ABSTRACT
This chapter analyzes the possibility of omitted variable bias, and considers various specification issues. It shows that altering the equation or the estimation technique of the original study, as critics have explicitly or implicitly suggested, either has no effect on the results or is not supported by the data. Munnell et al.’s Boston Fed study has elicited considerable reaction. Critics have raised serious questions about the quality of the data and the specification of the equation. The Boston Fed study has been implicitly criticized for not incorporating the variable in its analysis. Both econometric theory and common sense demand that this variable be omitted from the estimation since the variable is a measure of whether the application was actually denied a mortgage, not a determinant of mortgage denials. CM’s other major criterion was whether the obligation ratios calculated from the monthly expense and income variables equaled the obligation ratios lenders asserted they actually used in their lending decision.
