Skip to main content

Manufacturing Industries: Statistics

Volume 475: debated on Friday 9 May 2008

To ask the Chancellor of the Duchy of Lancaster what assessment he has made of the effect on the quality of labour market statistical information from (a) the manufacturing industry and (b) the furniture industry of the reduction in the sample size for the Monthly Production Inquiry. (201982)

The information requested falls within the responsibility of the UK Statistics Authority. I have asked the authority to reply.

Letter from Karen Dunnell, dated 9 May 2009:

As National Statistician, I have been asked to reply to your Parliamentary Question on the assessment of the effect on the quality of labour market statistical information from (a) the manufacturing industry and (b) the furniture industry of the reduction in sample size of the Monthly Production Inquiry (MPI). (201982)

The MPI collects turnover and employment information from businesses in Great Britain, which is used mainly for the compilation of National Accounts and Labour Market Statistics.

Monthly series for employee jobs in the manufacturing sector are published on the Office for National Statistics (ONS) website. In March 2008, when ONS published estimates for January 2008, the MPI sample of businesses was reduced by around 10 per cent, as part of a wider ONS efficiency programme and to meet ONS’ targets to reduce the burden it places on business. The sample was reallocated across the sample cells (defined by business size-bands and industries) to make the best use of the survey information. This offset the quality impact of reducing the sample size and enabled us to meet our aim of maintaining the quality of estimates at the aggregate level. At the more detailed level, some industries gained in quality at the expense of others.

Sampling errors are a key measure of survey quality, and were estimated to inform the sample changes and assess the impact on quality. The coefficient of variation is simply the sampling error given as a percentage of the actual estimate, which allows for a relative measure of error for comparative purposes. Please note that estimates of sampling error are themselves subject to a margin of uncertainty.

(a) The estimate of GB employee jobs in manufacturing in January 2008 was 2,806 thousand. The previous coefficient of variation of 1.2 per cent gave us 95 per cent confidence that the estimate was between 2773 and 2839 thousand. The changes resulted in a small increase in precision, with a coefficient of variation of 1.1 per cent, giving us 95 per cent confidence that the estimate was between 2776 and 2836 thousand.

(b) The estimate of GB employee jobs series in furniture manufacturing in January 2008 was 102 thousand. The previous coefficient of variation of 5.9 per cent gave us 95 per cent confidence that the estimate was between 96 and 108 thousand. The changes resulted in a small decrease in precision, with a coefficient of variation of 6.9 per cent, giving us 95 per cent confidence that the estimate was between 95 and 109 thousand.