by Sharyn O'Halloran
Here is an idea: why does the Fed not guarantee normal risk-return ratios from cataclysmic events like the collapse of a major counter party or aggressive naked short selling, much like FEMA is supposed to do with natural disasters? (OK-- bad example, but you get the idea.) These are exactly the cases when the normal functioning of markets fails and government can play an effective role.
In the current crisis this would mean that government would set aside funds for when a large institution's portfolio comes under attack, perhaps funded by premiums. Just like FEMA won't insure property that's too close to a flood zone, you do not have to back sub-prime or Alt-A loan vehicles, which raise the risk of moral hazard. Instead, they can guarantee those parts of the portfolio that represent normal risk ratios—perhaps those designated similar to qualified loan assets or investment grade.
If an institution comes under stress (like AIG), the guarantee/insurance on the core of its portfolio would minimize risk, making it less likely that it will be downgraded, which in turn prevents a run on its swap contracts and allows it to borrow from the markets at more reasonable rates so that it can raise capital. Maybe these can be written into loan agreements. This seems like the right balance between market-based incentives and regulatory intervention, without being pro-cyclical like higher reserve requirements.
As the devil is in the details, here are some details:
1) Measure of Risk: The risk-return ratio is just one measure of the riskiness of an investment; you could insert any other standard measure here (variance of returns, volatility, or expected absolute deviation, for example).
2) What is too risky? Take the market distribution of riskiness of assets, find its mean and then find the point that is, say, two standard deviations above the mean. We'd define investments with this level of risk or greater as too risky to be insured (the equivalent of building your house on the banks of a river that floods often).
3) Covered Assets: All other investments would be insured against catastrophic insurer default, meaning that if you have some type of insurance on the asset like a credit default swap, and the counterparty disappears in a financial tsunami, the government will step in and insure your assets. If you just make a bad bet, though, and you lose your money, the government won't insure you against that.
4) Moral Hazard: While this proposal similar to the money market insurance proposal, but those investments would be insured no matter what, similar to the role FSLIC plays with banks. Here, you can lose money and not be compensated. And so the moral hazard issues are mitigated, at least as far as the insurers themselves, who could, it's true, evaporate and still have the companies they insure be made whole.
5) Additional questions:
a) If a company uses best practices risk management should non-insured assets be covered as well?
(My initial take is that if you comply with Basel II risk sensitive reserve ratios, then all assets within the covered group would be insured.)
b) What would be an appropriate fund and premium schedule, e.g., risk-based vs. asset-based?
(I’m inclined toward a combination of the two.)
c) What would be the source of funding?
(I have the Resolution Trust Corporation model in mind here. They put into reserve about 1.25% of total deposits against default. I could see the same here. Say if total assets covered by participating institutions were $100 trillion. A 1% fund would place $1 trillion in reserve.)
d) Does this refer to catastrophic market risk or would it include credit risk as well?
(I am thinking of the former; most of the credit risk could be managed by an exchange for credit swaps and derivatives as opposed to over the counter deals.)
e) How can you take into account the possibility of correlated error (if one house goes into foreclosure, it makes it more likely that other homes will as well)?
(One of the main problems is determining the riskiness of the underlying asset. Each instrument would have attached to it a due diligence report that verifies the underlying portfolio risk. Usual practice would be that with a portfolio of 100,000 mortgages, for instance, you would randomly sample 1000 and verify the information.)
Dr O'Halloran, I would be interested to know if you have read Nassim Taleb's books on randomness, and whether his ideas factored into your commentary above. I ask because it is painfully clear to me that risk models such as those used for flood frequency and economic meltdowns are woefully ineffective. May I also refer you to this link addressing the $1.64B fine that AIG paid as a result of accounting practices that E Spitzer found unacceptable?
http://www.financial-planning.com/asset/article/527390/aig-and-spitzer-announce-landmark-settlement.html
Tony
Posted by: Tony Facade | September 28, 2008 at 03:40 PM