Credit Ratings and Mal-Investment
by Marc Joffe
Prices play a crucial role in a market economy because they provide signals to buyers and sellers about the availability and desirability of goods. Because prices coordinate supply and demand, they enabled the market system to triumph over Communism – which lacked a price mechanism.
Interest rates are also prices. They reflect investor willingness to delay consumption and take on risk. If interest rates are manipulated, serious dislocations can occur. As both Horwitz and O’Driscoll have discussed, the Fed’s suppression of interest rates in the early 2000s contributed to the housing bubble, which eventually gave way to a crash and a serious financial crisis.
Even in the absence of Fed policy errors, interest rate mispricing is possible. For example, ahead of the financial crisis, investors assumed that subprime residential mortgage backed securities (RMBS) were less risky than they really were. As a result, subprime mortgage rates did not reflect their underlying risk and thus too many dicey borrowers received home loans. The ill effects included a wave of foreclosures and huge, unexpected losses by pension funds and other institutional investors.
The mis-pricing of subprime credit risk was not the direct result of Federal Reserve or government intervention; instead, it stemmed from investor ignorance. Since humans lack perfect foresight, some degree of investor ignorance is inevitable, but it can be minimized through reliance on expert opinion.
In many markets, buyers rely on expert opinions when making purchase decisions. For example, when choosing a car we might look at Consumer Reports. When choosing stocks, we might read investment newsletters or review reports published by securities firms – hopefully taking into account potential biases in the latter case. When choosing fixed income most large investors rely on credit rating agencies.
The rating agencies assigned what ultimately turned out to be unjustifiably high ratings to subprime RMBS. This error and the fact that investors relied so heavily on credit rating agencies resulted in the overproduction and overconsumption of these toxic securities. Subsequent investigations revealed that the incorrect rating of these instruments resulted from some combination of suboptimal analytical techniques and conflicts of interest.
While this error occurred in market context, the institutional structure of the relevant market was the unintentional consequence of government interventions over a long period of time. Rating agencies first found their way into federal rulemaking in the wake of the Depression. With the inception of the FDIC, regulators decided that expert third party evaluations were needed to ensure that banks were investing depositor funds wisely.
The third party regulators chose were the credit rating agencies. Prior to receiving this federal mandate, and for a few decades thereafter, rating agencies made their money by selling manuals to libraries and institutional investors. The manuals included not only ratings but also large volumes of facts and figures about bond issuers.
After mid-century, the business became tougher with the advent of photocopiers. Eventually, rating agencies realized (perhaps implicitly) that they could monetize their federally granted power by selling ratings to bond issuers.
Rather than revoking their regulatory mandate in the wake of this new business model, federal regulators extended the power of incumbent rating agencies – codifying their opinions into the assessments of the portfolios of non-bank financial institutions.
With the growth in fixed income markets and the inception of structured finance over the last 25 years, rating agencies became much larger and more profitable. Due to their size and due to the fact that their ratings are disseminated for free, rating agencies have been able to limit the role of alternative credit opinion providers. For example, although a few analytical firms market their insights directly to institutional investors, it is hard for these players to get much traction given the widespread availability of credit ratings at no cost.
Even with rating agencies being written out of regulations under Dodd-Frank, market structure is not likely to change quickly. Many parts of the fixed income business display substantial inertia and the sheer size of the incumbent firms will continue to make the environment challenging for new entrants.
Regulatory involvement in the market for fixed income credit analysis has undoubtedly had many unintended consequences, some of which may be hard to ascertain in the absence of unregulated markets abroad. One fairly obvious negative consequence has been the stunting of innovation in the institutional credit analysis field.
Despite the proliferation of computer technology and statistical research methods, credit rating analysis remains firmly rooted in its early 20th century origins. Rather than estimate the probability of a default or the expected loss on a credit instruments, rating agencies still provide their assessments in the form of letter grades that have imprecise definitions and can easily be misinterpreted by market participants.
Starting with the pioneering work of Beaver and Altman in the 1960s, academic models of corporate bankruptcy risk have become common, but these modeling techniques have had limited impact on rating methodology.
Worse yet, in the area of government bonds, very little academic or applied work has taken place. This is especially unfortunate because government bond ratings frame the fiscal policy debate. In the absence of credible government bond ratings, we have no reliable way of estimating the probability that any government’s revenue and expenditure policies will lead to a socially disruptive default in the future. Further, in the absence of credible research, there is great likelihood that markets inefficiently price government bond risk – sending confusing signals to policymakers and the general public.
Given these concerns, I am pleased that the Mercatus Center has provided me the opportunity to build a model for Illinois state bond credit risk (as well as a reference model for Indiana). This is an effort to apply empirical research and Monte Carlo simulation techniques to the question of how much risk Illinois bondholders actually face.
While readers may not like my conclusion – that Illinois bonds carry very little credit risk – I hope they recognize the benefits of constructing, evaluating and improving credit models for systemically important public sector entities like our largest states. Hopefully, this research will contribute to a discussion about how we can improve credit rating assessments.