A Comparative Study of DEA-SFA for Industry-level in China

This paper aims to make a comparative analysis between Data Envelopment Analysis (DEA) and Stochastic Frontier Analysis (SFA) based efficiency scores and decomposed TFP (Total Factor Productivity) index, which are estimated from constructed industry-level data in China from 1985 to 2014. On one hand, the results are that DEA and SFA efficiency scores appear positively correlated and estimated TFP growth is in similar shape. On the other hand, for industry-level productivity in China, we found that according to the SFA, the estimation of TFP change and all decomposed elements showed a less noisy and much smoother shape when compared to DEA. The paper concludes that for an analysis of TFP of Chinese industry, the methodology of an SFA is more effective in explaining the changes and impacts as compared to the DEA. Since most of China’s industry level productivity studies have been done using DEA, we expect different and more practically significant results for future inter-industry studies. In this context, this paper would contribute to develop analytic methodology of China’s industry level productivity studies.