09 July 2009

The Trouble With Regulating Risk Concentration

In my view one of the main factors behind the severity of the financial crisis was the excessive concentration of aggregate risk in highly-leveraged financial institutions. Note that the emphasis is on the concentration of aggregate risk rather than on the much-hyped leverage. The problem in the current crisis was not leverage per se, but the fact that banks had held on to AAA tranches of structured asset-backed securities which were more exposed to aggregate surprise shocks than their rating would, when misinterpreted, suggest.

From here.

I've alluded to the issue a couple times when talking about FDIC regulation, which limits not only leverage per se, but also places some limits on types of investments permitted, which is a form of aggregate risk regulation.

The trouble is that this observation isn't too helpful from a regulatory perspective. Leverage is easy to monitor and regulate, even with rather complex risks. State insurance regulators and the Pension Benefit Guaranty Corporation both engage in this kind of reserve setting, and while it isn't perfect and can be stressed in extraordinary times, reserve regulation reduces systemic risk across whole industries, by imposing, if not more prudence than a typical industry executive would undertake, at least more prudence than a reckless industry executive would prefer, which in turn, discourages a race to the bottom scenario in terms of leverage (with part of the cost of any collapse paid by others).

No matter how much or little risk entities take on, when those risks go bad, entities with less leverage will be better equipped to absorb those risks than entities with more leverage. Even if leverage regulation doesn't prevent a bubble collapse or financial crisis, it will make it less severe.

Excessive concentration of aggregate risk, in contrast, is very hard to regulate directly.

Prudent investor rules, which require diversification of investments for fiduciary funds, help prevent concentrated risks in fiduciary investments, but it probably isn't reasonable to expect every player in an industry to be diversified the way that a fiduciary would be diversified. In the same vein, index funds work only because an important subset of the market doesn't index. The financial markets are all about taking risks on unbalance portfolios based upon relevant information that shows that current prices are wrong. Moreover, diversification often makes more sense at the investor level than at the investment vehicle level. Undiversified investment vehicles provide ways for ultimate investors to diversify. Nobody expects a mortgage backed security fund to diversify away from mortgage backed securities.

The AAA bond ratings assigned to bonds that, in fact, carried much more risk than a typical AAA bond illustrate how much skepticism and "read between the lines" effort is necessary for a regulator (who is probably very prone to regulatory capture) to distinguish between apparent risk and actual risk.

Transparency also has its limits. If AIG had disclosed all of its derivative positions, it would have done so in a couple of ways.

First, it might have disclosed the "worst case scenario" liability of its credit default swap obligations. But, given that these were credit default swaps on what were often investment grade bonds that the derivatives turned into AAA bonds, the numbers would have been simply a curiousity. Why use "worst case scenario" losses for a guarantee of a loss exceedingly unlikely to come due all at once? This would be like reporting the face value of all of a company's outstanding life insurance as a liability on the insurance company's books. Every once in a while, this number may even be meaningful, e.g., if you are Hiroshima Life and Casualty Co. when the Hiroshima bomb explodes, or write life insurance exclusively for gay men in California just before the AIDS epidemic sweeps in and dramatically increases mortality rates. But, investors and regulators are mostly going to right off this number as absurd. If they didn't, the industry couldn't exist at all.

Second, it might have disclosed anticipated risk. Somewhere out there, some CPA working for AIG prepared a document with an aggregate expected loss for its credit default swaps (probably based upon the bond rating assigned to the underlying investment whose default was covered) lined up against the price customers paid for those credit default swaps. This chart no doubt showed both a profit for AIG on each transaction and left AIG in the black on a balance sheet basis.

This is better than nothing. It isn't clear this kind of loss reserve disclosure, common at commercial banks and insurance companies, was on the books at all, and it would have at least provided a baseline from which to draw meaningful conclusions. But, this kind of analysis still doesn't solve the problem that sometimes the underlying assets that a credit default swap guarantees is riskier than it appears (indeed, adverse selection principles from insurance contexts makes it likely that the risks are higher than they appear), and it is quite unlikely that anyone would do the kind of counterparty default risk analysis needed, in part because the data was hard to secure and in part because counterparty risk normally seems remote.

Perhaps the best that one could do is to give someone smart with lots of easy access to good data, like a private mortgage insurance company, a very strong financial incentive to accurately assets the important components of downside risk, like poor underwriting and excessive housing prices in a market in the case of investments derived from mortgages.

Still, this was basically present in the mortgage backed security market. Typically, a mortgage backed security had standing behind its loans: (1) personal obligations to pay of the debtors, (2) brick and mortar collateral, (3) default rate triggered promises from the mortgage finance companies that supervised underwriting, (4) state law fraud liability for inaccurate applications and appraisals, (5) credit default swap guarantees from large financial companies that had been in existence for a long time, (6) reinsurance of credit default swap counterparty risks from large financial companies that had been in existence for a long time, and (7) the law of averages which made a simultaneous default at a level beyond that seen within the historic period within which good records were available extremely unlikely.

Defaults have cost debtors their homes and their credit. They have resulted in huge waves of foreclosures attempting to realize on the brick and mortar collateral. They mortgage finance companies turned everything they had over to their creditors in satsifaction of their contract obligations in the event of excessive defaults. Thousands of people who engaged in mortgage fraud or questionable disclosure practices have been prosecuted, faced suits for civil liability, or otherwise been financially ruined. The financial companies the issued the credit default swaps have taken massive losses, with many of them going out of business or getting federal bailouts that have decimated financial company equity.

While the housing bubble become obvious well before it collapsed, quantifying how much of the fallout from its inevitable collapse would be felt by investors in mortgage backed securities screened by layer affter layer of reassurances and risk mitigators was very difficult to do in advance. No regulatory agency can be trusts to catch this kind of Black Swam when it comes up through mere analysis and detective work.

Excessive concentration of aggregate risk is absolutely a key problem that goes to the heart of what caused this financial crisis and past financial panics. But, regulating it directly comes close to being a fool's errand. Pity the fellow assigned the job of systemic risk regulation at the Fed when the next crash strikes in some unexpected way.

This isn't to say that systemic risk can't be reduced. There are multiple sensible policy measures that can be taken. But, in order to work, systemic risk regulation must be indirect. Expecting regulators to identify problems waiting to happen before its too late is unrealistic. Instead, the regulation should give the smart people with inside information and insight who are in the game an incentive to avoid excessive risk, or, at least, to isolate the risk in the hands of ultimate investors who understand the risks they are taken and can afford to take them, so that losses don't spread and infect the entire financial system and economy.

In other words, the way to regulate excessive concentrations of aggregate risk is to implement measures that are broadly applicable across the economy, that encourages steps that make our economy more robust in hard times. This means removing artificial incentives to use leverage, and it means giving key economic actors, like financial firm managers, incentives that fairly weigh upside and downside risks. It also means identifying and discouraging arrangements and situations that turn independent random risks into synched correlated risks without intense and prudentially conservative regulation.

While preventing bubbles, or identifying the current existence and extent of specific economic risks, or promoting non-stop growth are goals that are nearly impossible to attain, using public policy to encourage actors in the economy to act in ways that make the economy generally more robust in bad economic times than it is now, does seem like an attainable goal.


Michael Malak said...

"it means giving key economic actors, like financial firm managers, incentives that fairly weigh upside and downside risks."

Or to put it in plainer language:
* No bailouts
* No bonuses
* No moral hazards

Andrew Oh-Willeke said...

Well, mostly, no stock options.