There are many ways of interpreting the complexities of published economic statistics, with the result that commentators can find snippets of data to support any story they want to tell.
You may have read reports that retail sales in Britain fell by 0.4 per cent between November and December last year. Economic commentators certainly did and cited it as evidence that recession was on its way.
But wait a moment. Less shopping in December than in November? That was not my experience and I doubt if it was yours. Or anyone’s. Sales of food and drink are usually about 20 per cent higher in December than in an average month, and sales of other goods about 40 per cent higher. And that is what happened.
You have to compare this December with previous Decembers. But the calendar is not static. Peak shopping for food is on Fridays and for other items on Saturdays, so that sales in any month depend on the number of Fridays and Saturdays, which varies. Christmas falls on a different day of the week every year, so shopping habits and opening decisions by shops vary year by year. Christmas was on a Tuesday in 2007, with the result that Monday – Christmas Eve – was a relatively quiet day while on Wednesday – Boxing Day – some shops were busy with January sales while others remained closed.
So the first, by no means simple, step is to adjust the crude figures for the number of shopping days. (December 2007 was deemed to start on November 25 and end on December 29). Then you either have to compare an average December week with an average December week in the preceding year – in which case you must allow for price changes in the intervening year – or else compute a “Christmas effect” and remove it from the series.
This latter process is called seasonal adjustment and it reduces the raw figures for December, overall, by something in the region of 25 per cent. This seasonally adjusted figure fell by 0.4 per cent in December relative to the seasonally adjusted November figure. That is the figure that grabbed the headline.
Over the years, careful effort has been devoted to refining the methods of seasonal adjustment. Still, it would be a brave and foolish statistician who would say with confidence that the right answer – whatever that means – can be determined to the nearest decimal point (or percentage point for that matter).
You might turn instead to the number from the British Retail Consortium, which also reported “a gloomy Christmas”. Sales were up by only 0.3 per cent with no adjustment made for inflation. But this report refers to “like for like” sales – takings at the same selection of stores as in December 2006.
While this measure makes a lot of sense for the individual retailer, it makes no sense as an economic indicator. What matters is what consumers are spending, not where they are spending it. If we look at all stores and not just the ones that were open a year before, the figure is not 0.3 per cent but 2.3 per cent. And that takes no account of sales by retailers outside the survey – such as the online sellers whose growth has been widely noted.
My guess is that this was not a very expansive Christmas for British consumers, although from the data one can distinguish that only very approximately. But gloom is the order of the day and gloomy headlines find their own justification. There are many ways of interpreting the complexities of published economic statistics, with the result that commentators can find snippets of data to support any story they want to tell. They often fail even to distinguish a decline in the rate of increase from a fall. Sometimes I long for the good old days when I was a graduate student and we had to type in data ourselves rather than summoning it from a file. We had to think about the information then and sometimes even summoned the curiosity to ask how it was compiled.
The most famous remark of Josiah Stamp, banker and statistician, was that the price of giving banks the capacity to create money was to pay the cost of your own slavery. He also popularised the observation of an Indian judge that, despite complexities of analysis and interpretation, economic statistics ultimately depend on the chowkidar, the village watchman, who put down how he felt. Both comments are relevant to our own times.
