New analysis by Tobias Peter, research analyst at AEI’s International Center on Housing Risk (ICHR), uncovered serious flaws with respect to the Federal Reserve’s quarterly Senior Loan Officer Opinion Survey on Bank Lending Practices when compared to ICHR’s much more comprehensive and timely National Mortgage Risk Index (NMRI).
Earlier this week the Fed released the closely-watched quarterly Senior Loan Officer Opinion Survey on Bank Lending Practices (SLOOS). The survey of loan officers is widely viewed as providing a key signpost for trends in mortgage lending standards in the United States. Unfortunately, the information provided by the survey has always been limited at best, and useless at worst. Limited because it only reports on the results based on about 60 loan officers. But even if these were representative for the fifty percent of mortgage lending originated by banks, it ignores the other half, consisting largely of much riskier originations by nonbanks. And useless because it showed no systematic loosening in mortgage lending standards in the run-up to the 2007/08 financial crisis, which runs contrary to everything we now know.
Fortunately, policymakers are no longer forced to rely on the SLOOS. A much better measure of mortgage lending standards now exists. With the creation of theNational Mortgage Risk Index (NMRI), published each month by AEI’s International Center on Housing Risk (ICHR) and covering an estimated 78 percent of all home purchase loans and 90 percent of all primary owner-occupied home purchase loans, we are now able to quantify the risk in mortgage lending and provide accurate, timely, and in-depth tracking of trends in lending standards.
A quick comparison of the two measures clearly shows why the NMRI is superior to the SLOOS. First, and most important, the NMRI is based on hard data – millions of loan records – rather than opinions. Second, as noted above, the loan officers included in the SLOOS are all from commercial banks. Third, the SLOOS weights all survey responses equally, rather than by their share of originations. Wells Fargo alone accounts for nearly 15 percent of purchase loan originations market wide, yet it receives the same weight in the survey as banks that are barely on the radar screen. The NMRI, in contrast, includes the loans with a government guarantee originated by all lenders, and each lender is properly weighted by its share of originations. In addition, because the NMRI is based on loan-level data, it can focus on mortgages used to purchase primary owner-occupied homes, the type of lending toward which the Federal government’s housing policies are aimed. The SLOOS doesn’t distinguish between such loans and those that are used to buy second homes or finance investor purchases. Finally, the NMRI allows for separate tracking of first-time and repeat buyers, a key metric in any analysis of mortgage lending trends.
The SLOOS shows that mortgage lending standards have loosened on net over the past year. This is the right signal, but the SLOOS arrives there by accident, rather than by design. In a head-to-head comparison with the NMRI for only bank-originated loans, the results do not line up. Where the SLOOS shows a loosening of bank lending standards over the past year for loans guaranteed by Fannie Mae and Freddie Mac, the NMRI, with its nearly complete census of such loans, shows little change in standards. Roughly the same discrepancy holds for bank-originated loans backed by Ginnie Mae (principally FHA and VA loans).
By focusing solely on banks, the Fed’s SLOOS misses two crucial mix shifts underway in mortgage lending that are responsible for the easing in standards that is clearly documented in the composite NMRI, which covers the entire government-guaranteed market. The first shift is that large banks have ceded substantial market share to higher-risk nonbank lenders. This has been largely due to concern over the risks associated with government guaranteed lending (particularly FHA’s extremely expansive credit standards), past and future legal liability, reputational risk, and greater capital requirements. The second shift is from loans guaranteed by Fannie and Freddie toward higher-risk FHA loans, which resulted from HUD’s reduction in FHA mortgage insurance premiums earlier this year.
While the SLOOS can shed some light on mortgages without a government guarantee, which are not currently covered by the NMRI, the same flaws apply. Instead of basing evaluation of lending standards on a small survey of bankers that will send the correct signal only by accident or even worse, the wrong signal, policymakers and the public should direct their attention to an index grounded in facts, not opinions. Next year, the ICHR plans to add loans without a government guarantee to the NMRI.