Dependency and Savings

The South Korean experience can illustrate how high fertility contributes to large numbers of dependents. Total fertility was well above four children per woman until 1970, and for every 100 persons of working age (15-64) there were between 70 and 90 dependents (either younger or older). But fertility was declining rapidly, and the dependency ratio followed it down, so that now there are only 40 dependents per 100 persons of working age, fewer than the 50 per 100 typical in other industrial countries (Figure 5). The dependency ratio will rise again in the future as the population ages, but since the mid-1980s South Korea has enjoyed a relatively light dependency burden and will for several more decades.

A key argument in a landmark study 40 years ago was that a light dependency burden should be good for economic growth (Coale and Hoover, 1958). Among the effects of the dependency burden, the way it depresses savings has recently received renewed attention. Arguments about this have gone back and forth among economists, but improved models now confirm this effect cross-nationally (Kelley and Schmidt, 1996). An analysis of household income and expenditures in Great Britain, Taiwan, Thailand, and the United States concludes that "in all countries except Thailand, more children depress the saving rate" (Deaton and Paxson, 1997, p. 106).⁶ Careful analysis of the experience of East Asian countries suggests that their reductions in fertility in the past decades relieved not only dependency burdens but also dependence on foreign capital by contributing to high saving rates (Higgins and Williamson, 1997; Williamson and Higgins, 1997; Lee et al., 1997; Bloom and Williamson, 1997).

⁶However, Deaton and Paxson go on to try to estimate the aggregate effect of population growth on the saving rate and find it is trivial (and actually has the reverse sign). They achieve this contradictory result with a model that assumes a stable population (contrary to the typical situation in a developing country, in which life expectancies are rising and fertility is falling) and unchanging age profiles of income and saving (contrary to their own empirical findings). These unusual assumptions may be largely responsible for their perverse aggregate results: A stable population, for instance, will have a different distribution of age groups - and therefore different proportions of high-saving and low-saving households - than a population nearing the end of fertility transition.

Figure 5 - Total Fertility Rate and Dependency Ratio, South Korea, 1950-2050

SOURCE: United Nations (1996) and World Bank (1997a).

Saving rates are indeed exceptionally high in East Asia: 35 percent of gross domestic product (GDP) in Northeast Asia and 33 percent in Southeast Asia, in contrast to 21 percent in the countries of the Organization for Economic Cooperation and Development (OECD) and only 10 percent in South Asia.⁷ Relying less on government services, the multitude of households save substantial amounts as their members save to provide for retirement; possibly to fund bequests to the next generation; sometimes to hedge against financial setbacks; and occasionally to anticipate large expenditures, such as those for children's education. Over their lives, people ordinarily save most in their middle years when they are most productive, provided their child-rearing expenses have declined. And as life expectancies rise, Asian workers, increasingly less confident of being able to depend on children in old age and unprotected by elaborate social security schemes, have also been pressed to save more for retirement.

⁷For present purposes, Northeast Asia covers Japan, South Korea, Taiwan, Hong Kong, and Singapore; Southeast Asia covers Indonesia, Malaysia, the Philippines, and Thailand.

Figures 6 and 7 show trends in dependency ratios, savings, and investment in Northeast and Southeast Asia.⁸ As expected, savings and investment generally moved in the opposite direction from dependency, rising as lower fertility reduced the size of dependent cohorts. To determine whether this apparent link is causal requires investigating many possible complicating factors. For instance, strong economic growth itself can produce greater savings, as households find their incomes augmented by unexpected windfalls. When such relationships are accounted for, however, the results still suggest that dependency has a strong effect on savings.

⁸The graphs show unweighted averages across countries, leaving out Taiwan for dependency ratios. Savings and investment rates cover five-year periods centered around the given dates, except that the last point is for 1990-1992 (Williamson and Higgins, 1997). Dependency ratios are for the middle of each period (United Nations, 1996).

For Northeast Asia, the net effect of a dependency ratio above 60 percent in the early 1970s was to depress savings by 5.2 percentage points (as a share of GDP), whereas the net effect of a dependency ratio close to 40 percent in the early 1990s was to increase savings by 8.4 percentage points (Williamson and Higgins, 1997).⁹ These percentages are not trivial. For South Korea, for example, an 8.4-point increase was worth additional savings of close to US$25 billion a year in the early 1990s. This figure exceeds the total official development assistance received by all of East Asia and the Pacific in 1991 (US$17 billion) and is equal to more than half the assistance received by all developing countries combined (US$47 billion).

⁹That is, savings were depressed or increased by the amounts given relative to what they would have been if "population age shares" had been at the 1950-1992 means.

Net effects of dependency on investment were similar: A reduction of 3.7 percentage points in the early 1970s contrasts with an increase of 5.4 percentage points in the early 1990s. The dependency burden contributed to these countries' international debt from 1950 to 1980, but the lighter dependency burden has contributed to positive current account balances since then. A similar, but slightly more moderate, pattern of net effects on savings and investment has been demonstrated for Southeast Asia (Williamson and Higgins, 1997).

The net effect of dependency ratios was large enough to produce all the decline in foreign capital dependence after 1970 in both Northeast and Southeast Asia, by itself turning these regions from net debtors to net creditors on world capital markets. It also produced substantial capital deepening: From 1965 to 1990, capital per worker grew 2.7 percent annually in the United States, but 6.6 percent annually in Thailand, 7.6 percent in Japan, 8.6 percent in South Korea, and 8.7 percent in Taiwan (Summers et al., 1995). (Although the United States appears disadvantaged in this comparison, it actually benefited by using East Asian savings to finance its perennial trade deficit.) The substantial liquidity these savings created, plus the additional investment attracted from foreign sources, could have contributed to the financial excesses, given lack of effective bank regulation, that led to the Asian currency crisis in 1997.¹⁰ Clearly, a savings bonus from favorable demography can cut two ways: Failing to use it productively can be, eventually, as injurious as investing it well can be beneficial.

¹⁰Krugman (1998) provides an interpretation of the crisis that explains its untraditional character: Implicit government guarantees for banks and finance companies, coupled with poor regulation, led to distorted investment decisions and an eventual collapse in asset prices.

Figure 6 - Dependency, Savings, and Investment in Northeast Asia

SOURCE: Based on Williamson and Higgins (1997).

Figure 7 - Dependency, Savings, and Investment in Southeast Asia

SOURCE: Based on Williamson and Higgins (1997).

The large effects of reduced dependency on savings owed something to the fact that these economies were rapidly growing, and the impact of dependency changes could be magnified by other factors. But this issue aside, East Asia does not appear to be a special case. Modeling suggests that the effect of a given demographic shock on the current account balance has been no different in the rest of the world. However, in such regions as South Asia, where dependency reductions have so far been smaller, the impact has been correspondingly less positive (Williamson and Higgins, 1997; Bloom and Williamson, 1997).