ESTIMATING THE INPUT-OUTPUT BASED INDUSTRY MULTIFACTOR PRODUCTIVITY
The ABS publishes a variety of productivity measures in the Australian System of National Accounts (ASNA, Cat. No. 5204.0). The most comprehensive measure at present is the index of multifactor productivity (MFP) for the market sector of the economy. This aggregate MFP index is a ratio of the index of real gross domestic product over the index of combined capital and labour inputs in the market sector. In this context, the "market sector" excludes Property and business services, Government administration and defence, Education, Health and community services and Personal and other services, owing to the difficulty of estimating real output for those industries. The growth of this MFP index reflects productivity and technical change in the market sector as a whole.
There is, however, an ongoing, strong user demand for the MFP estimates dissected by industry, which the ABS is not yet able to meet. At present, users must either compile their own MFP measures, or rely on the labour productivity measure in ASNA, which is a partial measure of productivity and unsatisfactory in a number of ways. To fill this gap, Analysis Branch, in consultation with the National Accounts Branch, has initiated a project to estimate the industry MFP growth using the input-output (I/O) based approach. The project initiation has taken account of several improvements made recently in the measurement of inputs and outputs, as well as the integration of supply and use tables in both current prices and in the prices of the previous year and the national accounts.
The methodology of estimating industry MFP growth based on the I/O framework has been developed and adopted by Statistics Canada for its productivity accounts. It provides a unified framework under which the aggregate as well as the industry MFP growth can be derived. Under this methodology, several classes of industry MFP measure can be estimated simultaneously. These classes of industry MFP measure capture the different levels of productivity growth by industry as a result of various degrees of interconnection among the industries. Thus the estimates can generate new interpretations and insights to enrich our understanding of productivity dynamics of different industries in the economy.
The recent debate on the role of information and communications technology (ICT) played in the productivity surge since the second half of the 1990s seems to largely focus on the results from the studies using aggregate productivity measures. Since different measures of MFP growth at the industry level can be derived under the I/O based methodology, it is appropriate to use in the estimation of productivity growth and spill-over effects among industries in which the impact of ICT is typically observed. This methodology may also be useful to investigate the issues of whether use or production of ICT is the main driver of the recent productivity surge.
To apply this methodology to the Australian MFP estimation requires much research effort to understand the methodology and the data issues that are unique to Australia. Thus the project has been planned to consist of several stages. At the first stage, we have focussed on the conceptual and theoretical aspects of the I/O based MFP methodology to reach a good understanding of the relationships between this and other methodologies of estimating MFP growth in economics literature. Based on this understanding, the second stage of the project concentrates on empirical work, where the methodology is applied to estimate the indices of industry MFP growth for the market sector using the 1996/97 and 1997/98 supply and use tables and other relevant data. Currently, we are in the middle of this stage.
As the quality and plausibility of the results are dependent on the quality of the I/O tables and other data provided, we have been working very closely with the Input-Output Section, and Capital, Production and Deflator Section of the National Accounts Branch.
The second stage is also very critical for our project, because the results generated from the work will largely determine the feasibility of a full-scale implementation of the methodology in the final stage of the project. Currently, our pilot study has produced some preliminary results and they are undergoing various diagnostics and validation work.
For more information, please contact Simon Zheng on (02) 6252 6019
Please note there was no December 2002 issue of the Methodological News