The artifice of an infinitely-lived representative agent is commonly invoke
d to balance the present costs and future benefits of climate stabilization
policies. Since actual economies are populated by overlapping generations
of finite-lived persons, this approach begs important questions of welfare
aggregation. This paper compares the results of representative agent and ov
erlapping generations models that are numerically calibrated based on stand
ard assumptions regarding climate-economy interactions. Under two social ch
oice rules - Pareto efficiency and classical utilitarianism - the models ge
nerate closely similar simulation results. In the absence of policies to re
distribute income between present and future generations, efficient rates o
f carbon dioxide emissions abatement rise from 15 to 20% between the years
2000 and 2105. Under classical utilitarianism, in contrast, optimal control
rates rise from 48 to 79% this same period.