Back in the 1930s John Maynard Keynes taught America to look away from the economic producer and focus on the economic consumer. America has kept its eyeballs glued to that consumer ever since.

Seven decades later, the US consumer seems all-powerful. When he buys, the world economy grows. When he does not, the whole mechanism stalls. The consumer's empire even extends to US electoral politics. When he is feeling good, incumbent presidents tend to win. When he is out of sorts, they lose. Long live the demand side!

But this line of reasoning is partly based on a fallacy. In the economy, other factors - production, investment, interest rates, the stock market, tax rates - are also important. There have been points when the consumer lost heart even as the economy accelerated. As the Chicago-based trading house Griffin, Kubik, Stephens and Thompson points out in a report this month, scrutiny of consumer confidence can cause observers to miss crucial economic inflection points. If you took consumer confidence as your only indicator in the early 1980s, you missed the start of a great recovery. In the winter of 1998-99, consumer confidence fell but the boom went on.

Then there is the question of the consumer and politics, and the belief that economic mood determines political choices. It's clear where that belief derives from - political reporters find it easier to relate to the notion of the consumer-voter than to study the investment habits of companies (which, of course, don't vote). But it is also possible to argue that politics can determine consumer confidence, rather than the other way around.

Consider ABC News/Money magazine's Consumer Comfort Index. A rating above zero indicates confidence, one below indicates a lack of it. Six weeks before the election, the poll shows confidence at -9, about the average of the past two decades.

The ABC pollsters, however, also look at consumers' political allegiances. On September 18, the confidence rating of the average Republican was a merry +39. But the average Democrat felt bleak. His confidence in the economy stood at -36.

This 75-point differential is enormous and contrasts with data for earlier elections. In September of the 2000 campaign the difference between Republicans' and Democrats' attitude was only nine points; in September of 1996 it was a mere 15 points. Even in 1992, a year when the economy was the central issue of the campaign, the differential was 41 points.

There are some obvious explanations for the 2004 chasm. Democrats tend to appeal to collectivists, and Republicans to individualists. The shift to a service economy from an industrial one is hard on collectivists; so is the dawning realisation that Social Security, America's pension system, will not pay for younger workers' retirements. These realities have collectivists down in the dumps.

Individualists, meanwhile, think that the Bush tax cuts spell opportunity - and another is coming this autumn.

But the extent of today's confidence gap suggests that voters' political feelings can overwhelm their economic perceptions. After all, the 2004 campaign is the most polarised since the Nixon-McGovern race of 1972. Political rage is so strong that it spills over into other areas and economics is no exception. Democratic consumers are saying they aren't confident because they don't want to give the gift of their approval to Mr Bush; Republicans are saying they are confident because they want to boost the president.

What is more, if we want to choose our president based on the performance of the economy, then setting too much store by consumer confidence as a measure can lead to perverse results. Back in September 1992, as Mr Bush doubtless recalls, there were signs of recovery. Businesses were reporting that they were spending 4.3 per cent more on capital goods - equipment, factories - than they had in 1991. September1992 saw an enormous increase in spending on machine tools. An autumn survey of companies revealed that they projected a 5.3 per cent rise in investment spending the following year.

But these were non-consumer measures, and so they were barely noted. What mattered was the consumer measure. That September the ABC/Money consumer comfort figure, for example, stood at a frigid -42. James Carville, the Democratic strategist, capitalised on negative sentiment with the motto "It's the Economy, Stupid". Voter-consumers believed the politicians; politicians believed voter-consumers; the two sides reinforced one another. President George H.W. Bush duly lost.

But not in a poor economy. As Mr Clinton's own economists reported 13 months later, real third-quarter growth for 1992 was actually 3.4 per cent; in the fourth quarter, when disillusioned consumers defeated Mr Bush, the growth rate was more than 5 per cent. Keynes' demand-focused economics had contributed to a political outcome that he would never have predicted.

What about this year? Pollster Gary Langer of ABC points out that the average consumer confidence rating for 2004 fits into the range in which postwar incumbents have tended to get re-elected. Still, it would be wrong to take that trend too seriously. Perception is only reality when we make it so. And the truth is that the consumer can be a tyrant. When he dominates the stage, he obscures crucial events - economic and political both. Why let him?

© Copyright 2004 Financial Times