2012 Graduate Student Research Poster Session

Graduate Student Research Poster Session
Princeton University
300 Wallace Hall
12:00 pm - 2:00 pm
January 12, 2012

The Partisan Dynamics of Medicaid Expenditure

Carolyn Abott & Mateusz Tomkowiak

Optional Medicaid expenditures in generous New York are six times as high as frugal Colorado. What explains such massive state variation in the nation's largest welfare policy? Kousser (2002) showed that partisan control of the state legislature is a meaningful predictor of the level of state Medicaid expenditures. We investigate the claim by expanding his dataset and offering several robustness checks. First, we extend Kousser's panel OLS regression by introducing fixed effects and by placing the data into a non-nested hierarchical model. Second, through a series of cross-sectional regressions we also test the theoretical claim made by Beamer (1999) that the political dynamics of Medicaid spending underwent a fundamental shift in the wake of the managed care revolution of the 1990s.

Swords or Ploughshares: Production, Predation & the Origins of the Territorial State

Scott F Abramson

To explain both cross sectional and temporal variation in the size of states this paper develops a model of state formation characterized by wealth-maximizing agents who devote some fraction of resources to the production of economic output and the remaining fraction to the conquest of territory from which resources necessary for economic production are drawn. If all territory produces similar output or if returns to territorially fixed inputs are low, a world of equally sized states is predicted. However, if there is inequality in the distribution of the geographically fixed inputs then a world of unequal state sizes is predicted. Moreover, it is predicted that states initially located in resource poor regions will devote more eort to territorial expansion and consequently be of larger geographic scale. These empirical predictions are then tested using a unique data set that combines observations of European boundaries measured every five years between 1100 and 1800 AD and pale-climatic data that captures the capacity of land to produce agricultural output.

A Model of Learning on The Supreme Court
Deborah Beim

Studies of the judicial hierarchy have assumed the Supreme Court to be a monitor policing the lower courts for non-compliance. In those models, justices know their ideal doctrines and struggle to see them implemented. This paper shows how justices learn their ideal doctrine from their subordinate courts. I propose a formal model that explicates which cases the Supreme Court would take if its goal were learning, not monitoring. The Supreme Courts derived optimal strategy noticeably changes if one assumes its goal is not error-correction. Even with ideologically identical subordinates, the Supreme Courts position as doctrinal leader will lead it to take and overturn cases. In fact, in a monitoring framework ideological homogeneity is optimal but in a learning model it may not be. Furthermore, the Supreme Courts cert strategy is dependent on whether it can overturn only the case it takes or whether it can set a universally applicable doctrine.

An Experimental Approach to the Study of Judicial Elections
Alex Bolton & P.J. Gardner

Critiques of judicial elections by high-prole public gures has increased attention to selection of judges in state courts. Recent research focuses on new-style judicial campaigns, particularly the impact of campaign contributions and attack ads on judicial legitimacy. Using an original survey experiment, we broaden the scope of this research in two ways. First, we consider the effect of selection mechanisms more generally for judicial legitimacy, finding that non-partisan elections are associated with the highest levels of legitimacy, followed by merit-selection plans and partisan elections. Second, we find that these effects are conditional on the quality of judges. When respondents were presented with information suggesting that judges were of some quality, the effects of selection mechanisms on legitimacy are negligible. Additionally, we use evidence from our own survey and a survey of Kansas residents to identify potential individual level factors that influence support for judicial elections.

Parliamentary or Presidential Government
Peter Buisseret

I compare parliamentary and presidential government in a model with an executive and a decisive parliamentarian, whose ability to submit bills of high quality, or learn their quality, respectively, is private information. When the parliamentarian passes a bill submitted to it by the executive, the voter learns its quality; when he vetoes it, the observation of its quality is censored. After observing the actions of each politician, the voter chooses whether to retain each politician or replace them with a randomly drawn alternative. Under presidential government, the voter is allowed to choose distinct replacement strategies for each politician, whereas under parliamentary government, the voter is committed to using the same replacement strategy for both politicians. Whilst the richer set of strategies available to the voter under presidential government makes it a superior choice for the voter in an environment of complete information, in the case of private information and re-election motives the parliamentarian in a presidential system establishes a reputation for competence at the expense of the executive, by vetoing executive bills.

Learning and Collective Signaling Before Agenda Setting

Jidong Chen

This paper develops a two-period Romer-Rosenthal monopoly agenda setting model with strategic voting. Finite committee members (i.e. voters) have private information about their preferences over policy (e.g. public expenditure). An exogenous proposal is voted against the status quo in the first period. An agenda setter makes a new proposal when the initial one fails. The revised proposal will replace the reversion only when it is accepted by the committee. Existence of separating equilibrium (i.e. interior cut-point equilibrium) depends crucially on voters' "demand elasticity" (the percentage change in ideal policy demanded in response to a one percent change in the marginal cost of policy). (1) When the "demand elasticity" is sufficiently low, voters' incentive in signaling effect is severely distorted so that the unique equilibrium is pooling (i.e. each voter casts a no vote in the first period no matter what his preference is), in which case no information is disclosed before the second period and the bargaining is delayed with probability one; (2) When the "demand elasticity" is sufficiently high, voters' incentive in signaling eect is not distorted and satises "single-crossing condition" so that there is a separating equilibrium, in which the revised proposal is decreasing in the total negative votes of the initial election; (3) When the "demand elasticity" lies in certain intermediate range (including quadratic utilities), the single crossing condition is endogenously unsatised in voter's signaling effect (namely the strategic utility dierence between voting for and against the first proposal is not a monotonic function in each voter's type), in which case sufficient conditions are made for the existence a separating equilibrium. In addition, we characterize other properties of equilibrium and make comparisons with classical agenda setting model. Empirical and policy implications are also derived.

The Questionable Validity of Income Measures in the WorldValues Survey

Michael Donnelly and Grigore Pop-Eleches

TheWorld Values Survey is a widely used data sets in comparative politics. By providing comparable surveys in many countries over a long time period, it has encouraged investigation into such important topics as the relationship between development and democracy, the changing role of ideology, and the causes of happiness. In this paper, we examine the WVS measure of income, arguing that a variety of problems arise when drawing inferences - descriptive or causal, individual or aggregate - using the standard, 10-category measure. To do
this, we first examine and describe the data, the codebooks, and where available, the original questionnaires. Together, these call into question the concept validity of this measure. We then propose and implement a number of possible corrections to the potential biases we describe. Finally, we examine the tradeoffs between these corrections, urging researchers to select the most theoretically appropriate solution for their theoretical context.

Examining the Effect of the Feeney Amendment

Cody Grey & Edward Engelhardt

In 2003, Congress passed the Feeney Amendment to the PROTECT Act, establishing mandatory guidelines for criminal sentencing in the federal courts. In 2005, the Supreme Court effectively rendered those guidelines advisory with its decision in U.S. v. Booker. Previous studies have examined the system-wide effects of Feeney and Booker on sentencing outcomes, but these studies fail to consider two important determinants of such outcomes. First, the Supreme Court struck down state-level mandatory guidelines in Blakely v. Washington in mid-2004, causing some court circuits and districts to change or abandon their application of the federal guidelines. Second, implementation rates for all three sentencing interventions likely varied among circuits and districts. We use a multilevel model to analyze differences in federal sentencing outcomes across the jurisprudential regimes created by Feeney, Blakely, and Booker, as implemented across the various circuits and districts of the judicial hierarchy.

A Hierarchical Approach to Ethnic Power
Michael Homan, John Chin, & Matthew Tokeshi

We replicate and extend Cederman et al. (2010), which introduces the Ethnic Power Relations (EPR) Dataset. We take advantage of the hierarchical and temporal structure of the data, which we believe is not fully considered in Cederman et al. The original article uses traditional econometric methods for panel data which do not properly account for the multi-level nature of the data. The key addition of the EPR Dataset is its inclusion of sub-national (ethnic group-level) data in addition to standard country-level variables, and our paper will incorporate this structure in order to obtain more precise estimates. We estimate a multi-level model that includes the national, group-level, and temporaldimensions of the data. This model allows ethnic group-specific parameters to borrow strength from the corresponding parameters from other ethnic groups.

Participation During Hard Times: Estimating the Effect of Unemployment on Voter Turnout

Matthew Incantalupo

How does economic hardship aect political participation? In this paper, I examine the effect of unemployment on voter turnout in recent national elections. To overcome problems of selection bias in who experiences unemployment, I leverage the panel structure of the Current Population Study Voting and Registration Supplement and limit my sample to individuals who experience job loss in the weeks before and after the arbitrary date of Election Day. This produces a sample with improved covariate balance on key demographic variables known to influence political participation. I estimate the causal eect of unemployment on voter turnout using this pruned sample. My results indicate that unemployment has a positive effect on turnout during periods of high unemployment, and a negative eect on turnout during periods of low unemployment. Further analysis indicates that this pattern is consistent with a mobilization-based theory of unemployment, and not other factors such as leisure time.

Revaluating the Relationship Between Land Inequality and Vote Fraud in Imperial Germany
Rohan Mukherjee, Kevin Mazur, & Alienor Van den Bosch

Daniel Ziblatts 2009 APSR (103:1) paper, Shaping Democratic Practice and the Causes of Electoral Fraud shows that the functioning of democratic institutions in a newly democratizing country is shaped by the underlying distribution of pre-existing social power. Looking at 13 consecutive elections following the unification and democratization of Germany in 1871, he uses a time-series cross sectional analysis and separate models for each election period to capture the effect of pre-existing social power (measured as rural land inequality) on the incidence of electoral fraud. We make three extensions to his work: we use a multi-level model to investigate social power effects not captured by land inequality; we test Ziblatts causal mechanism by looking at the content of election petitions in two elections for which there are data; and we investigate the impact on his analysis of a critical juncture in Germany history: Bismarcks exit from German politics in 1890.

Press Coverage and Campaign Finance: How Media Markets Affect Political Donations
Bryn Rosenthal, Baxter Oliphant, & Lizette Taguchi

We build and empirically test a formal model in which interest group contributions may signal a politician's willingness to pursue policies that diverge from the preferences of the median voter, leading to fewer donations in high-information districts. Our theory builds on a standard result in formal models of electoral competition that campaign contributions are increasing in the share of uninformed voters, since uninformed voters are more likely to be swayed by political advertising. Using Center for Responsive Politics data on campaign contributions over the period 1990-2006, we nd evidence not only that interest groups are more likely to donate to candidates who represent low-information districts, but also that the effect of news coverage depends on the type of interest group: Whereas corporate PACs give substantially less in high- information districts, this relationship does not hold for groups representing more encompassing interests, such as membership and labor PACs.

Responsiveness, Accountability, and the Regulatory State

Alex Ruder

Should presidents have more control over the bureaucracy? Does more presidential control increase the responsiveness and accountability of the regulatory state? We investigate the thesis of presidential administration by analyzing two U.S. agencies: the cabinet-level Department of Justices Antitrust Division and the independent Federal Trade Commissions Bureau of Competition. Using a variety of policy outputs, we find no difference between the two agencies responsiveness to external demands for antitrust enforcement. We also show that the president maintains control over independent regulatory commissions primarily through the appointment power. Finally, to test the accountability thesis, we use content analysis on a database of media articles covering antitrust enforcement, determining whether the actions of a cabinet-level agency are more closely associated with the electorally-accountable chief executive. The results clarify some common assumptions about agency institutional design and provide support for reputation-based theories of bureaucratic behavior.

Reappraising a Reappraisal of the Resource Curse
Yuki Shiraito, Jennifer Dennard, Gabriel Lopez-Moctezuma, & Devin Incerti

We reexamine two ndings from Haber and Menaldo (2011): first, that there is little evidence of a negative relationship between resource wealth and democracy, and second, that resource wealth may even increase the likelihood of democratization. Although they create a new valuable data set about resource wealth that goes back to 1800, there are serious mistakes in their analysis. We correct their methodological flaws and find that their results suggesting a resource blessing do not hold. We further explore the effect of resource wealth on regime type by analyzing their data with more appropriate methods. In particular, we propose (1) employing a correctly specied dynamic panel data model that explicitly considers structural breaks in
the data, (2) accounting for endogeneity by using an instrumental variables approach, (3) addressing the multilevel structure of the data with a hierarchical model and (4) measuring regime type as both a continuous and discrete variables.

A Theory of Presidential Signing Statements with a Test

Sharece Thrower

George W. Bushs controversial use of signing statements has given this previously unknown power recent attention. However, this practice, allowing the president to essentially nullify sections of legislation he deems unconstitutional, has been used strategically by presidents well before Bush. Existing political science research has primarily focused on how the presidents relationship to Congress influences his use of signing statements. However, these studies have neglected to address how the presidents relationship to the Court could constrain his use of these statements, a potentially important relationship given his often dubious use of constitutional objections. In this paper, I examine how the presidents distance to the Court influences his use of signing statements. Utilizing correlated random effects models to account for variation across presidencies, I find that the further the president is ideologically located from Court, the less likely he is to issue a signing statement raising constitutional objections.

The Onset of Ethnic Secession: Consolidation and Contestation
Jaquilyn Waddell Boie

Since the 1980s, ethnic self-determination movements have accounted for an increasing proportion of armed conflicts. In many states, numerous ethnic groups have expressed a desire for autonomy or have sought autonomy through violent means. While studies on self-determination movements have increased over recent years, there remains much debate regarding the conditions associated with ethnic secession. The question remains: why do some ethnic groups choose to engage in ethnic secession while others do not? This project seeks to contribute to the study of self-determination movements by analyzing the conditions associated with the onset of ethnic secession. Initial results from the formal and quantitative models suggest that previous explanations may not fully account for the empirical patterns of ethnic secession onset. An alternative explanation, namely that windows of opportunity created by space and leverage for political contestation may facilitate the onset of ethnic secession, provides a promising avenue for ongoing research.

Bringing the Firm into Political Analysis: The Distributional Implications of Basel III Regulatory Announcements upon Financial Services Firms
Meredith Wilf

How can international relations scholars move from country-level to firm-level political inquiry? To test hypotheses about domestic cleavages in a given issue-area, instead of analyzing aggregate country or industry data, I apply basic financial models to political science to estimate the effects of new policy information upon firm-level outcomes. Specically, I estimate firm-level abnormal stock returns in response to new policy information using the capital asset pricing model (CAPM) at the event-level, seemingly unrelated regression (SUR) to estimate an ongoing time series across events, and rolling regression to estimate the volatility of moving-window estimations over the time series. Substantively I apply this design to test the distributional implications of Basel III regulatory announcements, multilateral rules for international banks, in 2009 and 2010 upon a sample of 282 publicly-traded banks, securities, and insurance firms. The findings imply the within-industry cleavages that may be expected to arise in continued negotiations over rules-based international governance in financial services.

The Benefits of Monetary Policy-Making by Committee: The Case of England and Sweden 1999-2008
Carlos Velasco Rivera

Two important benets of monetary policymaking by committee emphasized in the literature are: (a) protection against extreme views and (b) pooling different processing methods available to individuals. But do extreme views exist in monetary policy-making? And if so, which type of committee member holds such views? Further, which type of committee member process information in the most efficient way? I present evidence tackling these questions based on voting data available for the Bank of England (BoE) and Sweden's Riksbank for the period 1999-2008. For England, the results show that only external members have better precision of information. In addition, the analysis shows that external members, along with members holding a PhD, are more dovish relative to their peers. In the case of Sweden, no member type has better quality of information, and only those with government experience are more dovish relative to their counterparts at the Riksbank.