|
Article Excerpt
Much policy attention has been given to promote fledgling energy technologies that promise to reduce our reliance on fossil fuels. These policies often aim to correct market failures, such as environmental externalities and learning-by-doing (LBD). We examine the implications of the assumption that LBD exists, quantifying the market failure due to LBD. We develop a model of technological advancement based on LBD and environmental market failures to examine the economically efficient level of subsidies in California's solar photovoltaic market. Under central-case parameter estimates, including nonappropriable LBD, we find that maximizing net social benefits implies a solar subsidy schedule similar in magnitude to the recently implemented California Solar Initiative. This result holds for a wide range of LBD parameters. However, with no LBD, the subsidies cannot be justified by the environmental externality alone.
1. INTRODUCTION
In light of the increasing scientific consensus on global climate change and the desire for greater energy security, many governments have recently set ambitious targets to increase the share of renewable energy in the total energy mix. To meet these targets, policymakers are deploying a variety of policy instruments, including technology subsidies. Along with wind, solar energy has been one of the largest beneficiaries of these policies, particularly in Germany, Japan and California. Appropriately assessing the economic efficiency of such policies is important as many other governments are planning on following suit.
This paper models the optimal photovoltaic (PV) solar subsidy policy in California, and compares this policy to one of the largest PV energy incentive programs in the world, the recently implemented California Solar Initiative (CSI).
Economic arguments for policies to promote renewable energy often include an assertion that the renewable energy technology will substitute for fossil fuel technologies that have important environmental externalities, in particular externalities associated with the atmospheric release of carbon dioxide. Although this argument is qualitatively correct, as will be discussed at a later point, it is not quantitatively correct for PVs: the externality appears to be far smaller than the proposed subsidy. Thus the current externality alone cannot justify large subsidies for renewable energy, including for PVs.
A second argument is based on an appropriability market failure if the production of the new technology may have spillover benefits from learning by doing (LBD). Learning by doing characterizes technical progress for a technology as related to the cumulative experience with the technology: costs may decrease as cumulative experience increases. With LBD, a positive externality occurs because increased output (e.g., of solar panels) by one firm today contributes to a lower production cost in the future, benefiting that firm as well as other firms or consumers in the market. As firms cannot appropriate this entire spillover effect, the private market under-provides the product of interest.
Such a LBD externality could provide an economic justification for a subsidy. The individually optimal production level of a firm would take into account the impact on future cost reductions for that firm alone. The socially optimal production level of that firm would take into account the impact on future cost reductions for all firms together, an amount that could be expected to be many times higher in a competitive industry characterized by substantial LBD. The question, however, remains whether such an externality is quantitatively consistent with the proposed subsidies.
Weighing the benefits of fostering a fledgling technology against the costs of the policy is complicated by difficulties of modeling the technology policy in inducing technological change. There is an extensive literature on the modeling of induced technological change in energy technologies, which tends to fall into the following camps: direct-price induced technological change, research and development (R&D)-based technological change, and learning-by-doing. Modeling technological change as directly price-induced assumes that price increases of an input, such as energy, induce technological change that economizes use of that input. Modeling technological change as R&D-based assumes a specified relationship between R&D investment and improved technology (Clark and Weyant, 2002; Edmonds, et al., 2000; Loschel, 2002).
The literature on technological change in solar energy primary focuses on learning-by-doing, with numerous studies empirically estimating the learning rate (LR), or the percent decrease in costs with a doubling of cumulative experience, where experience is often modeled simply as the capacity installed. Williams and Terizen (1993) estimate that solar photovoltaic (PV) module (i.e., solar panel) prices on the global market followed a learning rate of 18% between 1976 and 1992. Watanabe (1999) finds a 20% learning rate in installation costs in the Japanese market between 1981 and 1995. IEA (2000) and van der Zwaan and Rabl (2004) update the global learning rate with more recent data and both find a learning rate of around 20%.
McDonald and Schrattenholzer (2001) bring together estimates in the literature of learning rates in a wide range of energy technologies, including solar. Not surprisingly, they find that solar technology has had a relatively high rate of learning, especially when compared to mature fossil energy technologies. This result corresponds with Jamasb (2007), who finds that mature technologies such as coal-fired electricity generation and large hydropower have had much lower learning rates than "evolving" technologies such as nuclear power, waste to electricity, and wind power. Solar photovoltaics in California are arguably also an "evolving" technology.
Of course, such characterizations of LBD summarize all of factors associated with cost reductions into one simple functional relationship between the capacity installed and unit cost. This simple characterization leads to a common criticism: the lack of a "natural law" forcing such a relationship or theory explaining it (Junginger, et al., 2005). The intuition for learning described in the seminal paper on LBD by Arrow (1962) is that knowledge can be gained by hands-on experience with a problem and that learning occurs through action. But while the functional relationship between experience and costs may be an empirical observation, attributing all of the cost reductions to learning neglects any other sources of cost reduction (Clark and Weyant, 2002).
This criticism can also be expected to apply to the global solar PV module market. Nemet (2006) examines learning in the global PV market and finds that learning only weakly explains cost reductions in the most important factors in the cost of solar PV modules. Papineau (2004) finds that the effect of cumulative experience on total solar PV cost reductions is highly significant, but becomes insignificant when a time trend is added. However, the effect of R&D on total solar PV cost reductions is less significant than the effect of experience.
Duke, Williams, and Payne (2005) and Duke (2002) suggest a feature of the solar PV market that may provide an...
|
