We are seeking an explanation for complex economic phenomena. The framework for that explanation is economic theory, and whether the explanation is convincing depends on whether theoretical predictions are consistent with empirical observations. If a theory clearly explains an economic phenomenon, we do not doubt the validity of that theory. However, if we only assess the validity of a theory with simple observations, the theory may generate inaccurate beliefs.
In a series of studies with Taro Ohno, including the RIETI discussion paper, "Trends in Japan's Household Savings Rates by Demographic Type: An Update and Discussion" (18-J-024), the author points out that the usual explanations for trends in Japan's savings rates may produce such beliefs. In the following, we discuss why the usual explanations are wrong and the type of mistaken beliefs that are generated.
Life-Cycle Theory and Japan's Savings Rate
Japan's household savings rate has declined significantly over the long term. The savings rate is an indicator of the percentage of disposable income that is saved (i.e., not consumed) and is an indicator of resources people have stored for the future. In macroeconomics, the higher the savings rate, the higher the rate of economic growth. According to the National Accounts (SNA), the most representative statistic of that savings rate, the macro savings rate, which hovered around 20 percent in the 1980s, is now almost zero percent. While there has been an upward trend in the savings rate over the past few years, the increase has been minimal, at a few percent, and it is striking that the trend of the savings rate is not as high as it once was.
The typical explanation given for the decline in the savings rate employs a combination of life-cycle theory and an aging population. According to life-cycle theory, households try to maintain a certain level of consumption throughout their lives, saving a lot during their working years when their income is higher and then withdrawing their savings as their income declines in retirement. In other words, the savings rate is high for young people and the savings rate is low (or negative) for older people.
Thus, an aging population means that the proportion of people who are withdrawing their savings will increase. This combination of life-cycle theory and an aging population can explain the macroscopic decline in the savings rate even if the savings rates of individual households do not change.
This explanation has been considered quite compelling, since it is clear that the population is aging, and life-cycle theory is recognized by many economists as an important part of modern economics. According to this explanation, the decline in Japan's savings rate is the result of micro decision-making in terms of household optimizing behavior and the demographic phenomenon of aging, both of which are uncontrollable elements of macroeconomic policy.
This consideration has led many economists to believe that a decline in the savings rate is inevitable. The aging of the population, in addition to the decline in the working population, will lead to a slowdown in capital accumulation, which has been the basis for the pessimistic view of Japan's economic growth.
Is an aging population the main cause of the decline in savings rates?
While this explanation is quite convincing at first glance, a closer look at the data reveals that it is not necessarily true. The logic of this life-cycle theory-based explanation is that the age profile of savings rate is nearly constant and the decline of macro savings rate is caused by the changes in the age-population structure. However, this is not the case because, as shown in the figure below, the savings rate of some age groups has changed significantly.
This figure is the result of our estimates using the National Survey of Family Income and Expenditure from 1989-2014. While it has been difficult to accurately estimate the savings rate for each age group due to errors and anomalies in official statistics, Unayama and Ohno (2017a, 2017b, 2018) constructed a reliable data by exploiting information from multiple surveys.
According to this figure, for every year of the survey, the savings rate remained almost unchanged until the head of the household was about 60 years old, and the savings rate tended to decline as the household aged beyond that (see Note 1). In this sense, the age profile is consistent with life-cycle theory. However, there is a significant difference in the savings rate for the older age group between 1989 and 2014.Until about 2004, the savings rate was around 10% even after age 70, but in the 2014 survey, it fell to a negative level. This fact implies that the decline in the macro savings rate is not only due to "an increase in the proportion of older people with low savings rates" but also "a decline in the savings rate for older people."
Furthermore, since macro savings rates are weighted averages of those oh each age group, we can decompose it and calculate the extent to which each group contributed to the decline in the macro savings rate. Unayama and Ohno (2017a) show that changes in the composition of the population by age group account for up to 35% of the total decline in the savings rate, with more than 65% being the result of a decline in the age-specific (especially among the elderly) savings rate. In other words, the notion that the decline in the macro savings rate is "brought about by the aging of the population and is inevitable" is nothing more than a mistaken belief.
The true cause of the decline in the macro savings rate
Unayama and Ohno (2018) provide some factual information on the reasons for the decline in the savings rate of the elderly. What we know so far is that the decline in the savings rate of the elderly occurred due to the lower savings rate of later cohort; that is, the savings rates of the during/post-war cohort are lower than those of the pre-war. They also show that the low savings rate of the during/post-war generations is not caused by high consumption but by low income, and that income differences are caused by differences in income sources such as property income and public pensions.
The pre-war had retired before the bubble burst and had enjoyed higher asset income. In contrast, asset incomes have reduced drastically when the during/post-war generations who retired after the bubble burst, became a significant asset holders due to low- and zero–interest-rate policies. Additionally, as public pension funding has deteriorated, pension benefit levels have reduced, substantially lowering the income levels below those of the pre-war generation.
In sum, the aging of Japanese society is not the cause of the decline in the macro savings rate. The real cause of the decline in the savings rate is the decline in interest rates caused by the bursting of the bubble economy and the decline in pension benefits due to the deterioration of public pension finances. If the demographic phenomenon of an aging population is the actual cause, there is little that economic policy can do to address the decline in the savings rate. However, if economic factors such as interest rates and pension levels are the cause, then it should be possible to respond within an economic framework.
A decline in the macro savings rate is a major constraint on economic growth. If the Japanese economy is to grow further, an increase in the savings rate is essential. We need to abandon the mistaken belief that the decline of the savings rate is inevitably with the aging of the population. To do so, we must examine the determinants of individual household savings rates. In that sense, the current ability to calculate age-specific savings rates is only the beginning of the process of getting it right.
^ The increase in the savings rate in the 85+ age group may be due to (1) errors due to the small sample size, and (2) classification bias due to the rarity of the case where the head of household (defined as the primary breadwinner of the household), is categorized as such (the effect of special cases, such as corporate executives and self-employed, being the primary breadwinner even at age 85) despite their advanced age.