Prepayments on fixed-rate mortgage-backed securities

Scott F . Richard and Richard Roll M ortgage-backed securities (MBSs) are increasingly a part of the financial scene. To assess their relative value and the value of their derivatives, it is essential to predict mortgage prepayments accurately. It is important to understand how mortgage pass-through securities will prepay in today’s interest rate environment as well as how prepayments will fluctuate as interest rates fluctuate. This paper introduces our latest work on prepayment modeling. Recognizing that work in this area is never completed, we feel nevertheless that our model includes certain innovations and attributes that extend the understanding of mortgage prepayments. We begin by analyzing the economic theory underlying a homeowner’s decision to prepay a mortgage. This option-theoretic analysis serves as the basis for our empirical model of prepayment rates. Three aspects of our model are novel. First, we measure the mortgagor’s refinancing incentive as the ratio of the mortgage coupon to the current refinancing rate, not as the difference between these two rates. This idea comes from our economic analysis of the mortgagor’s prepayment decision. Second, we show that the seasoning process for mortgages depends importantly on this same ratio, the coupon relative to the refinancing rate. In particular, premium mortgages season more rapidly than current coupon mortgages, which, in turn, season more rapidly than discount mortgages. Finally, we examine the tendency of premium mortgages to slow or “burn out” over time. We introduce a measure of premium burnout that depends on the entire interest rate history since the mortgage was issued. We try to provide sufficient details so that readers can understand how the model works. Examples in the figures explain the Goldman Sachs prepayment model, without bogging readers down with unnecessary mathematical detail. We also report a detailed summary of our model’s predictions in relation to actual prepayment data.