Has been some twitter discussion as well.
Most of the people over here are used to administrative data, which measures 100% of whatever it measures (a moveable feast) and household surveys.
Business surveys are a bit of a dark art which aren't as well documented as the Labour Force Survey.
Are there reasons not to trust the LFS? @D_Blanchflower points out that most countries have a sample large enough to deliver single month estimates, whereas ours are QA to 3-month moving average. Sample response has been decreasing, but the sample frame is well-understood and continually updated. Weighting is to Census-based population projections. Datasets available from Data Archive if you want to play with the data (I do).
Generally, highly trustworthy. Some dodgy bits. Counting 165,000 Government programme participants as in employment is particular bugbear (as the nature of the programmes means jobsearch and availability are redoubled above the ILO unemployment criteria). 10 volumes of manual and quarterly performance reports on the ONS website.
Business surveys from ONS, sample frame updated annually in arrears. Weighting? We know stratified random sample. Manuals? Data available from data archive? Not usually, because of collection under mandatory powers.
My worry is that recessions (and this one, with zombie firms, may be less affected) with high rates of business deaths, births and mergers, will cause the business sample frame to drift much more than a household sample frame.
So why should we believe the business survey based estimates?
Associate Director of Analysis and Statistics
3rd floor, 89 Albert Embankment, London SE1 7TP
Inclusion website www.cesi.org.uk
See Inclusion's www.indusdelta.org.uk for the latest news and opinions in welfare to work
Show Original Message