An opportunity that should not be missed

RMA Journal, The, June, 2004 by George Pappadopoulos

How well are commercial bank real estate loans performing? Is your CRE loan portfolio performing better than your CMBS (commercial mortgage-backed securities) loans? Better than those of your peer institutions? If so, exactly how much better? How does each property type compare? Which geographic regions are outperforming others? What is the credit impact of recourse versus nonrecourse?

Although many of us have general opinions regarding these questions, no one can directly quantify the answers. The simple reason is that we can't point to specific empirical data to back up any hard quantitative conclusions. In other words, there is currently no performance benchmark for commercial bank CRE lending.

There are some arenas for commercial real estate within which these types of questions can be answered. The CMBS industry has a wealth of publicly available performance information. For example, we know that since 1992, historical losses for the CMBS universe have been a mere 15 basis points and that current delinquencies for securitized loans collateralized by multifamily properties average 1.77% versus 1.26% for offices. Since the American Council of Life Insurers (ACLI) tracks life company commercial real estate loans, we are aware of the low overall delinquency of 18 basis points and exactly how this compares to their delinquency peak of 7.53% in the early 1990s.

Certainly, this is useful information, but it is only the tip of the iceberg. Development of such databases allows for more granular differentiation of credit risk. Stratification can be accomplished across numerous dimensions, and the true impact of specific underwriting terms can be quantified. That is, one can examine various subsets of the data to better understand the differences in loss impact under alternative underwriting criteria, such as loan to value (LTV) and debt service coverage (DSCR).

Most important, Basel II-compliant risk models can be constructed and calibrated to more appropriately reflect differences across loan characteristics. The true impact of changes in terms, alternate loan type (e.g., construction, permanent, mini-perms, fixed rate, floating rate), property type, and geography can be assessed. The list goes on, and as a result, more accurate CRE loan grading and loss models will ensue.

Although there is some data available to aid this effort, the big question that remains is where do bank loans fit? The ACLI and CMBS databases are useful in that they cover commercial real estate loans, but there are some important differences between these loans and commercial bank CRE lending. First, how does the recourse-focused banking world benefit compared to the traditionally nonrecourse lenders? Borrower recourse in the event of default is not prevalent for life company and CMBS lending. All of us would agree that recourse has a loss-mitigating impact, but no one knows to what extent.

In addition, the existing databases primarily comprise long-term, fixed-rate lending on stabilized properties. Commercial bank CRE loans are mostly shorter-term, floating-rate products that can encompass nonstabilized properties or ground-up construction lending. A commercial bank CRE loan database would better explain some of the performance differences we know are inherent in these alternative types of lending.

Fortunately, much work has already been accomplished toward the establishment of a more appropriate database. RMA--The Risk Management Association, in conjunction with a steering committee of seven key commercial real estate lenders, has already set up the infrastructure for an ongoing study of CRE lending. Appropriate variables have been identified, expected output reports have been assembled, and RMA is now soliciting participants to join this consortium of data providers.

The objective is to deliver practical information that can be used to better assign risk ratings, set reserves, allocate capital, and strategically manage the composition of the real estate portfolio. The aggregated data set will be thorough enough, and the methodology flexible enough, to permit for confidential peer breakouts as well as numerous stratifications, such as geographic reporting.

At a minimum, the study would supply a badly needed benchmark for commercial banks. Answers to simple yet currently elusive questions, such as total origination volume and total volume outstanding, would be readily available. Furthermore, volume and loan count breakouts by loan type (e.g., permanent, interim, and construction), geography, property type, LTV, DSCR, origination period, recourse/nonrecourse, and a dozen different loan-status categories (e.g., current; 30, 60, and 90 days past due; REO; foreclosure; etc.) will all be available to participants.

The information will provide key insight into differentiating loan performance, not only outlining delinquency numbers but also cradle-to-grave default and loss-severity metrics. As noted earlier, stratification of performance information will increase our understanding of the true impact of underwriting terms, seasoning, loan type, geography, and property type on mortgage loss. This will lead to more accurate calibration of loss models and help address Basel II compliance.