Stephen King: As night follows day, causality is a problem

Causality is a very strange thing. When one event follows another on a systematic basis, people might think that there's a casual link. But, in reality, this can't be right. No one seriously believes that day is caused by night or, indeed, that night is caused by day. Yet both follow each other on a regular basis (unless, of course, you happen to be living either at the North or the South Pole). David Hume, the great Scottish philosopher, tried to get round the problem by coming up with the psychological idea of the "necessary connection", the mental leap that helps us to recognise that, only in specific instances, event B is caused by event A.

Causality is a very strange thing. When one event follows another on a systematic basis, people might think that there's a casual link. But, in reality, this can't be right. No one seriously believes that day is caused by night or, indeed, that night is caused by day. Yet both follow each other on a regular basis (unless, of course, you happen to be living either at the North or the South Pole). David Hume, the great Scottish philosopher, tried to get round the problem by coming up with the psychological idea of the "necessary connection", the mental leap that helps us to recognise that, only in specific instances, event B is caused by event A.

But although David Hume was doubtless on the right path, the idea of a "necessary connection" is inevitably rather vague. Why should I believe that Christmas, for example, is not caused by retail sales? After all, retail sales systematically strengthen in the months before Christmas, suggesting that Top Shop is more relevant to the Nativity than Jesus Christ, Mary, Joseph, the Archangel Gabriel, shepherds, wise men and the like. Strengthening retail sales and Christmas do have a necessary connection, but that, on its own, doesn't quite define the relationship.

We know that Top Shop wouldn't be seeing stronger sales unless Christmas was just around the corner. But the example shows that pinning down causality - defining causality - is a lot more difficult than appears to be the case at first sight. Expectations play a role: retail sales strengthen only because we expect Christmas to occur on 25 December and not on 3 July. More precisely, it is the knowledge that Christmas falls on 25 December that causes us to head for the shops in the weeks before Christmas.

Causality is a particularly big problem for policy-makers. Tried and tested relationships work precisely because they have been reliable in the past. But knowing that relationships work some of the time is not the same thing as knowing that these relationships work all of the time. Let me give you a relevant example. Economists monitor retail sales growth from month to month because they believe that retail sales provide an indication of demand pressures - and, hence, inflation pressures - within an economy. If retail sales are unusually strong, this is likely to be an indication of excessive demand pressures which, in turn, will give rise to higher inflation.

This approach reflects the use of "output gaps". Economists like to believe that economies are in a steady-state equilibrium when output is in line with "productive potential". Any level of demand above productive potential is likely to give rise to rising inflation. Any level of demand below productive potential is likely to give rise to falling inflation - at the extreme, deflation - and rising unemployment.

Of course, measuring productive potential is no easy thing. At any point in time, we have only the economy itself to look at. There is no "shadow" economic measure, something that might tell us where the economy ought to be. Instead, we're left with clues, indicators that might suggest that demand is, perhaps, a bit too strong or a bit too weak.

Last week's UK retail sales figures were quite a bit stronger than expected. They rose 1 per cent in September, the biggest monthly increase since January. For the third quarter as a whole, retail sales rose an annual 6.9 per cent. Admittedly, the underlying trend has slowed from the rapid pace of expansion seen earlier in the year, but these numbers still appear to be remarkably strong. They suggest that the interest rate medicine dosed out by the Bank of England over the past 12 months might not be working after all. More specifically, the strength of retail sales is, perhaps, suggesting that demand is still above the economy's productive potential, hinting that inflationary pressures might be lurking just around the corner.

The model of causality that's being employed here is simple enough: the stronger retail sales are, the more likely that inflation will eventually rise. But is this the right model to use? A closer inspection of the data suggests otherwise.

The charts provide some clues. The left-hand chart shows how total consumer spending and retail sales have developed in volume terms over the past four or five years. I've indexed the levels of both series to equal 100 at the beginning of 2000. By doing so, it's easy to show that retail sales have been rising a lot more quickly than overall consumer spending. Since 2000, retail sales have risen by more than 25 per cent, whereas overall consumer spending has risen by around 13 per cent: proof, perhaps, that the consumer's appetite for DVD players, plasma TVs and computers is the main driving force behind consumer spending. But the right-hand chart suggests that we shouldn't rush to quick conclusions about the implications for inflation. This chart shows what's been happening to the price of overall consumer goods and services and, more specifically, to the price of retail goods. At this point, causality alarm bells should start to ring. The chart shows that, over the past four or five years, overall consumer prices have risen, but retail prices have actually fallen, down by about 4 per cent over the period as a whole.

This suggests that the causality is very different from the one that economists are accustomed to. Retail sales aren't likely to be the cause of higher retail prices. Rather, falling retail prices are the cause of higher retail sales. Retail sales are strong in part because retail goods are getting cheaper.

If this is true, it's really rather difficult to argue that strong retail sales are likely to be the cause of higher inflation. If price declines are causing higher sales, it can't also be the case that higher sales are causing price increases.

This reverse causality suggests that retail sales are no longer quite so useful in identifying excessive demand pressures within an economy. Prices are coming down for all sorts of well-documented reasons - China's impact on the price of manufactured goods, greater competitive pressures both domestically and globally, more buying power for the retailers at the expense of their suppliers - and, in response, we're seeing people buying more retail goods than before. In other words, retail sales are rising because of an outward shift in the supply curve, not necessarily an outward shift in the demand curve.

All of this is good news for the inflation process. This doesn't mean, though, that the causal process is fully mastered. Falling retail prices will trigger other effects. Some areas of demand will soften, because people will substitute away from those areas towards retail. That points to more widespread downward pressure on prices. Some areas of demand, though, will strengthen because the boost to real incomes associated with lower retail prices implies higher demand for other goods and services. The final impact on inflation remains uncertain. Nevertheless, the strength of retail sales, on its own, seems to be more a consequence of falling prices than a cause of higher prices. As a result, the economists' assumed necessary connection may be neither necessary nor a connection.

Stephen King is managing director of economics at HSBC

stephen.king@hsbcib.com