Commercial Mortgage Defaults: From 1972 to 2011

CRE Finance World, Summer 2013

Commercial Mortgage Defaults: From 1972 to 2011 dentifying long-run loss expectations for commercial mortgage investments has been the quest of commercial mortgage market participants for decades, starting with the “Snyderman-Esaki” studies first published in 19911 and last updated in 20052. Here we extend those seminal studies and consider the implications of loan performance differences during the last recession. Since the last update, life insurers’ loan portfolios have performed well. This result is especially interesting because during the same period losses on loans in CMBS were high. The divergent behavior during a stressful economic period drives our conclusion that underwriting is the most important factor in whether a loan is likely to default. The importance of underwriting might not seem controversial, but since previous studies documented how defaults tend to bunch around economic downturns, it’s not been clear whether economic conditions are the key driver, or if other factors can trump economic conditions. Some will raise the question about whether the experience of life companies can be replicated by other lender types, since different types of lenders fill different market niches. Some proof that better underwriting can impact losses in more than just small segments of the broader market comes from the performance of GSE multifamily loans3. The GSEs have applied underwriting standards similar to life companies to the multifamily debt market as a whole, and have achieved similar results. There are a couple of differences in this work relative to the previous studies, but we generally focus on delivering information similar to the previous studies, knowing that many readers are familiar with how this data has been presented in that past. In our updated study we explore historic default performance, but have not updated the loss severity information. The previous study tracked performance only for cohorts with at least 5 years of seasoning. We track all cohorts given the interest in performance in recent years, but recognize that newly-originated loans have less exposure to losses than more seasoned cohorts. Background The commercial mortgage performance studies authored by Mark Snyderman, Howard Esaki, and others explored defaults at eight large insurance companies. These studies utilized loan performance data reported on individual loans each year, as required by regulators. The studies have been useful to the industry because the results inform both the level of losses and the timing of those losses, CRE Finance World Summer 2013 42 based on performance through a full economic cycle. In this work we updated commercial mortgage performance information in the same life insurance company portfolios from annual reports between 1999 and 2011 (collected by the National Association of Insurance Commissioners, NAIC). We then matched and aggregated that new data with the data underlying the previous studies†. Our focus is on the loans in the portfolios of the eight companies studied before because of the long history available. Default rates are lower than the previous study. Based on the historical performance over the period from 1972–2011 the cumulative default rate on the portfolio is close to 10%, down about a third from 15%. The drop is because we’ve added data covering a period of very low losses in the insurance company portfolios. Previous studies documented the high level of losses during the real estate crisis beginning in the late 1980s and continuing into the 1990s. The insurance industry responded to the previous losses with more disciplined risk management featuring conservative underwriting. The business practices they established have proven to be effective. The familiar time profile of default is consistent with previous studies, largely driven by the historical performance prior to including the more recent information. Losses start low, increase in years three through seven, and then drop off to low levels as loans season further. But looking only at the new data, the familiar shape is not as evident, as few defaults are insufficient to establish a pattern. There is a slight up-tick in defaults in year ten, reflecting a small amount of refinance risk at balloon maturity, but the information is insufficient to inform balloon loss expectations. Clearly, many factors drive defaults and practitioners need to consider macroeconomic conditions, vintage underwriting, and future interest rates – in addition to historical performance — to establish future expectations for loss timing. For the period from 1999-2001 we also had information on the broader universe of life insurance companies involved in commercial real estate from the NAIC. This gave us some ability to learn about how this sample of eight companies compared to the population. Generally, the sample companies represented just over a quarter of the overall universe of life company CRE debt. Although the sample is not large, it is fairly representative as measured by default rates conditional on origination cohort during the period of overlap. Performance of loans in the study was similar to those in the broader universe. I Jun Li Freddie Mac Yu Guan Freddie Mac Steve Guggenmos Freddie Mac Qi Gong Freddie Mac


CRE Finance World, Summer 2013
To see the actual publication please follow the link above