![Sample our Global Development journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5929/TOC-Subject-Sample-2021-Global-Development.jpg"})
Abstract
This paper addresses the important question of whether public investment spending on economic infrastructure enhances economic growth in Mexico. It estimates a Cobb-Douglas production function that includes public infrastructure capital. Using cointegration analysis, the paper estimates a vector error correction model (VECM) for the 1995 – 99 period. The results suggest that there is a long-term stable relationship among the variables included in the VECM. The evidence also indicates that both public infrastructure spending and private capital formation have a positive and highly significant effect on the rate of output growth. Finally, the impulse response functions (IRF) and the variance decompositions (VDC) of the endogenous variables in the VECM suggest that the response of private capital to public infrastructure is positive while the reverse causation is not affirmed. From a policy standpoint, the findings call into question stabilization policies that disproportionately reduce public infrastructure to meet targeted reductions in the fiscal deficit (JEL, O1, O47, O54).
Notes
For example, Devarajan and Zou (Citation1994) as well as Gramlich (Citation1994) have criticized the complementarity hypothesis because it assumes unrealistically that the positive spillover effects associated with public investment can be provided to additional users in a costless and efficient manner. In practice, government investment can substitute for private investment and inhibit future growth and productivity if it is undertaken by heavily subsidized and inefficient state-owned enterprises in key sectors of the economy. For example, in Mexico and elsewhere, many investments were undertaken in markets with high rates of effective protection, such as agriculture, mining and steel. In addition, the financing of public capital spending via external and internal indebtedness, the printing press, output taxes, and the repression of financial markets probably led to a significant crowding-out of private investment from potentially profitable sectors.
It can also be argued that the public sector need not provide these public goods directly; the goods can be contracted out to the private sector in accordance with government regulations and guidelines. In fact, many governments in Latin America (including Mexico) are in the process of awarding concessions to private firms to produce and provide quasi-public goods and services. However, as Prager (Citation1992) correctly observes, relatively little or no attention has been given to the monitoring or supervision costs of outsourcing public works projects. If these costs are substantial, particularly in the medium run, the bias in favour of privatizing these types of expenditures is removed.
To ensure the robustness of the results, other estimates of the rate of depreciation were used (10%), as well as different estimates of the initial capital stock (summing over 3 and 5 years), but the results were not altered significantly. The capital stock data are available upon written request.
The order of the lag length was determined by applying both the Akaike Information Criterion (AIC) and the Schwartz Bayesian Criterion (SBC). Lower values for these performance statistics indicate a better fit to the data. Between the two criteria, the SBC is more reliable because the AIC is known to be biased towards choosing an over-specified model. For all the variables in this study, the ADF tests with one lag showed the lowest value for both the AIC and SBC criteria.
A stochastic trend is one where the random component of the series itself, say variable xt , contributes directly to the long run pattern of the series, either upward or downward. However, in the case of a deterministic trend, the deviations from the non-stationary mean over time are quickly corrected. Thus, the former process is usually referred to as a random walk with drift, either positive or negative, and it is captured by including a constant in the DF or ADF tests. For further details see Charemza and Deadman (Citation1997: 84 – 92).
The AIC, SBC, and Hannan-Quin (HQ) information criteria suggested that the optimal lag value in the VECM is one period. That is, this specific lag ensures that the residuals are white noise.
The variables in question are also cointegrated with a constant term. Again, one cointegrating vector is present. The results are available upon written request.
This paper also carried out a pairwise Granger causality test to determine whether a group of endogenous variables can be treated as exogenous. The block exogeneity Wald test (available upon request) suggested that the variables l, kp and kgin can be treated individually and jointly as exogenous variables. That is, changes in these lagged (differenced) variables can be used to explain the variation in the (differenced) output variable. In all other cases, that is treating l, kp or kgin as ‘dependent variables’ and the remaining variables as ‘independent’, the Wald test indicated that each and all of the other lagged endogenous cannot be treated as exogenous, i.e. the null hypothesis of no exogeneity could not be rejected at the 5 per cent level.
In his estimation of a Cobb – Douglas production function, Looney (Citation1985: 29) uses a relatively high share for private capital (0.35 to 0.4) and argues that the high rate of return is explained by Mexico's ‘… numerous and obvious gaps in the capital structure to be filled’ as the industrialization effort was underway during the 1960 – 80 period. The presence of a positive public infrastructure-induced externality may help explain the frequently reported high estimate of the private capital elasticity in growth equations for developing nations (see Benhabib and Jovanovic, Citation1991).
The one-period VECM model for equation (6a) suggests that a deviation from long-run output growth during the current year is corrected by about 28.2 per cent in the next year (t-ratio = − 5.394). The results also indicate that the one-year lag in the growth rates of both public infrastructure and private capital are positive and statistically (and economically) significant (coefficients and (t-ratios) are, respectively, 0.74 (3.978) and 0.45 (3.03)). The Adj. R-squared is 0.664, the AIC = − 3.498, the SBC = − 3.092, and the F-statistic = 12.513. Stability tests were also undertaken to determine whether the null hypothesis of no structural break could be rejected at the 1 per cent level. The Chow breakpoint tests suggested that the hypothesis could not be rejected for the economic crises years 1976, 1982 and 1987. The complete results are available upon written request. Dynamic results for equation (5b) are basically the same as those of equation (5a) and are available upon written request.