Measuring and decomposing sectoral and economy-wide productivity growth: Assessing the accuracy of different approaches.
Productivity growth is on the forefront of European Union (EU) policy making. The Europe 2020 strategy's primary goal is to pursue "smart, sustainable and inclusive growth". To achieve that, the EU has targeted some key areas of 'Broad Economic Policy', which aim to assist the member states in achieving their productivity potential. With such increasing emphasis being placed on achieving productivity growth, it is crucial to first understand how productivity growth can be accurately measured.
There are various approaches that can be used to measure productivity growth; the most widely adopted approach is growth accounting (GA). GA is relatively easy to use and does not require information from outside the country or the sector examined, but it also requires the adoption of a number of simplistic assumptions, most notably those relying on the existence of perfect competition. If these assumptions are violated, GA could produce unreliable estimates. An alternative to the more traditional GA approach is to adopt one of the so-called frontier-based approaches. Such approaches relax the stringent assumptions required by GA and also allow for the decomposition of productivity growth, which could be of great interest to the users of productivity change estimates.
However, despite the theoretical advantages offered by frontier-based approaches, there has been very little research on whether these advantages do in fact affect the accuracy of the productivity growth estimates and by how much.
The aim of the research is to examine the various approaches that can be used to measure sectoral and economy-wide productivity growth and assess their robustness under different conditions. More specifically:
The research presents quantitative evidence on the overall robustness of a number of approaches (GA and frontier-based) using simulation experiments under various conditions that do not conform to the assumptions of perfectly competitive markets, such as the presence of technical inefficiency, measurement error, variable returns to scale, etc.
The research also provides quantitative evidence on the overall robustness of the decomposition of the productivity growth estimate, again through the use of simulation experiments and under various conditions that diverge from the assumptions of perfectly competitive markets.
Lastly, the research estimates and decomposes the productivity growth of a number of EU countries plus the USA, Japan and Australia, using a number of frontier-based approaches examined in the simulation exercises. This serves to provide robust, frontier-based productivity growth estimates and to demonstrate how the findings of the simulation exercises can be applied in a practical setting.
The findings of the research could be a great interest to national statistics offices and other organisations that examine productivity growth. They could lead to better estimates of productivity at country or sectoral level and these in turn could have significant policy implications (e.g. in the manner in which the impact of Structural funds is assessed).
Prof Emmanuel Thanassoulis