This post comes to us from Milton Harris, Professor of Finance and Economics at the University of Chicago, Christian Opp, Ph.D. Candidate in Finance at the University of Chicago, and Marcus Opp, Assistant Professor of Finance at UC Berkeley.
In our paper, Rating Agencies in the Face of Regulation – Rating Inflation and Regulatory Arbitrage, which was recently made publicly available on SSRN, we develop a rational expectations framework to analyze how rating agencies’ incentives are altered when ratings are used for regulatory purposes such as bank capital requirements. Rating agencies have been criticized by politicians, regulators and academics as one of the major catalysts of the 2008/2009 financial crisis. One of the most prominent lines of attack, as voiced by Henry Waxman, is that rating agencies “broke the bond of trust” and fooled trustful investors with inflated ratings. However, should sophisticated financial institutions be realistically categorized as trustful and fooled investors in light of the fact that they interacted with rating agencies not only as investors but also as originators of subprime mortgage securities? Why would these institutional investors care about ratings when they knew about rating agencies’ practices?
We argue that a first-order benefit of a high rating stems from financial regulations, such as minimum bank capital requirements. Over the last 20 years bank capital requirements (Basel I guidelines (1988) and Basel II guidelines (2004)) have become increasingly reliant on ratings as a measure of risk. For example, banks must hold five times as many reserves against BBB+ securities than against AAA securities. Moreover, the investment-grade threshold and the AAA threshold have become regulatory investment restrictions for pension and money market funds. Since these regulations are of first-order relevance for institutional investors’ capital management, a AAA label is economically valuable, independently of the underlying information it provides about the risk of a security.
Consistent with these observations, we develop a rational-expectations model of the “rating game” in which institutional investors face regulatory constraints that are contingent on ratings. The model reveals that regulation may, at least in part, reconcile rating inflation in select asset classes, low risk premia and investment by rational investors that are aware of the rating agencies’ practices. We show that regulatory benefits for highly rated securities distort the rating agency’s incentive to acquire information. If these benefits are above a threshold, the rating agency stops to acquire information and simply engages in rating inflation: the rating agency effectively becomes a regulatory arbitrageur rather than a provider of information. This extreme result is more likely to occur for complex securities that are costly to evaluate. Two observations may be explained by this result: the apparent low effort by rating agencies to create sophisticated models for the mortgage market, an area outside of the rating agency’s primary expertise, and the fact that exotic, structured securities receive a much higher percentage of AAA ratings (e.g., 60% for CDOs) than do corporate bonds (1%, see Fitch (2007)).
Interestingly, the effect of regulatory benefits is ambiguous below the threshold level at which rating agencies in our model produce no information. It is possible that the rating agency acquires more or less information in response to an increase in regulatory benefits. We show that these comparative statics depend on the distribution of types in the cross-section. If there is a large fraction of bad types and a few exceptionally good types, the rating agency will acquire less information, enabling more securities to be classified AAA. In the opposite case, the rating agency increases its information acquisition. A gradual shift in the distribution towards bad types, such as the inclusion of subprime mortgages, would therefore reduce information acquisition and hence would lead to a less efficient allocation of funds in the economy.
The full paper is available for download here.