As the number of renters in the United States grows, cities are under pressure to ensure a supply of affordable housing. One technique is to establish rent controls. But while this policy is intended to keep rents low, it can perversely worsen the situation by diminishing housing supply, generating market inefficiencies, and spurring gentrification, according to some economists.
After combing through many published econometric analyses of urban rent markets, Yijie Sun ’19 and her advisor, Associate Professor of Mathematics Jack O’Brien, decided they could examine the problem in a unique way, potentially using a combination of agent-based modeling and Bayesian statistics to suggest a solution. Bayesian inference is a mathematical procedure that uses probability to update beliefs about parameter values when new data emerges.
Most models for urban rental markets focus on abstract totals, such as the aggregate supply and demand, and tend to overlook the interaction between the agents involved — the tenants and landlords, according to Sun.
“My study focuses on the individual,” she said. “This is more of a bottom-up approach.” By creating a stochastic, or randomly determined, simulation that accounts for the complexity of decisions people make, Sun can incorporate the different characteristics (e.g., incomes, vacancy rate, and market trend) that influence movement into and out of city neighborhoods by renters. She can also factor in decisions made by landlords, including their rent-setting process and their tendency to favor higher-income tenants when they have a pool of applicants to choose from.
Sun can tweak her model to see what happens when the influence of these factors changes. “It is interesting to see what happens when landlord bias goes from 10 percent to 90 percent,” Sun said. “This makes the average income in the area jump from $65,000 to $85,000.”
This summer, Sun has a grant from Bowdoin’s Surdna Foundation Undergraduate Research Fellowship Program, one of Bowdoin’s more than 200 fellowships dedicated to supporting student research. Many of the summer fellows live on campus, pursuing independent research under the guidance of a faculty advisor.
Sun, who is majoring in economics and mathematics, said she wanted to pursue this particular project because it combines her academic interests, and will help prepare her for graduate school in statistics. “I wanted to do something that is at the intersection of economics and math, specifically statistics, which is like an applied science,” she said.
“I’m really interested in how people can use existing data to draw conclusions,” she said. “It is really powerful.”
The ultimate goal of her summer project, called “Bayesian Analysis of Urban Rent Dynamics,” is to develop a model in which users can input housing-related information from any city — San Francisco, New York, even Shanghai, where Sun is from — and come up with a more reliable prediction of how city policies, like setting annual rent control increases, will influence rental markets and city neighborhoods.
“I will use Bayesian methods to infer model parameters from the data so as to make predictions about potential outcomes of various rent stabilization policies and determine how they can be best implemented in practice,” Sun wrote in her fellowship application.
She added in person, “The ultimate goal of rent control is to make housing more affordable. The model might help [cities] set the best rate.”