Date created: 5/4/05 Last modified:5/4/05 Maintained by: John Quiggin John Quiggin
29 November 2004
In any assessment of the strengths and weaknesses of the US economy, productivity growth must play a central role. Most attention has been paid to positive assessments of US performance. There is plenty of good news on which to base an assessment. US productivity as measured by output per hour worked rose at an annual rate of 2.5 per cent during the dotcom boom from 1995 to 2000, considerably faster than the average for the previous 25 years.
Recessions are usually bad for productivity, as employers tend to keep workers on, even though there may be little for them to do. But no such effect was present during the US recession or the subsequent jobless recovery. Since 2000 output per hour in the nonfarm business sector has risen at an annual rate of more than 4 per cent.
However, there have been more critical interpretations of the data. A focus on productivity growth measured in terms of output per hour worked is common, but on this criterion, the United States is merely catching up to a number of European countries. The world leader in output per hour is Belgium, not normally considered an economic dynamo.
The problem here is that Belgium also has a low employment-population ratio. A lot of potential workers are either unemployed or not in the labour force, and these are presumably the less-skilled. The fact that these workers are excluded from the total increases the average, in the same way as British manufacturing productivity grew strongly under Margaret Thatcher, when large numbers of less efficient factories were forced to close.
There does not seem to be a great deal of merit in attaining high productivity levels by this route. Increased employment might reduce average output, but it would increase total output and most unemployed workers (as well as many of those who have left the labour force) would be better off in low-wage, low-productivity jobs than in their current position.
The ‘Belgium effect’ was not a problem for the United States during the ‘dotcom’ boom, which ran from 1995 to 2000. During the boom, output per hour grew strongly, at the same time as employment surged, and average hours of work increased.
On the other hand, there is increasingly strong evidence that the apparent acceleration of productivity since the end of the dotcom boom is due to the exclusion of low-productivity workers. Over the period since 2000, the employment-population ratio has fallen from a peak of nearly 65 per cent, to a low of 62 per cent. There has been a slight recovery recently, but this has been associated with a slowdown in productivity growth.
As was widely noted during the recent election campaign, George Bush is the first US president since Hoover to have presided over a decline in aggregate employment. This is a surprising outcome since the recession was a relatively mild one, and output growth has been reasonably strong.
It is a matter of simple arithmetic that increasing output and declining employment must imply strongly increasing output per worker. Given that average hours worked have remained fairly stable, or declined slightly, this also implies strongly increasing output per hour.
But standard economic analysis suggests that, if productivity is increasing for all workers, employment should be increasing, not declining. A popular explanation of the observed combination of increasing (measured) productivity and declining employment is that fewer workers are needed to produce the same output. But this would make sense only if total output was constrained, for example, by inadequate demand. In the United States, by contrast, growing demand has been reflected in rapidly increasing imports.
An alternative, but related, criticism has been made by The Economist magazine. The Economist looks at multi-factor productivity, the measure most commonly used in discussions of Australia’s productivity performance. On this measure, as assessed by the OECD, the United States looks much more like the European countries with which it is commonly compared. In particular, France looks almost identical to the United States.
In both France and the United States, the rate of growth of multifactor productivity accelerated in the late 1990s. France managed multi-factor productivity growth of 1.4 per cent between 1995 and 2002, compared to 1.2 per cent in the United States. These estimates are presented by the OECD, and don’t seem to match well with national estimates (the Australian numbers are different from those produced by the ABS, though both show strong growth). In the present case, this is inevitable, since the US multi-factor productivity statistics haven’t been updated past 2001.
The Economist’s focus on multifactor productivity has been criticised by economist Brad DeLong, of the University of California, Berkeley, who argues that a major source of productivity growth is technological progress ‘embodied’ in new and improved capital goods such as computers. An improvement in the quality of capital goods is reflected in a declining quality-adjusted price. This means that, for a given level of investment spending, the quality-adjusted addition to capital stock is greater. Hence, embodied technological progress is reflected in enhanced labour productivity, but no change in capital productivity.
DeLong is right to argue that the productivity growth of recent years has been driven, in large measure, by improvements in computer and telecommunications technology, and that a measure that takes no account of this cannot be regarded as satisfactory.
On the other hand, there are two (inter-related) reasons for using MFP measures, at least in the context of international comparisons. First, since computers are commodity items nowadays, the technological progress they embody is available, on more or less equal terms, to all developed countries, with only modest time lags.
A striking illustration of this point is given by looking at the proportions of households with access to PCs and the Internet. The gap between technological leaders like the United States, and transition economies like those of Eastern Europe is less than a decade. Household internet penetration in the Czech Republic, for example, is currently 15 per cent, the level achieved by the United States in 1997. The time taken for innovations in this area to diffuse among the leading developed nations is often a matter of months, rather than years. It follows that international comparisons are unlikely to be subject to any significant bias arising from failure to take account of embodied technical change.
A second, and related point is that there are big differences in the ways in which statistical agencies different countries take account of changes in the quality of computers, and other forms of embodied technological progress. The United States uses a method called ‘hedonic regression’ which involves pricing characteristics (in the case of computers, these would include processor speed, disk drive capacity and so on) rather than specific items. By contrast, most European countries use older methods involving matching specific models. Partly because of this difference in methods, US statistical agencies generally tend to be more aggressive in quality adjustment than their European counterparts.
Most economists think that the US approach is more accurate. Regardless of which approach is right, the effect of the different approaches is to produce an upward bias in estimates of US output growth, as compared to that of European countries. As The Economist points out, any bias of this kind is matched by a corresponding bias in estimates of the volume of capital investment. In the case of MFP estimates, these two biases largely cancel out, since an overestimate (or underestimate) of the value of computers biases both the input and output measures. When these biases are eliminated, international differences in estimates of productivity growth are greatly reduced.
The resolution of these seemingly technical debates is of great importance in projecting likely future developments in the world economy. The United States is currently experiencing large and growing deficits in goods and services trade, currently equal to about 6 per cent of GDP. Such deficits cannot be sustained for more than a few years, since consistent trade deficits inevitably produce exploding current account deficits and foreign debt.
The optimistic interpretation is that the United States can accumulate debt now because international lenders anticipate high rates of productivity growth, which will permit the rapid growth in output needed to produce trade surpluses in future, thereby permitting the debt to be serviced. Against this, it may be observed that private foreign investors are increasingly unwilling to invest in the United States or buy US securities. Their place has been taken by Asian central banks, particularly the People’s Bank of China, which is trying to avoid an upward revaluation of the yuan against the dollar.
The dollar has already declined against other currencies, such as the yen and euro, and has now reached a record low against the euro. Combined with strong productivity growth, relative to the rest of the world, this ought to produce a rapid turnaround in the US trade deficit. Whether it will do so remains to be seen.
John Quiggin is an Australian Research Council Federation Fellow in Economics and Political Science at the University of Queensland.
Read more articles from John Quiggin's home page
Go to John Quiggin's Weblog