Essays in risk and applied Bayesian econometric modeling

Abstract

My dissertation consists of three essays modeling economic agents' optimizing behaviors under various conditions, with special focus on the role of risk and Bayesian prior updating. Each essay combines in a multidisciplinary fashion tools from various fields to uniquely model agent's behaviors.

The first essay is an analytical investigation of the impact of risk on publication decisions by junior faculty members. It is motivated by the fact that an untenured member of a university faculty often faces pressure to publish or lose his position in the institution where publication is the ultimate criteria for advancement. A junior member faces risks during the probationary period for which a risk market is absent, and so he must manage his exposure to risk privately. Combining tools from the physics literature with modeling tools from probability theory and the economics of risk this essay analyzes ex-ante incentive properties of tenure from the perspective of the junior faculty member seeking tenure. The modeling principle used in this analysis is the first of its kind on the topic, and provides useful insights on the implicit cost of risk and its role in publication decisions.

The second essay introduces a new variant of the mixed logit model that relies on the multinomial logit formulation of the weighted logit formula. Current models in the literature are based on conditional logit representation. The model is used to investigate the consistency of revealed health insurance choices, with expressed attitudes towards health insurance cost, in an attempt to test for the rationality principle in the context of health insurance enrollment decisions by Medical Expenditure Panel Survey (MEPS) respondents.

The third essay models the effect of an individual's stated preferences for health insurance on their revealed choices of health insurance, using a framework in which stated preferences are assumed to be endogenous in the statistical sense. A discrete choice analysis is implemented where both stated preferences and choice outcomes are modeled through the multinomial probit and estimated in the Bayesian paradigm using a full Gibbs sampler. This estimation strategy is the first of its kind in the health economics literature, and improves computational efficiency and overall simplicity compared to related work using full information maximum likelihood. The analysis uses data from the 2007 MEPS.