Damned statistics!

I don’t know if the newly proposed English Baccalaureate exams will extend to statistics but if so plenty of us could do with sitting it. Yet again in recent days the Office for National Statistics (ONS) has been challenged over the reliability of a key economic indicator, though this time unemployment rather than GDP data, while so-called ‘government insiders’ are said to be fretting over which statistical index to use when uprating most welfare benefits. In both cases the level of public discourse leaves much to be desired.  

On benefits uprating, the government has already jettisoned the RPI as the reference index in favour of the CPI, which is typically lower. But ahead of this morning’s release of the latest monthly inflation figures (which show the CPI rate down to 2.5% and RPI down to 2.9%) the BBC reported that some government ministers think enabling benefit claimants to ‘enjoy’ CPI linked increases in welfare payments is still too generous at a time of fiscal austerity. According to the report the Treasury is looking at a two year benefit freeze followed by subsequent uprating in line with growth in average earnings rather than CPI. Had benefits tracked average pay growth rather than CPI inflation since recession first hit the jobs market in 2008/9, the Treasury is said to have reckoned, government coffers would be £14 billion richer.

This is all rather perplexing. If the government wants to cut the welfare bill it should simply do so rather than bother with the fiction of identifying some conveniently automatic, and thus supposedly more politically neutral, way of doing so. But in any case, average pay growth is not a good index to choose since one normally expects this to run comfortably ahead of price inflation. Assuming the economy at some point returns to the trend productivity performance enjoyed before the financial crisis, average pay should start to rise by around 4% to 4.5% each year, enabling the average worker to receive an annual real pay increase of around 2% above the Bank of England’s existing CPI inflation target. If this does not happen, either because trend productivity growth has truly slumped or target inflation proves very hard to achieve, the Treasury and the rest of us will have a lot more to worry about than how to uprate benefits.

Turning to unemployment, in an interesting article in yesterday’s Independent newspaper, former Monetary Policy Committee member Professor David (‘Danny’) Blanchflower contends that rather than puzzle over why UK unemployment is falling despite the double dip recession we should instead look more closely at ONS data, which on inspection actually show a rise in the number of jobless people actively seeking work.*

Professor Blanchflower is broadly right about this but perhaps protests a little too much. His argument is that the monthly estimate of unemployment provided by the Labour Force Survey shows a jump of 113,000 between June and July this year, which offers a very different picture of the current state of the jobs market than provided by the headline quarterly estimate which shows a fall of 7,000 between the three month average for May-July as compared with the previous three month average.

However, while the three month average comparisons can be confusing – and I agree with Professor Blanchflower that in due course the ONS should move to straightforward month on month comparison, say in a manner akin to what the US Bureau of Labor Statistics does – one simply can’t read too much into the current monthly estimates. The sampling variability for each month of data means that the ONS estimates are only fully statistically reliable when averaged over a quarter.

The most that one can say is that the labour market probably weakened in July and that this will show up in subsequent quarterly comparison. But there is as yet no way of telling how much the market has weakened. Moreover, however one reads the ONS data for this year it’s clear that there has been remarkably strong employment growth since the spring and nothing like the worsening in the unemployment situation that most economists, myself and I presume Professor Blanchflower included, expected at the start of the year. This remains a puzzle, albeit a welcome one.                 

* http://www.independent.co.uk/news/business/comment/david-blanchflower/david-blanchflower-unravelling-puzzle-of-jobless-fall-that-isnt-8143350.html?origin=internalSearch#