In the Banking column of the May/June 1993 issue of Business Geographics, Hass Tavakoli discusses GIS in the mortgage industry. Hass focuses on what I term "micro" GIS applications, that is, using GIS to efficiently and credibly appraise individual properties during the mortgage underwriting process. In this column I will describe "macro" applications -- essentially, relying on GIS to profitably manage entire portfolios of residential and commercial mortgages.

First, though, some background on the banking industry during the last decade is in order. Banks experienced a series of severe economic blows beginning in the early 1980s. Loan default rates from traditional, major business lines -- Third World , Oil Patch, agribusiness, and real estate -- all began to escalate at an alarming rate and, worst yet, at roughly the same time. The fact that newly deregulated savings and loans (S&Ls) were pumping vast sums into Sun Belt real estate exacerbated the problem tremendously. The result was the highest rate of bank failures since the Great Depression, to say nothing of the virtual demise of the S&L industry.

In many instances the very survival of banks -- then and now -- came to depend upon the perceived viability of the banks' commercial and residential mortgage portfolios. Government regulatory agencies such as the Office of the Comptroller of the Currency (OCC) and Federal Deposit Insurance Corporation (FDIC), under heavy fire from all sides for allowing the U.S. banking system to approach collapse, seemed to adopt a "guilty until proven innocent" posture when evaluating mortgage portfolios. Banks that could not clearly demonstrate to the regulators that the vast majority of their mortgages were and would continue to be "performing" (i.e., receiving mortgage payments from the borrowers on time and in the full amount) were prime candidates to be shut down.

How could bank officers -- again, both then and now -- demonstrate the viability of their mortgage portfolios? One way is to have every property reappraised every year. At least two problems, however, arise with reappraisals. The first problem is the cost, which can run into thousands of dollars for residential properties and tens of thousands of dollars for commercial properties. When issuing the mortgages, the banks did not assume that such expenditures would be needed each year, and accordingly did not set aside funds for such expenses. The second problem is getting the regulators to believe the reappraisals even if conducted. The appraisal industry has received at least as much criticism as the regulatory agencies for the problems in the banking industry; among the results have been Congressional initiatives to tighten appraisal standards.

Another approach for demonstrating viability is to evaluate in-depth a small, statistically selected sample of the mortgages. Accounting firms frequently employ this method when conducting annual audits. If the mortgages in the sample are viable, then the entire portfolio is viable, subject to the confidence intervals associated with the statistical methods used. At least two problems again arise. First, a statistically sufficient sample size can still be so large and so expensive to evaluate as to be cost-prohibitive. Second, the auditors also share in criticism levied at the regulators and the appraisers; if the auditors' statistical methods are reliable, why are so many banks suddenly in trouble?

All of which leads to the third, GIS-based approach, one which is only just starting to be used but which holds tremendous near-term and long-term potential. As pioneered by the Big Six accounting firm of Deloitte and Touche, in partnership with the real estate data vendor REIS Reports, Inc., the approach basically has five steps:

1. Import one or more bank MIS computer files containing the locations (e.g., situs address) of all the properties encompassed by the bank's mortgage portfolio into an appropriate GIS.
2. Use the GIS to identify spatial clusterings of the mortgages, either visually or with GIS tools used for trade area analysis.
3. For those real estate markets where the mortgages cluster, retrieve current and forecast information on the average asking rents, vacancy rates, operating expenses, capitalization rates, and similar standard inputs of real estate pro forma models. (The information might come from internal proprietary sources, from local brokers and appraisers, or from third party data vendors such as REIS Reports, DRI/F.W. Dodge, the Building Owners and Managers Association, etc.)
4. For each market, determine whether the properties covered by the mortgages will be able to cover the debt service on those mortgages if those properties experience the market's average asking rents, vacancy rates, operating expenses, etc.
5. Calculate the sum of all the mortgages whose properties cannot cover debt service, i.e., that already are or will become non-performing. This amount dictates the magnitude of the loan-loss reserves the bank will have to set aside.

From the perspective of the OCC and other regulatory agencies, the magnitude of the requisite capital reserves in comparison to the bank's total assets is a critical measure of the viability of the bank. The above approach provides an explicit, credible, cost-effective way for a bank to prove its "innocence."

As an added benefit, of course, the bank can use the same system for internal decision support. Periodically (say, quarterly or semi-annually) the bank can determine which markets are attractive for additional mortgage underwriting, which markets are stable but more risky, and which are so risky that current mortgages in those markets should be sold. (This is similar to the use of GIS for "market ranking" as discussed in my column in the July/August issue of Business Geographics.)

Incidentally, the same basic approach can also be used for assembling and pricing pools of mortgages in Collateralized Mortgage Obligations (CMOs), Real Estate Investment Trusts (REITs), and similar investment securities.

If enough banks adopt GIS for this purpose, and are sufficiently sophisticated in the design of their systems -- including appropriate data assembly on asking rents et al. -- the nation should be able to avoid a repetition of the nearly disastrous past decade.

____________

In the November/December issue of Business Geographics we will examine the use of GIS by private sector appraisers and public sector assessors for property valuation.