University of Southern California Dissertations and Theses

Bureaucratic politics and power building in the administrative state

(USC Thesis Other)

Bureaucratic politics and power building in the administrative state

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Bureaucratic Politics and Power Building in the Administrative State by A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (POLITICAL SCIENCE AND INTERNATIONAL RELATIONS) May 2023 Copyright 2023 Nicholas G. Napolio Acknowledgements I would like to express my gratitude to the following people who helped make this dissertation possible. First, I would like to thank my co-chairs, Christian Grose and Jeffery Jenkins, for their guidance and feedback throughout my time in graduate school. I am also grateful to the other members of my dissertation committee, Pamela J. Clouser McCann and Rachel Augustine Potter, for sharing their expertise and insights with me at critical times during the dissertation process. I would also like to thank Pablo Barber´ a and James Lo for their help in preparing me to undertake a project like this one. Finally, I would like to thank the countless individuals who have provided me with feedback and advice throughout the last five years at conferences and over coffee, drinks, or Zooms. ii Table of Contents Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix Chapter 1: Power Building in the Administrative State: Procedure, Persuasion, and Politics 1 1.1 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.2 Persuasion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.3 Politics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.4 Contribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 1.5 Outline of the Dissertation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Chapter 2: Procedure: Bureaucratic Control of Federal Spending . . . . . . . . . . . . . . 16 2.1 The Federal Grant Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.2 Expectations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.3 Data and Empirical Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.6 Postscript: A Pandemic Exception . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Chapter 3: Persuasion: Signaling to the White House . . . . . . . . . . . . . . . . . . . . . 46 3.1 Executive Coalition Building in the American System . . . . . . . . . . . . . . . . 49 3.2 Bureaucratic Strategy and Executive Coalitions . . . . . . . . . . . . . . . . . . . 52 3.3 Coalitions as Costly Signals and Insurance . . . . . . . . . . . . . . . . . . . . . . 54 3.4 Data and Empirical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 3.5 Empirical Test of the Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 3.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 Chapter 4: Politics: Exploiting Collective Action Problems in Congress . . . . . . . . . . . 81 4.1 Exploiting Collective Action Problems . . . . . . . . . . . . . . . . . . . . . . . . 84 4.2 Data and Empirical Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 iii 4.4 Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 A Appendix to Chapter 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 B Appendix to Chapter 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 iv List of Tables 1.1 Coalition Formation by Presidential Term . . . . . . . . . . . . . . . . . . . . . . 11 2.1 Agency Ideology, Politicization, and Presidential Particularism . . . . . . . . . . . 25 2.2 Substantive Effects, Agency–President Distance . . . . . . . . . . . . . . . . . . . 27 2.3 Placebo Test with Formula Grants . . . . . . . . . . . . . . . . . . . . . . . . . . 29 2.4 Typology of Twenty-First Century Pandemics . . . . . . . . . . . . . . . . . . . . 35 2.5 Effect of Copartisanship and District Urbanism on HHS Outlays . . . . . . . . . . 41 3.1 Most and Least Central Agencies . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 3.2 Cross-Tabulation of Probability of Coalition Formation . . . . . . . . . . . . . . . 69 3.3 Executive Coalition Building . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 3.4 OIRA Less Likely to Request Regulatory Changes from Coalitions . . . . . . . . . 74 3.5 Presidential Misalignment and OIRA Review of Coalitions . . . . . . . . . . . . . 75 4.1 Example of Regimes with Department of Labor Dyads in 112th Congress . . . . . 93 4.2 Proportion of Dyads Forming Coalitions by Regime and Presidential Term . . . . . 93 4.3 Coalition Building and Congressional Committee Gridlock. . . . . . . . . . . . . . 96 4.4 Coalition Building and Congressional Committee Gridlock. . . . . . . . . . . . . . 97 4.5 Proportion of Dyads Forming Coalitions by Regime and Presidential Term (Ma- jority Party Committee Medians) . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 4.6 Coalition Building and Congressional Committee Gridlock within the Majority Party. 99 A.1 Agencies in Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 A.2 Reanalysis with Alternative Clusters . . . . . . . . . . . . . . . . . . . . . . . . . 115 A.3 Reanalysis with High-Variance Programs . . . . . . . . . . . . . . . . . . . . . . 116 v A.4 Reanalysis Excluding Defense Agencies . . . . . . . . . . . . . . . . . . . . . . . 117 B.1 Agencies in Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 B.2 Centrality and Political Capital, 1998-2012 . . . . . . . . . . . . . . . . . . . . . 120 B.3 Reanalysis with Logit Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 B.4 Reanalysis with All Dyads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 B.5 Reanalysis at the Year Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 B.6 Reanalysis with Count Dependent Variable . . . . . . . . . . . . . . . . . . . . . 126 B.7 Reanalysis with Agencies with Overlapping Laws . . . . . . . . . . . . . . . . . . 127 B.8 Reanalysis with Alternative Operationalization of Presidential Misalignment . . . . 128 vi List of Figures 2.1 Particularism by Agency–President Distance and Politicization. Figure derived from model 4 in table 3.3. Left panel displays the marginal effect of presiden- tial co-partisan at all observed levels of agency–president distance and right panel displays the marginal effect of presidential co-partisan at all observed levels of agency politicization. The rug along the x-axis displays the densities of the respec- tive variables. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 2.2 Placebo Test: Particularism in Formula Grants. Figure derived from model 4 in table 2.3. The rug along the x-axis displays the density of each variable. . . . . . . 30 2.3 Federal Spending by V ote Share and Population by Pandemic. . . . . . . . . . . . 36 2.4 Comparing Effects of Copartisanship. Estimates derived from models controlling for dis- trict population, district median income, whether the member of Congress representing the district serves on the appropriations committee, whether they serve on the ways and means committee, whether they won their previous election with a margin of less than five points, state fixed effects, and—in the case of the models including all funding from 2007–2018—the president’s party. Error bars represent 95% credible intervals from the posterior distribution. . . . . . . . . . . . . . . 39 2.5 Department of Health and Human Services Spending. Estimates derived from models controlling for district population, district median income, whether the member of Congress rep- resenting the district serves on the appropriations committee, whether they serve on the ways and means committee, whether they won their previous election with a margin of less than five points, state fixed effects, and—in the case of the models including all and opioid funding—the president’s party. Error bars represent 95% credible intervals from the posterior distribution. . . . . . . . . . 43 3.1 Coalition Formation by Agency Type Combination, 1997-2012. Probability calcu- lated as the proportion of dyad-years of each combination that formed a coali- tion. Eighteen percent of observations are Cabinet-Cabinet, 51% are Cabinet- Independent, and 31% are Independent-Independent. . . . . . . . . . . . . . . . . 62 3.2 Probability of Coalition Formation by Number of Overlapping Laws. Points rep- resent dyads. Curves and ribbons estimated with bivariate logistic regression. . . . 64 vii 3.3 Marginal Effect of Agency Alignment. The rug on the x-axis displays the density of presidential misalignment. Marginal effects estimated from model 4 in table 3.3. The left y-axis plots the change in the predicted probability of coalition formation given a unit increase in agency alignment at the value of presidential misalignment on the x-axis, and the right y-axis displays the change in probability of coalition formation given a standard within-dyad change in agency alignment at the value of presidential misalignment on the x-axis. The white point indicates the marginal effect of agency alignment when presidential misalignment is one standard devia- tion above the mean and the horizontal line connects the point to the two y-axes to ease interpretation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 3.4 Relationship between Presidential Misalignment and OIRA Review of Coalitions. Ribbon represents 95% confidence interval of simulations. Rug along x-axis dis- plays the density of presidential misalignment. . . . . . . . . . . . . . . . . . . . . 77 4.1 Spatial Model with Two Committees . . . . . . . . . . . . . . . . . . . . . . . . . 85 4.2 Spatial Model with Three Committees . . . . . . . . . . . . . . . . . . . . . . . . 86 4.3 Three Oversight Regimes. Numbers in superscripts indicate which agency each committee oversees and letters in each subscript indicate each committee’s cham- ber. Brackets indicate ideological distance between each agency’s oversight com- mittees. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 4.4 Coalitions by Congress and Oversight Committees . . . . . . . . . . . . . . . . . 94 A.1 Testing Linearity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 A.2 Fixed Effects Adjustments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 A.3 Within-Agency Variation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 B.1 Delete-a-Group Jackknife Distribution . . . . . . . . . . . . . . . . . . . . . . . . 129 B.2 Delete-a-Group Jackknife Coefficients . . . . . . . . . . . . . . . . . . . . . . . . 130 B.3 Independent Variable Distributions . . . . . . . . . . . . . . . . . . . . . . . . . . 131 B.4 Within-Dyad Ranges of Independent Variables . . . . . . . . . . . . . . . . . . . . 132 viii Abstract Bureaucrats are political actors who pursue political power. They have preferences over policies, knowledge of the procedures legally allowed to them, and expertise on how to combine those pref- erences with that knowledge of procedures to get what they want -– even in the face of opposition from their overseers. However, bureaucrats derive no formal, legal authority independent of grants of authority from their overseers. Therefore, bureaucrats face a dilemma. This dissertation examines the tools bureaucrats use to manage this dilemma and build power even in the face of opposition from political overseers. Specifically, I ask how bureaucrats use different strategies in different situations to build power and manage this dilemma. I argue that agencies often push back against elected officials, are often successful at locking their preferred policies in place, and have been able to build power despite sometimes significant opposition from their overseers. The power-building tools bureaucrats use to push back against overseers, I argue, can be roughly divided into one of three types: procedure, persuasion, and politics. I make at least three major contributions. First, I expand the scope of inquiry into bureaucratic politics by conceiving of federal agencies as a constellation of entities able to form coalitions and work together. Second, I argue that bureaucrats use explicitly political strategies. Finally, I argue that bureaucrats often stymie efforts of their overseers to enforce overseers preferred policies, in contrast to many scholars of legislative politics who often view bureaucrats as simple agents of the legislature. ix Chapter 1 Power Building in the Administrative State: Procedure, Persuasion, and Politics Power is a prerequisite for all other goals among actors in the federal government. Since the U.S. Constitution disperses power across three coequal branches, interbranch struggles over power are recurrent motifs in American politics. Legislators and presidents cannot change policies without getting elected; judges cannot interpret, apply, and make the law without first being appointed to the bench through the legitimate, constitutional channels; bureaucrats cannot make rules without explicit legislative authorization and deference from the constitutional branches; and none can successfully enact enduring policies without the consent or at least passivity of the others. The explicitly constitutional bases of power for legislators, presidents, and judges confer legit- imate authority upon them almost automatically because elections and appointments – combined with centuries of norms surrounding the legitimacy of elections and of courts – leave few contest- ing the general legitimacy of the policymaking authority of Congress, the president, or courts. The federal bureaucracy, on the other hand, faces severe and persistent assaults on its legitimacy as a policymaker. Critics charge that “unelected bureaucrats” in a “Deep State” serve to undermine the democratic legitimacy of the constitutional branches and therefore that Congress, the presi- dent, or the courts ought to have and exercise authority to overturn bureaucratically made policies. 1 As a result, bureaucrats with any interest in changing policies must either convince the constitu- tional branches to defer to bureaucratic actors or stymie supervisory efforts to limit bureaucratic discretion. In a democracy, power emanates from the consent of the governed. Consequently, bureaucratic power in the United States must come through democratically and constitutionally sanctioned path- ways. While bureaucrats could wait passively to receive grants of power from Congress, the pres- ident, or the courts, the more proactive strategy is to take power from their overseers through legally or socially acceptable methods. And as power builders, bureaucrats prefer this more proac- tive strategy. Bureaucrats are political actors who pursue political power. They have preferences over poli- cies, knowledge of the procedures legally allowed to them, and expertise on how to combine those preferences with that knowledge of procedures to get what they want -– even in the face of op- position from their overseers. However, bureaucrats derive no formal, legal authority independent of grants of authority from their overseers. Therefore, bureaucrats face a dilemma. On the one hand, they want to satisfy or appease their overseers to avoid sanction and serious assaults on their power and authority, and on the other, they want to pursue their own preferences. This central tension in bureaucratic politics, between the desire to appease and the desire to pursue their own interests, plays out recurrently in American politics. It was on display during the battles between the Centers for Disease Control and President Trump in response to the coronavirus pandemic, during the Environmental Protection Agency’s battles with Congress over the Clean Air Act in the 1970s, during the legal assault on the authority of the Consumer Finance Protection Board, and on many more occasions. This dissertation examines the tools bureaucrats use to manage this dilemma and build as much power as possible even in the face of opposition from political overseers with divergent preferences. Specifically, I ask how bureaucrats use different strategies in different situations to build power and manage this dilemma. I argue that agencies often push back against elected officials, are often 2 successful at locking their preferred policies in place, and have been able to build power despite sometimes significant opposition from their overseers. The power-building tools bureaucrats use to push back against overseers, I argue, can be roughly divided into one of three types: procedure, persuasion, and politics. Procedure refers to legal avenues, often created by Congress, that give legitimate and formal power and authority to bureaucrats, such as formal rulemaking powers or jurisdiction over adjudication. Persuasion refers to how bureaucrats use information and reputation to induce changes in how overseers re- spond to bureaucratic policy choices and power building, for example through activating diverse networks of support for policies or controlling how much and what kind of information overseers receive. Finally, politics refers to how bureaucrats manipulate institutional rules in order to stymie overseers’ attempts to limit bureaucratic power, such as delaying choices to initiate policy changes until conditions are more opportune like leading up to or soon after an election. When exactly a particular strategy falling under one of these three types of power-building ac- tivities is the best way to achieve bureaucratic goals depends on what the particular task at hand is and who the particular overseers most responsible for evaluating bureaucratic policy choices are. For example, persuasion is a difficult strategy to use successfully when bureaucratic policy choices will be overseen by Congress, a collection of 535 legislators spread over two chambers, since bureaucrats would have to persuade a majority of legislators to allow a bureaucratic policy to stand. However, persuasion is a useful and appropriate tool when bureaucratic policy choices will be overseen by the White House, which is much more hierarchical and centrally organized with fewer individuals or offices that need to be persuaded. Likewise, procedure is a great tool when presenting policy changes to courts which defer greatly to choices made that follow legally sanc- tioned procedures, but may be less effective when the president is an audience since the president is likely more interested in the substance of a policy rather than what procedure was used to craft it. The remainder of this chapter discusses these three types of tools and provides examples, syn- thesizing disparate work on bureaucratic politics under this framework, and each of the following 3 chapters provides an original analysis of how bureaucrats use one of these three tools. Chapters 3 and 4 focus on one particular strategy – bureaucratic coalition building or interagency collabora- tion – which I argue is a persuasive tool vis-` a-vis the president as an overseer, and a political tool vis-` a-vis Congress as an overseer. Building coalitions when agencies have overlapping jurisdictions provides at least four poten- tial benefits to the constituent agencies’ pursuit of autonomy and policy goals. First, coalition building is a political tool that may induce overseers to underprovide oversight by creating collec- tive action problems among the congressional committees that oversee agencies (Gailmard 2009). Second, since coalition building involves transaction costs, it may serve as a persuasive tool, a costly signal to political overseers that the policy resulting from the coalition is particularly impor- tant, ripe, or well-supported by the public and therefore induce principals to let the rule stand as a matter of public policy or for electoral concerns. In other words, policymaking via coalitions may transmit credible information about the importance of, efficiency of, or public support for policy from a more informed agent to a less informed principal. Third, collaboration as a political tool forces overseers to distribute any sanction across multiple agencies thereby either diluting its effect on each individual agency or inducing the overseer to raise the severity of the sanction and incur a larger cost, both of which lower the probability that a sanction will have the deterrence effect desired by the principal. Last, coalition building may help agencies make a better or more efficient policy by combining resources and information (see, e.g., Austen-Smith and Banks 1996). From a scholarly perspective, studying coalition building in the executive is important as it moves beyond analyses of single agencies in an increasingly interconnected administrative state (Freeman and Rossi 2011, 2012) and it uncovers a tool bureaucrats have to push back against the authority of their principals. 4 1.1 Procedure The simplest type of power-building tools are those that have already been set up through legally sanctioned procedures. Congress regularly grants executive agencies legal authority to execute and interpret laws with a great deal of discretion, the President – through executive orders and other edicts – sets up procedures for agencies to legally change policies, and courts routinely create rules for agencies to legitimately create policies without interference from the elected branches or inferior courts. More than 99% of significant laws passed in the United States since 1947 delegate to at least one agency (McCann and Shipan 2022), thereby giving legal sanction for agencies to change poli- cies through the procedures outlined in the Administrative Procedure Act and other administrative law statues. And agencies use that discretion readily: the median number of rules promulgated pursuant to significant laws since 1947 is about thirty (Peterson and Napolio Forthcoming). The US Supreme Court is also extremely deferential to agency policy choices that interpret or apply congressional statutes, thereby allowing bureaucrats to move policy toward their ideal points with- out always having to reach into their political tool-belt, instead simply using procedures available to them from discrete grants of authority from Congress. Therefore, agencies can build power simply by following procedures set up for them by their overseers, like when the Equal Employ- ment Opportunity Commission ruled in 2015, through its legal powers to adjudicate, that sexual orientation and gender identity were protected classes under Title VII of the Civil Rights Act of 1964 despite congressional inactivity on expanding the scope of Title VII to include such bases. In addition to rulemaking and adjudication, a notable procedural tool available to bureaucrats is the allocation of federal grants. Allocating pork back home is an important part of a legislator’s reelection seeking and therefore members of Congress have a vested interest in seeing federal funds flow to their home districts. Despite this, Congress often creates grant schemes that afford bureaucrats significant discretion in how funds are eventually allocated. These program grants are consistently used to sent pork to politically friendly districts, but not always of the majority party in Congress; instead, to districts friendly to the president or bureaucrats (Berry, Burden and 5 Howell 2010; Bertelli and Grose 2011; Dynes and Huber 2015; Kriner and Reeves 2015). Like with rulemaking, bureaucrats do not have to reach into their political tool-belt to funnel money to districts represented by ideologically aligned members of Congress. They already have legal authority to do so by way of how Congress has structured grant programs. Congress, the president, and the courts, in other words, have provided some of the tools bureaucrats use to build power over time and bureaucrats readily use them to do just that. Chapter 2 provides evidence that agencies do use legally sanctioned procedures to build power. I show that agencies, using their legal authority over the distribution of federal grants, send more money to congressional districts that are friendly to those agencies’ goals. But simply using legally sanctioned procedures to build power only works when bureaucrats have those procedures readily available to them via explicit grants of procedural authority from Congress, the president, or the courts. When those procedures are not available to them, they must reach into their political tool- belts, using persuasion or politics to build power. 1.2 Persuasion Extracting power from the other branches through persuasion has been the subject of a vast liter- ature on bureaucratic politics spanning schools of thought and modes of inquiry from American Political Development and administrative law to public administration and more orthodox, con- temporary political science. Arguing that “[p]olitical legitimacy allows agencies to change minds” and that agency reputation facilitates such political legitimacy, Carpenter (2001, 15) showed that broad networks of support for bureaucratic policy goals are a necessary component of bureaucratic autonomy, or power, because broad networks make resistance to bureaucratic goals by the constitu- tional branches costly. The cultivation of autonomy-supporting networks by policy entrepreneurs in the Progressive Era over time facilitated an expansion in the power of the administrative state, not through force or violence but through a political, social, and cultural process of legitimiza- tion. Similarly, legal scholar (Schiller 2007, 406) argues that New Deal policymakers believed 6 that “[e]xpert administrators and a strong executive were best equipped to address a given societal problem and efficiently implement a solution” and ushered another expansion of administrative power by successfully convincing a majority of Supreme Court justices by 1940 that the adminis- trative state was indeed the proper organ to respond to societal problems and therefore deserving of deference, that is, more power. In both cases, and in both periods of growth in administrative power, agencies and executives persuaded the other branches to amplify bureaucratic power. Less developmentally inclined scholars have likewise argued that persuasion and legitimacy building are some of the predominant ways that the administrative state aggrandizes itself. Formal models of bureaucracy are often information or signaling games where agencies are assumed to have more or better information than their elected overseers and use that informational advantage to get what they want in either the short or long term. One interpretation of that body of work is that elected overseers respect the information and expertise embodied in bureaucratic agencies and are therefore willing to delegate with broad discretion, that is, empower bureaus even if it means giving up nominal power themselves. On this view, bureaucrats cultivate the expertise necessary to persuade with congressional blessing. Incentives align for Congress to grant discretion and job security (i.e., civil service protections) to bureaucrats and for bureaucrats to invest in expertise (Anderson and Potoski 2016; Gailmard and Patty 2007). Political scientists studying institutions often consider persuasion in the context of information asymmetries, where a party with more information attempts to influence the behavior of a party with less information by signaling the superiority of their understanding of the state of the world. Persuasion depends on two things, the quality (or perception of the quality) of the information, and the means by which that information is transmitted. Therefore, agents looking to persuade can follow one of two strategies: they can get better information, or they can develop better ways to transmit it. For example, conducting a cost-benefit analysis that uses agency resources and relies on rigorous analysis is a good tool to get better information, and providing the results of that cost- benefit analysis to Congress, rather than simply changing policy without explanation, is a good tool to convey that information convincingly. 7 Sometimes, agencies obfuscate information strategically. For example, bureaucrats share less analytical or useful information to Congress during periods of divided government, when the pref- erences of bureaucrats and of the majority party in Congress are at their most divergent (Ban, Park and You 2023). Likewise, agencies write harder to understand rule summaries when they expect opposition from overseers, effectively constraining how much information overseers can truly learn about the policy at hand (Potter 2019). Chapter 3 argues that one of the reasons bureaucrats form coalitions with each other is to persuade the White House to defer to their policies, allowing agencies to build power. Building policymaking coalitions is costly. It requires two or more agencies to agree to an exact wording of a rule or policy meaning each constituent agency of the coalition has to give concessions to the others in order to reach a compromise. Therefore, agencies should only form these policymaking coalitions when it truly improves the rule on some dimension. As such, successfully forming a coalition encodes information about the quality of the rule produced, signaling its quality to the President through the Office of Information and Regulatory Affairs. Persuasion, however, is not the only way that actors in the administrative state enhance their power. Good bureaucrats are also good politicians. And as good politicians, bureaucrats are famil- iar with the preferences of their overseers, lawmaking and judicial processes, and the substantive and procedural informational asymmetries inherent in the principal agent relationship. 1.3 Politics If neither procedure nor persuasion are likely to effectively build bureaucratic power, bureaucrats may turn to more Machiavellian politics. Rather than using legally sanctioned procedures or con- trol information streams, bureaucrats may manipulate or exploit institutional rules – like the ar- duous lawmaking process, Congress’ byzantine legislative organization, or biannual elections for Congress – to build power. For example, because passing legislation through Congress requires real or implied supermajorities in each chamber, status quo policies are very sticky. Agencies, like 8 political scientists and informed commentators, are aware of this and as first-movers can exploit the difficulty of passing legislation to make sticky policies that build their power. In her comprehensive book on “procedural politicking in the bureaucracy,” Potter (2019, 6) argues that agencies use procedures strategically “to insulate policies that are at risk of political interventions and ensure that bureaucrat-preferred policies endure.” Rather than persuading over- seers to defer to bureaucratic goals, Potter shows that, among other things, agencies slow-roll or fast-track rules depending on the configuration of congressional, presidential, and judicial prefer- ences in order to entrench bureaucratic policy goals and build bureaucratic power. In other words, agencies exploit the biannual turnover of majority coalitions in Congress and the longer timeline bureaucrats have in office to build power over time. Additionally, agencies recognize collective action problems in Congress that arise from su- permajoritarian lawmaking requirements and legislative organization, and can move policy toward their preferences in such a way that Congress cannot respond because of collective action problems (see, e.g., Boushey and McGrath 2020; Clinton, Lewis and Selin 2014; Gailmard 2009; Hammond and Knott 1996, 1999; MacDonald 2007; Shipan 2004). At the legislating stage, the required su- permajorities in Congress often cannot agree on a bill that would successfully change the status quo curb agency power; and at the oversight stage, having oversight committees in both the House and Senate can lead to freeriding and therefore an underprovision of oversight, meaning agencies can build power because Congress will not respond strongly enough to curb it. Chapter 4 argues that another reason bureaucrats form coalitions with each other is to induce collective action problems in Congress, making it difficult or impossible for Congress to success- fully overturn bureaucratic policies and allowing agencies to build power. Since each agency is overseen by two committees, one in the House and one in the Senate, collaboration brings in ad- ditional committees with a stake in bureaucratic policy choices, increasing the number of veto players able to kill bills during the lawmaking process and exacerbating the freeriding problem in the provision of congressional oversight after bureaucrats have made new policies. 9 1.4 Contribution This dissertation makes at least three major contributions. First, I expand the scope of inquiry into bureaucratic politics by conceiving of federal agencies as a constellation of entities able to form coalitions and work together, rather than the standard account of agencies working in isolation. Second, I argue that bureaucrats use explicitly political strategies, in contrast to a great deal of work that often conceives of bureaucrats as technocrats. Finally, I argue that bureaucrats often stymie efforts of their overseers to enforce overseers preferred policies, in contrast to many scholars of legislative politics who often view bureaucrats as simple agents of the legislature with minimal agency loss or as ruled by optimal procedural technologies that allow for effective control of the administrative state. 1.4.1 Bureaucratic Coalition Building and Power While existing research has illuminated the politics of bureaucracies in the United States, and more generally has helped us collectively understand how bureaucratic power is built and legiti- mated, it has been narrowly focused on individual agencies in isolation. Overlooked, however, is how bureaus can and have collaborated and built coalitions to empower the administrative state. This dissertation argues that executive coalition building is a strategy that agencies use to em- power themselves, and in so doing empower the administrative state. This section explains how executive coalitions build power in the administrative state through two of the aforementioned power-building mechanisms: persuasion and politics. Political scientists studying institutions often consider persuasion in the context of information asymmetries, where a party with more information attempts to influence the behavior of a party with less information by signaling the superiority of their understanding of the state of the world. Persuasion depends on two things, the quality (or perception of the quality) of the information, and the means by which that information is transmitted. Therefore, agents looking to persuade can follow one of two strategies: they can get better information, or they can develop better ways to 10 transmit it. Executive coalition building allows agencies to do both. By collaborating with other agencies, each agency can get more information than if they went at it alone, and can then transmit that information more convincingly by showing that multiple agencies have all come to the same conclusions, bolstering the credibility of that information. From intuitive pairings like the Departments of Defense and Veterans Affairs to perhaps less obvious pairings such as the Department of Labor and the National Aeronautics and Space Ad- ministration, executive agencies form hundreds of policymaking coalitions each presidential term. These coalitions are responsible for producing almost 3,000 rules between 1997 and 2016 rang- ing from financial regulation and the implementation of civil rights laws to responses to natural and environmental disasters. Table 1.1 displays the count and proportion of coalitions formed by presidential term from 1997 (Clinton’s second term) to 2016 (Obama’s second term). 1 The rate of coalition formation was highest during Clinton’s second and Bush’s first term, with about 36% of potential agency pairs forming coalitions. The rate of coalition formation then dropped to about 12% on average from Bush’s second to Obama’s second terms. Aggregating from 1997–2016, about 28% of potential agency pairs formed coalitions. Table 1.1: Coalition Formation by Presidential Term Presidential Term Coalitions Possible Proportion Formed Coalitions Coalitions Clinton II (1997–2000) 164 465 0.353 Bush I (2001–2004) 187 496 0.377 Bush II (2005–2008) 96 496 0.194 Obama I (2009–2012) 32 496 0.065 Obama II (2013–2016) 194 496 0.104 Aggregate 673 2,449 0.275 Note: Includes only presidential terms for which data from the first to last day of the term was available from the Federal Register’s API. There are fewer possible coalitions in Clinton’s second term because the Department of Homeland Security had not yet been created. Coalition formation, then, is commonplace in bureaucratic politics in the United States and yet has been understudied in political science. Work on networked governance has considered how 1 See Appendix for a description of the agencies included in the analysis. 11 agencies collaborate with each other and private entities (Freeman and Rossi 2011, 2012; McGuire 2006; Resh, Siddiki and McConnell 2014; Siddiki, Kim and Leach 2017), yet it often fails to consider the political environment in which agencies operate. Several studies do consider how overlapping jurisdictions affect bureaucratic policymaking, but they either focus on congressional incentives to concentrate or fragment authority (Bils 2019; Farhang and Yaver 2016; Peterson 2018; Ting 2003) or how overlapping jurisdictions might create inefficiencies like free-riding, turf wars, or preference cycling (Bils 2019; Hammond and Miller 1985; Herrera, Reuben and Ting 2017; Napolio and Peterson 2019; Ting 2003). Here, however, I argue bureaucrats take advan- tage of overlapping jurisdictions by building coalitions in order to build power. Agencies build coalitions actively to advance their goals in the face of political opposition. Much of the work on information sharing among federal agencies has been written by scholars in the public administration tradition, and often the theoretical or assumed purpose and effect of information sharing is increased capacity, better policy, or tighter relationships among agencies (Freeman and Rossi 2011, 2012; McGuire 2006; Resh, Siddiki and McConnell 2014). Missing from these accounts of information sharing or executive coalitions, however, is information shar- ing’s relationship to power in the administrative state, and specifically how information and per- suasion are tools to enhance bureaucratic power. A distinct goal of information sharing among agencies is to develop and communicate better information instrumentally to persuade overseers to defer to bureaucratic policy. In other words, an important goal of information sharing is to gain power. Instead of information as a tool to craft better policy for the sake of better policy, informa- tion is a tool to persuade political actors with the authority to effectively veto bureaucratic policy to instead defer to it. The institutional structure of the bureaucracy vis-` a-vis the White House and Congress affects the likelihood that persuasion via information successfully enhances bureaucratic power. Persuad- ing Congress, as a collection of 535 individual members, is a tall order. However, almost all bureaucratic policy is centrally overseen by the Office of Information and Regulatory Affairs, an 12 agency within the Executive Office of the President. Therefore, it is much less costly to at-tempt to persuade the White House as there is essentially only one liaison or point of access to convince. Congress is a collection of 535 individual members spread over two chambers who must form supermajority coalitions in order to advance any legislation to change the status quo. As such, “persuading Congress” is a misnomer. Congress, after all, is a they. Anyone wishing to “persuade Congress” must persuade hundreds of individuals to change the status quo, a very costly endeavor. Of course it is not impossible – if it were lobbyists would not spend billions of dollars of attempting to persuade legislators to vote a certain way or introduce specific legislation. However, it is very difficulty and politics, rather than persuasion, is likely the better technique to make changes to the status quo via Congress. On the other hand, the President – or more accurately the office of the presidency – is a rela- tively centrally organized hierarchy with the President at the top and levels of White House bureau- cracy nestled beneath the chief executive. Formal points of entry into the White House are much more centralized than those into Congress, especially for bureaucrats in the rest of the Executive Branch whose first formal submissions of policy change to the White House almost exclusively run through the Office of Information and Regulatory Affairs. As such, persuasion is a much more attractive tool when attempting to make changes to the status quo via the presidency compared to Congress because all attempts at persuasion can be directed at one office, rather than a majority of legislators offices. Building coalitions, then, serves different purposes when targeted toward the President or Congress. Coalitions aim to persuade the President that a policy is good and therefore should be upheld, even if the President may have chosen a different policy themselves. Coalitions are political when targeted toward Congress, exploiting or creating collective action problems in the national legislature. The effect of both purposes, however, is the same: keeping bureaucratic poli- cies in place even though elected officials would have selected a different policy if the bureaucracy did not exist. 13 1.4.2 Good Bureaucrats are Good Politicians Until recently, the orthodox view in political science has been that Congress has installed procedu- ral technologies that provide a good degree of control over the bureaucracy (McCubbins, Noll and Weingast 1987). Administrative procedures created by Congress in statutes like the Administrative Procedure Act of 1946, so the argument goes, install various “fire alarms” into the bureaucratic pol- icymaking process that allow aggrieved parties to alert Congress to any bureaucratic malfeasance (McCubbins and Schwartz 1984). These fire alarms cut down on monitoring costs for busy mem- bers of Congress by putting the responsibility for raising concern over bureaucratic activity on the regulated, but nonetheless results in efficient control of bureaucratic agencies. More recently, scholars of bureaucratic politics have argued that these forms of legislative control are not as effective as was commonly assumed, and have catalogued the various ways in which agencies circumvent top-down oversight. Fire alarms may still be pulled, for example, but agencies may proactively consult with potential fire alarm pullers in order to stave off legislative oversight (Potter 2019). This dissertation joins that line of work by arguing that good bureaucrats are good politicians and find clever ways to work around institutional rules that constrain them. 1.5 Outline of the Dissertation What follows is dedicated to providing novel analyses of each of the three types of power-building tools bureaucrats use. For each chapter, I collected and analyzed novel data related to bureau- cratic decisionmaking and policymaking. Chapters 3 and 4 focus on a particular tool that works as both persuasion and politics depending on the overseer: bureaucratic coalition building, or the phenomenon that agencies often collaborate in the policymaking process. Those chapters rely on a newly constructed network of bureaucratic coalitions and complicate the standard view of bureau- cratic politics that considers agencies as individual entities in isolation, rather than a constellation of entities that often cross jurisdictional boundaries to work together. 14 Chapter 2 focuses on a procedural tool available to bureaucrats to build power: the allocation of federal grants. I show that constellation of bureaucratic agencies responsible for allocating grants plays a key role in facilitating or frustrating presidential policy priorities. Using a dataset of 21 agencies over 14 years, I find that only agencies ideologically proximate to the president engage in particularism benefiting the president. I find no evidence that politicization influences agency implementation of particularism. Critically, the moderating effect of the bureaucracy on particularism only occurs for distributive programs over which agencies have discretion. When disbursing formula grants written by Congress but administered by the bureaucracy with little or no discretion, ideological distance between agencies and presidents has no effect on particularism. Chapter 3 focuses on how bureaucratic coalition building serves a persuasive tool enabling bureaucrats to build power vis-` a-vis the president. I present a theory that agencies form coalitions to optimize their autonomy given their subordinate position in a separation of powers system by signaling to overseers that their policies are efficient and should be maintained. Bureaucrats form coalitions actively to advance their policy goals in the face of political opposition. Using data on dozens of agencies over seventeen years, I find that agencies are most likely to form coalitions when their preferences are misaligned with the president but aligned with each other. I also find evidence that coalitions send credible signals that bureaucratic policies are efficient since OIRA is less likely to request regulatory revisions of policies produced by coalitions. Chapter 4 focuses on how bureaucratic coalition building serves as a political tool enabling bureaucrats to build power vis-` a-vis Congress. I present a theory arguing that agencies form coali- tions to optimize their autoomy given their subordinate position in a separation of powers system by exploiting and inducing collective action problems in Congress. Using data on dozens of agen- cies over seventeen years, I find that agencies are most likely to form coalitions when it helps them induce collective action problems among their overseers in Congress: namely, committee freerid- ing in oversight and grid-lock in lawmaking. Agencies form coalitions actively in order to insulate their policies against congressional oversight. 15 Chapter 2 Procedure: Bureaucratic Control of Federal Spending The power of the presidency is fundamentally dependent upon the actions of bureaucrats. Argu- ing that presidents direct distributive benefits to preferred constituencies like legislators do, the particularist view of the presidency finds substantial support in the literature (Berry, Burden and Howell 2010; Dynes and Huber 2015; Hudak 2014; Kriner and Reeves 2015; Lowande, Jenkins and Clarke 2018). However, extant studies largely overlook the implementation of particularistic policy by providing descriptive accounts of how the bureaucracy might moderate particularism in theory but nonetheless ignore heterogeneous implementation in empirical analyses either by ag- gregating spending across all agencies or focusing on a outlays from a single agency. Understand- ing the role of the bureaucracy in implementing presidential policy is integral to characterizing the modern administrative presidency. Regardless of presidential preferences, the effectiveness of presidential policy relies on compliant bureaucrats. The power of the presidency is, after all, the power to persuade (Neustadt 1991). Much of the work on particularism studies the distribution of federal grants and argues that presidents target outlays to co-partisans and other key constituencies. Grants, however, are dis- bursed by disparate agencies with varying institutional designs, preferences, and incentives (Ander- son and Potoski 2016; Arel-Bundock, Atkinson and Potter 2015; Berry and Gersen 2017; Bertelli and Grose 2009; Dahlstr¨ om, Fazekas and Lewis 2019). According to the Federal Assistance Award Data System (FAADS), about two hundred unique agencies have disbursed grants since 2000. 16 Agencies as different as the US Agency for International Development, Nuclear Regulatory Com- mission, and Department of Education disburse billions of dollars in grants every year with unique processes for drafting calls for applications and adjudicating between potential recipients devel- oped in-house by each agency. This chapter takes seriously the diversity of executive agencies responsible for disbursing grants and provides a nuanced account of a posited mechanism of pres- idential particularism: presidential control of the bureaucracy. I show that the constellation of agencies responsible for grantmaking varies in its implementa- tion of the president’s particularistic preference to target outlays to the president’s preferred con- stituencies. Specifically, this chapter interrogates whether ideological alignment with the president and agency politicization condition bureaus’ implementation of particularism benefiting the pres- ident. Using data from twenty-one agencies over fourteen years, I test whether agencies vary in their implementation of particularism, and specifically whether ideology and politicization account for such variation. I find that agencies ideologically proximate to the president engage in particularism while agen- cies ideologically distant from the president do not. I find no evidence that politicization influences agency implementation of presidential particularism. My findings suggest a more nuanced treat- ment of presidential particularism that takes seriously the implementation apparatus leveraged by the president to pursue their particularistic goals. Presidents do not have unilateral control over the disbursement of federal funds, even when Congress grants the Executive Branch discretion to allocate money. Instead, the individual agencies responsible for disbursing federal grants constrain the power of the president to target funds to key constituencies. In the following pages, I describe the process by which federal grants are administered, noting the various stages during the allocation process where agencies have broad discretion to disburse grants consistent with their preferences. Then, I test whether bureaucratic agencies implement the president’s particularistic goals uniformly and show that only agencies ideologically aligned with the president engage in particularism on the president’s behalf. Next, I conduct a placebo test on formula grants over which agencies have little, if any, discretion and show that the bureaucracy 17 does not influence the distribution of formula grants the same way it influences discretionary ones. I conclude with a discussion of the implications of my findings for the theory of presidential par- ticularism, the role of the bureaucracy in the American separation of powers system, and the power of the presidency. 2.1 The Federal Grant Process Federal grants come in many forms, but can be lumped into two broad categories: formula and pro- gram grants (sometimes called mandatory and discretionary grants, respectively). Formula grants are administered by federal agencies, but the recipients and amounts are determined by Congress. Examples of formula grants include: Medicaid, Temporary Assistance for Needy Families (TANF), and Special Needs Education grants. Program grants, on the other hand, are those for which the federal agencies administering them have discretion over both the recipient and amount based on an often competitive application process developed in-house by the agencies themselves (Chernick 2014; Kincaid 2008). Examples of program grants include: National Science Foundation Gradu- ate Research Fellowship Program grants, Department of Education State Personnel Development grants, and Department of Transportation Transportation Investment Generating Economic Recov- ery grants. Therefore, program grants represent opportunities for discretionary and distributive choices by bureaucrats, whereas formula grants represent the implementation of circumscribed distributive authority from Congress. The life cycle of a program or discretionary grant consists of four stages: pre-award, award, administration, and post-award. During the pre-award stage, federal agencies develop and publish criteria for evaluating applications while applicants prepare and submit applications. Agencies then evaluate the applications they receive, occasionally using panels of experts to help make deci- sions. During the award stage, the terms of the grant are established. Then, the agency administers the grant during the administration stage. Finally, during the post-award stage, the agency must comply with reporting requirements to oversight committees, oversight agencies like the Office 18 of Management and Budget or Government Accountability Office, and to databases like the Trea- sury’s online database hosted at usaspending.gov. Once the program has been administered in full, it is closed out and may be audited. (Keegan 2012) The pre-award stage, where agencies adjudicate among potential recipients, is particularly im- portant. For program grants, no entity but the agency distributing the grant formally makes the decision about where to allocate the outlay. Each agency establishes criteria for applicants and then selects both which applicant will receive the grant from the pool of contenders and how much the grant will be worth. Thus, agencies have discretion over how they will allocate program grants in (1) the development of criteria—which may target certain potential applicants—(2) the selection of recipients—which may favor certain constituencies—and (3) the determination of the value of the grant—which again may privilege certain partisans. The pre-award stage, then, offers multi- ple opportunities for agencies to make political and partisan decisions over the allocation of grant outlays: first in the development of criteria, second in the selection of recipients, and last in the valuation of the grant. Isolating program grants represents a break from extant studies of particularism in grantmak- ing which tend to pool all grants and then remove programs that do not meet an arbitrary level of within-program variation in funding over time and space. 1 More importantly, program grants are the appropriate venue for testing whether agencies vary in their implementation of presidential par- ticularism since bureaucrats enjoy broad discretion when deciding how to allocate program grants. Confining the analyses to program or discretionary grants also guards against drawing improper conclusions if, for example, an agency seems to follow its principal’s orders if grants are pooled simply because formula grants have been designed such that they benefit certain constituencies independent of the agencies responsible for disbursing them. 1 The analyses to come are robust to only including grants disbursed pursuant to high-variation programs, as in other work (see appendix “Reanalysis with High-Variance Programs”). This subsetting of the data, however, conflates high-variance formula grants, over which agencies have little discretion, with program grants over which agencies have discretion, and excludes low-variance program grants, leading both to false inclusions and exclusions, so it is suboptimal. 19 2.2 Expectations In the following empirical analyses, I test two expectations. First, ideological distance between agencies and presidents should decrease particularistic outlays from an agency. Second, agency politicization should increase particularistic outlays from an agency. Ideological disagreement with the president should condition an agency’s willingness to engage in particularism. If agencies are aligned ideologically with the president, those agencies likely want the president and their party to stay in power and to continue implementing policies consistent with their ideology. Ultimately, presidents’ aims in directing funds to co-partisans are to secure their support, win reelection, strengthen the party brand, and ensure that their party can gain or stay in power in order to implement its platform (Dynes and Huber 2015; Kriner and Reeves 2015). As such, one function of particularism is to cultivate the political capital necessary to enact ideological policy. Although bureaucrats face different incentives than presidents—influence from competing principals (Bertelli and Grose 2009; Gailmard 2009; Potter 2019), good government or efficiency (Gailmard and Patty 2007; Miller and Whitford 2016), ideology (Clinton et al. 2012; Potter 2019), and enlarging their budgets (Niskanen 1971)—agencies nonetheless care about pol- icy outcomes, particularly in their regulatory spheres, so helping the president direct funds to key constituencies may be mutually beneficial by aiding the reelection campaign of the president and that of their co-partisans. Additionally, since bureaucrats generally desire larger budgets and presidents typically request more in appropriations to agencies with whom they are ideologically aligned (Bolton 2020), agencies ideologically aligned with the president stand to gain if the in- cumbent stays in office and can help them do so by directing funds to districts important to their reelection. Therefore, agencies ideologically aligned with presidents should allocate more funding to pres- idential co-partisans by implementing the president’s particularist agenda. The logic for misaligned agencies is similar. Agencies distant from the president likely are not as invested in the incumbent president’s and their party’s success in future elections and therefore may not be willing to di- rect outlays to key constituencies and instead substitute in their own best judgment or standard 20 operating procedures despite the president’s preference for particularism. Additionally, agencies ideologically distant from the incumbent president stand to gain from their replacement with a friendlier chief executive who may seek more appropriations for those agencies (Bolton 2020). Politicization—the extent to which an agency is filled with political appointees rather than careerists—may also condition an agency’s willingness to engage in particularism. How politiciza- tion may affect the implementation of particularism is less clear however. Some studies suggest that politicization does result in agency responsiveness to the president since the president is able to stack agencies with loyalists (Berry and Gersen 2017; Lowande 2019), while others suggest the opposite since politicization often reduces bureaucratic capacity so much so that agencies cannot implement their principal’s policies even if they want to (Huber 2007; Huber and McCarty 2004; Kennedy 2015; Lewis 2010). The most straightforward expectation is that by inserting presidential loyalists into agencies, those agencies will be more likely to implement the president’s particularist agenda, but reasonable theories predict opposite results. 2.3 Data and Empirical Strategy To test whether the bureaucracy moderates presidential particularism, I compiled an original dataset containing information on discretionary grant outlays from each agency to each congressional dis- trict in each Congress from the 106th (1999–2000) to 112th (2011–2012) Congress. I collected data on grants from usaspending.gov which compiles FAADS data at the level of the outlay from each federal agency and reports, among other things, its amount, its recipient’s congres- sional district, and whether it was disbursed pursuant to a program or congressional formula. 2 The unit of analysis is the agency-congressional district-Congress and each observation represents how much in outlays (in 2018 dollars) each congressional district received from each agency in each Congress. I removed all grants that were allocated based on statutory formulas to isolate only those grants that were subject to agency discretion. 2 Since I am interested in the implementation process of appropriations, spending data are appropriate (Hammond and Rosenstiel 2020). 21 Formula grants, however, offer a nice placebo test. If the results are driven by secular trends in grant disbursement, then we should observe the same results from estimating the same models on formula grants, over which agencies have little, if any, discretion. But if the bureaucracy moderates presidential particularism in the allocation of grants, there should only be a relationship between particularism and agency ideological misalignment with the president or politicization for program grants. The dataset comprises twenty-one agencies and over $3.9 trillion in outlays. I aggregate outlays to agencies’ highest organizational level: the department or independent agency (Berry and Gersen 2017). The aggregation process resulted in a dataset of 63,075 observations. All agencies but the Department of Homeland Security were in operation throughout the entire period, each of which forms a pair with each of the 435 members of Congress (MC) in each of the seven Congresses (20× 435× 7= 60,900), then the Department of Homeland Security forms a pair with each of the 435 MCs for the five Congresses it was in operation (435 × 5 = 2,175), leading to a final dataset of 60,900+ 2,175= 63,075 observations at the agency-district-Congress level. The dependent variable is the logged outlays (+1) in 2018 dollars from each agency to each congressional district in each Congress. The first independent variable of interest is presidential co- partisan which takes the value of one if the MC representing the congressional district receiving the outlay is from the same party as the president in a given Congress, and zero otherwise. To construct the second independent variable of interest, agency–president distance, I take the absolute value of the difference between each agency’s Chen and Johnson (2015) campaign finance based ideal point estimate and the president’s DW-NOMINATE ideal point estimate in each Congress, which are measured on the same scale. 3 The third independent variable of interest is agency politicization, which I measure as the ratio of political appointees to the number of career senior executive serivce 3 Using the Chen and Johnson (2015) measure of agency ideal points is optimal since they are dynamic, varying with each presidential term. Alternative measures of agency ideal points are both static, thus obscuring the dynamic nature of bureaucratic ideology induced by changing appointees and civil servants, and not estimated on the same scale as other political actors, so calculating the spatial distance between bureaucratic and political ideal points requires the imposition of untestable assumptions to map political ideal points into bureaucratic ones (e.g., Richardson, Clinton and Lewis 2018). 22 members following previous work (see, e.g., Lewis 2010; Lowande 2019; Wood and Lewis 2017). I also include a host of control variables at both the agency- and legislator-levels. 4 With these data, I then estimate a series of generalized least squares regression models to assess agency implementation of presidential particularism. I estimate variants of the following general model: ln(Outlays) i jt =β 0 +β 1 Presidential Co-Partisan jt +β 2 Agency–President Distance it + β 3 Politicization it +β 4 Presidential Co-Partisan jt × Agency–President Distance it + β 5 Presidential Co-Partisan jt × Politicization it +ξ ξ ξ X i jt +α α α i j +δ δ δ t +ε i jt (2.1) where subscript i indexes agencies, subscript j indexes members of Congress, subscript t indexes Congresses, X is a matrix of covariates,ξ ξ ξ is a vector of coefficients attending the covariates, α α α is a vector of agency-legislator pair fixed effects, and δ δ δ is a vector of Congress fixed effects. I center each of the continuous independent variables to zero as its mean and standardize them by the standard deviation of their residualized values after absorbing variation from agency-legislator and Congress fixed effects (Mummolo and Peterson 2018), thus coefficients should be interpreted as the change in logged outlays given a standard, within-agency increase in the independent variables. Formally, if agencies ideologically distant from the president implement particularistic policy less vigorously than those ideologically proximate, we would expect β 2 +β 4 < 0 (i.e., increasing ideological distance decreases outlays to presidential co-partisans) and β 2 > 0 (i.e., increasing ideological distance increases outlay to presidential contra-partisans), which implies β 4 < 0. If more politicized agencies implement particularistic policy less vigorously than non-politicized one, we would expect β 3 +β 5 > 0 (i.e., increasing politicization increases outlays to presidential co- partisans) andβ 3 < 0 (i.e., increasing politicization decreases outlays to presidential co-partisans), which impliesβ 5 > 0. 4 Specifically, I include whether each member of Congress in each Congress is in the majority party, sits on the appropriations committee, sits on the ways and means committee, won their previous election with a margin less than 0.05, each district’s logged population, and logged median income, and the distance between each agency’s Chen and Johnson (2015) ideal point estimate and each member’s DW-NOMINATE ideal point estimate. 23 The agency-legislator fixed effects adjust for any time-invariant characteristics that may affect the distribution of federal grants including, but not limited to, agency structure and independence (Selin 2015), 5 workforce skill (Richardson, Clinton and Lewis 2018), legislator district, legislator gender (Anzia and Berry 2011), and the unobservable aspects of the relationship between each leg- islator and each agency. The Congress fixed effects adjust for any common shocks experienced by all agencies such as the president’s party (Reingewertz and Baskaran 2019), the majority party in the House of Representatives, and national economic health. This dual fixed effects design allows for identification from within-agency-legislator variation in the key independent variables. Identi- fication comes from the same representative changing from presidential co-partisan to presidential contra-partisan or vice versa, and changes in the ideological distance between the same agency and the president over time. 2.4 Results Table 3.3 reports the parameters estimated from the model in equation 2.1. 6 First, I report the aggregate level of particularism in the sample by aggregating the data to the legislator-Congress level. The coefficients on presidential co-partisan in models 1 and 2 indicate that, on average, congressional districts represented by legislators who share the president’s party receive more in outlays. This estimate is consistent with prior research on grants and particularism and indicates that my sample of agencies, on average, displays particularism (Berry, Burden and Howell 2010; Dynes and Huber 2015; Hudak 2014; Kriner and Reeves 2015). Models 3 and 4 disaggregate the data to the agency-district-Congress level and show that par- ticularism is primarily implemented by agencies ideologically proximate to the president. Figure 5 Agency independence is conceptually related, yet distinct, from politicization. My empirical strategy leveraging within-agency variation does not allow me to test how independence affects particularism since independence is time- invariant. However, in appendix “Reanalysis by Agency Structure Subsets”, I separate out cabinet departments from independent agencies and find similar results for both kinds of agencies. 6 Since estimated coefficients on “control variables generally have no structural interpretation themselves,” I de- cline to include them in tables (H¨ unermund and Louw 2020). 24 Table 2.1: Agency Ideology, Politicization, and Presidential Particularism Dependent variable: Logged Outlays (1) (2) (3) (4) Presidential 0.536 ∗∗∗ 0.451 ∗∗∗ 0.396 ∗∗∗ − 0.026 Co-Partisan (0.144) (0.135) (0.042) (0.113) Agency–President 0.350 ∗∗∗ 0.304 ∗∗∗ Distance (0.035) (0.088) Politicization − 0.064 ∗∗∗ − 0.081 ∗∗∗ Ratio (0.020) (0.020) Pres. Co-Partisan× − 0.805 ∗∗∗ − 0.713 ∗∗∗ Ag.–Pres. Dist. (0.057) (0.171) Pres. Co-Partisan× − 0.063 ∗∗∗ − 0.032 ∗∗∗ Politicization Ratio (0.008) (0.008) β 2 +β 4 − 0.455 ∗∗∗ − 0.409 ∗∗∗ (Ag.–Pres. Dist.) (0.0034) (0.088) β 3 +β 5 − 0.127 ∗∗∗ − 0.133 ∗∗∗ (Politicization) (0.019) (0.019) Congress FEs YES YES YES YES Legislator FEs YES YES Agency–Legislator FEs YES YES Time-Varying Covariates YES YES Observations 3,045 3,045 63,075 63,075 Adjusted R 2 0.610 0.620 0.592 0.596 ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 Note: Unit of analysis for models 1 and 2 is the district-Congress and for models 3 and 4 is the agency-district- Congress. Heteroskedasticity-corrected errors clustered by legislator (models 1 and 2) and agency-legislator (models 3 and 4) reported in parentheses. Model 2 controls for whether each member of Congress in each Congress is in the majority party, sits on the appropriations committee, sits on the ways and means committee, whether each member of Congress won their previous election with a margin less than 0.05, each district’s logged population and logged median income, and model 4 additionally controls for the distance between each agency’s Chen and Johnson (2015) ideal point estimate and each member’s DW-NOMINATE ideal point estimate. 25 2.1 displays the marginal effect of presidential co-partisanship on outlays at different levels of agency–president distance and politicization derived from model 4 in table 3.3. Agencies ideolog- ically proximate to the president allocate more to presidential co-partisans than contra-partisans. Ideologically distant agencies, on the other hand, allocate the same or less in outlays to districts represented by presidential co-partisans. -2 -1 0 1 2 -2 -1 0 1 2 Agency-President Distance (Scaled by within-agency standard deviation) Marginal Effect of Presidential Co-Partisanship -5.0 -2.5 0.0 2.5 5.0 0 10 20 Agency Politicization (Scaled by within-agency standard deviation) Marginal Effect of Presidential Co-Partisanship Figure 2.1: Particularism by Agency–President Distance and Politicization. Figure derived from model 4 in table 3.3. Left panel displays the marginal effect of presidential co-partisan at all observed levels of agency–president distance and right panel displays the marginal effect of presi- dential co-partisan at all observed levels of agency politicization. The rug along the x-axis displays the densities of the respective variables. The results provide no evidence that politicization influences agency implementation of presi- dential particularism since the marginal effect of presidential co-partisanship does not vary across differnet levels of agency politicization. Although the coefficient on politicization is negative, indicating that presidential contra-partisans receive less in outlays as agencies become more politi- cized, the sum of β 3 and β 5 is also negative, indicating that presidential contra-partisans also re- ceive less in outlays as agencies become more politicizied. Further, the effect sizes are quite small. A standard, within-agency increase in politicization reduces outlays to co-partisans by about 12% 26 and to contra-partisans by about 7.8%. If anything, more politicized agencies allocate more fund- ing to contra-partisans than co-partisans, but not by much. Ideological distance between agencies and presidents on the other hand has a much more pronounced effect. A standard, within-agency increase in agency–president distance increases outlays to contra-partisans by about 35% and de- creases outlays to co-partisans by about 34%. Table 2.2: Substantive Effects, Agency–President Distance Agency–President Co-Partisan Contra-Partisan Difference Distance Expected Outlay Expected Outlay (Co-Partisan Advantage) Aligned (− 1 SD) $12.6M $6.15M $6.42M Mean $8.32M $8.37M − $0.05M Misaligned (+1 SD) $5.51M $11.37M − $5.87M Note: Expected values calculated with estimates from model 4 in table 3.3 holding continuous variables at their means, binary variables at their modes, and adjusting for the estimated, median intercept of the agency-legislator fixed effects. Standard deviations measured as the standard devi- ation of the residualized values of agency–president distance with respect to agency-legislator and Congress fixed effects following Mummolo and Peterson (2018). Table 2.2 reports expected outlays for co- and contra-partisans, and the difference between the two, at different levels of agency–president distance. 7 An agency one within-agency standard de- viation below the mean of agency–president distance allocates, on average, $6.4 million more in funding to presidential co-partisans, while an agency one within-agency standard deviation above the mean allocates, on average, $5.9 million less over the course of an average two-year congres- sional session. 2.4.1 Placebo Test with Formula Grants In the main analysis above, I report results only from program grants over which agencies have much discretion. If the results are driven by secular trends in grant disbursement rather than bureau- cratic discretion, we should observe the same results from estimating the same model on formula 7 Since I find no evidence that agency politicization influences agency implementation of presidential particular- ism, I decline to calculate substantive effects. 27 grants over which agencies have little, if any, discretion. For the main analysis to pass the placebo test, it must be the case that the marginal effect of presidential co-partisan does not vary with agency–president distance. 8 In other words, it must be the case that particularism, or the advantage to presidential co- partisans, does not vary with the ideological alignment between the president and the disbursing agency. Formally, to pass the placebo test, it must be thatβ Placebo 4 <β Main 4 or, more conservatively, that β Placebo 4 ≤ 0. Table 2.3 reproduces the main analysis in models 1 and 2 and reports the parameters estimated for the placebo model on formula grants in models 3 and 4. For the placebo model with- out covariates, the effect (β Placebo 4 ) is only about half of the effect in the main model, passing the weaker placebo test. More importantly, once including covariates, the effect disappears and is null (p≈ 0.35), passing the more conservative test. The weak politicization findings for program grants are replicated for formula grants, so I cannot rule out that secular trends in grant disbursement, and not bureaucratic discretion, account for the politicization results in table 3.3. I can rule out that secular trends account for the ideology findings however. The main analysis of ideological distance on program grants passes the placebo test, indicating that ideological dissimilarity between agencies and presidents only affects the distribution of fed- eral grants when the agencies responsible for administering those grants have discretion over their distribution. When agencies do not have discretion, as with formula grants written by Congress, policy disagreement among presidents and agencies has no effect on whether agencies reward pres- idential co-partisans with federal funds. To see the difference more clearly, figure 2.2 displays the marginal effect of presidential co-parisan for formula grants, over which agencies have little dis- cretion. The slope is almost exactly zero, unlike for program grants in the main analysis in figure 2.1. 8 Since the effect of presidential co-partisanship does not vary with politicization even when agencies have discre- tion, this section focuses on probing the robustness of the ideology finding. 28 Table 2.3: Placebo Test with Formula Grants Dependent variable: Logged Outlays Program Grants Formula Grants (Placebo Test) (1) (2) (3) (4) Presidential 0.396 ∗∗∗ − 0.026 0.198 ∗∗∗ − 0.072 Co-Partisan (0.042) (0.113) (0.032) (0.079) Agency–President 0.350 ∗∗∗ 0.304 ∗∗∗ 0.307 ∗∗∗ 0.124 ∗∗ Distance (0.035) (0.088) (0.030) (0.056) Politicization − 0.064 ∗∗∗ − 0.081 ∗∗∗ − 0.168 ∗∗∗ − 0.175 ∗∗∗ Ratio (0.020) (0.020) (0.020) (0.020) Pres. Co-Partisan× − 0.805 ∗∗∗ − 0.713 ∗∗∗ − 0.403 ∗∗∗ − 0.042 Ag.–Pres. Dist. (0.057) (0.171) (0.053) (0.109) Pres. Co-Partisan× − 0.063 ∗∗∗ − 0.032 ∗∗∗ 0.0001 0.013 ∗ Politicization Ratio (0.008) (0.008) (0.007) (0.007) β 2 +β 4 − 0.455 ∗∗∗ − 0.409 ∗∗∗ − 0.096 ∗∗∗ 0.082 ∗ (Ag.–Pres. Dist.) (0.034) (0.088) (0.029) (0.056) β 3 +β 5 − 0.127 ∗∗∗ − 0.113 ∗∗∗ − 0.167 ∗∗∗ − 0.113 ∗∗∗ (Politicization) (0.019) (0.019) (0.020) (0.020) Congress FEs YES YES YES YES Agency–Legislator FEs YES YES YES YES Time-Varying Covariates YES YES Observations 63,075 63,075 63,075 63,075 Adjusted R 2 0.592 0.596 0.618 0.619 ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 Note: Unit of analysis is the agency-district-Congress. Heteroskedasticity-corrected errors clustered by agency- legislator reported in parentheses. Models 2 and 4 control for whether each member of Congress in each Congress is in the majority party, sits on the appropriations committee, sits on the ways and means committee, whether each member of Congress won their previous election with a margin less than 0.05, each district’s logged population and logged median income, and for each agency’s politicization ratio for each Congress, and the distance between each agency’s Chen and Johnson (2015) ideal point estimate and each member’s DW-NOMINATE ideal point estimate. 29 -2 -1 0 1 2 -2 -1 0 1 2 Agency-President Distance (Scaled by within-agency standard deviation) Marginal Effect of Presidential Co-Partisanship -5.0 -2.5 0.0 2.5 5.0 0 10 20 Agency Politicization (Scaled by within-agency standard deviation) Marginal Effect of Presidential Co-Partisanship Figure 2.2: Placebo Test: Particularism in Formula Grants. Figure derived from model 4 in table 2.3. The rug along the x-axis displays the density of each variable. 2.5 Conclusion In this chapter, I have shown that executive agencies do not uniformly implement the president’s particularistic policy preferences. Only agencies ideologically proximate to the president allocate more in grant funding to presidential co-partisans. Ideologically distant agencies do not engage in particularism. Critically, the bureaucracy only moderates presidential particularism when dis- bursing program grants, over which agencies have discretion to create application criteria, select recipients, and choose the value of the grant. When disbursing formula grants written by Congress, bureaucratic ideology has no effect on presidential particularism. e The findings presented here suggest a more nuanced treatment of presidential particularism, taking seriously the implemen- tation apparatus with which presidents must contend in order to pursue their particularistic goals. The power of the presidency is the power to persuade (Neustadt 1991), and conclusions about pres- idential power that rely on the president’s ability to direct benefits to their preferred constituents must take into account that presidents have not persuaded and cannot persuade every agency in the bureaucracy to implement their particularistic goals. Agencies do not implement policy blindly, nor do they acquiesce in every order from the president. Rather, federal agencies are active players 30 in the contestation and formulation of public policy and a critical part of the checks on presidential power and balances between institutions in the American political system (Miller and Whitford 2016). This chapter complements a growing body of work on presidential power that interrogates the agency problems inherent in the pursuit of presidential policy preferences. Regardless of presi- dential preferences, the effectiveness of presidential policy relies on compliant bureaucrats (see Lowande and Rogowski 2020 for an overview). Even for the hallmark of unilateral action, the executive order, presidential policy directives are not self-enforcing (Kennedy 2015; Rudalevige 2012). This chapter has shown that presidential power to funnel benefits to key constituencies also is not self-enforcing. Only agencies aligned with the president implement the president’s particularistic agenda. 2.6 Postscript: A Pandemic Exception “...we ought not to recklessly appropriate money, at least we ought, as a rule, to take the time to ascertain for what purpose it is going to be spent, how it will be spent, and as to whether the scope of the appropriation is within the possibilities of its helpful expenditure. ” —Senator Oscar Underwood concerning the flu pandemic in 1918 9 “The Executive Branch must be accountable to [the] taxpayers. Financial relief to address the coronavirus pandemic should not be turned into a slush fund for a pres- ident seeking reelection, with little accountability to the people whose money he is spending. ” — Senator Patrick Leahy concerning COVID-19 in 2020 10 As the two quotations uttered on the floor of the US Congress a century apart indicate, concerns over how funds appropriated by Congress to combat pandemics will be allocated by the Executive Branch have always pervaded congressional decisions to delegate and appropriate (Cf. Epstein and O’Halloran 1996, 1999; V olden 2002). During non-emergency periods in American politics, that 9 U. S. Congress 1918, 10896 10 U. S. Congress 2020, S2055 31 concern is indeed justified. Districts represented by members of Congress that share the president’s party consistently receive more federal money than those of the opposing party (Berry, Burden and Howell 2010; Christenson, Kriner and Reeves 2017; Rogowski 2016) and geographies that are important to the president’s reelection likewise receive more federal money than those less important (Dynes and Huber 2015; Kriner and Reeves 2015). But might the large and immediate societal cost stemming from the pursuit of a parochial distribution of lifesaving resources lead the Executive Branch to pursue a more efficient distribution of funds? The political economic literature on distributive politics makes clear that political actors with discretion over distributive programs use that discretion to further their own interests, be it re- election, coalition building, or strengthening the party brand for members of Congress and the president, or currying favor with and avoiding sanction by political overseers and interest groups for bureaucrats (Bertelli and Grose 2009; Bickers and Stein 1996; Evans 2004; Hudak 2014; Lee 2000, 2002, 2003; Mayhew 1974). However, empirical and theoretical treatments of distributive politics either aggregate over time or assume “normal” conditions. The potential devastation asso- ciated with pandemics produces good reasons to believe that pandemics should meaningfully alter elites’ distributive calculus. Studying distributive politics during pandemics is important for better- ing our understanding of how institutions made up of individuals with strong personal incentives respond to pandemics: problems of national scale that require collective action. In this section, I examine federal spending both in the aggregate and disaggregated across eight federal agencies both during non-emergency periods and during four twenty-first century pan- demics: H1N1 in 2009, Ebola in 2014, Zika in 2016, and COVID-19 in 2020. I find that although during normal, non-emergency periods, presidential co-partisanship is a significant predictor of federal outlays, during pandemics there is no evidence of a presidential co-partisan advantage. I then compare COVID-19 spending to spending on the opioid crisis, a public health problem with- out the public urgency of an infections pandemic, in order to isolate the independent effect of pandemics beyond general public health emergencies. 32 2.6.1 Pandemic Spending in the United States Despite the horizontally and vertically fragmented structure of the US government, public health crises require centralized, national responses. Legislators in the early 20th century knew as much when appropriating one million dollars to the Public Health Service to combat the flu pandemic in 1918 (about $17 million in 2020 dollars). “[O]rdinarily I would not be in favor of this bill, but with the emergency that is upon us at this time I see no other way to control it” admitted William R. Green, a representative from Iowa (U. S. Congress 1918, 11273). Although legislators knew they had to act, partisan battles over how the Democratic Wilson administration would implement their authorization took center stage during debate. Boies Pen- rose, Republican senator from Pennsylvania, asked to Thomas S. Martin, Democratic senator from Virginia, “Then the Senator’s thought is to appropriate this $1,000,000 and investigate the propri- ety of the appropriation afterwards?” Martin, President Wilson’s co-partisan, responded “Not to investigate it at all, but to leave it to the officers of the Government to expend it for the purposes in- dicated.” Unsatisfied with Martin’s laissez-faire attitude toward delegation, Penrose snapped back “Well, has the Senator any idea whatever as to how the money is to be expended?” Sticking to his initial retort, Martin maintained “The Medical Department of the Government is to expend it” (U. S. Congress 1918, 10895). One of the only recorded distributive policy choices to combat the 1918 pandemic is the Public Health Service’s creation of ten new hospitals in Palo Alto, CA, Greenville, SC, Alexandria, LA, Dansville, NY , Norfolk, V A, Chicago, IL, Washington, D.C., Jacksonville, FL, and East Norfolk, MA (Public Health Service 1919). Of the nine hospitals opened in states with electoral college votes, six were opened in states won by President Wilson. But of the thirty states Wilson carried, only twenty percent saw a new hospital while of the eighteen states Wilson’s Republican opponent Charles Evans Hughes carried, seventeen percent saw a new hospital, hardly constituting a presi- dential co-partisan advantage. Of the nine hospitals, six went to congressional districts represented 33 by Democrats and three to districts represented by Republicans. 11 Three percent of Democratic districts and one percent of Republican districts received a hospital. A century later, partisan battles over the implementation of distributive programs have only intensified. In 2020, Democrats held up what would become the Coronavirus Aid, Relief, and Economic Security Act (CARES Act) over concerns that there would not be enough oversight of a $500 billion appropriation to the Treasury to issue loans. 12 While information on spending in 1918 to combat the influenza pandemic are not readily available, systematic data on how the federal government allocated funds pursuant to four twenty-first century pandemics are. Table 2.4 displays a two-by-two table of those four pandemics by severity and the president’s party. I classified Zika and Ebola as not severe since each reported very few cases. Zika, a mosquito-borne virus associated with birth defects, resulted in only 5620 reported cases in the United States (Centers for Disease Control and Prevention 2019b,c) and Ebola, a potentially deadly virus spread though direct contact, resulted in only four reported cases in the United States (Cen- ters for Disease Control and Prevention 2015). On the other hand, both COVID-19 and H1N1, two respiratory diseases, resulted in tens of millions of cases and tens or hundreds of thousands of deaths in the United States. COVID-19 also resulted in an almost full shutdown of public gath- erings in the United States including school, government office, and business closures and H1N1 resulted in over 700 school closures (Klaiman, Kraemer and Stoto 2011). If pandemics sufficiently alter governmental officials’ distributive calculus, the moderating ef- fect of a severe one likely would much greater than for a mild one, implying that COVID-19 and H1N1 are the most likely cases for a reduction in parochialism. Additionally, Democratic presi- dents tend to favor direct spending, and a host of findings in the distributive politics literature hold 11 For all but the Palo Alto and Chicago hospitals, the district was represented by the party that carried the state’s electoral vote. Palo Alto, however was represented by a Republican and Chicago districts by Democrats. 12 Cochrane, Emily, Jim Tankersley, and Jeanna Smialek. “Emergency Economic Rescue Plan in Limbo as Democrats Block Action.” New York Times (March 22, 2020) https://www.nytimes.com/2020/03/22/us/ politics/coronavirus-economic-rescue-plan.html; Stein, Jeff. “Treasury’s power over $500 billion loan program becomes key sticking point in coronavirus aid bill.” Washington Post (March 22, 2020) https://www. washingtonpost.com/business/2020/03/22/treasury-coronavirus-senate-corporate-loan/. 34 only for Democratic presidents (Reingewertz and Baskaran 2019). Republicans’ core constituen- cies tend to be more fiscally conservative so more spending, even if it goes to those constituents, may not be Republicans’ preferred policy choice (Reingewertz and Baskaran 2019). Together, these imply that severe pandemics under Republican presidents should result in the smallest dis- crepancy in outlays to presidential co- and contra-partisans, while minor pandemics under Demo- cratic presidents should result in the most parochial distribution of federal funds. Specifically, spending to combat COVID-19 should be the least parochial and spending to combat Ebola should be the most parochial. Table 2.4: Typology of Twenty-First Century Pandemics Severity Severe Not Severe President’s Party Republican COVID-19 Zika (2020) (2016–17) Democratic H1N1 Ebola (2009–10) (2014) Note: At the time of writing, COVID-19 has resulted in 2.4 million cases and 121,809 deaths (Centers for Disease Control and Prevention 2020a). During the H1N1 outbreak in 2009 and 2010, the United States reported 60.8 million cases, 274,304 hospitalizations, and 12,469 in the United States (Centers for Disease Control and Prevention 2019a). During the Zika outbreak in 2016 and 2017, the United States reported 5620 cases (Centers for Disease Control and Prevention 2019b,c). During the Ebola outbreak in 2014, the United States reported four cases (Centers for Disease Control and Prevention 2015). Thus, two hypotheses emerge. First, during pandemics, there should no relationship, or a less robust relationship than during normal periods, between federal spending and presidential co- partisanship. Second, the relationship between federal spending and presidential co-partisanship should be conditioned by the severity of the pandemic and the president’s party. To test these hypotheses, I collected spending data from the Department of the Treasury on outlays related to COVID-19, Zika, Ebola, and H1N1. I then aggregated those data either to the state- or congres- sional district-level, and then further disaggregated them by the agency responsible for disbursing the funds. 13 13 Data were retrieved fromusaspending.gov. 35 Since only four major pandemics have occurred since the Treasury began releasing detailed spending data, the test of the conditional hypothesis from above is necessarily limited. That said, figure 2.3 displays the linear relationship between population and spending, and popular vote share for the governing president and all federal spending across the four twenty-first cen- tury pandemics. 14 The governmental responses to Zika, Ebola, and H1N1 were not as dramatic as the response to COVID-19, leaving me unable to conduct agency-level analyses of spending. In each case, a strong relationship between population and spending indicates a more or less efficient distribution of resources, while a positive relationship between popular vote share and spending indicates favoritism toward the president’s supporters. AL AK AZ AR CA CO CT DE FL GA HI ID IL IN IA KS KY LA ME MD MA MI MN MS MO MT NE NV NH NJ NM NY NC ND OH OK OR PA RI SC SD TN TX UT VT VA WA WV WI WY 0 100 200 300 0 10 20 30 40 Population (in millions) COVID-19 Funding (in millions) AL AK AZ AR CA CO CT DE FL GA HI ID IL IN IA KS KY LA ME MD MA MI MN MS MO MT NE NV NH NJ NM NY NC ND OH OK OR PA RI SC SD TN TX UT VT VA WA WV WI WY 0 25 50 75 100 30 40 50 60 70 2016 Trump popular vote COVID-19 funding per capita AL AK AZ AR CA CO CT DE FL GA HI ID IL IN IA KS KY LA ME MD MA MI MN MS MO MT NE NV NH NJ NM NY NC ND OH OK OR PA RI SC SD TN TX UT VT VA WA WV WI WY 0 100 200 300 0 10 20 30 40 Population (in millions) Zika funding (in millions) AL AK AZ AR CA CO CT DE FL GA HI ID IL IN IA KS KY LA ME MD MA MI MN MS MO MT NE NV NH NJ NM NY NC ND OH OK OR PA RI SC SD TN TX UT VT VA WA WV WI WY 0 10 20 30 30% 40% 50% 60% 70% 2016 Trump popular vote Zika funding per capita AL AK AZ AR CA CO CT DE FL GA HI ID IL IN IA KS KY LA ME MD MAMI MN MS MO MT NE NV NH NJ NM NY NC ND OH OK OR PA RI SC SD TN TX UT VT VA WA WVWI WY 0 200 400 600 0 10 20 30 Population (in millions) H1N1 funding (in millions) AL AK AZ AR CA CO CT DE FL GA HI ID IL IN IA KS KY LA ME MD MA MI MN MS MO MT NE NV NH NJ NM NY NC ND OH OK OR PA RI SC SD TN TX UT VT VA WA WV WI WY 0 25 50 75 40% 50% 60% 70% 2008 Obama popular vote H1N1 funding per capita AL AK AZ AR CA CO CT DE FL GA HI ID IL IN IA KS KY LA ME MD MA MI MN MS MO MT NE NV NH NJ NM NY NC ND OH OK OR PA RI SC SD TN TX UT VT VA WA WV WI WY 0 500 1000 0 10 20 30 Population (in millions) Ebola funding (in millions) AL AK AZ AR CA CO CT DE FL GA HI ID IL IN IA KS KY LA ME MD MA MI MN MS MO MT NE NV NH NJ NM NY NC ND OH OK OR PA RI SC SD TN TX UT VT VA WA WV WI WY 0 25 50 75 30% 40% 50% 60% 70% 2012 Obama popular vote Ebola funding per capita Figure 2.3: Federal Spending by V ote Share and Population by Pandemic. For COVID-19, a severe pandemic during a Republican presidency, the relationship between population and funding is quite strong and positive, while there is no relationship between vote share for President Trump, the governing president, and spending. For Ebola however, a mild pan- demic during a Democratic presidency, the relationship between vote share for President Obama, the governing president, and spending is positive, indicating favoritism toward the president’s sup- porters. These data are consistent with the conditional hypothesis that mild pandemics during 14 Locally estimated scatterplot smoothed (LOESS) curves allowing nonlinear relationships display substantively similar findings. 36 Democratic administrations are likely to result in a parochial distribution of funds, while severe pandemics during Republican administrations are likely to result in a more even distribution. Since the between-pandemics analysis essentially comprises four observations, no strong con- clusions can be drawn. However, the findings are suggestive of a moderating or dosage effect of severity and the president’s party. Since severe pandemics result in more deaths and illnesses, the societal costs of pursuing a parochial distribution of funds are much higher. Mild pandemics, on the other hand, may be insufficient to meaningfully alter officials’ distributive calculus, particularly during periods of Democratic control of the presidency. Dichotomizing pandemics into severe and not severe obscures important information concerning how severe a pandemic might be, but with a limited sample of pandemics such a classification is appropriate. Testing the main hypothesis that pandemic spending might be more evenly distributed than spending during normal periods of time, requires additional data and empirical strategies. The following section compares how each of eight agencies spent discretionary funds from 2007–2018 and how they spend funds appropriated to them by Congress to combat COVID-19. 2.6.2 COVID-19 Spending I collected data on how the Departments of Commerce, Defense, Health and Human Services, Justice, and State, National Credit Union Administration, National Science Foundation, and US Agency for International Development allocated funds from 2007–2018 to measure each agency’s baseline or normal level of presidential favoritism. Estimating baseline presidential favoritism in each agency is important since existing work has found that the implementation of presidential par- ticularism varies by agency (Berry and Gersen 2017; Napolio 2021). I then estimated the following least squares model for each agency: ln(Outlays it )=α+β All Presidential Co-Partisan it +ξ X it +δ +ε (2.2) 37 where subscript i indexes congressional districts, subscript t indexes Congresses, presidential co- partisan is a binary variable taking the value of one if the district receiving the outlay was repre- sented by the member of Congress from the same party as the president, X is a matrix of covariates, and δ is a vector of state-level fixed effects. Thus, β All identifies the within-state effect of presi- dential co-partisanship on all outlays. I also collected data on how those eight agencies allocated funds appropriated to them by Congress to combat COVID-19 and aggregated them to the congressional district level. Since COVID-19 is ongoing at the time of writing, I only include data on the first quarter of outlays. I then estimated a Bayesian least squares regression model via Markov chain Monte Carlo sam- pling of COVID-19 outlays on copartisanship and covariates with the estimated coefficients from estimating equation 3.3 serving as informative priors. I set each coefficient’s prior variance to the largest value that still ensures the coefficient is statistically distinguishable from zero in the direction estimated from equation 3.3. If the estimate was not distinguishable from zero, I use the estimated variance from equation 3.3, ensuring I do not shrink any variances. 15 Using these values as priors is a reasonably informative choice that nonetheless allows for the data to update estimates. Specifically, for each agency I estimate ln(Outlays i )∼ N (α+β COVID-19 Presidential Co-Partisan i +ξ X i +δ,σ 2 ) (2.3) where the prior onβ COVID-19 is β COVID-19 ∼ N β All ,max " |β All | 1.64 2 ,σ(β All ) #! (2.4) whereσ(β All ) is the estimated variance onβ All from equation 3.3. 15 For example, since the value of β All is positive for most agencies, I set the prior variance on β COVID-19 to the largest value such that the 90% credible interval touches but does not overlap zero. To find that value, I solve β All − 1.64× φ = 0 for φ. Since the normal distribution is symmetric, the prior variance is φ 2 = |β All | 1.64 2 . For estimates not distinguishable from zero, the variance, by definition, is sufficiently large that it overlaps zero, so finding φ 2 would shrink the variance. So, for null estimates, I use the estimated variance,σ(β All ). 38 Thus, β All serves as the prior, the COVID-19 spending data provides information with which to update the estimate of presidential particularism, and β COVID-19 is the mean of the posterior distribution of the effect of copartisanship on COVID-19 outlays. This empirical strategy allows me to estimate the effect of copartisanship on COVID-19 spending while incorporating prior be- liefs about each agency’s baseline presidential favoritism estimated from realistic models of the data-generating process. If spending during COVID-19 favors presidential co-partisans less than during normal periods of American politics, thenβ COVID-19 <β All for most agencies. In other words, the effect of shar- ing the president’s party on outlays should be less for pandemic spending than normal spending. Requiring β COVID-19 ≤ 0 is a more conservative test and would imply that there is no positive relationship between presidential co-partisanship and federal spending. 0.096 (0.147) 0.292 (0.096) -0.152 (0.25) -0.968 (0.424) 0.878 (0.42) 0.225 (0.088) -0.053 (0.01) 0.041 (0.065) 0.571 (0.204) 0.365 (0.197) 1.17 (0.251) 0.487 (0.207) 0.552 (0.143) 0.609 (0.22) -0.053 (0.05) 0.056 (0.106) HHS DOD NCUA NSF USAID DOS DOC DOJ -2 -1 COVID-19 Mean All Mean 1 2 Effect of Copartisanship on Logged Grant Funding All Funding (Prior) COVID-19 Funding (Posterior) Figure 2.4: Comparing Effects of Copartisanship. Estimates derived from models controlling for district population, district median income, whether the member of Congress representing the district serves on the appropri- ations committee, whether they serve on the ways and means committee, whether they won their previous election with a margin of less than five points, state fixed effects, and—in the case of the models including all funding from 2007–2018—the president’s party. Error bars represent 95% credible intervals from the posterior distribution. Figure 2.4 displays the effect of presidential co-partisanship on outlays for COVID-19 spending (β COVID-19 ) in blue and all spending (β All ) in black, along with 95% credible intervals. Each model controls for district population, district median income, whether the member of Congress 39 representing the district serves on the appropriations committee, whether they serve on the ways and means committee, whether they won their previous election with a margin of less than five points, state fixed effects, and—in the case of the models including all funding from 2007–2018— the president’s party. For all agencies except the Department of Justice, the effect of presidential co-partisanship on outlays is smaller for COVID-19 spending than all spending, with two of those differences statistically distinguishable from zero at the 0.05 level, implying a less parochial distribution of funds during COVID-19 than during normal periods. Additionally, the estimate for the Department of Health and Human Services is negative and statistically distinguishable from zero at the 0.05 level, passing the more conservative test ruling out a positive relationship between presidential co- partisanship and spending. Spending by the Department of Health and Human Services, the agency most closely related to the public health aspects of the governmental response to COVID-19, favors presidential co-partisans the least, instead favoring presidential contra-partisans. 16 The negative effect of copartisanship on spending by the Department of Health and Human Services deserves further investigation. The conservative interpretation of the estimate is simply that the estimate rejects a one-tailed hypothesis test of a positive effect, but since the model in- corporated information about the baseline level of presidential favoritism through an informative prior, the negative effect appears to be substantively important given that is not overwhelmed by the positive prior. One potential explanation for the negative effect of copartisanship for the Department of Health and Human Services is that COVID-19 cases were clustered in cities and other urban or metropoli- tan areas, and therefore funds had to flow to cities, which tend to be represented by Democrats. To examine whether congressional district’s urbanism is driving the results, I estimate new models using Bayesian regression as above including each district’s Census-estimated percent of residents 16 Although popular and media attention have focused on President Trump’s spars with contra-partisan governors (see, e.g., Shear, Michael D., and Sarah Mervosh. “Trump Encourages Protest Against Governors Who Have Imposed Virus Restrictions.” New York Times (April 29, 2020) https://www.nytimes.com/2020/04/17/us/politics/ trump-coronavirus-governors.html), similar models at the state level with an indicator variable for contra- partisan governors recover null results for both all spending and COVID-19 spending across all eight agencies. 40 living in a principal city and an interaction term between urbanism and copartisanship. 17 Table 2.5 reports results from these models. 18 Table 2.5: Effect of Copartisanship and District Urbanism on HHS Outlays Dependent variable: Logged Outlays (1) (2) Presidential − 0.377 − 0.355 Copartisan (0.441) (0.487) Percent 13.625 13.834 Urban (2.443) (2.518) Pres. Copart.× − 5.879 Percent Urban (4.756) Observations 435 435 State FEs YES YES Note: Unit of analysis is the congressional district. Estimates control for district population, district median income, whether the member of Congress representing the district serves on the appropriations committee, whether they serve on the ways and means committee, whether they won their previous election with a margin of less than five points, and state fixed effects. Standard deviation of the posterior distribution reported in parentheses. The first model shows that including a district’s urbanism does moderate the large negative effect of copartisanship, yet the relationship remains negative, still indicating no presidential fa- voritism. The second model including the interaction term, indicates that the effect of presidential copartisanship is negative at all levels of urbanism. The coefficient on presidential copartisan in- dicates that the effect of copartisanship on spending in districts where no residents live in cities is − 0.355. The effect of copartisanship on spending in districts at the 25th percentile of urbanism is − 0.688, at the median is− 1.034, and at the 75th percentile is− 1.616. The effect of copartisanship recovered from the main analysis is approximately the effect of copartisanship for districts at the 45th percentile of urbanism (10.4% of residents in a principal city in absolute terms). The average 17 Although measuring and conceptualizing “rurality” is difficult (Nemerever and Rogers Forthcoming), using the percent of residents living in a principal city is sufficient to capture the urbanism of each congressional district for my purposes. 18 I use uninformative priors on the new parameters for percent urban and the interaction between presidential copartisan and percent urban. 41 Republican district is almost exactly at the 45th percentile of urbanism, so the main results are not overwhelmed by the effects of urbanism, although urbanism did indeed play a large role. Together, the analyses imply that almost all agencies implementing distributive programs in response to COVID-19 have done so in a manner that privileges the president’s copartisans less than during “normal” periods in American politics. Of the eight agencies studied, the Department of Health and Human Services, one of the agencies most responsible for the scientific and medical response to COVID-19, privileged presidential co-partisans the least, with some evidence indicat- ing that it funneled more funds to Democrats, presidential contrapartisans, although that effect is somewhat moderated by congressional districts’ urbanism since COVID-19 cases were clustered in urban areas early on. 2.6.3 Counterfactual Public Health Emergency: The Opioid Crisis While comparing COVID-19 to all spending facilitates comparisons between pandemic spending and “normal” spending, using all spending likely includes outlays that are not appropriate coun- terfactual outlays to pandemic ones. Therefore, this section compares Department of Health and Human Services spending related to the opioid crisis and COVID-19. The opioid crisis represents a public health emergency without the widespread public urgency associated with infectious pandemics, offering a counterfactual distributive program implemented by the same agencies in the same policy area during a similar time period. The opiod crisis lacks the characteristics of a pandemic, having accelerated at a slower pace than COVID-19 and engen- dering a social and political etiology that has centered on medical doctors’ prescription practices and individual responsibility rather than unsolicited risk as with COVID-19 or other pandemics (Dasgupta, Beletsky and Ciccarone 2018; Meldrum 2016), yet it has resulted in over 450,000 deaths in the United States since 1999 (Centers for Disease Control and Prevention 2020b). Thus, the opioid crisis facilitates the isolation of the effect of pandemics on spending apart from general public health emergencies. 42 0.487 (0.207) 0.677 (0.202) -0.968 (0.424) COVID-19 (Posterior) All Funding (Prior) Opioid Crisis (Posterior) -2 -1 0 1 2 Effect of Copartisanship on Logged Grant Funding Figure 2.5: Department of Health and Human Services Spending. Estimates derived from models con- trolling for district population, district median income, whether the member of Congress representing the district serves on the appropriations committee, whether they serve on the ways and means committee, whether they won their previous election with a margin of less than five points, state fixed effects, and—in the case of the models including all and opioid funding—the president’s party. Error bars represent 95% credible intervals from the posterior distribution. Figure 2.5 displays the estimated effect of presidential co-partisanship on Department of Health and Human Services funding for COVID-19, all funding from 2007–2018, and spending related to the opioid crisis from 2007–2018. Estimates for the opioid crisis and COVID-19 were estimated using Bayesian regression as in the previous analysis. The effect of presidential co-partisanship on spending related to the opioid crisis is indistinguishable from the effect on all spending, indicat- ing that pandemic spending during COVID-19 favors presidential co-partisans both less than all spending and spending on a public health emergency without the social and political meaning as- cribed to pandemics. This suggests that severe pandemics, and COVID-19 specifically, engender a distributive calculus significantly different than did general public health emergencies, resulting in a less parochial distribution of funds lacking any clear favoritism toward presidential co-partisans. 2.6.4 Discussion and Conclusion The foregoing analysis provides evidence that spending during pandemics does not benefit presi- dential co-partisans as much as spending does during “normal” periods of American politics. For 43 severe pandemics, such as the COVID-19 outbreak in 2020, spending data reveal presidential fa- voritism neither in the aggregate at the state level, nor disaggregated by agency at the congressional district level. The moderating effect of pandemics on parochial spending cannot be explained by the idiosyncrasies of public health policy since spending within the Department of Health and Human Services during pandemics displays no presidential favoritism even though it does during “normal” periods. More convincingly, the moderating effect of pandemics cannot be explained by the idiosyncrasies of public health emergencies as demonstrated by the presidential favoritism associated with spending related to the opioid crisis, an ongoing public health crisis responsible for almost half a million deaths in the United States in the twenty-first century. This does not suggest, however, that spending during pandemics is somehow immune from traditional distributive politicking. Indeed, anecdotal evidence suggests that certain programs have benefited elected officials, 19 others have favored states important to President Trump’s reelection, 20 and still others have been directed toward furthering the preexisting policy goals of cabinet sec- retaries. 21 Additionally, at the time of writing, the federal government is still disbursing funds to combat COVID-19, so the analysis presented in this section should be interpreted as such. The evaluation of ongoing events necessitates ascribing more uncertainty to any analysis than the eval- uation of past events. Instead, I argue the large and immediate cost associated with socially inef- ficient spending during pandemics might induce a less parochial distribution of federal funds and 19 Daly, Matthew, and Brian Soldysko. “Congress created coronavirus aid, then reaped the bene- fits, data show.” Boston Globe (July 7, 2020) https://www.bostonglobe.com/2020/07/07/nation/ congress-created-coronavirus-aid-then-reaped-benefits-data-show/. 20 Olorunnipa, Toluse, Josh Dawsey, Chelsea Janes and Isaac Stanley-Becker.“Governors plead for medical equipment from federal stockpile plagued by shortages and confu- sion.” Washington Post (March 31, 2020) https://www.washingtonpost.com/politics/ governors-plead-for-medical-equipment-from-federal-stockpile-plagued-by-shortages-and-confusion/ 2020/03/31/18aadda0-728d-11ea-87da-77a8136c1a6d_story.html; Wire, Sarah D., and Jennifer Haberkorn. “Senate close to deal on more coronavirus funds for small businesses as critics say too much went to big firms.” Los Angeles Times (April 20, 2020) https://www.latimes.com/politics/story/2020-04-20/ senate-more-funding-small-business-program. 21 Meckler, Laura. “Betsy DeV os defends decision to direct stimulus money to private schools.” Washington Post (May 21, 2020) https://www.washingtonpost.com/local/education/ betsy-devos-stimulus-private-schools/2020/05/21/d790b926-9b99-11ea-ad09-8da7ec214672_ story.html 44 I find evidence that the first quarter of COVID-19 outlays were distributed less parochially than outlays during “normal” periods of American politics. 45 Chapter 3 Persuasion: Signaling to the White House “When we walk in the White House, we’re joined at the hip.” —Former Secretary of Defense James Mattis to former Secretary of State Rex Tiller- son (Woodward 2020, 21) In early 2020, a global pandemic broke out, leading to hundreds of thousands of deaths in the United States and millions worldwide. States and localities rushed to implement social distancing measures and prohibit economic activities that threatened public health while federal public health and emergency agencies began gathering information and developing plans to combat the novel coronavirus. Against the advice of experts, President Donald J. Trump was bullish on a quick return to normal. 1 Governors and other subnational leaders, as chief executives of sovereign entities, protested the President’s minimal response publicly and vociferously. 2 Bureaucrats in the Trump administration, on the other hand, faced high costs to speaking out against the President—officials from the Centers for Disease Control and Prevention were not allowed to speak to the media, 3 and 1 Wagner, John and Brady Dennis. “Trump wants U.S. economy ‘opened up and raring to go’ by Easter.” Washington Post (March 24, 2020). https://www.washingtonpost.com/ health/trump-wants-us-economy-opened-up-and-raring-to-go-by-easter/2020/03/24/ dced0a12-6d65-11ea-b148-e4ce3fbd85b5_story.html 2 Costa, Robert and Aaron Greg. “Governors and mayors in growing uproar over Trump’s lagging coro- navirus response.” Washington Post (March 22, 2020). https://www.washingtonpost.com/politics/ governors-and-mayors-in-growing-uproar-over-trumps-lagging-coronavirus-response/2020/03/ 22/98ac569a-6c49-11ea-a3ec-70d7479d83f0_story.html 3 Duhigg, Charles. “Seattle’s Leaders Let Scientists Take the Lead. New York’s Did Not.” The New Yorker (April 26, 2020). https://www.newyorker.com/magazine/2020/05/04/ seattles-leaders-let-scientists-take-the-lead-new-yorks-did-not 46 one official was even removed from his post for opposing one of the president’s policies. 4 In order to influence policy to reflect their expertise, bureaucrats had to turn to alternative approaches. One such approach was taken by The Centers for Disease Control and Prevention (CDC) and the Federal Emergency Management Agency (FEMA). A coalition of the two agencies—one staffed with public health experts and one with experts in disaster management—produced joint guidelines to reopen the economy that recommended a significantly longer return to normalcy than the president’s public position. 5 Although President Trump did not adopt the coalition’s recom- mendations wholly, his plan to reopen the economy moved toward the coalition’s recommended policy. 6 By collaborating, the CDC and FEMA successfully signaled to the president that he ought to move toward a more efficient policy. The coalition formed by the CDC and FEMA in response to the pandemic is not exceptional. From intuitive pairings like the Departments of Defense and Veterans Affairs to perhaps less obvi- ous pairings such as the Department of Labor and the National Aeronautics and Space Administra- tion, executive agencies form hundreds of policymaking coalitions each presidential term. These coalitions are responsible for producing almost 3,000 rules between 1997 and 2016 ranging from financial regulation and the implementation of civil rights laws to responses to natural and envi- ronmental disasters. This chapter asks: why do executive agencies form coalitions? Legislative coalitions in the form of pork-barrel majorities and political parties are widely theorized and studied, but less atten- tion has been paid to how and why agencies in the executive branch form coalitions. In contrast to legislator decisions to form voting blocs and parties, executive agencies are not fully autonomous agents engaged in divide-the-dollar type games, rather they are embedded in a separation of powers 4 Shear, Michael D. and Maggie Haberman. “Health Dept. Official Says Doubts on Hydroxychloroquine Led to His Ouster.” The New York Times (April 22, 2020). https://www.nytimes.com/2020/04/22/us/politics/ rick-bright-trump-hydroxychloroquine-coronavirus.html 5 Sun, Lenna H., Josh Dawsey and William Wan. “CDC, FEMA have created a plan to reopen America. Here’s what it says.” Washington Post (April 14, 2020). https://www.washingtonpost.com/health/2020/04/14/ cdc-fema-have-created-plan-reopen-america-heres-what-it-says/ 6 Freking, Kevin. “A look at new guidance to states on the coronavirus.” Washington Post (April 16, 2020). https: //www.washingtonpost.com/politics/a-look-at-new-guidance-to-states-on-the-coronavirus/ 2020/04/16/4e16fa7c-8030-11ea-84c2-0792d8591911_story.html 47 system that grants elected politicians substantial authority over them. Executive agencies’ depen- dence on the political branches calls for a distinctive theory of executive coalition building. This chapter presents such a theory, arguing that agencies form coalitions as costly signals to political overseers in order to optimize their autonomy given their subsidiary position in a separation of powers system. Bureaucrats form coalitions actively to advance their policy goals in the face of potential political opposition. I argue that coalition building serves as a costly signal to political overseers that certain bu- reaucratic policies are efficient. Agencies unlikely to see their preferred policies enacted without sanction if they act individually are likely to form coalitions when transaction costs associated with collaboration are sufficiently low. This stands in contrast both to the legislative motivation for coalition building of maximizing distributive benefits and overcoming social choice problems (Aldrich 2011; Baron 1989; Baron and Ferejohn 1989) and the technocratic or apolitical explana- tion for interagency coordination of information sharing or politically directed coordination where agencies simply implement presidentially led coordination (Freeman and Rossi 2011, 2012; Saito 2020). This chapter tests the strategic theory empirically with data on dozens of agencies using data from the Federal Register to construct a network of agency coalitions. I find that agencies form coalitions when doing so optimizes their autonomy given their subordinate position in the Amer- ican separation of powers system. Agencies form coalitions when the likelihood that individual action will go unsanctioned is low and the transaction costs associated with coalition formation are low, and they do so in order to actively pursue their policy goals in a federal system that grants politicians substantial authority over them. I supplement my theoretical and empirical analyses with interviews from civil servants involved in forming executive coalitions. 48 3.1 Executive Coalition Building in the American System As a first-order issue, I consider the legal, procedural, and political constraints on executive coali- tion building. The average law delegates to almost three executive agencies and about two-and-a- half clauses in the average law delegate authority to more than one actor for the same regulatory activity (Farhang and Yaver 2016, 441). Overlapping jurisdictions are commonplace from eco- nomic and financial regulation to food safety and border security (Freeman and Rossi 2012, 1134). Yet not all laws that delegate to multiple agencies compel coordination. Instead, a majority of interagency policymaking is voluntary initiated by agencies despite no formal requirement to do so either from Congress or the president (Saito 2020). For example, the Federal Deposit Insurance Act delegates to four agencies and provides that “more than one agency may be an appropriate Fed- eral banking agency with respect to any given institution” (Freeman and Rossi 2012), not requiring coordination but providing that coordination may be an option. Other laws, like the Dodd-Frank Wall Street Reform and Consumer Protection Act, authorize interagency coordination and pro- vide that a certain agency shall serve as the coordinator of joint policymaking, but do not require coordination. Further, broad laws governing administrative procedures, like the Administrative Procedure Act and Paperwork Reduction Act, have no explicit procedures or mandates for inter- agency policymaking, instead allowing agencies to form coalitions with few, if any, procedural requirements. Legislative authorization, however, is not the only legal basis for executive coalition build- ing. Executive orders and other presidential documents like memoranda may also authorize or facilitate interagency collaboration. For example, President Obama in 2011 issued Executive Or- der 13,563 which “emphasizes the importance of coordination to reduce regulatory burdens and to simplify and harmonize rules” (Freeman and Rossi 2012, 1180). However, the executive order does not mandate coordination, instead stating that agencies “shall attempt to promote...coordination.” Thus, while the order mandates that agencies consider coordination, it does not mandate that they do in fact coordinate, suggesting that decisions to form coalitions are discretionary choices. The president on occasion has mandated coordination: for example President Obama’s 2010 issuance 49 of a memorandum directing multiple agencies to collaborate to regulate carbon capture and seques- tration (75 FR 6087; Freeman and Rossi 2012, 1175). Yet it remains that the majority of executive coalition building is bureaucratically led rather than directed by the president (Saito 2020). Courts also constrain agencies’ attempts to build coalitions. Chevron v. National Defense Resource Council 467 U.S. 837 (1984) is the controlling law concerning whether agency inter- pretations of legislation are appropriate. Chevron requires first that a court determine whether a law delegates, second whether there is ambiguity in a statute, and third whether an agency’s in- terpretation of the ambiguous clause is reasonable. Legal scholars argue that agency coordination should make it easier for agencies to survive Chevron review since coordinating to produce a joint policy implies the agencies have all first understood the law to delegate, second that a statute is ambiguous, and last come to the same interpretation of the ambiguous clause (Freeman and Rossi 2011, 1203–9). Extant and controlling case law classifies statutes into one of three schemes: (1) “generic statutes like the [Administrative Procedure Act], [Freedom of Information Act], and [Federal Advi- sory Committee Act],” (2) those “where the agencies have specialized enforcement responsibilities but their authority potentially overlaps,” and (3) those “where expert enforcement agencies have mutually exclusive authority over separate sets of regulated persons” (Collins v. National Trans- portation Safety Board 351 F.3d 1246 (D.C. Circuit 2003)). When agency interpretations conflict, only in the first two cases may courts review the policy de novo; in the third case each agency is entitled to Chevron deference even if they come to opposing interpretations. Therefore, agencies implementing laws of the first two types, of which many statutes belong, may overcome legal chal- lenges by coordinating to set the same policy, signalling to courts that even under de novo review, the negotiated policy should stand. What is more, courts have the authority to interpret legislation only to allow some of the agen- cies delegated to to implement the law. In Couer Alaska v. Southeast Alaska Conservation Council 557 U.S. 261 (2009), the Supreme Court held that a rule jointly promulgated by the EPA and Army Corps of Engineers under the Clean Water Act was entitled to Chevron deference, but only 50 because the court viewed the Army Corps of Engineers as the appropriate regulatory body de- spite each agency coming to the same interpretation of the law and jointly promulgating a rule. Therefore, if two agencies have overlapping authority but there is uncertainty about whether each constituent agency has authority to implement a subset of the authorizing statute, collaboration may help those agencies survive a legal challenge since the union of their authorities may be larger than the intersection. Additionally, when agencies with overlapping jurisdictions come to different interpretations of the same law, courts may unilaterally decide which agency maintains the author- ity to regulate pursuant to the overlapping law (Martin v. Occupational Safety and Health Review Commission 499 U.S. 144 (1991)), so failing to form a coalition could result in the complete revo- cation of authority for one of the constituent agencies. These two cases, Coeur Alaska and Martin, jointly stand for the proposition that courts may decide which agency has authority to implement parts of a statue regardless of whether those agencies come to the same interpretation, highlighting the benefits of coalition building in the shadow of litigation. 7 Congress, the president, and the courts each constrain both agencies’ ability to form coalitions and the costs and benefits of engaging in coalition building. Yet contemporary research indicates that the majority of agency coalitions are bureaucratically led, rather than induced by Congress, the president, or the courts (Saito 2020). That is to say, while the three main branches of government constrain the behavior of bureaucrats, they do not determine it unilaterally. Therefore, executive coalition building is an often discretionary action taken by agencies and ought to be the focus of serious scholarly attention since, among other societal benefits, coalitions can help reduce regu- latory redundancies, standardize regulations, and facilitate information sharing among agencies occupying similar regulatory spheres. 8 Below I provide a theory of executive coalition building 7 My argument parallels the legal argument but instead of coalition building serving as a signal of compliance with Chevron or to avoid a court from stripping an agency of its authority, instead it serves as a signal to political overseers that the policy is efficient or best, a criterion not considered by courts when reviewing whether agencies followed the proper procedures or have the proper authority. 8 In consultation with an attorney and expert in administrative law, I also took a random sample of one hundred jointly produced rules in my data and examined whether they were promulgated pursuant to laws that required joint rulemaking. Only five percent of those one hundred rules were promulgated pursuant to laws that require coordination. Most laws did not require coordination either by omitting any requirements for coordination, or with explicit clauses releasing agencies from coordination requirements. For example, the Departments of Agriculture and Transportation promulgated a joint rule in 1996 pursuant to the 42 U.S.C. 106, which explicitly states the “administrator [of the 51 that takes seriously the institutional station of executive agencies in the American separation of powers system. 3.2 Bureaucratic Strategy and Executive Coalitions A growing body of literature on bureaucratic politics argues convincingly that bureaucrats act strategically, particularly with respect to policymaking (Lowande 2019; Potter 2017, 2019; Pot- ter and Shipan 2017; Shipan 2004). Of central concern to bureaucratic agents is the optimiza- tion of their autonomy since agencies derive no formal authority independent of the constitutional branches. Agencies can achieve autonomy through various means such as delaying policymaking until congressional, presidential, or judicial conditions are more favorable (Potter 2017, 2019) or leveraging diverse networks of political support (Carpenter 2001). The techniques to achieve bu- reaucratic policy goals and autonomy uncovered in prior work, however, are all confined to single agencies—they consider how individual agencies respond to political conditions and pursue their policy goals given political and legal constraints (see, e.g., Gailmard and Patty 2012). Overlooked, however, is how agencies might build coalitions and collaborate strategically to optimize their autonomy and achieve their policy goals. Work on networked governance has considered how agencies collaborate with each other and private entities (Freeman and Rossi 2011, 2012; McGuire 2006; Resh, Siddiki and McConnell 2014; Siddiki, Kim and Leach 2017), yet it often fails to consider the political environment in which agencies operate. Several studies do consider how overlapping jurisdictions affect bureau- cratic policymaking, but they either focus on congressional incentives to concentrate or fragment authority (Bils 2019; Farhang and Yaver 2016; Peterson 2018; Ting 2003) or how overlapping ju- risdictions might create inefficiencies like free-riding, turf wars, or preference cycling (Bils 2019; Hammond and Miller 1985; Herrera, Reuben and Ting 2017; Napolio and Peterson 2019; Ting 2003). Here, however, I argue bureaucrats take advantage of overlapping jurisdictions by building Federal Aviation Administration]...shall not be required to coordinate, submit for approval or concurrence, or seek the advice of views of the Secretary [of Transportation] or any other officer or employee.” 52 coalitions in order to forge autonomy and achieve their policy goals. 9 Agencies build coalitions actively to advance their goals in the face of political opposition. Building coalitions when agencies have overlapping jurisdictions provides at least three poten- tial benefits to the constituent agencies’ pursuit of autonomy and policy goals. First, since coalition building involves transaction costs, it may serve as a costly signal to political overseers that the pol- icy resulting from the coalition is particularly important, ripe, or well-supported by the public and therefore induce principals to let the rule stand as a matter of public policy or for electoral con- cerns. In other words, policymaking via coalitions may transmit credible information about the importance or efficiency of policy from a more informed agent to a less informed principal. Sec- ond, collaboration forces overseers, like the President, to distribute any sanction across multiple agencies thereby either diluting its effect on each individual agency or inducing the overseer to raise the severity of the sanction and incur a larger cost, both of which lower the probability that a sanction will have the deterrence effect desired by the principal. Third, coalition building may help agencies make a better or more efficient policy by combining resources and information (see, e.g., Austen-Smith and Banks 1996). In an interview with the researcher in November 2020, a civil servant involved in executive coalition building in a large independent regulatory agency stated that the policies their agency produces jointly are made better by learning from the expertise of the agencies with whom they collaborate, but that the process can be quite cumbersome. Agencies do not experience these coalitional benefits identically, however. Agencies that are ideologically aligned with their overseers can easily promulgate their preferred policies without collaboration (see, e.g., Shipan 2004), so coalition building introduces costs for little or no ben- efit as there is neither a need to signal nor dilute a sanction. Additionally, the transaction costs associated with collaboration almost certainly vary among potential coalitions where factors like ideology and capacity affect the cost of coalition building. Paraphrasing one bureaucrat: it can be difficult to bargain because the missions, motivations, and commitments of agencies can be quite 9 Some work in the public administration literature argues that coalition formation may be driven by power depen- dency, or stronger agencies coercing weaker agencies to collaborate to further the interests of the stronger one (see, e.g., Hjern and Porter 1983). In the appendix, I test this theory and find little evidence for power dependency. 53 different. 10 Together, these imply that those agencies best positioned to build coalitions are those for which the probability of achieving their preferred policies by policymaking on their own is low and the transaction costs associated with collaboration are low. Ideology—or preferences over policy alternatives—affects both of these conditions. Ideologi- cal proximity among agencies lowers transaction costs associated with coalition building and ide- ological distance from overseers decrease the probability that individual policymaking will go unsanctioned. Agencies aligned with each other must give up less to come to consensus since, in spatial terms, the bargaining region between each agency’s ideal points is small when those agencies are aligned and any policy in the bargaining region is relatively close to each agency’s ideal point. Here, I focus on the President as a political overseer since bureaus in the Executive Office of the President like the Office of Information and Regulatory Affairs (OIRA) and Office of Management and Budget (OMB) are the first major political hurdles agencies must face when pursuing policy (Bolton, Potter and Thrower 2016; Haeder and Yackee 2015, 2018; Potter 2017, 2019). Below I expand my theoretical argument, taking seriously the institutional station of bu- reaucratic agents and their desire for autonomy given their subsidiary positions in the American separation of powers system. 3.3 Coalitions as Costly Signals and Insurance Executive agencies are tasked with the implementation of policies passed by the political branches. Often, the political branches create policy with broad strokes, leaving room for bureaucratic inter- pretation to fill in the details. Thus, agencies must use their discretion and expertise to produce policies both consistent with their principals’ intent and their own preferences over policy, whether those preferences come from ideological leanings, professional expectations, expertise, or else- where. Therefore, bureaucrats face a constrained optimization problem where, on the one hand, they want to produce the most efficient policy from their informed perspective and, on the other, 10 Interview with federal civil servant, November 2020. 54 they must not produce policy that is too far from the preferences of their political overseers to avoid sanction or backlash. Bureaucrats often have an informational advantage in their policy area relative to their princi- pals. Bureaucrats can exploit that informational advantage to signal to political overseers that the decision the bureaucrats have made is efficient. For bureaucrats’ decisions to collaborate to serve as an informative signal to principals about the efficiency of their policy choices, it must be the case that, all else equal, it is less costly to collaborate when rulemaking is efficient than when it is not (Spence 1978). The procedural barriers to rulemaking in the Administrative Procedure Act and the Paperwork Reduction Act make the cost of rulemaking quite high since agencies have to engage in notice-and-comment rulemaking and submit their proposed policies to OIRA for review. If agencies wish to produce an inefficient or unnecessary rule, administrative procedures allow for affected parties to alert Congress that the agencies are engaged in superfluous policymaking (McCubbins, Noll and Weingast 1987; McCubbins and Schwartz 1984). Therefore, the cost of convincing other agencies to produce a superfluous policy are likely higher than the cost of con- vincing other agencies to produce a necessary one, as interested parties can easily alert political overseers that the agencies are engaged in regulatory overreach and principals can sanction the agencies. As a concrete example of a situation requiring policymaking, after the Deepwater Horizon oil spill in 2010—an environmental disaster requiring policy production—a coalition comprising the Department of Homeland Security and the Environmental Protection Agency formed a coalition to promulgate a joint rule to “suspend oil spill response time requirements, and certain identification and location requirements, for facilities and vessels whose response resources are relocated in support of the Deepwater Horizon [Spill of National Significance] response” (75 FR 37712). A response was clearly important and efficient and the transaction costs associated with forming a coalition to address the spill were likely low since both agencies knew that a policy response was necessary. 55 However, the cost of coalition building is not solely determined by the expected efficiency of the policy the coalition will produce. Ex ante ideological alignment among agencies also reduces the cost of coalition building. Two liberal agencies, like the EPA and the Department of Health and Human Services, can likely come to consensus about a best policy while incurring lower costs than one liberal and one conservative agency, like the EPA and the Department of Defense, would. In simple spatial models where multiple parties must agree to change policy, the smaller the space between parties, the more likely it is that the bargained outcome will be closer to the ideal points of at least one of those parties. The strategic opportunities for the agents, then, are first to form a coalition when they expect political opposition, second to bluff about the efficiency of the policy to avoid a sanction or, third, if a sanction is inevitable, to form a coalition as insurance to dilute the sanction. Agencies’ in- formational advantages offer them the opportunity to bluff about the efficiency of their policies and convince principals not to sanction a superfluous policy by manipulating principals’ informa- tion about the efficiency of the coalition’s policy choice. The following two sections discuss the implications of the theory through the mechanisms of costly signaling and insurance. 3.3.1 Costly Signals The political decision to delegate requires elected officials to forgo perfect information over the policies they create, setting up the principal-agent problem endemic to bureaucratic politics. Ex- ecutive agencies’ informational advantage provides them with an important means to achieve the policy goals they desire by making either their actions or information partially hidden from prin- cipals. Since bureaucrats have more information and expertise about certain policy areas than do elected officials, there are incentives for bureaucrats to provide biased information to principals 56 in order to move policy in the direction bureaucrats desire. 11 The incentive to provide biased in- formation, however, is only present when truthfully revealing information would result in a worse outcome for the agency. The information bureaucrats can provide to principals varies but often comprises technical in- formation about which policy or what level of regulation is most efficient or would result in the best social outcome. For example, the Centers for Disease Control and Prevention can provide information about the appropriate response to the outbreak of a pandemic, the Environmental Pro- tection Agency can provide information about the most efficient reduction in hydrofluorocarbon production among major firms, and the Department of Veterans Affairs can provide information about the appropriate number of beds per capita to allocate to V A hospitals. Occasionally, this information is transmitted directly from agencies via testimony or other for- mal communication so that Congress or the President can create an appropriate policy. Often, however, principals have delegated policymaking authority to agencies and installed procedural technologies to oversee agency policymaking processes and sanction agents who stray too far (McCubbins, Noll and Weingast 1987). In these situations, the policy itself, and the process used to create it, can convey information to political principals. Political and oversight decisions con- cerning the merit of delegated policymaking then occur after promulgation rather than before a policy is made. In situations where the process and policy themselves convey information, bureaucrats can signal strategically the efficiency, importance, ripeness, or appropriateness of the policies they produce. One way to do this might be through a cost-benefit analysis. For example, if the De- partment of Veterans Affairs wants to produce a rule that would allocate more beds to a district controlled by a member of Congress that has antagonized the President, the agency might conduct a cost-benefit analysis to signal to the President that the decision is efficient, even if the President would otherwise prefer more beds to be allocated to a politically friendly district. 11 The existence of such a bias, however, need not be undemocratic or inefficient. As Miller and Whitford (2016) argue, bureaucratic discretion—even when it results in policies that conflict with what a legislative majority might enact—is often efficient. 57 I argue that forming a coalition sends a similar signal about the efficiency, importance, ripeness, or appropriateness of bureaucratic policies. Rather than forming coalitions to maximize distribu- tive benefits or overcome social choice problems like legislators do, executive coalition building sends a credible signal to political overseers that the agencies’ expertise ought to be respected, and therefore that the coalition of agencies ought to be afforded the autonomy to produce policies that perhaps are not their overseers’ most-preferred alternative. However, like conducting a cost-benefit analysis, forming a coalition entails a cost. Agencies must seek out another agency with which to form a coalition, convince the other agency to form a coalition, and bargain over the policy that coalition will produce. In an interview with the re- searcher, a civil servant stated that the process for producing joint policy can be quite onerous. 12 First, a working group of regulators from each agency convenes to produce a draft of a regulation, then the working group sends it up the chain of command of both agencies for approval, next the group must reconvene to incorporate any edits from the senior civil servants of each agency. This process is continued until all relevant actors are satisfied with the policy. These procedures are layered on top of the already taxing process for promulgating rules. 13 Given the additional costs associated with executive coalition building, building coalitions to signal the appropriateness of a policy should only be undertaken by agencies when they are unlikely to achieve autonomy if they act alone, that is, when political overseers hold conflicting preferences over some set of alterna- tives. Therefore, misaligned preferences between agents and principals is a necessary condition for coalition building to be a best response. Misaligned political overseers, however, is not sufficient for executive coalition building to be a best response on the part of the agencies. As a simple counterexample, if two agencies are misaligned with their political overseers but the two agencies are extremely misaligned with 12 Interview with federal civil servant, November 2020. 13 As a basic check to validate the assumption that forming a coalition to make policy is costly, I consulted Potter’s (2017) dataset of time from when a rule is proposed to the time that it is finalized. The average time to finalization for rules promulgated by only one agency is 15 months (standard error of 0.11 months) and for rules promulgated by multiple agencies is 21.3 months (standard error of 2.99 months), implying that it takes more time and is costlier to form coalitions than it is to promulgate a rule individually. In fact, the mean time to finalization for rules promulgated by coalitions is longer than 81% of the time to finalization of rules promulgated individually. 58 each other, the cost of forming a coalition will be greater than any potential benefit from avoiding a sanction and the agencies will elect not to form a coalition. Therefore, the cost of coalition building must be sufficiently low as well. If the transaction costs associated with coalition building exceed the potential sanction, agencies are better off not coordinating. Even if agencies could increase the probability their policy goes unsanctioned by forming a coalition, if that increase is offset by the high cost of forming a coalition, the agencies are better off not collaborating. Therefore, the confluence of low transaction costs and misalignment with political overseers is necessary for agents to form policymaking coalitions. 3.3.2 Insurance: Sanction Dilution Executive coalitions may also be useful not as a costly signal, but as a means of diluting a likely sanction. If the principal is sufficiently misaligned with the agencies or has sufficiently high beliefs that the policy is not efficient, it will always sanction regardless of the signal it receives. Therefore, under certain conditions, agencies will form a coalition not to signal that the policy is efficient, but rather to brace themselves for the inevitable sanction and to dilute it by sharing the cost. Specifi- cally, if agencies are extremely misaligned with their political overseers but nonetheless face pres- sure from interest groups or the public to produce a policy, the agencies can reasonably be sure that they will be sanctioned if they make policy. Thus, anticipating a sanction, agencies can form a coalition as a sort of insurance to spread risk among multiple agencies and dilute a sanction. As with costly signaling, the incentive to form a coalition to dilute a sanction varies with the transaction costs associated with forming a coalition. Specifically, the transaction costs associated with forming a coalition must be less than the cost of the diluted sanction for the constituent agencies. Therefore, like in the case of costly signaling, principal misalignment and low transaction costs are both necessary for agents to form coalitions. When agencies are extremely misaligned with principals, coalitions are likely motivated by sanction dilution. 59 3.3.3 Empirical Implications The theory above implies two relevant hypotheses for agency behavior. First, agency decisions to build coalitions vary with the transaction costs associated with collaboration. As transaction costs rise, the probability of coalition building decreases. All else equal, as transaction costs rise coali- tions become increasingly unsustainable. On the other hand, all else equal, if transaction costs decrease, coalition building becomes more likely. Transaction costs are almost certainly lower among agencies with similar preferences over policies. Agencies with dissimilar policy prefer- ences have to compromise more than those with similar preferences since the policy dissimilar agencies can both agree too is likely farther from either agency than the compromise policy among agencies with similar preferences. Therefore: Hypothesis 1 Ideologically proximate agencies are more likely to form coalitions than ideologi- cally distant ones. Second, agency misalignment with the principal is a necessary condition for coalition building. In the case of costly signals, the agents must need to send a signal to convince their principals that they should let the coalition’s policy stand. If the principal agreed to the policy ex ante, the need to signal would be obviated since the principal would let the policy stand without sanction. And in the case of sanction dilution, only when the agencies are misaligned with the principal do they rationally expect that they cannot avoid a sanction. Therefore: Hypothesis 2 Among ideologically proximate agencies, those ideologically distant from the Pres- ident are the most likely to form coalitions. Hypothesis 2 is this chapter’s main theoretical contribution since it implies agencies form coali- tions only if they are misaligned with political principals in order to optimize their autonomy given their subsidiary positions in the American separation of powers system. Whereas legislators en- ter into coalitions to maximize individual distributive gains and overcome social choice problems, agents in the executive branch enter into coalitions to signal to overseers that their autonomy ought 60 to be respected, despite their dependence on political overseers who may otherwise disagree with the policy output of bureaucratic authority. In other words, upon observing political conditions un- favorable to their goals, agencies form coalitions to get what they want, but only if the transaction costs associated with collaboration are sufficiently low. The theory also implies one relevant hypothesis for presidential behavior. Since agencies mod- erately misaligned with the President are best positioned to signal successfully that their policies are efficient, the President should be least likely to sanction policies produced by coalitions mod- erately misaligned with the President. This implies the last hypothesis: Hypothesis 3 The relationship between presidential misalignment and the probability of sanction is U-shaped with a minimum for moderately misaligned agencies. 3.4 Data and Empirical Analysis Before presenting the results from my main analyses testing my theory, I provide several descrip- tive findings from the coalition network I generated with data from the Federal Register. The dataset comprises 496 pairs of thirty-two agencies over five presidential administrations. I define coalitions as groups of agencies aggregated to their highest levels that promulgate at least one joint rule in a presidential term. For example, a pair comprising the Agriculture Marketing Service and Agricultural Research Service, both in the Department of Agriculture, does not constitute a coalition, but a pair comprising the Agricultural Marketing Service and the Bureau of Economic Analysis in the Department of Commerce does. This suggests that the counts I have produced here are somewhat conservative, but using lower levels of agencies as units would likely present a confound in my analysis since sub-bureau independence from their parent agencies and authority to engage in rulemaking vary. With these data from the Federal Register, I then constructed a coalition network where each node or vertex is an agency and each edge or tie is the count of rules jointly promulgated by the 61 coalition comprising the two node agencies at any point from 1997–2012. 14 Eight of the thirty two agencies never formed a coalition from 1997–2012, but the pooled network density among the remaining thirty-two agencies is quite high at about 68%, meaning more than two thirds of all possible agency pairs formed a coalition together from 1997–2012. When including all agencies, including those that never formed a coalition, the network density is about 47%. 0.0 0.2 0.4 0.6 Cabinet-Cabinet Cabinet-Independent Independent-Independent Probability of Coalition by Agency Type Combination Figure 3.1: Coalition Formation by Agency Type Combination, 1997-2012. Probability calculated as the proportion of dyad-years of each combination that formed a coalition. Eighteen percent of observations are Cabinet-Cabinet, 51% are Cabinet-Independent, and 31% are Independent- Independent. Figure 3.1 displays the probability of coalition building for pairs of agency structures. Cabinet departments, on average, are more likely to form coalitions than other types of agencies, partic- ularly with other cabinet agencies. However, the network generally shows little evidence of ho- mophily, or the tendency of like units to form coalitions, with respect to agency structure. In fact, the assortativity coefficient—a measure of homophily which ranges from − 1 (if only dissimilar nodes form ties) to 1 (if only similar nodes form ties)—is only− 0.09, indicating that there is little support for homophily with respect to agency structure. Instead, the relatively large probability of cabinet departments forming coalitions with other cabinet departments is an artifact of cabinet 14 I limit the dataset I use for the main analysis to these terms since I am only able to collect sufficient data for my analysis during those years. Ideal point estimates at the agency level are only available through 2012. 62 departments’ general predisposition toward collaboration, perhaps due to their broader jurisdic- tions. 15 Table 3.1: Most and Least Central Agencies Agency Degree Agency Betweenness DOC 23 OPM 106 DOT 23 SSA 69 V A 23 CPSC 42 DOD 22 DHS 28 DOJ 22 TREAS 26 . . . . . . . . . . . . CPCS 12 RRB 0.50 RRB 4 NRC 0.42 EEOC 3 FCC 0.33 FCC 2 FTC 0 FTC 1 EEOC 0 Note: Table only includes agencies that formed at least one coalition between 1997 and 2012. Eight agencies never formed a coalition. Key: DOC = Department of Commerce; DOT = Department of Transportation; V A = Veterans Affairs Department; DOD = Department of Defense; DOJ = Department of Justice; CPSC = Con- sumer Product Safety Commission; RRB = Railroad Retirement Board; EEOC = Equal Employ- ment Opportunity Commission; FCC = Federal Communications Commission; FTC = Federal Trade Commission; OPM = Office of Personnel Management; SSA = Social Security Administra- tion; DHS = Department of Homeland Security; TREAS = Treasury Department; NRC = Nuclear Regulatory Commission. Next, I turn to agency-level measures. Table 3.1 displays degree and betweenness measures of centrality for the five most and least central agencies in the pooled network. The most central agency by degree—the number of unique connected agencies—is the Department of the Treasury and the most central agency by betweenness—a measure of how well each agency connects other 15 Formally, the assortativity coefficient is calculated as: Tr e−|| e 2 || 1−|| e 2 || where e is an adjacency matrix where each row and column is a dyad type and each entry is the proportion of edges realized among that dyad type. Tr e, the trace of the matrix, is the main diagonal where each entry is the proportion of edges realized among dyad types where each node belongs to the same dyad type. An assortativity coefficient of 1 would indicate that only like nodes form ties and an assortativity coefficient of -1 would indicate that nodes only form ties with dissimilar nodes. 63 agencies to the network—is the Office of Personnel Management. 16 The most central agencies uncover patterns that largely comport with conventional wisdom about the importance of different agencies. Agencies with the highest degree measure are mostly cabinet departments, which have broad jurisdictions and therefore likely have more opportunities to form coalitions. The agency with the highest betweenness measure is OPM, an agency that manages the US civil service. Clinton II Bush I Bush II Obama I 0 100 200 0 100 200 0 100 200 0 100 200 0 % 25 % 50 % 75 % 100 % Count of Laws Delegating to Both Agencies in Dyad Probability of Coalition Formation Figure 3.2: Probability of Coalition Formation by Number of Overlapping Laws. Points represent dyads. Curves and ribbons estimated with bivariate logistic regression. Last, I consider how overlapping jurisdictions influence the opportunity structure for coalition formation. As implied by the institutional powers granted to bureaucratic agencies, the opportunity to form coalitions should be limited only to those agencies with overlapping jurisdictions. To test this, I found the number of significant laws since 1947 that delegate to each agency dyad in each presidential term from McCann and Shipan (2022). No dyads with no overlapping laws formed a coalition from 1997–2012, and the probability of coalition formation is increasing in the number of overlapping laws, as evidenced by figure 3.2. 16 Formally, betweenness is calculated as: B(v)= ∑ s̸=v̸=t σ st (v) σ st where σ st is the number of shortest paths between nodes s and t and σ st (v) is the number of those paths that pass through node v. 64 3.5 Empirical Test of the Theory With these data from the Federal Register, I then created a panel dataset where each observation is an agency dyad during a presidential term. The data comprise 1,953 observations at the dyad- term level. In total, the analysis comprises thirty-two agencies, 496 dyads, and four presidential terms. Thirty one of the agencies appear each term (the only one that does not is the Department of Homeland Security, which was created during Bush’s first term). Combining each of the agencies that appear in all terms into pairs yields 465 dyads ( 31! 29!2! ) which when multiplied by four terms is 1,860. Then, the Department of Homeland Security forms a pair with each of the remaining thirty- one agencies over the three terms it was in operation, adding 93 to the number of observations to arrive at the final dataset of 1,953 observations. The dependent variable is a binary indicator of whether each dyad-term formed a Coalition to promulgate a joint rule, which I define as any coalition of two or more agencies from dif- ferent department-level organizations (e.g., Department of Agriculture, Environmental Protection Agency, Department of Defense). 17 The two independent variables of interest are agency alignment and presidential misalignment. I measure agency alignment as the Euclidean distance between the Chen and Johnson (2015) 18 ideal point estimates of agency ideology of the two agencies forming the dyad in each term multi- plied by negative one so that larger values represent more alignment among agencies in a dyad. I measure presidential misalignment as the average Euclidean distance between the Chen and John- son (2015) ideal point estimates of agency ideology of the two agencies forming the dyad and the President’s DW-NOMINATE ideal point estimate in each term. 19 17 All analyses are robust to an alternative dependent variable measuring the count of rules promulgated by each dyad-term coalition (see Appendix). 18 These are estimated on the same scale as DW-NOMINATE. 19 The anlysis is robust to an alternative operationalization of presidential misalignment as the distance between the agency closest to the President and the President. The appendix reports results from estimating the main models with this alternative operationalization. One concern with this measure is that if the President’s ideal point is between two agencies’, the average distance may be close to zero despite both agencies being distant from the President. However, there are no cases of the President between two agencies in the data. 65 I also include six control variables. The first set of control variables adjusts estimates for the political context and environment. First, I include overlapping laws, which I measure as the number of significant laws since 1947 that delegate to both dyads in a given presidential term (Mayhew 2005; Peterson 2018), and which controls for the statutory opportunity to form coalitions, proxying for congressionally mandated coordination. Second, I include presidential attention, which I measure as the logged count of presidential documents published in the Federal Register (e.g., executive orders, memoranda) that mention at least one of the agencies forming the dyad, which controls for how important those agencies are to the president’s agenda and proxies for presidentially mandated coordination. Third, I include House misalignment, which I measure the same way as presidential misalignment, but substituting the President’s DW-NOMINATE ideal point estimate for the US House of Representative’s median member’s DW-NOMINATE ideal point estimate, which controls for the expected congressional response to agency policy. Fourth, I include court misalignment, which I measure the same way as presidential misalignment, but substituting the President’s DW-NOMINATE ideal point estimate for the judicial common space ideal point estimate of the median justice of the Supreme Court (Epstein et al. 2007), and which controls for the expected judicial response to agency policy. The second set of control variables adjusts estimates for the relationship and similarities be- tween agencies forming the dyad. First, I include total rules, which I measure as the logged count of the total number of rules individually promulgated by each agency in a given presiden- tial term, and which controls for the baseline productivity of the agency-dyad. Second, I include employment difference, which I measure as the natural log of the absolute value of the difference between the total number of employees working within each agency forming the dyad in each term, which controls for the difference in capacity between the agencies forming the dyad. Third, I include politicization difference, which I measure as the absolute value of the difference between the politicization ratio of each agency forming the dyad in each term, and average politicization, 66 which I measure as the average politicization ration for the two agencies forming the dyad, 20 which controls for the difference in presidential attempts to control agency policy. 21 With these data, I first center agency alignment and presidential misalignment to zero as their means to ease in interpretation of the interaction effects, then I estimate the following general model via least squares: 22 Pr(Coalition it )=β 1 Agency Alignment it +β 2 Presidential Misalignment it + β 3 Agency Alignment it × Presidential Misalignment it +ξ ξ ξ X it +α α α i +δ δ δ t +ε it (3.1) where subscript i indexes agency dyads, subscript t indexes presidential terms, X is a matrix of dyad-level, time-varying covariates, ξ ξ ξ is a vector of coefficients attending X, α α α is a vector of agency dyad fixed effects, and δ δ δ is a vector of presidential term fixed effects. Standard errors are clustered by dyad. Hypothesis 1 implies the average effect of agency alignment is positive and hypothesis 2 implies the marginal effect of agency alignment is increasing in presidential misalignment, implyingβ 3 > 0. 23 This empirical strategy allows me to identify the effects of the independent variables within each dyad and absorbing any term-level exogenous shocks. The dyad fixed effects control for any time-invariant features of each agency in the dyad and the relationship between those agencies, 20 I measure politicization as the ratio of political appointees over the number of career senior executive service members following previous work (see, e.g., Lewis 2010; Lowande 2019). 21 Since many of these variables vary each quarter, year, or Congress, I take the average value of each over the full presidential term. All analyses are robust to estimating models at the dyad-year level where these vary more frequently (see appendix). 22 I use least squares, i.e., a linear probability model, for a few reasons. First, since I use dyad and year fixed effects, any maximum likelihood-based approach like logistic or probit regression would drop any year or dyad that never or always featured a joint rule, biasing the dataset (Beck 2018, 2020; Rodr´ ıguez and Goldman 1995). Second, the well- known problems with using least squares on binary dependent variables (heteroskedasticity, unrealistic predictions, and bias in small samples) are all inconsequential since I use heteroskedasticity-robust standard errors, am not interested in fitted values, but rather estimated coefficients, and the sample is quite large (see, e.g., Hellevik 2009; Wooldridge 2010). Third, logistic and probit regression render the interpretation of interaction terms unclear (Ai and Norton 2003). Last, parameters estimated with logistic regression are biased when variables are omitted even if the omitted variables are uncorrelated with the variable of interest (Mood 2010). That said, the results are robust to estimating a logistic regression (see appendix). 23 Formally, hypothesis 2 implies the mixed partial derivative of the function first with respect to agency alignment then with respect to presidential misalignment is greater than 0, orβ 3 > 0. 67 such as the structure of each agency, whether they meet the necessary condition for coordination by having overlapping jurisdictions, 24 and unobservable aspects of their working relationship, al- lowing me to estimate within-dyad effects. The presidential term fixed effects control for any common exogenous shocks such as the terrorist attacks in 2001, the financial crash in late 2000s, first and second term effects, and the unique administrative styles of each President. 3.5.1 Results First, table 3.2 reports the cross-tabulated probabilities of coalition formation splitting the sam- ple into four categories: ideologically aligned agencies aligned with the President, ideologically aligned agencies misaligned with the President, ideologically misaligned agencies aligned with the President, and ideologically misaligned agencies misaligned with the President. Observations are classified as misaligned if the value of the relevant variable is below the mean of alignment, and aligned otherwise. My theory implies that the probability of coalition formation should be greater when agencies are misaligned with the President, and should be highest for aligned agencies that are misaligned with the President. Table 3.2 shows that, on average, the probability of coalition formation is greater when agencies are misaligned with the President (0.277) rather than aligned (0.212), consistent with the implica- tions of the theory. The table also shows that the probability of coalition formation is greatest when agencies are misaligned with the President but aligned with each other (0.309), again consistent with the theory. The probability of coalition formation among aligned agencies with misaligned presidents is significantly higher than the other three conditions ( p < 0.001) and the other three conditions are statistically indistinguishable. Of course, cross-tabulated probabilities do not ac- count for confounds or the grouped structure of the data; therefore, below I present the results from the fully specified linear probability model from equation 3.1. 24 While this strategy almost certainly includes irrelevant dyads that share no overlapping jurisdictions, doing so will skew any coefficient toward zero and therefore not raise the probability of false positives. Further in the ap- pendix, I subset the data only to dyads with at least one law delegating to both agencies since 1947 and the results are substantively unchanged. 68 Table 3.2: Cross-Tabulation of Probability of Coalition Formation President Aligned Misaligned Agencies Aligned 0.199 0.309 (0.016) (0.019) Misaligned 0.236 0.226 (0.023) (0.022) Note: Cell entries report the probability of coalition formation. Standard errors reported in paren- theses. Observations are classified as misaligned if the value of the relevant variable is below the mean of alignment, and aligned otherwise. Probability calculated as the proportion of dyad-term’s forming coalitions in each of the four combinations of agency and presidential (mis)alignment. The probability of coalition formation among aligned agencies with misaligned presidents is sig- nificantly distinguishable from the other three conditions at the 0.001 level, and the other three conditions are statistically indistinguishable. Table 3.3 displays results from estimating variants of the general equation presented in equation 3.1. Models 1 and 3 do not include the interaction term and indicate that, on average, ideologically aligned agencies are more likely to form coalitions, although the coefficient on agency alignment is estimated with high uncertainty leaving me unable to reject the null for Hypothesis 1. However, my main hypothesis is a conditional one: when agencies are ideologically close to each other and ideologically distant from the President they will form coalitions because presidential misalign- ment is a necessary condition for agencies to turn to coalition building to optimize their autonomy given their subsidiary position in the American separation of powers system. 25 Models 2 and 4 include the interaction term between agency alignment and presidential misalignment. Interpreting these models is complicated since the parameter of interest results from the in- teraction of two continuous variable. To aid in interpretation, Figure 3.3 displays the estimated marginal effect (i.e., the change in the probability of coalition formation from a unit increase in the independent variable) of agency alignment at all observed values of presidential misalignment.The marginal effects plot, derived from model 4, supports Hypothesis 2 since the marginal effect of 25 Although the coefficient on presidential misalignment is negative and significant across all four models, any other specification or aggregation of the data renders the coefficient indistinguishable from zero, so I do not put much weight on the estimate (see appendix). 69 Table 3.3: Executive Coalition Building Dependent variable: Coalition Formation (1) (2) (3) (4) Agency 0.094 0.088 − 0.009 − 0.047 Alignment (0.070) (0.068) (0.087) (0.089) Presidential − 0.229 ∗ − 0.296 ∗∗ − 0.252 ∗ − 0.311 ∗∗ Misalignment (0.098) (0.100) (0.101) (0.102) Agency Alignment× 0.974 ∗∗∗ 1.225 ∗∗∗ Pres. Misalignment (0.267) (0.275) Overlapping − 0.009 ∗∗∗ − 0.010 ∗∗∗ Laws (0.002) (0.002) Presidential − 0.044 ∗ − 0.046 ∗ Attention (0.020) (0.020) House 0.274 0.159 Misalignment (0.267) (0.271) Court 0.073 0.233 Misalignment (0.221) (0.227) Log(Total Rules) − 0.096 ∗ − 0.114 ∗∗ (0.040) (0.039) Employment − 0.005 − 0.015 Difference (0.044) (0.044) Politicization − 0.017 − 0.018 Difference (0.070) (0.070) Average − 0.025 − 0.019 Politicization (0.016) (0.016) Observations 1,953 1,953 1,953 1,953 Dyad & Term FEs YES YES YES YES Adjusted R 2 0.453 0.458 0.480 0.486 ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001 Note: Unit of analysis is the dyad-year. Heteroskedasticity-corrected standard errors clustered by dyad reported in parentheses. 70 One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment Marginal Effect of Agency Alignment One Standard Deviation above the Mean of Presidential Misalignment -1.00 -0.75 -0.50 -0.25 0.00 0.25 0.50 0.75 1.00 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 -0.50 -0.25 0.00 0.25 0.50 Presidential Misalignment (Centered on Zero) Change in Probability of Coalition Formation Given Unit Increase in Agency Alignment Change in Probability of Coalition Formation Given Standard Within-Dyad Increase in Agency Alignment Marginal Effect of Agency Alignment on Probability of Coalition Formation Figure 3.3: Marginal Effect of Agency Alignment. The rug on the x-axis displays the density of presidential misalignment. Marginal effects estimated from model 4 in table 3.3. The left y-axis plots the change in the predicted probability of coalition formation given a unit increase in agency alignment at the value of presidential misalignment on the x-axis, and the right y-axis displays the change in probability of coalition formation given a standard within-dyad change in agency alignment at the value of presidential misalignment on the x-axis. The white point indicates the marginal effect of agency alignment when presidential misalignment is one standard deviation above the mean and the horizontal line connects the point to the two y-axes to ease interpretation. 71 agency alignment on coalition formation is increasing in presidential misalignment and is positive above the mean of presidential misalignment. When two agencies are both ideologically distant from the President, they become more likely to form a coalition as the two agencies themselves become ideologically closer to each other. At one standard deviation above the mean of presidential misalignment, the marginal effect of agency alignment on the probability of a joint rule is about 0.4, which, when scaled by the typical within- dyad change in agency alignment indicates about a 8.8 percentage point increase in the likelihood of collaboration, about a 36% increase from the mean probability of promulgation, from a 24.5% probability of joint promulgation to a probability of 33.3%, which corresponds to about forty four additional dyads forming coalitions, on average, each presidential term. Figure 3.3 also shows that agency alignment does not have a positive effect on the probability of coalition formation when two agencies are both aligned with the President. These findings are consistent with my theory that agencies are only incentivized to collaborate when they are both ideologically distant from the President in order to signal importance or efficiency and dilute sanctions. When agencies are ideologically aligned with the President, any rule they promulgate will likely face little opposition from the President or OIRA, obviating the need to take on the additional transaction costs associated with forming a coalition with another agency. 3.5.2 Mechanism Test The results presented above are consistent with the theory that agencies form coalitions in part as signals that their policies are efficient and should be maintained by political overseers and in part as insurance to dilute an inevitable sanction. Both mechanisms, signalling and insurance, imply that misalignment with the president and alignment among the agencies are necessary for agencies to prefer coalition building to solo policymaking; that is to say, both mechanisms are observationally equivalent when analyzing only agency behavior. But the two mechanisms imply agencies should expect different behavior from the president at different levels of misalignment. Signals are meant to avoid a sanction whereas insurance is meant to reduce the cost of oversight 72 by sharing the burden with coalition partners. Although agencies should form coalitions when they are misaligned with the president, they should expect the president, through OIRA, not to request regulations produced by coalitions be changed when agencies are slightly misaligned with the president since the coalitions send a signal of efficiency; but when agencies are extremely misaligned with the president, agencies should expect the president to request revisions to their proposed policies and form coalitions instead as insurance. To test this mechanistic expectation, I collected data on OIRA regulatory review from the Unified Agenda to analyze first whether OIRA is more likely to allow policies promulgated by coalitions to move along the regulatory process without revision, and second whether OIRA is least likely to request regulatory revisions when agencies form coalitions and are moderately misaligned with the president. The results provide evidence that coalitions do in fact send credible signals of efficiency. OIRA is less likely to request regulatory revisions to policies promulgated by coalitions. The results also provide evidence consistent with both the signaling and insurance mechanisms of the theory as the relationship between OIRA review of policies promulgated by coalitions is U- shaped, with OIRA least likely to request revisions of policies promulgated via coalition from agencies moderately misaligned with the President. 26 Table 3.4 presents ordinary least squares regression results from regressing a binary indicator for whether OIRA requested revisions to a rule on a binary indicator for whether that rule was promulgated by a coalition or not. Models one and two include year fixed effects, and model two additionally controls for whether OIRA deemed the rule economically significant, major, or whether it implicates federalism. On average, OIRA requests regulatory revisions from coalitions about 13.5 percentage points less than from agencies that promulgate rules on their own. This 26 Coalitions among agencies aligned with the president should not occur in equilibrium if agencies only form coalitions as signals or insurance since aligned agencies neither need to signal nor insure themselves since the president likely agrees ex ante with policies they produce. Therefore, when agencies aligned with the president form coalitions, the president relies on their prior belief that the policy is efficient and may sanction if that belief is sufficiently low. On the other hand, when agencies are moderately misaligned and the president observes a coalition, they never sanction in equilibrium. So, while counter-intuitive, presidents should be more likely to sanction agencies that form coalitions that are aligned with the president than those that are moderately misaligned. 73 Table 3.4: OIRA Less Likely to Request Regulatory Changes from Coalitions Dependent variable: OIRA Requests Change (1) (2) Produced by − 0.138 ∗ − 0.134 ∗ Coalition (0.057) (0.055) Economically 0.041 Significant (0.028) Major 0.067 ∗∗ (0.023) Federalism 0.017 Implications (0.031) Observations 8,622 8,622 Year FEs YES YES Adjusted R 2 0.099 0.106 ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001 Note: Unit of analysis is the rule. Heteroskedasticity-corrected standard errors clustered by year reported in parentheses. estimate is substantively large as it is about twice the magnitude of the effect of major regulations, those that are likely to result in an annual economic effect of at least one million dollars. Table 3.4 provides evidence that OIRA is more deferential to policies produced by coalitions, but more can be extracted from the data. The theory predicts that, when forming coalitions, agen- cies should expect little resistance when moderately misaligned with the president since the coali- tion can send a credible signal, yet agencies should expect OIRA to meddle in the regulatory process when they are extremely misaligned with the president. Therefore, table 3.5 presents both ordinary least squares and Heckman selection model results from regressing the proportion of policies an agency produced in a presidential term via coalition for which OIRA requested revi- sions on the distance and squared distance between each agency’s Chen and Johnson (2015) ideal point estimate and the president’s DW-NOMIANTE ideal point estimate for each presidential term. Since not all agencies formed a coalition each presidential term leading to an undefined propor- tion of OIRA requests—and the preceding analysis makes clear that decisions to form coalitions 74 Table 3.5: Presidential Misalignment and OIRA Review of Coalitions Dependent variable: Pr(Change Request): Coalition Policy OLS Heckman selection (1) (2) (3) (4) Presidential 0.110 ∗ 0.106 ∗ 0.107 ∗ 0.102 ∗ Misalignment (0.049) (0.048) (0.053) (0.052) Presidential 0.168 ∗ 0.210 ∗ Misalignment 2 (0.074) (0.085) Politicization − 0.222 − 0.227 − 0.158 − 0.078 Ratio (0.133) (0.145) (0.168) (0.166) Pr(Change Request): 0.445 0.226 0.436 0.141 Solo Policy (0.284) (0.345) (0.232) (0.255) Observations 42 42 83 83 Adjusted R 2 0.161 0.223 0.146 0.235 ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001 Note: Unit of analysis is the agency-presidential term. Models 1 and 2 report heteroskedasticity- corrected errors clustered by agency. Models 3 and 4 report the second stage of a Heckman selec- tion model as described in the text. 75 are strategic—the Heckman models allow me to account for selection into the sample of agencies that formed coalitions. The first step of the Heckman models estimate via probit regression the probability that an agency enters a coalition as a function of presidential misalignment, politiciza- tion, and agency and presidential term fixed effects and the second stage estimates the relationship between presidential misalignment and OIRA’s review rate. Models 1 and 3 do not include the polynomial and instead only estimate a linear effect which comports with a basic spatial model of oversight where OIRA requests changes from misaligned agencies. However, the theory proposed in this chapter predicts a nonlinear effect, so models 2 and 4 include the quadratic term. Each model additionally controls for each agency’s politicization ratio and the baseline probability that OIRA requests changes to policies produced by each agency. To ease interpretation of the polynomial models, figure 3.4 displays predicted values from 175,000 simulations of the second stage of the Heckman model summarized in model 4 of table 3.5. Consis- tent with the theory, OIRA requests that policies promulgated by coalitions of agencies moderately misaligned with the president be reviewed much less frequently than those promulgated by coali- tions that are either aligned or extremely misaligned with the president. The insurance mechanism helps explain the counterintuitive finding that agencies extremely misaligned with the president take on the additional costs of coalition building even when there is an almost one-hundred percent chance that they will face costly resistance from OIRA. By forming a coalition, each agency can dilute OIRA’s costly resistance that they know is incoming. Taken together, the analyses of why agencies form coalitions and how OIRA evaluates policies formed by coalitions are consistent with the theory that agencies form coalitions as signals about the efficiency of their policies or as insurance to dilute oversight. Agencies form coalitions when misaligned with political principals in order to convince principals that agency autonomy ought to be respected. If agencies are extremely misaligned with principals, however, they form coalitions not to signal but rather to insure themselves against an incoming sanction. 76 0.00 0.25 0.50 0.75 1.00 1.25 1.50 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 Presidential Misalignment (Centered on Zero) Predicted Probability of OIRA Review of Coalition Policy Figure 3.4: Relationship between Presidential Misalignment and OIRA Review of Coalitions. Rib- bon represents 95% confidence interval of simulations. Rug along x-axis displays the density of presidential misalignment. 77 3.5.3 Limitations Although the empirical results presented above are consistent with the theory presented in this paper, there remains an empirical limitation that is important to highlight. It is possible that joint rulemaking is required by some external actor (e.g., Congress, the president, or a court). To ac- count for this, I consulted McCann and Shipan (2022) to see how many laws delegate to the two agencies forming the coalition, tested a random sample of joint rules to see whether they cited a law requiring collaboration, and conducted interviews with federal bureaucrats to learn about the processes used to produce joint rules. These measures, however, are circumstantial – and the interviews do not provide a representative look into joint rulemaking procedures across the fed- eral government. We do not know for certain whether the formation of a particular coalition was mandatory. Ideally, for each dyad-year, the data would tell us whether there was some statute or unfulfilled delegated authority that the agencies could draw on. Future work should attempt to link each rule to its express statutory authority and determine whether that rule was indeed produced voluntarily by the agencies. 27 3.6 Conclusion Executive coalitions are commonplace in American politics. Coalitions responded to the coron- avirus pandemic in 2020, 28 the Deepwater Horizon spill in 2010 (75 FR 37712), the enforcement of Prohibition in the 1920s, and the management of reservations for indigenous people in the 1840s (Kaiser 2011). Despite their ubiquity in American politics, executive coalitions have received little attention from political scientists. This chapter represents an attempt to develop and test a theory of why executive coalitions form. 27 See Peterson and Napolio (Forthcoming) for a paper that does link rules to statutory authority, but does not test whether rules were mandatory. 28 Sun, Lenna H., Josh Dawsey and William Wan. “CDC, FEMA have created a plan to reopen America. Here’s what it says.” Washington Post (April 14, 2020). https://www.washingtonpost.com/health/2020/04/14/ cdc-fema-have-created-plan-reopen-america-heres-what-it-says/ 78 In this chapter, I have advanced a theory of executive coalition building that takes seriously the unique environment in which bureaucratic agencies operate. In contrast to legislative coalition building—where the goal is to maximize particularistic benefits to discrete and mutually exclu- sive constituencies or overcome social choice problems (Aldrich 2011; Baron 1989; Baron and Ferejohn 1989)—or technocratic explanations for interagency coordination—where the goal is in- formation sharing or the agencies simply implement presidentially induced coordination (Freeman and Rossi 2011, 2012; Saito 2020)—I argue that agencies build coalitions as costly signals to over- seers about the efficiency of their policy choices in order to optimize their autonomy given their subsidiary positions in the federal government. Incurring the costs associated with coalition build- ing, however, is only worthwhile when failing to form a coalition and acting alone is likely to be met with a political sanction. Therefore, misaligned preferences over policy alternatives between the agencies and their principal is a necessary condition for collaboration. The empirical analyses attending the model provide support for the strategic theory I advanced. Leveraging a new dataset of agency coalitions, I have shown that agencies form coalitions when the transaction costs associated with coalition building are low and the probability of individual policymaking going unsanctioned is low. This chapter represents a break with extant work on bureaucratic politics by considering networks of agencies rather than studying agencies in isolation either through in depth case studies or cross-sectional analyses. As with all social scientific theories that derive empirical implications from theory, the mech- anisms explicated in the theory—costly signaling and sanction dilution—are only sufficient for the outcomes of interest and therefore cannot explain every coalition ever formed in the federal executive branch. Agencies may form coalitions for more technocratic reasons like information sharing and reducing redundancies or they may be induced to form coalitions by political princi- pals or interest groups. They may also form coalitions for idiosyncratic reasons. In this chapter, however, I have detailed and defended a strategic and political explanation for coalition building in the executive branch and found robust empirical evidence across diverse agencies that such a mechanism both exists and can explain a significant portion of executive coalitions. 79 The theory I have advanced here implies several other empirical implications that future work ought to consider. Since I argue that one of the main purposes for coalition formation in the exec- utive branch is to manipulate the beliefs of political overseers and convince them that bureaucratic policy is efficient, future work should consider whether political overseers other than OIRA defer to agency policy more frequently when agencies form coalitions to produce those policies. For example, agencies should win more frequently in court when the policy in question was produced by a coalition, as implied by Freeman and Rossi (2011, 2012); Congress should sanction agencies less frequently when they form coalitions; and interest groups or other regulated entities should view policy as more efficient or informed when produced by coalitions. 80 Chapter 4 Politics: Exploiting Collective Action Problems in Congress In his influential book on the US Congress, Mayhew (1974) observed a simple yet important feature of the national legislature: “the organization of congress meets remarkably well the electoral needs of its members.” Going even further, he argued that “if a group of planners sat down and tried to design a pair of national assemblies with the goal of serving members’ electoral needs year in and year out, they would be hard pressed to improve on what exists” (81). But members’ electoral needs are only rarely so aligned as to facilitate national policy change or effective oversight of the Executive Branch: polarization, gridlock, and party gatekeeping can stall legislative activity even on issues where a majority of legislators and voters might prefer revision of the status quo. Despite the apparent ingenuity of US legislative design for members’ electoral needs, collective action problems plague the institution, often rendering it incapable of responding to the popular will or overseeing the Executive Branch. Legislating in the American system requires herculean efforts. Once introduced, bills must pass through at least two committees controlled by the majority party to make it to the floor of each chamber. Once on the floor, the House of Representatives must pass the bill by a simple majority vote, but a minority of senators in the upper chamber can halt the legislative process indefinitely. Even if the bill passes both chambers, the president may veto it which can only be overridden by supermajorities in both chambers. 1 Such a system, designed in the eighteenth century when the 1 Krehbiel (2010) explains how bicameralism, separation of powers, and supermajoritarian requirements in the U.S. Congress lead to gridlock, and Cox and McCubbins (2005) explain how parties exert negative agenda control 81 scope of the national government was contested, has at times seemed ill equipped to respond to national necessities. The rise of polarization since the 1970s has further highlighted the problems of legislative organization since Congress seems unable on occasion to respond to even the most basic national needs or hold presidents and other executives accountable for obvious violations of the public trust. So great are congressional collective action problems, in fact, that Congress has not infre- quently bestowed (or perhaps foisted) its authority to make policy upon the Executive Branch. Unlike Congress, executive agencies are hierarchical and centralized organizations usually headed by a single secretary or administrator at the top and filled with layers of careerists underneath. Although the bureaucratic policymaking process is neither unilateral nor without cost, agencies do not suffer from the same collective action problems that Congress does. Therefore, when congres- sional collective action problems become too much to bear, Congress delegates to the Executive Branch (Epstein et al. 2007). Over the last century or so, the Executive Branch has become respon- sible for such constitutionally legislative duties as budgeting, apportionment, tariffs, and military engagement (see, e.g., Dearborn 2021). But delegation as such does not always resolve Congress’ collective action problems; it may only delay them. After delegation, bicameralism, separation of powers, and the committee system continue to limit legislative responses to the administrative state. Congress is not the only institution respon- sible for overseeing the bureaucracy: the president and the courts also play an important role in overseeing federal agencies. These multiple principals, or overseers, create collective action prob- lems in the post-delegation stage of bureaucratic oversight. Even if majorities in each chamber of Congress support punishing an agency for its policies, the president can veto whatever sanction that legislative coalition supports or a court can nullify the law. The more disagreement among those principals, the more likely it is that agencies will be able to act with impunity since all must agree to punish an agency for such a punishment to be doled out (see, e.g., Boushey and McGrath 2020; Hammond and Knott 1996, 1999; MacDonald 2007). using the committee system. Combined, bicameralism, separation or powers, supermajoritarian requirements, and party gatekeeping limit the range of policies Congress can pass. 82 Even just within Congress, agencies are subjected to (or enjoy) oversight by multiple princi- pals. Both chambers of the U.S. Congress have organized themselves into several committees, each with a specific policy jurisdiction that members guard jealously (see, e.g., Weingast and Marshall 1988). For example, both the House and Senate have an Agriculture Committee, both an Armed Forces Committee, and both a committee handling education and labor policy. When Congress delegates the power to make policy to the Department of Agriculture, for example, both represen- tatives in the House Agriculture Committee and senators in the Senate Agriculture Committee are responsible for overseeing the Department’s activity. Such duplication leads to yet another col- lective action problem since each committee may free ride off the other’s supervisory activity and each committee must agree to any legislative action curbing the agencies (Clinton, Lewis and Selin 2014; Rezaee, Gailmard and Wood 2019; Shipan 2004; Woolley 1993). Collective action problems in Congress, in short, benefit agencies because legislative sclerosis limits legislative responses to agency actions. The politically astute bureaucrat recognizes congressional collective action problems at both the policymaking and oversight stages and exploits them to their benefit. This chapter consid- ers how executive coalition building offers bureaucrats a tool to exploit congressional collective action problems and argues that one of the key political functions of contemporary executive coali- tion building is to induce collective action problems among congressional overseers. In so doing, bureaucrats can promulgate policies that overseers cannot repeal by increasing the number of over- seers (and therefore veto players), compounding both the free rider and collective action problems endemic to legislative institutions. In short, bureaucrats understand how the legislative process and legislative organization limit congressional responses to agency actions and how to manipulate the legislative process in order to further limit Congress’s ability to limit agency behavior. 2 2 A not insignificant proportion of oversight occurs through channels that do not require collective action — for example contact from individual members of Congress (Lowande 2019). Although oversight does not occur exclu- sively through committee action, committee action remains the most public and costly form of oversight for agencies. Additionally, committees have a great deal of power over ex ante control of agencies (i.e., legislation. 83 4.1 Exploiting Collective Action Problems Agencies are not passive actors in the American political system. They recognize opportunities to achieve their policy goals in the face of political opposition. They may leverage their superior knowledge of the regulatory process to get what they want, wait to produce certain policies un- til congressional, presidential, or judicial conditions are more favorable (Potter 2017, 2019), or activate diverse networks of support to lobby or otherwise convince overseers to defer to agency desires (Carpenter 2001). To this, I add that agencies collaborate strategically to induce collective action problems in Congress and achieve their preferred policy outcomes. Much of the work on multiple principals studies either congressional outcomes like hearings held (McGrath 2013; Rezaee, Gailmard and Wood 2019), individual forms of oversight that do not require collective action (Bolton 2022; Lowande 2018a; Lowande and Augustine Potter 2021), or regulatory output from individual agencies given different partisan or ideological arrangements among principals (Boushey and McGrath 2020; Palus and Yackee 2022; Shipan 2004). But all of these previous studies overlook the ability of agencies to induce collective action problems among multiple principals. Agencies do not simply observe whether their principals are divided and choose to act – although waiting until political conditions among overseers are friendlier to agency action is a strategy agencies do use (Potter 2017, 2019) – they also work actively to amplify political divisions among their overseers, activating additional oversight committees by collabo- rating with other agencies with different oversight committees. Consider a simple spatial model in one dimension with three actors, an agency with ideal point A, a House committee whose median’s ideal point is C H , and a Senate committee whose median’s ideal point is C S , represented in figure 4.1. 3 The agency first promulgates a policy, A, pursuant to some grant of authority. Then, a member of Congress can introduce a bill revising the agency 3 A fully fleshed out spatial model of this process would require also including floor medians and filibuster pivots in the House and Senate respectively, but for the purposes of this example, assuming each chamber passes any bill that a committee reports favorably is sufficient to show the logic. This example is a slight modification of the model in Shipan (2004). Additionally, although my argument concerns agencies collaborating with each other, the simple spatial models present only one agency for simplicity and to present the basic logic of the argument with less clutter. A could be interpreted as the policy the coalition of agencies bargains to produce rather than the ideal point of an individual agency. 84 action which will be assigned to committee C H in the House and C S in the Senate. Either committee can kill the bill reverting the policy to whatever the agency did in the first place. If a committee kills the bill, then the policy outcome is the agency’s action. If both committees pass the bill, then the policy outcome is the bill introduced in Congress. (This is a simplified version of the model presented in Shipan (2004).) 4 Figure 4.1: Spatial Model with Two Committees Under standard assumptions, each player has a single-peaked ideal point such that each actor’s most preferred policy is represented by their ideal point and their utility decreases sym-metrically as policy outcomes diverge from their ideal point. Since preferences are fixed in (at least) the short-run, agencies can only manipulate outcomes by bringing in more veto players such that at least one prefers a policy outcome closer to the agency than any proposed bill that could pass all committees and both chambers. To see why, suppose two committees in the Senate are responsible for a bill to revise the agency’s policy, leading to the preference configuration represented in figure 4.2. As before, A represents the agency’s ideal point and C H represents the ideal point of the House committee’s median. Now, however, C 1 S represents one of the Senate committee median’s ideal point, and C 2 S represents the second’s. Under this preference configuration, the agency can set policy at exactly its ideal point. Even if the first Senate committee median would set policy at C 1 S instead of A, the second Senate committee prefers A to C 1 S so it will block the bill, reverting policy to A, the agency’s original proposal. As a general matter, if agencies can expand the scope of conflict to include at least one veto player closer to its ideal point than existing veto players, the agency can be no worse off and may often fare better. 4 Following, Shipan (2004), I do not include a status quo policy since agencies are first movers and therefore set the status quo for Congress to choose either to overturn or uphold. 85 Figure 4.2: Spatial Model with Three Committees But agencies do not refer bills to committees: Congress does. House rules enacted in the 1970s prescribe that each bill introduced in the House must be referred to all committees with jurisdiction over the subject matter of the bill. 5 For example, the Department of Homeland Security Blue Campaign Authorization Act, designed to aid the Department of Homeland Security in addressing human trafficking and signed into law by President Trump in 2018, was referred to both the House Homeland Security and Judiciary committees as it dealt with both issues of homeland security and law enforcement. 6 The United States Parole Commission Extension Act of 2018, extending the authority of the U.S. Parole Commission by two years and also signed into law by President Trump in 2018, on the other hand, was referred only to the House Judiciary Committee as it dealt solely with law enforcement. 7 The Senate has a longer history of multiple referrals (Davidson 1989; Sinclair 2016). Unsurprisingly, research indicates that multiply referred bills are less likely to be reported to the floor of the House than singly referred bills due to the increasing number of veto players created by multiple referrals (Davidson, Oleszek and Kephart 1988; Krutz and Cullison 2008; Young 1996). Since bills must be referred to all committees with reasonable claims to jurisdiction over the issue, any bill that affects agencies with disparate jurisdictions should be referred to all com-mittees with those same jurisdictions. For example, if Congress wishes to address a recent action by the Department of Homeland Security, a member can introduce a bill which will be referred to the 5 When the House changed its rules in the 1970s to allow for multiple referrals, it created three types of multiple refer-rals (Davidson 1989). The first is the joint or concurrent referral where a bill is referred to more than one com- mittee simultaneously. The second is the split or divided referral where each of the multiple committees responsible for a bill are responsible for different sections or titles of that bill. The last is the sequential referral where a bill is referred to multiple committees sequentially such that no committee considers the same bill at the same time. Upon taking control of the House for the first time in decades in 1995, Republicans changed House rules to require that the Speaker of the House designate one committee the primary committee, a change that somewhat lessened the effect of multiple referrals on bill progression, but only at the pre-floor stage (Krutz and Cullison 2008). 6 Department of Homeland Security Blue Campaign Authorization Act, H.R. 4708, 115th Cong. (2018). 7 United States Parole Commission Extension Act of 2018, H.R. 6896, 115th Cong. (2018). 86 Homeland Security committees in the House and Senate. But if Congress wishes to address a recent action by both the Department of Homeland Security and the Department of Energy, the bill will likely be referred to four committees: Homeland Security in the House and Senate, and Energy and Commerce in the House and Senate. Therefore, by forming a coalition, the Depart-ments of Homeland Security and Energy can increase the number of veto players required to overturn their action. As a concrete example, in response to a rule promulgated by the Environ-mental Protection Agency and Army Corps of Engineers related to the Clean Water Act in 2015, David Rouzer (R- NC) introduced a bill prohibiting the EPA from using its appropriated funds for the year 2015 until the rule was rescinded, and that bill was referred to four committees in the House. 8 A similar bill was introduced in the Senate and referred to two committees, only one of which even held hearings on the bill. 9 Neither bill passed either chamber. Oversight committee medians for pairs of agencies may be arranged in three possible ways. First, two agencies can have the same oversight committees. For example, the Equal Em-ployment Opportunity Commission and the Department of Labor are both overseen by the same agencies in the House (House Committee on Education and Labor) and the Senate (Senate Committee on Health, Education, Labor, and Pensions). 10 For agencies with oversight committees arranged in this way, collaborating will make no difference in the ideological disagreement or gridlock between committees since no additional veto players are brought in. This regime is represented in figure 4.3 in the first panel: (a) Same Oversight Committees. Agency 1 is overseen by C 1 S and C 1 H and agency 2 is overseen by C 2 S and C 2 H in each regime. Second, two agencies can have different oversight committees, but the ideological distance between the most liberal and most conservative of the four committees jointly responsible for overseeing those two agencies is only larger than the ideological distance between the most liberal and most conservative of each pair of committees for one agency. This occurs when the ideal points of both House and Senate committee medians for one agency lie between the ideal points of the 8 Don’t Ignore the Will of the People Act, H.R. 2599, 114th Cong. (2015). 9 Defending Our Rivers from Overreaching Policies Act of 2015, S. 1178, 114th Cong. (2015). 10 Throughout this chapter, I consider the primary oversight committee of each agency as those responsible for con-firming nominees to each agency in the Senate, and their counterparts in the House. 87 other agency. For agencies with oversight committees arranged in this way, the agency with less ideologically diverse committees benefits from coalition building, but the agency with more distant oversight committee medians may not. This regime is represented in figure 4.3 in the second panel: (b) Larger for One. Finally, two agencies can have different oversight committees, and the ideological distance between the most liberal and most conservative of the four committees jointly responsible for overseeing those two agencies is larger than the ideological distance between the most liberal and most conservative of each pair of committees for both agencies. This occurs when the interval between the House and Senate committee medians for each agency overlap but neither shis wholly contained in the other. 11 For pairs of agencies with oversight committees arranged in this way, both agencies benefit from collaboration by guaranteeing more committee gridlock if Congress attempts to overturn their policy. This regime is represented in figure 4.3 in the final panel: (c) Larger for Both. Figure 4.3: Three Oversight Regimes. Numbers in superscripts indicate which agency each com- mittee oversees and letters in each subscript indicate each committee’s chamber. Brackets indicate ideological distance between each agency’s oversight committees. 11 The two sets of committees theoretically could not overlap at all, but committee medians in each chamber belong to the majority party and the two parties have been polarized with almost no overlap in the contemporary era. 88 Agencies have the most to gain in the third regime since they can induce a collective action problem in Congress by forcing greater disagreement among legislative overseers. Therefore, the first empirical implication of my theory of executive coalitions vis- ` a-vis Congress is: Hypothesis 4 When the distance between the most liberal and most conservative committee me- dians of the four committees overseeing two agencies is larger than the distance between the most liberal and most conservative committee medians for each agency’s standard oversight committees (regime c), those agencies are most likely to collaborate. But legislative control of the administrative state is not only achieved through legislation. Com- mittees also serve an important role in overseeing agency implementation of legislative policy. And just like in the legislative process, collective action problems plague congressional oversight. Bicameralism again limits the responsiveness of committees to agency behavior since committees have incentives to free ride off their counterparts (Gailmard 2009; Rezaee, Gailmard and Wood 2019). In fact, agencies report less congressional influence in their affairs when they are subject to oversight by multiple committees (Clinton, Lewis and Selin 2014). Policing the administrative state is costly for members of Congress. The committee system helps cut down on information seeking costs since each committee is only responsible for a subset of federal agencies, but committees still oversee multiple agencies responsible for regulating activ- ity in many policy areas. If members of Congress had to actively monitor every agency under their committees’ jurisdictions, there would be no time for any of the many other activities members of Congress must do like legislating, case work, and campaigning. Therefore, Congress, with the Administrative Procedure Act and other statutes regulating the administrative state, has installed procedural technologies that allow interested parties like interest groups to alert Congress is an agency engages in undesirable behavior. These “fire alarms” reduce oversight costs for Congress, thereby providing a more efficient means of oversight (McCubbins and Schwartz 1984, but see Lowande 2018b for evidence that members of Congress do engage in some unprompted monitor- ing). 89 Relying on fire alarms, however, means that committees only receive allegations of agency malfeasance but the members sitting on committees do not directly observe agency behavior un- less they call for a hearing with agency witnesses or subpoena agency records. Calling a hearing, subpoenaing agency records, or otherwise seeking to audit agency actions, however, is costly. The benefits of such an audit are informational: the committees can learn whether and to what extent an agency has misbehaved. Because once that information is public, all members of Congress can access it, committees have incentives to free ride off the auditing activity of other committees in order to learn the information they seek without taking costly action (Gailmard 2009). Since each agency is overseen by a committee in the House and the Senate, each committee can theoretically free ride off the auditing activity of at least one other. Critically, even if both committees agree perfectly that the agency should be audited, oversight may be underprovided due to the collec- tive action problems. Therefore, the institutional design of bicameralism is sufficient to lead to inefficient oversight. But crafty agencies can make the problem even worse. Executive coalition building helps agencies avoid oversight. Agencies can induce an even larger collective action problem in oversight by collaborating with each other. If two agencies have different oversight committees, then by collaborating, agencies can expand the number of principals responsible for oversight from two to four. With four instead of two oversight commit- tees, free riding should increase since each committee may anticipate any of the three – rather than only one – other committees may audit the agencies. Mapping this theory onto the three regimes in figure 4.3 yields the second hypothesis. Pairs of agencies with the same oversight committees (regime a) do not stand to gain from collaborating since acting on their own or as a pair would result in the same number of committees responsi- ble for overseeing them. However, pairs of agencies with different sets of oversight committees (regimes b and c) can make freeriding among oversight committees more likely by collaborating and increasing the number of committees responsible for oversight. Therefore: Hypothesis 5 Pairs of agencies with different sets of oversight committees (regimes b and c) are more likely to collaborate than pairs of agencies with the same oversight committees (regime a) 90 This hypothesis is counterintuitive without theory. Agencies with the same oversight com- mittees inhabit similar policy areas and therefore might naively be expected to collaborate more frequently than those from different policy areas. If agency coalitions served only technocratic purposes, that might be the case. However, if agency coalitions are intended to induce collective action problems in Congress in order for agencies to achieve their desired policy outcomes, we should observe agencies collaborating when they have different oversight committees even though this means they are responsible for different policy areas. 4.2 Data and Empirical Strategy To test these hypotheses, I leverage the novel dataset of agency coalitions that I built for this dis- sertation and described in Chapter 3. Each observation is an agency dyad-year since commit-tee compositions change each Congress and occasionally within the same Congress. The dependent variable is a binary indicator for whether each agency dyad in each year formed a coalition or not. I then matched agencies to oversight committees in the Senate by selecting the committee responsi- ble for first considering nominees to that agency, and in the House by selecting the committee anal- ogous to the Senate oversight committee. Next, I calculated the absolute value of the difference in the DW-NOMINATE estimate of each agency’s two oversight committee’s median member’s ideal point. Then, I calculated the absolute value of the difference in the DW-NOMINATE estimate for the most liberal and most conservative of each of the four committees overseeing the two agencies forming the dyad. Finally, I created a variable that can take one of three values corresponding to the regimes in figure 4.3. This variable takes the value Same Over-sight Committee if the two agencies forming each dyad have the same oversight committee; it takes the value Larger for One if the absolute value of the difference in the DW-NOMINATE estimate for the most liberal and most conservative of each of the four committees overseeing the two agencies is larger than only one of the absolute values of the differences of the pairs of committees overseeing each individual 91 agency; last, it takes the value Larger for Both if the absolute value of the difference in the DW- NOMINATE estimate for the most liberal and most conservative of each of the four committees overseeing the two agencies is larger than both of the absolute values of the differences of the pairs of committees overseeing each individual agency. Table 4.1 displays examples of dyads with the Department of Labor in the 112th Con-gress to clarify the measurement. The first row displays information on the Department of Labor. From left to right, the second and third columns display the committee in the Senate overseeing the De- partment of Labor and the DW-NOMINATE estimate of that committee median’s ideal point. The fourth and fifth column display the same information but for the House committee overseeing the Department of Labor. The sixth column displays the distance between the Senate and House com- mittee medians’ ideal points. The second through fourth rows display the same information but for three other agencies during the 112th Congress. For these rows, the seventh column displays the distance between the most liberal and most conservative of all the commit-tees responsible for overseeing both that agency and the Department of Labor. For example, the Senate median for the Department of Labor is the most liberal of any, and the House median for the Department of Agriculture the most conservative. Therefore, the cell under joint distance for the Department of Agriculture shows the absolute value of the difference between the Senate median for the Depart- ment of Labor and the House median for the Department of Agriculture. Finally, the eighth column displays to which regime each dyad belongs by comparing the joint distance to the individual dis- tances in columns six and seven. Table 4.2 displays raw percentages of dyads forming coalitions in each regime during each presidential term. Agencies with the same oversight committees were consistently less likely to form coalitions than agencies with different coalitions. Aggregating across the entire time- frame, only 3.6% of dyads with the same oversight committees formed coalitions and about 11% of dyads with different oversight committees formed coalitions, a difference of about seven per- centage points, consistent with hypothesis 5. During Clinton’s two terms, agency pairs whose joint oversight committee ideological distance is larger for both individual agencies were most 92 Table 4.1: Example of Regimes with Department of Labor Dyads in 112th Congress Senate House Compared to Department of Labor Joint Agency Committee Median Committee Median Distance Distance Regime Department of Health, Education, Education Labor Labor, and Pensions − 0.215 and Labor 0.252 0.467 —– —– Equal Employment Opportunity Health, Education, − 0.215 Education 0.252 0.467 0.467 Same Oversight Commission Labor, and Pensions and Labor Committees Department of Armed Armed Defense Services − 0.043 Services 0.233 0.276 0.467 Larger for One Department of Agriculture Agriculture − 0.121 Agriculture 0.314 0.435 0.529 Larger for Both likely to collaborate, those whose joint oversight committee ideological distance is larger for only one agency were second most likely to collaborate, and those with the same oversight commit- tees were the least likely. The number are particularly striking in Clinton’s second term where 26.4% of dyads for whom collaborating increased ideological disagreement among overseers for both committees formed coalitions whereas only 17% of dyads for whom collaborating increase ideological disagreement among overseers for only one committee formed coalitions, consistent with hypothesis 4. The differences between these last two regimes dissipates in the raw numbers from Bush onward, but it remains across all time periods that agencies with the same oversight committees are the least likely to form coalitions. Table 4.2: Proportion of Dyads Forming Coalitions by Regime and Presidential Term Clinton I Clinton II Bush I Bush II Obama I Obama II Trump (1995–6) ∗ (1997–2000) (2001–4) (2005–8) (2009–12) (2013–6) (2017–8) ∗∗ Same Oversight Committees 3.8 7.7 4.8 1.0 0.96 2.7 5.8 Larger for One 4.0 17.0 15.8 7.4 5.1 6.1 19.1 Larger for Both 7.6 26.4 14.9 6.7 3.4 9.8 10.5 Note: Cell entries are percentages ∗ Clinton I only includes the last two years of that term. ∗∗ Trump only includes the first two years of that term. Figure 4.4 displays the proportion of dyads forming coalitions each Congress for agency pairs that share oversight committees and those that do not. In every Congress, agencies with different 93 Figure 4.4: Coalitions by Congress and Oversight Committees oversight committees were more likely to collaborate, consistent with hypothesis 5. Agencies with different oversight committees can compound the free rider problem that plagues committee oversight of agency actions by increasing the number of committees responsible for oversight, thereby making it more likely each individual committee believes it can free ride of the other committees’ oversight activities. Figure 4.4 provides evidence that agencies do in fact behave in this way. However, these raw numbers are only suggestive given the repeated observations in the data and dyad- and Congress-level confounders. Therefore, I estimate a series of linear probability models to estimate the effect of different regimes on agency coalition building. Specifically, to test hypothesis 4, I regress whether each dyad formed a coalition in each year on an indicator variable, Larger for One, which takes the value of one if collaborating increases gridlock for only one agency (regime b) and zero otherwise, dyad and year fixed effects, and the same control variables as in the previous chapter. These models exclude agencies with the same oversight committees (regime a) due to the dyad fixed effects. To test hypothesis 5, I include all dyads and regress the same dependent variable on a binary variable, Same Oversight Committees, which takes the value of one if the dyads share the same oversight committees (regime a), year fixed effects, and the same control variables. For both tests, the coefficients on the relevant independent variable should be negative. 94 4.3 Results Tables 4.3 and 4.4 reports the results of the linear probability models. Since, as in previous chap- ters, all the covariates are only available from the 105th (1997-1998) to the 112th Congress (2011- 2012), model 1 in each table presents results for the 104th (1995-1996) through 115th (2017-2018) Congresses without covariates — only with dyad and year fixed effects -– and model 2 presents results using only those covariates available for the full timeframe, though not for all dyad-years. 12 As expected, the coefficient on Larger for One across specifications in table 4.3 is negative, indicating that agencies are more likely to form coalitions when they can induce collective action problems by widening the ideological gap between oversight committees. Since these models in- clude dyad fixed effects, the coefficient estimates the within-dyad change in probability of coalition formation when a dyad changes from regime c to regime b and that change is negative, consistent with hypothesis 4. Also as expected, the coefficient on Same Oversight Committees across most specifications in table 4.4 is negative, indicating that agencies are more likely to from coalitions when they can compound the free rider problem among oversight committees by introducing addi- tional overseers. These models include year fixed effects, so the coefficients estimate the difference in the probability of coalition formation within the same year among dyads with and without the same oversight committees, consistent with hypothesis 5. Models 3 through 6 include the full suite of covariates included in the regressions in the previous chapter and uncovers similar results. 12 See the discussion of the data in Chapter 3. Specifically, I am able to include employment and politicization covariates in model 2, but not all agencies have a defined politicization measure because they do not always em-ploy career senior executive servants. 95 Table 4.3: Coalition Building and Congressional Committee Gridlock. Dependent variable: Coalition 104th–115th 105th–112th Congresses Congresses (1) (2) (3) (4) (5) (6) Larger for One − 0.020 ∗∗∗ − 0.037 ∗∗∗ − 0.038 ∗∗∗ − 0.038 ∗∗ − 0.036 ∗∗∗ − 0.036 ∗∗∗ (vs. Larger for Both) (0.006) (0.009) (0.008) (0.008) (0.009) (0.009) Agency Alignment 0.009 ∗∗ 0.007 − 0.058 − 0.012 ∗∗ (0.004) (0.005) (0.039) (0.006) Agency Alignment× 0.005 0.011 Larger for One (0.006) (0.008) Observations 10,583 6,893 7,771 7,771 6,250 6,250 Dyad FEs Yes Yes Yes Yes Yes Yes Year FEs Yes Yes Yes Yes Yes Yes Time-Varying Covariates No Limited No No Yes Yes Adjuster R 2 0.389 0.378 0.389 0.384 0.397 0.397 ∗ p<0.1, ∗ ∗ p<0.05, ∗∗∗ p<0.01 Note: Unit of analysis is the agency dyad-year. Standard errors clustered by dyad reported in parentheses. Substantively, the effects are notable. The unconditional average rate of coalition formation for the full sample is about eleven percent. Among agency dyads with different oversight committees, when collaborating guarantees increased gridlock among committees, they are about two to four percentage points more likely to form coalitions than when collaborating only in-creases gridlock for one agency forming the dyad, consistent with hypothesis 4. Among all agency dyads in a given year, those with different oversight committees are about four to eight percentage points more likely to collaborate than those with the same oversight committees, consistent with hypothesis 5. The results provide evidence in favor of both hypotheses 4 and 5. First, dyads for whom collab- orating increases ideological disagreement for both sets of oversight committees are most likely to form coalitions, consistent with hypothesis 4. Second, dyads with different sets of oversight committees are more likely to collaborate than those with the same oversight committees, consis- tent with hypothesis 5. These results are consistent with my theory that agencies form coalitions 96 Table 4.4: Coalition Building and Congressional Committee Gridlock. Dependent variable: Coalition 104th–115th 105th–112th Congresses Congresses (1) (2) (3) (4) (5) (6) Same Oversight Committees − 0.076 ∗∗∗ − 0.040 ∗∗ − 0.075 ∗∗∗ − 0.079 ∗∗∗ − 0.012 − 0.013 (0.018) (0.017) (0.018) (0.008) (0.018) (0.018) Agency Alignment 0.258 ∗∗∗ 0.038 ∗∗∗ 0.026 0.004 (0.035) (0.005) (0.038) (0.005) Agency Alignment× − 0.026 ∗∗∗ − 0.009 Same Oversight Committees (0.009) (0.010) Observations 11,191 7,285 8,215 8,215 6,867 6,867 Year FEs Yes Yes Yes Yes Yes Yes Time-Varying Covariates No Limited No No Yes Yes Adjuster R 2 0.001 0.144 0.132 0.132 0.219 0.219 ∗ p<0.1, ∗ ∗ p<0.05, ∗∗∗ p<0.01 Note: Unit of analysis is the agency dyad-year. Standard errors clustered by dyad reported in parentheses. when doing so can induce collective action problems in Congress by introducing larger ideolog- ical cleavages among principals (hypothesis 4) and worse free riding problems among oversight committees (hypothesis 5). The previous analysis assumes that committee medians are decisive in advancing bills through the legislative process and that committee medians vote their sincere preferences without any pres- sure from their parties. However, party leaders influence the rank-and-file by providing positive and negative incentives for members if they act in the interest of the party or not (Cox and Mc- Cubbins 2005, 2007). Parties exist partly to overcome the collective action problems inherent in legislative politics (Aldrich 2011). In addition, some argue committee power in Congress has waned in the postreform era when the Democratic Party leadership consciously reduced the power of committee chairs and other senior members since they were dominated by the Southern wing of the Party that was out of step with party leadership on many issues. After the Republican Party gained control of the U.S. House for the first time in forty years in 1995, committee power was reduced further, shifting agenda setting and legislative power from committees to centralized party leadership (see, e.g., Deering and Smith 1997; Rohde 1991). 97 This view of power in Congress implies one of two potential observable phenomena. First, if committees are weak and centralized party control strong, then agencies should have trouble induc- ing collective action problems among committees since each committee’s majority party members should vote the party line and advance bills that satisfy their party’s median member. If that were the case, then there should be no difference in agency collaboration across the three regimes. The previous analysis shows that not to be the case. Instead, I provided evidence that agencies antic- ipate that they can induce collective action problems, meaning they act as if committee medians can vote sincerely and committee medians from different committees could vote differently on the same bill. Second, if parties structure their members’ behavior but nevertheless allow some degree of freedom to their delegates on committees, then it is not committee medians that matter, but the median of each committee’s contingent of majority party members that does. If that were the case, then the regimes of overlapping oversight should be constructed by com-paring majority party medians on committees, not general committee medians. This expectation is tested below. Table 4.5: Proportion of Dyads Forming Coalitions by Regime and Presidential Term (Majority Party Committee Medians) Clinton I Clinton II Bush I Bush II Obama I Obama II Trump (1995–6) ∗ (1997–2000) (2001–4) (2005–8) (2009–12) (2013–6) (2017–8) ∗∗ Larger for One 5.2 21.0 15.1 6.3 3.8 8.5 21.0 Larger for Both 6.2 23.4 15.9 7.6 3.9 7.6 10.9 Note: Cell entries are percentages ∗ Clinton I only includes the last two years of that term. ∗∗ Trump only includes the first two years of that term. Table 4.5 displays raw percentages of dyads forming coalitions in the two regimes for dyads who do not share oversight committees. If agencies share oversight committees, the position of the party or committee median is irrelevant as collaborating cannot induce more ideological gridlock, so I decline to include those numbers in this table as they are the same as in table 4.2. Unlike the findings for general committee medians, table 4.5 shows that dyads do not collaborate more fre- quently when collaboration would induce more ideological conflict among majority party medians on committees. During each presidential term, dyads that can increase gridlock among majority 98 party medians on committees by collaborating were about as likely as those that could not, with the exception of the Trump presidency where the differences in probabilities are large but in the opposite direction of hypothesis 4. This implies that agencies care about general committee me- dians (who belong to the majority party because of how committees are constituted) and not the majority party medians on committees. Table 4.6: Coalition Building and Congressional Committee Gridlock within the Majority Party. Dependent variable: Coalition 104th–115th 105th–112th Congresses Congresses (1) (2) (3) (4) (5) (6) Larger for One 0.003 − 0.010 − 0.015 ∗∗ − 0.015 ∗∗ − 0.010 − 0.010 (vs. Larger for Both) (0.006) (0.007) (0.007) (0.007) (0.009) (0.009) Agency Alignment 0.069 ∗∗ 0.012 ∗∗∗ − 0.056 − 0.002 (0.0285) (0.005) (0.039) (0.006) Agency Alignment× − 0.006 − 0.012 Larger for One (0.006) (0.008) Observations 10,583 6,893 7,771 7,771 6,250 6,250 Dyad FEs Yes Yes Yes Yes Yes Yes Year FEs Yes Yes Yes Yes Yes Yes Time-Varying Covariates No Limited No No Yes Yes Adjuster R 2 0.388 0.375 0.387 0.387 0.394 0.395 ∗ p<0.1, ∗ ∗ p<0.05, ∗∗∗ p<0.01 Note: Unit of analysis is the agency dyad-year. Standard errors clustered by dyad reported in parentheses. Table 4.6 report results of the linear probability models analogous to those in table 4.3 but using majority party medians on committees instead of general committee medians. The prelim- inary findings in table 4.5 are borne out in the regression results. Although for some models the coefficient on Larger for One is negative and significant, the effects are much smaller than those in table 4.3. These results imply that agencies consider general committee medians rather than the median of the majority party’s contingent on each committee. As a results, agencies collaborate when doing so amplifies gridlock among general committee medians. 99 4.4 Discussion and Conclusion Existing theories of multiple principals overseeing the bureaucracy have ignored strategies agen- cies can use to exploit legislative collective action problems. Bureaucrats do not always have to wait for gridlock in Congress resulting from biannual elections. Instead, they can amplify gridlock between electorally induced changes in partisan and ideological coalitions by collaborating with other agencies to create ideological divisions among existing overseers. Examining agency collaboration from the 104th Congress (1995-6) to the 115th Congress (2017-8), I find that agencies collaborate when doing so increases ideological disagreement among overseers, frustrating attempts at legislatively addressing those agencies’ actions. Additionally, I find that agencies with different sets of oversight committees collaborate more frequently than those with the same set of oversight committees because those with different over-sight commit- tees can compound the free rider problem endemic to decentralized oversight of agencies by con- gressional committees. The arguments presented here acknowledge that agencies are experts not only in their subject matters and the procedures they can use to achieve their policy aims, but also demonstrate considerable knowledge of the legislative process, the members of Congress most re- sponsible for oversight, and the set of other agencies with whom they can collaborate to amplify collective action problems among congressional overseers. Taken together with the previous chapter’s argument and findings that agencies collaborate to achieve their policy goals in the face of opposition from the president and OIRA, these findings fur- ther demonstrate that agencies collaborate strategically and in explicitly political ways. Agencies form coalitions with each other strategically in order to sidestep oversight and political control by Congress. By collaborating, agencies induce collective action problems among overseers. Specifi- cally, collaboration creates or exacerbates the free rider problem among oversight committees and can increase ideological disagreement among overseers, making legislative responses less likely by amplifying gridlock. The results presented in this chapter strongly imply that agencies do not 100 collaborate for purely technocratic reasons like information sharing or the crafting of more effi- cient policy, but instead collaborate for political reasons, to achieve their policy goals, and increase their power within the American political system. 101 Bibliography Ai, Chunrong and Edward C. Norton. 2003. “Interaction Terms in Logit and Probit Models.” Economics letters 80(1):123–129. Aldrich, John H. 2011. Why Parties?: A Second Look. Chicago, IL: University of Chicago Press. Anderson, Sarah E and Matthew Potoski. 2016. “Agency Structure and the Distribution of Federal Spending.” Journal of Public Administration Research and Theory 26(3):461–474. Anzia, Sarah F. and Christopher R. Berry. 2011. “The Jackie (and Jill) Robinson Effect: Why Do Congresswomen Outperform Congressmen?” American Journal of Political Science 55(3):478– 493. Arel-Bundock, Vincent, James Atkinson and Rachel Augustine Potter. 2015. “The Limits of For- eign Aid Diplomacy: How Bureaucratic Design Shapes Aid Distribution.” International Studies Quarterly 59(3):544–556. Austen-Smith, David and Jeffrey S. Banks. 1996. “Information Aggregation, Rationality, and the Condorcet Jury Theorem.” American Political Science Review 90(1):34–45. Ban, Pamela, Ju Yeon Park and Hye Young You. 2023. Bureaucrats in Congress: Information Sharing in Policymaking. Technical report. Baron, David P. 1989. “A Noncooperative Theory of Legislative Coalitions.” American Journal of Political Science 33(4):1048–1084. Baron, David P. and John A. Ferejohn. 1989. “Bargaining in Legislatures.” American Political Science Review 83(4):1181–1206. Beck, Nathaniel. 2018. “Estimating Grouped Data Models with a Binary Dependent Variable and Fixed Effects: What Are the Issues.” Working Paper . URL: https://arxiv.org/abs/1809.06505 Beck, Nathaniel. 2020. “Estimating Grouped Data Models with a Binary-Dependent Variable and Fixed Effects via a Logit versus a Linear Probability Model: The Impact of Dropped Units.” Political Analysis 28(1):139–145. Berry, Christopher R., Barry C. Burden and William G. Howell. 2010. “The President and the Distribution of Federal Spending.” American Political Science Review 104(4):783–799. 102 Berry, Christopher R. and Jacob E. Gersen. 2017. “Agency Design and Political Control.” Yale Law Journal 126:1002. Bertelli, Anthony M. and Christian R. Grose. 2009. “Secretaries of Pork? A New Theory of Distributive Public Policy.” The Journal of Politics 71(3):926–945. Bertelli, Anthony M. and Christian R. Grose. 2011. “The Lengthened Shadow of Another Insti- tution? Ideal Point Estimates for the Executive Branch and Congress.” American Journal of Political Science 55(4):767–781. Bickers, Kenneth N. and Robert M. Stein. 1996. “The Electoral Dynamics of the Federal Pork Barrel.” American Journal of Political Science 40(4):1300–1326. Bils, Peter. 2019. “Policymaking with Multiple Agencies.” American Journal of Political Science 00(00):1–15. Bolton, Alexander. 2020. “Ideology, Unionization, and Personnel Politics in the Federal Budget Process.” Journal of Public Administration Research and Theory pp. 1–18. Bolton, Alexander. 2022. “Gridlock, Bureaucratic Control, and Nonstatutory Policymaking in Congress.” American Journal of Political Science 66(1):238–254. Bolton, Alexander, Rachel Augustine Potter and Sharece Thrower. 2016. “Organizational Capac- ity, Regulatory Review, and the Limits of Political Control.” The Journal of Law, Economics, and Organization 32(2):242–271. Boushey, Graeme T and Robert J McGrath. 2020. “Does partisan conflict lead to increased bureau- cratic policymaking? Evidence from the American States.” Journal of Public Administration Research and Theory 30(3):432–447. Carpenter, Daniel P. 2001. The Forging of Bureaucratic Autonomy: Reputations, Networks, and Policy Innovation in Executive Agencies, 1862-1928. Princeton, NJ: Princeton University Press. Centers for Disease Control and Prevention. 2015. “Ebola Report: Ebola by the Numbers.”. URL:https://www.cdc.gov/about/ebola/ebola-by-the-numbers.html Centers for Disease Control and Prevention. 2019a. “2009 H1N1 Pandemic (H1N1pdm09 virus).”. URL: https://www.cdc.gov/flu/pandemic-resources/2009-h1n1-pandemic. html Centers for Disease Control and Prevention. 2019b. “2016 Case Counts in the US.”. URL:https://www.cdc.gov/zika/reporting/2016-case-counts.html Centers for Disease Control and Prevention. 2019c. “2017 Case Counts in the US.”. URL:https://www.cdc.gov/zika/reporting/2017-case-counts.html Centers for Disease Control and Prevention. 2020a. “Cases in the U.S.”. URL: https://www.cdc.gov/coronavirus/2019-ncov/cases-updates/ cases-in-us.html 103 Centers for Disease Control and Prevention. 2020b. “Opioid Overdose.”. URL:https://www.cdc.gov/drugoverdose/epidemic/index.html Chen, Jowei and Tim Johnson. 2015. “Federal Employee Unionization and Presidential Control of the Bureaucracy: Estimating and Explaining Ideological Change in Executive Agencies.” Journal of Theoretical Politics 27(1):151–174. Chernick, Howard A. 2014. The US Grant System. In Public Economics in the United States: How the Federal Government Analyzes and Influences the Economy , ed. Steven Payson. ABC-CLIO pp. 341–374. Christenson, Dino P., Douglas L. Kriner and Andrew Reeves. 2017. “All the President’s Senators: Presidential Copartisans and the Allocation of Federal Grants.” Legislative Studies Quarterly 42(2):269–294. Clinton, Joshua D., Anthony Bertelli, Christian R. Grose, David E. Lewis and David C. Nixon. 2012. “Separated Powers in the United States: The Ideology of Agencies, Presidents, and Congress.” American Journal of Political Science 56(2):341–354. Clinton, Joshua D, David E Lewis and Jennifer L Selin. 2014. “Influencing the bureaucracy: The irony of congressional oversight.” American Journal of Political Science 58(2):387–401. Cox, Gary W and Mathew D McCubbins. 2005. Setting the agenda: Responsible party government in the US House of Representatives. Cambridge University Press. Cox, Gary W and Mathew D McCubbins. 2007. Legislative leviathan: Party government in the House. Cambridge University Press. Dahlstr¨ om, Carl, Mih´ aly Fazekas and David E. Lewis. 2019. Agency Design, Favoritism and Procurement in the United States. Working Paper 2019:4 QoG Institute. Dasgupta, Nabarun, Leo Beletsky and Daniel Ciccarone. 2018. “Opioid Crisis: No Easy Fix to its Social and Economic Determinants.” American Journal of Public Health 108(2):182–186. Davidson, Roger H. 1989. “Multiple referral of legislation in the US Senate.” Legislative Studies Quarterly pp. 375–392. Davidson, Roger H, Walter J Oleszek and Thomas Kephart. 1988. “One bill, many committees: Multiple referrals in the US House of Representatives.” Legislative Studies Quarterly pp. 3–28. Dearborn, John A. 2021. Power Shifts: Congress and Presidential Representation. University of Chicago Press. Deering, Christopher J and Steven S Smith. 1997. Committees in congress. CQ Press. Dynes, Adam M. and Gregory A. Huber. 2015. “Partisanship and the Allocation of Federal Spend- ing: Do Same-party Legislators or V oters Benefit from Shared Party Affiliation with the Presi- dent and House Majority?” American Political Science Review 109(1):172–186. 104 Epstein, David and Sharyn O’Halloran. 1996. “Divided Government and the Design of Adminis- trative Procedures: A Formal Model and Empirical Test.” The Journal of Politics 58(2):373–397. Epstein, David and Sharyn O’Halloran. 1999. Delegating Powers: A Transaction Cost Politics Approach to Policy Making under Separate Powers. New York: Cambridge University Press. Epstein, Lee, Andrew D. Martin, Jeffrey A. Segal and Chad Westerland. 2007. “The Judicial Common Space.” The Journal of Law, Economics, and Organization 23(2):303–325. Evans, Diana. 2004. Greasing the Wheels: Using Pork Barrel Projects to Build Majority Coalitions in Congress. New York: Cambridge University Press. Farhang, Sean and Miranda Yaver. 2016. “Divided Government and the Fragmentation of Ameri- can Law.” American Journal of Political Science 60(2):401–417. Freeman, Jody and Jim Rossi. 2011. “Agency Coordination in Shared Regulatory Space.” Harvard Law Review 125:1131–1211. Freeman, Jody and Jim Rossi. 2012. “Improving Interagency Coordination in Shared Regulatory Space.” Admininstration and Regulatory Law News 38:11–14. Gailmard, Sean. 2009. “Multiple Principals and Oversight of Bureaucratic Policymaking.” Journal of Theoretical Politics 21(2):161–186. Gailmard, Sean and John W Patty. 2007. “Slackers and Zealots: Civil Service, Policy Discretion, and Bureaucratic Expertise.” American Journal of Political Science 51(4):873–889. Gailmard, Sean and John W. Patty. 2012. “Formal Models of Bureaucracy.” Annual Review of Political Science 15:353–377. Haeder, Simon F. and Susan Webb Yackee. 2015. “Influence and the Administrative Process: Lobbying the US President’s Office of Management and Budget.” American Political Science Review 109(3):507–522. Haeder, Simon F. and Susan Webb Yackee. 2018. “Presidentially Directed Policy Change: The Office of Information and Regulatory Affairs as Partisan or Moderator?” Journal of Public Administration Research and Theory 28(4):475–488. Hainmueller, Jens, Jonathan Mummolo and Yiqing Xu. 2019. “How Much Should we Trust Es- timates from Multiplicative Interaction Models? Simple Tools to Improve Empirical Practice.” Political Analysis 27:163–192. Hammond, Ben and Leah Rosenstiel. 2020. “Measuring the Influence of Political Actors on the Federal Budget.” American Political Science Review 00(0):1–6. Hammond, Thomas H. and Gary J. Miller. 1985. “A Social Choice Perspective on Expertise and Authority in Bureaucracy.” American Journal of Political Science 29(1):1–28. 105 Hammond, Thomas H and Jack H Knott. 1996. “Who controls the bureaucracy?: Presi- dential power, congressional dominance, legal constraints, and bureaucratic autonomy in a model of multi-institutional policy-making.” The Journal of Law, Economics, and Organization 12(1):119–166. Hammond, Thomas H and Jack H Knott. 1999. “Political institutions, public management and policy choice.” Journal of Public Administration Research and Theory 9(1):33–86. Hellevik, Ottar. 2009. “Linear versus Logistic Regression when the Dependent Variable is a Di- chotomy.” Quality & Quantity 43(1):59–74. Herrera, Helios, Ernesto Reuben and Michael M. Ting. 2017. “Turf Wars.” Journal of Public Economics 152:143–153. Hjern, Benny and David O. Porter. 1983. “Implementation Structures: A New Unit of Administra- tive Analysis.” Organization Studies 2(3):211–227. Huber, Gregory A. 2007. The Craft of Bureaucratic Neutrality: Interests and Influence in Govern- mental Regulation of Occupational Safety. Cambridge University Press. Huber, John D and Nolan McCarty. 2004. “Bureaucratic capacity, delegation, and political reform.” American Political Science Review pp. 481–494. Hudak, John. 2014. Presidential Pork: White House Influence over the Distribution of Federal Grants. Washington, D.C.: Brookings Institution Press. H¨ unermund, Paul and Beyers Louw. 2020. “On the Nuisance of Control Variables in Regression Analysis.” arXiv preprint arXiv:2005.10314 . Kaiser, Frederick M. 2011. “Interagency Collaborative Arrangements and Activities: Types, Ra- tionales, Considerations.” Congressional Research Service R41803. Keegan, Natalie. 2012. “Federal Grants-in-Aid Administration: A Primer.” Congressional Re- search Service . Kennedy, Joshua B. 2015. ““‘Do This! Do That!’ and Nothing Will Happen” Executive Orders and Bureaucratic Responsiveness.” American Politics Research 43(1):59–82. Kincaid, John. 2008. Federal Grants. In Political Encyclopedia of U.S. States and Regions, ed. Donald P. Haider-Markel, Michael A. Card, Ronald Keith Gaddie, Gary Moncrief and Ken- neth T. Palmer. V ol. 2 CQ Press pp. 764–766. Klaiman, Tamar, John D Kraemer and Michael A Stoto. 2011. “Variability in School Closure Decisions in Response to 2009 H1N1: A Qualitative Systems Improvement Analysis.” BMC Public Health 11(1):73. Krehbiel, Keith. 2010. Pivotal politics: A theory of US lawmaking. University of Chicago Press. Kriner, Douglas L. and Andrew Reeves. 2015. The Particularist President: Executive Branch Politics and Political Inequality. New York: Cambridge University Press. 106 Krutz, Glen and Courtney Cullison. 2008. “Multiple referral and US House legislative success in the 1990s.” American Review of Politics 29:65–81. Lee, Frances E. 2000. “Senate Representation and Coalition Building in Distributive Politics.” American Political Science Review 94(1):59–72. Lee, Frances E. 2002. “Representational Power and Distributive Politics: Senate Influence on Federal Transportation Spending.” US Senate Exceptionalism pp. 283–301. Lee, Frances E. 2003. “Geographic Politics in the US House of Representatives: Coalition Build- ing and Distribution of Benefits.” American Journal of Political Science 47(4):714–728. Lewis, David E. 2010. The Politics of Presidential Appointments: Political Control and Bureau- cratic Performance. Princeton, NJ: Princeton University Press. Lowande, Kenneth. 2018a. “Delegation or Unilateral Action?” The Journal of Law, Economics, and Organization 34(1):54–78. Lowande, Kenneth. 2018b. “Who Polices the Administrative State?” American Political Science Review 112(4):874–890. Lowande, Kenneth. 2019. “Politicization and Responsiveness in Executive Agencies.” Journal of Politics 81(1):33–48. Lowande, Kenneth and Jon C. Rogowski. 2020. “Presidential Unilateral Power.” Annual Review of Political Science . Lowande, Kenneth and Rachel Augustine Potter. 2021. “Congressional oversight revisited: Politics and procedure in agency rulemaking.” The Journal of Politics 83(1):401–408. Lowande, Kenneth S., Jeffery A. Jenkins and Andrew J. Clarke. 2018. “Presidential Particularism and US Trade Politics.” Political Science Research and Methods 6(2):265–281. MacDonald, Jason A. 2007. “Agency design and postlegislative influence over the bureaucracy.” Political Research Quarterly 60(4):683–695. Mayhew, David R. 1974. Congress: The Electoral Connection. V ol. 26 New Haven, CT: Yale University Press. Mayhew, David R. 2005. Divided We Govern: Party Control, Lawmaking and Investigations, 1946–2002. New Haven, CT: Yale University Press. McCann, Pamela J. Clouser and Charles R. Shipan. 2022. “How Many Major US Laws Delegate to Federal Agencies? (Almost) All of Them.” Political Science Research and Methods 10(2):438– 444. McCubbins, Mathew D., Roger G. Noll and Barry R. Weingast. 1987. “Administrative Procedures as Instruments of Political Control.” Journal of Law, Economics, & Organization 3(2):243–277. 107 McCubbins, Mathew D. and Thomas Schwartz. 1984. “Congressional Oversight Overlooked: Police Patrols versus Fire Alarms.” American Journal of Political Science 28(1):165–179. McGrath, Robert J. 2013. “Congressional oversight hearings and policy control.” Legislative Stud- ies Quarterly 38(3):349–376. McGuire, Michael. 2006. “Collaborative Public Management: Assessing What We Know and How We Know It.” Public Administration Review 66:33–43. Meldrum, Marcia L. 2016. “The Ongoing Opioid Prescription Epidemic: Historical Context.” American Journal of Public Health 106(8):1365. Miller, Gary J. and Andrew B. Whitford. 2016. Above Politics: Bureaucratic Discretion and Credible Commitment. New York: Cambridge University Press. Mood, Carina. 2010. “Logistic Regression: Why We Cannot Do What We Think We Can Do, and What We Can Do about It.” European sociological review 26(1):67–82. Mummolo, Jonathan and Erik Peterson. 2018. “Improving the Interpretation of Fixed Effects Regression Results.” Political Science Research and Methods 6(4):829–835. Napolio, Nicholas G. 2021. “Implementing presidential particularism: bureaucracy and the distri- bution of federal grants.” Political Science Research and Methods pp. 1–11. Napolio, Nicholas G. and Jordan Carr Peterson. 2019. “Their Boot in Our Face No Longer? Administrative Sectionalism and Resistance to Federal Authority in the US South.” State Politics & Policy Quarterly 19(1):101–122. Nemerever, Zoe and Melissa Rogers. Forthcoming. “Measuring the Rural Continuum in Political Science.” Political Analysis . Neustadt, Richard E. 1991. Presidential Power and the Modern Presidents: The Politics of Lead- ership from Roosevelt to Reagan. Simon & Schuster. Niskanen, William. 1971. Bureaucracy and Representative Democracy. London and New York: Routledge. Palus, Christine K and Susan W Yackee. 2022. “When does the multiple principals hypothesis hold? The politics of US agency policymaking autonomy.” Governance 35(1):43–64. Peterson, Jordan Carr. 2018. “All Their Eggs in One Basket? Ideological Congruence in Congress and the Bicameral Origins of Concentrated Delegation to the Bureaucracy.” Laws 7(2):19–33. Peterson, Jordan Carr and Nicholas G Napolio. Forthcoming. “Regulatory Productivity in Federal Administration: An Empirical Analysis of the Deep State.” Rutgers University Law Review . Potter, Rachel Augustine. 2017. “Slow-Rolling, Fast-Tracking, and the Pace of Bureaucratic De- cisions in Rulemaking.” The Journal of Politics 79(3):841–855. 108 Potter, Rachel Augustine. 2019. Bending the Rules: Procedural Politicking in the Bureaucracy. Chicago, IL: University of Chicago Press. Potter, Rachel Augustine and Charles R. Shipan. 2017. “Agency Rulemaking in a Separation of Powers System.” Journal of Public Policy 39(1):89–113. Public Health Service. 1919. “Annual Report of the Surgeon General of the Public Health Service of the United States for the Fiscal Year 1919.” (Document No. 436). Reingewertz, Yaniv and Thushyanthan Baskaran. 2019. “Distributive Spending and Presidential Partisan Politics.” Public Choice pp. 1–21. Resh, William, Saba Siddiki and Will R. McConnell. 2014. “Does the Network Centrality of Government Actors Matter? Examining the Role of Government Organizations in Aquaculture Partnerships.” Review of Policy Research 31(6):584–609. Rezaee, Janna King, Sean Gailmard and Abby Wood. 2019. Decentralized Legislative Oversight of Bureaucratic Policy Making. Technical report Technical report Working paper. Richardson, Mark D., Joshua D. Clinton and David E. Lewis. 2018. “Elite Perceptions of Agency Ideology and Workforce Skill.” The Journal of Politics 80(1):303–308. Rodr´ ıguez, Germ´ an and Noreen Goldman. 1995. “An Assessment of Estimation Procedures for Multilevel Models with Binary Responses.” Journal of the Royal Statistical Society: Series A (Statistics in Society) 158(1):73–89. Rogowski, Jon C. 2016. “Presidential Influence in an Era of Congressional Dominance.” American Political Science Review 110(2):325–341. Rohde, David W. 1991. Parties and leaders in the postreform House. University of Chicago Press. Rudalevige, Andrew. 2012. “The contemporary presidency: Executive orders and presidential unilateralism.” Presidential Studies Quarterly 42(1):138–160. Saito, Takaharu. 2020. “Supervision among Agencies for the President: Selection of Agencies through Interagency Coordination.” Working Paper . URL:https://drive.google.com/file/d/15nOsnQootsITJnufBd4PgQOmaLWQbOVK/ view Schiller, Reuel E. 2007. “The era of deference: Courts, expertise, and the emergence of new deal administrative law.” Mich. L. Rev. 106:399. Selin, Jennifer L. 2015. “What Makes an Agency Independent?” American Journal of Political Science 59(4):971–987. Shipan, Charles R. 2004. “Regulatory Regimes, Agency Actions, and the Conditional Nature of Congressional Influence.” American Political Science Review 98(3):467–480. Siddiki, Saba, Jangmin Kim and William D. Leach. 2017. “Diversity, Trust, and Social Learning in Collaborative Governance.” Public Administration Review 77(6):863–874. 109 Sinclair, Barbara. 2016. Unorthodox lawmaking: New legislative processes in the US Congress. CQ Press. Spence, Michael. 1978. Job Market Signaling. In Uncertainty in Economics. Elsevier pp. 281–306. Ting, Michael M. 2003. “A Strategic Theory of Bureaucratic Redundancy.” American Journal of Political Science 47(2):274–292. U. S. Congress. 1918. “Prevention of the Spanish Influenza.” Congressional Record. . 65th Cong., 2nd sess.. U. S. Congress. 2020. “Coronavirus.” Congressional Record. . 116th Cong., 2nd sess.. V olden, Craig. 2002. “A Formal Model of the Politics of Delegation in a Separation of Powers System.” American Journal of Political Science 46(1):111–133. Weingast, Barry R and William J Marshall. 1988. “The industrial organization of Congress; or, why legislatures, like firms, are not organized as markets.” Journal of Political Economy 96(1):132– 163. Wood, Abby K. and David E. Lewis. 2017. “Agency Performance Challenges and Agency Politi- cization.” Journal of Public Administration Research and Theory 27(4):581–595. Woodward, Bob. 2020. Rage. New York: Simon & Schuster. Wooldridge, Jeffrey M. 2010. Econometric Analysis of Cross Section and Panel Data. Cambridge, MA: MIT press. Woolley, John T. 1993. “Conflict among regulators and the hypothesis of congressional domi- nance.” The Journal of Politics 55(1):92–114. Young, Garry. 1996. “Committee Gatekeeping and Proposal Power under Single and Multiple Referral.” Journal of Theoretical Politics 8(1):65–78. 110 Appendices A Appendix to Chapter 2 A.1 Agencies in Analysis The following agencies are included in the analysis. Table A.1: Agencies in Analysis Agency for International Development Air Force Army Department of Agriculture Department of Commerce Department of Defense Department of Education Department of Energy Department of Health and Human Services Department of Homeland Security Department of Housing and Urban Development Department of Justice Department of Labor Department of the Interior Department of the Treasury Department of Transportation Department of Veterans Affairs Environmental Protection Agency Navy Nuclear Regulatory Commission Small Business Administration A.2 Testing Linearity The data meet the assumptions of multiplicative interaction models. First, the marginal effect of presidential co-partisan on outlays is linear in agency–president distance, and second, there is common support for presidential co-partisan across agency–president distance (Hainmueller, Mummolo and Xu 2019). Figure A.1 displays diagnostics for the multiplicative interaction as- sumptions. The five dot-and-whiskers represent the binned estimates ( i.e., not assuming functional 111 form) and the lines and ribbons represent the marginal effect of presidential co-partisan over agency–president distance assuming linearity.The bins fit quite well with the lines and ribbons -5.0 -2.5 0.0 2.5 5.0 -5.0 -2.5 0.0 2.5 Agency-President Distance Marginal Effect of Presidential Co-Partisanship Figure A.1: Testing Linearity. Figure created with the interflex package developed by Hain- mueller, Mummolo and Xu (2019). and are monotonically decreasing, indicating that the marginal effect of presidential co-partisan approaches linearity over agency–president distance. The histograms at the bottom of the graphs indicate that, across all values of agency–president distance, observations share common support for placement into either presidential co- or contra-partisanship, meaning placement into presiden- tial co-partisan is not restricted to certain levels of agency–president distance. A.3 Fixed Effects Adjustment Interpreting substantive effects from these models with dual fixed effects requires some additional explanation. First, presidential co-partisan is about evenly distributed across observations, so within each agency-legislator a shift from not shared to shared partisanship with the president is plausible (see Figure A.1 for visual evidence of common support). agency–president distance on the other hand, is constrained within agencies, calling for a more nuanced discussion of the sub- stantive effects. Employing the method proposed by Mummolo and Peterson (2018) (residualizing 112 After FEs SD = 0.102 Original Distribution SD = 0.263 0 2 4 6 −0.50 −0.25 0.00 0.25 Agency−President Distance (Centered on Zero) Density Figure A.2: Fixed Effects Adjustments. agency–president distance with respect to the agency-legislator and Congress fixed effects), I am able to identify a plausible counterfactual (see figure A.2). The standard deviation of the residu- alized values of agency–president distance with respect to the fixed effects is 0.102 (the standard deviation of agency–president distance before adjusting for the fixed effects is 0.263). This repre- sents a typical deviation from the mean of agency–president distance. Further, Figure B.4 shows the distribution of within-agency ranges in agency–president dis- tance, which has a median of 0.639, indicating that, on average, the within-agency range of agency–president distance is about 0.639. The substantive effect of agency–president distance and its interaction with presidential co-partisan may therefore be derived from a counterfactual move of about 0.639. 113 Median 95th Percentile 0 1 2 3 4 5 0.25 0.50 0.75 Within−Agency Range in Agency−President Distance Count Figure A.3: Within-Agency Variation. 114 A.4 Reanalysis with Alternative Clusters In the main analysis, I report heteroskedasticity-corrected errors clustered by agency-legislator since the variation in the independent variables occurs at this level. However, the results from the main analyses are robust to estimating heteroskedasticity-corrected errors clustered by agency or legislator. Table A.2 reports results from estimating the main models clustering standard errors by agency (models 1 and 2) and legislator (models 3 and 4). Table A.2: Reanalysis with Alternative Clusters Dependent variable: Logged Outlays Clustered by Agency Clustered by Legislator (1) (2) (3) (4) Presidential 3.665 ∗∗∗ 3.479 ∗∗∗ 3.665 ∗∗∗ 3.479 ∗∗∗ Co-Partisan (0.612) (1.109) (0.560) (1.321) Agency–President 2.822 2.959 2.822 ∗∗∗ 2.959 ∗∗ Distance (2.101) (2.112) (0.573) (1.254) Pres. Co-Partisan − 6.491 ∗∗∗ − 6.926 ∗∗∗ − 6.491 ∗∗∗ − 6.926 ∗∗∗ Ag.–Pres. Dist. (1.185) (2.039) (1.069) (2.445) β 2 +β 3 − 3.668 ∗ − 3.966 ∗ − 3.668 ∗∗∗ − 3.966 ∗∗∗ (2.234) (2.464) (0.549) (1.215) Congress FEs YES YES YES YES Agency-Legislator FEs YES YES YES YES Time-Varying Covariates YES YES Observations 63,075 63,075 63,075 63,075 Adjusted R 2 0.591 0.596 0.591 0.596 ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 Note: Unit of analysis is the agency-district-Congress. Models 2 and 4 control for whether each member of Congress in each Congress is in the majority party, sits on the appropriations committee, sits on the ways and means committee, whether each member of Congress won their previous election with a margin less than 0.05, each district’s logged population and logged median income, and for each agency’s politicization ratio for each Congress, and the distance between each agency’s Chen and Johnson (2015) ideal point estimate and each member’s DW-NOMINATE ideal point estimate. 115 A.5 Reanalysis with High-Variance Programs This section replicates the main results but subsetting outlays to only those disbursed pursuant to high-variance programs. Table A.3: Reanalysis with High-Variance Programs Dependent variable: Logged Outlays (1) (2) Presidential 1.303 ∗∗∗ − 0.384 Co-Partisan (0.165) (0.310) Agency–President 1.687 ∗∗∗ 0.243 Distance (0.195) (0.308) Pres. Co-Partisan× − 2.361 ∗∗∗ 0.400 Ag.–Pres. Dist. (0.326) (0.570) β 2 +β 3 − 0.675 ∗∗∗ 0.642 ∗ (0.193) (0.298) Congress FEs YES YES Agency-Legislator FEs YES YES Time-Varying Covariates YES Observations 63,075 63,075 Adjusted R 2 0.587 0.591 ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001 Note: Unit of analysis is the agency-district-Congress. Heteroskedasticity-corrected errors clustered by agency- legislator reported in parentheses. Model 2 controls for whether each member of Congress in each Congress is in the majority party, sits on the appropriations committee, sits on the ways and means committee, whether each member of Congress won their previous election with a margin less than 0.05, each district’s logged population and logged median income, and for each agency’s politicization ratio for each Congress, and the distance between each agency’s Chen and Johnson (2015) ideal point estimate and each member’s DW-NOMINATE ideal point estimate. High-variance programs are defined as those with a coefficient of variance greater than 0.75 (?). Using high-variance programs as a proxy for programs over which agencies have discretion is suboptimal since it conflates high-variance formula grants with agency-allocated program grants, and excludes low-variance program grants. While model 1 in table A.3 reports similar results to the main specification, the model with covariates does not. Since some formula grants are included 116 in this model, the legislator-level variables swamp out the agency-level ones, since Congress is ultimately responsible for allocating those grants. A.6 Reanalysis Excluding Defense Agencies There is notable missingness in FAADS data with respect to defense agencies (see, e.g., Hammond and Rosenstiel 2020), Table A.4: Reanalysis Excluding Defense Agencies Dependent variable: Logged Outlays Program Grants Formula Grants (Placebo) (1) (2) (3) (4) Presidential 3.814 ∗∗∗ 3.278 ∗∗∗ 2.331 ∗∗∗ 0.259 Co-Partisan (0.271) (0.731) (0.274) (0.677) Agency–President 2.376 ∗∗∗ 1.978 ∗∗ 6.248 ∗∗∗ 4.017 ∗∗∗ Distance (0.479) (0.779) (0.446) (0.709) Pres. Co-Partisan× − 7.002 ∗∗∗ − 6.741 ∗∗∗ − 4.336 ∗∗∗ − 0.777 Ag.–Pres. Dist. (0.558) (1.335) (0.555) (1.240) β 2 +β 3 − 4.627 ∗∗∗ − 4.764 ∗∗∗ 1.912 ∗∗∗ 3.240 ∗∗∗ (0.487) (0.763) (0.439) (0.703) Congress FEs YES YES YES YES Agency-Legislator FEs YES YES YES YES Time-Varying Covariates YES YES Observations 50,895 50,895 50,895 50,895 Adjusted R 2 0.599 0.603 0.614 0.616 ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 Note: Unit of analysis is the agency-district-Congress. Heteroskedasticity-corrected errors clustered by agency- legislator reported in parentheses. Models 2 and 4 control for whether each member of Congress in each Congress is in the majority party, sits on the appropriations committee, sits on the ways and means committee, whether each member of Congress won their previous election with a margin less than 0.05, each district’s logged population and logged median income, and for each agency’s politicization ratio for each Congress, and the distance between each agency’s Chen and Johnson (2015) ideal point estimate and each member’s DW-NOMINATE ideal point estimate. 117 so in this section, I reanalyze the data but dropping the four defense agencies in the main sample (Department of Defense, Air Force, Army, and Navy). The results are robust to dropping defense agencies and the reanalysis passes the placebo test as the main analysis does, suggesting that the main results are not driven by data that are missing systematically. B Appendix to Chapter 3 118 B.1 Agencies in Analysis The following agencies are included in the analysis. Bold agencies are those included in the main analyses. Table B.1: Agencies in Analysis Agriculture Department Labor Department Commerce Department Merit Systems Protection Board Commodity Futures Trading Commission National Aeronautics and Space Administration Consumer Product Safety Commission National Credit Union Administration Defense Department National Labor Relations Board Education Department National Science Foundation Energy Department National Transportation Safety Board Environmental Protection Agency Nuclear Regulatory Commission Equal Employment Opportunity Commission Peace Corps Export-Import Bank Pension Benefit Guaranty Corporation Farm Credit Administration Personnel Management Office Federal Communications Commission Railroad Retirement Board Federal Deposit Insurance Corporation Securities and Exchange Commission Federal Election Commission Selective Service System Federal Labor Relations Authority Small Business Administration Federal Trade Commission Smithsonian Institution General Services Administration Social Security Administration Health and Human Services Department Transportation Department Housing and Urban Development Department Treasury Department Interior Department Homeland Security Department International Trade Commission Veterans Affairs Department Justice Department Note: Bolded agencies are those used in the main analysis, all agencies are used in the analyses in appendix B.5. B.2 Power Dependency Some literature in the public administration tradition argues that strong agents force weaker ones into coalitions to further the goals of the stronger ones (see, e.g., Hjern and Porter 1983). System- atic analysis shows no support for the claim that more politicized agencies, agencies ideologically aligned with the President, or large agencies are more likely to be central in the network, contrary to the power dependency model of networked implementation (Hjern and Porter 1983). The power 119 dependency theory predicts powerful agencies important to or aligned with the President, those with more political capital and power, will induce agencies with less political capital to collab- orate in order to further the goals of the more powerful agencies. Each of these three variables reasonably proxies for political importance to the President and political capital generally. Politi- cization affords agencies a more direct line to the President and the President generally politicizes those agencies important to their political success (Lewis 2010), ideological alignment with the President may provide leverage in negotiations with other agencies, and a large workforce affords agencies greater capacity. However, I find no evidence to suggest power dependency explains agency decisions to collaborate. Table B.2: Centrality and Political Capital, 1998-2012 Dependent variable: Degree Betweenness Logged Strength (1) (2) (3) (4) (5) (6) Politicization − 0.034 − 0.132 0.009 0.038 − 0.135 − 0.037 (0.409) (1.241) (0.016) (0.024) (0.229) (0.211) Ideological Proximity 1.205 0.711 0.072 0.066 0.347 0.188 to President (1.717) (1.591) (0.063) (0.063) (0.527) (0.305) Logged Employees 0.961 ∗∗∗ − 6.437 ∗ 0.040 ∗ − 0.002 0.431 ∗∗∗ − 0.885 (0.126) (2.641) (0.016) (0.049) (0.081) (0.631) Mean & St. Dev. of 3.357 0.085 1.379 Dependent Variable (5.563) (0.228) (1.703) Observations 465 465 465 465 465 465 Agencies 32 32 32 32 32 32 Year FEs YES YES YES YES YES YES Agency FEs NO YES NO YES NO YES Adjusted R 2 0.471 0.532 0.111 0.531 0.332 0.809 ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001 Note: Unit of analysis is the agency-year. Heteroskedasticity-corrected standard errors clustered by agency reported in parentheses. Table B.2 displays estimates from least squares models regressing measures of centrality on each agency’s politicization ratio (the ratio of political appointees to careerists), 13 ideological prox- imity to the President (the negative Euclidean distance between the President’s DW-NOMINATE ideal point estimate and the agency’s Chen and Johnson (2015) ideal point estimate), the logged 13 I measure politicization as the ratio of political appointees over the number of career senior executive service members following previous work (see, e.g., Lewis 2010; Lowande 2019). 120 count of employees within that agency, and fixed effects for year in all models and agency in even models. Degree is measured as the number of other agencies each agency has promulgated at least one joint rule with, betweenness is a measure of how well each agency connects other agencies to the network, 14 and strength is a measure of the total count of rules each agency has promulgated with other agencies. I normalize betweenness within-year to lie between zero and one and take the natural logarithm of strength. Table B.2 shows no evidence of a positive relationship between any of the measures of and centrality and is thus inconsistent with the power dependency hypothesis. The logged count of employees is the only variable with a consistently positive and significant relationship with net- work centrality with a standard deviation increase in employees resulting in effects of 0.33 stan- dard deviations of degree, 0.35 standard deviations of betweenness, and 0.05 of logged strength. The relationship does not hold when including agency fixed effects and therefore leveraging only within-agency variation in employment. Last, in the main regression analysis in the text, the coefficients on the control variables fur- ther cast doubt on the ability of power dependency theory to explain coalition formation. The agency-level analysis in table B.2 showed that measures of political capital do not predict network centrality, and the dyad-level analysis presented in the main analysis in the manuscript shows that differences in these measures of political capital (politicization and employment) do not predict the promulgation of joint rules. If the power dependency theory did explain the data, we might expect small agencies to collaborate often with large ones, and non-politicized agencies to collaborate 14 Formally, betweenness is calculated as: B(v)= ∑ s̸=v̸=t σ st (v) σ st where σ st is the number of shortest paths between nodes s and t and σ st (v) is the number of those paths that pass through node v. I then normalize the measure within-year: normal(B t (v))= B t (v)− min(B t ) max(B t )− min(B t ) 121 often with politicized ones, which would suggest the larger more politicized agencies inducing or coercing cooperation. Instead, the strategic theory presented here better explains the data. 122 B.3 Robustness Checks B.4 Reanalysis with Logit Estimator I justify in the text why I use a linear probability model, but table B.3 shows that main results on the interaction are robust to a logit specification as well. The number of observations is lower since the logit estimator drops any dyads that always or never formed a coalition due to the fixed effects. Table B.3: Reanalysis with Logit Estimator Dependent variable: Coalition Formation (1) (2) (3) (4) Agency 2.711 1.966 − 4.223 − 5.814 Alignment (1.575) (1.696) (2.861) (3.357) Presidential − 2.541 − 2.867 − 7.502 ∗∗ − 8.578 ∗∗ Misalignment (1.594) (1.656) (2.516) (2.958) Agency Alignment× 10.526 ∗ 19.409 ∗ Pres. Misalignment (5.153) (8.211) Overlapping − 0.039 − 0.069 Laws (0.060) (0.065) Presidential − 1.352 ∗ − 1.379 ∗ Attention (0.568) (0.582) House − 30.061 − 29.109 Misalignment (15.362) (14.946) Court 37.516 ∗∗ 38.598 ∗∗ Misalignment (13.487) (13.375) Log(Total Rules) − 6.395 ∗∗∗ − 7.159 ∗∗∗ (1.588) (1.592) Employment − 0.240 − 0.361 Difference (0.892) (0.929) Politicization − 1.020 − 1.445 Difference (1.989) (1.925) Average − 2.292 ∗∗∗ − 2.101 ∗∗∗ Politicization (0.591) (0.570) Observations 858 858 858 858 Dyad & Term FEs YES YES YES YES Log Likelihood − 303.529 − 300.979 − 219.893 − 215.488 Note: ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001 Note: Unit of analysis is the dyad-year. Heteroskedasticity-corrected standard errors clustered by dyad reported in parentheses. 123 B.5 Reanalysis with All Dyads The main text reports results culling dyads to only those that are politically relevant, that is, likely able to form coalitions in the first place. Those agencies were selected by eliminating any agencies that did not employ at least one career Senior Executive Service manager from 1996–2012. The results in table B.4 shows that the results hold when using all agencies. Expanding the dataset to irrelevant dyads deflates the mean of the dependent variable to 0.154 with a standard deviation of 0.361. Table B.4: Reanalysis with All Dyads Dependent variable: Coalition Formation (1) (2) (3) (4) Agency 0.036 0.021 0.040 0.009 Alignment (0.032) (0.033) (0.041) (0.044) Presidential − 0.065 − 0.096 − 0.113 ∗ − 0.128 ∗ Misalignment (0.049) (0.051) (0.056) (0.056) Agency Alignment× 0.466 ∗∗∗ 0.429 ∗∗ Pres. Misalignment (0.134) (0.141) Overlapping − 0.009 ∗∗∗ − 0.009 ∗∗∗ Laws (0.002) (0.002) Presidential − 0.010 − 0.011 Attention (0.013) (0.013) House 0.175 0.100 Misalignment (0.130) (0.135) Court − 0.121 − 0.044 Misalignment (0.093) (0.100) Log(Total Rules) − 0.023 − 0.024 (0.018) (0.018) Employment − 0.002 − 0.005 Difference (0.026) (0.026) Observations 3,570 3,570 3,570 3,570 Dyad & Term FEs YES YES YES YES Adjusted R 2 0.460 0.462 0.479 0.481 ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001 Note: Unit of analysis is the dyad-year. Heteroskedasticity-corrected standard errors clustered by dyad reported in parentheses. 124 B.6 Year-Level Analysis The analyses in the main text aggregate data to the presidential term level, but I indicated in the text that all results are robust to year-level analyses. Table B.5 displays results from the same specifications as the main analysis in the text but with a dyad-year dataset. The politicization ratio for some agencies is undefined in certain years because they employed no career Senior Executive Service managers, leading to an undefined politicization ratio. Table B.5: Reanalysis at the Year Level Dependent variable: Coalition Formation (1) (2) (3) (4) Agency 0.053 0.050 − 0.060 − 0.089 ∗ Alignment (0.029) (0.029) (0.039) (0.040) Presidential − 0.042 − 0.055 − 0.057 − 0.087 ∗ Misalignment (0.040) (0.041) (0.042) (0.044) Agency Alignment× 0.245 ∗ 0.643 ∗∗∗ Pres. Misalignment (0.112) (0.126) Overlapping − 0.019 ∗∗∗ − 0.020 ∗∗∗ Laws (0.004) (0.004) Presidential − 0.040 ∗∗∗ − 0.041 ∗∗∗ Attention (0.009) (0.009) Court 0.243 ∗∗∗ 0.302 ∗∗∗ Misalignment (0.067) (0.070) House − 0.105 − 0.132 ∗ Misalignment (0.064) (0.065) Log(Total Rules) − 0.061 ∗∗∗ − 0.067 ∗∗∗ (0.013) (0.013) Employment 0.007 0.003 Difference (0.013) (0.013) Politicization 0.009 0.010 Difference (0.019) (0.018) Average − 0.018 − 0.015 Politicization (0.018) (0.018) Observations 7,750 7,750 6,867 6,867 Dyad & Term FEs YES YES YES YES Adjusted R 2 0.385 0.385 0.390 0.393 ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001 Note: Unit of analysis is the dyad-year. Heteroskedasticity-corrected standard errors clustered by dyad reported in parentheses. 125 B.7 Count Dependent Variable The analyses in the main text use a binary measure of coalition, but I reported that all analyses were robust to a count measure of the coalition formation. Table B.6 displays results from estimating the same models as the main text but where the dependent variable is the logged count (plus one) of the number of joint rules promulgated by each agency dyad-term. The mean of the dependent variable is 0.307 and the standard deviation is 0.702. Table B.6: Reanalysis with Count Dependent Variable Dependent variable: Logged Count of Joint Rules (1) (2) (3) (4) Agency 0.088 0.085 − 0.176 − 0.210 Alignment (0.078) (0.077) (0.111) (0.116) Presidential − 0.093 − 0.136 − 0.070 − 0.123 Misalignment (0.105) (0.108) (0.104) (0.105) Agency Alignment× 0.626 ∗ 1.111 ∗∗∗ Pres. Misalignment (0.288) (0.315) Overlapping − 0.011 ∗∗∗ − 0.011 ∗∗∗ Laws (0.003) (0.003) Presidential − 0.0001 − 0.001 Attention (0.023) (0.022) House − 0.284 − 0.388 Misalignment (0.374) (0.383) Court 0.741 ∗ 0.885 ∗∗ Misalignment (0.301) (0.315) Log(Total Rules) − 0.118 ∗∗ − 0.135 ∗∗ (0.045) (0.044) Employment 0.006 − 0.003 Difference (0.051) (0.051) Politicization − 0.125 − 0.126 Difference (0.074) (0.075) Average − 0.003 0.003 Politicization (0.017) (0.017) Observations 1,953 1,953 1,953 1,953 Dyad & Term FEs YES YES YES YES Adjusted R 2 0.716 0.717 0.731 0.733 ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001 Note: Unit of analysis is the dyad-year. Heteroskedasticity-corrected standard errors clustered by dyad reported in parentheses. 126 B.8 Reanlysis on Agencies with at Least One Overlapping Law The analyses in the main text subset the data only to those dyads where each agency employed at least one career Senior Executive Service manager from 1996–2012. An alternative subsetting of the data to cull the analysis to only politically relevant dyads is to only include dyads where at least one significant law since 1947 delegates to both agencies. The results in table B.7 shows that the results hold when using only agencies with at least one overlapping law. Table B.7: Reanalysis with Agencies with Overlapping Laws Dependent variable: Coalition Formation (1) (2) (3) (4) Agency 0.131 0.091 − 0.144 − 0.211 Alignment (0.100) (0.102) (0.115) (0.117) Presidential − 0.558 ∗∗∗ − 0.614 ∗∗∗ − 0.511 ∗∗∗ − 0.572 ∗∗∗ Misalignment (0.136) (0.137) (0.144) (0.144) Agency Alignment× 1.015 ∗∗ 1.468 ∗∗∗ Pres. Misalignment (0.321) (0.330) Overlapping − 0.005 ∗ − 0.005 ∗∗ Laws (0.002) (0.002) Presidential − 0.072 ∗∗ − 0.072 ∗∗ Attention (0.026) (0.025) Court 1.511 ∗∗∗ 1.569 ∗∗∗ Misalignment (0.397) (0.395) House − 1.185 ∗ − 1.159 ∗ Misalignment (0.475) (0.469) Log(Total Rules) − 0.149 ∗∗ − 0.180 ∗∗∗ (0.056) (0.055) Employment − 0.006 − 0.015 Difference (0.052) (0.052) Politicization − 0.055 − 0.069 Difference (0.093) (0.093) Average − 0.026 − 0.014 Politicization (0.022) (0.023) Observations 1,584 1,584 1,584 1,584 Dyad & Term FEs YES YES YES YES Adjusted R 2 0.448 0.452 0.477 0.485 ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001 Note: Unit of analysis is the dyad-presidential term. Heteroskedasticity-corrected standard errors clustered by dyad reported in parentheses. 127 B.9 Reanlysis with Alternative Operationalization of Presidential Misalignment The analyses in the main text use the average distance between each agency forming a dyad and the president in a given term. To show that the results are not driven by quirks in using an average as an independent variable, table B.8 estimates the same models as in the main analysis, but op- erationalizes presidential misalignment as the ideological distance between the agency closest in ideological space to the President and the President. Table B.8: Reanalysis with Alternative Operationalization of Presidential Misalignment Dependent variable: rule (1) (2) (3) (4) Agency 0.197 ∗∗ − 0.043 0.109 − 0.282 Alignment (0.074) (0.102) (0.102) (0.144) Presidential − 0.192 − 0.250 ∗ − 0.207 ∗ − 0.262 ∗∗ Misalignment (0.099) (0.100) (0.100) (0.102) Agency Alignment× 0.774 ∗∗ 1.074 ∗∗∗ Pres. Misalignment (0.241) (0.255) Overlapping − 0.010 ∗∗∗ − 0.010 ∗∗∗ Laws (0.002) (0.002) Presidential − 0.044 ∗ − 0.047 ∗ Attention (0.020) (0.020) House 0.293 0.100 Misalignment (0.265) (0.279) Court 0.086 0.302 Misalignment (0.219) (0.237) Log(Total Rules) − 0.105 ∗∗ − 0.119 ∗∗ (0.040) (0.040) Employment − 0.007 − 0.013 Difference (0.044) (0.043) Politicization − 0.071 − 0.059 Difference (0.066) (0.065) Observations 1,953 1,953 1,953 1,953 Dyad & Term FEs YES YES YES YES Adjusted R 2 0.453 0.455 0.478 0.483 ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001 Note: Unit of analysis is the dyad-presidential term. Heteroskedasticity-corrected standard errors clustered by dyad reported in parentheses. 128 B.10 Delete-a-Group Jackknife This section shows the distribution of coefficients for the interaction term between agency align- ment and presidential misalignment estimated from models with the same specification as Model 4 in the main text when dropping one dyad at a time one agency at a time. With grouped data, this process allows me to check whether results are driven by one agency or dyad. 0 10 20 30 40 50 0.0 0.5 1.0 1.5 2.0 Estimated Coefficient Dropping One Dyad at a Time 0 1 2 3 0.0 0.5 1.0 1.5 2.0 Estimated Coefficient Dropping One Agency at a Time Figure B.1: Delete-a-Group Jackknife Distribution. Each panel displays the distribution of coefficients estimated using the same specification as model 4 in the main text but dropping either one dyad or agency at a time. Figure B.1 displays the distributions of coefficients estimated from these specifications. Solid black lines display the main coefficient reported in the text and dotted black lines display the mean value of the distribution coefficients recovered from using delete-a-group jackknife resampling. The two lines are almost identical in each panel. Taken together, these analyses indicate that the results I report in the main text are not driven by certain dyads or agencies. Figure B.2 displays the estimated coefficients and 95% confidence intervals recovered from the models dropping one dyad or agency at a time. 129 0.0 0.5 1.0 1.5 2.0 2.5 Estimated Coefficient Dropping One Dyad at a Time 0.0 0.5 1.0 1.5 2.0 2.5 Estimated Coefficient Dropping One Agency at a Time Figure B.2: Delete-a-Group Jackknife Coefficients. Each panel displays the coefficients and 95% confi- dence intervales estimated using the same specification as model 4 in the main text but dropping either one dyad or agency at a time. 130 B.11 Fixed Effects Adjustments When drawing inferences about the substantive effects of some independent variable using panel data, it is especially important to consider how the independent variable is distributed within units in order to avoid drawing inferences from extrapolation. In order to identify a plausible counter- factual change in agency alignment and presidential misalignment, I follow the procedure outlined by Mummolo and Peterson (2018). First, I compare the distributions of the independent variables before and after absorbing variation from the dyad and year fixed effects. Figure B.3 displays the original distributions in grey and the distributions after absorbing variation from the fixed effects in red. Agency alignment experiences a small reduction in variance when adjusting for the fixed effects, while more of the variation in presidential misalignment can be explained by the dyad and year fixed effects, suggesting attenuation of the substantive effects, like I report in the main text. After FEs SD=0.103 Original Distribution SD=0.135 0 1 2 3 4 −1.0 −0.5 0.0 0.5 1.0 Density Agency Alignment (Centered on Zero) After FEs SD=0.087 Original Distribution SD=0.307 0 1 2 3 4 5 −0.6 −0.3 0.0 0.3 0.6 Density Presidential Misalignment (centered on Zero) Figure B.3: Independent Variable Distributions. The gray plots display the distribution of agency align- ment (left) and presidential misalignment (right) without accounting for the dyad and year fixed effects, while the red plots display the distribution of the independent variables after adjusting for the dyad and year fixed effects by residualizing the variables with respect to the fixed effects (Mummolo and Peterson 2018). Identifying an appropriate counterfactual is aided by one further step: finding the average, within-unit range in the independent variables. Figure B.4 displays the distribution of within-dyad 131 ranges in the independent variables. The left panel shows that the median within-dyad range in agency alignment is about 0.21, and the right panel shows that the median within-dyad range in presidential misalignment is about 0.72. Thus, each coefficient in the main tables can be multiplied by these values in order to estimate the effect of the median, maximum within-unit shift in the independent variables. Median 95th Percentile 0 50 100 150 200 0.0 0.2 0.4 0.6 Frequency Within−Dyad Range of Agency Alignment Median 95th Percentile 0 50 100 150 200 250 0.4 0.6 0.8 1.0 Count Within−Dyad Range of Presidential Misalignment Figure B.4: Within-Dyad Ranges of Independent Variables. The distributions here do not account for the dyad and year fixed effects, but they do display the distribution of within-dyad ranges of agency alignment (left) and presidential misalignment (right), thus providing a useful counterfactual with which to estimate substantive effects (Mummolo and Peterson 2018). Dotted lines represent the median and 95th percentiles of the distribution to allow readers to determine an appropriate counterfactual if the one discussed in the text (median) is unconvincing. 132

Abstract (if available)

Abstract Bureaucrats are political actors who pursue political power. They have preferences over policies, knowledge of the procedures legally allowed to them, and expertise on how to combine those preferences with that knowledge of procedures to get what they want -– even in the face of opposition from their overseers. However, bureaucrats derive no formal, legal authority independent of grants of authority from their overseers. Therefore, bureaucrats face a dilemma.

This dissertation examines the tools bureaucrats use to manage this dilemma and build power even in the face of opposition from political overseers. Specifically, I ask how bureaucrats use different strategies in different situations to build power and manage this dilemma. I argue that agencies often push back against elected officials, are often successful at locking their preferred policies in place, and have been able to build power despite sometimes significant opposition from their overseers. The power-building tools bureaucrats use to push back against overseers, I argue, can be roughly divided into one of three types: procedure, persuasion, and politics.

I make at least three major contributions. First, I expand the scope of inquiry into bureaucratic politics by conceiving of federal agencies as a constellation of entities able to form coalitions and work together. Second, I argue that bureaucrats use explicitly political strategies. Finally, I argue that bureaucrats often stymie efforts of their overseers to enforce overseers preferred policies, in contrast to many scholars of legislative politics who often view bureaucrats as simple agents of the legislature.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Private interests in American government institutions

PDF

Can election reforms improve democracy?

PDF

The strategy of firms towards political elites and consumers in polarized western democracies

PDF

Three essays on the political economy of federal spending

PDF

Signaling identity: how race and gender shape what representatives say online

PDF

Voices unheard, stories untold: Black women, police violence and political participation

PDF

La eleccion de la pandemia: analyzing Latino political behavior during the 2020 election

PDF

Like father, like son? A succession-based explanation for conflict initiation by authoritarian regimes

PDF

The small dollar political donor: why regular folks give money to politics

PDF

"Hum so the devil doesn't hear you": rural Black Southern political existences as the otherwise

PDF

Challenges from below: the origins of status competitions in world politics

PDF

Politics is something we do together: identity and institutions in U.S. elections

PDF

International politics and domestic institutional change: the rise of executive war-making autonomy in the United States

PDF

Developing countries in the digital era: state and business interactions for industrial upgrading

PDF

Habits and policy: the socio-cognitive foundations of foreign policy stability

PDF

When passions run high: emotions and the communication of intentions in face-to-face diplomacy

PDF

The paradox of personalism: internal political constraints on nuclear posture in authoritarian regimes

PDF

Pernicious racism and the impacts of intentional racist in a color-blind society

PDF

In/visible constituents: the representation of undocumented immigrants

PDF

Securitizing language: political elites and language policy securitization in South Korea

Asset Metadata

Creator Napolio, Nicholas Gianni (author)

Core Title Bureaucratic politics and power building in the administrative state

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Political Science and International Relations

Degree Conferral Date 2023-05

Publication Date 03/15/2023

Defense Date 03/14/2023

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag American politics,Bureaucracy,OAI-PMH Harvest,political institutions,Political science,Public administration

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Grose, Christian R. (committee chair), Jenkins, Jeffery A. (committee chair), McCann, Pamela J. Clouser (committee member), Potter, Rachel Augustine (committee member)

Creator Email napolio@usc.edu,NicholasNapolio@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC112839108

Unique identifier UC112839108

Identifier etd-NapolioNic-11504.pdf (filename)

Legacy Identifier etd-NapolioNic-11504

Document Type Dissertation

Format theses (aat)

Rights Napolio, Nicholas Gianni

Internet Media Type application/pdf

Type texts

Source 20230320-usctheses-batch-1010 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu