one agent benefits others by enriching their intellectual environment
and raising the return to their own search activities. We then formulate
the problem of a hypothetical planner who can allocate people’s time
so as to internalize this external effect. We show how the decentralized
algorithm can be adapted to compute the planning solution as well and
compare it to the decentralized solution. We then consider tax structures
that implement an optimal solution. Finally, we provide three examples
of alternative learning technologies and show that the properties
of equilibrium allocations are quite sensitive to these variations.
All of this is carried out in a starkly simple context in order to reveal
the economic forces involved and the nature of their interactions and to
build up our experience with a novel and potentially useful mathematical
structure. But we also believe that the external effects we study here
are centrally important to the understanding of economic growth and
would like to view our analysis as a step toward a realistically quantitative
picture of the dynamics of production and distribution.17