Diffusion of Energy-Efficient Technologies

Diffusion of Energy-Efficient Technologies

MARKETS/TECHNOLOGY INNOVATION/ADOPTION/ DIFFUSION Contents Diffusion of Energy-Efficient Technologies Energy-Efficiency Gap Impacts of Innovation: Le...

335KB Sizes 4 Downloads 104 Views

MARKETS/TECHNOLOGY INNOVATION/ADOPTION/ DIFFUSION

Contents Diffusion of Energy-Efficient Technologies Energy-Efficiency Gap Impacts of Innovation: Lessons from the Empirical Evidence Modeling Technological Change in Economic Models of Climate Change Policy Incentives for Energy and Environmental Technological Innovation: Lessons from the Empirical Evidence Technological Change and Climate Change Policy Technological Change and the Marginal Cost of Abatement Technological Lock-In Technology and Environmental Policy

Diffusion of Energy-Efficient Technologies T Fleiter and P Plo¨tz, Fraunhofer Institute for Systems and Innovation Research, Karlsruhe, Germany ã 2013 Elsevier Inc. All rights reserved.

Glossary Barriers to diffusion Mechanisms that inhibit adoption of a technology that appears to be both energy-efficient and economically efficient. Diffusion The gradual adoption of an innovation by individuals, firms, or other organizations over time. Economies of scale A reduction of per-unit costs of an industrial good when the scale of production of the good increases. Energy efficiency The ratio between energy input and an output of performance, service, goods, or energy. Energy-efficient technology A technology that delivers an energy service or good with less energy input compared to a reference technology.

Introduction Why has it taken 30 years from market introduction for compact fluorescent lamps (CFLs) to reach saturation levels, at least in some countries, despite their clear economic and environmental benefits? Questions such as this form the focal point of this article and are addressed by studying the diffusion of energy-efficient technologies (EETs). Saving energy and improving energy efficiency are key strategies in the development of a more sustainable global energy system. Increased efficiency is seen as a major option for lowering greenhouse gas emissions in the energy sector, as it reduces the demand for fossil fuels. It also has the potential to

Encyclopedia of Energy, Natural Resource and Environmental Economics

Learning by doing Improvement of a process or product resulting from the manufacturing process. Learning by using Improvements of a process or product resulting from the use of the technology or product by end users. Network effects An increase in the utility of a product to a user when the number of users of the same product increases. Technology lock-in A situation in which a switch to a new technology (paradigm) that would be potentially superior in the long term is precluded, because the new technology is inferior in the short term. Technology lock-in arises when at least two technologies experience increasing returns to adoption.

significantly reduce the dependency on energy imports, address the scarcity of energy resources, and finally, contribute to improving the competitiveness and productivity of firms. Given these benefits, energy efficiency is at the top of the policy agenda of numerous governments worldwide and is also receiving a lot of attention from researchers and analysts. The International Energy Agency, for example, predicts that global greenhouse gas emissions could be significantly reduced simply by using the currently best available technology and that additional potential reductions are available because of new, emerging technologies. Thus, the spread or diffusion of EETs through society is a highly relevant research field. Even the most revolutionary

http://dx.doi.org/10.1016/B978-0-12-375067-9.00059-0

63

Markets/Technology Innovation/Adoption/Diffusion | Diffusion of Energy-Efficient Technologies

innovations will have no effect on energy demand if they do not find users. It is also a very complex field, as numerous and often interrelated factors affect the diffusion of EETs. For policy makers, however, it is crucial to understand the determinants of diffusion in order to effectively steer or accelerate it where it is too slow from a social-optimum perspective. Thus, the study of the diffusion of EETs analyzes its determinants and aims to derive patterns that may help to predict the diffusion of new EETs ex ante and provide useful suggestions for policy design. This article gives a short overview of the diffusion of EETs and its determinants. We begin by summarizing the theory behind the diffusion of innovations and then discuss the specific features relating to EET diffusion. In the section ‘Determinants of the Adoption of EET,’ we focus on the adoption decision and present empirical findings on EET adoption, before we discuss the role of policies aiming to accelerate the diffusion of EET. In the section ‘From Adoption to Diffusion: The Time Dimension, Feedback Loops and Diffusion Dynamics,’ we extend the view from individual adoption decisions to the diffusion perspective. The section ‘Technology Case Studies’ presents selected case studies to illustrate the determinants and dynamics of EET diffusion and the potential role of policies.

Diffusion Theory Diffusion of Technologies The theory of the diffusion of innovations also forms the basis for analyzing the diffusion of EETs. Rogers’ ‘diffusion of innovation’ provides a heuristic framework for analyzing the diffusion of innovations and defines an innovation as ‘an idea, practice or object that is perceived as new by an individual or other unit of adoption.’ He continues by emphasizing that the ‘newness’ of an innovation depends only on the perception of the potential adopter. In this sense, a technology that uses energy efficiently can be considered an innovation and the diffusion theory of innovations can be applied to the diffusion of EETs. Research on the diffusion of innovations started with a number of studies analyzing the diffusion of hybrid corn across US farms. These studies observed that the cumulative number of farms using hybrid corn follows an s-shaped curve over time in each state of the United States. The s-shaped or sigmoid diffusion curve implies that, during the early diffusion stage of an innovation, the number of users is only a relatively small proportion of all potential adopters. The adoption rate, measured as the share of new users in a given time interval compared to all potential adopters, increases continuously until it reaches a maximum at the point of inflection (of the cumulative number of adopters). Beyond this point, it continuously decreases and the diffusion curve slowly saturates toward an asymptote, given by the total number of potential adopters (see Figure 1). Although varying mathematical descriptions have been proposed to describe this pattern, including symmetrical and asymmetrical diffusion curves, empirical evidence for the s-shaped development has cumulated over the past decades, and it has now become widely accepted. When discussing the diffusion of innovations, one usually refers to the decision of agents, or potential adopters, to

Cumulative share of adopters

64

100%

Rate of adoption Cumulative share of adopters

75%

50%

25%

0%

Time

Figure 1 s-Shaped diffusion curve and rate of adoption over time.

acquire or use an innovation as ‘adoption.’ The diffusion of an innovation is the result of many adoption decisions over time and the cumulative share of adopters represents the diffusion curve, which is often s-shaped. There are numerous explanations for the s-shaped curve. A prominent one builds on information flows among potential adopters, which result in innovations spreading like an epidemic. In the early stage of the diffusion process, only a few users can spread information about the superiority of the new technology, but increasing numbers of users can access the information until after the point of inflection it becomes increasingly unlikely that users are in contact with remaining potential adopters, as their number decreases, and the diffusion process decelerates. Thus, the adoption rate would be proportional to the number of adopters and the number of remaining potential adopters. This is simply the logistic differential equation, leading to a logistic sigmoid function for the cumulative number of adopters. While this epidemic model accounts for the empirically observed s-shaped diffusion curve, it excludes other factors that certainly affect diffusion as well, such as heterogeneity among potential adopters, postinnovation improvements of a technology, and changing numbers of potential adopters. Furthermore, the epidemic approach is often criticized on the basis that it does not include a theoretical analysis of the decision to adopt a technology and consequently only allows for very restrictive policy conclusions. Alternative explanations for the shape of the diffusion curve have also been put forward. A very common one is based on the assumption of heterogeneity among potential adopters. Potential adopters differ in their characteristics, resulting in different potential benefits from adopting a technology, and technology adoption follows rational decisions based on the benefits experienced. For instance, adopters might be subject to different electricity tariffs, which result in a varying profitability of electricity-saving techniques. Assuming that profitable technologies are adopted, they spread through the market according to changes in the cost structure of firms and changes in the costs of technologies over time. Following the above example, rising electricity tariffs would result in higher numbers of adopters. Changes in technology costs may happen exogenously but may also be induced by learning effects and economies of scale. Such effects also result from increased technology diffusion, completing the feedback cycle of increasing returns to adoption. This explanation is referred to as the probit model of technology diffusion and is also able to reproduce the empirically observed s-shaped diffusion curve.

Cumulative share of adopters

Markets/Technology Innovation/Adoption/Diffusion | Diffusion of Energy-Efficient Technologies

100%

65

Diffusion of EETs

75%

50%

25%

0% Time

Figure 2 Diffusion curves with different speeds of diffusion.

Different disciplines have conducted empirical studies of a huge range of different innovations, adopters, and regions. The majority of studies show that diffusion paths and dynamics vary widely across technologies. However, diffusion is typically slow and a diffusion cycle from market introduction to saturation often takes decades, particularly for long-lived industrial technologies. Many technologies have taken longer than 30 years from market entry to reach final saturation. In this context, the speed of the diffusion process can be measured using the width of the rate-of-adoption curve (see Figure 1). Diffusion trajectories resulting from varying diffusion speed are shown in Figure 2. The speed of diffusion and its saturation level vary widely depending on a huge number of factors that can be divided into four classes: the characteristics of the innovation, the characteristics of the adopter, information channels, and contextual factors. While profitability is probably the most researched innovation characteristic and is generally viewed as increasing the speed of diffusion, other characteristics also have a significant impact. For example, it has been shown that the complexity of an innovation is negatively correlated with the speed of diffusion. When looking at the impact of adopter characteristics on diffusion, firm size is the most commonly researched parameter and is generally expected to have a positive impact on the adoption rate. The implication here is that larger firms are more likely to adopt innovations. Contextual factors can also significantly shape the diffusion curve, such as the regulatory framework and diffusion-oriented policies. They can stimulate or slow down diffusion by providing information, for example, or financial support to potential adopters. Technology diffusion is a complex issue due to the frequent interaction of technologies, that is, the fact that many technologies are related rather than independent. The possible influence that technologies have on each other ranges from mutual support, for example, symbiotic diffusion, to strict competition, where the technologies exclude each other and only one will eventually dominate the whole market. The substitution of one technology by another is a simple and common example of interaction. Thus, the diffusion of innovations is a complex process and depends on a variety of different factors that make it difficult to derive generalizations. However, some conclusions can still be drawn. Diffusion is a gradual process that can take decades from market entry to saturation. It often follows an s-shaped curve, which can, in principle, be explained by elements of information flow and also by heterogeneity among potential adopters.

Within the context of the diffusion of innovations, EETs can be regarded as a particular type of innovation. Potential users of EET are typically not interested in energy efficiency itself, but rather in an energy service or good. Examples of energy services or goods are lighting, motion, or refrigeration, for example, for cold beer. In these terms, different technologies can fulfill this need, using different energy inputs. Accordingly, energy efficiency is the ratio between energy input and an output of performance, service, goods, or energy. An EET is then defined as a technology that delivers an energy service or good with less energy input compared to a reference technology, no matter what the main reasons were for adopting the technology. Thus, even if some equipment is adopted not only in order to improve energy efficiency but also for another purpose, it is still regarded as an EET as long as it improves energy efficiency. Energy efficiency is often distinguished from energy saving by noting that the former does not necessarily imply a reduction of total energy demand, for example, due to rebound effects in the form of increased use of the more efficient technology. Here, we generally use the term EET. We furthermore disregard the fact that changed behavior can also be considered an innovation and can spread through society, that is, falls under the diffusion of innovations, and do not discuss it here, since our focus is on technologies. Furthermore, EETs are always defined in comparison to a reference or competing conventional technology. A fluorescent lamp is an EET only if compared to an incandescent light bulb, but not if compared to a light-emitting diode (LED) lamp. This example indicates that the perception of EETs is dynamic over time, as increasingly efficient technologies emerge. A central, much debated aspect in the diffusion of EETs is the seeming discrepancy between the observed share of users who actually adopt an EET and the share for which it is expected to be profitable. This observation of the only partial diffusion of apparently cost-effective EETs is referred to as the ‘energyefficiency gap.’ To illustrate this: Why did customers not purchase compact fluorescent light bulbs despite their clear cost advantages? In the context of society, the energy-efficiency gap refers to the discrepancy between the observed level of diffusion and the economically optimal state of diffusion. Clearly, while different social optima can be discussed, the general observation that even cost-effective EETs are often not adopted has been frequently observed and is only rarely questioned. The reasons for the nonadoption of cost-effective EETs are manifold and summarized under the heading of barriers to energy efficiency. Barriers have been researched for more than three decades, drawing on fields such as orthodox economics, transaction-cost economics, and behavioral economics. Some researchers argue that the existence of barriers is not surprising considering the fact that all (economically superior) innovations diffuse only gradually through the capital stock. In this broader sense, barriers to energy efficiency are regarded as factors slowing down the diffusion speed of EETs. Thus, one can distinguish different speeds of diffusion depending on different adoption behaviors by the underlying agents. Typically, even cost-effective EETs are often not adopted by potential users, resulting in slow market diffusion. Another diffusion rate results if all agents purchase EETs that have acceptable payback times. Diffusion is slightly faster

66

Markets/Technology Innovation/Adoption/Diffusion | Diffusion of Energy-Efficient Technologies

if users decide to adopt cost-effective technologies. Very rapid diffusion would be achieved in the counterfactual case of all actors adopting the most EETs available, irrespective of their costs. These distinctions not only imply a varying speed of diffusion but probably also result in a different level of adoption in the saturation phase. Particularly, in the first case, many EETs never diffuse through the entire market. For EETs that require replacing older equipment, the slowest diffusion is determined by capital stock turnover, that is, agents simply replacing their old equipment after the end of its lifespan by new commercially available equipment. Accelerating the diffusion through the stock would require replacing the old equipment before the regular replacement cycle and thus, reducing the lifespan of the old equipment. Although these distinctions of diffusion speeds are mainly theoretical, they are often used in the literature to estimate future energy-saving potentials of EETs. Since the diffusion of EETs is the result of many adoption decisions, the potential adopters or agents are also the subject of research. Here, many different adopters with different characteristics have to be considered. EETs can be adopted by such diverse entities as private customers, households, firms, parts of firms, or organizations, in general. These differ widely in their decision-making processes, aims, agendas, preferences, number of members, access to capital and information, and other factors that may influence adoption decisions. For example, EETs are usually not related to the core business of firms and have only a slight effect on firms’ competitiveness, particularly in firms with low energy costs. Firms with limited resources allocate these to the most important investment projects first. Thus, the priority that firms place on investing in EETs is often low, and many firms do not even actively search for EETs. For consumers, too, energy efficiency is usually not the central motivation for buying a product. Often, energy efficiency is not even considered as a criterion for the adoption decision. To conclude, the diffusion of EETs is a complex process and depends on a number of factors that can vary by technology and adopting agent, and also depend on the overall framework and other contextual factors. These factors are discussed in more detail in the following section.

Determinants of the Adoption of EETs Barriers to and Drivers of Adoption Throughout this work, we define barriers as “mechanisms that inhibit a decision or behavior that appears to be both energyefficient and economically efficient.” In other words, barriers are obstacles to the adoption of cost-effective EETs. Costeffectiveness is assessed from the perspective of the adopter and can vary significantly. Other, broader definitions of barriers define cost-effectiveness from the perspective of society and, for example, include external environmental costs. Such a broad scope typically results in a higher number of potential barriers, including macroeconomic factors such as artificially low energy prices and similar things. However, such barriers go beyond the decision to adopt EETs by individual users and are therefore not included here. While varying classifications of barriers are used in the literature, the following classes are widely applied: risk,

imperfect information, hidden costs, access to capital, split incentives, and bounded rationality. We briefly discuss each category in the following text. The importance of imperfect information as a barrier has often been shown empirically. This term covers knowledge about the availability of an EET, and also about its characteristics such as costs and saving potentials, as well as the actual energy consumption of current equipment. Several studies have shown that even firms are not aware of many EETs available on the market. Transaction costs for the search and information-gathering process are regarded as one reason for imperfect information. Transaction costs for the implementation of EETs, however, have only rarely been quantified. The few studies available indicate that transaction costs are only weakly correlated with the price of the equipment and, thus, cannot be measured as a fixed percentage value. The share of transaction costs compared to the total initial investment costs of an EET typically falls with increasing investment costs. Hidden costs prevent firms from undertaking energyefficiency projects, although the costs are generally not quantified by firms or consumers and are difficult to determine by outside observers. They may result, for example, from poorquality equipment, which leads to production disruptions or from replacing staff required to operate the new machinery. On the other hand, EETs may also be accompanied by hidden benefits, the so-called cobenefits of energy efficiency, which are often not accounted for when making decisions. Limited access to capital is also frequently found to be an important barrier to investments in energy efficiency for both firms and consumers. This concerns access to external capital, and also to the use of internal capital and priority setting among alternative investment projects. Firms often have sufficient access to external capital, but internal budgeting rules allocate this to other investment projects, as energy-efficiency investments have lower priority. Barriers related to risk and uncertainty range from uncertainty about future energy prices or technology development to the risk of production interruptions and impacts on product quality. Uncertainty related to future energy prices and technology development has received a lot of attention in the literature. Particularly, in the case of irreversible, expensive, and long-lived equipment, the potential adopter might want to delay in case a superior technology becomes available or energy prices fall. This investment behavior is referred to in the literature as the option value of waiting and can be regarded as one element of a rational adoption decision process. Another type of risk is related to the performance of the EET. Particularly when such technologies are applied to core industrial processes, such as in the pulp-and-paper or foundry industries, firms are reluctant to invest, because they fear production interruptions or dips in product quality. Split incentives can hamper the adoption of EETs at very different phases in the diffusion process. The most prominent example of split incentives is the investor/user dilemma, where the landlord is responsible for building refurbishment, but the tenant receives the benefit in the form of lower energy bills. In this case, the landlord would not receive a financial reward for the expensive building refurbishment. On the other hand, the tenant, who would receive the benefit and thus have an incentive to refurbish the building, does not own the building

Markets/Technology Innovation/Adoption/Diffusion | Diffusion of Energy-Efficient Technologies

and cannot refurbish it. This phenomenon is further exacerbated if information about the energy performance of buildings is incomplete or unreliable, because the landlord of an energyefficient building cannot (easily) prove its efficiency and the tenants cannot take the energy consumption of the building into account when deciding to rent. Such split incentives have also been observed in other areas, where they may be less obvious, such as in the electric motor market, where split incentives have been observed between different market actors, and also between different units within a single firm. When original-equipment manufacturers (OEMs) integrate a motor into a pump, fan, or similar equipment, they do not demand energy-efficient motors, because they mostly compete on the basis of price and reliability when selling their products. As they do not pay for the motor’s electricity bill, they have no interest in integrating energy-efficient motors into their products. The end-consumers of the equipment, however, either do not have sufficient information about the energy efficiency of the equipment or also focus mainly on the price of the equipment and neglect life cycle costs. For both examples, policies such as labeling and minimum energy performance standards (MEPS) have attempted to tackle split incentives and information gaps. Bounded rationality can also be regarded as a type of barrier, although this is not specific to energy efficiency. While bounded rationality for consumers may not come as a huge surprise, it is also often observed that even firms do not follow the rationale of cost minimization. For several decades, certain strands of economics have argued that observed business decision making conforms better with the assumptions of bounded rationality than with the dominant economic theory of rational choice. Decision makers are said to base their decisions on rules of thumb or heuristics. Such rules of thumb are observed in firms, for instance, when they invest in a new electric motor to replace a broken one in the production line. In the case of a broken motor, smaller firms especially do not have the capacity to compare alternative motor types. Their focus is on getting a new motor as quickly as possible, because even short interruptions cost the firm more in lost production than a new motor. As a consequence, they tend to replace a broken motor with a new motor of the same brand and type. But even larger firms, which generally have a stock of replacement motors, usually decide based on purchasing price rather than life cycle costs. Besides barriers, drivers of the diffusion of EETs are also observed and discussed in the literature, although they receive far less attention. Examples of drivers include the environmental awareness of consumers, the presence of motivated personnel, or the green image of EET. Important drivers are the so-called cobenefits of EETs. Such benefits comprise all the positive effects accompanying the adoption of EETs beyond energy savings. Examples include improved indoor air quality after insulating a building or reduced noise emissions due to the installation of triple-glazed windows. While these comfortrelated cobenefits are difficult to quantify, energy efficiency in industrial processes often entails considerable (financial) cobenefits such as waste reduction, reduced material consumption, lower maintenance needs, lower emissions, improved reliability, better product quality, and higher productivity. For some technologies, cobenefits may even exceed the cost savings because of improved energy efficiency.

67

To conclude, the barriers to energy efficiency are manifold and complex, and they overlap. In real life, none of the above six classes of barriers occurs in isolation. On the contrary, barriers are related to one another and also act as catalysts for each other, for example, in the case of information deficits in the presence of split incentives. Furthermore, barriers vary by the type of adopter and type of EET.

Adopter Characteristics Two general classes of adopters have to be distinguished when analyzing the diffusion of EETs: consumers and organizations (mostly firms). While the former are expected to adopt EETs depending on their preferences, the latter are expected to be more influenced by rational arguments, based on profitability as the main decision criterion. Research on the effect of firm characteristics on the adoption of EET often emphasizes the role of firm size. It is generally accepted that larger firms tend to have higher adoption rates than smaller firms. If, however, intrafirm diffusion is included in this comparison, the picture becomes more complex, as this often takes longer in larger firms. Firms also differ in terms of energy intensity, measured as the energy-cost share of the firms’ turnover. Energy-intensive firms typically focus more on energy efficiency and regard it as an important factor for their competitiveness. On the other hand, energy-intensive firms also typically have access to lower energy tariffs, rendering many EETs less profitable and thus reducing the incentive to adopt them. Furthermore, firms also differ regarding the extent to which energy management is integrated into their official routines. If it is officially integrated, the adoption of EETs and searching for new EETs are much more systematic and are given higher priority within the firm, resulting in higher adoption rates. These and many other factors potentially influencing the adoption decision have been analyzed, mainly in terms of case studies and econometric analyses. Consumers are far less heterogeneous than firms, that is, they fall into fewer subcases. The electricity tariffs they pay are more or less the same, and energy intensity also varies less across households. However, they do differ significantly in terms of their beliefs and preferences with regard to, for example, environmental awareness. Consumers who are more aware of the environmental consequences of energy consumption will put greater effort into searching for energy-efficient equipment and generally accord the aspect of energy efficiency greater weight when deciding to buy new equipment. Depending on the user’s preferences and the EET under consideration, the profitability of the investment can also play a role. Furthermore, the perceived image of a technology, the income and financial lucidity of a potential adopter, and his or her age, education, and social status can all influence the potential adopter’s decision. To account for these factors and the heterogeneity of adopters, different adopter classes are distinguished, such as innovators, early adopters, early majority, late majority, and laggards. The different groups can then be specifically addressed to accelerate the diffusion process.

Technology Characteristics Although EETs are sometimes treated as a rather homogenous group of technologies, they in fact constitute such a broad group of technologies that it is extremely challenging to draw

68

Markets/Technology Innovation/Adoption/Diffusion | Diffusion of Energy-Efficient Technologies

general conclusions about their diffusion patterns. For example, fluorescent lamps, building insulation, energy-efficient electric motors, new production plants for steel or aluminum, and the recovery of waste heat from various sources all fall under the term ‘EET.’ Obviously, however, the diffusion of these technologies is driven by completely different dynamics and factors. Thus, to better understand the diffusion of EETs, it is obligatory to consider their specific characteristics. Again, the theory of the diffusion of innovations provides a good starting point for analysis and suggests the following widely used innovation characteristics: the relative advantage, the complexity, the compatibility (with the existing system), the trialability, and the observability of an innovation as perceived by the potential adopters. As EETs mostly compete with other (less efficient) technologies, these characteristics need to be interpreted in comparison to a reference technology. Below, we discuss more concrete characteristics of EET, which we relate to these five broader groups whenever possible. The characteristics discussed are chosen on the basis that they have a significant impact on the adoption rate. The relative advantage of an EET consists of various elements, the most important of which is the profitability of the EET. It is often assessed using capital appraisal methods such as internal rate of return or more simply the payback period. Empirical evidence shows that higher profitability generally increases adoption rates. Beyond the classical investment appraisal, the relative advantage of an EET is also determined by its cobenefits and possible transaction costs. The initial investment sum required is another important characteristic of EET, which also is closely related to the discussion of barriers. The total investment in an EET may require the adopter to acquire external funds, which is often difficult and results in lower adoption rates. It also determines the investment decision rules applied by the adopter. While for small investments rather ad hoc rules of thumb are applied, larger investments representing a substantial share of the adopter’s budget require more sophisticated investment appraisal rules and typically shift the focus onto the EET’s profitability. Furthermore, most EETs replace equipment in the current capital stock, such as washing machines, electric motors, or lamps. In this case, adoption is either bound to the rate of capital turnover, so that new equipment only enters the capital stock when old equipment is being decommissioned, or it requires the premature replacement of existing equipment. In the first case, the appliance’s lifespan determines the speed of diffusion. In the second case, diffusion may be faster and is also more expensive. There are some EETs, however, such as insulating steam pipes, that do not require replacing existing equipment and thus have the potential to diffuse faster. The many EETs implemented in industrial firms are particularly varied. The more complex an EET, the more knowledge is required for its implementation. This might restrain firms from adopting if they do not have sufficient internal knowledge and are reluctant to involve external experts. The implementation risk varies considerably across technologies. Particularly, technologies used in the core production processes of firms often entail significant risks concerning product quality losses or production interruptions. On the other hand, EETs in ancillary processes such as building heating or compressed-air generation are perceived as less risky by firms and adoption rates here are often higher.

Furthermore, it is helpful to distinguish between EETs that imply the replacement of single components and those that require optimization of the overall system. While the latter also often entail a higher relative potential for energy-efficiency improvement, they are also more difficult to address via policies, because they are always site specific. In general, system optimization is expected to be more complex than replacing single components. To summarize, the range of EETs that can be adopted is as broad as their potential adopters. The EETs’ main characteristics can be partially summarized by their degree of complexity, relative advantage, compatibility, trialability, and observability, but account also has to be taken of the diversity of their adopters, which range from private individuals to households, firms, or larger organizations. Both aspects have to be considered to understand adoption processes, as it is the adopters’ perception of the technology that determines the decision to adopt.

The Regulatory Framework and Policies As a result of the huge potential that energy efficiency holds to address global challenges that are very high on the agenda of most governments, numerous energy-efficiency programs have been developed in the majority of countries. However, other political goals can also be strong reasons for fostering the diffusion of EETs. Governments may want their economy to play a leading role in the development of certain technologies and may stimulate the diffusion of these technologies in order to strengthen that role. A different example is provided by the United States, which lists reduction of dependence on foreign oil as one of its reasons to promote the use of alternative vehicle technologies. Early programs aimed at speeding up the diffusion of EETs mainly consisted of classical price-based policies, such as energy taxes aiming to increase the profitability of EETs. But policies soon became more diversified. Parallel to the observation that even cost-effective EETs often diffuse only slowly, various policies emerged, which specifically aimed at overcoming barriers responsible for the slow diffusion. Examples of such policies are soft loans, information campaigns and labeling programs, audit programs, energy management schemes, contracting support, MEPS, or voluntary agreements with industry. All these policies aim to overcome different barriers to adoption and address different adopter categories and many are even technology specific. Often, governments apply a policy mix that exploits synergies among policies and assures that policies match the adopter characteristics. Instead of discussing each of these policies in detail, we describe the role and impact of selected policies in technology case studies in the section ‘Technology Case Studies.’ Besides individual policies, the diffusion of EETs is also affected by a country’s general regulatory framework and how committed the political agenda is to energy efficiency. Clear commitment is perceived by decision-makers as a basis for reliable long-term planning and thus is expected to increase adoption rates. Figure 3 summarizes this section by illustrating the decision to adopt an EET and its determinants, including adopter characteristics, EET characteristics, policies, and other

Markets/Technology Innovation/Adoption/Diffusion | Diffusion of Energy-Efficient Technologies

Policies

Price-based policies (energy tax, grants)

Policies addressing individual barriers (standards, labeling, audits, soft loans, energy management)

Decision-making

Utility/profitability

Barriers and drivers (risk, imperfect information, hidden costs and benefits, split incentives, bounded rationality, long-term strategy)

69

Determinants of adoption

Contextual factors (energy prices, political stability, social norms)

Adopter characteristics (firm size, income, preferences)

EET characteristics (energy savings, complexity, risk)

EET adoption/ non-adoption Figure 3 Illustration of the determinants of EET adoption.

contextual factors such as the aforementioned broader political agenda. The figure, however, does not consider all potential interactions between the elements included; it only gives a general overview of the main elements.

From Adoption to Diffusion: The Time Dimension, Feedback Loops, and Diffusion Dynamics As mentioned earlier, the diffusion of a technology can be regarded as a sequence of individual adoption decisions by heterogeneous users. The heterogeneity and spread of information among users determine the resulting diffusion curve. While these factors certainly influence the progression and speed of diffusion, they are not sufficient to explain it entirely. The diffusion trajectory is also shaped by the interaction of competing or coevolving technologies, by interactions between supply and demand, and generally by effects from increasing returns to adoption. These factors result in very dynamic diffusion patterns. Particularly, factors summarized under ‘increasing returns to adoption’ critically affect the diffusion path for most technologies. These comprise technological learning in the form of learning by doing and learning by using, economies of scale, network effects, and the spread and availability of information. Together, these factors typically result in significant technological improvements, which are directly observed as falling specific costs. If these factors are intense and the technology is more mature and upscaled as a result, the entire system can become ‘locked-in’ to this technology. Emerging technologies compete on an unequal footing, as they are often less mature compared to the reference technology, resulting in lower quality and higher specific costs. This is particularly important for EETs, which, by definition, compete with the mature technologies they aim to replace.

As a result of technology learning, technologies also change significantly during the course of diffusion. Not only do they become less costly but their quality also improves and they become better adapted to the particular needs of the users. Thus, the technology supply side and its interaction with demand play an important role, particularly in the early diffusion phase. During this phase, technologies are often relatively immature, and quick quality improvements are required to avoid disillusionment after the first euphoria of technology development. Policy conclusions derived from these diffusion dynamics accord a key role to niche markets, which allow learning and scale effects to improve the technology. Consequently, policies need to address technologies in this early phase of diffusion by, for instance, generating artificial niche markets. Once the new technology becomes more mature, the policy support can be withdrawn step by step, until the technology is competitive even without support. It is, however, a difficult task to estimate which technologies will outperform the reference technology in the long term. If this does not occur, an artificial market is established that will not survive without policy support and that might have negative welfare effects.

Technology Case Studies A few case studies are presented here to illustrate the general discussion concerning the diffusion of EETs, its drivers and determinants.

Compact Fluorescent Lamps The past diffusion pattern of CFLs has to be studied in the light of the competition with conventional incandescent lamps, which accounted for nearly the entire residential lighting

70

Markets/Technology Innovation/Adoption/Diffusion | Diffusion of Energy-Efficient Technologies

market when CFLs entered the market in the 1980s. CFLs consume about 80% less energy than incandescent light bulbs. While prices for CFLs are significantly higher than for conventional light bulbs, their life cycle costs are much lower because of lower energy consumption and a longer lifespan. Despite these advantages, the diffusion of CFLs has been slower than expected considering the automatic capital stock turnover of incandescent light bulbs in most countries. Three main barriers have been identified. First, consumers focus on the initial purchase costs rather than lifecycle costs, and the higher price of CFLs made consumers reluctant to buy them. Second, the first generation of CFLs had significant quality losses compared to conventional light bulbs. They gave off a cold light, the size of the bulb was often too large, and many bulbs flickered and needed a relatively long time to achieve their full light output after being switched on. Although these problems were less pronounced in later generations of CFLs, consumers retained these negative associations and remained skeptical. Third, it was difficult for consumers to compare the efficiency of different types of bulbs, as information was often not available on the products, or not comparable. In combination, these factors constituted substantial barriers and kept diffusion of CFLs to a low level. Consumers preferred conventional incandescent bulbs, which had undergone about a century of incremental improvements in product design and production techniques and represent a very mature technology. Various policy programs with different types of instruments were established in many countries worldwide to try and accelerate the diffusion of CFLs and overcome this lock-in situation. Such policies ranged from broad information campaigns stressing the benefits of CFLs to consumers, to subsidy programs that aimed to lower the initial investment barrier, and CFL certification programs to assure product quality standards. Owing to the broad range of barriers, information programs alone were rarely effective in promoting CFLs. However, such policies did have a considerable impact in several countries (particularly, when several policies were applied in combination) and were able to generate niche markets and facilitate technological learning and economies of scale, resulting in more mature CFLs with lower costs, higher luminous efficacy, and better light quality. Despite the success of such policies and the continuing development of improved CFLs, incandescent bulbs still owned a large share of the market in many countries. As a reaction, the European Union recently decided to introduce MEPS for lamps, which will result in a phase out of incandescent light bulbs. The standards were first introduced for bulbs of at least 100 W in 2009 and then stepwise for lower wattage bulbs, until all bulbs are due to be covered by the end of 2012. With the market entry of LEDs, these dynamics started again and relatively immature LEDs with higher initial costs have to compete with more modern CFLs. A main difference, however, can be observed: The new reference technology, CFLs, is far less mature than the incandescent light bulbs were in the example above.

Electric Motors In the European Union, electric motors are responsible for about 70% of electricity consumption in the industrial sector and about 40% in the service sector. These shares are comparable in all industrialized countries. Thus, even minor

improvements in the efficiency of electric motors can yield substantial energy savings if applied throughout the entire motor stock. In general, motor efficiency depends on the rated power of the motor, with smaller motors being considerably less efficient than larger ones. While motors above 10 kW rated power show efficiencies of more than 90%, motors with a rated power of less than 1 kW often have efficiencies of less than 80% or even 70%. However, motor efficiencies also depend on materials used (e.g., increased use of copper resulting in lower resistance losses and higher motor efficiency), coil quality and design, so that efficiencies can vary considerably even for the same size of motor. Typically, energy-efficient electric motors are similar to conventional motors and have often undergone only incremental improvements such as the use of a copper rotor. Thus, the quality of even early-stage products is similar to that of conventional electric motors. Prices are a little higher for energyefficient motors, but life cycle costs are typically a lot lower for most applications (except for applications with very low annual running hours), resulting in very short payback times of a few years or, in some cases, a few months only. Despite their high cost-effectiveness, the market share of energy-efficient electric motors grew only slowly in the 1980s and 1990s. The main barrier was a combination of split incentives and imperfect information. For motor consumers, it was hardly possible to compare the efficiency of different electric motors, due to different testing standards and missing information on the motor plate. The diffusion of energyefficient electric motors did not take off even when in 1998 a label was introduced in Europe indicating their efficiency. There are several reasons for this, an important one being market structure. Most motors are bought by OEMs, who then integrate the motor into other products such as pumps, compressors, or refrigerators, which are then sold to the final consumers. OEMs, however, have less incentive to buy efficient motors because they do not pay the electricity bill and compete mostly based on the prices rather than the life cycle costs of their products. On the other hand, it is hard for consumers, who do pay the electricity bill, to identify the motor incorporated into the product they buy. As information programs are less effective in overcoming such structural barriers, more and more countries have introduced MEPS for electric motors, which continuously require higher levels of efficiency. Figure 4 shows the evolution of the market shares of different efficiency classes of electric motors within Europe. From 2011 onward, motors sold on the EU market need to meet the IE2 efficiency requirements or better. From the perspective of energy savings, optimizing the system in which electric motors are used has the potential of delivering much larger gains in efficiency than just using highly efficient motors. Since users are not directly interested in the rotating motion delivered by the electric motor, but rather in transporting goods or compressed air, it is possible to widen the scope of the system considered to include more efficient technologies capable of providing this energy service. Variablespeed drives, for instance, have received a lot of attention in energy-efficient motor-driven systems. Variable-speed drives are devices that can regulate the speed and rotational force of an electric motor depending on the system load and thereby

Markets/Technology Innovation/Adoption/Diffusion | Diffusion of Energy-Efficient Technologies

100% 16%

Sales share of energy efficient motor classes

90% 80%

12% 10%

8%

Below IE1

43%

4%

IE1

3%

IE2

2%

2%

1%

71

1%

IE3

53%

70% 68% 60% 50%

80%

40% 30%

85% 83% 84%

87%

78.5% 82% 82% 85% 86%

54% 44%

20%

30%

10% 9%

0%

17% 20% 12% 12% 16%

2% 3% 3% 4% 5% 6% 7% 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Figure 4 Diffusion of different electric motor efficiency classes in Europe. http://www.CEMEP.org

reduce energy consumption. In many cases, an efficient motor equipped with steering and control, for example, using variable-speed drives, is a more efficient system than an efficient motor without steering and control. However, the details depend on the specific load of the system, and it becomes increasingly difficult to make general statements. This is in fact a problem that also affects attempts to influence the diffusion of energy-efficient motor-driven systems, since MEPS can only be defined with reference to the system. In many respects, the notion of the system involved is one of the current core problems of EETs.

Condensing Gas Boilers Condensing gas boilers are designed to make use of the latent heat contained in their water vapor flue gas, which results in a conversion efficiency that is about 20% higher than noncondensing gas boilers. In the Netherlands, condensing gas boilers entered the market in the early 1980s. Despite their significant benefits in terms of energy efficiency, they only had a market share of around 8% by 1987. Representing an investment sum of more than 2000 Euros (including installation), a boiler is clearly an important investment decision for households and they are expected to take profitability into account in the form of the life cycle costs. Furthermore, boilers are not a lifestyle product and, because they are usually installed in the cellar or elsewhere out of sight, they do not have a high degree of visibility and prestige. As a result, households are expected to follow a more rational investment decision based on profitability. The main reasons for the low adoption rate were the lack of training and experience on the part of the installers, additional requirements of the household infrastructure, lower reliability than conventional noncondensing gas boilers, and higher boiler prices. Furthermore, the reference technology also changed with the market entry of improved, noncondensing gas boilers that reduced the efficiency difference

with condensing gas boilers and made them seem less attractive to consumers, that is, reduced their relative advantage. However, in the end, these reasons only slowed down the diffusion of condensing gas boilers, which finally gained momentum in 1991, when the Dutch government provided substantial investment grants for condensing gas boilers. The accelerated diffusion gave rise to learning effects, which resulted in falling specific costs for condensing gas boilers that, together with rising natural gas prices in the following years, helped to further push diffusion. In the years up to 2000, condensing gas boilers gained a market share of more than 80%, which continued to rise to more than 90% by 2006. To conclude, the overall diffusion path shows the frequently observed sigmoid pattern (Figure 5). The probit model of diffusion can help to explain the diffusion pattern, which is mainly driven by profitability. Policies in the form of investment grants substantially improved the profitability of condensing gas boilers in comparison to conventional boilers. In total, the diffusion of condensing gas boilers has taken about 25 years from market entry to a rather saturated market share (in terms of sales). This is already a long time frame, but the actual diffusion through the housing stock will take much longer, depending on the lifespan of gas boilers and the age distribution of the existing stock.

Diffusion of Efficient Propulsion Technologies Another interesting example of the diffusion of EETs is provided by efficient and alternative propulsion technologies for road transport. Road transport accounts for large shares of global CO2 emissions and local emissions of noise and dust particles. The reduction of such emissions is a priority for many governments. Lessons learnt from existing efficient or alternative propulsion technologies, such as diesel and gas engines, can be used in policy actions promoting the introduction of electric or fuel cell vehicles with their high potentials to reduce local and global emissions.

72

Markets/Technology Innovation/Adoption/Diffusion | Diffusion of Energy-Efficient Technologies

100 90 80 70 60 50 40 30 20

Noncondensing gas boilers Improved efficiency noncondensing gas boilers

10

Condensing gas boilers 06 20

04 20

02 20

00 20

98 19

96 19

94 19

92 19

90 19

88 19

86 19

84 19

19

82

0

Figure 5 Evolution of the market share of condensing gas boilers in the Netherlands showing the typical s-shaped diffusion curve. Reproduced from Weiss M, Dittmar L, Junginger M, et al. (2009) Market diffusion, technological learning, and cost-benefit dynamics of condensing gas boilers in the Netherlands. Energy Policy 37(8): 2962–2976.

Many countries have passenger transport based on diesel engines, albeit to strongly varying degrees, and this provides a good example for studying the different factors influencing the diffusion of EETs. It represents the classical case of an EET requiring a slightly higher investment than the reference technology (the gasoline engine) but with lower fuel consumption and thus lower variable costs. Diesel technology has been around almost as long as the gasoline engine but only gained significant market shares in the passenger vehicle market after the introduction of the turbocharger. Furthermore, the market for vehicles is divided into different categories: private purchasers, company fleet vehicles, and company cars that may also be used privately. These groups obviously assign different priorities to the total life cycle costs as, for example, commercial vehicle fleets with high annual mileages show high shares of diesel vehicles. On the other hand, the general perception of a technology plays an important role for private customers. For example, diesel engines are much less common in private vehicles in the United States than in Europe, since North American car buyers perceive diesel technology as ‘dirty’ rather than efficient and clean. But even within Europe, it is possible to observe the effect of financial incentives on energy prices: where diesel fuel is financially subsidized by governments, for example, via tax reduction, diesel vehicles reach higher market shares than elsewhere. In addition, efficient or alternative fuel vehicles often require special infrastructure for refueling. This is true for diesel vehicles, and also for the electric and fuel cell vehicles currently under discussion, although in different ways, since the electricity required is already widely available at present. That said, the case of compressed natural gas and liquefied petroleum gas shows that roughly the same infrastructure coverage does not lead to similar

market shares. Germany, Italy, and the Netherlands show similar ratios of gas vehicles per gas fueling station, that is, their infrastructure coverage seems roughly the same, but the market shares of these vehicles differ significantly between the three countries. These examples illustrate some of the complexity of diffusion processes as well as the interplay between the technology’s properties, adopters’ characteristics, perception, and policy intervention. They provide a useful knowledge base when developing policies to try and influence the diffusion of electric and fuel cell vehicles in the future.

Conclusions The diffusion of EETs is a complex process that can exhibit rapid dynamics depending on diverse factors that vary by adopter and by technology and also depend on the contextual framework. It is very difficult to make simple generalizations across all EETs, because each EET has its own peculiarities and dynamics. However, some general conclusions can still be drawn as follows:





The successful diffusion of EET often follows an s-shaped curve and shows both epidemic and probit effects, that is, effects due to the spread of information as well as elements of a cost-based investment choice of heterogeneous adopters. Technology learning, that is, improving technologies over time, is often a crucial factor, because most EETs have to compete against relatively mature conventional technologies with decades of incremental improvements behind them.

Markets/Technology Innovation/Adoption/Diffusion | Diffusion of Energy-Efficient Technologies



• • •

Policies addressing technology learning can create niche markets and act as a catalyst for a very dynamic development if the EET has the potential to outperform the reference technology in the long term even without policy support. The existing capital stock of the reference technology and its lifespan directly affect the speed of diffusion if the EET requires the replacement of existing equipment. Available information about the energy efficiency of an appliance is a prerequisite for many policies to be effective, while a lack of information can enhance other barriers. The barriers slowing EET diffusion can be very complex and interrelated, but if the major barriers are identified for a given technology, it is possible to design policies to effectively overcome them.

See also: Markets/Technology Innovation/Adoption/Diffusion: Energy-Efficiency Gap; Modeling Technological Change in Economic Models of Climate Change; Policy Incentives for Energy and Environmental Technological Innovation: Lessons from the Empirical Evidence.

Further Reading Dieperink C, Brand I, and Vermeulen W (2003) Diffusion of energy-saving innovations in industry and the built environment: Dutch studies as inputs for a more integrated analytical framework. Energy Policy 32: 773–784.

73

Hekkert MP, Harmsen R, and de Jong A (2007) Explaining the rapid diffusion of Dutch cogeneration by innovation system functioning. Energy Policy 35(9): 4677–4687. Jaffe AB (1995) Dynamic incentives of environmental regulations: The effects of alternative policy instruments on technology diffusion. Journal of Environmental Economics and Management 29(3): S43–S63. Jaffe AB and Stavins RN (1994) The energy-efficiency gap – What does it mean? Energy Policy 22(10): 804–810. Kemp R (1997) Environmental Policy and Technical Change. Cheltenham: Edward Elgar. Lund P (2005) Market penetration rates of new energy technologies. Energy Policy 34: 3317–3326. Menanteau P and Lefebvre H (2000) Competing technologies and the diffusion of innovations: The emergence of energy-saving lamps in the residential sector. Research Policy 29(3): 375–389. Nill J (2008) Diffusion as time-dependent result of technological evolution, competition, and policies: The case of cleaner iron and steel technologies. Journal of Cleaner Production 16(1): S58–S66. Rogers EM (2003) Diffusion of Innovations, 5th edn. New York: The Free Press – A division of Macmillan Publishing Co., Inc.. Sorrell S, O’Malley E, Schleich J, and Scott S (2004) The Economics of Energy Efficiency. Cheltenham: Elgar. Stoneman P (2002) The Economics of Technological Diffusion. Oxford: Blackwell Publishers. Waide P (2006) Light’s Labour’s Lost – Policies for Energy-Saving Lighting. Paris: International Energy Agency. Weiss M, Dittmar L, Junginger M, Patel MK, and Blok K (2009) Market diffusion, technological learning, and cost-benefit dynamics of condensing gas boilers in the Netherlands. Energy Policy 37(8): 2962–2976.