The World Bank A u g u s t PREMnotes 1 9 9 9 n u m b e r 2 8 Economic Policy Saving--what do we know, and why do we care? A recent World Bank research project shows that saving has important interactions with income and growth--with resulting implications for policy. Saving rates vary widely around the world. higher growth and further increases in sav- EastAsiasavesmorethan30percentofgross ing (see below). nationaldisposableincome,whileSub-Saha- The social value of saving could also ran Africa saves less than 15 percent. More- exceed its private value because of imper- over,between1966and1994thesedifferences fections in global financial markets. A widened, with saving rates climbing steeply national saving rate broadly in line with inEastAsia,stagnatinginLatinAmerica,and an economy's investment rate reduces vul- declining in Africa (figure 1). Higher sav- nerability to sudden shifts in international ing rates tend to accompany higher income capital flows driven by uncontrollable herd growth--suggestingvirtuouscyclesofsaving behavior or self-fulfilling investor expec- and prosperity, along with poverty traps of tations. Still, as East Asia's recent crisis undersaving and stagnation. shows, high saving alone does not provide complete insurance against the conse- Why saving matters quences of weak financial systems or unsus- Should such disparities make saving a mat- tainable exchange rate policies. ter of policy concern? In principle there is little reason people and countries fac- Figure 1 East Asia has led the way in regional saving ing different shocks and income streams should strive for similar, optimal saving Percent rates. But in practice the intertemporal 40 choices that underlie saving are subject to 35 1974­84 externalities, market failures, and policy 30 distortions that can cause saving rates to 1966­73 1985­94 25 differ from welfare-maximizing levels. 20 Some market imperfections--a lack of 15 risk-sharing instruments, overly stringent mandatory saving schemes, Soviet-style 10 rationing--can lead to socially excessive sav- 5 ing. Other imperfections--insufficient gov- 0 ernment saving, the negative effect on China East Asia Latin Middle South Sub-Saharan Industrial retirement saving of an anticipated public and America East and Asia Africa countries Pacific and North Africa bailout of the elderly poor--can cause too Carribean little national saving. More generally, the Note: The figure shows gross national saving rates (including net current transfers) as a percentage social value of saving exceeds its private value of gross national disposable income. Regional medians are shown at current prices for 1966­94. if higher saving sets off a virtuous cycle of Source: World Saving Data Base. from the development economics vice presidency and poverty reduction and economic management network Although a large amount of literature has ity to save rises sharply only after income shed light on various aspects of consump- covers subsistence consumption. Income tion and saving behavior, many policy-rel- has a larger influence in developing than in evant questions remain largely unanswered. industrial countries, tapering off at medium A recently completed World Bank research and high income levels. In developing coun- project on "Saving across the World" tries it is estimated that, other things being addresses many of these questions: equal, per capita income has to double to ˇ Why do saving rates differ so much across raise the long-term private saving rate by countries and over time? 10 percentage points of disposable income. ˇ What is the relationship between saving But development also brings changes in Policies that and growth--and which way does the demographics and urbanization, some of causal link go? which tend to reduce saving. Thus the long- spur development ˇ Which policy measures have the largest term effect of income on saving may be more effect on national saving--and which modest than this estimate indicates. Still, are an indirect should not be expected to work? the overall implication is that policies that spur development are an indirect but effec- but effective What drives private saving? tive way to raise private saving. The research project yielded a rich set of way to raise empirical results on the behavior of pri- Growth vate (and national) saving rates over the past Muchoftherelationshipbetweengrowthand private saving 30 years. It also provided a comprehensive saving runs from the former to the latter-- data set on saving aggregates and related rather than the other way--particularly over variables (box 1). So what factors account short horizons. Empirically, a 1 percentage for private saving? point increase in the growth rate raises the private saving rate by a similar amount. Persistence Although part of this effect may be transitory, Private saving rates show inertia--that is, they sustained accelerations in growth are associ- are highly serially correlated even after con- ated with permanent hikes in saving. trolling for other factors. Thus the effects On the other hand, sustained increases of a change in a given saving determinant are in saving are often followed by accelerations fullyrealizedonlyafterseveralyears,withlong- in growth that persist for several years but termresponsesestimatedtobe2.3timeslarger eventually disappear. The result is a per- than short-term (less than a year) ones. manent rise in incomes rather than in growth rates--in line with the predictions Income of neoclassical growth theory. Private saving rates rise with real per capita income, supporting the view that the abil- Demographics Cross-country evidence confirms the pre- dictions of the life-cycle hypothesis, indi- Box 1 Measuring saving: The World Saving Data Base cating that a 3.5 percentage point increase in the youth dependency ratio reduces the Most empirical studies of aggregate con- Though it is far from perfect, the private saving rate by about 1 percentage sumption and saving use data with major new World Saving Data Base (available shortcomings--shortcomingsthatrelate at http://www.worldbank.org/research/ point. The negative impact of an increase to the use of incorrect definitions of the projects/savings/data.htm) introduces in the elderly dependency ratio is more than privateandpublicsectors,neglectofcap- major improvements over publicly avail- twice as large. ital gains on assets and liabilities, and the able databases. It represents the largest use of too narrow a concept of net worth and most systematic collection of annual Uncertainty (excludingitemssuchasaccumulationof time series on country saving and sav- Theory predicts that higher uncertainty consumer durables and human capital ing-related variables, spanning 1960­94 assets).Theselimitationscandistortmea- and 112 developing and 22 industrial should raise saving as risk-averse consumers sured saving considerably. countries. set aside resources to protect against pos- sible adverse changes in income and other PREMnote 28 factors. International evidence supports this Pension reform prediction. Inflation--conventionally taken Somedevelopingcountries,especiallyinLatin as a summary measure of macroeconomic America, are replacing pay as you go pension volatility--increases private saving, holding systems with fully funded schemes, a reform other factors constant. A similar logic often advocated for its favorable effect on sav- explains the fact that urbanization is nega- ing.Timeseriesevidenceconfirmsthatdevel- tively associated with saving. Rural incomes oping countries that increase the funding are more uncertain and, in the absence of in their public retirement programs tend to financial markets through which risks can achieve higher saving. Countries with pay as be diversified, rural residents tend to save yougosystemstendtohavelowersavingrates, a larger portion of their income, other things andthiseffectincreaseswithsystemcoverage. Tax incentives being equal. Nevertheless, the impact of pension reform on saving is not a given and in practice hinges have only small Private saving and public policy on how the transition deficit is financed and What implications do these findings have for on the reform's efficiency gains. positive effects policy? Four areas merit particular attention. Pension reform should have little short- term effect on private saving if it is financed on national saving Fiscal issues by issuing public debt--a move that sim- A permanent increase in public saving of ply entails converting an implicit govern- 4 percent of gross national disposable ment liability into an explicit one. But if the income is estimated to raise national sav- transition is financed by reducing the non- ing by about 3 percent of gross national pension public deficit (through reduced disposable income within a year and by about benefits to retirees and higher taxes on the 1.5 percent over the long term. Thus even active labor force), saving levels of future in the long term, increases in public sav- generations will increase, though not nec- ing are only partly offset by declines in pri- essarily their saving rate. vate saving. But there is considerable In the long term, pension reform can have variation in the degree to which private additional effects on saving through manda- saving offsets changes in public saving-- tory saving requirements. An example is Sin- from less than 30 percent in India to nearly gapore's Central Provident Fund, which 80 percent in Mexico. requires workers to contribute at least 25 per- The evidence also consistently shows that cent of their salaries. Such requirements may cutting spending is a more effective way to significantly raise the saving of low-income, increase saving than raising taxes. Never- borrowing-constrained workers, though the theless, while public saving may have the welfare implications of such a change are quickest and most direct positive effect on open to question. But pension reform can national saving, other policies that affect sav- also have other positive, indirect effects on ing indirectly--through faster income savingifitraisespercapitaincomeandgrowth growth--may have bigger and more last- by reducing labor market distortions and ing saving effects in the long term. inducing capital market development. The evidence on the effectiveness of tax incentives to increase saving is mixed and, Financial liberalization overall, not promising. The elasticity of sav- Financial liberalization involves a bundle of ing to net rates of return is usually found to measures, including liberalizing interest beloworevennegative.Evidencefromindus- rates, eliminating credit ceilings, easing entry trial countries on the effectiveness of tax by foreign financial institutions, developing incentives for voluntary retirement saving new capital market sectors, and enhanc- instruments is also mixed. Overall, tax incen- ing prudential regulation and supervision. tiveshaveonlysmallpositiveeffectsonnational Until recently a widely held view--not least saving,particularlywhentheirnegativeeffects in World Bank reports--was that financial on public saving are taken into account. liberalization should raise saving. August 1999 Analytically, the effect of financial liber- 2percentofgrossnationaldisposableincome alization on private saving rates can be intheexogenouscomponentofforeignlend- divided into a direct short-term effect, which ing reduces private (and national) saving by is generally negative, and an indirect long- about 1 percent of gross national disposable term effect, which is generally positive. The income over the long term. direct effect consists of price and quantity The evidence on the effect of foreign aid channels. The price channel works through on private saving is mixed, however. Evidence higher interest rates and, although popu- from Africa suggests that higher foreign aid larly advocated in operational documents tends to reduce national saving. But this may and the financial press, empirically is sel- overstate the case, because aid largely accrues dom effective in raising private saving--sug- to poor countries at times of adverse income Financial gesting that the negative income effect of shocks, when saving is lowest. Closer scrutiny higher interest rates tends to neutralize their of countries experiencing a lasting transition liberalization positive intertemporal substitution effect. from low to high saving rates indeed reveals The quantity channel works by expanding that increases in foreign aid are positively may increase thesupplyofcredittopreviouslycredit-con- associated with higher private and national strained private agents, allowing households saving. This finding could reflect a situa- private saving andsmallfirmstousecollateralmorewidely tion where countries undertake reforms that and reducing down payments on loans for both invite aid and induce higher investment rates in the housingandconsumerdurables.Theorypre- and growth, so that aid and saving rise dicts that this move should reduce private together. The conclusion is that aid need not long run savingbecauseindividualsareabletofinance invariably crowd out national saving. higher consumption at a given current income level. This prediction is well sup- Further reading ported by cross-country evidence, which indi- The papers produced by the research pro- cates that a 1 percentage point increase in ject are available at the "Saving across the the ratio of private credit flow to income World" website (http://www.worldbank.org/ reduces the long-term private saving rate research/projects/savings/policies.htm). See also: by as much as 0.7 percentage points. Schmidt-Hebbel, Klaus, and Luis Servén. Nevertheless, the indirect positive effect 1997. Saving across the World: Puzzles and of financial liberalization on saving should Policies. World Bank Discussion Paper 354. notbeunderemphasized.Liberalizingdomes- Washington, D.C. tic financial markets--particularly if done by ------, eds. 1999. The Economics of Saving. strengthening domestic banks--increases Cambridge: Cambridge University Press. the efficiency of financial intermediation and hence investment, contributing to higher This note was written by Luis Servén (Lead growth. Thus it is through faster income Specialist, PREM Unit, Latin America and growth that financial liberalization may Caribbean Region), Norman Loayza (Senior increase private saving rates in the long run. Economist, Central Bank of Chile), and Klaus Schmidt-Hebbel (Research Manager, Central External borrowing and foreign aid Bank of Chile). As with easier domestic credit, higher foreign If you are interested in similar topics, consider saving also tends to substitute for private (and joining the Growth, Investment, and Savings The- hence national) saving. Taking the interna- maticGroup.ContactMilanBrahmbhatt,x33960, tional evidence at face value, an increase of or click on Thematic Groups on PREMnet. This note series is intended to summarize good practice and key policy findings on PREM-related topics. PREMnotes are distributed widely to Bank staff and are also available on the PREM website (http://prem). If you are interested in writing a PREMnote, email your idea to Sarah Nedolast. For additional copies of this PREM- note please contact the PREM Advisory Service at x87736. Prepared for World Bank staff