Analyzing and testing the structure of China’s imports for cotton : a bayesian system approach

Abstract: China is the biggest importer of cotton in the world, and it is of great interest to analyze its import demand system for cotton. Noteworthy is that cotton is deemed an intermediate commodity for textile production instead of a direct consumption one, hence this thesis adopts a two-stage cost minimization procedure resembling the Armington Model to study China’s demand for cotton differentiated by their countries of production with a focus on the substitution effects and weak separability between various sources. The CDE functional form is utilized for the cost function to generalize the Armington settings. Unlike former works, this thesis estimates the demand system with the Bayesian Bootstrap Multivariate Regression, and report the posterior distributions of coefficients and Allen elasticities of substitution. The results demonstrate that the CDE functional form may still be too demanding to be completely consistent with the data. Three weak separable structures are tested, and it can be concluded that the assumption of different separable structures improves the consistency between the model and data to various extents. The U.S. cotton is the least sensitive to its own price changes while the cotton from Egypt and Sudan is the most sensitive. The overall substitution effects between various cotton import sources appear to be relatively small.