The recent drop in market yields has led to a discussion on the "true" level of consumer mortgages rates, i.e., a realistic and consistent estimate for the level of mortgage rates at the consumer level. This rate is clearly important, whether it's being estimated from secondary market rates or is being measured empirically (i.e., through a survey such as the Freddie Mac Primary Mortgage Market Survey rate). Unfortunately, no single rate exists, even among individual products. As we've noted in the past, rates themselves are not calculated directly. Rather, the all-in execution (which accounts for all costs, including hedging and margins) at each rate point is calculated, and the difference between the all-in execution and par are the points associated with that rate. For example, if the all-in execution for a loan with a 6.375% rate is 99.5, the 6.375% note rate has 1/2 point associated with it. Therefore, at any given time, the "mortgage rate" is really a matrix of points and rates, as demonstrated in Exhibit 1.
Interestingly enough, however, the relationship between rates and points is neither linear nor constant. Negative points (i.e., note rates where a rebate is paid to the borrower) are associated with above-market rates and are used to price "no-cost" loans. Because of the difference in how servicing at different rate levels is valued, the relationship of rates to points is significantly different for rates with negative and positive points. Essentially, lenders become wary about putting up much up-front money in exchange for a higher rate, given the ability of the borrower to prepay the loan and extinguish the servicing strip.
Exhibit 2, which graphs the data from Exhibit 1, attempts to illustrate the difference in the rate/point relationship for discount and premium note rates. Note that the slope of the line changes at the zero-point mark, i.e., the rate is more sensitive to changes in points when points are in negative territory than when they are positive. In addition, the relationship is not constant, because the all-in execution is strongly dependent on servicing valuations. Note that the complexity of the rate/point relationship is glossed over in surveys such as the Freddie Mac PMMS rate, which basically averages one rate/point combination from each lender.
These factors have a number of implications for the mortgage market. At the consumer level, areas in the rate/price relationship can be classified as distinct market segments. Different lenders can be stronger in one segment than another, depending on both their willingness to retain servicing as well as their aggressiveness in valuing it. For example, lenders that are more willing to acquire servicing would be able to offer relatively attractive no-cost loans. Furthermore, from the perspective of prepayment modeling, extrapolating the rate on a no-cost loan from a survey rate is subject to differences in the rate/point relationship, both between lenders and over time.
An issue that consistently raises concerns in high-volume environments is capacity. As application and funding volumes have soared over a very short time frame, constraints on capacity are once again a problem. Using the Mortgage Banker Association's Composite applications index as a proxy, the index value of 1059 reported for the week ending 8/30/02 (close to the all-time high of 1127, set the week of 8/16) was roughly twice the average of the index for the first six months of this year. Clearly, any industry would be stretched at having to deal with a doubling in volume in the space of a few months, especially since capacity constraints were just beginning to ease this spring. The lack of slack capacity, in turn, tends to push rates higher; lenders generally don't have an incentive to be highly competitive in terms of rates if they have all the business they can handle. In addition, however, the need to extend commitment periods to deal with the volumes has direct and indirect costs.
An obvious cost of extending the commitment period is the cost of hedging. We estimate that every 15 days beyond a 45-day commitment costs between 4 to 7 ticks for a 30-year loan. This is fairly high historically, in part due to the high levels of market volatilities. In addition, having to carry loans past the monthly notification date costs the lender the incremental value of the roll, which in turn is reflected in the rate offered to the consumer. For example, imagine a case where the lender must extend the commitment period from 45 to 60 days and, by doing so, has to pool the loans in question for October rather than September settlement. The estimated costs could be as follows:
* Hedging cost - 6/32s (or .188% of the face value)
* Roll value (i.e., how much the roll is trading over carry-3/32s (or .09375% of the face value)
* Total cost - 9/32s (or .2813%)
Depending on the points the borrower is willing to pay, the additional cost translates into rates that are between 0.125% and 0.25% higher than they would otherwise have been. This can be estimated from Exhibit 1. For example, a drop in points from .50 to 0 suggests a 12.5 bp. increase in rate, while a drop in points (or, put differently, an increase in the rebate) from 1.625 to 2.0 implies an increase in rate of 25 bps. (Note that this is only an estimate, since both rates and points are generally quoted in even increments. Also, since different rate points use different coupons for their pricing benchmarks, the impact of the extension in time would depend on the price of the roll for that coupon.)
We draw the following conclusions:
* Capacity issues impact pricing in a number of different ways. In addition to normal issues of competitiveness, extending the commitment period can significantly boost rates at the consumer level.
* The costs of extending the commitment period can vary over time. This is due both to changes in the cost of hedging (which would generally be a function of both time and implied volatility levels) and the point in time in the allocation cycle. If, for example, an extension in the commitment period did not extend past the allocation date, the value of the roll does not come into play. If it did extend past notification day, however, the change in rates can be significant.
* Because the sensitivity of rates to changes in points is greater for note rates at a premium (i.e., those rates associated with negative points), extended commitment periods due to capacity constraints make it more difficult and expensive for lenders to offer no-cost loans. (Keep in mind that the costs are priced as a rebate to the borrower.) Also, this relationship can impact the rate value of add-ons, which in turn will impact the pricing of loans with either alternative documentation or other Alt-A characteristics.
