The Contribution of the Capital to Productivity: A Panel Data Analysis

Abstract

Much of the evidence on the impact of capital deepening on productivity in Turkish manufacturing industry comes from studies using growth accounting method which weighs growth in inputs, including capital, by their share in the value of output. The empirical analysis in this study extends the existing evidence on Turkish manufacturing sector productivity using econometric model that account for cross-section dependence and heterogeneity of production technology in a panel setting. The common correlated effects (CCE) type estimator is applied, which makes use of cross-sectional averages of the dependent and explanatory variables of the regression equation to remedy the cross-section dependence problem arising from unobserved common effects and/or error spill-over effects due to spatial or other forms of local dependencies. Overall, the following main conclusions are derived. 1. Individual industry regression results convey apparent technology heterogeneity across the industries. However, some of the individual industry estimates seem implausible whilst most of them are imprecise. 2. Imposing slope homogeneity restriction in the pooled models lends a lot of precision to the capital productivity estimate. However, when tested, the industries are not poolable. But, interestingly, the mean-group and pooled estimates of technology coefficients are close. 3.The technology estimates are sensitive to the presence of observed and unobserved common factors, justifying the use of CCE estimators. We identify four observed factors that render the coefficient estimates sensitive which, as a result, are included in the regressions. 4. Compared to pre-2002 period, labor productivity is widely dispersed across the industries in post-2002 period. By the end of 2011, in some industries labor productivity is well below 1980 levels.