Documents on Mexican Politics.

Public and private investment in Mexico, 1950-90: an empirical analysis.

Public and private investment in Mexico, 1950-90: an empirical analysis.

by Miguel D. Ramirez

Southern Economic Journal. July, 1994. SECTION: Vol. 61 ; No. 1 ; Pg. 1; ISSN: 0038-4038

I. Introduction

During most of the post-World War II era, public investment was the driving force behind the general strategy known as Import Substitution Industrialization (ISI) in Latin America.(1) Policymakers in countries such as Mexico and Brazil recognized that investment played a crucial role not only as a component of final aggregate demand, but also in terms of determining the size of a country's capital stock, and thus, its future source of growth and employment opportunities. It was also generally believed that private investors would be reluctant to channel needed resources to key industrial projects because of the region's lack of social and economic infrastructure, as well as the absence of fully developed markets for equity, insurance, and information. Government investments in infrastructure and basic industry, with their attendant positive spillover effects, were viewed as necessary by policymakers for achieving optimal rates of investment and growth.

However, with the onset and aftermath of the debt crisis in 1982, most countries of the region, particularly Mexico and Chile, have radically changed their overall development strategy. Instead of concentrating on inward-directed growth, under the auspices of state-directed investments, the new growth model is outward-oriented in nature, and more importantly, heavily reliant on market forces as evidenced by the ongoing deregulation of product and factor markets and the privatization of most state-owned enterprises.(2) The unprecedented streamlining of the public sector in countries such as Mexico can, in part, be traced to the limited external and internal resources available to governments of the region during the 1980s, but, more importantly, it is also the result of past inefficiencies and failures generated by the public sector's attempt to undertake too many investments in state-owned enterprises producing goods in which the private sector had (and has) a comparative advantage (e.g., steel, chemicals, trucks and buses, cellulose products, food processing, sugar, fertilizers, textiles, banking and tourism services, etc.).(3)

Nevertheless, in their justified efforts to diminish the government's excessive involvement in economic affairs, policymakers seem to have forgotten that the withdrawal of the state from the economy should not endanger those minimum essential duties that it must perform for the optimal functioning of a market economy, namely, the provision of essential social and economic infrastructure.(4) Having said this, the application of severe economic adjustment measures in many countries of the region, notably Mexico, has led to across-the-board cuts in public spending that have generated unprecedented reductions in the level of public investment in the nation's railways, highways and roads, bridges, sewerage systems, modern power plants, dams, airport fields, and port facilities. These investments, as a number of researchers have recently shown |1; 2; 3; 4; 5; 6; 9; 11; 30; 34; 35; 40; 41; 46; 50; 52; 53, usually complement, rather than crowd out private investment, a nd their continued and future neglect may well undermine the efficiency gains promised by the country's ongoing liberalization and privatization process.

In light of these developments, this paper examines the still controversial question of whether public investment complements or displaces private capital formation in one of the larger countries of the region: Mexico. The choice of Mexico to test the hypothesis of complementarity or substitutability between public and private investment spending is warranted for two reasons. First, it is one of the more industrialized, and from the standpoint of the U.S., strategically important countries of the region--other than Chile--that has implemented far-reaching market-oriented policies; and secondly, it is one of the few countries of the region for which there are reliable time-series data on public and private investment for a number of years.

The paper is organized as follows. First, it gives a brief overview of the role of the Mexican state in the investment process during the post-World War II period. Next, it develops a conceptual framework of analysis for including the public capital stock as an argument in a standard neoclassical production function, thereby generating a set of testable hypotheses. The third section presents a modified flexible accelerator model that not only explicitly incorporates the impact of public investment expenditures, but also circumvents some of the investment data problems one finds when dealing with developing nations such as Mexico. Statistical estimates of various specifications of the modified accelerator model are then presented in the fourth section. This section also attempts to determine whether there is a long-term relationship between the relevant regressors by performing Engle-Granger Cointegration (EG) tests on the various regression specifications. It therefore goes beyond other empirical discussions of the complementarity hypothesis--both for developed and developing nations--by testing whether the residuals of the relevant cointegrating regression equations are stationary, thus addressing the important question of spurious correlation. Lastly, pairwise Granger causality tests are performed to determine the direction of correlation between public and private investment expenditures. The last section summarizes the paper's major findings.

II. The State's Changing Role in the Investment Process

During most of the post-World War II period, the Mexican state played a pivotal role in the investment process. It is no exaggeration to say that during the heyday of the ISI strategy pervading most of Latin America, it was the driving force behind the rapid economic growth the country experienced. Table I below shows that during the fifties and sixties the state's share of overall investment averaged between .46 and .66, and as the yearly data indicate, it even exceeded 70 percent during the populist administration of Adolfo Lopez Mateos (1958-64). With the election of conservative president Gustavo Diaz Ordaz (1964-70), the government placed greater reliance on the private sector and its role in the investment process diminished accordingly, reaching a record low of .27 in 1971.(5)

   Table 1. Mexico: Participation of the State in the Process of  Capital      
   Formation, 1950-91 (as a percentage of total gross fixed  capital formation)
   Year  Coefficient  Year  Coefficient  Year  Coefficient  Year    Coefficient
   1950     .58       1960     .65       1970     .35       1980        .43    
   1951     .64       1961     .58       1971     .27       1981        .45    
   1952     .64       1962     .57       1972     .34       1982        .44    
   1953     .63       1963     .55       1973     .39       1983        .39    
   1954     .61       1964     .55       1974     .37       1984        .37    
   1955     .66       1965     .30       1975     .43       1985        .36    
   1956     .72       1966     .31       1976     .41       1986        .34    
   1957     .70       1967     .36       1977     .41       1987        .30    
   1958     .70       1968     .36       1978     .46       1988        .28    
   1959     .68       1969     .36       1979     .45       1989        .27    
                                                            1990        .26    
                                                            1991(*)     .23    
   Period        Average Coefficient of Participation                          
   1950-59                     .66                                            
   1960-69                     .46                                             
   1970-79                     .39                                             
   1980-85                     .41                                             
   1986-90                     .29                                             
   *Preliminary data.                                                          
Source: Computed from NAFINSA, La Economia Mexicana en Cifras |36, Tables 3.1 and 3.2, pp. 68-70; and NAFINSA, op. cit., |1992.

The government's pro-business policies, however, would be short-lived, especially in light of the country's deteriorating social indicators which generated mounting political pressure from key sectors of society, primarily urban labor, peasants and students, to confront and resolve them. Beginning in 1972, the Echeverria administration (1970-76), in an effort to diminish both social tensions and maintain the rate of economic growth, embarked upon an ambitious populist program which increased the state's expenditures on collective consumption goods while, at the same time, raising the rate of economic growth and profits via increased real spending in the capital goods sector. This explains the dramatic rise in the state's coefficient of participation of 16 percentage points between 1971 and 1975.(6)

However, the country's inefficient and regressive tax system was not up to the task of financing this dramatic increase in real government spending, so, by the end of the Echeverria term the country was caught in the vicious circle of rising public-sector deficits, excessive money growth, accelerating inflation, rapid capital flight, and mounting foreign debt. Had it not been for the discovery of vast deposits of oil and gas reserves in 1976-77, the country would have had to adopt many of the market-oriented measures that it eventually implemented under the administrations of Miguel de la Madrid (1982-88) and Carlos Salinas de Gortari (1988-1994).(7) As it turned out, these badly needed reforms of the country's fiscal and productive apparatus, as well as the export-import sector, were postponed during the petroleum-led boom of the 1977-81 period (this accounts for the increase in the state's coefficient of participation from .41 in 1977 to .45 in 1981.

The onset of the debt crisis in 1982, and its immediate aftermath, however, left the Mexican government with no choice but to alter radically its overall development strategy. Under the auspices of the IMF and World Bank, and in close consultation with the country's key economic grupos, it began a process which has culminated in the unprecedented opening of the Mexican economy to foreign competition and investment, as well as the total withdrawal of the state from key sectors of the economy.(8) Table I shows clearly this trend, particularly after the election of Carlos Salinas de Gortari.

In a market economy, where private investment is the driving force of economic growth, these are welcome developments, but, there is danger in too much of a good thing. The across-the-board cuts in government spending called for by expenditure-reducing policies tend to fall most heavily on central government capital expenditures because the negative effects are not evident to the public immediately. However, what is politically expedient in the short run may turn out to be, in the long run, very bad economics, particularly if empirical evidence suggests that significant complementarities are present between certain types of public and private investment. Moreover, expenditure-switching policies such as real devaluation may further aggravate private capital formation in countries such as Mexico where imports of key intermediate goods and capital inputs comprise a large fraction of prime costs; i.e., real devaluation may reduce profits and investment levels since it may increase overall costs prohibitively.(9) Table II above shows that the harsh and indiscriminate austerity measures adopted not only in Mexico during the eighties, but in many other Latin American nations as well, have had a particularly negative effect on the level and rate of capital formation. Mexico's investment growth rate experienced an overall deterioration of over 12 percentage points, and its investment coefficient fell from an average of 23 percent during the seventies to 19.6 percent in the "lost" decade of the eighties.

Perhaps more revealing--and of particular relevance to this study--is what has happened to the country's public and private investment during the 1970-90 period. Figure 1 below displays annual percentage changes for both public and private gross fixed investment during the 1970-90 period. It can be seen that throughout the eighties public and private investment experienced negative growth rates except for 1984 and 1988-89. The figure also reveals that changes in public investment expenditures are closely mirrored by changes in private investment, thereby suggesting that the drastic cuts in public investment, particularly in social and economic infrastructure, may have been responsible for the overall deterioration in the country's investment performance.


III. Public Investment and the Complementarity Hypothesis

In recent years, a number of researchers |6; 4; 5; 3; 35; 16; 30; 41; 43; 46h ave focused their attention on the complex and controversial question of whether public and private investment are complements, substitutes, or, as assumed in most macroeconomic models, independent of one another. The renewed interest in endogenous growth theory, for example, has also refocused the attention of prominent investigators such as Easterly |15, on the potential impact of public investment on long-term economic growth.

The economic rationale behind most of these studies resides in the idea that public investment is confined, by and large, to those goods and services that the private sector will not produce in optimal amounts because it is both hard to ration their use and benefits to paying customers (the free rider problem), and, they are subject to substantial start-up costs (lumpy and indivisible investments) . These public goods are, nevertheless, of critical importance to the proper functioning of a market system because they tend to generate large and widespread spillover benefits.(10) For example, investments by the state in social and economic infrastructure tend to complement private capital formation because they facilitate the implementation and realization of private agents' investment plans through their elimination of transportation, communications, and educational bottlenecks. Public investment also has an indirect effect on private capital formation because it not only augments overall aggregate demand for the goods and services produced by the private sector, but also influences private investors' future profit and sales expectations. Lastly, public investment in streets and highways, airports, ports, electrical and gas facilities, mass transit, water and sewers, health facilities, and education services, generates increases in both total factor productivity and labor productivity (output per worker hour).

On the negative side of the ledger, public investment undertaken by heavily subsidized and inefficient state-owned enterprises in agriculture, manufacturing, capital goods, energy, banking and financial services, has, more often than not, reduced the possibilities for private investment and long-term economic growth. This is particularly the case if these investments are undertaken, as many were during the ISI era, in markets with high rates of effective protection. In addition, the financing of public capital expenditures via external and internal indebtedness, the printing press, and in general, the repression of what is, in many developing countries (including Mexico), a structurally weak private financial system, has crowded-out the private sector from potentially profitable investment outlets. The combined negative effects of these price and quantity constraints can, in developing nations, become binding in nature very quickly, thus outweighing any positive direct and indirect effects claimed by public investment enthusiasts.

Thus, the impact of public investment on private capital formation is uncertain, complex, and, as indicated above, subject to a number of important caveats. Nevertheless, it is possible to formalize and analyze the potential impact of public investment on private capital formation in Mexico by appealing to the modified neoclassical production function given in equation (1) below.

|Mathematical Expression Omitted

where Y is the level of real output; N denotes employment; |K.sub.p is the stock of private capital; |K.sub.g refers to the public capital stock; and | Epsilon denotes a shift parameter of the production function.

By treating the public capital stock as a separate input in the production function, a ceteris paribus increase in public investment gives rise to three conceptually distinct effects. Consider first the case where the public capital stock is productive and complements the private capital stock: a ceteris paribus increase in the public capital stock will increase output directly in the same way that an increase in any other factor of production raises output (|F.sub.3 |is greater than 0). Secondly, it will indirectly increase private investment and output by raising the marginal productivity of the private capital stock (|F.sub.23 |is greater than 0) relative to the real interest rate. Third, it will increase output via its positive impact on the marginal productivity of labor; i.e., by increasing the amount of both private and public capital per worker (|F.sub.12 and |F.sub.13 |is greater than 0).

Of course, in the case where public and private capital are direct substitutes, an increase in public investment via, say, the operations of state-owned enterprises, generates a positive direct effect, but a negative indirect effect that could more than offset it; i.e., when the following condition arises: |(|F.sub.3 + |F.sub.13) + (|F.sub.23) - (|F.sub.12) |is less than 0. Lastly, in the case where private and public capital are independent, a ceteris paribus increase in public investment will generate a direct positive effect on output.

IV. Empirical Model

Following the lead of Blejer and Kahn |9, Calderon |11, Looney |30, Sundararajan and Thakur |50, and Wai and Wong |53, one can proceed to estimatea flexible accelerator investment model that captures some of the institutional and structural characteristics of developing nations such as Mexico. The choice of the modified accelerator model over more conventional formulations of the neoclassical investment model resides in the fact that in countries such as Mexico there are, to begin with, no published capital stock series or reliable estimates for the rate of depreciation; and secondly, given the well-documented intervention of Mexican government officials in the operation of financial and factor markets (thereby preventing investors from equating the marginal product of their capital to its rental price), the available quantity of resources, rather than their cost or price, tends to be the binding constraint.(11) This has been particularly the case in the allocation of financial resources, where Central Bank officials have relied extensively on changes in the marginal reserve requirements of commercial banks and private financieras (development banks) to channel investible funds to government-approved projects.(12)

In the formulation of flexible accelerator models it is a fairly standard procedure to assume that the desired capital stock is proportional the level of expected output:

|K* = |Alpha|Y*.sub.t

where |K* is the capital stock that the private sector desires to have in period t, and |Y*.sub.t is the expected level of output in period t.(13) The actual stock of private capital is assumed to adjust to the difference between the desired stock in period t and the actual stock in the previous period:

|Delta| = |Beta(|K* - | - 1)(3)


| = |Beta|K* + (1 - |Beta)|


where |Beta is the coefficient of adjustment, such that 0 |is less than or equal to |Beta |is less than or equal to 1, and |Delta| is the change in the actual capital stock between two periods, or simply, net private investment. Equation (3) postulates that net private investment in any given time period t is some fraction |Beta of the desired change for that period. Although it is theoretically possible for the actual capital stock to adjust instantaneously to its desired level (|Beta = 1) or for no adjustment to take place at all (|Beta = 0), in practice, |Beta lies somewhere in between these two extremes. Equation (3a), on the other hand, is a convenient way of rewriting the partial adjustment model because it explicitly shows that the actual capital stock in period t is a weighted average of the desired capital stock and the capital stock in the previous period.

Since the data on investment for most developing nations (including Mexico) is in gross terms, equation (3) must be transformed into gross investment terms for estimation purposes.(14) Gross private investment, I|P.sub.t, is defined as,

I|P.sub.t = |Delta| + |Sigma|


where |Sigma is the rate of depreciation of the private capital stock. In l ag-operator notation, equation (4) can be written as,

I|P.sub.t = |1 - (1 - |Sigma)L|


L is the lag operator, and it is defined as, L| = | Inverting equation (5), we can relate the stock of private capital to the level of gross private investment as follows:

| = I|P.sub.t/|1 - (1 - |Sigma)L. (6)

Equation (6) can now be substituted in equation (3a) to obtain,

I|P.sub.t/ (1 - |Sigma)L = |Beta|K* + (1 -|Beta)I|P.sub.t-1 / |1 - (1 - |Sigma)L (7)

multiplying both sides of equation (7) by the term in brackets, it has solution,

I|P.sub.t = |Beta|1 - (1 - |Sigma)L|K* + (1 - |Beta)||Phi.sub.t-1. (8)

If we now substitute for |K* from equation (2), we obtain an expression for a basic accelerator model for gross private investment:

I|P.sub.t = |Alpha|Beta|1 - (1 - |Sigma)L|Y*.sub.t + (1 - |Beta)I|P.sub.t-1.(9)

Equation (9) has the advantage of not requiring information on the capital stock or net private investment, and therefore, it can be readily applied to available gross investment data for the Mexican economy.

For estimation purposes, it is perhaps more convenient to derive equation (9) by directly specifying a partial adjustment function for I|P.sub.t rather than |,

|DeltaI|P.sub.t = |Beta (I|P*.sub.t - I|P.sub.t-1)(10)

where I|P*.sub.t is the desired level of gross private investment. In the s teady state, I|P*.sub.t is given by:

I|P*.sub.t = |1 - (1 - |Sigma)L|K*.sub.t.(11)

Combining equations (10) and (11) and solving for I|P.sub.t, and then using e quation (2), yields an equation exactly equal to equation (9) above.

At this juncture, one can use equation (11)--after substituting from equation (2) for the desired capital stock--to specify desired gross private investment not only as a function of the desired level of real output, but also of a number of relevant variables such as, the output gap, changes in real bank credit to the private sector, the real exchange rate, public investment, and qualitative factors.

In this study, the output gap, or the difference between actual and trend output, is incorporated to pick up cyclical factors. The impact of this variable on gross private investment is ambiguous because, on the one hand, when aggregate demand is buoyant, i.e., when output is above trend, private investors can be expected to adjust their actual investment more quickly towards their desired level. On the other hand, investment could respond more rapidly in situations of moderate excess capacity because financial and real resources are relatively plentiful and little upward pressure is exerted on input costs.

Changes in real bank credit to the private sector, or the real interest rate, can also be included as arguments in equation (11) depending on whether one believes quantity or price (cost) constraints are relatively more important. In the Mexican case, the institutional evidence for the period in question suggests that the binding constraint is the availability of financing;(15) it is expected to have a positive and significant impact on gross private investment.

Next, public investment is added to equation (11) to test whether overall public investment complements or displaces private capital formation. In view of the fact that public investment includes not only expenditures on social and economic infrastructure, which tend to "crowd in" private investment, (16) but also investment expenditures undertaken by state-owned enterprises in sectors that produce marketable output in direct competition with the private sector thereby crowding out private investment --it is not altogether clear what the net effect will be.

The real exchange rate is also included as an argument in the estimation of equation (11) because in countries such as Mexico, where dollar-denominated imports of intermediate goods and capital inputs comprise at least 80 percent of the country's total imports, real devaluation will increase import costs,(17) and ceteris paribus, lower profit and investment levels, particularly in those industries that lack mark-up rules for passing rising costs to consumers in the form of higher prices.

Lastly, the impact of qualitative factors such as the petroleum-led boom of the late seventies, and the dramatic real devaluation of the peso during the 1982-83 and 1986-87 years, are included via the addition of dummy variables.

Adding these variables to equation (11) yields the following expression,

I|P*.sub.t = |Alpha|1 - (1 - |Sigma)L|Y*.sub.t + |a.sub.1|GAP.sub.t + |a.sub. 2|DeltaRC|P.sub.t + |a.sub.3|RGI.sub.t + |a.sub.4|EX.sub.t + |a.sub.5|D.sub.t (12)

here GAP is the output gap, |DeltaRCP refers to the flow of real credit to the private sector, RGI is public investment, EX denotes the real exchange rate, and D is a dummy variable. Substituting equation (12) into equation (10), and solving for the actual level of gross private investment, I|P.sub.t, yields:

I|P.sub.t = |Alpha|Beta|1 - (1 - |Sigma)L|Y*.sub.t + |Beta|a.sub.1|GAP.sub.t + |Beta|a.sub.2|Delta|RCP.sub.t + |Beta|a.sub.3|RGI.sub.t + |Beta|a.sub.4|EX.sub.t + |Beta|a.sub.5|D.sub.t + (1 - |Beta)I|P.sub.t-1 (13)

The only remaining unobservable variable in equation (13) is |Y*.sub.t it can be estimated in a variety of ways. For example, one can compute moving averages of lagged values of real output or fit a more complicated autoregressive process (where the predicted values of the regression are defined as expected output). Unfortunately, these methods of estimating desired output sacrifice needed degrees of freedom for estimating the relevant parameters--a problem that assumes added importance when dealing with the limited set of investment data for developing nations in general, and Mexico in particular.

Thus, for estimation purposes, this study fitted a first-order autoregressive process of the form,

|Y.sub.t = ||Tau.sub.0 + ||Tau.sub.1|Y.sub.t-1 + |Epsilon (14)

where ||Tau.sub.0 is the average level of real output and ||Tau.sub.1 is the autoregressive parameter. The predicted values from equation eq. (14) were defined as expected output.(18)

Lastly, a direct estimate of the rate of depreciation, |Sigma, is needed to e stimate equation (13). Since there are no official published capital stock data for Mexico, we followed the lead of Reynolds |44 and Looney |30 and used th eir arbitrarily chosen value of 5 percent. As it turned out, other estimates of the rate of depreciation did not alter significantly the robustness of the statistical results.(19)

V. Results

Various specifications and functional forms of equation (13) were estimated using annual data for Mexico during the 1950-88 period. They are reported in Tables III and IV below. Real output (Y) is defined as real gross domestic product (RGDP), and TRGDP refers to the trend level of real output;(20) both variables are measured in millions of 1970 pesos; RPI and RGI are, respectively, gross private investment and public investment expenditures in millions of 1970 pesos; TRGI denotes the trend level in real government investment spending, and |DeltaRGI refers to the change in public sector investment; RCPS is the stock of real credit available to the private sector at the end of each calendar year (in billions of real pesos), while |DeltaRCPS denotes the flow of credit to the private sector; EX represents the real exchange rate (1970 pesos per dollar of 1970); DUM1 is a dummy variable that takes on a value of 1 for 1978-81 (petroleum boom years) and 0 otherwise, while DUM2 takes on a 1 for 1982-3 and 1986-7 (economic crisis years); the symbol "In" denotes the natural logarithm, i.e., log to the base e; and finally, the t-ratios are given in parentheses.

Lags for the government investment variable were included in the estimated equations because it takes time for government expenditures on social and economic infrastructure to have their full effect on private capital formation. The nature and length of the lag structure was determined from examining the relevant data and fitting a second degree, three-lag polynomial to the basic investment equation. The results indicated that public investment was most significant when lagged one period.(21)

Turning to Tables III-IV, it can be seen that the various equations have a relatively high degree of explanatory power as measured by their adjusted coefficients of determination, and more importantly, the Durbin-h statistics suggest that serial correlation is not a major problem in the sample data. The variables RGI and RCPS have a positive and significant impact on gross private capital formation, thereby suggesting that both the availability of credit and public investment are important determinants of private investment. The accelerator effect is captured by the coefficient of the |DeltaRGDP term, and as expected, it has the correct sign and is highly significant for the various estimated regressions. Table III-IV also indicate that real private investment is negatively related to the real exchange rate (EX), implying that the real cost of the country's substantial imports of capital and intermediate goods rises with devaluation. The robustness of these results are not altered when including the output gap variable, whose sign and statistical significance suggests that when real output is below trend, gross private investment rises, ceteris paribus. Lastly, the estimated results are significantly improved when greater variability is inserted into the model in TABULAR DATA OMITTED the form of changes in public investment, as well as deviations from its trend value.

Table III also presents results that incorporate the effect of changes in government policy on private capital formation, as well as the impact of the economic crisis and the accompanying stabilization measures implemented during the decade of the eighties. The coefficient for DUM1 TABULAR DATA OMITTED in equation (6) suggests that the unprecedented promotion of basic industry (primarily petroleum and gas) during the Lopez Portillo administration (1976-82) increased private capital formation by a cumulative 10.4 billion pesos. On the other hand, the negative coefficient for DUM2 in equation (7) indicates that the debt crisis and its aftermath have had a devastating cumulative impact on private capital formation. Notice that the inclusion of these qualitative variables does not alter the sign or significance of the credit and public investment variables.

For ease of interpretation, Table IV displays results of the accelerator model in logarithmic form since it allows us to read the coefficients as elasticities, and also because logarithmic functions minimize the impact of outliers in the sample data. As can be seen from equations (1) thru (5), the coefficient for ln RG|I.sub.t-1 suggests that a ceteris paribus increase of 10 p ercent in public investment generates an increase in gross private capital f ormation of between 2-3 percent within one year. The results also indicate thatt he output gap, real exchange rate, and the deviations of public investment from its trend value, contribute to explaining variations in gross private investment.

   Table V. Pairwise Granger Causality Tests                                   
                                                 F-statistics     Probability  
   Two lags:                                                                   
   RPI is not Granger-caused by RGI                6.309(**)         0.006     
   RGI is not Granger-caused by RPI                0.529             0.596     
   Three lags:                                                                 
   RPI is not Granger-caused by RG                17.517(**)         0.001     
   RGI is not Granger-caused by RPI                0.174             0.913     
   Four lags:                                                                  
   RPI is not Granger-caused by RGI                5.874(**)         0.003     
   RGI is not Granger-caused by RPI                0.352             0.839     
   Five lags:                                                                  
   RPI is not Granger-caused by RGI                4.753(**)         0.008     
   RGI is not Granger-caused by RPI                0.862             0.527     
   ** Denotes significance at least at the 5 percent level.                    

The coefficients of adjustment are displayed in the last row of Table IV. For example, |Mu = 1 - .68 = .32 for equation (2), implying that 32 percent of the a djustment of actual investment towards its desired level takes place within one year. To obtain the long-run elasticity for the public investment variable in equation (2), all one needs to do is divide equation (2) by |Mu and drop thel n RP|I.sub.t-1 term. The estimated long-run elasticity of public investment is now .63, which is substantially greater than the corresponding short-run elasticity of .20. Lastly, the Durbin-h statistics for the various specifications in Table IV suggest that serial correlation is not a major problem in the data sample.

An important omission from most recent studies investigating the relationship between public and private investment expenditures is their failure to test for cointegration among the relevant variables. In this study, Augmented Dickey-Fuller (ADF) tests are applied to the residuals from the cointegrating regressions to determine whether there is a long-term relationship between the variables in question, thus addressing the important question of spurious correlation.(22) For most of the equations displayed in Tables III-IV, the ADF t-statistics lead to a rejection of the null hypothesis of a unit root (nonstationarity) in the residuals when compared to relevant MacKinnon critical values (at the 1%, 5% and 10% levels);(23) i.e., the series in question are cointegrated, implying that the relevant regressors are both stable and statistically important in explaining the variation in gross private investment.

Before concluding, this study also addressed the question of whether changes in private capital formation precede changes in public investment rather than, as assumed in this and other studies, the other way around. The rationale here is that initial investments by private entrepreneurs in, say, the assembly-line industry along the U.S.-Mexico border will, in due time, induce the public sector to provide essential investments in social and economic infrastructure. Although examination of the relevant data, as well as institutional information about the Mexican economy, suggest the estimation of pairwise Granger causality tests with one and two lags, it is better to use more rather than fewer lags to ensure the robustness of the results over different lag structures.(24) As shown in Table V above, the results for two, three, four and five lags indicate one can reject the null hypothesis that gross private capital formation is not Granger caused by public investment but not the other way around, implying that changes in public investment do precede and add significantly to the explanation of variations in private investment expenditures.

IV. Conclusions

This paper developed a conceptual and empirical model to test the hypothesis of complementarity between public investment spending and private capital formation in the Mexican case. It briefly reviewed the role of the Mexican state in the process of capital formation: a role which, in view of the debt crisis and the demise of ISI, has diminished significantly. It was argued that the government's indiscriminate cuts in public spending, although politically expedient, may in the long run jeopardize the efficiency gains of the ongoing deregulation and privatization program because they have fallen disproportionately on the country's capital expenditures.

Next, the paper proceeded to show how the stock of public capital could be treated as a separate input in a standard neoclassical production function, thereby generating direct effects on output and indirect ones on the other factors of production, such as the marginal productivity of the private capital stock. It was shown that when the public capital stock is productive and complements the private capital stock, a ceteris paribus increase in public investment would have overall positive effects on factor productivity and output.

For estimation purposes, a modified accelerator model was derived that circumvented some of the data problems present when dealing with developing nations such as Mexico. The estimated results indicated that, in addition to other relevant variables, public investment had an overall positive and significant effect on gross private investment expenditures during the 1950-88 period. The results also suggested that the deflationary measures implemented throughout the eighties had a negative impact on gross private investment. In light of these estimates, policy-makers might do well to pay close attention not just to the level of government expenditures but also to its composition between consumption and investment goods, and insofar as the latter is concerned, a concerted effort should be made to maintain adequate levels of investment in social and economic infrastructure.

To ensure the robustness of the statistical results, various specifications of the basic accelerator model were run, and more importantly, cointegration tests were performed to determine whether the underlying relationships among the regressors were stable and reliable. Most of these tests led to the rejection of a unit root in the residuals from the cointegrating regressions. Lastly, the paper performed pairwise Granger "causality" tests to determine the direction of correlation between public and private investment expenditures. Using different lag structures, the estimated results suggested that changes in public investment precede changes in private capital formation in the Mexican case.

1. For further detail on the rationale, evolution, and impact of ISI, see Kay |25; Kuczynski |27; Edwards |18; Ranis and Chenery |12; 44, and Hirschman |23.

2. For a thorough discussion of the growth and structure of the parastate sector in Mexico, see chapters 5 and 6 in James M. Cypher |14; the Chilean casei s discussed in Pinera and Glade |39; lastly, a good overview of the changing role of the state in Latin America during the 1980s is provided by Fishlow |20 and Meller |33, 169-206.

3. A good overview of the privatization process in Mexico over the last seven years is given by Oscar H. Ferrer |19. For the Chilean case, see Luders |31.

4. In regard to the essential duties of the "sovereign," see Adam Smith |49, 208-19 Book IV, Chapter IX.

5. An analysis of the economic policies implemented during this period and their impact on the Mexican economy can be found in Brothers and Solis |10, Hansen |24, and Reynolds |45.

6. Carlos Bazdresch and Santiago Levy |8 provide an excellent analysis of the populist policies pursued under the Echeverria administration. See also Cypher |14, ch. 4.

7. An informed, if not complete, discussion of the petro boom years in Mexico and the ensuing economic breakdown of 1981-1982 is provided in Miguel D. Ramirez |41, 85-95. See also Looney for further detail |30.

8. For further details see Cypher |14, 107-26, Ferrer |19, and Ramirez |42,85 -92; the Chilean case is discussed by Corbo and Solimano |13, Fontaine |21 and Larrain |29.

9. Paul Krugman and Lance Taylor |26 argue that real devaluation can generate contractionary effects on output and real wages in small open economies that import a significant fraction of their overall intermediate goods and capital inputs. For further detail see Taylor |51, 12-27; 52.

10. The now classic explanation for market failure in a developing country and the need for state intervention in the provision of social overhead capital is found in Nurske |37, Rosenstein-Rodan |47, and Scitovsky |48.

11. For further detail see Brothers and Soils |10, Basch |7, and Ramirez |41, 153-63. Also, examination of the data for nominal interest rates on government bonds (CETES) revealed that the rate during the fifties, sixties, and seventies was maintained at fixed levels for several years before being abruptly changed by government fiat. Since the inflation rate was highly variable during this period, the real interest rate was essentially picking up changes in the price level. Interest rate data is available upon request.

12. Basically, the central bank induced privately owned deposit banks (and subsequently private development banks or financieras) to meet their marginal reserve requirements by acquiring government bonds and/or issues of public enterprises. For example, beginning in 1968, 40 percent of each additional peso deposited in a commercial bank had to be held either in government securities and/or government-approved investments. Otherwise, it became subject to a 100 percent legal reserve requirement. See Basch |7 and Ramirez |41.

13. |Y.sub.t can also be viewed as future aggregate demand.

14. The investment data used in this study has been obtained from Nacional Financiera, S.A. (NAFINSA), La Economia Mexicana en Cifras |36, various issues,a nd OECD Economic Surveys |38. Other relevant economic data has been obtained from the Central Bureau of Statistics (CBS) of the National Institute of Statistics, Geography and Informatics (INEGI), and is available upon request from the author.

15. As indicated earlier, the Mexican government has intervened heavily in financial markets through its placement of CETES (government bonds) and its extensive control over the legal reserve requirements of the private banking system. For example, beginning in 1956, 15 percent of additional deposits of commercial banks had to be held as cash in the Banco de Mexico; 25 percent had to be invested in government-endorsed growth industries; 35 percent had to be invested in government bonds; and 25 percent could be invested freely. The yields on the regulated investments were not as competitive with those realized by the banks on other investments, however, failure to comply with these requirements meant that any additional funds raised were subject to a 100 percent legal-reserve requirement. See Ramirez |41, 154 and Basch |7,53.

16. Unfortunately, the data on public investment is not sufficiently disaggregated and/or available over a long enough period to be able to determine separately the impact of expenditures on social and economic infrastructure.

17. Computed with data obtained from Nacional Financiera, S.A. |36. Import data is available upon request.

18. In practice, the estimated results did not differ significantly whether one used the predicted or the actual values for real gross domestic product.

19. For example, the estimated results did not differ both in terms of significance or expected sign when one used a rate of depreciation of 0, 5, 8 or 10 percent.

20. The trend value of real gross domestic product (RGDP) was calculated as follows: TRGDP = RGD|P.sub.0|, where RGD|P.sub.0 is the initial valueof RGDP, g is the instantaneous growth rate of RGDP, and t is a linear time trend. For estimation purposes, we take the natural logarithm of both sides and regress real output on a time trend.

21. The lag distribution of RGI is available from the author upon request.

22. For further detail see Engle and Granger |17, 251-76.

23. In a recent paper, MacKinnon |32 has generated a much larger set of replications than those found in standard Dickey-Fuller tables. TSP has incorporated these critical values for any sample size in its estimation procedures. The null hypothesis of a unit root in the residuals of the cointegrating regression implies that the expected value of the t-statistic is zero. If the residuals are stationary, however, the t-statistic will be negative and greater, in absolute value, than the MacKinnon critical values at the 1%, 5% and 10% levels.

Most of the E-G cointegrating regressions run in this paper included up to three lagged first differences of the residuals because inspection of the test regression revealed they were highly significant in explaining the variation in the first difference of the residuals. (The results are available upon request.)

24. For further detail see Granger |22, 424-28. The pairwise Granger causality tests in this study were estimated over the 1950-81 period because, with the onset and aftermath of the debt crisis in 1982, both the public and private investment data were highly unstable.


1. Aschauer, David A., "Government Spending and the Falling Rate of Profit." Economic Perspectives, May/June 1988, 11-17.

2. -----, "Public Investment and Productivity Growth in the Group of Seven." Economic Perspectives, September/October 1989, 17-25.

3. -----, "Does Public Capital Crowd Out Private Capital? " Journal of Monetary Economics, September 1989, 171-88.

4. Barro, Robert J., "Output Effects of Government Purchases." Journal of Political Economy, December 1981, 1086-1121.

5. -----, "Government Spending in a Simple Model of Endogenous Growth." Journal of Political Economy, October 1990, 103-25.

6. Barth, James R. and Joseph J. Cordes, "Substitutability, Complementarity, and the Impact of Government Spending on Economic Activity." Journal of Economic and Business, Spring 1980, 235-42.

7. Basch, Antonin. El Mercado de Capitales en Mexico. Mexico City: Centro de Estudios Monetarios Latinoamericanos, 1968.

8. Bazdresch, Carlos and Santiago Levy. "Populism and Economic Policy in Mexico, 1970-1982," in The Macro-economics of Populism in Latin America, edited by Rudiger Dornbusch and Sebastian Edwards. Chicago: The University of Chicago Press, 1991.

9. Blejer, Mario I. and Mohsin S. Khan, "Government Policy and Government Investment in Developing Countries." IMF Staff Papers, June 1984, 379-403.

10. Brothers, Dwight S. and Leopoldo M. Soils. Mexican Financial Development. Austin: University of Texas Press, 1966.

11. Calderon, Francisco, "La Inversion Privada en Mexico, 1970-1987." CIDE: Documentos de Investigacion, Division de Economia, September 1988, 1-39.

12. Chenery, Hollis. "Comments on 'Challenges and Opportunities Posed by Asia Superexporters: Implications for Manufactured Exports from Latin America,'" in Export Diversification and the New Protectionism, edited by Werner Baer and Malcolm Gillis. Urbana, Illinois: Bureau of Economic and Business Research, 1981.

13. Corbo, Vittorio and Andres Solimano. "Chile's Experience with Stabilization, Revisited." PRE Working Paper #579. Washington, D.C.: The World Bank, January 1991.

14. Cypher, James M. State and Capital in Mexico. Boulder, Colorado: Westview Press, 1990.

15. Easterly, W. R. "Endogenous Growth in Developing Countries with Government-Induced Distortions," in Adjustment Lending Revisited: Policies to Restore Growth, edited by Vittorio Corbo and Stanley Fisher. Washington, D.C.: I.B.R.D., 1992.

16. Eberts, Randall, "Estimating the Contribution of Urban Public Infrastructure to Regional Growth." Federal Reserve Bank of Cleveland, 1986, Working paper No. 8610.

17. Engle, R. F., and Granger, C. W. J., "Cointegration and Error Correction: Representation, Estimation and Testing." Econometrica, March 1987, 251-276.

18. Edwards, Sebastian. "Structural Adjustment Policies in Heavily Indebted Countries," in Developing Country Debt and the World Economy, edited by Jeffrey Sachs, Chicago: The University of Chicago Press, 1989.

19. Ferrer, Oscar H. "The Political Economy of Privatization in Mexico," in Privatization of Public Enterprises in Latin America, edited by William Glade. San Francisco, California: ICS Press, 1991.

20. Fishlow, Albert, "The Latin American State." Journal of Economic Perspectives, Summer 1990, 61-74.

21. Fontaine, J. A. "The Chilean Economy in the 1980s: Adjustment and Recovery," in The Debt Crisis, edited by S. Edwards and F. Larrain. London, England: Blackwell, 1989.

22. Granger, C. W. J., "Investigating Causal Relations by Econometric Models and Cross-Spectral Methods." Econometrica, July 1969, 424-438.

23. Hirschman, Albert O. "The Rise and Decline of Development Economics," in Essays in Trespassing: Economics to Politics and Beyond, edited by Albert Hirschman. Cambridge: Cambridge University Press, 1981.

24. Hansen, Roger D. Mexican Economic Development: The Roots of Rapid Growth. Washington, D.C.: National Planning Association, 1971.

25. Kay, Cristobal. Latin American Theories of Development and Underdevelopment. New York: Routledge, 1989.

26. Krugman, Paul, and Lance Taylor, "Contractionary Effects of Devaluation." Journal of International Economics, August 1978, 445-56.

27. Kuczynski, Pedro-Pablo. Latin American Debt. Baltimore, Maryland: The Johns Hopkins University Press, 1988.

28. Kuznets, Simon. "Towards a Theory of Economic Growth," in National Policy for Economic Welfare at Home and Abroad, Bicentennial Conference Series, edited by Robert Levachman. New York: Doubleday and Co., 1955.

29. Larrain, M. "How the 1981-83 Chilean Banking Crisis was Handled." PPR Working Paper #300. Washington, D.C.: The World Bank, 1989.

30. Looney, Robert E. Economic Policymaking in Mexico. Durham, N.C.: Duke University Press, 1985.

31. Luders, Rolf J., "Massive Divestiture and Privatization: Lessons From Chile." Contemporary Policy Issues, October 1991, 1-19.

32. MacKinnon, James G. "Critical Values for Cointegration Tests." Working Paper. San Diego: University of California, January 24, 1990.

33. Meller, Patricio. "IMF and World Bank Roles in the Latin American Foreign Debt Problem," in The Latin American Development Debate, edited by Patricio Meller. Boulder, Colorado: Westview Press, 1991.

34. Munnell, Alicia., "Why Has Productivity Growth Declined? Productivity and Public Investment. " New England Economic Review, January/February 1990a, 3-22.

35. -----, "How Does Public Infrastructure Affect Regional Economic Performance?" New England Economic Review, September/October 1990, 11-33.

36. NAFINSA. La Economia Mexicana en Cifras. Mexico City: 1984, 1986, and 1991.

37. Nurske, R. Problems of Capital Formation in Under-Developed Countries. Oxford University Press, 1953.

38. OECD. OECD Economic Surveys: Mexico. Pads, France: OECD, 1992.

39. Pinera, Jose and W. Glade. "Privatization in Chile," in Privatization of Public Enterprises in Latin America, edited by William Glade. San Francisco: ICS Press, 1991.

40. Ram, Rati. "Government Size and Economic Growth." American Economic Review, March 1986, 191-203.

41. Ramirez, Miguel D. Development Banking in Mexico: The Case of Nacional Financiera, S.A. New York: Praeger, 1986.

42. -----. Mexico's Economic Crisis: Its Origins and Consequences. New York: Praeger, 1989.

43. -----, "Stabilization and Trade Reform in Mexico, 1983-1989." Journal of Developing Areas, January 1993, 173-90.

44. Ranis, Gustav, "Challenges and Opportunities Posed by Asia's Superexporters: Implications for Manufactured Exports from Latin America." Quarterly Review of Economics and Business, Summer 1981, 205-26.

45. Reynolds, Clark W. The Mexican Economy. New Haven: Yale University Press, 1970.

46. Riedel, James. "Public Investment and Growth in Latin America." Draft Manuscript. Baltimore, Maryland: The Johns Hopkins University School of Advanced International Studies, July 1992.

47. Rosenstein-Rodan, P. N., "Problems of Industrialization of Eastern and South-Eastern Europe." Economic Journal, June-September 1943, 202-11.

48. Scitovsky, Tibor, "Two Concepts of External Economies." Journal of Political Economy, April 1954, 143-51.

49. Smith, Adam. An Inquiry into the Nature and Causes of The Wealth of Nations, edited by Edwin Cannan. Chicago: The University of Chicago Press, 1976.

50. Sundararajan, V. and Subhash Thakur, "Public Investment, Crowding Out, and Growth: A Dynamic Model Applied to India and Korea." IMF Staff Papers, December 1980, 814-855.

51. Taylor, Lance. Structuralist Macroeconomics. New York: Basic Books, Inc., 1983.

52. -----. Varieties of Stabilization Experience. Oxford: Oxford University Press, 1991.

53. Tun Wai, U. and C. Wong, "Determinants of Private Investment in Developing Countries." Journal of Development Studies, October 1982, 19-36.