Using Hedonic Models to Measure Racial Discrimination and Prejudice in the U.S. Housing Market

Given the longstanding goal of racial equality in the USA (and elsewhere), there have been many attempts to measure the presence of racial discrimination in the housing market. In this context, racial discrimination is an action whereby nonwhites are treated differently than whites in some aspect of the housing market. Given the complexity of the housing market, racial discrimination can manifest itself in a number of ways. First, suppliers of housing can price discriminate and charge nonwhites more than whites. Second, whites can by force, threat, or collusion prevent nonwhites from living in certain areas. This can include some forms of zoning or racial covenants that can restrict the types of individuals that can purchase houses in certain areas or towns. Third, real estate agents can steer nonwhites away from white neighborhoods and hence deny nonwhites access to these areas. Fourth, nonwhites can be denied mortgages at a higher rate than whites, all else equal. Fifth, lenders can refuse to write loans in certain high minority areas; this is known as redlining. Sixth, lenders can charge higher prices to nonwhites for mortgages by offering higher interest rates or by forcing them to apply for private mortgage insurance. The focus of this chapter is on the fourth of these methods for detecting discrimination in the housing market. Initially, a general framework for detecting discrimination based on the hedonic model of house prices will be established. Then, this framework will be used to evaluate the literature. This paper is limited to an analysis of the U.S. housing market since this is the basis of most of the research on discrimination in the housing market (but see Harrison et. al. (2005) for an analysis of housing discrimination in the European Union). The burst of energy in the 1970s devoted to estimating discrimination and prejudice in the housing market using hedonic house price models has been followed by a relative dearth of such studies in the past twenty-five years. This might be due to a change of focus to other forms of discrimination in the housing market (i.e. forms three through six above). Also, with the advent of estimable forms of the general equilibrium urban model, it is now possible to analyze the impact of racial preferences on residential patterns in urban areas in a general equilibrium framework rather than the inherently partial equilibrium framework that underlies the studies based on hedonic models.3 Another reason is that the data requirements for accurately estimating discrimination using hedonic models are particularly onerous. Given these difficulties, along with other econometric issues that arise in specifying and estimating the hedonic model, it is recommended that the focus should be on trends in racial discrimination in the housing market rather than on point estimates from crosssection data. Of course, this only adds to the data requirements for estimating these trends.