The latest Scottish labour market data published today for the quarter January to March should be welcomed. Unemployment fell by 10,000 over the quarter as employment rose by 24,000. The Scottish unemployment rate moved into line with the UK rate of 8.2% and the employment rate, at 71.2%, stayed above the UK rate of 70.5% for the 16-64 age group. For all aged 16 and over, the employment rate in Scotland moved from slightly below the UK rate to parity at 58%. But within these numbers male employment is rising (+29k) while female employment is falling (-5k).
At first sight these data from the Labour Force Survey (LFS) appear to conflict with what we know about the real economy. GDP has fallen in the UK in successive quarters to the first quarter and fallen in Scotland in the final quarter of last year. We will not know what has happened to Scottish GDP in the first quarter until mid July. However, as I noted in this recent post many analysts believe that the ONS is underestimating recent GDP growth. Business surveys such as the PMI suggest a stronger outturn performance. Against that must be set the continuing impact of fiscal consolidation, the inflation rate is falling back more slowly than the Bank of England expected due to the effects of earlier rises in energy prices, and the Eurozone crisis is worsening again.
So, there is no guarantee that unemployment will continue to fall. Indeed, it could worsen again.
Our estimates at the Fraser of Allander Institute suggest that there needs to be Scottish GDP growth of 2% per year - approximately 0.5% per quarter - for unemployment to stabilise. Faster growth for it to fall. Slower GDP growth and the unemployment rate will rise. We do seem to be quite a bit below the 2% per annum threshold at the moment.
Another reason for caution about these LFS data is the high sampling variability - for some of the issues see this Dutch research paper on the Dutch LFS. There is a huge range within which one can have 95% confidence that the Scottish LFS estimates fall as the chart below shows:
So, while LFS estimates that jobs rose by 24,000 over the last quarter, the actual change could have - with 95% certainty - been between minus 34,000 and plus 83,000. For unemployment, the LFS estimate is a 10,000 fall but the actual change in unemployment might have been anywhere between a fall of 42,000 and a rise of 22,000. A similar range of variation applies to the change in the numbers economically active and inactive.
We should also be aware that the data are produced quarterly on a rolling monthly basis. So, today we were given the data for Jan-March. Last month the data showed the situation between December and February. Hence, there are two common months, January and February between the two releases. It is therefore incorrect to say as is implied here that unemployment has been falling in Scotland for two successive quarters. It hasn't. In the first quarter of this year, Jan - March it fell by 10,000. In the final quarter of last year, Oct- Dec, it rose by 16,000.
Finally, we should be concerned that the data as currently released do not allow us to drill down and ask how the 24k jobs change is broken down into part-time jobs and full-time jobs, or hours worked. Nor do we know age composition of the recent unemployment change. Yes, we are given data in the latest release on part-time and full-time jobs, hours worked and the age distribution of employment and unemployment. But these data are only provided up to the year Oct 2010 - September 2011. So, we just don't know what's happening recently.
It is clear from the data to September 2011 that full-time jobs are falling while part-time jobs are rising. This is significant, because it is not impossible that labour demand has fallen with GDP even though the number of jobs rose and unemployment fell. This would be the case if the loss of labour input through the fall in full-time employment was greater than the gain in labour services from the rise in part-time employment. It is a possibility. But we just don't know.
Surely, the ONS can do better than this?
Comments