Implications of Uncertainty for Expected Efficiency Gains
Maximizing the efficiency gains from pollution control requires that marginal damages from
emissions (or marginal benefits from emissions reductions) equal society’s (each firm’s)
marginal costs of emissions reductions. However, a regulator seeking to maximize efficiency
gains will not have perfect information about marginal abatement costs, a reflection of the
inability of the regulator to know each firm’s current capabilities for input-substitution and
end-of-pipe treatment. There is even more uncertainty as to future abatement costs, as these
will depend on additional variables that are difficult to predict, such as fuel prices and the
extent of technological change.
In the presence of abatement cost uncertainty, the choice of instrument affects the expected
efficiency gains.11 In a static context, the relative efficiency impact of a “price” policy such as
an emissions tax compared to a “quantity” policy such as an aggregate emissions cap depends
on the relative steepness of the aggregate marginal abatement cost curve and the marginal
damage curve.12 In a limiting case, where the marginal damage curve is perfectly elastic,
expected net benefits are maximized under the emissions tax, with the tax rate set equal to the
(constant) marginal damages. In this case the tax automatically equates marginal damages
to marginal abatement costs, regardless of the actual location of the marginal abatement
cost schedule. In contrast, if an aggregate emissions cap is employed, with the cap set to
equate marginal damages with expected marginal abatement costs, abatement will be too high
ex post if marginal abatement costs turn out to be greater than expected, and too low ex post if
marginal abatement costs are lower than expected. The relative efficiency gains are reversed in
the other limiting case: when marginal damages are perfectly inelastic, expected net benefits
are maximized under the emissions cap. For intermediate cases, either the tax or the cap
could offer higher net benefits, depending on whether the marginal damage curve is flatter
or steeper than the marginal abatement cost curve (Weitzman 1974).
These results carry over to a dynamic setting, where environmental damages depend on
the accumulated stock of pollution. Some dynamic analyses (see Kolstad 1996; Pizer 2002;
Newell and Pizer 2003) suggest that in the presence of uncertainty, a carbon tax (a “price”
policy)might offer substantially higher expected efficiency gains than a cap-and-trade system
(a “quantity” policy).
Uncertainty and Policy Flexibility
The analyses just discussed do not consider differences across instruments in the speed
at which they can adjust to new information. However, an emissions allowance system
that includes provisions for the banking and borrowing of allowances might have a slight
advantage over emissions taxes in this regard. For example, suppose that, under a carbon
cap-and-trade system, new evidence emerges that global warming is occurring faster than
projected. Speculators would anticipate a tightening of the future emissions cap, which
11Policymakers are also uncertain about the marginal damage schedule. However, as discussed in Weitzman
(1974) and Stavins (1996), this does not have strong implications for instrument choice unless marginal
damages are correlated with marginal abatement costs.
12The aggregate emissions cap policy could involve either fixed quotas on individual pollution sources, or
a set of tradable emissions allowances, where the total number of allowances in circulation represents the