This has expanded homeownership opportunities but has led to a segmented credit market where borrowers fall into three categories; prime borrowers, i.e. those who always got credit and now get the best deals, and "subprime" or "near prime" borrowers. Those in the subprime category typically pay the highest rate because of the risks they pose; not only a risk that they won't pay on their obligation but that they will be more likely to prepay their mortgages ahead of schedule once they have cleaned up their credit, build up equity or increased their income. These subprime borrowers can also be more costly to service, requiring more monitoring or greater collection efforts. But, the thresholds that separate these market segments can change as market interest rates move, as lenders' appetites for interest rate or credit risk change, as technological improvements allow for more precise risk assessments.

The report states that the nonprime market has grown dramatically in recent years and quotes one source as saying that, from 1994 to 2005 the dollar volume of subprime loans increased from $35 billion to over $600 billion and subprime loans are now estimated to make up 20 percent of all mortgage originations compared to less than 5 percent in 1994.

Subprime loan originations do not correspond exactly with those loans defined as having prices exceeding the threshold as defined by Regulation C, however the latter increased significantly in 2005 over 2004 figures. For example, the incidence of higher-priced lending for conventional, owner-occupied, first-lien home-purchase loans rose from 11.5 percent in 2004 to 24.6 percent in 2005. This increase, however, was driven in part by the flattening of the yield curve, by the yield curve coupled with an artifact of the way APRs on adjustable rate loans are determined, and borrower or lender-specific changes in the risk characteristics of lending. These changes in risk characteristics were in part because substantial increases in house prices in some parts of the country caused more borrowers to stretch financially to obtain loans.

Analysis of the HMDA data revealed substantial variations in the incidence of higher-priced lending across racial and ethnic lines and the report states that these differences could not be fully explained by information in the HMDA data. The 2005 HMDA data indicate that black and Hispanic borrowers are more likely, and Asian borrowers less likely to obtain loans that cost more than the pricing thresholds than are non-Hispanic white borrowers. The gross mean incidence of higher-priced lending for home purchases was 54.7 percent for blacks and 17.2 percent for non-Hispanic whites. When the analysis took into account variables other than race that effect either borrower or lender this 37.5 percentage point difference is reduced to about 10 percentage points. When it comes to refinancing the gross difference between blacks and non-Hispanic whites is 28.3 percent but this is reduced to 6.2 percentage points after controlling for borrower and lender factors.

Analysis of the HMDA data across several years has indicated that loans are denied at different rates when applicants are grouped by race or ethnicity. 2005 was no exception. For each loan product category, American Indians, blacks, and Hispanic whites had higher denial rates than non-Hispanic whites; blacks generally had the highest denial rates and non-Hispanic whites the lowest; Hispanic whites fall about midway between the two other groups. The denial rates for Asians relative to other ethnic groups varied across loan products.

As with pricing, controlling for other borrower and lender variables reduced the disparity in denial rates. For example, blacks had a gross denial rate for first-lien home purchase loans of 27.5 percent compared to 12.3 percent for non-Hispanic whites. Accounting for income, loan amount, and other borrower-related factors reduces the difference by 3.1 percentage points and adding lender factors to the control reduces the gap to 7.0 percentage points. Refinancing patterns were very similar.

Edwina Baniqued