Date & Time：Dec. 16, 2020 14:00- 15:30pm
China has been experiencing a rapid expansion of the service sector since the mid-2000s. This paper uses a dual accounting approach to investigate the underlying forces for the structural change across sectors and regions. We first use development accounting to back out region-sector TFP growth from regional and sectoral employment shares. We then use the direct method that aggregates Hsieh-Klenow statistics to the region-sector level and apply it to a large-scale firm data set. Development accounting and the direct method will deliver the same estimates if the sectoral demand structure is correctly specified for development accounting. We find three main results. First, both development accounting and the direct method reveal fast manufacturing and producer services TFP growth. The sectoral TFP growth estimated by the two methods is also highly correlated, in line with the internal consistency of the dual accounting approach. Second, we decompose sectoral TFP growth by the direct method into efficient sectoral TFP growth, changes in sectoral wedges, and changes in within-industry misallocation. We find the manufacturing and producer services TFP growth to be primarily driven by efficient TFP growth but substantially slowed by the worsening of misallocation. Third, we infer misallocation from the dual accounting approach, which turns out to be highly correlated with the directly measured misallocation. The strong correlation not only reinforces the deteriorating misallocation by the direct method but also provides another validity check for our dual accounting approach. Holding within-industry misallocation constant would increase China's service employment share by 29 percentage points.
Xilu CHEN is a Ph.D candidate of economics at CUHK. His research interests focus on Macroeconomics, Chinese Economy, and Economic Growth. His papers have been published at Brookings Papers on Economic Activity, Economic Research Journal《经济研究》, and Journal of Sun Yat-sen University (Social Science Edition)《中山大学学报（社会科学版）》.