ABSTRACT
Taking advantage of annual data on the production of rice, wheat, maize (corn), soybeans, and oil-bearing plants in China, we calculate scale efficiencies of product yields and product output values from 1992 to 2016 using a variable return to scale–super efficiency model (VRS-SEM). The trade factors from supply and demand are verified using an error correction model (ECM). Apart from the description of different characteristics of products, the findings show the weak competiveness of grain and oil-bearing plants in international market, especially in soybean products as well as cost problems in scale production in the three major staple foods.
Supplementary Material
Supplemental data for this article can be accessed here.
Notes
1. High cost causes fiscal burden for the government. China’s grain distribution costs are over twice that of the average level in developed countries. Because of the government’s price subsidies, inventories are piling up in government storage facilities. Declining prices for food products, rising labor costs, and limited government subsidies have prevented farmers from planting cash crops.
2. State Administration of Grain in China.
3. The numbers in parentheses are all data in 2017.
4. Studies that use a DEA model (Jia and Xia Citation2017; Timpanaro, Urso, and Foti Citation2018; Wanke Citation2012; Xiao and Wang Citation2012; Xu, Luo, and Liu Citation2014) calculate scale efficiency based on input–output principles. If the input variables continue to be considered influential factors, the results of the regression must be significant. Therefore, the new variables are generally regarded as influential factors analyzed in the next step.
5. Before the regression model is predicted, we conduct an adjusted Dickey–Fuller (ADF) test on the time series. We find that variables are all integrated of order 1, and their long-term integration relationship is verified by a co-integration test. In the end, we adopt the ECM model.
6. The model passes both the autocorrelation test and the heteroskedasticity test, and the results are not shown due to space limitations. In addition, variable percentages are used to replace the original variables for a robustness test. The overall robustness rate is 85%, in which 90% of the import and export variables we focus on had in the same directions of influence as those in the original model.