Cold facts collide with new paradigm

The contribution of technical change to US growth has scarcely ever been as low as in the 1990s

IT IS NOW a commonplace to observe that we stand on the threshold of a new industrial revolution, based on the application of information technology in every corner of the economy. This, it is said, will transform productivity performance, creating a new era of rocketing GDP growth with minimal inflation pressures. Because more than three-quarters of all the connections to the Internet across the globe are located in America, it is usually taken for granted that the United States will benefit most from this new wave of technological advance. And because a new commercial infrastructure, based on information technology, is being spawned in the US, many people see the boom on Wall Street as a natural consequence of the "Third Industrial Revolution".

The only problem with this enticing tale (or "new paradigm, as it is known to its aficionados) is that it collides with a large number of cold facts about the way in which the US economy has actually behaved in the past decade. As we noted here a few weeks ago, American productivity growth during the 1990s has been no higher than the average for previous cycles, though in the past few quarters there have been some encouraging signs that this might have taken a turn for the better. More intriguingly, recent economic research has confirmed a surprising fact, normally kept well hidden from public view, which is that the contribution of technical change to American economic growth has scarcely ever been as low as it has been during the 1990s.

For those who find this scarcely credible, given the degree of hype which has surrounded the digital age, it is important to realise that technological advance is scarcely a novelty in a modern economy. In fact, the latest burst of technical change has been considerably smaller than that which occurred from 1870-1910, the application of which ushered in an exceptional period of productivity growth from 1913-72, and especially from 1950-64.

Robert J Gordon has just published a paper arguing that the whole of the period since 1870 should in fact be viewed as a single huge wave of technology, involving a bunching of technical progress up to 1910, the application of that change up to 1972, and then a return to a much more sedate pace of advance since then. (See The American Economic Review, Papers and Proceedings, May 1999.)

Gordon reckons that the key technical advances of the past 130 years or so all came in four clusters around the turn of the present century. These clusters were the development of electricity (motors, light and household appliances); the internal combustion engine (which created motor and air transport, transforming the economic geography of the world); the chemical revolution (creating industries such as petro-chemicals, man-made fibres and pharmaceuticals); and, finally, the radiowave revolution (giving birth to telephony, radio and television). A fifth cluster - centred around indoor plumbing, running water and sanitary waste disposal - was of a different variety, but may have been just as important in transforming economic welfare.

Although the outline of these innovations was faintly visible by about 1910, their profound combined effects on the structure of the economy were far from clear for many decades. No doubt this is partly because the investment boom of the 1920s was rather rudely interrupted by the Great Depression and the Second World War, meaning that it was not until the 1950s that the economy was able to benefit in full from the application of scientific advances which had occurred more than five decades previously. But, for whatever reason, there is no doubt that the delayed effects of scientific change boosted economic growth from 1950-72 to a far greater extent than it has done at any time either before or since.

This broad pattern has been long recognised by specialists in economic growth theory, but Gordon presents new calculations which confirm the point. Essentially, the contribution of technical change to GDP growth is calculated as follows. First, the contribution from extra labour (population growth plus the impact of education) and extra capital (new machinery, computers etc) is estimated. Then the remainder of growth is attributed to technical change. Gordon's new contribution to the debate is to present new calculations which adjust the labour and capital inputs for quality changes on a consistent basis throughout the period from 1870 onwards.

Obviously this methodology has the key disadvantage that all of the errors in the technique for measuring labour and capital inputs are automatically attributed to "technical change", but with Gordon's adjustments this should be less of a problem. And in any case the results are too dramatic to be explained away solely by errors in measurement or technique.

The shape of the great wave of technical advance during the 20th century follows a clearly delineated head-and-shoulders pattern. The "head" comes in the 1950-72 period, when technical change contributed a huge 1.8 percentage points per annum to the growth of GDP. Meanwhile, the "shoulders" come in the periods from 1913-1950, and 1972-79, when the contribution is around 1.0 percentage points a year.

This leaves the periods before and after the great wave. In these more "normal" periods, technical change has typically contributed an unimpressive 0.3 percentage points to the growth rate of GDP each year. In the latest period from 1988-96, when the Third Industrial Revolution has supposedly been in full swing, this has fallen to an almost-derisory 0.26 percentage points a year.

How can productivity growth have fallen in the digital age, given that computers now pervade every corner of our lives? The answer comes in another research paper (also published in The American Economic Review, May 1999) by Dale W Jorgenson and Kevin J Stiroh. This paper uses similar techniques to the Gordon study, but with the added feature that it specifically attempts to measure the direct contribution of computer technology to the growth of GDP in the postwar period. Not surprisingly, there was no contribution at all before 1973. Then, from 1973-90, computers contributed 0.11 percentage points per year to the GDP growth rate. And finally, from 1990-96 the contribution rose to a still-rather-miserly 0.16 percentage points per year. In other words, the computer age has indeed boosted the growth rate, but not by much.

We are therefore faced with two facts which are certainly at odds with those who believe in the "new paradigm". First, the pace of technical change has decelerated during the 1990s rather than the reverse. Second, the massive investment in computers has boosted growth by only a small amount. Take both of these forces together and they produce a total growth contribution of around 0.4 per cent a year, which is only a fraction of the contribution made by the "great wave" of technical application in the 1950s and 1960s.

This leaves one last refuge for those who believe in a Third Industrial Revolution. Just as the scientific advances of the late Victorian age did not have their full effect until the middle of the 20th century, it is possible that computer, Internet and digital revolutions will not be felt in the productivity figures until sometime in the 21st century. Maybe. But the one thing we know for sure is that it has not happened yet.