Friday, May 26, 2006

Science, voodoo… or just ideology?

The past few weeks have been a time of turmoil for economic markets. They have been lurching and plunging all over the place, prompting a rash of ‘explanations’ from market analysts. ‘The noise in the markets is the sound of everyone and his dog coming up with post-facto justifications for the apparently random movements in assets from gold to equities, copper and the dollar’, says Tom Stevenson in his 23 May Investment Column in the Daily Telegraph. ‘And while the latest rationalisation sounds plausible enough when you’ve just heard it, so too does the next believable but contradictory explanation.’

Full marks to Stevenson for his honesty. But then he goes on to suggest that ‘what drives financial markets is not the ebb and flow of investment ratios and economic statistics but the fickle and often lemming-like workings of investors’ minds… In other words, asset prices are falling simply because they’ve been rising sharply and investors have become more nervous.’ Now, I am not a highly paid investments analyst, but I can’t help feeling that Stevenson is hardly dropping a bombshell here. Weather forecasters would be unlikely to gain many plaudits by telling us that ‘tomorrow it will rain because some water vapour has evaporated and then condensed up in the sky.’

I suppose we should be grateful that Stevenson is at least not recycling one of the many voodoo tales to which he alludes. But his comments are a symptom of the extraordinary state of economic punditry today – which in itself is a reflection of the bizarre state of economic theory itself. When Thomas Carlyle called it the ‘dismal science’, he was (contrary to popular belief) making a judgement not about its quality but about its seemingly gloomy message. But ‘dismal’ hardly does the situation justice today – there is no other ‘science’, hard or soft, that has got itself into a comparably strange and parlous state.

Everyone knows that market statistics, such as commodity values, fluctuate wildly over a wide range of timescales (while, in the long term, showing generally steady growth). There is nothing particularly remarkable or surprising about that: clearly, the economy is a complex system (one of the most complex we know of, in fact), and such systems, whether they be earthquakes or landslides or biological populations or electronic circuits, show pronounced and seemingly random noise. What is unusual about economic noise, however , is that an awful lot of money rides on it.

That is why, rather than regard it indeed as noise, economists and market analysts are desperate to ‘explain’ it. Imagine a physicist looking through a magnifying glass at the wiggles in her data, and deciding to find a causal explanation for each individual spike. But that is precisely the game in market analysis.

The standard approach to this aspect of economic theory is as revealing as it is disturbing. Economic noise is a ‘bad thing’, because it seems to undermine the notion that economists understand the economy. And so it is banished. Noise, they say, has nothing to do with the operation of the market. In the ‘neoclassical’ theory that dominates all of academic economics today, markets are instantaneously in equilibrium, so that they display optimal efficiency and all goods find their way effectively to those who want them. So the marketplace would run as smoothly as the Japanese rail network – if only it did not keep getting disrupted by external ‘shocks’.

These shocks come from factors such as technological change – an idea that stems back to Marx – which force the market constantly to readjust itself. The very language of this process, in which economists talk of ‘corrections’ to the market, betrays their insistence that none of this is the fault of the market itself, which is simply doing its best to accommodate the nasty outside world. “Nothing more useless than listening to a newscaster tell us how the market just made a little ‘correction’”, says Joe McCauley of the University of Houston, who believes that ideas from physics can help explain what is really going on in economics. (I have a forthcoming article in Nature on this topic.)

“The economists incorrectly try to imagine that the system is in equilibrium, and then gets a shock into a new equilibrium state”, says McCauley. “But real economic systems are never in equilibrium. There is, to date, no empirical evidence whatsoever for either statistical or dynamic equilibrium in any real market. In their way of thinking, they have treat one, single point in a time series as ‘equilibrium’, and that is total nonsense. It’s completely unscientific.”

Unscientific perhaps – but politically useful. While the idea of perfect market efficiency rules, it is easy to argue that any tampering with market mechanisms – any regulation of free markets – is harmful to the economy and therefore pernicious. And that is the suggestion that has dominated the climate of US economic policy since the Reagan administration, as Paul Krugman points out in his excellent book Peddling Prosperity (W. W. Norton, 1994). Of course, the existence of an unregulated economy suits big business just fine, even if there is no objective reason at all to believe that it is the optimal solution to anything.

A little history makes it clear how economics got into this state. Adam Smith’s idea of a ‘hidden hand’ that matches supply to demand was a truly fundamental insight, elucidating how a market can be self-regulating. It chimed with the Newtonian tenor of Smith’s times, and led to the notion (which Smith himself never expressed as such) of ‘market forces’. Then at the end of the nineteenth century, the architects of what became the standard neoclassical economic theory of today – men like Francis Edgeworth and Alfred Marshall – were highly influenced by the ideas on thermodynamic equilibrium developed by scientists such as James Clerk Maxwell and Ludwig Boltzmann. Concepts of equilibration and balancing of forces, imported from physics, are to blame for the misleading notions at the heart of modern economics.

Ah, but there was still no escaping those fluctuations – certainly not in the 1930s, when they led to the biggest ever market crash and the Great Depression that followed. But at that time the scientific concept of noise, pioneered by Einstein and Marian Smoluchowski, was in its infancy. Much better understood was the concept of oscillations – of periodic movements. And so economists decided that, because sometimes prices rose and sometimes they fell, they must be ‘cyclic’, which is to say, periodic. This gave rise to the most extraordinary proliferation of theories about economic ‘cycles’ – there were Kitchin cycles, Juglar cycles, Kuznets and Kondriateff cycles… The Great Crash was then simply an unfortunate piling up of troughs in different cycles. Some were more sophisticated: in the 1930s US accountant Ralph Elliott proposed that markets wax and wane in waves based on the Fibonacci series, a piece of numerology that even today leaves analysts making forecasts based on the Golden Mean, as though they have been spending rather too long with The Da Vinci Code.

As a result, we are now seemingly stuck with the concept of the ‘business cycle’, a piece of lore so deeply embedded in economics that it is almost impossible to discuss market behaviour without it. So let’s get it straight: there is no business cycle in any meaningful sense – nothing cyclic about ‘bull’ and ‘bear’ markets. Sometimes traders decide to buy; sometimes they sell. Sometimes prices rise; sometimes they fall. Things go up and down. That is not a ‘cycle’. “There is no empirical evidence for cycles or periodicity”, says McCauley. “And there is also no evidence for periodic motion plus noise. There is only noise with [long-term] drift. So the better phrase would be 'business fluctuations'.” But ‘fluctuations’ sounds little scary – as though there might be something unpredictable about it.

And anyway, there is a neoclassical explanation for business cycles, so no need to worry. It was concocted by Finn Kydland and Edward Prescott and they won the 2004 Nobel prize for it. Surely that counts for something? Not in McCauley’s view: “The K&P model is totally misleading and, like many other examples in economics, should have been awarded a booby prize, certainly not a Noble Prize.” This model gives the favoured explanation: those pesky fluctuations are ‘exogenous’, imposed from outside. The market is perfect; it’s just that the world gets in the way.

You’d think we must have a good understanding of business cycles, because some economic policies depend on it. One of the five criteria for the UK entering into the European monetary union – for adopting the euro – is that UK business cycles should converge with those of countries that now have the euro. This, I assumed, must be predicated on some model of what causes the business cycle – surely we wouldn’t make a criterion like this if we didn’t understand the phenomenon that underpinned it? So I wrote to the UK Treasury two years ago to ask what model they used to understand the business cycle. I’m still waiting for an answer.

Economics as a whole is not a worthless discipline – indeed, some of the recent Nobels (look at 1998, 2002 and 2005, for example) have been awarded for genuinely exciting work. But McCauley is not alone in thinking that its core is rotten. Steve Keen’s Debunking Economics (Pluto Press, 2001) demonstrates why, even if you accept the absurdly simplistic first principles of neoclassical microeconomic theory, the way the theory unfolds is internally inconsistent. Economists interested in using agent-based modelling to explore realistic agent behaviour and non-equilibrium markets have become so fed up with their exclusion from the mainstream that they are starting their own journal, the Journal of Economic Interaction and Coordination. Such models have made it very clear that Tom Stevenson’s hunch that market fluctuations come from herding behaviour – that the noise is intrinsic to the economy, not an external disturbance – is right on the mark. So much so, in fact, that it is odd and disheartening to see commentators still needing to disclose this as though it was some kind of revelation.

No comments: