Home » Publications & Resources » Article Library » 2007-08 » Credit Stress Testing

Publications & Resources

April/May 2007
Focus: Risk Assessment & Disaster Planning

Credit Stress Testing: A Tool for Assessing Concentration Risk A Path to Improving Risk Management

By Larry V. Sorensen

Challenging times are often the most instructive. For those institutions heavily reliant upon commercial real estate (“CRE”) lending, the current environment is decidedly more difficult than just a few years ago. Add to that, the heightened regulatory anxiety over commercial real estate concentrations, and the time is ripe for a serious look at the industry’s risk management tools, practices and capabilities.

On December 6, 2006, the federal banking regulators issued their final joint Guidance on Concentrations in Commercial Real Estate Lending and Sound Risk Management Practices. The clear message is that risk management practices must improve and CRE concentrations must be prudent and well defended. What is not so clear, is how.

One place to start that may be appropriate for your institution is credit stress modeling[1]. Credit stress testing can be the foundational tool to accomplish the larger objectives of quantifying your level of concentration risk, establishing sensible concentration risk limits, and addressing the concerns of the banking regulators. Below is an outline of one such approach.

The basic concept of credit stress modeling is to evaluate key credit performance parameters (typically debt service coverage and loan to value ratios) under adverse economic conditions. The “adverse economic conditions” are the input variables to the model, and different models can use different variables. For purposes here, the three variables that will “stress” the loan will be interest rates (assuming the loan has an adjustable rate feature), market capitalization rates and property vacancy rates[2]. The initial data input into the model is from the current characteristics of the loan (interest rate, loan amount, amortization period, etc.) and the underlying collateral property (income, expenses, vacancy, property type, geography, etc.).

For stress testing and benchmarking purposes, it makes sense to evaluate the base case, meaning current conditions, and to compare the calculated DSC and LTV ratios to moderate and severe scenarios based upon prescribed stress increments to interest, vacancy and cap rates. A moderate stress scenario should be a fairly adverse market environment that would not be unusual in the normal course of business cycles, while a severe stress scenario[3] should be an extremely adverse environment that is unlikely in the normal course of events, but still possible (i.e., the thrift crisis).

There are a variety of sources for interest, cap and vacancy rate benchmarking data you would need to establish defensible stress increments[4]. While some customization by property type and geography makes sense, standardizing the stress increments in the model allows you to roll-up the individual loan results to evaluate the impact of adverse market conditions upon specific portfolios.

A way to extend the benefits of the analysis would be to integrate into the stress test model a bifurcated loan grading system. DSC and LTV ratios are not only key credit performance indicators, they also map directly to the risk management framework being developed by the Basel Committee on Banking Supervision. Two key components of the Basel II framework for quantifying credit risk are probability of default (“PD”) and loss given default (“LGD”). A property’s DSC ratio, or the ability to meet scheduled loan payments from property cash flow, can be viewed as a key indicator of a loan’s PD. Similarly, a loan’s LTV ratio, or the relationship between the market value of the collateral property and the outstanding loan balance, can be viewed as a key indicator of a loan’s LGD (see Table 1). Taken together, the credit stress model can be used to express a loan’s possible credit performance (default risk and loss exposure) under adverse market conditions.

TABLE 1

Once you have run your loan portfolio through the model, you will have a treasure of data you can use to help evaluate your level of risk exposure under moderate and severe stress conditions. Based upon how you segregate your loan portfolio for concentration risk purposes (i.e., by property type, geography, etc.), you now can estimate how much of each sub-portfolio fails a 1.00 DSC ratio under stress, and therefore presents a higher risk of default, and of those loans, how many have LTV ratios so high that losses are probable in the event of default.

You are now in a position to evaluate potential loan defaults and loss exposures, by sub-portfolio, under various market and economic environments. The final step is to size up the potential losses under moderate and severe stress relative to your earnings, reserves and capital. In addition to possible loan losses, remember to consider the impact of non-performing loans on interest income, and all the attendant costs of workouts, foreclosure and sale. From this exercise, senior management and the board can establish defensible concentration risk limits that are compatible with the institution’s risk tolerance and financial condition.

What is outlined above is by no means the end of the potential benefits that credit stress testing can provide. Improved risk based pricing is one example. With a credit stress model and two-tiered credit grading, you can look at weighted average loan yields across both DSC and LTV credit grades to evaluate the risk and return characteristics of your portfolio. You may well find that your loan pricing doesn’t always align well with risk. Additionally, by instituting quantitative loan grading standards, you will be positioned to evaluate credit migration over time and under various stress scenarios, strengthening your risk mitigation capabilities. It may be that the model’s results help highlight certain property types or geographic sectors that are particularly susceptible to adverse market conditions. That information can translate into tightening underwriting standards or pricing changes that will help redirect the credit exposures of the portfolio.

Integrating credit stress testing into your lending function can bring valuable insights to your organization and elevate your risk management capabilities. The knowledge gained will help make you better risk managers and equip you to have a more substantive and favorable concentration risk discussion with the regulators the next time the examiners visit.

[1] See Industry Studies in the Research section of the Dominion Bond Rating Service website (www.dbrs.com) for an interesting approach to stress testing applied to a hypothetical bank for purposes of assessing CRE concentration risk. The study’s title is “Rising Exposure to Commercial Real Estate in Bank Loan Portfolios - Time Bomb or Fire Cracker?” (referenced herein with DBRS permission)

[2] If you think through the mathematics, you will find that while changing vacancy rates impacts both DSC and LTV, changing interest rates only impacts DSC, and changing cap rates only impacts LTV.

[3] For example, a severe stress scenario given current market conditions might be up 200 bps in interest rates, up 100 bps in cap rates and up 10 percentage points in vacancy rates.

[4] The Federal Reserve publishes historical interest rate data on its website (www.federalreserve.gov). Torto Wheaton Research sells historical vacancy rate data by property type and geography (www.twr.com). Global Real Analytics sells historical capitalization rate data by property type and geography (www.graglobal.com).

Larry Sorensen is a banking professional with more than 20 years experience in the industry, most recently as executive vice president at Sonoma National Bank in Santa Rosa , Calif. He can be reached at Larrysorensen@hotmail.com or 415-297-8648.

Unauthorized reproduction of all or part of this material without the express written consent of the author is strictly prohibited. All rights reserved.

© 2012 Western Independent Bankers, All Rights Reserved.