50th Congress of the European Regional Science Association: "Sustainable Regional Growth and Development in the Creative Knowledge Economy", 19-23 August 2010, Jönköping, Sweden
Aims of the study are to estimate 1) contribution of infrastructure capital in productivity growth in Russia, and 2) level and spatial extension of the spillovers for different categories of infrastructure. We measure stock of the traditional infrastructure sectors: railways, highways, communication. We use data for 79 Russian regions, covered period is 1992 - 2007. The basic idea of econometric estimates is to expand a production function including infrastructure capital stock. We examine several different categories of public capital. The log-linear Cobb-Douglas form of the production function gives empirical model. We assume existence of spatial spillovers of the infrastructure elements and dependence of regional productivity on public capital of neighboring regions introducing into the model spatial weights matrix and a spatial lag component. There are several problems of econometric estimates relating to the model. The first one is correct specification of the spatial dependence, what includes construction of the spatial weights matrix. The proposed strategy is to run series of regressions using different spatial weighs matrixes. The second one is common trends of output and public capital. One of the proposed ways to resolve the problem is to use some forms of differences. The third problem involves missing variables; panel data and taking of the differences to some extent lessen this problem. Another problem is causality: does absence of progress in infrastructure capital reduce economic growth or does low growth of output decrease the demand for infrastructure? The endogeneity poses question of instrumental variables, the choice of a spatial lag of the predicted values of the dependent variable or of spatially lagged exogenous variables is considered. Infrastructure capital is a public good however its effects can be distributed uneven among different sectors of economic activity. To take this fact into account the proposed set of growth model estimations are done for two alternative production functions focusing on gross regional product and on manufacturing sector only. The results are as follows: - estimates of contribution of infrastructure capital in productivity growth in Russia; - evidence of infrastructure externalities; - estimates of infrastructure spatial spillovers for different categories.