Secondary marketing done better.

Seat-of-the-pants secondary marketing doesn't make it anymore. Sophisticated techniques for assessing your risk exposure from hedging, now mean there are no more good excuses for simply playing it by ear.

ONE OF THE KEYS TO SUCCESS IN MORTGAGE BANKING is a strong secondary marketing operation. Correct pricing and efficient management of risks is prerequisite to consistent profitability. Success requires mastery of finance, trading, logistics, computer systems, data management and accounting.

Recently there has been much progress in several areas of secondary marketing, such as data collection and the availability and use of real-time market information. However, while the days of running secondary marketing from the back of an envelope may be long gone, most mortgage banks still have a long way to go to meet the major unmet challenge of the secondary marketing department. That challenge is to transform a mountain of raw data into informed analysis that can be quickly turned around in the critical area of interest rate risk management. This article reviews the common obstacles encountered in meeting this challenge, and offer some potential solutions.

This review is divided into six somewhat overlapping areas related to the secondary marketing function: loan tracking, position analysis and reporting, hedge position management, trading analytics, mark-to-market procedures and pricing. Following is a summary of problems typically found in these areas. A detailed discussion follows the summaries.

Loan tracking

Most mortgage banks have in place loan-tracking systems capable of providing the data needed to manage interest rate risk. However, these data are rarely used to the fullest extent possible. In addition, few mortgage banks have adequate controls over the flow of information between their loan-tracking system, trade-management system, exposure management system, delivery/settlement system and market information source.

Position analysis and reporting

Many mortgage bankers have a reasonable understanding of interest rate risk. However, individuals charged with running the secondary marketing function rarely, if ever, have the background, the time or the resources to develop the models and systems needed to optimize the interest rate risk management of a mortgage pipeline.

Systems actually being used for managing interest rate risk are particularly inadequate when it comes to analyzing option or option-like positions. There is considerable confusion about what options are worth (particularly implicit options embedded in rate locks), when options are needed in a hedge position and, if they are needed, how they should be incorporated, and how options should be represented in a position analysis report. As a result of these omissions and limitations, secondary marketing departments, for the most part, are unable to produce brief and accurate summaries of interest rate risk exposure that are understandable by senior management without further explanation.

Hedge position management

Hedge positions are typically arrived at on an ad-hoc basis, through trial and error or rule of thumb. There is particular confusion concerning how to define and measure fallout, and how different fallout patterns affect the cost of hedging and the optimal hedge strategy.

Few, if any, secondary marketing departments have the combined statistical and financial algorithms needed to optimize the placement of hedge coverage between forward programs, between optional and mandatory delivery positions and between settlement months.

The main obstacle in doing so is confusion as to how to integrate rate locks on the exposure side and options on the coverage side, due to uncertainties, respectively, about their closing or being in-the-money on the option expiration date.

Trading analytics

Most secondary marketing operations could benefit from having more sophisticated tools for making trading decisions, particularly in the area of mortgage-backed securities (MBS) and Treasury options. In addition, we find that traders continue to use crude statistical models for determining the relations between and among different MBS types and coupons and Treasury futures contracts. Far more precise results could be obtained by using state-of-the-art option-adjusted spread models, which have been widely favored by sophisticated money managers for several years.

Mark-to-market

Typically, mark-to-market reports are based on subjective fallout estimates, do not properly incorporate option values and are not produced in a timely manner. Yet, it is very important for senior management to get regular and timely feedback to warn them if something is going wrong. There is often fuzziness or arbitrariness in accounting for servicing values being created.

Pricing

Most mortgage bankers continue to use old rules of thumb or play "follow the leader" in pricing. Instead, pricing should be guided by careful profitability analysis performed each day. Rate-lock pricing should be better differentiated according to differences in hedge costs and values of servicing. We often find that mortgage bankers ignore these and other important variables (such as interest spreads from expected closing dates to expected forward settlement dates) in pricing rate locks.

Loan tracking

Loan records must contain information relevant to fallout behavior. To make good secondary marketing decisions, it is essential that current closed loan warehouse and rate-lock inventory data always be available. To develop an accurate profile of the interest rate risk of a given pipeline, rate-lock loan commitments in the pipeline must be categorized according to those attributes that are known to affect fallout behavior. At minimum, those attributes include mortgage purpose (i.e., refinance vs. purchase), source of business (e.g., wholesale vs. retail), geographical location, processing status (e.g., application stage, approved stage) and spread between rate-lock commitment price and underlying mortgage price.

Relocks must be tracked, currently and historically. Another important capability in a loan-tracking system is the ability to monitor renegotiated interest rates or points on rate locks. Many mortgage bankers have tracking systems that make it difficult or impossible to determine whether or not a repricing has occurred.

Renegotiated rates must be tracked not only to control current profitability, but also to sharpen understanding of fallout sensitivity to interest rate movements. This is because a partial or full rate concession is equivalent to a partial or full fallout of a loan from the pipeline.

It is not possible to formulate an efficient hedge position or strategy without a good understanding of how fallout relates to interest rate movements. Effective risk management requires that this relationship be monitored on a continuing basis. Relocked loans should contain an audit trail in the tracking system making it possible to later perform historical analysis.

A loan-tracking system can't take the place of an expert interest rate risk management system. It is a major error to expect your loan-tracking system to be your risk-management system as well. Loan-tracking systems are superb for sorting, searching and matching items. However, they do not contain the financial models required to manage the interest rate risk of the rate-lock agreement or the other instruments that make up the exposure and coverage of a mortgage bank.

Risk-exposure analysis, option trading analytics, hedge position management, true optimization of the forward sale and delivery process, pricing and assigning mark-to-market values to option and option-like positions, are all beyond the ability of these systems. Considering the availability of more-advanced systems, only lack of knowledge or inertia can explain the heavy reliance on these loan-tracking systems for interest rate risk management. For this reason, it is important that it be easy to export data from your loan-tracking system. Otherwise you will be at the mercy of the system for any desired analysis or reports.

Position analysis and reporting

Rate-lock modeling is a science, not an art. The interest rate risk of a mortgage pipeline is difficult for a non-options expert to model, particularly because of the inefficient way in which borrowers exercise their option to "walk away" from a higher-than-market rate loan. This complexity is compounded by the fact that the underlying asset created by the actual takedown of a loan is also a complex financial instrument. Ad-hoc rules and seat-of-the-pants methods for risk management that might be successful under one set of circumstances will not work universally because different types of rate locks and underlying mortgages exhibit significantly different risk characteristics (e.g., fixed- vs. adjustable-rate mortgages, retail vs. wholesale rate locks, short-term rate locks vs. long-term builder commitments). Properly measuring rate-lock fallout levels and elasticities and integrating these assumptions into a valid, option-based risk model is prerequisite to establishing and maintaining optimal hedging strategies.

Fallout behavior should be expressed as a function of interest rate changes; point estimates should not be used. A well-designed risk-management system contains predefined fallout functions (not point estimates) for each category of rate lock. Then, as market movements take place, the system should automatically determine the impact of price changes on risk profiles. A system that requires entry of new fallout projections each time there is a market movement is bound to fail because risk profiles and hedge positions must contemplate, before the fact, how fallout ratios would change if rates were to change.

Fallout can occur even when a loan closes. Fallout occurs when a loan does not close or when a loan closes at a lower rate or fewer discount points in a falling interest rate (rising price) market scenario. In the case of a change in rate or price, a rate lock is said to have fallen out to the extent that the market improvement was passed through to the borrower.

If you don't know your fallout, it is easy to formulate a conservative estimate until you have more data and experience. Assume the highest closing ratio you have ever experienced will occur for any down market and the lowest closing ratio you have ever experienced will occur for any up market. Then begin to relax this assumption as you get more relevant information.

Do not confuse rate-lock delta and closing ratio. (Rate-lock delta is the change in the value of a rate lock given a change in the value of the mortgage underlying the rate lock.) This is by far the most serious error committed by secondary marketing professionals. This error seems to arise from a misplaced concern about having the right amount of coverage on the day the loan closes, instead of a proper concern for hedging profits and losses.

Profits are a function of positions held throughout the life of the hedge, not a function of arriving on the loan closing date with the same amount of coverage as closed loans. It is surprising that so many practitioners believe this approach yields the optimal hedge ratio, considering the fact that it is relatively easy to show that this approach is not equivalent to hedging the profit/loss resulting from changes in underlying mortgage prices.

At the heart of this error is a confounding of the marginal and the average. Holding a short position equal to the closing ratio will work only if this size position is held throughout the life of the rate lock. And this could happen only if the hedger were able to predict what interest rate will exist at the

end of the hedge period, because the actual closing ratio is a function of the ultimate market interest rate.

Consider an example of a sustained period of falling interest rates (rising prices). During such a period, the expected closing ratio will be repeatedly revised downward, reflecting constantly increasing fallout expectations. As a result, the average position held will be larger than the position held at the end of the hedge period.

In this example, a hedger whose practice is to continuously maintain a position equal in size to the current expected closing ratio will inevitably lose more on the hedge position than he or she makes on the loans that close. If, on the other hand, the hedger had calculated on each step of the way the marginal profit/loss from a small additional change in the market price and set the size of the position based on the marginal effect of a change in price on his or her total profit/loss, then the hedger would have been consistently holding a position smaller than the expected closing ratio.

Such an approach would have resulted in cumulative losses that would not exceed profits on the volume of loans that actually closed at the end of the period.

The difference in the two approaches derives from the fact that there are two distinct effects on rate-lock profits when mortgage prices change: the first is the change in profit or loss on rate locks whose closing status is unaffected by the marginal price change; the second comes from the elimination or addition of all previous unrealized profits or losses on rate locks that fall out or close as a result of the marginal price change.

A rate lock, while containing elements of a put option, is not a standard put option. Recognizing that, in principle, a rate-lock commitment is a quasi-put option because borrowers have the right, but not the obligation, to take down the rate-lock loan commitment, some practitioners attempt to solve the option problem by simply assuming that the rate-lock commitment is a standard put option, equivalent in substance to those traded on exchanges or over-the-counter.

The problem with this approach is that standard option models assume efficient exercise (i.e., upon expiration, an in-the-money option will be fully exercised and an out-of-the-money option will not be exercised at all). It follows that a standard put-option model will accurately evaluate the delta of a rate lock only where fallout behavior is such that all rate locks close for any rise in rates and all rate locks fall out for any drop in rates.

The Black Scholes model was designed in the 1970s to evaluate options on stocks. It was not designed for evaluating MBS options, much less rate locks. In addition to the problem due to the efficiency of exercise factor, using the Black Scholes model to evaluate an option on a mortgage introduces another problem. It arises because the Black Scholes model assumes that the asset on which the option exists has a symmetric return distribution (i.e., percentage increases of a given magnitude are equally likely as percentage decreases of the same magnitude). This is clearly not the case with a mortgage asset.

Because of the embedded prepayment option, a mortgage has a very limited upside compared to its downside. Figure 2 compares the delta of a put option on a mortgage measured using the Black Scholes model versus the delta of the same option measured using a model that uses an accurate assumption with respect to the underlying asset return distribution.

The bias of the Black Scholes model can be characterized as follows. When the put option is out-of-the-money, the Black Scholes model overestimates the delta (implicitly overestimating the likelihood that the option will be in-the money by the expiration date). When the put option is in-the-money, the Black Scholes model underestimates the delta (implicitly underestimating the likelihood that the option will remain in-the-money on the expiration date).

These biases arise from the fact that the Black Scholes model assumptions are not consistent with the compression (falling volatility) in the price of a given mortgage that occurs in rising markets and the decompression (rising volatility) in the price of a given mortgage that occurs in falling markets.

A rate lock is, in fact, a variable-quantity option. The variable-quantity option model differs from a standard option model in that it allows for different degrees of option exercise, depending on the relation between the market price of the underlying asset and the option strike price. Thus, it is ideally suited for evaluating rate locks because it captures the optional nature of rate locks, while at the same time allowing for any degree in exercise efficiency the hedger wishes to assume.

As such, it overcomes the problems inherent in both of the static closing ratio approaches, as well as the standard option models. As a result, it is the only approach that can provide the correct rate-lock delta. It provides the same delta as would be obtained by evaluating a portfolio of put options, call options and mandatory forward delivery positions that collectively provide a profit/loss profile exactly equal but opposite to the profit/loss profile of the rate-lock commitment.

Exposure management reports often cover too narrow a range of interest rate movements. We recommend projecting value changes for all exposure and coverage items for a range of MBS price changes of at least 5 to 10 points up or down (e.g., relating to 100 to 200 basis points up or down in the Treasury yield curve). Moves of these magnitudes have occurred several times in the last decade. There is no reason to believe they will not again in the future.

The impact of out-of-the-money options is often mistakenly left out of position reports. Out-of-the-money options are often not included in basic reports that purport to measure overall longness or shortness of the mortgage banker's position. This is the case even though they may afford a substantial degree of protection or risk, particularly if they are near-the-money, if they have a long remaining life or if bond market implied volatility is high. In a long/short report, option positions should be included on the basis of a delta weighting.

Exposure reports that exclude the current effects of out-of-the-money options may indicate rather substantially long positions when, in fact, actual positions may be net short or vice versa.

Prices of different MBSs do not move together point for point. A very common error committed in the production of consolidated exposure reports is the assumption that all mortgage prices (Fannie Mae, Ginnie Mae, 8 percent, 9 percent) move together point for point. As everyone knows, MBS prices do not behave in this manner.

Some practitioners use regression analysis to establish a relationship; this too is incorrect because the relation between two mortgage assets changes as interest rate levels change. Thus, the relation found in a regression is backward looking and, in any event, will not hold for broad movements in rates (making it useless for a broadly scoped shock analysis).

An option-adjusted spread model is needed to accurately determine price elasticity relationships, over a broad range of interest rate scenarios, between different mortgage types(and coupons) so as to enable the construction of accurate consolidated exposure reports.

Use of a standard benchmark, such as the Treasury yield curve, also makes it possible to produce management reports that show on a consolidated basis, the effect on the risk profile of positions in other hedge instruments, such as Treasury options. Unless the highest-level risk-management report consolidates the effects of all positions taken, it becomes very difficult to express the true net exposure of the institution.

Include your exposure to pending rate locks. An estimate of rate locks issued, but not yet reflected in computer files or reports available to the secondary marketing department, should be included in exposure-management reports so that interest rate risk from those items can be reflected before hedge adjustment decisions are made.

It should be recognized that, not only do you have a current exposure to any rate locks that are likely to be issued prior to your next price resetting, but also, your exposure will be greater in falling price markets as borrowers rush to beat anticipated price changes.

Figures 3 and 4 (Risk Exposure Summary--MBS Price Shift and Risk Exposure Summary--Derivatives) show ideal formats for communicating the interest rate exposure of a mortgage bank. Reports like these should be produced daily and made available to top management. The first is a shock report; the second is a derivatives report.

Hedge position management

Secondary marketing managers should not be forced to speculate on market direction. But secondary marketing managers are sometimes led to speculate on market direction when unreasonable profit expectations exist. For example, some mortgage banks impose the arbitrary standard that secondary marketing "break even" (with no credit for the value of servicing created or interest income).

Competition has in many instances driven average total origination profits (including the value of servicing) below the value of servicing. This, of course, does not mean that mortgage origination is no longer profitable on a net basis. Secondary marketing managers should not be required to make secondary marketing profits (as just defined) if it is not possible to do so without taking interest rate risk.

Taking interest rate risk obviously can't improve profitability unless the mortgage banker is able to "call the market" on a consistent basis. The ultimate goal of risk management decisions should be to maximize long-term average profits, while minimizing the variability of profits from month to month.

Actual results should be measured against an objective performance benchmark based on quantifiable variables. This benchmark should be based on results that are attainable without taking interest rate risk.

What is attainable depends on the marketing spread (the difference between the commitment price and the forward price), servicing values, fees collected, operating costs and the determinants of hedge cost. Determinants of hedge cost are rate-lock period, fallout function elasticity and interest rate volatility (the greater the rate-lock period, elasticity of fallout and/or volatility of interest rates, the greater the hedge cost).

Your risk-management system should measure this expected "all-in, ex ante profitability" on a rate lock-by-rate lock basis. Because it includes the cost of hedging, it is, by definition, an amount that is attainable without taking interest rate risk.

If interest rate risk is permitted, there should exist clearly articulated risk limits. Management should establish limits on the amount of interest rate risk exposure that is allowable in terms of the amount of profit and loss that would be expected to result from shifts in interest rates of various magnitudes in either direction. For example, upper management might stipulate that expected losses not exceed $25,000 for a 25 basis point rate shock or $50,000 for a 100 basis point rate shock.

As far as the use of options is concerned, the two extreme approaches to risk management are global (static) and delta (dynamic) hedging. Understanding the difference between global hedging and delta hedging is critical. This decision is particularly important for a mortgage banking operation that has fallout ratios that are highly sensitive to interest rate changes. Aside from whether or not to allow market calls, this is the most important risk management policy decision to be made.

Global hedging involves establishing a hedge position that has a return profile that is equal but opposite to the profit/loss profile of the exposure (e.g., the profit/loss profile of the rate-lock pipeline). Global hedging may also be referred to as static hedging because a global hedge does not automatically require changes in response to market movements. The more sensitive fallout is to rate changes, the more asymmetric will be the profit/loss profile of the rate-lock pipeline, requiring a higher proportion of option purchases to forward sales in a hedge position designed to flatten the net profit/loss profile.

Delta hedging involves not the purchase, but rather the replication, of options. Actual market price volatility is what determines the cost of replicating options. If a mortgage bank does not actually purchase options, it must then replicate them, because it starts from a position of being short options (i.e., from rate-lock commitments).

Whether a strategy of avoiding options turns out to be cheaper than a strategy of purchasing options, depends simply upon whether or not actual volatility over a given period is higher or lower than was the volatility implied by option prices at the beginning of the given hedge period. Delta hedging is sometimes referred to as dynamic hedging because a delta hedge automatically requires changes in response to market movements.

Global hedging advantages

Global hedges do not require constant adjustment because a global hedge is designed to self-adjust as interest rates (and correlated fallout) change. Thus, a global hedger avoids the well-known portfolio insurance problem of not being able to adjust in time due to large sudden price movements. Global hedging allows you to know your hedge cost in advance. In short, global hedging provides more consistent, and more predictable results.

Delta hedging advantages

In option theory, the delta of an option is the change in the value of the option with respect to a very small change in the value of the asset on which the option exists. It follows from this that an exposure to an option may be offset, for very small changes in the underlying asset price, by taking a reverse fractional position (equal to the delta) in the underlying asset. With time, however, slippage occurs in execution of this strategy because the delta changes after even very small changes in the market price of the underlying asset and the delta hedger is always at least slightly behind in adjusting his or her position to the theoretical delta of the option. The delta hedger is, in essence, always buying high and selling low.

In an efficient market, the cost of replicating an option through delta hedging (as a result of slippage) will be the same as the cost of buying the option outright (i.e., the cumulative trading profits/losses and transaction costs will equal the premium cost of an outright purchase). Therefore, a delta hedger is simply making a bet that actual volatility will be lower than implied volatility. The problem is, it is never known in advance what actual volatility will be over the period of the option replication.

A quick test of market efficiency is whether or not, on average, implied volatility equals actual volatility. Historically, on average, they have been very close, except for a two-year period following the stock market crash (1988 to 1989) when implied volatilities on bond options were higher than actual volatility (premiums were 10 percent to 20 percent greater than fair value).

For a mortgage originator with significant fallout risk, truly conservative risk management requires the purchase of options. Delta hedging involves several risks. The cost of the hedge is not known in advance; it will be determined by the actual volatility experienced over the life of the hedge. It introduces a human behavior risk, namely, that the delta hedger will freeze during a large market move and end up infusing a market view into his position hoping the market will come back, instead of religiously following the delta. The strategy involves frequent monitoring and recalculation of deltas to determine if an adjustment is needed.

Transaction costs also can be significant. You pay the bid/ask spread each time you adjust your position. Finally, there is the portfolio insurance problem. The success of the delta hedging strategy depends on the hedger's ability to adjust his or her position for each market movement. If there are large gaps in market prices, adjustments are missed, and replication costs (i.e., hedge losses) will be greater than expected.

Non-global hedgers should always have a contingency plan in place. Secondary marketing departments should have a contingency plan in place at all times, in the form of a standard report, informing the head position manager what specific trades would be required to bring net exposure back into line should significant market movements occur. For global hedgers, by definition, this report would not be of great importance because their hedge positions are designed to self-adjust to market movements. However, for followers of dynamic hedging strategies, this report is critical.

Transactional efficiency should be considered each time the position is adjusted. The secondary marketing department should determine for each forward type, settlement month and coupon, what the mortgage banker's net long or short position is, based on a loan-by-loan analysis of likely settlement month and market value sensitivity analysis. When hedge position adjustments are needed, they should be taken based on where the mortgage banker's current net long or short position is. Following this process will substantially reduce pair-off/roll costs.

Trading analytics

Hedge instruments should be selected based on relative value. Management should be familiar with the different types of hedge instruments available, their risk/reward characteristics, their appropriateness within the context of their operations and the relative advantages and disadvantages of each. In general, specific hedge instruments should be chosen based on a joint consideration of the company's desired profit-and-loss profile and the current relative cost of the alternative instruments.

Choose option strike prices based on relative value. Options of various types/strike prices are often mispriced and should be chosen based on relative value. The same hedge objective can often be accomplished by mixing over-the-counter (OTC) and exchange-traded options on different, but similar, assets of different strike prices and expirations, which provide the same basic protection but at a lower total cost.

Treasury-based hedges should be used more when option-adjusted spreads (OASs) on mortgages are high. Policy statements that permit varied usage of Treasury futures should require a prior examination of spreads between MBSs and Treasuries, on an option-adjusted basis (i.e., taking into consideration the effect of changing values of embedded prepayment options). On average, it is best to use Treasury futures when option-adjusted spreads on MBSs are high because high OASs indicate relative cheapness of mortgages and low OASs indicate relative richness.

Option-adjusted spread models should be used to derive hedge ratios. Statistical models for determining the relations between and among different MBS types and coupons and Treasury futures contracts should be avoided. Far more precise results can be obtained by basing hedge ratios on option-adjusted durations derived from state-of-the-art option-adjusted spread models that have been widely favored by money managers for several years.

Mark-to-market

The mark-to-market process should provide timely feedback. Marking a mortgage rate-lock pipeline to market can be an arduous task. Many mark-to-market systems are based on only the "intrinsic value" of options and rate locks, ignoring the uncertainty (or option "time value") element. This can lead to large errors and often opens the door to arbitrary accounting. For example, in the case of rate-lock commitments, most mark-to-market systems use expected closing percentages that can be arbitrarily changed from day to day, either ignoring completely or only loosely considering the impact of the relationship between market rate levels and existing rate-lock rates on fallout probabilities.

Fallout ratios should be estimated based on processing status and the difference between commitment price and market price. A mark-to-market that is based on option evaluation techniques and a constant rate-lock fallout function provides an objective measure of daily profit and loss.

A reserve for future hedge costs should be established. Although it is now well recognized that rate-lock commitments issued to a borrower are short put options, marks-to-market invariably do not evaluate rate-lock commitments based on their option-like characteristics (e.g., implied volatility, time to expiration, dispersion of in-/out-of-the-money amount). Many practitioners acknowledge that higher volatility results in higher hedging expense. This should be reflected by the establishment of fluctuating reserves for hedge costs. Your mark-to-market system should be able to compute and report this.

The secondary marketing department should produce income statements that include the effects of closed positions. Secondary marketing departments should produce daily and month-to-date income statements that show the change in the value of inventories plus the gain/loss on positions liquidated between marks.

A secondary marketing department should receive credit for interest earned on mortgages while they are in the warehouse. By not giving interest credit to the secondary marketing department, at least two distortions occur. First, when placing forward coverage, an incentive to bet on the possibility of an early settlement is created (because forward prices are generally higher the closer in the settlement date). Second, pricing glitches will occur once a month when the projected settlement month for new rate-lock commitments rolls back a month, unless the secondary marketing department factors in interest carry from loan closing date to forward settlement date. This is something they may be disinclined to do if they do not receive credit for interest accrued during the warehousing period.

Pricing

A profitability analysis should be performed each day on a loan-by-loan basis taking into consideration hedging costs and the value of servicing. Relative pricing of different rate locks should take into consideration differences in hedge costs and servicing values.

Forward drops should be calculated to the expected day of closing rather than to the settlement date following the expected closing date. This is the case because if the loan closes prior to the settlement date, an interest spread (carry) is earned or paid.

Mortgage banks should establish internal transfer prices for servicing that are equal to the economic value of the servicing to the mortgage bank based on its cost of capital and cost of servicing. The effect of these variables, along with macroeconomic parameters such as prepayment functions and interest rate volatility, can be determined by periodically running option-adjusted spread model evaluations. This would thereby discourage the transfer of excess servicing to third parties for a price that is less than the economic value of the excess servicing to the mortgage bank.

Stephen R. Rigsbee, Sirri S. Ayaydin and Charles A. Richard III are principals of Quantitative Risk Management Group, based in Chicago.