Individual borrowers control the fate of their loans, and sometimes make decisions that on their face seem inefficient. For example, a borrower with a low mortgage rate may decide to refinance into a larger loan (i.e., take a cash-out refinancing) with a higher rate if they either need the proceeds for personal reasons (e.g., to pay off credit-card debt) or fund expenditures. Alternatively, borrowers with relatively high rates may not avail themselves of refinancing opportunities due to lack of knowledge, limited equity, or poor credit. Furthermore, they may decide that the monthly savings on their loans do not offset the upfront costs and aggravation of taking out a new loan.
A segment of the MBS market called “specified pool trading” attempts to take advantage of refinancing inefficiencies by identifying pools or categories of loans that are expected to experience either lower refinancing rates (i.e., slower prepayment speeds) or, depending on market prices, faster turnover speeds. In many cases, lenders are rewarded for segregating and pooling loans with attractive prepayment characteristics separately, with investors willing to buy such pools at a premium or “pay-up” to TBA prices.
Originally, specified pool trading involved sifting through large numbers of pools and trying to identify securities with favorable speed histories. In the late 1990s, however, lenders began to create pools comprised of loans with favorable attributes, leading to the growth of the so-called custom pool market. That market has evolved to the point where a number of standardized attributes are traded at a pay-up to TBA prices. The most popular categories are:
- Maximum (“Max”) Balance pools. These pools are created by limiting the size of the largest loan in the pool. The reason that they are categorized by maximum (and not average or weighted average) balance is that the dispersion can have a big impact on prepayment performance. As a simple example, a pool where all loans have a balance of $125K will behave much differently than a pool with a weighted average balance of $125k but a few very large loans tucked in among the smaller loans. Standard stratifications include max balances of $85K, $110K, $125K, $150K, and $175K.
- High LTV pools. Loans with high LTVs have historically prepaid relatively slowly, although their speeds can increase during periods of rapid home price appreciation. Standard stratifications include loans with LTVs of 80-90%, 90-95%, 95-105%, 105-125%, and greater than $125%. (Note that pools backed by loans with LTVs greater than 105% cannot be delivered into TBA transactions and are pooled separately into programs with different prefixes.)
- Investor pools. Loans made to investors are generally less responsive to refinancing incentives than so-called owner occupied loans. Investor loans are both underwritten differently and often are associated with higher rates and/or additional fees.
- “Geography” pools. New York has long imposed a recording tax on new loan transactions which acts as an incremental cost to a refi and thus suppresses the prepayment speeds of loans originated in the state. Loans issued in Puerto Rico also tend to prepay very slowly; the territory’s current financial difficulties have served to exacerbate this effect.
- FICO pools. As with LTV pools, loans to borrowers with weak credit tend to prepay slowly. As these borrowers pay significantly high rates and fees for loans, borrowers’ abilities to achieve significant savings on their monthly payments are limited. The threshold for “low FICO” pools is currently 700.
There are also loan categories that are expected to prepay very fast. Chief among these are “super-conforming” or “jumbo-conforming” loans, i.e., loans issued in high-cost areas where the regular conforming-balance limit (currently $417K) is overridden. These loans tend to exhibit very fast prepayments, and their speeds are very sensitive to relatively small refinancing incentives; the fixed refinancing costs are easily offset by the monthly savings on the large loan balances. Deliverable pools can only be backed by a maximum of 10% of super-conforming loans, with the remainder pooled and traded separately at a concession to TBAs.
Prepayment Profiles and the S-Curve
It’s important to understand that “favorable” prepayment performance does not necessarily translate to “slow” speeds. For example, holding pools or securities that consistently prepay slowly would serve as a major drag on performance when rates rise significantly and the MBS in question are trading at a discount to par. This is especially true for pools that cannot be delivered into TBAs, such as those backed by loans with LTVs greater than 105%, as discussed below. A more useful way to consider prepayment performance is based on the overall prepayment profile of either a single security or a cohort. This involves measuring and/or predicting the prepayment speeds of the assets in question in different interest rate environments, i.e., scenarios where rates are assumed to both decline and rise. The best way to gauge a profile is through a charting technique called an “S-curve,” which shows prepayments on the vertical axis and “rates” (or, more accurately, refinancing incentives) on the horizontal. The accompanying illustration shows hypothetical “steep” and “flat” prepayment S-curves. The flat curve is distinguished by relatively fast out-of-the-money speeds (i.e., in negative incentive scenarios) and relatively slow in-the-money speeds where the note rates of existing loans are higher than the prevailing level of rates. By contrast, the steep S-curve shows slow speeds in rising rate scenarios and very fast speeds when the assets have a refinancing incentive. Given the impact of prepayments on returns and risks discussed in previous reports, the flatter profile is clearly more desirable. Therefore, while investors in the current low rate regime typically seek loan attributes that mute prepayment speeds, managers seeking enhanced performance in a variety of rate scenarios will try to identify loans that have the flattest possible profile and, more to the point, exhibit relatively fast speeds in the absence of a refinancing incentive.