Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 52
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop ENERGY EFFICIENCY COST CURVES: EMPIRICAL INSIGHTS FOR ENERGY-CLIMATE MODELING Jayant Sathaye and Amol Phadke Lawrence Berkeley National Laboratory Berkeley, California Abstract In this paper, we report on implications of key energy efficiency issues such as barriers that include market failures; policies and programs; co-benefits; and historical changes in costs. We show examples for their representation in selected energy climate (EC) models and other analytical approaches. We report on two approaches for the quantification of barriers. Our analysis of California utilities’ efficiency policies and programs exemplifies a cost effective approach to overcoming these barriers. The analysis of energy efficient clothes washers shows the importance of including water savings as an invaluable co-benefit, and the examples of iron and steel and cement show significant reduction in costs of energy efficiency options from inclusion of multiple co-benefits. Lastly, we illustrate the changes in costs of efficient products in U.S. industrial and residential end uses over time, which call for the use of dynamic as opposed to static cost curves in EC models. Keywords Energy efficiency, cost of conserved energy, empirical insights, energy-climate modeling, co-benefits, dynamic cost curves. Introduction Adoption of efficient end-use technologies is one of the key measures for reducing GHG emissions. In many cases, these are cost effective investments that an energy consumer could make for improving energy productivity while reducing GHG emissions. With the rising interest in policies and programs to reduce GHG emissions, estimating the costs of energy efficiency options and managing them has become increasingly important for policy makers and consumers around the world. Energy-climate (EC) models are classically used for analyzing the costs of reducing carbon and other GHG emissions for various types of technical and policy measures. An increasing number of models3 are now representing energy efficiency measures because an accurate estimation of these costs is critical for identifying and choosing the measures, and for developing related policy options to accelerate their market adoption and implementation. However, the accuracy of assessing GHG-emission reduction costs by taking into account the adoption of energy efficiency technologies will depend on how well these end-use technologies are represented in the models. For example, if the models do not include end-use technologies with an appropriate level of detail in their modeling framework, it will be difficult to estimate the costs and benefits of reducing GHG emissions with certainty. In this paper, we review three topics related to the representation of energy efficiency improvement opportunities in energy-climate models. These include the treatment of (1) barriers to energy efficiency as a no- regrets option,4 (2) co-benefits and costs, e.g., due to water and labor savings, and (3) changes in energy efficiency costs over time. In the second section, we review the literature on energy efficiency cost curves as a way of representing cost and benefits associated with energy efficiency options and also broadly describe the representation of energy efficiency in EC models. 3 For example, those participating in the EMF24 exercise. 4 No-regrets options are those whose benefits such as reduced energy costs and reduced emissions of local and regional pollution equal or exceed their cost to the society, excluding the benefits of avoided climate change (IPCC 2001, p. 21).
OCR for page 53
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop Representation of energy efficiency as a no-regrets option requires an estimate of the costs indirectly imposed by the barriers to their adoption or conversely of the costs of policies and programs that are established to address these barriers. In the third section, we review these issues and summarize the results of the empirical analysis we have conducted on this topic on the quantification of the principal agent barrier in the U.S. residential sector. We also undertake an empirical assessment of the costs and benefits for overcoming these barriers by policies and programs by analyzing the energy efficiency programs implemented by utilities in California, which are one of the world’s largest energy efficiency programs. The section also includes a brief description of the use of such costs and benefits in EC models. In the fourth section, we focus on the importance of considering the co-benefits of adopting energy efficiency measures. Based on the empirical analysis of the U.S. iron and steel industry, and residential clothes washers, we show how the estimates of the potential of no-regrets options changes when co-benefits are included in the analysis. In the fifth section, we report on the historical changes in the cost curves in industrial and residential sectors, and how these dynamic changes are likely to influence the costs of carbon emissions reduction. The last section summarizes our findings and provides recommendations for future work. Representation of Energy Efficiency: Cost Curves and Energy-Climate Models In this section, we first review the use of cost curves, one of the commonly used approaches to represent and compare various GHG mitigation options including energy efficiency improvements. We then broadly review how energy efficiency improvement options are represented in energy climate models. Energy Efficiency Cost Curves The energy sector comprises the major energy demand sectors (industry, residential and commercial, transport, and agriculture), and the energy supply sector, which consists of resource extraction, conversion, and delivery of energy products. Greenhouse gas emissions occur at various points in the sector, from resource extraction to end use, and accordingly, options for mitigation exist at any of these points. The bottom-up approach involves the development of scenarios based on energy end uses and evaluation of specific technologies that can satisfy demands for energy services. One can compare technologies based on their relative cost to achieve a unit of GHG reduction and other features of interest. This approach gives equal weight to both energy supply and energy demand options. A variety of screening criteria, including indicators of cost-effectiveness as well as non-economic concerns, can be used to identify and assess promising options, which can then be combined to create one or more mitigation scenarios. Mitigation scenarios are evaluated against the backdrop of a baseline scenario, which simulates the events assumed to take place in the absence of mitigation efforts. Mitigation scenarios can be designed to meet specific emission reduction targets or to simulate the effect of specific policy interventions. The results of a bottom-up assessment can then be compared to a top-down analysis of the impacts of energy sector scenarios on the macro-economy. In this paper, we will focus on energy efficiency options and the representation of their marginal cost curves, typically referred to as cost of conserved energy (CCE) curves. CCE curves were developed about three decades ago to place energy-efficiency cost estimates at a level comparable to that for supply-side options (Meier, 1982). A CCE curve is made up of a combination of several options and can be sector-specific or economy-wide. The CCE is estimated for each mitigation option and plotted against its resulting energy or emissions savings.5 A combination of such calculations yields a curve of CCE for a suite of mitigation options. The CCE calculation is based on investment theory6 and it is expressed as: 5 One criticism that has been directed at the calculation of CCEs is that they may err by aggregating across the entire stock, without differentiating costs, and discount rates for various classes of consumers. One way to avoid this mistake is illustrated by Sathaye and Murtishaw (2004) in the clothes washer analysis by disaggregating potential adopters by income class. 6 Stoft (1995) offers an alternative expression that avoids the problem of double counting due to dependence among measures on a curve.
OCR for page 54
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop (1) (2) Where: CCE = Cost of conserved energy for an energy-efficiency measure (or mitigation option), in $/GJ I = Incremental Capital cost ($) q = Capital recovery factor (yr−1) ES= Annual energy savings (GJ/yr) d = Discount rate n = Lifetime of the mitigation option (years) These savings are estimated by comparing the electricity use of an efficient and an inefficient end-use technology, a compact fluorescent lamp (CFL) vs. an incandescent bulb for example. The comparable device has to be such that it offers the same service in terms of the lumen output, color rendering, and robustness to quality of electricity supply. The above expression can easily be converted to a cost of avoided GHGs by applying emission factors to the affected energy source. Appropriate terms can be added to the equation (1) to take into account various other costs such as operating and transaction costs (such as search costs), and benefits such as material and labor savings and environmental benefits. Representation of Energy Efficiency in Existing Energy-Climate Models Integrated assessment models originally emerged primarily from economic and energy modeling approaches that were for the most part developed for, and applied to, industrialized economies (Sanstad and Greening, 1998). Increasingly, however, these models have been enhanced and extended over time, and in many cases created, to encompass the global economy at various levels of regional and sectoral disaggregation. Integrated assessment modeling of climate policy uses various top-down models that describe the general economy and its interactions, and the effects of price changes. Many of these models include a sectoral representation of the economy (see Ross, 2005) for the description of the ADAGE model, as an example of a top-down model). The existing empirical basis for modeling of technologies that represent these sectors is often weak, and largely arises from limited literature at the sectoral level rather than technology-level. There is a need to constantly investigate and improve the representation of end-use technologies in energy-climate models, in coordination with EC modelers who will stand to benefit from this research. Bottom-up models, on the other hand, have detailed representation of GHG mitigation technologies including those of the demand side. However, often the representation of supply side options is more rigorous and detailed than that on the demand side. Further, the representation of various barriers to adoption of efficiency measures that appear to be cost effective from a societal perspective is often indirect. For example, most bottom-up models calibrate the adoption of energy efficiency measures predicted by the model to that observed in reality by artificially increasing the discount rate (also known as the hurdle rate) used by consumers to evaluate these investments or by introducing artificial limits to cost effective technology updates. The effect of policies and programs is indirectly modeled by changing the discount rate (see for example Latiner and Hanson, 2006, for such analysis using the AMIGA model). These methods are indirectly trying to represent various barriers faced by consumers in adoption of energy efficiency options. Only a limited number of analyses attempt to model the impact of energy efficiency policies and programs which address some of the barriers to adoption of energy efficiency and have limited empirical basis of how the effect of policies and programs is modeled (see for example, Roland-Host, 2006 which considers the effect of efficiency policies and programs while undertaking energy climate modeling of California’s economy using the BEAR model).
OCR for page 55
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop Although most energy-climate models consider technological change in energy supply and use, modeling of technological change in demand-side energy efficiency technologies endogenously is a new topic. Given the growing importance of technological improvement (e.g., energy efficiency) as an avenue to mitigate long-term climate change, it is critical that technology characteristics, their evolution and response to energy and carbon price be understood better than has been the case to date. This is also particularly true of developing countries where obsolete technologies are likely to see a more rapid transformation, as their markets integrate into the global economy, while newer technologies are likely to be adopted faster due to evolving global markets and more availing policy support. In the following sections, we discuss how the analysis of energy efficiency as a mitigation option can be improved in analyses performed using energy climate models or other methods, by addressing some of the issues raised above. Energy Efficiency as a No-Regrets Option The Intergovernmental Panel on Climate Change defines no-regrets options in its Third Assessment as “…those options whose benefits such as reduced energy costs and reduced emissions of local and regional pollution equals or exceeds their cost to the society, excluding the benefits of avoided climate change” (IPCC 2001, p. 21). This definition suggests that no-regrets options should be pursued even without considering their benefits of avoided climate change. Alternatively, the benefits of avoiding climate change do not result into net costs to the society but in fact result in net benefits to the society. Hence no-regrets options are also known as negative-cost (equivalently net-benefit) options for avoiding climate change. There is a lot of debate on the availability of no-regrets options (Ostertag, 2006). Those who posit that significant no-regrets options do not exist (see for example, Sutherland, 1991) argue that if no-regret options were available, they would have been pursued by the market. They typically argue that engineering-economic studies which show a large potential for no-regrets options such as cost effective energy efficiency measures often do not take into account many indirect costs of these measures such as search costs and transaction costs. Further, these studies do not take into account the effect of various market failures and consumer preferences, which hinder the adoption of no-regret options that appear to be cost effective in an engineering-economic type of analysis. Researchers who posit that significant no-regrets options do not exist argue that cost effective ways of reducing these indirect costs and correcting various market failures which hinder the adoption of energy efficient technologies are rarely available. On the other hand, those who posit that significant no-regret options exist generally argue that there are various cost effective policy measures which could reduce the indirect costs associated with no-regret options and correct some of the market failures which hinder their adoption. We review some of the literature on barriers (including transaction costs) and approaches to characterize the potential for energy efficiency as a GHG mitigation option and the associated policies and programs that address these barriers Barriers, Potentials, and Policies: Overview Earlier reports have enumerated lists of several factors (barriers) affecting the penetration of energy-efficient devices by customer class or tariff category, region and/or sector (Reddy, 1991; Golove and Eto, 1996; Eto, Prahl, and Schlegel, 1997; Sathaye and Bouille et al., 2001). These factors include lack of information, lack of access to capital, misplaced incentives, flaws in market structure, performance uncertainties, decisions influenced by custom and habits, inseparability of features, heterogeneity of consumers, hidden costs, transaction costs, bounded rationality, product unavailability, externalities, imperfect competition, etc. The extent of their inclusion affects both costs and the mitigation potential of a technology or a mix of technologies. Sathaye and Bouille (2001), following on the work of Jaffe and Stavins (1994), classify factors into two categories. The first category refers to factors that economists may typically classify as “market failures,” the second, to factors that are manifestations of consumer preferences, custom, cultural traits, habits, lifestyles, etc. Associated with each category is the concept of potentials for GHG mitigation (Figure C.6). Each concept of the potential represents a hypothetical projection that might be made today regarding the extent of GHG mitigation. The leftmost line, labeled market potential indicates the amount of GHG mitigation that might be expected to
OCR for page 56
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop FIGURE C.6 Penetration of mitigation technologies: A conceptual framework. occur under forecast market conditions, with no changes in policy or implementation of measures whose primary purpose is the mitigation of GHGs. At the other extreme, the technical or physical potential describes the maximum amount of GHG mitigation achievable through technology diffusion. This is a hypothetical projection of the extent of GHG mitigation that could be achieved over time if all technically feasible technologies were used in all relevant applications, without regard to their cost or user acceptability. Definitionally, we can say that whatever physical, cultural, institutional, social, or human factors are preventing us from reaching the technical potential are barriers to the mitigation of GHG via technology diffusion. Since our ultimate goal, however, is to understand policy options for mitigation, it is useful to group these factors in a way that facilitates understanding of the kinds of policies that would be necessary to overcome them. As we create these different categories of factors, we correspondingly create intermediate conceptions of the potential for GHG mitigation. Starting at the left in Figure C.6, we can imagine addressing factors (often referred to as market failures) that relate to markets, public policies, and other institutions that inhibit the diffusion of technologies that are (or are projected to be) cost-effective for users without reference to any GHG benefits they may generate. Amelioration of this class of market imperfections would increase GHG mitigation towards the level that is labeled as the economic potential. The economic potential represents the level of GHG mitigation that could be achieved if all technologies that are cost-effective from consumers’ point of view were implemented. Because economic potential is evaluated from the consumer’s point of view, we would evaluate cost-effectiveness using market prices and the private rate of time discounting, and also take into account consumers’ preferences regarding the acceptability of the technologies’ performance characteristics. Some of the market failures listed above can be broadly grouped together as cognitive factors affecting product diffusion. By this, we mean that there are limitations to consumers’ ability to gather and process information. Before any consumer can make the decision to adopt a technology, he or she must at a minimum be aware of its existence. Once aware, a consumer needs to make some effort to gather the information needed to make an informed decision about whether a given technology provides more benefits than it costs. In order to do this, an individual needs the analytic capacity to fairly accurately quantify the benefits and costs. Even an aware, informed, capable consumer must ultimately make the effort to assess benefits and costs before making the decision to adopt. A consumer who has made the decision to adopt needs to find a vendor for the product in question. Relatively new technologies are likely to be less widely available than their more standard counterparts. Thus, limitations on cognitive resources are described by factors such as performance uncertainty, information costs, and bounded rationality. Elimination of all of these market imperfections would not produce technology diffusion at the level of the technical potential. That is, even if these factors are removed, some GHG-mitigating technologies may not be widely used simply because consumer preferences operate against their acceptance. These factors, which define
OCR for page 57
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop the gap between economic potential and technical potential, are usefully placed in two groups separated by a socio-economic potential. The socio-economic potential represents the level of GHG mitigation that would be achieved if all technologies that are cost effective on the basis of using a social, rather than a private, rate of discount (including externalities, with the use of appropriate prices devoid of taxes and subsidies) were implemented. The socio-economic potential may or may not require a change in consumer preferences. Finally, even if all market, institutional, social, and cultural factors whose removal is cost-effective from a societal perspective were removed, some technologies might not be widely used simply because they are still too expensive. Elimination of this requirement would therefore take us to the level of the technical potential, the maximum technologically feasible extent of GHG mitigation through technology diffusion. Moving from right to left, the figure shows that factors increase costs and reduce the savings potential of a mitigation technology. Another commonly considered factor in the development of cost curves is the increased use of energy that might be prompted by the reduction in total cost of energy use due to the adoption of efficient devices. Indeed, several papers show that a rebound effect can range from 10 to 40% (Sorrell 2009; Geller and Attali 2005) A rebound effect, however, would also be accompanied by increased consumer welfare or increased production of goods due to the increased energy use. Moreover, Ehrhardt-Martinez and Laitner (2010) suggest that the magnitude of rebound can be mitigated by the adoption of smart “people-centered” or behavioral initiatives that enable households and businesses to more effectively manage their overall levels of energy consumption. Figure C.6 also presents a snapshot in time of the factors and potentials for the penetration of technologies. Over time, technological progress, discoveries of new resources and/or technologies, and cost-effective government policies and programs could eliminate some of the factors and hence move the potential lines (including the technical potential) to the right, thereby increasing the savings from a mitigation option. The figure also shows that transaction costs add to the cost of the mitigation option. As a market matures, the decline in transaction costs caused by learning by doing and standardization will push the cost curve lower, which will increase the market penetration of a technology. The focus of this paper is on the factors affecting the realization of GHG reductions from energy efficiency. The price of energy plays a role in determining the energy savings potential—the higher the price, the larger the potential, and vice versa. The price or tariff line is seen to intersect the marginal cost curve of energy efficiency savings. Should the tariff be higher, then more of these savings would be cost effective than would be the case otherwise. Barriers, Policies, and Programs: Insights from Empirical Analysis As discussed above, various barriers cause the market potential for energy efficiency improvements to be smaller than its economic potential. Quantifying the impact of diverse barriers on energy efficiency cost and potential is a challenging task but it provides a better understanding of the reasons for the difference between the market and economic potentials and the benefits of addressing various barriers through policies and programs. An earlier report by Sathaye and Murtishaw (2004) predicted several likely barriers (split incentives, access to capital and efficient products, lifetime and consumption uncertainty, limited product and vendor information, and consumer preferences) that prevent the purchase of efficient residential lamps and clothes washers. For each barrier, they estimated its effect on the capital cost, annual costs, equipment lifetime, or discount rate in the CCE Equation 1, and on the number of consumers that are likely to adopt this measure, which influences the total potential energy savings. Barriers such as split incentives for example do not directly increase the costs of adopting energy efficient products; however, they limit the consumer base that is likely to adopt this measure. Barriers such as limited product and vendor information and uncertainty in performance directly add to the cost of adoption. The analysis showed the relative share of these costs in shifting from the socio-economic potential to the market potential shown in Figure C.6 above. The cost of accessing information and the uncertainty about equipment lifetime turned out to be the largest contributors to the costs imposed by various barriers in their analysis. A second approach quantified the impact of the principal agent problem on the sales of efficient U.S. residential appliances (Murtishaw and Sathaye, 2007 and De La Rue du Can and Sathaye, 2008). This analysis and its results are described in the section “Quantifying the Impact of Barriers” below.
OCR for page 58
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop It has been demonstrated that policies and programs which address various barriers lead to increased adoption of energy efficiency (see section entitled “Costs and Benefits of Policies and Programs That Address Barriers”). These policies and programs also incur some costs such as administrative, marketing, installation, and financial incentives (which are more of a transfer than actual cost). From a societal perspective, the desirability of these programs depends on whether the benefits of these policies and programs outweigh their costs and empirical evaluation of the same is warranted. Further, it is important to analyze the impact of policies and programs to estimate how the impact of energy efficiency gets adopted. We report the results of our empirical analysis of utility energy efficiency programs in California, which are one of the largest and most comprehensive energy efficiency programs in the world. Quantifying the Impact of Barriers Multiple barriers have been listed in the previous section, but until recently few studies to our knowledge had attempted to quantify the “excess” energy consumed due to any particular barrier or the savings potential from mitigating it. Murtishaw and Sathaye (2007) and De La Rue du Can and Sathaye (2008) focused on the principal agent (PA)7 or split incentives problem because quantifying its extent and the excess energy consumption it causes was perceived to be more tractable than for other market failures.8 In residential energy use, this commonly occurs in two critical transactions, one between home builders and prospective buyers, and the second between renters and landlords (Jaffe and Stavins, 1994). Home builders may have difficulty conveying the benefits of energy efficiency technologies to prospective buyers because these technologies and their future energy use consequences are not observable. Likewise, landlords may not be able to recover all of the value of such investments in the form of higher rents, where renters pay fuel bills, and tenants who make these investments in cases where the landlord pays the energy bill may not be able to get reduced rents. From a policy perspective, split incentives can block or delay utility price signals from reaching the end-user. Murtishaw and Sathaye (2007) thus provide a quantitative basis for supporting other forms of government interventions that complement price policy for increasing the penetration of cost-effective energy efficient products. Such interventions may include the provision of additional and targeted information, energy performance standards and labels, and building codes. In order to determine whether any particular end use is affected by the PA problem, three questions must be answered. First, who uses the device? Second, who selects the device? Third, who pays the energy cost? Theoretically, if the answer to these questions is not the same person or entity, a PA problem exists, albeit of two different types. If the person paying the utility bill is not the person using the device, the user may consume more energy services than if he were not shielded from the price of energy. Similarly, if the person paying for energy is not the person choosing the device, the buyer will generally choose among the cheapest, and often least efficient, options. Thus, the PA problem can arise from two kinds of split incentives, one concerning usage (demand for energy services) and the other concerning the technical efficiency of the end-use device. For any given device, determining the cases involving a PA problem may be conceptualized as a two-by-two table that classifies the device according to a user’s ability to choose the device and the user’s responsibility for paying associated energy costs (Table C.1). The analysis confirms that price signals alone may have a limited effect on inducing energy conservation in the U.S. residential sector because a significant share of energy is consumed by end users who either have little or no control over the efficiency of energy-using equipment (Case E) or who are shielded to some extent from 7 The PA problem arises in many spheres of economic activity, when one person, the principal, hires an agent to perform tasks on his behalf but cannot ensure that the agent performs them in exactly the way the principal would like (Bannock et al. 1992). The efforts of the agent are impossible or expensive to monitor and the incentives of the agent differ from those of the principal. Thus, the PA problem is a function of incentives, information asymmetry, and enforcement capacity. 8 Results of this study are also reported in an International Energy Agency (IEA) publication, Mind the Gap (IEA 2007). The IEA publication reports on case studies that used the same methodology in Japan, the Netherlands, and Norway primarily in the commercial and residential sectors.
OCR for page 59
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop TABLE C.1 Shares of Site Energy by End Use Affected by Principal-Agent Problems Can Choose Device Cannot Choose Device Direct Energy Payment Case N: Case E: No PA Problem Efficiency Problem Refs: 72% Refs: 25% WH: 31% WH: 59% SH: 64% SH: 31% AC: 63% AC: 31% Indirect Energy Payment Case B: Case U: Usage and Efficiency Problem Usage Problem Refs: <1% Refs: 3%a WH: negligible WH: 10% SH: negligible SH: 5% AC: <1% AC: 6% a Refrigerators are an exception since no usage problem exists in Case U, assuming same agent (e.g. landlord) chooses the device and pays for energy. the costs of their energy consumption (Cases B and U). Table C.1 highlights the fact that a conspicuous share of energy use falls into Cases E and U. The bulk of the energy affected is characterized by Case E. In order to assess the potential energy savings by overcoming PA problems, we applied to PA affected households the same Energy Star appliances penetration rate that was applied to non-affected households for each end use. Table C.2 provides a summary of the results by end use. Space and water heating were estimated to have the largest potential. This is in part due to the fact that these end uses represent the largest share of energy use in the residential sector (47% for space heating and 17% for water heating in 2001) and also because the penetrations of Energy Star boilers and furnaces are much higher than for the other appliances considered, thus increasing the potential savings for PA-affected households. A recent statistical evaluation of the U.S. residential energy consumption survey (RECS) data by Davis (2010) reconfirms the above analysis. Its results show that, controlling for household income and other household characteristics, renters are significantly less likely to have energy efficient refrigerators, clothes washers and dishwashers. Can this type of information be included in the types of EC models that were discussed above? A model such as NEMS for example generates projections of total residential energy demand and appliance stocks. It is a complex model, with a detailed accounting system for tracking appliance stocks over time in each of nine different census regions. However, new appliance purchase decisions in the model are based on a fairly simple methodology, including three housing types and three market share “logit” equations. Appliance energy use for any given household type varies as a function of the efficiency of appliances (as determined above), the rebound effect and the price elasticity of energy demand. Models such as NEMS that are designed to include or already include a representation of end uses can be modified to provide an explicit consideration of the PA problem. The household groupings shown in Table C.1 TABLE C.2 Estimated Energy Savings per End Use End Use Energy Savings (Tbtu) Site Energy Primary Energy Space heating 8.65 9.86 Water heating 4.60 5.90 Refrigerators 0.16 0.49 Air conditioning 0.20 0.60
OCR for page 60
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop have different price elasticities of demand to purchase appliances and to use them. Group N households readily change both purchase and use behavior in response to a shift in energy price (high price elasticity to buy and use). Group E households change use when the price shifts, but are slow to change purchase behavior (low price elasticity to buy; high price elasticity to use). Group U households are slow to change use and purchase behavior in response to price shifts (low price elasticity to buy and use). Data permitting, a more detailed representation of the four categories of households and associated different elasticity values would give explicit representation of the PA problem in these models. Costs and Benefits of Policies and Programs That Address Barriers As shown in Figure C.6, cost effective policies and programs can address some of the barriers and increase the energy efficiency potential that can be achieved. Hence understanding the cost, benefit, and potential for these cost effective policies and programs is critical for accurately estimating/representing the role EE can play as a mitigation option. In the United States, utility programs address these barriers to capture some of the “negative-cost” potential, and expenditures on these programs provide an estimate of the cost of overcoming some of these barriers. Hence by adding these program costs to the cost of efficiency measures typically estimated by engineering economic studies, one can potentially estimate the societal cost of energy efficiency as a mitigation option. Many energy climate models, which focus on analyzing the role of various mitigation options in climate stabilization scenarios, can consider these program costs and model accelerated adoption of cost effective measures due to these programs. Most energy-climate models do not undertake this exercise given the lack of empirical information on program costs especially at the measure level. We present the results of the analysis of efficiency programs implemented by California utilities during 2006-2008. California has long been a national leader in promoting energy efficiency: the state’s policies, programs, and standards have served as a model for federal policies as well as initiatives in other states. California utilities undertake one of the largest ratepayer funded energy efficiency initiatives in the world. The state’s utility energy efficiency programs are numerous, costing about one billion dollars per year. These programs are funded by resources collected from ratepayers and are overseen by the California Public Utilities Commission (CPUC), which regulates the investor owned utilities (IOUs) in the state. IOUs are required to file monthly, quarterly, and annual reports, which provide various details on the outcomes of their energy efficiency programs to the California Public Utilities Commission. Table C.3 shows the average net consumer cost per kWh, rebate and non-rebate utility expenditures per kWh, and the societal cost of conserved energy. The incremental marginal cost borne by consumers (IMC) ranges from 1.73 to 2.85 cents/kWh. The rebate offered by each utility company is shown in the second column (0.9-1.22 cent/kWh). The third column shows the net cost to consumers (Net Cost = IMC cost – rebate cost) (0.83-1.63 cents/kWh). Non-rebate utility expenditures vary widely from 0.5 cent/kWh for SCE up to 1.1 cent/kWh for SDG&E. TABLE C.3 Cost of Conserved Energy (cents/kWh), 2006-2008 Utility Consumer Incremental Marginal Cost / kWh Rebate / kWh Net Consumer Cost / kWh Non-rebate Utility Exp/kWh Societal Cost / kWh PGE 1.73 0.90 0.83 0.99 2.72 SCE 2.85 1.22 1.63 0.50 3.35 SDGE 2.34 0.98 1.36 1.10 3.44 PGE—Pacific Gas and Electric, SCE—Southern California Edison, SDGE—San Diego Gas and Electric.
OCR for page 61
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop The societal cost of conserved energy, which is the sum of the rebate, net consumer cost, and non-rebate utility expenditures, ranges from 2.72 cents/kWh for PG&E to 3.44 cents/kWh for SDG&E, within the expected range for an energy efficiency portfolio. The societal cost is much lower than the cost of supply from various generation technologies in California. A CEC report calculated that the levelized cost of IOU-owned plants brought online in 2007 is 9.6 cents/kWh for integrated gasification combined cycle, 6.7 cents/kWh for Class 5 wind, and 19 cents/kWh for concentrated solar PV (CEC 2007). The societal CCE is also much lower than electricity tariffs; PG&E’s average rate was 17.6 and 17.9 cents/kWh for residential and commercial consumers respectively. The non-rebate utility expenditure on incentives and other program implementation items such as providing information and installation services, which address some of the barriers to adoption of energy efficiency measures, are substantially lower than their benefits indicating that such programs are a cost effective way of addressing barriers to adoption of energy efficiency measures. These numbers are consistent with findings of other studies of the cost effectiveness of energy efficiency programs. A ACEEE report lists the cost of conserved energy in California as 3 cents/kWh (Kushler et al., 2004).9 The CCE for New Jersey and Vermont is also similar at 3 cents/kWh, while Connecticut is at the low end with 2.3 cents/kWh and New York at the high end with 4.4 cents/kWh.The rebate (subsidy) provided to consumers ranged between 42-52% of the total IMC. The substantial rebate decreases the payback period, or the amount of time needed for consumers to recover their initial outlay. For example, the simple payback period (without discounting) for PG&E’s energy efficiency measures is 1.7 years for electricity savings, and roughly 80% have a simple payback period of 1 year or less. In contrast, if no financial incentives were provided, the simple payback period would be 3.1 years.10 The utilities are thus attempting to overcome consumer myopia by providing rebates and reducing their payback period. If we believe that the utilities are using incentives optimally—that is, providing the minimum amount of incentive to achieve a certain target of savings over a given period—then the implication is that the payback period must be brought down substantially (to 1 year or less in this particular instance) for large-scale promotion of energy efficiency measures. Indeed, given the many barriers to adoption of energy efficiency measures, such as limited availability of information, uncertainty about performance, split incentives, and limited product availability, this reduction in the payback period may be essential. Utility expenditures, which are categorized into administrative, marketing/advertising/outreach, and direct implementation costs, averaged 1.72 cents/kWh to 2.08 cents/kWh over the 3-year program cycle (see below). As mentioned above, direct implementation costs include labor for installation and service, hardware and materials, rebate processing, and rebates for customers. The average rebate per kWh of energy savings is represented by the dotted line in Figure C.7.11 For PG&E and SDG&E, the average rebate is about 47% of total expenditures, while SCE’s average rebate is 71% of total expenditures, at 1.2 cents/kWh. Energy-climate models can use the empirical information on policy and program cost and impacts to evaluate the role that energy efficiency policies and programs can play in GHG reduction. For example, the analysis of GHG mitigation options by Roland-Holst (2006) using the BEAR model illustrates one approach to incorporating these costs in an EC model. The cost of implementing these programs and policies is assumed to be borne entirely by the government and is added to the government expenditure. The estimated gains in energy efficiency due to these policies and programs are used to modify the assumptions about the energy intensity of economic output in the model. Roland-Holst (2006) estimates that implementing no-regret options in California could lead to a reduction of 83 Mt CO2eq. by 2020 (which is 50% of the emission reduction target set by California Assembly Bill 32) and would result in a net economic benefit of $58.8 billion. Bottom-up models can take advantage of some of the additional empirical information available at each energy efficiency measure level while modeling the adoption of energy efficiency technologies. Bottom-up models like NEMS estimate the effect of programs such as providing rebates to consumers to accelerate the adoption of 9 Kushler et al. (2004) Five Years In: An Examination of the First Half-Decade of Public Benefits Energy Efficiency Policies. ACEEE. 10 Payback periods were calculated using PG&E’s average residential, commercial, industrial, and large agricultural rates in the second and third quarter of 2007. 11 These rebate numbers are calculated from the rebate amount offered for each energy efficiency measure and do not include expenditures for rebate processing and applications.
OCR for page 62
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop FIGURE C.7 Conservation supply curves with and without including non-energy productivity benefits, U.S. steel industry (Worrell et al., 2003). energy efficiency technologies. Such models can utilize the information available on the rebates provided in energy efficiency programs and their impact of increasing penetration levels. Further, bottom-up models can incorporate empirical information on other program costs to accurately model the cost, benefit, and impact of energy efficiency programs. Consideration of Non-Energy Benefits and Costs Accounting for “hidden benefits” requires that bottom-up models look beyond the energy markets and examine the cost considerations in light of their impact on other resource markets. Below we illustrate an example of the inclusion of such costs in the residential clothes washer, and iron and steel industry sectors. Residential Clothes Washers Residential clothes washers use a large amount of water whose costs in a California study were higher than the cost of energy used in the washer (Sathaye and Murtishaw, 2006). Most of the energy saving in efficient washers is due to avoided energy use from reduction of hot water consumption. The analysis offered a quantitative explanation of reasons that might prevent certain consumers from purchasing efficient washers and other products. Due to the differences in the cost of water, as well as electricity, the analysis segmented the market into low-, middle-, and high-income households. The inclusion of non-energy (water savings) benefits has a crucial impact on the cost-effectiveness of efficient washers. The initial weighted average CCE for all households when water savings were not counted was $0.136/kWh, compared to $0.049/kWh when they were included. Efficient washers that use electricity are significantly more cost-effective that those using natural gas-heated water due to the higher cost per unit of energy for electricity. For example, although the initial CCEs for wash-
OCR for page 63
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop TABLE C.4 Non-energy Benefits from Efficiency Improvements (Worrell et al., 2003) Waste Emissions Operation and Maintenance Use of waste fuels, heat, gas Reduced dust emissions Reduced need for engineering controls Reduced product waste Reduced CO, CO2, NOx, SOx emissions Lower cooling requirements Reduced waste water Increased facility reliability Reduced hazardous waste Reduced wear and tear on equipment/machinery Materials reduction Reductions in labor requirements Production Working Environment Other Increased product output/yields Reduced need for personal protective equipment Decreased liability Improved equipment performance Improved lighting Improved public image Shorter process cycle times Reduced noise levels Delaying or reducing capital expenditures Improved product quality/purity Improved temperature control Additional space Increased reliability in production Improved air quality Improved worker morale ers using gas and electric-heated water in high-income households are virtually the same ($0.016 and $0.017 respectively), the benefit-cost ratios differ significantly, due to the reduced benefit of saving a cheaper fuel (1.4 versus 2.5). The CCE varies by income groups due to differences in the costs of water and power among these groups, the differing shares of residency in units with washer/dryer hook-ups, and the fact that some lower income households may not be able to secure credit for the resource-efficient washer. Iron and Steel Sector Cost Curves Worrell et al. (2003) reported cost effective annual primary energy savings of 1.9 GJ/tonne for the U.S. iron and steel industry in 1994 which may be compared to the primary energy price of $2.14/GJ in 1994. Corresponding to the implementation of an array of 47 measures, the cost of supplied energy conservation are generally reduced when productivity benefits associated with labor and material cost savings are included in the calculation during the operation of an efficient iron and steel plant. Table C.4 shows the non-energy benefits that were included in their analysis. These included waste streams, emissions, O&M costs, production costs, working environment and other items. Inclusion of such productivity benefits increased the potential from cost-effective measures to 3.8 GJ/tonne at the same unit price of primary energy ($2.14/GJ in 1994). When including productivity benefits, the CCE ranking of technologies also changed dramatically. Inclusion of all resource benefits thus is crucial to understanding the full cost impacts of a technology. This may be particularly relevant to end-use energy efficiency technologies whose main goal often is not only saving energy but also providing some other form of service for the production of an industrial product. How different are the emissions reductions using the two different types of cost curves in a EC model? We tested this using the COBRA model12 (Wagner and Sathaye, 2006). The model tested results from U.S. cost curves for the iron and steel and cement sectors (Sathaye et al., 2010; Xu et al., 2010). Since the cost curves included negative cost options (no regrets options (NRO)), which are often not captured in efficiency scenarios of a top-down model, we tested the results with and without inclusion of NRO. The difference between the results with 12 COBRA is a bottom-up global energy model that uses linear optimization to design cost-minimal long-term scenarios on the basis of exogenous demand projections. COBRA can be operated at various levels of geographical, technological and temporal resolution. For the present study we use a version with ten-year time steps until 2100 and ten world regions (USA, EU, Rest of OECD90, REF, India, China, Rest of ASIA, Brazil, South Africa, Rest of ALM—where OECD90, REF, ASIA and ALM are the four world regions of the SRES study (Nakicenovic et al., 2000)). COBRA distinguishes the electricity producing sector from the rest of the sectors and further divides the latter into residential, transport, industry, and other. In order to address specific questions for individual industries, for the present study we have further divided the industry sector into iron and steel, fertilizer production, paper and pulp, cement production, aluminum production, and other industries.
OCR for page 64
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop FIGURE C.8 Effect of demand-side efficiency improvements on U.S. industry emissions. and without inclusion of the NRO options is not very large for the cement sector, but it is quite significant (13%) for the iron and steel sector. The inclusion of other non-energy savings benefits yields further reduction in energy use at the same CCE (Figure C.8), which translates into larger CO2 emissions reductions for the cement sector than for the iron and steel sector. The dotted line in Figure C.8 illustrates savings in cases where implementation is slower than what a techno-economic potential might imply. Historical Change in Energy Efficiency Cost Curves In this section, we provide examples of changes in cost curves for the industrial and residential sectors over time. While no analysis exists of extrapolating these changes over future time periods, it is evident that their inclusion will significantly impact the estimated cost of future emissions reductions if static cost curves are used in EC models. One of the popular approaches to take into account the future changes in the cost of technologies (including energy efficiency technologies) is using learning or experience curves which specify a relationship between cumulative installed capacity and the cost of a technology. This relationship is typically specified as a learning rate or a progress ratio which is the reduction in costs per doubling of cumulative installed capacity. Many studies have estimated learning rates for demand side technologies. A review of these studies by Weiss et al. (2010a) finds a widespread trend towards declining prices at an average learning rate of 18 ± 9%. Experience curve analysis has many limitations in explaining and predicting the decline in prices of clean energy technologies. Many factors such as learning, economies of scale, innovation, competition, and decline in prices of factors of production (due to reasons such as shifting of production to regions with lower labor costs) can potentially contribute to decline in prices of clean energy technologies. Learning curve approach is a black box approach at it lumps the effect of all these factors in only one explanatory variable, the cumulative installed capacity. Further, learning curve estimates have significant uncertainty depending on the time period considered for the analysis. The question of whether the learning curve analysis is a useful input for policy formulation thus needs to be examined given the uncertainty in learning curve estimates (Nemet, 2010). Only a few studies attempt to address these limitations of learning curve analysis by qualitatively examining or empirically estimating the effect of various factors contributing to the decline of prices of specific energy consumption in case of demand side technologies (see for example, Nemet, 2006; Ramírez, and Weiss et al., 2009; Worrell, 2006). Nemet (2006) is the only study that estimates the effect of factors such as learning, economies of scale, and R&D using empirical analysis and finds that they contributed 8%, 43%, and 35% respectively to the decline of PV costs in the United States during 1979-2000. More such studies need to be conducted in order to gain policy relevant insights into factors that contribute to decline in prices of clean energy technologies.
OCR for page 65
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop FIGURE C.9 U.S. cement sector example with other benefits included: changes in energy savings potential between 1994 and 2004 at 30% discount rate. One potential approach to better understand the factors contributing to improvement of energy efficiency (decline in specific energy consumption) is to evaluate the trends in the incremental cost of various component technologies that contribute to efficiency improvement options. This approach is more appropriate for examining technological change in end uses and sectors where multiple component technologies can contribute to efficiency improvement options. For example, improvements in the efficiency of compressors, motors, and drives contribute to the decline in the specific energy consumption of refrigerators. Economies of scale and learning in these technologies may be driven not just by cumulative sales of refrigerators but by various other products such as air conditioners, which use similar component technologies. Standard experience curve approach, which only examines the trend at a single appliance level cannot capture this effect. Analyzing trends in the incremental cost of various efficiency improvement options becomes even more important for end-uses that are more complex such as industries that have several component systems, with each of them having potential for efficiency improvement. The first step in such approach is to review trends in the incremental cost of various efficiency improvement options (which are due to improvements in different component technologies or systems) and examine potential factors contributing to their improved performance and declining costs. We present our analysis of the trends in incremental cost of various efficiency improvement options in select appliances and industries. Figure C.9 shows two cost curves, one that was developed for 1994 and another for 2004 for the U.S. cement sector (Sathaye et al., 2010). Each of the two curves shows the costs of conserved energy versus energy-savings potential for one time period. It shows that the energy-savings potential in 2004 was larger than that in 1994 when given the same cost of conserved energy (i.e., exhibited by a same Y-value in the chart). The energy-savings potential for instance at the cost of $40/GJ increased from 1.06 GJ/tonne to 1.24 GJ/tonne (by approximately 15%) over this decade. Such historical changes in the magnitude of savings potential may become useful for predicting future trends in energy climate modeling. Sathaye et al. (2010) also provide details of the cost of efficiency improvements in various component technologies or systems. Analysis of time trends in each of these enables the assessment of the degree to which various component technologies or systems have contributed to efficiency improvements in the U.S. cement sector. Changes in cost curves over time have been estimated for residential and commercial sectors as well. Figure C.10 shows such a set of curves for the U.S. commercial air conditioning equipment and heat pumps. In both
OCR for page 66
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop FIGURE C.10 Efficiency Improvements in U.S. Commercial Air Conditioning Equipment between 1999 and 2009. SOURCE: Sathaye, EMF 25 Presentation, 2009. cases, the curves demonstrate that the cost of installation (including the product cost) for achieving the same level of energy efficiency ratio (EER) was lower in 2009 compared to 1999 and ranged from 8% to 15%. Similar data exist for several other types of products that are used in both these sectors. The primary sources for these data are the Technical Support Documents (TSDs) published by the U.S. Department of Energy that develop estimates of incremental cost of efficiency improvements in appliances and equipment by assessing the cost and potential of efficiency improvements in their components and systems.13 These estimates are based on engineering and empirical analysis and form the basis of minimum energy performance standards (MEPS) set in the United States. By assessing the trends in the incremental cost and saving opportunities in various component technologies and systems, one can estimate the degree to which each of these have contributed to the changes in the cost and potential of efficiency improvement opportunities in appliances and equipment. This is the focus of our ongoing research focused on understanding various factors contributing to changes in the cost and potential of efficiency improvement options in key energy consuming appliances and equipment. Summary and Conclusions It is important to evaluate what role improvements in energy efficiency can play in GHG mitigation especially when it has been demonstrated that significant efficiency improvements that lead to GHG mitigation can be realized at a net societal benefit unlike many other GHG mitigation options. In this paper, we analyzed some key issues in evaluating energy efficiency as a GHG mitigation option and how its representation can be improved in energy-climate models. We report on an empirical analysis of the cost imposed by various barriers to the adoption of energy efficient end-uses in California as an example of how these barriers impose large enough costs that might prevent consumers from adopting technologies that appear highly cost effective from a societal perspective. We also demonstrate an approach to quantifying the cost imposed by the PA problem on a national scale, and show how the quantification 13 See for example TSD for residential air conditioners and heat pumps at http://www1.eere.energy.gov/buildings/appliance_standards/residential/central_ac_hp.html.
OCR for page 67
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop could be included in NEMS and MARKAL type models. Further, improved understanding of the costs imposed by these barriers would lead to a better understanding of the benefits of policies and programs that address these barriers. It has been demonstrated that policies and programs that deal with various barriers lead to increased adoption of energy efficiency. These policies and programs also incur some costs such as administrative, marketing, installation, and rebates. From a societal perspective, the desirability of these programs depends on whether the benefits of these policies and programs outweigh their costs and empirical evaluation of the outcomes of these policies and programs is warranted. Further, it is important to analyze the impact of policies and programs in terms of their performance in accelerating the adoption of energy efficiency measures. We report the results of our empirical analysis of utility energy efficiency programs in California, which are among the largest and most comprehensive energy efficiency programs in the world with budgets of close to one billion dollars per year. We find that program implementation and financial incentive costs are relatively minor compared to their benefits. We also find that a majority of the costs of the utility programs are financial incentive costs, which reduce the payback period of consumers on their efficiency improvement investments to less than a year in most of the instances. Financial incentives are a transfer and not an actual cost from a societal perspective. We evaluate various ways using the results of such type of analysis to improve the representation of energy efficiency as a GHG mitigation option in energy climate models and find that both bottom-up and top down energy-climate models can incorporate information on policy and program costs to model their impact. Many times energy efficiency improvements result in reduction not only of energy costs but also other costs such as maintenance and materials. Estimation of these co-benefits is required for accurately estimating the cost and potential of EE improvements. We present the results of our empirical analysis for the residential clothes washers and iron and steel sector to show the consideration of co-benefits influences the cost and potential of EE improvements. For $40/GJ, we find that the cost effective potential for iron and steel sector increases by 15% when some of the co-benefits are considered. Technological change influences the cost and potential of various GHG mitigation options. Learning or experience curve approach, one of the popular approaches to model technological change, is a black box approach as it lumps the effect of various factors such as learning, economies of scale, innovation, competition, and decline in prices of factors of production into one explanatory variable, the cumulative installed capacity. One potential approach to better understand the factors contributing to improvement of energy efficiency (decline in specific energy consumption) is to evaluate the trends in the incremental cost and efficiency improvement opportunities in component technologies and systems. This approach is more appropriate for examining technological change in end uses and sectors where multiple component technologies can contribute to efficiency improvement options. We present our analysis of the trends in incremental cost of various efficiency improvement opportunities in component technologies and systems for some key energy consuming industries and appliances in the form of cost curves and find that efficiency improvement opportunities have increased and costs of the same have decreased over time. We conclude that analysis of energy efficiency as a mitigation option, undertaken using energy climate models or any other methods or tools, needs to consider the issues identified in this paper to improve its assessment. Empirical analyses which forms the basis of improved understanding of these issues needs to be expanded significantly, potentially drawing on some of the analyses presented in this paper. Acknowledgments Authors would like to acknowledge the contribution of Skip Laitner, ACEEE for providing information and data on rebound effect. This paper was prepared with the support of the Office of Science, Integrated Assessment Models, Department of Energy under contract No. DE-AC02-05CH11231. References Davis, L.W. (2010) “Evaluating the Slow Adoption of Energy Efficient Investments: Are Renters Less Likely to Have Energy Efficient Appliances?” Energy Institute at Haas, WP 205.
OCR for page 68
Modeling the Economics of Greenhouse Gas Mitigation: Summary of a Workshop de la Rue du Can, S. and J. Sathaye (2008) “Quantifying the Effect of the Principal-Agent Problem on U.S. Residential Energy Use: Refrigerators, Space Heating and Air Conditioners.” LBNL. Ehrhardt-Martinez, Karen and John A. “Skip” Laitner. 2010. “Rebound, Technology and People: Mitigating the Rebound Effect with Energy-Resource Management and People-Centered Initiatives.” Proceedings of the 2010 ACEEE Summer Study on Energy Efficiency in Buildings, Washington, DC: American Council for an Energy-Efficient Economy. Eto, J., Prahl, R., and Schlegel, J., 1997. A scoping study on energy-efficiency market transformation by California utility DSM programs. Lawrence Berkeley National Laboratory, LBL-39058. Geller, Howard and Sophie Attali. 2005. “The Experience with Energy Efficiency Policies and Programmes in IEA Countries: Learning from the Critics.”International Energy Agency: Paris, France. Golove, W. H. and Eto, J., 1996. Market barriers to energy efficiency: a critical reappraisal of the rationale for public policies to promote energy efficiency. Lawrence Berkeley National Laboratory, LBL-38059. Hanson, Donald A. and John A. “Skip” Laitner. 2004. “An Integrated Analysis of Policies that Increase Investments in Advanced Energy-Efficient/Low-Carbon Technologies.” Energy Economics 26:739-755. International Energy Agency (IEA) 2007. Mind the Gap. Jaffe, A. and Stavins, R., 1994. ‘The Energy-Efficiency Gap: What Does it Mean?” Energy Policy. 22(10): 804-810. Laitner J. and D. Hanson. 2006. “Modeling Detailed Energy-Efficiency Technologies and Technology Policies within a CGE Framework.” Energy Journal Special Issue on Hybrid Modeling of Energy-Environment Policies: Reconciling Bottom-Up and Top-Down. Laitner, John A. “Skip,” Donald A. Hanson, Irving Mintzer, and Amber J. Leonard. 2006. “Adapting in Uncertain Times: A Scenario Analysis of U.S. Energy and Technology Futures.” Energy Studies Review 14(1): 120-135. Laitner, John A. “Skip,” Stephen J. DeCanio, and Irene Peters. 2000. “Incorporating Behavioral, Social, and Organizational Phenomena in the Assessment of Climate Change Mitigation Options.” Society, Behavior, and Climate Change Mitigation, edited by Eberhard Jochem, Jayant Sathaye, and Daniel Bouille. Dordrecht, The Netherlands: Kluwer Academic Press. Manne, A. and Barreto. L., 2002. Learn-by-Doing and Carbon Dioxide Abatement, http://www.stanford.edu/group/MERGE/lbd.pdf, visited 14 June 2002. Meier, A., 1982. “Supply Curves of Conserved Energy.” Lawrence Berkeley National Laboratory, LBL-14686. Murtishaw, S. and J. Sathaye, “Quantifying the Effect of the Principal-Agent Problem on U.S. Residential Energy Use,” LBNL-59773, August 2006. Nakicenovic et al. (Eds), 2000. IPCC Special Report on Emission Scenarios (SRES). IPCC/UNEP/WMO, Cambridge University Press, Cambridge, United Kingdom. Nemet, G.F. (2006). “Beyond the learning curve: factors influencing cost reductions in photovoltaics.” Energy Policy 34(17): 3218-3232. Ramírez, C.A., Worrell, E. (2006): Feeding fossil fuels to the soil—An analysis of energy embedded and technological learning in the fertilizer industry. Resources, Conservation and Recycling 46, pp. 75-93. Reddy, A.K.N., 1991. Barriers to Improvements in Energy Efficiency. Energy Policy 19, 953- 961. Roland-Holst, D (2006) Economic Assessment of some California Greenhouse Gas Control Policies: Applications of the BEAR Model. In Managing Greenhouse Gas Emissions in California, University of California Berkeley, available at http://calclimate.berkeley.edu/managing_GHGs_in_CA.html. Ross, M. (2005) Documentation of the Applied Dynamic Analysis of the Global Economy (ADAGE) Research Triangle Institute Working Paper 05_01 November 2005. Sathaye, J., Bouille et al., 2001. Barriers, opportunities, and market potential of technologies and practices. In Climate Change 2001: Mitigation, (Eds., Metz, B., Davidson, O., Swart, R., and Pan, J.) Cambridge University Press for the Intergovernmental Panel on Climate Change. Sathaye, J. and Murtishaw, S., 2004. Market Failures, Consumer Preferences, and Transaction Costs in Energy Efficiency Purchase Decisions. Lawrence Berkeley National Laboratory for the California Energy Commission, PIER Energy-Related Environmental Research. CEC-500-2005-020. Sathaye, J., T. Xu, C. Galitsky. 2010. Bottom-up Representation of Industrial Energy Efficiency Technologies in Integrated Assessment Models for the Cement Sector. LBNL. Sorrell, S. 2009. “The Evidence for Direct Rebound Effects.” In Energy Efficiency and Sustainable Consumption: The Rebound Effect. Horace Herring and Steve Sorrell (eds.). Palgrave Macmillan (St. Martin’s Press): New York, NY. Wagner F. and Sathaye J. (2006). Sharing the burden of carbon mitigation: A long-term global perspective of the energy sector. Energy Policy, Volume 34, Issue 15, October, pp. 2217-2231. LBNL-52911. Weiss M., M. Junginger, M. Patel, K. Blok (2010) “A review of experience curve analyses for energy demand technologies.” Technological Forecasting and Social Change, Volume 77, Issue 3, March 2010, Pages 411-414. Weiss, M., Dittmar, L., Junginger, H.M., Patel, M.K., Blok, K. (2009b): Market diffusion, technological learning, and cost-benefit dynamics of condensing gas boilers in the Netherlands. Energy Policy 37, pp. 2962-2976. Worrell, E., Laitner, J., Ruth, M., and Finman, H., 2003. Productivity benefits of industrial energy efficiency measures. Energy 28:1081-1098. Xu, T., J. Sathaye, C. Galitsky. 2010. Bottom-up Representation of Industrial Energy Efficiency Technologies in Integrated Assessment Models for the Iron and Steel Sector. LBNL.