44th Congress of the European Regional Science Association: "Regions and Fiscal Federalism", 25th - 29th August 2004, Porto, Portugal
Labor productivity convergence is a key factor in the catching up process of less developed regions. For the regional economies as a whole labor productivity differentials can be traced back to three distinct determinants: - composition effects due to the peculiar structure of the regional economy; a lower than average productivity level could, for instance, be due to the fact that a greater share or the regional labor force is employed in sectors that are denoted by lower productivity at the aggregate level; - different regional endowments, within each given industry, of physical and human capital per worker; - differing levels of total factor productivity (TFP). The study aims at explaining substantial and persistent regional differentials in labor productivity in Italy providing: 1. an assessment of the role played by the three factors above outlined in the variuos regions; 2. an empirical evaluation of the role played by some of the relevant factors suggested in the related literature (e.g., public and social capital, R&D expenditure, international openness, financial markets development, agglomeration and diversification economies, geographic factors), in explaining regional TFP differentials. The empirical analysis makes use of a particularly rich data set including annual regional accounts and capital stock data for 17 industries covering the period 1970-1994. Estimates of human capital broken down by region and industry are produced by the authors pooling information from the Labor force survey and Bank of Italy’s Survey of households income and wealth. The analysis of structural composition effects is carried out by means of the shift-share technique proposed by Esteban (2000), while a cointegrated panel model is used to estimate total factor productivity by region and sector. In an attempt to assess the relevance of spatial externalities in explaining regional TFP levels the final regression analysis makes use of spatial econometric techniques.