Reputation Games and Political Economy

Abstract

This dissertation studies the applications of reputation games in social media and finance as well as decision games in political economy. Chapter 1 develops a reputation game in which a biased but informed expert makes a statement to attract audiences. The biased expert has an ideological incentive to distort his information as well as having a reputation concern. The expert knows that his expertise may vary in different topics, while the audiences cannot identify such differences. The biased expert is more likely to announce his favorite message when he knows less about it. Moreover, the biased expert is less willing to lie when the audiences have better outside options, and such improvements in outside options may benefit both the expert and the audiences.

Chapter 2 studies a credit rating game with a credit rating agency(CRA), an issuer and an investor. The privately informed and biased CRA provides a rating on the issuer's project, and the investor decides to purchase the project or not according to the report. As long as the CRA obtains a contract, he will inflate the rating. When the default risk is high, the CRA tells the truth. Moreover, he is more likely to tell the truth when the issuer's private benefit is larger. When the default risk is low, the CRA sends a good rating. He is more likely to inflate the rating if the issuer has a higher private benefit.

Chapter 3 presents a model in secessions and nationalism, with a special emphasis on the role of civil war. In our model, a disagreement on secession between the central government and the minority group leads to disastrous military conflicts. As a result, the tremendous potential cost of the war distorts the political choice of the minority group, and helps the central government to exploit them both economically and politically. Several key ingredients, such as population, per capita income and perceived winning chance of the civil war, play an essential role in the decision making process of the minority group. I also conduct an empirical test of this model, which supports the major findings stated above.