Some Caveats for the Use of Forecating Models for Assessing Rates of Return in Workers’ Compensation

Abstract

A model, in the most general sense, is a simplified representation of reality, as it is, or as it is to come to be. The reality of interest may be a phenomenon, a system, a process, a living thing, that is, virtually anything. The simplification embodied in the representation may range from slight to extensive, from simple to highly sophisticated. All models are incorrect in some way; some models are useful; all, even good and correct models, are capable of being used in an inappropriate way. Complex economic models that have not been appropriately validated, preferably in at least two ways, are particularly dangerous.

We are indebted to the Massachusetts Rating Bureau for providing data concerning the workers’ compensation line and in particular to Dr. Richard A. Derrig for much stimulating discussion concerning the modelling process.