Statistics, damned statistics and value added


Starting from a flawed productivity comparison, John highlights the difficulties in interpreting economic statistics.

I have been reading a report from the Sector Skills Development Agency (now happily defunct). It claims to compare the efficiency of industrial sectors in Britain with the same activities in other countries. Britain is second from bottom in financial services. We are slightly behind the US, but well below stellar performers such as Italy and Belgium. The report notes that the large size of the industry means that this poor showing holds back Britain’s overall economic performance.

Alarm bells might have rung. If Britain’s financial services companies are so bad, why are they so successful? But no: the report notes that Britain’s retailers, feared and respected around the world, also do badly on these productivity measures.

Output of financial services is difficult to compute. The usual economic definition looks at value added, the difference between the value of the goods you sell and the cost of the materials you buy. This definition excludes profits from securities, interest paid or received and all transactions associated with the financing, rather than the operations, of the business. These principles work well for manufacturers, but not for a financial services company, which typically pays its bills through profits from trading and investment returns.

The difficulty of measuring value added is the reason financial services are exempt from value added tax. It is easier to leave such activities out of the net than to devise special rules for their inclusion. Measuring financial services is also an embarrassment for economic statisticians. They introduce a fudge to reconcile their own estimates of the output of the sector with the numbers reported in company accounts. In Britain, this adjustment amounts to almost 5 per cent of national income.

I have not started on the problems of measuring capital employed in financial services. Is the capital of a bank the size of its balance sheet, its shareholders’ funds, or the value of the premises it occupies and the computers it uses? What is the output of a retailer? You might take the view that selling a pair of socks is the same service whether the socks are in Wal-Mart or Neiman Marcus, but if you do you have not understood retailing. Or you might think that the value of a retailing service is measured by the retailer’s gross margin, in which case you will not notice that the protected independent retailing sectors of many European countries are indeed inefficient.

Few economic statistics are facts: most require thoughtful analysis and interpretation. The common sense check that the Sector Skills Development Agency failed to apply is even more important in economics than in physical sciences. As a research student, I learnt that when I found an anomaly in the data, the usual reason was that I had made a mistake. If a conclusion contradicts common sense, you check it again, and again. If the observation still stands you may – just may – have discovered something interesting.

Today’s researchers import large databases into econometric packages without seeing the numbers on which their analysis is based. The frequent outcome is unreflective data crunching that displays sophistication of technique but is devoid of meaning.

So I find on my desk an analysis of the economic benefit of computers to economic growth which, on examination, only measures their costs. The analysis assumes that the benefits equal the costs, otherwise a rational company would not have bought them. There is also a lengthy description of competition policy in UK civil aviation since 1993 that omits any mention of EasyJet and Ryanair. These articles could be written only by people who have never been to an office or an airport – or, more likely, who fail to connect professional activities to everyday experience.

Sir Josiah Stamp, a founder of economic statistics, observed that “the government are very keen on amassing statistics – they collect them, add them, raise them to the nth power, take the cube root and prepare wonderful diagrams. But what you must never forget is that every one of these figures comes in the first instance from the village watchman, who just puts down what he damn well pleases.”

Print Friendly, PDF & Email