Energy technology learning through deployment in competitive markets.

INTRODUCTION

Market deployment can make currently high-cost, low-carbon energy technologies cost efficient. Recent high-level policy documents embrace the insights from experience and learning curves into the crucial role of this mechanism (IEA/OECD, 2006, 2008: Stern, 2006). The curves show that cumulative market experience, e.g., in the form of cumulative sales, provides a steady, progressive decrease in cost and increase in technical performance for a technology. This technology learning has major implications for policy design: e.g., market support lot low-carbon technologies.

Since 2000, academic journals have published an increasing amount of papers using experience and learning curves to analyze development of low-carbon energy technologies and technology-led strategies to manage the risk of global warming. A major energy journal has dedicated a special issue to endogenous technological change and the economics of atmospheric stabilization (Grubb et al., 2006; Edenhofer et al., 2006). In 1999, a group of scientists and analysts from academia, industry, and government agencies participating in a workshop arranged by the International Energy Agency (lEA) observed that experience and learning curves "are underexploited for public policy analysis." They recommended that the curves "are used to analyse the cost and benefits of programmes to promote environment friendly technologies" and "explicitly considered in exploring scenarios to reduce C[O.sub.2] emissions and calculating the cost of reaching emissions targets" (IEA/OECD, 2000, Appendix B). It would appear that the call from the IEA Workshop is being answered.

However, even if the policy reports acknowledge the learning effect, their recommendations do not embrace experience and learning curves as operational policy tools. In their report to the 2006 G8 meeting of Head of States, the International Energy Agency finds:

The Stern report observes:

The comments point to the fact that the curves appear to express purely empirical relations between cost, price, or technical performance and cumulative production or use. Theoretical grounding is needed to identify causes, legitimize extrapolations, and explain learning rates.

Several mechanisms have been proposed to explain technology learning and the observed relationships (Abell and Hammond, 1979; Arthur, 1988; Argote and Epple, 1990; Adler and Clark, 1991; Nemet, 2006). Investigations have been made on firm and industry levels. The "bottom-up" efforts to unfold experience and learning curves provide insights into specific features, events, and processes, or FEPs, (1) driving technology learning at different periods and on different levels, but they fail to reconstruct the shape of the curves or explain the observed learning rates.

For energy technologies, the role of R&D processes is of special interest. In his analysis of solar cell technology, Nemet (2006) concludes that "experience" is insufficient as the explaining variable and points to R&D and knowledge spillovers as other important explaining variables. Jamasb and Kohler (2008) call for a specific focus on the influence of R&D on observed technology improvements. Kouvaritakis et al. (2000) propose a two-factor learning curve with cumulative R&D efforts and cumulative production as explaining variables. More sophisticated variants of this approach include capacity diffusion into the model (Soderholm and Klaasen, 2007; Jamasb, 2007).

The two-factor learning curve has considerable attraction for a policy analyst. A well-defined curve would make it possible to find an efficient balance between spending on R&D and on deployment programs. However, the data problems are large and, considerably more importantly, the approach can be seriously questioned both from the point of view of organisational theory and scientific methodology. Ockham's razor is a good guiding principle for scientific research and it stipulates that "what can be done with fewer assumptions is done in vain with more." There is overwhelming empirical support for the simple relation between cost, price, or technical performance and the cumulative production or use expressed by the single-factor experience and learning curves. R&D processes certainly influence cost and technical performance, but this influence seems too tightly bound to cumulative output from the learning system to be projected out and analyzed by adding an independent variable to the learning curve. Following organizational theory, there is also a need to distinguish between the different identities and functions of government-funded public R&D and private industry R&D. The first R&D process uses taxpayers' money to support effective action within society, the second process uses private business resources to support effective market action of a firm or group of firms. The effects of the two different R&D processes on cost and technical performance of a specific technology may be quite different.

The single-factor experience curve was firmly established within the management sciences through the work of the Boston Consulting Group (BCG) during the 1960s (Boston Consulting Group, 1968; Abell and Hammond, 1979). They found that the experience curve should include "all of the cost elements which may have a trade-off against each other" (BCG, 1968, p. 12). The cost elements include private industry R&D investments in the learning system. For the photovoltaic (PV) industry, Watanabe et al. (2000) find that increasing markets stimulated solar cell producers in Japan to invest in private R&D, establishing a tight relation between the output from a technology learning system and industry R&D efforts. Adler and Clark (1991) point to the importance of second- and third-order learning.

Analysis shows that the classic, single-factor experience and learning curves do capture the effects of both private industry R&D ("learning by searching") and increasing skills in management, production, and use ("learning by doing"). Other approaches than regression analysis are needed to separate the effects of the two types of learning on cost and technical performance.

This article looks at the single-factor experience and learning curves through the prism of a theoretical approach based on cybernetics (Wene, 2007, 2008). The goal of the approach is to explain observed learning rates and provide a framework to analyze the balance between government R&D and deployment programs.

The theoretical approach is presented in the following section, followed by a comparison of the theoretical predictions for learning rates with measured distributions. The next section discusses the implications of the theoretical approach for the balance between government R&D and deployment programs, followed by a section that illustrates how market deployment of a radical innovation made available through government R&D can transform the market for photovoltaics.

A CYBERNETIC MODEL FOR EXPERIENCE AND LEARNING CURVES

Figure 1 shows an example of what theory has to explain. The figure presents the showpiece experience curve for energy technology, namely photovoltaic modules. Since 1976, prices have been reduced from over US$(2001) 60/Wp to around US$(2001) 3/Wp today. The straight line is the experience curve fitted to the time series. The price scale is logarithmic, so the experience curve can be written as

Price(t) = [C.sub.0] x X[(t).sup.-E] (1)

Price at time t is equal to a constant, Co, times the cumulative sales X(t) at time t raised to the power of -E. E is a constant and will be referred to as the experience or learning parameter. The value of this constant is to be explained by the theory.

The literature uses not E but learning rate, LR, or progress ratio, PR, to characterize the steepness of the curve. The learning rate is the relative reduction in price for each doubling of cumulative sales. The relation between E, LR, and PR is given by

[FIGURE 1 OMITTED]

PR = 1 - LR = [2.sup.-E] (2)

The learning rate for PV modules is constant at 20% over three decades and almost four orders of magnitude in cumulative global sales. The stability of learning is quite impressive, considering the great swings in market growth caused by instability in government deployment programmes. The example in Figure 1 raises two issues, which are important for the theoretical analysis of experience and learning curves, namely how to measure the effect of technology learning and how to set the boundaries of the learning system.

The learning effect in Figure 1 is measured by price, which is set by the actors in the market. Competitive markets are necessary to foster learning, but the observed learning is the result of internal operations within the learning system, which in Figure 1 is the PV module production system. Technology learning should be measured by cost rather than price. The theory will use cost as the variable to be explained. However, cost data are usually very difficult to obtain and the experience and learning curve literature usually measures the learning effect by price series. It is therefore crucial to clarify the relationship between cost and price. The analysis in BCG (1968, pp. 19-22: see also IEA/OECD, 2000, pp. 35-40) shows that the ratio between price and cost remains constant in equilibrium markets; i.e., performance measured by price and cost have the same learning rates in this case. However, new products or rapidly changing growth rates may induce price-cost cycles, which show up as systematic deviations from the experience curve measured by price.

[FIGURE 2 OMITTED]

The analysis in Figure 1 assumes that the learning system for PV modules is global. But not all technologies have global system boundaries. Schaeffer et al. (2004) conclude that the learning system for balance of system (BOS) for the complete photovoltaic installation is national or regional with considerable spillover between countries and regions. Early analysis of wind turbines was made on a national basis (Neij, 1999; Durstewitz and Hoppe-Kilpper, 1999). But more recent analysis of arrays of wind power plants in wind farms suggests global boundaries (Junginger et al., 2005). A theory for technology learning system should give system boundaries.

Print Get the Mag Weekly Updates