The WSJ this weekend came out with its post mortem of Sowood Capital, the hedge fund started by former Harvard high-flyer Jeffrey Larson which imploded spectacularly in July. There's nothing particularly surprising in the story (too much leverage, not enough capital), but I did get an email from a reader asking about this passage:

By the beginning of this year, Mr. Larson was worried about many kinds of riskier debt investments, according to people familiar with the situation. To protect himself and take advantage of those risks, he bought senior debt securities and sold short, or bet against, a range of investments generally viewed as more risky, such as junior debt securities and various stocks.
Mr. Larson's strategy depended heavily on using borrowed money, or leverage. Because he was betting on small movements -- such as whether a company's senior debt would go up more than its junior debt went down -- he borrowed as much as six times the firm's capital to generate respectable returns when his bets were right.

My reader asks, sensibly enough:

The first paragraph seems to be referring to senior and junior debt of different issuers. However the second paragraph seems to be talking about senior and junior debt of the SAME company. Under what circumstances would a company's senior and junior debt move in opposite directions?

It's a good question. And the simple answer is that markets are unpredictable, and that just about anything can happen, often for no good reason at all.

But there is actually a reason why the Sowood trade went sour, and it centers on the whole phenomenon of highly-rated debt. Many investors are very risk-averse, and buy only debt with high credit ratings; they often restrict themselves to AAA-rated securities. Indeed, the demand for AAA-rated paper is so high that an entire securitization industry has essentially sprung up in order to create enough supply of such stuff, which always trades at tighter spreads than capital-structure theory would imply.

Or always used to, anyway.

In July, a lot of market participants lost faith in the credit ratings in general, and in their AAA ratings in particular. Suddenly, it came to light that some (not all) AAA-rated securities were much riskier than the markets had previously thought. People had understood the risk inherent in the lower-rated tranches, and so they paid lower prices for them, or demanded higher coupons. So to a certain extent the risk was priced in, with those. But because AAA-rated securities were erroneously considered risk-free by some of their buyers, they often traded with no credit-risk premium embedded. And to make matters worse, many of those securities were also extremely illiquid.

What's more, risky securities are generally held by investors with some non-zero risk appetite. If they go down, they go down: that's a known risk. But AAA-rated securities are often held by investors with very, very little risk appetite. If they go down, those investors are liable to want to sell, and sell quickly.

And then comes the icing on the cake: the fact that in most securitization structures, there are many, many more AAA-rated bonds than there are bonds lower down the ratings scale. If some small percentage of the holders of BBB-rated bonds, for instance, decided to sell, the impact would be relatively low, because the total value of the sale would be small, and some dealer somewhere would probably happily make a market in that security. But if a similar percentage of the holders of AAA-rated bonds decides to sell, then all hell can break loose, because now we're talking large sums of money.

Oh, and did I mention that AAA-rated bonds had recently become flavor of the month among highly-levered hedge funds like Sowood? Leverage, of course, always increases risk.

Put all that together, and it's easy to understand why highly-rated debt might crater overnight, even as lower-rated securities emerged relatively unscathed.

But in any case in order for Sowood to lose a lot of money it wasn't necessary for an issuer's low-rated debt to move up while its high-rated debt fell. This was just a relative-value trade, and so all that was needed was for the spread between the low-rated and high-rated debt to narrow, rather than widen. So long as the high-rated debt fell more than the low-rated debt, Sowood was in trouble.

Of course, I doubt that the WSJ reporters know exactly what kind of trades Larson was putting on when his fund blew up: Larson refused to talk to them, and it could be that they got some of the specifics wrong. But the big losses this summer were generally suffered by funds which went short risky securities (small-cap equities, low-rated bonds) and long securities which were perceived to be safer, like large-cap equities and high-rated bonds. So what the WSJ reported rings true to me.