Beyond The Learning Curve: Factors Influencing Cost .

3y ago
331.01 KB
15 Pages
Last View : 21d ago
Last Download : 6m ago
Upload by : Brady Himes

ARTICLE IN PRESSEnergy Policy 34 (2006) 3218– the learning curve: factors influencing cost reductionsin photovoltaicsGregory F. Nemet Energy and Resources Group, University of California, 310 Barrows Hall 3050, Berkeley, CA 94720-3050, USAAvailable online 1 August 2005AbstractThe extent and timing of cost-reducing improvements in low-carbon energy systems are important sources of uncertainty infuture levels of greenhouse-gas emissions. Models that assess the costs of climate change mitigation policy, and energy policy ingeneral, rely heavily on learning curves to include technology dynamics. Historically, no energy technology has changed moredramatically than photovoltaics (PV), the cost of which has declined by a factor of nearly 100 since the 1950s. Which changeswere most important in accounting for the cost reductions that have occurred over the past three decades? Are these resultsconsistent with the notion that learning from experience drove technical change? In this paper, empirical data are assembledto populate a simple model identifying the most important factors affecting the cost of PV. The results indicate that learningfrom experience, the theoretical mechanism used to explain learning curves, only weakly explains change in the most important factors—plant size, module efficiency, and the cost of silicon. Ways in which the consideration of a broader set of influences,such as technical barriers, industry structure, and characteristics of demand, might be used to inform energy technology policyare discussed.r 2005 Elsevier Ltd. All rights reserved.Keywords: Photovoltaics; Learning curves; Experience curves1. IntroductionThe cost of photovoltaics (PV) has declined by afactor of nearly 100 since the 1950s, more than any otherenergy technology in that period (Wolf, 1974; McDonald and Schrattenholzer, 2001; Maycock, 2002).Markets for PV are expanding rapidly, recently growingat over 40% per year (Maycock, 2005). Future scenariosthat include stabilization of greenhouse-gas (GHG)concentrations assume widespread diffusion of PV. Ina review of 34 emissions scenarios, Nakicenovic andRiahi (2002) found a median of 22 terawatts (TW) of PVdeployed in 2100 for those scenarios that include GHGstabilization. At present however, PV remains a nicheelectricity source and in the overwhelming majority ofsituations does not compete economically with conven Tel.: 1 415 218 1728; fax: 1 510 642 1085.E-mail address: - see front matter r 2005 Elsevier Ltd. All rights reserved.doi:10.1016/j.enpol.2005.06.020tional sources, such as coal and gas, or even with otherrenewable sources, such as wind and biomass. Theextent to which the technology improves over the nextfew decades will determine whether PV reaches terawattscale and makes a meaningful contribution to reducingGHG emissions or remains limited to niche applications.The learning curve is an important tool for modelingtechnical change and informing policy decisions relatedto energy technology. For example, it provides a methodfor evaluating the cost effectiveness of public policies tosupport new technologies (Duke and Kammen, 1999)and for weighing public technology investment againstenvironmental damage costs (van der Zwaan and Rabl,2004). Energy supply models now also use learningcurves to endogenate improvements in technology. Priorto the 1990s, technological change was typically included either as an exogenous increase in energy conversion efficiency or ignored (Azar and Dowlatabadi,

ARTICLE IN PRESSG.F. Nemet / Energy Policy 34 (2006) 3218–32321999). Studies in the 1990s began to use the learningcurve to treat technology dynamically (Williams andTarzian, 1993; Grübler et al., 1999) and since then it hasbecome a powerful and widely used model for projectingtechnological change. Recent work however has cautioned that uncertainties in key parameters may besignificant (Wene, 2000), making application of thelearning curve to evaluate public policies inappropriatein some cases (Neij et al., 2003). This paper examinessome of these concerns. After a review of the advantagesand limitations of the learning curve model, theapplicability of learning curves to PV is then assessedby constructing a bottom-up cost model and comparingits results to the assumptions behind the learning curve.1.1. The learning curve modelCharacterizations of technological change have identified patterns in the ways that technologies areinvented, improve, and diffuse into society (Schumpeter,1947). Studies have described the complex nature of theinnovation process in which uncertainty is inherent(Freeman, 1994), knowledge flows across sectors areimportant (Mowery and Rosenberg, 1998), and lags canbe long (Rosenberg, 1994). Perhaps because of characteristics such as these, theoretical work on innovationprovides only a limited set of methods with which topredict changes in technology. The learning curve modeloffers an exception.The learning curve originates from observations thatworkers in manufacturing plants become more efficientas they produce more units (Wright, 1936; Alchian,1963; Rapping, 1965). Drawing on the concept oflearning in psychological theory, Arrow (1962) formalized a model explaining technical change as a functionof learning derived from the accumulation of experiences in production. In its original conception, thelearning curve referred to the changes in the productivity of labor which were enabled by the experience ofcumulative production within a manufacturing plant. Ithas since been refined, for example, Bahk and Gort(1993) make the distinction between ‘‘labor learning’’,‘‘capital learning’’, and ‘‘organizational learning’’.Others developed the experience curve to provide amore general formulation of the concept, including notjust labor but all manufacturing costs (Conley, 1970)and aggregating entire industries rather than singleplants (Dutton and Thomas, 1984). Though different inscope, each of these concepts is based on Arrow’sexplanation that ‘‘learning-by-doing’’ provides opportunities for cost reductions and quality improvements.As a result, these concepts are often, and perhapsmisleadingly, grouped under the general category oflearning curves. An important implication of theexperience curve is that increasing accumulated experience in the early stages of a technology is a dominant3219strategy both for maximizing the profitability of firmsand the societal benefits of technology-related publicpolicy (BCG, 1972).The learning curve model operationalizes the explanatory variable experience using a cumulative measureof production or use. Change in cost typically provides ameasure of learning and technological improvement,and represents the dependent variable.1 Learning curvestudies have experimented with a variety of functionalforms to describe the relationship between cumulativecapacity and cost (Yelle, 1979). The log-linear functionis most common perhaps for its simplicity and generallyhigh goodness-of-fit to observed data. The centralparameter in the learning curve model is the exponentdefining the slope of a power function, which appears asa linear function when plotted on a log–log scale. Thisparameter is known as the learning coefficient (b) andcan be used to calculate the progress ratio (PR) andlearning ratio (LR) as shown below where C is unit costand q represents cumulative output: bqCt ¼ C0 t,(1)q0PR ¼ 2 b ,(2)LR ¼ ð1 PRÞ.(3)Several studies have criticized the learning curvemodel, especially in its more general form as theexperience curve. Dutton and Thomas (1984) surveyed108 learning curve studies and showed a wide variationin learning rates leading them to question the explanatory power of experience. Argote and Epple (1990)explored this variation further and proposed fouralternative hypotheses for the observed technical improvements: economies of scale, knowledge spillovers,and two opposing factors, organizational forgetting andemployee turnover. Despite such critiques, the application of the learning curve model has persisted withoutmajor modifications as a basis for predicting technicalchange, informing public policy, and guiding firmstrategy. Below, the advantages and limitations of usingthe more general version of the learning curve, theexperience curve, for such applications are outlined.The experience curve provides an appealing model forseveral reasons. First, availability of the two empiricaltime series required to build an experience curve—costand production data—facilitates testing of the model.As a result, a rather large body of empirical studies hasemerged to support the model. Compare the simplicityof obtaining cost and production data with the difficultyof quantifying related concepts such as knowledge flows1Cost is often normalized by an indicator of performance, e.g. /W.Alternative performance measures are also sometimes used such asaccident and defect rates.

ARTICLE IN PRESS3220G.F. Nemet / Energy Policy 34 (2006) 3218–3232Fig. 1. Experience curves for PV modules and sensitivity of learning rate to underlying data. Data: Maycock (2002) and Strategies-Unlimited (2003).and inventive output. Still, data quality and uncertaintyare infrequently explicitly assessed and as shown belowcan have a large impact on results. Second, earlierstudies of the origin of technical improvements, such asin the aircraft industry (Alchian, 1963) and shipbuilding(Rapping, 1965), provide narratives consistent with thetheory that firms learn from past experience. Third,studies cite the generally high goodness-of-fit of powerfunctions to empirical data over several years, or evendecades, as validation of the model. Fourth, thedynamic aspect of the model—the rate of improvementadjusts to changes in the growth of production—makesthe model superior to forecasts that treat change purelyas a function of time.2 Finally, the reduction of thecomplex process of innovation to a single parameter, thelearning rate, facilitates its inclusion in energy supplyand computable general equilibrium models.The combination of a rich body of empirical literatureand the more recent applications of learning curves inpredictive models has revealed weaknesses that echoearlier critiques. First, the timing of future costreductions is highly sensitive not only to changes inthe market growth rate but also to small changes in thelearning rate. Although, an experience curve R2 value of40:95 is considered a strong validation of the experience curve model, variation in the underlying data canlead to uncertainty about the timing of cost reductionson the scale of decades. Fig. 1 shows experience curvesbased on the two most comprehensive world surveys ofPV prices (Maycock, 2002; Strategies-Unlimited, 2003).The Maycock survey produces a learning rate of 0.26while the Strategies Unlimited data give 0.17.3 What2An example of the opposite, a non-dynamic forecast, is autonomous energy efficiency improvement (AEEI) in which technologiesimprove at rates exogenously specified by the modeler (Grubb et al.,2002).3Note that the largest differences between the price surveys are in theearly stages of commercialization when using experience curves may beleast appropriate.may appear as a minor difference has a large effect. Forexample, assuming a steady industry growth rate of 15%per year, consider how long it will take for PV costs toreach a threshold of 0.30/W, an estimate for competitiveness with conventional alternatives. Just the difference in the choice of data set used produces a crossoverpoint of 2039 for the 0.26 learning rate and 2067 for the0.17 rate, a difference of 28 years. McDonald andSchrattenholzer (2001) show that the range of learningrates for energy technologies in general is even larger.Neij et al. (2003) find that calculations of the costeffectiveness of public policies are very sensitive to suchvariation. Wene (2000) observes this sensitivity as welland recommends an ongoing process of policy evaluation that continuously incorporates recent data.Second, the experience curve model gives no way topredict discontinuities in the learning rate. In the case ofPV, the experience curve switched to a lower trajectoryaround 1980. As a result, experience curve-based forecasts of PV in the 1970s predicted faster technologicalprogress than actually occurred (Schaeffer et al., 2004).Discontinuities present special difficulties at early stagesin the life of a technology. Early on, only a few datapoints define the experience curve, while at such timesdecisions about public support may be most critical.Third, studies that address uncertainty typicallycalculate uncertainties in the learning rate using thehistorical level of variance in the relationship betweencost and cumulative capacity. This approach ignoresuncertainties and limitations in the progress of thespecific technical factors that are important in drivingcost reductions (Wene, 2000). For example, constraintson individual factors, such as theoretical efficiencylimits, might affect our confidence in the likelihood offuture cost reductions.Fourth, due to their application in planning andforecasting, emphasis has shifted away from learningcurves based on employee productivity and plant-levelanalysis, to experience curves aggregating industries and

ARTICLE IN PRESSG.F. Nemet / Energy Policy 34 (2006) 3218–3232including all components of operating cost. While thestatistical relationships generally remain strong, theconceptual story begins to look stretched as one mustmake assumptions about the extent to which experienceis shared across firms. In the strictest interpretation ofthe learning-by-doing model applied to entire industries,one must assume that each firm benefits from thecollective experience of all. The model assumes homogenous knowledge spillovers among firms.Fifth, the assumption that experience, as representedby cumulative capacity, is the only determinant of costreductions ignores the effect of knowledge acquiredfrom other sources, such as from R&D or from otherindustries. Earlier, Sheshinski (1967) wrestled with theseparation of the impact of two competing factors,investment and output. Others have addressed thislimitation by incorporating additional factors such asworkforce training (Adler and Clark, 1991), R&D(Buonanno et al., 2003; Miketa and Schrattenholzer,2004), and the interactions between R&D and diffusion(Watanabe et al., 2000). The amount of data requiredfor parameter estimation has so far limited widespreadapplication of these more sophisticated models.Finally, experience curves ignore changes in qualitybeyond the single dimension being analyzed (Thompson, 2001).4 The dependent variable is limited to costnormalized by a single measure of performance—for example, hours of labor/aircraft, /W, or b/megabyte. Measures of performance like these ignorechanges in quality such as aircraft speed, reliability ofpower generation, and the compactness of computermemory.1.2. ApproachThis study seeks to understand the drivers behindtechnical change in PV by disaggregating historic costreductions into observable technical factors. The mechanisms linking factors such as cumulative capacityand R&D to technological outcomes, while certainlyimportant, are at present not well understood. Many ofthe problems mentioned above arise because theexperience curve model relies on assumptions aboutweakly understood phenomena. Rather than makingassumptions about the roles that factors like experience,learning, R&D, and spillovers play in reducing costs, aset of observable technical factors are identified whoseimpact on cost can be directly calculated.This study includes the period from nascent commercialization, 1975, to 2001. During this 26-year period,there was a factor of 20 cost reduction in the cost of PVmodules. Only PV modules are examined and balanceof-system components such as inverters, storage, and4Payson (1998) provides an alternative framework that incorporatesboth changes in quality and cost improvements.3221supporting structures are excluded.5 The focus here is onexplaining change in the capital cost of PV modules,rather than on the cost of electricity produced, mainlydue to data quality considerations and to be able toexclude influential but exogenous factors such as interestrates. The study is limited to PV modules manufacturedfrom mono-crystalline and poly-crystalline silicon wafers because crystalline silicon has been the overwhelmingly dominant technology for PV over this period.Crystalline silicon PV comprised over 90% of production over this period and its share increased in thesecond half of the period.6 While photovoltaic electricityhas been produced from a wide variety of othermaterials, such as cadmium-telluride and copper-indium-diselenide, during the study period these competing technologies remained in the development stage andwere not commercially relevant. The price data used inthe study are weighted averages of the two types ofsilicon crystals. The study uses worldwide data ratherthan country-level data because over this time period themarket for PV became global. Some of the change oftenattributed to within-country costs is due to theglobalization of the industry, rather than learning fromthat country’s experience. Junginger et al. (2005)articulated the need for such an international view andas a result developed a global experience curve for windpower. This study adopts a similarly global view. Thescope of this study thus addresses the concerns raised bySchaeffer et al. (2004) regarding the importance of dataquality, system boundaries, and sufficient historical timeperiod for assessing experience in energy technologies.Finally, the technological characteristics of PV providetwo simplifying aspects that help restrict the influence ofpotentially confounding factors in the study. First, therehas been no significant change in per unit scale in PVpanels. PV panels have been sized on the order of onesquare meter per panel for three decades. Compare thisto wind turbines in which the size of individual units hasincreased by almost two orders of magnitude overthe same period (Madsen et al., 2003; Jungingeret al., 2005).Second, there are essentially no operation and maintenance costs associated with PV, otherthan regular cleaning and inverter replacement. Thislimits the role of ‘‘learning-by-using’’, which wouldnormally be an important additional factor to consider(Rosenberg, 1982).The analysis began by identifying factors thatchanged over time and had some impact on PV costs.Using empirical data, the annual level of these seven5Inverters and other components have similar progress ratios tomodules and have exhibited cost decreases by factors of 5 and 10respectively.6Crystalline silicon makes up close to 100% of the market forapplications of 41 kW, a definition of the market that includeshousehold-scale and larger power generation and excludes consumerelectronics.

ARTICLE IN PRESS3222G.F. Nemet / Energy Policy 34 (2006) 3218–3232factors over the study period, 1975–2001, was compiledand a model to quantify the impact of the change ineach factor on module cost developed.2. Cost model methodologyThis cost model simulates the effect of changes in eachof seven factors on manufacturing cost in each year, t, asfollows.2.1. CostAverage module cost (C) in Wpeak is the dependentvariable in the model.7 The time series for cost uses anaverage of the two most comprehensive world surveys ofPV prices (Maycock, 2002; Strategies-Unlimited, 2003).Using prices as a proxy for costs is a widespread practicewhose validity is discussed below. The model usesmodule cost, rather than cost of energy produced, toavoid the large uncertainties associated with makingassumptions about capacity factors, lifetimes, andfinancing mechanisms.2.2. Module efficiencyImprovementsintheenergyefficiency(Z ¼ W out W in ) of modules sold have nearly doubledthe rated power output of each square meter (m2 )of PV material produced (Christensen, 1985

Energy Policy 34 (2006) 3218–3232 Beyond the learning curve: factors influencing cost reductions in photovoltaics Gregory F. Nemet Energy and Resources Group, University of California, 310 Barrows Hall 3050, Berkeley, CA 94720-3050, USA

Related Documents:

May 02, 2018 · D. Program Evaluation ͟The organization has provided a description of the framework for how each program will be evaluated. The framework should include all the elements below: ͟The evaluation methods are cost-effective for the organization ͟Quantitative and qualitative data is being collected (at Basics tier, data collection must have begun)

Silat is a combative art of self-defense and survival rooted from Matay archipelago. It was traced at thé early of Langkasuka Kingdom (2nd century CE) till thé reign of Melaka (Malaysia) Sultanate era (13th century). Silat has now evolved to become part of social culture and tradition with thé appearance of a fine physical and spiritual .

On an exceptional basis, Member States may request UNESCO to provide thé candidates with access to thé platform so they can complète thé form by themselves. Thèse requests must be addressed to esd rize unesco. or by 15 A ril 2021 UNESCO will provide thé nomineewith accessto thé platform via their émail address.

̶The leading indicator of employee engagement is based on the quality of the relationship between employee and supervisor Empower your managers! ̶Help them understand the impact on the organization ̶Share important changes, plan options, tasks, and deadlines ̶Provide key messages and talking points ̶Prepare them to answer employee questions

Dr. Sunita Bharatwal** Dr. Pawan Garga*** Abstract Customer satisfaction is derived from thè functionalities and values, a product or Service can provide. The current study aims to segregate thè dimensions of ordine Service quality and gather insights on its impact on web shopping. The trends of purchases have

behringer ultra-curve pro dsp 24 a/d- d/a dsp ultra-curve pro ultra- curve pro 1.1 behringer ultra-curve pro 24 ad/da 24 dsp ultra-curve pro dsp8024 smd (surface mounted device) iso9000 ultra-curve pro 1.2 ultra-curve pro ultra-curve pro 19 2u 10 ultra-curve pro ultra-curve pro iec . 7 ultra-curve pro dsp8024 .

Chính Văn.- Còn đức Thế tôn thì tuệ giác cực kỳ trong sạch 8: hiện hành bất nhị 9, đạt đến vô tướng 10, đứng vào chỗ đứng của các đức Thế tôn 11, thể hiện tính bình đẳng của các Ngài, đến chỗ không còn chướng ngại 12, giáo pháp không thể khuynh đảo, tâm thức không bị cản trở, cái được

Food outlets which focused on food quality, Service quality, environment and price factors, are thè valuable factors for food outlets to increase thè satisfaction level of customers and it will create a positive impact through word ofmouth. Keyword : Customer satisfaction, food quality, Service quality, physical environment off ood outlets .