Political Compromise and Bureaucratic Structure: The Political Origins of the Federal Reserve System

472 JLEO, V25 N2 Political Compromise and Bureaucratic Structure: The Political Origins of the Federal Reserve System Gyung-Ho Jeong* Claremont Graduate University Gary J. Miller Washington University in St. Louis Andrew C. Sobel Washington University in St. Louis What is the origin of the structural independence of the Federal Reserve System? Unlike existing explanations on central bank independence, we show that the structural independence of the Fed is not the result of intentional design but a product of compromise among disparate groups. Using agendaconstrained ideal point estimation techniques to estimate both the preferences of senators on key questions of Fed structure and the locations of alternative forms of the bill with respect to those preferences, we show that the structural features of the Fed in the final bill differed markedly from the original preferences of legislators representing competing groups. The result was a compromise that offered the prospect of significant independence for the new agency. The Fed case shows that political compromise can provide useful bureaucratic insulation when the short-term incentives of political principals promote unstable, selfseeking policy choices (JEL N41, N21). 1. Introduction A major divide separates the political control of bureaucracy literature and the central bank literature. The political control of bureaucracy literature, based on principal-agent theory, emphasizes the control and responsiveness of bureaucracies to political masters as a matter of democratic accountability and limiting bureaucratic inefficiency and mismanagement. Thus, some scholars raise concerns about the inefficiency resulting from political compromise that renders bureaucratic organizations independent of direct oversight or coordination (Moe 1989; Lewis 2003). Yet, the central bank literature highlights the importance of bureaucratic insulation for effective monetary policy, arguing *Email: gyung-ho.jeong@cgu.edu. The authors are grateful to the participants in Nate Jensen s class, the editor, and three anonymous reviewers for their helpful comments. The Journal of Law, Economics, & Organization, Vol. 25, No. 2, doi:10.1093/jleo/ewn010 Advance Access publication June 12, 2008 Ó The Author 2008. Published by Oxford University Press on behalf of Yale University. All rights reserved. For permissions, please email: journals.permissions@oxfordjournals.org

Political Compromise and Bureaucratic Structure 473 that central bank independence is an outcome of intentional design to increase credible commitment to stable monetary policy, to enhance informational transparency, or to represent dominant financial interests. In this literature, it is the lack of bureaucratic independence, the responsiveness of bureaucrats to politicians, which leads to bureaucratic inefficiency and mismanagement. Further, this body of research provides ample evidence that central banks insulated from political influence prove effective in controlling inflation and stabilizing the expectations of economic actors (Grilli et al. 1991; Cukierman 1992; Cukierman et al. 1992; Alesina and Summers 1993; Franzese 1999; Keefer and Stasavage 2003). In this article, we use the case of the creation of the US Federal Reserve System to gain leverage on this divide across literatures. In particular, we examine which of the two literatures provides a better explanation of the origins of the structural independence of the Fed. As a stronghold against inflationary policies in the United States and thus the world economy, the Fed is an important case for the central bank independence literature. Also, with its unusually insulated structure, 1 the insulation of the Fed calls for an explanation from the political control of bureaucracy literature. Our examination of the Fed s creation shows that, as the political control of bureaucracy literature argues, the structure of the Fed was heavily impacted by interest group, partisan, and institutional competition. This competition was resolved by compromise that rendered the organizational structure of the Fed independent of direct oversight or coordination. On the other hand, the existing explanations of central bank independence cannot explain the origin of the structural independence of the Fed. In the creation of the Fed, the concern for increasing the credibility of the commitment to stable monetary policies did not play a dominant role, neither did the concern for increasing informational transparency (Bernhard 1998, 2002) nor the dominance of financial interests (Posen 1995; Broz 1997). In fact, the celebrated structural independence of the Fed was a product of political compromise among disparate groups represented in the US Congress the very explanation the political control of bureaucracy literature anticipated. However, unlike the predictions of the political control of bureaucracy literature, this process did not lead to bureaucratic inefficiency and mismanagement. In the Fed s case, this insulation contributed to the government s credible commitment to policies that limited political micromanagement, which exacerbated investor risk and that of other economic actors. We emphasize this point because the political control of bureaucracy literature focuses on solving the moral hazard of the bureaucratic agencies, assuming that the moral hazard of the bureaucratic agencies is the only source of the problem. Advocates of what has been called the congressional dominance approach to bureaucracy argue that responsiveness is the most significant concern in the design of 1. The members of the Federal Reserve Board enjoy 14-year terms, second only to a 15-year term for the head of the Government Accounting Office. The Reserve Banks are owned privately. The Fed enjoys budgetary autonomy, making it free from congressional appropriations.

474 The Journal of Law, Economics, & Organization, V25 N2 bureaucracies. McCubbins and Schwartz (1984) state that whatever the original intent, it is no longer plausible in most cases to suppose that the public interest is best served by a bureaucracy unaccountable to Congress and, therefore, unaccountable to the electorate (p. 169). The fact that structural independence of the Fed was essential for its management of anti-inflationary monetary policies belies the key assumption underlying the political control of bureaucracy literature. In the next two sections, we review the literature on bureaucratic structure and the existing explanations of the origins of central bank independence. Then, we give background on central banking history in the United States leading up to the election of 1912, emphasizing the discrete and conflicting preferences as regards central banking. The rest of the article consists of a legislative history of the Federal Reserve Act of 1913, featuring a method for estimating the preferences of US Senators for two characteristics of the new institution centralization and independence. An application of the method to the Senate roll call data shows that different interest groups and the Progressive movement had significant and competing influences on the structure of the Fed and that the final bill was a product of political compromise and quite distinct from the original preferences of any of the relevant interest groups and political parties. By providing new evidence, our findings substantiate the shared view of many existing historical accounts of the Fed (Cushman 1941; Wiebe 1962; Clifford 1965; West 1974; Woolley 1984; White 1985; Kettl 1986; Greider 1987) that the Fed was as an outcome of political compromise. Our findings disprove the view that the Fed was an outcome of bankers dominance (Kolko 1963). This compromise, as predicted by the bureaucracy literature, offered the prospect of significant independence for the new agency, which eventually transformed into credible commitment. Yet, contrary to the bureaucracy literature and consistent with the central banking literature, history demonstrates that the resulting independence reduced bureaucratic inefficiency and mismanagement, helped stabilize monetary policy, and constrained the increasing bouts of financial distress that were plaguing the US economy. 2. Structural Politics and Bureaucratic Insulation Despite an expressed preference for the design of bureaucracies that are accountable and responsive to their political principals, much of the political control of bureaucracy literature has argued that bureaucratic structure is not so much the result of any grand administrative design, as it is the result of interest group, partisan, and institutional competition. Politicians and interest groups would both like to be in a commanding position that would allow them to design a bureaucracy that was infinitely responsive to their preferences. However, for the most part, politicians and interest groups find that their opponents have roughly as much blocking power as they do. Even if their political allies are in a position of control, they realize that, after the next election, their opponents could potentially have as much or more influence as they currently do. Political competition threatens the political survival of the currently

Political Compromise and Bureaucratic Structure 475 ascendant majority and its interest group allies; it therefore increases the incentives to compromise with political opponents and to create bureaucratic structures that guarantee the current majority party continued impact after losing power. The resulting insulated bureaucratic structure prevents the opposing party from immediately reversing course when it comes to power. Moe (1989) notes that The driving force of political uncertainty, then, causes the winning group to favor structural designs it would never favor on technical grounds alone... The group has to protect itself and its agency from the dangers of democracy, and it does so by imposing structures that appear strange and incongruous indeed when judged by almost any reasonable standards of what an effective organization ought to look like (p. 275). He also states, legislative victory of any consequence almost always requires compromise. This means that opposing groups will have a direct say in how the agency and its mandate are constructed (p. 275). This is complemented by institutional rules such as the Senate filibuster or separation of powers, which are institutions that encourage centrist and compromise alternatives. Unless a majority party is extremely large and encompassing in terms of the institutional divisions of governance, it must find a way to modify its proposal under such institutional constraints so as to get some support from the opposing party. Lewis (2003) has constructed a data set of the creation of 180 new agencies in the United States from 1946 to 1997. He was in particular interested in bureaucratic insulation, which he argued was the result of a series of structural variables including location outside a cabinet department, multimember commission, specific qualifications for appointees, fixed terms for commissioners, and independence (meaning an absence of bureaucratic layers above the organization s head). All these structural features (and more), for example, are present in the Federal Reserve. Lewis documents that expectations about presidential durability, indicators of electoral uncertainty in Congress, and the need for a compromise due to the size of the majority party all play a role in settling on a more or less insulated bureaucratic structure. Bureaucratic insulation is more likely to result from more need for compromise and more uncertainty. For both Moe and Lewis, the consequences of this compromise are largely negative. Moe interprets bureaucratic insulation as an indication of bureaucratic failure. Moe (1989) begins his article on bureaucratic structure with the claim that American public bureaucracy is not designed to be effective. The bureaucracy arises out of politics, and its design reflects the interests, strategies, and compromises of those who exercise political power (p. 267). Lewis regards the study of insulation as being crucial to understanding the pathologies of the modern administrative state (p. 7). Both feel that bureaucratic insulation prevents the president from being able to coordinate the performance of executive agencies in the broader national interest. We regard the argument of Moe and Lewis as a compelling explanation regarding the origins of bureaucratic structure and insulation, if not regarding its consequences. Unlike their views that disdain such independence, we note

476 The Journal of Law, Economics, & Organization, V25 N2 that the consequences of bureaucratic insulation may depend on the role of the bureaucratic agency in question. If the agency is responsible for determining key aspects of economic policy, such as monetary policy, then dominance and coordination by a president or any other political actor may be a problem for social welfare and insulation from political influence may in such a case be a solution. The politics of bureaucratic structure, involving compromise in the context of political uncertainty, may be the means by which self-interested politicians settle on a bureaucratic structure that ties their own hands. 3. The Existing Explanations of the Origins of Central Bank Independence The literature on central banking has by now established that short-term electoral interests of politicians tend toward inflationary surprises that detract from long-term investment and growth. Elected officials face temptations to enact stimulatory polices that result in inflation (Cukierman and Melzer 1986). Studies also demonstrate that central bank independence systematically helps control inflation (Alesina and Summers 1993). The literature on central banking tends to posit independence as an intended consequence of rational institutional design. Such explanations for central bank independence tend to fall into three categories: credible commitment, informational transparency, and the dominance of financial interests. First, policy makers create independence to signal a credible commitment to stable prices and monetary policy (Maxfield 1997). This line of reasoning argues that states seeking to address recurring financial distress and uncertainty, in need of international capital, wanting to play a larger role in the global political economy, or those experiencing reputational difficulties attempt to signal their creditworthiness or commitment to economic stability by creating an independent central bank. This category can include both domestic and international motivations. Maxfield suggests that the differing needs for credible commitment can explain variation of central bank independence across developing countries. However, the problem of this type of explanation is that the consequence of central bank independence is considered as the cause of central bank independence. For example, Moser (1999) argues that countries with multiple veto players tend to have independent central banks because multiplicity of veto players makes their commitment to stable economic policies more credible. In these explanations, independent central banks are intentionally chosen by politicians to credibly commit to creditworthiness. However, to show that central bank independence is effective for credible commitment is one thing. And to argue that political players intentionally design independent central banks to make a credible commitment to overcoming a time inconsistency dilemma is another matter. Central bank independence may well have such effects, but without carefully examining the process of the creation of independent central banks, we cannot conclude that such motives were decisive in the creation of central bank independence. Moreover, the need for credible commitment seems to explain the variation in those countries precisely because their regimes fit the formal models that

Political Compromise and Bureaucratic Structure 477 assume a single politician (Kyland and Prescott 1977; Barro and Gordon 1983; Lohmann 1992). Leaders of less democratic regimes may enjoy long enough time horizons to anticipate reaping some of the benefits of signaling creditworthiness with an independent central bank. Yet, the benefits of credible commitment are not necessarily a high concern for politicians in more democratic regimes where politicians face electoral uncertainty and shorter time horizons. Claiming credit for some future, yet unattained, long-term benefits generated by an independent central bank rarely translates into immediate electoral gains. Faced with electoral uncertainty and constraint on credit claiming, democratic politicians may well anticipate getting little or no benefit from forgoing the ability to manipulate monetary policy. The political agent, in other words, may suffer from moral hazard; she may actually prefer less economic growth and an assured reelection to better economic prospects and loss of power. In a second type of explanation, Bernhard (1998, 2002) focuses primarily upon domestic politics and argues that countries with multiple political actors create independent central banks because they alleviate potential conflicts among politicians conflicts between cabinet and parties in parliament, between the executive and backbenchers, and between coalition partners by providing transparency. According to Bernhard, the government creates an independent central bank to provide informational transparency, where backbenchers and coalition partners may blame the government for inflationary outcomes beyond the government s control and withdraw their support... (1998: 314 5). This places a focus on the political source of central bank independence rather than the economic source. Consequently, they link the variation in central bank independence to the variation in the amount and structure of political conflict. Yet, assuming that politicians create an independent bureaucracy to avoid the costs of controlling the policy-making authority being blamed for poor policy outcomes ignores that politicians may care as much if not more about the political benefits of controlling the policy-making authority claiming credit for good policy outcomes. Since there is evidence in the economic voting literature that voters punish incumbents for bad economic policy outcomes relatively independently of the locus of responsibility for those outcomes (Kramer 1971; Lewis-Beck 1988; Powell and Whitten 1993; Samuels 2004), we should not ignore the benefits of maintaining control over monetary policy. Also, this blame-shifting motivation overlooks interest groups and politicians needs for their electoral support. Can politicians ignore policy demands from interest groups, which, unlike the public, are attentive to details of bureaucratic structures and responsibilities? The third category of explanations focuses upon the dominance of financial interests in explaining the emergence of independent central banks (Maxfield 1994; Posen 1995; Broz 1997). From this perspective, the most important factor in explaining central bank independence is the dominance of financial interests. The problem we find with the financial interest dominance hypothesis is that it assumes that only the dominance and preferences of a specific interest

478 The Journal of Law, Economics, & Organization, V25 N2 group lead to an independent central bank. For example, Broz (1997) argues that it was the incentive and efforts of city bankers that produced the Fed. This begs the question, if such financial interests were so dominant, then why would they want bureaucratic insulation when they can already exercise control through their influence over politicians? As in the case of Broz (1997), this might account for having a central bank but not necessarily an independent central bank. This explanation ignores the fierce conflicts over important characteristics of the proposed central bank: who would control the bank, would it be public or privately dominated, how would it address the division between southern/western agricultural interests and money center elites, and should such a mechanism be insulated from political influence and manipulation? The political process of creating the Fed had to overcome a variety of interests, which dotted the political landscape with different views over the type and structure of a central bank. In the remaining sections, we show that the independence of the Fed was an outcome of political compromise, unlike the explanations of the existing central bank independence literature. Then, in the conclusion, we discuss how political compromise of competing principals can be a useful solution to the moral hazard problem on the part of political principals. 4. Before the Fed: Recurring Bouts of Illiquidity, Instability, and Financial Crisis Prior to the establishment of the Fed, banking in the United States was characterized by a lack of coordination among banks, absence of centralized oversight and clearing, and chronic banking crises. The Jeffersonian tradition in American politics feared that a concentration of economic power would undermine the basis of the republic, advancing the interests of a few over the many. In line with this tradition, Andrew Jackson s veto of the Second US Bank in 1837 was grounded in his and his supporters distrust of large city bankers and the power of capital. Jackson s Democrats feared the concentration of wealth and economic influence in the banks, arguing, banks have been the known enemies of our republican government from the beginning (Sharp 1970: 313). Without a central bank, however, banks reserves were scattered about in individual banks vaults and on deposit at correspondent banks. The economic incentives for local and regional banks were to deposit their reserves in the larger money center banks, creating an upward flow, a pyramiding, of banking reserves into the major New York banks (Broz 1997: 28). This practice made individual banks vulnerable to seasonal fluctuations in demand for credit, which proved especially problematic in communities where economic cycles and capital needs were tied to planting and harvest. When heavy seasonal demand appeared simultaneously across most agricultural areas, New York banks, the last bastion of lender of last resort credit, could not always provide enough liquidity to meet the demands of all the banks down the capital food chain.

Political Compromise and Bureaucratic Structure 479 As a consequence, increasingly frequent banking crises threatened the banks and economy during the late 19th century and the early 20th century. Following the civil war, the United States was the only major industrial power to experience intense seasonal financial fluctuations and crises on a regular basis: 1873, 1884, 1890, 1893, and 1907 (Borda 1986; Broz 1997). After the most severe of these crises, 1907, almost all parties private and public, urban and rural, bankers and nonbankers recognized a need to address the instabilities in the US banking system and its inability to reconcile money supply and credit with economic conditions. The 1907 crisis led to the congressional Pujo hearings, which focused on bankers power in the US economy or the Money Trust. The 8 months of hearings revealed that three major banks of the Money Trust Morgan, First National, and National City had acquired 341 directorships in 112 firms, which controlled over $22 billion in assets (Brandeis 1914). The public discovered that 18 institutions of the Money Trust enjoyed influence over two-thirds of US Gross National Product via their directorships. This fueled fears about concentrated economic and political influence in the hands of those who controlled the mechanisms of credit in the United States. Deliberations about reforming the existing banking system took place in the shadow of the 1907 financial panic, the Pujo hearings, and populist and progressive concerns. 5. Conflicting Interests over Reform Given the frequent banking crises that affected urban banks and rural credits, many social interest groups had a stake in the structure of the new agency. Among those, three groups played important roles: city bankers, farmers/ populists, and Progressives. 5.1 Large City Bankers New York bankers had the most global perspective on the creation of a central bank. Many of the major New York banking houses could see the benefits of a state agency with control over credit and money supply. With economic expansion and the increasing participation of the US economy in international exchange, the advantages of having an agency managing money supply and credit became increasingly appealing to those who worried about the growing frequency and severity of downturns and who viewed stability as essential to having the dollar accepted as an international currency. The Jewish houses and those with strong European connections were mostly in favor of a strong central bank....they realized that U.S. economic growth in the United States was always in danger of sharp downturns and depressions as long as the dollar was inelastic (Geisst 2004: 134). Despite wanting a central bank, New York bankers wanted a central bank relatively free of intervention by the federal government and one that would exercise monetary discipline. Much like the Whigs that railed against Jackson s Democrats, they wanted a central bank dominated by bankers and their expertise.

480 The Journal of Law, Economics, & Organization, V25 N2 5.2 Farmers and Populists Small country bankers, farmers, and populists feared a central bank dominated by the large city banks. They opposed the tight monetary discipline favored by the eastern bankers who saw such discipline as a necessary condition for elevating the dollar to international currency status. Thus, they preferred decentralized control of local funds (Broz 1997: 178). However, they also recognized that coordination up the banking hierarchy was essential to avoiding the chronic liquidity crises that plagued the system. Pure local control would not eliminate the problem of seasonal liquidity crises that burdened local interests. As a solution, the populists preferred federal regulation of the banking system because the federal government was the only center of power that could counter the influences of the power of Wall Street (Greider 1987). They sought to limit the influence of the large city banks by advocating a central bank that was controlled by their political representatives. As a Wisconsin country banker testified in a Senate hearing, they were more willing to take our [their] chances with the government than with big bankers (US Congress, Senate, Committee on Banking and Currency, Hearings on H.R. 7837, 63rd Congress, 1st session, 1913). 5.3 Progressives While city bankers and farmers had conflicting interests on the issues of central bank centralization and control, Progressives brought the issue of bureaucratic independence into the debate. Progressives like Theodore Roosevelt demanded a civil service system and politically independent regulatory commissions as the ideal form for new bureaucracies (Knott and Miller 1987). They viewed neutral bureaucracy as an alternative to governance by party politics, which during that period primarily meant machine politics. Thus, they preferred public, rather than private, control over the Fed. Neither the New York bankers nor the populists originally preferred an independent Fed. That is, they preferred, as their first choice, a Federal Reserve System that would be structured to be dependent on their own preferences, and they came to see the advantages of independence only instrumentally as a solution to stalemate and as protection against the exigencies of future political development. Progressives, however, had an ideological preference for a Fed independent of the influences of party politicians. Thus, they could play a special role as catalyst for a compromise that created a Fed that was neither as responsive to banks as the bankers wanted nor as responsive to public control as the populists wanted. 6. A Brief Legislative History of the Federal Reserve Act of 1913 After the 1907 panic, city bankers produced a central bank strategy called the Aldrich plan during the Taft administration. The Aldrich bill of 1911 proposed the creation of a National Reserve Association, which was to be wholly owned by subscribing banks. However, even under the Republican administration, city bankers had to compromise. To reduce the fears of farmers and populists that the Wall Street banks would dominate the central bank, the bill allowed

Political Compromise and Bureaucratic Structure 481 representation of local interests by providing for 15 regional districts. Local associations of banks would select the board of directors of each district. To reduce public suspicions that this organization would be a profit venture meant to enrich banks, the maximum return for subscribing banks was limited to 5%. The remaining surplus would feed the government treasury as a franchise tax. In other words, the provisions for decentralization and limited profit taking were geared to build a coalition behind the new central bank. Despite these efforts to build a central bank, the Money Trust investigation and democratic victories in the 1912 presidential and senate elections prevented the passage of the Aldrich bill, which was sponsored by a Republican senator who had a close connection with New York bankers. The populist wing of the Democratic Party, representing farmers and small businesses in the west and the south, demanded a democratic version of the reform. 6.1 The House Carter Glass (D), the Chairman of the House Banking and Currency Committee, presented a draft banking reform bill to president-elect Wilson in the winter of 1912. The draft was similar to the Aldrich bill, but with a little more decentralization. Glass did not include any central board (the Federal Reserve Board), and in an effort to reduce public fears of the dominance of large bankers, he increased the number of local reserve banks from 15 to 20. However, Western and Southern Democrats opposed the draft because they thought it gave too much power to the bankers. They demanded government control from top to bottom of the system. As president and party leader, President Wilson negotiated a compromise between bankers and farmers. On the issue of centralization, he supported a degree of centralization. He suggested installing the Federal Reserve Board at the top of the system as a capstone (Glass 1927). On the issue of control, Wilson originally took a position of allowing a minor representation of bankers on the Board, but he changed after populist Democrats expressed strong opposition to the idea of such private control, especially private control by bankers. Instead, he guaranteed bankers some participation in the operation of the central board by adding a Federal Advisory Council, which was to be solely composed by bankers for the purpose of policy recommendation. Thus, when introduced to the House, the Glass bill had been significantly modified in response to pressures from populist Democrats and by Wilson s mediation. Whereas the Aldrich bill proposed a central board dominated by bankers, the Glass bill had a central board composed solely of presidential appointees and government officials like the Secretary of Treasury. 6.2 The Senate When the Glass bill passed in the House with limited debate, the city bankers felt that Glass inadequately addressed their interests and they started a fusillade against the Glass bill when the Senate took up the bill (Glass 1927: 163 4). Particularly, they focused their criticism upon insufficient centralization. Frank A. Vanderlip of the National City Bank (later CitiBank) proposed

482 The Journal of Law, Economics, & Organization, V25 N2 a plan with more centralization. The Hitchcock substitute, proposed by Gilbert Hitchcock of Nebraska (D), was built on this proposal (Glass 1927: 166 7). The Hitchcock substitute proposed limiting the number of reserve banks to four and having the central board elect the majority of their boards. From the city bankers perspective, a lot more centralization with a small sacrifice of banker representation in the reserve bank boards trumped less centralization with more banker representation in reserve bank boards. Due to heavy lobbying by city bankers, the mood of the Senate Banking Committee turned critical of the Glass bill. Committee members criticized the Glass bill for insufficient centralization. Thus, the Senate Democratic Caucus adopted a substitute put forward by Robert Owen, the chairman of the Senate Banking Committee (Glass 1927: 205 7). The Owen substitute looked similar to the Glass bill except that it favored greater centralization by limiting the number of reserve banks to a minimum of eight and maximum of 12 where Glass specified a minimum of 12. The Owen substitute with some modifications passed the Senate with the support of all Democratic senators and six Republican senators. The Conference Committee ironed out the differences between the Senate bill and the House bill to recommend 12 reserve banks. The Conference report passed in both houses. In the House, most Democrats and 42 Republicans voted for the Conference report. All Democratic senators and four Republican senators voted for the report. In sum, the Glass bill and its two successive alternatives led to the final political compromise by moving the structure of the new central bank closer to what bankers and farmers could agree was a workable compromise. Thus, William Jennings Bryan, the leader of populist Democrats, said, the business interests will, I think, welcome this bill as an unalloyed blessing. It gives them, through their banks, a promise of relief in any time of stringency, and it gives this promise without putting in the hands of banks a power that might be used against the public (Glass 1927: 143). Bankers also responded positively to the passage of the Federal Reserve Act. Paul Warburg (1930), a New York banker with influential input in the design of the Aldrich bill, concluded that these compromises demonstrated progress of great significance for freedom from politics and governmental control (p. 127). In private correspondence before the passage of the bill, Frank Vanderlip of the National City Bank wrote, It is workable and I think bankers should recognize the difficulties of getting an ideal measure... I think the measure is vastly better than the present law under which we are operating (Frank Vanderlip to J. M. Smith, June 19, 1913, Frank A. Vanderlip Papers. Recited from Broz 1997: 201). 7. Roll Call Evidence for Interest Group Influences In this section, we analyze the roll call voting in the Senate on the proposed legislation establishing the Federal Reserve System. 2 Our purpose is to 2. Since there was very limited debate and votes in the House and district-level data are not available, we focus on the Senate roll calls.

Political Compromise and Bureaucratic Structure 483 Table 1. Roll Call Votes and Agenda Structure of the Federal Reserve Act of 1913 Control Central j ICPSR Issue Pass? Yea Nay Yea Nay Brief description of Yea 1 160 Control Pass OW 1 h 1 OW 2 OW 2 To table Hitchcock s amendment to the Owen bill (OW) that increases the tenure of the Board members 2 166 Central Pass OW 1 OW 1 OW 2 d 1 To table Hitchcock s amendment to OW that reduces the number of Reserve banks 3 167 Central Pass OW 1 OW 1 OW 2 d 2 To table Hitchcock s amendment to OW that increases the influence of the Board on Reserve banks 4 170 Control Pass h 2 OW 1 OW 2 OW 2 To amend OW by suspending civil service requirements for the Fed employees 5 178 Control Fail h 3 h 2 OW 2 OW 2 To amend OW by placing the Fed employees under civil service 6 179 Control Fail h 4 h 2 OW 2 OW 2 To amend OW by placing the Fed employees under civil service 7 180 Control Pass h 5 h 2 OW 2 OW 2 To amend OW by authorizing the president to place Fed employees under civil service 8 182 Both Fail h 1 h 5 d 3 OW 2 To pass the Hitchcock substitute 9 184 Both Pass h 5 OW 2 To pass the amended OW OW 1 and OW 2 denote the positions of the Owen bill in the control dimension and the centralization dimension, respectively. ICPSR, The Interuniversity Consortium for Polotical and Social Research. examine how different political forces affected legislators voting decisions and the voting outcome. 7.1 Roll Call Data In the Senate roll call votes, city bankers push for more centralization was reflected in several perfecting amendments and in one substitute. In the Senate, there were 10 votes regarding the structure of the Fed (see Table 1 for the detailed description of the votes). Some of them were related to the degree of centralization. Others were about the control over the Fed. Whereas the centralization issue was about the number of regional reserve banks and the authority of the Federal Reserve Board over the reserve banks, the control issue was mostly about the degree of presidential control over the

484 The Journal of Law, Economics, & Organization, V25 N2 new agency. The House version excluded bankers from the Federal Reserve Board. New York bankers realized that achieving both private control and centralization of the Fed was impossible. Thus, instead of trying to reverse the House outcome, city bankers focused on increasing the degree of centralization. The control issue was raised by Progressives in the form of independence from the president. Democratic senators, with a Democrat in the White House, were willing to give the president significant control over the new agency. They supported an amendment by Senator Owen (D, chairman of the Senate Banking Committee) that exempted the Fed employees from civil service requirements, thus giving the president a great deal of leverage in appointments to the Fed. Naturally, most Republicans wanted to ensure more independence from the president by seeking civil service protection for Fed employees and longer tenure for Board members. A statement by Senator Burton (R) during the Senate debate illustrates the importance of civil service requirements for Fed employees. He criticized the amendment by Senator Owen that exempted the Fed employees from civil service law requirement by saying that the authority given in both paragraphs (referring to the clause in the amended Owen bill that exempted the Fed employees from the civil service law requirements) is unprecedented. It takes away the control of Congress and gives to an executive body the determination as to the number of appointments they shall make, and also the determination of the salaries, without the usual rule that authority must first be given by Congress (Congressional Record, Vol. 51, No. 2, p. 1215). Thus, final debate in the Senate was about key aspects of control including the tenure structure of Federal Reserve Board members and civil service law requirements for Fed employees. Thus, we set the first dimension to be the degree of independence from the president (the control dimension). The second dimension is set to be the degree of centralization. 7.2 The Model and Hypotheses A typical model used to analyze the roll call votes is a logit or probit model. However, logit or probit models involve an inefficient use of the data because the model is separately fit for each of the roll call votes and, in some votes, inclusion of a party dummy variable creates a separation problem 3 due to party line voting in some votes. An alternative technique is to pool roll call votes and then fit a logit or probit model. However, when roll call votes represent different dimensions, pooling roll call votes is problematic. To overcome such problems, we adopt an ideal point estimation technique. Nevertheless, we do not assume ideal points are exogenous. Instead, we assume that constituency interests, interest group lobbying, and political parties affect the ideal point of a legislator. Thus, like a logit/probit model, we use 3. When a variable perfectly predicts binary outcomes, this leads to infinite coefficients and standard errors (Zorn 2005).

Political Compromise and Bureaucratic Structure 485 covariates to explain the variation of ideal points. This approach allows us to test the hypotheses on interest group/constituency influences as well as to estimate the ideal points of the legislators. 4 We start by modeling each legislator s ideal point or preferred position on the structure of the Fed as a function of each legislator s party membership, congressional committee membership, and the presence of various interest groups in her district (see the Appendix for details of the model and estimation). Since we hypothesized that three groups bankers, farmers, and Progressives affected the structural politics of the Fed, we model the ideal points of legislators using these variables. In the control dimension, we argued that Progressives played an important role in increasing independence of the Fed from the president. Thus, we include Theodore Roosevelt s vote share in the 1912 presidential election in each senator s state to measure the level of support for the Progressive movement in each state. Since Theodore Roosevelt was the major advocate of the Progressive movement, his vote share in the 1912 election represents how much support for the Progressive movement existed in each state. Thus, we expect the variable to have a positive effect on legislators support for more independence in the control dimension. The second variable in the control dimension is the membership in the Senate Banking and Currency Committee. We used this variable to capture the effects of city bankers lobbying. The members of the Senate Banking Committee were exposed to serious lobbying by bankers. According to Glass, after the House Bill excluded bankers from the Federal Reserve Board the bankers concentrated their lobbying efforts on the Senate Banking Committee. The bankers who had been unsuccessful in their effort to impress the House committee with the desirability of establishing a central bank seemed to take courage. They swept down on the Capitol in troops and produced the same arguments to a more sympathetic audience on the Senate side (Glass 1927: 183). Since the Senate Banking and Currency Committee was established in 1912, we doubt that the difference between the committee members and other senators can be attributed to the committee members accumulated expertise or any committee-specific factors. We expect this variable to have a positive effect on legislators support for more independence because 4. However, combining ideal point estimation with a probit/logit model brings in some of the problems of a probit or logit model. For instance, explaining the variation of ideal points using a couple of covariates can result in poorly predicted ideal points, which will cause low classification rates. Nevertheless, it does not necessarily mean that the model estimates are wrong, just as low R-squares cannot be the basis for rejecting a linear regression model. Further, including more covariates which will improve the classification rates, just like adding more variables will only increase R-squares in regression models was not practical because we do not have any theoretical guidance for what kind of additional covariates were needed and because most of other data are not available given the time period (year 1913) we are dealing with. Therefore, in the next section where we do a spatial analysis using ideal point estimates, we estimate ideal points without equation (1) in Appendix. This improved the overall classification rate from 53% to 94.7%.

486 The Journal of Law, Economics, & Organization, V25 N2 bankers wanted to minimize the interference of the government in the management of the Fed. The final variable in the control dimension is the Democratic Party. We use this dummy variable to capture the effects of partisanship. As many studies of delegation have documented (Epstein and O Halloran 1999; Huber and Shipan 2002; Lewis 2003), members of Congress in the president s party tend to allow presidential control over bureaucratic agencies. Thus, we expect this variable to have a negative effect on legislators support for more independence in the control dimension. In the centralization dimension, we argued that city bankers had a stake in creating a central bank because they expected to benefit from having the US dollar as an international currency. Thus, we include the Banking Committee membership to capture the effects of bankers lobbying. We expect this variable to have a positive effect on legislators support for more centralization. The second variable potentially impacting the centralization dimension is the number of farms in each state (Source: 1910 Census). We use this variable to capture the effect of farmers. Since farmers demanded local control of local funds, we expect this variable to have a negative effect on legislators support for more centralization. The final variable in the centralization dimension is a dummy variable for the Democratic Party. Since the Bryan wing of the Democratic Party has its base in the south and the west, we can expect that this populist wing had an influence on the party s position on the structure of the Fed. Thus, we expect this variable to have a negative effect on legislators support for more centralization. Note that this variable is correlated with the number of farmers because many Democrats were from agricultural states. Thus, inclusion of this variable will affect the performance of the number of farms. 7.3 Results The results of the model estimates are reported in Figure 1. The results in Figure 1 present the marginal posterior distributions and the highest density regions of the parameters of the model. The results are generally consistent with our interpretation and predictions. First, the Democratic Party induced legislators to support a more dependent Fed (see the left panel in the second row). As members of the president s party, Democrats were opposed to increasing the independence of the Fed from the president. Nevertheless, we should note that the Federal Reserve Act contained components, like budget autonomy and long tenure for board members, that made the Fed the most independent bureaucratic agency in the United States. That is, Democrats supported a Fed that was not dependent in an absolute sense but was dependent relative to the position of the Republicans. Indeed, the crucial Democratic spokesman for the bill, Carter Glass, promised during debate that the Fed would be an altruistic institution...a distinctly nonpartisan organization whose functions are to be wholly divorced from politics (Greider 1987: 281).

Political Compromise and Bureaucratic Structure 487 Intercept Intercept density density 0.0 0.5 1.0 1.5 density 0.0 1.0 2.0 density 0.0 0.2 0.4 0.0 0.4 0.8 density 0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Dependent < > Independent Democratic Party 3.0 2.5 2.0 1.5 Dependent < > Independent Banking Committee 0.0 0.5 1.0 Dependent < > Independent T. Rooservelt s Vote Share 0 2 4 6 Dependent < > Independent density 0.0 0.2 0.4 density 0.0 0.2 0.4 density 0.0 0.4 0.8 1.0 1.5 2.0 2.5 3.0 3.5 Local < > Central Democratic Party 10 8 6 4 2 Local < > Central Banking Committee 0 2 4 6 Local < > Central Number of Farms 3 2 1 0 1 Local < > Central Figure 1. Posterior Distributions and Highest Density Regions of the Parameters of the Senate Voting Model. Two Horizontal Bars under the Densities Indicate 95% and 80% Highest Density Regions of the Posterior Distributions, respectively. The Left Panels Are Coefficients in the Control Dimension, whereas the Right Panels Are Coefficients in the Centralization Dimension. Then, the question is why even Democrats agreed to the basic independence of the Fed, rather than giving total control over the agency to the president. This is where the Progressive movement came in. As seen in the coefficient estimate of Theodore Roosevelt s vote share in the 1912 presidential election, the effect of the Progressives in the control dimension is significant at the 95% level (see the left panel in the fourth row). Specifically, the popularity of the Progressive movement measured by Roosevelt s vote share had a positive effect on the legislators support for an independent Fed. This effect is significant even after we control for the effect of political parties. Thus, it supports

488 The Journal of Law, Economics, & Organization, V25 N2 our argument that Progressives played a unique role in making the Fed an independent agency. The support for a more independent Fed came also from bankers. Measured by the Banking Committee membership, the city bankers had a positive effect on legislators support for a more independent Fed. On average, the Banking Committee members had 0.5 point higher ideal points in the control dimension. Since most ideal points in the control dimension are in-between 2 and 3, this effect is substantial. Accordingly, in the control dimension, the Democratic Party s tendency to grant more control over the Fed to the president was offset by the influences of the Progressive movement and by city bankers lobbying. On the other hand, in the centralization dimension, the conflict was between bankers and their Republican allies who supported more centralization of the Fed and farmers or Democrats who supported less centralization of the Fed. First, as a party that gained support from the south and the west, the Democratic Party supported farmers interests by supporting a localized Fed. The posterior density of its coefficient is far from zero. That is, there is no doubt that Democrats supported less centralization of the Fed. The coefficient of the number of farms is also negative, although its 95% credible interval includes zero. It excludes zero only at the 80% level. This is due to its correlation with the Democratic Party. Thus, a proper interpretation of this result is that farmers had a negative influence on legislators support for more centralization of the Fed in part through the political party and in part independent of the political party. Finally, the city bankers had a positive effect on legislators support for more centralization of the Fed. Members of the Banking Committee, who were exposed to the city bankers lobbying, had 1.5 unit higher ideal points on average in the centralization dimension. Since most ideal points in the centralization dimension are between 2 and 3 in the space (see Figure 2), this 1.5- unit change is a substantial change. Also, note that the coefficient of the Banking Committee membership in the centralization dimension is much larger than the coefficient in the control dimension. This reflects that the issue of centralization was more salient to bankers than the issue of control. In sum, conflicting interests affected the creation of the Fed. Farmers and the Democratic Party pulled legislators to support less centralization of the Fed, whereas city bankers pushed legislators to support more centralization of the Fed. In the control dimension, the president s party pushed legislators to grant more control over the Fed to the president, whereas the Progressive movement and city bankers pulled the legislators to make the Fed independent from the president. In other words, the structural politics of the Fed s creation was characterized by the conflict of interests. 8. A Spatial Analysis of the Structure of the Fed The legislative history of the Fed is open to a variety of interpretations. Carter Glass, in his memoir of the Federal Reserve Act, proudly reports that the hostile forces of bankers were repeatedly beaten back by Wilson and his allies.