Bayesian Learning and Statistical Discrimination in the Market for Small Business Loans

2015-11-25 09:13:30

Dear students, faculty and staff,

Nazarbayev University Graduate School of Business invites you to Research Seminar


by Atanu Rakshit & Jonathan Peterson, Assistant professors of the Graduate School of Business, Nazarbayev University

WHEN: Thursday, December 3rd, 11.30 – 12.30

WHERE: Block C-3/new GSB/GSPP Building, 3rd floor, Room 3.037


In this paper we show that when applying for small business loans in the U.S., Black, White and Hispanic owned businesses are treated differently from each other. In particular, the manner in which banks use other information about the firm (such as credit rating) differs across these groups. We argue that these differences are consistent with race based statistical discrimination.

We develop a simple theoretical model that shows that the effect of signals on loan approval rating can differ when banks believe there is an underlying difference in riskiness between two groups (all else equal). Some of the obvious signals are standard credentials such as the company financials. However, the banks may also be tempted to use other information such as the race of the owner. When the financials are average, the information about race can have a significant effect on the anticipated probability that the loan will be repaid. When the financials are tremendous (or abysmal), race should be less informative and should have very little effect on the anticipated probability of repayment. Companies that are statistically discriminated against based on race should see their approval rating increase more rapidly as their financials improve. That is for firms who have faced statistical discrimination and have good financials; an increase in the quality of the financials should have a greater effect on the bank’s anticipation of repayment than for firms that were already seen as better risks. Other types of discrimination, such as taste based discrimination, should not have this effect. We also test this theory using publicly available data from the Federal Reserve’s Survey of Small Business Finances (2003 and 1998), finding results that support our hypotheses. 

image (3)


Atanu Rakshit

Assistant Professor of Economics

Dr. Atanu Rakshit received his PhD in Economics and M.S. in Applied Economics from Virginia Polytechnic Institute and State University, and his research and teaching interests include International Macro-Finance, Time Series Analysis, International Economics and the Economy of Emerging Markets. Atanu’s current research is focused on non-linear time series econometrics, financial volatility and exchange rates and the impact of fiscal policy on business environment.

Prior to commencing his graduate studies in the U.S., Atanu worked as a Project Associate on a Central Statistical Organization, Govt. of India project on Natural Resource Accounting in India (2003-2005). Before joining the Graduate School of Business at NU, Atanu was a Visiting Assistant Professor of Economics at the Williams School of Commerce, Politics and Economics at Washington and Lee University, U.S. Atanu is also an avid reader and a die-hard soccer fan.

Jonathan Peterson

Assistant Professor of Managerial Economics and Competitive Analysis

Jonathan Peterson earned his PhD in Economics at Cornell University. His research interests are applied microeconomic theory in Industrial Organization and Organizational Economics. His currently works on firm responses to the threat of employee turnover, adverse selection in the used car market, and planned obsolescence in the textbook market. Prior to joining NU Graduate School of Business, Jonathan worked as a postdoctoral fellow at The Johnson School of Business, then as an assistant professor at the Faculty of Management, NRU Higher School of Economics, Moscow.