Jamasb and Kohler 2007 Learning curves for energy tech

Jamasb, T. and J. Kohler (2007). "Learning Curves for Energy Technology: A Critical Assessment," in Delivering a Low Carbon Electricity System: Technologies, Economics and Policy, Ed. M. Grubb, T. Jamasb, and M. Pollitt. Cambridge: Cambridge University Press, Forthcoming. Available online at (accessed 30 Nov 2007).

• INTRO
• 1 Cites Schumpeter's invention-innovation-diffusion paradigm of technical change
• 1-2 Developments in macro (endogenous technical change) and micro (learning curve)
• 2 Need for innovation in environmental and energy technology revived interest in learning curve concept
• 3 Learning curve originally had labor as dependent variable and manufacturing/production as independent. Using price (DV) and capacity (IV) requires attention to determinants of innovation
• LEARNING CURVES AND TECHNICAL CHANGE
• 5 Bar chart showing estimated learning rates in electricity production technologies, from survey of studies. Varies widely. Generally higher for newer technologies.
• 6 Incorporating learning curves can change outcomes of cost of climate change models, but all depends on assumptions
• 6 Important assumption in experience curve is what "floor cost" is thought to be - so cost doesn't decline to 0 as time goes to infinity, floor cost must be specified. But how do we know??
• 7 Benefits of endogenous technical change beyond direct Pigouvian benefits of carbon abatement - spillover/technology diffusion effercts. "This will result in a positive spillover that will offset the negative spillover usually hypothesized to result from the migration of polluting industries."
• 7 Need menu of policies in addition to carbon caps
• THEORY-INFORMED MODELS OF TECHNOLOGY LEARNING
• Questions remain about causal links between experience and cost
• Authors really like two-factor learning curves (2FLCs) that incorporate R&D.
• This is what Klaassen (2005) did, but since R&D funding was only public funding, seemed like results gave too much credit to R&D.
• 10 "A simultaneous equations model with capacity and R&D as well as endogeneity of capacity on cost transforms single-factor models from partial empirical functions into learning-innovation-diffusion models that conform to basic elements and feedback of technical change process and invention-innovation-diffusion paradigm."
• 11 Problems: development of technologies unlikely to look list past progress; lack of long-term, detailed data.
• 12 Still, evidence for some degree of experience-based cost reduction overwhelming.
• LEARNING CURVES FOR LOW-CARBON ELECTRICITY SECTOR
• Can help determine whether funds for tech promotion allocated in proportion to their relative effectiveness (assumes 2FLC - help tell us relative effectiveness of R&D and deployment, i guess)
• Can also be used to estimate total required investment on R&D and capacity support for bringing tech cost down to given level
• CONCLUSIONS
• 15 Incorporation of learning curves can change estimates of costs of stabilization and policy conclusions (e.g., used in Stern review)
• 16 Recommends models that include R&D, but acknowledges lack of suitable data
• 17 Learning models can be used to analyze effect of international policy coordination and pooling R&D resources or deployment initiatives in order to accelerate technical progress. (How?)
• 17 Extension: non-electricity energy sources, other environmental technologies, energy storage technologies. Also, use to answer shorter-term questions (not necessarily doubling of capacity).