Back-to-Basics: Adjustable Rate Mortgage-Backed Securities

Teri Geske
Senior Vice President, Product Development

In an upcoming version of BondEdge we will release a new ARM/Hybrid ARM prepayment model and new ARM pricing, so we thought it would be a good time to re-visit the subject of ARMs in this "Back to Basics" article (updated from an article originally published in February 2000).

Adjustable rate mortgage pools (ARMs) are an important investment alternative for commercial banks and other depository institutions, as well as property & casualty insurance companies and many short- to intermediate-term funds. ARMs have attributes and price sensitivities that are quite different than those of fixed rate mortgage pools and their risk/return characteristics are not necessarily intuitive. The burgeoning market for hybrid ARMs, those with extended initial reset periods, represents yet another type of investment with its own unique features, and there are an increasing number of CMO deals backed by hybrid ARM collateral. This "Back to Basics" article discusses some of the salient features of the ARM market and offers some insight into the price behavior of these securities.

Probably the most important characteristic of ARMs is that the coupon rate adjustment, which occurs either annually, semi-annually, or monthly, may be subject to periodic and lifetime caps and floors that protect the borrower from dramatic changes in the required loan payment from one period to the next. Hybrid ARMs, which comprise the majority of new ARM issuance, have an initial “fixed” period of 3, 5, 7 or even 10 years, during which time the homeowner’s interest rate (and the hybrid ARM’s coupon rate) does not change, regardless of what the underlying index may do.

When it does adjust, the coupon rate on an ARM is based on a specified index value plus a spread (margin). The most common indices in the ARM market are the One Year Constant Maturity Treasury (1 yr CMT), 6-Month LIBOR and the 11th District Cost of Funds Index (COFI). Unlike the other indices, COFI is not a market interest rate; it is the average interest expense of savings institutions in California, Arizona and Nevada, and reflects the cost to those institutions of maintaining deposits of varying maturities. Since the rate paid for deposits does not adjust until the “old” deposits mature, the COFI index is not immediately affected by changes in market rates. Instead, changes in COFI lag changes observed in other market rates. When rates have peaked and begin to fall, COFI will continue to rise for some time before trending downward, and vice versa. As a result, prices of ARM pools backed by COFI-based loans respond differently to changes in interest rates than other types of ARM pools¹.

To an investor, an ARM may be thought of as a combination of a purely floating rate instrument (whose rate adjusts freely in response to changes in the index value), along with short and long positions in a series of interest rate caps and floors and the prepayment option in the underlying mortgages: ARM Pool = Pure Floater – (Periodic Caps + Lifetime Caps) + Periodic Floors – Prepayment Option. An ARM’s periodic caps and floors are “path-dependent” options; their value depends not upon the absolute level of interest rates at a particular point in time, but upon the path that rates follow. In valuing the security today, we therefore need to forecast the expected future cash flows to the investor, so we need to make some assumptions about how rates might evolve over time², since the coupon payments depend upon the level of interest rates. For example, consider an ARM with a 1.0% annual reset cap, meaning the coupon rate cannot move up or down by more than 1.0% per year, with a coupon formula of 1yr CMT + 150 bps. When we compute the coupon rate to be paid in Year 2, it is not sufficient to simply observe the level of the 1-yr CMT rate at that point in time; we need to know the path interest rates followed up to that point. Consider the pattern of coupon rates that would be paid on this ARM, given the following two interest rate paths:

Even though the values of the 1yr CMT index at T₀ and T₂ are the same under both paths, the coupon rates at T₂ are different, because the reset cap prevents the coupon rate under Path #1 from changing by the full amount of the change in the underlying index.

Since homeowners with adjustable rate mortgages are at least somewhat sheltered from changes in interest rates, ARM securities exhibit prepayment patterns that differ considerably from fixed rate MBS. Unlike fixed rate mortgages, which homeowners tend to refinance fairly quickly and often dramatically when rates fall and abruptly stop refinancing when rates rise, prepayments on ARMs are less predictable. Homeowners with ARMs that are going to reset in the near term do not respond to a refinancing opportunity the way homeowners with fixed rate mortgages do, because the adjustable-rate borrower’s loan is going to automatically reset at a lower rate, greatly reducing or perhaps eliminating any refinancing incentive. Hybrid ARM prepayments display somewhat different patterns. Hybrid ARMs may be thought of as balloon mortgages without a forced refinancing, and hybrid ARM borrowers appear to act more like fixed rate borrowers if the initial reset date is some years in the future, except that hybrid ARM borrowers expect to move sooner than 30yr borrowers, suggesting that we should see a greater amount of turnover (non interest rate-sensitive prepayments) with hybrid ARMs compared to fixed rate mortgages. As the first reset date approaches, hybrid ARMs appear to behave more like traditional ARMs, and once a hybrid ARM’s first reset date has passed, there is virtually no difference in prepayment expectations between that security and a traditional, non-hybrid ARM, except to the extent that the hybrid ARM’s coupon after the initial reset may still be substantially different from current market levels due to the reset cap.

Prepayments affect the interest rate sensitivity (i.e., effective duration and convexity) of ARM securities to the extent that the price on an ARM pool differs from par. If the price of the ARM pool is equal to par, the investor is indifferent as to the speed with which principal is returned at par in the form of prepayments. Estimating an ARM’s effective duration in a “back of the envelope” way is quite difficult because of the interplay of the periodic caps/floors and prepayments. If we assume that a pure floating rate security is always priced at par (assuming no credit risk), as interest rates move an ARM’s price will deviate from par due to the net value of the long and short positions in the embedded caps, floors and prepayment option. For example, as rates decline the lifetime cap becomes less valuable, while the prepayment option and lifetime floor become more valuable; the net effect of these rate changes will cause the security’s price to move by a small amount or by a meaningful percentage. As rates rise, the reset and lifetime caps increase in value, and since the ARM investor is short these options, the price of the security falls; therefore, the effective duration lengthens as rates rise. All other things being equal, the tighter the reset caps (e.g., 1% vs. 2%), the greater the percentage price decline in a rising rate environment. Even if there were no caps on the ARM’s coupon rate, a change in interest rates would affect its price, since the investor has to wait for some period of time until the coupon rate adjusts to reflect the new interest rate environment. Some ARM pools (known as “WAC ARMs” or “Flex Pools”) are backed by loans which reset on different dates throughout the year. For example, 20% of the loans might reset in March, 50% in July and 30% in November. This collateral reset or “roll” schedule affects the value of the security, as a change in rates will affect the coupon rate paid on some of the mortgages sooner than on others. The effective duration also reflects the time to the next coupon reset date.

ARM prices are often quoted using the concept of “Effective Margin”. This spread is the difference between the ARM’s yield-to-maturity (assuming a given CPR%) and the value of the underlying index off which the ARM’s coupon resets. For example, if a 1-yr CMT-based ARM has a yield of 3.50% and the current 1-yr. CMT rate is 1.50%, the Effective Margin is 200 bps. This is similar to the concept of a Discount Margin (DM) used to value other floating rate securities (such as CMO floaters); the difference is, Effective Margin assumes semi-annual compounding, while a DM is based on monthly compounding. The notion of an Effective Margin is easy to understand, and appeals to depository institutions as a measure of the income derived from the investing in the ARM versus the cost of liabilities which are based on the same underlying index. However, Effective Margin has a number of drawbacks as a valuation measure. It assumes prepayments are fixed and known, and that interest rates will remain unchanged. In other words, it ignores the full value (the time value) of the caps and floors, and of the prepayment option. Also, it can only be used as a comparative measure for ARMs which are all based on the same underlying index. Option-Adjusted Spread is a more robust valuation tool than Effective Margin, because it incorporates the impact of future interest rate volatility on the ARM’s caps, floors and prepayment option, which all affect the amount and timing of cashflows to the investor.

We hope this brief review of ARMs has been useful. If there are other topics you would like us to cover in a "Back-to-Basics article", we’d like to hear from you. Please contact marketing at fia.marketing@interactivedata.com or by phone at (310) 479-9715, with your suggestions.

______________________________________

¹In BondEdge, effective duration calculations and return simulations for COFI ARMs reflect the lagging nature of this index.
²Even if we assume there will be no change in interest rates from current levels, that itself is a rate forecast. In practice, a Monte Carlo process that generates a number of possible future rate paths is used. For more information about Monte Carlo valuations, see "Back-to-Basics": Why Do We Use Monte Carlo Simulations?, available on the BondEdge Private Client Website or at www.bondedge.com