PUBLICATIONS & REPORTS

Contents Introduction

This study develops a simple forecasting model using Japanese prefectural data. The Markov chain, known as a stochastic model, corresponds to a first-order vector auto-regressive (VAR) model. If the transition probability matrix can be appropriately estimated, a forecasting model using the Markov chain can be constructed. This study introduces a methodology for estimating the transition probability matrix of the Markov chain using least-squares optimization. The model is used first to analyze economy-wide changes encompassing all Japanese prefectures up to 2020. Second, a shock emanating from one prefecture is inserted into the transition probability matrix to investigate its influence on the other prefectures. Finally, a Monte Carlo experiment is conducted to refine the model’s predicted outcomes. Although this study’s model is simple, we provide more sophisticated forecasting information for prefectural economies in Japan. JEL classification: C15, C53, C61, O53, R12 Keywords: Prefectural economy, Japan, Stochastic model, Markov chain