PsyFi Search

Wednesday 29 December 2010

Economics & Psychology: Reconciliation?

Continued From Economics & Psychology: The Divorce ...

By the early 1970’s, as the long bull market of the post war years collapsed in a welter of unforeseen problems, financial professionals confronted the real meaning of risk on a systemic basis. As markets crumbled in the face of economic uncertainty trading companies turned to economists in academia in the hope of finding a way through the mess, or at least some excuse to get people to buy stocks.

What they discovered was a way of measuring risk that appeared to offer the option of quantitatively managing investments in a rational way, rather than relying on the intuitions of individuals. This approach has come to dominate the securities industry ever since. At the same time, though, a small revolution was brewing in psychology. And it's been fermenting revolution ever since.

Friedman's Not For Turning

You might have thought that the problems encountered by markets would have led economists to rethink their models, constructed as they were to remove any last vestige of psychology. However, mostly they followed the dictums of Milton Friedman who, as far back as 1953, in The Methodology of Positive Economics, had argued that economic models were to be judged not on how accurately they described the underlying human behaviour but on how well they predicted economic events. In what turned out to be the unusually placid times of the 50’s and 60’s, these models worked well.

When the oil crisis and rampant inflation devastated the markets in the early '70's the securities industry came looking for risk management techniques. By then economists were ready armed with a host of models that happily expunged any trace of human behaviour or unmeasureable uncertainty. Harry Markowitz’s Portfolio Theory was leapt upon as a risk management tool without, it seems, any recognition of its author’s admonition that this was a model of how people should behave, not how they actually went about managing their investments.

Economics With The People Taken Out

Portfolio Theory begat CAPM and CAPM begat a whole slew of dubious children, amongst them the Black-Scholes option pricing model and Value at Risk. Dig deep enough and you’ll find a common genealogy. This was, and still is, economics with the people taken out. Now, because the models generally operate statistically, as long as we all behave rationally on average then the predictions are pretty accurate. This idea of averaging out behaviour is about as close as these models come to taking human behaviour into account: of morals and anything else you can’t measure monetarily they tell us nothing.

However, although for most of the time we do behave normally and rationally on average, occasionally we all wake up and decide to annoy economists by behaving like stupid, spoilt imbeciles. When markets go mad the models break down.

Meanwhile, even as the securities industry started trying to build ever bigger and better quantitative models to cope with insipient human irrationality, a few psychologists started wondering whether the cognitive revolution that was happening in psychology was equally applicable to economics. They wondered if people were systematically irrational in a way they would undermine the economic models.

Enter the Psychologists

Of course, the first people to begin looking at this seriously were Amos Tversky and Daniel Kahnemann, who started to point out some rather inconvenient facts about real human behaviour and its relationship to the assumptions behind the standard economic models. In particular they showed that people are asymmetric in their approach to risks – we chase losses but protect gains and, moreover, we’re anchored on the starting point at which we buy. From this starting point the whole topic of behavioral bias snowballed as researchers have uncovered ever more evidence that people don’t behave in the rational way required of the standard economic theories.

You might expect that this news would result in a deep rethinking of the ideas behind economics but, in fact, what happened was that people started tinkering with their preference based utility models. This was less revolution and more incrementalism of the kind we saw in Copernicus, Muddling Through.

Hedonic Incrementalism

This type of conservatism is typical of all subjects, it’s very rare for new knowledge to cause an immediate revolution – the only extant example of this may be the changes to physics caused by the quantum shift at the beginning of the 20th century – and even that’s subject to debate. So, economists tried to fit the new behavioural evidence into their existing models despite their explicitly anti-behavioural heritage.

Initially Tversky and Kahnemann introduced a weighting function which changed the utility of the existing models dependent on whether gains or losses were being experienced. Although this may not seem like a huge break with the past it was, in fact, winding back the clock half a century as allowing evidence gathered from working with actual people had been virtually outlawed since the introduction of revealed preference theory.

And, of course, in noting that people react differently to gains (pleasure) and losses (pain) they re-introduced hedonic utility by the back door. But what a strange creature this had become.

Utility Trembles

Tversky and Kahnemann’s work led to an explosion of research in behavioral economics and subsequent modifications of utility theory. By 2000, as Chris Starmer explains in Developments in Non-Expected Utility Theory: The Hunt for A Descriptive Theory of Choice under Risk, this was becoming ridiculous:
“Readers of this article will no doubt be familiar with Expected Utility Theory (EUT), the standard theory of individual choice in economics. Many, I expect will know of a few alternatives to this model. But how many, I wonder, will be aware that these so-called non-expected utility models now number well into double figures?”
To get an idea of how this approach to modelling works, consider the idea of trembles. No, this isn't the result of too heavy a night on the town: basically the idea is that when I make a decision, say by choosing a meal from a menu, I’m revealing my preference, but there’s a random element – a tremble – involved in this choice. Some of these models predict that this randomness disappears with experience, as an individual discovers their expected utility preferences.

Now what this suggests is that expected utility theory is correct, but only after we’ve undergone some education. So the suggestion is that the theory, carefully constructed to ignore psychology, is correct because it’s actually how people behave if you subtract their limited cognitive powers. Whether you think this is likely or not it’s an interesting example of how economists have been trying to take the psychological evidence into account without actually giving up a utility model approach specifically designed to ignore it.

Sociology, Satisficing and Neuroeconomics

Perhaps unsurprisingly not everyone agrees that it’s possible to modify existing economic models to take real behaviour into account. There are at least three different strands, all intermingled and often entwined with the standard economic models, all of which we’ve investigated at some point. Firstly there’s the idea that these models simply assume that society is a mass of individuals without taking into account their interactions. This is the issue of society that we described in Arbitraging Embeddedness and sociologists trace the issue all the way back to Adam Smith’s original formulation of economics – and see in this hole a fundamental problem with applying economics to the world.

Then there are the ideas of Herb Simon, based on bounded rationality, which have been taken up by Gerd Gigerenzer and other researchers, who simply believe that the psychological research suggests that people make economic decisions in a way that the standard models don’t so much capture as completely ignore. In this line of reasoning utility models fail at all levels and no amount of fiddling with different utility functions can save it. We don’t work that way, rather we satisfice using so-called good enough rules of thumb.

Finally there’s neuroeconomics, where researchers are trying to relate how we behave financially to how we actually process information. There’s a vague hope that the mentally deficient models of economics will somehow translate into neural processing but it’s a bit of a stretch, to be honest.

From Hedonics and Back Again

The strange and circuitous route that economic thinking has taken has gone from a basic hedonic model of human behaviour – the pain-pleasure principle – to a world completely divorced from psychology, back to something that recognises human feelings, relationships and emotions but doesn’t quite know how to include them without throwing away all of the carefully thought through models.

Unfortunately a spot of Schumpterian creative destructionism may be called for, tearing down the old assumptions in order to rebuild anew. In the meantime the fact that governments, regulators and the whole of the securities industry are operating on models that are known to be wrong, if only through the evidence of their results, ought to be a serious concern for everyone, everywhere.

See The Latticework for more articles like this.


  1. Super article.

    It's my view that a model that ignores irrationality ENCOURAGES it by doing so.

    Say that we had a model of nutrition rooted in a premise that everyone always eats what he or she knows is best for him. This model would teach that there is zero risk in eating too much chocolate cake because the possibility that someone might eat too much chocolate cake simply does not compute under a model that ignores irrationality.

    Someone eats so much chocolate cake that he weighs 300 pounds. He goes to a nutritionist and happens to mention that he eats six pieces of chocolate cake each night. The nutritionist makes no response. He has studied all the literature and has never seen a single reference to the problem of too much chocolate cake consumption (because it is an impossibility under the dominant theory).

    The overweight individual suspects that the chocolate cake might be a problem. But he notices that the nutritionist does not mention it and concludes that his concerns are silly. If there had been a possibility that eating too much chocolate cake could cause problems, surely a nutritionist would have mentioned it.

    We expect experts in a field to know something about the ABCs. This is not a safe assumption in the investing field today, given the models that remain dominant.


  2. I appreciate your post, thanks for sharing the post, i would like to hear more about this in future

  3. "fermenting revolution" in the second paragraph should be "fomenting revolution".

  4. Hi Will

    "fermenting revolution" in the second paragraph should be "fomenting revolution"

    Sadly it's just a terrible play on words: "brewing" is linked to "fermenting" and "fermenting" sounds like "fomenting". I do it all the time, it's a very bad habit :-)

  5. You will have to wait another 40.000 years before the great Hari Seldon finally gets the statistics and math right establishing the ultimate model integrating psychology, economics and political science in one great general theory.