Dynamics of the Comparative Advantage Structure and their Differences between Goods and Services

This study investigates state transitions of the comparative advantage structures by use of a Balassa index and one of the applicative methods of the stochastic process. The analytical framework consists of estimating non-parametrically the initial distribution of the transition probability matrix (TPM) that is well known as Markov matrix, evaluating the mobility, which means the degree of structural change in the analysis, of the initial distributions, and computing numerically the limit distributions of the TPM. The analysis also attempts to improve methods of the existing studies, and proposes a new evaluation index. Results of a series of analyses based on empirical data provide the following facts: firstly, the comparative advantage mobility differs between trade in goods and in services. Secondly, the state persistency of the comparative advantage indicates the higher probabilities in both strong and weak edge regions and the lower probabilities in intermediate regions, relatively. Thirdly, a convergence speed of the stochastic process depends on conditions of the initial distributions of the TPM. The convergence speed is related to the mobility, and differs between the goods and the services. Finally, external shocks stimulate the mobility. This study consequently brings out features of the comparative advantage structure dynamics.