X . , Wi'S- 1'38,a

POLICY RESEARCH WORKING PAPER 1382 Public Disclosure Authorized

Saving, Investment, promote investment and Growth innowaton;and gro, h,is to in . . C'provirdea'supporv poicy'--'l Public Disclosure Authorized Developing Countries an d -,--,!,. h essentiaoc are An Overviewieni iidstbrrepicswl- ' - defined and wellenlbrced Klaus Schmidt-Hebbel ,prperl -rightsa Luis Senien n ta keeps Andrds Solimano busnn

Public Disclosure Authorized -politcalinsu% nt~totr~ -soalconsensusand politcal-- S~~~ , -;. ~ -f, .tbility ,Governmetshoul

~~~~~~~~~~~~~~~~. als pcicl&aEqu fnfrastrujctur#enduma . F -- -= -s 4f

asybl&cproj&ecsI6wdt,

h-lpheo terivat& ,e . - Public Disclosure Authorized

The World Bank PolicyReseach Deparmtent 'Macoeconomicsand Growth Division November 1994 IPOLICY RESEARCH WORKING PAPER 1382

Summaryfindings The 1990s have seen renewed interest in themes of disequilibria or both. And encouragingprivate saving economicgrowth and development. This is a welcome may be essential to expand investment,considering changeafter a decade and a half during which capital market imperfectionsand liquidityconstraints on macroeconomicswas dominated by a concern with short- firms and households in many devclopingeconomies. term adjustmentand stabilizationissues - and basic Four policy conclusions cmerge: problems of growth, capital accumulation,and the * Public savingdoes not crowd out private savingone- generation of savingswere largely ignored. to-one, so increasingpublic saving is an effectivedirect Schmidt-Hebbel,Serven, and Solimanodraw three way to raise nationa saving general lessons from recent literature on saving, * Foreign saving should be allowed and encouraged to investmnent,and growth: support domestic investment-even if it also helps Despite empiricalevidence about virtuous circles of finance consumption - as long as the macroeconomic heavy saving and investment and rapid growth, the and regulatory frameworkis adequate. relationship between the three is complex, with causality Higher private saving should not be expected in running in several directions. response to the liberalizationof interest ates. Market- * Still,saving often seems to follow, rather than determined interest rates will improve financial precede, investmentand growth, contrary to the Mill- internediation, the quality of portfolio choices, and the Marshall-Solowinterpretation quality of investment- but not necessarilythe volume - Investment and innovation are the centerpiecesof of savings. Pension reform may be a better way to growth. In this regard, the new literature on growth mobilizedomestic resources. represents a decided (if unintended) return to the * Potentially large externalities associatedwith tradition initiatedby Marx, Schumpeter,and Keynes. investmentwould seem to suggestthe need for an Saving may not be the chief driving force behind "activistf investmentpolicy. But a better way to promotc growth, but ensuring an adequatc savings level must investmentand growth is a supportive policy and remain a central policy concern - to ensure enough institutional environment, ensuring macroeconomic financing for capital accumulationand to prevent stability, social consensus,and a low cost of doing inflationarypressures or balance of payments business.

This paper - a product of tne Macroeconomicsand Growth Division, PolicyResearch Department-was presented at the EconomicGrowth and Long-TermDevelopment conference, held in El Escorial,Spain, July 11-13,1994. Copies of this paperareavailable freefrom theWorld Bank, 1818 HStreetNW, Washington,DC 20433. PleasecontactEmilyKhine, room N1-061, extension 37471 (48 pages). November 1994-

The Ply Rearch Woitn Papa Saks disemiate the findigs of zwk h prays to eacouage the change of ideas about deuclopmentssuAn objectie of the seie is toga the indinsgsot qukly, cvwfthe presenation are kss tbh fidly polshedt Thc paperscary the nanes of the autborsand should he usedand citcDcrdigly. The indgs interpretations.and connsv are the authors'ow and shouldnot be attributed to the World BPk its Eecuaiv Boardof Diors, orany of its meber couti

Produced by the PolicyResearch Dissemnination Center SAVING, INVESTMENT AND GROVVT IN DEVELOPING COUNTRIES: AN OVERVIEW

Klaus Schmidt-Hebber*,Luis Serv&n and Andrds Solimano'

World Bank. 'Inter-American DevelopmentBank.

This paper was presented at the conference 'Economic Growth and Long-Term Developmentwheld in El Escorial, July 11-13, 1994. We thank Peter Mantiel and Paulo Vieira da Cunha for detailed comments. We are also indebted to Felipe Laffain, Lance Taylor, and conferce participants for comments and discussions. 1. INTRODUCTION

The 1990s have seen a resurgence of interest in themes of economic growth and development.

This is certainly a welcome change from the last decade and a half, during which macroeconomicswas dominated by a concern-with short-term adjustmentand stabilizationissues, leaving largely aside the basic problems of growth, capital accumulationand saving generation.

Material progress and improving living standards require sustained output expansion. Even though questions of capital formation, technical progress and supportive saving have been at the core of economic analysis for two or three centuries, their links are far from being clearly understood still today.

Causes and effects among them are not unambiguous and several factors can start and support growth.

The transformation of an initial growth spurt into a sustained process of output expansion and prosperity requires accumulation of capital and its corresponding financing. This sets in motion a self-reinforcing process in which perceived prosperity invites investmnent,actual investment supports growth, and saving flows are generated along the way as income rises.

These links, however, can be weak. As the Latin American experience of the 1980s has shown the resumption of investment and growth after a period of adjustment and depressed economic activity is far from automatic. Coordinationfailures, pervasive uncertainties and irreversibilities in investrent, all make the recovery of growth a complex process. Moreover, the ability of governments tD boost saving is often limited, and scarce foreign financing can pose a binding constraint. Booms are not problem-free either, as they often generate unsustainablegrowth trajectories that lead to busts. These transitional problems are analytically exciting but constitute a nightmare for policy-makers. The bliss of steady growth still remains elusive. Historically, sustained growth paths have lasted at best a few decades, as in East Asia since the 1960s. By contrast, the high growth enjoyed-by Brazil and Mexico between 1940 and 1980 has been followed by persistently sluggish growth thereafter.

This paper provides a policy-orientedoverview of recent dteoretical and empirical work on saving and physical investment and their relationship with growth. Because the research and policy agenda in 2 these areas is quite broad, the paper focuses only on major issues and broad shifts in perspective

highlightedin the recent literature, leaving aside a number of potentially interesting topics. Nevertheless,

the discussion covers a variety of issues, such as the interaction between saving and investment, their

links with growth, the sensitivity of -saving to policy measures, or the implications of uncertainty and

instabilityfor capital accumulation-and growth. Needless to say, some of these questions - such as the

complexitiesof the link between-incomedistribution and capital accumulation, or the practical importance

of the extemalities associated with physical investment - are still far from being fully understood.

The paper is organized in seven sections. First, it reviews recent world and regional trends in

saving, investment and growth (section 2). The paper then re-examines the critical saving-investment-

growth links in the light of the recent theoretical and empirical literaure (section 3). Next, the discussion

turns to the behavior of saving, focusing on the links to fiscal policy, foreign saving, and financial and

tax incentives(section 4). The paper then examinesthe main features of investmentbehavior highlighting

new developments on the impact of irreversibility and uncertainty, political instability and institutional

factors, and the links between income distribution, profitability and investment (section 5). Next, the

discussion turns to the saving and investment response to the different phases of adjustment and refonra

programs (section 6). Finally, the paper's concluding section turns to the issue of appropriate policy

interventions to foster accumulation and growth (section 7).

2. FACTS AND PUZZLES: LONG-TERM REGIONAL TRENDS

Saving, investment, and growth have shown in most world regions a 30-year downwardtrend that

is both puzzling and worrisome (see Table 1). World saving and investment rates have declined by

Xhe data source for Tablel is the World Bank's BESD datbase, which in tnm is based an national accountsdata publishedby countrysources or by the IMP's InternationalFinancial Statistics. Regionalmeans for eachperiod are unweightedaverages of countrymeans for the correspondingpiod. Regionalstandarrl deviations for 1965-92are unweightedaverages of countrystandard deviations for 1965-92(or the availableperiod, if shorter ). Notethat the regionalavcnges of sbndard deviationsreflect the averageinstability over time that countriesfiac 3 some 3 percentage points of GDP during 1982-92 in comparison to 1974481. Annual per-capita world

GDP growth shows a marked long-run decline from 2.7% in 1965-73 to 2.0% in 1974-81 and to a bare

0.5% in 1982-92.2

These negative trends are more marked in LDCs than in OECD countries. And among developing regions, large disparities in performance are observed, with Africa and East Asian providing the two extreme cases of regional divergence. While in Africa gross national saving rates were more than halved between 1965-73and 1982-92 - falling to a record low 3% of GNP in 1990-92 - the East-Asian tigers' national saving rates rose from 20% of GNP to 32% during the same period - actually reaching

35% in 1990-92. With massive increases in foreign saving inflows (from 6.7% of GNP in 1965-73 to

15% in 1982-92), Africa's investment performance did not suffer much despite the reduction in national saving; while lower than in the 1970s, Africa's recent investmentrates exceed those achieved in 1965-73.

The opposite trend in the contribution of foreign saving to financing domestic investment is observed in the group of East Asian tigers, where foreign saving has declined to a tiny 1.5% in 1982-92- This, however, does not make East Asia's investmentperformance any less impressive than its saving record: at 33% of GDP in 1982-92 (and 36% in 1990-92), East Asia's investment rate exceeds by 12 percentage points of GDP the rates prevalent in both OECD countries and bther LDCs. It is not coincidental that

East-Asia's 1982-92per capita GDP growth, at 4.8% per year, -is ten times as high as the world's average

in any given region - and not the regional dispesion oF countr performance indicators. Sample sizes vary slightly by variables and yetrs within the 1965-1990 span, with the, world sample varying between 99 and 104 countries, and declining more strongly for 1991 and 1992 (when the world sample includes only between 87 and 102 countries). The main regions considered here are Sub-Saharan Africa. (Africa), Iatin America and the Caribbean (LAC), six East-Asian Countries (the 'tigei, comprised by Hong Kong, Indonesia, Korea, Malaysia, Singapore, and Thailand), Other DevelopingCountries (Middle-Eastem and other Asian countries), All LDCs, and OECD countries. Former socialist countries are excluded from the sample.

2 One should be aware of the limitations of regional means and standard deviations: variable definitions and data quality vary across countries, some countries are excluded fron each region, and the use of unweighted averages - while useful when treating each country experienceas one distinct observation - is not representative of individual country (and regional) economic weights and -is inconsistent with world adding-up constraints (for instance, that the world's foreign saving should be zero). 4 of 0.5%. At the other extreme, Africa experienced an average annual 0.4% decline in per capita GDP during 1982-92.

Latin America and the Caribbean's (LAC) investment and growth performance is disturbingly similar to Africa's - both in magnitude and time pattern. In contrast with Africa, however, LAC's investment is financed to a larger extent by national saving, which results in lower current account deficits.

Another important aspect of countries' performance is the stability of their saving, investment and growth patterns. The last column in Table 1 reports regional averages of country standard deviations for

1965-92, that reflect the instability that countries have faced in their performance indicators during the last three deiades.

Instability is much more acute in LDCs than in OECD countries - standard deviations of the five indicators are typically twice as large in the former than in the latter. While these measures of instability should be viewed with some caution,' their regional pattern does suggest that bad performance goes together with high instability. Africa is again the extreme case, with the largest standard deviations.

Higher instability in the poorer and worst-performiing regions may well reflect their larger exposure to adverse foreign shocks, the higher incidence of Various domestic shocks (ranging from wars to droughts), and their lower capability to cope with these shocks. This inference is supported by annual figures for regional performance (not presented here), that reveal the importance of worldwide shocks for decade-long performance trends. Saving, investment and growth suffered in all regions in the aftermath of the first (1973) and second (1980) oil shocks. But Esct Asia's performance suffered less and recovered more quickly from the adverse oil shocks than most other oil-importing developing countries.

3 Note that the standard deviation is larger when time trends or step-wise structural breaks are present - as actually is the case in many variables. Hence successful development - as reflected by uptrending saving, investment and growth like in East Asia - could raise our measure of instability. Nevertheless, we prefer to report straightforward standard deviations instead of standard deviations from time trends, as the latter would be based on necessarily arbitraiy trend estimates. 5

Table 1 World Saving, Investment, and Growth Rates by Major Regions and Sub-periods, 196S-1992 (Means and Standard Deviation)

1965-73 1974-81 1982-92 1965-1992 1965-92 Mean Mean Mean Mean St. Deviation

GROSS NATIONAL SAVING (% of GNP, including transfers)

AFRICA 10.7 8.3 4.8 7.7- 5.7 LAC 16.8 18.6 13.0 IS.8 6.0 EAST-ASIAN TIGERS 19.6 27.5 32.4 26.9 6.6 OTHER LDCS 15.8 20.5 16.5 17.4 5.2 ALL LDCS 14.1 15.0 11.4 13.3 7.1 OECD 25.7 23.5 21.2 23.3 3.1 WORLD 16.6 -16.8 13.3 15.4 6.3

GROSS DOMESTIC INVESTMENT (C of GDP)

AFRICA 16.9 21.8 19.2 19.2 6.2 LAC 20.3 23.5 19.3 20.3 4.8 EAST-ASIAN TIGERS 22.8 30.4 32.6 28.8 6.2 aTHER LDCS 18.7 23.4 21.9 21.3 4.6 ALL LDCS 18.7 23.2 20.7 20.8 5.5 OECD 25.6 24.7 21.6 23.8 . 3.2 WORLD 20.2 23.5 20.9 21.4 5.0

CURRENT ACCOUNT BALANCE (% of GNP, including transfers)

AFRICA 6.7 14.4 - 15.2 12.3 9.3 LAC 4.2 5.6 7.9 6.1 5.3 EAST-ASIAN TIGERS 3.9 3.5 1.5 2.8 5.S OTHER LDCS 2.7 3.1 5.8 4.0 5.8 ALL LDCS 5.0 8.9 10.3 8.2 7.3 OECD -0.1 1.2 1.1 0.8 3.3 WORLD 3.9 7.3 8.3 6.6 6.4

GDP GROWTH RATE (5)

AFRICA 4.3 3.5 2.4 3.3 5.8 LAC 5.5 3.1 1.9 3.4 4.5 EAST-ASIAN TIGERS 8.0 6.6 -6.2 6.9 3.3 OTHER LDCS 5.0 2.9 3.3 3.7 4.7 ALL LDCS 5.0 3.5 2.7 3.7 5.0 OECD 5.3 4.5 3.1 4.2 2.8 WORLD 5.1 3.7 2.8 3.8 4.6

PER CAPITA GDP GROWTH

AFRICA 1.2 1.1 -0.4 0.6 5.6 LAC 2.6 1.5 -0.7 1.1 4.6 EAST-ASIAN TIGERS 5.2- 5.0 4.8 4.8 3.1 OTHER LDCS 3.7 3.3 1.1 2.7 4.8 ALLLDCS 2.4 2.0 0.2 1.5 5.0 OECD -- 4.0 2.2 1.7 2.8 2.9 WORLD 2.7 2.0 0.5 1.8 4.6

Source: The World Bank (see fotnote 1). 6 Wedraw four conclusionsfrom thesefacts. First, all worldregions - includingOECD countries but with the remarkableexception of East Asia - show a disturbingdownward trend in their saving, investmentand growthperformance during the last three decades. Second,the regionaldiversity within

LDCsis growing:the poorestcountries and regions(Africa) are gettingeven poorer while middle.-income countries and regions (East Asia in particular)grow richer. Third, the poorer regions are not only a¶fectedby bad performancebut also by much higher degrees of instability. This suggeststhat better policiesand a matureinstitutional framework - encounteredmore often in East Asia than in Africaand

Latin America - are paramountto overcomespells of bad luck more rapidly and effectively. And, finally, the facts strongly suggest the existenceof vicious circles of low saving and investmnentand deterioratinggrowth (e.g., Africa), along with virtuouscircles of vigorous saving and investmentand rapid growth (e.g., East Asia). The nature of these links is the subject of the next section.

3. SAVING, INVESTMEN AND GROWER

Thetraditional wisdom of developmenttheory since at least WorldWar IHhas been that the long- run rate of economicgrowth is largelydependent on the saving rate: saving determinesthe financeable rate of capital accumulation, which in turn is the basic determinant of long-run growth.' Recent theoreticaland empiricalresearch has shed new light on, and also uncoveredsome puzzles, concerning this mechanism. Belowwe draw from this work to examinethree key questionsthat refer to each of the logicalbuilding blocks of the conventionalwisdom: first, what is the relationship(and, specifically,the directionof causality)between growth and saving ? Second, does countries' nationalsaving really get

4 This notion rests on what is termed the 'Marsha-Mill view' (adopted later by Solow as well; see Chaklavarty1993, chapter 3). For an alternativeinterpretation of post-World War EIgrowth, see Marglin and Schor (1990). 7 translated into domestic investment- or, more generally, what is the saving-investmentlink ? And third, what is the contribution of capital accumulation to growth - most importantly, is investment really the key to growth ?

3.1 - Saving-growth causality: which way?

The conventional wisdom would seem to be supported by the long-term patterns of saving and growth in developing countries described in the preceding section, which as noted earlier hint at the existence of a virtuous cycle between developmentand saving - as well as low-saving and poverty traps

- across countries and over time. Indeed, during the 1960s Asia and Sub-Saharan Africa had similar average gross domestic savingrates (12% of GDP); twenty years later Asian saving rates almostdoubled to reach 22% of GDP and African rates were halved to 6% of GDP. At the country level, Korean householdsaving rates started at levels close to zero in the mid-1960sand, after 20 years of high growth, rose above 20% of household income.

The apparent policy implicationwould be that raising domestic saving levels is of high priority to ensure a sustainable path of high growth. Yet recent research has shown that the relationship between private saving and income growth is a complex one. It is important to distinguish two aspects of the saving-incomelink: first, the saving response to incomefluctuations; second, its response to trend income or growth.

The saving-growth link over the cycle

Consider first income fluctuations. Overwhelming evidence for industrial and less systematic results for developingcountries show that both public and private saving are strongly pro-cyclical: most fiscal policies are counter-cyclicaland private consumers tend to smooth consumption spending over the business cycle. If households are forward-looking, not credit constrained and temporary income 8 fluctuationsdo not change permanentincome much, consumptionwould respond only marginallyto temporaryfluctuations. But households,particularly poorer ones in developingcountries or those below a certainthreshold income, tend to be credit-onstrainedand therefore respond strongly even to temporary

incomeshocks (Deaton 1989a,b). Campbelland Deaton(1989) also arguethat a household'sperception of its permanentincome is stronglyaffected by currentshocks, with no evidentdistinction made between currentand permanentincome flows. Bothpreceding arguments imply that householdswill consume out of current shocks,saving muchless than predictedby permanent-incomeand life-cycletheories, giving

rise to what in the U.S. literatureis termed'excess sensitivityof consumption"(Flavin 1981,Hall and

Mishkin1982). Few empiricalstudies of developingcountries have looked closely at the effectof income

fluctuationson saving. Exceptionsare Gupta (1987)and Schmidt-Hebbel,Webb and Corsetti(1992),

whichfind that savingresponds significantly and positivelyto temporaryincome shocks but by muchless

than whatis predictedby the permanent-incomehypothesis.

The longer term

Let us turn now to the relationbetween trend incomeand private saving, which is the more

relevantone from the viewpointof long-termgrowth. It is a bit ironicthat while the strongcorrelation

betweensaving and growth is a firmly establishedempirical fact, researchershave found it hard to

identifrthe preciselinks betweensaving and growth.

Mostcross-country empirical studies that includeinwme growthas a determinantof savingreport a significantpositive effect of growth rates on the saving rate. This result is obtainedfor exampleby

Modigliani(1990) in a large study based on combined cross-countrytime-series data, performed separatelyfor 21 OECDcountries and 85 developingeconomies. Jappelli and Pagano(1994) for OECD countries,and Edwards(1994) for a samnpleof both OECDand LDC economies,confirm this finding.

The same result arises in cross-countryempirical work for developingeconomies - such as those by 9 Collins(1991), Fry (1978, 1980),Giovannini (1983, 1985),Mason (1987, 1988),Mason et al. (1989), and Schmidt-Hebbel,Webb and Corsetti(1992).5

From the viewpointof textbookconsumption theory, this poses a puzzle: both the conventional infinitely-lived,representative-agent model and the overlapping-generationsmodel (wheregrowth takes place mostlywithin cohorts' lifetimes) would predict a negativeeffect of growthon saving, as individuals adjusttheir present consumptionupward in anticipationof higher future income. The exceptioncould be a life-cycleoverlapping-generations framework where growth takes placebetween cohorts (rather than withineach cohort's lifetime);in such conditions,growth increases aggregate saving for the simple reason that incomeof the activecohorts (and hence their saving) is larger than income(and dissaving)of the retired cohorts. However, Carroll and Summers(1991) show for three OECD countries, and Deaton

(1989b)for LDCs,that actualage-consumption profiles are not consistentwith what is predictedby life- cycle theories,undermining the empiricalimportance of this mechanism.

A numberof less-conventionalhypotheses on consumerbehavior have been advancedto explain the positivesaving-growth link - but have scarcely been tested (see Carroll and Weil 1993, for an overview). One is concentrationof gr-owthin householdswith high savingrates, such as rich or middle- aged households(Collins 1991)-- but its empiricalrelevance appears limited. A related, possiblymore relevantexplanation is that growthpushes consumers beyond the thresholdlevel of incomeunder which they are borrowing-constrainedor myopic. Slowly-changingconsumption habits could also contribute to higher saving rates in the face of rapid growth. Combinationsof these approachesmay also help explain the empiricalevidence; Carroll and Weil (1993) suggest a mixture of habit formatiob with uncertainincomes (giving rise to precautionarysaving) as a promisingavenue for further research. A

5 Interestingly,Carroll and Well (1993)show that this maybe partly due to the outlyingobservations from fast- growing, high-saving East-Asian countries: when theseare omitted from their sample, the positive saving-growth correlationbecomes much smaller and insignificant.However, Carroll and Weil's cross-countrysample might also be biasedby the exclusionof manylow-income countries. 10 final unconventionalhypothesis is that bothconsumption and wealth (or capital)enter the utilityfunction

- an idea advancedin differentways by "classical"economists (from Smithand Marx to Keynesand

Schumpeter,see Zou 1993b)and-which is resurfacingin somevery recentliterature (Cole, Mailath and

Postlewaite1992, Fershtmnan and Weiss 1993,Zou 1993a, 1993b):higher growth raises wealth, but - due to wealth/consumptionsubstitutability - increases consumption-less than proportionately, thereby raising saving.

Of course, an alternativeexplanation would just go back to the Marshall-Milltradition: saving is automaticallytranslated into capital ac-umulationand thereby incomegrowth, and this is simplythe mechanismunderlying the positivecorrelation between saving and growthobserved in practice. Yetsome empiricalevidence that this may not be the whole story is providedby Carroll and We'l (1993), who argue that savingtypicallyfoglows growth, rather than precedingit.

Fromthe growthviewpoint, the main conclusionis that causalitybetween saving and growthruns both ways. Thus, modelsthat fail to take such two-wayendogeneity into account likely overstate the contributionof savingto growth - and this even under the maintainedhypothesis that saving is fully translatedinto investment..We now turn to this issue.

3.2 - The saving-investment relationship

Understandingthe saving-investmentlink is importantfor at least two reasons: first, as just argued, it may hold the key to the positivecorrelation between saving and growth. Second, if capital accumulationis indeed the centerpieceof the growth engine, the interactionbetween saving and investmentis crucialfor assessingthe validityof the traditionalrecipe that raising saving is the surestway to increase growth -which involvesthe implicit-assumption that each country's extra saving.is necessarilytranslated into higherdomestic investment. 11 Textbookmacroeconomics emphasizes that the determinantsof saving are differentfrom those of investment. The former depends mainly on income and wealth, whereas the latter depends on profitabilityand risk.' Being the result of two independentdecisions, saving and investmentcan obviouslydiffer ex-ante, as emphasizedby the Keynesiantradition, in which discrepanciesbetween plannedsaving and investmentare at the .earEof macroeconomicfluctuations and growthcrises. Indeed,

Keyn,es'well-known 'paradox of thrift" - accordingto which an ex-anteincrease in savingmay lead via multiplierto an ex-postdeclinein real output,investment and savingitself- illustratesthatpolicies aimed at raising investmentand growth by encouragingsaving mightactually yield the oppositeresult.

Nevertheless,in the closedeconomy national saving and domesticinvestment must be identically equalat least in an ex-postsense, so that if saving effectivelyincreases investment must rise as well. But matters are more complicatedin the open economy, as capital flows introducea distinctionbetween nationalsaving and domesticinvestment. National saving need not be used to investdomestically; it can also be devotedto financeinvestnent abroad. Ideally, in a world of unrestrictedcapital mobility,each country'ssaving would flow to the most productiveuse in the world; thus, an increasein nationalsaving would be primarilyreflected in an improvementin the current accountbalance, rather than in higher domestic investmentand growth. And this mechaniismseems all the more relevant in view of the substantialdecline in barriers to internationalcapital flows (especiallyamong industrial countries) over the last two decades.

This reasoning,however, is in direct contradictionwith the empiricalevidence first reportedby

Feldsteinand Horioka (1980),and recendyupdated by Feldsteinand Bacchetta(1991), that in the long- run savingand investmentrates show a strongpositive correlation. On a sample of industrialcountries,

Feldsteinand Horioka find a correlationcoefficient close to 0.9 (virtuallythe same fbund in Feldstein

6 This,of course, was one of the fundanmtal contnbutionsof Keynes(1936). In the classical,pre-Keynes tradition, there was no independent investment function, and therefore investmentjust followed saving passively. 12 and Bacchetta's' update). Other studies (Dooley, Frankel and Mathieson 1987; Summers 1988) find a

-similarly strong correlation for LDCs, although somewhat lower in magnitude than for industrial countries-

Low capital mobilit or omitted factors?

*Whether' this result is evidence .of international capital immobility has been the subject of considerable debate. One view (defended by Feldstein and his associates) holds that international capital mobility is indeed far from perfect, and therefore national saving increases crowd-in domestic investmnent

(albeit dtrugh unspecified mechanisms). The alternative view is that the observed saving-investment correlation says little about international capital mobility7 and is mostly due to policy reactions and/or common factors that cause both saving and investment to move together in the longer term.

Along these lines, a wide variety of mechanisms have been proposed that could give rise to a strong saving-investment correlation even in the presence of high capital mobility (Obstfeld (1994) provides an extensive review; see also Dornbusch (1991a) and Blecker (1994)). For example, Frankel

(1993) argues that, even under perfect capital mobility, shifts in saving alter the real interest rate, which in turn tends to move investment in the same direction as saving; however, this would imply that real interest rates differ across countries even in the long run- or, in other words, with uncovered interest parity real exchange rates would have to display long-run trends. A second explanation underscores the

' The validity of saving-investment correlations as a measure of intemational capital mobility has been -extensivelydiscussed by Frankel (1992, 1993),who examines ree alternativemeasures: covered interest parity, uncovered interestparity, and real interest rate equalization. Frankel (1992) argues that the first of these is the most adequate,as it compensatesfor courtty risk he reportsa significantdecline (nearly a completedisapp ) in coveredinterest differentials among industrial countries in the 1980s, which implies that capital mobility increased steadily over the same period. In turn, Montiel (1993) provides a comprehensive analysis of capital mobility in LDCs. More generally, the finding reported in the text that saving-investment correlations are typically lower for 'suples of LDCs than for industrial countries - in spite of the extensive capital controls and external borrowing ntras in many developing economiesduring the 19 70s and 1980s- raises strong suspicions about their validity as measures of capital mobility. 13 role of slowly-changingdemographic and technologicalfactors affectingboth saving and investmentin the samedirection (Obstfeld 1986, Tesar 1991). A third alternativeattributes the saving-investmentlink to the operationof the economy'slong-run budget constraint (Obstfeld, 1986): in the long-mn, if the economyis closeto a stationaryforeignrasset/GDP ratio, it is possibleto showthat savingand investment

ratioscannot show much divergence.

An alternativeview shifts the focus from internaional to domesic capital immobility,by underscoringthe close linkbetween corporate investment (which in industrialcountries accounts for the bulk of private investment)and retained earningsempirically found in most OECD economies(and highlightedby Feldsteinand Horioka themselves). Indeed, in industrialcountries retained earnings typicallyrepresent the major source of corporateinvestment financing; the strong correlationbetween

both variables(documented by Murphy(1984) fbr U.S. firms, and by Blecker(1994) for a sampleof

OECD countries)could be the key to the aggregate saving-investmaentcorrelation.' This type of

mechanismseems particularly relevant for LDCs, in whichcapital market imperfections are widespread and borrowingconstraints are the norm, notonly for organizedcorporations but especiallyfor households and firms in the informalsector, whichin manydeveloping economies account for the majorityof private

investnent.

Finally,another leading explanation (first proposed by Summers1988) is basedon the existence of restrictionson countries' current accountimbalances, due either to lendingconstraints imposed by world capitalmarkets on deficitcountries, or to systematic(and successful)current accounttgeting by policy-makers. In both cases, the result would be a strong ex-post correlationbetween saving 3nd

investment.In particular,under the extremesituation of a perfectlyinelastic supply of fioeign borrowing

However,as Obstfeld(1994) notes, this requiresnot onlythat fims financetheir investment with coprae earnings,but also that householdsfail to pirce the -"corponteveil, thus ailingto offseta rise in firm.' sving with a declinein householdsaving, and tharebyallowing a risein corporateinvestment to be matchedwith a rime m aggregatesaving. 14

-as faced by many LDCs during the 1980s-- national saving and domestic investmentwould be highly

Cindeedperfectly) correlated. Moreover, in such conditionsthe supply of foreign saving would play a causal role in affecting domestic saving and investment - in stark contrast with the textbook-case of perfect capital mobility in which foreign saving is determined residually by the excess of domestic investment over national saving, and any extra gross capital inflows are completely offset by gross outflows. Along these lines, Argimon and Roldan (1994) find empirically that in EC countries enfbrcing capital controls (presumably to target the external balance), saving and investment display virtually unit correlation, and causality runs from the former to the latter.

To sum up the discussion, the saving-investmentpuzzle remains far from resolved. From the policy viewpoint, the key conclusion is perhaps that underlined by Dombusch (1991a): unless the empirical saving-investmentlink is better understood, its existence does not justify any strong inferences about the investment and growth response to saving policies.

3.3 - The investment-growth link

Growth and development theory have long regarded the accumulationof physical capital as the engine of growth. Indeed, the notion that raising the investmentrate is key to increasing long-rungrowth has been at the heart of growth thinking since the times of David Ricardo.

The strong associationbetween investment ratios and long-term growth performance is a well- establishedempirical fact (see e.g. Kuznets, 1973). Indeed, most counitryexperiences of sustainedgrowth tend to stress the link between capital accumulation and GDP growth. The case of East Asia, the most successful regional experience in tenms of rapid and sustained growth of the last three decades, provides a good example- As documented in the preceding section, the East-Asian economies have been able to maintain rates of GDP expansion on the order of 7 to 8 percent per year, supported by rates of capital 15 formationaround 30 percentof GDP. Clearly,high growthand high investmentshares have gone hand-

in-hand.

However,the key-roleof investmentin the growthprocess was challenged in the 1960sand 1970s by neoclassicalgrowth theorists. In the neoclassicalmodel (Solow, 1956),capital accumulationaffects growthonly during the transitionto the steadystate; by contrast, long-rungrowth is determinedonly by populationgrowth and the rate of technicalchange, which was assumedexogenous. This viewattracted

considerablecriticism from a numberof authors(e.g., Kaldor 1957,Robinson 1962) on the groundsthat

the separationbetween investment and innovation(or technicalchange) was artificial,as mosttechnical

innovationtends to be embodiedin new machineryand equipment.9 Growth-accountingexercises based on the neoclassicalmodel (Solow 1957; Denison 1962, 1967)appeared to confirmthat cross-country differencesin investmentratios could explainonly a limited portion of the differencesin per capita growth performanceover long periods, suggestinga crucial role for technologicalchange (or, more honesdy, unidentifiedresidual factors) as a majorsource of long-rungrowth.

The arithmeticof the Solowmodel, however,does not square well with the strong correlation betweeninvestment ratios and growth performanceobserved in practice(see Romer, .1987). Recent

researchaddressing this issuehas broughtcapital accumulation back to centerstage of the growthprocess,

suggestingan enhanced - albeit more indirect - role for investment as a key growth determinant. One

lineof researchfocuses on the complementaritiesbetween investment in physicaland humancapital: new

and technologicallyadvanced machines and equipmentneed to be operated by workers with adequate skills and education. Likewise, the identificationand design of profitableand innovativeinvestmente projectsrequires also the existenceof an entrepreneurialclass with innovativeskills and awarenessof businessopportunities. Alongthese lines, Mankiw,Romer and Well (1992)extend the Solowmodel to

-See alss Schumpeter,1934, who emphasizedthe factthat most innovationswere embeddedin the production of new capitalgoods and businesspmctices. t16 include human capital and, under the assumptionthat its accumulationis guided by that of physical capital, find that investmentperformance can accountdirectly and indirectlyCi.e., through the parallel accumulationof humancapital) for the bulk of the variationin growthperformance across countries.

A secondline of research,that has featuredprominently in the "new" growth literature (e.g.,

Romer 1986,1987, among many others) emphasizes the close linksbetween the accumulationof physical capital and technologicalchange (a point that had been underscoredalready by Allyn Young back in

1928, and also by Kaldor-1957 and Arrow 1962). If productivitygrowth is endogenousrather than exogenous,and relatedto the accumulationof physical(or human)capital, then an increasein the rate of investmentagain raises the rate of growth in the steadystate.

The strongcorrelation between technical progress and investment(or the capitalstock) has been amply documented. Baumolel. at. (1989), for example, show that in industrialcountries technical change is highly correlated with capital/laborratios. De, Long and Summers(1993) estimatepanel regressionsof total factor productivitygrowth (1FP) for a large sample of developingcountries.

Consistently,their results show a positiveand statisticallysignificant correlation between the ratio of equipmentinvestment to GDP and total factorproductivity growth; they also find a negativecoefficient for structuresinvestment in the regressions.

If technicalprogress is drivenby investment,then capitalaccumulation by any one firm benefits other firms in the economy,creating an externalitythat opens a gap betweenprivate and social returns to investment.In fact, De Long and Summers(1993) calculate a net socialrate of returnon equipment investmentthat is about 35 percentper year: 10 percentis the privately-appropriatedreturn on capital and 20 percent correspondsto the external effects.induced by the correlation between equipment investmentand TFP growth. Thesefigures, as the authorsrecognize, are probablyon the high side.but still provide an order of magnitudeof the possibledivergence between private and social rturns to 17 investment.Earlier work by Romer(1987) likewise estimated that the socialmarginal product of capital could be over twice as high as the privatemarginal return.

Fromcorrelation to causality

The stronginvestinent-growth correlation referred to abovehas been extensivelycorroborated by a numberof econometricstudies. Levineand Renelt(1992) conducted extensive tests on the robustess of alternativespecifications of the growth process. In particular,they lookedfor the specificationleast sensitiveto changesin additionalexplanatory variables, the sampleof countriesand the choice of time periods. Their main result is that the only robust regressor in the different growth equations- both across countriesand over time - is the ratio of physicalinvestment to GDP. Other variables,such as indicesof externalopenness and price distortions,and indicatorsof fiscal and monetarypolicy stance, all are ultimatelyfragile as regressors. Moreover, this general result also holds in other empirical

'analysesof growthconducted for Latin Americaand East Asia.t -

DeLongandSummers(1991, 1993)haveadvancedonestep furtherbydisaggregatinginvestment into its structures and equipmentcomponents. They present abundantempirical evidence,for both developingand developedeconomies, of a crucialnexus between investment in machineryand equipment and the rate of GDP growth. Estimatingpanel growth regressions for a sampleof 88 non-oilexporting countriesfor the period 1960485,DeLong and Summersfound a muchhigher contributionof machinery and equipmentinvestment than structures investmentto GDP growth. The result also holds for sub- samplesarranged according to differentincome levels and continents.

A relatedline of work is that of Easterlyand Rebelo(1993), who explore the relationshipwith growth of differenttypes of public investment. They gathered data on aggregateand sectoral public

'e See De Gregorio(1992), Corbo and Rojas (1993), Lefort and Solimano(1993) for Latin America and Young (1994)for East Asia- 18 investment for nearly 100 countries. Subject to some reservations about the quality of their information, the authors find that central govermmentinvestment is positively correlated with both growth and private investment, while investmentby public enterprises (which presumably competes more closely with private investment projects) is negatively correlated with growth. Sectorally, they also find that the strongest growth impact corresponds to public investmentin infrastructure (transport and communications). These findings agree well with analytical and empirical results reported by Serven and Solimano (1993) on the complementarity between public and private investment. Public investment in ports, roads, telecommunicationsand the like, provides the basic infrastructure needed for private investmentprojects to operate profitably; thus, such types of public projects are likely to have a maximum impact on growth.

Empirical studies of private investment for Latin America (Serven and Solimano, Chapter 7) and East

Asia (Chapter 8) show a positive and significant correlation between public and private investment.

While all these empiricalresults underscore potentiallyimportant aspects of the investment-growth link, in principle they say little about the direction of causation. And, like with saving- it is necessary to distinguish.between the cycle and the long term. In the short run, -investmenthas been empirically shown to depend on the rate of output growth and/or the rate of capacity utilization, as. indicators of future demand pressure and/or the severity of liquidity constraints faced by firms - two key variables in their decision to expand productive capacity. Thus, over the business cycle investment can be led by output in an accelerator-like fashion (see Serven and Solimano 1993, Chapter 2 for further discussion and references).'"

In the long term, the traditional,wisdom of development economics has long been that capital accumulation is a fundamental cause of the growth process. Recently, however, this view has been

It is worth noting that this sensitivity of investment to cyclical variations in output. (or other short-term. factors)opens the possibilitythat short-termrecessions could have long-term effects, by causinga deepinvestment slump that leaves the economy permanently trapped in a low-growth, low-investmentequilibrium. In other words, thegrowth process could be path-dependeat. SceeBruno(1994) for further discussion and somecounty experiences. 19 challengedby several authors(e.g., Benhabiband Jovanovic1991, King and Levine 1994)who argue that the co-movementof investmentratios and growthrates may be largelydue to the action of a third factor- technologicalinnovation - drivingboth capital accumulation and outputexpansion. An extreme interpretationof this viewwould hold that capitalaccumulation is just a consequence,rather than a cause, of the growthprocess. Alongthese lines, empiricalevidence recently provided by Blomstrom,Lipsey and Zejan (1993)using a sampleof 100 countriesshows that growthGranger-causes investment, but not conversely.

is investment sufficient for growth?

While the extremeview just describedseems rather implausible,the lessonto be drawn is that the investment-growthrelationship is more complexthan suggested by the conventionalwisdom. Indeed, the simplistic(albeit popular) view that capital accumulation is the onlypre-requisite for economicgrowth

- a notionlabelled 'capital fiindamentalism"by Yotopoulosand Nugent(1976) - is untenableas well.

As the recentliteraturehas emphasized, other complementary inputs (notably, human capital and technical knowledge)are also paramountfor growth.

Buteven if these other inputsare available,the quality i.e., productivity)of investmentis a key determinantof its ultimate reward in terms of growth. Brute-forcecapital accumulationin mindless pursuitof scale economiesin a frameworkcharacterized by severedistortions is unlikelyto have much of a growthpayoff. Indeed,under extremedistortions massive capital accumulation can ultimately result in reducedgrowth when measured at shadow(i.e., distortion-free)prices. Examplesof this situationare not difficultto find amongcountries with heavilyregulated investment regimes and/or extensive barriers to marketforces; an extreme case is that of the former SovietUnion (see Easterlyand Fischer, 1994).

Thus, an incentiveframework free of major distortionscan be instrumentalto realize the full growth potential of capital accumulation. 20 However,the concldsionthat investmentmay not be the sole engineof growthdoes not alter the fact that capitalaccumulation remains a centerpieceofthat engine. In general, it is hardto find countries that have been able to grow at high and sustainedrates for long time periodswithout an importanteffort of capital formation - a fact noted long ago by economic historians such as Rostow (1958) or

Gerschenkron(1962). In fact, one of the mainmerits of the recent literaturelies perhaps in the renewed focus on technologicalchange and externaleffects associated with physical investnent as a major aspect of the growth engine- filling what Romer(1993) labels the "idea gap", as opposedto the 'tobjectgap" that can be filled just by.accumulation of material inputs. As noted earlier, this notion that technical progressis largelyembodied in new capitalgoods is hardlynew in growththeory. From this perspective, the recent literature'semphasis on the majorrole of productivitygains in long-tenngrowth contributes to enhance, rather than lessen,.the importanceof capital accumulationfor the growth process.

Nevertheless,it is importantto note tat the experienceof East Asia over the last two decades suggests.thatthe'recent literature's emphasis on "ideas' may have been overstated:as Young(1993) has convincinglyshown, East Asia's stellar growth-performancecan be fullyexplained by its unusuallyhigh investnent ratios, and does not arise from any extraordinaryproductivity gains. 12 Indeed, Young estimates.that over the last two decadesproductivity growth in East Asian countriesproceeded at a rate cdoseto.the world average. This result is corroboratedby DeLong and Summers(1993), in whose frameworkEast Asia's rapid growth is likewiseunsurprising given its rapid pace of investmentin machinery and equipment.

32 It is importnt to recall,however, that EastAsia's high rates of physicalinvestment were also accompanied by an importanteffort at hunmncapitalformation.: 21 4. SAVING

Regardless of whether saving is the chief force driving growth, ensuring adequate levels of saving

remains a central policy concern in order to provide sufficient financing for investment and to avoid

balance-of-paymentsdisequilibria. Yet saving perhaps is the macroeconomic aggregate that is likely to be the most elusive to both the economist's understanding and the policymaker's influence. Whatever

saving definition is used,'3 it shows massive dispeLsionthroughout the world and over time. For

instance, it is puzzling why two similar experiences of stabilization and reform in Latin America.have

implied such diverse saving response: in Chile gross domestic saving rates grew from 17.0% of GDP in

1974-87 to 27.0 % in 1988-93, while in Mexico they declined from 26.3% of GDP in 1980-87to 20.2%

in 1988-92. For the U.S. many different explanations are offered for its secular reduction in net national

saving rates from 8% in 1950-1979 to 4.5% in the 1980s and to less than 2% sinue 1990, including

budget deficits, population ageing, and lower growth.

Having discussed the relations between saving and growth above, we fos next on tiree

unresolved questions on the behavior of saving that are important for both research and policy design in

developing countries.' They refer to the influence of public saving on national saving, the influence

of foreign resource inflows on domestic saving, and the effectiveness of financial and tax incentives in

raising private saving.

'3 Empiricil work on saving is hauntedby the low qualityof savingdata and the inadequacyof saving definitions.Aggregate saving data are oftenreported in nationalaccounts statistics as a residualof anotherresidual: consumption. Saving flows are typicallyincozisistent wit changes in wealth stocks beas of incomplete accounting,under-reporting of capitalincome, and exclusionof capitalgains. Savingdoes not haveany identifiable deflatorrelated to it, hence makingit difficultto constructconstant-price time series. Finally,researchers often fail to adequatelycorrect raw savingdata, for instance,for the infltion componentof nominalinterest payments.

" Someof theseunresolved puzzles have been pointedout by recentsurveys of savingbehavior in developing countries, see Gersovitz1988, Deaton 1989b,and Schmidt-Hebbel,Webb and Corsetti1992.

"Issues relatedto the sectoraldisaggregation of savingand differences,in sectoralbehavior are considered explicitlyby the fixstand second quesfions. We will not discuss,however, a furtherdisaggrgton of privaxtsaving amonghouseholds and firms, although some evidenceexists. that-households are not able to 'see through the corporateveil" (for instance,Bosworth 1993 estimates that out of 9 OECDcountries, thehousehold saving offset 22 4.1 - How Effective is Fiscal Policy in Raising National Saving?

At a theoreticallevel, the two dominantparadigms in modemneoclassical macroeconomics offer two strikingly opposing views on this issue.14 Accordingto the life-cycleoverlapping-generations model, publicsaving affectsnational saving (andhence capitalformation in a closedeconomy) because it shifts resourcesacross generationsthat are only weakly (or not at all) connectedto each other. In contrast, the infinite-horizon(or Ramsey)model predictsthat when privateconsumers internalize the government'sintertemporal budget constraint, changes in publicsaving are exactlyoffset by changesin privatesaving because of strongintergenerational linkages (the well-known Ricardian equivalence, Barro

1974).

Whatdoes the evidenceshow? Mostsurveys of the empiricalevidence for OECDcountries reject

Ricardianequivalence (Bernheim 1987, Hayashi1985, Hubbardand Judd 1986,Leiderman and Blejer

1988); the notable exception is Seater 1993 who concludes that Ricardian equivalence holds approximately.Recent studies for developingcountries also reject the Ricardiannotion in its pure form

(Haqueand Montiel 1989, Corboand Schmidt-Hebbel1991,'Easterly, Rodrfguez and Schmidt-Hebbel

1994),but concludethiat some offsettingof publicsaving by privatesavers occurs. -Empiricalrejection of Ricardianequivalence could implythat real-worldconsumption is boundby many featuresassumed away by Ricardian equivalence: weak intergenerational links, financial-market impections Cmduding borwowingconstraints), consumer myopia, and precautionarysaving under uncertainty.

Hcwever,the existingestimates of offsetcoefficients are few, too differentfrom each other, and belongto widelydiffering samples. Even whenusing a commonmethodology, country estimates vary widely. For example, Corboand Schmidt-Hebbel(1991) develop a nested-hypothesisframework that

coefficient for corporate saving is (not) significantly diffrent fiom 1 in 6 (3) countries). However, there are vexy few. saving studies for LDCs using household data, and only one using aggregate (national accounts) data for *households (Schmidt-Hebbel,Webb and Corsetti 1992).

16 See for instance Azariadis (1993), Blanchardand Fischer (1990). and Deaton (1992). 23 distinguishesbetween Keynesian, permanent-income and Ricardianconsumption hypotheses and apply

it to a paneldata sampleof 13 developingcountries, rejecting each of these tiree simplehypotheses in their pure forms. In the overallsample, they find a public-privatesaving offset coefficientof 0.47; this

result,however, conceals a large variationin individual-countryoffset coefficients. The same is true for

Bosworth's(1993) results for a panelof 12 OECDcountries, with average offset coefficientsin the 0.24-

0.33 range(depending on the savingdefinition), but witha muchwider range of offsetcoefficients among

countries. Japelliand Pagano(1994) report offset coefficients for a panel of OECDcountries that vary

hetween0.38 and 0.48, and Edwards(1994) reports offset coefficients for a sampleof OECDand LDC

countriesthat vary similarly-- between0.43 and 0.58 -- dependingon the regressionspecification.

The implicationof the frequentrejection of Ricardianequivalence is chat public saving is an

effectivetool in raising national saving, at least in the short to mediumterm. However,just how

effectivefiscal policy is, is likelyto be muchdisputed until more refined measuresof offsetcoefficients

are obtained.

4.2 - Do Foreign Resource Inflows crowd out DomesticSaving?

While in the abovediscussion on public and privatesaving decisionscausality runs from the former to the latter variable, foreign and national saving decisions can affect each other in either direction. As noted in section 3.2, the higher is internationalcapital mobility,the larger is the degree of endogeneityof foreignsaving to domesticinvestment and saving decisions. Whatdoes the empirical

literaturesay about the influenceof foreignresource inflows on domesticsaving (and investment)?We

will start by referringto studieslooking at the consequencesof overallforeign saving,to summarize subsequentlysome resultson the effectsof foreign aid.17

17 Two analyticallyessential (but seldom acknowledged)points shouldbe addressedby empirical studies. First, a distinction should be made between non-connessionalforeign resource inflows (foreign investment plus non- concessionallending) and foreign aid (unilateraltransfers plus the grant component of concessionallending). While 24 With a non-econometricanalysis, Chenery and Strout (1966) found a negative initial effect of capital inflows on domestic saving, although second-roundeffects on capacity growth tend to increase saving. Giovannini(1983) finds coefficientson foreign saving for LDCs to have mixed signs and to be insignificantin domesticsaving regression equations. Fry (1978, 1980) and Giovannini (1985) find the effect significantand negative, although small. Schmidt-Hebbelet at. (1992) show a significant negative effect of foreignsaving on householdsaving for a panel sample of 10 developingcountries. Gupta (1987) obtains the most extreme results: he reports crowding-in Cinsteadof out), as reflected by positive coefficients on foreign saving, which are significant for Asia and Latin America but not for Asia.

Studies on the effects of foreign aid focus typically on how it is spent on consumption and investrnent,i.e., on empiricalestimates of the marginal propensitiesto consume (MPC) and invest (MPI) of foreign aid transfers. The cross-countryliterature on the relation between foreign aid and saving (and investment) was started by Griffin (197-0, 1971), who found from OLS regressions fdr a cross-section sample of LDCs a negative correlation between saving and aid, with an implied MPC of aid equal to

0.73. Subsequentstudies by Weisskopf (1972), Papanek (1972, 1973), Chenery and Eckstein (1970),

Chenery and Syrquin (1975) generally found far lower consumptionpropensities, implyingthat foreign aid was more effective in financing investment. However, much of this literature - like that on the domestic saving effects of foreign resource inflows - is haunted by severe biases stemming from data measurementerror, mis-specification,and simultaneity(including selectivity) bias as well as an inadequate

the former embodyan external fmancingsource, the latter impliesa wealth transfer. Second,it is importantto deal explicitly with resource fungibility at three different levels: external, domestic macroeconomic,and domestic microeconomic. External resource fungibilityrefers to the degree of intemational capital mobility. Perfect capital mobility is equal to extreme external resource fingibility and implies complete endogeneityof foreign resource inflows. Domesticmacroeconomic resource fungibilityimplies that external resources lead to higher spending - in the absence of external resource fungibility - on both consumptionand investmentaccording to a domestic rationing mechanism(for instance interest rates) that is independentof the programmed resource use. Finally, domestic microeconoumicresource fuigibility implies that external resources lead to higher spending - in the absence of both external and domesticmacroecononic resourcefungibility - on the broad spendingcategory they are intended for (for instanceon consumptionif intended for food consumption),but not necessrily on the specific project or sector they are targeted to. 25 treatment of resource fungibility (see Papanek 1972, Levy 1987, White 1994, and Boone 1994 for methodologicaldiscussions).

More recently, Levy (1988) fbund a MPC from anticipated foreign transfers equal to 0.4. For a sample of Sub-Saharan countries, the Global Coalition for Africa (1993) claims a negative and significant effect of foreign aid on domestic saving. The World Bank reports a MPC of 0.4 and a MPI of 0.6 for net transfers-received by Sub-Saharancountries in the late 1980s (World Bank, 1993b). The most systematicstudy availableto date (Boone 1994) comes to the opposite conclusion, drawn from panel estimationsfor 97 developingcountries. When the sample is restricted to the 82 countries where aid is less than 15% of GNP -- that is, most LDCs -- all of aid is spent on consumptionwhile nothing goes to investment.'l However, when the full sample is used Cincludingthe 15 small SSA and island economies that receive aid equal to more than 15% of their GNP), Boone's estimatedfull-sample MPC drops to 0.45 and his MPI increases from zero to 0.35, suggestingthatthe lack of resource fungibilityis severe in small and poor countries receiving massive aid financingprimarily investmentprojects. His overall finding of a very high MPC in most other countries is consistent with his model that postulates a world comprised of heterogeneouscountries that do not converge in income levels. In such a model the MPC of fbreign

1 aid is 1.0 - but no convergence in income levels seems to be a rather stringent assumption.'

We draw three conclusions. First,. empirical estimates of saving effcts of foreign aid (and foreign saving at large) vary widely with samples, model specifications,and empirical methods. Second, the extent of extemal and domesticresource fungibility is the key determinantof the extentto which non-

" Boone (1994) reports that 93% of foreign aid is grants and that the portion of tied aid is quite small. Emergencyand food aid represent only 5% and 13% of foreign aid, respectively;over 40% is general balance of paymentssupport.

"Foreign aid could affectsaving and investmentindirectly. For instance, private investmentand growth could. rise when foreignaid conditionalityis successfulin promotingpolicy reformsand a better allocationof public expenditure,raising the productivityof private investment. However, Boone's cross-country growth regressions show that foreign aid does not play a positive role in most (82) developingeconomies - but aid helps growth in those 15 small LDCs where it exceeds 15% of GNP. 26 concessionalexternal resource inflows and foreignaid are channeledto higherdomestic consumption and investment.Finally, and most important,very little is knownabout howto measureresource fungibility and how tn identifyits underlyingcauses. Until the latter issue is resolved,the relationbetween foreign resourceinflows and domesticsaving-will be blurred.

4.3 - Do Financialand Tax Incentivesraise Private Saving?

Governmentshave attemptedto raise private saving through many means, includinginterest liberalization,direct tax incentives,and capital market reforms. Many of such attemptshave failed, however(Gersovitz 1988, Deaton 1989b),and this raises questionsabout the responsivenessof saving to such policies.

sbPrivate Saving Sensitive to the Real InterestRate and to Tax Incentives?

- There are two waysof lookingat this question. Oneis by testingfor theinterest effect on saving or consumptionlevels. Along this traditionalline of research, country and cross-countrystudies for

OECDcountries tend to show that saving is not much influencedby interestrates (see Deaton 1992, section2.2). Increasingevidence for developingcountries (Giovannini 1983, 1985;Corbo and Schmidt-

Hebbel1991; Schmidt-Hebbel,Webb and Corsetti1992) suggests that private saving (or consumption) typicallydoes not respondto the real interestrate. Edwards(1994) confirms the insensitivityof private savingto the real interestrate for a cross-countrysample of OECD and developingeconomies. In those exceptionalcases where a positiveresponse of saving to the interestrate is found (Gupta 1987, Fry

1988), it is quantitativelyvery small. There are many possible reasons for this finding.2? For

2' One possibility is that the findings might be spurious due to severe aggregation problems assumed away by the representative-agentmodel. FollowingDeaton (1992, pp.70), consideratwo-overlapping-generationslife-cycle model without bequests, where consumptionat the individual level of each generation is flat along the life horizon, but where life-time consumption of each generation is increased by productivity growth. Then agegate consumptionwill increase over time implying a biased estimate of the intertemporal elasticity of substitution of the 27 consumersnot facing liquidityconstraints, the substitution,income and human wealth effects of an interest rate rise roughly neutralizeeach other when the intertemporalblasticity of consumption substitutionis close to 1, as has been foundby Schmidt-Hebbel(1987) and Arrau (1989)for Southern

Cone countries: Second,liquidity-constrained individuals do not respondmuch to changesin saving incentives- and liquidityconstraints have a significantand positiveeffect on privateand nationalsaving rates in bothOECD and developingcountries (see Jappelli and Pagano1994 for the formerand Easterly,

Rodrfguezand Schmidt-Hebbel1994 for the latter)- In addition,there couldbe a non-monotonicrelation betweensaving and interestrates arisingfrom incomeconcentration when interest rates increase. This wouldimply that at low and negativereal interestrates, higherrates raise savingwhile at high interest rates the savingschedule bends backwards. Evidencefor this has beenprovided by Reynoso(1988) for somedeveloping countries.

A less conventionalway of looking at this issue is 'by testing how sensitive the rate of consumptiongrowth is to the interestrate, followingthe Euler equationapproach. This is equivalentto testingseparately for the intertenporalsubstitution effect Evidencereported by Deaton(1989b) suggests a weakpositive substitution effect across a large numberof countries. Surprisingly,econometric studies for the U.S. fail to find significantpositive effects. For developingcountries Giovannini (1985) finds positivesubstitution effects for 5 countries and no effectsin 13 countries.

If on averageprivate saving were insensitiveto the (after-tax)real interestrate, three policy implications would follow. Fiscal stabilization, if it lowers the real interest rate, would not depress privatesaving through this channel. Financialreform, if it raises the real interestrate, wouldnot raise the flow of private saving through this channel, although it may affect significantlythe portfolio compositionof the stock of savings,possibly from flight capital to domestically-heldfinancial assets.

And tax incentivesfor savingwould be ineffectivein raisingprivate saving. presumed representativeagent as obtainedfiom the macro data 28 On the latter point a sizableliterature has focusedon the savingeffectiveness of tax incentives in OECD countries,particularly in the U.S., making use of cross-sectionhousehold data typically unavailablein LDCs. The resultsof the studiesfor U.S. savingincentive plans (suchas IRAS, 401(k) plansand other schemesthat offer tax deductionson contributionsand accrualof interestbut impose limitson annualcontributions ancd withdrawils) are still inconclusiveon their effectson overallnational saving(see Gersovitz1988, Deaton 1989afor surveys). Amongthe more recent studies,Engen, Gale and Scholz(1994) find that U.S. saving incentiveplans have been ineffectivein raisingnational saving althoughthey may raise saving after two generationsas a result of income shifts toward future generations.

Could Finandal and Capital-MarketReforms raise Private Saving?

Financialand capital-marketreforms have potential effects on privatesaving - other thanthrough changes in real interestrates- throughvarious channels whose effects are of differentsigns. -First, capital-marketreforns (andmacroeconomic stabilization) could lead to reversecapital flight, raisingthe portfolioshare of domesticassets and increasingmeasured income, measured net exports,and measured domesticsaving, but withoutaffecting overall (correctly measured) private saving very much. Second, financial liberalizationand capital-marketdeepening could raise the efficiency of intermediation, increasinggrowth and hence(imdirectly) private saving (Easterly 1993, King and Levine 1993). Third, financialliberalization - and the consequentincrease in geographicaldensity of fiancial institutions, range of financial instruments,and quality of financial-sectorregulation and solvency-enhancing supervision- typicallyleads to financialdeepening, reflected in a permanentincrease in the stocks (and a temporary.increase in the flows)of financialsavings, such as fnancial and banking-sectorliabilities.

Still, this increase in financialsavings could simplyreflect a portfolioshift withoutaffecting overall privatesaving. Fourth, financialliberalization typically leads to expandedconsumer lending and, more 29 generally, less stringent constraints on consumer borrowing, affecting private saving negatively. It has been argued that, on balance, financialdeepening has contributedto the growth in overall savingobserved in East Asian economies(World Bank 1993a).

Empirical studies have typically proxied financial deepeningby a measure of broad money (e.g.,

M2); but broad money could be negativelycorrelated with consumer borrowing constraints (and hence positively with consumption), and positively correlated with consumer wealth (and again with consumptionas well). Hence its effect on saving is ambiguous - and so are the empirical results for cross-country samples. For instance, while Corbo and Schmidt-Hebbel (1991) and Schmidt-Hebbet,

Webb, and Corsetti (1992) report negative effects of broad money on saving in LDCs, Edwards (1994) reports positive effects for OECD and developing economies.

Variables that reflect borrowing constraints more closely are reported to have less ambiguous effects on saving in empirical cross-country saving studies. Jappelli and Pagano (1994) found a significant negative effect of loan-to-asset-valueratios on net national saving in OECD countries and

Edwards (1994) found a negative but insignificanteffect of consumer credit on private saving in OECD and developingcountries.

A final question arises about the potential effects on private and national saving of one important reform with significant capital-marketlabor-market and public-savingimplications: the introductionof mandatorypension schemes and pension refbrmsthat substitutepay-as-you-go (PAYG) systems by fiully- funded (FF) schemes. While the overall evidence on mandatorypension systems is ambiguousfor the

U.S. and Japan, there is some evidencethat Singapore's Central Provident Fund boosted aggregatesaving

(WorldBank, 1993a)and that govermnentexpenditure on social security benefits (typicallythrough state- run PAYG schemes) has reduced private saving (Edwards 1994). Pension reforms that substituteIT for

PAYG systemscould have long-termstatic effects on saving and the level of output and dynamiceffects via capital-marketdeepening and higher growth. Simulationresultsfor the U.S. (Auerbachand Kotlikoff 30 1987) and for representativeLDCs (Arrau and Schmidt-Hebbel1993, Valdds-Prietoand Cifuentes 1993) suggest that the long-term static effects are small. However, dynamic effects through capital-market deepening and growth could be significant, as suggested both by simulattons based on a endogenous- growth model as well as by regression results that explain the spectacular increase in private saving rates attained by Chile after financial market and pension reforms were adopted (Corsetti and Schmidt-Hebbel

1994).

5. INVESTSM: NEW DEVELOPMENTS

As noted above, the analysis of investment decisions was pioneered by Keynes (1936), who challenged the then-prevalentview of the inseparabilitybetween saving and investment decisions. He emphasizedthe distinction between investorsand savers, and viewed investmentas a decision made under highly uncertain conditionsthat required unavoidableguesses about prospective future returns and capital costs.

Since the 1960s investment theory has been in rapid evolution.' We begin this section by reviewing some recent developmentsin investment analysis. We focus on three issues: the implications of irreversibility and uncertainty for investment, the links between economic and political instabilityand investment, and the private investmentconsequences of ill-defined property rights and phenomena such as corruption.

5.1 Uncertinty and Irreversibility

Conventionalinvestrnent theory focused on different variants of the cost of capital (Jorgenson) and the replacement-versus-market-valueof new capital goods (Tobin) as centerpieces of investment decisions. These theories provided useful insights about the decision to invest but failed to fully consider

21 See Serven a;m Solimano (1993, Chapter 2) for a survey of investmenttheories. 31 three main features of investmentdecisions:22 (i) the partial or completeirreversibility of most investmentdecisions - oncethe capitalstock is installedit cannotbe put to a newuse withoutincurring a substantialeconomic cost; (ii) most investmentdecisions face inherentuncentainty regarding future benefitsand costs - the best investorscan do is to attach(subjective) probabilities to de net returnsof differentinvestnent projects; finally, (iii) investorscan controlthe timingof investment,waiting for relevantinformation that can reducethe uncertaintysurrounding investment.

Thesecrucial features have led to a new optionapproach that sees an investmentopportunity as an optionto buy an assetat differentpoints in time, balancingthe value of waitingwith the opportunity cost (m terms of foregonereturns) of postponinginvestment decisions.' Someempirical studies of investmentboth at microeconomicand aggregatelevels confirm the importanceof the valueof waiting.2?

5.2 - Economicand Political Crises and Investment

Economicand political instabilityis an importantfield of applicationof the option view of investment. Enparticular, politicaland economiccrises IRlarge external shock, a financial crash, a war, exacerbatedsocial and politicalconflict) typically create considerable uncertainty about variableswhich are key in the decisionto invest The result can be a huge increasein the value of waitingfar new informationin orderto revisethe probabilitiesattached to the variouspossible events; this is turn causes an investmentslump.

22 Se Dixit and Pindyck,1994, chapter 1.

z The option approach shows that the standard net present discountedvalue rule of investment(invest when the purchase and installation cost equals the expectedreturn) must be modified. The correct rule is that the value of the new capital good must acted the purchase andtinsaton cost, by an amount equal to the value of keeping the investmentoption alive.

24 See partic6larly chapter 4 in Serven and Solimano(1993). '32 Political instabilityreflected in rapid governmentturnover may also hamper investmnent,if it leads to an unstable incentive and policy framework. An adverse investmentimpact is especialiy likely when the political change involves a re-definitionof the basic "rules'of the game" (investmentcodes, property rights, tax laws) and, in particular, when it raises expropriation risks (e.g., Chile in the early 1970s,

Nicaragua in the 1980s). Empirically, recent studies (e.g., Alesina and Perotti 1993, Mauro 1993) confirm the important role of political instabilityas an investmentdeterrent.,

5.3 - Property Rights and the Cost of Doing Business

There is no question about the importance of property rights for investment. From the practical viewpoint, however, the formal definition of property rights needs to be complementedby other factors guaranteeing their effective enforcement. An important link in that enforcement are principal-agent relationships,that tend to be very complex. Monitoring of agents (managers, workers) by the principal

(the owner) often has to face imperfect informationand is costly. The mrereexistence of private propert is no assurance that the agents will act in the best interests of thieowners, a feature that is partiuLariy- relevant for investment.

Related to this is the issue of contract enforcement, whose effectiveness can be jeopardized by weak and inefficientjudiciary systems. Yet without proper contract enforcement, property rights would be good only on paper. In the end, the lack of impartial mechanisms to resolve contractual disputes results in an increased cost of doing business, as the probability that contracts will be respected diminishesand/or expensessuch as bribes become necessary.

This leads to the issue of corruption, wl-.ichis critical for investment projects because their implementation- particularly in highly regulated economies- may involve many administrativesteps:

-gettingan investmnentpermit, obtaininglicenses to importcritical inputs, gainingaccess to directed credit, assuring compliancewith labor standardsand environmentl regulations, etc. The cost of doing business 33 can rise substantiallyif the typicalinvestor is forcedto bribe governmentofficials-in order to complete

or speed up these administrativesteps. In this regard, corruptioneffectively amounts to a tax on

investment. Recentcross-country empirical work by Knack and Keefer (1994) shows that higher

measuresof corruption,just like weakerproperty rights, tend to be associatedwith a significantly

worsenedinvestment performance.

The eliminationof unnecessaryregulation can reducepotential sources of rents for bureaucrats

and thus corruption. Civilservice reform, includingcompetitive-pay polidies, career opportunitiesand

moralstandards is also an importantaspect of institutionalreform to boostinvestment and growth.

5.4 - Profits, IncomeDistribution and Capadty Utilization

A basic insightof Keynes(1936) is the dependenceof investmenton expectedprofits (relative

to the cost of capital). The rate of profit (profitsover capital),can be expressedas the productof the

- - profit share (profits over output) times the rate of capacity utilization (output over capital). This brings

out two key determinantsof investment:income distribution and the levelof aggregatedemand (or phase

of the businesscycle). Incomedistribution essentially refers to the shares of capital, labor and other

factors in national income. However, the relationshipbetween income distribution and capital

accumulationis a complexone. On the one hand, a distributionof incomebiased towards capital can

increasethe profitrate througha higherprofit share. On the other hand, if aggregatedemand depends

negativelyon the profit share - due to a higher propensityto spendfor labor than capital - then a

redistributionof incometowards capital will depress aggregatedemand, lower the rate of capacity

utilizationand depressthe profitrate. In sum, thepro abiltyeffect can operate in an oppositedirection

to the demandeffect, making the reationshipbetween income distribution and investnentambiguous.

Other channelsmight be relevanttoo for the relationshipbetween accumulation and distribution.

Incomedistribution affects the choiceof publicpolicies and the degreeof socialand politicalstability: 34

an uneven income distribution may encourage workers' demands and labor militancy, increasing the

degree of political conflict. An uneven income distribution might also compel governments to seek

populist policies of rapid redistribution of income that can be fiscally and macroeconomically

destabilizing, as has been witnessed several times in Latin America during the last few decades (see

Dornbusch and Edwards, 1991). Redistributivepolicies in turn, may be financed through higher taxes

on capital thereby discouraging investment and growth. The relationship between investment and

distribution is filtered by political inistitutions,and can be stronger in democracies than in non-

democracies (see Persson and Tabellini, 1992 and Alesina and Rodrik, 1992). Empirical evidence for

a cross-section of countries provided in Persson and Tabellini (1992) shows a positive and marginally

significantcorrelation between equality in incomedistribution and the investrnent-to-GDPratio.

Stronger results are obtained in Alesina and Rodrik (1992). In cross-section regressions using

the ratio of total physical investmentover GDP as the dependentvariable and controlling for the type of

politicalregime (democraciesversus non-democracies),they find a positivecorrelation between the degree of income equality and the investmentratio. A similar result arises when using average GDP growth

-ratesas the dependent variable, thus sugesting that investmentis an important channel through which

incomedistribution affects GDP growth.

6. ADJUSTMENT, REFORM, SAVING AND INVESTMENT

This section discusses the main linkages between investment adjustment and overall systemic

reform. Many adjustmentprograms seek to correct an excessof aggregatespending over real output that

creates a deficit in the current account of the balance of payments andlor inflationary pressures. From

simple national accounts identitiesthe aggregateexcess spending signifies an excess of investnientover saving. A cut in investmentspending often takes place during the first phase of an adjustmentprogram

(see below). This is an adverse trend if the investmentcompression lasts too long, as was .the case of 35 Latin America in the 1980s after the debt crisis. In this case, productive capacities are impaired and so is the economy's ability to grow.. However, if the productivityof investment is low then some decline in investmentwill not be that harmful.

National saving also tends to follow a distinct pattern according to the different phases of an adjustment and reform program (of varying length, overlap and partial recurrence). To understandthe interactionsbetween investment, saving, adjustment and reform it is useful -to separate.between the pre- and post-reform situations.

6.1 - The Pre-refonn Situation

There is a wide variety in initial conditions for reform. Some countries might experience instabilityand large macro imbalancessuch as high and erratic inflation, exchange rate volatility, fiscal deficits and balance of payments disequilibria. Examples abound, including Argentina, Brazil, Peru,

Poland, Russia in the late 1980s and the early 1990s. In contrast, other countries have initiated their reform programs from more stable macro conditions,?5 but with pervasive microeconomic distortions and an oversized state?26

When instability is rampant before reform, private .investmentis hurt. The key reason is that, as noted earlier, instability and uncertainty invite a delay in investment spending as most investment outlays are irreversible. Empirically, evidencefor five high inflation countries in Latin America from

1960 to 1990 shows that private investment is negatively correlated with the level (and sometimes the variance) of inflation and with the variance of the real exchange rate, two indicators of macroeconomic instability. Interestingly, when the same equations for private investment are run for (low inflation)

- In theseeconomies, the macrodisequilibria (e.g., inflation)can be repressed. In arn, fiscaland balance of payments deficits can be financed by external aid granted for some specific geopolitical reasons.

2 See Solimano, Sunkel and Blejer (1994, Chapter 1) for a discussion of inifial conditions before reform in a broad vanretyof countries. 36 OECDeconomies, investment is negativelycorrelated also with inflationand measuresof exchangerate volatility (see Pindyck and Solimano, 1993).Y These results dearly support the hypothesisthat

instabilityis a powerfuldeterrent of privateinvestment.

Pre-reforminstability can also affectpublic investment, in particularif there is a fiscalcrisis that forcesgovernments to cut spending.In general,capital expenditures are cut beforecurrent expenditures, as the lattercuts often createless political resistance. Of coursethe costsshow up in the futurethrough lack of infrastructureand other productivecapacities.

In the case of highly distortedbut more 'stable" macroeconomics,the problem is the low productivityof investmentrather than too low a levelof capitalaccumulation. The problembecomes one of "quality' ratherthan "quantity"of investment.

6.2 - The Reform Period

Let us turn now to the situationin whichreform policiesare adopted. For analyticalpurposes

we will identifythree differentphases of reformprograms that are relevantin terms of their impacton

investnentand saving.

Phase 1. Adjustmentand Stabilization

In economieswith large macro disequilibria, the firsttask of a reforminggovernment is to restore-

basicmacro balances and stabilizeinflation. The typicalpackage involves a combinationof expenditure-

reducingpolicies (fiscal and monetaryrestraint) along withexpenditure-switching policies (depreciation

of the real exchangerate). These policies are usually supplementedby the eliminationof various

subsidiesand by foreignfinancing coming from multilateralinstitutions.

Therelationship between investment and inflationis foundto benon linearin Pindyckand Solizuo (1993). Only when inflationmates exeed cerain threshoidsa significantdrop in investmenttakes place; moreover,these tdresholdsare muchhigher in LatinAmerica than in the OECD. 37 Aucteritymeasures tend to reducecapita formation,at least in the short run, for severalreasons.

Fiscaladjustment often involves a cut of publicinvestment, currency depreciation makes importedcapital goodsmore expensiveand monetarytightening pushes up real interestrates depressingprivate capital formation..'

On the savingside, during this first stage fiscal adjustmentis also reflectedin higher public saving. However,the contractionaryeffects of restrictiveaggregate demand policies typically lead to a steep declineof privatesaving, as consumersattempt - as longas they are not credit-constrained to maintaintheir consumptionlevels during the recessionaryphase. Nationalsaving often falls duringthis stage - a result of privatesaving cuts that exceedthe increasesin public saving. Dependingon how strong the declineof domesticinvestment is, the ex-antedemand for foreignsaving - satisfiedex-post only in the absenceof a foreignborrowing constraint - couldrise or fall duringthis period.

Phase II. Intensiricationof structural reformsand the InvestmentPause

A secondphase in reformprograms comes after the basicmacroeconomic disequilibria have been corrected. At this stagegovernments can acceleratete structuralreforms comprising restructuring and liberalization.Of course,in practicethere mightbe somesimultaneity between macro stabilization and the structuralreforms as the latterwill hardlyconsolidate if the macrofundamentals are not in place.

The responseof investmentin this secondphases is mixed. In Chileprivate investment responded forcefullyto the launchingof reformsin the secondhalf of the 1970s3' Likewise,private investment has surgedin Argentinaand Peru in the early 1990safter toe launchingof liberalizationpolicies. By contrast,the responseof privateinvestnent was modest in Israel, Mexicoand Bolivia,in the aftermath

- See Servenand Solimano(1993, Chapter 2) for a surveyof the effectsof fiscal,monetary and exchangerate policieson investment.

21 This responsewas reversedin the crisisof 1982483. 38 of stabilizationduring the secondhalf of the 1980s. A substantialdeal of work has beendone on the so- called"investment pause" after adjustment." The combinationof lack of confidencein the consolidation of structuralreforms and the fact that investmentis irreversible(in an economicsense) makesinvestors adopt a "waitand see" attitudein the aftermathof stabilization.

Publicinvestment often suffersduring the adoptionof reformpolicies. In part this is a byproduct of fiscal adiustment;however, the decline in public investmentalso obeys a deeper structuralchange.

In the new developmentstrategy that followsthe adoptionof market-orientedreform, a smallerrole for the state is envisagedboth as producerand providerof basic infrastructure. The privatizationof state- owned enterprisestypically leads to a declinein public investment. Also, the openingof new business opportunitiesto the privatesector in activitiesthat wereformerly in the realm of the publicsector - like powergeneration, roads, ports -- leads to a decline in public investnent in these activities. How does private saving respond to these structural reforms? If cerin reforms - such as trade opening or liberalizationof capital inflows- are not fully credible, they may spur a consumptionboom to take advantageof temporarilycheap importedgoods. And if they are credible and lead to an expectationof higher permanentincome, consumersmay spend today in anticipationof tomorrow's income gains.

Moreover,trade liberalizationfrequently allows consumers to adjusttheir holdingsof consumerdurables towarddesired levels after years of importdeprivation. For all these reasonsliberalization policies lead often to cn initialdecline in private saving that can be offset by an increase in public saving as fiscal adjustmentis deepenedin this phase.

Phasem. Maturity of Reforms and HRighGrowth

The third phase of a reformprogram reflects consolidation of the measurestaken before. This is a "happyperiod" as the fruits of reform start to accrue in terms of highergrowth, enhancedstability

s See Dornbuch (199lb), SoGmano (1992, 1993), Serven and Solimano (1993). 39 and improvementin living standards. Examplesof countriesin this stage would be Chile in the late

1980sand early 1990sand the rapidly-growingeconomies of East Asia. At this stage toe main problem is not how to get investmentand growthunderway, but rather how to preservethe dynamismalready achieved. In this case, the sensiblething would be to maintainthe "fundamentals'in place - low inflation,fiscal balance, external sustainability, adequate price signals,social stability -- and monitorthe sectoralcomposition of investment.In fact, duringmore advancedstages of the developmentprocess it becomescrucial to enhancethe economy'sinfrastructure and supportthe physicalinvestment effort with the provisionof better education,training and other formsof humancapital accumulation.

On the saving side, once durable stock adjustmentis well-advancedand real growth'is rising, privateconsumption tends to grow lessthan income,therefore generating a steadyrise in privatesaving, as exemplifiedby the increasein Koreanurban householdsaving rates from 3% of householdincome in

1965-1970to 23% in 1980-85(Collins 1991)and by the rise in the privatesaving rate in ChHefrom

1.0% of GDP in 1979-81to 12% in 1982-89and to 17% in 1990-92(Corsetti and Sclunidt-Hebbel,

1994).

7. POLICY ISSUESAND FINAL REMARKS

We concludeby summarizingour discussionof the saving-investment-growthmechanism and derivingsome policy implications for savingand investment.

Threegeneral lessons can be drawnfrom the recenttheoretical and empiricalliterature. The first one is that, in spite of the virtuouscircles of high saving and investnentand rapid growth apparentin the empiricalfacts, the saving-investment-growthrelationship is a complexone, in whichcausality runs in severaldirections. The secondlesson is that, nevertheless,saving seems to be mostlypassive and in manyinstances it appearsto follow- ratherthan precede- investmentand growth,contrary to the Mill-

Marshall-Solowtradition. Finally, the third lesson is that investmentand innovationremain the 40 centerpiecesof the growth process. In this regard, the new growth literature representsa decided(albeit perhaps unintended)return to the traditioninitiated by Marx, Schumpeterand Keynes.

Saving

Even if saving is not the chief driving force behindgrowth, ensuring adequatesaving levels still remains a central policy concern in order to guaranteesufficient financing for capital accumulationand avoid an excess of investmentover saving that can create inflationarypressures and/or balance-of- paymentsdisequilibria. Moreover,encouraging private saving may be essentialto expandinvestment in a frameworkof extensivecapital market imperfections and liquidityconstraints on firms and households, typicalof many developingeconomies.

From our discussionof saving determinants,we derivethree policy conclusions. First, since the evidenceshows that public saving does not crowd out private saving one-to-one,raising public saving is an effective,direct way, availableto policy makers in order to raise nadonal saving.

Second, as long as the domesticmacroeconomic policy frameworkis sustainable,the domestic bankingsector is effectivelyregulated and supervised, and the govermnentdoes not provideguarantees on foreign credit flows, foreign saving inflowsshould be allowedand encouragedto support domestic investment- even if they help in part to finance consumption.

Third, the empirical evidence indicates that one should not expect higher private saving in response to interest rate liberalization. Market-determinedinterest rates will improve financial intermediation,the qualityof portfoliochoices (henceavoiding or reverting capital Right and, possibly, raising measuredsaving flows) and the quality of investment,but not necessarilythe total volume of saving. Financialdeepening (as reflectedby higher financialsavings stocks) has ambiguouseffects on savingwhile a relaxationof constraintson consumerlending tends to depressprivate saving. By contrast, pensionreform can be much more importantin mobilizingdomestic resources. 41 Investment

If investmentand innovationare the key to growth,an importantpolicy questionis whetheran activist"investment policy' is neededto boostgrowth and, if so, which form it shouldtake. In theory, the need for an activistpolicy couldbe justifiedby the existenceof investmentexternalities, as implied by the positivecorrelation between investment and total factor productivitygrowth. To correct the externality,the conventionalpolicy response would involvesubsidies and/or tax exemptionsto capital accumulation.However, although this typeof direct-interventionpolicymight theoretically boostgrowth, there are, in practice,serious limitationsto the administrativeand institutionalcapacities of government agenciesto target the "right" investmentsand avoid rent-seeking.

A perhaps more effectiveway to promoteinvestment, innovation and growth is throughthe provisionof a supportivepolicy and institutionalfameworlk Such a frameworkrequires severn ingredients:macroeconomic stability, a relativelydistortion-free relative price structure,well-defined (and effectivelyenforced) property rights, an environmentconducive to a low cost of doing business,and adequatepolitical institutions that foster socialconsensus and politicalstability. To complementall this, the governmentshould ensure the provisionof adequate trte and humancapital investment - be it in the formof publicprojects or with privatesector participation. 42

REFERENCES

Aicina, A. and D. Rodrik (1992): "Distribution,Political Conflict and EconomicGrowth: A Simple Theoryand SomeEmpirical Evidence" in A. Cukierman,Z. Hercowitzand L. Leiderman(eds) PoliticalEconomy. Growth, and BusinessCycles, MIT Press.

Alesina, A. and R. Perotti (1993): "Income Distribution,Political Instability and Investment",NBER WorkingPaper 4486.

Argimon,I. and J. Roldan (1994): "Saving,Investment and IntemationalCapital Mobilityin EC Countries," EuropeanEconomic Review 3 8, p. 59-67.

Arrow, K. J. (1962) " The EconomicImplications of Learningby Doing" Reviewof EconornicStudies, 29.

Arau, P. (1989) "IntertemporalMonetary Economics: Evidence from the Southem Cone of Latin America",Ph-D. Dissertation,University of Pennsylvania,Philadelphia.

Amru, P. and K. Schmidt-Hebbel(1993) "Macroeconomicand IntergeneationalWelfare Effectsof a Transitionfrom Pay-As You-Go to Fully-FundedPension Systems," manuscript, The WorldBank, June.

Auerbachand L Kotlikoff(1987) DynamicFiscal Policy.Cambridge University Press.

Azaiadis, C. (1993)Intertemporal Macroeconomics Blackwell.

Barro, R (1974) "Are GovernmentBonds Net Wealth?",Joumal of PoliticalEconomy, 82: 1095-1117.

Baumol,W. et. al. (1989):Productivity and AmericanLeadership: The Lone!View. Cambridge:MUT Press.

Benbabib,J. and B. Jovanovic(1991): "Externalitiesand Growth Accounting",American Economic Review81: 82-113.

Bernheim,B.B. (1987) "RicardianEquivalence: An Evaluationof Theoiy and Evidence,"in NBER MacroeconomicsAnnual 1987. MIT Press, Cambridge,Mass.

Blanchard,0. and S. Fischer(1990) Lectures on Macroeconomics,The M.I.T. Press.

Blocker,R. (1994):"Policy Implications of the InternationalSaving-Investment Correlation," mimeo, American University.

Blomstrom,M., R. Lipseyand M. Zejan(1993): "Is Fixed investmentthe Key to EconomicGrowth?", NBERWorking Paper 4436.

Boone,P. (1994): "TheImpact of ForeignAid on Savingsand Growth",manuscript, London School of Economics,June.

Bosworth,B.P. (1993)Saving and Investmentin a GlobalEcnomy. Brookings,Washington, D.C. 43

Bruno, M. (1994): 'Short-Run Macroeconomic Managementand Long-mn Economic Growth mnimeo,The World Bank.

Campbell, J.Y. and A. Deaton (1989) "Why is Consumptionso Smooth?", Review of Economic Studies 56: 357-74.

Carroll, C. and D. Weil (1993): "Saving and Growth: A Reinterpretation", NEER Working Paper 4470.

Carroll, C. and L Summers (1991): "Consumption Growth Parallels Income Growth: Some New Evidence", in B.D. Douglas and J.B. Shoven, (eds.): "National Saving and Economic Performance. National Bureau of Economic Research, University of Chicago Press.

Chakravarty, S. (1993): Selected Economic Writings, Oxford University Press.

Chenery, H. B. and W. Strout (1966) 'Foreign Assistance and Economic Development", American Economic Review 66: 679-733.

and P. Eckstein (1970): "DevelopmentAltematives for WLinAmerica", Journal of Political Economy 78 (4): 966-1006.

and M. Syrquin (1975): patterns of Development: 1950-1970. Oxford University Press.

Cole, H.L, G.J. Mailath and A. Postlewaite (1992) "Social Norms, Savings Behavior, and Growth", Journal of Political Economy, 100 (6): 1092-1125.

Collins, S.M. (1991) "Saving Behavior in Ten Developing Countries, in B.D. Douglas and J.B. Shoven,- (eds.): "National Savine and Economic Performance. National Bureau of Economic Research, University of Chicago Press.

Corbo, V. and IC Schmidt-Hebbel(1991) "Public Policies and Saving in Developing Countries," Journal of DevelopmentEconomnics, 36: 89-115, July.

-___ and P. Rojas (1993) "Investment, Macroeconomic Stability and Growth: The Latin American Experience" AnaflisisEconomico S.

Corsetti, G. and K. Schmidt-Hebbel(1994): "Pension Refbrm and Growth", manuscript, presented at the Conference on Pensions: Funding, privatization, and Macroeconomic Policy, Santiago (Chile), January.

Deaton, A. (1989a) "Savingand Liquidity Constraints", manuscript, Princeton University Prss, Woodrow Wilson School, Princeton, N.J.

(1989b) "Savingin Developing Countries: Theory and Review", Proceedings of the World Bank Annual Conference on Development Economics, The World Bank, Washington, D.C.

(1992) Understanding Consumption. Oxford University Press.

De Long, J. and L Summers (1991): "Equipment Investment and EconomnicGrowth", Ouarterly Journal of Economics 106: 445-502. 44 (1993):'How StronglyDo DevelopingEoonomies Benefit from Equipment Investment ?', Jounal of Monetar Economics32: 395416.

Denson, E. (1962):Sources of EconomicGrowth in the United Statesand the AlternativesBefore Us. New York: Committeefor EconomicDevelopment.

(1967):WMy Grwth RatesDiffer. WashingtonDC: The BrookingsInstitution.

Do Oregano,G. (1992) "EconomicGrowth in Latin America"Journal of DevelopmentEconomics 39.

Dixit, A. and R. Pindyck(1994): Imnestment Under Uncertainty.Princeton University Press

Dooley, M.3 1. Frankel and D. Mathieson (1987): Inmteationa Capit Mobility: What do Saving- InvestmentCormlations Tell Us 7'. IMP Staff Papers34: 503-30.

Dombusch,R. (1991a):"Comment", in B. DouglasBemheim and . Shoven,eds., NationalSaving and EconomnicPerformance Chicago: university of Chicago Press.

(1991b):"Policies to MoveFrom Stabilization to Growth"frgWinjs of the WorldBank Annual -Confere on DevelopmentEcono-mi 1990. Washington, DC.

Easterly,W. (1993): 'How muchDo Distortionsaffet Growth?', Journalof MonetaryEconomics, 32.

and S. Fischer(1994): "The SovietEconomic Decline Historicaland RepublicanData", WorldBank PR WorkingPaper 1284.

Easterly,W. andS. Rebelo (1993):'FLn Policyand EconomicGrowth: an EmpiricalInvestigation", Joumal of MonetaryEconomics 32: 417-458.

Easterly, W., C.A. Rodriguez, and K. Schmidt-Hebbel(eds.) (1994): Public Sector Deficits and MacroeconomicPerformance. Oxford University Press, forthcoming.

Edwards, S. (1994): Why-are Latin America'sSaving Rates so Low?", manuct presente at the Confec on GrowthProspects in Lain America,Bogott (Colombia),June 1994.

Feldstein,M. and P. Baechetta(1991): "NationalSaving and Intetonal Investment",in B. Douglas Bemnheimand J. Shoven,eds., Nation Savin and Enomic Performance.Chicago: university of ChicagoPress.

Feldstein,M. and C. Horiokl (1980): "DomesticSaving and interational Capital Flows", Economic Journal 90: 314-29.

Fershtzn, C. and Y. Weiss(1993) 'Social Stats, Cultre and EconomicPerformance". The Economic eurial103 (July):946-959. -

Flavin, M. (1981) "The Adjustmnt of Consmption to ChangingExpectations about Future hicome", Journalof PoliticalEconomy 89; 974-1009.

Fan", 3. (1992): 'MeasuringInterationl CapitalMobility: A Review', AmericanEconomic Review 82 (2): 197-202.May. 45 (1993): 'QuantifyingInternational Capital Mobility in the 1980s', in On E3xchangeRates. Cambridge,MA: MIT Press.

Fry, M.J. (1988) Moneg. Interest and Bankinain EconomicDevelopinent. John HopkinsUniversity, Baltimore,Maryland.

Gerschenkron,A. (1962):Economic Backwardness in a HistoricalPerspective. Cambridge, MA: Harvard UniversityPress.

Gersovitz,M. (1988)"Saving and Development",in H. Cheneryand T.N. Srinivasan(eds.): Handbook of DeveloomentEconomics, North Holland.

Giovannini,A. (1983)'The InterestElasticity of Savingsin DevelopingCountries: the ExistingEvidence,' WorldDevelopment, 11(7):601-07.

(1985): 'Saving and the Real InterestRate in LDCs, "Journalof DevelogmentEconomics, 18, August.

GlobalCoalition on Africa (1993)'Annual Report", manuscript, November.

Griffin, K. (1970): 'Foreign Capital,Domestic Savings, and EconomicDevelopment', Bulletin of the OxfordInstitute of Economicsand Statistics32: 99-112.

(1971): 'Reply', Bulletinof the OxfordInstitute of Economicsand Statistics33: 156-61.

Gupta, K.L (1987) "AggregateSavings, Financial Intermediation,and Interest Rate,' Review of Economicsand Statistics.69(2):303-11.

Hall, R.E. and F. S. Mishkin(1982): 'The Sensitivityof Consumptionto TransitoryIncome: Estimates fwomPanel Data on Households',Econometrica 50: 46181.

Haque, N.V. and P. Montiel (1989): 'Consumptionin DevelopingCountries: Tests for Liquidity Constraintsand FiniteHorizons," Review of Economicsand Statistics71 (3): 408-15.

Hayasbi,F. (1985) 'Tests for LiquidityConstaints: A Critica!Survey, 'NBER WorkingPaver 1720, Cambridge,Mass., October.

Hubbard,RG. andKLL Judd (1986)'Liquidity Constraints, Fiscal Policy and Consumption,' Brookinfs Paners on EconomicActivity. 1.,

Japelli, T. and M. Pagano (1994): 'Saving, G;rowth,and Iiquidity Constraints",Ouarterlv Journal of Economics, : 83-109.

Kaldor,N. (1957) 'A Modelof EconomicGrowth' EcosiomicJounal, 67

King,R. and R. Levine(1993): 'Fmance, Ent ship and Growth', Journalof Monetr Economics 32: 513-542

- (1994):'Capital Fundamentlism, Economic Development and EconomicGrowth', mimoo,Tec World Bank. 46- Knack, S. and P. Keefer (1994): 'Institutionsand Economic Performance",miimeo, University of Maryland.

Kuzets, S. (1973):PODulation. Canitai and Growth.Selected Essays New York: Norton.

Lefort, F. and A. Solimano(1993) "Economic Growth After Markst Based Reforms in LatinAmerica: The Cases of Chile and Mexico', mimeo,The World Bank.

Leiderman,L. and M. Blejer(1988) 'Modelingand TestingRicardian Equivalence: A survey," IMF Staff Payers 35. No.l, March.

Levine, R. and D. Renelt (2992): 'A SensitivityAnalysis of Cross-CountryGrowth Regressions", AmericanEconomic Review 82: 942-963.

Levy, V. (1988):'Aid and rowth in Sub-SaharanAfrica The RecentExperience", European Economic Review32(9): 1777-95.

Mankiw, G., D. Romer and D. Weil (1992): 'A Contributionto the Empiricsof EcononilcGrowth', Ouarterlylournal of Economics107: 407-438.

Marglin,S. and J. Schor (1990):The GoldenAMe of Capitalism:Reiutervreting the Post-WarExperience. Oxford: ClarendonPress.

Mason,A. (1987) 'National Saving ates and PopulationGrowth: A New Modeland New Evidence",in D.G. Johnsonand R. Lee (eds.) PopulationGrowth and EconomicDevelopment: Issues and EjvideneeUniversity of WisconsinPress, Madison,Wis.

(1988) 'Saving, EconomicGrowth and DemographicChange", Populationand Development Review 14 (1).

Mauro, r. (1993): 'PoliticalInstability, Growth and Investment",mimeo, Harvard University.

Modiglini, F. (1990):'Recent Declinesin the Savingrate: A Life CyclePerspective", manuscript, Frisch Lecture, Sixth World Congressof the EconometricSociety, Barcelona(August).

Montiel, P. (1993): 'Capital Mobility in Developing Countries: Some MeasurementIssues and Empirical Estimates', World Bank PRE Worcing Paper 1103.

Murphy,R (1984):"Capital Mobility and the RelationshipBetween Saving and Investmentin OECD Countries", Journal of IntenationalMoney and Finance3: 327-342.

Obstfeld, M. (1986): 'Capital Mobilityin the World Economy:Theory and Measarement",Carnesie- RochesterConference Series on PublicPolicy. Amsterdam.-North-Holland.

-____ (1994):'International Capital Mobility in the 1990s', Boardof Governorsof the FederalReserve SystemIntemational Finance Discussion Paper 472.

Ppanek, G.F. (1972): 'The Effect of Aid and OtherResource Transfers on Savingsand Growthin Less DevelopedCountries", Economnic Journal 82: 934-51. 47

-____ (1973): "Aid, Foreign Private Investment, Savings and Growth in Less Developed Countries', Joumal of Political Economy, 81(1): 121-30.

Persson, T. and G. Tabellini (1992): "Growth, Distributionand Policies' in A. Cukienmmn,Z. Hercowitz and L. Leiderman (ods) Political Economy. Growth, and Business Ccles The MIT Press.

Pindyck, R. and A. Solinano (1993): 'Economic instability and Aggregate Investment', JER MacroeconomicsAnnual.

Robinson, J. (1962) Essays in the Theory of Economic Growth London: Macmillan.

Romer, P. (1986): 'Increasing Returmsand Long-run Growth', Journal of Political Economy 4, 1002- 1017.

(1987): "Crzy Explanations for the Productivity Slowdown', NBER MacroeconomicsAnnual.

(1993): 'Idea Gaps and Object Gaps i EconomicDvelopment', Journal of Monetar Economics 32, 543-574.

Rostow, W. (1958): The Stages of Economic Growth. London: Macmillan.

Schmidt-Hebbel,K. (1987) 'Foreign Shocks and MacroeconomicAdjustment in Small Open Economies', Ph.D. Dissertation, MassachusettsInstitute of Technology, Cambridge, Mass.

Schmidt-Hebbel,K., S. Webb and G. Corsetti (1992) 'Household Saving in Developing Countries: first Cross-Country Evidence," World Bank Economic Review. 6(3): 529-547, September.

Schumpeter, L (1934) The Theoa of EconomicDevelopment Harvard University Press.

Seater, 3.J. (1993) 'Ricardian Equivalence,' Journal of Econoniic Literature, XJXX (1): 142-190, March.

Serven, L and A. Solimano (1993): Striving for Growth After Adiustment the Role of Capital Formation. WashingtonDC: The World Bank.

Solimano, A. (1992): "Understanding the Investment Cycle in AdjustmentPrograms. Evidence from Rcfoming Economies. WPS 921, The World Bank

(1993): 'How Private InvestmentReacts to Changing MacroeconomicConditions. The Case of Chile in the 1980s? in A. Chhibber et. al. (edi) Reviving Private Investment in Developing Countries. North Holland.

Solimano, A., 0. Sunicl and M. Blejer (ads.) (1994): Rebuildinz Capitalism. Ann Arbor. The University of Michigan Press.

Solow, R. (1956): 'A Contribntionto the Theory of Economic Growth', OuarterlyJournal of Economics70: 65-94.

-(1957): 'Technical Change and the Aggregate Production Function', Review of Economics and Slatistics 39: 312-320.

Summers, L (1983): 'Tax Policy and InternationalCompetitiveness', in J. Frenkel, ed., Internaonal Apectso Fiscal Policies. Chicago: Universityof Chicago Press. 48

Tesar, L (1991): "Savings, Investment, and International Capital Flows', Journal of International Economics 31: 55-78.

Vald6s-Prielo,S. and R. Cifuentes (1993): 'Credit Constraintsand Pensions', manuscript, Catholic University of Chile, Santiago, December.

Weisskopf, T.E. (1972): 'The Impact of Foreign Capital Inflows on Domestic Savings in Underdeveloped Countries', Journal of InternationalEconomics, 25-38.

White, H. (1992a) 'The MacroeconomicImpact of DevelopmentAid: a critical survey", Journal of Develonment Studies 28: 163-240.

White, H. (1992b) "What do we know about Aid's MacroeconomicImpact?', Journal of International Development 4:121-137.

White, H. (1994) 'The Country-WideEffects of Aid: Report for the World Bank", manuscript, Istitute of Social Studies, The Hague, January.

World Bank (1993a) The East Asian Miracle: EconornicGrowth and Public Policy, Oxford University Press.

World Bank (1993b) "Adjustment in Africa: Reform, Results, and the Road Ahead', Discussion Draft, Policy Research Department, October.

Yotopoulos, P. and J. Nugent (1976): Economics of Development: Emnirical Investigations. New York: Haper and Row.

Young, A. (1928): 'Increasing Returns and Economic Progress" Economic Jou.rnal, 38.

(1993): 'Lesons from the East-AsianNICs: a Contrarian View', NBER Working Paper 4482.

Tou, Heng-fu (1993a) "The Spirit of Capitalismand Savings Behavior', manuscript. The World Bank, October.

(1993b) "The Capitalist-SpiritApproachto Growth: A Perspectivefrom the History of Economic Analysis', manuscript, The World Bank, November.

(9t28:upai4S Policy Research Working Paper Series

Contact Thits Author Date for paper

WPS1358 Patternsof Behaviorin Andrew Metick September1994 A. Marafion Biodiversity Preservation Martin L Weltzman 39074

WPS1359 When MethodMatters: Toward a Martin Ravallion September1994 P. Cook Resolutionof the Debateabout Binayak Sen 33902 Bangladesh's Poverty Measures

WPS1360 Are PortfolioFlows to Emerging SudarshanGooptu September1994 R. Vo MarketsComplementary or 31047 Competitive?

WPS1361 External Shocks and Performance F. DesmnondMcCarthy September1994 M. Divino Responsesduring Systemic ChandrashekarPant 33739 Transition:The Case of Ukraine KangbinZheng GiovanniZanalda

WPS1362 Regulation, Institutions,and PabloT. Spiller October 1994 B. Moore Commitment The Jamaican Cezley I. Sampson 38526 TelecommunicationsSector

WPS1363 Brazil's SugarcaneSector. A Case Brent Borrell October1994 M. Bale of Lost Opportunity Jos6 R. Bianco 31913 MalcolmD. Bale

WPS1364 Why Do Some EconomiesAdjust lshrat Husain October 1994 J. Schwartz More SuccessfullyThan Others? 32250 Lessonsfrom SevenAfrican Countries

WPS1365 The Macroeconomicsof Adjustment lshrat Husain October 1994 J. Schwartz in Sub-SaharanAfrican Countries: 32250 Results and Lessons

WPS1366 DistributiveConcems When Salvador Vald6s-Prieto October 1994 E. Khine Replacinga Pay-As-You-GoSystem 37471 vwitha Fully FundedSystem

WPS1367 The Economicsof Cash Shortage Patrick Conway October 1994 L Suki 33974

WPS1368 Sustained Inflationin Responseto Patrick Conway October1994 L Suhi Price Uberalization 33974

WPS1369 Economic Policy Reform, Philip L Brock October 1994 P. Sintim- GovemmentDebt Guarantees,and Aboagye Financial Bailouts 38526

WPS1370 Is East Asia Less Open than North SumanaDhar October 1994 J. Ngaine Americaand the EuropeanEconomic Aruind Panagariya 37959 Community?No PolIcy Research Working Paper Series

Contact Title Author Date for paper

WPS1371 The Evolutionof Trade Treaties and Sarath Fiajapatirana October 1994 J. Troncoso Trade Creation: Lessons for Latin America 37826

WPS1372 Administrative Charges In Pensions Salvador Vald6s-Prieto October 1994 E. Khine In Chile, Malaysia, Zambia,and the 37471 United States

WPS1373 Firm Behaviorand the Labor Market Simon Commander October 1994 B. Keller in the HungarianTransiffon Janos Kollo 35195 CociuiaUgaz

WPS1374 Infrastructure Finance: Issues, Anand Chandavarkar November 1994 M. Geller Institutions, and Policies 31393

WPS1375 Policy Lessons from a Simple Shantayanan Devarajan November1994 C. Jones Open-Economy Model Delfin S. Go 37699 Jeffrey D. Lewis Sherman Robinson Pekka Sinko

WPS1376 How Trade, Aid, and Remittances Maurice Schiff November1994 J. Ngalne Affect Intemational Migration 37947

WPS1377 Macroeconomic Adjustment to Vittorio Carbo Novenmber1994 A. Vo Capital Inflows: Latin American Style Leonardo Hemrndez 31047 versus East Asian Style

WPSI 378 Mexico after the Debt Crisis: Daniel Oks November 1994 S. Harbi Is Growth Sustainable? Sweder van Wijnbergen 37143

WPS1379 Financing Infrastructure in Developing Barry Echengreen November 1994 WDR Countries: Lessons from the Railway 31393 Age

WPS1380 Transfers and the Transitionfrom Kathie Krumm November 1994 K. Krrnm Socialism: Key Tradeoffs Branko Milanovic 34263 Michael Walton

WPS1381 Welfare Economics, Political Ravi Kanbur November1994 M. Youssef Economy and Policy Reform in Ghana 34614

WPS1382 Saving, Investment and Growth in Klaus Schmidt-Hebbel November1994 E. Khine Developing Countries: An Overview Luis Serv6n 37471 Andr6s Solimano