For most of us in the MBS markets, the process of mortgage securitization is assumed to begin with the mortgage rate as the independent variable. Rarely does our thought process focus on the fact that issues and developments in the MBS markets may lead to the determination of the mortgage rate as a dependent variable of capital market dynamics. This paper will examine the links between the rates consumers are charged for residential mortgage loans and the role of capital markets instruments, and demonstrate that capital market dynamics heavily influence the rate generation process.

Determination of mortgage rates

With respect to conforming loans, originators have three basic options in selling such production. The most important mechanism for monetizing production is through the creation of MBSs. This methodology, for loans underwritten to the GSEs' standards, involves paying the agency a fee to guarantee the loans and liquefying the loans as tradable securities. A second option is cash execution (sometimes referred to as the "cash window"), where loans are sold directly to the GSE. The usefulness of this option is generally limited to either smaller originators (who do not have the infrastructure to price, create, and deliver securities), or products which are not TBA-eligible. The third option for an originator is to sell their production in whole-loan form. In the non-conforming market, by comparison, the originator has only two options; the production can either be sold in whole loan form or securitized as a private label execution. This discussion will focus on the generation of conforming loan rates.

While the specifics of the markets for the different loans may vary, there are several common steps involved in using capital markets information to determine mortgage lending rates. These steps may be broadly defined as: calculating the optimal security coupon by evaluating the components of a loan; using the optimal coupon execution to determine the points associated with the given note rate; creating a rate/point grid for the product.

A key point to remember is that the dependent variable is points, not rates. As we will demonstrate, any reasonable rate is available to a borrower; the question becomes that of the points charged (or rebated, for an above-market rate) to the borrower.

Determining the

security coupon

From the point of view of the rate-setting process, the determination of the optimal security coupon is arguably the most important step. The key elements in the process are:

1) MBS TBA prices; 2) the agencies' guaranty fees for a program;

3) guaranty fee buy-up and buy-down multiples; and 4) valuation of excess servicing.

In order for each loan to be securitized, a guaranty fee (referred to as a "g-fee") is paid to the agency pooling the loan. The amount of the guaranty fee represents the cost of credit insurance that the agency charges the originator to monetize the loan into a component part of a tradable security. Note that guaranty fees can be capitalized and bought or sold by the originator on a present value basis; the process is similar to trading IOs, in that the prices are quoted in terms of multiples. Servicing values are also calculated on a present value basis, after making prepayment and return assumptions; the pricing of this asset involves the valuation of factors such as tax and insurance float, principal and interest float and the value of cross selling opportunities associated with servicing the loan. As a result, there can be variations between Trust IO multiples, base and excess servicing multiples, and multiples implied by the pricing of buy-up and buy-down fees.

The decision into which MBS coupon the mortgage is pooled into is heavily determined by the trade-off between 1) the value of excess servicing, 2) the capitalized value of guaranty fees, as reflected by the value of buy-ups and buy-downs, and 3) the value of incremental interest income in the passthrough market, as reflected by inter-coupon swaps. Exhibit 1 illustrates the difference in execution for a hypothetical mortgage originated with a 7.75% note rate and securitized either as a 7% or a 7.5% pass-through MBS.

The differences in valuation between the two potential coupons come from netting the greater proceeds realized by selling the loan into the 7.5% coupon passthrough with both the cost of buying down the guarantee fee and the value of excess servicing which is not retained. One of the pricing dynamics that skews the pooling into a 7% MBS is, in this example, due the fact that the pricing of the MBS coupon stack reflects weakness in inter-coupon spreads, relative to the value of servicing. During periods of weakness in inter-coupon swaps, issuers may find it optimal to hold servicing, rather than sell it in the form of coupon.

Once the optimal execution coupon has been determined for each note rate, the actual calculation of a rate/point grid is fairly straightforward. In fact, the notion of "calculating rates" is actually fallacious. Any reasonable rate is available to borrowers; the issue is that of the appropriate number of points to charge for each rate (or, in the case of an above-market rate, how many points should be rebated to the borrower). Exhibit 2 shows a calculation of points for two different note rates. For the purposes of the analysis, items like profit margins and hedging costs are assumed to be the same. In practice, hedging costs will vary depending on the coupon, and margins will vary over time depending on coupons and programs. This analysis forms the basis for periodic computations of a matrix of rates and points for the various programs offered by the originator. The calculations require that the optimal execution for each possible note rate is determined, and then the associated points for each rate is computed to populate the rate/point grid. (Note that in some cases, the amount of a potential rebate is capped.)

While the above discussion has served to highlight the role of capital market dynamics in the pricing of generic rates, there are several factors which, while subject to competitive considerations, are determined at the local level to reflect the idiosyncratic risk of the borrower. Over the last few years, the industry has increasingly shifted to a regime of risk-based pricing. As a result, individual borrowers are quoted rates based on the perception of their individual riskiness, using attributes such as credit scores, loan-to-value ratios, the type of loan, and other obligor and/or property characteristics. The result has been a proliferation of programs that are distinguished by borrower characteristics, limitations, and specifications. Borrowers that apply for loans are directed to optimal programs, which provide the lowest applicable rate for each borrower given their credit profile.

The general assignment of a borrower to a program is then augmented by so-called "add-ons." Add-ons are additional points added to the cost of a loan based on certain designated credit characteristics, such as higher LTVs, reduced documentation, cash-outs, second homes non-standard property types, or some combination of attributes. Add-ons are generally quoted in terms of points. However, many borrowers convert at least some of the points into a higher loan rate. The conversion is fairly straightforward. As discussed previously, a rate/point grid would have been computed for the program in question. Once the total add-ons for the loan are calculated, the current rate/point matrix is used to determine the appropriate rate, given the borrower's desired points. For example, if the appropriate add-ons for a particular customer total 1.25 points, and the borrower wanted a zero-point rate, the lender would refer to a rate/point grid for the program. The rate associated with negative 1.25 points (or, more accurately, a rebate of 1.25 points) would be the borrower's rate.