Sunday 5 October 2008

Experts can be bad for your economy

This point is made by David M. Levy and Sandra J. Peart in an article at reasononline. Levy and Peart call their piece An Expert-Induced Bubble: The nasty role of ratings agencies in the busted housing market.

They open their article by asking a question many people are asking given the current financial crisis,
How is it that assets built out of mortgages which just yesterday were worth so much are worth so little today?
Levy and Peart explain,
Experts are constantly telling investors what to buy. Sometimes they give us good advice and sometimes not. So surely the fact that there are experts who give investment advice can't explain the trillion-dollar bubble and subsequent meltdown we're now witnessing. The key additional fact is that experts were selling advice about mortgage-backed assets as if those assets were independent when, in reality, they weren't at all independent assets. Only once investors realized that the housing market is a national market—not a local one—did it become clear that these securities were extraordinarily risky. Hence the collapse.
They then go on to note an important point about housing markets in the US,
Until very recently it was widely believed that all housing markets were local. If this were so, then assets constructed by pooling mortgages across different localities would consist of pooled independent assets. And these new assets would be dramatically less risky to hold than a single mortgage of similar worth: Combine a bunch of diverse mortgages and sell shares of the new security and those shares represent much less risk than holding a single mortgage of the same value as the share. Or so the story went.
But all of this depended on housing markets being local. Critically it was thought that housing markets being local meant that the assets in the pooled security didn't move together. Even Alan Greenspan argued this.
[He] testified to this effect before Congress in 2005, when the housing bubble was well under way: "The housing market in the United States is quite heterogeneous, and it does not have the capacity to move excesses easily from one area to another. Instead, we have a collection of only loosely connected local markets."
Why does this matter? Levy and Peart explain,
... following Greenspan's advice, a firm could build highly rated investment portfolios of purportedly uncorrelated assets out of nothing but mortgages from different parts of the country. Once these portfolios were built, it would become easier to finance houses even for buyers of dubious credit. The problem was that these new securities, and the money which flowed into all housing markets, were sufficient to generate correlation in housing values across the country. As everyone followed the experts' advice—and invested in these new mortgage-backed assets—we began to observe correlated behavior in the housing market, nationwide.
This raises another question, So how did the securities maintain their high investment grades? Once these correlations were observed, once people realised that there were interconnections between housing markets nationwide, why didn't another set of experts, the rating agencies, step in and downgrade the securities? Clearly these securities were not as safe as was being assumed.

The answer, like so much of economics, has to do with incentives.
In this case, the incentives weren't there to obtaining unbiased estimates of security values. Instead, incentives favored "rating shopping" and so, unsurprisingly, rating shopping became the norm. The Securities and Exchange Commission's 1994 report, Concept Release: The Nationally Recognized Statistical Ratings Organization (NRSRO), contained the following sentences:"A mortgage related security must, among other things, be rated in one of the two highest rating categories by at least one NRSRO." The phrase "one of the two highest rating categories" authorized the firm holding a mortgage backed security to shop for ratings. If one rating agency failed to produce a desirable rating, the firm could look for another, more favorable rating.
So the firms could get the ratings they needed, without the buyer of the securities knowing about any other lower rating. This meant that the buyer didn't have the ability to average out different ratings to help remove the bias caused by the rate shopping. Levy and Peart note,
As long as experts were trusted and the market didn't know the difference between unbiased and biased estimates, the trick worked marvelously. The collapse followed suddenly as we have all come to understand that the ratings were miserably biased.
Upshot, don't necessarily trust the experts, or at least don't trust just one expert.

2 comments:

Eric Crampton said...

David tells me that he was called to testify on Capitol Hill because of that article....

Paul Walker said...

Cool. But I can't help thinking the only thing worse than the advice of the experts is, government trying to "fix" the advice of the experts.