Public Investment and Economic Growth in Latin America: an Empirical Test

Public Investment and Economic Growth in Latin America: an Empirical Test

Miguel D. Ramirez and Nader Nazmi*

AbstractThe paper analyzes the impact on economic growth of public investment spending and other relevant vari-ables (such as human capital) for nine major Latin American nations over the 1983–93 period. The resultssuggest that both public and private investment spending contribute to economic growth. Overall centralgovernment consumption expenditures, on the other hand, are found to have a negative effect on privateinvestment and growth. Finally, public expenditures on education and healthcare are found to have a posi-tive and statistically significant effect on private capital formation and long-term economic growth. From apolicy standpoint, the results suggest that indiscriminate cuts in public and private investment spending arelikely to be counterproductive in the long run, and more importantly, scarce public expenditures should bechanneled to the promotion of new human capital (via primary and secondary education) and the mainte-nance of existing human capital (through healthcare expenditures).

1. Introduction

In contrast to the “lost decade of the eighties,” the decade of the 1990s has been labeled“a decade of hope” by contemporary observers of the Latin American economic andsocial landscape. They point to the renewed growth in real gross domestic product(GDP) for the region over the 1991–94 period, with an average of 3.6%, and the dra-matic fall in inflation during the 1994–96 period (from 340% to 28%), primarily as a result of the success of the Brazilian stabilization program (ECLAC, 1996). The optimistic outlook for the region is tempered somewhat by the severe recession andinflationary pressures experienced by the Mexican economy following the 1994 peso devaluation, and the poor economic performance of Argentina, Uruguay, andVenezuela. Nevertheless, the conventional wisdom regarding the future economicpanorama of Latin America and the Caribbean is that after anemic economic growthin 1995 (0.4% in real terms and an estimated 2.8% in 1996), the region is poised toresume vigorous and sustained growth rates in output and investment comparable tothose registered during the decade of the 1960s and 1970s.

This paper takes a less sanguine assessment of the long-term economic prospects ofthe region by noting that the overall performance of public and private investmentspending in the major Latin economies during the 1980s and early 1990s was disap-pointing and erratic at best. The implications of this unfavorable trend in the invest-ment data are sobering for, if there is anything on which economists agree, it is that arobust and sustained increase in overall capital formation is absolutely essential forcreating future opportunities for income and employment creation and ensuring thesuccess of the far-reaching market-oriented reforms that are currently being imple-mented in the region. In light of these developments, it is the objective of this study to

Review of Development Economics, 7(1), 115–126, 2003

* Ramirez: Trinity College, Hartford, CT 06106, USA. Tel: (860)-297-2487; Fax: 297-2163; E-mail:[email protected]. Nazmi: Lake Forest College, Chicago, IL 60045, USA. Tel: (847)-735-5147; Fax:735-6291; E-mail: [email protected].

© Blackwell Publishing Ltd 2003, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA

analyze the link between public investment spending and economic growth in LatinAmerican both in theory and in practice.

Section 2 provides an overview of the disappointing performance of the rate ofcapital formation for nine major Latin economies. Section 3 develops a dynamic opti-mization model that explicitly incorporates the potential impact of public capital for-mation and technology on the rate of private investment spending undertaken by arepresentative firm. Section 4 presents an empirical model that assesses the impact ofpublic investment spending and other relevant economic variables on private invest-ment spending and the rate of economic growth during the 1980–93 period. Section 5summarizes the main conclusions.

2. Investment Performance in Latin America in the 1980–95 Period

The debt crisis of the 1980s and its economic and financial aftermath had a devastat-ing effect on Latin America’s overall rate of capital formation, the region’s futuresource of income and employment creation. The severity and duration of the debt-induced economic crisis of the 1980s forced many governments of the region to imple-ment severe economic adjustment and stabilization programs sponsored by the IMFand the World Bank. In the short run, these programs called for a simultaneous adjust-ment of the external and internal sectors via a real devaluation of the domestic currency and stabilization of the rate of inflation through across-the-board cuts in government spending. Initially, these measures amounted to a de facto strategy of generating trade surpluses through a massive reduction in the level of economic activity and imports in order to secure the hard currency needed to service the stag-gering debt accumulated by the Latin economies during the 1970s. Alongside theseshort-term measures, structural reform policies were recommended and implementedto liberalize trade, dismantle the state-led sector of the economy by privatizing leadingstate-owned enterprises, and finally, deregulate financial and labor markets to improvethe “flexibility” of the Latin economies (Nazmi, 1997). The conventional wisdomunderlying market-oriented reforms is that, in the long run, the thorough implemen-tation of these policies will pave the way for sustained economic growth with price stability by improving the overall allocative and productive efficiency of the Latineconomies.

In the meantime, as most economies of the region wait for robust economic expan-sion promised by neoliberal reforms, the decline in the level of economic activity andin essential imports generated by the short-term adjustment measures implemented inthe aftermath of the debt crisis has had a particularly negative effect on the rate ofcapital formation in Latin America and the Caribbean. For example, the investmentcoefficient for Latin America and the Caribbean fell from 24.7% in 1980 to a low of15.7 in 1990, and although it rose to 18.8% in 1995, it still remains far below the 1980and even the 1970 levels (ECLAC, 1996).

This troubling decline in investment was caused partly by stabilization measures thatreduced sharply the level of public investment in economic and social infrastructure,as well as public expenditures in areas such as education and healthcare where thegovernment has traditionally supported the maintenance and creation of humancapital. The public investment coefficient in Latin America and the Caribbean hasfallen dramatically relative to the early 1980s and is far below that of all other regionsof the world, including East Asia where it has risen by a full three percentage pointssince 1988.1 Public investments are, by and large, confined to projects that are “lumpy”and indivisible, such as roads, bridges, and ports, and whose gestation period is rela-

116 Miguel D. Ramirez and Nader Nazmi

© Blackwell Publishing Ltd 2003

tively long. Although, in principle, the private sector could undertake these types ofinvestments, which are essential for the long-term growth and development of thecountries of Latin America, they often do not because of the well known freeriderproblem. Moreover, public spending on highways, bridges, ports, airfields, education,health, and energy supplies tends to generate direct positive effects on private sectorprofitability and investment which implies, ceteris paribus, that pronounced cuts in public investment may further decrease the rate of private capital formation(Aschauer, 1989; Cardoso, 1993; Mouguillansky, 1996).

The anemic nature of investment by the private sector during the “lost decade” canbe attributed in part to IMF-sponsored stabilization and adjustment policies. Thesemeasures often call upon governments of the region to slash expenditures across-the-board to meet prescribed targets for the fiscal deficit (as a percentage of GDP), whileadjustment measures such as real devaluation alter relative prices for a given level orrate of aggregate spending. These policies have adversely affected investment spend-ing through three major channels.2 First, since the region’s import-substituting andexport-oriented firms rely heavily on imported capital goods embodying sophisticatedtechnology, IMF-sponsored stabilization measures that generate a sharp fall in the levelof economic activity and imports immediately translate into a decrease in the demandfor investment goods. Second, investment demand and the rate of capital formationhave been adversely affected by the squeeze on profit levels generated, on the demandside, by draconian reductions in domestic spending and the lowered prospects forfuture economic recovery that result from the deflationary measures themselves.Third,in order to meet the prescribed cuts in the fiscal deficit, the governments of the regionhave resorted to the politically expedient policy of disproportionately reducing publicinvestment in economic and social infrastructure. However, what is politically expedi-ent in the short run may be economically foolish in the long run if, as a number ofinvestigators have recently shown, there are significant complementarities betweenpublic and private investment spending.3 Finally, on the supply side, investment in newplant, machinery, and equipment is constrained by the high borrowing costs forworking capital generated by the stabilization-induced reduction in credit and the com-petitive (real) devaluations of the domestic currency which make imports of capitalinputs and intermediate goods more expensive in domestic currency terms. Given thatexport prices (in domestic currency terms) do not match the rise in import prices, thereis a deterioration in the terms of trade and a further contraction in real income andspending (Taylor, 1997).

The severity and widespread nature of the investment crisis has also been reflectedin the experience of the individual countries. Argentina’s investment coefficient fellfrom 25.0% in 1980 to 13.4% in 1990, and despite its increase in recent years it yet hasto recover its 1980 level. In Brazil it dropped from 23.3% in 1980 to 13.2% in 1992,only to be followed by slight increases in 1993 and 1994. Chile’s investment coefficientdecreased sharply from 21.0% in 1980 to 13.9% in 1985, and despite the fact that Chilehas been the fastest growing economy in the region it barely attained its 1980 level in1992. Mexico’s investment coefficient, after falling by close to nine percentage pointsbetween 1980 and 1990, registered a slight increase between 1991 and 1994 only to fallsharply again in 1995. Finally, Venezuela experienced a precipitous drop in its invest-ment coefficient of close to 20 percentage points between 1980 and 1992, and in 1995it had yet to recover its 1980 level (ECLAC, 1996).

The disappointing and erratic behavior of investment during the decade of the 1980sand early 1990s is particularly disturbing because it has been accompanied, with theexception of Chile,4 by relatively low growth rates of real GDP per capita, low savings

PUBLIC INVESTMENT AND GROWTH IN LATIN AMERICA 117

© Blackwell Publishing Ltd 2003

rates, increasing poverty rates, and dwindling employment opportunities for the mil-lions of new entrants into the workforce each year (ECLAC, 1996) Without sustainedgrowth and substantial rises in investment, the Latin American governments will notbe able to meet the huge backlog of unmet social needs created by the “lost decade”of the 1980s. In fact, failure to address the rapidly deteriorating social and political sit-uation in several countries of the region (notably Colombia, Ecuador, and Venezuela)may seriously compromise the political and economic viability of market-orientedreforms and dash the expectations raised by the so-called “decade of hope.”

3. Dynamic Optimization and Optimal Capital Formation

This section analyzes the impact of public sector policies and investment on economicgrowth in Latin America. It develops a model for the optimal path of private capitalformation in a small open economy with competitive markets under conditions of certainty. The model is based on a modified neoclassical production function of the following form:

(1)

where Y is the level of real output, F is a concave production function satisfying theneoclassical properties of positive and diminishing marginal products of each input(the Inada conditions) and constant return to scale, L denotes labor, Kp is the stock ofprivate (physical) capital, Kg stands for the public (physical) capital stock, and A rep-resents the technological factor.

This model differs from neoclassical production functions in two important ways.First, it does not assume that technology grows exogenously at a given rate; instead ittakes technology as an explicit factor of production with its own price.The key assump-tion about productivity growth here is that a typical developing country purchasestechnological knowledge abroad from various suppliers. What technology will be pur-chased depends on the price of foreign technology as well as trade and exchange ratepolicies that impact the final cost of the imported technology. In this respect, our modelalso differs from Romer (1994)-type endogenous-growth models where technology isendogenously advanced to eliminate diminishing returns. In these models new tech-nology is introduced by research and development (R&D), but in our model it isexogenous and imported. Thus the old center–periphery dichotomy is subjected to anew and important twist: the center countries can escape from diminishing returnsthrough R&D activity while developing countries remain captive to it as they are con-sumers and not producers of technology. This distinction between center and periph-ery production functions also has significant negative implications for the notion of(beta-type) convergence: despite lower capital stocks and GDPs, poor countries willnot grow faster than rich countries as they are subject to diminishing returns and richcountries are not.

Second, the above model separates capital into public and private capital to help usbetter answer the complex and controversial question of whether public and privateinvestment are complements, substitutes, or, as assumed in most macroeconomicmodels, independent of one another (Aschauer, 1989; Cardoso, 1993; Khan and Reinhart, 1990; Moguillansky, 1996; Nazmi and Ramirez, 1997; Ramirez, 1994). Theinclusion of the public capital stock as a separate input in the production function gives rise to three conceptually different effects. If the public capital stock is comple-mentary to the private capital stock, an increase in the former will increase outputdirectly in the same way that an increase in any other factor of production raises output

Y F L K K A F Fp g= ( ) ¢ > ¢¢ <, , , , ,0 0

118 Miguel D. Ramirez and Nader Nazmi

© Blackwell Publishing Ltd 2003

(F3 > 0). In such a case, an increase in public capital stock will increase investment spending and output indirectly by raising the marginal productivity of the private capitalstock (F23 > 0) or by increasing the marginal productivity of labor (F12 and F13 > 0).

In the case where public and private capital are direct substitutes (F23 < 0), anincrease in public capital formation will generate a positive direct effect on output, buta negative effect on the marginal productivity of private capital and labor that couldmore than offset it.5 Finally, if private and public capital are independent of oneanother, as assumed in the conventional model, then an increase in public investmentspending will raise output and marginal productivity of labor in the public sector.

The dynamics of the modified model are examined via the behavior of a represen-tative firm that attempts to maximize profits over the time interval [0, T]. In thisdynamic optimization framework, there is a single state variable Kp and two controlvariables, L and A. The public capital stock, Kg, is determined by the government andis therefore exogenously given to the representative firm.The amount and type of tech-nology transferred depends on various factors including the price of technology, tradepolicies, and income levels. A typical firm starts out at time 0 with a capital stock Kp0

and ends with capital stock that is left open or free, K(T). At each instant in time, therepresentative firm is assumed to maximize the present value of its profits subject tothe appropriate choice of one of its control variables. The optimal control problem cantherefore be stated as maximizing

(2)

subject to

(3)

A dot above a variable denotes the derivative with respect to time. Kp is the rate ofchange of private (physical) capital, r is the discount rate (real interest rate), l is therate of depreciation of private capital stock, and Ip is gross private capital formation.The representative firm’s profits (P) in each time period are determined as

(4)

Here P represents the general price level, F is the economy-wide production function,and w, PA, and rg are, respectively, the wage rate, price of imported technology, andprice (cost) of obtaining one unit of public capital. Finally, rp denotes the cost of invest-ment in private capital goods.

To solve the dynamic optimization problem outlined in equations (2) and (3) wemake use of the maximum principle embodied in the following Hamiltonian:

(5)

Differentiation of H with respect to the two control variables yields

(6)

(7)

Equations (6) and (7) imply that the maximizing firm will employ additional unitsof labor and imported technology until the value of their perspective marginal pro-ducts is equal to the corresponding prices of its inputs. In this formulation, open tradepolicies that reduce PA through the elimination of quotas, tariff barriers, and over-valued exchange rates directly foster economic growth. The equations of motion for

d dH A e PF P F P PrtA A= -[ ] = =-

4 40, .

d dH L e PF w F w Prt= -[ ] = =-1 10, ,

H e PY wL P A r K r K K I KrtA g g p p p p p= - - - - +( )[ ]+ -[ ]- l m l .

II = ( ) - + +( ) -PF L K K A wL P A r K r Ip g A g g p p, , , .

K K K Tp p p0 0( ) = ( ) =, .free

K̇ I Kp p p= - l

II II= ( )-Ú e t dtrtT

0

PUBLIC INVESTMENT AND GROWTH IN LATIN AMERICA 119

© Blackwell Publishing Ltd 2003

the state and costate variables are obtained, respectively, by differentiating H withrespect to Kp and m (which represents the shadow price of private capital at everyinstant in time):

(8)

(9)

Equation (8) specifies how the firm determines its rate of capital formation overtime. Equation (9), on the other hand, expresses the rate of depreciation of the shadowprice of private capital, or the rate at which private capital is contributing to the currentand future profits of the representative firm. At this juncture, it is convenient to definea new Hamiltonian that is free of the discount factor. In general, this so-called current-value Hamiltonian is expressed as

(10)

where m is the companion concept of the current-value Lagrange multiplier, and it isdefined as

(11)

To revise the equation of motion for the costate variable, m = -dH/dKp, each side of(9) is expressed in terms of m and the results are equated. Differentiating the left-handside with respect to time and using (10) we obtain

(12)

Using equations (10) and (12) and eliminating the discount term yields

(13)

To see the economic interpretation of (13), we substitute for the term dHc/dKp from(9) which, after rearranging terms, gives

(14)

The first term of the right-hand side of (14) shows the marginal contribution of capitalto current profits, while the second denotes the marginal increase in the future mone-tary value of capital generated by an additional unit of capital. The left-hand side of(14) expresses the rate of decrease of the imputed or shadow price of private capitalover time. The maximum principle requires that the imputed price of capital shoulddecrease at the rate at which capital is contributing to current and future profits in therepresentative firm. Taking the derivative of the imputed price of private capital, m,with respect to Kg in (14) yields

(15)

According to (15), the rate of depreciation of the imputed value of private capitalaccelerates if public capital is complementary to private capital (F23 > 0). It will remainunchanged as long as public and private capital are independent (F23 = 0). Finally, ifpublic capital crowds out private capital (F23 < 0), the rate of depreciation of theimputed value of private capital decelerates. Under the first scenario, the more publiccapital (infrastructure) private capital has at its disposal, the greater the value itimparts to the representative firm at every instant in time. When private and public

- =d d˙ .m k PFg 23

- = -( ) - +( )˙ .m PF r r mp2 l l

˙ .m H K rmc p= - +d d

˙ .m = -- -me rmert rt

m e mert rt= =( )-m mimplying .

H He G mfcrt= = ( ) + ( ). . ,

˙ .m ml l- = - -[ ]-e PF rrtp2

˙ ,K H I Kp p p= = -d dm l

120 Miguel D. Ramirez and Nader Nazmi

© Blackwell Publishing Ltd 2003

capital are substitutes, however, increases in public capital reduce private capital’s con-tributions to the firm’s profits at every point in time.

4. Data and Empirical Results

Before examining the estimation procedure and the empirical results several featuresof the countries used in this study need to be emphasized. First, it must be remem-bered that these economies have evolved quite differently over the last two decades.The growth experiences of the countries under investigation included the strong per-formance of Chile (6.3% average annual growth rate), the dismal record of Peru(0.97% average growth rate), and the respectable showing of Colombia (3.89%) andUruguay (3.18%).

Second, the role played by the public sector varied significantly across the economiesin question. For the entire region private investment amounted to an average of11.35% of GDP, almost double the size of the public sector investment share of 5.41%.In Bolivia, however, the private sector’s investment share was only 4.38% of GDP compared to the public sector’s share of 7.61%. At the opposite end of the spectrum,Brazilian private and public investment amounted to 14.62% and 2.06% of GDP,respectively. Third, public funds are used for very different programs in these coun-tries. For example, for the 1983–93 period more than half of Uruguay’s public expen-ditures were spent on social programs, while in Ecuador a large share of the centralgovernment’s expenditures was earmarked for education, and Peru devoted consider-able public resources to military expenditures. These differences in the role of thepublic sector in the countries considered in this study make the results reported belowsuggestive and preliminary in nature. Nevertheless, these results are drawn from thefirst comprehensive study of the effect of public investment expenditures on privateinvestment and economic growth in Latin America, and as such, they offer valuableinsights into the role of the public sector in the economies of Latin American countries.

The data used in this study are for the 1983–93 period, and they include the nineLatin American countries of Argentina, Bolivia, Brazil, Chile, Colombia, Ecuador,Mexico, Peru, and Uruguay. The panel dataset contains 99 observations on each vari-able. It is possible to focus on the impact of private and public investment expendi-tures on economic growth rate by holding constant the impact of other variables thattheoretically affect a country’s growth rate. We thus estimated the following model forthe growth rate of real per capita GDP (y):

(16)

The variable log(Y1980) is the logarithm of real per-capita GDP in 1980 and is includedto acknowledge the theoretical prediction of convergence as posited by standardgrowth models: holding other variables constant, economies with lower per-capitaincome will grow faster as lower per-capita output corresponds to higher capital pro-ductivity and growth. The variable H is a measure of educational attainment and isused as a proxy for human capital.6 It measures the average number of years of edu-cation by those older than 25 years of age reported by Barro and Lee (1994) for theperiod 1960–85 and updated by us for 1985–93. The variables Ip/Y and Ig/Y are, respec-tively, the ratios of real gross private investment and real gross public investment to

y Y HIY

IY

Open Erovert tp

t

g

tt t t= + ( ) + + Ê

ËÁˆ¯̃ + Ê

ËÁˆ¯̃ + + +a a a a a a a e0 1 1980 2 3 4 5 6log .

PUBLIC INVESTMENT AND GROWTH IN LATIN AMERICA 121

© Blackwell Publishing Ltd 2003

real GDP. Finally, variables Open and Erover are proxies for the price of importedtechnology as they measure, respectively, the openness of the economy to foreign tradeand the black market premium on foreign exchange.

When assessing the impact of public investment on economic growth and on privateinvestment, one has to be mindful of simultaneity bias.7 In addition, the inclusion ofan equation for the private investment variable offers valuable insights regarding howpublic investment affects private investment. Even when public investment expendi-tures impact output positively and significantly, their contribution could come—as indi-cated in endnote 5—at the expense of private investment; that is, public investmentcould crowd out private investment (Nazmi and Ramirez, 1997).

To address these issues, the following equation for private investment is added toequation (16):

(17)

where GC/Y measures the ratio of government consumption to real GDP. The humancapital variable was included in this equation following the model for imbalancesbetween human capital and physical capital developed by Mulligan and Sala-i-Martin(1993) where increases in human capital increase the marginal productivity of physi-cal capital relative to human capital and results in a rise in the investment/GDP ratio.

Equations (16) and (17) were estimated simultaneously using the seemingly un-related (SUR) procedure. Tables 1 and 2 give the initial estimation results. For eachestimated equation insignificant variables were eliminated systematically and themodel was re-estimated each time. The final estimated equation for the GDP growthvariable is of the form

(18)

Thus for a given initial value of per-capita GDP, both private and public investmenthave a positive and statistically significant impact on output growth. The estimatedinitial per-capita GDP variable has a negative and significant effect on output growth,indicating a “conditional convergence” of about 0.7% per year to the long-run growthtarget.

Equation (1) offers an explanation of this convergence in a way different from theexplanation put forth by Barro. For a given production function, lower levels of H and

y I y I y yt p t g t= ( ) + ( ) -( ) ( ) ( )2 887 5 613 0 007

0 875 2 12 0 003

0. . . .

. . .

I y y I y GC Y H vp t t g t t t t( ) = + + ( ) + ( ) + +b b b b b0 1 2 3 4

122 Miguel D. Ramirez and Nader Nazmi

© Blackwell Publishing Ltd 2003

Table 1. Panel Regression for Growth Rate of Real Per-CapitaGDP, 1983–93

Independent variable Estimated coefficient Standard error

Constant -0.125 0.088Initial GDP 0.005 0.008Human capital 0.007 0.393Private investment 0.009 0.002Public investment 0.014 0.003Open trade 0.005 0.006Real effective -0.002 0.002

exchange rate

R2 = 0.43, T ¥ R2 = 24.04.

K encourage increased technology transfer (a rise in A) which elevates both labor andcapital productivity and output. At this juncture, it should be pointed out that the con-vergence rate reported above is slower than that reported by Barro (1990) and Barroand Lee (1994) for other regions of the world, the US, and a cross-section of 73 countries. This can probably be explained by the limited number of observations usedin this study which reduce the observed gap between the initial output level and subsequent levels.

Our results also differ from results reported for other regions of the world in thatthey suggests that the human capital variable is an insignificant determinant of outputgrowth.Three factors can explain this unexpected result. First, since our data span onlyone decade, the change in human capital is negligible. Second, there are well-knownshortcomings associated with the Barro–Lee data (and hence our extension of theirdata) used in this study as a measure of human capital (Stokey, 1994). Finally, thehuman capital variable can enter the growth process indirectly through the physicalinvestment channel.

To explore this possibility, consider the final estimated private investment equation:

(19)

The estimated coefficient on the variable H is positive with a relatively low standarddeviation, showing that the human capital variable has a positive and significant impacton output growth.This result is consistent with Barro and Sala-i-Martin’s (1992) findingthat countries rich in human capital invest relatively more heavily in physical capitalwhich, in turn, spurs faster economic growth. Equation (19) also shows that outputgrowth has a positive and significant impact on private investment spending; that is, afaster growing economy spurs investment by the private sector. The estimated coeffi-cients and their standard deviations for the public sector investment (Ig) and con-sumption (GC) variables show that the public sector crowds out the private sector.Thefinding that private investment is negatively impacted by public investment is consis-tent with results reported for Mexico (Nazmi and Ramirez, 1997) but inconsistent withresults reported for a group of 72 countries (Barro, 1990).

The conclusion that public-sector expenditures crowd out the private sector is toogeneral and as such not very useful. One needs to distinguish between various invest-ment and consumption projects that governments sponsor. To partially remedy thisdrawback, we examine the impact of government consumption on private investmentand growth by considering data on the government consumption variable disaggre-

I y H y I y GC yp t t t g t t( ) = + + - ( ) - ( )( ) ( ) ( ) ( ) ( )12 484 1 065 0 314 1 769 2 136

1 039 0 211 0 078 0 003 0 005

. . . . . .

. . . . .

PUBLIC INVESTMENT AND GROWTH IN LATIN AMERICA 123

© Blackwell Publishing Ltd 2003

Table 2. Panel Regression for Ratio of Real Private Investment toReal GDP, 1983–93

EstimatedIndependent variable coefficient Standard error

Constant 12.484 1.039Human capital 1.065 0.211GDP growth 0.314 0.078Public investment -1.769 0.003Government consumption -2.136 0.005

R2 = 0.96, T ¥ R2 = 53.87.

gated into six categories: education (Ge/Y), health care (Gh/Y), social/welfare (Gs/Y),economic projects (Gp/Y), defense (Gd/Y), and other. All these variables are measuredin terms of the ratio of real government expenditures to real GDP. We then estimatea regression model of the form

(20)

where the “other” category is left out to avoid a multicollinearity problem. After elim-inating insignificant variables, we obtain a final model for the ratio of private invest-ment to GDP of the form

(21)

The human capital (H) and government expenditure on healthcare (Gh/Y) variablesimpact the private investment variable positively and significantly.These two variables,reflecting the role of human capital, increase output through their positive impact onprivate investment. Government investment and defense expenditures, however, havesignificant and adverse effects on private investment.

5. Conclusion

This paper has analyzed the impact of public investment spending and other relevanteconomic variables (such as human capital) on the growth rate of nine major LatinAmerican countries. It first developed a dynamic optimization framework that showedthat public investment spending can affect the rate at which private capital formationcontributes to the current and future profits of the representative firm. The modifiedmodel also indicated that while the developed world may be free from diminishingreturns owing to endogenous technology, Latin American countries remain subject to it, thereby precluding them from closing the development gap as suggested by theconvergence hypothesis. Finally, the conceptual framework developed warns againstthe introduction of trade and exchange rate policies that introduce biases againstimported technology given that these measures are likely to retard further economicgrowth.

Second, the empirical results suggest that both public and private investment con-tribute to economic growth. Thus, IMF-sponsored policies which result in a dispro-portionate reduction in private and public investment are, ceteris paribus, detrimentalto long-term economic growth. Admittedly, the model estimated in this paper cannotcontrol (directly) for the timing and sequencing of IMF-sponsored stabilization andadjustment measures implemented in the region during the period under review.However, it is important to note that many of the countries included in this studyundertook both sharp and sustained reductions in government spending and repeateddevaluations of their currencies at a time (1980s and early 1990s) when markets in thedeveloped world were either stagnant or expanding at modest rates through reces-sionary pressures. Under these less than auspicious conditions, repeated (real) deval-uations had a beggar-thy-neighbor effect which further contributed to the fall in theregion’s terms of trade and export revenues.

The results also suggest that public investment expenditures tend to crowd outprivate investment spending. This somewhat unexpected result may be due to the levelof aggregation of the public investment data which includes not only expenditures on

I y H y I y G Y G Yp t t t g t h t d t( ) = + + - ( ) + ( ) - ( )( ) ( ) ( ) ( ) ( ) ( )12 751 0 083 0 201 0 082 6 038 7 054

0 931 0 014 0 081 0 027 2 245 2 051

. . . . . . .

. . . . . .

I y H y I y G Y G Y

G Y G Y G Y

p t t t g t e t h t

s t p t d t t

( ) = + + + ( ) + ( ) + ( )+ ( ) + ( ) + ( ) +

d d d d d d

d d d h0 1 2 3 4 5

6 7 8 ,

124 Miguel D. Ramirez and Nader Nazmi

© Blackwell Publishing Ltd 2003

economic infrastructure which, a priori, should have a positive impact on privateinvestment, but also expenditures by relatively inefficient state-owned enterprises andpublic trust funds that often tend to substitute for private investment spending.

Finally, in contrast to the public investment data, the government consumption data can be broken down into various important spending categories with the follow-ing results. First, expenditures on education, social, and welfare programs have no discernible impact on private investment. Second, defense spending reduces privateinvestment significantly, while public expenditures on health services have a positiveand significant impact on private investment. Third, the human capital variable wasalso found to have a significant and positive impact on private investment spending.From a policy standpoint, this finding suggests that the promotion of new humancapital (via primary, secondary, and vocational education) and the maintenance ofexisting human capital (through healthcare) are likely to affect private investment pos-itively, thereby contributing to long-term economic growth.

References

Aschauer, D., “Public Investment and Productivity Growth in the Group of Seven,” EconomicPerspectives 13 (1989):17–25.

Barro, R. J., “Government Spending in a Simple Model of Endogenous Growth,” Journal ofPolitical Economy 98 (1990):103–25.

Barro, R. J. and J.W. Lee,“Sources of Economic Growth,” Carnegie–Rochester Conference Serieson Public Policy 40 (1994):1–46.

Barro, R. J. and X. Sala-i-Martin, “Convergence,” Journal of Political Economy 100(1992):223–51.

Cardoso, E., “Private Investment in Latin America,” Economic Development and CulturalChange 41 (1993):833–48.

ECLAC, Statistical Yearbook for Latin America and the Caribbean. Santiago, Chile: UnitedNations (1996).

Glen, J. D. and M. A. Sumlinski, Trends in Private Investment in Developing Countries,Washington, DC: International Finance Corporation (1996).

Green, Joshua and D. Villanueva, “Private Investment in Developing Countries: an EmpiricalAnalysis,” IMF Staff Papers 38 (1991):33–58.

Khan, M. S. and C. M. Reinhart, “Private Investment and Economic Growth in DevelopingCountries,” World Development 18 (1990):19–27.

Killick, T., M. Malik, and M. Manuel, “What Can We Know About the Effects of IMF Programmes?” World Economy 20 (1992):575–97.

Mouguillansky, G., “The Macroeconomic Context and Investment: Latin America Since 1980,”Cepal Review (April 1996):79–94.

Mulligan, C. B. and Sala-i-Martin, “Transitional Dynamics in Two-Sector Models of EndogenousGrowth,” Quarterly Journal of Economics 108 (August 1993):739–73.

Nazmi, N., “Exchange Rate-Based Stabilization in Latin America,” World Development 25(1997):519–35.

Nazmi, N. and Miguel D. Ramirez, “Private Capital Expenditures, Public Investment and Eco-nomic Growth in Mexico,” Contemporary Economic Policy 15 (January 1997):65–75.

Ramirez, M. D., “Public and Private Investment in Mexico, 1950–90: an Empirical Analysis,”Southern Economic Journal 61 (1994):1–17.

Romer, P. M.,“The Origins of Endogenous Growth,” Journal of Economic Perspectives 8 (Winter1994):3–22.

Stokey, N. L., “Comments on Barro and Lee,” Carnegie–Rochester Conference Series on PublicPolicy 40 (1994):1–46.

Taylor, L., “The Revival of the Liberal Creed: the IMF and the World Bank in a GlobalizedEconomy,” World Development 25 (1997):145–52.

PUBLIC INVESTMENT AND GROWTH IN LATIN AMERICA 125

© Blackwell Publishing Ltd 2003

Weyland, K., “Growth with Equity in Chile’s New Democracy?” Latin American ResearchReview (Spring 1997):37–67.

Notes

1. During the 1980–89 period both the simple and weighted investment coefficients for EastAsia were substantially higher (26.9% and 27.7%, respectively) than those for Latin America(18.1% and 20.6%). In the more recent 1990–95 period, the gap has widened even more asattested by the following figures: East Asia’s simple and weighted investment coefficients stoodat 30.8% and 34.3%, respectively, while those for Latin America came in at a disappointing18.4% and 19.1% (Glen and Sumlinski, 1996).2. Killick et al. (1992) find that the “brunt (of IMF adjustment programs) falls on the fixedinvestment ratio, which declines substantially and significantly over the whole period” (p. 593).3. In the case of Latin America, Cardoso (1993) and Moguillansky (1996) have found that acontraction in public investment spending explained an annual drop in private capital forma-tion equivalent to 0.2–0.5% of GDP.4. Contrary to neoliberal claims, Chile’s “exceptionalism” in recent years can be attributed inpart to the pursuit of unorthodox economic and social policies by the civilian administrations.For example, the state has redefined its role by channeling resources to projects that tend tocomplement, rather than compete, with the economic activities of the private sector, such as thebuilding of roads, bridges, and ports, and the promotion of education and health (Weyland, 1997).5. Government investment spending can crowd out private investment spending if it is financedvia greater indebtedness, printing money, higher current and future taxes, and higher real inter-est rates via the repression of the financial system (Green and Villanueva, 1991).6. Not only are better-educated and trained workers more productive in their own right, butthrough their synergistic interaction with other workers they also contribute to raising their productivity even though their level of education remains the same (Romer, 1994).7. It is also possible that public investment is an endogenous variable because expenditures onhighways, bridges, and ports tend to behave procyclically. However, in our empirical model wedo not treat this variable as an endogenous variable for three reasons. First, government spend-ing in general and, by extension, public investment are virtually always taken as exogenous vari-ables in the extant macroeconomic development literature. Second, our theoretical model treatspublic capital, technology, and labor as exogenous variables. Finally, while public investment iscertainly sensitive to economic conditions, it is unclear which factors impact it more forcefully.For example, while public investment spending might rise in response to economic weakness asa way to stimulate growth, yet if large budget deficits and resulting high interest rates are iden-tified as the source of the economic weakness, public investment might be reduced as an expe-dient way to cut the budget deficit.

126 Miguel D. Ramirez and Nader Nazmi

© Blackwell Publishing Ltd 2003