Streetwise Professor

Yesterday the WSJ published a long article about Morgan Stanley’s near death experience in September.   Although the byline reads Susan Pulliam, Liz Rappaport, Aaron Lucchetti, Jenny Strasburg, and Tom McGinty, it reads like it was ghosted by Chris Cox because it raises the specter of evil speculators ganging up to destroy Morgan by buying credit protection against the bank, and short selling its stock:

Trading records reviewed by The Wall Street Journal now provide a partial answer. It turns out that some of the biggest names on Wall Street — Merrill Lynch & Co., Citigroup Inc., Deutsche Bank and UBS AG — were placing large bets against Morgan Stanley, the records indicate. They did so using complicated financial instruments called credit-default swaps, a form of insurance against losses on loans and bonds.

A close examination by the Journal of that trading also reveals that the swaps played a critical role in magnifying bearish sentiment about Morgan Stanley, in turn prompting traders to bet against the firm’s stock by selling it short. The interplay between swaps trading and short selling accelerated the firm’s downward spiral.

. . . .

Pressure also mounted on another front. There was a surge in “short sales” — bets against the price of Morgan Stanley’s stock — by large hedge funds including Third Point LLC. By day’s end, Morgan Stanley’s shares were down 24%, fanning fears among regulators that predatory investors were targeting investment banks.

. . . .

New York Attorney General Andrew Cuomo, the U.S. Attorney’s office in Manhattan and the Securities and Exchange Commission are looking into whether traders manipulated markets by intentionally disseminating false rumors in order to profit on their bets. The investigations also are examining whether traders bought swaps at high prices to spark fear about Morgan Stanley’s stability in order to profit on other trading positions, and whether trading involved bogus price quotes and sham trades, people familiar with the probes say.No evidence has emerged publicly that any firm trading in Morgan Stanley stock or credit-default swaps did anything wrong. Most of the firms say they purchased the credit-default swaps simply to protect themselves against potential losses on various types of business they were doing with Morgan Stanley. Some say their swap wagers were small, relative to all such trading that was done that day.

. . . .

But swaps were also a good way to speculate for traders who didn’t own the debt. Swap values rise on the fear of default. So traders who believed that fears about Morgan Stanley were likely to intensify could use swaps to try to turn a fast profit.

Note the use of the word “bet” (three times) in the first three quoted paragraphs.   The article also throws about terms like “financial Frankenstein” to describe credit default swaps.

I agree with Felix Salmon’s verdict:

In fact it tells a much simpler tale: Morgan Stanley’s counterparties were forced to buy CDS protection to hedge their exposure to the bank, and Morgan Stanley’s hedge-fund clients withdrew a lot of money, not in a bear-raid attempt to kill it off, but because prudence demanded that they do so, and also because they were understandably upset about John Mack’s role in getting the short-selling ban put in place. That ban devastated many hedge-fund relative-value and convertible-arbitrage strategies and caused a lot of anger in the hedge-fund community.

. . . .

So yes, it’s possible that those demonic short-sellers were responsible for the fall not only in Morgan Stanley’s share price, but also in Citi’s at the end of last week. But it’s also possible that it was just old-fashioned sellers, who weren’t selling short but were rather selling down their existing long positions. And it’s also possible that it wasn’t selling at all, but simply a lack of buyers willing to place bets on a fragile institution with a possible leadership vacuum. We simply don’t know — which is why it’s silly to assume that short-sellers were to blame.

That is, it was individually rational for the Merrills, Citigroups, Deutsche Banks, UBSs, etc., to hedge their counterparty risk.   This was especially true in the immediate aftermath of the Lehman collapse, which had demonstrated that (a) a big investment bank could collapse with stunning swiftness, and (b) there was no guarantee that the government would intervene to prevent the collapse.   Put differently, one can explain what went on without positing that malign speculators were making these trades with the specific intent of destroying the firm.

There is still an issue that deserves thought, but which is likely impossible to resolve.   That is whether this individually rational behavior was economically efficient–or collectively rational.

For instance, the activities of those hedging against (or betting on) a Morgan Stanley collapse could be consistent with an informational cascade.   The individual banks had information on Morgan Stanley’s creditworthiness, but were also drawing inferences from the actions of other potentially informed parties.   A given bank, observing others buying protection or short selling MS stock, would infer that these parties had received adverse information about Morgan’s creditworthiness.   Even if a given bank’s own information wasn’t that bearish, the inference drawn from the behavior of others could be sufficient to convince it that the firm was on the brink of failure, leading it to buy protection against an MS default.   And the next bank, seeing what that bank had just done, would likely make the same inference.   And the next bank, and the next one.   Such a “cascade” can become self-sustaining, leading everyone, regardless of their information, to join the wave of those “betting” against MS’s survival.

The literature on cascades demonstrates that these cascades can lead to the “wrong” outcome.   Specifically, if the first couple of traders act on signals that present an overly pessimistic view of Morgan’s condition, an unjustifiably bearish cascade can occur.   In such a cascade, even those with more optimistic signals will make bearish trades. Thus, the ultimate outcome of the cascade may not aggregate information accurately.   A healthy firm can be brought to its knees by a cascade.

So it is possible that Morgan Stanley’s agonies were the result of an inefficient informational cascade in which everybody was individually rational, but the result was inefficient and collectively irrational.

This raises the question: What can be done to stop such inefficient cascades?

The literature on cascades suggests that cascades are very fragile.   They can be stopped quickly by the release of accurate, credible information.   It is hard to imagine this occurring in the frenzied conditions of the market on those September days.   Who could have provided this credible, accurate information in real time?   Morgan Stanley itself was hardly a credible source of information.   Given the opaque nature of the firm’s balance sheet, moreover, it is hard to imagine that there was any credible source of information.

Perhaps some sort of circuit breaker–or, perhaps, “cascade breaker”–would have done the trick.   Stopping trading would have stemmed the cascade’s momentum.   This stoppage would have had to have been coordinated across several markets, including the CDS, stock, and options markets, in the US and abroad.   Moreover, how would the breaker be triggered?   Would it be some non-discretionary trigger based on observable market prices, e.g., a fall in a company’s stock price of X percent?   Or a discretionary action?   If discretionary, who would exercise it?   A government regulator?   There are obvious pitfalls with any of these alternatives.   (Look at the chaos in the Russian stock market in recent weeks, with myriad market closures.)

This “cascade breaker” could provide an intellectually respectable justification for a restriction on short sales.   (Not that the opponents of short sales have made such a justification, preferring instead to portray short sellers as financial Snidely Whiplashes.)

But not so fast.   Although it seems that bank runs–and what happened to Morgan Stanley is analogous to a bank run–are inherently bad, and that stopping them would be good public policy, the banking literature does not support this conclusion.   As Diamond and Rajan and others have shown, bank runs can be a valuable disciplining advice.   Absent the threat of runs, financial institutions that supply liquidity can act opportunistically towards their creditors.   This opportunistic behavior can be costlier than a run.

Thus, there is no immediately obvious policy response to the events described in the WSJ article.   The cartoonish demonization of short sellers and credit protection buyers is certainly not a reasonable basis for intervention into the markets.   The details presented in the article do not prove that speculators intent on driving Morgan Stanley out of business were responsible for the short sales and the CDS purchases; individually rational hedging behavior could have produced the same behavior.   Individual rationality does not imply efficiency because of the possibility of inefficient information cascades, but even then, the optimal policy response is not obvious.     Unless policy makers can reliably distinguish between inefficient and efficient runs/cascades, intervening to prevent all cascades through the imposition of a circuit breaker can eliminate a valuable disciplining mechanism that reduces the costs of contracting.

In the end, I imagine that one’s priors about the likelihood of inefficient cascades determines one’s policy recommendation.   Since information cascades are the result of myriad individual decisions, and the optimality of these decisions depends on private information unobservable to any investigator, it is impossible to determine ex post (let alone in real time) whether a particular run/cascade was efficient or inefficient.   Thus, there is inevitably a paucity of evidence that people can use to revise their priors.   Moreover, any policy action is likely to have large, and virtually impossible to quantify, costs.   (How does one quantify the costs arising from less efficient contracting when the elimination of disciplining runs rules out the “best” contracts?)   Thus, cascade breakers, or any policy response, are little more than shots in the dark, based on quasi-religious views about the efficiency of markets.

I know that’s not a very palatable diagnosis, but there it is.   We are ignorant, and the ignorance cannot be remedied.