Balancing the Forces of the Market
New assistant professor of economics Yi Xin loved math growing up but then later, in college, realized she was interested in how people interact with markets. Her work focuses on the economics of industrial organizations—for instance, she is studying the mechanics of peer-to-peer lending, or microloan, markets. One of her goals is to make markets such as these more efficient and beneficial to all parties involved.
Xin grew up in Xi'an, China. She received two bachelor's degrees from Peking University in 2012, in economics and psychology, and in 2018 earned her PhD from Johns Hopkins University in economics. She joined the faculty in Caltech's Division of the Humanities and Social Sciences in the summer of 2018.
We sat down with Xin to learn more about markets and incentives, and to hear about her favorite new pastime in the Pasadena area.
What exactly is peer-to-peer lending?
In peer-to-peer lending, a borrower will go to a website instead of a traditional bank or credit card and will submit a loan proposal. They will receive an interest rate based on their credit score and other information, and then their proposal will be listed online where individual lenders, not banks, can take a look and decide whether they want to fund the project. These are usually small personal loans; a borrower may be paying off credit cards or taking out a loan for home improvement. The lenders take a high risk in funding these microloans, but they also charge high interest rates, so overall, they do make a profit.
How do you study these markets?
I develop models to explain why borrowers repay or default on their loans and how lenders make investment decisions. The goal is to gain a microlevel understanding of both the borrowers' and the lenders' behaviors. From the decisions people are making, we can understand more about their preferences and hence predict their behavior in scenarios where different mechanisms are implemented. This eventually helps us to assess the impact of policies and regulations.
In particular, I'm interested in learning more about the market mechanisms that facilitate the transactions between borrowers and lenders. These peer-to-peer markets typically use something called the reputation/feedback system. Specifically, if somebody repays the first loan on time, then they can return to the same website and obtain a lower interest rate for a second loan. But if a borrower defaults on a loan, they are banned from receiving another loan. This gives borrowers incentives to behave well.
I'm trying to evaluate how important this reputation mechanism is. One of my papers describes what would happen when the mechanism is removed. I found that about half of the lending interactions would go away. Basically, without the reputation mechanism, lenders have no reason to believe that the borrowers will repay. The reputation mechanism facilitates trust between the two parties, which is mutually beneficial for both sides. But I'm also looking at whether this is the optimal system or whether it might be improved to work more efficiently.
What other projects are you working on?
In general, I'm interested in markets with asymmetric information. This means that when you have two parties involved, such as borrowers and lenders, one group knows more than the other. For example, take the car insurance market. There are two parties—the insurance company and the people insured, or what we call the agents. The agents know more about themselves: they know whether they are good drivers or not. But the insurance company does not know as much about the agents' driving skills.
One project I'm working on now is to analyze data from a mobile app that keeps a record of people's driving patterns. We want to understand whether drivers engage in less-risky behavior after experiencing adverse events, such as accidents or near misses. If drivers indeed become more risk averse, what are the implications for insurance pricing? Potentially, these driving data could be used by the insurance company to distinguish between bad and good drivers. Some auto insurance companies in the U.S. have adopted similar technologies to monitor drivers' behavior and adjust prices accordingly.
Do you think a lot of people will want to download and use this app?
If you are a good driver, you would use the app and get discounts from the insurance company. The bad drivers are less willing to participate because they are worried that their true driving behavior will be revealed. Overall, the goal is to incentivize the drivers to do what they are supposed to, such as drive well.
Are you working with insurance companies on this study?
Not at this time, but I hope to connect with companies in the future.
How do you like Pasadena so far?
It has a little too much sunshine! I miss the rain in Baltimore. But I like the city—it's not too big or small, and I can find whatever I need here. I've also been enjoying a table tennis club in Rosemead where my husband and I have been learning to play Ping-Pong with a professional player.
I also really like the environment at Caltech. The senior people in my division are very supportive of us junior faculty.