Conflicts of interest and reputation of rating agencies Jean Charles Rochet Toulouse School of Economics AMF conference, Paris 2 October 2008 This presentation is based on on-going research with Jérôme Mathis (TSE) and Jamie McAndrews (New York Fed). The views expressed here are those of the authors alone, and not necessarily those of the Federal Reserve Bank of New York or the Federal Reserve System.
Rating Agencies and the Crisis Credit Rating Agencies (CRAs) are accused of bearing a strong responsibility in the Subprime crisis, e.g.: Mason- Rosner (2007): How misapplied bond ratings cause mortgage backed securities and C.D.O. market disruptions. Many people argue that CRAs understated risks (in particular default correlation) Calomiris (2008) criticizes the «6% solution»: «CRAs assumed unrealistically low expected losses on subprime MBSs pools prior to the crisis and failed to timely revise them upward»
Conflict of interest? Bulk of CRAs revenue comes from the issuers: referee paid by one of the teams. Moody s net income grew dramatically with structured finance development ($159 m in 2000, $754 m in 2006, $701m in 2007) CESR report (May 2008), «the fee model for structured products is transaction based and dependent upon the completion of the rating process most respondent agreed that this fee model created a conflict of interest»
Rating Corporate Bonds Vs CDOs: what s the difference? In spite of initial skepticism of Basel Committee, there are important differences between rating corporate bonds (a well established methodology) and structured finance products: Different migration behavior: less downgrading for ABSs but larger magnitude Higher sensitivity to model error (systematic risk and default correlation, both very difficult to estimate) Different incentives for raters: stakes are higher for ABSs Goodhart (2008) «corporate rating fees are quite small-why risk the company s reputation to gain a fee?»
Rating Inflation: «unforced errors?» March 2007, FPC discovered that Fitch used a model that assumed home prices would appreciate forever! May 2008: Moody s acknowledged a bug in one of their ratings model WSJ (April 11, 2008): Brian Clarkson (was in 1999 head of Moody s structured finance business, and later became president) changed the code of conduct of the firm : it s important to socialize. Moody s market share of structured finance ratings grew in two years from 14% to 64%. Moody s was later forced to downgrade 31% of all asset backed CDO tranches it had rated and 14% of AAAs.
CRAs and reputation(1) CRAs argue they cannot be too lax since they risk losing their reputation (parallel with Enron/Arthur Andersen). We examine the validity of this assertion in a formal model (Mathis, McAndrews and Rochet 2008) By inflating its ratings, a CRA increases its current revenue but faces the risk of a decrease in future income, due to a loss in reputation: basic trade-off
CRAs and reputation(2) Terms of the trade off depend on several parameters like fraction of income that comes from structured products. We show that reputation concerns can discipline CRAs only if this fraction is small enough If fraction becomes large, we show that «honesty does not always pay» Reputation cycles may emerge: CRAs are initially very cautious (in order to build a reputation), but ultimately become lax (in order to cash in their reputation), increasing the probability of a collapse.
Reputation cycles Issuing volume (stock price of CRA) Reputation building Inflation rating Collapse Time
Source: Coval, J., J. Jurek and E. Stafford (2008) The Economics of Structured Finance
Moody s stock price (blue) VS S&P 500 (red)
Possible Policy Responses Public supervision of CRAs: seems difficult (complex) and counterproductive (why have regulators «out-sourced» the monitoring of banks and funds in the first place?) Make CRAs legally liable for their ratings (so far viewed as «opinions»): would probably kill the business (Goodhart 2008):Create an independent agency in charge of assessing private ratings (CRAAC)
How to eliminate the conflict of interest? Conflict of interest comes from the «issuers pay model»: CRAs want to please their customers, namely the issuers Difficult to return to the «investors pay model»: investors are dispersed and can easily resell the information. Ratings are a private good for issuers but a public good for investors (free riding, information leakage) Only possibility to eliminate conflicts of interest: cut the commercial links between issuers and CRAs; interpose a «platform», supposed to aggregate the interests of buy-side and sell-side
Our proposal: The platform pays model(1) Consider first securities that are traded on an exchange. Our proposal would work as follows: Any candidate issuer would pay a pre-issuing fee to the exchange. The exchange would then select one or several CRAs that would rate the security. After knowing the results of the rating process, the candidate issuer would decide to go on or not.
Our proposal: The platform pays model(2) Rating fees would be paid by the exchange to the CRAs. They would depend on the reputation of CRA but not of outcome of rating process and of the fact that the issue finally takes place or not. This would eliminate perverse incentives for rating inflation. This could solve the conflict of interest, provided the exchange is able to aggregate the interests of the two sides of the market. However this would not work if there is also a conflict of interest between fund managers and final investors
Our proposal: The platform pays model(3) What would happen for OTC transactions? Platform could be a Clearing House or a Central Depository Regulators should impose the view that if supervised institutions want to benefit from reductions in capital charges when they use complex financial instruments, they have to accept a certain degree of standardisation and centralisation in the issuance, clearing and settlement of these instruments. This would improve financial stability and transparency but also would promote a healthier business model for rating industry.
Conclusion The recent rating inflation behavior of CRAs was not accidental: it was a predictable consequence of the issuer pays model Reputational concerns are not enough to discipline CRAs as soon as the stakes are high Our proposal is to reform the industrial organization of the ratings industry, by making a central platform (Exchange, Clearing House or Central Depository) responsible for hiring CRAs More generally, giving more power to central platforms in the trading, clearing, settlement and rating processes would improve financial stability and transparency but would also promote a healthier business model for the rating industry.