Essays on network effects, unobserved heterogeneity, and job search.

摘要

This dissertation is composed of three essays: "The Effect of Migrant Networks on Mexican Migration," "Testing for the Presence and Correct Specification of the Unobserved Heterogeneity in Likelihood Models," and "Benefit Durations and Job Search Behavior.";The first essay analyzes the factors determining Mexican migration decision to the United States, focusing on the causal effect of family networks on the timing of a Mexican household head's migration. This essay contributes to the migrant network literature on relating family members' migration as a channel to the migrant networks and showing the econometric problems if the correlated migration behavior between family members is not considered. This essay proposes an econometric model incorporating the family factors based on the duration models, while no instrumental variables are needed. After correcting for endogeneity, we find that the substitution effect of family members' migration outweighs the complementary effect of family members' propensity to migrate, and therefore, the estimated network effect is not understated, as opposed to the downward biased network effect measured in the models without taking endogeneity into account.;The second essay proposes a test for detecting misspecification of unobserved heterogeneity in a wide class of models. The test can also be used to detect the presence of unobserved heterogeneity and is based on the Hausman specification test. In particular, this study shows the conditions under which the maximum rank correlation estimator is a regular estimator and compares this estimator to an efficient estimator.;The third essay focuses on the implications of the remaining/elapsed Unemployment insurance (UI) benefit durations and the changes of the UI entitlements, and investigates the job search behavior of the recipients in a structural model framework. This study advances the literature on quantifying job search efforts and providing evidence to show that the benefit recipients find jobs quicker when benefits are running out. Moreover, this study takes the beneficiaries' expectation for benefit extension into account and deals with the endogenous extension problem. We find that those recipients more likely to find new jobs possess characteristics unobserved by econometricians that also drive them less likely to get benefit extension.