The American economy - Slow road ahead

Oct 26th 2006 | WASHINGTON, DC
From The Economist print edition
America's long-term potential rate of growth is falling, perhaps to its lowest pace in over a century

EVERYONE knows that America's economy is slowing. Thanks to the bursting of the housing bubble, overall GDP growth has fallen back sharply. The biggest short-term uncertainty for the world economy is whether American consumers stop spending and drag the country into recession. But beyond the business cycle, another slowdown has received scant attention. America's potential rate of growth—that is, the pace at which annual output can expand without pushing up inflation—is also falling. By some estimates, it could drop to 2.5% over the next few years, which would be the slowest pace in over a century.

If that happens, the consequences will be serious. Tax revenues will grow more slowly than expected. Monetary policy will become harder to manage: as the 1970s showed, inflation can get out of control if central bankers do not realise that an economy's speed limit has fallen. Financial markets will be disturbed as conventional wisdom adjusts from an assumption of 3-3.5% potential output growth, and investors downgrade their expectations.

Potential output is hard to estimate, let alone predict. That is because an economy's trend growth rate cannot be measured directly. It has to be inferred. Over the long run economic growth depends on two things: increases in the supply and productivity of labour. The growth of labour supply, in turn, depends on the growth of the working-age population, the proportion of people who work and the number of hours they put in. The pace of productivity growth depends on capital investment, improvements in business processes and technological innovation. By looking at such trends, economists can estimate future potential output.

Although it generates precise-looking forecasts, this kind of “growth accounting” is fraught with difficulty. Both labour supply and productivity growth bounce around during business cycles. The share of people willing to work may fall in a recession, for instance, as discouraged people temporarily drop out of the workforce. Once job prospects pick up, they might return. Productivity growth is usually higher at the beginning of an expansion than at its end as firms work their existing employees harder before hiring new people. As a result, potential output can temporarily diverge from its underlying trends, making it even harder to estimate.

Nonetheless, the broad post-war history of America's underlying growth rate is clear. In the 1960s potential output accelerated to around 4% a year, largely because more women got jobs. In the early 1970s, for reasons that are still ill-understood, productivity growth slowed sharply, pulling down the trend rate of growth. Between the mid-1970s and mid-1990s America's economic speed limit was about 3%. Around half that growth came from an expanding workforce; the other half from productivity growth.

Thanks mainly to higher productivity growth, but also to a rise in the number of Americans working, trend growth rose suddenly in the mid-1990s. After the 2001 recession, productivity growth accelerated again, while the growth of labour supply slowed sharply. The share of working-age Americans in jobs fell after rising almost continuously for over four decades (see chart 1). This fall was widely interpreted as temporary, a sign that the recession was deeper than it appeared. But after five years of expansion, it has not been reversed, suggesting (although the evidence is still tentative) that structural changes are afoot. These labour-markets shifts are the main reason to be pessimistic about America's potential output growth.

So why is the proportion of Americans who work falling? For three reasons. First, the baby-boomers are heading towards retirement. The share of people aged between 55 and 64 has risen from 10.5% in 1995 to 13.3% in 2005 and is likely to reach over 16% by 2015. People over 55 tend to work much less than younger folk.

Second, the rush of women into the workplace has stopped. The proportion of women working rose from below 40% in 1960 to a peak of over 60% in 1999. It has subsequently fallen slightly.

Third, the rate of teenage employment has plunged. In the 1990s over 50% of young people aged 16-19 had jobs. Today just over 40% do, the biggest drop since records began. This decline is a bit of a mystery, since job growth in the kinds of industries that tend to employ young people—restaurants and shops—has been well above the national average. It may have happened because teenagers are staying at school or college longer, and are working less on the side. More education may mean higher future productivity, but in the medium term it cuts the number of available workers.

If economists at the Federal Reserve in Washington, DC, are right, these three components are likely to result in a bigger change than has hitherto been expected. A recent study* suggests that America's trend rate of labour-force participation could drop by a further 1.4 percentage points in the next four years, to just over 64%. By combining these projections with the Census Bureau's estimates for the growth in the working-age population, they calculate likely changes in the overall supply of workers. By 2010, the Fed economists reckon, labour supply in America will be rising by a mere 0.4% a year, well under half its current rate.
The geezer difference

These projections could be too gloomy. Washington's official number-crunchers, such as the Congressional Budget Office, predict that labour-force growth will slow down, but not that dramatically. An unexpected jump in what Dick Berner of Morgan Stanley calls the “geezer labour force” could make a difference as ageing baby-boomers work longer. Men in their 60s are already the fastest-growing segment of the workforce, although this has not made up for the overall decline.

A rise in immigration could increase the supply of labour, too, particularly since the proportion of immigrants who work is higher than that of native-born Americans. As the labour shortage begins to bite, the demand for immigrants should rise. But the politics of the issue argue against a big rise in the next few years. The government is under increasing pressure to crack down on employers who hire illegal workers, and Congress recently passed legislation to build a fence along a large stretch of the Mexican border.

It seems very likely, then, that America's labour supply will grow more slowly. And if that happens, potential output growth will too, unless productivity growth accelerates.

Unfortunately, the latest evidence suggests that, if anything, productivity growth is slowing unexpectedly. Over the past two years, in the non-farm business sector, it slowed to an annual average of 2% from an average of almost 4% in the previous three years.

And even that figure may be too high. The Bureau of Labour Statistics recently said 810,000 more jobs were created between March 2005 and March 2006 than they first thought. Thanks to those extra jobs the productivity figures for 2005 may be revised down by a further half of a percentage point. For the economy as a whole, the figure is some 0.3-0.5 percentage points lower than the official productivity figures suggest, because productivity growth in farming and government, which are left out of those figures, is even lower.

The longer the economy's expansion goes on, the slower productivity growth is bound to be. But the pace of the deceleration has begun to raise concerns. The small band of economists who study these things now agree that underlying productivity growth rose sharply in the late 1990s, from around 1.5% to some 2.8%. Three of those experts, Dale Jorgenson, of Harvard University, Kevin Stiroh, of the New York Fed and Mun Ho, also of Harvard, have calculated that over 60% of the late 1990s productivity surge was related to information technology. The industries that saw the biggest productivity gains were those that used IT most intensively.

The 2001-04 productivity surge is now the focus of an argument with big implications. Only 30% of that acceleration was related to IT, says Mr Jorgenson and his colleagues. Optimists take that as a good sign. It shows improvements in business processes spawned by the IT revolution are spreading through the broader economy. The economists reckon underlying business productivity growth is now around 2.5% (or around 2.2% in the overall economy). This is slower than in the late 1990s, but still far above historical averages.

Others are more worried. Robert Gordon, of Northwestern University, reckons that the post-2001 acceleration was the result of cost-cutting, not innovation. Other economists note that the pace of investment, particularly in IT, is much lower than it was in the 1990s. JPMorgan's calculations, for instance, show that firms' spending on IT equipment has grown by only 5.5% a year in this expansionary period, compared with over 20% a year in the late 1990s. Lower capital spending, they fear, could be a harbinger of slower productivity growth. Morgan Stanley's Mr Berner worries that high oil prices may also have hurt underlying productivity growth, by shifting the relative profitability of different sectors of the economy (much as the oil shocks of the 1970s did).

Mr Gordon sees the underlying rate of business productivity growth slowing to below 2.5% which, by his calculations, implies a rate of 2.1% for the overall economy. Coupled with the slowdown in labour supply, he concludes that America's potential growth rate could fall to 2.5%. Although their predictions for productivity and labour supply are different, JPMorgan's economists agree with his figure.

Mr Gordon's calculations suggest that 2.5% would be America's slowest economic speed limit in over a century (see chart 2). That is less surprising than it sounds. Though the statistics are less reliable for the years before 1950, it seems that in the early 20th century potential output grew rapidly thanks both to technological innovation and rapid immigration. Although actual output tumbled during the 1930s, America's potential growth rate did not. Some scholars argue that the 1930s were among the most technologically innovative years of the century.

Since no one foresaw the productivity revolution of the mid-1990s, these predictions could prove too pessimistic. The next “killer application” could be just around the corner. Even without rapid investment, the IT revolution may yield more productivity gains. But without some such happy chance, it looks as though America's potential growth rate is heading for a slowdown, at least for the next few years. The sooner that investors and policymakers wake up to this, the smoother the adjustment is likely to be.