Whilst I was away, there was a debate on Twitter about why so many Oxford PPEists become newspaper opinion writers. The question was ill-posed: in fact only a teeny fraction of PPE graduates enter that profession - much less than 1%.
The idea that PPEists become opinion-mongers is an example of the representativeness heuristic - our tendency to over-estimate a probability if something seems representative of a group. In this case, because bashing out half-cocked opinions quickly seems representative of PPEists, people over-estimate the probability of PPEists becoming opinion-writers. In doing so, they neglect base rate probabilities: there are thousands of us PPEists but only a few opinion-writers.
For years, doctors have made this same mistake. Back in 1982 David Eddy showed that they over-estimated (pdf) the chances of patients having a disease if they had tested positive for it: because a positive test is representative of the disease, they neglected the basal probability of the disease. In Bayesian terms, people mistake P(A|B) and P(B|A): we all do.
The same mistake explains the financial crisis of 2008, according to a new book, A Crisis of Beliefs, by Nicola Gennaioli and Andrei Shleifer. They argue that the market stability of the 90s and early 00s caused investors to over-estimate the probability of stability continuing – stability being representative of a healthy financial system.
In other words, they say, investors have neither the naïve adaptive expectations attributed to them by early Keynesians, nor fully rational ones. Instead, they update their beliefs in light of new evidence (such as prolonged stability) but do so in a quasi-rational way, to use Richard Thaler’s phrase.
This theory fits a Big Fact – that the crisis was not a bolt from the blue. In a neat narrative of the events, Gennaioli and Shleifer show that the crisis was bubbling up for months before Lehmans collapsed: US house prices began falling in 2006 and some sub-prime lenders and hedge funds holding mortgage-backed securities collapsed in 2007. Such events did not trouble equity investors much, however, because they took past stability to be representative of future stability and so thought that the risks posed by the housing market were manageable. It was only when Lehmans collapsed and was not bailed out that this belief changed. And – in retrospect – investors over-reacted to that collapse because of the same representativeness heuristic: a banking crisis is representative of a dysfunctional economy, which led to shares falling too far.
Personally, I like this explanation, but with four caveats.
One is that it is not a full explanation of the crisis. Gennaioli and Shleifer do not ask why demand for “safe” assets was so high in the mid-00s. The deeper structural failings of capitalism that created that demand are outside the purview of their book.
Secondly, they fit this theory to only the 2007-08 crisis without applying it to others. This invites the old criticism that economists see something fail in practice and wonder if it can fail in theory.
I suspect, however, that it does have wider applicability. It might fit the tech bubble: sharply rising share prices are representative of great future businesses, thereby leading people to over-estimate the likelihood of companies becoming the latter. It might also be consistent with economists’ longstanding failure to foresee recessions: growth is representative of normal stable economies, which leads us to under-estimate the chance of downturns.
Thirdly, they pay insufficient attention to the question: why doesn’t the smart money correct this inferential error? The answer, in fact, lies in one of Shleifer’s earlier papers – that there are tight limits on how far one can short-sell; if you’re right at the wrong time, margin calls can wipe you out. There’s a reason why John Paulson and Michael Burry were rarities in profiting from the slump in mortgage securities.
Finally, I fear they underplay mechanisms which reinforce inferential errors. For me, one of the most revealing statements of the crisis was that of Chuck Prince, then head of Citigroup, in July 2007:
When the music stops, in terms of liquidity, things will be complicated. But as long as the music is playing, you’ve got to get up and dance. We’re still dancing.
This speaks to institutional and psychological pressures. The cautious manager who avoided sub-prime mortgages or shorted them under-performed his peers in the mid-00s and so faced calls from investors to change strategy; that scene in the Big Short where big investors demand their money back because being early is the same as being wrong rings true. And even without such demands, it’s hard to disagree with the herd; as a great philosopher said, you must be strong if you’re to go it alone. Again, though, Shleifer’s own work speaks to this: his paper Chasing Noise describes the herding mechanism well.
You might think all this is minor stuff. It’s not. Although the representative heuristic seems like a small deviation from rationality, it has a devastating effect. Gennaioli and Shleifer say:
We expect new financial products to load up precisely on the risks investors do not fully appreciate. Under rational expectations, financial innovation would improve risk-sharing and welfare. With neglected risk, in contrast, it creates inefficient risk-sharing and the possibility of crises. (p87)
In this sense, the psychological question of how expectations are formed raises a much more general question about the stability of the financial system. It is, therefore, a deeply political question.