Labour market statistics — particularly employment, unemployment and inactivity — are often a topic of fraught and politicised debate.
This article considers the trends in UK employment, to test various claims made about labour statistics, over the past quarter of a century. Many of these claims are found to be false or likely to be misleading.
An International Definition
Like other statistics offices in the European Union and the OECD countries, the UK’s Office for National Statistics (ONS) follows the International Labour Organisation (ILO) definition of employment, for those aged 16 and over:
Employment: worked one hour or more in the week (either as an employee or self-employed), those with a job that they are temporarily away from (such as through illness and holidays), those on government-assisted training and work programmes, and those doing unpaid family work.
Unemployment: out of work, but actively seeking work in the past four weeks and available to start in the next two weeks, or have found a job and are waiting to start in the next two weeks;
Inactive: Without a job, and have not active sought work in the past four weeks, or are not not available to start work in the next two weeks.
‘Unpaid family workers’ means those who work for a family business, whilst not receiving a salary and profiting from the business. It does not mean unpaid family carers, baby-sitters or volunteers are classed as employed.
This definition is not set by the government, nor has there been any recent redefinition. Headline figures have been calculated in a consistent manner, and are comparable to when such records began in 1971. Additional information has been collected from the Labour Force Survey since 1992.
Some countries start their labour force population at a different age to the UK (e.g. France begins aged 15). Other statistics offices, such as the US Bureau of Labour Statistics, provide alternate measures of labour under-utilisation.
These definitions do not refer to receiving state unemployment payments or working credits. Whether you are unemployed in the Labour Force Survey does not depend on receiving unemployment benefits.
Some unemployed people do not collect such benefits — even when they are entitled — and it would be wrong to exclude them.
Like other surveys, the Labour Force Survey provides estimates — not counts. It is subject to numerous sources of potential error, including sampling variability.
The estimated employment rate reached 76.1% in November-January 2019, the highest since comparable records began in 1971. The unemployment rate estimate was 3.9%, which has not been lower since November-January 1975.
With its criterion that only one hour per week is required to be classed as employed, the ILO definition does have its detractors. This definition can lead to an overestimate of labour utilisation.
Note: due to an error in a BBC Reality Check article, people continue to falsely claim it is ‘one hour per fortnight’. That article has been corrected.
However, in the UK, few people work few hours. For November-January 2019, it was estimated that only 1.4% of those in employment usually work less than six hours a week — glacially subsiding from about 2.0% in 1992.
The estimated share of those usually working between 31 and 45 hours of all in employment was 54.1% in November-January 2019. That figure was under 50% in the late-1990s.
Some claims on social media relate to the minor categories of ‘unpaid family workers’ and those on ‘government training programmes’.
In November-January 2019, about 172,000 people were estimated to work in these two combined categories: less than 0.6% of the total workforce. That estimate is also dramatically lower than in 1992.
Self-employment has disproportionately risen in recent years: from an estimated 13.7% of the workforce in February-April 2010 to 14.8% in November-January 2019. In February 2019, the ONS published an analysis of this rising trend in self-employment.
Whether someone in the Labour Force Survey is working full-time or part-time is determined by self-classification.
The share of workers in part-time roles has marginally fallen since February-April 2010, from 26.9% to 26.3% in November-January 2019.
Temporary Work and Part-Time Underemployment
Another assertion is that employment growth is due to low-quality jobs, such as temporary work and part-timers who could not find full employment.
Temporary employment as a share of the total workforce has been fluctuating around 6%. Since February-April 2010, the estimated proportion of temporary workers who could not find permanent work peaked in 2013, and has now fallen below its 2010 rate. We see a similar trend for the proportion of part-time workers who say they could find full-time employment.
Additionally, the ONS has asked about time-based underemployment since January-March 2002. We see that the latest reading for the time-based underemployment rate has fallen below its rate in October-December 2008.
Zero-hours contracts have no minimum hours, and offer no guarantee of work.
Prior to 2013, the ONS estimate for this type of employment was vastly different to those produced by the Chartered Institute for Personnel and Development, estimated through a business survey. The CIPD estimated there were 1m employed on such contracts, about four times the level that the Labour Force Survey had suggested around that period.
In 2013, the ONS estimate rose of employment in these contracts sharply. As stated in their April 2018 report:
Comparisons with 2012 and earlier years are complicated by a large increase between 2012 and 2013 that appeared to be due mainly to increased recognition and awareness of “zero-hours contracts”.
As with other parts of the Labour Force Survey, the employment estimate of zero-hours contracts is concerned with the main employment of that person.
It is likely that pre-2013 levels were substantially under-estimating employment on zero-hours contracts, as people answering the Labour Force Survey were failing to recognise the term. (In survey statistics, this is one kind of measurement error.)
The LFS estimate peaked at 907,000 people in October-December 2016. The latest reading in October-December 2018 estimated 844,000 people were on such contracts in their main employment (2.6% of the total workforce).
By Age Group
As society changes, the labour market changes.
About half of the net employment growth between February-April 2010 and November-January 2019 was in the employment of those aged 50–64. The changes in employment rate have been particularly differentiated by age for women.
Between those two snapshots, the estimated employment rate for women aged 16–17 fell by 4.2 points. For women aged 50–64, the employment rate rose by 9.6 points.
This is likely to be partially driven by government policy: with longer education keeping young people from participating in the labour market, and an increased pension age meaning women work for longer. The employment rate among women aged 50–64 has been steadily rising since the early 1990s.
To the future
Analysing the labour market considers more than headline employment, unemployment and inactivity estimates. Pay and productivity are also key lenses to understand the labour market. The following are both important: the highest employment rate since comparable records began, and real average weekly earnings still sitting under its 2008 peak a decade later.
Since the labour statistics reports have broken into ‘modules’, a further report on under-employment, long-term unemployment and flexible working could be added. Additional data collection is needed to understand the pay of self-employed workers, estimated to constitute over 14% of the workforce. Whilst explicit references to uncertainty are welcome, graphical expressions of uncertainty in headline estimates are surely needed.
Separate suggestions that the record employment rate estimate is caused by the inclusion of family workers, or people working a few hours each week, or temporary workers, or some government redefinition, are all false. The ILO definition is used internationally, and has been for over three decades.
Spuriously undermining trust in official statistics for partisan reasons should be excoriated.
Data used in this article was drawn from the ONS labour market statistics reports and related data-sets. I have published my collation in a Google Sheet. The R Code used to produce the graphs is also available online. I added the time-based underemployment graph to the article after initial publication.