James P. Scanlan, Attorney at Law

Disparities by Sector

(June 7, 2014)

I first explained the pattern implications of the pattern whereby the rarer an outcome, the greater tends to be the relative difference in experiencing it and the smaller tends to be the relative difference in avoiding it in the fair lending context in April 17, 1992 American Banker article titled  “Bias Data Can Make the Good Look Bad.”  The twenty-plus subsequent articles on the subjects are collected on the Lending Disparities page of jpscanlan.com, the most recent of which is the May 2014 Mortgage Banking article “The Perverse Enforcement of Fair Lending Laws.”

One of the points I made in the 1992 American Banker article was that minorities could more efficiently use their time seeking mortgage loans by first approaching lenders with large relative differences in mortgage rejection rates.  For those would be the lenders where minorities, like whites, would have the greatest chance of having their applications approved.   Relative differences in approval rates would tend also to be smaller among the lenders with comparatively large relative differences in rejection rates. 

An interesting illustration of that point can be found in a November 18, 2013 article in Mortgage Compliance Magazine titled “All Lenders Are Not Alike: Making Sense Out of you HMDA Data.”  The article contained information on denial rates by lending sector (for 26 sectors) as well as ratios of black denial to white denial rates and Hispanic to white denial rates. 

Table 1 below show the averages of the black/white and Hispanic/white rejection ratios according to whether the lending sector was among  the 13 sectors with highest denial rates or the 13 sectors with the lowest denial rates. 

Table 1.  Ratios of Black to White and Hispanic to White Mortgage Denial Ratios by 13 Lending Sectors with Highest and 13 Lending Sectors with Lowest Overall Rejection Rates.

The Table shows the common pattern whereby the relative differences in adverse outcomes are larger in the settings with the lower frequency of the adverse outcomes.  Presumably, the underlying would show an opposite pattern for approval rates.  That is, the relative differences in approval rates would be larger in the settings with the higher denial rates.