This section first describes a broad framework for analyzing the behavior of private saving and the major factors that may have contributed to the general decline in the saving rates of the industrial countries since the early 1970s. These factors include budgetary policies, the age structure of the population, changes in the level and distribution of household resources, interest rates and inflation, changes in certain relative prices, enterprise saving, and financial liberalization. While the discussion focuses on the response of saving to changes in these factors, it should be recognized that some of these factors are themselves influenced by saving behavior or by exogenous factors that influence saving. The interactions between these factors and private saving should properly be analyzed in a general equilibrium framework, but the development of such a framework is beyond the scope of the present study.
A major motivation for saving by individuals is to smooth out consumption over time on the basis of anticipated lifetime income. This idea is the basis for the development of the modern theory of saving behavior, which is expressed most familiarly in the “life-cycle” approach.6 saving also enables people to provide for bequests, to purchase housing or other major durable goods, or to cover contingencies such as medical emergencies. In order to account realistically for the influence of these various motives, the life-cycle approach has been extended to incorporate liquidity constraints and to allow for the fact that individuals have finite lives and may have limited planning horizons (Blanchard and Fischer, 1989).
The general life-cycle model suggests that individuals smooth consumption over time by taking into account anticipated changes in household resources, as well as the expected rate of return on savings. In this framework, private sector saving behavior is directly influenced by assessments of how the future pattern of income and consumption would be affected by government policy, particularly government saving or dissaving. Another important implication of the life-cycle theory is that the saving rate changes systematically over the individual’s lifetime. Consequently, the aggregate saving rate is directly influenced by the distribution of income by age. Because prime age workers who are near their earning peak have the highest saving rates, the life-cycle theory predicts that aggregate saving rates are higher, the larger the fraction of income going to prime age workers. The desire to accumulate assets for retirement or bequests will be affected by anticipated changes in income and wealth and by the availability of pensions. The need to save for contingencies will be affected by the extent of insurance coverage. And the motivation to save for large purchases of consumer goods will be influenced by the extent to which people can borrow for that purpose.
Factors influencing saving generally operate through two channels: changes in lifetime incomes and substitution between current and future consumption. To the extent that a given factor induces a rise in the present discounted value of lifetime income, it increases current consumption. Whether the saving rate will rise or fall will depend on the extent to which the factor in question affects current income. In many cases, however, factors that have a positive income effect on consumption also exert a negative substitution effect by lowering the price of future consumption relative to current consumption. Perhaps the best example of a factor with opposing income and substitution effects is a change in the interest rate. A rise in the interest rate will increase future income and wealth of households, prompting an increase in current consumption. On the other hand, a higher rate of interest implies that forgoing a unit of current consumption will yield a larger amount of future consumption, thus making current saving more attractive. The net effect of a rise in the interest rate on the saving rate is therefore ambiguous. This issue is explored in more detail in the section on “Interest Rates and Inflation” below.
Government Saving
The most important single component of the secular decline in national saving rates has been the drop in government saving in a number of countries (Section I above). The reasons for changes in government budgetary policies have been examined regularly in the Fund’s World Economic Outlook exercise, and are not repeated here. What is relevant for the present discussion is the extent to which shifts in government saving induce offsetting shifts in private saving.
The relationship between government and private saving has been a central issue in much of the modern literature on fiscal policy. Within the general life-cycle approach sketched above, there are three ways of looking at this relationship, each of which contributes to an overall understanding of the effects of recent shifts in government saving.7
Perhaps the most widely accepted view today is the general “neoclassical” version of the life-cycle model. A decline in government saving will tend to raise consumption and discourage saving by shifting the tax burden from present to future generations. In a closed economy, the resulting increase in the real interest rate lowers domestic investment and raises private saving, although the latter effect is likely to be relatively small. In an open economy, the rise in the domestic interest rate will attract foreign capital, while the associated appreciation of the currency accommodates the corresponding deficit in the external current account. In this case, the decline in domestic investment will be moderated by an inflow of foreign saving. In both cases, a decline in government saving will cause a decline in national saving.
The second variant is a Keynesian view, which is based on the premise that if there are unutilized resources in the economy, higher government dissaving will raise consumption and income through the multiplier effect. The increased income will, in turn, generate additional saving. In this framework, an initial decline in government saving may eventually be offset by the higher level of saving associated with the faster growth of employment and income. It should be noted that, in the Keynesian framework, shifts in the budget deficit are assumed to be temporary. The effect on private saving of a permanent change in the government deficit would be similar in this framework to that implied by the neoclassical approach.
The third view is based on the Ricardian paradigm, which begins with the notion that “for a given path of government spending, a deficit-financed cut in current taxes leads to higher future taxes that have the same present value as the initial cut.”8 Thus, to the extent that individuals are rational and farsighted, they realize that government spending must be paid for either now or later. In this framework, a rise in the budget deficit (dissaving) resulting from a tax cut would not affect the national saving rate because private saving would rise by an equivalent amount—thus the term “Ricardian equivalence”—in anticipation of future tax liabilities.
The Ricardian paradigm has been criticized for its unrealistically strict assumptions, which stipulate that (1) consumers are farsighted, (2) successive generations are linked by altruistically motivated bequests, (3) consumers do not face a liquidity constraint, and (4) taxes are nondistortionary. Not even the strong proponents of this theory would go so far as to say that the choice between deficit and tax financing is irrelevant (Barro, 1989b). They would point out, however, that any analysis of fiscal policy needs to take into account the underlying features of the Ricardian approach, while allowing for certain modifications.
The relative merits of the above approaches ultimately need to be determined empirically. Findings on this issue, however, are not conclusive (see Leiderman and Blejer, 1988). At one extreme, some studies, such as P. Evans (1988), have concluded that the data cannot reject the hypothesis of full Ricardian equivalence. Similarly, a study by Darby, Gillingham, and Greenlees (1987) indicates that in the United States, a dollar of tax income might yield only a 25 cent increase in national saving. At the other extreme, Ebrill and Evans (1988)—who point out a number of conceptual problems with the specification used by Darby and others—estimate “close to zero Ricardian equivalence” (p. i). More generally, most studies that focus directly on consumption have found an offset coefficient between 0.2 and 0.5 (Bernheim, 1989). Notwithstanding the serious difficulties of measurement and estimation faced by empirical researchers, the broad consensus seems to be that a significant portion of a change in government saving is likely to be reflected in national saving.
Age Structure of the Population
In general, young people save relatively little as they anticipate increases in their future incomes. Middle-aged people, who are nearing the peak of their earnings, tend to save the most, in anticipation of relatively low incomes after retirement. The elderly tend to have a low, or even a negative, saving rate, although the desire to leave a bequest or to cover the contingency of living longer than expected could provide a motivation for saving even after retirement. It follows from this discussion that the aggrgate saving rate falls in response to an increase in either the youth-dependency ratio (that is, the ratio of under-20 age group to the 20–64 age group) or in the ratio of the elderly to the working age population.
With declining birth rates and increases in life expectancy, the average age in most industrial countries has been rising during the past two decades. In most cases, this rise in average age has taken the form of a decline in the youth dependency ratio. Although the ratio of the elderly to the total population has also risen, the ratio of the elderly to the working-age population has been fairly stable in most industrial countries. Consequently, the effect on saving of the increasing average age has so far been small, except in the United Kingdom, where this aging process has been relatively advanced (see Currie, Holly, and Scott, 1989). Nonetheless, in a recent paper, Bovenberg and Evans (1989) find that, in the United States, demographic changes made an important contribution to the decline in the household saving rate in the 1980s. More generally, there has been a pronounced shift in relative incomes toward the elderly. For example, Summers and Carroll (1987) note that from 1950 to 1985 the ratio of per capita income of the elderly to that of employed persons rose by more than 10 percentage points in the United States. They conclude that this factor was probably the single most important cause of reduced saving in the United States. 9
As noted in the section on “Projections” above, shifts in the age structure of the population in these countries are likely to affect saving positively during the next decade, but this trend will be sharply reversed by the early years of the next century (see Heller, 1989, as well as Masson and Tryon, 1990). In addition, there may be a “generational” effect that could depress saving to some extent. Boskin and Lau (1988a and 1988b) find strong statistical support for the proposition that generations born after 1939 save substantially less than those born before 1939 (who presumably have a better recollection of the Great Depression). According to this “vintage effect” of saving, the aggregate saving rate will fall substantially as the share of pre-1939 generations in total population diminishes. The implication of these trends is that, in the absence of positive influences from other sources, further declines in private saving rates are in the offing, albeit only gradually.
Income and Productivity Growth
An important implication of the life-cycle view of consumer behavior is that faster growing countries should have a higher aggregate saving rate (Modigliani, 1986). The rationale for this proposition is that countries with higher rates of economic growth display a larger gap between the lifetime income of current workers and that of retired workers, resulting in a higher aggregate saving rate—one that is closer to the saving rate of the more numerous and prosperous workers. This proposition is consistent with the positive correlation between the saving rate and growth observed across countries; most notably, Japan (with gross saving of more than 30 percent of GNP) grew at an average rate of 4 percent from 1980 to 1988, while the United States (with gross saving of about 17 percent of GNP) grew at a rate of 2¾ percent. This correlation could also be attributed to a number of other factors, including the reverse causation running from saving to investment and growth. (The latter issue is discussed in Section III below; also see Carroll and Summers, 1989.)
Wealth Revaluation
In the life-cycle framework, one of the main determinants of saving is the value of wealth, including the discounted present value of future labor income as well as the value of the individual’s net assets. For a given level of current income, the saving rate will be reduced by an increase in the value of wealth, since less saving will be needed in order to accumulate assets to any target level.
In most industrial countries, owing to revaluations of equities and housing, the value of household wealth increased substantially through most of the 1980s and diminished the need for saving out of current income. This effect was particularly prominent for Japan where, according to Fund staff estimates, a near quadrupling of the ratio of household net wealth to GNP during 1982–87 accounted for about three quarters of the total decline (of 3.2 percentage points) in the saving rate. In the United States, Fund staff estimates indicate that the rise in the wealth-income ratio could explain a 2 percentage point decline in the personal saving rate from the 1970s to 1986–88. Although it is perhaps too early to form judgments about the most recent period, it is noteworthy that private saving rates seem to have recovered somewhat since the sharp decline in equity prices of October 1987.
Pension Plans and Insurance
The increased availability and benefits of public pension and social security plans lower the private saving rate through three channels: by redistributing income to the elderly; by reducing the need to save for retirement (unless it contributes to a lowering of the retirement age); and by curbing the need for precautionary saving to cover the contingency of living longer than expected (see O. Evans, 1983). The impact of increased social security benefits on national saving will depend on the corresponding developments in public saving. To the extent that the social security system is unfunded, public saving will not rise directly to offset any induced decline in private saving. It could be argued that in this case the private sector will not be prepared to reduce its current saving as it would anticipate that taxes will eventually have to be raised to finance future pension outlays. Empirical evidence suggests, however, that increases in public pensions result in a decline in private and national savings, albeit by less than the full increase in pension benefits.
In contrast to social security plans, private pensions are likely to add to total saving. In principle, individuals should regard their contributions to funded private pensions as a perfect substitute for other forms of saving. But, in practice, individuals do not seem to fully take into account their pension contributions in determining their saving behavior. In this context, and to the extent that private pension plans are based on defined benefits, a surge in the bond and stock markets (such as that observed in the early 1980s) would lead to a reduction in required contributions by both employees and employers and thus lower the private saving rate. Empirical estimates for the United States suggest that about one third of any addition to private pension plans would be added to total private saving (Smith, 1989).
The increased availability of various kinds of insurance, such as health, liability, unemployment, personal loss and liability, and life, may also have influenced saving behavior. To the extent that insurance plans limit expected outlays for contingencies and emergencies, they reduce uncertainty and therefore the need for precautionary saving.
Relative Prices
Preferences for saving may be affected by shifts in relative prices if the shifts are thought to be temporary or if they involve capital goods. A temporary rise in the price of consumer goods may induce individuals to postpone consumption, while a rise relative to capital goods may shift aggregate expenditure from consumption to investment. These effects—like those of changes in interest rates, discussed below—are very difficult to evaluate empirically. Expectations are hard to assess, there may be offsetting income and substitution effects, and there is likely to be some reverse causation as relative prices are themselves affected by shifts in saving behavior. There are, however, some prominent examples that seem to have been important in recent years.
The price of capital goods has declined relative to consumer goods during the 1980s, in part because of technological advancements, particularly in electronics and office automation. For example, in the United States the deflator for nonresidential fixed investment declined by more than 10 percent relative to the consumption deflator from 1980 to 1987. Such a decline will have a positive wealth effect and a negative substitution effect on private saving. An indirect estimate of the net effect is provided by a study by Blinder and Deaton (1985) of consumption functions for the United States. The study suggests that a 10 percent decline in the price of durable goods relative to nondurable goods would ultimately reduce consumption of nondurables (that is, raise saving) by 2.7 percent, although in the short run the saving would fall.
Another important relative price movement that has occurred in many industrial countries during the 1970s and 1980s is the sharp increase in the real price of housing. This increase has had two offsetting effects on household saving. More expensive housing is likely to have increased the saving rate of the young (who aspire to purchase a house), but also to have shifted the distribution of wealth from the young to the old, thus reducing the aggregate saving rate. The net effect is therefore undetermined.
The realignment of the major currencies and the associated changes in the terms of trade during 1985–86 also represent a significant relative price change. The substantial depreciation of the U.S. dollar raised the cost of imports in the United States and reduced them elsewhere. Although a permanent shift in the terms of trade would not necessarily have a relative price effect on the saving rate, a temporary rise in import prices in the United States may well induce a rise in demand for current consumption and thereby cause the U.S. saving rate to fall, with offsetting shifts in the saving rates of other countries. 10
The relationship between exchange rate changes and saving rates is affected by the source of the exchange rate change as well as by its permanence. A depreciation could result from an expansionary monetary policy, a contractionary fiscal policy, or a shift in portfolio preferences away from securities denominated in that currency. While a temporary depreciation could raise the national saving rate regardless of its source, the effect will generally be much stronger if it is associated with a fiscal consolidation, which in any case will raise national saving directly (see Boughton and others, 1986).
Interest Rates and Inflation
As noted earlier, the effect of a change in the real interest rate on saving is theoretically ambiguous. Over the years, substantial empirical work has been undertaken to identify the sign and quantify the magnitude of the interest elasticity of saving. An interesting study of saving in Canada by Beach, Boadway, and Bruce (1988) finds that the interest elasticity is positive for the younger age groups and negative for the older age groups. In this framework, the sign and magnitude of the interest elasticity would depend, inter alia, on the age structure of population; the study estimates a positive aggregate elasticity for Canada. However, the weight of the empirical evidence—which is concentrated on the United States but also includes studies for the other industrial countries—supports the view that the partial correlation between the interest rate and the saving rate is likely to be small, irrespective of its sign. 11
Saving rates may also respond to changes in the rate of inflation, although again the net effect of the changes that have taken place in inflation during the 1970s and 1980s has probably been small. An increase in the inflation rate, even when accompanied by a commensurate increase in the nominal interest rate, erodes the real value of household wealth and raises the saving rate. On the other hand, a higher rate of inflation is generally associated with higher uncertainty on the rate of return, which could have a depressing effect on the desire to save. The effect of inflation on saving will also depend on the country’s tax system. For example, in those countries in which such expenses are deductible (for example, the United States), a higher rate of inflation will lower the real after-tax rate of interest and encourage borrowing, thus lowering saving. This issue is discussed more fully in Section V.
Enterprise Saving
In analyzing the behavior of private saving, a pertinent issue is whether enterprise and household saving are determined jointly or separately. Recent trends described in Section I suggest a tendency for changes in these two components to offset each other in some countries. One explanation for this pattern is that an increase in enterprise saving generally induces an upward revaluation of enterprise equity which, through a wealth effect on consumption, leads to a decline in household saving. Thus, households may “see through the corporate veil” and adjust their own saving to compensate for fluctuations in enterprise saving. Empirical studies of this issue confirm that there is a significant, negative correlation between enterprise and household saving, but that the offset coefficient is below unity. 12
The composition of private saving between the enterprise sector and the household sector is also affected by inflation. To the extent that the enterprise sector is a net borrower and the household sector a net lender, a rise in inflation (with unchanged real interest rates) would lower real liabilities of the enterprise sector to households. This inflation effect contributes to the negative correlation between enterprise and household saving.
Financial Liberalization
During the last two decades, financial markets in the industrial countries have undergone substantial reforms. 13 These reforms include relaxation or elimination of ceilings on interest rates, the opening of money and capital markets to small savers, relaxation of controls on international capital markets, and expansion of the range of activities permissible to banks and other financial institutions. While these reforms have been aimed mainly at strengthening the mobilization and allocation of financial resources, and have thereby improved the efficiency with which savings are used, they may have also contributed to the observed decline in national saving rates.
The desire by households to smooth consumption over time would be constrained by the limited access to credit (see Hubbard and Judd, 1986). Consequently, a relaxation of this constraint will allow individuals to bring forward their consumption (and thereby reduce saving) over their working life through borrowing. An obvious example is greater availability of finance for purchases of housing and expensive consumer durables, which has obviated the need for households to save in preparation for making large down payments for such purchases.
The increased availability of certain financial instruments—particularly futures and options—has provided insurance against some forms of risk. As with any form of insurance, these instruments reduce the uncertainty associated with future income and serve as a substitute for precautionary saving. At the same time, to the extent that these instruments reduce the risk associated with the expected rate of return, they could encourage saving. The net effect of these instruments on saving is not, therefore, expected to be large.
Financial deregulation has also had a direct effect on the rate of return to saving. Insofar as improved financial intermediation stimulates economic growth and lowers the cost of intermediation, it has probably increased returns to saving. Such an increase, however, would not be large and, in any event, it is not likely to have a significant, positive effect on private saving (see section on “Enterprise Saving” above).
A noteworthy development with implications for private saving has been the introduction of leveraged buyouts of corporations in the 1980s. Some argue that leveraged buyouts have increased the efficiency of firms through corporate restructuring. But leveraged buyouts have also increased cash payments from the enterprise sector to households. There is some evidence that these payments have had a supplementing, rather than a supplanting, effect on dividend payments. While households may have reinvested part of their additional cash, they are likely to have increased their overall consumption. According to an estimate by Summers and Carroll (1987), corporate repurchases and takeovers in the United States may have lowered the U.S. private saving rate by 1 to 2 percentage points.
