Resisting the Allure of Large Projection Models

Abstract

This article examines the value of developing large-scale projection models. Although its primary focus is on the methods that would integrate demographic and economic models to yield housing stock forecasts, the considerations can apply to large scale modeling efforts in general. A case study provides the context for practical considerations. Technical, theoretical, institutional, practical, ethical, and planning issues are considered. The institutional context within which large models are developed and used can affect the information content of forecasts. Realistic contexts can make simpler projection models backed by qualitative studies, produce better forecasts than the large projection models that absorb most of an agency's research budget. Theoretical and econometric considerations as well as track record may favor the use of simpler projection models. The commitment to modeling, however, can help staff development and reduce the role of expediency, wishful thinking, and political manipulation of forecasts. The article stresses the need for a balance between quantitative and qualitative methods.