As someone who has spent close to a decade hunched over a PC, hurriedly writing briefings on labour market statistics in dimly lit offices, I couldn’t help but find entertainment in the tangle some respected news organisations got themselves recently into over the latest UK unemployment data.
Reuters reported a “surprise increase” in unemployment, while the Guardian first noted an increase in their rolling business blog, then corrected themselves a few minutes later, citing a “statistical quirk”. This was not before several commentators were quoted explaining why unemployment may have unexpectedly jumped up.
All the while the Office for National Statistics (ONS), who collect and publish the data from the Labour Force Survey, maintained unemployment had clearly fallen.
This is a very easy mistake to make. As Nick Palmer, a senior statistician in the ONS’ Labour Force Survey department, helpfully explained to the Guardian, this confusion occurred due to the way the ONS aggregates and publishes the data.
Each monthly release of Labour Force Survey data is based on three months’ worth of observations. This three month period “rolls forward” one month at a time, and thus has two months overlapping with the previous release. Because of this overlap, the ONS never directly compares two consecutive monthly releases. They effectively scribble out the estimates released the previous month. Instead, they publishes estimates for a new previous non-overlapping three months alongside the latest data to enable a meaningful picture of change over time.
So, the latest data covers the period October to December 2013. It reports an estimated unemployment rate of 7.2% of the economically active population, equivalent to 2.34m people. This is a fall from the previous non-overlapping quarter, July to September 2013, of 0.4 percentage points and 125,000 individuals.
But the Labour Market Statistics for January 2014, published on the 22nd of January, included an unemployment rate of 7.1% – tantalisingly close to the 7% threshold identified in the Forward Guidance provided by the Governor of the Bank of England Mark Carney, at which the Monetary Policy Committee may consider raising interest rates. Clearly this attracted a lot of attention, and contributed to the bank issuing revised forward guidance in this month’s inflation report.
As the January rate made front page headlines, it is not surprising that commentators looked at yesterday’s slightly higher rate and initially reported an increase. But the January estimate was based on the period September to November 2013 and, like any monthly labour stats release, should not be directly compared with this month’s data.
The ONS has suggested that data for December alone may indeed indicate a slight increase in unemployment within that month. But it also cautions against drawing too much from a single month’s data – the relatively small sample size increases volatility.
Looking at a non-overlapping time-series of quarterly data, which is revised each month and published in the tables at the end of the ONS’ Labour Market Statistics briefing, yesterday’s data shows a fall in the numbers unemployed in every quarter since January to March 2013.
Another area of potential confusion came from the Department of Work and Pension’s Quarterly Statistical Summary. The data appeared soon after Channel 4’s live televised debate on welfare that coincided with the final episode of the controversial Benefits Street, and in the midst of a row over cuts ignited by the Archbishop of Westminster that drew in the prime minister, and spokespeople from the church, charities and the Opposition.
During the televised debate, commentators on the left, such as author and columnist Owen Jones, highlighted the relatively large share of the welfare budget attributed to people in work but on low incomes – principally through housing benefit. Those on the right, including columnist Allison Pearson and DWP Minister Mike Penning, countered that work was always the best way out of poverty and that the benefit system encouraged too many claimants to remain out of work.
On the face of it, yesterday’s DWP data suggests a large majority of working-age benefit claimants are out of work. In August 2013, there were 5.4m working-age claimants accessing either out-of-work benefits or in-work income support. The largest proportion claimed Employment and Support Allowance (the out-of-work benefit which replaced Incapacity Benefits and the Severe Disablement Allowance in 2008), accounting for 45% of total working-age claimants (2.4m). Job Seekers’ Allowance accounted for a further 23% (1.3m). The remainder, 1.7m, received benefits that could be claimed by those in work, such as lone parent or carers’ support, other income support, and disability living allowance.
But these totals do not include housing benefits, which, according to the DWP’s own annual report, account for the second largest share of the UK’s total benefit bill after pensions. The DWP statistical release includes analysis of Housing Benefits in a separate section: they are not included in either the up-front breakdown or the total for the “working-age claimant” analysis. This is important, as out of the 5m housing benefit claimants, 3.7m were working age. Of these, more than a million housing benefit claimants were in employment and not on other benefits (such as income support). When looking specifically at new housing benefit claims registered since 2010 – the vast majority of new claimants (more than 90%) were in work.
This suggests there is merit in the arguments made by Owen Jones et al, but – much like the monthly unemployment statistics – it unfortunately requires either some digging or significant familiarity with the data. Neither of which is ideal if we are to have a well informed debate.