The US employment figures will, as usual, be the centre of financial market attention when they are published on Friday. But whatever this month’s figures show, they will not alter the fact that something has gone badly wrong in the American labour market in recent years. Once the envy of the developed world, US unemployment has risen much more sharply than in other economies since 2008, and this has come after a prolonged period in which the number of jobs in the economy has been considerably less responsive to GDP growth than used to be the case. So what has changed?
The most obvious possible explanation for the poor performance of the US labour market since 2008 would simply be that the economy has fallen into a very deep recession. But this is not confirmed by examining the link between the rise in the unemployment rate and the behaviour of real GDP. While unemployment in the US has risen by much more than it has in other countries, real GDP has fallen by much less. (See this piece on the subject by the St Louis Fed.)
The link between GDP growth and unemployment is the subject of Okun’s Law, first estimated by Arthur Okun in 1962 (and well explained by the IMF here.) The Law enables us to estimate what should have happened to US unemployment, given what has actually happened to real GDP. I have estimated the relationship in the two decades ending in 2007, and have then used the results to calculate what should have happened to unemployment since then, given the actual behaviour of GDP during the recession and the subsequent recovery. 
As the graph shows, unemployment rose much more sharply than it “should” have done in 2009. In fact, the GDP recession accounts for only around half of the rise in unemployment during the recession, with the rest needing to be explained by other factors. This is worrying, since around 2 percentage points of unemployment may not disappear, even if real GDP were to return to its previous trend (which in itself does not seem very likely in the near future).
I have seen four possible explanations for the breakdown of Okun’s Law since 2008.
1. The Impact of the Financial and Housing Crashes
The IMF (see above) has suggested that much of the unexplained rise in US unemployment has occurred because labour shake-outs tend to be larger in recessions which have been precipitated by a financial crisis, and which are accompanied by a housing crash. Firms which are deprived of access to credit become eager to improve their net liquidity position, and the quickest way to do this is to shed jobs. Furthermore, these firms seem to remain risk averse during the subsequent GDP recovery, so the pick-up in the number of jobs is slower than usual.
If the financial crash is accompanied by a housing crash, then the situation is worse, since the construction sector tends to fire more workers than other sectors when in trouble. This negative effect does not normally persist during a subsequent housing recovery, but at present that seems to be a very distant prospect indeed.
2. Increased labour market flexibility
In a flexible labour market, it is easier for firms to react to a recession by reducing their work force, so the rise in unemployment may be greater than in a highly regulated labour market. While this probably explains much of the difference between the performance of the US and European labour markets since 2008, it is hard to see how it can explain the change in the behaviour of the US market relative to its own history. There has been no increase in US labour market flexibility in recent times. Furthermore, this very interesting recent study by researchers at Amherst College shows that the sectoral pattern of job changes since 2008 does not fit the pattern which would be predicted by this hypothesis. (Industries with the highest numbers of temporary workers have not shed the most jobs.)
3. The measurement of output in financial services
This is also suggested by the Amherst researchers. Basically, they argue that a rise in the relative pay of workers in financial services is measured in the US GDP figures (probably misleadingly) as an increase in real GDP. When this happens, as it has done throughout the past decade or two, real GDP is therefore shown to have increased relative to employment, and Okun’s Law breaks down. Since the fall in output in job-creating industries was larger than that in financial services during the recent recession, it is possible that this has contributed to the overall rise in unemployment, according to this research. I am not sure that I find this argument convincing for the period since 2008, but it might explain some of the puzzle about US employment in the previous two economic cycles.
4. The effect of globalisation
There is evidence that the off-shoring of jobs to the emerging economies has substantially reduced employment in the US economy since about 2000. (See this recent article by David Wessel.) In 2005, the New York Fed estimated that 300,000-400,000 jobs were being lost to off-shoring each year, and more recent data show that 2.9m jobs were shifted out of the US by multinationals in the last decade. Since these jobs had probably been originally filled by relatively low productivity workers, this shift would increase the average level of productivity in the US economy, but could leave more workers unemployed.
It is not clear whether this effect has become more pronounced during and after the recent recession, but it is certainly conceivable that firms may have cut jobs where they will not need them in the future (i.e. the US), and subsequently hired new workers where they believe they will in fact need them (i.e. the emerging economies). If so, the recession could have increased the pace of off-shoring, though I am not aware of any firm evidence which confirms this.
So where does all this leave us? It is clear that there was an abnormally large shake-out of jobs in the US economy after 2008, and that the bounce-back in the past 18 months has been no faster than occurred in the two “jobless” recoveries in the 1990s and 2000s. The first explanation outlined above for this phenomenon is probably the most convincing in the context of the change since 2008, though the others may also play important roles over longer periods.
Although all this has had the effect of appearing to raise productivity in the part of the economy which has remained in work, it has also left a large residue of unemployment workers which may not disappear when GDP recovers. And that is a problem which will not be easy to solve.
