A model estimates the impact of economic variables on the pricing of prepayment risk
The ability of homeowners to prepay their mortgages presents a tricky pricing challenge in the $8 trillion agency mortgage-backed securities (MBS) market. While the backing of Fannie Mae, Freddie Mac and Ginnie Mae effectively removes any credit risk, calculating a prepayment risk factor is central to properly valuing MBS, but not exactly easy.
Prepayment risk is essentially the risk that the mortgage-backed security buyer will receive, say, seven years of interest income at an agreed-upon rate, on top of principal repayment, instead of 10 years of such interest. Prepayment forces the buyer to reinvest the principal, often at a lower rate of return.
Changes in interest rates certainly account for most prepayment risk, as anyone who has ever refinanced into a lower-rate mortgage can understand. But interest rate changes alone can't fully predict prepayment activity. A recent case-in-point is the hordes of homeowners who would have loved to refinance into rock-bottom mortgage rates manufactured by the Federal Reserve's Quantitative Easing push that nearly halved the 30-year fixed mortgage rate between 2007 and 2016, but were shut out amid falling home values and incomes.
In a paper published in the Review of Financial Studies, UCLA Anderson's Mikhail Chernov, Brett R. Dunn, a Ph.D. student, and Francis Longstaff developed a model based on a cross section of all MBS to explore the market's pricing-in of prepayment risk, and what factors beyond interest rates are responsible for the prepayment risk premium.
Using MBS data from 1998 to September 2014, the researchers' model computed an average implied prepayment rate of 25 percent, which is significantly higher than the actual average prepayment rate of 21 percent during that stretch. That four-percentage-point gap represents a significant prepayment risk premium with real-world pricing implications. The higher implied prepayment rate would cause mortgages with below-market rates to be priced higher than the historical prepayment rate would suggest, and vice versa for mortgages with above-market rates.
Chernov, Dunn and Longstaff then dive into why there is a significant gap between implied and actual prepayment rates.
They estimate that more than 90 percent of the prepayment risk premium is driven by the market's taking into consideration (pricing in) macroeconomic factors such as unemployment rates and changing home valuation trends. For instance, in rough economic times, rising unemployment can lead to homeowners' moving or being foreclosed upon, or choosing to default on an underwater mortgage. In robust economic times when home prices are rising, homeowners may opt for a cash-out refinance that puts money in their pocket. Both the good-news and bad-news economic trends have the same prepayment end result for an MBS investor, and thus are measured together as the "turnover rate."
The turnover rate explains the bulk of the gap between the implied and actual prepayment rates. The authors note that refinancing because of changes in interest rates is "of first order importance" in sussing out prepayment risk, given its large-volume footprint. In their model, they find that nearly 70 percent of prepayments were a result of "rate response"-related prepayments.
Over the entire study period, there was little difference between the implied and actual rate response levels. The authors note that the 2001–2005 period may skew the results, as rapidly rising home prices and lax lending standards triggered a sharp increase in cash-out refinancings.