Devoted readers, you’re no doubt wondering where I’ve been. The truth is, I’ve been anxious to post, because my late readings have mostly been either boring textbooks that don’t leave much room for stimulating criticism or popular business books that I was bashful to be seen thumbing on the subway, much less blogging about—namely, Andrew Ross Sorkin’s Too Big to Fail and Roger Lowenstein’s When Genius Failed.
I started with Too Big to Fail, the wunderkind NYT financial journalist’s play-by-play of the collapse of Lehman and financial crisis from (mainly) August through November 2008. At that time, I was still in the phase of my life when ancient controversies in ontology seemed more urgent than looming global economic crises and I would have been viscerally opposed to reading the business section of a newspaper—so there were some really basic gaps in my knowledge of the financial crisis that I had to fill in, and this book served as an accessible narrative supplement to the more analytic Financial Crisis Inquiry Report. TBTF is a great act of journalism, but it wasn’t chock full of analytically interesting stuff. My big take-away from the whole 600 pages was a gestalt shift—i.e., a look at how financial markets move from the micro-view, inside board rooms, as opposed to from the macro-view of trend lines of graphs of asset prices. This is one of those truths that seems obvious when written down, but takes time to really internalize: A statement like “Markets pushed Asset X higher/lower in a renewed flight to/from risk,” is several levels of abstraction removed from the plain truth about the world. What has really happened when that has been reported is that one group of people, bound by various legal and other institutional constraints and social pressures, has agreed that it needs to buy more of Asset X; they have called up a bunch of other people with whom they happen to be on friendly terms and expressed said need; the two groups, bound by all kinds of reputational concerns, social sympathies, fears, opportunisms, ass-covering needs, and ignorances, have negotiated a price (which must be some amount higher than recent sale prices if that asset is now considered more desirable). From a micro-level we can call this “a deal” or “a trade”; from a macro-level we call it “a price change.” In other words, the simple curves we see in the newspaper the next day often conceal a great deal of discontinuity and the multiplicity of factors that influence people’s willingness to sell/buy any asset aside from price. And then the curves that we find in financial theory textbooks are still one degree of abstraction more beyond the level of abstraction in the newspaper. These curves are extremely useful theoretical constructs—but they are, again, not actually the truth about how the world works.
My understanding was that his has a lot of bearing on how the Lehman crisis happened. In summer 2008, Lehman still had a lot of valuable assets, tangible and intangible. So theoretically, there really should have been some outside, better-capitalized firm that could have really benefited from purchasing the investment bank at a low share price. If this had happened, if Lehman had been able to find a buyer at a price that reflected both the reality of its mistakes (entailing realistic markdowns on its mortgage-backed securities) and its potential, Lehman’s shareholders would have ended up a lot better off and so would the whole financial system/global economy. But what was reasonable in theory couldn’t happen in practice—because of those discontinuities that meant that it simply wasn’t the case that the “demand for Lehman curve” met the “supply of Lehman” curve at a particular price where a purchase would necessarily happen. Some of these discontinuities include: (1) Legal constrains: i.e., capital requirements (good things on the whole) meant that, even if buying all of Lehman’s assets would be a profitable proposition in the long run, many potential buyers that were already pushing their lower capital limit wouldn’t be able to absorb the short-run losses from, say, a run on its mortgage-backed assets; (2) human status anxieties—a lot of deals founder on the “social issues” question of who should run the new, merged firm; (3) other, more admirable social anxieties—Dick Fuld not wanting to face the humiliation of selling of Lehman, in which no doubt many of his friends were shareholders, at a low price, even if that low price was the best he could get; (4) the self-sustaining madness of crowds–even e.g., financially savvy portfolio managers who knew that mortgage-backed securities had some value were limited from purchasing them even at very low prices because their own investors were caught up in a media-fueled panic over the very compound term “mortgage-backed,” and would, consequently, have made withdrawals; and (5) a million others I can’t list now.
If I can very broadly generalize, my big takeaway from Too Big to Fail, is that the reality of markets, particularly in times of crisis, is that they are very discontinuous, and we must take account of this.
More intellectually interesting to me was Roger Lowenstein’s When Genius Failed (which I was moved to read by all the references to the failure of Long Term Capital Management in Too Big to Fail) because the book, as I read it, gestures at a lot of commentary on economics, the philosophy of social science, and even theory of knowledge more generally. Indeed, I would say that this commentary is the most intellectually interesting part of the book, and also the book’s greatest flaw. From the very beginning, Lowenstein sets up a pretty tropeish economic cosmogony: He reminds us, three or four times per chapter, that Long-Term Capital Management was run by academics with theories that centered on the rationality of markets and led them to believe that all trading should be driven by models and risk could be measured scientifically. We hear constantly of their “naive” “faith” in markets and their models. Occasionally, some wise, old, experienced trader who trades from his gut is introduced as their alternative. And yes, as you could have predicted in advance, there really is a chapter called “The Human Factor.”
And, as you may gather from my tone, I think this cosmogony is idiotic. Every trade — indeed, every action in financial markets — entails a prediction about the future: A purchase is a prediction that the asset will go up in value, a short sale is a prediction that it will go down in value. And any good prediction about the future, in turn, requires an accurate (scientific) picture of the world as it is today, and an idea about how that will drive changes in the future. In other words, it requiers a theory of cause and effect — a model. So we all have, implicitly, models about how the world works; and we have nothing else to guide our choices that relate to the future. Some of us, like the academics at LTCM, render our models more rigorously and scientifically; others less so. Predictions that aren’t modeled are either (1) trivial (“people are getting nervous, so they’re going to go for Treasuries, and I can capture some of that value on the way up!”) or (2) not actually knowledge at all (“this asset’s just gotta turn around soon”). For every one old experienced gut trader who has been a success, there are many more who were unexceptional or ruined despite identical general ideas and feelings about markes — the successful one being distinguished solely by his luck. In short, if we find a “gut trader” who is successful, there is a much simpler explanation than ‘his wise rejection of the scientistic preoccupation with mathematical modeling’: luck.
So, to review: The economic cosmogony which places “academics” with “theories” and “faith” in the “rationality” of “markets” versus “gut” traders is idiotic. The only way to make rigorous, intelligent, and non-trivial predictions about markets is to use models and math.
So, what did Long-Term Capital Management get wrong? Well, I think a quote from John Maynard Keynes that Lowenstein pulled out for a chapter epigraph actually gets it right: “The market can remain irrational longer than you can remain solvent.”
In truth, LTCM was correct in its major predictions — bond spreads would have to decrease, eventually, because, well, investors had incentives to bring bond spreads together. And eventually, this prediction did come true, such that if LTCM had faced no constraints on its ability to hold its positions, it would have profited massively. So, again, the problem wasn’t that LTCM’s positions were wrong, or that markets were irrational, but that institutional constraints rendered LTCM insolvent in the short run, even while its positions were profitable in the long run. The main things that made this happen were: (1) unpredictable geopolitical stability caused, in part, by Boris Yeltsin’s alcoholism, (2) legal and institutional restraints such as margin requirements and capital requirements that drained the fund of cash and credit, even though it had a very persuasive argument that it must profit eventually, (3) other portfolio managers who were forced to abandon positions that they thought made sense, under pressure from withdrawals from their own, irrationally pessimistic investors and (4) other, predatory traders who saw how all of these together could bring LTCM down, and therefore bet against LTCM’s positions, making the fund’s downfall a self-fulfilling prophecy. To put it more plainly, LTCM faced a classic bank-run — a phenomenon which has been stopped in commercial banks due to the Federal Deposit Insurance Corporation, but which still afflicts the “shadow-banking” sector (as seen memorably during the recent financial crisis).
So, to really belabor the point, the problem with Long Term Capital Management, was not that “they believed that markets were rational and eventually always become more efficient” or “they relied on academic models and theories,” it was that they didn’t predict all of the freaky shit that could happen in the short run to prevent them from cashing in on positions that were rational in the long run.
My main takeaway from all of this is that the big danger to economists is not that we assume that financial markets are rational. In the economic meaning of the word ‘rational’ – very smart – financial markets are very rational indeed. This is why, even though most of my friends are theoretically smart people, I am sure most would not make money by picking stocks – because any true and profitable idea they might have about particular stocks has already been had by thousands of traders who follow financial markets much more closely than they. The very basic claims about the efficiency and rationality of markets, however much they have been sneered at in recent years, are true.
The problems are that financial markets (1) are not continuous, (2) that market participants are constrained by institutional and legal factors, external to the market itself, that can force them to abandon positions that would be profitable over the long-term, and (3) the very classic, very basic problems of self-fulfilling panics and bank runs that afflicted commercial banks until the FDIC have still not been solved for the shadow-banking sector. The problem with Merton’s approach wasn’t (1) that it “naively” and wrongly “assumed” that markets were rational, or (2) that it was “scientistic” in relying on predictions from mathematically rigorous models. It was that the models didn’t properly the things outside of the market, and the way they can make markets behave not like markets in the short run.