Benchmarking the Probability of Default for Commercial Mortgages

RMA Journal, The, Nov, 2001 by George J. Pappadopoulos

The tragic events of September 11 and continued unhealthy economic news have investors deeply concerned about the future direction of their various interests. Many credit portfolios have felt the sting of the downturn, and risk managers need to ascertain the effects of continued deterioration across many lines of business. Unfortunately, proper allocation of capital becomes an even tougher issue when a consistent measure of credit risk across the entire portfolio does not exist. Although available risk management systems are helpful for a number of asset classes, none provides an adequate solution for real estate credit risk.

The issue lies in the shortfall of appropriate inputs. The credit models are certainly mathematically robust, but the real-estate-specific insights are lacking. Although traditional fixed-income markets benefit from a wealth of high-quality data, there is an unfortunate scarcity of information regarding commercial mortgage defaults. Because the vast majority of real estate is privately held, meaningful performance data arc difficult to come by. While the emergence of the CMBS market has made positive inroads, its true value from a data-generation perspective has yet to be realized. The 1998 crisis posed severe capital markets constraints, but the underlying mortgage collateral has benefited from an up real estate market for almost the entire period in which a meaningful CMBS market has been in place.

This lack of actual data on commercial mortgage defaults begs an alternative approach to understanding expected probability of default. Because individual property market and property type cycles are a major driver of default, credit risk needs to be assessed through three key systematic risk factors:

1. Market volatility.

2. Expected growth of NOI and value.

3. Loan structure protection. I The mathematical combination of these thr4 e factors allows the development of an informed opinion about the relative riskiness of different loan structures in different markets and property types. [1]

Although integration of the three key risk factors is the strength behind this metric, another, perhaps more helpful feature is its standard deviation-based scale. As probability theory dictates, volatility outlines the likelihood of default. Therefore, this important metric can be directly translated into an expected frequency of default. In other words, calibration to known historical default outcomes can be directly accomplished and then used to forecast defaults on a differentiated basis.

The only available data for this process come from a seminal study of defaults originally conducted by Mark Snyderman, [2] and recently updated by Esaki, L'Heureux, and Snyderman. [3] The study tracks average cumulative default rates for a commercial mortgage pool of more than 15,000 loans garnered from American Council of Life Insurers (ACLI) data. The important outcome is a portfolio-level benchmark for the overall cost of default. This study is quite useful in that it tracks each loan throughout its life span and it currently provides the best benchmark for observing actual cumulative loan default.

Lacking specific, detailed loan information utilized in the study, summary ACLI statistics were used to develop a starting point for modeling a pool as similar as possible. The simulated portfolio contained more than 8,000 loans with underwriting terms equivalent to ACLI averages, by time period and by property type, for coupon rate, underwritten LTV, and DSGR. To match the origination timeframe of the updated study, modeled mortgages were underwritten every quarter from the first quarter of 1982 through the fourth quarter of 1992. Further, the relative size of each loan was weighted by an estimate of the capitalized value of the real estate in each respective market and property type in place at the time of each loan origination. In other words, each quarter's originations reflect the structure of the U.S. real estate property market, as well as the average loan terms that were in effect at each point through history.

Overall, the process yielded results that were quite striking. In fact, the performance of this modeled "market basket" of loans fit the outcome of the broad-based, actual mortgage pool extremely well. The study by Snyderman and others concluded that 18.1% of the loans defaulted, and this model's benchmarking process exhibited an overall weighted average frequency of default of 18.07%.

Even more encouraging are the results at a more disaggregated level. The chart below breaks out the modeled default frequencies by year of origination and compares them directly with those encountered by the actual loan pool. Given the difficulties encountered, the calibration is extremely well matched to the actual default experience described by Snyderman and others. Overall, the series are very highly related to each other as they exhibit a correlation coefficient of 82%. Perhaps more important, with only a few exceptions, the individual levels are very tight at each point along the curve.