Search

This dissertation consists of three chapters in applied behavioral economics and machine learning applications in economics. The first chapter studies how reference-dependent utilities influence people's behaviors on crowd-sourced review websites and cause attribution bias. Using data from Yelp, I tested how potential disappointments may affect customers' reviews by applying a regression discontinuity design to control for unobserved factors that may also simultaneously influence ratings. This chapter links to an emerging literature of attribution bias in economics and provides empirical evidence and implications of attribution bias on online reputation systems. The second chapter extends the work of first study and explores attribution bias when both reference dependence and state dependence are possible to appear. I specifically use the scenario of special occasions to test two leading theories of attribution bias empirically. The empirical results can be explained by one theory of attribution bias where people have higher expectations about restaurants on special occasions and then misattribute their disappointments to the qualities of the restaurants. From the connection between our empirical analyses and theories of attribution bias, this chapter provides another piece of evidence of how attribution bias influences people's perceptions and behaviors. The third chapter connects machine learning with financial forecasting. I construct a model with recurrent neural networks and focus on the point forecasting of the yield curve to explore the possibility of having better forecasts for the term structure. While allowing similar interpretation as previous econometric methods, the neural network model in this paper shows better forecasting accuracy.