2011 Graduate Student Research Poster Session
300 Wallace Hall
11:00 am - 1:30 pm
January 7, 2011
Feudal Institutions and Enclosure in England
Legal acts of enclosure have been thought to indicate a fundamental change in the English aristocracy's attitude towards the treatment of land, a change demonstrative of an abandonment of medieval economic values and the adoption of a system of beliefs concordant with commercial capitalism (Moore, 1966; Polanyi 1944). I show that the probability enclosure between 1500 and 1900 in two southern English counties was a function of the medieval institutions in existence half a millennium prior. Where serfdom prevailed landlords were less restricted in their later ability to enclose on common land. Where peasants maintained political rights, acts of enclosure were less likely to occur. I propose an instrument to show that this relationship is causal.
Do Congressional Elites Respond to Voter Signals? A Regression Discontinuity Approach
Michael Barber and Aram Hur
Most studies in American political behavior have focused on how the public responds to elite signals, but not the other way around. Yet elite response to the public is the key to a healthy democracy. The challenge of studying elite response is that usually, voter signals are endogenous to prior elite influence. Using a regression discontinuity design, we are able to move beyond this problem. We take advantage of precipitation on Election Day and its documented boost for Republican presidential candidates. By estimating the effect of precipitation on GOP vote share, we are able to identify states that, because to precipitation, were carried by a Republican presidential candidate. In other words, we can identify states where the Electoral College outcome is essentially a randomized, and therefore exogenous, voter preference signal. Using regression discontinuity, we then analyze change in DW-NOMINATE scores of congressmen in these states to see if and how much they respond to voter signals.
Accounting for Accounting: What Governments Do with Unanticipated Revenue
Omar Bashir and Darren Lim
How do governments make use of unanticipated revenue inflows? Building on the work of Kuziemko and Werker (2006) and Bueno de Mesquita and Smith (2010), we employ election to the United Nations Security Council (UNSC) as an exogenous source of variation in US foreign aid. Given that temporary UNSC membership triggers a short-term spike in US aid flows, we seek to investigate how this unanticipated revenue alters the spending patterns of governments. In contrast to previous studies which analyze macro-level variables such as GDP growth and democracy score, we focus on how money is spent, asking whether additional funds are diverted toward social spending, capital works, the military, bureaucracy or simply towards patronage. The goal of this project is to understand what, if anything, explains variation in the spending arising from unanticipated revenue inflows, and ultimately to discern which patterns are healthiest for economic and political development.
What is the causal eect of single-party majority government on fiscal policy? Evidence from a regression-discontinuity analysis
A prominent view in comparative politics and political economy is that single-party majority governments produce lower budget decits than coalition governments. Because the type of government is an endogenous feature of representative democracies, it is difficult to estimate the causal effect of the type of government on social policy from observational data. This paper uses a natural experiment to estimate the effect of single-party government on budget deficits in a large panel of parliamentary democracies. In particular, it exploits the discontinuity provided by majority rule in the legislature and compares social policy outcomes in cases where the largest party in the lower house just wins enough seats to form a single-party majority government with cases where the largest party narrowly fails to obtain a majority by a few seats, and a coalition or minority government emerges. The (fuzzy) regression discontinuity analysis shows that single-party government lowers the decit by approximately 2 percentage points, plus or minus 1 percentage point.
Bargaining Over Opinion Content on the Supreme Court
Judicial politics has focused extensively on who controls the content of opinions on the Supreme Court. Most studies have focused on observed behavior, usually votes; more recently, others have studied attributes of the opinions themselves, like their linguistic content or citations within them. I link these two strands of research by using data from the production of opinions, including length of time in days to produce the majority opinion, number of drafts produced, and number of memos concerning the opinion circulated. I ask whether the median justice receives fewer memos from their colleagues, write fewer drafts of the majority opinion, or write the majority opinion more quickly. Next, I use the Clark and Lauderdale scores, which put justices and opinions on the same ideological scale, and ask whether an opinion is closer to a justice's ideal point when it is written in fewer drafts or with fewer received memos.
Methods for Estimating the Causal Effect of the Discovery of Natural Resources on Violence
The workhorse theory relating resource revenues and civil conflict suggest that weakness in the state caused by the reliance on natural resource revenues instead of taxes leads to challenges both at the center and for control of remote areas. Yet a competing theory proposes that the higher returns to capturing the resource region increases the probability of challenges to the state's control of that region. Current methods do not allow us to adjudicate between these theories, because they fail to account for the location of resources and conflict. I present results using a simple matching estimator with data from 1km grids across the world from 1946 to 2005, with the causal quantity being the discovery in a grid square-year of crude oil, and civil conflict onset in a grid square-year as the outcome. Contrary to expectations, the presence of natural resources does not appear to be related to civil conflict.
Protection for Workers or Aggressive Unilateralism?: the Generalized System of Preferences and Human Rights Conditions
Jaquilyn Waddell Boie
In 1971, the EU became the first nation to offer a Generalized System of Preferences (GSP) to promote preferential trade policy with developing nations. In 1976, the United States became the seventh nation to implement the GSP. Under the 1984 GSP Renewal Act, the US became the first nation to formally condition GSP benefits on countries' protection of fundamental workers' rights. The EU followed in the 1990s, first conditioning the receipt and continuation of GSP benefits on the protection of workers from forced labor and later providing incentives to ratify conventions of the International Labour Organization. Though hailed as a means of promoting workers' rights, critics have argued that GSP sanctions are ineffective and merely a guise for protectionism. We seek to address this debate by exploring the economic, political, and human rights conditions associated with US GSP country beneficiary status as well as the application of workers' rights sanctions.
Modes of Hierarchical Communication in Political Parties
This paper considers the positive and normative basis for communication and deliberation in political parties. Activists possess private information about an unknown state and have an opportunity to communicate their information to any subset of activists before a second stage in which a pre-specied set of party members take actions on behalf of the party. Each activist cares about the proximity of the action profile to the true state plus an agent-specic bias. The leadership cares about the congruence of actions with the unknown state, but also wishes to minimize the heterogeneity of the action profile. I study the incentives for agents to communicate with one another under both private and public deliberation, as well as how these incentives change when the leadership delegates policy authority to agents with more proximate bias to those holding relevant information. The model relates the degree of oligarchy in parties to the `voice' of its members, and provides conditions under which centralized parties may provide for more effective information aggregation than those which are decentralized.
Identifying Causal Mediators for the Effect of Multi-member Districts on the Election of Female Representatives
The goal of the project is to assess what impact, if any, the use of multi-member districts (MMDs) has on the behavior of voters, relative to the more common single-member districts (SMDs). Observational evidence suggests that MMDs result in the election of more women, but the reason for this increase in unknown. While there may be an eect on party leaders and candidates, this research focuses on the behavior of voters. The underlying assumption of this research is that some voters will desire gender balance of their elected representatives and in MMD elections promote the female candidate into the top two, when they otherwise would not have done so under SMD election scenarios.
Pooling Ordered Data with Dierent Scales
A common problem in political science research is that different polling organizations or different polls ask the same question, but code responses using a dierent scale, requiring the researcher who wants to pool the data for the purpose of increasing statistical power or analytical leverage to recode the data. In this paper, I argue that the traditional solutions that applied researchers use to solve this problem can be improved upon using Bayesian imputation, which takes advantage of the latent variable representation of ordered data. First, I formalize and categorize the some traditional approaches. I then compare the results of two new techniques to the outcome of traditional approaches on Monte Carlo simulated data. Results suggest that these are promising approaches to a problem that is common both in American Politics and Comparative Politics.
Open Trade for Sale
In Song Kim
In addition to the big states' beggar-thy-neighbor economic incentives, political motivation to protect import competing industries exist even within small countries who cannot manipulate their terms of trade. In this regard, "Protection for Sale" well summarizes the political dynamics against trade liberalization. This paper examines theoretical conditions under which counteracting political forces may also exist. Rather than introducing distortions into economic policy, political pressures from organized lobby groups can induce governments to reduce distortions. Motivated by Melitz (2003), this paper presents a formal model of an industry with differentiated goods under monopolistic competition. First, consumer's love of variety decreases political incentives to protect domestic industries since doing so may increase economic deadweight loss. Second, only small numbers of highly productive firms can export, thereby owners of specific factors for such firms credibly solve their collective action problem in organizing lobbying groups. In this respect, this paper identies an important political channel through which "open trade for sale" occurs within industries of differentiated goods due to diverse consumer tastes and firm level differences.
Who Supports FDI? A Multilevel Approach to Individual Preference Formation
Jason McMann and Justin Simeone
The existing literature on foreign direct investment (FDI) focuses on the economic and political consequences of investment inflows, yet largely fails to examine the variables that shape individual preferences for FDI. Using three discrete years of Latinobarometer survey data across 7 to 18 countries, we advance and test a theory to account for variation in individual preferences for FDI. In so doing, we make two key theoretical contributions to existing work. First, whereas prior research emphasizes individuals' skill level, we argue that preferences also vary by country-level characteristics. Second, given that individuals likely possess limited information about the effects of FDI, we advance a sociotropic mechanism based on individuals' perceptions of national economic performance. We test these hypotheses using a multi-level probit model. While skill level remains signicant, individuals' perceptions of national economic performance, as well as country-level economic characteristics, are also important predictors of individual preferences for FDI.
Press Coverage and Campaign Finance: How Media Markets Affect Political Donations
Baxter Oliphant, Bryn Rosenfeld, and Lizette Taguchi
In a recently published study of media coverage and political accountability, Snyder and Stromberg (2010) find that the structure of U.S. House districts' media markets affects voter participation and the behavior of congressional representatives. Where newspapers cover multiple districts (low congruence), voters are less well-informed about their representative and representatives do less for their districts. Our paper extends these findings, using Snyder and Stromberg's measure of exogenous variation in press coverage to investigate how media market structure influences campaign finance. Employing a dataset from the Federal Election Commission (FEC), we examine two possibilities: First, special interests are more likely to donate to candidates who represent low congruence districts where representatives may have more leeway to pursue special interest agendas. Second, members in low congruence districts are more likely to rely on larger individual donations, since the incentives for a representative to cultivate a signicant network of small donors are weaker in such an information environment.
Finding Optimal Balance: Support Vector Machines as a Matching Method
Contemporary matching algorithms leave much to the applied researcher's discretion. These arbitrary decisions relating to functional form, caliper size, or bin size, can lead to different matched subsets, with no sense as to which is optimal. In this paper, I use support vector machines to characterize and identify a subset of the data that is optimally matched, in the sense of minimizing a stated objective function. The method is implemented both parametrically and nonparametrically. I prove that minimizing this objective function produces uncorrelatedness between assignment to treatment and covariates. Illustration on the Lalonde data illustrates dramatic improvement over existing methods.
Were there returns to political office in Postwar Britain?
Timo Thoms and Carlos Velasco
Estimating causal effects with regression discontinuity designs and matching, Eggers and Hainmueller (2009) and that serving as Members of Parliament causes Conservative politicians to accrue more wealth than unsuccessful Conservative candidates. We revisit their results by addressing two methodological issues in their analysis. First, we follow Imbens and Kalyanaraman's (2009) proposed rule for deriving the optimal bandwidth for the RDD, and reestimate the causal effect. This choice yields a much smaller optimal bandwith than that used by Eggers and Hainmueller (2009), and puts into question some of the paper's results. Second, we note that Eggers and Hainmueller (2009) introduce post-treatment bias by controlling for the year of death of candidates in the analyses. To address this issue, we suggest reconceptualizing the dependent variable as the average rate of wealth accumulation between treatment assignment and a candidate's end of life, and based on this new dependent variable, we reestimate the causal effects with both RDD and matching.
Voting, Income Inequality, and Polarization
Existing models of income-based voting tend to overlook societal factors which may critically condition the way income is related to preferences about redistribution. In this paper, we focus on two such factors: economic inequality and political polarization. We extend the model of income-based voting developed by McCarty, Poole and Rosenthal (2006) to analyze how inequality and polarization may modify the effect of income on redistribution preferences in multiparty systems. The model predicts that preferences about taxation will be more strongly associated with income in systems where party platforms are more polarized. Moreover, greater inequality will further enhance this association for those voters who support the most extreme parties. To test these hypotheses with crossnational survey data, we use a Bayesian hierarchical approach in which vote choices are modeled as function of national, party, and individual-level predictors.