Growth and degrowth in a Learning by Doing model[Croissance et décroissance dans un modèle Learning by Doing]

This article examines the likelihood of a peak in global GDP occurring in the coming decades, using a "learning-by-doing" growth model that distinguishes between two factors of production: physical capital generated by economic activity and natural capital, which encompasses all natural resources. This model is based on the concepts of ecological footprint and biocapacity and assumes productivity with an upper bound. Three main types of GDP trajectories are possible: (I) monotonic growth; (II) transient growth up to a GDP peak, followed by monotonic decline; (III) transient growth, followed by a transient decline to a minimum, followed by a rebound in economic activity. The probabilities of occurrence for each of these trajectory types show that type II is by far the most frequent. If biocapacity is assumed to be constant, the cumulative probability distribution of GDP peak dates indicates that (i) the probability of a reversal in global growth before 2050 is around 15\%, (ii) there is more than a 50\% chance that the peak will occur within 40 years, and (iii) the probability of a global GDP peak during the 21st century is greater than 95\%. The vast majority of simulations are characterized by rates of decline (calculated relative to the peak) ranging from 0 to 25\%. Even if biocapacity grows linearly (in which case infinite growth is possible), the results remain similar. The assumption of a finite biophysical world is therefore not necessary for a peak in GDP to be highly probable before 2100.