What’s left?
May 10th 2001
From The Economist print edition

This year’s fall in American productivity might be seen as the death knell of the “new economy”. But its obituary would be premature

THE announcement on May 8th that America’s productivity declined in the first quarter of 2001 at an annual rate of 0.1%, compared with growth of more than 5% during the year to June 2000, is a blow for the IT-powered new economy. One by one, its claims to be special are being exposed as myths. Now it seems that the widely-held belief that America’s sustainable rate of productivity growth had doubled to around 3% was also mere myth. That does not mean, however, that the new economy was entirely hot air.

Its cheerleaders have certainly been muffled this year by the plunge in the Nasdaq high-tech stockmarket index, by the collapse of dotcom firms, by the slump in the profits of Internet giants such as Cisco (which this week reported its first quarterly loss in 11 years and a 30% decline in sales), and by signs that the American economy may be heading into recession. The new-economy sceptics, who long argued that computers and the Internet did not rate in the same economic league as electricity or the car, are now grinning smugly. They have a long list of unfulfilled promises to point to.

•The cycle. Top of the list is the idea that the traditional business cycle is dead. Now that America’s economy is slowing sharply and unemployment is rising, everybody is trying to deny that they ever made such a claim. Instead, they argue that IT helps to smooth the cycle. But even that claim looks suspect: if America’s GDP growth this year slows to the average forecast of 1.5%, then the decline in growth from 5% last year would be very abrupt indeed. And because of the excesses that have built up during the boom years—such as too much investment and too little saving—there is a high risk that the downturn could be much deeper. If America’s GDP growth were to fall below 1% this year, it would be the sharpest slowdown between any two years since the 1974-75 oil crisis. So much for smoothing.

•Inventories. One reason why IT was supposed to smooth out the economic cycle was that fancy computer systems and instant information would allow firms to ensure that production never got too far ahead of sales, and would thus avoid an excessive build-up of inventories. Yet despite spending millions on cutting-edge software, firms failed to spot the slowdown late last year, and now they have warehouses bulging with unwanted stocks. Cisco was supposed to be the very model of modern electronic business, using the Internet to make sales, order components and monitor inventories. But it, too, has been badly caught out. The firm was forced to write down $2.2 billion-worth of excess stocks in the three months to April—equivalent to almost 50% of its sales during the period.

Across the economy as a whole, “just-in-time” inventory management has reduced the ratio of inventories to sales. But it cannot guard against excessive stocks when shocks occur. It can only ensure that for any given shock, the error is smaller than it would otherwise have been. So far, the build-up in stocks is modest given how sharply demand has fallen. But it is far from negligible.

•Investment. The third myth of the new economy was that IT spending was recession-proof. Demand for IT equipment, it was argued, would continue to grow briskly throughout any downturn in the old economy because the pace of innovation was so fast that existing IT equipment would rapidly become obsolete. Firms would be forced to keep investing merely to stay competitive.

In reality, as profits have fallen, firms have decided that they can easily delay buying new PCs or upgrading their e-mail systems. In the first quarter of this year, investment in IT equipment and software in America fell at an annual rate of 6.5% in real terms, after rising at an average annual rate of 25% during the period 1995 to 2000. New orders for electronic goods fell at an annualised rate of almost 20% in the first quarter, signalling that bigger cuts are in the pipeline.

•Profits. Myth number four was that corporate profits would continue to rise rapidly for years to come. Average profits now look set to fall by at least 10% in 2001. Late last year, equity analysts were forecasting average long-term profits growth of 19%. That has now been revised down to 12%, but it still looks far too optimistic. It implies that profits will account for an ever-rising share of GDP. Yet historically, profits have tended to increase broadly in line with nominal GDP. If anything, the ratio of profits to GDP should now be falling slightly, because IT is likely to trim profit margins by increasing price transparency and shifting power from sellers to buyers.

•Share prices. The fifth (and silliest) claim was that in this new world of rapid technological change, old methods of share valuation had become irrelevant. Profits were for wimps, it claimed. Falls of 90-100% in the share prices of loss-making dotcom firms show that profits do matter after all. During previous technological revolutions (spurred by the invention of the railways, electricity or the automobile, for example) share prices similarly soared and then tumbled. But this time prices rose to a higher level in relation to profits than ever before in history.

The productivity puzzle

What about the most important claim of the new economy, namely that investment in IT has lifted productivity growth? The fall in productivity this year appears to confirm that this was also a myth. But America’s productivity gains cannot be dismissed so easily. Recent estimates of sustainable productivity growth were almost certainly too optimistic, but there is still reason to believe that some of the productivity gains will survive even as the IT boom turns to bust.

Labour productivity growth in America’s non-farm business sector rose to an annual average of almost 3% over the five years to 2000, up from 1.4% between 1975 and 1995. This faster productivity growth has been the lifeblood of claims about the new economy. It helped to deliver faster GDP growth with low inflation, higher profits and a large budget surplus.

One of the hottest debates in economics for the past few years has been about how much of this increase in productivity growth was structural and how much of it was cyclical. During boom times firms tend to work their employees harder, producing a cyclical rise in productivity growth, which then declines during the subsequent recession.

Robert Gordon, an economist at America’s Northwestern University, and one of the most outspoken new-economy sceptics, reckons that outside the manufacture of computers and other durable goods there has been no increase in labour productivity growth after adjusting for the effects of the economic cycle. As a result, he estimated last year that America’s structural productivity growth was somewhere around 2%.

At the other extreme, two studies published early this year—one by the president’s Council of Economic Advisers and the other by the Federal Reserve—both conclude that virtually all the increase in labour productivity growth since 1995 has been structural, putting the sustainable rate at close to 3%.

It is hard to believe that none of the increase in productivity growth was cyclical, given the strength of the recent economic boom. On the other hand, microeconomic studies appear to confirm that a sizeable chunk of the increase has been structural. A study by Kevin Stiroh, an economist at the Federal Reserve Bank of New York, examined productivity growth in individual industries. Mr Stiroh found that those industries which invested most in IT in the early 1990s saw the biggest productivity gains in the late 1990s. If the increase in productivity growth had been largely cyclical, the gains should have been more equally spread across industries.

By late last year, as the reported productivity growth figures looked stronger and stronger, a consensus emerged among most economists, including those at America’s Federal Reserve and the OECD, that America’s structural rate of productivity growth had increased to around 3%.

By itself, the fall in productivity in the first quarter sheds little new light on the issue because quarterly figures tend to jump about a lot. Also, productivity growth always dips as growth slows because firms delay cutting jobs—although the sharp rise in joblessness in America last month suggests that employers there are now wielding the axe. Productivity has held up quite well so far compared with previous downturns: it is still 2.8% higher than a year ago. But we need to wait for a full economic cycle to get the complete picture.

If the fall in productivity growth over the past year is only cyclical (and due to weaker demand) then, once the economy recovers, productivity growth should resume its faster trend. What is becoming increasingly clear, however, is that the productivity gains in recent years were not only exaggerated by the cyclical impact of the economic boom, they were also inflated by an unsustainable IT investment boom that has now turned to bust.

Investing in IT can increase labour productivity either by increasing the amount of capital employed per worker (ie, “capital deepening”) or by speeding up total factor productivity (TFP—the efficiency with which both capital and labour are used). Unlike previous spurts in America’s productivity growth, the recent one has been unusually dependent on capital deepening rather than on TFP. A study last year, by Stephen Oliner and Daniel Sichel at the Federal Reserve, concluded that almost half of the acceleration in productivity growth between the first and second halves of the 1990s was due to capital deepening. If that is so, a decline in IT investment will have worrying implications for future productivity growth.

Deepening concerns

The IT investment boom was unsustainable. The initial rise in productivity growth during the 1990s generated bumper profits, which encouraged firms to become over-optimistic about future returns. At the same time, the stockmarket bubble pushed capital costs down to virtually zero. The inevitable result was over-investment. Credit Suisse First Boston (CSFB) estimates that American firms have overspent on IT equipment to the tune of $190 billion over the past two years.

Now, as profits plunge, firms are being forced to cut their investment plans. CSFB suggests that if firms try to eliminate the surplus over two years, real IT investment will need to fall by an average of 16% this year and next. If firms try to extend the life of their existing IT equipment during the downturn, then new investment would have to fall by even more.

One year’s dip in investment would have only a modest impact on structural productivity growth if capital spending were then to bounce back strongly. More important is the longer-term trend in investment. America’s business capital stock has recently been growing at a real annual rate of over 5%, well in excess of the growth in GDP. Jan Hatzius, an economist at Goldman Sachs, estimates that the trend growth in the capital stock will now slow to 3-3.5%. If so, estimates Mr Hatzius, the contribution of capital deepening to labour productivity growth will be reduced to 0.75%, half its recent rate.

Mr Hatzius also expects the growth in total factor productivity to slow as the cyclical boost to productivity fades, and as efficiency gains in computer manufacturing diminish. One measure of the latter is the pace of price deflation for IT equipment. Price deflation accelerated in the late 1990s, partly because of increased competition and partly because of increased efficiency in the manufacture of chips. But it has since slowed. Computer prices fell by 16% in the year to the end of March, down from an average annual decline of 25% in the previous five years.

Putting all this together, Goldman Sachs has reduced its estimate of underlying productivity growth to a more modest 2.25%. That would be barely half the rate of productivity growth in 2000, but it would still be significantly higher than the 1.4% average annual rate recorded between 1975 and 1995. Only if investment stagnated for several years would productivity growth revert to its pre-1995 pace.

And that seems unlikely. The IT boom was mainly driven by falling IT prices that were following Moore’s law, which states that the processing power of a silicon chip roughly doubles every 18 months. This encouraged firms to substitute IT equipment for labour or other capital. Scientists believe that Moore’s law should hold for at least another ten years, allowing IT prices to continue to fall and hence encouraging further IT capital deepening, albeit at a slower pace than in recent years.

Yet to be revealed

Not only is the pace of technological innovation likely to continue to support productivity growth for at least another decade, but firms have yet to exploit fully the potential of their existing IT. It is not just the direct impact of computers and the Internet on productivity that matters, but also the ability of firms to organise their businesses more efficiently as a result. Just as the steam age moved production from the household to the factory, railways allowed the development of mass markets, and electricity made possible the assembly line, so IT allows for a more efficient business organisation.

Tossing aside all the hype, there are sound reasons for expecting IT to continue to boost productivity. By increasing access to information, it helps to make markets work more efficiently and it reduces transaction costs. Better-informed markets should ensure that resources are allocated to their most productive use. The most important aspect of the new economy was never the shift to high-tech industries; it was the way in which IT could improve the efficiency of old-economy firms.

A study by Alice Rivlin and Robert Litan at the Brookings Institution, a Washington think-tank, considers ways in which the Internet might further lift productivity growth. It examines how leading-edge firms are using the Internet to reduce transaction costs, to boost the efficiency of supply-chain management, and to improve communications with customers and suppliers. It then projects best practice across each sector.

The study’s tentative conclusion is that the economy as a whole can look forward to productivity gains from the Internet of 0.2-0.4% a year for the next five years. The potential cost savings look especially large in the health-care industry, where there is plenty of scope to improve the management of medical records and to communicate better with patients. Only 3% of Americans have so far corresponded with their doctor by e-mail; the rest still use more time-consuming methods, such as the telephone or a personal visit.

It is also worth remembering that not all of America’s productivity growth in recent years has come from IT. Some of it has been the pay-off from earlier economic deregulation designed to make labour, product and capital markets work more efficiently. Whatever happens to IT spending, these benefits will endure.

High-tech investment and productivity growth are, however, unlikely to return to the giddy pace of the past two years. The best guess is that productivity growth will average 2-2.5% per year over the next decade, still well above the pace of the 1970s and 1980s. Those who claimed that 3-3.5% growth was sustainable will be disappointed, but their expectations were far too bold. They implied that IT would have a much bigger economic impact than cars and electricity did in the 1920s, when annual labour productivity growth in the non-farm business sector averaged 2.5%.

Slower-than-expected growth in productivity has two implications for policymakers, however. First, after taking account of labour-market changes, it means the rate at which the Fed can safely allow the economy to grow without pushing up inflation is 3-3.5%, much less than in recent years. Second, it implies a smaller budget surplus—and hence less room for tax cuts. Back-of-the-envelope calculations by Goldman Sachs suggest that if productivity growth falls to 2.25%, the cumulative budget surplus in the years 2002-11 would fall to just over $2 trillion, as against the $3.1 trillion estimated by the Congressional Budget Office.

In recent years, economists have tended to exaggerate about IT. Either they have denied that anything has changed, or they have insisted that everything has changed. The truth, as ever, lay somewhere in the middle.

Future productivity gains from computers and the Internet will probably not be enough to justify current share prices, even after this year’s slide, but they will still matter. An increase in productivity growth of 0.5-1.0 percentage points per annum may not sound very exciting, but it will lift future living standards and make it easier for the government to pay tomorrow’s pensions. That, rather than the get-rich-quick culture of the dotcoms, is the true stuff of the new economy.

Sources

“Does the New Economy Measure up to the Great Inventions of the Past?”, by R. Gordon. Journal of Economic Perspectives, Volume 14, No.4, Fall 2000.

Economic Report of the President, January 2001.

“Estimates of the Productivity Trend Using Time-Varying Parameter Techniques”, by John Roberts, Federal Reserve Board working paper No. 2001-08.

“Information Technology and the US Productivity Revival: What do the Industry Data Say?”, by K. Stiroh, Federal Reserve Bank of New York Staff Report No. 115, January 2001.

“The Resurgence of Growth in the Late 1990s: Is Information Technology the Story?”, by S. Oliner and D. Sichel. Journal of Economic Perspectives, Volume 14, No.4, Fall 2000.

“The Economy And The Internet: What Lies Ahead?”, by R. Litan and A. Rivlin. Internet Policy Institute, November 2000.