V Saving in a Growing World Economy
- International Monetary Fund. Research Dept.
- Published Date:
- May 1995
Saving is the process by which an economy sets aside part of its output and uses it to generate income in the future. Individuals, corporations, and governments save. Households set aside money for home ownership and for retirement, businesses store up earnings to construct new factories, and governments build up assets in public pension systems and infrastructure. Aggregate savings provide the ultimate constraint on global investment spending, and therefore play a critical macroeconomic role as well. Since investment spending provides a key link to productivity and real income growth, there must be an adequate supply of saving for the world economy to advance at an acceptable pace.
With today’s large potential investment demands, especially in emerging market economies in the developing world and among the countries in transition from central planning, the question naturally arises whether the supply of saving will be adequate to finance worthwhile projects around the world. The question is made sharper because of growing pressures on savings. Since the early 1980s, government absorption of private sector saving has increased sharply, and the retirement-age populations of some industrial countries are starting to increase rapidly. Meanwhile world real interest rates in the mid-1990s are already high by historical standards. What are the trends in world saving patterns and how important is saving for economic growth? What are the main determinants of saving and why do saving rates vary across countries and over time? To what extent can international capital flows allow countries to borrow savings from abroad? To what extent is a “saving shortage” already putting upward pressure on world interest rates? And what policies might be used to encourage more saving? These issues are the focus of this chapter.
Saving is an enormously complex issue and many aspects of the process are still not well understood. There are disagreements even about the basic definition of saving, and there are myriad data problems.44 Most definitions, for example, view saving as a residual concept—as income less consumption, or in an international context, as investment demand minus net capital inflows. As a result, measures of saving can vary greatly because of small differences in the measurement of other difficult-to-measure concepts. Further, many assumptions that are necessary to test key hypotheses about saving, such as assumptions that financial markets offer flexible opportunities for borrowing and saving, or that world capital markets are well integrated, may simply not be justified. In short, it is not surprising that analyses of the saving process do not produce the clean, conclusive results that policymakers might like.45
Trends in Saving Patterns
Long-run historical data on saving patterns suggest that before the 1920s most industrial countries, including Australia, Canada, France, Japan, and the United Kingdom, had very low gross national saving rates—in the range of 10 to 15 percent of GDP.46 The exception was the United States, which as early as the 1870s had a saving rate of almost 20 percent. While the U.S. saving rate stayed remarkably constant over most of the twentieth century until the 1980s, virtually all other industrial countries and fast-growing developing countries have seen their saving rates move up to the 20 percent range. A few, such as Japan, China, and some of the newly industrializing Asian economies have had rates that exceeded 30 percent.
The world saving rate climbed steadily over the 1960s, driven partly by rapid increases in Japan’s saving rate. The saving rate registered twin peaks during the oil price spikes in the 1970s, and has fallen noticeably since 1980 (Chart 23). Statistical tests find evidence of structural breaks in the level and trend of the world saving rate after 1972 and again after 1980, so it is convenient to look at saving behavior over three time periods: the pre-oil shock years, 1960–72; the period of adjustment to the oil shocks, 1973–80; and the post-oil shock period, 1981 to the present. Using this breakdown, the world saving rate averaged 23 percent in the first period, increased to 25 percent in the oilshock period, and then declined to 22½ percent in the most recent period. Over 1992 and 1993, the most recent years for which there are historical data, the world saving rate averaged just 21¾ percent—down over 2 percentage points from the late 1960s and 1970s.
What sectors were responsible for this decline in the saving rate in the 1980s? Breaking down the national saving data for the major industrial countries—which account for roughly half of world saving—into private and public saving reveals that virtually all of the decline took place in public sector saving (Chart 24). For these countries, the private saving rate averaged 20 percent in both the 1960–72 and 1981–93 periods.47 But the public saving rate fell from 4 percent of GDP in the 1960–72 period, to just ½ of 1 percent in the 1981–93 era. Declines in the public saving rate occurred in every major industrial country except Japan.
Chart 24.Major Industrial Countries: Gross National Saving Rates1
1 Data prior to 1963 exclude France.
In contrast to the industrial countries, the average national saving rate in developing countries has shown a sharp upward trend—from about 19 percent of GDP in 1970 to 27 percent today (Chart 25). Much of the boost has been due to increases in Asian countries. The saving rate for developing Asian economies rose by over 10 percentage points, from 21 percent in 1969 to over 31 percent in 1993, mainly because of increases in private saving rates. Some developing countries that have been successful in sustaining high growth rates have even seen their saving rates double over the past twenty years or so, including Chile, Korea, Malaysia, and Thailand. In contrast, there was a decline in the saving rate in Africa, largely attributable to a falling public saving rate. The aggregate saving rate in developing Western Hemisphere countries rose during the 1970s, but has been declining since 1989, mainly because of decreases in private saving rates. Among the factors responsible for the fall in the saving rate in these countries were the surge in capital flows and financial liberalization, which resulted in increased access to domestic and foreign borrowing.
Chart 25.Saving Rates
Saving trends in many eastern European countries in transition are hard to analyze because of data limitations. Saving rates in these countries were high before transition, largely because limited consumption opportunities led to so-called forced saving. Private saving rates have probably been declining during the early phase of transition. Household saving rates have fallen because of negative income shocks, and business sector savings have almost certainly diminished, although data limitations make quantification difficult.48 Most of these countries have also experienced significant government dissaving—on the order of 5 to 15 percent of GDP a year. There is little doubt that aggregate saving rates in these countries are much lower than they were before transition.
Chart 26 shows the relationship between per capita income levels and the gross national saving rate for the period 1989–93. These data suggest that national saving rates tend to increase sharply as per capita income increases, and that saving rates level off or even decline somewhat in high-income economies. Very low saving rates are most often observed in low per capita income countries, in part because subsistence limits the ability to save. The tendency of saving rates in middle-income countries to rise substantially is noticed particularly among newly industrializing Asian economies. And the leveling off of saving rates (or even declines), are most noticeable in such high per capita income countries as Canada, the United Kingdom, and the United States.
Chart 26.Gross National Saving Rates and Per Capita Income Levels1
1 The shaded area illustrates a range that captures the experience of most countries. The pattern implied between saving rates and per capita income levels is consistent with statistical evidence reported later in the chapter.
When thinking about the overall supply of saving to meet world investment demands, it is both the actual quantity of saving and the ability for it to move across international borders that matters. Predictably, the largest suppliers of gross saving are also the largest economies (Chart 27). Japan and the newly industrializing Asian economies supply proportionally more saving than the size of their GDPs would suggest, while the United States supplies less. The fact that the industrial countries supply about half of total world saving clearly understates their role, since most of the developing countries and the transition countries are still less than fully integrated into world financial markets. It is still the industrial countries that dominate world capital markets, an issue that will be taken up below.
Chart 27.Share of World Output and Saving, 1993
Important Questions About Saving
Does Saving Cause Growth or Does Growth Cause Saving? One of the most striking regularities in cross-country data is the relationship between the rate of saving and the growth of output. High-saving countries generally grow faster than do low-saving countries (Chart 28). Fourteen of the world’s 20 fastest growing economies over the past ten years had a saving rate over 25 percent, and none had a saving rate under 18 percent. Meanwhile 8 of the world’s slowest-growing 20 economies over the same period had a saving rate below 10 percent, and 14 were below 15 percent.49 But the linkage from saving to economic growth is not ironclad. Some countries have had very high saving rates and comparatively low economic growth—Switzerland, for example—while others have experienced healthy economic growth despite a relatively low saving rate, at least for a period of time. Chile, for example, has enjoyed over 6 percent average annual real GDP growth over the past ten years, while its saving rate has averaged just 18 percent.
Chart 28.Saving Rate and Per Capita Real GDP Growth1
1 The shaded area is merely indicative of a positive correlation between the saving rate and the real per capita GDP growth.
Understanding the direction of causation that underlies the relationship between saving and growth is difficult—indeed, there is good reason to believe that there are positive effects running in both directions. The positive effect of saving on growth is the more straightforward: higher saving raises the growth rate of output by increasing capital accumulation.50 The traditional policy recommendation flowing from this view is that in order to increase the pace of economic growth, countries need to think first about boosting their saving in order to spur capital formation.
Empirical evidence suggests that income growth also has a positive effect on saving. Visual examination of data from the high-saving, high-growth countries of east Asia, for example, suggests that these countries experienced high growth before their saving rates rose. In the 1950s, 1960s, and early 1970s, for example, increases in saving rates in Japan and Korea lagged a few years behind increases in GDP growth rates (Chart 29). And the slowdown in Japanese growth in the late 1960s was followed by a decline in the saving rate. Recent research, including statistical causality tests, increasingly points in the direction of a link from growth to saving.51 The experience of the industrial countries, where saving rates fell in the presence of slow output and productivity growth following 1973, is also consistent with this view.
While the latest research on the causality between growth and saving may not be totally conclusive, the data suggest that there may be a virtuous circle between growth and saving. Increases in growth raise the saving rate, which in turn feeds back to increase growth. This has potentially broad ramifications. It might suggest that an acceleration of growth in developing countries is possible even in the absence of an initial jump in saving, if say, the boost came from technology transfer.52 It would also suggest that much of the saving that fast-growing developing countries will need in the future to fund their investment needs will probably be self-generated.
What Is the Right Amount of World Saving? Is It Possible to Save Too Much? According to some, the world economy saves and invests too little, and more saving would always be better. The world economy today would benefit from higher saving and investment rates, but it is theoretically possible for the world to save too much. The question of optimal saving belongs in the domain of welfare economics and growth theory, where the answer depends to a large extent upon subjective judgments, such as the weights one would give to the interests of different households, generations, or countries. Theoretical growth models have been used to identify “golden rules”—saving and investment rates that would lead to optimal economic growth under hypothetical conditions. In one formulation, additional increases in the saving rate increase welfare as long as the rate of return (determined by the supply of saving) exceeds the rate of population growth. In another variation, a higher saving rate improves welfare as long as the dividend flow from investment projects is positive. Both of these conditions easily hold for the world economy today, suggesting that too much saving is hardly a problem.
While oversaving is theoretically possible in a closed economy and might lower domestic welfare, excess saving by one country may be invested in other countries in a world of open capital markets. The other countries would benefit from lower interest rates and more capital formation, and hence would enjoy higher economic growth. However, such a situation would also be characterized by trade imbalances that could fuel perceptions of unfair trading practices and lead to political frictions and financial market pressures. In practice, there may therefore be limits on the extent to which a country can accumulate assets abroad.53
Why Do Some Countries Save So Much? Will They Continue to Do So?Chart 30 shows the world’s highest- and lowest-saving economies in each decade since 1960.54 At the top of the list are Japan, Korea, Taiwan Province of China, and China. The extremely high saving rates of a set of east Asian economies is remarkable: Japan’s saving rate averaged 34 percent in 1990–93, Korea’s 35 percent, China’s 37 percent, and Taiwan Province of China’s 29 percent. In the absence of these “supersavers,” the world saving rate would be significantly lower. For example, if Japan’s saving rate in 1993 had fallen to the 16 percent average rate of the other major industrial countries, the world saving rate would have been 1½ percentage points lower. Among the lowest-saving countries are Zaïre, Côte d’Ivoire, and Cameroon. Perhaps the most interesting case is Korea, which was one of the world’s lowest-saving countries in the early 1960s, and one of the highest-saving countries in the 1980s and 1990s.
Chart 30.Highest- and Lowest-Saving Economies1
1 Excluding economies in transition and major oil exporting countries.
The importance of these high-saving countries raises the questions of why this phenomenon is observed, and whether it can be expected to continue. Numerous causes have been suggested for high saving. In several east Asian countries, government policy has favored saving, either through public saving, mandatory private saving, restrictions on consumer access to credit, or the provision of widely accessible savings vehicles. Many observers have suggested that cultural factors are at work, and while that may be part of the explanation, it is not the complete story. For example, there is little evidence of high saving in east Asia before World War II, while in the late nineteenth century the United States stood out as a relatively high-saving country.55
No fully convincing case can be made for any one explanation for high saving, but a natural suspect is high growth, as described earlier. If fast growth does promote saving, then future saving may be expected to fall, at least in one high-saving country, Japan. Japanese growth over the period 1970–93 averaged 4¼ percent a year. While a return to high growth in Japan cannot be ruled out, it seems unlikely since the process of catching up has been largely completed. In addition, the population in Japan is starting to age more rapidly, and this will also work to reduce the Japanese saving rate. Overall it seems very likely that the Japanese saving rate will show steady erosion over the coming years. On the other hand, there does not appear to be any strong reason why China’s saving rate will not remain high for some time, especially if growth remains high.
Key Factors Affecting Saving
To better understand the likely future supply of world saving, the key factors that influence saving behavior should be examined. There have been two fundamentally different theoretical views about the prime motivation for saving. In one view, saving is seen as resulting from a choice between present and future consumption.56 Individuals compare their rate of time preference to the interest rate, and smooth their consumption over time to maximize their utility. The interest rate is the key mechanism by which saving and investment are equilibrated. The other view sees a close link between current income and consumption, with the residual being saving. In this view, saving and investment are equilibrated mainly by movements in income, with the interest rate having a smaller effect. A hybrid view attempts to reconcile consumption smoothing with the income determination of consumption, by seeing a concept of permanent income as driving the consumption process.
The extensive literature on the determinants of saving is still open on many key questions, including whether there is substantial intertemporal smoothing of consumption, as held by the first view, or whether consumption (and hence saving) is mainly determined by current income patterns, as held by the second view. The thorny question of causality between income growth and the saving rate, as discussed earlier, also remains indeterminate. And the impact of changes in the real interest rate on the saving rate is theoretically and empirically ambiguous: there is both a positive substitution effect and a possibly negative income effect (if, for example, a rising interest rate causes a drop in the required contributions to a defined benefit pension plan), and empirical studies usually find little effect. Public pensions also might affect private saving, because they might substitute for private pension contributions, but again the empirical evidence is mixed.
The literature has reached tentative conclusions on other saving issues. Empirical research has tended to reject the notion that swings in government fiscal position are fully offset or matched by private savers (so-called Ricardian equivalence) for both industrial and developing countries.57 The typical result is a roughly one-half offset, although some researchers find a larger offset among countries with particularly high rates of government dissaving. The literature generally finds a negative impact of financial liberalization on saving behavior in industrial countries, although the effect should be transitory. New outlets for financial saving may become available and encourage saving, but the loosening of constraints on borrowing is usually viewed as the dominant factor at least in the short run.58 The literature also finds generally ambiguous effects of taxation on saving behavior. This is because taxes and saving subsidies work through their effects on the real after-tax interest rate, and as mentioned above, the saving rate shows little sensitivity to changes in the real interest rate.
Age-specific effects on saving are generally found to be significant. The life-cycle model, where saving is seen as providing for retirement, predicts that the saving rate should decline as the retired proportion of the population increases. This result is often extended to those below labor-force age as well, with the general result that a higher share of dependents implies a lower saving rate. Finally, a temporary improvement in the terms of trade is considered to increase the saving rate, because it suggests a boost in transitory income. This effect can be huge if the swing in the terms of trade is large. For example, the oil shocks of the 1970s represented a massive shift in the terms of trade in favor of oil exporters and led to sharp though temporary increases in the world saving rate. The funds that flooded into OPEC (Organization of Petroleum Exporting Countries) coffers simply could not be spent as rapidly as they mounted.59
For developing countries, there are some further determinants of saving. Outlets for financial saving are more limited because capital market imperfections are more widespread, financial markets are less developed, a lack of confidence in the banking system may exist, and interest rates tend to be heavily regulated. As a result, and because of liquidity constraints and subsistence considerations, the effect of the interest rate on saving tends to be weaker. This effect may also vary with income and wealth, with countries closer to subsistence levels less able to substitute intertemporally (Box 10).
In many developing countries, foreign saving can be a determinant of total saving because foreign borrowing is rationed. Although foreign saving can promote growth by increasing the total saving available for investment, it may partially replace national saving by raising domestic consumption. There is some empirical evidence of a negative relationship between domestic saving and foreign saving in certain developing regions.60 Some studies suggest that capital flows to major Latin American countries on average have lowered national saving rates, whereas flows to high-growth Asian countries have arguably added to national saving and increased overall investment.61 Another important source of external saving in poorer developing countries is foreign aid. Empirical evidence on the impact of foreign aid on saving is mixed, although it tends to be negative. Recent studies suggest that about 40 percent of foreign aid on average goes into consumption, reflecting the fact that in most cases foreign aid goes to countries where per capita income is low and subsistence considerations dominate saving decisions.62
Box 10.Saving and Real Interest Rates in Developing Countries
There is little consensus on the size or significance of the response of saving to changes in real interest rates in developing countries. This may reflect data problems, the lack of sophistication and depth in domestic financial markets, or regulations that limit the scope for market-determined interest rates in many low-income developing countries. Most important, perhaps, is that interest rates may have little or no impact simply because household saving may be essentially zero in countries that are at or near subsistence income levels.
Recent studies have attempted to estimate the interest rate sensitivity of household saving in developing countries using macroeconomic data. This sensitivity depends on how easily households can substitute their consumption over time—technically, the intertemporal elasticity of substitution in consumption. In contrast to other studies, the approach adopted does not require the interest sensitivity of saving to be equal across countries with different per capita incomes, which is consistent with subsistence considerations, and allows relative prices of imports and home goods to enter into households’ saving decisions.1 A temporary reduction in import prices may reduce saving since the price of imports is a component of the consumer price index, whose changes over time (like the change in interest rates) affect incentives to consume today or to save in order to consume more in the future.
The results suggest that the household saving rate in upper middle-income developing countries is likely to increase significantly as interest rates move up, and the response is unlikely to be very different from what would typically be observed in industrial countries. However, the interest elasticity of saving varies considerably with the level of wealth in the way predicted by the subsistence model: in low-income countries, the interest elasticity of private saving is close to zero, but it rises markedly in low middle-income countries, and increases further in upper middle- and high-income countries. Financial liberalization—and the resulting increases in interest rates—may have a number of positive effects, including increasing the efficiency of investment and strengthening economic growth, but the direct impact of such policies on household saving behavior is likely to be relatively small in low-income countries. Increasing national saving in such countries may require an alternative strategy. The experience of a number of high-performing east Asian economies highlights the role of prudent fiscal policies, low inflation, and macroeconomic stability as important means of increasing national saving.2 In addition, trade liberalization and other structural reforms strengthen price signals, help financial resources flow more efficiently, stimulate the growth of productivity and output, and hence promote saving.1See Jonathan D. Ostry and Carmen M. Reinhart, “Private Saving and Terms of Trade Shocks,” Staff Papers (IMF), Vol. 39 (September 1992), pp. 495–517; and Masao Ogaki, Jonathan D. Ostry, and Carmen M. Reinhart, “Saving Behavior in Low- and Middle-Income Developing Countries,” IMF Working Paper 95/3 (January 1995).2See World Bank, The East Asian Miracle: Economic Growth and Public Policy, Policy Research Report (New York: Oxford University Press for the World Bank, 1993). Compulsory saving programs may also have been effective in boosting saving in some Asian countries. Other factors that may have contributed to high saving rates in the Asian countries included effective prudential supervision of banks, which enhanced confidence in the financial system, and regulation of interest rate spreads, which may have lowered the costs of financial intermediation.
A background study carried out by the staff for this chapter has tested how these kinds of factors might influence the private saving rate.63 The study performed both time-series and cross-section regressions using panel data for 21 industrial countries and 64 developing countries over the period 1971–93. The analysis found that most standard economic variables are correctly signed and generally significant. Higher output growth, higher real interest rates, and improved terms of trade tended to increase the private saving rate, whereas increases in government surplus, greater wealth, higher per capita income relative to the United States (a proxy for catch-up effects in growth), and a higher fraction of young and old all suggested a lower private saving rate. Generally speaking, the study found less support for standard saving theories among lower-income developing countries, perhaps because of less reliable data.
Box 11.Pension Reform in Developing Countries
Pension systems differ in the way they finance and provide pensions, in whether they fix benefits or contributions, and in whether they are run by the state or by the private sector. In most countries, pension systems are state run and offer benefits that are determined by law rather than by lifetime contributions. These unfunded, defined benefit systems operate on a pay-as-you-go basis that transfers resources from the contributors to the beneficiaries.
Although unfunded public pension systems may have advantages over privately run, fully funded systems in terms of the extent of coverage, the protection of the retirees, and equity considerations, there are well-known problems with these systems. The lack of a direct relationship between benefits and contributions can seriously strain government budgets; wage taxes may distort labor markets and encourage tax evasion; and the aging of populations, especially in industrial countries, is increasingly making these systems unsustainable in the long run.
Several developing countries, in particular in Latin America, have increasingly moved toward fully funded pension schemes, while some east Asian countries have used such schemes for a number of years.1 Chile, where the traditional pay-as-you-go system had developed a host of problems by the late 1970s, has adopted the most far-reaching reforms. Pension reform began in 1981 when the fiscal surplus was sufficiently large to finance the transition to a fully funded system, and after the retirement age was increased to 65. A privatized savings plan was introduced whereby workers were required to place 10 percent of their earnings in any one of a number of highly regulated intermediaries; at retirement, they could choose between a sequence of phased withdrawals or an annuity. There are indications that the increase in the retirement age and the fiscal consolidation that accompanied the reforms contributed importantly to the subsequent surge in private saving and to a strengthening of potential and actual growth.2
Moreover, Chile’s capital markets have been deepened by creation of a market for indexed annuities, by the steady flow of contributions into investment funds, and by significant improvements in the functioning of capital markets. The reforms have also insulated the pension system against the possible budgetary risks of a state-run defined benefit systems. Administrative costs—roughly 30 percent of mandatory saving—are high relative to a single compulsory system, however, partly because of the costs of competitive attempts to attract customers. Moreover, in contrast to a defined benefit system, uncertainties regarding the length of working life, the duration of retirement, and the return on invested contributions make pension benefits somewhat unpredictable.
Pension reform is also being discussed or implemented elsewhere in Latin America. In Colombia, the 1993 pension reform overcame many of the flaws of the previous system by reducing net benefits paid to future pensioners and by gradually phasing in a dual pay-as-you-go, fully funded system.3 In Brazil, one proposal based on the Chilean experience would raise the retirement age to 65 and substitute a fully funded system based on individual accounts for the current pay-as-you-go system; another proposal would combine defined benefit and defined contribution schemes in order to protect retirees and reduce the fiscal cost by preserving a partial pay-as-you-go system. In Argentina, the minimum retirement age was raised to 65, phased in over a number of years, and a privately capitalized system was introduced in July 1994. The privately administered pension funds, which presently cover some 6.3 million enrollees, co-exist with the pay-as-you go system, predominantly for older contributors who preferred to remain in the existing system.
Several east Asian countries have used variations of the defined contribution system, in part as a tool of development policy.4 The pension systems in Malaysia and Singapore rely primarily on provident-fund, fully funded, defined contribution schemes, where compulsory contributions are maintained in a central fund with separate accounts for individual contributors. Upon retirement, benefits consist of accumulated contributions plus interest and take the form of a lump-sum payment. Compulsory coverage is generally restricted to wage earners in the formal sector, and those with low wages or with short employment records receive limited coverage. One of the virtues of provident-fund systems is that they may generate higher saving. While it is possible that voluntary private saving may be reduced as compulsory saving increases, the experience of Malaysia and Singapore, where national saving rates typically exceed 30 percent, suggests that provident funds may have contributed to high overall national saving rates.
The experience of developing countries that have reformed their pension systems indicates that a combination of mandatory fully funded schemes with supplementary defined benefit arrangements relieves pressures on budgets while providing protection for retirees. The mandatory, fully funded pillar of the system may lead to higher saving by making people more aware of the need to save for the future, and through forced saving. It is also likely to deepen capital markets through private sector participation in the investment of pension funds. To the extent that reforms lead to higher saving, they are likely to have a positive effect on growth. The increased capital market deepening that results from the reforms is also likely to contribute to higher output growth, which in turn would generate higher saving.51See Patricio Arrau and Klaus Schmidt-Hebbel, “Pension Systems and Reforms: Country Experiences and Research Issues,” Revista de Analisis Economico, Vol. 9 (June 1994), pp. 3–20.2See Peter Diamond, “Privatization of Social Security: Lessons from Chile,” Revista de Analisis Economico, Vol. 9 (June 1994), pp. 21–33; and references in The Chilean Economy: Policy Lessons and Challenges, ed., by Barry Bosworth, Rudiger Dornbusch, and Raul Laban (Washington: Brookings Institution, 1994).3See Klaus Schmidt-Hebbel, “Colombia’s Pension Reform: Fiscal and Macroeconomic Implications,” Policy Research Department (Washington: The World Bank, October 1994).4See Frederick Ribe, “Funded Social Security Systems: A Review of Issues in Four East Asian Countries,” Revista de Analisis Economico, Vol. 9 (June 1994), pp. 169–82.5The experience of reforms in relatively high-income developing countries may not be applicable to low-income countries because of the less-developed institutional and financial environment.
Among the staff study’s main conclusions are that changes in government fiscal position seem to lead to only a partial offset in private saving. A simple average of both industrial and developing country results suggest a value of about three fifths—considerably below the unity implied by full Ricardian equivalence and in line with other estimates in the literature. Another conclusion is that demographic effects are significant and have generally worked to increase saving rates over the past twenty-five years. A third finding is that there does appear to be a positive association between GDP growth and the saving rate, and in addition, there is some evidence that the saving rate may flatten out or even decline at high per capita income levels. The study also finds that changes in the terms of trade have a strong positive effect on saving. For developing countries, foreign saving reduces private saving, but increases total saving available for investment. Finally, the study concludes that a 100-basis-point increase in the real interest rate causes the saving rate to increase by a bare 0.2 percent of GDP in industrial countries.
With respect to changes in public saving, three factors can be distinguished: business cycle effects in the short run, political will to contain the accumulation of public debt in the medium term, and structural government obligations in the long run. Business cycle effects will not be discussed because of the medium- to long-run focus of this study. Over the medium term, however—say five to ten years—it seems that reasonably major shifts in public saving are possible if political will is strong. Over the longer term, the most important government obligation by far is public pension liabilities.64 As of 1990, it was estimated that the present value of net pension liabilities for the major industrial countries on average exceeded 130 percent of their GDP.65 With aging populations, a financing crunch seems unavoidable, and only a few choices are available. Countries might increase contribution rates, which would tend to increase directly the public saving rate. They could reduce pension benefits over time, which would probably cause the private saving rate to increase as workers sought to provide some private offset for the reduced benefits. They could try to shift responsibilities back to the private sector (Box 11). Or countries could raise the retirement age—another way to reduce benefits—which would cut the dependency ratio, lead to fewer aggregate retirement years, and likely give an upward boost to the private saving rate.66 The overall outcome is hard to predict, but as countries begin to address their pension difficulties, it seems likely that both public and private saving rates will have to increase.
Future Supply of Saving
Overall the prospects for world saving over the next fifteen or twenty years are not necessarily grim, especially if rapid growth in high-saving countries can be sustained. The key will be what happens to public saving. Private saving rates, as mentioned, have remained roughly flat over the past three decades, and will probably not experience any sharp upward or downward movement over the next couple of decades. The world public saving rate fell sharply after 1980, however, and was the dominant cause of the overall decline in the world saving rate over the period 1981–93. Even a modest swing in the world’s public saving position over the next couple of decades will probably dominate other movements.
Perhaps surprisingly, world population trends are not likely to have a major overall effect on the world private saving rate over the next twenty years. There will be gradual further declines in the aggregate private saving rate in industrial countries because of steady increases in their dependency ratios.67 On average, the dependency ratio in industrial countries will increase from about 66 percent in 1995—about its trough level—to roughly 70 percent by 2015, which should tend to reduce the industrial country saving rate by less than 1 percentage point (Chart 31).68 The effect on Japan will be much more noticeable: its dependency ratio will increase by roughly 20 percentage points, which should cause the Japanese saving rate to decline by roughly 3 percentage points—say from 33 percent to 30 percent if other things remain equal.69
Chart 31.Dependency Rates1
1 The blue shaded area indicates the United Nations population projections.
But this drop in the private saving rate in the industrial countries is likely to be offset by steadily falling dependency ratios in developing countries.70 After peaking at more than 120 percent in 1970, the average dependency ratio in developing countries has been dropping steadily, and should continue to drop over at least the next twenty years according to United Nations demographic projections. Between 1995 and 2015, for example, the dependency ratio will drop from about 90 percent to about 70 percent, which could raise the developing country private saving rate by about 3 percentage points. Given that developing countries represent approximately one third of world GDP on a PPP basis, this increase should roughly offset any private saving decline in the industrial countries. Moreover, the developing country share of world GDP is likely to increase substantially over the next twenty years, and that in itself should also boost the world saving rate.
The impact on the saving rate of increases or decreases in real interest rates is likely to be fairly small. The effect of economic growth—as captured through the stages of economic development effect—is hard to quantify. While the private saving rates in maturing economies may decline somewhat, new groups of countries should achieve economic takeoff and experience sharp increases in their saving rates. It is difficult to predict which countries will grow fastest, but it certainly is not inconceivable that fast growth and saving surges in countries like Brazil, China, India, or Indonesia could offset or even dominate saving declines in countries like Japan or France.
The key factor that will influence the future supply of world saving is the fiscal position of governments, which remains a big unknown. Suppose, on the one hand, that governments cut their deficits and increased their public saving rates to one half of the major industrial country rate in the 1960–72 period—to 2 percent. That would represent an increase in the public saving rate of about 1½ percentage points of GDP (compared with the 1981–93 average saving rate of ½ of 1 percent of GDP), and with roughly a one-half private sector saving offset, would suggest an increase in the world saving rate of about ¾ of a 1 percentage point. Or suppose that governments increased their public saving rates all the way to the 1960–72 average rate of 4 percent of GDP. That would represent an upward shift of government saving of 3½ percentage points, which after the private saving offset would give an upward boost of roughly 1¾ percentage points to the world saving rate.
On balance, the future supply of world saving depends upon two critical variables: the pace of economic growth among developing countries and the amount of public dissaving among industrial countries. If some large developing countries grow rapidly and experience high saving rates, and if the industrial countries return to their public saving rates of the 1960–72 era, the world saving rate could easily increase by 2 to 4 percentage points—despite steady declines in private saving rates in Japan and other industrial countries. This might be called the optimistic saving scenario. On the other hand, a pessimistic saving scenario might develop: in the face of stagnant economic growth in key developing countries and even larger budget deficits in industrial countries, the world saving rate could decline by an additional 2 or 3 percentage points over the next fifteen to twenty years.
Limits on Borrowing the Saving of Others
It has been argued that the amount of saving supplied and demanded by developing countries and countries in transition may be a key force that will influence world real interest rates. If countries with high demands for saving enter the world capital market, the world real interest rate will rise, while the interest rate will fall if countries with potential surplus saving enter the market. Recent attention has focused on the possibility that eastern Europe, the former Soviet Union, and developing countries with newly liberalized capital markets will absorb large amounts of world saving over the coming decades.
In assessing the importance of these new capital demands, a key question is the potential volume of such saving flows. Generally speaking, the net flow of saving to a given country will be a function of the gap between domestic savings and domestic investment demand at the world interest rate, the country’s size, and the openness of its capital market to international flows (Box 12). This openness, in turn, will depend on institutional characteristics, such as legal restrictions on profit repatriation or foreign ownership, and on the premium over world interest rates demanded by foreign investors.
Massive net flows of saving from one country to another remain the exception rather than the rule even after substantial financial liberalization. Feldstein and Horioka, for example, found in 1980 that national rates of saving and investment were highly correlated—a fact that has been interpreted as evidence that changes in domestic savings translate mostly into corresponding changes in domestic investment, and not into capital flows.71 More recent research has found that the correlation between national saving and investment has been slowly falling over time.72 Work done at the Fund suggests that part of the reason for the high correlation between domestic saving and investment ratios is that some governments attempt to target the current account and adjust fiscal policy accordingly.73 An additional reason is that financial markets may be unwilling to finance large current account deficits on a sustained basis. (A corollary is that investment demands in most successful developing countries will almost always be funded largely by domestic saving; given the apparent link between economic growth and saving described earlier, this may be natural to expect and not necessarily worrisome.)
Historical experience may suggest an upper bound on the potential magnitude and sustainability of net capital inflows, and their possible effects on world interest rates.74 Measured relative to GDP, Korea was perhaps the most extreme case of a net capital importer in the postwar period. Over the 27 years from 1953–80, Korea’s net capital inflows averaged 9 percent of GDP a year. Although the eventual outcome for Korea has been quite favorable, sustained capital inflows of this magnitude did lead to a severe balance of payments problem in the late 1970s, which required an adjustment effort supported by the Fund. Mexico’s net capital inflow averaged almost 7 percent of GDP over the period from 1991–94 and, as illustrated by recent events, also appears to have been in the danger zone. On the other hand, net capital inflows into Canada averaged 7 percent of GDP a year over the 43-year period from 1870 to 1913, and Australia has habitually recorded net capital inflows of between 1½ and 4 percent of GDP over most of the past century. These data suggest that in terms of broad orders of magnitude, a net capital inflow in the range of 4 to 6 percent of GDP may be sustainable under the right circumstances.75 Even so, those circumstances are rarely present. A key factor that influences the sustainability of large net inflows is the degree to which the external resources add to overall investment and are invested profitably. Experience also shows that the form of the capital flow—foreign direct investment, portfolio investment, or bank lending—might make some difference.
Box 12.Cross-Border Capital Flows and Capital Market Integration
Capital market integration is best thought of as the cross-border integration of markets for tradable financial securities, and to a lesser degree, as the integration of markets for physical capital (direct investment) and bank loans and deposits. Capital market integration means accessibility to foreign financial markets, where accessibility is measured by the direct and indirect costs associated with transacting in a foreign financial market relative to the costs of a similar transaction in the domestic market. Capital market efficiency has a number of meanings, many of which are viewed as synonymous with capital market integration, but the efficiency of financial markets is only part of the integration process. Capital flows provide a measure of the desire to invest capital abroad but are only indirect barometers of accessibility, efficiency, and thus of capital market integration.
The scale and growth of securities transactions between residents and nonresidents provide a good measure of accessibility to foreign capital markets. As a rough indicator, close to 50 percent of equity transactions for firms located in the European Union takes place outside the home country, and one out of every seven trades worldwide involves a foreigner as a counterparty.1 In addition, cross-border financial transactions in most industrial countries expanded from less than 10 percent of GDP in 1980 to well in excess of 100 percent of GDP in 1992 (see table). Issuers of securities have increasingly turned to international securities markets. The outstanding amount of international bond issues rose from $574 billion in 1985 to $2 trillion in June 1994. Over the same period, international bank loans tripled to $4 trillion.2
Secondary market turnover of fixed income securities through the international securities clearers, Euroclear and Cedel, has also accelerated in recent years. Turnover of $4 trillion in 1987 had increased to $21 trillion by 1993, an average annual growth of 31 percent.3 Turnover of equities in domestic markets by, or on the behalf of, foreign-based investors rose from $73 billion in 1979 to $1½ trillion in 1990.4 In comparison, turnover of equities and bonds on the New York Stock Exchange fell from $1.87 trillion in 1987 to $1.76 trillion in 1992, an average annual decrease of 1.4 percent.
Trading in the global foreign exchange (forex) market has accelerated in tandem with international securities transactions. Most estimates put current turnover at around $1 trillion a day in the global forex spot market, a 25-fold increase since 1980.5 The current level of activity in foreign exchange markets is much too large to be accounted for by the growth in world trade in goods and services. Rather, it reflects the tremendous increase in world trade in financial assets. Use of exchange-traded and over-the-counter derivative instruments also points to the integration of capital markets. The outstanding amount of currency swaps grew by five and a half times between 1988 and 1994, reaching over $1 trillion. The notional amounts outstanding of currency forwards, futures, and options displayed similar growth, reaching over $10 trillion in 1994.6
A driving force behind the growth in cross-border financial market activity has been the institutionalization of savings. The most important nonbank financial participants in the forex market are institutional investors. Total assets of the 300 largest U.S. institutional investors rose from 30 percent of U.S. GDP in 1975 ($535 billion) to more than 110 percent of GDP in 1993 ($7¼ trillion).7 Similar growth of total assets under management has occurred in other industrial countries. The importance of the institutionalization of savings for capital market integration is that it underpinned a shift toward international diversification of portfolios. This factor may be at least as important in the capital market integration process as financial deregulation, technological developments, and other “supply-side” factors. Cross-border equity holdings in the United States, Europe, and Japan increased from $800 billion in 1986 to $1¼ trillion in 1991, while total cross-border ownership of debt and equity holdings is estimated at $2½ trillion in 1991. Foreign ownership of government debt securities also points to the integration of capital markets (see table opposite).
1985 and 1991, respectively.
1985 and 1991, respectively.
Cross-border positions as a percent of portfolio assets differ markedly by country. For example, mutual funds in Germany and the United Kingdom had foreign asset holdings that were well over 30 percent of managed assets in 1991, whereas U.S. mutual funds’ foreign securities holdings were under 7 percent of managed assets. These cross-country differences in foreign security holdings may reflect the infancy of the capital market integration process. However, despite the relatively small fractions of foreign assets in institutional investors’ portfolios for some countries, the magnitude of the underlying assets translates into potentially large international portfolio holdings—the 100 largest institutional investors in Europe and in the United States had about $1 trillion of foreign investments in tradable securities in 1991. Based on recent trends, their holdings of foreign securities as a percent of gross assets may well double between 1991 and 1996. In addition, institutional investors have diversified significantly into emerging markets during the 1990s.
There are a variety of formal approaches to measuring the trend and level of the capital market integration process. One approach is simply to identify barriers to capital flows, including regulatory barriers such as capital controls and limits on foreign ownership, and also less tangible barriers such as language, information, and relevant cultural differences. But because barriers can often be avoided to varying degrees, a key problem with this approach is that it does not provide a measure of capital market integration that is easily quantifiable.
A second approach is to examine the extent to which returns on similar assets that are traded in different markets have been equalized, presumably by capital flows. Most of these “law-of-one-price” tests study the equality of nominal or real interest rates on short-term government bonds across countries. Tests for equality of nominal interest rates when currency risk is hedged in the forward market—covered interest parity—suggest that there has been a convergence of (covered) interest rates in the 1980s, especially for countries that have removed capital controls. Industrial countries are further along in the integration process than developing countries, based on departures from covered interest parity, although many developing countries exhibit a high degree of capital mobility as well.
A third approach is to compare actual investment portfolios with the highly internationally diversified portfolios that standard finance theory predicts in an integrated global capital market. The increase in cross-border transactions and holdings of foreign securities does suggest a trend toward integrated capital markets. The interesting question, however, is, How far have portfolios moved toward an “optimally diversified portfolio?” The answer, it turns out, is not very far.
Empirical studies indicate that investors in major industrial countries still have a strong “home-asset preference,” implying limited international diversification. U.S. investors, for example, hold about 94 percent of their equity holdings in the form of U.S. securities; in Japan, the United Kingdom, and Germany, the percentages exceed 85 percent; and the 300 largest pension funds in the world had only about 7 percent of their assets in foreign securities in 1991. In addition, domestic ownership on the five largest stock exchanges is very high: 92 percent in the United States, 96 percent in Japan, 79 percent in Germany, 89 percent in France, and 92 percent in the United Kingdom.8 Many reasons have been put forward to explain this home-asset preference—transactions costs, externally imposed prudential limits on foreign assets, uncertainties about expected returns, and unfamiliarity with foreign markets and tax laws—but the fact remains that the capital market integration process has a way to go.
Another approach to draw inferences from these co-movements about capital mobility and financial market integration is to study co-movements of certain variables either within a country (e.g., savings and investment) or across countries (e.g., consumption and stock indices). Although the hypothesis of high capital mobility is typically rejected, all of these tests rely on a number of strong assumptions that may color conclusions about capital market integration.
Despite the clearly impressive trends in the capital market integration process, we are a long way from a world capital market. One reason for this is that, as Morris Goldstein and Michael Mussa put it: “Nearly fifty years after Bretton Woods, less than one fifth of the IMF’s 178 member countries and territories voluntarily refrain from either restricting payments or using separate exchange rates for some or all capital account transactions.” Even for the industrial countries with few barriers to cross-border investment, the scale of international diversification is limited. Except for the wholesale markets for heavily traded, highly liquid, largely default-free financial assets, international integration of asset markets for the broader categories of world saving and wealth appears to be limited.1See Morris Goldstein and Michael Mussa, “The Integration of World Capital Markets,” IMF Working Paper 93/95 (December 1993).2See Joseph Bisignano, “The Internationalization of Financial Markets: Measurement, Benefits and Unexpected Interdependence,” Cahiers Economiques et Monétaires, Banque de France, Vol. 43 (1994), pp. 9–71; and Bank for International Settlements, International Banking and Financial Market Developments (Basle, various issues).3See Eurostat, “Ecu Statistics,” Monetary and Financial Statistics (Luxembourg: Statistical Office of the European Communities, April 1994).4See Philip Turner, “Capital Flows in the 1980s: A Survey of Major Trends,” BIS Economic Papers No. 30 (Basle: Bank for International Settlements, April 1991).5See Morris Goldstein and others, International Capital Markets: Part I. Exchange Rate Management and International Capital Flows, World Economic and Financial Surveys (IMF, April 1993).6Bank for International Settlements (Basle).7The figures on institutional investor behavior are taken from Institutional Investor (July 1994) and the sources cited in previous footnotes.8See Kenneth R. French and James M. Poterba, “Investor Diversification and International Equity Markets,” NBER Working Paper No. 3609 (Cambridge, Massachusetts: National Bureau of Economic Research, January 1991).
Of course, the other factor that determines the size of net capital flows is the supply of saving by capital exporters. Before World War I, the big net capital exporter was the United Kingdom. Over the 1870–1913 period, its net investment abroad averaged 4½ percent of GDP a year. In recent decades, Japan and Germany have been large saving exporters. But the biggest sustained net capital outflow in history as a percentage of GDP was arguably from Taiwan Province of China in the 1980s, which averaged 9½ percent of GDP. Today the net supply of capital from industrial countries that is available for developing countries is constrained because most capital goes to other industrial countries. Chart 32 shows which countries supplied net capital to the world (defined as average current surplus) over the five-year period from 1989–93, and which countries were the largest users of those resources. These data suggest that Japan was by far the world’s dominant supplier of net capital to the world over this period, and that the United States was the dominant net user. Indeed, the magnitude of the bilateral capital flow from Japan to the United States—about $80 billion a year—represented about a third of the total world net flows of capital over this period.76 Other heavy users of world capital have been the United Kingdom, Canada, Mexico, Spain, Italy, and Australia. The chief reason for such heavy capital flows to these and other industrial countries has been large government budget deficits. Over the 1989–93 period, net capital flows to developing countries averaged close to $100 billion a year—only about 1 percent of developing country GDP and only about 18 percent of the aggregate industrial country budget deficits over these same years.77
Chart 32.Largest Suppliers and Users of Net Capital Flows, 1989–93
With the discipline of financial markets limiting the effective demand for capital by developing countries, and with the large absorption of industrial country saving by other industrial countries limiting the supply, net capital flows to developing countries are likely to remain relatively modest. As a result, capital flows to developing countries should have only small effects on world interest rates. Much more important will be the future budget paths in the larger industrial countries. Although capital flows to developing countries increased substantially over the 1991–93 period and hit $161 billion in 1993, the flow slowed to $134 billion in 1994 and will slow even further in 1995. The surge likely was related to the sharp and prolonged industrial country recessions over this period.78 With sharply reduced industrial country investment demands and historically low interest rates, returns in developing countries looked very good and attracted saving from industrial countries. But with the industrial countries now showing good forward momentum again and the uptick in interest rates in the United States and some other industrial countries, flows to traditional industrial country markets have been picking up, and flows to developing countries have started to decline.79 This phenomenon is reducing capital flows to emerging market stock and bond markets.
In any case, it is far from obvious that developing countries need large amounts of foreign capital to develop successfully. While the benefits from large inflows of foreign direct investment can be substantial, it is important to note that the investment projects that attract foreign direct investment often tend to be financed to a large extent by complementary financing raised in the host country.80 In contrast, history suggests that even if large portfolio or debt-raising capital flows were available, as they were in the wake of the oil price increases of 1974 and 1979, for example, they probably would not be sustainable. Besides, the large capital inflows may increasingly substitute for national saving and might also cause the stock of foreign debt to increase relative to GDP.
Admittedly, the special capital requirements of the countries in transition may justify a relatively large-scale use of foreign saving. However, if the conditions and incentives are not in place to encourage national saving and investment, then economic performance will remain unsatisfactory, and that will bring into question the ability of these countries to service their foreign debt. As mentioned, when capital inflows exceed a range of 4 to 6 percent of GDP, even highly successful developing countries, such as Korea, have sometimes had trouble making principal and interest payments on their foreign debts. On the other hand, once the proper institutions, financial market instruments, and incentives to save have been established, domestic saving is likely to increase on its own because of the dynamics of the growth process, and should go a long way toward meeting domestic investment needs. In short, if developing and transition countries maintain a good overall policy mix, foreign saving can provide a boost to capital formation and can help to promote economic growth, but most investment spending is likely to be self-financed.
Global Real Interest Rate as an Indicator of Saving Adequacy
Economic theory and national accounting conventions require that saving has to equal investment, and one of the equilibrating mechanisms that brings this equality about is the interest rate. A high real interest rate suggests that investment demands are putting pressure on saving. A low real interest rate suggests an adequate supply of saving relative to investment demands. With current real interest rates high by the standards of the past one-third century, many feel there is at least prima facie evidence of a saving adequacy problem (Chart 33).
Chart 33.Global Short-Term and Long-Term I Interest Rates
1 GDP-weighted average of ten-year (or nearest maturity) government bond rates for the United States, Japan, Germany, the United Kingdom, Canada, Belgium, the Netherlands, and Switzerland minus long memory inflation estimate.
2 GDP-weighted average of three-month treasury bill note for same countries minus actual inflation.
Rather than using a spectrum of national interest rates to analyze trends, the increased integration of the world’s capital markets suggests using a single global interest rate. Such a global interest rate is of course synthetic—there is no world interest rate to be observed because each country issues its own bonds in its own currency. Instead, a global interest rate can be thought of as the underlying dominant factor that ties together the various observed national interest rates and implicitly equilibrates supply and demand in all markets simultaneously.81 There are several ways of constructing such a global interest rate. The proxy for the global interest rate used below is a simple weighted average of interest rates for eight countries that were deemed to have effectively open capital markets over the period 1960–93. A short-term (three-month) real interest rate was constructed using ex post inflation, and a long-term (usually ten-year) real government bond rate was constructed using a long-memory measure of inflation expectations.
Both short- and long-term global real interest rates were markedly higher in the 1980s and 1990s than in the 1960s and 1970s. In keeping with the breaks in saving trends noted earlier, three distinct regimes appear during this 35-year period: (1) a regime of low and fairly stable real interest rates between 1960 and 1972, (2) an oil-shock-adjustment regime between 1973 and 1980, and (3) a regime of high real interest rates from 1981 to the present. The global long-term real interest rate averaged about 3 percent in the first regime, about ½ of 1 percent during the second regime, and about 4¾ percent in the third regime. The global long-term real interest rate declined somewhat over the early 1990s, probably reflecting industrial country recessions. The rate seems to have returned to the range of 4 to 5 percent since 1994 and shows no statistical evidence of a regime shift since 1981. The global short-term real interest rate averaged about 1¼ percent during the 1960–72 regime, fell to about negative 1¼ percent during the turbulence of the oil-shock regime, but increased dramatically to average 4 percent in the 1981–93 period. Real short-term interest rates have diverged considerably across countries since the beginning of this decade, with some countries pursuing more expansionary monetary policies than others. Nevertheless, the global short-term real interest rate has been between 2 and 3 percent since 1991.
While the oil-shock period presents an analytical challenge because of the possibility that markets persistently underestimated inflation, a more relevant comparison is between the 1960–72 and 1981–93 periods. The evidence suggests that the global short-term real interest rate has been about 250 basis points higher in the latter period, and that the long-term real interest rate has been about 175 basis points higher. How can this increase in short- and long-term global real interest rates be explained? One obvious answer, that there was an exogenous upward shift in world investment demand, cannot be completely ruled out but seems unlikely. The ratio of ex post world investment to world GDP was lower in the 1981–93 period than it was in the 1960–72 period (see Chart 23). Instead, it appears that the high degree of public dissaving over the 1980s and 1990s has been a key factor. To quantify this and other effects, the staff constructed a simple model of the global short-term real interest rate that looks at three categories of determinants: economic environment factors, government budget policy, and monetary policy.82
Among the economic environment factors are a change in the world rate of return on productive capacity (perhaps due to a technological shift or some other innovation), financial market liberalization, and the removal or relaxation of capital controls (both internal and external). Increases in the world economy’s rate of return, measured via an aggregate stock market return measure, would tend to increase the world real interest rate. In the simplest terms, if stock markets start offering higher returns to investors, bond issuers will have to raise interest rates to attract funds.83 Financial liberalization, as discussed earlier, might also tend to increase the world real interest rate, although this relationship is complex and there would be economic forces working in both directions. A newfound availability to borrow, whether within countries or across countries, would tend to reduce the incentive to save, and as saving fell, the interest rate would rise. Financial liberalization also might transform some agents with notional demand for saving into effective demanders, leading to higher interest rates.84 These two factors, rate of return and financial liberalization, are probably interrelated. The move toward more liberal world capital markets may have contributed to an upward movement in the world rate of return, because making capital freer to flow to projects with the highest returns—wherever they exist—rather than keeping it captive in inefficient uses, should boost returns.
A rise in government budget deficits and increased levels of government debt are generally associated with higher real interest rates. There is an active literature about the complex relationship between government budget deficits and the response of private savers. As described earlier, however, the empirical evidence suggests that there is less than a one-for-one increase in private saving when governments dissave, so overall, national saving and world saving decline when governments run higher budget deficits. Over shorter time periods, deficits could be expected to help explain movements in real interest rates, whereas the debt-to-GDP ratio may be thought of as a long-run factor.
The relationship between money supply and real interest rates is not straightforward. To the extent that monetary authorities expand the money supply faster than expected, they could encourage higher long-term real interest rates by increasing perceived risk premiums beyond the normal “expected inflation” premium, which would be a component of the nominal interest rate, but not the real interest rate. On the other hand, over shorter time periods, it is possible to envision a negative relationship between money supply and a real short-term interest rate, if say, a countercyclical role for monetary policy is allowed.
Empirical estimates of the effects of these factors on the global real interest rate are not wholly conclusive, and the results are not always robust with respect to specification. Some researchers find large effects of budget deficits or government debt on interest rates, while others find effectively no effects.85 Studies that find relatively small interest rate movements often concentrate on U.S. (or other single-country) data, and probably understate deficit effects because capital moves to the United States so freely. The scale of international capital movements today clearly requires that analysis of the relationship between budget deficits and government debt and interest rates be done in an international context.
Econometric results based on a model that looks at the issue in a global context suggest that shifts in government debt can explain the bulk of the increase in real interest rates from the 1960–72 period to the 1981–93 period. Gross government debt averaged 45 percent of world GDP in the 1960–72 period, over 55 percent over 1981–93, and ended 1994 at over 70 percent. This shift seems to explain roughly 200 of the 250-basis-point increase in the real short-term interest rate between the two periods, or about 145 of the 175-basis-point increase in the real long-term interest rate. On average, each percentage point increase in the ratio of world government debt to GDP adds about 14 basis points to the global long-term real interest rate. Alternative specifications using the ratio of world government fiscal surplus to GDP found that each percentage point increase in government dissaving caused the short-term real interest rate to increase by 50 to 75 basis points. The increase in the rate of return on productive capacity is also statistically significant in explaining a share of the movement of the global real interest rates. The rate of return increased from 3½ percent a year in 1960–72 to 7¼ percent a year in the 1981–93 period, and this change explains about 20 basis points of increase in both short- and long-term real interest rates.
For purely illustrative purposes, it is of interest to explore what the optimistic and pessimistic saving scenarios sketched out earlier would suggest for global real interest rates over the next fifteen to twenty years. Using the implied relationship between saving rates and real interest rates developed in this section, global real interest rates could well decline by 100 to 200 basis points if the supply of world saving increased as described in the optimistic scenario. On the other hand, if the pessimistic scenario came to pass, the global real interest rate could easily increase by another 100 or 150 basis points. These measures obviously should be treated with caution given the uncertainties about the precise magnitude of effect of government saving on interest rates. However, the crucial point is that government fiscal positions unquestionably do matter for the level of real interest rates, and that the effects of significant changes in fiscal policy probably are important.
Box 13.Effects of Increased Government Debt: Illustrative Calculations
Net public debt in the industrial countries has increased from 20 percent of GDP in 1979 to 42 percent in 1994 (see chart). Gross debt has increased from 40 percent of GDP to 71 percent of GDP over the same period. Many analyses examining the implications of higher public debt have focused on a particular country and implicitly assume that the effects on the equilibrium world real interest rate will be insignificant. Although such an assumption may be appropriate when a country is small, the fact that so many countries together have run persistent government deficits suggests that the combined effects of the debt buildup on world real interest rates may have been large.
The long-term effects on real interest rates will depend on the extent to which consumers view government debt as wealth.1 If consumers are connected to all future generations and can borrow against their future income streams, changes in government debt will not crowd out private consumption and investment because consumers change their saving behavior today to prepare for tax liabilities in future. This invariance proposition is referred to as Ricardian equivalence, although David Ricardo himself did not believe that the economic consequences of deficits were unimportant. Indeed, Ricardo was concerned that if government expenditures were financed by issuing government debt, the private sector would underestimate their future tax liabilities and in such circumstances there would be an incentive to overconsume available resources.2 In an economy with scarce resources, an incentive to overconsume would result in higher real interest rates and a lower capital stock. Because a lower capital stock diminishes the level of output and, hence, the real income that the world economy can generate, an increase in government debt also reduces the sustainable level of consumption in the future.
|Output (in percent)||8.39||6.21||4.08||2.01||—||-1.94||-3.81|
|Consumption (in percent)||4.30||3.27||2.21||1.11||—||-1.12||-2.24|
|Capital stock (in percent)||25.87||18.78||12.10||5.85||—||-5.44||-10.50|
|Real interest rate (basis paints)||-203||-154||-104||-53||—||54||109|
The effects of government debt can be illustrated using an extended version of Blanchard’s model.3 The model has been used extensively to study fiscal policy issues, because it captures essential intertemporal aspects and has well-defined long-run properties. In order to develop some illustrative quantitative estimates of the effects of governmentdebt, the basic model has been extended in two ways. First, it is assumed that a significant proportion of the population, 40 percent, consume out of their current disposable income. This is consistent with time series evidence that attempts to measure the “excess sensitivity” of consumption to innovations in disposable income. The second extension is to incorporate more realistic income profiles, which initially rise with age and then decline as workers reach retirement, into the basic model. In line with the available evidence, it is also assumed that consumption is fairly insensitive to changes in the real interest rate.
Industrial Countries: Net Public Debt
Source: OECD, Economic Outlook data base.
1Excludes Australia, Austria, Greece, Ireland, New Zealand, Portugal, and Switzerland.
To establish a benchmark, a long-run equilibrium solution for the real interest rate was computed assuming that the current net public debt ratio of about 40 percent of GDP in the industrial countries would stabilize at this level. Several scenarios with higher and lower levels of government debt were considered to illustrate the potential long-run effects of government debt. At one extreme, the implications of eliminating all government debt in the OECD countries, and at the other extreme, the implications of allowing the government debt ratio to continue to drift up by an additional 20 percentage points, were examined. The latter would be similar to repeating the experience of the past two decades. The table reports illustrative estimates for real interest rates, output, consumption, and the capital stock for these two extreme cases and for intermediate cases.
These illustrative estimates suggest that eliminating all government debt in the industrial countries would reduce real interest rates by 200 basis points in the long run. The calculations imply that the buildup in government debt since the late 1970s has caused an increase in real interest rates of about 100 basis points. The effects of government debt on the cost of capital permanently lower both the capital stock and the sustainable level of consumption. In equilibrium, the higher real interest rate on financial assets is supported by a higher marginal product of capital owing to the lower capital stock. These simulations ignore other important channels that may result in even larger effects on the capital stock and real interest rates. For example, the model simulations assume that interest payments on the higher level of government debt can be financed with nondistortionary labor taxes. If the higher interest burden is financed through distortionary capital taxes or reduced infrastructure expenditures, there may be additional adverse effects on thecapital stock and real interest rates. Some recent econometric evidence suggests that the permanent effects of the debt buildup may be even larger than what is suggested by these simulations, although the empirical evidence is far from conclusive.1See Robert Barro, “Are Government Bonds Net Wealth?” Journal of Political Economy, Vol. 82 (November-December 1974), pp. 1095–117.2See Piero Sraffa, The Works and Correspondence of David Ricardo, Vol. 4, ed. by Piero Sraffa and M.H. Dobb (Cambridge: Cambridge University Press, 1951). For a discussion of why Ricardian equivalence does not hold, see Willem Buiter, “Death, Birth, Productivity Growth and Debt Neutrality,” Economic Journal, Vol. 98 (June 1988), pp. 279–93.3See Olivier Blanchard, “Debt, Deficits and Finite Horizons,” Journal of Political Economy, Vol. 93 (April 1985), pp. 223–47. The extensions are discussed in Hamid Faruquee, Douglas Laxton, and Steven Symansky, “Government Debt, Life-Cycle Income, and Liquidity Constraints: Beyond Approximate Ricardian Evidence” (mimeograph, IMF, 1995).
World Economic Performance Under Different Saving Scenarios
The long-term implications of increasing government debt or of a fundamental change in saving behavior are worth considering. As just described, the buildup of government debt over the last twenty-five years has important consequences for real interest rates. But it also has implications for the world capital stock and welfare over the long term, since a lower capital stock will result in a decline in the sustainable level of consumption. In order to illustrate the effects of government debt, the staff has examined saving issues with an extended version of Blanchard’s model. This model has been used extensively to study fiscal policy issues because it captures intertemporal aspects and gives rise to well-defined long-run properties. Consequently, this model provides a useful framework to examine the long-run effects of government debt.
One scenario for the world economy suggests that a continued rise in government debt in the industrial countries, equal to the increase over the last twenty-five years, would result in a permanent increase in the world real interest rate of at least 100 basis points (Box 13). This increase would lead to a permanent 12 percent reduction in the capital stock, which in turn would reduce the sustainable level of consumption by 2 percent—a reduction that the model sees as persisting forever. This would be a substantial reduction in world welfare—equivalent, say, to the total loss of a year’s output for each generation.
Another scenario considers the effects of a hypothetical reduction in world private saving. Since Japan, which accounts for about 15 percent of world saving, is widely viewed as the world’s supersaver, this scenario assumes that the private saving rate in Japan gradually falls by 10 percentage points.86 To put this shock in perspective, it would be similar to an assumption that the world saving rate fell by 1 percentage point. In the short and medium term, the shock would lead to an increase in aggregate demand in Japan (Chart 34). Higher consumption in Japan would cause real GDP to rise above a baseline level for about fifteen years. This higher level of output would not be sustainable, however, because a lower saving rate would cause an increase in world real interest rates and a reduction in the capital stock. Results are reported for a single representative country, the United States, but the impacts on the other industrial economies would be very similar. The shock would increase the world real interest rate by about 75 basis points in the long run. An increase in real interest rates of this magnitude would induce a permanent reduction in the capital stock of 6 percent and a permanent reduction in real GDP of about 1 percent in the United States and in most other countries.
Chart 34.The Effects of a Lower Saving Rate in Japan: A Simulation Exercise
Both of these potential developments, a sharply higher government debt-to-GDP ratio and a significant downward movement in private saving in one or more industrial countries, show the importance of saving for the world economy. Both would substantially cut the consumption path of all future generations and mean a sharp long-term welfare loss.
Conclusions and Policy Considerations
There is some evidence of a world saving adequacy problem. No single indicator alone proves this conclusion, but a variety of evidence points in this direction, including relatively high global real interest rates, relatively low world saving and investment rates, relatively high rates of return on productive capacity, and relatively low world real income growth, compared with pre-1973 experience. The period between 1960 and 1972 is arguably a relevant reference period—GDP growth was strong, the ratio of investment to GDP was high, productivity performance was excellent, and real income gains were healthy. Compared with that period, global real interest rates today are noticeably higher and the ratio of world investment to GDP is lower. Real returns on productive capacity, as measured by the world’s stock markets, were nearly twice as high in the 1981–93 period as they were in the 1960–72 period, suggesting more relative scarcity of capital. And productivity and real income growth have been noticeably lower over the 1981–93 period than they were in the 1960–72 period—at least in part because of the lower investment rate. The declining trend in the world saving rate draws much of the blame.
A dramatic decline in the public saving rate since about 1980 seems to be the root cause of the saving slowdown. Private saving has remained roughly steady—in the 20 percent range—over most of the past three decades. Public saving in the industrial countries by contrast has fallen sharply since the 1960s—averaging 4 percent in the 1960–72 period, but only ½ of 1 percent on average over the years from 1981 to 1993. Meanwhile gross public debt increased from 45 percent in the 1960–72 period, to 55 percent on average during the years between 1981 and 1993, and to over 70 percent by 1993. Econometric analysis suggests that this increase in government debt has been a key factor contributing to the rise in global real interest rates since the 1960s. The economic cost is substantial. An illustrative staff scenario suggests that the cost of this pattern of lower saving, higher interest rate, and lower investment is a permanent loss of roughly 2 percent of world consumption.
Where are saving and interest rates heading on present trends? Demographic swings—toward higher dependency ratios in the industrial countries, and toward lower dependency ratios in the developing countries—will probably roughly cancel each other out in terms of their impact on the world private saving rate over the next fifteen to twenty years. But two issues will be key. First is the rate of economic growth in developing countries, especially large countries such as China, India, and Brazil, and the extent to which they generate large quantities of saving. And second is the degree to which government dissaving in industrial countries is reversed. Global real interest rates could decline noticeably if developing country saving picked up strongly and if public saving rates in industrial countries returned to their levels of the 1960s and early 1970s. On the other hand, global real interest rates could easily increase further if negative outcomes occur instead.
Firm and committed policy actions are necessary to reverse current pressures on saving. The 1960s started out with a high ratio of world government debt to GDP. But as the decade progressed and as governments enjoyed strong economic growth, they used the opportunity to run fiscal surpluses, cut the ratio of government debt to GDP sharply, and saw the world saving rate increase steadily. That is because government budget deficits do matter for overall national and world saving. While there is some tendency for private savers to adjust their saving upward in the face of increased government dissaving, the offset is only partial. Although the causality runs both ways, it probably was no coincidence that the strong fiscal positions in the 1960s were associated with relatively affordable investment funds, a high ratio of investment to GDP, and good macroeconomic performance. With the industrial world now poised for solid growth again, there appear to be opportunities for a replay if governments choose to take advantage of them. There are growing dangers of not acting. With increasingly integrated world capital markets, the global real interest rate rises for everyone when individual governments run budget deficits—that is, the burden gets shared. This suggests the need for a cooperative effort to alleviate the problem—countries should set good examples for others to follow.
Who must act? World capital flows and financial conditions are largely determined by the industrial countries and the trend toward public dissaving is also heavily an industrial country problem. These are the countries where fiscal consolidation can help boost world saving the most. The case of the United States is noteworthy—after four consecutive years of healthy GDP growth, the federal budget deficit remains near 3 percent of GDP. Most other industrial countries are roughly in the same league as the United States in terms of fiscal deterioration compared with the 1960s or 1970s.87
Developing countries cannot be complacent. Many have enjoyed large inflows of capital over the past few years as the industrial countries posted poor economic growth. This has buffered developing countries from the growing world saving squeeze. But with the industrial countries now on track for more healthy economic growth, their investment demands are picking up again, reducing the availability of funds to developing countries. Increasing pressure on saving will reveal policy imbalances and structural economic weaknesses everywhere, and each country must put its own house in order. From some perspectives, for example, Mexico’s current situation is not just a simple crisis of confidence. Mexico has had a very low saving rate, and arguably, an unsustainable dependence on foreign saving.
The scope for microeconomic policies to encourage private saving, especially in industrial countries, appears to be more limited. Cross-country data suggest a natural evolution of the private saving rate, which increases sharply in the high-growth phase, levels off as an economy matures, and may decline somewhat as populations age. This means that as some countries mature and see their saving decline, other new industrial countries will be created, and should help to boost the world supply of saving. Some of the emerging countries may become net suppliers of saving to the world. Because private saving has such a strong internal dynamic, government policy can probably make mainly indirect contributions through structural reforms that promote economic growth. With private saving behavior only mildly sensitive to changes in the real interest rate, for example, the scope for tax policy to have a major impact on saving rates would seem to be limited. Of course, tax policy changes that would improve allocative efficiency should always be considered. In developing countries in particular, there is evidence that tax changes and structural reforms may be important in promoting allocative efficiency. Overall it seems clear, however, that fiscal responsibility is likely to offer most governments the biggest potential payoff for total saving.