Bolton, Freixas & Shapiro (2012)

Citation: Bolton, P., Freixas, X., & Shapiro, J. (2012). “The Credit Ratings Game.” Journal of Finance. (Working paper version 2009)

Bolton, Freixas, and Shapiro model the fundamental conflicts of interest in the credit rating industry. The paper addresses why rating agencies systematically understated credit risk in the run-up to 2008, and evaluates regulatory proposals to fix the industry.

The model’s key elements:

1. Issuer-pays creates conflicts. CRAs are paid by the firms whose products they rate. This creates obvious incentives to inflate ratings to attract business.

2. Issuers can shop for ratings. If a CRA gives an unfavorable rating, the issuer can refuse to publish it and solicit another. The market doesn’t see rejected ratings.

3. Two types of investors. Naive investors take ratings at face value. Sophisticated investors understand CRA incentives and discount accordingly. The fraction of naive investors (α) is a key parameter.

4. Reputation provides some discipline. CRAs face future costs if caught lying—but only if the investment defaults AND investors can trace the failure to the CRA’s misrepresentation.

flowchart TD
  subgraph CRA Incentives
    SR[Short-run: fees from issuers]
    LR[Long-run: reputation costs if caught]
  end
  SR -->|Favors inflation| I[Inflate ratings]
  LR -->|Disciplines| T[Tell truth]
  
  subgraph Key Parameters
    N[Naive investor fraction α]
    R[Reputation cost ρ]
    P[Signal precision e]
  end
  N -->|High α| I
  R -->|Low ρ| I

Central results:

1. CRAs inflate ratings when naive investors are prevalent and reputation costs are low. The paper derives a precise cutoff: inflation occurs when the gains from fooling naive investors exceed expected reputation losses (probability of default × reputation cost).

2. Ratings inflation is pro-cyclical. In booms, more investors are naive (less incentive for due diligence), and reputation costs are lower (harder to attribute failures to CRA error when everything is rising). This predicts exactly what happened pre-2008.

3. Competition can make things worse. This is the paper’s most counterintuitive result. A CRA duopoly can be less efficient than a monopoly in terms of both total surplus and investor surplus. The reason: with two CRAs, issuers have more opportunities to shop. They can solicit both ratings, publish only the favorable one(s), and exploit naive investors more effectively.

flowchart LR
  subgraph Monopoly
    M1[One rating] --> MS[Less shopping opportunity]
  end
  
  subgraph Duopoly
    D1[Two ratings] --> D2[Publish best one only]
    D2 --> DS[More shopping, more exploitation]
  end
  
  MS --> MW[Higher welfare conditional on same info regime]
  DS --> DW[Lower welfare due to shopping]

1. Tranching and restructuring make it worse still. For structured finance products, issuers can restructure deals to get better ratings. This is pure rent extraction from naive investors—no real value is created by the restructuring, but it convinces investors to overpay.

Regulatory analysis:

The paper evaluates three reform proposals:

1. Cuomo Plan (upfront payment). Requiring issuers to pay CRAs before seeing the rating eliminates CRA incentive to inflate. But it doesn’t stop issuer shopping—issuers can still refuse to publish unfavorable ratings.

2. Revised Cuomo Plan (upfront + mandatory disclosure). Adding a requirement that any rating produced must be disclosed eliminates both CRA conflicts and issuer shopping. This achieves the efficient outcome but has an unintended consequence: it makes a second CRA worthless to issuers, so the market reverts to monopoly.

3. Investor-pays model. Funded by transaction tax, this could work as well as the revised Cuomo plan but requires substantial government supervision.

The moral hazard problem:

All three reforms share a flaw: if CRAs are paid regardless of rating quality, they have no incentive to invest in information-gathering and due diligence. The reforms solve the inflation problem but may create an accuracy problem.

Key implications:

• Entry and competition in rating markets is not obviously beneficial. More CRAs can mean more shopping opportunities, not better information.

• The issuer-pays model has inherent conflicts that cannot be fully eliminated without mandatory disclosure.

• Rating accuracy requires some skin in the game for CRAs—either through reputation (which requires naive investors to eventually learn) or through regulatory oversight of methodology.

• Pro-cyclical ratings inflation is a structural feature of the industry, not just a failure of individual CRAs.

CRA conflicts of interest, issuer shopping, and naive investors interact to produce systematic ratings inflation—and competition can exacerbate rather than solve the problem. Regulatory fixes must address both CRA incentives and issuer shopping, but doing so creates new moral hazard problems. The failure of rating agencies in 2008 was structural, not accidental.