≡ (1 + ikmax,2 )

k

k

Lemma 5.1 (Minimum Fraction of Type A Borrowers, α ∗ ) For any value Ck , there is an α ∗ ≥ 0, such that for α > α ∗,ρi max,1 > maxρi switch ,ρr . k

k

Proof See Appendix.

5.2.2.4

Collateral and Credit Allocation

Proposition 5.1 (Bank’s Expected Return Order) A. There is a critical collateral level, C low ≤ C ∗ , such that for Ck < C low , the bank’s expected returns are ordered as follows: ρi max,1 > ρr > ρi switch k

k

For Ck ≥ C low , the bank’s expected returns are ordered as follows: ρi max,1 ≥ ρi switch ≥ ρr k

k

(continued)

110

C. W. Calomiris and C. D. Ramírez

Proposition 5.1 (continued) B. For any C,

ρr > ρi max,2 k

Where C low is defined as: C low ≡ C ∗ −



Rc −X 1+r



Proof See Appendix.

Proposition 5.2 (Unit Bank’s Optimal Interest Rate Offer and Optimal Allocation of Credit) a. For Ck < C low : Invest θkunit (Ck < C low ) < 1 on loans at interest rate ikmax,1 ; invest the rest on government securities. b. For C ∗ > Ck ≥ C low : Invest θkunit (C ∗ > Ck ≥ C low ) = 1 on loans. Offer some loans at interest rate ikswitch , and the rest at interest rate ikmax,1 . Since the bank cannot distinguish between Type A and Type B borrowers, it will randomize among applicants when deciding who receives loan offers at interest rate ikswitch or ikmax,1 . c. For Ck ≥ C ∗ : Invest θkunit (Ck ≥ C ∗ ) = 1 on loans at interest rate ikmax,1 . Proof (a) Proposition 5.1 indicates that when Ck < C low , risk-return combinations 2 and 4 (from Table 5.1) are strictly dominated by alternative 1 (from Table 5.1). Although alternative 3 offers a higher return for the bank, it is riskier than alternative 1. According to Tobin’s (1958) optimal portfolio allocation model, a risk-averse bank will select a portfolio that invests a fraction of its assets in alternative 1, and the rest in alternative 3. (b) For C ∗ > Ck ≥ C low , risk-return combinations 1 and 4 (from Table 5.1) are strictly dominated by alternative 2 (from Table 5.1). Although alternative 3 offers a higher return for the bank, it is riskier than alternative 2. According to Tobin’s (1958) optimal portfolio allocation model, a riskaverse bank will select a portfolio that invests a fraction of its assets in alternative 2, and the rest in alternative 3. (continued)

5 The Political Economy of Bank Entry Restrictions: A Theory of Unit Banking

111

Proposition 5.2 (continued) (c) For Ck ≥ C ∗ , none of the alternatives carry any risk. Hence, the bank will select the highest interest rate compatible with borrower participation in the market, ikmax,1 .

Corollary 5.1 (Optimal Project Choice for Type B Borrowers) Given banks’ optimal interest rate offers, a Type B borrower will choose to do a Type 2 project if his collateral is below C ∗ and the interest rate offer is larger than ikswitch . He will choose to do a Type 1 project if his collateral is above C ∗ or the interest rate offer is below ikswitch . Proof Follows from Figs. 5.1 and 5.2.

5.2.2.5

Branch Bank Behavior: Interest Rate Offers and Credit Allocation Across Regions

The branch bank must decide how to allocate funds between the two regions, and how much to invest in the risk-free alternative. Since a branch bank can invest in either region, it has a total of seven possible interest rate alternatives from which to choose: i1switch , i1max,1 , or i1max,2 , in Region 1; i2switch , i2max,1 , or i2max,2 , in Region 2; or r in government bonds. Because the branch bank’s risk-return choices include those available to unit banks, it can replicate the unit bank’s portfolio in each region, and therefore, its expected profits. In practice, however, the risk-return choice set is much larger for branch banks than for unit banks. In fact, because of the perfectly negatively correlated outcomes across regions, the choice set will include the zero-risk portfolio alternative. Both ikswitch , and ikmax,1 increase with Ck , while ikmax,2 decreases with Ck .8 Hence, the level of collateral determines interest rate offers as well as the allocation of

8 Note

that

∂(1+ikmax,1 ) ∂Ck

> 0, since Rc > (1 + r)X; similarly,

It is straightforward to note that

∂(1+ikswitch ) ∂Ck

> 0.

∂(1+ikmax,2 ) ∂Ck

< 0, since pRs < (1 + r)X.

112

C. W. Calomiris and C. D. Ramírez

funds. It follows that equilibrium outcomes will also depend on collateral levels. In particular, conditions on C low determine the branch bank’s expected return order, while conditions on C ∗ determine credit allocation, as established in the following propositions.

Proposition 5.3 (Branch Bank’s Expected Return Order) When collateral level C low is: Expected Return Order is: 1. C2 < C1 < C low ρi max,1 > ρi max,1 > ρr > ρi switch > ρi switch 2. C2 < C low < C1

1

2

1

2

1

1

1

2

1

2

ρi max,1 > ρi max,1 > ρr ρi max,1 > ρi switch > ρr > ρi switch

3. C low < C2 < C1

2

ρi max,1 > ρi max,1 > ρi switch > ρi switch > ρr 1

2

For any C,ρr > ρi max,2 > ρi max,2 . 2

1

Proof The result follows from Proposition 5.1, and the fact that ikmax,1 increases with Ck .

The following proposition establishes how the branch bank will behave assuming the same degree of risk aversion as that of the unit bank.

Proposition 5.4 (Branch Bank’s Optimal Interest Rate Offers and Credit Allocation Across Regions) 1. C2 < C1 < C ∗ : Invest 0 < θ1branch (C2 < C1 < C ∗ ) < 1 in Region 1 and 0 < θ2branch (C2 < C1 < C ∗ ) < 1 in Region 2 with θ1branch (C2 < C1 < C ∗ ) + θ2branch (C2 < C1 < C ∗ ) = 1. Offer interest rate i1max,1 on loans in Region 1 and i2max,1 on loans in Region 2. 2. C2 < C ∗ ≤ C1 : Invest θ1branch (C2 < C ∗ ≤ C1 ) = 1 in Region 1 on loans at interest rate i1max,1 . 3. C ∗ ≤ C2 < C1 : Invest θ1branch (C ∗ ≤ C2 < C1 ) = 1 in Region 1 on loans at interest rate i1max,1 . Proof See Appendix.

5 The Political Economy of Bank Entry Restrictions: A Theory of Unit Banking

113

5.2.3 Analysis of Voting Period At the beginning of period 1, voters (borrowers) must decide on the banking regime they prefer. Borrowers in Region k prefer to be serviced by the type of regime that gives them the highest expected profits, λbk (θkb,Ck ) × πj,k (ik ,Ck ), where j indicates the type of project undertaken, and b the banking regime. In our model thus far, Type A borrowers sometimes do not earn positive profits because, by assumption, in many states of the world banks are able to extract all profits from them through the loan contracts that are offered (e.g. πl,k (ik ,Ck ) = 0 when ik = ikmax,1 ). Type B borrowers, by contrast, are able to earn positive profits in more states than Type A borrowers, because they sometimes find it advantageous to pretend to undertake the riskless project, when actually undertaking the risky project. Of course, more realistically, borrowers would earn expected “control rents” from undertaking projects whenever they receive funding. We define that control rent as ε for borrowers receiving loans. Thus, in the context of our model, a Type A borrower who is offered a loan at rate ik = ikmax,1 will accept the loan since he will receive ε in control rents, even though π1,k (ik ,Ck ) = 0. A Type B borrower who chooses to undertake a Type 2 project when ik = ikmax,1 , will earn π2,k (ikmax,1,Ck ) ≡ s > 0, plus the control rent ε. We adopt this assumption in order to ensure that borrowers have sufficient stake in choosing between branching and unit banking. In our discussion of the voting, we will allow collateral to be either “high” or “low” (that is, below or above the critical value C low ), but we will assume, for simplicity, that collateral is always below C ∗ . Our central conclusion about loan markets and bank entry barriers that low levels of collateral will lead borrowers to prefer branch banking over unit banking, and that at high levels of collateral, that preference is reversed is not dependent on the assumption that C < C ∗ .9 For simplicity, and without loss of generality, we will also assume in this section that the levels of collateral are the same in the two regions. Finally, we also make the simplifying assumption, again without loss of generality, that banks are sufficiently risk-averse. Specifically, we assume that when forced to undertake risky lending, banks withdraw from the loan market (i.e. θkb is zero). This assumption simplifies the computation of λbk , but is not necessary to generate our qualitative results. We now consider the expected profits of Type A and Type B borrowers. We begin with the case where initial collateral is low. As shown in Table 5.2, in this “low-collateral” case, under the above simplifying assumptions, Type A borrowers face two possibilities: receiving credit when the loan interest rate is set at ik = ikmax,1 (and earning control rent ε), or not receiving credit, and thus earning no profit. Type 9 Specifically, for very high range of collateral values (those in excess of C ∗ ), where collateral levels

are identical across regions, borrowers are indifferent between choosing unit or branch banking.

114

C. W. Calomiris and C. D. Ramírez

B borrowers face a similar scenario, but with different profits: receiving credit at ik = ikmax,1 and earning s plus ε, or receiving no credit, and hence, no profit. Both types of borrowers face the same probability of being granted credit. If branching is chosen, that probability is F¯ /XN k . If unit branching is chosen, the probability of being granted credit is zero. Thus, in Case 1, Type A borrowers expect to earn (F¯ /XN k )ε while Type B borrowers expect to earn (F¯ /XN k )(s + ε under branching, and both types expect to earn zero under unit banking. Clearly, in this low-collateral case, both types of borrowers prefer branch banking. When initial collateral is in the “high” range (above C low ), the result is the reverse: both types of borrowers prefer unit banking. In this case, the probability and payoffs from being granted credit, if branching is chosen, are the same as in the case of low collateral. But unit banking delivers higher expected profits to both types of borrowers than it does in the low-collateral case, and that level of profit is also higher that the expected profit from branching in the high-collateral case. The reason is that under unit banking, the payoffs conditional on being granted credit are higher, and the probability of being granted credit are the same as under branching. The reason profits are higher is that, in the high-collateral case, the interest rate on

Table 5.2 Borrower expected returns in high and low collateral cases order Collateral Value of θkb 1.

C2 = C1 θ1unit = 0 ∗ < C low

Value of λbk λunit =0 1

θ1branch = 0.5 λbranch = 1 θ2unit

=0

λunit 2

C low < θ1unit = 1 C2 = C1

λunit = 1

θ2branch

=1

λunit 2

=

F¯ XN 2

F¯ XN 1

θ1branch = 0.5 λbranch = 1

F¯ XN 1

F¯ XN 2

θ2branch = 0.5 λbranch = 2

Π1A,unit = 0ε = 0 Π1A,branch = Π2A,unit

=0

θ2branch = 0.5 λbranch = 2 2.

F¯ XN 1

Borrowers expected returns Type A Type B

F¯ XN 2

F¯ XN 1 ε

= 0ε = 0

Π1B,unit = 0(s + ε) = 0 Π1B,branch = F¯ XN 1 (s

+ ε)

Π2B,unit

= 0(s + ε) = 0

Π2B,unit = 0(s+ε) = 0 Π2B,branch = F¯ XN 2 (s

Π1A,unit =

F¯ switch,C ) 1 XN 1 π1,1 (i1

Π1A,branch =

F¯ XN 1 ε

Π2A,unit = F¯ switch,C ) 2 XN 2 π1,2 (i2 Π2A,branch =

F¯ XN 2 ε

+ ε)

Π1B,unit =

F¯ switch,C ) 1 XN 1 π1,1 (i1

Π1B,branch = F¯ XN 1 (s

+ ε)

Π2B,unit = F¯ switch,C ) 2 XN 2 π1,2 (i2 Π2B,branch = F¯ XN 2 (s

+ ε)

5 The Political Economy of Bank Entry Restrictions: A Theory of Unit Banking

115

loans is set at ik = i switch , implying a positive amount of rents earned by borrowers in addition to ε.10 The intuition for the result that borrowers prefer unit banking when collateral is high is as follows: When collateral is high, under unit banking, banks will choose to lower interest rates as a means of solving the moral-hazard problem (since low interest rates induce borrowers to invest in the good project).11 Branch banking would induce banks, instead, to keep interest rates higher and rely on diversification of risk across regions to limit losses from moral hazard. Thus, when collateral is high, borrowers prefer unit banking. In contrast, when collateral is low, under unit banking, banks will not supply credit, while under branch banking, diversification will induce banks to continue to supply credit. In summary, the empirical implications of our model are as follows: 1. In “poor” states of the world (when collateral levels are low), branching will be unambiguously preferred; 2. In relatively “rich” states of the world, unit banking will be preferred.

5.3 Conclusions We develop a model to show why some bank borrowers (those who own immobile factors of production) find unit banking attractive in high net worth states of the world. In particular, we argue that the unit bank earnings premium (the higher loan interest rates and lower deposit interest rates of unit banks) could represent a payment for the implicit cost of providing credit insurance to these borrowers. Although we have argued that branching restrictions were beneficial to certain segments of the population, we emphasize that it does not follow that branching restrictions were beneficial to society as a whole. Calomiris (2000, 2010) review in detail the reasons to believe that branching restrictions were highly socially costly from the standpoint of macroeconomic growth and stability. Nevertheless, in states where a critical mass of borrowers existed that supported these entry barriers, those borrowers were able to successfully enact or retain unit banking. Our interpretation, which is consistent with evidence about historical differences across states reported in Calomiris (2000) and Rajan and Ramcharan (2011, 2016), shifts attention away from unit bankers as the prime special interest group to support unit banking, and focuses instead on certain classes of bank borrowers.

10 Note

that is true even for Type B borrowers, since for ik ≤ ikswitch , π1,k (ik ,Ck ) ≥ π2,k (ik ,Ck ) >

π2,k (ikmax,1,Ck ) ≡ s. 11 Technically, as Proposition

5.2 indicates, the unit bank will offer loans at ikswitch and ikmax,1 , and randomize among borrowers as to who gets which interest rate. The proportion of borrowers receiving credit at ikswitch will increase with the degree of risk-aversion of the bank. Thus, even if the bank is not very risk-averse, borrowers will prefer unit banking as long as some of them receive credit at interest rate ikswitch .

116

C. W. Calomiris and C. D. Ramírez

Beyond its historical application, our theory offers some implications for today’s global wave of mergers and consolidation in the banking industry. Despite the fact that branching is superior to unit banking on macroeconomic growth and stability grounds, it is likely that some countries will continue to limit entry by foreign banks, perhaps at the behest of domestic borrowers that own immobile factors of production. Acknowledgements We would like to thank, without implicating, Bryan Caplan, Mark Crain, Tyler Cowen, Claudia Goldin, Kyle Kauffman, John Matsusaka, Allan Meltzer, David Moss, Peter Temin, Jeff Williamson, as well as seminar participants at Harvard University, George Mason University, the FDIC, and the 2004 European Business History Association for their helpful comments and suggestions.

Appendix

1. Proof that ikswitch < ikmax,1 < ikmax,2 under Case 1: First, note that ikmax,1 < ikmax,2 if: pRs − (1 = r)C Rc − (1 + r)C < X−C p(X − C) ⇒ pRc − p(1 + r)C < pRs − (1 = r)C ⇒ p(Rs − Rc ) − (1 − p)(1 + r)C > 0 and ikswitch < ikmax,1 if: Rc − (1 + r)C Rc − pRs < (1 − p)(X − C) X−C ⇒ Rc − pRs < (1 − p)(Rc − (1 + r)C) ⇒ p(Rs − Rc ) − (1 − p)C(1 + r) > 0 Case 2 can be analogously derived.

Q.E.D.

5 The Political Economy of Bank Entry Restrictions: A Theory of Unit Banking

117

2. Proof of Lemma 1 (identification of α ∗ ): Consider first the case when C < C ∗ . The bank’s expected gross returns from investing at rate r, and lending at rates ikswitch , and ikmax,1 are: ρr = (1 + r) ρi switch = (1 + ikswitch ) k

ρi max,1 (α) = (α + (1 − α)p)(1 + ikmax,1 ) k

To establish the proposition, we must show that there is α ∗ ≥ 0, such that for α ≥ α ∗ , ρi max,1 (α|α ≥ α ∗ ) > ρi switch , and ρi max,1 (α|α ≥ α ∗ ) > ρr . This is k

k

k

clearly true for α = 1, since ikswitch < ikmax,1 for C < C ∗ , and since r < ikmax,1 (because Rc ≥ (1 + r)X). To establish the existence of α ∗ it is enough to notice that ρ max,1 (α) > 0, because this implies that the bank’s return from ik

lending at rate ikmax,1 decreases as α declines. Consider next the case when C ≥ C ∗ . In this case, the bank’s expected gross returns from investing at rate r, and lending at rates ikswitch , and ikmax,1 are: ρr = (1 + r) ρi switch = 0 k

ρi max,1 = (1 + ikmax,1 ) k

Note that for any α, r < ikmax,1 since Rc ≥ (1 + r)X. Also, for any α, it is trivially true that ρi max,1 > 0. Q.E.D. l

3. Proof of Proposition 1: Proposition 5.1 establishes that for α ≥ α ∗ , ρi max,1 (α|α ≥ α ∗ ) > ρi switch , k

k

and ρi max,1 (α|α ≥ α ∗ ) > ρr . Hence, we need to show that for C < C low , k

ρr > ρi switch , and that for C ≥ C low , ρr ≤ ρi switch ; where C low is defined as: k

k

C

low

∗

≡C −



Rc −X 1+r



(continued)

118

C. W. Calomiris and C. D. Ramírez

If C < C low , then Rc −X C ρi max,2 k holds for any C since pRs ≥ (1 + r)X. Q.E.D.

4. Proof of Proposition 4: 1. C2 < C1 < C ∗ Under this collateral order, Proposition 5.3 establishes that ρi max,1 > 1 ρi max,1 > ρr . Although loans in Region 1 offer the highest expected return 2 to the bank, this return is risky because Type B borrowers will choose to do Type 2 projects. Investing in loans in Region 2 is also risky for the same reason. However, it is possible to find a loan portfolio combination of the two regions that eliminates all risk since outcomes of Type 2 projects are perfectly negatively correlated across regions. Thus, it is possible to find a loan portfolio that, from a risk-return perspective, strictly dominates investment in government bonds. According to Tobin’s (1958) optimal portfolio allocation model, a risk-averse branch bank will select a portfolio that invests some of its assets in Region 1 and some in Region 2. 2. C2 < C ∗ ≤ C1 According to Proposition 5.3, the branch bank’s loan return is highest in Region 1. Since C1 ≥ C ∗ , Type B borrowers in this region will choose to do Type 1 projects only (as Corollary 5.1 establishes). Thus, investing in loans in Region 1 is riskless. Hence, this investment alternative dominates all other risk-return combinations available to the branch bank. 3. C ∗ ≤ C2 < C1 According to Proposition 5.3, the branch bank’s loan return is highest in Region 1. Since C1 ≥ C ∗ , Type B borrowers in this region will choose to do Type 1 projects only (as Corollary 5.1 establishes). Thus, investing in loans in Region 1 is riskless. Hence, this investment alternative dominates all other risk-return combinations available to the branch bank. Q.E.D.

5 The Political Economy of Bank Entry Restrictions: A Theory of Unit Banking

119

References Berger AN, Kashyap AK, Scalise JM, Gertler M, Friedman BM (1995) The transformation of the US banking industry: what a long, strange trip it’s been. Brook Pap Econ Act 1995(2):55–218 Calomiris CW (2000) US bank deregulation in historical perspective. Cambridge University Press, New York Calomiris CW (2010) The political lessons of depression-era banking reform. Oxf Rev Econ Policy 26(3):540–560 Calomiris CW, Haber SH (2014) Fragile by design: the political origins of banking crises and scarce credit. Princeton University Press, New York Calomiris CW, Hubbard RG (1990) Firm heterogeneity, internal finance, and ‘credit rationing’. Econ J 100(399):90–104 Economides N, Hubbard RG, Palia D (1996) The political economy of branching restrictions and deposit insurance: a model of monopolistic competition among small and large banks. J Law Econ 39(2):667–704 Kroszner RS, Strahan PE (1999) What drives deregulation? economics and politics of the relaxation of bank branching restrictions. Q J Econ 114(4):1437–1467 Mengle DL (1990) The case for interstate branch banking. Fed Reserve Bank Richmond Econ Rev (November/December):3–17 Pyle DH (1971) On the theory of financial intermediation. J Finance 26(3):737–747 Pyle DH (1972) Descriptive theories of financial institutions under uncertainty. J Financ Quant Anal 7(5):2009–2029 Rajan RG, Ramcharan R (2011) Land and credit: a study of the political economy of banking in the United States in the early 20th century. J Finance 66(6):1895–1931 Rajan RG, Ramcharan R (2016) Constituencies and legislation: the fight over the McFadden Act of 1927. Manag Sci 62(7):1843–1859 Santomero AM (1984) Modeling the banking firm: a survey. J Money Credit Bank 16(4):576–602 Tobin J (1958) Liquidity preference as behavior towards risk. Rev Econ Stud 25(2):65–86 White EN (1985) Voting for costly regulation: evidence from banking referenda in Illinois, 1924. South Econ J 51(4):1084–1098 World Bank (2018) Global financial development report 2017/2018: bankers without borders. World Bank, Washington

Chapter 6

Partisanship and Electoral Reform: Change in Congressional Cohesion, 1877–1932 Rick K. Wilson

Abstract Explanations for the decline of partisanship in the early part of the twentieth century are at odds. The received wisdom holds that a set of electoral reforms led Congressmen to break their partisan ties, engaging in more familiar modes of personalistic behavior. This view has recently been challenged noting that the bulk of the reforms passed in the Populist and Progressive periods eliminated factional strife within parties and led to increased partisanship. This paper looks at a wide variety of reforms introduced in a 40 year period. While a set of early ballot reforms did result in increased levels of state delegation partisanship, subsequent reforms, combined with the passage of time, undermined partisan strength in the U.S. House of Representatives.

6.1 Introduction The ebb and flow of party strength remains a puzzle for political scientists. This is especially true for partisanship within the US Congress. At the outset of the twentieth century levels of party cohesion were extremely high, but steadily declined in the subsequent two decades. Reaching a low water mark in the 1960s, party cohesion resurged in the 1980s, with measurable levels of party voting approaching that observed in the heyday of “boss rule.” These swings in Congress have been well documented (Brady et al. 1979; Hurley and Wilson 1989; Patterson and Caldeira 1988). While the resurgence in the 1980s has been explained by many, in particular Rohde (1992), our explanations remain incomplete. Key to understanding partisan strength in the US Congress is explaining the precipitous decline of partisanship at the turn of this century. Numerous explanations have been offered, and one which has gained widespread currency points to fundamental changes in electoral rules. In particular, ballot reform rules that

R. K. Wilson () Rice University, Houston, TX, USA e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2018 J. Hall, M. Witcher (eds.), Public Choice Analyses of American Economic History, Studies in Public Choice 37, https://doi.org/10.1007/978-3-319-95819-4_6

121

122

R. K. Wilson

were adopted to overcome local party factionalism in the 1890s and Progressive reforms designed to both expand and constrict electoral participation, were thought to undermine partisan linkages between the electorate and those holding office. In turn, it is argued that the electoral linkage was transformed such that office holders became more closely tied to their constituents. The net effect was a slow erosion of party controls over the behavior of office holders. This paper first details the current explanations for partisan decline around the turn of the twentieth century. While there is a standard story focused on the role of rules, recently this view has been challenged (Reynolds 1995, 1997). In the second section several hypotheses are offered that are derived from these competing approaches. The third section details a set of reforms that dominated the political landscape during the period 1876–1932. Finally, a set of models estimate the impact of these reforms on levels of partisanship among Congressional state delegations throughout this period.

6.2 Explanations for Partisan Decline in the ‘Golden Age of Party’ The dominant explanation for the decline of partisanship at the turn of the twentieth century focuses on changes in the electoral rules pressed by Populists and Progressives.1 By changing the “rules of the game” reformers broke down the ability of party bosses to control elections. Weakened partisan bonds meant that elected officials were now more closely attached to constituents, and constituent interests often diverged from purely partisan politics. Consequently, elected officials were less likely to toe a party line and observable measures of partisan attachment were likely to decline. A number of studies have shown that changes in electoral rules, whether it be the introduction of the Australian ballot (Rusk 1970) or the direct primary (Galderisi and Ginsberg 1986) were correlated with a decline in partisan attachment. Katz and Sala (1996) go so far as to argue that the introduction of the secret ballot transformed the institution of Congress, cementing into place a norm in which re-elected members had a property right claim to committee appointments. It has been proposed that each change in the rules had an impact on eroding the control of party bosses. As Galderisi and Ginsberg (1986) argue, the direct primary was one of the more interesting of the reforms introduced during the Progressive period.

1 This

sweeping statement is not intended to shortchange the pioneering work by Cooper et al. (1977), among others, who noted the rise and fall of partisanship in the U.S. House of Representatives over time. Their approach produces a complex explanation that points to large-scale changes occurring in the general political environment as well as within the House. Given the discipline’s recent focus on electoral reforms, my focus reflects such concerns.

6 Partisanship and Electoral Reform: Change in Congressional Cohesion, 1877–1932

123

The primary can be seen as an antiparty reform on three separate counts. First, by weakening party leaders’ capacity to control nominating processes, primary elections undermine the organizational coherence of established parties. Second, primaries tend to direct the attention of voters and political activists toward the nominating contests of the party most likely to win the general election, and away from the interparty race. . . . Last, and most interesting, primary elections have the effect of inhibiting the formation of new parties (Galderisi and Ginsberg 1986, p. 116).

The argument for the latter point is that contested primaries allow dissidents to oppose within established parties rather than challenge from the outside. Consequently, attention turns away from building a new third party in order to opt to work from within. The principal evidence offered for the impact of changing electoral rules has focused on the Australian ballot. The argument typically offered holds that the secret ballot directly removed the ability of partisan bosses to control votes. Prior to the introduction of the secret ballot, parties (or even splinter groups) printed their own ballots which were handed to voters and which could be dropped into a ballot box. Because only one such voting slip could be dropped into the ballot box, this required that someone watch to prevent more than one vote being cast. Obviously, because ballots were distinctive, this also allowed partisan observers to monitor who was for and against them. The Australian ballot, which required that the state or local government print a standard ballot and required voters to mark their ballot, ensured secrecy when casting the ballot and removed partisan monitoring. If a voter was ‘bought’ it was no longer clear whether he stayed that way when casting his vote. Recently this view has been challenged. Reynolds (1995) points out that ballot reforms were aimed at curbing abuses created by urban political machines. Rather than behaving as tightly controlled national political parties these local machines were fractionated and factional. Reynolds’ claim is that the reformist movements were led by state party leaders who viewed these reforms as an opportunity to retrench the two major parties in the political process. By requiring a centralized, standardized ballot, the major parties could squeeze out third party competition and they could curtail factions within their own party. Indeed, Reynolds provides very strong evidence that the introduction of the Australian ballot and mechanisms for voting via party list led to increased cohesion within the parties—particularly at the national level (Reynolds 1997). Reynolds may well be correct in arguing that the net effect of populist (and later progressive) reforms strengthened national political parties, rather than undermined them. However, the Australian ballot was only one of many reforms pursued by reformers. As Burnham (1970), among many others, notes the reform movements that gained steam in the cities in the 1890s pressed for a wholesale transformation of the American electoral system. At least two general sets of reforms were sought. The first tried to undermine urban political machines by implementing the Australian Ballot, direct primaries and at- large, non-partisan elections. The net effect of these major reforms (and a host of minor experiments) was, in Burnham’s mind, to erode party functions. In turn, as Rusk (1970) and others claim, this weakened the grip of Congressional partisans in both Chambers.

124

R. K. Wilson

The second set of reforms, rather than breaking down control by party elites, turned to curbing electoral abuses by citizens. These reforms included allowing women’s suffrage, thereby enfranchising a group that was expected to have a leavening effect on the political process, the introduction of poll taxes and other barriers to voting, effectively disenfranchising a different group and lessening competition in the South, and finally by requiring personal registration of voters to limit the amount of fraud due to individuals casting multiple ballots. Burnham is quite deliberate in his conclusions about the consequences of these reforms. As he argues, it was not one or the other of these reforms that lead to the decline of partisan attachment at the outset of the twentieth century. Instead, he notes: . . . most if not all of these fundamental changes in the ‘rule of the game’ were in effect devices of political establishment and control, with strongly conservative latent consequences if not overt justifications, and with an overwhelmingly antipartisan bias (Burnham 1970, p. 74).

Burnham is only echoing the arguments offered by many Progressives at the turn of this century. For example, Ernst Meyer in a self-published treatise advocating the primary system as a means for wresting control from partisan bosses, argued: Closely related to the Australian ballot and the primary election reforms safeguarding our nominations and elections, is that which aims to punish the corrupt use of money and other forms of bribery in securing either nominations or elections to office. These reforms are all complementary. Neither one is complete in itself, and all cooperate for a common end,— the establishment of more perfect institutions for the selection of our public service (Meyer 1902, p. 444).

Changing a single rule of the electoral game was not regarded as sufficient. Instead, a galaxy of electoral reforms were necessary in order to return to political order (of course, see Wiebe (1967) for a trenchant analysis as to whose order was being restored). Regardless of the reformist claims, perhaps there is truth in the matter that an array of reforms, which were implemented at different times, had an impact on voter’s partisan attachments and consequently on the attachments of members to party-in-the-Congress. Certainly, Rusk (1978) provide strong evidence for at least the first part of this equation in southern states. They analyze the confluence of a large number of electoral reforms and links their impact to Congressional state delegations. Rather than resting with Reynold’s conclusions about the impact of the Australian ballot on cohesion among partisans in Congress, I press as to whether the change in the ballot, coupled with a myriad of other reforms, had an impact on shifting levels of cohesion in the US Congress.

6 Partisanship and Electoral Reform: Change in Congressional Cohesion, 1877–1932

125

6.3 Hypotheses The principle hypothesis is very straightforward. The cumulative effect of ballot and electoral process reforms was to break down levels of cohesion in state congressional delegations. While several of the ballot reforms in the 1890s may have boosted state party control over the nomination and election of members to the US Congress, over time that control shifted from the party to the individual representatives. With the addition of Progressive reforms that controlled access to the ballot and enhanced the ability of voters to mask their partisan allegiance, representatives increasingly had to tie their fortunes to the voters and were less bound to the dictates of party interests. This meant that where the electoral rules provided new incentives to respond to voters, members of congress did so. All of this occurred at the same time that the tenure for members of congress was increasing. Hypothesis 1 The net effect of ballot and electoral reforms was a decline in state delegation partisan attachment. If the hypothesis is incorrect then the reforms should have no independent effect. Instead, the observed shift in aggregate partisan attachment is due to some other combination of variables. Because these reforms were adopted at different times in different states, the effects of different reforms will be negligible. Hypothesis 2 The short-term effects of ballot reforms was to strengthen state delegation partisan attachment. Hypothesis 2 extends directly from the discussions in Reynolds (1995, 1997). Her arguments and data point out that fundamental changes in the ballot structure meant increased control by state party leaders and eliminated the factionalism that had often been present in parties. However, these reforms, while eliminating factional strife within the party, also planted the seeds for individual candidates to build their own voting coalitions. If similar findings are not uncovered here, then the measures I use can be called into question. Hypothesis 3 The short and long-term effects of Progressive electoral system reforms was to weaken state delegation partisan attachment. Hypothesis 3 distinguishes between ballot reforms and electoral reforms. The latter aimed to eliminate electoral fraud and to ‘cleanse’ the pool of eligible voters. This in turn decreased the ability of partisan machines to control who was placed on the ballot and led to a steady decline in partisan attachment. Fortunately, from the standpoint of a social scientist trying to disentangle these different effects, not all reforms were adopted in all states nor were they adopted at the same time. Consequently, by taking a lengthy time series and tracing when reforms were differentially implemented, it is possible to tease out the impact of these different reforms and address these different hypotheses.

126

R. K. Wilson

6.4 Data The data used for this analysis cover the period 1876 (when members of the 45th Congress were first elected) to 1932 (when members of the 73rd Congress were elected). Two distinct types of data were used. The first constitute measures of partisanship within the U.S. House of Representatives. These are recovered from recorded votes provided by the ICPSR. The second type of data concerns the electoral rules adopted by each state during this period. The key dependent variables are derived from measures of party voting in the US Congress. Two types of measures are used in order to tap levels of delegation cohesion. The first is a standard measure that often has been used. The Party Voting measure is based on the percentage of the time that 50% of one party voted against 50% of the other party.2 As such this measure simultaneously (but imperfectly) captures inter-party conflict as well as the degree to which parties cohere. Usually this measure is reported in the aggregate for the Congress. In this analysis, however, the focus is with state delegations. For each state an average Party Voting score was calculated for each party. The score was derived from the percentage of times individual members voted with their party on party votes. Someone who rarely voted with the party majority when the parties were in conflict would have a low score (a score below 50 was quite rare), while someone always backing the party would have a score approaching 100. Consequently, highly partisan delegations had average scores approaching 100 and weakly partisan delegations had lower scores. Overall rates of party voting are high for both parties from 1876 to 1932, climbing during the 1890s and then declining in the second decade of the twentieth century. A second measure is borrowed from Poole and Rosenthal (1997). Party Nominate is a measure of partisan distance calculated from the two-dimensional individuallevel scores produced by Keith Poole and Howard Rosenthal from all votes cast in a Congress. Their two dimensions are used to construct a new dimension that yields the best fitting measure of partisanship for the two dominant parties.3 Here too state delegation means, derived from member’s scores, are calculated for both parties in each Congress. Because of the way in which the Poole and Rosenthal data are normalized, scores close to zero reflect low partisan attachment—that is such members tend to be as close to one party as the other in their voting behavior. Scores toward a value of 1.0 are extreme Republicans, while scores tending toward −1.0 are extreme Democrats. Figure 6.1 plots the overall averages for both parties across time. These data show a similar (and smoother) trend to what could be observed from overall

2 This measure excludes

third parties from the analysis. The focus here is with major party conflict. probit was estimated for an individual’s party affiliation as a function of Poole and Rosenthal’s DIM1 and DIM2. The resulting coefficients were used to calculate a single, underlying partisan dimension. This is easily done with a simple rotation of the two-dimensional Poole and Rosenthal data by an angle given by the estimated coefficients. See the discussion in Posler and Rhodes (1997).

3A

6 Partisanship and Electoral Reform: Change in Congressional Cohesion, 1877–1932

127

Fig. 6.1 Party distance by congress

rates of party voting. In the 1890s partisan allegiance, as recovered through roll call votes, increased. This is clear from the widening distance between the two parties in this period. By the second decade of the twentieth century the party averages move toward zero (and one another), indicating that partisanship was in decline. The independent variables, by and large, are dummy variables that are turned on when particular voting rules were implemented in the state and turned off otherwise. At the same time, in order to assess the long-term impact of the reform, counter variables are included. Prior to implementing a reform the appropriate counter variable was coded as zero. In the first Congress for which the reform was implemented, the counter was assigned a value of one and then was incremented by a value of one for each subsequent Congress. Consequently, the dummy variable can be thought of as the immediate impact of the reform (an intercept effect) while the counter is the long-term impact of the reform (the slope effect). The first set of variables focused on balloting procedures that sought to lessen the influence of organized political machines. For example, Australian Ballot is a variable indicating whether or not the secret ballot was adopted by a state. Because these data are organized by Congress and state delegation, the variable is coded as a zero in those years prior to the adoption of the secret ballot and coded as one when the ballot system was implemented in time for elections to that Congress.4 4 The dates when states adopted the Australian ballot are widely available. Most recently those dates

were published in Katz and Sala (1996, Table 1). With respect to the other variables coded in this

128

R. K. Wilson

The Australian ballot was implemented under a number of variations by the states. Nonetheless it shared the common feature that balloting would be secret and that the state or local government would print the ballots. Much of the movement for the Australian ballot occurred in the 1890s with most of the states quickly adopting this procedure (see discussions by Fredman (1968) or Rusk (1970). Coupled with the Australian ballot were two additional features of ballot reform. Prior to the introduction of the Australian ballot political parties printed their own slates which served as ballots. This enabled voters to cast their ballot for a straight party ticket. Voting a split ticket usually meant cutting and pasting a new name over a name for a candidate for a particular party. Of course, parties and splinter groups within parties often appropriated one another’s colors and party symbols such that there was a hodgepodge of ballots. With the introduction of the Australian ballot, fundamental questions were raised as to whether candidates parties could be listed under a PARTY COLUMN and whether STRAIGHT TICKET voting was allowed. The reform movements pressing the Australian ballot viewed many local political parties as corrupt machines and aimed to eliminate all references to party on ballots. While the Australian ballot was adopted quickly in most states, it took some time for states to allow candidates to be listed under a party column. Even when candidates could be listed in such a manner, many states refused to allow straight ticket voting. The coding rules were relatively straightforward. The Party Column dummy variable was switched on when the state allowed candidates to be listed under a party name or party emblem. Straight Ticket was triggered when states allowed a voter to check a single box to vote for the entire party slate. A final reform aimed at undermining the grip of party bosses over elections involved the routinization and oversight of primary elections. Many reformers in the late nineteenth century thought party conventions (when they occurred) to be dominated by political machines that were able to ratify their slates without problem (Meyer 1902). A primary system was thought to be a way of letting rank and file partisans have the opportunity to nominate the ‘best’ candidates. Moreover, when overseen by state government, a primary was viewed as overcoming corrupt election practices. Primary is a dummy variable indicating when a state adopted a state-wide primary election process. It should be noted that, as with many of these reforms, requiring the state as a whole to use primaries often lagged behind requiring large municipalities implement the reform. As was often the case at the turn of the century, many state legislatures pointed to urban machine politics as the chief source of corrupted elections, while ignoring the fact that the same reforms could be more broadly applied. A second set of reforms aimed at making it more or less difficult for citizens to vote. The poll tax was directed at blocking black electoral participation in the south. study, the dates for implementing new reforms came from a variety of sources. These include U.S. Government documents like Senate (1897, 1917), Merriam and Overacker (1928), Harris (1929), Ludington (1911). These sources were supplemented and double checked using WestLaw Digest for each of the states. WestLaw produces a useful point of reference for a variety of election laws and is usually thorough in presenting the statutes and case law pertaining to elections.

6 Partisanship and Electoral Reform: Change in Congressional Cohesion, 1877–1932

129

Poll taxes were generally introduced at the turn of the twentieth century and involved pre-paying a fee in order to cast a ballot. These fees were not exorbitant, usually in the range of $1.50–$5.00 per year. However, the fee was often waived for whites and the time period during which the poll tax could be paid was often only a few days out of the year. Poll Tax then is a dummy variable indicating when a state implemented poll tax laws. Registration laws had a different aim. While registration requirements were nominally aimed at eliminating individuals from voting more than once, the net effect was to limit access to the polls by the ill-educated working poor. Registration laws required that voters prove they were eligible prior to an election. Registration is a dummy variable indicating when state-wide voter registration was required. The final reforms considered expanded the franchise. Many states adopted legislation providing women the right to vote well prior to the adoption of the 19th Amendment. This obviously had the effect of expanding the franchise to a large population. The dummy variable Sex is coded to reflect when women were granted the right to vote. Finally, a movement for absentee voting was also pressed in many of the states. While absentee voting had long been extended to soldiers at war, reformers thought this right should be extended to all citizens. Absentee is coded one when a state adopted absentee voting rules extending to all eligible voters. Several additional control variables are included in the analysis. First, all estimations are performed separately for the two major parties. This aids considerable in interpretation. Second, a dummy variable for South is included. This variable takes on a value of one for those states that were part of the Confederacy. As many have shown, in this period the South is different and has an impact on estimations. Third I control the extent of turnover in state delegations. Return calculates the proportion of members from the same party in a state that returned to Congress. This captures the continuity of members to the delegation and enables me to estimate whether changes in state delegation averages are a function of replacement of members or due to changes by the members themselves. Finally, the size of the delegation is controlled for in some of the estimations. It is likely that the delegation averages are sensitive to the number of members in a delegation—in states where there is only a single Democrat or Republican and that member turns over, this can yield considerable volatility in the time series.

6.5 Analysis All of the analysis reported here is based on OLS. Because this data set is a pooled time series with multiple cross sections there are a number of cautionary provisos that need to be kept in mind. Obviously, some set of corrections should be added to the data to correct for serial correlation in the errors, among other problems. These data pose real problems for standard correction techniques. Because the analysis proceeds by estimating the parties separately there are numerous ‘holes’ in the time series in some cross sections. It is simply a fact of life that in some years there were no members of one party elected to Congress in some states.

130

R. K. Wilson

Table 6.1 Ballot reforms only Variable Intercept Australian ballot Australian ballot counter Party column Party column counter Straight ticket Straight ticket counter South R2 N

Party voting Democrats 75.409** (0.70) 3.456** (1.113) −0.170 (0.086) 2.594 (1.389) −0.245* (0.118) 3.836* (1.516) −0.222 (0.147) 12.591** (0.717) 0.29 893

Republicans 87.208** (0.558) 2.525* (1.043) −0.218** (0.083) 2.370 (1.244) −0.206 (0.116) −0.677 (1.270) −0.131 (0.124) – 0.06 948

Poole Rosenthal Democrats Republicans −0.285** 0.417** (0.008) (0.007) −0.030* 0.090** (0.012) (0.013) 0.007** −0.009** (0.001) (0.001) −0.033* 0.070** (0.015) (0.016) 0.002 −0.005** (0.001) (0.002) −0.017 −0.012 (0.016) (0.017) 0.0004 0.001 (0.002) (0.001) −0.166** – (0.008) 0.43 0.19 900 955

Note: standard errors in parentheses **p < −0.01; *p < 0.05

Not correcting these data and resorting to OLS may not be too devastating a problem. As Beck and Katz (1995) point out, many of the standard corrections are unstable when the ratio of cross sections to time series is large. These data have exactly this problem, with upwards of 48 states in the cross section compared with 29 points in the time series for each cross section. They argue that OLS is robust—certainly with respect to the coefficients—although the standard errors will be unreliable. Some corrections (using GLM techniques with Parks corrections) were tried over a subset of these data where the series was complete, with no changes in the direction or relative strength of the parameter estimates. As such, the analysis proceeds with OLS, Keeping in mind concerns over the reliability of the standard errors. Table 6.1 produces estimates for both the Party Voting and Poole/Rosenthal partisanship measures (Party Nominate) regressed on the set of ballot reforms that took place primarily in the 1890s. While not estimated in the same fashion as that presented in Reynolds (1997), many of the patterns are consistent with her results. Taking the Party Voting measure first and focusing the intercept term it is clear that there were high levels of party voting irrespective of ballot reform changes. Consistent with Reynolds the introduction of the Australian ballot increased partisanship for both Democrats and Republicans. So too did the introduction of Party Column voting (although there is a great deal of variability). Straight

6 Partisanship and Electoral Reform: Change in Congressional Cohesion, 1877–1932

131

Ticket voting, while strong for Democrats, had no effect for Republicans. In sum, by looking at just the dummy variables, the impact of ballot reforms on party voting was positive and indicates increased partisanship. These findings are reproduced when using the Poole/Rosenthal measure of partisan distance. However, because of the differences in the signs for these variables, it is important to consider the parties separately. The intercept term for the two parties is correctly signed. For the Democrats the introduction of the Australian ballot moves state delegation averages away from the Republicans (hence the negative coefficient), while for Republican the opposite is true. Likewise, the effect of Party Column lists is equivalently signed, and the effects are roughly the same. For neither party are the effects of Straight Ticket voting very strong. While these effects may seem strong, support Reynold’s findings and are consistent with Hypothesis 2, the counter variables call into question whether these reforms were long lasting. For the Party Voting variable the counters are all negatively signed, pointing out that as time passed from implementing the reform, its effect eroded. For example, a state adopting a Party Column ballot in 1894, and showing an increase in Democratic party voting of almost 2.6% effectively lost that gain in 20 years (10 Congresses). A similar finding holds when looking at the Poole and Rosenthal partisan distance measure. For Democrats the counters are all positively signed, pointing to an erosion of partisan distance and a movement toward zero. For Republicans the signs are opposite but also moving Republican positions toward zero. In sum, the estimates in Table 6.1 indicate that these early ballot reforms cemented stronger partisan positions. This confirms Reynold’s finding and supports Hypothesis 2. At the same time, these findings point out that the impact of these ballot reforms declined over time. Is this because these reforms had little permanence or was it because other reforms had an intervening effect? If the latter, such a finding supports both Hypotheses 1 and 3. To answer this question, all of the reforms and their respective counters are estimated. These estimates are given in Table 6.2. Because Republicans rarely turn up in Southern states and in states imposing a poll tax during this period, those parameters are omitted. Otherwise the estimates are calculated in the same manner as the coefficients in Table 6.1. What is notable about these estimates is the relative stability of the ballot reform coefficients. Their strength and direction are maintained with the addition of these new variables. It is clear that the electoral process reform variables have a consistent, negative impact on partisanship. The only exception is the Poll Tax which contributes positively to party cohesion among the 10 southern states that adopted this change at the turn of the century. In part the Poll Tax variable is picking up aspects of Democratic cohesion in the South.5 Otherwise, these participation variables quickly 5 In

analysis not reported here, estimates were run including only the 13 states that were members of the Confederacy. There, the Poll Tax variable was not significant for Party Voting and had a weaker, but significant effect on the Poole/Rosenthal party distance measure. Given that 10 of 13 Southern states imposed the Poll Tax, there is a high degree of intercorrelation between that variable and the dummy variable representing the South.

132

R. K. Wilson

Table 6.2 All electoral reforms Variable Intercept Australian ballot Australian ballot counter Party column Party column counter Straight ticket Straight ticket counter Women’s vote Women’s vote counter Poll tax Poll tax counter Registration Registration counter Absentee voting Absentee voting counter Primaries Primaries counter Returning members South R2 N

Party voting Democrats 77.296** (0.854) 4.244** (1.103) −0.005 (0.105) 2.056 (1.363) −0.160 (0.126) 3.909* (1.565) −0.208 (0.160) −4.459** (0.854) 0.294 (0.175) 1.153 (1.682) −0.022 (0.147) −1.347 (0.994) −0.118 (0.082) −1.878 (1.242) −0.001 (0.134) −0.306 (1.134) 0.132 (0.144) −1.472 (0.954) 11.858** (0.872) 0.32 892

Note: standard errors in parentheses **p < −0.01; *p < 0.05

Republicans 87.644** (0.687) 1.342 (0.961) 0.145 (0.099) 2.845* (1.159) 0.229* (0.114) −0.879 (1.228) −0.078 (0.121) −3.087** (1.039) 0.185 (0.121) – – 0.272 (0.803) 0.063 (0.066) −2.755** (0.993) 0.240* (0.010) −3.300** (0.953) −0.217 (0.128) 0.396 (0.810) – 0.13 946

Poole Rosenthal Democrats Republicans −0.300** 0.399** (0.009) (0.009) −0.034** 0.090** (0.012) (0.013) 0.002 −0.004** (0.001) (0.001) −0.013 0.057** (0.014) (0.016) −0.000 −0.005** (0.001) (0.002) −0.046** 0.003 (0.016) (0.017) 0.003 −0.000 (0.002) (0.002) 0.057** −0.063** (0.013) (0.014) −0.001 0.004* (0.002) (0.002) −0.060** – (0.018) 0.009** – (0.002) 0.013 −0.007 (0.010) (0.011) −0.0004 0.002* (0.0009) (0.0009) 0.025 −0.032* (0.013) (0.014) −0.002 0.004** (0.001) (0.001) 0.039** −0.077** (0.012) (0.013) 0.0002 −0.0004 (0.002) (0.002) 0.023* 0.043** (0.010) (0.011) −0.183** – (0.009) 0.52 0.27 900 955

6 Partisanship and Electoral Reform: Change in Congressional Cohesion, 1877–1932

133

swamp the effects of ballot reforms—both in terms of the dummy variables and the cumulating time series counters. For Democrats and Republicans alike, opening voting to women, constraining participation through registration requirements, allowing for absentee voting and implementing a system of primaries for the nomination of candidates, decreased levels of Party Voting cohesion among state delegations. Each of the coefficients for the dummy variables are negatively signed and so too are the counter variables. This indicates, with respect to Party Voting, that delegation cohesion levels steadily declined. The table also shows that the impact of these reforms was much greater for Republican delegations. A similar finding holds when looking at the Poole/Rosenthal partisan distance measure. For Democrats the dummy variable coefficients are positively signed, pulling delegations toward zero. For Republicans the coefficients are negative, also pulling state delegations toward zero. In both instances these variables close the gap between the two parties, implying a decrease in partisan attachment and partisan conflict. It may be the case that these estimates are largely driven by fluctuating values due to small sized state party delegations. To explore this possibility the equations from Table 6.2 were re-estimated, including only delegations with three or more members. Table 6.3 provides these estimates. Clearly, the findings from Table 6.2 are quite robust. Some minor changes are worth note. First, the effect of the Australian ballot is dampened for Democratic and Republican delegations alike. Second, the impact of Absentee voting is heightened (and leads to a decrease in partisanship). Even so, the basic findings from above are supported—ballot reforms led to shortterm increases in partisan cohesion. However, consistent with Hypotheses 1 and 3, reforms tied to participation led to declining partisanship. In order to clarify the complex relationships between these reforms which occurred at quite different times, Figs. 6.2 and 6.3 plot the predicted values for Party Voting under a variety of conditions. Figure 6.2 is for Democrats, while Fig. 6.3 is for Republicans. The first line, marked by squares, produces the estimated impact of ballot reforms alone on Party Voting, while including counter variables controlling for the passage of time. The second line, denoted by diamonds, shows the joint effects of ballot reforms and electoral participation reforms, again controlling for variables accounting for time. Finally, the third line, marked by triangles, estimates Party Voting strictly as a function of the passage of time. It is important to keep in mind that the lines are only predicted values, denoted by fixing a variety of reforms at different times. The choices reflected here are fixed by the median time at which the reforms were implemented. The times picked for Ballot reforms are: Australian ballot (1892); Party Column (1896); and Straight Ticket Voting (1900). Reforms linked to participation are given by: Women’s Suffrage (1914); Poll Tax (1900); Registration (1898); Absentee Voting (1914); and Primaries (1906). Figure 6.2 points to levels of Democratic Party voting. In this figure, only nonSouthern states are considered (for southern states the level of party voting would increase by approximately 12 points). The overall time trend is practically zero, although increasing slightly. Ballot reforms alone, as noted above, are correlated with increasing levels of Party Voting. The entire series of reforms, ranging from

134

R. K. Wilson

Table 6.3 OLS estimates of all reforms for large state delegations Variable Intercept Australian ballot Australian ballot counter Party column Party column counter Straight ticket Straight ticket counter Women’s vote Women’s vote counter Poll tax Poll tax counter Registration Registration counter Absentee voting Absentee voting counter Primaries Primaries counter Returning members South R2 N

Party voting Democrats 76.865** (0.904) 1.737 (1.123) −0.067 (0.111) 3.390* (1.520) −0.308* (0.125) 3.246 (1.703) −0.024 (0.158) −5.327** (1.361) 0.687** (0.223) 2.178 (1.445) −0.111 (0.127) −0.291 (0.966) −0.112 (0.079) −2.989* (1.284) 0.178 (0.133) 1.257 (1.108) 0.101 (0.134) 0.637 (1.191) 11.099** (0.808) 0.42 629

Note: standard errors in parentheses **p < −0.01; *p < 0.05

Republicans 88.600** (1.026) 1.957 (1.338) −0.248 (0.142) 1.936 (1.448) −0.129 (0.135) −0.425 (1.519) −0.248 (0.143) −2.481 (1.369) 0.369 (0.199) – – 0.841 (1.025) −0.0003 (0.076) −3.065* (1.203) 0.258* (0.113) −2.830* (1.220) 0.155 (0.158) 1.399 (1.310) – 0.19 560

Poole Rosenthal Democrats Republicans −0.300** 0.421** (0.010) (0.012) −0.011 0.119** (0.012) (0.016) 0.001 −0.011** (0.001) (0.002) −0.042** 0.046** (0.016) (0.018) 0.003 −0.003 (0.001) (0.002) −0.025 −0.020 (0.018) (0.018) 0.001 −0.002 (0.002) (0.002) 0.055** −0.57** (0.015) (0.017) −0.003 0.008** (0.002) (0.002) −0.066** – (0.016) 0.009** – (0.001) 0.005 −0.014 (0.010) (0.012) −0.0004 0.001 (0.0009) (0.001) 0.047** −0.027 (0.014) (0.015) −0.004** 0.004** (0.001) (0.001) 0.016 −0.056** (0.012) (0.015) 0.002 0.004 (0.001) (0.002) 0.006 0.063** (0.013) (0.015) −0.171** – (0.009) 0.63 0.43 629 560

6 Partisanship and Electoral Reform: Change in Congressional Cohesion, 1877–1932

135

Fig. 6.2 Predicted values based on OLS estimates—Democrats

the Australian ballot to Straight Party Voting, leads to an increase in levels of Party Voting. However, as can be see from the dips in Party Voting, the compounded effect of time results in a steady decline. When participatory reforms are put into place there is a similar trend. Peaking around 1898, there is a steady decline in Party Voting. There is an especially pronounced drop around 1914, which is a function of the Women’s Suffrage and Absentee Voting effects. No matter how these reforms are dissected, following a brief upturn in Party Voting, there is a steady, compounded decline following the major ballot reforms. A similar finding holds for Republicans. Figure 6.3 points to a steady decline in Party Voting over time when controlling for nothing else. Ballot reforms lead to a short upturn in Party Voting, with a subsequent decline by 1898. By comparison with Democrats, the short-term increase from the Ballot reforms for Party Voting is smaller, while the overall decrease is the same. In general, these findings support all three hypotheses and not their alternatives. Reynolds’ prediction, contained in Hypothesis 2, is confirmed. The short-term effect of Ballot reform was to strengthen partisan attachment in Congress. At the same time, Hypothesis 3 is borne out. The longer-term effect of a variety of electoral reforms was associated with a decline in partisanship. Finally, the principal hypothesis for this research is supported. The net effect of electoral reforms in the late nineteenth and early twentieth century was strongly associated with a decline in partisanship.

136

R. K. Wilson

Fig. 6.3 Predicted values based on OLS estimates—Republicans

6.6 Conclusion In a sense both the received wisdom and its current challenge are both correct. A decline in partisanship did indeed occur contemporaneous with a large number of reforms to the U.S. electoral system. On the other hand, there was an increase in partisanship prior to the general decay. That increase was most closely associated with one sort of electoral reform: balloting. Because electoral reforms were implemented at different times in different states, econometric techniques ought to be able to test whether specific reforms had any impact. The findings here clearly show that both ballot and participatory reforms had differential effects—even though they were widely dispersed through time. This leads to the tentative conclusion that Reynolds has part of the story correct. Ballot reforms, which targeted local political machines, strengthened the upper hand of state and national party leaders. By centralizing the voting process partisan brokers exacted greater allegiance by party members in Congress. However, reforms that affected who could participate had exactly the opposite effect. Such reforms, while constraining some (notably blacks and the lower classes) opened the process to others (notably women). The net impact of these reforms, coupled with the lingering effects of Ballot reforms, was to undermine the linkage between party leaders and partisans in Congress. In this sense the standard story as told by Burnham (1970)

6 Partisanship and Electoral Reform: Change in Congressional Cohesion, 1877–1932

137

(among others) is probably correct as well. Progressive reforms implemented during the first two decades of the twentieth century hurried a decline in partisanship in Congress. Ballot reforms were insufficient to bring about the changes that were observed, in fact, they initially contributed toward increasing levels of partisanship as state and national leaders reasserted their prominence in the U.S. Congress. The electoral reforms noted here are very blunt instruments. The reader ought to be skeptical that these reforms alone resulted in the changes noted above. As Rohde (1992) notes, in order to get a complete picture of shifting levels of partisanship, it is important to understand the context. “Conditional party government” depends importantly on the issues that come before a legislature. Those issues are pressed by both leadership and by national needs. The same point was raised long ago and extensively discussed by Cooper et al. (1977). Clearly the period leading up to 1896 was nothing less than a battle for the soul of the nation. As all political historians have noted, the 1896 election clearly defined the Republican party and carried it into the twentieth century, while the Democratic party collapsed about itself. Such a national election could not help but have an effect on partisan attachment in Congress. Carefully disentangling what was due to the political agenda presented by partisan leaders and what was due to member’s responses to new electoral incentives remains unanalyzed. However, such analysis is needed before we can speak confidently of the overall impact of reforms in the Age of Reform.

References Beck N, Katz JN (1995) What to do (and not to do) with time-series cross-section data. Am Polit Sci Rev 89(3):634–647 Brady DW, Cooper J, Hurley PA (1979) The decline of party in the US House of Representatives, 1887–1968. Legis Stud Q 4(3):381–407 Burnham WD (1970) Critical elections and the mainspring of American politics. WW Norton and Company, New York Cooper J, Brady DW, Hurley PA (1977) The electoral basis of party voting: patterns and trends in the US House of Representatives, 1887–1969. In: Maisel L, Cooper J (eds) The impact of the electoral process. Sage, Thousand Oak, pp 135–167 Fredman LE (1968) The Australian ballot: the story of an American reform. Michigan State University Press, East Lansing Galderisi PF, Ginsberg B (1986) Primary elections and the evanescence of third party activity in the United States. In: Ginsberg B, Stone A (eds) Do elections matter? ME Sharpe, Armonk Harris JP (1929) Registration of voters in the United States. Brookings Institution, Washington Hurley PA, Wilson RK (1989) Partisan voting patterns in the US Senate, 1877–1986. Legis Stud Q 14(2):225–250 Katz JN, Sala BR (1996) Careerism, committee assignments, and the electoral connection. Am Polit Sci Rev 90(1):21–33 Ludington AC (1911) American ballot laws, 1888–1910. New York State Library, Albany Merriam CE, Overacker L (1928) Primary elections. University of Chicago Press, Chicago Meyer EC (1902) Nominating systems: direct primaries versus conventions in the United States. Author, Madison Patterson SC, Caldeira GA (1988) Party voting in the United States Congress. Br J Polit Sci 18(1):111–131

138

R. K. Wilson

Poole KT, Rosenthal H (1997) Congress: a political-economic history of roll call voting. Oxford University Press, Oxford Posler BD, Rhodes CM (1997) Pre-leadership signaling in the US House. Legis Stud Q 22(3): 351–368 Reynolds LA (1995) Reassessing the impact of progressive era ballot reform. PhD thesis, University of California, San Diego Reynolds LA (1997) The Australian ballot and party voting: re-examining the electoral origins of 20th century party decline. Mimeo, New York Rohde DW (1992) Agenda change and partisan resurgence in the house of representatives. In: Hertzke AD (ed) The atomistic congress, ME Sharpe, Armonk Rusk JG (1970) The effect of the Australian ballot reform on split ticket voting: 1876–1908. Am Polit Sci Rev 64(4):1220–1238 Rusk JG (1978) The effect of southern systems of election laws on voting participation: a reply to V.O. Key. In: Silbey JH (ed) The history of American electoral behavior. Princeton University Press, Princeton Senate USC (1897) Provisions of election laws in different states. In: 54th congress, 2nd session (Document 91). US Government Printing Office, Washington Senate USC (1917) Summary of statutes and constitutional provisions in force in vaious states, November 1916. In: 64th Congress, 2nd Session (Document 659). US Government Printing Office, Washington Wiebe RH (1967) The search for order, 1877–1920. Macmillan, New York

Chapter 7

The Economics and Politics of Unit Banking: Evidence from the McFadden Banking Bill of 1927 Marcus M. Witcher

Abstract This chapter recounts the debate over branch banking in the 1920s. It demonstrates that there was intense and overwhelming opposition to branching in any form. This hostility was both materially and ideologically motivated. Tracing the debate of the McFadden Banking bill, this chapter demonstrates that its passage was due to special interests—in particular the American Bankers’ Association— who ultimately realized that while the bill would allow limited branching of national banks, it would also end branching among the state banks and therefore ensure the continuation of a unit-banking system. While most economists agree that branch banking would have prevented the type of widespread bank failures the United States experienced during the Great Depression, this chapter shows that it was politically impossible to enact such a system in the 1920s.

7.1 Introduction There has been a great amount of work describing how the United States banking system would have benefited from following the Canadian model and allowing branch banking, however, little attention has been paid to the political impediments that existed to branching. This article demonstrates that there were extensive interests that made branching an impossibility. Using the McFadden banking bill as a case study, it demonstrates how special interests, including the unit banking lobby, limited the opportunity for reform during the 1920s. The McFadden banking bill had two main intentions: to make national banks more competitive with state banks and to limit the spread of branch banking. Even though the McFadden bill was an anti-branching piece of legislation, it attempted to make national banks competitive against state banks by allowing national banks

M. M. Witcher () West Virginia University, Morgantown, WV, USA e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2018 J. Hall, M. Witcher (eds.), Public Choice Analyses of American Economic History, Studies in Public Choice 37, https://doi.org/10.1007/978-3-319-95819-4_7

139

140

M. M. Witcher

to branch in states where the laws allowed state banks to do so. This measure resulted in a barrage of special interests who opposed even limited reforms towards branching. At the time of the bill’s debate, many congressmen believed that the passage of the McFadden bill was essential to preserving the national banking system and that without its passage branch banking would create a highly centralized banking system.1 In the wake of the Great Depression, in which undiverse unit banks— those the McFadden bill aimed to protect—failed en masse, the importance of the McFadden bill is magnified. Studying the legislative process of the bill demonstrates that special interests who opposed branch banking were so engrained by the 1920s that any meaningful reform of the American banking system towards a branching system, similar to the Canadian model, was dead upon arrival.2

7.2 Historical Background on the American Banking System, 1783 to 1926 The debate over how the American banking system should be structured has been a highly divisive issue since the founding of the nation. In the early days of the American Republic, Alexander Hamilton, realizing how fragile the economy was after the Revolutionary War, proposed that the United States establish a central bank to strengthen and modernize American finance. Hamilton was countered by another ardent American patriot: Thomas Jefferson. Jefferson argued that the Constitution did not provide for a central bank, and he made a distinction between “necessary” and convenient, labeling the establishment of a national bank as convenient, at best. In his elegant prose, Jefferson took Hamilton’s logic to its apparent conclusion, “it would be still more convenient that there should be a bank, whose bills should have a currency all over the world.” Hamilton countered that the secretary of state’s definition of necessary was too limited and that necessary often meant “needful, requisite, incidental, useful or conductive to.” In Hamilton’s view it was a matter of life and death for the fledgling republic that the word “necessary” be expanded

1A

branch system is a banking system in which a single bank in one locality opens banks other than the parent bank. The second bank is still under the management of the parent bank, but it is located in a separate place. Branch banking allows capital to reach small communities that might not have the capital to open a unit bank. A unitary banking system is a system where each bank is a single operating bank, meaning that the bank has no branches or offices in separate localities from itself. Many unit banks are small in scale and develop in small towns and rural communities to meet the credit needs of those communities. They are locally run and do not need the permission of a parent bank to make loans. 2 I would like to thank Dave Welky and Joe Horton for serving as my advisers during my undergraduate studies. This article was part of my honors thesis, which they advised. I also owe Spencer Ayers a debt of gratitude for accompanying me on what was my first archival research trip and for taking excellent photographs of the materials I needed to make this paper a success.

7 The Economics and Politics of Unit Banking: Evidence from the McFadden. . .

141

beyond Jefferson’s definition. Hamilton, who had the ear of President Washington, won out. Washington’s decision to establish a national bank set the precedent that the federal government had the right to establish corporations—including banks. The debate, however, between those who favored a decentralized state bank system and those who believed that the national government had a role to play in regulating and structuring the financial system was just beginning. Throughout the nineteenth century, the conflict between two competing ideologies would come to bear heavily on the issue of bank structure in America (Wilentz and Earle 2008, pp. 62–64). The First Bank of the United States enabled well-ordered credit expansion to take place and also made great strides to modernize the monetary system.3 In 1811, however, the ideological struggle came to a head and the bank’s charter was not renewed. As a result, the monetary conditions in the U.S. at the time of the War of 1812 were sadly lacking and financing the war proved difficult. The Second Bank of the United States was chartered in 1816, in response to America’s experience during the war (Robertson 1968, pp. 19–20). The Second Bank, like the first, acted as a lender of last resort to banks that were faced with the threat of failure. The bank, under Nicholas Biddle’s leadership, countered the business cycle, established a currency on par with gold, and enabled business to be more prosperous by decreasing the costs involved with transferring money via its branches around the country. In 1832, Henry Clay and the Whigs made the rechartering of the Second Bank of the United States central to their campaign against Andrew Jackson. Jackson despised the national bank, and when he achieved reelection, he took it upon himself to ensure that the central bank cease to operate. The Whigs extended the charter through legislation, and Jackson, keeping his campaign promise, vetoed the measure. In his veto message to Congress, Jackson cited that the bank had many foreign stock holders and claimed that the bank was not interested in the welfare of the American people, but only in its own profits. Furthermore, he claimed, the Second Bank of the United States violated states’ rights (Wilentz and Earle 2008, pp. 324–328). Because of Jackson’s veto, the banking system was set free to develop naturally within the market. Branch banking systems developed in the South and West, whereas a unitary system developed in the East. Different areas of the country developed differently depending on the availability of capital and the opportunity for profit. The branch banking system developed naturally, without restrictive regulations on branching until 1850 when the first anti-branching legislation was passed in Massachusetts, Connecticut, and Rhode Island. Over time, branching became associated with centralization and soon a link developed in the minds of many Americans between branching and monopoly (Robertson 1968, pp. 27–29). About midway through the Civil War, the National Currency Act and the National Bank Act were passed into law. The laws established a uniform currency throughout the country and created the Office of the Comptroller of the Currency who had the authority to charter national banks. Before this act was passed, all banks had been

3 For

a comprehensive analysis of banking in the Early Republic see Hammond (1991).

142

M. M. Witcher

state banks. To curb the power of the states, Congress attempted to coerce state banks to become national banks by levying a 10% tax on all non-national banknotes. This tax had the desired effect and most banks joined the new national system. The national system, however, placed restraints on banks—“imposing minimum capital and reserve requirements, restricting real estate loans, and prohibiting branching” (White 1982, p. 34). When the National Bank Act became law, branching was commonplace. The early comptrollers, however, interpreted the act’s singular pronoun language in section six, which states “the place,” as an indication that the act was intended to prohibit branching. The first comptroller did not interpret this language as a prohibition on branch banking, but his successor, Freeman Clarke, did. In accepting a bank in Maryland into the national system, the comptroller forced the bank to give up its branch. Clarke’s August 18, 1865 letter to the bank explains his decision: “The sixth section [of the law] requires that [the organizers] shall specify in their organization certificate the particular place (not places) at which their operations of discount and deposit shall be carried out.” As a result of Clarke’s interpretation, national banks were prohibited from branching for decades to come (Robertson 1968, pp. 82–83; White 1983, p. 14). Whereas the banking system before the National Banking Act had been able to develop freely to meet the needs of the country, the newly imposed restrictions prevented the supply of banks from filling the demand. Rural areas in the everexpanding West desired credit institutions, but the restrictions on branching, coupled with minimal size requirements for national banks, led to the reemergence of the state banking system. The state bank system, which had bottomed out with 277 members in 1873, posted gains almost every year for the rest of the century. By 1900 it had a total of 5007 members (Robertson 1968, p. 67). The restrictions placed on national banks ensured that they would not be able to meet the demand for banking facilities. Furthermore, the state legislatures that had grown hostile to branching—insisting that credit allocation in a community be made by members of that community—ensured that branching would not be the means by which the demand was filled. Instead, new capital had to be raised and new unit banks established to meet the demand. It was soon discovered that even the capital requirements set by the states were too high to ensure that the demands of the public were filled. In response, from 1890 to 1907 there was a widespread movement in banking to reduce capital requirements on banks pursuing charters. As a result, thousands of unit banks were established. The surge in the membership in the state system led the comptroller in 1896 to amend the National Bank Act to lower capital requirements for national banks and to allow national banks to open offices in towns with fewer than 1000 residents. The actions of the comptroller and the state legislatures resulted in the creation of thousands of unit banks that would form the unit bank lobby that would ensure that branching remained illegal (White 1983, p. 14, p. 23; Robertson 1968, pp. 66–68, p. 83). Anti-branching sentiment grew throughout the 1890s and early twentieth century. Branching became fundamentally evil and un-American; in the mind of many bankers and legislators it was a force that must be crushed. By 1910, it appeared as if

7 The Economics and Politics of Unit Banking: Evidence from the McFadden. . .

143

the unit bank lobby had been successful. At the Alabama Bankers’ Association convention one of the speakers declared, “the days of the branch bank are numbered.” At a New York bankers’ meeting, E.B. Vreeland, another banker, concurred, “No one will ever live to see the day when the branch banking system which prevails in Canada and in Germany and in England and in France will be tolerated by the people of the United States.” Vreeland went on to explain why branching must be opposed: “The economies of the branch banking system are such that no other system can live beside it. It is just as sure as the sun will rise to-morrow that the branch banking system, if taken up in the United States, would in the end drive out of existence all the banks in every city and town in the country outside of the great financial centers. This is the experience of the world.” Vreeland’s statements demonstrate a fear of monopoly and centralization, a fear that stretches back to Jefferson’s condemnation of the First Bank of the United States. Mr. Vreeland’s statements were also an admission: an acknowledgement that the unit banking system was wasteful, and that branch banking would have been much more efficient (Eckardt 1910, pp. 626–627). In 1910, a small group of academics began to challenge the claim that branching would lead to monopoly in the banking system. H.M.P. Eckardt (1910) argued that branching would not cause monopoly, but rather would provide for excellent facilities, as well as better quality management undoubtedly resulting in better loaning practices and increased solvency. The argument for branching, while still uncommon, was beginning to become more acceptable. In 1916, the Federal Reserve Board encouraged Congress to pass legislation that would allow branching of national banks in cities with more than 100,000 inhabitants. The board’s justification for requesting branching privileges concentrated on the fact that some states that allowed branching were putting national banks—which were restricted from branching—at a competitive disadvantage. The board’s support did not derive from the superiority of a branch system. Regardless of the justification for the implementation of branching, at least the issue was being discussed.4 As World War I ended, the authorities on banking began to openly discuss the need for establishing a limited branch banking system in the United States. In 1919, the Federal Reserve Board through its annual report made its position clear, “[the board] recommends once more that national banks be permitted to establish banks in the cities in which they are located.” Legislation that would have granted national banks the privilege described above was passed in the Senate in 1916, but defeated in the House. In 1922, the Joint Commission of Agricultural Inquiry prepared a report on the poor credit situation in rural communities. The report declared that “a system of limited branch banking might furnish a possible solution to the problem.” The board responded to the report by issuing a statement which pointed out that such a system already existed in some parts of the country and that these regions had 4 RG 82, Records of the Federal Reserve System, Board of Governors, Central Subjects File, 1913–

1954, Box 151. File: 111.1-T-5 Branch, Chain & Group Banking Legislation H.R. 6855, 68th Congress located at National Archive II. Specific information found in a report titled “Recommendations, Regulations and Administrative Policies re Branch, Chain and Group Banking.” Issued by Walter Wyatt on March 1, 1930 and addressed to the Federal Reserve Board.

144

M. M. Witcher

seen interest rates decline and security for rural investments increase. The board continued to say that these branch banks “appear to be serving their communities well.” 5 In 1922, two bills addressing the issue of branch banking were presented to Congress. The McFadden Banking bill, believed by many to be the more responsive to real economic problems, became Congress’ solution to the disparities between national and state banks. The bill also attempted to provide a legislative solution on the issue of branch banking. The bill died in the sixty-seventh and sixty-eighth Congresses. In 1926, Congressman McFadden brought the bill to the floor of the House of Representatives once more for debate (Chapman and Westerfield 1942, p. 102).

7.3 The House Debate As Congressman Louis T. McFadden walked through the Capitol on Tuesday, January 12, 1926 he must have known it would be very difficult to pass his legislation. McFadden had grown up in a small town near the mountain region of northeastern Pennsylvania, graduated from Warner’s Commercial College in 1892 and was hired shortly after by the First National Bank of Canton, Pennsylvania. In 1899, McFadden was elected cashier of the bank and from 1906 to 1907 he served as treasurer of the Pennsylvania Bankers’ Association. In 1914, McFadden was elected to the U.S. House of Representatives and in 1916 became President of First National—a position he would hold until 1925 while serving as a U.S. Representative from Pennsylvania. By all measurements Representative McFadden had been a great success.6 His time in Congress, however, had proved frustrating. His principle piece of legislation, the McFadden banking bill had been introduced in the previous Congress but had died in committee in the Senate when the final session expired. McFadden, determined to give the bill enough time in the Sixty-Ninth Congress, moved quickly to get the bill, H.R. 2, on the agenda in the House. By early afternoon on January 12, Congressman McFadden had referred the bill to the entire House.7 Despite Congressman McFadden’s effort to fast track H.R. 2, over 2 weeks passed before the bill came up for debate in the House. On Tuesday, January 26, McFadden was recognized by the speaker and he began summarizing the bill. The legislation would strive to reduce the competitive edge that state banks 5 RG 82, Records of the Federal Reserve System, Board of Governors, Central Subjects File, 1913–1954, Box 151. File: 111.1-T-5 Branch, Chain & Group Banking Legislation H.R. 6855, 68th Congress located at National Archive II. Specific information found in a report titled “Recommendations, Regulations and Administrative Policies re Branch, Chain and Group Banking.” Issued by Walter Wyatt on March 1, 1930 and addressed to the Federal Reserve Board. 6 Biographic Directory of the United States Congress, Louis T. McFadden. 7 U.S. Congressional Record, vol. 67, pt. 1–3, 69th Congress, 1925.

7 The Economics and Politics of Unit Banking: Evidence from the McFadden. . .

145

had over national institutions by allowing national banks to branch within their parent city and by reducing the restrictions on national banks’ loaning practices. Altogether, the measures in the bill aimed to reverse the declining membership in the national banking system. While the bill permitted national banks limited branching privileges, it prohibited state banks from bringing any branches into the Federal Reserve System. Overall, Congressman McFadden and his bill were overwhelmingly opposed to branching and only permitted national banks to do so to stop numerous national banks from leaving the system. Following his summary of the bill, McFadden listed the numerous supporters of the bill. Those in support included the American Bankers’ Association, the Comptroller of the Currency, the Association of Federal Reserve Governors, the Association of Federal Reserve Agents, the Federal Reserve Board itself, and many state bank associations. The congressman then read a letter from the Federal Reserve Board concerning the bill: “The urgent importance of liberalizing the law so as to enable national banks to compete more effectively with state institutions has long been recognized by the Board.”8 After expounding on his bill, McFadden turned his attention to the opposition. He highlighted section nine of the bill as the major section of debate. This part of the bill, McFadden explained, was the teeth of the bill against branch banking. Out of a fear that national banks would pressure state legislatures into legalizing branch banking in every state, Representative Hull of Illinois had introduced section nine. The section established that the McFadden bill would grant national banks the right to branch only in states that at the time of the passage of the legislation already permitted branching. In other words, if a state legalized branching after the McFadden bill was passed the national banks in that state would not be allowed to branch. Section nine, as well as several other phrases scattered throughout the bill, became known as the Hull Amendments. McFadden demonstrated that he and the banking community favored the Hull Amendments. McFadden then placed the opponents of the amendments into two camps. On the one hand, there were the actual proponents of branch banking and on the other, were the states’ rights advocates who claimed that the Hull Amendments took the power to draw legislation away from states. McFadden dismissed the real supporters of branch banking as being “so few in number as not to cause great concern.” Those who fought the Hull Amendments on the grounds of states’ rights, however, were a force to be reckoned with—at least in the Senate.9 McFadden discredited the pro-branching faction by declaring branch banking as being monopolistic in nature and damaging to the economy. Turning to what he saw as the real threat to the Hull Amendments—those who invoked states’ rights as a reason for opposition—McFadden declared that the Hull Amendment made “no infringement upon the rights of any state legislature, or interferes with any State banking system.” McFadden assured members of Congress that the bill was an anti-

8 U.S. 9 U.S.

Congressional Record, vol. 67, pt. 1–3, 69th Congress, 2827–2829. Congressional Record, vol. 67, pt. 1–3, 1925–1926, 69th Congress, 2831.

146

M. M. Witcher

branch piece of legislation and that he was pleased that communities were able to make credit decisions in their own community. McFadden concluded by praising America’s banking system as “a system which has shown a vitality and adaptability unparalleled in the financial history of any other country.”10 It is interesting that Congressman McFadden claimed that the U.S. bank system should be seen by the rest of the world with envy considering that from 1921 to 1925 almost 3000 U.S. banks had failed. With an average of 582 banks failing a year—most of them the unitary banks in small communities his legislation aimed to protect—some may find it hard to understand how the congressman was so inclined to believe that the U.S. banks system was elite. To put these nominal numbers into perspective, the average percentage of banks failing per year from 1900 to 1920 was 0.33% of all banks. The average number of bank failures from 1921 to 1925 was 1.64%, or nearly five times as high. Bank failures in the U.S. had increased 500% while Congressman McFadden was drafting his legislation. Most of these failures were occurring in agricultural areas and in states where unit bank lobbies had successfully prohibited branch banking (Spahr 1932, p. 237). Applause broke out on the floor of the House of Representatives as Congressman McFadden concluded that his bill would fix the archaic restrictions of the National Bank Act and offer long overdue relief to the national banks. After McFadden relinquished the floor, several members voiced concern over state banks being forced to terminate their branches upon becoming members of the Federal Reserve and it was concluded by the House that those banks should quickly become members of the Federal Reserve System before the McFadden bill passed. The Congressmen’s concern demonstrated the harsh anti-branching sentiment in the bill. The bill, as proposed by the House in 1926, would have practically prevented banks with branches from becoming members of the Federal Reserve System, however, it would have forced other banks without branches to be sure that they had no intention of ever joining the Federal Reserve System before they established a branch.11 Despite McFadden’s repeated assurances that his bill was opposed to branching, some congressmen still dismissed the legislation on principle. These representatives, while a minority, loudly voiced their opposition to allowing national banks to branch even within their parent locality. Representative Nelson of Wisconsin best represented the dissenters’ position: “No man needs to be a prophet to foretell that this bill will inevitably lead, with increasing speed, straight to branch banking, State and National. Let us not deceive ourselves.” Nelson spoke of the St. Louis Supreme Court case and reminded the House that it was the Wall Street banks that aided in the First National Bank of St. Louis’ defense.12 According to Nelson, Daniel 10 U.S.

Congressional Record, vol. 67, pt. 1–3, 1925, 69th Congress, 2832. Congressional Record, vol. 67, pt. 1–3, 1925–1926, 69th Congress, 2832–2833. 12 In 1922, the First National Bank of St. Louis, Missouri, opened a branch several blocks from its parent bank. The bank did not consult the Comptroller of the Currency. The state of Missouri, which did not allow branch banking commenced with a suit against the bank. The Missouri Supreme Court ruled in favor of the state and declared that the prohibition on branching, within Missouri, applied to both state and national banks. First National appealed and the case proceeded 11 U.S.

7 The Economics and Politics of Unit Banking: Evidence from the McFadden. . .

147

R. Crissinger,13 the Federal Reserve Board and the money trust were behind the McFadden bill and that the goal was a national branch system.14 Representative Nelson picked up steam and, in all likelihood, volume as he went. At one point he denounced the McFadden bill by using allusions from sop’s Fables (claiming the McFadden bill to be a wolf in sheep’s clothing) and The Iliad: “In like manner this bill is a wooden horse filled with branch banking, which we are asked to take into the Federal bank system, and when we do it will be the end of independent banking.” Nelson, showing his verbosity as well as his learnedness, drew upon a biblical theme next: Holy Writ tells us that Satan, when he is about to do mankind great harm, always appears as an angel of light. The crime of history was accomplished with the kiss of Judas. Therefore, it is not safe to accept what men say; we must carefully scrutinize what they do. Men’s deeds speak louder than their words! . . . When did a compromise ever stop anything? Never has evil been stopped by a compromise. Hell itself is paved with compromise Once we surrender the principle what are we going to stand on?

By this point Representative Nelson had gotten many of the other representatives worked up, and as he concluded, applause broke out from the floor: “I conclude, Mr. Chairman, as I began. This is one of the most objectionable bills before Congress. This measure means, if it passes, inevitably the end of our independent banking system and in its stead the Canadian, the British and the Continental System of branch banking in this country.”15

to the United States Supreme Court. The Supreme Court upheld the decision of the lower court. The Attorney’s General for Illinois, Connecticut, North Dakota, Washington, Wisconsin, Iowa, Arkansas, Minnesota, Indiana, and Kansas took the position with Missouri (all the before mentioned states either prohibited branching or did not have a law that explicitly stated legality). What gave Rep. Nelson the license to say what he did on the floor was that three New York Banks: The National City, Chemical National, and National Bank of Commerce, along with the Solicitor General of the United States, all challenged the courts authority to pass judgments on national banks (these three banks were part of the so called “money trust” so feared and disdained by progressive politicians throughout the early Twentieth Century). Chapman and Westerfield describe the Courts decision as such: “The decision of the Supreme Court (January 28, 1924) upheld the power of a state to enforce state legislation prohibiting national banks from establishing branches, unless the authority of national banks to establish branches is expressly conferred by Congress. It also decided that the state was not trying to interpret and enforce a Federal Statute, but merely to vindicate and enforce its own law, which prohibited branch banking; and since this law did not frustrate the purpose for which the bank was created or interfere with the discharge of its duties to the government or impair its efficiency as a federal agency, the state statute as applied to national banks was valid, obligatory, and enforceable” (Chapman and Westerfield 1942, pp. 98–99). 13 Crissinger served as Comptroller of the Currency from 1921 to 1923. While he was in power, Crissinger encouraged the spread of branch banking and as such drew the ire of the unit banking lobby that had become almost invincible in Congress. He also was the Governor of the Federal Reserve Board from 1923 until 1927 (Katz 1992, pp. 60–65). 14 U. S. Congressional Record, vol. 67, pt. 1–3, 1925–26, 69th Congress, 2840–41. 15 U. S. Congressional Record, vol. 67, pt. 1–3, 1925–26, 69th Congress, 2842, 2845.

148

M. M. Witcher

Representative Stevenson, a Democrat from South Carolina and a member of the Banking and Currency Committee, spoke after Congressman Nelson and made fun of his extreme positions. Stevenson labeled those who would see a nationwide branch system as part of the fringe. He then discredited the impassioned position of Representative Nelson: “[A] small but very respectable element who are so averse to branch banking or branch offices or branch anything that they would protest against a national bank teller carrying change across the street to a one-legged widow selling peanuts to a paralyzed cripple of the corner.” The chamber echoed with laughter.16 After reemphasizing that the purpose of the bill was to put national banks on par with state banks, Stevenson relinquished the floor to Congressman Steagall from Alabama. In response to Stevenson’s pragmatic reasoning for allowing branching, Representative Steagall framed him as clouding the issue and hiding behind the equity between the two systems when really the McFadden bill was a pro-branching measure. Steagall condemned the proponents of the McFadden bill stating that, “when opponents meet you in the open, you know where you are and how to fight, but when false theories are clothed in misleading terms, when evils are hidden in a wooden horse, those who fight them are always at a disadvantage.” Steagall continued by correctly claiming that the purpose of the bill was to allow national banks to branch, however, the representative fumed that this bill would give branching a “wedge” that, in the future would lead to increased branching. The debate on the floor ended at four o’clock in the afternoon with Representative Steagall insisting that “branch banking is un-American.”17 On Thursday, February 4, at noon, the House met to vote on the McFadden bill. Representative Steagall offered several amendments in hope of somehow defeating the bill and when his amendments failed, he made a last-ditch effort to have the bill recommitted to the House Committee on Banking and Currency. His efforts were defeated 62 to 14. Steagall continued to press the issue citing no quorum. The congressmen were rounded up and a second vote was held on Steagall’s proposal to send the bill back to conference. The representative from Alabama was once again defeated by a vote of 291 to 91, with 49 not voting.18 McFadden stood after the vote and triumphantly said, “Mr. Speaker, I ask for the yeas and nays.” A few minutes passed but it was clear from the vote on Steagall’s proposal that the bill would pass. The final vote was 293 yeas, 90 nays and 48 not voting. The bill was passed in the House, but with the Hull Amendments still in place. The real test would come in the Senate.19

16 U.

S. Congressional Record, vol. 67, pt. 1–3, 1925–26, 69th Congress, 2847. S. Congressional Record, vol. 67, pt. 1–3, 1925–26, 69th Congress, 2850. 18 U. S. Congressional Record, vol. 67, pt. 1–3, 1925–26, 69th Congress, 3302. 19 U. S. Congressional Record, vol. 67, pt. 1–3, 1925–26, 69th Congress, 3302–03. 17 U.

7 The Economics and Politics of Unit Banking: Evidence from the McFadden. . .

149

7.4 The Senate Debate On April 27, Senator Pepper asked that the bank bill be taken up for consideration. After a short time in which other bills were discussed, Pepper was able to open the floor for discussion of the McFadden bill. At the opening of the debate, there was much confusion over the purpose of the legislation. Senators continued to ask general and uninformed questions about the legislation and it is unlikely that many of them had read the bill before the debate began. Finally, after 45 min of debate, Pepper outlined the major differences between the House and the Senate measures. The House bill, with the Hull Amendments, would restrict national branch banking to states that at the time of passage of the legislation allowed branching and was not responsive to future state legislation. Pepper explained what was wrong with the Hull Amendment in the House bill: “It has seemed to the committee that to make any such iron-clad provision, crystalizing or freezing the situation as of the date of the passage of the bill was an unreasonable restriction by Congress on the liberty and action of the State.” Pepper went on to explain that the Hull Amendments were illogical and cited that the purpose of the bill was to make national banks more competitive with state banks. If that was to be the goal, he claimed that national policy should be responsive to state policy changes.20 Senator Pepper yielded to Senator McLean, a Republican Senator from Connecticut, who questioned what the results would be if the Hull Amendments were left in the legislation: If the so-called Hull amendments were adopted would it not be probable, if not inevitable, that many of the national banks not in the system would be compelled to retire and take advantage of State privileges, and is it not our purpose and should it not be our purpose to so sustain the Federal reserve system and the national banking system that we may invite State banks into the system and keep the national banks from retrieving therefrom?

Senator Pepper, who had undoubtedly set up this exchange with his ally Senator McLean, who also served on the Senate Committee on Banking and Currency, replied, “Undoubtedly it would be most unfortunate in the judgment of the committee if we were, by making an arbitrary and it seems to me an unreasonable restriction, to defeat the primary purpose of the legislation.”21 As the debate picked up, a senator suggested that the Hull Amendments might have to be accepted to pass the bill and that most of the letters he had received were insisting that “they would prefer to have half a loaf [of bread] than no loaf.” At this point Carter Glass, a Democratic senator from Virginia, and a member of the Senate Committee on Banking and Currency responded, “In other words, the contention has been that the Senate ought not to do the right thing for fear that the

20 U.

S. Congressional Record, vol. 67, pt. 1–3, 1925–26, 69th Congress, 3341, 6215, 8290–8293, 8294–8297. 21 U. S. Congressional Record, vol. 67, pt. 6–8, 1925–26, 69th Congress, 8298.

150

M. M. Witcher

House would not concur?” Debate ended shortly after Senator Glass’ remarks and was scheduled to resume the next day.22 Glass was filled with indignation by the Hull Amendments and by the letters that he received from countless bankers insisting that the amendments be passed into law. Most of the letters received by congressmen were sent at the request of the American Bankers’ Association (ABA), which in a conference in Chicago had formally announced its support for the Hull Amendments. A letter received by Carter Glass on April 14, from the president of the Mercantile Trust Company is a good representation of the type of material that was being sent out by members of the ABA. The ABA urged the banker to write his Senators and Congressmen and inform them of his bank’s support for the bill as proposed by the House. The letter even included a template where the bank’s president would simply need to fill in his name and the name of the bank that he served. It is probable that each Senator and Congressman was bombarded, just as Senator Glass was, with letters of this sort.23 Glass and George J. Seay, the governor of the Federal Reserve Bank of Richmond, Virginia, corresponded on a regular basis. On April 20, 1926, Glass wrote Seay to express his frustration with the American Bankers’ Association: There is no little indignation in the Senate over the effrontery of the general counsel of the American Bankers’ Association and his legislative committee in practically denying the right of the Senate to make any amendment to the bill. Their willingness to even sacrifice indeterminate charters for the Federal Reserve banks in order to get their utterly selfish and atrocious Hull Amendments through is an exhibition of which they should be rather ashamed. They’ll find that they have greatly miscalculated the situation at Washington.24

Senator Glass had been in Washington since 1901, when he was appointed to the House of Representatives. Glass went on to win reelection to the House for the next eight congresses, after which he was appointed as the Secretary of the Treasury in 1918 under President Wilson. With the election of President Harding, Carter Glass found himself out of a job—but not for long. Glass was elected to the Senate in early 1920 and was instantly recognized as an expert in financial matters.25 The next day, April 28th, debate on the McFadden bill resumed. Senator King of Utah opened the second day of debate by citing that many agricultural banks had failed in recent years due to the low capital requirements imposed by state regulators. He hoped that the McFadden bill would increase capital requirements for national banks. In response to King’s proposal, Senator McLean argued that according to a Federal Reserve Board report it was state banks with low capitalization that had failed. McLean pointed out that Congress could not impose increased capital requirements for state banks, but that the Senate Committee on Banking

22 U.

S. Congressional Record, vol. 67, pt. 6–8, 1925–26, 69th Congress, 8298. dated April 14, 1926 from the President of the Mercantile Trust Company to Senator Glass. Small Special Collection Library, Papers of Carter Glass, Box 248. 24 Letter dated April 20, 1926 from Senator Glass to George J. Seay. Small Special Collection Library, Papers of Carter Glass, Box 7. 25 Biographic Directory of the United States Congress, Carter Glass. 23 Letter

7 The Economics and Politics of Unit Banking: Evidence from the McFadden. . .

151

and Currency had determined that it was preferable “to have a small national bank, carefully regulated by the Comptroller of the Currency, and examined two or three times a year, than to leave the whole field to the state banks.” Senator King responded: “I had hoped, if the Federal Government fixed a higher limit, the wisdom of that having been demonstrated, that it might be an example the States would be inclined to follow.” King was correct that low capital requirements allowed small, undiverse banks to populate the country and that these banks were very likely to fail, but his idealistic notion that the states would follow the example of the national bank system was unrealistic. Ever since the establishment of the dual banking system, the state bank regulators had been deregulating beyond what the federal government had done as to attract more banks to the state system. If the Senate had adopted King’s proposal it is a safe bet that even more banks would have chosen state charters over national charters.26 Debate concluded not long after Senator King’s statements and in the afternoon of May 12, 1926, the Senate went forward with the amendment process. The Hull Amendments, which Senator Glass detested, and the Senate Committee on Banking and Currency had stripped from the bill, still had some support in the Senate. Senator Lenroot from Wisconsin expressed his dedication to the House bill lamenting, “I am very sorry, personally, that the plan of the House bill was not followed, known as the Hull amendment. I am very much opposed to branch banking as a system.” Mr. Lenroot labeled branch banking as evil and contended that enabling national banks to branch in already established branching states and not having the Hull Amendments would lead to national banks lobbying state legislatures to allow branch banking.27 Senator Glass rose to defend the modifications that his committee had made to the bill. Glass claimed that the bill as it passed the House “will not only not induce a single State bank to become a member of the Federal Reserve system, but if concurred in by the Senate it will, in my considered judgment, drive out of the Federal Reserve system 75% of the State banks now members of the system.” Glass was granted the floor and he went on the offensive against the Hull Amendments: I venture to say that not since the foundation of this Republic was there ever before attempted, in any Federal legislation proposed, such a deliberate invasion of the rights of the States of this Union as is proposed in the so-called Hull amendments as passed by the House. That House bill proposes to do for banking what Mr. Lincoln said could not be done for slavery, to establish a nation one-half branch banking and one-half unit banking I am not now speaking for branch banking, whether corporate or state-wide. I am advocating the right of the States to establish their own banking systems, free from coercion by Federal Statute. I am unalterably opposed to undertaking, by Federal law, to coerce 26 States not now permitting branch banking, into forever precluding the system, no matter how much they might desire to adopt it. I favor letting these 26 States determine for themselves whether or not hereafter they will have branch banking.28

26 U.S.

Congressional Record, vol. 67, pt. 6–8, 1925–26, 69th Congress, 8352. Congressional Record, vol. 67, pt. 6–8, 1925–26, 69th Congress, 9295,9296. 28 U.S. Congressional Record, vol. 67, pt. 6–8, 1925–26, 69th Congress, 9296, 9297. 27 U.S.

152

M. M. Witcher

The supporters of the Hull Amendments blasted any sort of branching as monopolistic. Carter Glass, however, unflinching in his denunciation of the amendments proclaimed that the Hull Amendments, “originated with a little coterie of scheming bankers which, pretending to want to safeguard unit banking against monopoly, is, indeed, appealing to Congress to give these unit bankers a monopoly of the credits of their respective communities.”29 Glass went on to speak about the public hearings that were held on the Hull Amendments by the Senate. At these hearings, almost every banker, including representatives of the American Bankers’ Association, declared the Hull Amendments not to be fair or in line with good policy. The opinion among those bankers was, however, that the anti-branch lobby would only permit passage of McFadden with the Hull Amendments. The following is an excerpt from the Senate hearings on the Hull Amendments: SENATOR GLASS: “Were these bankers charged with the function of national legislation?” MR. HINSCH, A BANKER FROM CINCINNATI: “No; but they exercise sufficient influence over Members of the House to defeat, in my opinion, any bill that deviates materially from Section 9 [the Hull amendments] of the McFadden bill.” SENATOR GLASS: “That remains to be seen. It may be that there are enough Senators who think for themselves to deny the right of those bankers to control legislation.” MR. HINSCH: “Unfortunately most legislation is based upon compromise, and so it is, as we see it, with the McFadden bill.” SENATOR GLASS: “You want the compromise made for us by the American Bankers’ Association instead of permitting the Senate to make the compromise for itself. Is that the idea?” MR. HINSCH: “If you can make it and get it through so that you will relieve the national banks so they can compete on more equal terms with state banks, we will be very glad to present you with a Carnegie medal.”30

Carter Glass continued to denounce the Hull Amendments and their unequivocal hatred of branch banking. Senator Glass tried to dismiss some of the claims that branch banking was ‘evil.’ In the hearings on banking, Glass reported to the Senate that the bankers from California “presented fact after fact and an abundance of statistics to prove that branch banking in California had been a blessing to commerce and to industry. They pointed out that since the establishment of branch banking in the state not a depositor in a branch bank had lost a dollar in twelve years.”31 With that, Senator Glass yielded to Senator Ashurst from Arizona, who asked for clarification from Glass on the matter at hand: “I desire to be certain as to this particular subject before I commit myself, and I wish information. I sit, as to fiscal affairs, at the feet of the Senator from Virginia [Glass], even as Saul set at the feet of Gamaliel.” Senator Glass modestly responded by saying, “Well, a good many people who do that ought not to,” and then proceeded to further clarify the bill as reported to the Senate.32

29 U.S.

Congressional Record, vol. 67, pt. 6–8, 1925–26, 69th Congress, 9297. Congressional Record, vol. 67, pt. 6–8, 1925–26, 69th Congress, 9297. 31 U.S. Congressional Record, vol. 67, pt. 6–8, 1925–26, 69th Congress, 9298. 32 U.S. Congressional Record, vol. 67, pt. 6–8, 1925–26, 69th Congress, 9299. 30 U.S.

7 The Economics and Politics of Unit Banking: Evidence from the McFadden. . .

153

After Senator Glass relinquished the floor, Senator Lenroot and Senator Shipstead, a member of the Farmer Laborite party from Minnesota, again brought up the possibility of monopoly arising out of a branching system. These senators determined that branching would concentrate credit in the hands of a few and small banks would be forced into failure as a result. Senator Glass refused to allow Lenroot and Shipstead to shift the debate and interrupted: “Without seeming to transform myself into an advocate of branch banking, I may say to the Senator from Minnesota [Shipstead] that of the 526 banks that failed in the Minneapolis Federal reserve district within the last few years, 90% of them were small banks, and if some of the strong banks had been conducting branches where these small banks had failed were located, there would not have been one-tenth that many failures.”33 Despite Glass’ hour long speech denouncing the Hull Amendments, Senator Lenroot—who had the floor at this point—continued to rail against branch banking and the version of the bill reported to the Senate. Lenroot claimed that the passage of the Senate version would lead to pressure being put on state legislatures to legalize branching. The House bill, he insisted, would do the opposite; it would put pressure on states that presently permitted branching to abolish it. After attacking the Senate proposal, Lenroot again attacked branching and spoke of the Canadian system: “I think it is exactly analogous to the difference between local strike and a general strike. A local strike would affect only the community. But if we should contemplate in the future a banking system, such as they have in Canada, of four or five systems doing all of the banking of the country and if any one of them should fail, there would be a disaster in every part of our country growing out of that failure, while to-day it is local merely in its effect.” Senator Glass once again responded, “I can point the Senator to two middle western states where the depositors in banks have lost ten times as much money within the last 2 years as the depositors in all the Canadian banks lost in the last 10 years.”34 The day’ debate came to a close as Senator George of Georgia also stood in opposition of Lenroot. Congressman George articulated a counter argument: “In place of weakening the banks and in place of making more liable the collapse of the whole banking system, if we permit branch banking on a wide scale we get further and further away from the danger which may come to the system in a period of great 33 U.S.

Congressional Record, vol. 67, pt. 6–8, 1925–26, 69th Congress, 9300. Congressional Record, vol. 67, pt. 6–8, 1925–26, 69th Congress, 9302. It should be noted that in reference to the most recent financial meltdown in 2008, Mr. Lenroot would seem to have a valid point. While the financial system was considerably stronger in 2008 than in 1929, due in large part to the legalization of branching in 1994, the United States suffered a financial meltdown. Both catastrophes have a common feature, the government—both in 1929 and in 2008— passed legislation that significantly manipulated the markets. In the 1920s there was an unnatural expansion of credit that allowed investors to make poor investments that eventually caused the collapse of the banking system. In the late nineties and early two-thousands, Federal legislation encouraged bad loans by banks to people who did not have the ability to pay off their mortgages. Furthermore, throughout the nineties and early twenty-first century interest rates were kept at arbitrarily low rates causing once again unnatural expansion—especially in housing - that led to the destruction of the financial system.

34 U.S.

154

M. M. Witcher

depression or in a period of perhaps business failures throughout a large section of the country.” The words of Senator George come off the page as prophetic to the historian who knows and understands the great consequences of having a weak bank structure in the middle of an economic contraction.35 On May 13, 1926 debate resumed and Senator Glass sat back and allowed Senator McLean to defend the Senate’s version of the McFadden bill. McLean touted the McFadden bill’s ability to keep national banks from pursuing state charters. McLean went a step further and claimed: “Some of them have gone out of the Federal Reserve System and reorganized as State banks in order that they might have branch-banking privileges.” McLean’s comment drew a challenge from Senator Smoot, a Republican Congressman from Utah. “Mr. President, the Senator does not claim that this is the reason why many of the national banks have gone out of the Federal Reserve System, does he?” Senator Edge, a Republican Congressman from New Jersey, was able to head off what probably would have been a divisive argument by demonstrating that both were correct. Banks were not leaving the national system, Edge insisted, solely due to restrictions on branching but some banks, 166 in fact, left the national system to engage in branch banking under state charters.36 Senators Glass, Edge, and McLean continued to defend the bill as reported by the Senate Committee on Banking and Currency. Senator Glass discredited the idea that if the Hull Amendments were stripped from the bill that national banks would pressure state legislatures to pass legislation permitting branching. Glass pointed out that without the Hull Amendments, banks had the right to lobby the state legislatures and state-wide branching systems had not been enacted. The McFadden bill, as amended, passed the Senate with a vote of sixty yeas, seventeen nays, and nineteen not voting.37

7.5 The Conference Committee Congressmen from both houses determined, after the passage of the McFadden bill in the Senate without the Hull Amendments, that a conference committee should be established to work out the differences between the two bills. Senators Pepper, Edge, and Glass were appointed by the Senate to the committee. On May 25, 1926, Congressman McFadden once again went to the floor of the House to suggest that the House elect three members to the committee. In response to Congressman McFadden’s request, Representative Morton D. Hull of Illinois, the author of the Hull Amendments, rose: “Now I do not want to make a personal issue with the chairman of the committee or anybody else but I would welcome from the chairman

35 U.S.

Congressional Record, vol. 67, pt. 6–8, 1925–26, 69th Congress, 9302. Congressional Record, vol. 67, pt. 9–11, 1925–26, 69th Congress, 9338. 37 U.S. Congressional Record, vol. 67, pt. 9–11, 1925–26, 69th Congress, 9348. 36 U.S.

7 The Economics and Politics of Unit Banking: Evidence from the McFadden. . .

155

of the committee some declaration as to what his attitude would be in conference committee with reference to the branch-banking provisions in the bill.” McFadden responded that the Hull Amendments had passed the House because it was thought they limited the spread of branch banking. “The Senate,” McFadden pointed out, “however, not believing that this was the best plan in considering this matter have stricken out a portion of the Hull amendments.” Representative Hull interrupted, “They have stricken out all of it.”38 Congressman McFadden refrained from a direct confrontation with Hull and instead attempted to frame the debate: “Now, the question is, is the House more interested in the so-called Hull amendments, or is it more interested in the restriction, or a better restriction, on branch banking that the Hull amendments provide?” At this point a representative rose and drew applause by disagreeing with McFadden. He demanded instead that the House stand by the Hull Amendments and force the Senate to accept them.39 Representative McFadden, his time expired, relinquished the floor to Representative Wingo, a Democrat from Arkansas. Congressman Wingo, a member of the Committee on Currency and Banking, represented the greatest threat to McFadden’s goal of framing the Senate bill as a compromise piece of legislation that House members could accept. Wingo questioned the Senate’s judgment in stripping the Hull Amendments from the bill, claiming that misinformation and false statements derived from such misinformation led many senators to vote for the Senate’s amendments. Wingo continued, “Some of the gentlemen at the other end of the Capitol have admitted the facts frankly, stating that they knew nothing about the bill and relied upon Senator Glass.” Wingo claimed most of the senators as agreeing with the House position: “By conviction they are against branch banking and want to preserve independent unit banking in America. Knowing this, I say to you that this bill is dead unless the Senate recedes from its amendments.” Applause broke out from the floor, but Representative Wingo was not finished: “You can take me off the list of conferees if you like, but I state to you that I will never surrender to the Senate on this question, and if the Senate conferees insist upon it that kills the bill, because, by the eternal gods, I will never by any act of mine encourage the further spread of branch banking.”40 The House exploded with applause and Congressman McFadden scrambled to regain control. Before McFadden was able to have a vote taken on who should be appointed to the conference committee, Representative Hull was allowed to clarify on further point: “I know it has been argued that the bill in the form it passed the House is an attempt to interfere with States’ rights. That is a sham objection. No State right is interfered with. The States have still the rights to legislate as they please. All that has been done is to remove the motives for such legislation on the part of the banks located in such states. To call it an attempt to coerce the

38 U.S.

Congressional Record, vol. 67, pt. 9–11, 1925–26, 69th Congress, 10020, 10021. Congressional Record, vol. 67, pt. 9–11, 1925–26, 69th Congress, 10021. 40 U.S. Congressional Record, vol. 67, pt. 9–11, 1925–26, 69th Congress, 10024, 10025. 39 U.S.

156

M. M. Witcher

states is pure buncombe.” Indeed, there is little evidence to support Senator Glass’ suggestion that somehow the Hull Amendments would have restricted states from passing legislation that enabled their banks to branch. What the Hull Amendments would have restricted was national bank branching to states that allowed branching at the passage of the bill. Ironically, the Hull Amendments, by restricting national banks, would have maintained the incentive for state regulators, concerned with increasing the number of banks in their system, to legalize branching. As the debate drew to a close, it was determined that Congressmen McFadden, King, and Wingo would represent the House in the conference committee.41 While the conferees from the Senate and House were working to come to a conclusion on the McFadden bill, Senator Glass was communicating with members of the American Bankers’ Association. In a letter dated May 26th, a representative of the ABA wrote Glass to explain that Senator Glass did not “appreciate the difficulties under which [the ABA] has had to work in dealing with this particular question of branch banks.” The representative attempted to demonstrate how far the ABA had come in its position on branching in the last 5 years and cited that there were two distinct factions in the organization and that branching had a tendency to “swallow” the group of small banks and “curtail their privileges.” Senator Glass was not persuaded and continued to insist that the McFadden bill should be stripped of the Hull Amendments despite the ABA’s endorsement of those amendments.42 By mid-June it was apparent that the two versions of the bill could not be consolidated. On June 15th, Congressman McFadden presented the conference report concerning H.R. 2. The report had the signatures of five members. One signature was not present—that of Congressman Wingo. Representative Hull requested that the debate over the conference report occur at a different time as to allow a fuller discussion. Debate was rescheduled for June 22, 1926.43 Representative Wingo was granted 40 min at the opening of the debate of the McFadden bill on Tuesday, June 22nd. While the bill had been in committee, allegations had been whirling that Congressman Wingo was attempting to kill the bill by not attending committee meetings. He was also accused in letters of being a supporter of branch banking. Congressman Wingo blamed these attacks on the branch banking lobby in Washington.44 Throughout his defense of his position he pointed to the American Bankers’ Associations support for the Hull Amendments. Speaking of the Senate proposal Wingo said: “With a fine Italian hand they have injected an amendment here and another there, with the idea of producing hopeless confusion, and with the idea that you would yield upon the Hull amendments and 41 U.S.

Congressional Record, vol. 67, pt. 9–11, 1925–26, 69th Congress, 10026, 10028. dated May 26, 1926 from Thomas McAdams to Carter Glass. Small Special Collection Library, Papers of Carter Glass, Box 46. 43 U.S. Congressional Record, vol. 67, pt. 9–11, 1925–26, 69th Congress, 11316. 44 From my research, I see little reason to believe that anyone in Washington was actively working towards the establishment of branch banking—at least not in the open. There existed no branch banking lobby. There was, however, a unit banking lobby that was very active and engaged in behavior that Congressman Wingo attributed to the branch banking forces. 42 Letter

7 The Economics and Politics of Unit Banking: Evidence from the McFadden. . .

157

abandon the McFadden bill and accept the bill as it passed the Senate, the so-called Pepper bill.”45 Representative Wingo, after denouncing the Senate amendments, turned his attention to Congressman McFadden, who was willing to yield to the Senate amendments. Wingo demanded that the House vote on the conference report. The conference report, he stated, was in line with the Senate and blamed McFadden for going into the conference willing to give up the Hull Amendments saying that at that point the Senate “would be foolish to yield, would they not?” Wingo then recounted how he was willing to give the Senate all their other amendments, including the extension of the Federal Reserve charter, as long as the Hull Amendments were retained. The conference report instead was a complete acceptance of the Senate proposals. In closing Congressman Wingo rallied the opposition: The House passed this bill almost unanimously, and now those in favor of branch banking are going to ask you to betray the House and fly in the face of the unanimous endorsement of the American Bankers’ Association Monopoly of credit will enable those who desire and have the capital and credit necessary to drive out the independent drug stores, the independent factories, and independent groceriers, as well as the small independent banker, and then these independent concerns will join forces and call to account these men who have deserted them and gone over this branch banking monopoly.

The floor erupted with applause.46 At this point, McFadden, who had sat silently through Wingo’s speech, took the floor and presented the report of the conference committee. One could not think of a worse introduction for the report, and McFadden from the beginning was on the defensive. After the reading of the report, McFadden reminded the House that the purpose of the bill was to make national banks more competitive with state banks. The debate, however, had almost completely centered on branching. McFadden took a shot at the Federal Reserve System, claiming that the spread of branch banking began with the establishment of the “FED” and maintained that branch banking should be “considered an evil.” McFadden went on to state, “This bill absolutely confines branch banking to forty-four cities located in 15 States. Under the Hull provision, which the House passed on, branch banking will be permitted in 199 cities.” McFadden concluded by requesting the support of those Representatives that desired to truly check branch banking.47 McFadden, while answering a question from another Congressman, stated that Representative Wingo had “opposed this legislation from the beginning, both in committee by his votes and on the floor by his speeches and votes.” Wingo was infuriated and rose to defend himself: “Does the gentleman say that I have not fought for the House bill and against the Senate bill at all times?” Congressman McFadden responded: “The gentleman has voted and spoken against this bill every time it has been up in the House.” To which Wingo replied: “No, sir. The matter in controversy

45 U.S.

Congressional Record, vol. 67, pt. 9–11, 1925–26, 69th Congress, 11762. Congressional Record, vol. 67, pt. 9–11, 1925–26, 69th Congress, 11763, 11764. 47 U.S. Congressional Record, vol. 67, pt. 9–11, 1925–26, 69th Congress, 11767, 11768. 46 U.S.

158

M. M. Witcher

was that I have always stood by the House bill. The gentleman is the one who has been against his own bill. He has been voting for his bill and is against it. Does the gentleman know of a single day in conference where he ever stood by his bill, and has it not been like his speech today, defending the Senate bill?” McFadden, who had the floor, put the matter to rest with a final reply: “I will say, in answer to the gentleman from Arkansas, that I have been very consistent for this measure. The gentleman from Arkansas has been standing and insisting on one of the minor things in this bill, a provision he knew the Senators were unalterably opposed to, that which was the least important, in my judgment, of any provision in the bill. I have fought consistently to get this important banking measure enacted. The gentleman from Arkansas is in a very consistent position in opposing the measure today.”48 McFadden, in preparation of the vote on the conference report, again stated the main purpose of the bill and reassured the House that the conference report “arrived at a far more restrictive plan than is embodied in the Hull amendments” with regard to branch banking. Congressman McFadden then had three letters read into the Congressional Record that supported his position. The letters of support for the conference report were from Andrew Mellon, the Secretary of the Treasury; J.W. McIntosh, the Comptroller of the Currency; and Edmund Platt, representing the Federal Reserve Board. Once the letters were read into the Record, Congressman Hull made a point of order that McFadden had been speaking for more than an hour—far more time than he had been allotted. The House agreed that further time for debate was needed and a vote was postponed.49 On June 24, 1926, the conference report was once again brought to the floor of the House. Representative Hull was granted the floor and standing by his amendments he asked House members to “consider the public interest, to keep open the avenues of credit for young men of courage, vision and character. The branch banking system closes it to those who are not able to reach the head. It shuts out competitive and community banking. I hope you will vote down the Senate amendments.” Applause broke out in support for Hull’s position.50 Congressman McFadden, instead of making the closing arguments for the conference report, allowed Representative Beedy from Maine to rally support for the bill. Beedy made the same arguments that McFadden had made 2 days before for the report and concluded: “Congress is a human institution. The pending proposal therefore is not perfect. It does not and cannot in the nature of things suit everybody. We should vote for this compromise motion, however, and give the country much needed legislation.” Beedy recounted the support from the Treasury, the Federal Reserve Board and the Comptroller of the Currency and concluded that the measure should be adopted. Congressman Wingo, who undoubtedly knew the political winds

48 U.S.

Congressional Record, vol. 67, pt. 9–11, 1925–26, 69th Congress, 11769. Congressional Record, vol. 67, pt. 9–11, 1925–26, 69th Congress, 11770–11773. 50 U.S. Congressional Record, vol. 67, pt. 9–11, 1925–26, 69th Congress, 11917. 49 U.S.

7 The Economics and Politics of Unit Banking: Evidence from the McFadden. . .

159

favored him, called for the vote. The House rejected the conference report by a vote of 118 ayes, 197 nays, and 115 not voting.51 It became apparent after the failure of the compromise to pass the House that the McFadden bill was in real danger of not passing at all. The Senate, in a memo to the House, stated: “We are prepared either to recede or confer on every point in controversy except the point last mentioned [the Hull amendments]. But as the House conferees are bound by instructions which preclude them from receding upon this point, we see no way to prevent the disastrous failure of the legislation.” Representative Wingo cast the fault for the bill’s failing at the feet of one man: Carter Glass.52 A day after McFadden announced that the bill would remain in Congress until the next session in December, Senator Glass sent a letter to McFadden. In the letter, Glass was unwilling to yield to the House’s position on the Hull Amendments and wrote that he saw little point in further discussions due to the instructions that the House had placed on the House conferees to maintain the Hull amendments. It was clear in the summer of 1926 that it would take an Earth shattering event to alter the political stalemate.53

7.6 The ABA, Lobbying, and the Passage of the McFadeen Banking Bill In October, the American Bankers’ Association at their annual conference dropped a bombshell that shifted the political alliances concerning the McFadden bill. On the sixth, Senator Glass received a telegram from a member of the ABA. The telegram read simply: DEFEAT OF HULL AMENDMENTS BY AMERICAN BANKERS ASSOCIATION LAST NIGHT STOP GREAT VINDICATION OF YOUR MAGNIFICENT FIGHT IN WASHINGTON ON MCFADDEN BILL

The telegram’s revelation gave the McFadden bill new life, but with the good news came a new fight: the interpretation of the convention’s decision. Forces in

51 U.S.

Congressional Record, vol. 67, pt. 9–11, 1925–26, 69th Congress, 11923, 11926–11927. Wingo does not call out Senator Glass by name in the statement but earlier in one of his comments he mentions that Glass did not allow the Senate to vote on the Hull amendments. Then Wingo states: “I feel that the blame for the killing of the McFadden bill, the blame for having no restrictions upon branch banking, the blame for failing to renew the Charter of the Federal Reserve banks, can be placed squarely at the door of one man who even refused to confer upon the only single point which was left in disagreement.” Therefore, I deducted that Glass must be the conferee that Wingo is alluding to in the above statement. U.S. Congressional Record, vol. 67, pt. 9–11, 1925–26, 69th Congress, 13089, 13091. 53 Letter dated July 3, 1926 from Carter Glass on behalf of the Senate Conferees to Congressman McFadden. Small Special Collection Library, Papers of Carter Glass, Box 248. 52 Congressman

160

M. M. Witcher

favor of the Hull Amendments quickly rose to counter that the conference’s decision represented the will of California branch bankers and that most members of the ABA had not attended.54 In the aftermath of the ABA convention in Los Angeles, Representatives Wingo and Hull went on the offensive to discredit the endorsement of the ABA in favor of the McFadden bill as it passed the Senate. Both Congressmen were quoted in letters issued by the Committee of One Hundred—a group of bankers who formed an organization to support the Hull Amendments—discrediting the convention. Representative Hull claimed that the issue of the amendments was kept quiet as to ensure that only pro-branching forces would be present when a vote was placed on the amendments: “The branch banking question was not on the original official program of the convention. At the close of Tuesday morning general session of the convention a motion was passed—quickly and without debate—to hold an extra, unprogrammed meeting of the Association Tuesday evening. Official notice of this meeting was not given to the newspaper until nearly 6 o’clock—too late for publication in any newspapers which would appear before the convening of the meeting.”55 Congressman Wingo also encouraged the idea that the decision reached by the convention was unethical: They were careful to arrange the program so that this question was handled at a special snap meeting, at night, when it was known that fully three-fourths of the delegates would be absent The plan was well managed by the branch bankers, who had full control and the only thing necessary to transform it into a typical musical comedy would be for the band to play what under the circumstances would be very appropriate, that lifting, lively tune, The Parade of the Wooden Soldiers.’ These branch bank managers were nothing but wooden soldiers, voted by their leaders in blocks according to the latest approved methods of ward politics.56

Melvin A. Traylor, the president of the American Bankers’ Association, spoke at the Los Angeles convention in favor of the McFadden bill without the Hull Amendments and in light of the misstatements put forward by the Illinois based Committee of One Hundred, he approved a pamphlet to be prepared explaining to members of the ABA how the events at the convention transpired. The pamphlet countered the allegation that Los Angeles had been chosen by members who were pro-branching because they intended to overturn the Hull Amendments and

54 Telegram

dated October 6, 1926 from John S. Drum to Carter Glass. Small Special Collection Library, Papers of Carter Glass, Box 248. 55 Memorandum, “Facts verses Misstatements Re The McFadden Banking Bill,” prepared by Edmund S. Wolfe,Chairman, Committee on Federal Legislation, and approved by the President of the American Bankers’ Association, Melvin A. Traylor. Small Special Collection Library, Papers of Carter Glass, Box 240. 56 Memorandum, “Facts verses Misstatements Re The McFadden Banking Bill,” prepared by Edmund S. Wolfe,Chairman, Committee on Federal Legislation, and approved by the President of the American Bankers’ Association, Melvin A. Traylor. Small Special Collection Library, Papers of Carter Glass, Box 240.

7 The Economics and Politics of Unit Banking: Evidence from the McFadden. . .

161

California would provide a base of support for branching. It also pointed out that the decision to have the convention in Los Angeles was arrived at during the convention in Chicago, a year before, and at that time the issue of the Hull Amendments was settled in the minds of almost all the members of the ABA with the ABA supporting the Hull amendments. The pamphlet continued that it was the Committee of One Hundred that wanted to put the ABA on record as being unmistakably in favor of the Hull Amendments at the Los Angeles convention. Countering Representative Hull’s assertions, the pamphlet demonstrated that ABA President Traylor made a motion in open convention on Tuesday morning to have a special session to deal with the amendments. The motion passed a little after noon on that day and typewritten notices were sent to the press at 12:30 in the afternoon. To further spread the news of a special session, bulletins were posted in hotel lobbies and in the headquarters of the convention. Furthermore, verbal announcements were made in afternoon sessions of the convention.57 The pamphlet also confronted the idea that the special session was held at the request of the branch bankers. Both those for and against the Hull Amendments were involved in determining the rules for the session. It was agreed that there would be at least 3 h for debate and that the time would be split evenly between proponents and opponents of the amendments. The special convention was attended by 681 delegates of the 1439 delegates who attended the convention. The pamphlet also demonstrated that the idea that all the branch bankers voted for the abolition of the Hull amendments was incorrect. Some voted for the amendments with the hope that if they were adopted by the ABA the McFadden bill might not be passed at all. After all, the McFadden bill in any form would limit branching. When the final votes were cast 268 delegates voted to retain the Hull Amendments and 413 voted to endorse the McFadden bill as it was reported out of the Senate. Addressing the statements of Congressman Wingo, the pamphlet declared that “the attempt by Congressman Wingo to characterize a serious session of the American Bankers’ Association as a musical comedy performance will not be dignified by comment.” It seems likely that the ABA altered its position because it realized that while the bill would grant limited branch banking privileges to national banks in the short term, it would also stop any attempt to establish interstate branching. As such, they embraced the McFadden bill as a powerful anti-branching measure that would protect unit banks from future branching legislation.58 With the ABA renouncing its support for the Hull Amendments, the opportunity for the passage of the McFadden bill was once again a reality. After the ABA

57 Memorandum,

“Facts verses Misstatements Re The McFadden Banking Bill,”prepared by Edmund S. Wolfe,Chairman, Committee on Federal Legislation, and approved by the President of the American Bankers’ Association, Melvin A. Traylor. Small Special Collection Library, Papers of Carter Glass, Box 240. 58 Memorandum, “Facts verses Misstatements Re The McFadden Banking Bill,” prepared by Edmund S. Wolfe,Chairman, Committee on Federal Legislation, and approved by the President of the American Bankers Association, Melvin A. Traylor. Small Special Collection Library, Papers of Carter Glass, Box 240.

162

M. M. Witcher

convention, Senator Glass wrote that he “was glad to note that the American Bankers’ Association, at its annual meeting in Los Angeles, completely reversed itself on the proposed Hull amendments to the McFadden bill, which shows how foolish it was ever to have been drawn into an endorsement of such an atrocious position. There is still work to be done to get the House of Representatives to change its attitude.” Luckily for Senator Glass, the same force that had flooded his office with letters in support of the Hull Amendments now sent letters to congressmen urging the passage of the McFadden bill without the amendments. The ABA also mailed each member of Congress a copy of the debate and conclusions of the Los Angeles convention regarding the Hull Amendments. The letters also urged each congressman to vote for the measure as it passed the Senate. Congressmen, with their local bankers behind their votes, were more willing to vote the McFadden bill into law.59 On January 17, 1927, the conferees from the Senate and House met to work out the remaining differences in the two versions of the bill. Two days later McFadden had drafted the version of the bill he would present to the House. McFadden mailed the Senate conferees a copy of the bill and Glass responded: “We find ourselves unanimously in accord with its terms, and if the resolutions adopted by the House in form submitted, we, as conferees on the part of the Senate, will give it our unqualified support in the Senate.”60 Hull made a final attempt to oppose the passage of the McFadden bill but even he, in his final statement, acknowledged defeat: “I know if it was left to the judgment of the House, without any outside pressure, what the result would be; but the whip has been cracked and the lash has been laid on the backs of members.” The McFadden bill passed the House after support for the bill was exchanged for support of the McNary-Haugen bill for agricultural relief. The bill then went back to the Senate, where Senators Pepper and Glass were forced to enact cloture to bring the bill to a vote. A newspaper on Monday reported: “This banking bill, the most important since the Federal Reserve Act, was approved by the Senate, 71 to 17, on the day after the adoption of cloture was sent to President Coolidge.”61 Senator Glass, speaking to the press after the passage of the bill in both Houses of Congress, praised the bill: I think the country is to be congratulated that the perplexing branch bank question has apparently been equitably settled for a long time to come and that the Federal Reserve System is given such an overwhelming vote of confidence in both branches of Congress;

59 Letter

dated October 18, 1926 from Carter Glass to Mr. Paul M. Warburg, Box 8. This letter is misdated as October 18, 1936. This date is impossible due to the fact that Mr. Warburg died in 1932; letter from Thomas B. Paton, the General Council for the ABA to Carter Glass dated December 2, 1926. Small Special Collection Library, Papers of Carter Glass, Box 248. 60 Letter dated January 14, 1927 from Congressman McFadden to Carter Glass and a letter dated January 20, 1927 from Carter Glass to McFadden. Small Special Collection Library, Papers of Carter Glass, Box 240. 61 U. S. Congressional Record, vol. 68, pt. 1–3, 69th Congress, 2169.

7 The Economics and Politics of Unit Banking: Evidence from the McFadden. . .

163

for it must be recalled that a separate vote on the charter extension provision demanded in the House resulted in 298 ayes and 22 nays, whereas in the Senate the bill receives its final endorsement by a vote of 71 to 17.62

A newspaper report also concluded that the extension of the Federal Reserve charter was a big victory of that system as it “puts the positive stamp of approval of a Republican administration on the greatest domestic achievement of the Wilson regime.” Senator Glass emerged from the debate on the McFadden bill as the big winner. He was able to strip the bill of the Hull Amendments and was able to tack onto the bill an indeterminate charter for the Federal Reserve System that he helped to create. Charles S. Hamlin, a member of the Federal Reserve Board wrote Senator Glass to congratulate him on the passage of the McFadden bill: “I want to congratulate you upon your wonderful achievement in getting the McFadden bill finally through. I remember the old problem—‘what happens when an irresistible projectile meets an impenetrable object.’ You are the irresistible projectile, and in this case the impenetrable object has been penetrated.” The McFadden bill had been signed into law, but there would only be a short reprieve until further legislation concerning banking would be needed.63

7.7 Conclusion Under the McFadden bill, national banks were permitted to branch in states that allowed branching. The bill had the effect of halting the spread of interstate, and in many states, the spread of intrastate branching. The legislation did achieve its goal of making national charters more appealing to state banks. From 1884 to 1926, the amount of commercial bank resources in the hands of national banks had decreased from 75% to 46%. The McFadden bill righted the national banking system, which the New York Times described as a “sinking ship” in 1922.64 For all its success, however, the McFadden bill did little to tackle the causes of bank failures that were occurring at a rapid and alarming pace throughout the twenties. Chapman and Westerfield conclude that the McFadden Act “ignored entirely the condition of the country banks and not only did nothing whatever to relieve the situation which had resulted in an increasing number of failures of small banks but tended to make the situation worse by preventing within the Federal

62 Chapman

and Westerfield (1942, p. 107); “Bank Bill,” National Affairs, February 28, 1927. by Carter Glass concerning the passage of the McFadden bill dated February 16, 1927. Small Special Collection Library, Papers of Carter Glass, Box 240; Letter from Charles S. Hamlin to Carter Glass dated February 17, 1927. Small Special Collection Library, Papers of Carter Glass, Box 64. 64 Robertson (1968, p. 97), Meltzer (2003, p. 217), Chapman and Westerfield (1942, p. 108); “National Banking System Rapidly Disintegrating,” New York Times, March 7, 1922. Small Special Collection Library, Papers of Carter Glass. 63 Comments

164

M. M. Witcher

Reserve System further consolidations outside of city limits in branch banking states” (Chapman and Westerfield 1942, p. 108). The McFadden bill ultimately did little to address the weak and undiversified American banking system, however, it achieved what little could be achieved given the political circumstances. There was simply no way any legislation that explicitly embraced branch banking could have passed either the House or the Senate given the power of the unit banking lobby and the American Bankers Association. Indeed, without the ABA switching its lobbying efforts in favor of the McFadden bill the legislation would have failed. In short, while economists are correct that a more diverse banking system, similar to the one Canada had at the time, would have better prepared the United States for the monetary contraction that ensued during the Great Depression, it was not a possibility in 1927. The Federal Reserve raised the discount rate in 1928, and again in 1929. In all, the discount rate doubled from 1927 to 1929. This increase in interest rates caused contraction in the economy and led to a downturn in business. Unit banks that had been shielded from competition by bills that restricted branching were undiverse and contained limited portfolios—primarily consisting of agricultural loans. When contraction occurred, and the easy credit that had marked the twenties disappeared, all the malinvestment of the period was exposed. The hardship was further increased by the passage of the Smoot-Hawley tariff and the country’s poor economic conditions were worsened by the Hoover tax hikes and his administration’s efforts to keep prices from falling. From 1929 to 1933, almost ten thousand banks failed, the majority being small unit banks that served rural areas. Thirty-nine percent of all banks in existence in 1928 had failed by 1933. It is unclear whether or not the bank failures of this time period experienced a contagion effect. Most of the bank failures were regional and there is not clear evidence that points to panic causing the widespread collapse of the entire structure. What is clear from economic research is that states with some sort of state deposit insurance experienced higher rates of bank failures than other states. Economic research has also discerned that states that allowed branch banking suffered fewer bank failures than states that restricted branching. Furthermore, Canada, the country that many congressmen claimed to be monopolistic during the debate on the McFadden bill, experienced zero bank failures during the Depression despite having a similar decline in production and overall economic performance (Wicker 1996, p. 2; Friedman and Schwartz 1963, p. 352; Grossman 1994, p. 655). For Senator Glass, the passage of the McFadden bill ensured that the Federal Reserve System would not share the fate of the First and Second Banks of the United States. In 1932 and 1933, Glass again demonstrated his leadership abilities by drafting legislation aimed at stabilizing the failing American financial system. The

7 The Economics and Politics of Unit Banking: Evidence from the McFadden. . .

165

merit of that legislation is up for debate, but Senator Glass continued throughout his tenure in Washington to offer leadership and ideas concerning matters of finance.65 Congressman McFadden, unlike Carter Glass, would not demand that further powers be granted to the Federal Reserve to combat the depression. Instead, McFadden went on the offensive against those, who he believed, had caused the economic downturn. McFadden called for the impeachment of Herbert Hoover and for an audit of the Federal Reserve Board. In McFadden’s words, the Board of Governors had “treasonably conspired to destroy constitutional government in the United States” and had “caused the greatest depression we have ever known.” Many thought that McFadden had lost his mind, but the founder of the National Press Club, George Stimpson, had a different view of McFadden’s accusations: “It was incredible. This town went into a state of shock, we couldn’t believe what we were hearing. Of course, they said right away that he had lost his mind.” When asked if McFadden had indeed “gone off the deep end,” Stimpson replied, “Oh, nobut it was too much, too much for one man to do” (McFadden 1970, pp. v–x). Despite McFadden’s calls for an audit of the Federal Reserve System, no such action was taken and instead the American people willingly embraced an increased role for government in the economy. Congressman McFadden lost reelection in 1934, and died shortly after, on October 1, 1936.

References Chapman JM, Westerfield RB (1942) Branch banking: its historical and theoretical positions in America and abroad. Harper, New York Eckardt H (1910) Branch banking among the state banks. Banking problems. Ann Am Acad Pol Soc Sci 36(3):148–161 Friedman M, Schwartz AJ (1963) A monetary history of the United States: 1886–1960. Princeton University Press, Princeton Grossman RS (1994) The shoe that didn’t drop: explaining bank stability during the great depression. J Econ Hist 54(3):654–682 Hammond B (1991) Banks and politics in America from the revolution to the civil war. Princeton University Press, Princeton Katz BS (1992) Biographical dictionary of the Board of Governors of the Federal Reserve. Greenwood Press, New York McFadden LT (1970) Collective speeches of congressman Louis T. McFadden. Omni Publications, Hawthorne, CA Meltzer AH (2003) A history of the Federal Reserve: Volume I, 1923–1951. University of Chicago Press, Chicago Robertson RM (1968) The comptroller and bank supervision: a historical appraisal. McCall, Washington, DC Spahr WE (1932) Bank failures in the United States. Am Econ Rev 22(1):208–238

65 Letter from Carter Glass to Mr. Walter Edward Harris of the Rotary Club of Petersburg, Virginia,

dated June 28, 1927. Small Special Collection Library, Carter Glass Papers, Box 4; Biographical Dictionary of the United States Congress, Carter Glass.

166

M. M. Witcher

White EN (1982) The political economy of banking regulation, 1864–1933. J Econ Hist 42(1): 33–40 White EN (1983) The regulation and reform of the American Banking System, 1900–1929. Princeton University Press, Princeton Wicker E (1996) The banking panics of the great depression. Cambridge University Press, New York Wilentz S, Earle JH (2008) Major problems in the early republic, 1787–1848. Houghton Mifflin, New York

Chapter 8

Immigrant Ethnic Composition and the Adoption of Women’s Suffrage in the United States Ho-Po Crystal Wong, J. R. Clark, and Joshua C. Hall

Abstract This paper seeks to understand the role played by immigrant ethnic composition in the process of women’s suffrage in the United States. Any theory of the extension of voting rights to women must explain why native men voted to extend the franchise to women. In this paper, we consider what we call the “ethnic group threat.” To the extent that native males believed that the political preferences of native women were better aligned with theirs than new (primarily male) immigrants, male voters would be willing to grant women voting rights to secure their social and political status. We use a hazard model and immigration data from 1870 to 1920 to investigate the impact of immigrant ethnic composition on women suffrage, we find that states with a higher proportion of immigrants from Italy, Eastern/Southern Europe, and Mexico gave women the the right to vote faster.

8.1 Introduction The economic theory of rights is a purely positive theory of the rights that individuals are willing and able to claim (Holcombe 2014). The ability of people to claim and enforce rights, such as voting rights, is dependent on others respecting their claims, either through mutual agreement or through force. With respect to extensions of suffrage, a number of arguments have been put forth regarding why those with political power would extend the voting franchise. Most of the arguments

H.-P. C. Wong National Tsing Hua University, Hsinchu, Taiwan e-mail: [email protected] J. R. Clark The University of Tennessee at Chattanooga, Chattanooga, TN, USA e-mail: [email protected] J. C. Hall () West Virginia University, Morgantown, WV, USA e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2018 J. Hall, M. Witcher (eds.), Public Choice Analyses of American Economic History, Studies in Public Choice 37, https://doi.org/10.1007/978-3-319-95819-4_8

167

168

H.-P. C. Wong et al.

rest on the threat of violence by the disenfranchised. For example, Acemoglu and Robinson (2000) argue that the political elite extend the franchise in order to avoid revolution.1 Using a data set covering 187 countries during most of the twentieth century, Przeworski (2009) finds that in most cases those outside of power had to fight their way to suffrage. One of the few exceptions was Great Britain, where the voting franchise was gradually extended over time to non-elites without a revolutionary threat. Women’s suffrage is another exception. As noted by Doepke et al. (2012) in their review article on the economics and politics of women’s rights, the literature on the origins of women’s rights is small. Much of the the U.S. literature focuses on property rights. For example, Geddes and Lueck (2002) looked at two statutory changes affecting women from 1850 to 1920: laws giving them control of their earnings and laws allowing them to manage estates separate from men. Using a property rights perspective, their analysis focuses on the incentives facing men to extend property rights to women. Primarily their work focuses on how rising living standards and changes in the U.S. economy over the period increase the value of women having property rights. Their empirical analysis finds that changes in wealth, city growth, and female human capital accumulation all positively explain the expansion of these two statutory changes. A handful of papers have followed up on Geddes and Lueck (2002). Doepke and Tertilt (2009) present a theoretical model where men would prefer that their wives have no property rights but that other women have property rights. The intuition behind this model is that men care about their daughters and an expansion of women’s rights leads to a greater investment in children’s human capital, especially females. Geddes et al. (2012) look at school attendance rate for girls between 15–19, and find that states that gave women property rights had higher relative rates of female school attendance. Fernández (2014) develops a similar model that highlights the role of fertility in this dynamic. In her model, lower fertility exacerbates the welfare levels between daughters and sons. Fernández (2014) finds, using data on the same period as Geddes and Lueck (2002), a negative correlation between fertility and reform. Finally, Lemke (2016) highlights how interjurisdictional competition between states for women was a contributing factor to the expansion of married women’s property rights. States and territories in the Western United States where male settlers substantially outnumbered women faced stronger incentives to pass reforms that enhanced the private property rights of married women, especially when local legislators benefitted from a larger population. Teele (2018) also emphasizes the importance of political competition and mobilization in Western states. The literature on the extension of suffrage rights, which are but a small part of legal rights, is equally small.2 McCammon et al. (2001) focus on the role

1 For

a purely theoretical approach consistent with this argument, see Conley and Temimi (2001). is, however, a large literature on the effects of women’s suffrage, especially with respect to the amount and composition of government spending. In their well-cited paper, Lott and Kenny (1999) find that women’s suffrage was associated with increases in state government expenditures

2 There

8 Immigrant Ethnic Composition and the Adoption of Women’s Suffrage in. . .

169

played by political decision-makers towards women suffrage. They argue that policymakers are willing to support policy change that strengthens or preserves their own institutional positions. They provide empirical evidence that shifting gender relations in favor of more active women’s roles outside the private sphere of the household increased political decision makers’ willingness to support women suffrage. In economics, Bertocchi (2011) was the first published paper specifically on the expansion of voting rights for women. Her story is one of human capital accumulation. Rising levels of schooling for women led to a reduction in gender wage gap, narrowing gender differences in the preferred tax rate. Using crosscountry data, she finds empirical evidence to support her hypothesis. Braun and Kvasnicka (2013) confirm Bertocchi’s (2011) findings but find that imbalances in the adult sex ratio are more important to explaining the adoption of women’s suffrage laws by state. They argue that the relative scarcity of women lowered the potential costs for men in terms of devaluing men’s own political influence and the local marriage market squeeze might have led legislators to view granting women’s rights as a viable tool to lure female settlers, similar to the story of Lemke (2016) but for political rights. Using discrete time duration models, Braun and Kvasnicka (2013) show that the general scarcity of women indeed increased the hazard of women’s suffrage. In this paper we seek to improve our understanding of the timing of women’s suffrage laws in the United States by exploring the role of ethnic group “threats.” As argued in Andrews and Seguin (2015, p. 2) “demographic changes that challenge the economic, political, and cultural standing of established groups encourage widespread mobilization and support for new policies to maintain or restore group status.” The male dominated mass migration period of the United States provides an ideal social setting in which immigrants from diverse cultural backgrounds produced drastic demographic changes that could threaten the socioeconomic and political position of those already established. To the extent that male settlers that were native or from the better established ethnic groups believed that the political preference of women from their own groups were better aligned with theirs than the new male immigrants, the mass of male voters might be willing to grant the rights to vote to women if the majority of the male voters felt that such enfranchisement would secure their social and political status. We call this the “ethnic group effect.” We examine using hazard models the effect of changes in the proportion of the major old immigrant groups in the population including English,

and revenues. In the same vein, Aidt and Dallal (2008) look at a sample of six European countries from 1869–1960 and find that social sending increased in both the short and long run after women had the right to vote. Abrams and Settle (1999) look at the expansion of the voting franchise to women in Switzerland in 1971 and find social welfare spending increased by 28% after extension. Krogstrup and Wälti (2011) take advantage of the fact that women’s suffrage occurred in different Swiss cantons at different times and find that women’s suffrage led to reduced budget deficits. Miller (2008) finds that changes in women’s suffrage laws led to big changes in legislative behavior, especially spending on local public health.

170

H.-P. C. Wong et al.

German, and Scandinavian and the new immigrant groups such as Italian, Mexican, and Eastern/Southern European, on the timing of women’s suffrage passage in the US states.

8.2 Data and Empirical Approach We perform Cox Proportional Hazard models to estimate the relationship between women’s suffrage and ethnic composition of immigrants as well as other state demographics. Let Ti be the year that state s passed women’s suffrage, which is a random variable and t represents the year. The baseline hazard function h0 (t) is left unspecified. The hazard function is then given by: 





h(t) = P (Ti = t|Ti ≥ t) = h0 (t)expβ EthnicSizest +ψ Xst +ρ Regions +st

(8.1)

where s denotes state and t takes the value corresponding to the calendar year. Therefore covariates enter the hazard model through the linear predictor and they shift the baseline hazard function multiplicatively by: φst = β  EthnicSizest + ψ  Xst + ρ  Regions + st

(8.2)

Where EthnicSizest is a vector of ethnic time-variant covariates capturing the size of the ethnic group and is given by: EthnicPopist ∗ 100 Population aged 15 − 49

(8.3)

Where EthnicPopist is the estimated number of individuals from ethnic group i aged 15–49 in state s in year t and Population aged 15–49st is the estimated state population aged 15–49 in state s in year t; Xst is a vector of state demographic controls; Regions represent a vector of regional dummies including West, Midwest, Northeast and South (omitted). We make use of the variation in ethnic immigrant concentration to the United States across states and over time to estimate the effect of immigrant group threat on women suffrage. Brits, Irish, Scandinavian and Germans arrived in large numbers prior to the nineteenth century while Italians, other Eastern and Southern Europeans, and Mexicans began to arrive in large scale from 1900 onward. This is also a period that corresponds to a time of widespread introduction of women suffrage across states (see Appendix Table 8.4). In addition, immigrants of different European ancestries tended to concentrate in different areas of the United States, for instance, historically Italians tended to concentrate in ethnic enclaves on the East Coast such as New York City, while Germans concentrated in the Midwest. The differences in ethnic immigrant composition across states thus provide another source of variation in ethnic population to test our ethnic group threat hypothesis. Our hypothesis

8 Immigrant Ethnic Composition and the Adoption of Women’s Suffrage in. . .

171

Table 8.1 Descriptive statistics Covariates (British aged 15–49/population aged 15–49)*100 (Irish aged 15–49/population aged 15–49)*100 (German aged 15–49/population aged 15–49)*100 (Scandinavian aged 15–49/population aged 15–49)*100 (Central European aged 15–49/population aged 15–49)*100 (Eastern-Southern European aged 15–49/population aged 15–49)*100 (Italian aged 15–49/population aged 15–49)*100 (Mexican aged 15–49/population aged 15–49)*100 (Other population aged 15–49/population aged 15–49)*100 Proportion of literate population aged 15–49 (Literate woman aged 15–49/women aged 15–49)*100 (Female natives aged 15–49/population aged 15–49)*100 Proportion of female population aged 15–49 Proportion of women aged 51–49 in the labor force Proportion of working population aged 15–49 in the brewing industry Proportion of rural population Proportion of African-American population

Min Mean Max St. Dev. 0.176 4.163 19.10 2.888 0.065 7.352 29.05 6.971 0.167 10.31 38.48 7.959 0 2.858 32.68 5.718 0 1.299 5.502 1.018 0

4.574

20.47

4.950

0 0 0.035

1.359 0.343 4.066

11.10 37.63 40.53

2.164 1.646 5.911

37.67 19.17 15.72 23.86 3.950 0

88.11 88.02 49.73 49.11 23.11 0.252

18.17 0

60.38 12.02

99.17 10.17 98.27 10.86 53.26 1.827 53.00 2.179 50.24 7.479 1.104 0.208 100 61.28

21.15 16.82

Note: N=1624. The means are weighted by state population

suggests that an increase in the population share of ethnic groups that are culturally and religiously more distinct from the established Anglo-Saxon ethnic groups posed a political threat to incumbents groups, and thus will raise the likelihood of native males being in favor of women’s suffrage. The data for our analysis come from the 1880, 1900, 1910 and 1920 Census IPUMS, 1% sample. The individual data is aggregated at the state level weighted by the person weight. As annual data on the state population is unavailable, the values of all the state variables employed in the hazard model analysis are estimated by linear extrapolation using the above census data. The resulting time to event data consists of 1624 observations. Table 8.1 presents the summary statistics of the covariates used in analysis. The focused ethnic groups include “British”, “Irish”, “Scandinavian”, “German”, “Central European”, “Eastern/Southern European”, “Italian” and “Mexican”. “British” is made up of individuals from England, Scotland, Wales and United Kingdom. “Scandinavian” comprises individuals from Sweden, Finland, Denmark, Norway and Iceland. We also control for the state population that fall outside of these ethnic groups and the native stock. This group of population is identified as “others”. The ethnic counts include both first and second generation immigrants. For second generation native born individuals with both parents being foreigners, they would be categorized into the ethnic group of his mother as second generation unless

172

H.-P. C. Wong et al.

the father is the only parent that is foreign. If threats from relatively more recent immigrants that are culturally more distant from the native population are indeed significant determinants of female suffrage, we will expect a rise in the population share of ethnic groups such as “Irish”, “Italian”, “Eastern/Southern European” and “Mexican” to increase the hazard rate of female suffrage. Appendix Table 8.3 details how country of origin in the Census was translated into our ethnicity groupings. As state sex ratios could also affect competition in the marriage market, states with a scarcity of women might compete by granting women the right to vote. To see whether and to what extent the relative abundance of women in the population would affect our results, we control for the state female population aged 15–49 and native female population aged 15–49 as a proportion of the total population, depending on specifications. The list of other covariates introduced in our hazard model include the proportion of women aged 15–49 in the workforce, the proportion of literate women aged 15–49, the proportion of literate population aged 15–49, the state level proportion of workers in the brewery industry in the workforce, the state level rural population and the state level black population. Women are expected to have a stronger demand for political rights as they became economically more independent. A more literate population, especially a more literate female population could enhance the promotion of gender equality and therefore could speed up the acquisition of women suffrage. Also as women suffragists traditionally had been strong supporters of prohibition, states with a larger share of population working in the brewery industry might be less willing to grant women voting rights, we therefore also add the state level proportion of workers in the brewery industry in the workforce. Furthermore immigrants tend to concentrate in cities, and therefore a larger rural population is likely to reduce the hazard rate of women suffrage as they face less immigrant threat. Lastly we include the state level proportion of African-American population in the hazard model because states with higher levels of African-Americans faced stronger opposition to female suffrage (Braun and Kvasnicka 2013).

8.3 Empirical Results The ethnic group hypothesis suggests that an increase in the size of the “new immigrant” ethnic groups such as Italians, Southern/Eastern European, Mexicans will increase the hazard rate of women’s suffrage. Table 8.2 presents our empirical results. We present the estimated hazard ratios eβi : a one unit increase in the covariate will change the hazard risk by (eβi −1)*100%. For the vector of covariates of interest related to EthnicSize, this means when the size of ethnic group i increase by 1%, the hazard of women suffrage is changed by (eβi − 1)*100%. Numbers greater than one indicate that the variable is positively associated with a state granting women suffrage rights before passage of the 19th Amendment. Numbers less than one indicate the opposite.

8 Immigrant Ethnic Composition and the Adoption of Women’s Suffrage in. . .

173

Table 8.2 Hazard models for women’s suffrage in the United States, 1870–1919 Covariates British share

Total female proportion

(1) 0.588 (0.236) 1.276 (0.567) 0.903 (0.112) 0.971 (0.037) 1.294 (0.587) 1.242* (0.159) 1.716*** (0.291) 1.405*** (0.179) 1.198** (1.150) –

Native female proportion

–

(2) 0.680 (0.316) 1.203 (0.603) 0.874 (0.132) 0.988 (0.048) 1.279 (0.647) 1.310* (0.211) 1.788*** (0.324) 1.486*** (0.217) 1.174** (0.085) 1.360 (0.576) –

State demographics Regional dummies LR χ 2

X X −68.30

X X −68.07

Irish share German share Scandinavian share Central European share Eastern/Southern European share Italian share Mexican share “Others” share

(3) 0.911 (0.415) 1.047 (0.411) 0.875 (0.133) 1.019 (0.050) 1.132 (0.625) 1.297* (0.181 1.929*** (0.482) 1.642* (0.424) 1.142** (0.069) 5.522 (8.719) 0.274 (0.336) X X −66.98

Notes: Estimates are hazard ratios. Robust standard errors are in parentheses and Native Share is the omitted group. Regional Dummies are for West, Midwest, and Northwest, with South the omitted region. State demographic controls include proportion of females aged 15–49 in the labor force, proportion of rural population, proportion of literate women aged 15–49, proportion of African America population, and proportion of working population aged 15–49 in the brewing industry. N=1624 and the number of failures is 25 in each specification *** denotes variable is statistically significant at the 1% level; ** at the 5% level; and * at the 1% level

Column (1) of Table 8.2 includes just the shares of our ethnic groups, state demographics, and regional dummies. The results support the ethnic group threat hypothesis. Eastern/Southern European Share, Italian Share, Mexican Share, and “Others” Share are all positively and statistically significantly associated with a state adopting a women’s suffrage law earlier. In column (2) we add Total Female Proportion to the hazard model and we find the variable to be statistically insignificant and our primary finding unchanged. The proportion of 15–49 year olds that are women in a state does not seem to have made it more likely for a state to have adopted woman’s suffrage earlier. In column (3) we provide our full model that includes all of our ethnic group threat variables, the total female proportion and the native female proportion along with all of our demographic controls and regional

174

H.-P. C. Wong et al.

dummies. Our results remain unchanged. The proportion of native women in a state did not seem to impact the the passage of female suffrage. There are good reasons to think that the social and political preference of the majority of men from the established ethnic groups were better aligned with women than with the new immigrants that were primarily male during a period with a massive influx of immigrants from Eastern and Southern Europe. For example, O’Neill (2013) illustrated that the women’s suffrage movement over time became increasingly connected to nativism. Increasingly more emphasis was placed on the preservation of Anglo-Saxon or Nordic supremacy as a justification for enfranchising women. Such emphasis was founded on the growing fear of the middle class around the large influx of immigrants from countries with very different cultures and ideologies, which the old settlers believed to be inferior to the mainstream American values.3 Given that migration during that period was largely dominated by men, enfranchising women would secure an electoral majority for the Americans of Northern or Western European ancestry. Cohen (1996) also argued that the women’s suffrage movement was a realization of nationalism through an alliance between two dominant groups in American society: white American women and men. In achieving the political goals of woman’s suffrage, the suffragists were often explicit in their opposition to nonwhite and foreign-born women and men. He suggested that suffragists convince the ruling white men to vote for women’s suffrage by emphasizing the threats posted by nonwhites and European immigrants (primarily from Southern and Eastern Europe) to the “hegemonic” white national identity at the turn of the twentieth century. An example of legislations that illustrate the contrasting political position of the old and new immigrants is the series of immigration quota restrictions that were introduced in the 1920s, after the ratification of the 19th Amendment that granted the right of suffrage to women in all states. The 1921 Emergency Quota Act restricted the number of immigrants in any year to no more than 3% of the number of foreign-born persons from a nation residing in the United States in 1910 based on the 1910 census.4 In 1924, the annual quota was further reduced to 2 per cent of the number of foreign-born persons from a country and the base census year was moved to 1890, which further privileged the ethnic groups from Western and Northern Europe, as the largest influx of immigrants of Anglo-Saxon and Nordic ancestry arrived prior to 1890 (Bloch 1929). The aim of this series of quota acts was to limit the number of immigrants and control the ethnic composition of the new immigrants. An immediate effect of the quota law was that the number of immigrants from Southern and Eastern Europe had been drastically reduced, as they only began to arrive in the United States on a large scale after 1890. Historical studies on the Immigration Act of 1924 also suggested that the law was nativist in

3 Goldin

(1994) notes that those of German and British ancestry opposed immigration during this period. 4 For more on the Emergency Quota Act and the political economy of immigration restrictions from 1890 to 1921, see Goldin (1994).

8 Immigrant Ethnic Composition and the Adoption of Women’s Suffrage in. . .

175

nature. “The central theme of that process [passage of the law] was a race-based nativism, which favored the ‘Nordics’ of northern and western over the ‘undesirable races’ of eastern and southern Europe (Ngai 1999, p. 69).” The passage of these quota acts provided a clear picture of the opposing political preferences of the new and old immigrants. And the fact that the enfranchised women were mostly from the established ethnic groups and generally were nativists themselves made them important political allies of their male counterparts in an era with strong anti-immigrant sentiment. Our results are consistent with these new immigrants, primarily from Italy, Mexico, and Eastern/Southern Europe, being seen as a threat and women’s suffrage was a way for natives to deal with this threat.

8.4 Conclusion From 1850 to 1915, more than 30 million immigrants came from Europe to the US (Abramitzky et al. 2012). Over time the composition of that those immigration flows changed from English, German, and Irish to Italian, Eastern and Southern European, and Mexican. A number of prominent economic historians have written about the composition and consequences of this “Age of Mass Migration” (Hatton and Williamson 1998; Abramitzky et al. 2014, 2016). Recently Tabellini (2018) finds considerable political backlash in areas with high levels of immigration between 1910 and 1930, in spite of considerable economic benefits for natives in those area. He attributes this political backlash between natives and new immigrants to cultural distance between immigrants and natives and finds empirical support for it.5 Although not explored in his paper, our findings that increasing shares of immigrants from Italy, Eastern and Southern Europe, and Mexico were more likely to result in native men giving women the right to vote is consistent with this nativist political backlash.

5 While

Abramitzky et al. (2016) do not have data for Mexican immigrants or Eastern/Southern Europeans, they find that Italians were the least likely to marry outside their ethnic group in the first-generation. Only 11% of first-generation Italians married a non-Italian contrasted with 72% of Scots (the highest rate of out-group marriage).

176

H.-P. C. Wong et al.

Appendix See Tables 8.3 and 8.4. Table 8.3 Ethnicity grouping detail

Country code in census 200 400 401 402 404 405 410 411 412 413 414 420 421 423 425 426 433 434 436 438 450 452 453 454 455 456 462 542

Country

Ethnicity

Mexico Denmark Finland Iceland Norway Sweden England Scotland Wales United Kingdom Ireland Belgium France Luxembourg Netherlands Switzerland Greece Italy Portugal Spain Austria Czechoslovakia Germany Hungary Poland Romania Lithuania Turkey

Mexican Scandinavian Scandinavian Scandinavian Scandinavian Scandinavian British British British British Irish Central European Central European Central European Central European Central European Eastern/Southern European Italian Eastern/Southern European Eastern/Southern European Eastern/Southern European Eastern/Southern European German Eastern/Southern European Eastern/Southern European Eastern/Southern European Eastern/Southern European Eastern/Southern European

Note: Individuals are categorized into different ethnic groups using codes of their places of birth in the IPUMS Census data. Native-born individuals whose parents are foreign are categorized into their mothers’ ethnic group In cases where the mother is native while the father is foreign, the individual was categorized into his/her father’s ethnic group

8 Immigrant Ethnic Composition and the Adoption of Women’s Suffrage in. . .

177

Table 8.4 Year of women’s suffrage, by state and region State West: Arizona California Colorado Idaho Montana Nevada New Mexico Oregon Utah Washington Wyoming Midwest: Illinois Indiana Iowa Kansas Michigan Minnesota Missouri Nebraska North Dakota Ohio South Dakota Wisconsin

Suffrage year 1912 1911 1893 1896 1914 1914 1920 1812 1870 1883 1869

1913 1919 1919 1912 1918 1919 1919 1917 1917 1919 1918 1919

State Northeast: Connecticut Maine Massachusetts New Hampshire New Jersey New York Pennsylvania Rhode Island Vermont

Suffrage year 1919 1920 1920 1920 1920 1917 1920 1917 1920

South: Alabama Arkansas Delaware Florida Georgia Kentucky Louisiana Maryland Mississippi North Carolina Oklahoma South Carolina Tennessee Texas Virginia West Virginia

1920 1917 1920 1920 1920 1920 1920 1920 1920 1920 1918 1920 1919 1918 1920 1920

Sources: Carter et al. (2006) and Braun and Kvasnicka (2013)

References Abramitzky R, Boustan LP, Eriksson K (2012) Europe’s tired, poor, huddled masses: self-selection and economic outcomes in the age of mass migration. Am Econ Rev 102(5):1832–1856 Abramitzky R, Boustan LP, Eriksson K (2014) A nation of immigrants: assimilation and economic outcomes in the age of mass migration. J Polit Econ 122(3):467–506 Abramitzky R, Boustan LP, Eriksson K (2016) Cultural assimilation during the age of mass migration. National Bureau of Economic Research No. w22381 Abrams BA, Settle RF (1999) Women’s suffrage and the growth of the welfare state. Public Choice 100(3–4):289–300 Acemoglu D, Robinson JA (2000) Why did the West extend the franchise? Democracy, inequality, and growth in historical perspective. Q J Econ 115(4):1167–1199

178

H.-P. C. Wong et al.

Aidt TS, Dallal B (2008) Female voting power: the contribution of women’s suffrage to the growth of social spending in Western Europe (1869–1960). Public Choice 134(3):391–417 Andrews KT, Seguin C (2015) Group threat and policy change: the spatial dynamics of prohibition politics, 1890–1919. Am J Sociol 121(2):475–510 Bertocchi G (2011) The enfranchisement of women and the welfare state. Eur Econ Rev 55(4): 535–553 Bloch L (1929) Facts about Mexican immigration before and since the quota restriction laws. J Am Stat Assoc 24(165):50–60 Braun S, Kvasnicka M (2013) Men, women, and the ballot: gender imbalances and suffrage extensions in the United States. Explor Econ Hist 50(3):405–426 Carter SB, Gartner SS, Haines MR, Olmstead AL, Sutch R, Wright G (2006) Historical Statistics of the United States: Millennial edition, vol 3. Cambridge University Press, Cambridge Cohen PN (1996) Nationalism and suffrage: gender struggle in nation-building America. Signs J Women Cult Soc 21(3):707–727 Conley JP, Temimi A (2001) Endogenous enfranchisement when groups’ preferences conflict. J Polit Econ 109(1):79–102 Doepke M, Tertilt M (2009) Women’s liberation: what’s in it for men? Q J Econ 124(4):1541–1591 Doepke M, Tertilt M, Voena A (2012) The economics and politics of women’s rights. Ann Rev Econ 4(1):339–372 Fernández R (2014) Womens rights and development. J Econ Growth 19(1):37–80 Geddes R, Lueck D (2002) The gains from self-ownership and the expansion of women’s rights. Am Econ Rev 92(4):1079–1092 Geddes R, Lueck D, Tennyson S (2012) Human capital accumulation and the expansion of women’s economic rights. J Law Econ 55(4):839–867 Goldin C (1994) The political economy of immigration restriction in the United States, 1890 to 1921. In: Goldin C, Libecap G (eds) The regulated economy: a historical approach to political economy. University of Chicago Press, Chicago, pp 223–258 Hatton TJ, Williamson JG (1998) The age of mass migration: causes and economic impact. Oxford University Press, New York Holcombe RG (2014) The economic theory of rights. J Inst Econ 10(3):471–491 Krogstrup S, Wälti S (2011) Women and budget deficits. Scand J Econ 113(3):712–728 Lemke JS (2016) Interjurisdictional competition and the married women’s property acts. Public Choice 166(3–4):291–313 Lott JR Jr, Kenny LW (1999) Did women’s suffrage change the size and scope of government? J Polit Econ 107(6):1163–1198 McCammon HJ, Campbell KE, Granberg EM, Mowery C (2001) How movements win: Gendered opportunity structures and US women’s suffrage movements, 1866 to 1919. Am Sociol Rev 66(1):49–70 Miller G (2008) Women’s suffrage, political responsiveness, and child survival in American history. Q J Econ 123(3):1287–1327 Ngai MM (1999) The architecture of race in American immigration law: a reexamination of the Immigration Act of 1924. J Am Hist 86(1):67–92 O’Neill WL (2013) The woman movement: feminism in the United States and England. Routledge, New York Przeworski A (2009) Conquered or granted? A history of suffrage extensions. Br J Polit Sci 39(2):291–321 Tabellini M (2018) Gifts of the immigrants, woes of the natives: Lessons from the age of mass migration. Technical report, Massachusetts Institute of Technology Teele DL (2018) How the west was won: competition, mobilization, and women’s enfranchisement in the United States. J Polit 80(2):442–461