Degree Type


Date of Award


Degree Name

Doctor of Philosophy



First Advisor

Tanya Rosenblat

Second Advisor

Markus Mobius


I use experimental methodology to study interactions in the labor market which are otherwise unobservable. In my experimental labor market, textquotedblleft workerstextquotedblright perform a real effort task by solving character puzzles. The worker first solves a single practice puzzle and then is paid to solve as many puzzles as possible in a 5-minute task period. I interpret the puzzles solved in the 5-minute task period as the worker's actual productivity, and the time to complete the single practice puzzle as a noisy signal of that productivity. Based on this noisy signal and other labor market characteristics such as ethnicity, gender, and urban/rural, "employers" are given incentives to estimate the productivity of workers.

Compared to observational data, my experimental methodology has four key advantages. First, I precisely measure the workers' productivity through their ability to solve puzzles. This design eliminates the unobservable factors such as non-cognitive skills, which are likely to affect real labor market outcomes. Second, I build a direct link between the signal and the productivity using the uniform measure of puzzle-solving ability. Third, I explicitly measure the workers' self-confidence with a self-evaluation of their puzzle-solving ability. This measure allows me to study the role of self-confidence in the labor market. Finally, I can construct informative resumes for workers and observe how employers interpret this information when evaluating workers.

In Chapter 1, "Diversity and Discrimination in Experimental Labor Markets," I use this experimental framework to study how stereotyping discrimination against ethnic minorities depends on the shares of ethnic groups in the population. To this purpose, I conduct the experiment with university students in two Chinese provinces: (1) a diverse province, where 60% of the population is Han Chinese; and (2) a non-diverse province, where 99% of the population is Han Chinese. The stereotype against ethnic minorities is measured by the employer's estimate of minority workers' productivity.

I find that: (1) Han and minority workers are equally productive; (2) in the non-diverse province, Han employer productivity estimates are significantly lower for minority workers; (3) in the diverse population, a minority worker's productivity is equally estimated by Han and minority employers.

This research furthers our understanding of the economic effects of diversity. It establishes a negative relationship between labor market stereotypes and diversity. Such findings may also provide an explanation for why the inflows of immigrant workers in some US states, like California, have continuously increased. My work suggests that the immigrant workers are looking for diverse communities with lower stereotypes in the labor market.

In Chapter 2, "Self-confidence and Wage in Experimental Labor Markets," I study how signaling self-confidence to employers increases the worker's wage. Self-confidence is an example of a non-cognitive skill, that is likely to be important in the labor market. My experimental framework provides an explicit measure for self-confidence: the worker's evaluation of their own productivity.

I find that for workers, being self-confident is a channel to signal high productivity to employers. Specifically, signaling 1% higher self-confidence to employers increases the employer estimate by 0.09%-0.21%, controlling for other labor market characteristics. The results establish the signaling value of self-confidence in wage negotiations, and highlight the importance of non-cognitive skills in the labor market.

Chapter 3 proposes a methodology to measure the value of worker characteristics. In the design, employers buy worker characteristics in the Becker-DeGroot-Marschak (BDM) market. Specifically, employers claim a willingness to pay (WTP) for a characteristic. This characteristic is displayed on the resume if the WTP is higher than or equal to a randomly determined price. The value of a characteristic is measured by the magnitude of the WTP.

This methodology can be applied in pricing discrimination. The common method to do so is an ex post approach, in which we study discrimination with a wage regression. In such regressions we measure the discriminatory wage differential of a characteristic, which is not related to the actual productivity, by looking at its coefficient in the regression. In our design, the discriminatory wage differential is measured by the WTP of a characteristic.

Although Chapter 4 is not based on the experimental labor market, it serves as a complementary study to Chapter 1. To demonstrate how indirect contact can influence economic behavior, in this chapter, I study intergroup cooperation after observing in-group members interacting with out-group members.

In the control treatment, a student is matched with someone from the other major in a two-player public goods game. In another treatment, the game players watch intergroup contact prior to the public goods game. The intergroup contact is defined by playing a jigsaw puzzle with someone from the other major. I find that relative to the control treatment, the contribution to the public goods after observing intergroup contact is significantly higher. To distinguish intergroup contact effect from simply putting subjects in a cooperative mood, the game players in a third treatment watch random contact. The results show that it is important to have in-group members in the contact.

The results suggest that indirect contact can be applied when direct contact is restricted. When intergroup cooperation is desired, yet one or more groups are not available, we can select some members from each group and perform demonstrations on the rest. This is particularly useful for majority-minority intergroup cooperation, and for groups that are segregated in many dimensions. Indirect contact also implies financial freedom, as getting every group member involved in direct intergroup contact is very costly.

To summarize, my dissertation contributes to the growing experimental labor market literature. Relative to data from the real labor markets, the experimental labor markets allow us to study otherwise unobservable interactions. With such experimental labor markets, I study the relationship between stereotyping discrimination in the labor market and diversity, the signaling value of self-confidence in wage negotiations, and an alternative methodology to price worker characteristics. In addition, I study the application of indirect contact in raising intergroup cooperation.

Copyright Owner

Qiqi Wang



File Format


File Size

147 pages

Included in

Economics Commons