Speed of Convergence And Economic Policy Instruments

This article presents an innovative methodology to empirically evaluate the response of the rate of convergence in the neoclassical growth model to changes in the conditional factors. We apply this method to test the sensitivity of the speed of convergence in the EU-15 during the period 1980-2002 to increases in human and public capital. INTRODUCTION The relation between private productivity and policy instruments has been a topic of considerable interest in the literature on economic development. A key aspect in the literature on convergence has been the determination of the speed of convergence towards the stationary state. This is determined by its own structural characteristics, among which technology, public policies and the population growth rate are included. Recently, in the debate on convergence, the role played by these conditioning factors has gained prominence – primarily public policy instruments – with the objective of establishing whether part of the observed convergence (if any) can be attributed to these instruments. In the context of the EU, development and cohesion policies have primarily supported public investment in infrastructures and in education, in addition to support policies for the private sector during the last 20 years. The analyses carried out suggest that these efforts have contributed positively to the process of convergence in productivity in the EU (Salinas et al., 2006). However, recent difficulties encountered in relation to the reduction of disparities in the EU have led to questioning to what extent the speed of convergence could be altered by means of public policies in the European Union. In spite of a clear interest in this matter, few papers have analyzed the sensitivity of the speed of convergence and the majority are theoretical research papers (Turnovsky, 2002, Gokan, 2003, Klump et al., 2008). Unlike these works, in this article a methodological proposal is put forth to estimate the response of the speed of convergence in the EU to changes in economic policy instruments. The analysis proposes two scenarios of convergence: the first one will make it possible to estimate the effect on the speed of convergence arising from changes in the instruments in the EU as a whole; and a second scenario where the effect on the speed of convergence resulting from changes in each one of the countries is considered separately. In this way an attempt will be made to identify not only the potential of various public policies to influence European convergence, but also the role they may play in each country toward achieving convergence in the EU. The proposal develops various scenarios with Matlab that will make it possible to evaluate the sensitivity of the parameter of convergence in the face of variations in the analyzed determining factors, while the other factors remain ceteris paribus. The article is structured as follows: in the next section, the proposal for empirically evaluating the sensibility of the speed of convergence in the face of changes in the conditioning factors is presented. Section 3 describes the data and presents the findings. Finally, in Section 4, the primary conclusions are presented. CONDITIONAL CONVERGENCE AND SENSITIVITY ANALYSIS As of the research carried out by Mankiw et al. (1992) and Barro y Sala-i-Martin (1995), the neoclassical growth model is commonly used in order to contrast Proceedings 23rd European Conference on Modelling and Simulation ©ECMS Javier Otamendi, Andrzej Bargiela, José Luis Montes, Luis Miguel Doncel Pedrera (Editors) ISBN: 978-0-9553018-8-9 / ISBN: 978-0-9553018-9-6 (CD) conditional convergence. The metodology proposed to examine the response of the rate of convergence consists of the estimation of the convergence equation, recovery of the structural parameters and evaluate the sensibility of the parameter of convergence under specific assumptions about changes in the model. In addition, we simulate those changes In the estimation of the model we used the routine programming in Matlab available in http://www.spatialeconometrics.com/. The simulations programming in Matlab for this work are available on requestby introducing an extension in the neoclassical growth model. To do so, the fixed effects model provides the basis -. The estimation method is determined by the Matlab programming used to perform the sensitivity analysis. Since GMM approach is suitable to deal with growth models in empirical work, we have also estimated the model with this method and we have reached similar results -, where it is assumed that each explanatory variable has a single coefficient; that is, it has the same effect on the dependent variable, while each individual variable has a different constant which represents the individual effect. Formally, the model to be estimated is the following: , , 1 , , , 1 ln ln( ) ln( ) i t i t i t i i t i t y y x u y                    ;(1) where it y is the annual GDP per employee of 1, ,   i N countries (N=15) and for an annual period of 1, ,   t T (T=22 from 1980 until 2002); , i t x is a vector of 1  k explanatory variables; i  is the individual effect; and , i t u is a disturbance term. In this case, the concept of conditioned convergence is analyzed by introducing heterogeneity in the model by means of the introduction of fixed effects and/or the effect of other variables, , i t x , assuming that the fixed effects i  are correlated with them. The estimation of the model (1) is carried out in deviations with respect to the mean using Ordinary Least Squares (OLS), making it possible to consistently estimate the parameters of the model associated with convergence-β and with the explanatory variables since its consistency will not depend on the specification of fixed effects i  which have been eliminated with the transformation. This estimation is also equivalent to estimating with the OLS model (1) or the model with the variables transformed by orthogonal deviations, a transformation proposed by Arellano (1988) Given the model , , , i t i t i i t y x u      he defines the transformation in orthogonal deviations as: , , ,( 1) , ( 1) 1 ( ) ( 1) i t i t i t i T T t y y y y T t T t                   -. The extension of the estimated model (1) takes place through the introduction of changes (simulated increases) in exogenous variables of interest. The simulated increases are based on the minimum of the selected variable and they continue increasing with a fixed value, inc , until reaching a maximum value that, in this case, has been fixed at 100%, since the variables are taken as the percentage of GVA; but both the minimum value as well as the maximum value could be modified with absolute flexibility. The variations generated with Matlab are then placed in rows, until they complete a column, and they continue being placed in the first row of the next column until it has been completed as has been mentioned above and the same process continues likewise until a maximum value of 100% is reached. Therefore, the number of columns will depend on the increase inc selected, which in our paper is 5%. Furthermore, the analysis of the variations has been approached in two ways: a. Evaluating variations of all of the countries at the same time: 1,1 2,1 1, ,1 1, 1,2 1,1 2,2 2,1 ,2 ,1 1 1, 1 2, 2, 1 , min( ) ( ) 100% t T S S T S S t T S T T T T S T X inc inc inc inc inc X inc inc 