2002 Japan Conference: A Summary of the Papers

This paper provides a summary of the portfolio allocation of Japanese households in recent years and studies the relationship between portfolio choice and age for Japanese households. It uses micro data and pays particular attention to the interaction between decisions to hold stocks versus real estate.

The relationship between age and portfolio structure has been a focus of attention among economists for several reasons. First, this relationship is directly related to the various issues of an aging economy. The potential effects of the aging of the population on the level of national saving and the level of household wealth have drawn much attention. At the same time, the composition and the riskiness of the wealth of older Japanese are equally important in understanding the welfare implications of an aging economy. On a more practical level, how individuals allocate their portfolios is relevant to the debate concerning defined contribution pension plans that allow participants some discretion in their investment choices. How Japanese households allocate their wealth, and how that may change, are also very important in understanding the Japanese Big Bang, that is, the ongoing structural change in the Japanese financial system. Many macro and financial economists, most notably Hoshi and Kashyap, view the bubble economy in the second half of the 1980s and the prolonged economic and financial turmoil since the early 1990s as intimately related to the structural changes in the Japanese financial system --- specifically, the shift from a bank-oriented system to a market-oriented system. In current discussions, changing corporate financing decisions and corporate governance have been the main focus of analysis. However, if the way that firms raise funds for their business (that is, the supply structure of financial assets) changes, then the way households allocate their funds (the demand structure of financial assets) also must change. So, studying the portfolio structure of Japanese households is essential to understanding the changing Japanese financial system as a whole. Finally, dynamic portfolio choice recently has re-emerged as a major research topic in finance. Recent theoretical developments allow us to analyze dynamic portfolio choice when stock returns are predictable and in the presence of uninsurable labor income risk. In response to these theoretical developments, some recent empirical studies -- such as Amerkis and Zeldes (2001), Bodie and Crane (1997), Poterba and Samwick (1995, 1997), and the papers in the volume edited by Guiso, Haliassos, and Jappelli (2001) -- investigated household portfolio choice in the United States and major European countries by emphasizing its relationship to age.

My paper investigates the relationship between age and portfolio structure of Japanese households using annual survey data published by Nihon Keizai Shimbun, data known as Nikkei Radar. Nikkei Radar has some important limitations: in particular, the data covers only the Tokyo metropolitan area. For Japan as a whole, there is some work on household asset allocation that uses Nikkei Radar and other datasets. But the previous work emphasizes the uniqueness of Japanese household portfolios, or structural changes in investment behavior, from a microeconomic point of view. This paper is the first of its kind, using Japanese data, on the age-related pattern of stock investments and its relationship to real estate holdings.

Stock holdings by Japanese households, measured by shares of equities in households financial wealth, increased during the years of the bubble economy, from 25 percent in 1987 to more than 35 percent in 1990. After the bubbles in the Japanese asset market burst in the early 1990s, the shares of equities declined continuously: 18 percent in 1993, 12 percent in 1996, and as low as 7 percent in 1999. Participation in the stock market, measured by the fraction of the population owning any equity, also increased in the second half of 1980s, from 26 percent in 1987 to 30 percent in 1990. It then decreased in the first half of 1990s, down to 24 percent in 1996, and has remained at a similar level since then. Given the very poor performance of the Japanese stock market through the 1990s, the decline of stock holdings by Japanese households may seem natural. But Japan is an exception among OECD economies. In most countries, household participation in the stock market has increased in the last ten years.

On the other hand, the basic pattern of the relationship between portfolio choice and age for Japanese households has been stable and is quite similar to that of western countries. Equity shares in financial wealth have a humped-shape pattern: they increase with age among young households, peaking in middle age, then decline. The significant difference in the data reported in Amerkis and Zeldes (2001) is that the peak of equity shares comes at a much later stage of life. Also, the decline of equity shares after retirement age seems to be much slower in Japan than in the United States. The proportion of the U.S. population owning equity displays a hump-shaped pattern with age, and equity shares in financial assets conditional on ownership are mostly constant with age. I confirm that these patterns also exist in the Japanese data. As in the United States, the age-related pattern in the share of equity in Japanese households financial wealth can be explained mostly by the decision to own (or not to own) stocks.

I go on to discuss the pattern of real estate holdings by Japanese households. The same age-related pattern, with the life-cycle variation, is found for real estate holdings in Japan. The age-related pattern of real estates shares in total household wealth is hump-shaped and is explained mostly by the decision to own (or not to own) real estate (mostly owner-occupied housing). In fact, this age-related pattern is even clearer with real estate shares/holdings.

On the other hand, no age-related pattern in equity holding is observed for households that do not own real estate. These findings suggest that the age-related pattern observed in stock holding can be explained by household's tenure choice of housing: younger households tend to accumulate their wealth in safe assets to save for purchasing houses. After they purchase a house, they are restrained from taking risks in financial investments because of their highly leveraged positions in housing loans. The demand for equity must be more elastic in terms of wealth level for homeowners than for non-owners. Only after they have purchased places to live do households begin to make risky financial investments. This can explain why stock holdings by very young households are so low in Japan. Also, since Japanese land prices are so high, average down payments and housing loans are larger in Japan than in the United States. So the shares of real estate in total wealth naturally are higher in Japan than in the United States. This explains why Japanese households own less equities than U.S. households and why the peak of their stock holdings comes at a later stage of life. Therefore, any serious attempt at modeling Japanese households' dynamic portfolio choice should incorporate the effect of housing tenure choice.

Despite high land prices, there are strong incentives for Japanese households to own rather than to rent housing. In the second half of my paper, I emphasize and discuss the sources of such incentives. First is overprotection of tenants in the Japanese legal system. It is difficult for landlords to raise rents and even more difficult for them to remove tenants. Such overprotection makes landowners afraid of large investments that might turn sour and of the re-development of old, existing rented houses. As a result, the supply of rental housing in Japan is limited and the quality of that supply is worse than for owner-occupied houses. The average size of owner-occupied houses is almost the same in Japan, France, and Germany, but the average size of rental houses in Japan is only two-thirds that of those in Europe. Such inefficiencies in the Japanese housing market limit the supply of quality rental housing, forcing households to hold very large shares of their assets in the form of owner-occupied housing and to take risky positions in their portfolios. Thus, the willingness of households to take risky positions in the financial market is intimately related to their positions in the housing market.

My paper also emphasizes the problem with the Japanese bequest tax. The inheritance tax burden in Japan is much heavier than in the United States and most developed economies. At the same time, if one plans a bequest, it is preferable from the standpoint of tax saving to hold real estate rather than financial wealth. This is because financial assets are valued at market value and real estate historically has been valued at below market value in bequest tax assessment until the early 1990s. So, there is a strong tax incentive for Japanese households to hold real estate and to take out housing loans: the latter is tax deductible at market value if one is planning a bequest. Also, for residential real estate there are huge tax deductions in general. This tax system explains why Japanese households prefer to hold owner-occupied houses rather than to rent houses. It also helps to explain why the elderly in Japan retain houses and other real estate until their death.

In Japan, policy recommendations have been made in recent years to promote household stock investments by changing the tax systems and removing obsolete regulations. However, one potentially important explanation for why Japanese make few stock investments is that, since they have already taken extreme positions by purchasing their residence, they simply cannot take risky positions in financial investments. If so, then removing obstacles to stock investment alone will not be sufficient to induce individual investors' equity holdings. Legal and structural reforms in the housing market might be important in promoting stock investments by individuals and in enhancing the efficiency of the financial system in Japan. It has been suggested that high real estate prices and large down payments provide some explanation for the high household saving rate in Japan (Hayashi, Ito, and Slemrod, 1988). My paper suggests that, in addition to the effect on savings pointed out by Hayashi et.al., high land prices and housing market imperfections very likely affect the allocation of Japanese households' financial wealth. This result has policy implications and should be explored more carefully in future research.

Finally, the investment behavior and the tenure decisions of Japanese households are related to the Japanese employment system. When the average household takes out housing loans, the lender has to be convinced about the prospects of the household's future loan payments. In Japan, it is not unusual that the monthly repayment of housing loans exceeds half of household expenses. Although the housing loan is usually backed by a mortgage, if the household becomes unemployed, it will immediately experience serious financial trouble because housing accounts for such a large proportion of living expenses. So, given the high real estate prices in Japan, it is very important that workers have a relatively stable and safe future labor income in order to finance housing purchases. Without the conventional lifetime employment system, this will be very difficult. The collapse of the lifetime employment system, combined with the lingering recession of the 1990s, is thus likely to depress demand for stocks by Japanese households.