The beauty is in the data

229

Last week, Alan Greenspan described the increase in the share of the financial services sector in gross domestic product around the world. He applauded its contribution to the growth of economic activity and world trade.

But there are two ways of interpreting such data. The shares of both health expenditure and information and communications technology in GDP have also increased. But while public and politicians welcome the expansion of ICT, recognising its contribution to daily life and the efficiency of business, they fear the growth of health spending. The benefits, though real, are hard to measure, do not necessarily increase in line with the costs of provision, and threaten to squeeze other elements of spending. Are financial services like computers or like medicine?

The share of ICT spending has risen less than you might expect because the prices of these services have been falling. The share of health spending, however, has risen largely because the price of these services has been rising.

Mr Greenspan’s own data also note that the contribution of financial services to US GDP rose sharply between 2008 and 2009. This was true in Britain also: Andrew Haldane of the Bank of England has pointed out that the largest increase ever recorded in financial services output was in the fourth quarter of 2008, during the Lehman bankruptcy and the collapse of two major British banks. But the economic contribution of the financial sector to the growth of national income in 2009 was plainly negative: dislocation of the supply of financial services led to the largest fall in GDP since the second world war.

To understand what is going on it is, unfortunately, necessary to dig in to how such data are compiled. Standard accounting conventions do not work well when applied to financial services. The problems fall into two main groups.

National accounts distinguish operating profits from investment and trading gains. Unlike industrial and commercial companies, for financial businesses, that distinction is not clear. The modern investment bank relies heavily on trading profits, and the distinction between the profits from trading and those from market-making is blurred. An insurance company relies on returns from invested funds derived from premiums paid well before claims. A bank offers “free” banking paid for by the interest spread between deposit interest and lending rates.

Each of these transactions poses reporting problems which officials who meet to agree the United Nations system of statistical accounts have debated for years. Their recent solution is an adjustment called financial services indirectly measured (FISIM). This adjustment is a major component of the data Mr Greenspan cites and FISIM is the main explanation of the data peculiarities of 2008-9.

The second issue is that accounting systems are not well adapted either to measuring the costs of risk-bearing or the rewards to it. There is frequent asymmetry between the treatment of gains and the treatment of losses. Returns to increased leverage tend to show up simply as an increase in output. This is a general problem but one which is focused on financial services.

The figures we have for the contribution of financial services to GDP are best regarded as a record of growing inputs, like the shares of health expenditures, rather than a measure of growing outputs, like ICT spending. The rise in recorded share largely replays the well-known fact about financial services – that earnings from it have risen a great deal. British statisticians are engaged in an important attempt to find measures of public service output directly. It would be useful to try to do the same for financial services.

Until that happens, it is well to recall the warning of Josiah Stamp, once both director of the Bank of England and president of the Royal Statistical Society. The statistics governments relied on were, he observed, ultimately derived from the records of the village night watchman, “who just put down whatever he damn pleased”.

Print Friendly, PDF & Email