Working Paper 96-10
by Geoffrey M. B. Tootell

Several articles in the popular press have asserted that a simple comparison of average mortgage default rates for white and minority applicants is necessary and sufficient to uncover discrimination in mortgage lending. The fallacy of this assertion has been examined in Peterson (1981), Tootell (1993), and Yinger (1993). These papers show that a failure to account for the financial characteristics of each application or loan makes a simple comparison of average rates meaningless. However, recent empirical work on discrimination in mortgage lending has examined both application denial and mortgage default rates conditional on the strength of each application, not average rates for whites and minorities. This paper assesses the information about discrimination contained in these conditional rates. It is found that the debate over denial versus defaults is misdirected; examining denials is a marginally better method to uncover discrimination. Much of the apparent debate was really over the potential importance of omitted variables.

PDF version of paper