With GDP, facts depend on how you measure them

US gross domestic product was increased last year by much more than the growth of the economy. This sleight of hand was achieved by changing the way GDP is measured. The UK is due for a similar make-over this year. Reality won’t change and we need to be alert to the comments of those who will think it has.

“Nearly all scientists believe that there is a clear-cut distinction between fact and theory…William Whewell (1794-1866) denied that any such sharp distinction existed.” Peter Medawar, the Nobel laureate, whom I am quoting, agreed with the denial; and, as GDP data are generally considered to be facts, the revisions show that Whewell was spot on. The calculation of GDP depends on the theoretical model on which it is based. The change in GDP involves a change in the model being used and, in my view, the new model is worse than the old one.

Two major changes were made last year in the presentation of US GDP data. One redefined research and development as final rather than intermediate output; the other was to include in the assets of the household sector the value of their pension assets. In this blog I am going to consider only the first of these.

In its preview of the changes, the US Bureau of Economic Analysis wrote: “Conceptually, the value of an enterprise’s R&D is equal to the present value of the future benefits that the company derives from R&D.” I doubt whether the author realised that this statement is contentious on two grounds. First, it implies that the value of other sorts of spending designed to provide future benefits, such as advertising, should not be included; and, second, it assumes that any value added to an individual company is not offset by reductions to the value of others. That is, it assumes that the implied model does not involve a “fallacy of composition”.

It is interesting to compare the BEA’s valuation of R&D with the comment attributed by Lucy Prebble, in her play, Enron, to the character of Jeffrey Skilling, who is discussing his company’s accounting policies. “ …mark-to-market. What does that mean?… if you sign a deal … that future income can be valued, at market prices today, and written down as earnings the moment the deal is signed.”

Unkind people might claim that Enron accounting has now gone national in the US and is about to do so in the UK.

If R&D is worth the present value of the future profits that the individual company is expected to make from it, we need to ask why this should not be applied more widely. Advertising is one obvious example and so, to a large extent, are the pay packets of chief executives. They are seldom out there selling or back at base working the equipment; they are largely paid on the assumption that the company will, because of their current stewardship and guidance, be more profitable in the future than it would be if a less well paid person was in charge.

Advertising provides an excellent example because the idea of including it, like R&D, as a final rather than an intermediate output has already received semi-official blessing. In 2006 the US Federal Reserve Board in its Economics Discussion Series published a paper by Carol Corrado, Charles Hulten and Daniel Sichel called Intangible Capital and Economic Growth. This proposed that both advertising and R&D should be treated in GDP data as final rather than intermediate expenditures.

Most of us would, I think, consider that increasing GDP to include money spent on advertising would be silly as it is part of the normal expense of running a company in a competitive world. Those that don’t advertise lose out to those that do. It’s a zero sum game in which money may be well or badly spent, and the winners gain at the expense of the losers, but there is no net gain to the economy.

Is money spent on advertising different from money spent on R&D? I think it is, in an important way, but nonetheless it should not be designated as final output in the GDP data. R&D is different from advertising because it can increase the future rate of growth of real wages but it remains a zero sum game for companies. Its positive value for the winners is matched by the negative impact on the losers. The value of successful R&D is in improving the economy’s potential growth. Enron’s accounting confused today with the future; national accounts should not, I think, make the same error.

Successful R&D will typically take one of two forms. It may lower the cost of production of existing products or introduce a new and better product. Either way, it will improve the profits of the successful company but reduce the profits of the rest. This is fairly obvious when new and better products are brought to market and displace current ones; but it also applies when the result is an improvement in productivity. Real wages rise over time with output and this depends crucially on the rate at which productivity improves. The wages that companies pay are broadly similar to those paid by others. If one company improves its productivity, then it will benefit because the wages it pays will barely rise as a result. A rise in productivity will, however, increase wages generally. R&D that is successful in improving one company’s productivity is therefore just as much a zero sum game as advertising. An increase in productivity will raise real wages but, spread over the whole economy, it will go largely unnoticed. There will, however, be a large rise in profits of one company at the expense of tiny falls in all the others. Valuing R&D on the basis of its future profits is thus a fallacy of composition. It consists of counting one obvious benefit but forgetting the numerous but individually tiny reductions in the profits of others.

The result of these changes is not very great. Chart one shows that GDP has been increased by a bit over 3 per cent and private investment has been increased by between 200 and 220 basis points of GDP.

Even though the effect is small, the new presentation of the “facts” is, I think, a retrograde step.

(i) The model that underlies the new data compares badly in my view with its predecessor. It is subject to a fallacy of composition.

(ii) Unless and until similar changes are introduced in other economies, it gives a false view of the US, and will shortly of the UK, in comparative terms. For example, the US now overstates the proportion of its GDP that is invested compared with that of other countries.

(iii) We can be confident, I think, that changes would not have been introduced had they led to a lower measure of GDP, and this will encourage further changes. Unless such changes are challenged now, we will find GDP becoming increasingly meaningless as a measure of national resources, being boosted by another redefinition of the difference between final and intermediate output – with advertising as one prominent candidate