In this paper, we apply Kazuo Sato's target wealth hypothesis to saving for life after retirement and analyze the impact of social security wealth, retirement payments, permanent income, and other factors on people's retirement saving using micro data from the Survey of Social Security and Self Help, which was conducted in 1996 by the Japan Institute of Life Insurance. Our findings provide strong confirmation of the target wealth hypothesis and of the life cycle model and imply that the Japanese take account of their future social security benefits and retirement payments, their permanent income, etc., when saving for life after retirement. J. Japan. Int. Econ., June 2001, 15(2), pp. 131–159. Graduate School of Economics, and Institute of Social and Economic Research, Osaka University, 6-1, Mihogaoka, Ibaraki, Osaka, 567-0047, Japan
Nowadays, retirement security system programs such as social security and
employer-provided pensions and lump-sum retirement payments play an important
role in providing old-age security in Japan. However, these programs are expected
to undergo dramatic change in the coming years. For example, the rapid aging
of the population will put severe strains on the public pension system. Moreover,
401(k)-type defined contribution company pension programs will be introduced
beginning in 2001 and are expected to partially or largely displace current defined
benefit pension and lump-sum retirement payment programs. Under such circumstances,
we urgently need to improve our understanding of retirement saving and
of the impact of social security and employer-provided pensions and lump-sum
retirement payments on retirement saving.
In this paper,we apply Kazuo Sato’s (1995) target wealth hypothesis to saving for
life after retirement and analyze the impact of social security wealth, retirement
payments, permanent income, and other factors on people’s retirement saving
using micro data from the 1996 Survey of Social Security and Self Help (Kouteki-
Hoshou to Jijo-Doryoku ni kansuru Ishiki-Chousa), which were provided by the
Japan Institute of Life Insurance (JILI) and the Information Center for Social
Science Research on Japan, Institute of Social Science, University of Tokyo (SSJ
Data Archive).
According to the life cycle hypothesis of Modigliani and Brumberg (1954),
as extended by Feldstein (1974), people save for life after retirement, taking
account of their own expectations concerning their living expenses after retirement,
social security benefits, retirement payments, etc. In Japan, a large number of detailed
studies have been conducted on the relationship between social security and
saving since the 1980s (Tachibanaki and Sasaki, 1985; Homma et al., 1987; Dekle,
1990; etc.). For example, Dekle analyzed the impact of social security on household
saving using micro data from the 1983 Survey on the Living Behavior of the Aged
(Roujin Ishiki Chousa). He finds that the Japanese elderly are not dissaving and that
social security does not appear to displace private tangible wealth in Japan. One
of the most careful of these studies is Takayama et al.’s (1990b), which uses micro
data from the National Survey of Family Income and Expenditure (Zenkoku Shouhi
Jittai Chousa), conducted by the Management and Coordination Agency of the
Government of Japan. There is no information on the amount of future social
security benefits in this survey, but the authors estimate the amount thereof and analyze the relationship between saving and social security benefits and confirm
the existence of a replacement effect of social security and retirement payments
on household saving. Two other relevant studies are Horioka and Okui (1999) and
Horioka et al. (2000). Horioka and Okui (1999) uses micro data from the Comparative
Survey of Savings in Japan and the United States, a binational household
survey conducted in 1996 by the Institute for Posts and Telecommunications
Policy of the Ministry of Posts and Telecommunications of the Government
of Japan, to analyze the importance and determinants of retirement saving (not
total saving) in the United States and Japan. They found that retirement saving
is far more important quantitatively in the United States than it is in Japan
and that, in both the United States and Japan, retirement saving is influenced
by some (but not all) of the factors identified by the extended life cycle model,
especially expected living expenses during retirement. Horioka et al. (2000) does
a similar analysis for Japan using micro data from the Survey on the Financial
Asset Choice of Households, a household survey conducted by the same
entity.
This paper improves upon earlier studies in at least four ways2: first, the survey
we used for our analysis focuses on retirement and collects information on
living expenses during retirement, saving for retirement, etc., making it ideal for
the purposes of our analysis. Second, we analyze the impact of social security and
other factors on people’s retirement saving, not their total saving. Most previous
authors analyze people’s total saving, but social security should affect primarily
people’s retirement saving, not their total saving.3 Third, whereas previous studies
typically estimate future social security benefits from the provisions of the social
security system and the individual’s earnings history, we use information on the
respondents’ expectations concerning their future social security benefits because
the estimated amounts do not necessarily approximate people’s actual expectations
and because data on earnings histories are not available in the case of Japan, as they
are in the case of the United States (see, for example, Bernheim, 1988). Fourth,
we examine the replacement effect not only of social security wealth but also of
retirement payments on people’s retirement saving.4 A large number of detailed
studies have been conducted on the relationship between social security and saving,
but very little has been written on the relationship between retirement payments
and saving due in large part to the unavailability of data on retirement payments in
2 The first three also apply to Horioka and Okui(1999) and Horioka et al. (2000).
3 Some might argue that saving is not fungible and that it does not make sense to speak of saving for
a specific motive, but I disagree for the following two reasons: first, household surveys consistently find
that respondents are able to allocate their saving and wealth among specific motives (see, for example,
Horioka andWatanabe (1997)). Second, in Japan, there are many types of saving accounts for specific
motives such as housing purchase and retirement, and penalties are imposed if the funds are used for
a different purpose.
4 There are two types of retirement payment systems in Japan. One is a lump-sum retirement payment
system (in Japanese, taishokukin), and the other is an employer-provided pension system (in Japanese,
kigyou nenkin) In this paper, we refer collectively to lump-sum retirement payments and company Japan.We believe that it is important to consider the impact of retirement payments
on saving for the following four reasons: first, retirement payments are one of the
most important sources of income after retirement (in 1997, 88.9% of companies
had a retirement payment system). Second, in the future, retirement payments will
play an even more important role in providing old-age security because the rapid
aging of the population has made people feel uneasy about their future social security
benefits. Third, the proportion of companies with a company pension system
has increased over time, and because company pensions are often portable, this has
increased the importance of retirement payments as an income source during retirement.
Fourth, the role played by retirement payments is different from that played
by social security benefits. For example, homeowners often pay off their outstanding
housing loans using their retirement payments. No information is available on
the amount of retirement payments in the survey we used in our analysis, so we
estimate the amount thereof from information on income at retirement and firm
size.
To previewour main findings, we obtain strong confirmation of the target wealth
hypothesis and of the life cycle model and find that the Japanese take account
of their future social security benefits and retirement payments, their permanent
income, etc., when saving for life after retirement.
This paper is organized as follows. In Section 2 we discuss theoretical considerations,
in Section 3 we describe the data source, in Section 4 we describe the
variable definitions and calculation method, in Section 5we describe the estimation
model and estimation method, in Section 6 we present some descriptive statistics,
in Section 7 we present our estimation results, and Section 8 concludes.