The US slowdown
I missed this piece by Brad de Long when it came out in December last year, but Jack Strocchi sent it to me recently, and I thought it would be worth responding to now. The headline The Slow Countries and lead-in
US productivity keeps growing – right through the bust. So what’s wrong with Europe?
give the general flavor. Here and elsewhere, Brad focuses on technology-driven change in ICT-producing and ICT-using manufacturing manufacturing has grown even faster, and argues that the US has proved clearly superior to Europe in this respect. To the extent that he’s a pessimist, Brad worries about the fact that, with slow output growth, all this productivity growth is driving employment down in the US. Brad suggests that Europe may catch up to the US in the next growth cycle, but worries that ‘something in the water’ (that is, a combination of restrictive labor market policies and deflationary macro policies) will prevent this. I disagree with a lot of this and, to sharpen things up, I’ll argue for the polar opposite case, that we’re going to see the US converging towards Europe in important respects over the next few years.
I’ll begin with some observations that are needed to make the case. I’m drawing heavily on the work of Stefano Scarpetta, Andrea Bassanini, Dirk Pilat and Paul Schreyer at the OECD (Brad has also cited Scarpetta’s work recently, in this piece about firm dynamics. He draws very different conclusions but is clearly impressed witht he quality of the work).
First, productivity in the leading European countries [such as France, Germany & Benelux], as measured by output per hour, caught up to the US around 1990.
Second, labour isn’t homogenous. Typically, the least-skilled workers are most likely to be unemployed so an increase in the employment-population ratio is often associated with a reduction in the ratio of the effective input of labour to the measured input. This means that strong employment growth tends to bias MFP productivity measures downwards. The bias in labour productivity measures is even stronger if capital inputs don’t keep pace with labour. The opposite effect known as the Thatcher effect, produces a spurious improvement in productivity associated with shedding of inefficient labour and closure of inefficient plants.
Third, the US had strong growth in employment during the 1990s and anecdotal evidence suggests that, at the height of the boom, even very marginal workers were being drawn into the workforce. Equally striking the US experienced increases in the average number of hours per worker, contrary to the trend of the past century and the experience of all other countries.
Fourth, the most relevant measure of productivity is not output per worker or output per hour but multifactor productivity (which takes account of capital as well as labour inputs). It’s hard to get direct comparisons of MFP between countries. However, it seems likely that the ratio of capital to labour inputs as of 1990 was at least as high in the US as in Europe. This implies that MFP was about equal in the US and Europe by 1990. From 1990 to 2000, as Scarpetta et al show, MFP growth rates in the US and Europe were about equal
Fifth, and the opposite of the situation with labour productivity, MFP growth has not held up through the recession in the US. As I pointed out recently, US MFP actually fell in 2001, the latest year for which data is available. The reason is simple. Output and employment were falling, but capital stock was not.
Sixth, and even more striking, MFP growth over the entire course of the bubble and bust was very weak. To make life easy for myself, I’ll choose 1996 (the year Allan Greenspan warned of irrational exuberance and the base year for the Bureau of Labor Statistics MFP series) as the beginning of the bubble. In the five subsequent years, MFP rose by 3.3 per cent, an annual rate of 0.65 per cent, which is very modest.
Seventh, we can reinforce the MFP data by observing that a technologically improvement in MFP in a given sector producing traded goods should be associated with an expansion of employment and capital in that sector, and an expansion of net exports. This hasn’t happened in US manufacturing, even in ICT-producing and ICT-using sectors. Admittedly the overvaluation of the $US has been relevant here until recently.
I’d summarise all of this evidence as follows. The boom of the 1990s was investment-driven and reflected a belief that US MFP would rise strongly as a result of developments in ICT and also because of superior economic flexibility. The growth in capital inputs produced strong demand for labour, reflected in rising participation rates and hours per worker. This produced strong growth in output per capita in advance of the anticipated rise in MFP.
Since MFP growth didn’t increase but slowed the investments haven’t paid off and are now being liquidated. This implies a sustained reduction in labour demand. The best way of meeting this reduction in demand is through a reduction in hours per worker with the US resuming the long-term downward trend and catching up with Europe. This will entail a lengthy period of slow growth in output per capita. However, given that hours per worker are declining, economic welfare will actually be improving.