sectoral economic costs and benefits of ghg mitigation - IPCC

Renewable Energy
2.3. Endogenous TC and the experience effect: an analysis of past learning rates
For any estimates of future learning rates of energy conversion technologies, it is essential to
understand past learning performances in this area. This sub-section first summarises the concept
of technological learning and discusses the main assumptions behind it. Then it presents learning
rates found in the manufacturing sector and possible causal factors. We then proceed to report
learning rates observed for energy conversion technologies.
The concept of technological learning
For the purpose of the following discussion, technological learning is meant to describe
reductions of specific investment costs of a technology, assumed to accompany the increasing
use of the technology in question.
A learning curve, or experience curve, describes the specific (investment) cost as a function of
the cumulative capacity for a given technology. It reflects the fact that technologies may
experience declining costs as a result of increasing adoption into society due to the accumulation
of knowledge through, among others, processes of learning-by-doing and learning-by-using
(Dutton and Thomas, 1984; Grübler, 1998b). The cumulative capacity is used as a measure of the
knowledge accumulation occurring during the manufacturing and use of one technology
(Christiansson, 1995).
The most common concept to express technological learning is to postulate a constant relative
reduction of technology costs for each doubling of total installed capacity. Expressed in
mathematical form, specific capital costs are therefore an exponential function of the total
cumulative capacity installed:
Cost(CCap) = A* CCap -b (1)
where: Cost (.) ……. Specific capital costs
A …………... Specific capital costs at a total cumulative (initial) capacity of 1
CCap ………. Total cumulative installed capacity
-b …..……… Learning elasticity
The learning elasticity b can be used to calculate the progress ratio or vice versa. The progress
ratio (pr) expresses the rate at which the cost declines each time the cumulative production
doubles:
−b
pr = 2
E.g., a progress ratio of 0.8 means that the costs per unit of newly installed capacity decrease by
20% for each doubling of cumulative installed capacity. The parameter b thus constitutes one of
the key assumptions describing technological progress because it defines the speed of learning
for the technology. It is important to note that an alternative but equivalent parameter, the
learning rate, is often used which is defined as ‘1 - pr’. The three indicators (elasticity, progress
rate, and learning rate) are therefore equivalent in the sense that any two of the three can be
calculated from the third. In the following survey, we will mainly use learning rates. As with any
model, the learning concept as presented here is a simplification, in particular the assumption of a
constant learning rate. As will be shown below, several authors relax this assumption by
considering learning curves that are only piece-wise linear on a double-logarithmic scale.
Learning rates observed in manufacturing
The concept of technological learning was first researched at the firm level. In an overview
paper, Dutton and Thomas (1984) reported observed learning rates in 108 cases analysing
technological learning at the level of individual firms. A histogram of these rates is presented in
116