Your Loan Portfolio’s Private Parts

Quantifying Lending and Credit Risk

Three weeks ago we gave bankers a quick way to translate commercial real estate portfolios into a common rating using simplified methodology (HERE) in order to quickly access their portfolio or a portfolio for purchase. We received many comments from bankers wanting to know more of a detailed approach to accessing credit by loan type and wanting to know how C&I fit into a ratings construct.  As one banker put it, “You told us how a public rating might shake out, but we don’t care about public ratings. Can you tell us how we should think about it privately?” So, to that end, we take more of a historically-based approach in order to give bankers more of a granular framework by loan type, as to how to think of the statistical risk of their loans.

To do this, we analyzed the last 10 years of loan performance for banks (commercial and savings), $100mm to $10Bn in asset size to track actual performance.  We then compare community bank loan “private” default rates to publicly rated loans.  The results may surprise you.

Data for Community Banks

In the graph below we considered 11 loan categories for banks ranging in size from $100mm to $10Bn (both commercial and savings institutions).  Default rates are plotted from 2005 to 2014.   A few comments on this graph: First, of the 11 loan categories that we track, “all construction and land development” and “1-4 family construction and land development” show the highest historical default rates in down cycles. The default rates for construction and land development were at least four times higher than other categories during the last recession (as an aside, recovery rates were especially abysmal).  In fact, these two loan categories skew axis so much that they make other lines unreadable on the graph.  Second, and this is most important for bankers, in non-recession periods, the default rates on construction and land development loans fall to the average for all loans at banks.  This creates the insidious perception that this loan category default rates are benign and comparable to other categories.  We are now witnessing some banks rapidly expanding their construction lending.

Community Bank Loan Default Rates  

In the graph below, we eliminate construction and land development loans to observe how the remaining categories fared between 2005 and 2014.  First, by far the lowest default rates were observed for agricultural loans, followed by lease financing, then consumer.  While real estate default rates were higher than any of the remaining categories (and statistically significantly higher than C&I), interestingly there was no statistical difference in default rates between owner-occupied and non-owner-occupied CRE.   

Historic loan default rates for community banks

Default rates for the 11 categories of loans compiled for all banks between $100mm and $10Bn from 2005 to 2014 are shown below (sorted from lowest to highest):

Loan Average Probability of Default Rates

Data for Corporate Loans (Source Moody’s Investor Services)

While we have spent some time in this blog on commercial real estate lending, due to popular request, we also wanted to highlight the mapped ratings for commercial and industrial (C&I) loans. The graph below shows corporate default rates by letter rating.  The graph lines for loans rated AAA, AA, and BBB are invisible in the graph (the default rates are bumping near zero for the entire graph period). The only meaningful observed default rates are for loans rated BB and below. Default rate variations are observable primarily for single B-rated credits.  In other words, in stressed economic environments single B-rated credits default rates increase substantially. 

Commercial Loan Default Rates

The table below shows commercial default rates (in percent) by alphanumeric rating and for the period from 2005 to 2014 (to match our available data period for community banks).  

Commercial Loan Ratings Matrix

We can now map community bank loan category default rates to the loans with public credit ratings. We observe that the average construction and land development loan for the last 10 years defaulted at the rate of an average CCC+/CCC credit for that same period.  The average real estate loan at community banks exhibits default rates of the average B-/CCC+ rated credit.  The average agriculture loan at community banks (the best-performing credit category) is the default equivalent of a BB+/BB rated credit.

Conclusion

While the average community bank loan is the equivalent of a single B-rated credit, as measured by default rates (recovery rates will be the subject of another blog), the key conclusion in the data is more nuanced.  The more important observation in our analysis is that within any loan category (including construction and land development) there are loans that exhibit default rates very different than the category average.  However, banks can improve by differentiating the credit risk between different types of loans.  Understanding these differences can help bankers more accurately price, structure and manage these loans. A general purpose CRE loan, with diversified leases, in a stable and established MSA, with 2.0X DSC ratio, and 60% LTV may well show default rates equivalent to an A-rated credit (which today is priced at LIBOR + 1.50%).  On the other hand, a spec construction loan will probably, in the loan run, default at the average rate of a CCC-type credit (which today is priced at LIBOR + 6.50%).      

We make our PC and Smartphone-based loan pricing model, called Smart Loan Express, available to all financial institutions at our cost, so there is little excuse for not pricing credit, interest rate and liquidity risk correctly. The above data underscores the fact that banks need to incorporate more granularity into pricing, and it is difficult to do without a risk-adjusted model. Even though banks aren’t going for public ratings on their loans, understanding the “private” risk is equally important in order to optimize your capital.