Three essays on conflict and financial markets, and political methodology

Thesis (Ph. D.)--University of Rochester. Department of Political Science, 2015.This dissertation consists of three distinct papers. In the first paper, I estimate the effects of battles in the American Civil War on the bond yields of U.S. government bonds. Since the yields of U.S. government bonds were primarily a function of the expected outcome of the war, they are a proxy for the expected cost of war to the United States. While Confederate victories increased the expected cost of the war to the United States, U.S. victories had little effect on war cost expectations. The effects of battles on expected war costs are inconsistent with a private information theory of war, since the largest effects occur late in the war, in the summer of 1864. Surprisingly, major battles explain little of the variation in expectations of war costs. In the second paper, I estimate the market implied ex ante probability of the onset of the American Civil War using U.S. government and state bonds. Surprisingly, financial markets were surprised by the Battle of Fort Sumter and the start of the war. Prior to Abraham Lincoln’s election in November 1860, the market assigned almost no probability to a war. Even after the secession of several states, the week before the Battle of Fort Sumter, the market assigned a negligible probability, approximately 5%, to war onset. In the third paper, I provide a general method to estimate change points using Bayesian sparse shrinkage priors, such as the horseshoe distribution. This method is flexible and can estimate change points in a variety of models without assuming a fixed number of change points. Since many of these models can be represented as Gaussian dynamic linear models, I provide a method to efficiently sample these models using a partially collapsed Gibbs sampler. I implement this method, as well as functions to perform Kalman filtering, smoothing, and sampling in the probabilistic programming language Stan