Ok, now that I’ve gotten the Miller-Moore battle out of my system, Reihan Salam, in trying to navigate a potential left-right program for building on and reforming Dodd-Frank, offers us Charles Calomiris’ short paper Beyond Basel and the Dodd-Frank Bill. Salam asks ”Is this a program around which we can build a [left-right] consensus?” In addition to specific creditor hits, there are three additional parts. It’s an interesting debate:
(2) Use loan interest rates when measuring loan default risk. Loan interest rate spreads compensate banks for the risk of default on the loan. If U.S. regulators had followed the example of countries that use interest rate spreads to measure loan risk, they would have required substantial additional capital to be budgeted against high-interest subprime loans in 2004-2007, which would have discouraged the overinvesting in housing ex ante, and would have insulated the banking system from the losses on those loans ex post.
(3) The SEC should reform the use of credit ratings to require NRSROs to estimate the probability of default, and provide that number as their rating for regulatory purposes, rather than give letter grades, and then hold NRSROs accountable for the accuracy of those estimates. Letter grades have no objective meaning, and thus, rating agencies cannot be held accountable for those letter grade ratings. If, instead, NRSROs were required to provide numbers, representing their estimates of the five-year probability of default on the debt, then regulators could construct (generous) confidence intervals for those estimates, and penalize rating agencies for grossly underestimating the probability of default.
(4) Establish a minimum uninsured debt requirement for large banks in the form of a specially designed class of subordinated debt known as “CoCos,” or contingent capital certificates. These CoCos would convert to equity based on observable market triggers (when the ratio of the market value of equity relative to the market value of assets falls below some pre-established threshold).
My thoughts in a different order:
(3) The letter grades from the rating agencies do in fact refer to specific probabilities of default, and there’s even well-quantified transitions between these states. Here’s a sample transition matrix I grabbed off the internet:
that you can plug into your various financial engineering state diagrams. The probability of default on a AA-rated debt over one year is the AA->D(efault) spot, or 0.01%. A quick monte carlo simulation will get you the five year estimate. (Trust it?)
The problem as I read it for the ratings agencies here isn’t that they said the number was 10% and it was 20%, it’s that they said it was zero and that it wasn’t zero (Notice how the one-year AAA->D is actually zero here). As Jerome Fons pointed out in this interview, there has always been conflicts with the ratings agencies. The new problem with securitization is that there would have been no demand for the assets unless they were AAA same-as-cash rated, so it wasn’t the same conflict as bumping a BB to a BBB. Whether this was because of regulatory arbitrage or demand for “informationally-insensitive” assets required for the repo shadow banking market (I currently find the second more convincing), the issue isn’t whether a slightly more accurate number would get us further.
The penalizing ratings agencies part is interesting. Right now putting them on the hook slightly through Dodd-Frank shut down new issuances in the bond market (which made Alan Greenspan gloat about the negative consequences of financial reform). No clue what is going to come of that.
(2) This I read as a call for CDS prices as default metrics for regulators and perhaps more broadly to replace the ratings agencies. Christopher Papagianis suggested similar things at Economics 21. I think it’s a good idea, but as one of many tools for regulators. For replacing the ratings agencies, it obviously doesn’t help you on new issuances where the information is most important. And for regulators there isn’t a ton of real early warning, as Stephen Lubben pointed out on Lehman:
Calomiris may also mean this as a response to the recourse rule, where a securitized bond went from being capitalized from what was in it (say subprime) to what the ratings agencies gave it (as if it was a corporate AAA bond if the ratings agencies gave it a AAA, even if it was full of the first claims to subprime). I’m torn on this. If it’s a bond it’s a bond, and it kind of makes the securitization process bad faith if we say it isn’t a bond. I’m for getting rid of this in terms of eliminating regulatory arbitrage, but worry it opens new doors. I’m good with closing some of the 2005 BAPCPA expansion of safe harbor for repo, which is probably equally important.
(4) The first step in evaluating coco bonds is to figure out how this is different than preferred stock, which performed horribly in the crisis. As Raj Date wrote in his excellent “Now More Absorbent! Five Principles to Make ‘Contingent Capital’ More Like Capital, and Less Contingent:”
From the point of view of senior creditors or depositors, in bad times, traditional preferred stock, in theory, should have functioned just like contingent capital: it should not have received dividends, and it would have been subordinated to a bank’s creditors and depositors….In reality, though, hybrid capital did nothing of the kind. The preferred market failed in every important way: its pricing mechanism failed to discipline banks’ risk-taking behavior (preferred stock tended to price relatively close to more senior debt, under the tacit assumption, encouraged by issuers and Wall Street alike, that banks would protect investors); it did not adequately absorb credit losses (banks continued to pay preferred dividends well after the magnitude of the crisis was clear); it created, instead of mitigated, systemic contagion between firms (because banks held large quantities of each other’s hybrid securities)…
Convertible bonds could easily fail. There are ways around this, like pre-writing the conversion structure, moving the debt towards traditional long investors, removing management discretion for triggers, a punitive level of conversion price, etc. Read Date’s excellent paper for more. But all of these take a bite out of the largest financial institutions. It makes these bonds more painful to issue when they are issued right – they’ll be costly to financial firms, and as such a ton of lobbying to make them toothless.
So in practice, they need regulators who are empowered to make sure that these coco bonds resemble their intended goal and aren’t simply being manipulated. But that’s going to required a stronger, better funded and less in-the-constant-crosshairs-of-Congress regulatory body than what we have now.