Economics: The dangers of stirring up chaos

DEEP IN the bowels of the London School of Economics sits an extraordinary machine made of glass tubes, measuring chambers, valves and sluice gates. By pressing buttons or pulling levers, different coloured liquids can be made to flow round it to represent flows of money and goods in the economy.

The machine was built in the 1950s by Bill Phillips, an engineer turned economist. He intended it as a teaching aid, but it also possesses symbolic significance: it epitomises the fundamentally flawed methodology on which mainstream economics now relies.

Economists too often behave as though they were engineers. They treat the object of their study as a large and intricate machine, complicated to understand but ultimately consistent and predictable in its behaviour. But while machines are made from inanimate components, economies are collections of millions of people who are irrational, inconsistent and easily influenced.

This makes forecasting the behaviour of the economy as a whole very difficult. The economics profession failed to predict the boom of the late 1980s, the depth and duration of the recession or the timing and strength of the recovery. And economists are still arguing about the impact of the tax rises announced in last year's Budgets. Some believe they will slow the recovery while others expect them to pass by unnoticed.

The imperfections of economic forecasting are recognised by practitioners as well as critics. Many top economists - even those who take the trouble to test their theories against real world evidence - argue that far too much attention is paid to forecasting. But they have also had only limited success understanding history. After decades economists are still arguing whether the amount of money in the economy determines the rate of inflation, or vice versa.

Some economists none the less feel hard done by, relative to their counterparts in other fields. David Hendry, the Oxford economist who quietly debunked the statistical foundations of monetarism in the 1980s, complains that the Government bought the Meteorological Office a new supercomputer after it failed to forecast the hurricane of 1987, while economists were threatened with cuts in their funding after failing to warn of the dangers of the Lawson boom.

Economists, like all scientists, are interested ultimately in identifying cause and effect. But this is not easy. Any given action might have more than one outcome and any given outcome could have a number of causes - economics is 'non-linear', to use the jargon. It is as though pulling a lever on Bill Phillips's machine had a different result each time you tried it.

Economists have usually dealt with this problem simply by pretending that it does not exist. They approximate these complicated relationships with simplified 'linear' alternatives in which any given cause can have one and only one effect. If events do not fit these linear approximations exactly, the difference is assumed to be made up of random disturbances.

So, for example, an economic model might assume on past evidence that a 1 per cent rise in tax rates will result after a few months in a 1 per cent fall in high street spending. In reality, however, the impact of any given tax rise would depend on whether people felt it would be reversed in the future and on how they thought their neighbours would react.

Simplifying complicated relationships in this fashion has another attraction: linear relationships are simply the sum of their parts. So if you understand how one consumer or stock market investor behaves, you can simply aggregate the individual behaviour to understand consumer spending or stock market movements as a whole. Most mainstream economists feel theories about the behaviour of the economy are suspect unless they are built up from theories about the behaviour of individuals.

But this neglects the way in which people tailor their behaviour to their expectations of the future or of how other people will behave. Consumers often resist cutting their spending when their incomes fall, so that they can 'keep up with the Joneses'.

Similarly, speculative players in financial markets buy shares and bonds simply because they think other people will buy them too and keep pushing the price up. Unfortunately, the mugs join in just as the experts pull out.

Economists are prone to mock sociologists for woolly thinking, but sociologists have at least grasped that groups do not necessarily behave in the same way as any lone member of that group - as watching the crowd at Millwall soon demonstrates. As Friedrich von Schiller observed: 'Anyone taken as an individual is tolerably sensible and reasonable - as a member of a crowd, he at once becomes a blockhead.'

But simplifying a non-linear relationship into a linear one can be dangerously misleading if the interrelationships and feedback between individuals and companies are too complicated. Cause and effect may in fact be impossible to identify, as the economy moves in patterns that never repeat themselves, but are none the less recognisable. Interfering in tiny ways can then have unpredictable and dramatic consequences that are quite out of proportion to the event that triggered them.

Most natural sciences faced up to this problem years ago as 'chaos theory' came into fashion. In meteorology, for example, it is well known that feedback between different parts of weather systems is so complicated that a butterfly flapping its wings in Tokyo could have a series of unpredictable knock-on effects causing a hurricane in the Caribbean six weeks later. This means forecasts over anything more than the very short term are probably doomed to failure.

The same could happen in the economy. David Carrier of the University of Notre Dame argued in a recent paper that 15 minutes' discussion by the US Congress Ways and Means Committee of a proposal to raise dollars 400m by taxing company takeovers may have been the last straw that caused the 1987 crash in world stock markets.*

Similarly, the Federal Reserve's decision to raise US interest rates by a quarter of a percentage point in early February should have had little direct effect on the US economy, but it triggered a dramatic slump in world bond and stock markets that has greatly magnified its impact.

Economics has long been obsessed with the idea of equilibrium - the belief that the economy, if undisturbed, will settle down to some welcome stable state. The upturns and downturns that we all recognise as part of economic life are regarded as the result of random external events - such as changes in technology or consumer tastes - buffeting the system.

But chaos theory shows that weird fluctuations can arise very easily from the sort of non-linear relationships we find in the real world, without the need to fall back on outside disturbances to explain them. Patterns that are stable for a while can change suddenly before settling down again. Hence the regular upturns and downturns in the economy may change in duration and severity. Examples of chaos are all around us: look at the way smoke rises in a straight line from a cigarette before suddenly becoming turbulent for no apparent reason.

But what do the insights of chaos theory mean for economics? Does it have practical consequences or is it merely a mathematical curiosity to explain formerly inexplicable movements in economic statistics?

Chaos theory should change the way we look at economics, but unfortunately it poses as many new questions as it provides answers. It does not tell us whether governments should intervene or not.

On one hand we cannot regard the economy as a reliable and controllable machine on which we can press a few buttons to produce a desired result. On the other, chaos theory suggests free markets will not necessarily provide a stable environment in which resources are used as efficiently as possible.

David Parker and Ralph Stacey argue in an admirably clear exposition of chaos and its implications for economics, published last week, that on balance governments should intervene less**. They say that small changes in interest rates or tax policy could have seriously destabilising effects. Policy-makers should therefore leave consumers and businesses as unfettered as possible to cope with, and participate in, an unavoidably chaotic world. Entrepreneurs should be liberated with low taxation and less regulation.

But laissez-faire economists should not be allowed to claim chaos for their own. Carrier argues it is only because of government regulation and intervention that the economy does not behave in an even more peculiar and unpredictable fashion than it does already.

The case for intervention can also be made in the foreign exchange market. Exchange rates appear to behave chaotically because of the interaction between traders buying for fundamental reasons (a country's growth, inflation and interest rate prospects) and Chartists who look for buying opportunities by extrapolating past exchange rate movements.

There is a case for central banks to intervene in the currency market by 'leaning against the wind' to dampen any rises or falls. This could help move exchange rates from chaotic to relatively stable movements by reducing the weight given to Chartist predictions, helping businesses plan and calming swings in inflation.

Chaos deserves a wider audience than mathematicians and readers of popular science books sold at airports. It undermines the idea that free markets are efficient, casts doubt on forecasting and warns that government intervention may have unforeseen results. Chaos may not offer easy answers, but if it instilled greater humility in the economics profession that would be no bad thing.

*'Will Chaos Kill the Auctioneer?' David Carrier, Review of Political Economy, 1993.

**'Chaos: Management & Economics', Parker & Stacey, Institute of Economic Affairs, 1994.