skip navigation

Sunday, November 23, 2014

[ – ] Text Size [ + ]  |  Print Page

The Quality of FHA Lending in Pennsylvania, New Jersey, and Delaware

1 | 2 | 3 | 4 | 5

Credit Scores

Print Version of Complete Report

FHA Lending Patterns Nationally and in the Third District States PDF (27 pages, 858 KB)

Credit scores are used as an indicator of the quality of the borrower pool in the FHA portfolio. Although they are an important indicator for this purpose, it should be noted that they are by no means the only factor that is important in predicting the probability of a mortgage default. This has been particularly true during the housing crisis. For example, among the factors that have been important during this time period have been unemployment rates, loan to value (LTV) ratios, housing price trends, and mortgage loan structures. Since these factors change over time, the probability that an individual with a particular credit score will default will also vary at different points in time. While such factors are noted in the discussion, they have not been formally incorporated into the analysis. (See, for example, the discussion of home purchase loan performance in 2006 vs. home purchase loan performance in 2008, below.) In turn, the data provided here on credit scores help to provide a picture of the relevant borrower cohorts but are not meant to be used as a sole predictor of default probabilities.

We used data from LPS to calculate mean borrower credit scores for all FHA first-lien mortgage originations for each year from 2005 through 2009 and, for comparison purposes, for prime and subprime originations as well; all means presented in this section were calculated using data on both purchase and refinance loans. Chart 2 shows mean credit score for each cohort by loan type (prime, subprime, and FHA) for the nation as a whole.11 As would be expected, both FHA and subprime borrowers have mean credit scores substantially lower than those borrowers receiving prime loans in all years. However, the credit scores of borrowers in the FHA program are higher than those of subprime borrowers in all years, including years in which the share of subprime loans was nontrivial.

Between 2005 and 2007, mean credit scores were relatively flat for prime loans and fell for both FHA and subprime loans, both nationally and in the Third District states (Table 3). After the advent of the housing crisis in 2007, the mean credit scores of borrowers for all loan types increased, again for both the nation and the Third District states. In part, this increase reflects an overall tightening of lending standards due to the housing crisis. Evidence that lending standards were tightening comes from a time series from Haver Analytics,12 based on data collected by the Federal Reserve Board, on the extent to which loan officers at mortgage lending institutions report whether they are easing or tightening credit standards for home mortgages (Chart 3). For this purpose, Haver calculates an index,13 which is the difference between the share of respondents tightening credit and those easing standards; its value was particularly high in 2008.14

By 2009 the mean borrower credit scores for FHA originations were in prime territory for the nation as a whole and in the Third District states. As credit standards have tightened with the onset of the housing crisis, the FHA is likely financing borrowers who might once have been able to obtain financing in the prime market. This may explain why the gap between FHA cohorts’ credit scores and credit scores for prime loan cohorts narrowed from 2007 through 2009. In addition, data from LPS suggest that lenders may have been reluctant to make loans to borrowers with credit scores below 620 in 2009, further contributing to the rise in average credit scores for this FHA cohort and to the narrowing of the gap between credit scores for prime borrowers and FHA borrowers in this year.

While all of the Third District states exhibit the same basic pattern for mean credit scores over time, there are nonetheless differences among the states. For example, the mean credit score for FHA borrowers in Pennsylvania is lower than the corresponding national mean in only one of the five FHA cohorts examined and then only by a small amount. In contrast, in New Jersey, the mean credit score for FHA borrowers is below the corresponding national mean for all five of the cohorts, and in four of the five cohorts, this difference is substantial (Table 3).

  • 11 In evaluating the figures in this chart, it should be noted that LPS data in general provide considerably less coverage of the subprime sector than of the prime sector and FHA. In addition, in 2008 and 2009, the actual number of subprime loans in the database is very small. For example, the LPS database contains fewer than 500 subprime loans for 2008 and 2009 in the three states, and nationally, in LPS, there were fewer than 7,000 subprime loans in 2008 and fewer than 1,000 subprime loans in 2009.
  • 12 Haver Analytics, a New York-based provider of time series data for the economic research community, maintains more than 200 databases from government and private sources.
  • 13 More specifically, Haver bases its index on a question asked in the Federal Reserve Senior Loan Officer Opinion Survey. Prior to the second quarter of 2007, the Federal Reserve Board of Governors calculated the index. After that date, the Board discontinued the series. Haver picked up construction of the index, using survey data that the Board continues to report and a methodology similar to the Board’s. Haver used earlier Board calculations to calibrate its own estimates.
  • 14 From 2006 through 2010, more financial institutions indicated that they were tightening standards than easing them.
  • Last updated: June 3, 2011