I expound a model of a one-good economy where long-run growth and output fluctuations are endogenous consequences of the decisions taken by entrepreneurs on the allocation of their resources between production and innovation in a Markovian sequence of one-period games. I show, first, that the log of output follows a process with random walk characteristics; second, that, contrarywise to the mechanism at work in Real Business Cycle models, the variability in the rate of productivity growth does not imply, by itself, the occurrence of any recession; third, that a recession is the consequence not of a Kydland-Prescott (1982) negative shock on technology, but of the reallocation of factors in the face of an 'increased opportunity' ex ante which the entrepreneurs fail to exploit ex post; fourth, and finally, that, although any generation could make itself unilaterally better off by reducing its level of research, the onegood economy is intertemporally efficient in that such a move would imply a reduction in the level of welfare of future generations.