Employment data: when there’s no story – try not to tell one
24 June 2019
SUGGESTED
Government and Institutions
21 June 2019
Society and Culture
25 June 2019
Trade, Development, and Immigration
4 November 2024
People can get a very misleading picture from statistics which are not properly understood. Take this, from The Telegraph:
“Around 357,000 jobs were added to the economy in the 12 months to April but the jobs boom has largely been driven by the 50-64 age group and even the over-65s. Of those jobs, 304,000 were given to workers above the age of 50.”
Possibly the author really knows better, but the impression given is that there was a stock of jobs in April 2018 which was augmented by the creation of 357,000 jobs over the next twelve months. For some reason, the vast majority of these jobs went to older workers. The author of the piece (illustrated by a pair of gnarled hands grasping a walking-stick) then goes on to speculate about changes in the nature of jobs suiting older workers, who are fitter than ever and staying on at work longer to augment their depleted pensions.
But this is not how it works. Over a year there is huge churn in the labour market. A BEIS study a few years back showed that over the decade to 2008, 2.5 million private sector jobs were lost on average each year as companies reduced their headcount or closed, while at the same time 2.67 million jobs were created by existing businesses and new entrants.
Overlaid on top of this, hundreds of thousands of individuals entered the labour market and took jobs (school and university leavers, immigrants, women returners etc), while many others left work as a result of retirement, going to university, having babies, illness and death, emigration and so on. In addition, hundreds of thousands more changed jobs, moving to different employers or obtaining promotion to different jobs within the same organisation. So even where ‘jobs’ remained unchanged, their occupants may well have changed – possibly more than once – over a twelve-month period.
All this means that the jobs and job-holders of April 2019 differ in many ways from those of April 2018. Those taking new jobs during this 12-month period would likely be a random draw from those in the workforce during that period. A slightly higher proportion in work at the end of the period would be over 50 than was the case at the beginning of the period. This is probably almost entirely the effect of demography, with the bulge of older workers (with a slightly higher proportion of females than previously), plus some delay in taking retirement associated with actual and prospective changes in state pension age.
Employers taking on a new employee would be very slightly more likely to take on an over-50 for the job – just as they would be very slightly more likely to be taking on a woman and very slightly more likely to take on an applicant from an ethnic minority.
You can infer little from this about the types of jobs done, the suitability of these jobs for older workers (speculation on which is riddled with ageist assumptions, incidentally) or the health and attitudes to work of older people.
In this case, speculation is probably harmless. But too often misleading inferences from employment data lead on to poor policymaking. And this should concern us.
“Around 357,000 jobs were added to the economy in the 12 months to April but the jobs boom has largely been driven by the 50-64 age group and even the over-65s. Of those jobs, 304,000 were given to workers above the age of 50.”
Possibly the author really knows better, but the impression given is that there was a stock of jobs in April 2018 which was augmented by the creation of 357,000 jobs over the next twelve months. For some reason, the vast majority of these jobs went to older workers. The author of the piece (illustrated by a pair of gnarled hands grasping a walking-stick) then goes on to speculate about changes in the nature of jobs suiting older workers, who are fitter than ever and staying on at work longer to augment their depleted pensions.
But this is not how it works. Over a year there is huge churn in the labour market. A BEIS study a few years back showed that over the decade to 2008, 2.5 million private sector jobs were lost on average each year as companies reduced their headcount or closed, while at the same time 2.67 million jobs were created by existing businesses and new entrants.
Overlaid on top of this, hundreds of thousands of individuals entered the labour market and took jobs (school and university leavers, immigrants, women returners etc), while many others left work as a result of retirement, going to university, having babies, illness and death, emigration and so on. In addition, hundreds of thousands more changed jobs, moving to different employers or obtaining promotion to different jobs within the same organisation. So even where ‘jobs’ remained unchanged, their occupants may well have changed – possibly more than once – over a twelve-month period.
All this means that the jobs and job-holders of April 2019 differ in many ways from those of April 2018. Those taking new jobs during this 12-month period would likely be a random draw from those in the workforce during that period. A slightly higher proportion in work at the end of the period would be over 50 than was the case at the beginning of the period. This is probably almost entirely the effect of demography, with the bulge of older workers (with a slightly higher proportion of females than previously), plus some delay in taking retirement associated with actual and prospective changes in state pension age.
Employers taking on a new employee would be very slightly more likely to take on an over-50 for the job – just as they would be very slightly more likely to be taking on a woman and very slightly more likely to take on an applicant from an ethnic minority.
You can infer little from this about the types of jobs done, the suitability of these jobs for older workers (speculation on which is riddled with ageist assumptions, incidentally) or the health and attitudes to work of older people.
In this case, speculation is probably harmless. But too often misleading inferences from employment data lead on to poor policymaking. And this should concern us.
SHARE
