The empirical research on the impact of infrastructure on economic growth began at the end of the last century with Aschauer (1989). This author examined the relationship between these two variables for the USA and observed a relevant and positive effect of investment in public infrastructure on growth. From there on, international empirical research on that topic remained not conclusive. Bom and Ligthart (2014), based on a meta-analysis on almost 70 studies for the period 1983-2008, found mixed evidence regarding the sign of the effect, and even when this is positive, they observed a broad range of variation of the estimated elasticities from growth to infrastructure investment.
Uruguay, a small Latin American country, has traditionally had a strong instability in its investment rate. Along the last century, the country has gone through periods of strong investment expansion and by others in which investment was less than the required replacement rate. (Román & Willebald, 2012). According to Calderon & Servén (2004), during the '80s and '90s, when other Latin American countries have made progress in transport infrastructure investment, Uruguay cannot avoid to undergone a negative tendency in road investment. For the period 200 to 2010, Capurro et al. (2014) stated that in term of transport infrastructure, has accumulated a backlog of investment in the national road network of almost 3% of GDP.
This paper empirically explores the causal relationship between public investment in road infrastructure and economic growth in Uruguay. We analyze whether road infrastructure investment leads -in the long run- to Uruguayan economic growth, or, alternatively, economic expansion drives this investment, or a bidirectional relationship exists between the two variables.
Based on aggregated data of public expenditure for the period 1988-2014, we perform a cointegration analysis and estimate of VEqCM models (Autoregressive Vectors with Mechanism of Correction to Equilibrium models). This methodology has the convenience of allowing analyze the long term and short term dynamics, simultaneously (Juselius, 2006). Additionally, the Granger test is implemented in order to identify the direction of the causality between both variables (Granger, 1988; Hendry, 2017). Finally, impulse-response simulations and variance decomposition are performed to analyze the response dynamics of the variables as a co0nsequence of exogenous shocks.
Two alternative models were estimated. Firstly, a simplistic model where the relationship between the road investment and economic growth was analyzed, without including any additional covariables within the endogenous vector. Secondly, and with the objective to capture the specific demand for infrastructure services, an alternative model was estimated. The second model introduces as a covariate (or control variable) an indicator of transported tons. In both models, a unique cointegration relationship (long-term equilibrium equations) between growth and investment was found, following the Johansen procedure.
Likewise, in both models, results of Granger causality test indicate that the direction of causality goes from GDP growth to road investment, but not conversely. This means that variations in GDP may cause changes in infrastructure investment, but the latter fails to have a significant effect on GDP, at least in the sense of Granger causality test. This contradicts the starting hypothesis of the presence of a bidirectional causal relationship between economic growth and investment in road infrastructure in Uruguay. The hypothesis of bidirectionality and feedback between economic growth and road investment can then be rejected for the Uruguayan economy in the period analyzed.
An additional interesting finding arose from the analysis of the equilibrium equations. In both models, residuals of the cointegration equations show significant deviations from the equilibrium, at the end of the period analyzed. More precisely, from 2010, in the first model, and from 2013, in the second model, the relation between growth and public road investment set away from equilibrium. The evidence that in both cases, relations located below the equilibrium at the end of the period shows a mismatch between road investment and the GDP Uruguayan dynamics. This result could be interpreted as a deficit in the transport infrastructure sector, given the economic growth dynamic.
Moreover, the impulse-response results, as well as the variance decomposition, present similar results. Both analyses show that the response of road investment to a growth innovation (an exogenous shock) is slow and delayed. Nevertheless, the responses are statistically significant in both cases. The response of road investment takes 10 quarters (in the first model) and 15 quarters (in the second one) to fully absorb a shock from economic growth.
Note that the lag of the response of investment in transport infrastructure with respect to economic growth strengthens the evidence that a long period of economic expansion is required for road public investment to take place.
This seems to indicate that for the impulse coming from certain activities related to transport, such as the wood load (which has increased significantly in the last decade), is translated into an increase in investment in infrastructure, it can take between two or three years.
The evidence of the existence of a cointegration relationship between road investment and economic growth in Uruguay is consistent with a broad part of the international and empirical findings for other economies (see, Achour & Belloumi, 2016; Chang & Nieh, 2004; Mbulawa, 2017; Serdaroğlu, 2016; Badalyan et al., 2014; Beyzaltar et al., 2014; Pradhan et al., 2013). Nevertheless, results from the Granger causality analysis attested that causality is unidirectional and goes from growth to investment, and not inversely (neither bidirectional) as is found in most of the reviewed precedents. However, as mentioned before, the international evidence is not conclusive in this regard and our results seem to corroborate this indeterminacy.
This inverse causality between
public road investment and growth ascertained found in the case of Uruguay
could be related to several factors, such as Achour & Belloumi (2016)
suggest for the Tunisian economy. From our point of view, one of the most plausible
explanations is related to what Kustapeli et al. (2012) called "the
inefficiency in the use of the investment". This inefficiency may be
caused because, in the analyzed period, economic growth has preceded public
expenditure in Uruguay. Nevertheless, this topic should be the subject of
future research.
Classification-JEL: R1
Keywords: Inversión en Infraestructura Vial y Crecimiento, Causalidad, Cointegración, Modelos de Corrección al Equilibrio, Uruguay, Road Infrastructure Investment And Growth, Causality, Johansen Cointegration Test, Equilibrium Correction Models, Uruguay
Pages: 177-211
Volume: 2
Year: 2020
File-URL: http://www.revistaestudiosregionales.com/documentos/articulos/pdf-articulo-2596.pdf
File-Format: Application/pdf
Handle: RePEc:rer:articu:v:2:y:2020:p:177-211