Dynamic graphics and model validation: an application to best-practice production functions

Abstract

The aim of this work is to show how dynamic graphics can supplement, and in some instances even surrogate, techniques of model validation traditionally adopted in economic and business applications. A case study considers a production function estimated over an unbalanced panel of technically efficient countries, where standard validation tools cannot be applied. Yet, the dynamic use of the summary plot and the added variable plot allows even in this case to assess the role of potentially influential observations. Through these procedures, a simple intervention model is implemented, obtaining coefficient values more in line with a priori expectations. Copyright © 1999 John Wiley & Sons, Ltd.