The interaction between theory and observation in economics

Abstract

In economics both theories and observations tend to be very heterogeneous. This paper discusses the methodological difficulties that are involved in relating theories and observations and the characteristics of each that influence the way they interact. We examine the factors that influence the impact of observation on theory in terms of the precision of measurement of the observations; the correspondence of the measurements to the theoretical concepts; the applicable domain of the theory, i.e. the extent to which it has implications for observable features of the economy; and the importance of the results to decision makers. I n this interaction between theory and observations, statistical models play a central role and we discuss how this role evolved in the context of four statistical models: single equation and multivariate regressions, ARIMAs and VARs. The paper concludes with a brief discussion of an approach to modelling theories which do not impose standard restrictions on these statistical models.