that the trade-off between wellbeing in the first two periods be separable from the
remainder of the stream. Regarding (i), there is a large literature, starting with
Diamond (1965), which has established a conflict between imposing equity conditions
(like equal treatment and “Hammond Equity for the Future”) on the one hand, and
remaining sensitive to the interests of every generation on the other. Asheim et
al. (2010, Section 4) present a formal analysis showing how imposing on the set of
all streams the separability property of DU mentioned under (ii) is in conflict with
equity conditions that respect the interests of future generations.
The present paper proposes SDU as an alternative criterion to DU for the evalu-
ation of climate abatement policies, and it seeks to illustrate that substituting SDU
for DU matters for empirical evaluation of such policies. We do so by considering
the DICE integrated assessment model, built by William Nordhaus, but where we
run risk analysis, including alternative specifications for the important parameters
determining the climate sensitivity and damage function. Weitzman (2009, 2010a,b)
in particular has raised doubt concerning the climate sensitivity and damage func-
tion used in the standard DICE model. Our alternative specifications lead to a
non-negligible probability that some generation is better off than its descendants, in
which case adopting SDU instead of DU matters for the evaluation, so much so that
we are able to show aggressive emissions abatement increases social welfare under
SDU, even when the utility discount rate is relatively high. By contrast, we confirm
that such abatement policies fail to increase social welfare under DU. We also show
that the optimal abatement policy is more stringent under SDU than DU.
Our analysis is ethical in nature, asking what our generation as a collective should
do to serve the interests of all generations from an impartial perspective. This is
different from taking a strategic perspective, asking what contemporary countries or
individuals should do to serve their own interests when such actions influence the
future strategic actions of other countries and individuals. Nevertheless, since SDU
bridges the gap between DU and undiscounted utilitarianism, we believe it is realistic
to suggest that its recommendations are of interest for current decision-makers.
The paper is organised as follows. In Section 2 we present formally the concept
of SDU. While Asheim and Mitra (2010) have already done so in a deterministic
setting and without explicit consideration of population growth, empirical evaluation
of climate policies does not permit either of these simplifications, so Section 2 extends
SDU to variable population and uncertainty. In Section 3 we present risk analysis
with DICE, discussing in particular our choice of climate sensitivity and damage