Garbage in, garbage out: our jobs figures are not working any more

It is very hard to work out what is going on in the UK labour market because the quality of the statistics is basically junk – garbage in, garbage out describes the lack of quality of the data well. I really am not exaggerating.

Bad Labour Market Data Part 1 is that every other major country, including the euro area as a whole, is able to produce timely estimates, but not the UK. Currently unemployment rates for February 2014 are available for Australia, Austria, Belgium, Bulgaria, Canada, Croatia, Cyprus, the Czech Republic, Denmark, Finland, France, Germany, Hungary, Iceland, Ireland, Israel, Italy, Japan, Lithuania, Luxembourg, Malta, Netherlands, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, and the United States. Data for April 2014 were released by the United States on Friday. The UK stands out as the only country out of 31 that has no data available for February, March or April 2014.

Pathetic. The national statistic that pretends to be for January is actually an average of December of 2013 and January and February of 2014. The reason for this is simply because the sample sizes are too small to generate accurate monthly estimates. The Office for National Statistics does in fact publish a single-month estimate of the unemployment rate but that jumps around all over the place. Let me illustrate the problem. The ONS makes the supporting micro data on individuals available for researchers like me to examine. They take out identifiers so we can’t work out who anyone is. The latest micro data we have is for the three-month period October to December 2013.

In total over these three months 77,657 people between ages 16-98 were interviewed. Of these, 39,761 were employed 6,995 were self-employed and 3,347 were unemployed. The overall unemployment rate, once the data have been weighted and seasonally adjusted is 7.2 per cent, but the relatively small sample size means this estimate is measured with lots of error.

For the technically minded, the 95 per cent confidence interval for the monthly national change is ± 0.3 per cent, which means that any monthly difference smaller than that is not statistically significantly different from zero. The unemployment rates that were calculated, for example, for East Anglia (5.7 per cent), East Midlands (6.4 per cent), Scotland (7.1 per cent), Wales (7.1 per cent), Northern Ireland (7.4 per cent) as reported by the ONS for October-December were based on ridiculously small samples of 114, 246, 281, 153 and 142 unemployed people respectively. Given the very small sizes the result is that the regional unemployment rates are measured with even more error than the national rate and bounce around like a rubber ball from month to month.

The reason why the ONS struggles to report unemployment rates by month becomes obvious rather quickly.

So the single-month estimate for December of 7.2 per cent that it reports is only based on a sample of 1,198 unemployed people, of whom 632 were male and 452 were under the age of 25.

The number of unemployed people in each of the five regions identified above in December is East Anglia (34), East Midlands (91), Scotland (105), Wales (51), Northern Ireland (55), hence why no single-month disaggregated estimates can be produced.

Bad Labour Market Data Part 2. The government has claimed recently that based on earnings growth of the national statistic called Average Weekly Earnings (AWE) for the whole economy of 1.9 per cent in February 2014 and the fact that the Consumer Price Index has been steadily falling, this means that real wages are set to rise.

If only that was true. But sadly it seems most unlikely given the fact that the Monthly Wages and Salaries Survey (MWSS) on which the estimate is derived has two major sample exclusions whose wages are likely to be growing much more slowly than that, if at all.

First, the ONS has no earnings data, as in none, on the 4.5 million self-employed workers, including large numbers who have set up in business recently. The only earnings data we have available from HMRC are over two years old. What we do know is that the typical self-employed person earns less than the typical employee and some have zero earnings or even losses; there is every prospect earnings growth of the self-employed will be low.

Second, it also turns out that the MWSS doesn’t sample workers employed in firms with fewer than 20 employees that are the least likely to have strong earnings growth given the difficulty small firms have had in raising capital. The ONS simply makes an adjustment based on the Annual Survey of Hours and Earnings (ASHE), which was last available in April 2013 and which itself excludes the lowest earners below the National Insurance threshold. The ONS computes an average over the previous three years that it imposes on the AWE monthly data. So the ONS just guesses that what happened in the past applies now. But maybe it doesn’t.

The ONS admitted to me that “ideally, we would sample businesses with fewer than 20 employees in the MWSS. However, we do have to pay close attention to minimising the burden on respondents, and we believe that using the adjustment factor from the ASHE strikes an appropriate balance between this and accuracy of the estimates.”

Really? So making it up as you go along is OK? It turns out that this amounts to approximately 20 per cent of all employees, or another 5.2 million workers whose wages we know zippo about. So the national wage measure excludes 10 million out of the UK’s 30 million workers and my working assumption, for the sake of argument, is that their average pay rise over the past year is zero (it’s a maybe not-so-wild guess that the ONS can’t disprove)!

There is supporting contradictory evidence of strong earnings growth from the latest UK Job Market Report from Adzuna.co.uk, showing that average advertised salaries have slipped £1,800 in the past year down to £31,818 in March 2014, 0.6 per cent lower than in February, and 5.3 per cent lower than in March 2013.

A survey carried out by the Federation of Small Businesses at the end of 2013 reported that “after several years of wage restraint, it is encouraging that the vast majority of small firms are beginning to raise wages again”. They found that 29 per cent of firm owners said that over the next year they would raise wages for all staff, 35 per cent for some staff, 8 per cent for those on the minimum wage. 22 per cent said they would freeze wages, 2 per cent said they would lower them and the rest didn’t answer.

So the AWE is an upward-biased estimate of wage growth. Garbage in, garbage out. The UK’s labour market data are not fit for purpose.