Application of Kernel Estimation in Analysis of Labour Productivity of the Largest Polish Firms in 2004-2008

The parametric methods of statistical and econometric analysis are not always useful in examination of labour productivity of economic entities. In previous works the Authors found that the labour productivity is characterized by the lack of stable regularities in the range of structure and interdependency. In that case it is possible to apply non-parametric methods. In the paper the Authors tried to model the distributions of the labour productivity in time by means of kernel estimation using classical approaches (Epanechnikov, Rosenblatt) and the new proposition called kernel B. It seems that proposed approach is a useful merger of the statistical modeling theory and economic practice which allows to analyze the changes in the labour productivity - the essential factor for long-term economic growth and the welfare of society. The empirical results show that the labour productivity in the largest Polish companies had increased in 2004-2008 but the growths had not the same dynamics in different economic sectors.