Interactions Between Climate Policies in the Power Sector - Green Energy and Efficiency: An Economic Perspective - page 280

5.2 Non-appropriable Technology Externalities

A second argument to justify the existence of renewable energy support mecha-
nisms would be the existence of non-appropriable technological externalities.
Positive externalities of innovation exist in several sectors and per se do not justify
the existence of speci c subsidies. Innovation externalities justify horizontal sup-

port to R&D, but not sector-targeted support. However, in the case of climate
technologies, innovation can decrease abatement costs. Thus, supporting positive
innovation externalities would help to reduce the cost of the emissions externality.
Innovation in climate technologies is a tale of two market failures: synergies
between the innovation externality and the environmental externality help to reduce
the abatement costs (See [ 17 ]).
Innovation externalities can arise from the investment in R&D or from learning-
by-doing. Learning by doing occurs when a technology becomes more ef

cient the
more it is used. Investment in R&D can reduce the cost of non-emitting technol-
ogies or can reduce the emissions of fossil-fuel technologies through better carbon
ef ciency or through carbon sequestration. The appropriate policy is different in

each case, and also the interaction between policies.

5.2.1 Learning by Doing

Learning by doing implies that the costs of producing renewables are reduced the
more renewable energy is produced. To analyse the effect of learning by doing we
need to add a second period to our model: During the rst period the monopolist

decides how much energy from renewables sources to produce. The more renew-
ables it produces during the rst period the less costly will be to produce renewables

in the second period.
This would be equivalent to adding a second stage to our model where the
renewables cost function is gðq 2 Þc 2 ðq 2 Þ 0

and q 2 and q 2 are the renewable energy production in periods 1 and 2 respectively.
That is, g () would re

where g (0) = 1, g ( q ) ≤ 1and g ′ ( q ) ≤

ect the decrease in the costs of producing renewables due to
the effect of learning by doing i.e. the more renewables are produced in period 1,
the lower the cost of producing renewable energy in period 2.
In the case of a monopolist, learning-by-doing effects would provide more
incentives to produce renewable energy during the

rst period that in the absence of
such effects (even if there is no subsidy). Since all the bene ts will be captured by

the rm in the second stage, the rm will produce more renewable energy during

the rst period than in the absence of learning by doing (or, equivalently, the

necessary subsidy to reach a speci

c target R will be lower) in order to reduce the

costs in the second period.
Therefore, learning-by-doing effects do not justify the existence of a subsidy to
renewables when rms will be able to capture the bene ts from learning by doing

effects during the second period. In such a case, rms will have incentives to