Thursday, November 20, 2008

Economist's Notebook: Risk


Nobel winner Michael Spence gave a talk on the current crisis that inspired this rumination on risk.

Risk in the case of the meltdown of the balance sheets of the world’s most important financial institutions is quite different than the type of risk that financial institutions and insurance agencies were used to dealing with. What characterizes what we might term “normal risk” are three things: it is exogenous, stationary and uncorrelated. What this means is that risk is not affected by the actions of the participants in the market, that the risk is not changing through time and that the risks are not correlated with each other. House insurance is like this. Consider an insurance market for hurricane risk in the Gulf of Mexico. The risk of Hurricanes is not affected my the actions of participants (though the risk of loss is due to moral hazard – those with insurance are less likely to do things to prevent damage when there is severe weather –but this is fairly straightforward to deal with), global warming aside, the risk of a hurricane is fairly stationary meaning it is not changing through time very much, and risks are uncorrelated (a hurricane in Florida does not make a different hurricane in Texas more likely) thus the market works through diversification (no company should concentrate on only one location like, e.g., Homestead Country in south Florida).

The risks involved in securitizing assets and insuring them had none of these aspects. They were endogenous – the behavior of participants in the market significantly altered the risk profile of the assets, from underwriters to credit rating agencies to the institutions themselves. They were non-stationary – the risks were getting worse and worse through time (and quite rapidly). And they were correlated – more risky MBSs made for more risky CDOs and on and on. Thus these institutions were faced with new and non-standard risks that one could argue they were simply not able to deal with. But I think the true answer was that there was such strong short-term pressure to be willfully ignorant that this is what they remained. The profits made on these new securities were so large that to not take part in the market or to pull back at or near the peak would have been very difficult (especially for publicly held companies). This risk was also poorly understood by credit rating agencies and regulators - even when some in the industry raised the alarm, they were largely ignored. Understanding this mew type of risk is key to understanding the way forward.

1 comment:

Greg T said...

Nice observations on risk. Although non-standard risks are difficult to plan for and manage, we must do so nonetheless. Risk models that breakdown when risks are endogenous, moving, or correlated is like building a car whose seat belts when the driver falls asleep, or whose car bag fails over 10Mph.

Transparency is the issue. Risks cannot be managed if they cannot be measured. If government is to become the insurer of last resort, then it has the right (and obligation) to ensure transparency of the markets. The CDO and DSO markets are very opaque, to the benefit of nobody except a small handful of inside players.