How do political actors learn about their environment when the “data” provided by political processes are characterized by rare events and highly discontinuous variation? In such learning environments, what can theory predict about how learning actors will take costly actions that are difficult to reverse (e.g., eliminating programs, approving a risky new product, revising a security policy, firing or recalling an appointed or elected official)? We develop a formal model for this problem and apply it to the termination of bureaucratic agencies. The conventional wisdom that “the older a bureau is, the less likely it is to die” (Downs 1967, Inside Bureaucracy) persists but has never been properly tested. This paper offers a learning-based stochastic optimization model of agency termination that offers two counterintuitive predictions. First, politicians terminate agencies only after learning about them, so the hazard of agencies should be nonmonotonic, contradicting Downs's prediction. Second, if terminating agencies is costly, agencies are least likely to be terminated when politicians are fiscally constrained or when the deficit is high. We assess the model by developing a battery of tests for the shape of the hazard function and estimate these and other duration models using data on U.S. federal government agencies created between 1946 and 1997. Results show that the hazard rate of agency termination is strongly nonmonotonic and that agencies are less likely to be terminated under high deficits and divided government. For the first 50 years of the agency duration distribution, the modal termination hazard occurs at five years after agencies are enabled. Methodologically, our approach ties the functional form of a hazard model tightly to theory and presents an applied “agenda” for testing the shape of an empirical hazard function. With extensions, our model and empirical framework are applicable to a range of political phenomena.

Social scientific theories frequently posit that multiple causal mechanisms may produce the same outcome. Unfortunately, it is not always possible to observe which mechanism was responsible. For example, IMF scholars conjecture that nations enter IMF agreements both out of economic need and for discretionary domestic political reasons. Typically, though, all we observe is the fact of agreement, not its cause. Partial observability probit models (Poirier 1980, Journal of Econometrics 12:209–217; Braumoeller 2003, Political Analysis 11:209–233) provide one method for the statistical analysis of such phenomena. Unfortunately, they are often plagued by identification and labeling difficulties. Sometimes, however, qualitative studies of particular cases enlighten us about causes when quantitative studies cannot. We propose exploiting this information to lend additional structure to the partial observability approach. Monte Carlo simulation reveals that by anchoring “discernible” causes for a handful of cases about which we possess qualitative information, we obtain greater efficiency. More important, our method proves reliable at recovering unbiased parameter estimates when the partial observability model fails. The paper concludes with an analysis of the determinants of IMF agreements.

We present a simulation technique for sorting out the size, shape, and location of the uncovered set to estimate the set of enactable outcomes in “real-world” social choice situations, such as the contemporary Congress. The uncovered set is a well-known but underexploited solution concept in the literature on spatial voting games and collective choice mechanisms. We explain this solution concept in nontechnical terms, submit some theoretical observations to improve our theoretical grasp of it, and provide a simulation technique that makes it possible to estimate this set and thus enable a series of tests of its empirical relevance.

The dominant methodology for short-term forecasting of electoral outcomes uses trial-heat polls, where respondents report their current electoral preferences (not their election-day predictions). Election markets, where self-selected participants trade shares of candidates at prices predictive of election-day results, provide an alternative method that often produces more accurate forecasts. Consequently, increasing attention is being paid to this methodology. However, it is poorly understood and lacks theoretical justification. Surprisingly, the rationale for forecasting using trial-heat polls has not been completely developed either. We develop the justification for using both election markets and public opinion polls to forecast electoral outcomes, giving conditions under which each method performs ideally. For the ideal case, we prove (under the reasonable assumption that market participants are aware of the poll results) that the mean square prediction error for the market forecast is smaller than that of any forecast based on one or more polls. The case in which the assumptions supporting each method fail is also considered. It is often reasonable to expect that the best case results hold approximately, in which case the market forecast should also beat any poll-based forecast. We also compare the bias and variance of market and poll-based forecasts; our results suggest the utility of using the series of market prices to study the course of campaigns.

Existing research on electoral politics and financial markets predicts that when investors expect left parties—Democrats (US), Labor (UK)—to win elections, market volatility increases. In addition, current econometric research on stock market volatility suggests that Markov-switching models provide more accurate volatility forecasts and fit stock price volatility data better than linear or nonlinear GARCH (generalized autoregressive conditional heteroskedasticity) models. Contrary to the existing literature, we argue here that when traders anticipate that the Democratic candidate will win the presidential election, stock market volatility decreases. Using two data sets from the 2000 U.S. presidential election, we test our claim by estimating several GARCH, exponential GARCH (EGARCH), fractionally integrated exponential GARCH (FIEGARCH), and Markov-switching models. We also conduct extensive forecasting tests—including RMSE and MAE statistics as well as realized volatility regressions—to evaluate these competing statistical models. Results from forecasting tests show, in contrast to prevailing claims, that GARCH and EGARCH models provide substantially more accurate forecasts than the Markov-switching models. Estimates from all the statistical models support our key prediction that stock market volatility decreases when traders anticipate a Democratic victory.