Thursday, October 30, 2008

Harvard/Princeton Paper on Mortgage Backed Securities

In my numerous discussions on the cause of the housing crisis, one of the ideas that I have come to believe is that human beings have a tendency to extrapolate. This might seem trivial, but it is where I differ from the Milton Freidman school of thought.

Not since the Great Depression had home prices fallen considerably nationwide, so believing that home price increases would stay positive even given the obvious bubble, was rational, but incorrect. Homebuyers, occupants and speculators alike, were extrapolating the recent abnormally high returns and assuming no downside risk. As I said here, “What naïve investors often do is look at trends (i.e. they extrapolate).”

In a May op-ed for Real Clear Markets, I wrote:

“Ratings are given to mortgage-backed securities based on backward looking analysis of defaults. In an environment where rapidly rising home prices mask foreclosures, risk premiums were too low and values too high on these securities. This practice had been a reliable model due to decades of steady trends. “

A paper from two professors of the Harvard Business School, and another from Princeton write a thorough examination of the credit crisis. (p. 21,23)

First, the securities’ credit ratings provided a downward biased view of their actual default risks, since they were based on the credit ratings agencies’ naïve extrapolation of the favorable economic conditions. Second, the yields failed to account for the extreme exposure of structured products to declines in aggregate economic conditions (i.e. systematic risk).

In March 2007, First Pacific Advisors discovered that Fitch used a model that assumed constantly appreciating home prices, ignoring the possibility that they could fall.

So we have homebuyers extrapolating, and we have rating agencies extrapolating. That's a whole lot of extrapolating going on, to ignore it as part of economic theory.


Summary of the Rest of the Paper

The paper is a little dense if you aren't familiar with higher finance, but the other main point they make reveals that the basic model for pricing risk was flawed. Mortgages were split up into two or more parts: The junior tranche, the risky part, and the senior tranche, the safe part. If someone defaulted, the junior tranche had to pay first. The junior tranche was like the deductible in an insurance policy. If defaults got too big, then the senior tranche would have to pay the rest.

Like life insurance, it was believed that defaults would be fairly well dispersed across the country and were caused by the same typical random financial problems that always cause people to foreclose. The pricing completely ignored the systematic risk of a broad national drop in home prices. Furthermore, the riskiest of sub-prime mortgages and their tranches were far more often to originate in California, Nevada, Arizona, and Florida, the worst victims of the housing crunch.

No comments: