Polarized America, second edition: The Dance of Ideology and Unequal Riches

The Walras-Pareto Lectures, at the École des Hautes Études Commerciales, Université de Lausanne

Mathias Dewatripont and Jean Tirole, The Prudential Regulation of Banks, 1994

David M. Newbery, Privatization, Restructuring, and Regulation of Network Utilities, 2000

Stephen L. Parente and Edward C. Prescott, Barriers to Riches, 2000

Joseph P. Newhouse, Pricing the Priceless: A Health Care Conundrum, 2002

Nolan McCarty, Keith T. Poole, and Howard Rosenthal, Polarized America: The Dance of Ideology and Unequal Riches, 2006

Andrei Shleifer, The Failure of Judges and the Rise of Regulators, 2012

Nolan McCarty, Keith T. Poole, and Howard Rosenthal, Polarized America: The Dance of Ideology and Unequal Riches, second edition, 2016

Polarized America

The Dance of Ideology and Unequal Riches

second edition

Nolan McCarty, Keith T. Poole, and Howard Rosenthal

The MIT Press

Cambridge, Massachusetts

London, England

© 2016 Massachusetts Institute of Technology

All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from the publisher.

This book was set in Palatino by Toppan Best-set Premedia Limited. Printed and bound in the United States of America.

Library of Congress Cataloging-in-Publication Data

Names: McCarty, Nolan M., author. | Poole, Keith T., author. | Rosenthal, Howard, 1939- author.

Title: Polarized America : the dance of ideology and unequal riches / Nolan McCarty, Keith T. Poole, and Howard Rosenthal.

Description: second edition. | Cambridge, MA : MIT Press, 2016. | Series: Walras-Pareto lectures | Includes bibliographical references and index.

Identifiers: LCCN 2015038321 | ISBN 9780262528627 (pbk. : alk. paper)

eISBN 9780262334068

Subjects: LCSH: Equality–United States. | Polarization (Social sciences)–United States. | Income distribution–United States. | United States–Politics and government–1989-

Classification: LCC HN90.S6 M37 2016 | DDC 305.50973–dc23 LC record available at http://lccn.loc.gov/2015038321

10 9 8 7 6 5 4 3 2 1

d_r0

From NMc to Janis, Lachlan, and Delaney

From KP to Janice

From HR to Illia, Manu, and Gil

Table of Contents

Series page

Title page

Copyright page

Dedication

Preface to the Second Edition

Acknowledgments

1 The Choreography of American Politics

2 Polarized Politicians

3 Income Polarization and the Electorate

4 Immigration, Income, and the Voters’ Incentive to Redistribute

5 Campaign Finance and Polarization

6 Polarization and Public Policy

7 Where Have You Gone, Mr. Sam?

References

Index

List of Tables

Table 2.1 Republican vote share in the poorest and richest congressional districts

Table 2.2 Partisan control of southern House seats by income

Table 2.3 Roll call weights

Table 2.4 Change in Senate polarization (standard errors in parentheses)

Table 2.A1 Determinants of NOMINATE scores, 113th House (2013–2014) (standard errors in parentheses)

Table 2.A2 Determinants of NOMINATE scores, 108th House (2003–2004) (standard errors in parentheses)

Table 2.A3 Determinants of NOMINATE scores, 103rd House (1993–1994) (standard errors in parentheses)

Table 2.A4 Determinants of NOMINATE scores, 98th House (1983–1984) (standard errors in parentheses)

Table 2.A5 Determinants of NOMINATE scores, 93rd House (1973–1974) (standard errors in parentheses)

Table 3.1 NES sample means of variables (presidential election years)

Table 3.2 Effects of relative income on Republican partisanship, ordered probit (standard errors in parentheses)

Table 3.3 Demographic and income shifts compared

Table 3.4 Characteristics of income quintiles, 1960, 2000, and 2008

Table 3.5 Determinants of party-income stratification

Table 3.6 Republican identification and the change in real income

Table 3.7 Effects of relative income on presidential vote choice, probit (standard errors in parentheses)

Table 3.8 Republican support by county, 1968–2012 (standard errors in parentheses)

Table 3 A.1. Estimates of the log-normal income distribution, by election year

Table 4.1 Median (50th) to 80th centile income comparisons (t-statistics in parentheses)

Table 4.2 Sample percentages by voting and citizenship

Table 4.3 Comparisons of median noncitizen and median nonvoter incomes to median voter income (t-statistics in parentheses)

Table 4.4 Comparisons of median voter income (50th centile) to 80th centile income of all families (t-statistics in parentheses)

Table 5.1 Summary PAC statistics, 2012 elections, House and Senate

Table 5.2 Contributions to independent expenditure groups, 2002–2012 (standard errors in parentheses)

Table 5.A1 Estimates of the Dependence of S on F for 1980−2012

Table 6.1 Polarization and the gridlock interval, 1946–2014 (standard errors in parentheses)

Table 6.2a State decisions on ACA exchanges

Table 6.2b State decisions on Medicaid expansion

List of Illustrations

Figure 1.1 Income inequality and House polarization, 1947–2012. Note: Polarization is measured as the difference between the Democratic and Republican Party mean NOMINATE scores in the U.S. House. The Gini index and polarization measures correspond to the first year of each biennial congressional term. Source: Gini index data from the U.S. Census Bureau (2014).

Figure 1.2 Top 1 percent income share and House polarization, 1913–2012. Source: Alvaredo et al. (2015).

Figure 1.3 Percent foreign-born and House polarization, 1880–2012. Note: Each observation of foreign-born population corresponds to a U.S. decennial census or an annual observation from the Census Bureau’s Current Population Survey.

Figure 1.4 Attention to social welfare in the Democratic Party platform as indicated by percentage of paragraphs in platform devoted to social welfare policies, 1908–1992. Source: Author calculations based on Gerring (1998), figure 14. Used by permission.

Figure 2.1 Classification of roll call votes, 1879–2014. Source: Computed from the two-dimensional NOMINATE model. Only roll calls with at least 2.5 percent in the minority are included.

Figure 2.2 Classification gain of second dimension, 1879–2014. Source: Computed from the two-dimensional NOMINATE model.

Figure 2.3 Party means in the U.S. House by region, 1879–2014. Note: The range of NOMINATE scores is approximately −1.0 to +1.0. The South is defined as the eleven states of the Confederacy plus Kentucky and Oklahoma. Source: Computed from NOMINATE scores, available at PooleandRosenthal.com.

Figure 2.4 Party means in the U.S. Senate by region, 1879–2014. Note: The range of NOMINATE scores is approximately −1.0 to +1.0. The South is defined as the eleven states of the Confederacy plus Kentucky and Oklahoma. Because of small numbers, the southern Republican mean is computed only after 1956. Source: Computed from NOMINATE scores, available at PooleandRosenthal.com.

Figure 2.5 Standard deviations of House NOMINATE positions, 1879–2014. Note: The range of NOMINATE scores is approximately −1.0 to +1.0. Source: Computed from NOMINATE scores, available at PooleandRosenthal.com.

Figure 2.6 Standard deviations of Senate NOMINATE positions, 1879–2014. Note: The range of NOMINATE scores is approximately −1.0 to +1.0. Source: Computed from NOMINATE scores, available at PooleandRosenthal.com.

Figure 2.7 Average two-dimensional distances in the House, 1879–2014.

Figure 2.8 Average two-dimensional distances in the Senate, 1879–2014.

Figure 2.9 Percent party overlap in Congress, 1879–2014.

Figure 2.10 Distribution of ideal points in the 113th House (2013–2014).

Figure 2.11 Explanatory power of the models.

Figure 2.12 Effect of district income on left-right position. Note: Estimated difference between richest and poorest district is provided for each model.

Figure 2.13 Partisan difference on left-right scale. Note: Estimates of party coefficient in model B are provided.

Figure 2.14 Difference in Republican share of two-party presidential vote between South and North, 1948–2012. Note: The South is defined as the eleven states of the Confederacy plus Kentucky and Oklahoma.

Figure 2.15 Republican percentage of seats in U.S. Senate, 1948–2014. Note: The South is defined as the eleven states of the Confederacy plus Kentucky and Oklahoma.

Figure 2.16 Republican percentage of seats in U.S. House, 1948–2014. Note: The South is defined as the eleven states of the Confederacy plus Kentucky and Oklahoma.

Figure 2.17 Republican percentage of seats in state legislatures, 1948–2014. Note: The South is defined as the eleven states of the Confederacy plus Kentucky and Oklahoma.

Figure 2.18 Southern effect on polarization in U.S. House, 1879–2014. Note: Measures of distance between two parties (differences in mean NOMINATE scores) with and without southern members.

Figure 2.19a Civil rights votes, 1937–2012. Note: Congresses with fewer than two votes were not included in the calculations.

Figure 2.19b Labor and minimum wage votes, 1937–2012. Note: Congresses with fewer than two votes were not included in the calculations.

Figure 2.19c Government management votes, 1937–2012.

Figure 2.20 Polarization when roll call margins are held constant.

Figure 2.21 Cut-point models.

Figure 2.22 Classification gain, one cutpoint versus two cutpoints. Note: The classification gains are for a one-dimensional voting model. All representatives were scaled together, and a separate cutpoint was then estimated for each party. The classification gains are similar to those scalings in which each party has an independent rank order of ideal points as well as a separate cutpoint.

Figure 2.23a Distribution of the 2012 Romney two-party vote by counties and congressional districts. Note: Counties are weighted by population. Both densities were estimated using bandwidth = 0.025.

Figure 2.23b Distribution of the 2000 Bush two-party vote by counties and congressional districts. Note: Counties are weighted by population. Both densities were estimated using bandwidth = 0.025.

Figure 2.23c Distribution of the 1992 Bush two-party vote by counties and congressional districts. Note: Counties are weighted by population. Both densities were estimated using bandwidth = 0.025.

Figure 2.23d Distribution of the 1980 Reagan two-party vote by counties and congressional districts. Note: Counties are weighted by population. Both densities were estimated using bandwidth = 0.025.

Figure 2.23e Distribution of the 1972 Nixon two-party vote by counties and congressional districts. Note: Counties are weighted by population. Both densities were estimated using bandwidth = 0.025.

Figure 3.1 Ideological self-placement. Note: Respondents to National Election Study. Respondents identifying as liberal or conservative (including strongly and slightly) are grouped into the ideological category.

Figure 3.2 Party stratification by income. Note: Stratification is computed as the Republican proportion of the top income quintile divided by the Republican proportion of the bottom quintile. The 2012 sample includes only face-to-face interviews and omits the oversample of blacks and Hispanics. Source: American National Election Study, Survey Research Center, University of Michigan, Ann Arbor. Various years.

Figure 3.3 Income effect on party identification: fourth-order polynomial estimates. Note: The dashed line is an estimate excluding responses to the 2012 NES. The dotted lines represent the confidence interval for the estimates using all years.

Figure 3.4 Income effect on party identification: fourth-order polynomial estimates for southern non-African Americans. Note: Dashed line is an estimate excluding responses to the 2012 NES. The dotted lines represent the confidence interval for the estimates using all years.

Figure 3.5 Income effect on party identification: fourth-order estimates for northern non-African Americans for the estimates using all years. Note: Dashed line is an estimate excluding responses to the 2012 NES. The dotted lines represent the confidence interval.

Figure 3.6 Annual income effect. Note: From PEW surveys conducted between 1997 and 2014 (N = 325,704). The dotted lines represent the confidence interval.

Figure 3.7 Republican identification, religion, and relative income. Note: Computed from Pew surveys conducted between 1997 and 2014. The figures are based solely on the responses of white, non-Hispanic respondents.

Figure 3.8 Party-income stratification by state. Note: Computed from Pew surveys conducted between 1997 and 2014. The figures are based on the responses of white respondents.

Figure 3.A1 Fit of log-normal income approximation. Note: If the fit were perfect, each plotted line would be horizontal at the corresponding percentile.

Figure 3.A2 Box plot of stratification measures in presidential election years. Note: The bottom of each box is the 75th percentile, the top is the 25th percentile, and the middle bar is the median.

Figure 4.1 Citizenship and voting in 2008 and 2010. Source: Figure redrawn from from Bonica et al. (2013) with permission.

Figure 4.2 Median/mean income ratios. Source: November Current Population Survey, various years.

Figure 4.3 Ratio of median income to 80th centile income, all families, various categories. Source: November Current Population Survey, various years. Years 1998–2002 not shown because data was top-coded.

Figure 4.4 Ratio of median income to 72nd centile income, all families, various categories. Source: November Current Population Survey, various years.

Figure 4.5 Income ratios, 1972–2012. Source: November Current Population Survey, various years.

Figure 4.6 Noncitizens by race and ethnicity. Source: November Current Population Survey, various years.

Figure 4.7 Ratio of median voter income to mean family income and mean citizen income. Source: November Current Population Survey, various years.

Figure 4.8 Ratio of median income to 80th centile income of citizens and families compared. Source: November Current Population Survey, various years. Years 1998–2002 not shown because data was top-coded.

Figure 4.9 Ratio of median income to 72nd centile income of citizens and families compared. Source: November Current Population Survey, various years.

Figure 5.1 Real spending on House and Senate campaigns, 1980–2012. Note: Adjusted for inflation, spending has nearly tripled in thirty years. The inflation adjustment is to 2012 using the CPI-U series. Source: Data from the Federal Election Commission (http://www.fec.gov).

Figure 5.2 Real soft-money receipts of national parties, by donor type, 1992–2012. Note: Spending deflated to year 2012 using the CPI-U series. Soft money represents contributions to party organizations. Source: Data from the Federal Election Commission (http://www.fec.gov).

Figure 5.3 Measures of inequality in individual campaign contributions, 1980–2012. Note: Measures based on the inequality of contributions sizes for all individual contributions over $200. Source: Database on Ideology, Money in Politics and Elections.

Figure 5.4 PAC contribution patterns. Note: Each token in the plot represents a PAC. The letter denotes the PAC type—corporate (C), labor (L), trade (T), nonconnected (N), unaffiliated (U), cooperative (V), and corporation without stock (W). Moderate PACs, with F near zero, spread their contributions more widely than do more extreme ones. Almost all PACs, however, disperse their contributions less than the random giving benchmark of 0.577, shown as a dark black line. The horizontal dashed lines indicate the mean S. The figure includes committees making contributions to thirty or more distinct candidates.

Figure 5.5 Estimated standard deviations (S) as a function of F, 1980–2012. Note: F and S are defined in the text.

Figure 5.6 Total hard-money contributions to incumbents by PACs in 1988, 1996, 2004, and 2012. Note: Each token in the plot represents a PAC. The letter denotes the PAC type—corporate (C), labor (L), trade (T), nonconnected (N), unaffiliated (U), cooperative (V), and corporation without stock (W). Groups contributing more than $2 million are excluded from the figure.

Figures 5.7a–c Candidate receipts by candidate ideological position in 1988, 1996, 2004, and 2012. Note: Figure 5.7a orders House members by their NOMINATE scores. Figure 5.7b uses F scores for House candidates and figure 5.7c for Senate candidates. There is little correlation between ideological position and fund raising. The NOMINATE scores are from DW-NOMINATE scores, found at voteview.com.

Figure 5.8 Individual contributions by F and S in 1988, 1996, 2004, and 2012. Note: Each point represents an individual donor who made more than eight contributions in four election cycles. The solid horizontal lines represent the random contribution benchmark while the horizontal dashed lines indicate the mean S.

Figure 5.9 Individual contributions to House members in 1988, 1996, 2004, and 2012. Note: The figure plots individuals making more than eight contributions.

Figure 5.10 Large individual contributions by ideology (F) of contributor. Note: The figure plots individuals contributing more than $250,000 in the 2012 election cycle.

Figure 6.1 Real minimum wages, 1938–2015. Source: U.S. Department of Labor. Chart available at http://www.dol.gov/whd/minwage/chart.htm. Deflator: CPI-U.

Figure 6.2 Estate tax, 1916–2014. Note: Those dying in 2010 were given the option (1) pay no estate tax but receive a smaller step up in the capital gains basis or (2) pay 35% on the value of the estate over $5m. The figure assumes option 2. Source: For 1916–2010, Darien B. Jacobson, Brian G. Raub, and Barry W. Johnson, The Estate Tax: Ninety Years and Counting, http://www.irs.gov/pub/irs-soi/ninetyestate.pdf. For 2011–2015, http://www.bankrate.com/finance/taxes/estate-tax-and-gift-tax-amounts.aspx, http://www.irs.gov/Businesses/Small-Businesses-&-Self-Employed/Estate-Tax, http://en.wikipedia.org/wiki/Estate_tax_in_the_United_States, http://wills.about.com/od/understandingestatetaxes/a/estatetaxchart.htm. Deflator: CPI-U

Figure 6.3 Top federal income tax rate, 1913–2014. Note: This table contains a number of simplifications and ignores a number of factors, such as a maximum tax on earned income of 50 percent when the top rate was 70 percent and the current increase in rates due to income-related reductions in the value of itemized deductions. Perhaps most important, it ignores the large increase in the percentage of returns that were subject to this top rate. Source: Historical Parameters Tax Policy Center pdf, downloaded from http://www.taxpolicycenter.org/taxfacts/displayafact.cfm?DocID=543&Topic2id=30&Topic3id=39.

Figure 6.4 Social Security benefits and taxes, 1938–2015. Note: In thousands of 2014 dollars. Beginning in 2003, the full retirement age was gradually increased from sixty-five years to seventy years. But benefits are still calculated for someone who retires exactly at sixty-five years. Source (except CPI-U): Social Security Administration, Annual Statistical Supplement 2014. Available at http://www.ssa.gov/policy/docs/statcomps/supplement/2014/index.html.

Figure 6.5 Polarization and legislative production, 1947–2012. Note: Data on significant legislation are from Mayhew (1991). The figure combines the Sweep I and Sweep II laws from his original analysis, supplemented by his subsequent list of significant laws from 1990 to 2012. Source: Available at http://campuspress.yale.edu/davidmayhew/datasets-divided-we-govern.

Figure 6.6 What if there were no polarization? The effect of polarization on producing major legislation. Note: Based on the estimates of model described in text.

Preface to the Second Edition

Both political polarization and economic inequality have accelerated into record territory since the publication of Polarized America in 2006. Money has poured into campaign finance at unprecedented levels. Immigration has turned into a hot button issue; noncitizens continue to represent a substantial fraction of the adult population. Government policy, impeded by polarization and gridlock, has left income and estate taxes at relatively low levels. The intent of the Affordable Care Act to reduce health care inequality for the poor has been blunted by the refusal of most red states to expand their Medicaid programs.

In the first edition of this book, our data series stopped in 2002 or 2004. Here we document and interpret the trends of the past ten years using data that run through 2012 or 2014. All the figures, tables, and statistical analyses have been updated. (The sole exception is the technical appendix to chapter 3.) The text has been modified or added to reflect the latest evidence.

The most important changes are in chapter 5, which deals with campaign finance. In the first edition, we estimated the ideological position of a campaign contributor by simply taking the money-weighted average of the ideological positions of incumbents who received money from the contributor. Our colleague Adam Bonica has more recently elaborated and extended that method to jointly estimate the positions of contributors and candidates, including the losers of primary or general elections. He has joined us, as a full partner, in writing the current version of chapter 5.

In large part, America today is what it was a decade ago, only worse. The relationship of voting to county median income is stronger than ever. The economic position of the median voter, for years maintained by the arrival of noncitizens at the bottom of the income distribution and by the low participation of poor citizens, has eroded. On the other hand, by comparison to first edition results, the last two elections, those of 2012 and 2014, show a weakened connection of income to voter behavior and to congressional representation. These changes could conceivably be short term, with low-income Republican voting by poor whites and higher participation by nonwhites both being a response to an African American president. Or they could be long term, with politics being permanently distorted by the campaign contributions of the plutocracy.

Acknowledgments

Work on this book officially started almost a decade ago when Howard Rosenthal gave the Walras-Pareto Lectures, based in part on our joint work, at the University of Lausanne. He is very grateful for the warm reception from his hosts, particularly Alberto Holly, Damien Neven, and Elu von Thadden, and for Alberto’s persistence in insisting that the book be finished.

The intellectual origins of the book are even older. Our central obligation is to the analytical approach to politics that developed in the 1960s and 1970s at four universities, Carnegie Mellon, the University of Rochester, Washington University in St. Louis, and the California Institute of Technology. Rosenthal owes much to the stimulus of his early colleagues at Carnegie Tech—Otto Davis, Melvin Hinich, James Laing, and Peter Ordeshook—and to his later collaborators, Thomas Romer, Thomas Palfrey, and Alberto Alesina.

Keith Poole was the most influenced by the Rochester school, as he was trained by the two central figures in the development of the analytical approach, William Riker and Richard McKelvey. McKelvey stopped at Carnegie Mellon on his way to Caltech. Most unfortunately, we have lost both Bill and Dick. We hope they are looking and reading from above.

Nolan McCarty began his research career as a graduate student in the Graduate School of Industrial Administration (GSIA) at Carnegie, his dissertation being supervised by Poole and Rosenthal. His interest in American politics was in no small part kindled by Larry Rothenberg, a visitor from Rochester. McCarty’s early papers with Poole and with Rothenberg are the origins of the campaign contributions analysis in chapter 5, which also continues work that Poole and Rosenthal conducted with Romer.

Shortly after Poole and Rosenthal began their collaboration, Rosenthal spent a year as a Fairchild Scholar at Caltech. During this stay Poole and Rosenthal wrote their 1984 Journal of Politics article The Polarization of American Politics, the protoplasm of this book.

Poole’s first gig at Carnegie was as one of the two inaugural fellows of the postdoctoral program in political economy at GSIA. This program and some of the research reflected in this book found financial support thanks to the efforts of Allan Meltzer. Many of the other postdoctoral fellows, especially Zeev Maoz, David Austen-Smith, Randy Calvert, Fritz Schneider, Guido Tabellini, Peter Van Doren, Alesina, Jim Snyder, and Rothenberg, were an integral part of our research environment.

As our joint work progressed, we benefited from a lot of release time and research assistance provided by think tanks. McCarty and Rosenthal were both National Fellows at the Hoover Institution. All three of us have been fellows at Stanford University’s Center for Advanced Study in the Behavioral Sciences (CASBS). If this book conveys a message successfully, the kudos go to Kathleen Much, the center’s editor par excellence (a.k.a. the book doctor). Rosenthal was also a visiting scholar at the Russell Sage Foundation, where sitting next to Frank Bean would lead to valuable contributions to chapter 4. He and McCarty also participated in the RSF-sponsored Princeton working group on inequality. The other members of that working group—Larry Bartels, Paul Dimaggio, Leslie McCall, and Bruce Western—have provided lots of feedback on much of this manuscript. The RSF support is reflected in the work reported in chapter 6. The broader inequality project of the RSF also led to Rosenthal’s collaboration with Christine Eibner, which is reflected in chapter 4.

Poole and Rosenthal have also benefited from multiple NSF grants from the Political Science program and the supercomputer program. Indeed, this project is very much technology-driven. The NSF supercomputer project gave us the capacity—today available in high-end PCs—for a dynamic analysis of the 11 million individual roll call votes that took place between 1789 and 1986. Without this support, chapter 2 of this book would not have been imaginable, and the rest of the book would not have followed.

Special thanks are owed to Robert Erikson and David Rhode, two superb editors of the American Journal of Political Science who encouraged this research at a time when the APSR had its head and somewhat more of its anatomy buried in the tar pits of traditional congressional scholarship.

Research on this project was both facilitated and hindered by the fact that the three of us accumulated eleven institutional affiliations, as visitors and regular faculty, during the gestation of this project: Brown University, CASBS, Carnegie Mellon, Columbia University, the Free University of Brussels, the University of Houston, the Hoover Institution, New York University, Princeton University, the University of Southern California, and the University of California–San Diego. Our colleagues at these institutions provided insight and advice on this project that more than compensated for the productivity lost in moving around so much.

Rosenthal would especially like to thank his colleagues from ECARES at the Free University of Brussels—Erik Berglof, Patrick Bolton, Mathias Dewatripont, and Gerard Roland—for the insights that came from eating lots of baguette sandwiches in the Canadian ambassador’s residence. He also thanks his fellow MIT alums Sam Popkin and Susan Shirk. McCarty extends a special thanks to Doug Arnold, Larry Bartels, Chuck Cameron, Tom Gilligan, John Huber, Ira Katznelson, Keith Krehbiel, John Matsusaka, Bob Shapiro, and Greg Wawro, both for advice on this project and for support early in his career. He’d also like to thank his junior colleagues Josh Clinton, David Lewis, Adam Meirowitz, and Markus Prior for helping to keep his mental faculties perking during our daily Starbucks run. Poole would like to thank his former colleagues at the University of Houston—Ray Duch, Harrell Rogers, Bob Lineberry, Kent Tedin, Ernesto Calvo, Noah Kaplan, and Tim Nokken—who gave valuable feedback on portions of this book while he was the Kenneth L. Lay Professor of Political Science. Poole would also like to thank