Regular commenter Doug made a particularly insightful comment to my post last week about safe assets. His basic point is that
Many people would be able to shift economic consumption across time and different states of the world. If I’m a pension fund I want to secure access to aggregated medical care for my pensioners between 65 and death. If I’m a homeowner I want to obtain be able to get a new house in case of natural disaster. If I’m saving for retirement I want a condo in Florida 30 years from now.
It is the function of the financial system to do this.
We simply choose our maturity, risk, terms of payment, etc from an array of standard contracts at listed prices. Then … we can rely on major banks or insurance companies to deliver what they’ve promised 99.99% of the time.
We can’t expect this to happen 100% of the time. If the Borg make it to earth, or the Yellowstone caldera blows, or there is a civil war in China, all bets are off.
The point, though, is that there are tail risks which we have a choice about. We can either back them by the state (assuming that the state can afford them), bear them without backing, or redesign the system to mitigate them. A good example is pensions: we try to help people save for retirement through a combination of regulation and tax breaks, but we also provide the pensions protection fund to mitigate the tail risk of fund insolvency.
The key observation in the safe asset hypothesis (that the supply and demand for safe assets are important macro economic variables) is that if genuinely safe assets are not available to support a financial system function, then pseudo-safe ones will be substituted, possibly leading to disaster – or at least calls for a bail-out. (Some) states can fix this by proving safety, either in the form of a guarantee, or by providing more safe assets. I’m not saying that they always should: merely that there is a choice about which risks the government backstops, and that by observing the risks that the system tries to hedge, useful macro prudential information can be gleaned.
I see CCPs as a (flawed) attempt to move risk further out in the tail. Instead of the failure of one OTC derivatives dealer, we have the failure of enough of them to bring down the CCP. Naively* we have substituted a higher probability lower (not low) impact risk for a lower probability higher impact one. The interfluidity proposal on stochastic failure I posted about a week or so ago is equivalent to the argument that we should go the other way – try to create a financial system with more, lower impact failures.
Doug’s right, we can’t reduce failure probability of either institutions or the system to zero. But we can change both who bears the costs of mitigation, and how much mitigation, to what risks, is done. Yes, a sufficiently big shock will cause any system to fail, but clever design can help improve robustness under less severe shocks. In many ways we over-emphasize the study of crises: we also need to remember when something went wrong and the system coped, for those occasions have valuable lessons too.
*Whether it is lower probability or not is of course debatable.