Understanding Australia’s industry-level productivity dynamics: from measurement to econometric estimation

Abstract

This thesis investigates a range of issues related to the industry-level productivity growth in Australia. It approaches the issues from three different, but related angles. First, it derives the estimates of the industry-level multifactor productivity (MFP) index for the 12 market-sector industries in Australia using the non-parametric, growth accounting approach. It then seeks to explain the industry-level productivity changes by estimating the relationship between R&D and MFP growth for several one-digit level industries in Australia. The third major part of the thesis attempts to provide some systematic explanations for the poor productivity performance that has been reflected in measured MFP for the mining industry during the recent mining boom in Australia.While there have been numerous studies on industry-level productivity dynamics in Australia, the work for this thesis is unique in two respects. First, it takes measurement issues seriously. The emphasis of the study is placed on the understanding of the conceptual basis and statistical limitations of the industry-level MFP measures by exploring the links between production economics and economic measurement. Second, the study decomposes the measured MFP growth into the effects of the ‘true’ MFP growth, returns to scale, capacity utilisation and natural resource inputs, and then empirically estimates these components for Australia’s mining industry. This contributes to the analysis and debate on mining productivity in Australia, which so far seems to have attracted limited attention.The estimation of the MFP index and the proposed solutions to several econometric specification issues presented in the study are consistent with Australia’s System of National Accounts, an economic measurement framework used by the Australian Bureau of Statistics (ABS). Based on the econometric results, the study finds some evidence of a positive relationship between the industry’s own R&D capital stock and its productivity growth in several industries in Australia. However, doubts remain about precision and reliability of the estimated returns to industry R&D. Based on the components in the MFP decomposition estimated from a variable cost function of the translog form, the study concludes that the so-called mining ivproductivity paradox observed recently in Australia is in fact created by the measurement issues, particularly by the omission of the natural resource inputs, that have caused various biases in measured MFP, while the ‘true’ productivity growth in Australian mining has remained positive and stable over the sample period.