An argument for Big Data to provide meaningful employment figures
Wed 16 Dec 2015
Today Britain’s largest trade union warned Prime Minister David Cameron’s government against ‘popping out the champagne’ at the news that UK’s unemployment, at 5.2%, has fallen to its lowest level in nearly a decade. Speaking via a release on the Unite union’s web site, general secretary Len McCluskey commented: ‘The continued rise of insecure self-employment and part-time work, coupled with the slowing down in an already painful and slow recovery in wages, means that too many in the workforce will be struggling to get by this Christmas.’
Unemployment statistics, certainly in the UK, use two methods which are astonishingly unreliable, and in an age where Uber-style ‘contracting’ and zero hours contracts are reducing millions back to the pre-Beveridge days of ‘casual’ or unreliable labour, both methods would be considered poor source data in any analytical experiment.
The ‘unpeople’ who cannot claim or be counted
On the one hand the Claimant Count Method calculates the unemployment rate by reporting the number of people in receipt of two types of benefit – Contribution Based Job Seeker’s Allowance (equivalent to ‘unemployment’ payments in the U.S.) and Income Based JSA (equivalent to ‘Welfare’ Stateside). The claimant count criteria has been changed by successive governments 30 times since 1979, making historical like-for-like comparisons utterly meaningless on a per-decade basis, since no-one has taken the trouble to create conversion indexes or other tools which might allow statisticians to approximate long-term comparisons.
Even trying to compare unemployment figures over the last ten years is a deceptive pursuit, since the last five years have been hallmarked by an unprecedented and frenzied sequence of controversial government initiatives to lower the claimant count, still seen as the ultimate index of UK unemployment.
As a measure of the increased fervour the Tory-led coalition/government has shown in this regard, consider that John Major’s government was criticised in the 1990s (even the Daily Mail admits this) for allowing too many unemployed claimants to transit to sickness benefit, since being on such benefits removes UK citizens entirely from the jobless figures, and thereby massages the figures in the government’s favour; and consider then that the current government has instead been savagely criticised for seeking to remove those too ill for work not only from sickness benefits, but from that embarrassing jobless tally – and from all economic refuge.
Additionally consider the highly publicised government initiatives to ‘sanction’ as many claimants as possible, no matter how spurious the reason, thereby removing the ‘offenders’ both from the claimant count and from all economic assistance (beyond foodbanks).
The low-budget straw poll
The second resource behind UK employment statistics is the Labour Force Survey, a representational poll of 60,000 individuals which is likely to reveal more about the true state of UK employment, since it takes circumstances into account rather than using successful benefit claims as units. On the plus side, the LFS sticks to international standards and therefore at least nominally provides figures which can be compared to other countries that apply the same criteria. Negatively the numbers involved in the survey are absurdly small compared to the 64 million people they are drawn from, and thus the LFS is perceived, even by those who are in support of current government policy, as a crude and unreliable yardstick of UK employment.
Six of one, half a dozen of the Uber
UK unemployment figures refer to a lost world of husbands trying to get another hour in the pub before going home to the wife and kids, in a three-bedroom house they were able to get a mortgage for on a tradesman’s sole wages; to a world which peaked in the late 1950s and was getting pretty fractured even by the early 1980s.
The paranoid presumption is that governments are afraid to leave the atavistic umbrella of outdated methodologies for arriving at these figures – that a terrible truth lies under that data iceberg
This method of reckoning is pure Sith-binary – you either have a job or you don’t. You’re either doing okay (job, car, family, future) or you’re not (‘scrounging’, ‘skiving’, foodbanks, desperation). It doesn’t take into account whether or not you receive tax credits because of low income which is recognised by the government as not being adequate to live on (even though that particular safety net is getting increasingly threadbare); or if you’re a zero-hours Uber driver in your ‘spare time’ not because you’re saving up for life’s little treats, but because you can’t make enough out of that ‘job’ to live; or you’ve been sanctioned off from claimant figures and are now a statistical ghost haunting foodbanks; or on a zero hours contract where you don’t work for five or six days at a time, none of which ‘downtime’ periods ever enter into the calculations for unemployment figures. You could be unemployed from the knees down, the neck up, the elbow outwards, or any combination of these over any given period of time.
A mathematical proposition
Britain is gleefully envisaging a data-driven future where bridges and supporting structures report their condition every half hour; where wearable technology promises to balloon our medical records into terabytes of crunchable long-term data; where our biometric characteristics will be analysed and acted upon to within a fraction of a data-burst.
So it is quite amazing that we are sticking our knobbly elbows into the bathwater to determine the state of UK employment in this way. It would only take a NoSql blitz on two or three tax office data streams to get better source data than we are currently using, to establish genuinely meaningful and informative employment figures.
The paranoid presumption is that governments are afraid to leave the atavistic umbrella of outdated methodologies for arriving at these figures – that a terrible truth lies under that data iceberg. But, as with all statistics, it would only be a matter for interpretation in the end; and there is even the possibility of retrieving so much data that Parliament could spin itself into the small hours all year arguing what it means. But at least we would have some ‘modern’ information on the issue.
Unless we call for more reliable data-driven numbers about what is happening in the social economy of our country, future economists and historians are going to have to work out what was really happening in this period on the basis of oral histories, hearsay, approximations and guesses – despite Britain’s supposed ‘vanguard’ place in an era of governmental cloud computing and ambitious digital initiatives.
Whether you think Ian Duncan Smith and David Cameron are the saviours of the British economy or the very ushers of the apocalypse, do you not want better source information than this?