Heterogeneity of Capital Stocks in Japan: Classification by Factor Analysis

This paper examines the heterogeneity of capital stocks using financial statement data of publicly listed Japanese firms. We conduct factor analysis on investment rates among various capital goods and estimate factor loadings of each as its reactions to common factors like total factor productivity (TFP) shocks. Then we estimate the uniqueness for each investment rate, which is the percentage of its variance that is not explained by the common factors. If the estimated factor loadings are similar between some of the heterogeneous capital goods, it may well imply that the adjustment cost structure of these investments is also similar. Further, if some of the estimated values of uniqueness are small, it suggests that certain theoretical models may track the dynamics of the investment rates well. Our estimation results show that Building and Structure have similar factor loadings as do Machinery & Equipment, Vehicles & Delivery Equipment, and Tools, Furniture, & Fixture. This suggests that we could remedy the Curse of Dimensionality by bundling the investments that have similar factor loadings together and that identifying the functional structures of each group of capital goods can greatly improve the performance of empirical investment equations.