Public Disclosure Authorized

POLICY RESEARCH
WORKING PAPER

1215

Public Disclosure Authorized

How Much Do Distortions Affect Growth?

Wil'liam Easterly

Contraryto the traditona

view that distorions of

relative priceshave modest

effects, Easterlyfinds that distortions can greatly affect

growthandwelfare.

Public Disclosure Authorized

Public Disclosure Authorized

The WorldBank Policy Research Departrnent andGrowthDiviLsionI Macroecononmics November1993

PouIcy

RESEARC(H WORKIN(.

PAI'ER 1215

Summary findings
Fasterly prsecrits a simple encogcnoos growth moiodel are hiwnired, no mnatterhow highl the -att of'

(with two tYpes of capital) that showssthe sizable long.

runi effects on growth of distortionary model afpplies to manly dLifferent types relative prices conimmon in developinig exanipe, price coritrols, black market and differential taxes and tariffs. policies. 'the of distortions of tor cotiintries exchange

distortion.
Suobssidizing inp uts and investment goods canir incrcase gro .th, even thxotughi it worsenis wclfarv. But a suhsidy to witt capit;ul gookl fintauncetd hy a 1tx o redixc growthi ainahmbiguously another )ratcs capiral good resuilts show strong negative effects f'ro Inopirical

'I'he mnodelshows that distortions of relative input prices can greatly affect growth and welfare. I'hl
magnitude of tht eCffectdCepends on the plroduLctionl elasticity of substittition. With high -UsStitUtabiliry, te effects on growth and welfare, although possibly large,

variance in thle relative prices of insestniniit goods across sectors, shile also cotifiriniig and extenrding earlier result. showing tihatpenalizing investment
g)ods and disituorting finatncial markets reduce growth,

This paper- a producr effort in the department Support B3udget under Copies of this pa!per are Rebecca Martin, room

vResea rc h )epartrmienrt-- is pa t of a larger Growth D)ivision, Policy of the Macroeconomics antid to rclate long-riun growth to national policies. 'I'h stIdy was funLded hy the Bank's Research the researclh project "How Do National PoliciCs Affect Long-Run Growth?" (RPO 67f-66). available free from thb World Batik. 181 8 H Street NWV,Washington, DC'20433. Please contact N1 1-0.35,extensiol 39065 (3() pages). NoVemrber 199 3.

Fe'oPlwc

Research ,W rking Paper Sf'ies

Jisss -inate5

tle

tind;ogs

'

in mi pk

s tu e'nc'

,ofa

tlr efxcht aec

-,

ileas ca

b,it

development issues.An objective oftthe series is ,., get the lindvigs out 1i4i,kl) cu

i thc ie'i 1/"esentations are les1tOwn ul//v polisred, I h,e

he fndings, interpretations, ,7nd con(lusi 'ns are the| papers carry the nates of the authors and s*ou,l bleused and cited accordlingI. J authors' own and should not be attributed to the World Bank, its F-cecuti:e Board o! I)!rtr tors, or an) of !ts memb cr countries

IProducedby the Policy Research Dissemination Center

torthcoming, Journalof Monetary Economics

HOW MUCH DO DISTORTIONS AWECT GROWTH? WilliamEasterly

World Bank. WashingtonDC 20433

correcpondence to: William Easterly, Room N-I 1-043, World Bank, 1818 H St. NW, Washington DC 20433 USA

'I am grateful to participants in presentationsat the NBER Workshop on Economic Growth and at the World Bank and to an anonymousreferee for useful comments.

1. INTRODUCTION The economicliterature on developingcountries has long containedargumentsthat the large growth differentialsamong these countries could be explained by policy differences. Policies that directlydistort relative prices and resource allocation - such as tariffs, import quotas, controls on prices and interest rates, and discriminatorytaxes - have been particularly singled out to explain differentialperforrnance of developingcountries (Balassa(1985), Krueger (1978, 1983), Chenery et al. (1986)). However, according to the standard neoclassicalgrowth model due to Solow (1956), distortionarypolicie- affect only the level of income (and that modestly as dramatizedby the famous Harberger triangles) and not its rate of growth. The alleged long-run growth effects of distortionar policy have also been dismissed in some "new" growth models (Lucas (1988) and Young (1991)). in explairing large The natural conclusion is that distortionarypolicies are not that imnportant differencesin economicperformance (Rodrik (1991), Sachs (1987)).' In contrast, authors such as Romer (1986, 1989), Barro (1990), Barro and Sala-i-Martin (1993), Jones and Manuelli (1990), and Rebelo (1991) have presented models in which policies have significanteffectson long-run growth. For example, in the area of fiscal policy, Barro (1990) has discussedhow tax rates can distort savings decisions and lower growth, while governmentse-vices financed bv those taxes can potentiallyraise private productivitv and increase growth. A recent body of work has calculatedthe quantitativeimpact of tax policy on growth rates. with results that range from small (Lucas(;990)) to large (King and Rebelo (1990), Rebelo and Stokey (1993)) to very large (Jones. Ma.nuelli.and Rossi (1993)). However, the endogenousgrowth literature has so far not givenmuch attention to the distortionsof resource allocation commonly induced by policies in developingcountries. including
Thegrowth effects of distorionary policiescouldbe preserved in the Solow modelif we focused on the trnsition to thestdy state. explain very l1t1.of the long-rungrowth demcnstrated that theSolow traition can Rebelo(1993) have convsncingly However,King and rolefor implicationsfor the interet rte. Mankiw. Romer, end Weil (IM) genete a larger rate withoutgeneratingcounterfactual transitional dynamicsby rupposinga high capitalshareincludinghuman capital.

2 both tax and nontax policies.' Tax systems in developingcountries often have a very narrow base, becauseof widespreadtax evasion and the small size of the formal sector (World Bank (1988)). Generationof revenue from this narrow base often implies very high tax rates. A few examples help illustratethis. More than 80 percent of income is .id to go unrepo.ted in Argentina (ibid.). Employmentin the privat^ formal sector in Cote d'Ivoire amounts to only 1.4 percent of the population. Repeatedattemptsto increase tax revenue from the formal sector in Cote d'Ivoire have :net with failure, as there is large-scale evasionof taxes - which have an average effectivetax rate of 48 percent (Chamleyand Ghanem (1991)). Relianceon taxes on internationaltrade is heavy in developingcountries (particularlyin Africa where it amountsto 35 percent of revenues), implying inputs in production (World Bank (1988)). Although that the tax system penalizes the use of imporLed tax distortionsare also significantin industrial countries (King and Fullerton (1984)), these examples suggest anotherorder of magnitudefor deveiopingcountry distortions. Besidesexplicit fear'ires of the tax system, there are numerous oth ways that governmentsin

developingcountries interveneto create large distortionso. resource allocation. In additionto explicit taxes on trade, there are a host of other trade interventionsfrequently used, such as import quotasand licensesand multipleexchange rates, which imply distortionsof prices of domesticversus foreign goods,as well as among different types of foreign goods (World Bank (1987)). Foreign exchange controls lead to the creation of blacKmarkets in foreign exchange, where the exchangerate is a large multipleof the officialexchange rate. Prior to reforms in Ghana in 1983. for example. the black market exchangerate was 44 times the official rate (Islam and Wetzel (1991)). Government marketingboards pay prices to domestic producersof export commoditiesthat are a small fraction of the international price.

:An exception is Murphy, Shleifer, and Vishny ( 1991), which shows how rent-seekng in the presence of distortionscan significantly lower growth

3 Controls on nominal interest rates often result in large negative real interest rates, while quantitativecredit allocation rules imply large subsidiesfor some types of capital and heavy implicit taxes on others. For example, credit for mortgage lending in Turkey was neavily subsidizedin the early 1980s,so that these borrowers paid highly negative real interest rates, while commercialcredits for other types of loans carried real in!erest rates as high as 50 percent (World Bank (1989b)). However, a large informal sector has sprung up side by side with the formal economy in many developingcountries in response to distortionarygovernment policies. De Snto (1989) has convincinglyshownhow government regulationled to a thriving informal sector in Peru. Over half sector (World Bank (1989a)). The illegal of urban employmentin Africa is said to be in the int.-,rmal economy in Ghana in 1982 was estimated to have accountedfor 32 percent of GDP (Islam and Wetzel (1991))The informalsector often seems to provide a cushion to soften the impactof even the most policies. The depredationsof the Amir. government in Uganda caused formal extreme govermment sector output to fall at 2 percent per year in 1970-78,but subsistenceproducticn continuedto grow at 3.4 percent per year (Reynolds(1985)). The multipledistortionsdescribedhere cften cause severe misalignmentof relative prices. For example domesticprices of some goods will be driven up through exorbitant tariffs or restrictive quotas whileothers receive preferential tariff or quotatreatment, or are held down by domestic price controls. The model in this paper will show that the distortion of compositionof the aggregatecapital stock resultingfrom such interventionscan have large growth effects, just as the tax-induced distortionof the ratio of physical to iiulnan capital affe.ctsgrowth in Jones, Manuelli, and Rossi (1993)and Rebelo and Stokey (1993). The price distortionscan be measured di-ectly with the price data collectedby Summers and Heston (1988). This paper will commodity-by-comnodity proposea measureof relative price distortion relevantto long-run growth determination.and empiricallyexamineits effects on growth.

The paper is organized as follows. Section II will present the basic model relating distortions to growth, and show how it covers a variety of commondistortionary policies. Section m will present empiricaltesting of some of the implicationsof the mode', deriving a new measure of the severity of distortion of the price system. Section IV cancludes.

POLICEES [I. A MODELOF ENDOGENOUSGROWTHWITH DISTORTIONARY This sectionpresents an endogenous growth model that attempts to capture some of the special features of distortionar:,government policy in developingcountries. It postulates a modelof constantreturns to scale in reproduciblecapital, as in Rebelo (1991) and Barro (1990).3 Canital is broadly defined to include both physical and human capital. Two generic types of capital are defined io producethe single output - the types will be initially interpretedas formal and informal stor capital. The distortionwill be defined initially as a sales tax on investmentpurchases in the formal sector. The relationshipbetweenthis distortion rate and growth will then be discussed. Applications to other policies will then be considered. Throughout the paper, we assume that the economy is insulatedfrom internationalfinancial markets, although trade in goods is considered. While the model is highiy stylized, the results seem to give insight into some of the developingcountry characteristicsdiscussedabove.

The Model Equation(1) shows the production function for our analysis:
1
(1) Y = A (y K1 + (1 -y) K2 )

'Tis type of modelcan alsobe seenasa converuent approximaution to a Solow model with a very high share of capital including hnan capital), i:l which policies have suchlarge level effects that they appearto be growth effects. Consunt returns in reproducible capita.allowsthe denvaton of a rteadv statethat approximatea the transcuionalbehavior of a Solow model with high capitalshare.

5 Output is a CES function of the two generic types of capital, with elasticityof substitution l/(e-1). For simplicity, we assume populationto be fixed throughoutthe analysis. .aitially we interpret the two types of capital as being formal and informal sector capitil, which are defined in terms of their location, ownership, and visibilityto the tax authorities. We assume that the two types of capital can both be formed fTomthe domestic good with zero installation cost, and that the economy is closed both to foreign trade and capital flows (in the next section, we will show how some alternativeinterpretationsao not change the model). The distortion is a sales tax on formal sector investmenm purchases, which informal sector investmentpurchases evade. We assume that ideritical,infinitely-liveddynastiesof producer-consumersmaximizethe present discountedvalue of the utility of future consumption:
(2)

e -Pt

cl

-a

Utility is given as an isoelasticfunction of total consumptionC where the intertemporalelasticity of substitutionis I/a. Consumptionis given by the excess of income over investmentspending:
(3) C = Y (1+r) I12 -

12

+ T

where r is the rate of sales tax on purchases of investmentgoods of the first type. The variableT is the lump-sumtransfer of tax revenues back to consumers, ex-post equal to -r1, but treated by the consumeras fixed. Growth is unaffect, !, however, if the revenues are not rebated to the consumer. The equationsof accumulationof capital are:
(4) K1 = I6K

(5)

K2

12

where 6 is the rate of depreciation,ass_med equal across capital types to simplify the algebra. We do not explicitly imposethe conditionthat investmentis irreversible.

6 The t. t-order conditionsimply rhat the distortion will act as a wedge betweenthe marginal products of the two types of capital, as follows:
(6)+76)
= 2

~1

where T is the rate u.ax on investmentin type I capital. This implies the followingratio of type 2 to type 1 capital (denoted '1), from (1) and (6):

K2
KI
=

(1*Y)

(1+7)

The distortionr inducesmore of type 2 cnpitalto be held relative to type 1 capital than is socially optimal. The elasticity of subs.itution(1/(e-1))determineshow strongly the ratio of capital inputs will respond to increasesin the distortion. In steady state, consumptionand output will grow at the same rate, which will be given by:
r2 -j_p

(8)

g --

where r, is the marginal product of type 2 capital, given by the following:

(9)

= A (I-y)

(y 4 6+ 1-Y)

and 4)is the ratio of K. to K, given by (7). Growth is giver. by the familiar conditionthat optial growth of consumptionis equal to the difference between the net marginal product of capital and the rate of time preference. times the intertemporalelasticity of substitution. Since the marginalproduct of type 2 capital goes down when the ratio of type 2 to typt I capital increases, which occurs when type I capital is taxed more heavily, growth is unambiguouslva negative functionof the tax rate r. A policy equivalencethat follows immediatelyis that between a sales tax and an incometax. An incometax on the income from formal sector capital that is evaded by holders of informal sector

7 capital will enter in an exactly analogousway as above (the income tax rate t equivalentLo a sales tax
7

is t = r/(l +r)).' The nature of lie relationshipbetweei.growth and distortion depends on the production

elasticityof substitutionbetween the two types of capita'. Figure 1 shows a simulationof growth rates
5 The graphs are shown in terrms for plausibleparameter values with alternativeproduction elasticities.

^f the equivalentincome tax. With a high elasticity of substitutionin prudaction, even an income tax
rate close to unity could still permit positive per capita growth. This is because when the elasticity of substitutionis greater than one, neither input is essentialto production. As the tax rate approaches one, investrnentin typu I capital asymptoticallyapproacheszero. However, production is still possiblerelying only on type 2 capital, and so the rate of return to type 2 capital remains positive. W'hilea high elasticity of substitutionmay seem implausibleir a model with only two inputs, it is more prlausible in a multiple input production function where many types of taxed capital goods will have close substitutesin the informal sectoe. A high elasticity of subsLitution betweeninformal and fcrmal capital could give insight into why even highly distorted economies could have less than catastrophicgrowth performancc (such as Amin's Uganda, as cited earlier). If the elastic..y of productionis less than or equal to one, then the level of output and the rate of return to type 2 capital go to zero as the tax rate approaches one. As the tax rate goes to unity, economicagents will decumulatecapital (assumingno irreversibility in investment)at a rate
((6 +p)/a).

Growthalso falls off more steeply at lower levels of the tax rate, as shown in figure 1.

There is an analogouseffect of the elasticity of substitutionon the welfare loss associated with the distortionarytax. Figure 2 shows how the welfare loss is small at first, regardlessof the

'this equivalence is noted in another context by Atkinson and Stiglitz (1980). nThe parameter values used for the simulationare A= .5. d .05. y = .5, =2/3 = 2. and a = I. These values are arbitrarily chosen to reproduce plausible ranges of growthrates for illustrative purposes. The shapeof the growth-distortion relationship is robust to changes in these parameers except for the production elasticity of subsitution (as discussed in the text) and the intertempormlelasticitv of subritution (not discussed here, sinceit is well known to control the str-ngth of the saving respons. t t.L changes).

8 elasticity.' This follows from the fact that a laissez-fairezero tax policy is first-bes,zwhich must imply that the der,vativeof welfare with respect to the tax rate is zero at a zero tax rate. As the tax rate increases, welfare 'ils off more sharply in the high elasticity case, then bottomsout at a finite level as the tax rate goes to one. This occurs because the taxed input is inessential,so even eiirrmination of the input's use only leads tO a finite, albeitsubstantial, welfare loss. As the use of the inputapprcacheszero, the marginal welfare loss of a higher tax cn that input is minimal. In the low substitutabilitycase, the welfaru loss is not bounded since the taxed input is essential for production. This idealizedframeworksuggest that one's strategy in a highly distorted economyshould be influencedby one's assumptionabout substitutabilityof inputs in the production function. If one believes substitutabilityis high, one should concentrateon large reductionsin large distortions, sincz moderatereductionshave little benefit - a "Big Bailg" is needed. If one believes substitutabilityis low, then any reduction in a large distortion will be beneficial - "gradualism" is a viable strategy. It is also apparent from the maximizationproblem above that a sales tax that applies only to
7 With exogenouslabor supply, investment consumptionpurchases will have no effect on growth.

price distortionslower growth, but consumptionprice distortionsdo not. Since this result will be tested in the empirical analysis, it is worth examininghow it compares with others iL the literature. Consumptiontaxes that are rebated lump-sumto consumers affect growth in the endogenouslabor supply modelsoi Rebelo and Stokey (1993) and Jones, Manuelli, and Rossi (1993) (with Lucas

'Welfare in steady

t can be calculated as:

- a

Ip-g(i-a)

where K, is the iniual level of the total capital stock and4 is the ratio of consurnption to total capital stock, given as:

+ K_ C,

(a

It can be shown,however, that taxes on intermedistegoods act much like taxeson investment goods and aode affect ::rowth.

9 (1988, 1990) "effectivelabor"). However, Rebelo and Stokey (1993) show that consumptiontaxes do not affect growth if they are not rebated to consumers.' The fall in growth as a result of the tax on capital is similar to results obtained by Barro (1990), King and Rebelo.(1990),and Jones and Manuelli (1990). What is different here is that the fall in growth occurs not only through a fall in ; vestment, but also in the efficiencyof investment. An increase in the distortionarytax rate will lower growth even for a fixed rate of total investmentto output (definedas the sum of the two types of investmentas a ratio to output). Rebelo and Stokey of tax policy on the (1993) and Jones, Manuelli, and Rossi (1993)show related compositioncle&fects ratios of human to physical capital. It is also apparent from (7)-(9) that a negativevalue of r, i.e. a subsidy to investmentin type I capital financed by lunp-sumn taation, will raise growth. Although lump-sumtaxation is unrealistic, this result would also occur with a subsidy to one type of investmentfinancedby a tax on consumption,since the latter has no growth effect. This is an example of the well-knownpossibility growth,.since welfare is worsened but growth increasesdue to the subsidy. It is still of imrniserizing true, however, that a higher subsidy will lower growth for a given rate of saving and in,,tsnment;the increase in growth comes aboutthrough higher investment,partially offset by lower efficiencyof investment.

Self-financing subsidy and tax schemes Thus far we have consideredonly policies that implicitlytax one form of capital. assuming the proceedsare dissipated in lump-sumtransfers. Other policies have the character of taxing one

'This result is a consequence of the restrictions on the utility funcuon rcquired to generte a tAdy oute growth path with a constnt fraction of time devoted to leisure, which requires that income and substitution effects of wage changes cancel out, as pointed out earliie by King, Plosser, and Rebelo (1988). Rebelo and Stokey (1993) also show that if leisure is *quality-adjuated (i.e. multiplied by the level of human capital), then consumptiousixes do not affect growth regardless of whether the taxes are rebated or not.

10

type of capital and using the proceeds to subsidizeanother. For example, price controls act as a
9 subsidyto the use of the controlled-pricegoods, implicitlyfinanced by taxes on other goods.

Using the framework of the previous section, we now assume investmentin capital type I is taxed at rate r, and that the proceeds are used to finance a subsidy to capital type 2 at rate s. The first-ordercondition (7) for the allocation of capital will be modified to be: (II)
1... ~ r1~K [ [

) (

1= 1 ry-I I I, sl I]

The requirementthat the scheme be self-financingimposes the condition: (12)

2=

For a given subsidyrate s, (11) determinesthe allocation of capital betweentypes and (12) determines the tax rate necessaryto finance the subsidy. The growth rate expression (8) will now be modified to be:
:2

-p

(13)

ls 1=

An increase in the self-financingsubsidywill have two offsetting effects on the growth rate. The increase in the subsidy makes accumulationof type 2 capital more attractive. and tends to raise growth, as in the previous section. However, the before-subsidymarginal product cf type 2 capital will fal: becauseof the substitutionfrom type 1 to type 2 capital, which is inducedby the subsidy and by the tax necescaryto finance the subsidy. AppendixI shows that the second effect always outweighsthe first: the net effect is that growth unambiguouslyfalls with an increase in the rate of subsidyfinanced by a tax on capital.

'Mis is presumrng that the pnce controls are accompanued by budgetary transfers to compensate the producer for the lower pnce. Otherwise, shorages and black market pnces anse.Justasin the case of foreign exchange controls (6ee below).

11 Other policy applications The analysishas been presented thus far in terms of a sales tax or subsidy (or equivalent incometax) on formal sector capital in the presence of an informal sector. We now show some other examplesof policies that give similar results, illustratingthat the framework is general enough to considera range of distortionary policies. 1. Financialrepression Controls on interest rates below market levels generally lead to quantitativecredit allocation by the formal financialsystem, combined with a flourishing"curb" market formed by consumersor producerswho lend to each other at market rates. The quantitativecredit allocation rules are of en designedto promote certain kinds of activities, for example, small-scaleenterprises or agriculture (WorldBank (1989b)). In terms of this model, we can ti 'k of a credit subsidy being given to all who invest in a particular kind of capital goods (type 2 capital). We can assume that the subsidy is financedby an implicit tax on the non-subsidizedcapital (type 1 capital). 2. Tradepolicy We could next redefine the two types of capital to be imported capital (K,) and domestically producedcapital (KY),and redefine the tax rate r as a tariff on imported goods. We assumethat the domestic good is also traded internationallyand that the country is a price-taker (the international price ratio of the two goods is fixed at unity for convenience). The tariff then acts as a tax on type I investment goods. Equaily, both types of goods could be imported, and the tariff structure imply a tax-cum-subsidy scheme in favor of some types of investmentgoods. Thus, both a higher level and a greater dispersionof tariffs (or equivalent quotas) will result in lower growth according to this model. 3. Exchangerate policy Whenexchange rate controls in developingcountries lead to black markets in foreign exchange, foreignexchange is typically allocated at the official rate so as to favor certain activities.

12 In terms of this model, a foreign exchange allocationthat grants foreign exchange at the official rates for goods that make up capital type 2, and does not grant any for investmentin type 1 goods, will imply a tax-cum-subsidyschemewith the wedge betweenthe two goods equal to the black rnarket exchange rate premium. 4. Price controls Price controlson particular goods are a common feature of developingcountry policies. For example,coal prices are often controlled to be below internationallevels in many countries (World Bank (1988)). The coal subsidy is financed by other revenuesfror- the govemment budget. This is equivalentto the tax cum subsidyscheme discussed in the previous section. Since distortions2. and 3. above also affect the price system, this suggeststhat a comprehensivemeasure is needed to capture all of the complicatedand possiblyoffsetting relative price distortionsinherent in government policies.

M. EMPIRICAL RESULTS
Althougha large literature exists relating policy variables to growth, there have been few attemptsto measureprice distortionsdirectly. This section proposesa new measure of relative price distortionand tests its relationto growth, along with other distortion measuresused in other studies. We discuss first the variables, and then discuss the results.

Variables 1. Price distortions One way to measure indirectdistortionsof prices, like those analyzedin the previous section, wouldbe to look directly at observed retail prices across countries. Summersand Heston (1988) collected data on 1980 benchmarkprices relative to the U.S. for 57 countries and 151 commodities.

13 We construct a measure of the varianceof relative prices across commoditiesfor each country and interpretthis as a measure of overall distortion of the p;ice system. There are two adjustmentsthat need to be made. First, the variance of each commodity's price across countries due to differences in natural resourceendowmentsneeds to be removed. Although theoreticallythis is necessary only for nontradedcommodities,even traded commodities' domesticprice may contain a nontraded component due to retailing and transportationcosts. This is done by isolating the orthogonal component of each of the log of the 151 prices with respect to per capita income. Secondly,the distortionsin prices of intermediateinputs and investmentgoods matter. We i ii.; predicts that only implicationof

the theory by computingseparatelycross-commodityvariances for consumptioL ,oc ' and for goods (humancapital inputs such as health and educationwere al'o included investment/intermediate under investment- Appendix m containsthe list of commodities).We use the same data to compute mean of the orthogonalizedlogs of relative input/investmentprices. We the expenditure-weighted sum of squared deviations of each then compute for each countrythe expenditure-weighted commodity's orthogonalizedlog price from the log mean.'0 AppendixII shows the variance and mean of input prices for each country. The countries with the highestvariance of relative input prices in 1980 are Sri Lanka, Venezuela, Peru, Bolivia, and Indonesia. All of these countries partially compensatewith cheap prices for inputs and investment goods. Cote d'Ivc.re, Nigeria, Senegal, and Carneroonhave the highest average input prices. Industrialcountriesgenerally have low distortions of relative input prices, and neither very high nor very low average input prices. Table I shows the mean and variance of these and the other distortion measures.

to the clever procedure followed by Aitken ( 1991) to compute a measure of distoruon. However, Aitken aggregates This is aimrilar commodities into 16 categones before computing cross-commodity vanances, he does not distinguish between congumpuon andinvestment goods. and he does not relate his index to growth. I am grsteful to Brian Aitken for supplying the Summers-Heston price data in convenient format.

14 Table 1: Descriptivestatistics for distortionmeasures Mean across input categoriesof log of ratio of inputprices to U.S. price (orthogonal componentto income), 1980 Varianceacross input categoriesof log of ratio of input prices to U.S. price
(orthogonal component to income), 1980

Sample mean -0.230

Sample standard deviation 0.458

Number of countries in sample 57

0.241

0.158
_

57

Mean across consumptioncategoriesof log of ratio of consumptionprices to U.S. price (orthogonalcomponent to income), 1980 Variance across consumptioncategoriesof log of ratio of input prices to U.S. price (orthogonalcomponent to income), 1980 Log of ratio of black market to official
exchange rate, average 1970-85

-0.160

0.333

57

0.230

0.149

57

0.194 -3.58

0.251 8.95

106 53

Averagereal interest rate, 1970-85(used to define financialrepression dummy)

15

Earlier results for growth using the Summersand Heston price data were found by Dollar (1992)and De Long and Sumners (1991). Dollar tested a measure of exchange rate overvaluation constructedas the average price level in each country relative to the US, corrected for differences in resourceendowments,and found overvaluationto significantlylower gru..th. De Long and Summers found that high relative equipmentprices significantlylower growth. 2. Financialdistortions We extend a dataset from Gelb (1989) of real interest rates for developingcountries, adding observationsfor developed countries. A highly negativereal interest rate is interpretedto imply the existenceof nominal interest rate controls (moderatelynegative real interest rates have been observed for reasonablylong periods even in countries with uncontrolled interest rates). We follow World Bank (1989b) in defining "financialrepression" as an average real interest rate below -5 percent over 1970-85'period varies depending on data availability)- there are 14 such countries (out of 53). We will examinerobustnessto other cutoff points as well as to the use of a continuousvariable. Gelb (1989) found real interest rates to be significantlycorrelated with growth. Fischer (1983, 1991)and Kormendiand Meguire (1985) report a possibly related finding of a negative effect of inflationon growth. 3. Black marketpremium The black market premium is defined as the log of the ratio of the black market to the official rate over roughly 1970-85(period varies with data availability), based on data from Pick's Currency Yearbookand the World Bank. Levine and Renelt (1992) found this variable to have a significant effect on investment,althoughnot on growth, Easterly (1991) found it to be a good predictor of whethercountriesare caught in a low-level equilibriumtrap.

16
4. Governmentconswnpnon

Governmentconsumptionhas been found to be negativelyrelated to growth by Barro (1991), Grier and Tullock (1989), Landau (1986), and Murphy, Shleifer, and Vishny (1991). In terms of the model in section II, govermnentconsumptionis an indicator of high taxes on the formal sector. 5. Barro correlates In additionto these measuresof distortion of resource allocation, we also test for correlates reported by other studies, taking Barro (1991) as the benchmarkstudy. We focus on initial income, primary enrollment, and secondaryenrollment. We also use the same dependent variable (per capita growth 1970-85from Summersand Heston (1988)) as Barro for ease of comparison. Finally, the investmentrate (also from Barro/Summersand Heston) is added to the equation to test the implication of the model that distortionslower growth concrollingfor investment.

Empirical result

Table 2 shows the results. Both the mean input price and the variance of input prices across commoditiesare significant. The variance result is consistentwith the prediction of the model in section II - in a multiple input world, distortionsof relative prices away from world prices will lower growth. The magnitudeof the variance coefficient is such that increasingthe variance of relative prices from the sample mean of .24 bv one standard deviationto .40 will lower growth by 1.2 percentagepoints. The significanceof the mean input price is reminiscentof the De Long and Summers(1991) findingthat cheap investmentequipmentis good for growth. Here we find more generally that cheap inputs -- includingintermediateinputs, education,health, buildingsand equipmentinvestment- are good for growth. This result confirmsthe predictionof the model that subsidizinginputs is a feasible way to raise growth (althoughirnmiserizingunless there are externalities). When the equipmentprice

17 variableis added, neither price is significant-- we cannot distinguishbetween the two - but the v?rianceof input prices continuesto be significant(regression VI). The financialrepression dummy is also highly significant, with a coefficient indicatinga loss of about 1.5 percentage points in growth due to financial repression. Th does not appear to be a proxy for high inflation, which was insignificantwhen tested separately. The implicit taxes from highly negativereal interest rates apparentlyfall on productive investment. If we accept regression I as a valid reduced form, these three distortion variables (with governmentconsumption)explain half of the cross-sectionvar dz,.e in growth rates. The financialrepression dummy, and the mean and variance of input prices continue to be significantwhen investmentis added (regressionsII and EII). Growthfalls for a given rate of investmentfalls when relative input prices are heavily distorted, when inputs are expensive, and when interestrates are repressed. The significanceof the financial repression dummy and the variance of inputprices confirms the efficiencyeffects predicted by the model in section II. The significanceof mean input prices does not at first seem to conform to the prediction of the model that a uniform tax on all inputs should affect the level of investment,but not the efficiencyof investment. However, since investmenthere includesonly physical capital, the result may reflect higher productivityof physicalcapital with cheaper intermediateand human capital inputs. Becauseinvestmentis likely endogenous,we show also a regression (Ia) with an instrumentalvariables estimator, using as instrumentsfor investmentthe other right-handside variables as well as continentdummies for Africa and Latin America, a dummy variable for oil producers, and the populationsize in 1960. The coefficienton investrnentis increased, but the coefficientson the other variables are roughly unchanged.

18 Table 2: Regression resulls, growth and distortion

Lkp.nkll
H,

oAruabk P.. 'ap,il

V-r.bk ~~....

rr.,*,
A-, ISS
-V.

9 le
P.-

9 rss85 (Srunwis. IudJ l*cskail(88))

.4
3

V-e.e o.

*60k.g
ph

.4
l

U-n-,e owsr.c
C

..
C

1 Vsjinumr'
mo -014

hn-...

37-.

be.uml

M-eo
P-

or.ass..

*9*0

V.16
p

repr I 0041 (6 14)

-. pn.e

wymqopel
UUSA

of *90

?A M pDC

.. *91100
Clow S

Psz9eIO

aMe
p.ewml

a".e

*a.1.1
IYeO

IYttO
bt.

Vao*o.a 140

*nn ..

d 19YOIddIYO O. -a
1940
ifto

o-

*)Yh(*AI2

N-..rA,.4

-0024 (2 67)

0a071 (-2 I5)

4-04

01 2

(-1.3)

3*

( 2 00)
0 46

0016 I0 (I 3)

0026 ( 305)

a0065 (-I 91)

.0301 (076)

0016 (231)

0092 059 (264)

3i

0s3
III Vio 0 010

-0024

-0066

0010
(I 0)

4)016

4n
( 3 19

0046 W3

40641
(2.18) (2 36) (2

05^4 993
0$4

38

1l)(14SiS)

0i1Y (a51)

-0021 (404M)

a034 (226

4-0017
(2571)

016 035 499) 045 30

IV

0023 O (
47

009 ( 3 11

4-004 ( 2 0)

-021 (0 75)

4000 (I I5)

4005 (0 7*)

-0006
(-23)1

0036a
(l 6

0035 (226)

a00 0 49

Iv. 4u' .uh

(I l .

.* cpi.-d m

0027 (201)

-01 (331)

-04 (1Y1)

-0131 (499)

iu)6

4021 (107)

4).006 (-275)

0.01 (175)

0036 (241)

63 52

36

(244)

0O2l (61)

0032 (-2 19)

4a 19 (-3 54)

a022 (2*2)

0012 (071)

0066 (I 61)

0 57 050

36

VI

00)2 (O22)

a00ld (o 0))

4073 (2.14)

-014 (205)

0013 (4046)4

050I

VIl

0040 (6 47

10 (-3 1(7)

4030 -0-0 012 (2 4) ( 2 6) 038 3

VIb (FI1L1L1 < 2)

040 (6 65)

10O* ( 30 O) isvest*rl is islilumcnlcd iuterecst itac<-S) is with

thc

-00oil (-I I1)

-077 1 47)

038 0 14 America. rate and po7pulaIon < -2. sit

CL

51

Note. In
rtprcsarn T

Rcgresssos dummy
(-

life.

Statistcs All esuhlE r,pomtcd using Wltc

I t r*al in parcfnih.ca

Fcprocdby
-

odler tight hand asdc variabics. a dummy the CLaL .cfcst rate iscltc. In rcgession

flo
Vila,

being an oil produccr. conieacsa dwunics the fiuancial prepssios duinmy is dcIfirid

fos Ainca

and Latin

s I df Me

rcal iboCcrt

icgiersstur

(va

dic

fincia*l

hcIcsvskceJ.l,.ity

cohiuisst covaranoc

iIlaiaccs

19 Table 2 Variabledefinitions: Averageinputprice: Expenditure-weighted average across categories of inputs and investmentgoods listed in AppendixIII for 1980 orthogonalizedlog of dolharprice relative to the U.S. Varianceof input prices: Expenditureweightedvarianceacross input and investmentgood categories for 1980 orthogonalizedlog of dollar price relative to the US. Governmentconsumption: Summers and Heston ratio of real government consumptionto GDP, 1970-85(Barro variable HSGOV) Financialrepression dummy: 1 if country has average real interest rate for 1970-85less than -5 percent, 0 otherwise (periods of average may vary dependingon data availability)[From Gelb (1989) and World Bank datal In regression IVa, the dummy variable is replaced by the actual average real interest rate over 1970-85(notice interpretationof the sign of coefficient will be reversed); in regressionVila, the dumnmy variable is replacedby a dummy variable where the dummy is equal to one if the average reai interest rate is less than -2 over 1970-85 Total investment: Summers and Heston ratio of real investmentto GDP (Barro variable HSINV) Mean consumptionprice: Exper-diture-weighted average across categories of consumptiongoods listed in AppendixII for 1980orthogonalizedlog of dollar price relative to the U.S. Varianceof consumptionprice: Expenditure-weighted variance across categoriesof consumption goodsfor 1980orthogonalizedlog of dollar price relative to the U.S. Mean of equipmentprice: Expenditure-weighted average across categories of equipment investment for 1980orthogonalizedlog of dollar price relative to the U.S. Black marketpremium: Log of ratio of black market to official exchange rate [World Bank data] Initialincome: Summers and Heston real per capita GDP in 1960 (Barro variable GDP60) Primaryenrollment: Barro variable PRIM60 Secondaryenrollment: Barro variable SEC60

Note: orthogonalization is with respect to log of per capita income.

20 Whenwe add the analogousmeasuresof the mean and cross-commodityvariance of consumptiongood prices, we find that they are insignificant,while the mean and varianceof input prices continueto be significant(regression V). This is consistent with the predictionof the model that consumptionprice distortionsdo not matter but input price distortionsdo.'" The significanceof the mean and variance of input prices is robust to the inclusionof correlatesreportedby Barro (regression IV). The same cannot be said for the financialrepression dummy,whose significanceis weakenedwhen the Barro correlate,sare added. We also examine the robustnessof the resu!i on the financial repression variable by consideringalternativedefinitions of "repression". Redefiningrepression as less than a -2 percent real rate of interest increasesthe number of repression cases from 11 to 19 countries in the sampleof 51 (regression VIla). The redefined repression variable has the same sign but generally falls short of significance.On the other hand, a continuousvaria3le for the real interest rate is still significantat roughly the same or higher level as the current dummyvariable. The continuousvariable retains significancein the Barro regression (regression IVa). Note that because of the redefinitionof the variable, the sign here is the opposite of the dummy vaziablebut still has the interpretationthat negativereal interest rates are bad for growth. The continuousvariable may be more likely than the dummyvariableto capture other sources of variation than just the degree of financialrepression, however. We concludethe evidencefor effects of financial distortionson growth is suggestive,but far from conclusive. There is some evidencefor effects of other distortionvariables on growth. The black market premium is significantin an equation with financial repression and governrment spending (regression VlT),but its significancedisappearswhen the mean and variance of relative input prices (and the

"We alsotried Dollar's (1992) measure of exchange rate overvaluation, which is closely related to the mean consumption price, and found it to be inuignaicant.

21

Barro correlates)are added (regressionIV). The same is true for the share of governrnent, whose significancealso disappearswhen investmentis added (regressionO). The strongest results are those for the new measures of price distortion. The results can be illustratedby lookingat three data points: Ch!le. Korea, and Zambia, using regression m as the benchmark. All three had roughly the same investmentrates, around 27 percent of GDP, but Korea had high growth (6%) and Chile (0%) and Zambia (-2%) had low growth. A significant portion of the difference in growth rates can be explainedby regression IH. Chile and Zambia both had high negativereal interest rates (in Chile's case, mainly during the 70-73 Allende regime and its aftermath), which itself lowers growth by 1.6 percentagepoints. Korea and Chile both implicitly subsidizedinput investmentgoodsheavily on averaig, while Zambia implicitlytaxed them - this explains over 2 percentagepoints of Zambia's lower growth compared to Korea. On the other hand, while Korea and Zambia only had moderately distorted prices between types of inputs, Chile's were heavily distorted. The high variance of input prices cost Chile 1.6 percentagepoints of growth compared to Korea. It is interestingthat Korea was not a case of "gettingprices right", but rather one of heavy subsidizationof inputs, which implies growth was "too high" by about one and a half percentagepoints if no externalitiesare assumed.

V. CONCLUSIONS

A simplemodel of endogenousgrowth with two types of capital shows that distortions of relative input prices can have large effects on growth and welfare. The magnitudeof the effect depends on the productionelasticity of substitution. With high substitutability,the effects on growth
and welfare are bounded -- albeit possibly large -- no matter how high the rate of distortion.

Subsidizationof inputs and investmentgoods can raise growth, even though they worsen welfare. But a subsidyto one capital good financed by a tax on another capital good unambiguouslylowers

22 growth. Empirical results show strong negativeeffects of the variance of relative prices of investment goods across sector, while also confirmingand extendingearlier results that penalizinginvestment goods and distorting ;.nancial markets lowers growth.

23 APPENDEI I: effect on growth of a self-rinancirng subsidy-cum-tax policy As outlined in the text, an increase in a subsidyto type II capital financedby an increased tax rate on type I capitalalways lowers growth. To show this, it is sufficient to s'"ow that the tax increase necessaryto financethe increased subsidy is greater than the tax increase that wouldjust maintainthe growth rate fixed (since a tax increase by itself unambiguouslylowers growth,. From (11) and (12) in the text, the balancedbudget requirement means that s and 7 must satisfy:
(A. 1)

The derivativeof r with respect to s from (A.1) (using (A.1) itself to simplify) is:

(A.2)

d.r a s
(I'S _(

(A)
)(t)

i-)

Note th it it (elasticity of substitution > 1), then as the tax rate approaches(1-e)/e, the further increase is= rate necessary to finance the subsidy goes to infinity. With elastic substitution possibilities,i,h!'reis a limit on the subsidy rate that is feasible under a subsidy-cum-taxscheme. If . s 0, then any subsidy rate less than one is financeable.
-0

From (3) in the text, the combinationof 7 and s that yield a constant growth rate g is given by: (A.3) E2k

(t ts}

+prg

where the {} denotes "functionof ". The right-handside of (A.3) is constant by assumption;g is the growth rate cnsistent with the initial J and s. The derivativeof T with respect to s from (A.3) can be radicallysimplifiedto:

(A.4) dt

ds

where we use the fact that

k2 =
k'

/s

at the initial point. Comparing the derivativesin (A.2) and

(A.4), we see that (A.2) is unambiguouslylarger. This says that the tax increasenecessary to raise the revenueto finance a given subsidyrate increase is larger than the tax rate increase that wouldjust maintainthe growth rate constant. Since an increase in the tax rate for a given rate of subsidyalways lowersgrowti. this says that an increase in the subsidy rate on type II capital financed by an increase in the tax rate on type I capital unambiguouslylowers growth.

24
Appendix 11 Measur of distorton by country
Country Code AXO
AUS

Di

my for mal infgrt l.=

rats < -5 pereerl

VAannce of log input pnice acrerw c0todiues 0.30340 0.11637 022637 0A.M 0.2090 0 29419 OM.Ml7.1E 0.449D 0.24175 0.339 02746 0.21J177 O.m=24 0.0 oM7e 0.26l91 0.10100
0.0913ii

Averqe of log mput pnrc 0.11346 02121 0our62 040131 4.0741 437146 40.9X0 0.51117 03477 4. 9 4.54929 0.21025 02m
4oJ717

O.OCOO

AVT BEL
BCD

O.OC 0.C
O.OODO

SOL SRA BWA CAN C1L clv CMX COL CR DEU DNK

1.00 O.OODO 0.0OO l.COD O.OOD 0.om 0.OO 0.wOD l.OODO O.OX 0.CX 0.0m
0.OOC

DOM
Wu
EOY

.0.7 47400
4019

ESP
Er

FIN FtA
GOX

0.1541 0.01334
O.OC92

4.12514 0.16t2
0.14W

OKA CRC
urM

l.OX 0.00

0.11019
0.25904

.O.1I0S
40.7559

KKO HND
HVO

0.0CO
O.OCCO

0.3ut 0.21711 0.45X4 02197

4.34221 -0.49242 4.74D11 .0.43C

mm tND

0.D O.OXO

rrA
IAM mmD )ax KOCt LKA
LSO

0.om Lam 0.= 0.0Z20 0.00 O.OD

0.11 0.I41 0.12181 0.21275 o g.sw 0.06M 0.24763 0.2X91i 4.14 0 am 0.19747 0.211W 0.ml13 0.18O 0J8I39 0.22161 0.5377 0360 0.102 0 3757 0 17139

4.61162 4.1s3 0ff" 43329 *.729 0m07541 OIM

LUX .Ait Moo

Mt#

O.OCO
I.000

MU MWm MYS NOA NLD NOR

OMN

0.o OODO 1.01 0.000 O.DDD

O.OOCO

0.19"31 4AM4 0M3 4 0.23171 0.1539 L12 4.40 4.9179 41m9O 40323 4.32105 0.46117

PAK PAN PER PfL POL PIT PRY SEN

SCP SLE

0.Ww l.OODO O.0OO I.OD

I MM

SLV
SWE rro O ODDO 0O.ODDO

0 0946

4.71495

T1N nm 77A UiRY USA VEN ZAR ZMs ZW!E

0.23726 1.0= 032011 0 221:1 0 04179 0 7DI

4.11C9i 022443 4354 4 044 4.01041 0.24114 0.16105

OOD 0.om I.a= I, o.mm 0om

022341

25
.kppendix m Goods for Calculation of consumotion cnd input price averates and varisence
ConsumnpDoo Goods lnatenediast laput andLnveazmeat Goods Rice Meal, ithercereals Bread Biscuits.cakes. etc. Cerel preparation Mc.aroru. spaghetti Freshbeaf, veal Fresh lanb, munon Fresh pork Freah poultry Other fresh meat Frozen and altedmeat Freah andfrozen fiah Canned fish Freh milk Milk producu Eggs,eggproducta Butuer Margerine,edible oil LArd, edible fat Fresh tropical fruita Otherfresh fruits Fresh vegetAblea Fruit other than freh Vegetables other than fresh Tubers, including pocAtoes Coffee TeA Cocoa Sugar lam, syrup,honey Chocolate, ice cream Salt, spices, sauces Non-alcoholic bevergea Spirits Wine, cider Beer Cigarettes Cigars, tobacco, *nuff Clothing mntariala Men's clothing Women'sclothing Boys' andgirls' clothing Men's and boys' urderwear Women's andgirOs'underwear Haberdashery, millinery Clothingrenta nd repair Men's foorwear Women's footwear Children's footwear Foorwear, repain Furniure Fixzures Floor coverings Household texules, etc. Refrigerators, freiere WashingappLiance Cooking Appliances Heating appliances Cleaning rppliance Other household appLiunces Household utensils Nondurablehousehold goods Domestic services Household setvices Barber andbeaury shops Toilet artcles Othc: :;:z:.-. -oods Restaurants. tafea Hotels, lodgings Other services Expendicurea of resident ebroad Ono andtw'--family dwellings Multifamily dwellings
Drugs, medical prepanrioos

Medical cuppLiea Therapeuuc equipment Physicians' equipment Denutu' servic4m Nurses' service HogpitAle Personal automobiles Other peron traport Til". tub"s, aiessories Automobile puairs Gasolne, oil. gpea. Pc.rking, 'olls, ec. Local uansport RAiltranport Bus tranporn Air tasport Posal communication Telephone. telegrnph Books, paper. magazin Sutionery Fun and econd-level teachers College teachers Physical facilities for education Educational books, supplies Other oduca ocAl expenditures Hotels Industrialbutidings Commercial buildings Ofice building Educanonal buildings Hospital buildings Agricultuml buildinpg Other buildingv Roadc, suee, highways Truport and udlizy lines Other conauction LAnd improvement Locomotives Other railwayvehice Pasenger aut4mobile Trucks, buse, trilers Aircft Ships, boats Other trranport equipenent Engine, turbines Tractor Other agriculturaL machinery Offic machinery Metalworking machiney Conatrucuon. mining machinery Special industnsl machinery Electrical tranamission equipment Commurncations equipment Other electrical equipment Inruments Furdtutre., Exures Other durable goods

26 REFERENCES Aitken,B., 1991, Measuring Trade Policy Intervention:A Cross-CountryIndex of Relative Price Distortion, Working Paper 838, World Bank. Atkinson,A., and J. E. Stiglitz, 1980, Lectures on Public Economics (McGraw-Hill,New York) Balassa,B., 1985, Exports, PolicyChoices, and Economic Growth in DevelopingCountriesAfter the 1973Oil Shock, Journal of DevelopmentEconomics 18, 23-35. Growth in a Cross Section of Countries, QuarterlyJoumal of Barro, R. J., 1991, Econormic Economics 106, 407A44.
_____,

1990, GovernmentSpending in a Simple Model of EndogenousGrowth, Journalof Political Economy98, 5103-5125. and Sala-i-Martin,X., 1993, Public Finance in Models of EconomicGrowth, Reviewof EconomicStudies, forthcoming.

_______

Chamley, C. and H. Ghanem, 1991, Fiscal Policy with Fixed Nominal ExchangeRates: Cote d'Ivoire, Working Paper 658, World Bank. Chenery, H., S. Robinson, and M. Syrquin, 1986, Industrializationand Growth: A Comparative Study (Oxford UniversityPress, New York) De Long, J. B. and L. H. Summers, 1991, EquipmentInvestment and EconomicGrowth, Quarterly Journal of Economics 106, 407 444. de Soto, H., 1989, The Other Path: the Invisible Revolution in the Third World, (Harper and Row, New Yo-k) Dollar, D., 1992 Outward-OrientedDevelopingEconomies Really Do Grow More Rapidly: Evidencefrom 95 LDCs, 1976-85,EconomicDevelopmentand Cultural Change 40, 523-544 Easterly, W., 1991, Economic Stagnation,Fixed Factors, and Policy Thresholds, mimeo, World Bank Fischer, S., 1983, Inflation and Growth, NBER WorkingPaper No. 1235, Cambridge, MA: NBER.
______

1991, Growth, Macroeconomics,and Development,in: S. Fischer and 0. Blanchard, eds., NBERMacroeconomicsAnnual 1991(MIT Press, Cambridge)

Gelb, A., 1989, A Cross-SectionAnalysisof Financial Policies, Efficiencyand Growth. Working Paper 202, World Bank. Grier, K. and G. Tullock, 1989, An Empirical Analysisof Cror<-NationalEconomicGrowth, 19511980, Journal of MonetaryEconomics24, 259-974.

27

Islam, R. and D. Wetzel. 1991, Macroeconomicsof the Public Sector Deficit: the Case of Ghana, WorkingPaper 672, World Bank Jones, L. and R. Manuelli, 1990, A Convex Model of Equilibrium Growth. Journal of Political Economy78, 1008-38. Jones, L., R.E. Manuelli, and P.E. RoSsi. 1993. Optimal Taxation in Models of EndogenousGrowth, Journal of Political Economv 101 (forthconing) King, M. A. and Fullerton, D., eds, 1984, The Taxationof Income from Capital: A Comparative Study of the U.S., U.K., Sweden aad West Germany, Chicago: Universityof Chicago Press. King, R. and S. Rebelo, 1990, Public Policy and EconomicGrowth: Developing Neoclass.cal Implications,Journal of Political Economy 98, S126-S150.
and
_ _,

1993, Transitional Dvnamics and Economic Growth in the Neoclassical

Mkdel, American EconomicReview, forthcoming. King, R., C. Plosser, and S. Rebelo, 1988, Production, Growth and Business Cycles f. The Basic NeoclassicalModel, Journal of Monetary Economics21, 195-232. Kormendi R. and P. Meguire, 1985, MacroeconomicDeterminantsof Growth: Cross- Country Evidence,Journal of Monetary Economics 16, 141-163. Krueger, A. O., 1978, Foreign trade Regimes and EconomicDevelopment: LiberalizationAttempts and Consequences(NBER, Cambridge, MA)
______

1983, Trade and Employment in Developing Countries, 3. Synthesis and Conclusions

(Universityof Chicago Press, Chicago) Landau. D., 1986, Governmentand EconomicGrowth in the Less Developed Countries: An Empirical Study for 1960-80, EconomicDevelopmentand Cultural Change 35, 35-75. Levine, R. and D. Renelt, 1992, A SensitivityAnalysis of Cross-Countrv Growth Regressions, Anierican EconomicReview 82, 942-63. Lucas, R. E., 1988, On the Mechanicsof EconomicDevelopment, Journal of Monetary Economics 22, 342. ,1990, Supply-sideEconomics:an AnalyticalReview, Oxford EconomicPapers 42, 293316. Manrkiw, N. G., D. Romer, and D.Weil, 1992, A Contribution to the Empirics of EconomicGrowth, QuarterlyJournal of Economics 107, 407438. Murphy,K., A. Shleifer, and R. Vishny, 1991, The Allocationof Talent: Implicationsfor Growth, QuarterlyJournal of Economics 106, 503-530.

28 Rebelo. S, 1991, Long Run Policy Analysisand Long Run Growth, Journal of Political Economv99, 500-521. and N. Stokey, 1993, Growth Effects of Flat-rate Taxes, mimeo, Universityof Chicago. Reynolds,L. G, 1985, EconomicGrowth in the Third World, 1850-1980, (Yale UniversityPress, New Haven) Rodrik, D., 1991, Closing the ProductivityGap: Does Trade LiberalizationReally Help?, in: G. Helleiner,ed., Trade Policy, Industrialization,and Development:A Reconsideration(Oxford UniversityPress, New York) Romer, P, M, 1986, IncreasingReturns and Long-Run Growth, Journal of Political Econo y 94, 1002-1037.
______

1989, CapitalAccumulationin the Theory of Long-Run Growth, In R. Barro, ed., Modern BusinessCycle Theory (Harvard University, CambridgeMA)

Sachs, J. D, 1987,Trade and exchange rate policies in growth-orientedadjustrnentprograms, In V. Corbo, M. Goldstein, and M. Xhan, eds., Growth-orientedAdjustmentPrograms (IMF and WorldBank, Washington,D.C.) Solow, R., 1956, A Contributionto the Theory of EconomicGrowth, Quarterly Journal of Economics70, February 1956, 65-94. Summers, R. and A. Heston, 1988, A New Set of InternationalComparisonsof Real Product and Price Level Estimatesfor 130 Countries, 1950-85,Review of Income and Wealth 34 , 1-25. Young,A, 1991, Learning by Doing and the DynamicEffects of InternationalTrade, Quarterly Journal of Economics 106, 369406. World Bank, 1987, World DevelopmentReport i987, Oxford UniversityPress. , 1988, World DevelopmentReport 1988, Oxford UniversityPress. , 1989a, Sub-SaharanAfrica: From Crisis to SustainableGrowth, Washington:the World Bank, 1989.
______

1989b,World DevelopmentReport 1989, Oxford UniversityPress.

Figure 1: DistortionaryTax Rates and Growth (Alternative Elasticities of Substitution in Production)

0.05

__'v

0.04

0.01 0
-~-0.01

-0S).02

>
t

Elasticity of substitution= 2.0

-0.04 -0.05 0
.__
-

Elasticity of substitution
_ .
_ _ ___-

.5

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

ImplicitTax Rate on Incomefrom Type I Capital

Figure 2: Welfare and distortion (Alternative elasticities of substitution in production)

0
_

CD~~~~~~~~~~~~~~~~~

Elasticity of substitution 0.5

.413

Elasticity of substitution
0i

3.0 -0.5

-0 .6

- -

----

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Implicit Tax Rate on Incomefrom Type I Capital

Pollcy Research Working Paper Series Title WPS1 194 How Fast HasChineseIndustry Grown? WPS1195 The EnterpriseSectorand Emergence of the PolishFiscal Crisis, 1990-91 WPS1 196 Corporate Tax Structureand Production WPS1197 Determinants of Inflationamong FrancZoneCountriesin Africa WPS1 198 Enterprise Reformin China:The EvolvingLegal Framework Contact for paper E. Khine 37471 E. Khine 37471

Author Tom Rawski

Date September1993

Mark Schatter

September1993

Joffrey Bernstein AnwarShati Bruno Boccara Shantayanan Devarajan NatalieLichtenstein

September1993

C. Jones 37754 C. Jones 37754 M. Rangarajan 81710 D. Evans 37496

September1993

September1993

WPS1199 Public PensionGovernance and Olivia Mitchell Performance: Lessonsfor Developing Countries WPS1200 The Life-CycleDistributional Consequences of Pay-As-You-Go and FundedPensionSystems WPS1201 FiveCriteriafor Choosingamong PovertyPrograms Jane Falkingham PaulJohnson

October 1993

October 1993

D. Evans 37496

MargaretE. Grosh

October 1993

M. Quintero 37792 RoseVo 33722 F. Smith 36072 K. Rivera 34141

WPS1202 Privatization and ForeignInvestment FrankSader in the Developing World,1988-92 WPS1203 Determinants of Value-Added Tax Zeljko Bcgetic Revenue A Cross-Section Analysis FareedHassan WPS1204 StructuralAdjustment,Economic NishaAgrawal Performance, and AiJ Dependency ZafarAhmed in Tanzania MichaelMered RogerNord WPS1205 Wageand Employment Decisions SimonCommander in the RussianEconomy:An Analysis LeonidLiberman of Developments in 1992 Ruslan Yemtsov WPS1206 EmpiricalPerspectives on National Index Funds IsnacDiwan VihangErrunza LemmaW. Senbet Bill H. Kinsey Hans P. Binswanger

October 1993

October1993

October 1993

October 1993

0. del Cid 35195

October1993

A. Yideru 36067

WPS1207 Characteristics and Performance of SettlementPrograms: A Review

October 1993

H. Binswanger 31871

Policy Research Working Paper Serles Contact for paper I. Conachy 33669

Title in WPS1208 PrimarySchool Achievement Englishand Mathematics in Zimbabwe:A Multi-Level Analysis WPS1209ShouldEastAsia Go Regional? No, No, and Maybe

Author Levi M. Nyagura AbbyRiddell

Date October 1993

Arvind Panagariya

October 1993

D. Ballantyne 37947 C. Jones 37699 R. Vo 31047 P. Attipoe 526-3003

WPS1210 The Taxation of NaturalResources: RobinBoadway Principlesand Policy Issues Frank Flatters WPS1211 Savings-Investment Correlations NlanduMamingi and CapitalMobilityin Developing Countries WPS1212 The LinksbetweenEconomicPolicy Rsvi Kanbur and Research: ThreeExamplesfrom Ghanaand SomeGereral Thoughts WPS1213Japanese ForeignDirect Investment: KwangW.Jun RecentTrends,Determinants, and FrankSader Prospects HaruoHoraguchi HyuntaiKwak WPS1214Trade,Aid, and Investmentin SubSaharanAfrica WPS1215How Much Do Distortions Affect Growth? WPS1216Regulation,Institutions,and Commitment: Privatization and Regulationin the Argentine Telecommunications Sector WPS1 217 Unitaryversus CollectiveModels of the Household: Timeto Shiftthe Burdenof Proof? Ishrat Husain

October 1993

October 1993

October 1993

November 1993

S. King-Watson 33730

November 1993

M. Youssef 34637 R. Martin 39065 D. Evans 38526

WilliamEasterly

November 1993

Alice Hill Manuel AngelAbdala

November 1993

Pierre-Andre Chiappori Lawrence Haddad John Hoddinott RaviKanbur

November1993

P. Attipoe 526-3002