Date of Award

5-2016

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Legacy Department

Economics

Committee Member

Dr. Curtis Simon, Committee Chair

Committee Member

Dr. Scott Baier

Committee Member

Dr. Tom Lam

Committee Member

Dr. Kevin Tsui

Abstract

The first chapter of this dissertation focuses on the heterogeneity in earnings losses of displaced workers. A large literature shows that workers, on average, suffer large and persistent earnings losses when displaced from a job. However, little attention has been devoted to the distribution of those losses. Using the 1994-2014 Current Population Survey displaced worker surveys, I find that averages, whether from simple summary statistics or calculated as expectations using ordinary least squares (OLS), give a misleading and incomplete picture. The average loss in earnings in my sample is 29%, compared with a median loss of just 12%. Using OLS, each year of tenure is associated with an additional 1.3% reduction in earnings. Quantile regression estimates reveal that this magnitude of effect corresponds to the 80th percentile, and that the effect of tenure at the median is just 0.7%. OLS estimates of the effect of changing industries and occupations are also considerably larger than quantile regressions at or below the median. I also attempt to correct for possible endogeneity in the choice to switch industry and occupation. Two-stage least squares estimates yield even larger estimated negative effects of changing industry and occupation than OLS, but instrumental variables quantile regression estimates, using the model developed by Chernozhukov and Hansen (2005), suggest that the marginal effects on such losses are highly concentrated among relatively few workers. Finally, I demonstrate that there is a positive relationship between the magnitude of earnings losses and the level of skill associated with the pre-displacement jobs. Workers displaced from jobs in the third skill quartile tend to find jobs lower down in the occupation distribution, consistent with Autor and Dorn's (2013) polarization hypothesis. However, workers displaced from jobs in the middle of the distribution tend to find jobs in the middle post-displacement, and workers displaced from jobs lower down in the distribution tend to find jobs higher up. The second chapter of this dissertation deals with the self-selection bias and unobserved heterogeneity present in the estimation of the effect of economic integration agreements (EIA) on international trade flows. As detailed bilateral trade flow data have become more readily available over time, panel data approaches to empirical bilateral trade flows employing the gravity model have allowed researchers to better estimate the impact of time varying trade costs. However, as pointed out by Helpman, Melitz, and Rubinstein (2008), there are numerous instances of zero (or non-reported) bilateral trade flows in any given year. If these zero bilateral trade flows arise from non-random events, the resulting selection may lead to biased coefficient estimates. We build on the contributions of Helpman, Melitz and Runbinstein by using panel data techniques to assess the importance of selection over time. We first show that the occurrences of zeros appear to be non-random. To control for selection we employ a method that combines time varying probit specification with a correlated random effects model in the second stage. We show that after controlling for selection and firm heterogeneity, the inclusion of bilateral fixed effects has a modest impact on the measured effects of free trade agreements. To our knowledge, we are the first to apply this technique to the gravity literature.

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.