According to economic theory, however, traditional physical investment is not all that important for an economy's long-term growth. Professor Robert Solow of the Massachussetts Institute of Technology famously discovered that only a fifth of the improvement in output per head in the US could be explained by higher levels of capital.
Far more important is investment in technology and humans - or "know- what" and "know-how". As these are more intangible than factories, they are the least likely to be measured and included in official investment figures.
Take computer software, for example. This is a key investment in new information technologies and a crucial method of extending knowledge. Spending on software is also high and growing rapidly. According to the Computing Services and Software Association, members' revenues were more than pounds 7.1bn in 1994, pounds 1bn higher than the previous year, and up from pounds 1.3bn in 1985.
But unlike the physical computer casings, software is not yet included in the national accounts definition of investment.
This will change in time for the new millennium. As a result of a European directive, EU countries will all have to publish estimates of investment in computer software. The Office for National Statistics plans to start in 1999, and will also publish historical data, leading to upward revisions in earlier estimates of GDP.
The US already includes software spending in its investment figures. These show that the share of investment in information processing in total investment has climbed fairly steadily from 18 per cent in 1981 to 41 per cent in the latest quarter. The information processing category is currently growing at an annual rate of about 20 per cent and rising, compared with about 5 per cent and falling for total investment.
So far the Office for National Statistics here has produced only estimates of the output of the computer industry, hardware and software combined. It also attaches a serious health warning to these because it is not clear that the available price figures fully reflect the extent of price discounting and reductions.
However, noting the warning, computer industry output has grown at an average of 9.2 per cent a year since 1988, compared with average GDP growth of 1.7 per cent. The industry had a share in GDP of 1.2 per cent in 1990, and this can be assumed to be rising.
There are two particular reasons why software spending in particular will surge during the rest of this century. One is the year 2000 problem. Most computers in use, which use a two-digit abbreviation for the year, will need reprogramming to make sure they do not turn the clock back to 1900 when the year 2000 dawns.
David Owen, an economist at the City bank Kleinwort Benson, estimates that British computer users might have to spend pounds 5bn during the next few years to tackle this problem. (He cites the forward-looking state of Nebraska, which has announced a two cent tax on every packet of cigarettes sold in the state from July 1997 to help fund its own $31m pre-millennium computer bill.)
This spending alone, on top of any existing software spending, could add 0.1 per cent a year to GDP when the ONS includes it in the investment figures, Mr Owen predicts.
This calculation excludes any other increase in software spending. But there are likely to be other boosts too. For instance, the City of London is expected to spend millions updating trading systems for the introduction of the euro from 1 January 1999. Even if Britain does not join, London will have to be able to trade in the new European currency.
Besides, it is clear that the role computers play in the industrialised economies will carry on increasing. A forthcoming report from the Organisation for Economic Co-operation and Development looks at the available figures to assess the spread of information technologies. It estimates that more than half of GDP in member countries is "knowledge-based". Spending on software, now bigger than purchases of hardware, has been growing by 12 per cent a year since the mid-1980s, according to the OECD's figures. Trade in technology-related services increased by a fifth between 1985 and 1993. World-wide, the Internet gets 160,000 new users a month.
Despite the availability of such statistics, it is intrinsically hard to measure the increase in an economy's knowledge base, its real investment in technology and human capital. Technical advances in information technology in particular take many forms: new products, small changes in old products, new combinations, new connections in people's minds about how to use them. The spread of technology and knowledge is very diffuse and easily crosses national boundaries. Even reliable figures on software spending would not reveal who was using the programmes and how. Some surveys have made a stab at measuring such innovation by asking firms about their sources of information and equipment and whom they co-operate with. But findings are preliminary.
Similarly with human capital, it is possible to measure the number of years most people spend in education, or firms' spending on training. Yet this does not directly show how much they know and how effectively they use it in their work.
The OECD concludes: "The relationship between the creation of knowledge and economic growth is still virtually unmapped. In many respects the knowledge-based economy remains more of a concept than a measurable entity."
This is not to say that it is not worth improving the available figures, as the ONS intends to do. Nor does it mean that it is impossible to draw any sensible policy conclusions. It is now standard to call for better education and training, and the Labour Party in particular has made this a centrepiece of its long-term economic policies. Indeed, it appears to be groping towards an updated version of the "white heat of technology" enthusiasm for the information age.
However, what a "knowledge-based" economy needs is not people who have been trained to do specific things - use one type of software programme, say. Knowledge workers need to learn how to learn and how to be creative. As the OECD puts it, in economists' language: "Individuals must upgrade their skills in both codified and tacit knowledge continuously so as to keep up with fast-moving technologies." Perhaps they will eventually be clever enough to measure what they are doing.
* The Knowledge-Based Economy, OECD Observer, June/July.
The spread of information technology % of households, 1994
US Japan UK Germany France
Personal computers 37 12 24 28 15
Fax na 8 2 4 3
PC modem 15 na 4 3 1
Digital main lines 65 72 75 37 86
Passed by cable 83 na 16 56 23
Satellite na 27 11 20 2
Share of high technology goods in manufacturing output, %
US 18.2 24.2
Japan 16.4 22.2
France 12.8 18.7
Germany 15.3 20.1
UK 16.4 22.2