Dynamic graphics and model validation: An application to best-practice production functions (Q2711693)

The use of graphical analysis for the validation of regression models in economic and business applications is limited to the examination of a few residual plots. However, further useful information can be gained through the use of other plots within a dynamic graphic framework. NEWLINENEWLINENEWLINEThe aim of this paper is to show a way in which dynamic graphics can provide useful insights in the estimation of economic relationships, and in particular in the estimation of best-practice production functions, an empirical tool widely used in applications concerning the analysis of productivity and performance. By dynamic graphics is defined a set of procedures allowing a faster evaluation of standard graphical tools through direct manipulation of graphical elements-and their virtually instantaneous modification-on a computer screen. Dynamic graphics can detect interesting features of the data not immediately visible otherwise, and it can supplement more traditional techniques of model validation. In some instances where traditional specification tests cannot be properly applied, dynamic graphics can still be used. In order to illustrate the capabilities of these techniques, it is considered the validation of a best-practice production function estimated over a sample of 43 countries. A case study involving country data from the Penn World Table is also illustrated.