Printer Friendly

Servicing's cost code.


Using four year's worth of data, Peat Marwick is trying to crack the code that hides the basic cost fundamentals of servicing.

Pleased and puzzled by the results of the cost analyses of the "1989 Mortgage Servicing Performance Study" (MSPS) (Mortgage Banking, September 1990), we elected to revisit the wealth of data provided to us by the participants. We did so using logic, seeking truth and, more importantly, hoping to gain a better understanding of predictors of performance in mortgage servicing.

Our preliminary cost analyses indicated that the lower-cost servicers shared certain characteristics, such as higher loans per full-time equivalent (FTE), lower salaries and a high proportion of loans serviced for GNMA. While higher-cost servicers did experience somewhat higher salaries and fewer loans per FTE, they did not necessarily have higher delinquency or prepayment ratios, nor did they share any other key characteristics that might have generally indicated a higher cost.

Using the Statistical Package for the Social Sciences (SPSS) as the analytical vehicle, we created a series of scenarios where certain study variables (such as portfolio composition) were used to "predict" study results (such as core cost). The SPSS mechanics calculated the strength of the independent variables' ability to "predict" results using a regression technique designed for smaller sample sizes.

Not surprisingly, the study's relationships were magnified using statistical techniques. Like the cost analyses, every test raised 10 more questions. Those presented here are only the beginning of the application of additional analytical techniques to some very credible data.


The preponderance of loans serviced for GNMA found among the lower-cost servicers highlighted an observation made during the course of this four-year study. In addition, our extensive valuation and merger and acquisition experiences have implied, time after time, that portfolio composition, or book of business, plays a key role in the performance of a servicing entity. Hence, portfolio composition was selected as an independent variable; the analysis was intended to measure the strength of the portfolio's composition to "predict" core cost. Chart 1 summarizes these analyses.

With a population of 2 million loans, 29 percent of those serviced for GNMA, our analysis indicated a strong negative relationship between the percentage of loans serviced for GNMA and core cost. From our limited sample, this suggests that an increasing percentage of GNMA servicing applies downward pressure on core cost. This statistical analysis is supported by the lower-cost servicers in the 1989 study, who, on average, serviced more than 42 percent of their portfolio for GNMA.

Observation and experience support the theory that a concentration of GNMA servicing may indeed produce lower servicing cost results. Fully escrowed loans demand state-of-the-art management of the tax and insurance functions. The VA no-bid exposure and its bottom line impact keep immense pressure on the collection and foreclosure processes. The timing of investor reporting and remitting allows only slight leeway in this function. Hence, large GNMA servicing operations do take on "factory-like" characteristics. However, other study portfolio factors might affect this lower core cost. For example, study participants with an above-average percentage of GNMA servicing also serviced less than the study's average percentage of ARMs, owned loans and loans for the parent.

How much impact does the percentage of GNMA servicing in the portfolio have on core costs? Chart 2 suggests that a high percentage of GNMA servicing lowered core costs dedicated to customer service among the 1989 participants--up to a point.

Interestingly enough, there was a very similar relationship between the servicing of a higher percentage of loans for private investors (13 percent of the study's portfolio) and core cost. Further, when the percentage of loans serviced for GNMA and those serviced for private investors were combined, the results provided an even stronger predictor of decreasing core cost.

Why would the servicing of GNMA and private investor loans have a very similar effect on core cost? These loans in the study displayed some very distinct differences in average loan size, prepayment ratios and delinquency ratios. One suggestion is that the servicing for these investors reflects somewhat parallel borrower characteristics. For the FHA/VA borrower, these programs may be the only source of financing. For the jumbo borrower, or borrower who will not meet conventional conforming standards, a loan meeting the requirements of the private investor may be the only source of financing. With no analysis of credit quality, the actual servicing of loans for GNMA and for private investors--and their costs--may have some similarities. Another possibility is that the servicing of loans for GNMA and private investors actually complement one another and, therefore, behave similarly when analyzed statistically.

Surprisingly, the study data indicates a strong positive relationship between the portfolio's 15 percent concentration of loans serviced for Freddie Mac and core cost. This analysis suggests that as the percentage of Freddie Mac servicing increases, upward pressure on core cost occurs, driving up the core cost.

Looking further into this, we isolated study participants who serviced between 21 percent to 50 percent of their servicing portfolios for Freddie Mac (see Chart 3). In these portfolios, core cost appears to have been driven by some other factors. For instance, portfolios with concentrations of Freddie Mac loans experienced higher levels of prepayments in 1989. Also, loans serviced for Freddie Mac were newer (one- to three-years old) than the overall average age (5.5 years) of the entire portfolio.

Traditionally, thrifts have been the primary servicer of loans for Freddie Mac. During the course of this four-year study, thrift participants have tended to show above-average core cost. Logic would suggest that this finding of higher core cost associated with Freddie Mac servicing is more a result of the fact that thrifts in this study have higher core servicing costs in general. They also tend to have higher concentrations of Freddie Mac servicing. This was not the case of the 1989 study, however.

The analysis does not suggest that it is more costly to service loans for Freddie Mac. What it does suggest is that a higher percentage of Freddie Mac servicing drove up core cost among study participants in 1989. Hence, it became necessary to identify specific events that may have created this effect last year.

Our experience in the financial institutions industry has heightened our awareness of the impact of regulatory and accounting changes on lenders' ongoing operations. Mortgage servicing entities are not exempt from these changes. While the impact of secondary market investor rule changes that affect the reporting of servicing activities is not directly measurable, it is logical that a flood of new rules in a short time span might reduce efficiencies, thus driving up core costs in the short run as well. Because Freddie Mac issued more bulletins in 1989 than either Fannie Mae or GNMA, these additional reporting requirements may have translated into higher investor reporting costs and, subsequently, higher core costs for those participants with higher concentrations of Freddie Mac servicing.

An operational observation supports this finding. Increasingly, we have noticed that investor reporting is performed within the payoff, collection and other pertinent functional areas, in addition to the traditional "investor reporting" functional area. This decentralized reporting approach, coupled with multiple reporting changes, would create more inefficiencies in the short term than if all investor reporting were done by a single functional area. With time, these inefficiencies would be reduced as the new reporting requirements become automated and part of the normal routine.

A similar relationship existed between the percentage of owned loans serviced (18 percent) and core cost, as well as the percentage of ARM loans (17 percent) and core cost. In each case, the percentage predicted a positive, or increasing effect on core cost. Several of those servicers with higher concentrations of loans serviced for Freddie Mac also had above-average percentages of both owned loans and ARM loans serviced.

The casual observer may assume that all large servicing operations are like "factories" and thus capable of lower core costs. These preliminary statistical analyses suggest that portfolio composition, and, more specifically, the percentage of GNMA servicing, may be a better predictor of factory-type cost characteristics for the larger servicers.

Another area that we believed deserved attention was the ability of portfolio composition to "predict" specific functional areas of core cost--in particular--customer service and collection and foreclosure. Because these two areas represented almost 70 percent of total core cost for the participants in the study, any statistically meaningful relationship between the two areas and portfolio composition would be extremely useful to today's mortgage servicer.

In the course of the analysis, several relationships became evident. In keeping with previous results when GNMA concentration was compared with total core cost, high portfolio concentrations of GNMA loans demonstrated a very strong negative relationship with the percentage of core cost devoted to customer service. This suggests that a higher concentration of GNMA servicing results in lesser customer service costs. This was particularly true of the lowest-cost servicers in the study who were also the servicers with the highest percentage of GNMA loans.

This relationship holds true as well, although to a lesser extent, for private investor loans; the higher the concentration of private loans in a servicer's portfolio, the lower the customer service core cost. The basis for the similarity of these two investors complements the relationship found between percentage of servicing and overall core cost.

Another relationship of interest included the concentration of owned loans and customer service core cost. As the percentage of owned loans in a servicer's portfolio increased, so did the core cost devoted to customer service. This is to be expected based upon the tendency of many thrifts to service their own loans, and upon the presence of thrifts in the 1989 study. These findings suggest that attempting to balance the customer service relationship-versus-efficiencies equation carries with it a tendency to increase customer service core costs.

As expected, the relationship between the concentration of adjustable rate loans in a servicer's portfolio had a moderately positive relationship with customer service expenditures, suggesting that as the concentration of ARM loans increased, the customer service costs increased as well.

In the collection and foreclosure area, the most predictable relationship was between GNMA concentration and core cost devoted to this area. The no-bid consequences of GNMA loans demand a healthy allocation of servicing core cost to collection and foreclosure activities. The statistical analysis run on the participants of the study confirmed this theory by showing a very strong, positive relationship between GNMA concentration and collection and foreclosure costs as well as between GNMA concentration and delinquency ratios.

A notable lack of correlation existed between the percentage of ARM loans in a given portfolio and collection and foreclosure costs. In practice, we have often found this to be true. This finding suggests that the theory that servicing ARM loans is not more costly than servicing fixed-rate loans has statistical merit as well.

The moderately negative relationship between owned-loan concentration and collection and foreclosure cost intimated tighter underwriting standards adhered to by servicers when the risk of default was their own. Higher customer service expenditures may have an impact on keeping collection and foreclosure costs lower, and the emphasis on customer relations may create the incentives to solve problems before they reach the collection and foreclosure stage.

As evidenced by the relationships discussed, the possibilities for statistical analysis with even the most basic of data are endless. Many more theories than those servicing costs discussed were suggested by our research, including the theory of a "critical mass" of servicing by investor type. This "critical mass" appears to be the basis for some servicers' ability to maintain or to lower core servicing cost. However, limitations with our data prevent us from an in-depth study of this theory and many others.

While it is certainly justifiable to draw the conclusions that we have provided from the study results during the last four years, as well as from prior industry experience, and from our present statistical capabilities, we are able, at this point, only to lay the basic framework for future analysis of the mortgage servicing industry. However, this is clearly a promising avenue for further research, and will get only more promising as it is pursued. The more data that is made available to us, the more precise our findings become and the more positive impact our conclusions will have on the mortgage servicing industry. [Charts 1 to 3 Omitted]

Linda C. Simmons is a principal at KPMG Peat Marwick's Washington, D.C. office. Dr. Simmons is an active participant in the firm's national mortgage banking practice. "The 1990 Mortgage Servicing Performance Study" will be conducted by KPMG's Washington D.C. office.
COPYRIGHT 1990 Mortgage Bankers Association of America
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1990 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:KPMG Peat Marwick's research on mortgage servicing
Author:Simmons, Linda C.
Publication:Mortgage Banking
Date:Oct 1, 1990
Previous Article:110 percent Swedish.
Next Article:Top dollar.

Related Articles
3-D computing a big boon.
Following the Money: The Inside Story of Accounting's First Mega-Merger.
Apples to apples.
Judge finds for KPMG Peat Marwick in Renaissance litigation.
REMICs emerge as popular new capital source.
Partner allowed to sue partnership following termination.
KPMG teams up with investment bankers.
Suing for secrets.

Terms of use | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters