Endogenous growth theory twenty years on: a critical assessment

Endogenous growth literature emerged from dissatisfaction with one result of the neoclassical growth model: the independence of the growth rate from the saving ratio, which is seen as a variable subject to policy influence. There are at least three generations of EGT models: the old one of the sixties; the new one of the late eighties; and the most recent one, from the second half of the nineties. EGT models of any vintage fall into one of two fields: neo-Solowian (or semi-endogenous models) or fully endogenous models. Models from the sixties would generally fall into the first class and for good reasons. Indeed, most of the early generation of fully endogenous models from the late eighties fell under the ‘Jones critique’ (Jones 1995b), which pointed out some of the difficulties of these models. The most recent models have found various ways to avoid those problems. It is shown that these stratagems were anticipated by Marvin Frankel in the sixties and by Lucas in the eighties. One suspects that these devices arose in order to fix the theory rather than from, say, some ex-ante empirical observation (which is often provided ex post). More importantly, this paper indicates some problems common to all vintages of EGT models, beginning with the Cambridge capital theory critique, and suggests some alternative routes for growth analysis outside neoclassical theory.