Abstract

Macroeconometric models are built on astonishingly precarious grounds and yet are used by policy makers to project precision and certainty. Econometricians use lagged dependent variables, “add factors,” and other techniques to make their models more accurate—at the expense of the integrity of the models. The reason for the unscientific nature of macroeconometric models is that, unlike the objects of controlled experimentation, real-world events are often unique and non-repeatable. Models that use repeatable events are poorly suited to accurate prediction or historical explanation.