The hypothesis that securitization spreads risk is a cliche. It might even have some truth to it. The problem is the price. Consider. Start with a group of individual loans representing an aggregate of $X of debt. What sort of markup is necessary to justify turning those loans into a mortgage backed security (MBS)? I don't know, but if it wasn't a profitable activity, nobody would have done it, at least not unless they were obliged to liquidate the individual loans to raise capital. So now the package is worth some function of X, aX, with a>1. The same question arises in packaging the MBS into a Collateralized Debt Obligation (CDO). So, now the original package has a new aggregate value, abX, with both a and b>1.

There isn't anything wrong with the concept. It's just a version of buying, loans in this case, in bulk wholesale and selling the pieces retail. That is, there's nothing wrong until the merchandise turns out to be defective and the customers are lined up at the customer service department to return it. If the customer service department isn't available, maybe they go to eBay or Craigslist. In the former case, the merchandise is going to have to be marked down by the returns department to move it if it can't be returned to the supplier. In the latter case, is there enough liquidity to absorb the supply at all, and if so, at what price? Probably it will not be retail.

Previously my focus has been on X and the property securing it. It is possible that the loans were never worth X because of bad underwriting, even erroneous valuation of the underlying security. Let us say that the underlying security, P for Property, was believed to be something like P=X/.9. It is even possible that some allowance was made for default, collection and so on in the original securitization process. Let's use the factor c, with c<1. So the package now has a value of abcX. But nobody probably knows whether abcX is equal to, greater than or less than X. My bet is abcX is at least equal to and probably greater than X. How much greater is dependent on efficiency of pricing, competitiveness. Just how competitive was this process?

What nobody knows yet is whether when the supply and demand curves for P shift to reflect increased supply (overbuilding and default/foreclosures) as well as reduced demand (tighter underwriting/higher mortgage rates) what relationship P will have to abcX. Depending on just which X you're talking about the relationship is likely to be different. But that is looking beyond pricing a security to valuing THE security. So far that topic is not widely discussed.

Now, what if the buyers of abcX have borrowed money to make those bets? Are we surprised that we're seeing volatility where the original idea was to spread risk?