Regional variation of mortgage yields and simultaneity bias

This paper has important methodological implications for the literature on regional variation of mortgage yields. Previous studies typically estimate models of regional disparities in mortgage yields using single‐equation regression techniques. It is well known, however, that such methods can lead to biased estimates when simultaneity exists. In this paper it is demonstrated that the single‐equation approach, when applied to mortgage data from the 1980s, produces radically different results from the previous literature. It is further demonstrated that simultaneous‐equations techniques resolve these apparent anomalies.