Externalities and growth models.

Summary This work is a critical analysis of the way externalities are introduced in some recent growth models. First, it systematically examines the concrete content and the specific allocative problems associated with informational externalities. In particular, we show that their diffusion is often mediated by market or public intermediaries, one of whose functions is to reduce the cost of appropriation for economic agents, that there are very specific and generally neglected processes of internalization, but that at the same time, some are not by nature internalizable because of their unpredictability. We then return in detail to three forms of introducing externalities into growth models: the hypothesis of an intergenerational transmission of human capital in education models, the attachment of the externality to physical capital in a Romer-type model [1986], and the attachment to human capital in a Lucas-type model [1988]. We show that it is difficult to find a conception of human capital that allows for its intergenerational transmission despite parental influence. We also refute the main arguments put forward to justify the idea of an externality on physical capital, in particular that of a Schmookler-like causality [1966] or that of the exploitation of intersectoral links. Finally, after having established that, with the exception of family action, there were no significant "educational" externalities for growth, we conclude that, even with the most favourable interpretation, the external advantage described by Lucas is not attached as simply to the average level of human capital as his model would suggest.