One of the biggest puzzles economists are dealing with in relation to the Great Recession is why it is that the United States had approximately twice the increase in unemployment or decrease in employment compared to the UK for approximately half the fall in output. In the recovery phase, US output has surpassed pre-recession levels while less than half of UK GDP has been restored. In contrast, employment in the US is below what it was at the start of the recession in January 2008, while in the UK employment is up slightly, from 29,482,000 in January 2008 to 29,560,000 in August 2012.

The two measures used to calculate employment in the US, based on a household survey and a firm survey, show slightly different amounts still to be restored. In the case of the former, civilian employment is down 2.3 per cent, while in the latter, non-farm employment is down 3.3 per cent.

Commuter traffic

The most unexpected figures recently were the big drops in the UK’s headline unemployment number and rate. However, the way the Office for National Statistics (ONS) reports the numbers – as three-month rolling averages – doesn’t really help us understand what is happening.
It turns out that this occurs because of concerns about the variability of sample estimates that arise because the survey from which the numbers are extracted – the labour force survey (LFS) – is underfunded and sample sizes are so small that there is a lot of month-to-month variation.

If, for example, you go to the latest data release on unemployment rates from Eurostat, the statistical agency for the EU, you find that there are unemployment rates for August for 22 EU countries, including even tiny Cyprus and Malta; Hungary has data up to July, while Estonia, Greece, Latvia and the UK have data through to June only. It’s time to fix this, Mr Cameron; it’s embarrassing.

Monthly data is available, however, on the ONS website, even though it is not designated as what we call “national statistics”, which are used to calculate the rolling averages. The table illustrates the problem. The headline numbers were that the unemployment rate was 7.9 per cent, down from 8.1 per cent, and unemployment was 2,528,000, down from 2,577,000, marked in red in the table. These are derived as rolling averages, so the 7.9 per cent is the average of June to August, while the 8.1 per cent is the average of March through May, and similarly for the unemployment numbers as well as the employment and inactivity rates.

What does stand out is that the drop in unemployment in August to 2,421,000 (from 2,633,000 in July) looks like an outlier, not
least because of the huge increase in inactivity of 280,000 and a fall in employment of 32,000. Something strange appears to be going on. Note that employment fell in the past two months, from a June peak that is probably distorted by the Olympics, as well as by young people who may not be entering education, in part because of the increase in tuition fees. So the unemployment rate is likely to rise; the Bank of England agents in their October report suggested that there would be “little job creation” in the private sector over the next six months and their past predictions have been reliable.

Why hasn’t employment been as weak in the UK as in the US? I have two plausible explanations. The first relates to eastern Europeans and the other to interest rates. In contrast to other EU countries, in 2004 the UK, along with Ireland and Sweden, opened its borders to workers from the “A8” accession countries (the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia and Slovenia). But the UK
restricted access to state benefits. On 1 January 2007, Bulgaria and Romania joined the EU; workers from these two countries were given much less open access to the UK labour market than those from the A8. (I refer to the group of ten eastern European countries as the “A10”.) Approximately a million workers from these countries registered to work in the UK and signed up for National Insurance numbers. The vast proportion of these workers, when interviewed at the borders, said they had no intention of staying permanently.

This represents a problem because they were not picked up in the LFS or in the population weights used to generate national estimates,
although their consumption would have been picked up. These folks from the A10 were, in essence, commuters – once the economy collapsed, they returned home and either did not work at all in the UK or severely restricted the numbers of days they worked. This would be picked up by a drop in GDP.

Mea culpa

By way of illustration, suppose there were a million A10 workers in the UK in 2008 who worked ten three-week spells. Assuming 50 weeks a year for simplicity, that is the equivalent of 600,000 workers who weren’t counted in 2008. In 2012, there are only 800,000 workers who show up; on average they work three spells of two weeks. That is equivalent to a drop in employment of about half a million. So the drop in employment since then has been understated, although by how much is unclear because we don’t have precise estimates.

Another contributing explanation is that wages have become more flexible in the UK, because of the high proportion of homeowners who have variable-rate mortgages in this country compared to those in the US. In the US, those with fixed-rate mortgages could refinance as interest rates fell but, as credit conditions tightened for most people, that option was off the table. In the UK, workers were less resistant to reductions in hours and pay freezes than they might have been if they hadn’t had big reductions in monthly payments on their mortgages. As a result, any increase in interest rates before growth is fully restored would have a major downward impact on consumption and GDP.

I was wrong on the latest figures, to the delight of some of my Twitter followers, but I suspect unemployment will rise sharply again, and soon.

David Blanchflower is economics editor of the New Statesman and professor of economics at Dartmouth College, New Hampshire