| Robi
Ragan Visiting Assistant Professor Department of Political Science Duke University |
| • Home • CV • Research • Teaching • Personal • Contact and Links |
 Research: Dissertation and Working Papers |
| My Research interests are in American political institutions, political economy, and computational political science. My general research agenda
consists of formulating formal/computational models of American
political phenomena and empirically testing the implications of those
models. On this page, you can find abstracts for all of my current projects. I have divided the projects into two groups. 1) Front Burner: Active projects that are under review, or which I hope to send out for review in the near future, and 3) Back Burner: Projects which are not actively being revised at this point, but that I hope to take back up once time permits. With each abstract, I have included details about where the paper is in the overall publication process, as well as information about presentations based on this paper. Working Papers - Front Burner: "The Implications of Using Direct Democracy Models for Cases of Representative Democracy" Revise and Resubmit, Journal of Theoretical Politics Representative democracy translates the preferences of the electorate into policy outcomes. Individual voters do not directly vote on policy; rather, their elected representatives create and establish policy. How well does representative democracy translate the preferences of the electorate into policy? Is there any systematic bias in a representative democracy system? I examine the policy implications of a representative democracy system itself absent these other effects. To explore this question, I have formulated a series of computational models that calculate policy outcomes of both a direct democracy and a representative democracy system. The results allow me to isolate any systematic deviation between the two systems. I find that there are two main factors that cause the policy outcomes of a representative democracy system to deviate from direct democracy outcomes. The first is the distribution of preferences across the general population of voters. The second is the degree of heterogeneity within legislative districts. When population preferences are normally distributed and preferences within legislative districts are extremely heterogeneous, there is little difference between the policies predicted under representative democracy and direct democracy. However, when the preferences in the population are distributed in a log-normal fashion and the distribution of preferences within districts is more lop sided, the deviation between the two systems can be as much as 30%. The findings are important because a direct democracy model is often used as the model of policy formation in social science research. This modeling convention is common within the political economy literature, beginning with Romer's (1975) work on the political determination of the parameters of an income tax. Modeling representative democracy systems using direct democracy was firmly established in the literature with Meltzer & Richard's (1981) model of the “rational” size of government. - Poster Presented at the Midwest Political Science Association Meeting (April 2007) "The Implications of Inherited Status Quos for Models of Congress" For this project, the aim is to look at the implications of making existing static game theoretic models of Congress fully dynamic. Each of the dominant congressional models achieves its equilibrium predictions for each policy at time t from a static game. When the existing models are made dynamic, and status quos are allowed to be inherited across time periods, radically different equilibrium predictions are made for each of the models. For example, the party cartel model (which normally predicts that outcomes will be at or near the median member of the majority party) begins to predict median outcomes. Conversely, the pivotal politics model (which in the static case tends to be more majoritarian) predicts outcomes that are significantly different from the location of the median member of the chamber. Next, rules for learning and adaptation are added to the decision making process for each member of Congress. For example, in the party cartel model, the party leaders take into account the centripetal nature of the dynamics of Congress and adopt a different set of strategies depending on the political environment in each time period. This decision rule is forward looking and adaptive with respect to its expectation formation. The results from these adaptive models produce outcomes that further deviate from the predictions of the static game theoretic models. "An Experimental Study of the Efficiency of Unanimity Rule and Majority Rule" with Keith Dougherty, Brian Pitts and Justin Moeller It is widely believed that unanimity rule is more capable of producing Pareto efficient outcomes than other voting rules, such as majority rule, in cases where transaction costs are negligible (Buchanan & Tullock 1962; Ordeshook 1986; Mueller 2003). As a result, scholars have recommended unanimity rule to create Pareto efficient constitutions and optimal procedural rules. Dougherty and Edward (2008) have challenged this claim in two dimensional spatial voting games with finite rounds of voting. In this paper, we test three hypotheses related to the work of Dougherty and Edward: 1. Subjects are more likely to select Pareto-optimal outcomes after multiple rounds of voting under majority rule than under unanimity rule (contrary to standard equilibrium analyses); 2. Subjects will select Pareto-optimal outcomes under majority rule in less rounds of voting than under unanimity rule (regardless of the information condition); and 3. Subjects are less likely to select Pareto-optimal outcomes with complete information than with incomplete information; The results can be contrasted with theorems by McKelvey (1976), Shepsle & Weingast (1984), and Austen-Smith & Banks (2005) as well as the experimental research of Bianco et. al. (2007). - Paper Presented at Midwest Political Science Association Meeting (April 2009). "Growing Conditional Party Government in an Agent Based Model" with Gregory Robinson The goal of this project is to formulate a formal statement of Aldrich and Rohde's "Conditional Party Government" theory of Congressional policy making. The theory of Conditional Party Government (CPG) does not lend itself well to the static and game theoretic formalizations that other theories of Congress have received. The Cartel Model (Cox and McCubbins), Pivotal Politics (Krehbiel) and the Distributive Model (Weingast and Marshall), have all been translated into formal models. CPG, however, has several aspects that make it difficult (if not impossible) to translate it into the same sort of frame work the other models have been expressed with. The tension between the electoral and policy goals of the party and their interaction with the electoral and policy goals of members create a complex system with heterogeneous agents, feedback loops, and an important dynamic component. Instead of trying to incorporate all of these important components into a traditional game theoretic model, we construct an agent based model of the members of Congress. Thus we are able to, in the words of Shepsle (1986), ``address both institutional equilibria and equilibrium institutions". Using this agent based model, we are able to examine several questions that have remained unanswered due to the lack of a fully specified model of CPG. Many empirical investigations of CPG (and other party models for that matter) face severe simultaneity and omitted variable biases. For example, if party leaders are endowed with resources sufficient to "twist arms" in their caucus/conference, the question remains as to why self-interested party members would accept policy losses created by that arm twisting, even if those losses are relative rather than absolute. Another puzzle that is difficult to untangle empirically is that it becomes more difficult to distinguish true party effects from preference effects as the conditions of CPG become more clearly met. The problem is that members are predicted to delegate authority to their party leaders precisely when the use of punishments and inducements on behalf of the party position are least necessary (i.e. when the difference between pure-preference and party-biased outcomes is at its least). Using different parameter choices for the agent based model, we are able to begin to unravel some of these endogeneities and get a sense of the possible direction of the causal mechanism. This paper is a part of our larger "Complex Adaptive Congress" project, and represents the first major step in this project. - Paper Presented at Midwest Political Science Association Meeting (April 2008). "A Congressional Politics Theory of the Size of Government" In this project, I use an EITM approach to bring a formal models of Congress to bear on the question of redistribution and to empirically test the resulting model. I posit that the flaw in the current models of redistribution in the political economy literature is that they rely on a general population median voter as the pivotal actor. The classic example of such a model is the Meltzer and Richards' 1981 JPE piece. Institutions do matter, and when considering redistribution, we need to incorporate the actual mechanism by which this type of policy is made. As of yet, there is no model of the redistribution (what Meltzer and Richard call the “size of government”) that actually incorporates the U.S. Congress and the presidency. The general population median voter model is not applicable to the question of redistribution policy formation. In this paper, I take the Pivotal Politics model of Congressional policy making and analyze its predictions about the size of the tax and transfer system. After incorporating this model into a model of redistribution policy formation, it becomes clear why the Meltzer and Richards model has failed to stand up to empirical tests. In their model, the policy outcome is the median voter's ideal point. However, in reality, the median voter in the U.S. only gets to vote for his or her member, and where that member sits in the ideological distribution of Congress may or may not be the median. Further, the median member of Congress is likely not the actor whose ideal point becomes the policy outcome. For the model, a set of testable predictions about the size of the tax and transfer system are derived. These predictions are then tested using a time series approach using Federal data on redistribution in the U.S. with respect to the location of pivotal members of Congress. - Paper Presented at Southern Political Science Association Meeting (January 2008) - Paper Presented at Midwest Political Science Association Meeting (April 2007) - Paper Presented at EITM V - Ann Arbor (July 2006) "The Complex Adaptive Congress" with Gregory Robinson The use of complexity science and agent based models has been growing within social science in general and specifically within political science. The tools of complexity have been used to examine a wide range of social science issues. However, as of yet, there has been little application of the tools of complexity science to issues associated with American Political Institutions. This paper advocates taking a complex adaptive systems approach to modeling the United States Congress. Traditionally, formal models of Congress have used a static, single dimensional and game theoretic approach to modeling policy formation. However, we believe that there are analytical gains to be made by using an agent based, complex adaptive approach. In order to illustrate the analytical leverage of using the complex systems approach, we first build a set of agent based generative models that computationally replicate the three most prominent models of congressional policy making: the distributive/committee model, the partisan model, and the informational model. Each of these models exogenously assumes which agents in the policy making system are pivotal to the policy outcomes. With the dominant models replicated, we then unify the three models into one overarching model. We allow the pivotal actors in the Congress to arise endogenously. We find that under certain conditions individually rational members of Congress cede power to their party and refuse to do so under other conditions (an example of this is the theory of ”Conditional Party Government” put forth by Aldrich and Rohde). This model allows us to explore other emergent behavior within the system which is yet uncovered due to the partial equilibrium approach of many formal models of Congress. In addition to providing a unifying framework in which the three dominant models can co-exist, this macro level approach to studying Congress enables us to move past the limitations of "ceteris paribus" and comparative statics, to a framework in which we can allow many parameters to move simultaneously. This approach also lends itself to a hybridization between the worlds of behavioral and institutional American politics. For example, we are able to investigate questions such as: what is the affect of a change in the policy mood of the mass electorate on policy in light of electoral pressure, party pressure, and the institutional rules of the Congress itself? - Paper Presented at Southern Political Science Association Meeting (January 2008). "Proportional Allocation versus Winner-Take-All in U.S. Presidential Primaries: A Computational Investigation" with Paul Gurian This paper uses a set of computational simulations of a single dimensional spatial voting model in order to compare the effects of proportional and winner-take-all delegate allocation mechanisms in the U.S. presidential primary system. First a parsimonious analytical model of the primary system is created. Then the model is simulated for all possible combinations of proportional and winner-take-all primaries. The results indicate that systems with higher levels of proportional allocation will usually take longer to winnow down to one nominee. In fact, it takes a relatively small number of proportional allocation primaries in a given system to significantly prolong the time it takes to select a presumptive nominee. - Poster presented at American Political Science Association Meeting (August 2008). "The Complexity of Presidential Primaries" with Paul Gurian The forces underlying the process by which the major parties pick their nominees to face off in the Presidential election are not well understood. Is it true that the person who is the leader in the polls goes on to win the nomination, and why is this so? Even more importantly, how did that candidate become the leader in the polls in the first place? What determines the dynamics of the nomination campaign, delivering the results that forecasting models have found? The presidential nomination process (and indeed the entire presidential campaign) is an event in which time is a major component. Therefore, any model that hopes to sufficiently describe this process must have time as a component. It must be a dynamic model. An optimization model with comparative statics that looks only at a snapshot in time would likely miss some of the key aspects of the nomination process. This paper updates Aldrich’s 1980 APSR paper on Presidential Campaigns by taking a complex adaptive systems approach to modeling presidential primaries. In order to illustrate the analytical leverage of the complex systems approach, we first build a model that attempts to replicate Aldrich's model of presidential primaries (1980 APSR), using autonomous adaptive agents. To these we then add increasing complexity in the form of positive and negative feedback loops, learning rules and endogenizing several of Aldrich's exogenous assumptions. With this model in place, we are able to explore the behavior of the system when key parameters are changed (front loading, the number of candidates, types of candidates, perceived strength of the other party). - Paper Presented at Midwest Political Science Association Meeting (April 2008). Working Papers - Back Burner: Note: If you have any interest in any of these "back burner" projects, I am always looking for co-author opportunities that will kick these projects back to the "front burner". "Ideological Shirking in the United States Congress: A Multitask Principal-Agent Model" Ideological shirking is the situation in which an elected representative votes based on her own ideological preferences rather than on the preferences of their constituency. The literature on ideological shirking in the United States Congress has largely been empirical, and the empirical results present mixed results. Many empirical works claim to find evidence of ideological shirking and, on the other hand, there are also many works claiming there is no evidence of ideological shirking in Congress. This paper seeks to develop a theory that can add clarity to the discussion. By using a multitask principal-agent framework, I develop a model that examines the types of issues on which a member of Congress may successfully shirk. Several testable predictions about the existence of ideological shirking are derived, and implications with an eye toward empirical testing are discussed. "Sticky Position Theory of Non-Convergence in Presidential Campaigns" The standard Hotelling-Downs spatial model predicts that candidates converge on the median of the distribution of voter preferences in order to win an election. There is also anecdotal evidence that candidates attempt to moderate themselves in order to appeal to median voters. However, many empirical studies provide little evidence of this convergence. Why don’t candidates converge? This paper makes the argument that some candidates are unable to move toward the center even if they want to. Candidates may be sticky (unable to move to the center) on issues for one of a variety of possible reasons. Sticky candidates are those with well-known positions on issues, either through their, press or m past political office or , media coverage. Non-sticky candidates are not well known for their stance on issues; however, they may be well-known for other accomplishments. In this paper, I seek to model how stickiness applies in the basic spatial framework. I then test the model using empirical methods. "Party as Credible Commitment" In this paper, I propose a model explaining why candidates affiliate with political parties even though party affiliation limits their choices while in Congress. Through affiliation with a political party, the candidate is solving a problem of asymmetric information. The voter has limited information about the candidate, but the candidate has full information about herself. The candidate knows whether or not she will shirk on her campaign promises, but the voter has no way of knowing this information. So, in order to induce the voter to vote for her, the candidate has to solve this problem. She has to credibly commit to the voter that she will not shirk. She does this through affiliation with a political party. I develop a rational choice framework within which to examine the question of why Congressional candidates affiliate with political parties. The model is then tested empirically using panel data from congressional elections. |