Select Committee on Treasury Minutes of Evidence

Supplementary memorandum from the Office for National Statistics

Question 118:  In which Len Cook was asked to provide a note giving an approximate timetable for enhancing productivity data.


  Productivity measures the output produced per unit of input, primarily focusing on the two inputs labour and capital. The output measure used is real value added, which at the whole economy level is real GDP. Headline productivity figures usually refer to labour productivity, measuring output per unit of labour. There are a number of ways of measuring the labour input: numbers of jobs, numbers of workers (who may have more than one job) or hours worked.

  Another productivity measure is total factor productivity (TFP), which takes account of inputs in addition to labour. A source of output growth is enhancement of the capital stock through investment and TFP takes account of this by combining both capital and labour inputs when dividing output. An alternative view of TFP is the growth in output that cannot be explained either from changes in labour input or from the capital stock. To quantify the input of the capital stock into production, a measure of the flow of productive input of the capital stock—called capital services—is calculated.

  Productivity can be measured for the whole economy, main subsectors such as manufacturing and a detailed industrial breakdown is also available. Industrial analysis of TFP is also possible. Such analysis can extend beyond capital and labour inputs, to take account of how one industry uses the products of another industry in its own production. Then productivity measures can take account of efficiency changes in the use of materials.

  There is considerable interest in comparing whole economy productivity measures across countries. To make international comparisons of productivity both outputs and inputs have to be computed in a comparable manner and adjusted for differences in price levels. Such statistics are taking advantage of the increased comparability of GDP and worker numbers across major economies.

  An expanding area of productivity research is the analysis of business microdata, analysing firms' survey returns. To allow this, secure systems for academic researchers to access confidential business data have been set up at ONS. This allows productivity to be computed for individual businesses (though analysis is published at highly aggregate level). A second statistical innovation is to link surveys together in an attempt to improve understanding of the determinants of productivity growth. For example, the ONS Annual Business Inquiry (ABI) covers total output and employment and, by linking this to the ONS e-Commerce survey, firm-level productivity measures can be related to that firms' e-commerce activities.


  Measurement issues are being addressed through improvements in methods and statistical sources in work across ONS. It is also being informed by debates outside ONS, through links with UK and international experts. These statistical issues in productivity can be differentiated from more complex conceptual issues:

    —  Measurement of output. Manufacturing and other production industries have traditionally had good output measures: the nominal value of output is well-defined and appropriate price indices are available to calculate real values. Regular—annual—updating of weights for each industry in aggregated output measures will improve these statistics. Service industries have more statistical issues remaining. Measures of price change are being developed. In some services, such as finance or some parts of government, output measurement is more complex, with conceptual issues about defining outputs.

    —  Measurement of inputs. Changes in the economy sometimes manifest themselves in the input measures. On labour input, changes in working patterns towards part-time work imply hourly productivity measures are necessary for labour input. When output in one industry increases more than another despite similar changes in hours worked it may be because more skilled labour is being used, suggesting further refinements. "Double-deflation", a process that measures the real changes in goods and services that input into production, can provide measures of the inputs other than labour and capital. This would help analyse how much of the growth in high technology output comes from a main input in production, the semiconductor. However, conceptual issues remain, such as the role of intellectual property, eg patents and brands.

    —  Consistency between the output and input measures. Productivity measures are very sensitive to the numerator and denominator referring as closely as possible to the same economic agents or sectors. Only then can output change confidently be related to the input change in the productivity measure. Such issues relate to statistical sources and development to sources is being informed by the needs of productivity measures. Some conceptual issues remain here, such as the effect of increased globalisation in production processes on relating domestic output to domestic inputs.


  By addressing the statistical measurement issues, the productivity measures available to analysts are improving. Currently, the ONS publishes each quarter (ref 1):

    —  labour productivity, as output per job, with detailed industrial breakdown for production industries;

    —  labour productivity as output per hour with industrial breakdown;

    —  regional productivity measuring nominal output per worker on a residence basis;

    —  unit wage costs, an index of wage costs in terms real output; and

    —  experimental series for output per job and hourly productivity for Agriculture, forestry and fisheries; Distribution (including wholesale and retail); and Total services.

  Half-yearly, ONS also publishes international comparisons of productivity: the output per worker series is a National Statistic, the hourly measure is experimental (1).


  The ONS expanded its productivity measures in 2001, largely due to the benefits of improvements in consistency between output and input measures:

    —  April 2001, productivity measures used the results of new survey instruments to improve the consistency between the output and labour input data (2).

    —  Increased consistency allowed new measures of output per hour to be published, also in April (2).

    —  ONS took over from the DTI the publication of international comparisons of productivity in October 2001 with improvements to the methodology (3).

    —  Improvements in the output measures for services—the Index of Services (4)—meant experimental quarterly productivity measures for the total services, agriculture and for wholesale and retail trades were published (5).


  In 2002, the developments have focused on improving output and input measures, taking on results of ONS development work in service sector measures and the results of work on capital input and skills. Also, increased use of business microdata by academic researchers seconded to ONS is providing first results.

    —  The ONS/DTI Productivity Workshop (November) saw results on total factor productivity published. The policy interest in TFP results from the accounting for growth in terms of the key drivers. The paper (6) indicates the contribution of skills to growth and the capital stock.

    —  Annually re-weighted output indices are to replace existing fixed based indices in the National Accounts in 2003. The structure of production changes over time and re-weighting regularly through the process of chain-linking can improve the precision of output growth estimates.

    —  At the ONS/DTI Workshop, labour productivity measures from the Annual Business Inquiry were also presented. The paper (12) discusses the potential of ABI in producing labour productivity measures at more disaggregate level including in some service sectors.

    —  Measurement of public sector productivity was discussed in a paper in May 2002 published in Economic Trends on improved output measures. The UK leads in the development of measures of non-market, output, using indicators that track output measured using counts of various government services (7).

    —  Productivity analysis by academic researchers based in ONS using plant-level microdata is continuing to extend our understanding of business level productivity change (8). The papers presented at the ONS/DTI Productivity Workshop explored the link between the innovations of a firm and its performance, as measured by productivity (10). Also, worker data was linked to business data (11).


  In April 2002, Productivity Measures: ONS strategy was published, which reflected the views gathered from a consultation with key users (9). The Strategy outlines works that ONS plans to take forward in the next three years. The table below draws on this and other work areas in ONS.
Input measuresDescription Timing
Labour hour measures, adjusted for skills A methodology so hours worked takes account of changes in the composition of labour skills over time has been developed. As experimental series is planned for end of 2003.
Constant price input-output tablesDouble deflation of output allows the real changes in goods and services used in production to be measured. Results in 2003.
Capital servicesThe methodology of a volume index has been developed. First set of estimates to be published in mid-2003.
Chain-linked output measuresAnnual re-weighting of industry output will mean that the changes in industrial structure will be up-to-date. Autumn 2003 in the Blue Book.

Services measures
Index of Services to be developed enlarging coverage of monthly indices. Work is ongoing (see ref 4).

General Government Output
Output measures have been developed for different public sectors such as education, police and road. Work is ongoing.
Data consistency
Reconciliation of output and input series Review the current methods and sources used in the National Accounts employment statistics studying other countries' experiences. A report by summer 2003.
Developing business microdata for productivity measures Micro analysis of business performance; some macro analysis by aggregating over businesses. The largest microdata project— Professor Jonathan Haskel's team funded jointly by DTI, HMT and ONS—is due to be completed by the end of 2003.

  (1)  Productivity First Release, Quarterly,

  (2)  Chris Daffin, 2001, "Introducing new and improved labour productivity data." Economic Trends no 570.

  (3)  Craig Richardson, 2002, "International comparisons of productivity." Economic Trends no 577.

  (4)  Steve Drew, Rob Pike, 2002, "Experimental monthly index of services." Economic Trends no 583.

  (5)  Chris Daffin, Geoff Reed, Prabhat Vaze, 2002, "Labour productivity measures for the non-production industries." Economic Trends no 579.

  (6)  ONS/DTI Productivity Workshop programme and abstracts events/future_events/downloads/productivity/_/workshop/_/programme/_/abstracts.pdf.

  (7)  Alwyn Pritchard, 2001, "Measuring productivity in the provision of public services." Economic Trends no 570. Alwyn Pritchard, 2002, "Measuring productivity change in the provision of public services." Economic Trends no 582. 1&Rank=112.

  (8)  Matthew Barnes, Ralf Martin, 2002, Business data linking: An introduction. Economic Trends no 581.

  (9)  Eunice Lau, 2002, Productivity measures: ONS strategy. Economic Trends no 581.

  (10)  Jonathan Haskel and Sonia Pereira, 2002, "Skills and productivity in the UK using matched firm and worker data".

  (11)  Chiara Criscuolo, Jonathan Haskel, 2002, "Innovations and Productivity Growth in the UK".

  (12)  Chris Daffin and Eunice Lau (2002) "Labour Productivity Measures from the Annual Business Inquiry".

November 2002

previous page contents

House of Commons home page Parliament home page House of Lords home page search page enquiries index

© Parliamentary copyright 2003
Prepared 6 January 2003