Environmental Engineering Reference
In-Depth Information
leaves out many important factors such as political, technological, demo-
graphic and social factors. He has similar concerns about the collective
optimum, pointing out that it neglects important side-ef ects of regimes
when accounting for regime consequences. Empirically, Young suggests
that the use of the counterfactual poses the same methodological prob-
lems discussed before since the use of expert judgements to estimate it are
insui cient especially when they do not account for social or technological
factors (Young, 2001, pp. 110-14). This critique led to a fruitful debate on
the issue and on potential ways to improve these approaches (Hovi et al.,
2003a, 2003b; Young, 2003).
Another approach to measuring ef ectiveness is given by Mitchell
(2004) who, in order to evaluate international environmental regimes, uses
regression analysis on panel data. He proposes a quantitative approach
by developing a model for a single regime's ef ects. In this model he uses
time-series data for one country at a time for the 1985 Sulphur Protocol
of the European Convention on Long-range Transboundary Air Pollution
(LRTAP). He specii es the following model to estimate national sulphur
emissions for the LRTAP case (Mitchell, 2004, p. 127):
EMISS
5 a 1 b
1
*
MEMBER
1 b
2
*
INCOME
1 b
3
*
POP
1 b
4
*
COAL
1 b
5
*
EFFIC
1 . . . 1 b
N
*
OTHER
1 e
where
EMISS
is annual emissions of sulphur dioxide and
MEMBER
is
coded as 0 in years of non-membership of the country to the regime and as
1 in years of membership. Generic drivers of emissions of most pollutants
are also included such as per capita income (
INCOME
) and population
(
POP
). Emission-specii c drivers are included, such as the country's coal
power plants (
COAL
) and their average ei ciency (
EFFIC
). The model
estimates dif erence in sulphur emissions and how these are explained
by the dif erent variables. For instance b
1
represents the expected dif er-
ence in emissions that would arise from a country becoming a regime
member, holding all other variables constant. The coei cients of the other
independent variables b
2
through b
N
correspond to the estimated increase
in emissions that would arise from a one-unit increase in that variable.
The
t
-statistic on the coei cients shows the statistical signii cance of the
independent variables, whereas the goodness-of-i t (
R
2
) of the model equa-
tion as a whole provides an estimate of how completely the analyst has
modelled the dependent variable.
Mitchell (2004, p. 129) advances his method by developing another
model that allows comparison by combining data from dif erent regimes.
He uses time-series data and data across regimes. As an example he devel-
ops a model to assess the simple claim that sanctions are necessary for a
