Date of this Version
This research seeks to provide a comprehensive picture of lifetime asset allocation in Japan. We evaluate patterns in the level and composition of assets by household type, taking account of home ownership and household claims on social security as well as financial assets and pensions. The analysis relies on a micro-data taken from the RADAR survey fielded by Nikkei, for the year 2000. The RADAR data are the only publicly available dataset to record financial asset holdings in any detail, with which we calculate housing equity as well as other forms of wealth. We supplement this with external information on pension and social security entitlements. The resulting picture attests to the importance of housing and social security in the portfolios of households approaching retirement. As well, we compare asset allocation patterns between those who are worse vs. better off, better vs. worse educated, married couples vs others, and dualearner vs. single earner couples. Our econometric analysis of asset allocation choice reinforces priors regarding Japanese lifecycle asset allocation patterns: households do invest conservatively, other than their family home; they are risk-averse in their allocation of financial assets; and they appear to over-invest in housing and life insurance, two assets which enjoy preferential bequest tax treatment. We argue that institutional, historical, and policy influences explain much of observed Japanese preferences for safe financial assets, and we offer suggestions for mechanisms that could increase household diversification.
Working Paper Number
©2004 Pension Research Council of the Wharton School of the University of Pennsylvania. All Rights Reserved.
Funding for this research was provided by the Economic and Social Research Institute, the Japan Society for the Promotion of Science, the Project on Intergenerational Equity (Hitotsubashi University), the Pension Research Council and Boettner Center at the Wharton School of the University of Pennsylvania, and the Australian Research Council. Without implicating them, we acknowledge helpful assistance and suggestions from Rochelle Belkar, Noriko Inakura, Henry Jin, Masaharu Usuki, and Takeshi Yamaguchi, as well as seminar participants at the ESRI. This paper was presented at the Tokyo, Japan, February 2004 International Collaboration Forum organized by the ESRI, Cabinet Office, Government of Japan.
Date Posted: 30 August 2019