Date of Award

Spring 5-15-2017

Author's School

Graduate School of Arts and Sciences

Author's Department


Degree Name

Doctor of Philosophy (PhD)

Degree Type



I explore the role of experimentation on the career choices of individuals deciding whether to be paid employees or entrepreneurs, and on the decisions of consumers deciding what medical treatment to buy. In the labor market, experimentation entails the accumulation of information that allows individuals to improve their occupational choices. In the product market, experimentation entails discovering the quality of new products and it may have an effect on the evolution of technology: less experimentation by individuals may slow down the process of innovation. I start in Chapter 2 with the observation that most individuals do not start a business and, if they do, tend to do so well into their thirties. While policies encouraging young, would-be entrepreneurs are popular, little is known about whether they are effective. Using data from the Panel Study of Income Dynamics, I estimate a dynamic Roy model with imperfect information about ability to evaluate the relative importance of various economic determinants of entrepreneurial participation. Risk-averse, forward-looking individuals sequentially select entrepreneurial and paid-employment occupations based on their returns to experience, information value, non-pecuniary benefits, and entry costs. Results show that the main barriers faced by young entrepreneurs are entry costs and information frictions. I consider two policy counterfactuals: a subsidy targeting entry costs and entrepreneurship education targeting information frictions. I extend previous literature providing a mapping from the information quality of entrepreneurship education into career choices and long-term outcomes. A subsidy for young entrepreneurs increases participation but has small long-term effects. Entrepreneurship education can have sizable effects on participation and present value of income flows, even for low information quality. Nevertheless, the value of any particular entrepreneurship education program will depend on its cost and its information quality. In Chapter 3 we develop and estimate a dynamic structural model of demand for a product line whose spectrum of characteristics evolves over time because innovation is endogenous to consumer demand. To achieve this goal, we provide a new approach to the econometric challenge of estimating the process of technological change where innovation under uncertainty includes both frequent and incremental modifications along with sporadic major breakthroughs. Quality in our model is a multidimensional object: new products that are superior in some dimensions might be inferior in others. For example, new medicines more effective in combating disease than existing products sometimes have harsher side effects. In our model, consumer choices determine both the speed and the direction of product innovation. Demand externalities arise because the aggregate choices of atomistic individuals drive innovation. We apply our framework to analyze consumer choice and the realized path of innovations over a long time horizon in a maturing product market: HIV drugs. In this market, we observe the introduction of hundreds of new products, marking mostly modest, but sometimes major innovations over existing technologies. Our estimates are obtained through simulations of alternative hypothetical worlds that might have arisen if the innovations had taken different paths to the ones we observe. We use our estimates to assess the effects of policies that internalize the externalities affecting innovation and consumer welfare by modifying consumer choices. We find that experimentation in clinical trials is one of the mechanisms through which the externality operates.


English (en)

Chair and Committee

Barton Hamilton

Committee Members

George-Levi Gayle, Limor Golan, Robert Miller, Carl Sanders,


Permanent URL: