Archer, W. R., P. J. Elmer, D. M. Harrison, and D. C. Ling, 1998, “Determinants of Multifamily Mortgage Default,” FDIC Working Paper No. 99–2.
Avery, R., R. Bostic, P. Calem, and G. Canner, 1996, “Credit Risk, Credit Scoring, and the Performance of Home Mortgages,” Federal Reserve Bulletin. Vol. 82, pp. 621–64.
Ball, M., and A. Wood, 1999, “Housing Investment: Long-Run International Trends and Volatility,” Housing Studies, Vol. 14, pp. 185–209.
Elmer, P. J., and S. A. Seelig, 1999, “Insolvency, Trigger Events, and Consumer Risk Posture in the Theory of Single-Family Mortgage Default,” FDIC Working Paper No. 98–3.
Jimenez, G., and J. Saurina, 2006, “Credit Cycles, Credit Risk, and Prudential Regulation,” International Journal of Central Banking, Vol. 2, pp. 65–97.
Quigley, J. M., R. Van Order, and Y. Deng, 2000, “Mortgage Terminations, Heterogeneity and the Exercise of Mortgage Options,” Econometrica, Vol. 68, pp. 275–307.
Vandell, K. D., W. Barnes, D. Hartzell, D. Kraft, and W. Wendt, 1993, “Commercial Mortgage Defaults: Proportional Hazards Estimation Using Individual Loan Histories,” AREUEA Journal, Vol. 21, pp. 451–80.
The authors thank Steven Ott, Stijn Claessens, and two anonymous referees for their comments on earlier drafts. Marcelo Pinheiro is an associate at Cornerstone Research, Washington, D.C. The material discussed herein may not reflect the opinions of Cornerstone Research.
See “Interagency Guidance on Nontraditional Mortgage Product Risks”, dated September 29, 2006, and “Concentrations in Commercial Real Estate Lending, Sound Risk Management Practices”, dated December 9, 2006, available at http://www.federalreserve.gov/boarddocs/press/bcreg/2006.
See Ruckes (2004), and references therein, for theoretical frameworks explaining cyclical variation in credit standards. Asea and Blomberg (1998) provide evidence on lax lending during the expansionary phase of the business cycle. Jimenez and Saurina (2006) present similar evidence for Spain.
Indeed, the correlation between credit movements and house prices was 0.74 between 1976 and 1990, but dropped to 0.13 between 1991 and 2006. Similarly, the correlation between income changes and house prices was 0.53 between 1976 and 1990, but dropped to 0.13 between 1991 and 2006.
At the time of the writing of the first version of this paper, the unwinding had not started, yet it is obvious now that the banking system is under distress as predicted.
In the period shown, for instance, the annualized volatility of commercial real estate price index was 3.44 percentage points while that of the residential real estate price index was 1.77 percentage points.
Subprime mortgages are defined as those loans offered to borrowers with higher risk profiles. According to the industry publisher Inside Mortgage Finance, the share of subprime lending increased from 9 percent in 2000 to 20 percent in 2006. In absolute terms, the change was from $120 billion to $600 billion.
These include interest-only mortgages, where the borrower does not pay any loan principal during the first few years of the loan, and payment-option adjustable-rate mortgages, where the borrower has flexibility over the amount he pays with the potential for negative amortization.
It should be noted that the stress testing exercise assumes that banks do not respond to changing market conditions, and hence, does not consider the dynamic nature of risk management practices. The implication is that the results of the stress testing exercise represent a worst case scenario of bank probabilities of encountering financial distress. Responsive managers can mitigate the problems by imposing more strict underwriting standards and enhanced scrutiny of loan applications.
CAMEL is a rating system used widely by bank supervisors. The main idea is assigning a score to each bank based on five factors: capital adequacy, asset quality, management, earnings, and liquidity. Lately, a sixth factor, sensitivity to market risk, was added, changing the acronym to CAMELS. For more information on CAMEL(S), see, among others, Lopez (1999).
Before proceeding with the regression analysis, we check if the data series we are using are stationary. All series are found to be I(1) at high levels of significance, with the slight exception of outstanding commercial real estate loans, for which the null hypothesis of the first difference having a unit root is rejected at the 10 percent significance level. Then, we determine the number of cointegrating relationships. The test statistics support the existence of two cointegrating relationships for real estate loans at the 5 percent significance level. For residential and commercial real estate loans, the number of cointegrating relationships are one and three, respectively. Tables summarizing the unit root and Johansen cointegration tests are available on request.
Higher interest rates here imply decreased ability to make payments because cost of borrowing is higher. This might occur, for instance, as economic activity reaches its peak and monetary stance tightens.
Summary statistics also point to a great degree of skewness in the distribution of key variables across banks. For instance, while the median delinquency rate stands at 2 percent, the mean is closer to 4 percent. Hence, it is possible that a small number of banks find themselves in serious distress, potentially spreading the problems elsewhere in the financial system through cross holdings.
For banks operating in multiple states, the data are rearranged at the state level. For example, Wachovia appears as several entities in the database as Wachovia California and Wachovia Arizona. Therefore, the location variable reflects the situation associated with the location of the loan activity and/or property.
An equally important development in mortgage markets has been the increased degree of securitization. The analysis here focuses on the traditional measures of exposure while supervisors should also watch the exposure through asset-backed security holdings.