May 01, 2002

Information Technology Sets the Pace

US productivity growth is bouncing along nicely. From the beginning of 1995 to March 2001 – the peak of the business cycle, according to the National Bureau of Economic Research – labour productivity grew at an annual rate of 2.77%, just over twice its tempo of 1.34% a year from 1973 to 1995.

Nearly all observers agree that the extraordinary wave of technological innovation in computer and communications equipment brought about the improvement. But will it last? I believe it will. Which begs another question: why is productivity growth in the European Union still lagging behind?

A chorus of voices agrees that America’s productivity "miracle" stems from the revolution in IT (information technology). Kevin Stiroh, an economist for the Federal Reserve Bank of New York, writes that "all of the direct contribution to the post-1995 productivity acceleration can be traced to the industries that either produce [IT capital goods] or use [IT capital goods] most intensively, with no net contribution from other industries… relatively isolated from the [IT] revolution."

Steven Oliner and Daniel Sichel, both Federal Reserve Board economists, declared in 2000 that "the rapid capital deepening related to information-technology capital accounted for nearly half of this increase" in labour-productivity growth, with powerful "additional growth contributions com[ing] through efficiency improvement in the production of computing equipment."

Robert Gordon of Northwestern University believes in the "capital-deepening effect of faster growth in computer capital...[that] in the aggregate economy accounts [for a] 0.60%… acceleration [in labour-productivity growth]."

Only McKinsey has an alternative explanation for faster US productivity growth. But on close examination, the consulting firm’s analysis seems based on the successful use, rather than just the raw installation, of data-processing and data-communications equipment.


Pundits and economists are divided over whether this IT-driven boost to productivity growth will continue. Sceptics, such as Julie Kosterlitz of the National Journal, talk of how growth in the late 1990s was unsustainably fast due to the Nasdaq stock market bubble. Joseph Stiglitz, a Nobel prize-winning economist, talks of how there must be "real restructuring" because the "capital overhang" produced by excessive investment has created "overcapacity" and "real restructuring takes time".

But a smaller group of economic forecasters think productivity is likely to continue growing as fast as it did in the 1990s. They argue that three factors in the IT revolution have together speeded up productivity growth.

First, the price of IT capital goods has fallen rapidly as the technology used in them has improved. Second, the share of spending on the IT capital stock leapt in the 1990s. This higher spending, coupled with the lower prices, accelerated the pace of growth of this stock. Third, the share of total income attributable to the returns on existing IT capital stock also jumped. This means that the faster pace of growth of the stock produced even faster growth in labour productivity. The impact on the economy of this continuing technological revolution is proportional to the strength of these three factors multiplied together. Productivity growth accelerated because all three factors improved rapidly in the 1990s.

If you subscribe to this analysis, the speed of growth can fall only if the IT revolution slows, the share of total expenditure of IT goods falls, or the share of total income attributable to the higher productivity made possible by IT declines. Optimists think none of these three things will happen.

On the contrary. They point to four factors that could further brighten prospects for future productivity growth. It is improbable that all of these four will materialise. But it is also highly unlikely that none will. The consensus forecast that US productivity growth will continue at the same clip as in the recent past is probably too pessimistic.

The four reasons for optimism are as follows. First, the demand for IT goods is highly price-sensitive. A 1% fall in the price of IT goods is estimated to lead to boost demand by 2-3%. There is no reason to believe this elasticity will fall soon. It means that as prices fall, the share of spending on IT rises, boosting growth rates above today’s estimates.

Second, the share of income attributable to the returns on existing IT capital stock has been rising for years. A sudden end to this trend would be surprising. A rising share of national income attributable to IT would boost growth further, as long as prices continue to decline.

Productivity usually falls in recessions. Yet US productivity rose at an annual rate of 5% in the fourth quarter of 2001, even though hours worked by US workers fell at an annual rate of 3.3%

Whether the share of spending on IT and the share of income attributable to returns on installed IT both continue to rise depends on how many computers people buy as prices drop.

So far, the evidence shows that each price decline brings forth a more than proportional increase in demand, as new uses for processors and networks are found. Just after the Second World War, when IBM’s president foresaw demand for "five, maybe six" computers worldwide, he did not imagine that cheaper mainframes would stimulate a host of new uses, such as running databases and performing calculations for personnel departments, airline-reservations systems and life-insurance companies. In the heyday of the batch-process mainframe, few imagined that declining computer prices would trigger a host of new, interactive uses, such as the cheap time-sharing machines that spread through business, education and government. Those who ran time-sharing mini-computers did not imagine that the next generation of computer-price declines would spur a host of new uses for the spreadsheet-equipped microcomputer. Sun Microsystems saw dimly – but was unable to capitalise on – the explosion of new uses that arose as more powerful microprocessors and better networking generated the narrowband Internet. We do not know what the killer applications of the next generations of faster, cheaper computers and networks will be. But it seems highly likely that they will exist – and hence that the expenditure and income shares of IT yet to peak.

Third, the sharp rise in the US economy’s IT capital-to-output ratio in the late 1990s raises the possibility that effort that would otherwise have been devoted to increasing output was diverted to figuring out how best to organise the new computer-intensive business systems. If true, this suggests that US growth in the late 1990s was below its long-run trend.

Finally, Paul David, an economist at Stanford university, has long argued that the boost to labour productivity from the computer revolution will show up in two stages. The first comes through "capital deepening", as the cheapness of machines leads us to use computers as improved typewriters and filing cabinets, of which we buy more; the second as a process of social learning about how to use these new technologies most efficiently, a process that drives rapid total factor productivity growth for a long time, but that cannot begin until computers are ubiquitous.

The course of US productivity growth during the recent recession is another reason for optimism. Productivity usually falls in recessions: firms hoard labour that they cannot use productively in order to have a pool of trained workers when demand picks up. Moreover, when demand is slack, the pressure to find ways to economise on resources in order to expand capacity is absent. Yet US productivity rose in the 2001 recession. In the fourth quarter of 2001, hours worked by US workers fell at an annual rate of 3.3%. But production rose at an annual rate of 1.7% and productivity rose at an annual rate of 5%.

A politician may claim that such growth proves that there never was a recession – although rising job losses suggest otherwise. But an economist interested in long-term forecasting sees this as evidence that the automation and investment in IT that fuelled such rapid real-wage and production growth in the late 1990s is still occurring. Since business is not under immediate pressure to boost capacity, this suggests that underlying computer-driven productivity growth is rapid.

But those who (like me) are optimistic about future productivity growth remain puzzled. If IT is revolutionising the economy, why is productivity growth in western Europe not accelerating?

One possible answer is that long-term US productivity growth has not in fact risen. Perhaps the US is experiencing a short, cyclical boost to output rather than a long-run upward movement in productivity. Another is that western Europe’s productivity miracle may yet be coming, but that it takes time for new ways to spread. A third – Alan Greenspan’s favourite – is that business will only invest in more efficient computer-intensive modes of organisation when they see potential cost savings from doing so. European business may not be investing enough in IT because it does not see any leeway to cut costs by firing workers. Thorough reform may therefore be needed before Europe’s economies can reap large aggregate benefits. A final explanation is that governments’ failure to cut red tape – not rigid labour markets and slack demand – is slowing the spread of the IT revolution in Europe. After all, the revolution is well under way on the periphery of western Europe, from Ireland to Finland.

As yet, there is no answer to this conundrum. But western Europe needs to find out soon why it is lagging behind if it is to match the strong productivity growth that the US seems to enjoy.

J. Bradford DeLong (2002), "" Worldlink (May/June).

Posted by DeLong at May 1, 2002 07:58 AM | TrackBack

Post a comment