On payroll day, the markets usually focus on the the nitty-gritty of the monthly data, searching for lessons on the near term movement of the US economy. This is frequently a forlorn task, since the initial estimates of US employment (covering more than 130 million workers, some of whom are just assumptions in the models of the official statisticians) are so uncertain. For example, in today’s better-than-expected numbers, the revisions to the last two months’ estimates were bigger than the estimated change in employment in August. Too much focus on the short term can obscure the bigger picture, which in the end is more important for the economy and markets. One crucial aspect of this bigger picture is the relationship between unemployment, long-term unemployment and structural unemployment.
An obvious feature of the recent recession in the US is not just the extent of the rise in unemployment, but the degree to which this is being translated into long-term unemployment. As the first graph shows, the number of Americans who have been out of work for more than 6 months now stands at 6.5 million, compared to peaks of only about 2 million in the previous two recessions. This is a much bigger increase in long-term unemployment than the previous behaviour of the US economy would have implied. In an examination of long-term unemployment published earlier this year by the OECD, the authors estimate a relationship between total unemployment and the long-term component. This enables them to estimate the expected change in long-term unemployment as a result of the recession. To their great surprise, they found that the rise in long-term unemployment, given the behaviour of total unemployment, was two and a half times higher than previous experience implied. They do not offer an explanation for this, but they ask whether it may have been due to the sectoral composition of the recession (hitting construction, cars, and real estate the most), and the role played by the housing market in limiting labour force mobility.
A separate question is whether this rise in long term unemployment has been translated into a rise in structural unemployment. These concepts are usually related in most countries, but that rule may not always apply, and the rule has not applied as much in the US as elsewhere. Structural unemployment is defined as the unemployment rate required to hold inflation constant in the medium-term. Long-term unemployment usually exerts less downward pressure on inflation than short-term unemployment, because the former group has less effect on wage bargaining than the latter group. Therefore, structural unemployment rises with the level of long term unemployment.
The OECD authors reckon that, on past experience, the recent rise in long term unemployment in the US “should” have raised structural unemployment by around 0.7 per cent. This is actually a rather small amount, compared to the rise of around 5 percentage points in total unemployment. This would leave the present level of structural unemployment at about 5.7 per cent, which is far below the actual unemployment rate of 9.6 per cent.
Other economists have taken a different view on the rise in structural unemployment. For example, in this recent blog, I quoted Edmond Phelps as saying that structural unemployment has risen to as much as 7.5 per cent of the labour force. Given this uncertainty, I have shown a possible range for structural unemployment in the second graph from 5.7 per cent to 7.5 per cent, which I admit is unhelpfully wide. Nevertheless, the key point is that this entire range is considerably lower than the actual level of unemployment, which means that on most measures there is a large amount of cyclical unemployment left in the economy today. This is the portion of unemployment which should be susceptible to government or central bank policy easing, since it can be reduced without raising the inflation rate. After that, supply-side action would be needed to get the rate down further.
However, there is an important caveat to this suggestion. As Martin Wolf points out in his FT column this week, the clock is ticking. European experience in the 1980s and 1990s clearly suggests that long-term unemployment tends to leak more and more into structural unemployment as time passes, because these workers effectively become irrelevant to the active labour market. Therefore the problem of long-term unemployment may become increasingly intractable as time passes.
Today’s employment data have brought with them a respite for the Fed, which is not likely to take emergency action while the economy looks fairly stable. But the underlying state of the labour market remains worrying, and will do so until long-term unemployment starts to fall.