12 March 2006
Scottish Executive Economic Research
This document is available in PDF format. Further information on recent Scottish Executive Economic publications is available from the Economy Webpage.
RESEARCH ON OPTIONS FOR THE DEVELOPMENT OF BETTER MEASURES OF SCOTTISH PRODUCTIVITY
This study was undertaken by Professor Richard Harris of the Centre for Public Policy and Research (CPPR), based at the University of Glasgow, and Professor Mary O'Mahony and Dr Catherine Robinson of the National Institute of Social and Economic Research (NIESR).
There were two main aims of the research. Firstly, it was intended to establish the feasibility of developing an aggregate and sectoral time series on Scotland's capital productivity and total factor productivity. Secondly, the research was to consider the most appropriate methodology in which constructing these two series.
Executive Summary
The main focus of this study (for the Office of the Chief Economic Advisor and the Department of Enterprise, Transport & Lifelong Learning, at the Scottish Executive) on the productivity of the Scottish economy is on establishing the feasibility of further research in this area. The key area investigated relates to the possibility of developing series on Scotland's capital productivity and TFP, broken-down into industrial sectors, and the most appropriate methodologies to use. These series need to be of sufficient quality to allow for comparisons with the UK and other countries.
It was agreed at the inception report stage that the project needed primarily to consider the feasibility of producing comparable data for Scotland to that available at the NIESR which allows the estimation of total factor productivity (using comparable methodologies) Thus, it was agreed that in line with the data available in the NIESR growth accounting dataset, the core data needed to produce TFP at the sectoral level are:
• Estimates of real gross value added
• Persons engaged in the workforce
• Annual average hours worked
• Capital stocks (plant, machinery, buildings, vehicles and ICT)
• Skills data on the workforce (proxied by qualifications achieved)
• Contribution of labour and capital costs in total costs
We begin (in Section 2) with a discussion of the methodological issues that arise when measuring total factor productivity (TFP). Based on the data requirements of this project Section 3 provides an assessment of the GVA data sources available from the ONS; the data used by the NIESR; and lastly the data that is available for Scotland. Section 4 looks at sources of data on the factor input labour, while Section 5 considers the information that is available to measure gross domestic fixed capital formation, and thus the capitals stock. Finally, in Section 6 there is a summary and conclusion.
The most common method used to measure total factor productivity (i.e. output growth due to changes in technical efficiency and technical change) is the growth accounting approach. However, this assumes that there is constant returns-to-scale, and perfect competition in factor and output markets. If this is not the case (as is likely for the majority of markets), estimates of TFP will be biased. In addition, omitting known factors that determine TFP from the growth accounting identity results in further bias. Thus, Section 2 argues that a production function approach has much to recommend it. However, using the growth accounting at a broad industry-level can also be advocated, as it is the most common method used, as well as being transparent and easy to understand. In essence, it depends on the purpose of producing estimates of TFP as to which approach to adopt; if estimates are required for publication and to provide overall comparisons with other localities, then growth accounting may be preferable. If we wish to analyse the causes of TFP (and/or obtain more precise estimates perhaps at firm or plant level), a production function approach needs to be used.
A number of other issues were raised with respect to the measurement of TFP. It was argued that ideally estimates should be based on a gross output measure (which involves separate deflation of output and intermediate inputs), rather than a gross value-added measure; underlying series (e.g. different types of labour) should be aggregated using the Törnqvist index approach, rather than simply added together; the prices indices used to deflate output and inputs need to take account of quality adjustments (whereas in some sectors - such as non-marketable services - adequate price indices and measures of output are difficult to define conceptually); and lastly, there are significant issues with measuring the capital stock (including the appropriate rate of depreciation to use, and how to take account of pre-mature scrapping of assets through plant closures).
These issues are not trivial, and impact significantly on whether TFP can be measured and the accuracy of the resulting estimates obtained. In all, it is probably accurate to say that measuring productivity is as much an 'art' as it is a 'science'.
Section 3 provides a detailed breakdown of the sources of gross value added (GVA) data for the UK, the NIESR database, and Scotland. Generally, reliable data is available for the production sector throughout from 1979 onwards. Information for most service sectors is generally good from around 1993 for Scotland, but poor to satisfactory for 1979-1992. In summary, we conclude that outside of the production industries, there is limited data for the whole of the period since 1979; better data is available for 1993 onwards but this limits any analysis of the data to a 12-14 year period at best.
Section 4 considers the availability of data on labour inputs for use in measuring TFP. It is argued that such data are readily available for both Scotland and the UK. There are no major issues with being able to construct the relevant series by industry for the 1979- onwards period.
Section 5 provides a detailed breakdown of the sources of gross investment data for the UK and Scotland. Generally, reliable data (with which to compute adequate measures of capital stocks) is available for the production sector, transport and communications, and non-market services throughout from 1961 onwards. Information for most of the remaining service sectors is generally poor until 1998.
In summary, sufficient information is available for measuring output and inputs in the production sector of the economy, plus transport, communications, and non-marketed services. Whether this is sufficient to undertake further research to construct the required datasets depends on a number of issues. We discuss two of these in Section 6: firstly, are the sectors that can be covered, in terms of being able to generate values of TFP, sufficient for policy purposes? Secondly, is using aggregate industry-level data the appropriate level of analysis (vis-à-vis micro-level plant or firm based data)?
We argue that presumably the sectors of the economy of most interest are those that contribute the most to self-reliant economic growth. This concept is discussed in terms of recent literature on regional growth, and leads us to conclude that the sectors that contribute most to self-reliant economic growth include the production sector and producer services (the latter being loosely defined here as including communications, and certain elements of financial and business services). This report demonstrates that data is available for Scotland for the former, but it is currently inadequate to analyse TFP for the producer services sector.
Much of the recent literature would arguably suggest that micro-based studies of productivity (and the drivers of productivity differences across plants, firms and regions) offer the most promising approach to understanding what determines (regional) growth, rather than aggregate data. Thus, there is considerable scope for using the micro-data that is (becoming) available for Scotland (and other UK regions) to analyse differences in industry and regional performance, which takes account of the heterogeneous nature of economic activities. Greater use of this data is likely to provide a rich source of information to Scottish policy makers.