Consumer Risk Appetite: How Loan Demand Drives Credit Risk

By Joseph Breeden and Kevin Coop

For lenders and regulators, one of the latest hot topics is risk appetite. Lenders are being asked to specify their appetite for risk and set guidelines to manage the risks they assume in their lending operations. However, with all the discussion around lenders, what about consumer risk appetite? We seem to have forgotten that consumers also manage their personal financial risk based on their willingness to assume debt.

Consumers — that includes all of us — don't often consider what their risk appetite is. If you ask your friends at a party what they think is a reasonable debt-to-income ratio for their family or how leveraged they should be entering a recession, they will most likely tell you to stop working and have a beer. However, at the same party, if you ask whether this is a good time to buy a car or a house, you may well get an engaging response and many strong opinions — as well as a beer.

For decades, the Federal Reserve Board (FRB) has been conducting surveys that address similar issues — presumably without the beer. The FRB Senior Loan Officer Opinion Survey (SLOOS) asks lenders whether consumer demand for loans is increasing or decreasing. Although not as direct as questioning consumers themselves, the survey still provides an interesting timeline of how consumer demand for mortgages — one measure of consumer risk appetite — has changed over the years.

A recent research study by Strategic Analytics[1] compared the FRB SLOOS measure of consumer demand with the credit risk of mortgages originated each month since 1990. Strategic Analytics measured credit risk using its proprietary Dual-time Dynamics technology to compare loans originated in different months. Credit quality was measured as the mid-delinquency performance of each vintage[2] adjusted for the economic environment and age of the loans.

Figure 1 below shows that the credit risk of the loans originated by the mortgage industry varies strongly over time. In 1995, 2000 through 2001, and 2005 through the end of this study, we observe periods of exceptionally risky loans that had delinquency rates twice the average level.

Figure 1
A Comparison of Credit Risk

A Comparison of Credit Risk

The same graph compares the observed credit quality with consumer demand. We see that when consumer demand is high, the mortgages originated at that time will have very low credit risk. When consumer demand is low, high-credit-risk loans will be originated. This is an industrywide relationship. Individual lenders may book higher- or lower-risk loans according to their credit policy changes, but those originations still draw from the pool of consumers interested in borrowing. One lender may adopt policies intended to improve the credit quality of their originations, but if only risky consumers want loans, the result will not be what was intended.

Credit bureaus don't have access to one of the most important financial attributes of the consumer: Does the consumer take risks? Our hypothesis for the strong correlation between demand and credit risk is that the pool of consumers interested in obtaining a mortgage changes over time. In finance as in daily life, we can describe consumers as risk-averse or risk takers. At certain times, risk-averse consumers appear to pull out of the market. Demand falls — not uniformly across all types of consumers, but primarily for those low-risk, fiscally conservative consumers.

By comparing the credit risk measure in the plot to macroeconomic factors, we can see what elements may influence consumer demand for mortgages. By correlating to changes in house prices and changes in mortgage interest rates, we obtain an even better fit than that in the FRB survey. We find that when mortgage interest rates have fallen over the previous two years, consumer demand is high and credit risk is low. Conversely, rising rates correlate to rising credit risk for the loans issued at that time.

Low-credit-risk consumers seek steady growth in house prices. Also, when house prices are rising rapidly or when house prices are falling, fiscally conservative consumers avoid buying new homes. When a home is expensive due to either rising prices or rising interest rates, the lender needs to ask, "Why would someone want to buy now?"

Ironically, the FRB survey also shows us that lenders are not accurate in assessing the risks they are taking. The most watched question on the survey is whether lenders are tightening or loosening underwriting standards. The Strategic Analytics study reveals that industrywide the measure shows no correlation to the credit risk being assumed. Lenders may raise lending standards, but if low-risk consumers don't want loans, no change in underwriting standards can assure that good loans are written.

Figure 2
Predicting Overall Credit Risk

Predicting Overall Credit Risk

Using either the FRB SLOOS survey for consumer demand or current macroeconomic conditions, we can predict the overall credit risk that the mortgage industry is originating. Figure 2 above shows the indices through to the present, capturing the mortgage bubble and the current better-quality loans. These indices are available quarterly from Strategic Analytics.

Although this discussion has been about mortgages, we have observed the same patterns of shifting consumer risk appetite across all retail loan products. However, the macroeconomic drivers of demand will differ by loan type.

Shifts in consumer risk appetite interfere with even the best plans of originators. It is the biggest reason that risk-based pricing has failed over the last decade. If lenders incorporate this knowledge into their pricing and underwriting strategies, they may yet weather future disruptions and be successful.

Joseph Breeden, Ph.D., is a vice president at Interthinx and cofounder of Strategic Analytics, an Interthinx company. Kevin Coop is president of Interthinx.

1. Contact Strategic Analytics at for a copy of the research report.

2. A "vintage," also known as a "static pool," is a group of accounts originated in the same time period. The performance of each vintage is tracked separately and compared using vintage analysis techniques such as Dual-time Dynamics.