Employment sizeband estimate update

Investigations have revealed an issue with the estimation methodology adopted when BRES replaced ABI/1 as the source for detailed employment estimates by industry in the UK. The core of the issue is that employment sizeband by business was not accounted for in the calibration weights for BRES (affects all years from 2009 onwards) but had been in the ABI/1 methodology (used for years prior to 2009).

Incorporation of employment sizeband into the calibration weighting methodology for BRES

As users will be aware, The Office for National Statistics (ONS) removed the BRES business size band datasets from Nomis on 15 August 2014 due to quality concerns about this information.

Analysis by the Department for Business Energy and Industrial Strategy (BEIS) of productivity estimates by sizeband showed unusual differences between those obtained when employment was sourced from the Annual Business Inquiry (part 1) (ABI/1) and those produced when employment came from the Business Register and Employment Survey (BRES). This analysis revealed a large increase in productivity for businesses with 0-9 employees, with a noticeable drop for businesses in the 10-49 and 50-249 size bands. The 250+ size band was much less affected.

Investigations have revealed an issue with the estimation methodology adopted when BRES replaced ABI/1 as the source for detailed employment estimates by industry in the UK. The core of the issue is that employment sizeband by business was not accounted for in the calibration weights for BRES (affects all years from 2009 onwards) but had been in the ABI/1 methodology (used for years prior to 2009).

In the meantime users are asked to consider the revised 2015 provisional employment estimates and provide comments if they wish by contacting ONS at bres@ons.gov.uk.