health variable to exogenous (e.g., pollution) and choice variables (averting and
mitigating behaviour). A health improvement corresponds with a fall in the number of
days with a certain degree of impairment. We consider a deterministic framework.
To keep things simple, our presentation assumes a one-period model, while
bearing in mind that GEM-E3 represents consumer behaviour by an inter-temporal
model of the household sector. The representative consumer’s utility function is a two-
level nested LES utility function as in the standard GEM-E3 model, but with one more
component linked to health.
The upper level utility function U0 is a LES function defined over excess
consumption ( C - C ), excess leisure ( l-l ) and excess health (H-H ). It is also a separable
function of the ambient concentration of the different air pollutants.
M -λ _ __ M
U0 = α0 ln (C - C) + α0 ln(l -1) + α? ln(H - H) - ∑αH,„A„ (1)
m =1
C, l and H are subsistence levels of consumption, leisure and health. αn0 (n =1,...,3)
are parameters of the LES function. αH,m0 is the marginal utility of a decrease in the
ambient concentration of pollutant m (m=1,...,M)(αH,m0 > 0). It reflects the separable
effects of air pollution. Am is the ambient concentration of air pollutant m w.r.t. the
reference equilibrium. It is assumed to be a function of the emissions of the various air
pollutants w.r.t. the reference equilibrium (EMpo with po=1,...,PO):
Am=Am(EM1,...,EMPO) ∀m (2)
The set of M air pollutants does not only contain the PO primary pollutants, but also the
secondary pollutants formed out of them in atmospheric transformation processes. The
individual considers himself to be small relative to the rest of the economy and
therefore takes Am as given.
An alternative approach considers different health states rather than a continuous health variable (see
Freeman (2003)). However, this approach is less straightforward to integrate in the GEM-E3