Alternative Mortgage Instruments / The Adjustable-Rate Mortgage

A mortgage with an adjustable interest rate has a substantially shorter duration than an SFPM. For a mortgage of a given maturity, the decline in duration is a function of the ability of the contract rate to adjust to new market rates. At the extreme, an adjustable mortgage whose contract rate change matches every change in market rates would always have a duration of zero. This is because the mortgage would always be worth the principal due and there would be no change in the value of the mortgage due to changes in market interest rates.

A more realistic situation is that the interest rate would change at intervals ranging from six months (ARMs) to five years (RRMs) with no restrictions on the magnitude of change. In these cases, the duration of the mortgage would range from slightly less than six months to somewhat more than four years. In other words, the process of adjusting the contract rate to market rates is analogous to the mortgage being paid off and the lender immediately relending the money. Since the original VRMs had relatively tight restrictions on the total change allowed, they did not substantially lower the durations of the mortgages.

In view of the radical difference in the duration of an SFPM and the more recently authorized adjustable mortgages, it would seem that these latter instruments virtually eliminate the interest rate risk exposure of SLAs. In fact, that is the case. The massive losses that SLAs are sustaining today accrue from the old, low-rate SFPMs in their mortgage portfolios. They are suffering no losses and bear little interest rate risk exposure from the more recent RRMs and AMLs in their portfolios.

It should be pointed out that the use of adjustable mortgages does not eliminate the interest rate risk exposure; it transfers that risk to the borrower. This transfer is a touchy political issue and represents one of the reasons that the regulatory authorities have taken so long to authorize adjustable mortgages. In fact, political displeasure with the adjustables continues to be voiced, as evidenced by recent testimony before the Subcommittee on Financial Institutions Supervision, Regulation and Insurance.

Yet, there is some evidence that all borrowers are not necessarily worse off with adjustable mortgages. Two key factors appear to affect the relative riskiness. One is the source of change in market interest rates, and the other is the composition of the borrower's portfolio. If the market interest rate changes due to a change in the real interest rate, then an adjustable mortgage is unquestionably riskier for the borrower. If, however, market interest rates change due to changes in the rate of inflation, then many borrowers will find the adjustable mortgage less risky than the SFPM. These would be the borrowers whose portfolios contain large portions of "variable coupon assets, common stocks, real assets, and fixed coupon liabilities.''

The initial coupon rate on an adjustable mortgage should be less than that on an SFPM that has an equal maturity. The reason is that an SFPM is equivalent to issuing an adjustable mortgage and simultaneously writing a put option on the level of bond prices. [A put option is an agreement whereby the buyer of the option has the privilege of selling a specified asset to the seller of the option at an agreed upon price prior to an agreed upon expiration date.] Since the put option has a positive value, the adjustable mortgage would have to be worth less than the SFPM. This reduction in value is achieved with a lower initial coupon rate. There are potential exceptions to the relative pricing of SFPMs and adjustable mortgages. For example, if everyone were convinced that interest rates were about to experience a substantial decline such that holders of SFPMs could profitably refinance even after paying the refinancing costs, then an adjustable mortgage might be priced slightly above an SFPM.

The selection of an index for adjustable mortgages can be quite a problem. The two basic choices are a cost of funds index and a mortgage rate index. If the lender specifies a cost of funds index, then he creates arbitrage opportunities for the borrower. For example, if the cost of funds index rises and the rate on new mortgages does not, then the borrower may find it profitable to refinance the mortgage. If the cost of funds index drops and the rate on new mortgages does not, then the borrower will likely continue to hold the mortgage and the lender suffers an opportunity loss. If the lender selects a mortgage index, then he still suffers from a form of interest rate risk exposure. For example, if the index drops while cost of funds does not, then the lender will have a squeeze on earnings. If the index rises and his cost of funds does not, then the lender will have an unexpected increase in profits.

Although these problems associated with the selection of an index may make it sound like the variable-rate mortgages are not really transferring interest rate risk, that is not the case. Interest rate risk, as we have described it so far, has been restricted to changes in the level of interest rates. The problems associated with the selection of an index are actually part of the risk associated with a "twist" in interest rates. A twist is defined as moves in opposite directions of long-term (mortgage) rates and short-term (deposit) rates, which leave average rates unchanged. Neither the SFPM nor any of the AMIs deal with the risk of interest rate twists.

One potential political problem with the use of a cost of funds index has been suggested by the Director of Economic Research at the FHLB of Boston, who fears that such practice may invite profit margin regulation. If the index chosen is highly or perfectly correlated with the lender's own cost of funds, then such an index generates a guaranteed and highly visible profit margin, which could then be easily regulated. Despite this political issue, the same writer notes that cost of funds indices tend to immunize management from errors in their decisions about deposits and deposit rate offerings. This risk of setting inappropriate deposits rates then gets transferred to the mortgage holders. The suggested implication is that regions such as the West Coast which have shortages of loanable funds could afford to incorporate cost of funds indices, and regions such as New England which have surpluses of loanable funds would have to use mortgage rate indices.

These arguments overlook several economic issues. The first is that as long as the prepayment privilege exists, no one SLA can transfer all of its risk of increases in its cost of funds to borrowers if there have not been corresponding increases in rates on new mortgages. The second is that SLAs have never been allowed to use their own cost of funds as an index. Thus, even if the managers at an SLA pay unnecessarily high rates for its funds, the cost of funds index would not necessarily change. Finally, it is not clear why the interest rates on mortgages would not fully reflect regional differences in the availability of funds, and thus why contract terms could be used to adjust for these differences.

The effect of adjustable mortgages on the incomes of SLAs would depend on the terms of the mortgage. Some simulation evidence suggests that the income effects would be beneficial but not dramatic. In one such simulation, Joseph McKenzie constructed a hypothetical mortgage portfolio for the SLA industry at the end of 1978. The parameters for the portfolio were based on "actual aggregate lending volumes, rates, and terms from 1954 through 1978, along with an assumed constant payment factor of 9 percent per year.'' The reasonableness of the simulation parameters can be judged by the fact that the simulation portfolio had a year-end 1978 average yield of 8.60 percent and an outstanding stock of $453.2 billion, compared to an actual yield of 8.54 percent and an actual stock of $432.9 billion. McKenzie found that the highest attainable portfolio yield at the end of 1978 would have been 9.34 percent. This could have been attained if all of the mortgages acquired since 1954 had been VRMs which used as an index the rates on new mortgages.

McKenzie then ran several simulations in which he varied the terms of the rollover mortgage with respect to the maximum allowable change each period and the maximum cumulative change. All of the VRMs had four-year adjustment periods. He also allowed for variations in the proportion of new mortgages that were made as VRMS. He found that under most of the simulations, the average mortgage portfolio yield would be only 10 to 40 basis points higher. Income effects were small because over the period in question, newly issued mortgages grew at a fairly high rate from year to year, especially in the last few years of the simulation. Thus, as McKenzie points out, "an estimated 58.8 percent of the mortgage balances outstanding at the end of 1978 were less than three years old and an estimated 67.5 percent were less than four years old.'' He concludes that if loan closings are able to grow at a rate of roughly 10 percent per year, then the flow of new mortgages will always be large relative to the stock of old ones. This means that adjustable mortgages with adjustment periods of at least four years will have little effect on the current yield of the mortgage portfolio.

Source: The Savings and Loan Industry: Current Problems and Possible Solutions