Risk Rating Reluctance

Managers of many financial organizations are arguably tempted to take too much risk with organizational investments.  Reputation-wise, such managers often gain more from stellar investments than they lose from disastrous ones.  Many regulations try to address this problem by limiting such organizations to “safe” investments, as determined by a few official ratings agencies.  This creates a demand by such managers for assets that are actually risky, but which are officially rated as safe.

Apparently the recent financial crisis was due in part to those official risk rating agencies supplying this demand; they are private firms and profited from applying too little skepticism to whether certain key assets really were as safe as some claimed.  No doubt those agencies say that this was all a terrible accident, that they never intended to profit from mistakenly calling risky things safe, but since they seem to have suffered little from the harm they assisted in creating for others, I have my doubts.

In all the financial reform discussions, I haven’t heard any proposals to address this specific problem.  And I’ve always wondered why we need investment ratings agencies anyway.  In principle, the right financial assets can give any elements you like from a full joint probability distribution over asset returns; how could a ratings agency expect to do consistently better?

At lunch today I asked my wise colleague Garett Jones about this, and he suggested that big financial orgs like being rated by people they can pressure. If they don’t like a rating, they can work their elite-school-alum networks of contacts to apply pressure to the ratings agencies to change unwanted ratings.  Such pressure is much less effective on financial market prices.  Garett also pointed me to this passage of his:

[Remember] the debate over subordinated debt in the early 2000’s, surveyed in Stern and Feldman’s Too Big to Fail. The Gramm-Leach-Bliley financial reform bill attempted to create a class of subordinated debt that would be explicitly banned from any future bailouts.  Major financial institutions would have been be required to hold some portion of their liabilities in the form of subordinated debt in order to give financial markets and regulators alike a market-based measure of firm health: If yields on a major firm’s subordinated debt spiked, that would be a warning sign.  But financial institutions and the Federal Reserve Board both pushed back against this market-based indicator, and so the subordinated debt requirement never made it through the regulatory process. … Firms will resist issuing debt that is bailout-free, and will overwhelmingly prefer debt that is bailout-qualified.

Bailouts mess up the connection between asset prices and their inherent risks.  Ordinarily, market prices only tell you the chance that a debt will be paid, not whether it would have been paid without a bailout.

This is all moderately bad news for prediction markets in such firms.  Apparently well-connected managers already know they prefer estimates by officials who respond to social pressure, over hard-to-manipulate market estimates, even if the later are more accurate.  Of course less well-connected managers should prefer the opposite, but who wants to signal their bad connections by endorsing independent markets?

Added 8a: Unnamed points us to Matt Yglesias responding in Sept. to Kevin Drum and an August Policy Report by Mark Calabria.  Kevin says:

Over the past decade ratings agencies were, at best, negligent, and at worst, perpetrators of outright fraud.

but doesn’t think raters were a big problem because other orgs used similar risk models and both buyers and sellers liked the mis-labeling.  Yet this is just what a corrupted regulator model predicts.  These folks consider switching who pays the raters, or making them a direct government agency, but not replacing them with direct market price risk estimates – why so blind to such an obvious solution?