Saving and the Real Interest Rate in Developing Countries

JONATHAN D. OSTRY AND CARMEN M. REINHART

to stimulate saving is unlikely to be success- rates. This feature of credit markets in the \AISING real interest ful in the poorest developing countries, least developed countries may itself make where consumption choices are heavily influ- saving less responsive to interest rates. rates has often been enced by subsistence considerations. Indeed, Another reason that might explain the cited as a way to the evidence suggests that the interest rate failure of existing studies to detect signifi- IN sensitivity of saving to real rates of return cant effects of interest rates on household I increase private saving, rises with a country's income level. Hence, saving in developing countries relates to the and thus provide the resources the same policies undertaken in different fact that—particularly in the poorest coun- countries could produce very different out- tries—subsistence considerations are likely for growth. But this may not be comes, depending on the country's level of to be a significant factor determining con- a viable approach in the development. sumption behavior. Given that, to be able to poorest developing countries, in save, households must first achieve a subsis- Is saving responsive? tence consumption level, the interest rate sen- which most people live at the There is little consensus in the empirical sitivity of private saving will be close to zero subsistence level. literature on the interaction between saving for countries where a large share of the popu- and the real rate of interest in developing lation lives at or near subsistence consump- countries. Some studies have concluded that, tion levels. Indeed, a subsistence model for a large number of developing countries, would predict a nonlinear relationship How does private saving respond to changes there does not appear to be any systematic between saving behavior and the level of in real interest rates in developing countries? relationship between rates of return and con- development, with the most significant The answer will influence judgments about sumption/saving behavior; others have sug- changes occurring when countries move from the effectiveness of a range of policies. For gested that there may be considerable low-income to low-middle-income status, and example, financial liberalizations, which gen- regional variation in this elasticity. relatively minor changes between middle- erate higher real interest rates, will result in One reason may simply be the poor qual- income and high-income countries. Of course, greater savings by households only if the lat- ity of the data in general and, more specifi- the interest rate sensitivity of saving could be ter decide to defer consumption; in other cally, the fact that there is considerable low even in middle-income countries if the words, if the sensitivity of consumption and variation in the economic significance and income distribution were sufficiently skewed saving to higher interest rates is significant. informational content of the data on real so that a large proportion of the population Similarly, when fiscal policy shocks con- rates of return. Lack of sophistication and lived at or near subsistence levels. tribute to movements in domestic interest depth in domestic financial markets or direct A frequently used proxy for the impor- rates, their impact on the external current regulation may result in interest rates that tance of subsistence in household budgets is account will depend on how responsive pri- do not adequately reflect expectations about the budget share going to food. It is indeed vate saving is to changes in real rates of the underlying economic fundamentals. In noteworthy that food consumption accounts return. many low-income developing countries, for a markedly lower share of total expendi- In this article, we present some evidence to there are few banks, and there is little scope ture in high-income than in low-income coun- suggest that raising domestic interest rates for true market determination of interest tries. As shown in Table 1, food consumption

Jonathan D. Ostry, Carmen M. Reinhart, a Canadian national, is a Senior Economist in the IMF's Southeast Asia and a US national, is an Economist in the IMF's Western Hemisphere Department. Pacific Department.

16 Finance & Development / December 1995

©International Monetary Fund. Not for Redistribution income developing countries, and the fact that tries than among the Latin American coun- Table 1 Food as a share of expenditure 1 the share of subsistence in total con- tries in the sample, despite similar income lev- (1990)' sumption—perhaps proxied by the food els. The more equitable income distribution share—declines as income increases. The sec- that characterizes the Asian countries may be Country groupings Percent ond prediction may also be related to subsis- a factor behind those differences. Institutional Low-income 55.6 tence considerations, since incentives to save arrangements (such as pension funds), which Lower-middle-income 32.1 over time should affect only that portion of the play a more prominent role in several Asian Upper-middle-income 30.5 budget left over after subsistence has been countries, may also be conducive to fostering High-income 13.0 achieved, that is, discretionary income. higher saving rates. Source: Judith Jones Putnam and Jane A. Allshouse, Among countries in various income groups, The role of real interest rates in saving Food Consumption, Prices and Expenditures, United the patterns of saving rates that emerge are behavior is more difficult to gauge. One prob- States Department of Agriculture, Statistical Bulletin No. 867. broadly consistent with the predictions of the lem—which is particularly important in 1 Expenditure on food as a percent of total personal subsistence model. As Table 2 shows, private Africa—is that financial markets remain thin expenditure. For details of the countries included in each country grouping, see World Bank, World saving is, on average, considerably lower for and governments often set interest rates at Development Report 1994: Infrastructure for the poorest developing countries, where the nonmarket levels. Nevertheless, there is some Development, Washington, DC, 1994. saving rate is about one half that of the high- evidence that financial saving increased as a 2 Data refer to averages for each country grouping. income group. In fact, such differences also result of the increase in real interest rates appear within regions. World Bank data show associated with liberalization of financial mar- that median gross domestic saving as a per- kets in developing countries, both in Africa accounts for an average of 13 percent of total centage of GDP (1987-91 average) was 5.6 per- and elsewhere. For example, increases in sav- expenditure in high-income countries and for cent for low-income African countries and 19 ing rates have been associated with higher only 8 percent of total consumption in the percent for middle-income African countries. real interest rates in Indonesia and the United States. For middle-income countries, Average gross domestic saving for both Republic of Korea and, more recently, in such as Mexico and Thailand, the share is groups was 7.7 percent of GDP. Argentina, Chile, and Pakistan. There is also often 30-40 percent while, for the poorest Furthermore, the relationship between the some evidence that reform programs in countries, the share of food is over 50 percent; level of income and the saving rate appears to Africa—which caused real interest rates to it is in the 60-70 percent range in some coun- be nonlinear, as the largest increases in saving move from sharply negative to mildly positive tries, such as Sierra Leone and Sudan. The rates occur in the transition from low- to levels—were successful in mobilizing domes- very sharp differences in food expenditure lower-middle-income levels, where the average tic savings. across countries with different per capita personal saving rate rises by 5.5 percentage income levels suggests that subsistence con- points. The average for the upper-middle- Empirical findings siderations may indeed be important in under- income countries is still 2.8 percentage points When households are assumed to maximize standing saving behavior, particularly in above that of the lower-middle-income group, utility, or welfare, subject to a resource con- low-income developing countries. but there appears to be relatively little differ- straint, the interest-rate sensitivity of house- There are, however, additional reasons why ence between the average saving rates in the hold saving depends on how easily saving may be less responsive to changes in high-income and upper-middle-income coun- households can substitute future consumption real interest rates in low-income than in mid- tries in the sample. for current consumption (technically known dle-income countries. Some have argued that It is noteworthy that liquidity constraints as the intertemporal elasticity of substitution low-income countries are characterized by and precautionary motives for saving could (IBS) in consumption). If a given change in pervasive liquidity constraints, which implies produce the opposite pattern in saving rates. real interest rates induces large shifts in con- that changes in consumption will be heavily If poor consumers cannot borrow but face sumption, then the IBS—one of the parame- influenced by changes in current income, an uncertain income stream, the demand ters describing household preferences—will rather than by changes in rates of return. for precautionary saving rises. Hence, the liquidity-con- Table 2 Savings and income straint/precautionary saving Income and personal saving rates Subsistence considerations offer two main hypothesis would predict that it predictions about saving behavior. First, sav- is the poorest consumers, who GNP per adult Personal saving as in 1985 dollars a percentage of GDP ing rates should increase with the level of real have no access to credit mar- Country groupings (1980-87 average) (1985-93 average) income at the initial stages of development, kets, who would save the most. with the largest increases in the saving rate Still, there appear to be Low-income 1,380 11.2 Lower-middle-income 2,806 16.7 occurring as a country moves from low- to marked cross-country differ- Upper-middle-income 4,896 19.5 middle-income levels. Second, saving should ences in saving rates that can- High-income 16,161 20.0 become more responsive to changes in real not be accounted for by a Sources: IMF, World Economic Outlook, October 1994, Washington; and interest rates as countries become richer. subsistence explanation. For World Bank, Adjustment Policy Research Report, Oxford University Press, The first prediction follows directly from the example, saving rates tend to be New York, 1994. role of subsistence consumption in the low- higher among the Asian coun-

Finance & Development / December 1995 17

©International Monetary Fund. Not for Redistribution be correspondingly large. In our work on Table 3 ings may shed some light on the wide this issue, macroeconomic data from a Interest sensitivity of saying under variation among countries in the response 1 sample of countries were used to evaluate alternative scenarios of the current account to fiscal policy the magnitude of the IBS for households changes that alter interest rates. from developing countries with diverse Initial real interest rate The failure of saving to respond to income levels. The low-income countries Country groupings 3 percent 4 percent 5 percent changes in real interest rates in many in the sample were Egypt, Ghana, India, low-income countries is particularly prob- Low-income Pakistan, and Sri Lanka. The low-middle- Average for group 0.312 0.306 0.300 lematic, since these are also the countries income countries were Colombia, Costa Average for 10 poorest 0.177 0.174 0.171 that have the least access to international Rica, Cote d'lvoire, Morocco, and the Lower-middle-income 0.532 0.522 0.512 capital markets and foreign savings. But Philippines. Three upper-middle-income Upper-middle-income 0.560 0.549 0.539 even for developing countries that can High-income 0.584 0.573 0.562 countries also were included in the analy- obtain foreign financing, achieving or sis: Brazil, Korea, and Mexico. Source: M. Ogaki, J.D. Ostry, and C.M. Reinhart, "Saving maintaining adequate domestic savings Behavior in Low- and Middle-Income Developing Countries: A In addition to not requiring that the Comparison," IMF Staff Papers (forthcoming). is essential for sustained growth. As the IBS be equal across countries with very 1 The data refer to the change (in percentage points) in the saving turbulence at the end of 1994 in Mexican rate owing to a 1 percentage point increase in the real interest rate. different per capita incomes (consistent For example, in high-income countries with a real interest rate of and other emerging financial markets with the subsistence model), our method- 3 percent, a 1 percentage point rise in the real interest rate would highlighted, foreign financing can be raise the saving rate by nearly two thirds of a percentage point ology allowed relative prices (of imports (0.584 of a percentage point). At higher baseline levels of the real volatile and reversals can be quite abrupt. and home goods) to enter into house- interest rate, the saving response diminishes slightly. Increasing national saving in lower- holds' saving decisions. To take one income countries may therefore require example, a temporary reduction in an alternative strategy. A recent World import prices should reduce saving, other moves up the income scale. Specifically, in Bank study of high-performing Asian things being equal, because the price of low-income countries, the IBS (and hence the economies highlights the role of lower public imports is a factor whose changes over time interest rate elasticity of private saving) is sector deficits (achieved through expenditure (like the changes in interest rates) affect incen- close to zero. In low-middle-income countries, cuts rather than tax increases) as an impor- tives to consume today or in the future. there is a marked rise in the IBS. The IBS is tant means of achieving higher national sav- Estimation of the IBS for countries with dif- found to increase again in the upper-middle- ing. Reducing the level and volatility of ferent per capita incomes is, however, an inter- income countries, though there is little differ- inflation, and promoting macroeconomic sta- mediate step. Ultimately, the aim of our ence between this group and the high-income bility more generally were also seen as useful analysis is to quantify the responsiveness of countries. ways of promoting saving. Mandatory saving saving to policies that alter domestic interest programs may also have been effective in rates, that is, the elasticity of saving with Implications boosting saving in some Asian countries. respect to the interest rate. It is possible to The main conclusion that emerges from our In contrast, in upper-middle-income devel- simulate how saving responds to changes in analysis is that much of the considerable vari- oping countries, the household saving rate is real interest rates using our empirical esti- ation among countries in both the level of sav- likely to increase significantly as interest rates mates of the IBS. By estimating how the IBS ing and the responsiveness of saving to the move up, and the response is unlikely to be varies with income level within our sample of real interest rate can be systematically very different from what would typically be 12 countries, we can infer values of the IBS explained by the country's income level. observed in industrial countries. outside the sample (using information on the Specifically, the hypothesis that the saving per capita income levels for other countries) rate and its sensitivity to interest rate changes and thereby simulate the interest rate elastic- are rising functions of income finds consider- ity of saving for a broad range of countries. able empirical support. In particular, our Our results (see Table 3) suggest that a 1 results may help to explain why the rising real per-centage point rise in the real rate of interest interest rates that typically accompany finan- should elicit a rise in the saving rate of only cial liberalization have failed to elicit an appre- about two tenths of 1 percentage point for the ciable rise in private saving in many countries. 10 poorest countries in the sample. For the Though financial liberalization policies—and wealthiest countries, by contrast, the rise in the resulting increases in interest rates—may have a number of positive effects (including References: the saving rate in response to a similar change J.D. Ostry and C.M. Reinhart, "Private Saving and in the real interest rate is about two thirds of a increasing both the efficiency of investment Terms of Trade Shocks," IMF Staff Papers, Vol. 39, percentage point. As Table 3 shows, the and economic growth), the results suggest September 1992, pp. 495-517; M. Ogaki, J.D. Ostry, results are not very sensitive to the initial level that the direct impact of such policies on and C.M. Reinhart, "Saving Behavior in Low- and of the real interest rate. household saving behavior is likely to be rela- Middle-Income Developing Countries: A In general, the cross-country variation in tively small in low-income countries. Other Comparison, "forthcoming in IMF Staff Papers; the responsiveness of saving to a change in policy questions—for example, the relation- G.L. Kaminsky and A. Pereira, "The Debt Crisis: the real interest rate is wide, consistent with ship between government deficits and the cur- Lessons of the 1980s for tin 1990s," forthcoming in the predictions of the subsistence consump- rent account of the balance of payments—will the Journal of Development Economics; World tion model and the wide variation in income also depend on the responsiveness of private Bank, Adjustment in Africa: Reforms, Results, and the Road Ahead, World Bank Policy Research levels in the sample of countries included in saving to real interest rates, to the extent that Report (New York: Oxford University Press, 1994); our study. As with saving rates, the relation- changes in public saving (increases or and World Bank, World Development Report 1994, ship between a country's income level and the decreases in the budget deficit or surplus) and World Development Indicators (New York: interest elasticity of saving varies as one alter domestic rates of return. Hence, our find- Oxford University Press, 1994).

18 Finance & Development /December 1995

Uganda: Adjustment with Growth, 1987-94, by Robert L. Sharer, Hema R. De Zoysa, and Calvin A. McDonald. No. 121. 1995. vii + 43 pp. This paper explores not only Uganda's recent adjustment efforts but also the country's prospects in the medium term. It provides an overview of recent economic performance with respect to growth, saving, and investment, and provides an analysis of Uganda's external adjustment efforts. The paper surveys fiscal adjustment and the prospects for a sustainable fiscal position, public enterprise reform, and army demobilization. Available in English, (paper) ISBN 1-55775-461-6 Stock #S121EA Capital Flows in the APEC Region, edited by Mohsin S. Khan and Carmen M. Reinhart. No. 122. 1995. viii + 76pp. The developing economies of the Asia Pacific Economic Cooperation Council (APEC) have been the recipients of a considerable volume of capital inflows in the 1990s. Given the increased integration of capital markets, it is not surprising that monetary control became more difficult for many developing APEC economies. Formulating an appropriate policy response has naturally been important. Each of the three papers that make up this Occasional Paper examine different aspects of these issues. Available in English, (paper) ISBN 1-55775-466-7 Stock #S122EA Comprehensive Tax Reform: Hie Colombian Experience, edited by Parthasarathi Shome. No. 123. 1995. viii + 67pp. This paper analyzes particular areas of tax policy that have concerned the Colombian authorities during the 1990s, albeit comprising a comprehensive approach to tax reform over time. It is intended to allow the reader to view in technical detail the type of analysis conducted in a representative tax reform study carried out by the IMF. Available in English, (paper) ISBN 1-55775-464-0 Stock #S123EA Saving Behavior and the Asset Price "Bubble" in Japan: Analytical Studies, edited by Ulrich Baumgartner and Guy Meredith. No. 124. 1995. vii + 79 pp. This volume brings together various analytical studies the IMF staff has undertaken on the Japanese economy, focusing on two areas of particular interest for both longer-term economic performance and recent cyclical developments. The first is Japan's saving behavior; the second is the remarkable swing in asset prices that occurred in the late 1980s and early 1990s. Available in English, (paper) ISBN 1-55775-462-4 Stock #S124EA I'nited Germany: The First Five Years: Performance and Policy Issues, by Robert Corker, Robert A. Feldman, Karl Habermeiei; Huri Vittas, and Tessa van der Willigen. No. 125. 1995. vii + 74 pp. This paper reviews economic and financial developments in Germany since its reunification nearly five years ago and analyzes some critical issues that have featured prominently in the policy debate over this period and are likely to continue attracting attention in the years ahead. Available in English, (paper) ISBN 1-55775-472-1 Stock #S125EA The Adoption of Indirect Instruments of Monetary Policy, by a Staff Team headed by William E. Alexander, Tomds J.T. Balino, and Charles Enoch. No. 126. 1995. viii + 67 pp. This paper examines the experience of implementing indirect instruments of monetary policy. The experiences of country studies illustrate the variety of circumstances under which indirect instruments of monetary policy have been introduced. Case studies are presented for Chile, Egypt, Ghana, Indonesia, Mexico, New Zealand, and Poland. Available in English, (paper) ISBN 1-55775-489-6 Stock #S126EA Road Maps of the Transition: The Baltics. Hie C/ecli Republic, Hungan. and Russia, by Biswajit Banerjee, Vincent Koen, Thomas Krenget; Mark S. Lutz, Michael Marrese, and Tapio O. Suavalainen. No. 127. 1995. viii+ 70 pp. Four papers analyze the process of transition to a market economy in the Baltics, the Czech Republic, Hungary, and Russia through early 1995. Available in English, (paper) ISBN 1-55775-519-1 Stock #S127EA Price: US$15.00 each. ($12.00 for full-time faculty members and students) TO ORDER, PLEASE WRITE OR CALL: International Monetary Fund Telephone (202) 623-7430 Publication Services Telefax: (202)623-7201 Box FD-504 Cable Address: INTERFUND 700 19th Street, N.W. INTERNET: PUBLICATIONS® IMF.ORG Washington, D.C. 20431 U.S.A. American Express, MasterCard, and VISA credit cards accepted. ©International Monetary Fund. Not for Redistribution