A meta-regression analysis of agricultural total factor productivity in China

The results of Chinese agricultural total factor productivity (TFP) growth rates reported in current literatures have dramatic differences. A metaregression analysis of 2132 observations of agricultural TFP growth rates in China from 56 studies covering 1982-2013 is undertaken to test whether the agricultural TFP heterogeneities are systematically affected by the specific characteristics of previous studies. The main findings are as follows: (1) a mean agricultural TFP growth rate of 2.89% is obtained from the current literatures and the average agricultural TFP growth rate after 1978 is higher than that before 1978; (2) study-specific characteristics such as estimation technique, inputs size, data type, language used as well as geographical region have significant impacts on agricultural TFP growth estimates. Compared with Solow Residual Method, the Arithmetic Index Number Approach tends to report lower agricultural TFP growth rates while the Stochastic Frontier Analysis method is likely to display higher agricultural TFP growth rates. Studies based on time series data seem to present lower agricultural TFP growth rates than those with panel data and more inputs tend to result in lower agricultural TFP growth rates. In addition, paper languages are significantly correlated with agricultural TFP growth rates results. The results also show that the agricultural TFP growth rates in East and Central China are higher than that in national-level.