220.127.116.11 Learning by Doing (LBD)
LBD as a source of technical change was first emphasized by Arrow (1962). Nakicenovic
(1996) discussed the importance of LBD in energy technology, and Messner (1995)
endogenizes the learning process in energy models. LBD is a happy consequence
of those investments in which learning is a result of cumulative experience
with new technologies. LBD typically refers to reductions in production cost,
in which learning takes place on the shop floor through day-to-day operations,
not in the R&D laboratory. The LBD component of change is significant too.
Kline and Rosenberg (1986) discuss industry studies that indicate that LBD-type
improvements to processes in some cases contribute more to technological progress
than the initial process development itself.
LBD models use the installed capacity or cumulative use as an indicator of
accumulating knowledge in each sector. The abatement costs are represented by
the specific investment costs in US$/kWh. The models are global and therefore
the diffusion process is not represented. The optimization problems are non-convex,
which raises a difficult computational problem to find an optimum. However,
pioneering work at the International Institute for Applied Systems Analysis
(IIASA) on the MESSAGE model and additional developments based on models like
MARKAL and ERIS; (MATSSON), Kypreos and Barreto (1999), Seebregts et al.
(1999a), (SKFB), Tseng et al. (1999), and Kypreos et al. (2000)
demonstrate progress in this direction. They show that several technologies
are likely to play a prominent role in reducing the cost of abatement, if ITC
is indeed taken into account when computing the equilibrium. A problem with
modelling endogenous technological change is that the traditional baseline scenario
versus optimal policy run argumentation is not feasible. This follows directly
from the path dependence. The most important results are: greater consistency
of model results with the observed developments of technological change;
- new technologies first appear in niche markets with rising market shares;
- the time of breakthrough of new technologies can be influenced by policy
measures (taxes and R&D) if they are strong enough;
- identification of key technologies, like photovoltaic modules or fuel cells,
for public R&D investments is difficult; and
- technological lock-in effects depend on costs.
The most important conclusion for the timing of a mitigation policy is that
early emissions-reduction measures are preferable when LBD is considered. This
is confirmed unambiguously by a macroeconomic modelling study (van der Zwaan
et al., 1999/2000) which finds also lower levels of carbon taxes than
those usually advocated.
These findings must be tempered by the fact that the models are not only highly
non-linear systems, and therefore potentially sensitive to input assumptions,
but also the quantitative values employed by modellers are typically drawn from
successful historical examples. Furthermore, the empirical foundations of LBD
are drawn from observations of the relationship between cumulative deployment
and/or investment in new technology and cost. This relationship is equally consistent
with the hypothesis that a third factor reduced costs, in turn leading to increases
in demand. The authors restrict their findings to more qualitative assertions,
because of the limitations of current models (Messner, 1997; Grübler and
Messner, 1998; Barreto and Kypreos, 1999; Seebregts et al., 1999a, 1999b).
The research so far has been limited to energy system models and ignored other
forms of endogenous, complex changes that are important for emissions, like
changes in lifestyles and social institutions.
18.104.22.168 The Distinction Between Action and Abatement
The key message from this discussion about technical change is that a clear
distinction has to be made between the timing of action and the timing of abatement.
As a result of inertia in technological innovation, short-term action is required
to abate more in the future, but a given amount of abatement at a given point
in time is not a good measure of the effort. The necessity of this distinction
is reinforced by the consideration of inertia in capital stocks. Mitigation
costs are influenced by assumptions about the lifespan of existing plants and
equipment (e.g., power plants, housing, and transport). Energy-related capital
stock is typically long lived and premature retirement is apt to be costly.
For example, an effort to change the transportation infrastructure will not
reduce carbon emissions significantly for two decades or more. Hence, a drastic
departure from the current trend is impossible without high social costs and
a delay of action in this sector will require higher abatement costs in the
more flexible sectors to meet a given target. Lecocq et al. (1999) found
that these costs would be increased by 18% in 2020 for a 550ppmv target and
by 150% for a 450ppmv target.
This irreversibility built into technological change is far more critical when
the uncertainty about the ultimate target is considered. In this case indeed,
many of the parameters that legitimize the postponing of abatement play in the
opposite direction. If indeed the concentration constraints turn out to be lower
than anticipated, there may be a need for abrupt reduction in emissions and
premature retirement of equipment. In other words, even if the permanent costs
of an option (in case of perfect expectation) are lower than those of an alternative
option, it may be the case that its transition costs are higher because of inertia.
For example, two ideal transportation systems can be envisaged, one relying
on gasoline, the other on electric cars and railways, both with comparable costs
in a stabilized situation; however, a brutal transition from the first system
to the second may be economically disruptive and politically unsustainable.
These issues are examined in more depth in Chapter 10
because the selection of the ultimate target depends upon the decision-making
framework and upon the nature of the damage functions. But, it matters here
to insist on the fact that the more inertia is built into the technical system,
and the less processes of learning by doing and induced technical change have
operated, the more costly corrections of trajectories in hedging strategies
will be, for example, moving from a 550ppmv concentration goal to 450ppmv (Ha-Duong
et al., 1997; see also Grubb et al., 1995; Grubb, 1997). This
possibility of switching from one objective to another is supported by current
material regarding climate damages, in particular (Tol, 1996) if the rate of
change is considered in the analysis and the delay between symptoms and the
response by society (see Chapter 10).