Do Ratings Matter?

The role of Credit Rating Agencies has come to the forefront of the policy debate in recent months. The sudden collapse of complex structured financial instruments that were once categorized as safe “investment grade” products has cast doubts on the way rating agencies do their business, the structure of incentives they face, and the appropriateness of the existing regulatory framework and surveillance mechanisms.

Yet one question that few people have asked in this context is if ratings really matter. In other words: do ratings provide information above and beyond what is already reflected in the prices and yields of traded financial instruments? If rating agencies add no new information to markets their actions are not a public policy concern.

Given the nature of their business, it is clear that rating agencies probably have more information than the average investor. What it is not clear, and where there is considerable debate in the literature, is whether the opinion of rating agencies matter for the determination of market prices of financial instruments, even after controlling for the fundamentals underlying the value of these assets. Whether they do or not is an important policy question; especially for policy makers concerned about financial stability and access to finance. In a recent paper, we explore this question by scrutinizing the sovereign ratings’ market.

Ratings are oftentimes anticipated by the market because rating agencies appear to try to signal when rating changes may occur. Financial instruments are either given a positive or negative outlook (suggesting an upgrade or a downgrade may be the next change respectively) and additionally may be placed on a “rating watch” (indicating that a decision may be about to be made). Moreover, agencies publish what a particular issuer would have to do to improve its rating and while targets may not be precise, the information required to make a judgment is generally public and indeed may become a focus of market research and analysis.

Market anticipation means that it is a real challenge to answer the question as to whether ratings agencies add value and standard “event study” techniques may not be appropriate. If rating agency actions are fully anticipated then we would see no contemporaneous effects of rating changes on, for example, bond spreads. But seeing no effects would not mean that rating agencies do not add value. It just implies that whatever effect ratings had on spreads may have already been incorporated into spreads before the rating change.

We suggest therefore that we need to seek novel methods to tackle this question that do not rely on unobservable economic fundamentals and that takes into account that rating changes are possibly anticipated. For this purpose, we first devise a simple specification test that is robust to these features of the data. The question we are interested in answering is to evaluate the informational content that the rating has in addition to the observed spread on marketable sovereign debt. In other words, the null hypothesis is that all the information in the rating is already reflected in the spread. This is equivalent to say that the spread is a sufficient statistic. The alternative hypothesis, on the other hand, implies that spreads and ratings are imperfect measures of the unobservable fundamentals of the economy; and therefore ratings provide information above and beyond what spreads reflect. Across several variants of the test, we find high rejection rates for the null hypothesis. In other words, there does seem to be informational content in country ratings that is not captured by the sovereign spreads.

Having established that spreads are not a sufficient statistic, we turn to estimating a new model in which we exploit the informational content of ratings in order to explain the variation in three macro variables (i.e., stock market indices, nominal exchange rates, spreads one day forward) using high frequency (daily) data. We consider a horse-race between ratings and spreads as to how well they are correlated to the other macroeconomic variables. We find that ratings typically explain part of the variation in the selected macro variables, even after controlling for the spread. The fact that the rating is significant after controlling for the spread is supportive evidence in favor of the hypothesis that spreads are not a sufficient statistic.

Our results across several methods and for the three main credit rating agencies (Standard and Poor’s, Moody’s and Fitch) are significant and highly consistent. We find that we cannot reject the view that rating agencies add information. We find that this is true for all rating changes, upgrades and downgrades, as well as only for changes in asset classes (rating changes from investment to non-investment grade and vice-versa). We also find that less anticipated rating changes (i.e., we suggest that if the outlook change precedes the rating change by only a reasonably small number of days, then the rating change may not be fully anticipated) have even more significant effects. We conclude that rating agencies do matter in the sense that rating changes contain information that is not fully incorporated into the observed prices of financial assets – in this case sovereign debt.  In turn this implies that the appropriate functioning of the rating market is indeed a valid public policy concern.

4 Responses to "Do Ratings Matter?"

  1. 888   October 21, 2008 at 9:18 pm

    Recent events demonstrate that credit rating agencies conduct elaborate exercises of numerical and qualitative masturbation resulting in rating scores that have been proven to be worse than useless. The credit debacle is underpinned by the pillars of credit ratings made of smoke. Think I’m full of S–t? What were the credit ratings of the MBS (toxic waste) that are now up for TARP?Since realistic credit rating scores would detonate a moribund financial fiasco into a Nukular Explosion, there is no real support from anyone of authority to do anything but blow smoke.Fear this?I pity the many talented analysts who’s work as published is

  2. Lu   October 23, 2008 at 9:32 pm

    The problem with rating is that they are used to drive the risk models. It is assumed that b- is inherently more risky than AAA. This may not be the case since the ratings (esp for companies) are largely based on financial information and not more qualitative judgement, e.g. operational efficiency or integrity.Ratings may be useful as a guide to buying an investment but should not be used as one of the measures to calculate risk.Ratings form one of the quantitaive parameters used to caculate the risk when it should inherently be a dark art.What ever happened to gut instinct?