RDP 2013-05: Liquidity Shocks and the US Housing Credit Crisis of 2007–2008 Appendix D: The Measurement of Subprime Mortgage Lending

There are two problems with using the share of high-priced loans as an indicator of risky (or subprime) lending. First, it may be biased as the share of high-priced mortgages can change over time due to changes in the yield curve rather than changes in bank lending policies (Mayer and Pence 2008). The HMDA does not collect information directly on the interest rate spread, but rather estimates the spread from information it collects on the interest rate of each loan. For example, to calculate the interest rate spread on an adjustable-rate mortgage (ARM) with a contract maturity of 30 years, the HMDA uses the interest rate on a 30-year Treasury bond even though the interest rate on the loan may actually be priced off a shorter-term security. In other words, the maturity of the loan is assumed to correspond to the maturity of the loan contract, not the expected maturity of the loan (which is more likely to be used by the lender). As short-term rates are generally lower than long-term rates, subprime ARMs are likely to be under-reported in the data (because there will be fewer loans reported with a sufficiently large interest rate spread).

Second, the extent of this bias shifts over time as the slope of the yield curve changes. For instance, if longer-term rates fall relative to short-term rates (i.e. the yield curve becomes flatter), the measured share of subprime ARMs will rise as a result of this bias. It is estimated that at least 13 per cent of the increase in the number of higher-priced loans in the HMDA data between 2004 and 2005 was attributable to a flattening of the yield curve (Avery, Brevoort and Canner 2007).