Ecological Economics 68 (2009) 2150–2159
Contents lists available at ScienceDirect
Ecological Economics j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e c o l e c o n
Analysis
Barriers to energy efficiency: A comparison across the German commercial and services sector Joachim Schleich ⁎ Fraunhofer Institute for Systems and Innovation Research (Fraunhofer ISI), Breslauer Strasse 48, 76139 Karlsruhe, Germany Virginia Polytechnic Institute and State University, Blacksburg, Virginia, USA
a r t i c l e
i n f o
Article history: Received 17 October 2007 Received in revised form 15 February 2009 Accepted 17 February 2009 Available online 11 March 2009 Keywords: Energy efficiency Energy consumption Energy policy Organizational investment behaviour Technology diffusion JEL classification: D20 Q48
a b s t r a c t Based on a large sample for the German commercial and services sector, this paper econometrically assesses the relevance of various types of barriers to energy efficiency at the sectoral level and across fifteen subsectors. The results at the level of entire sectors suggest that the lack of information about energy consumption patterns and about energy efficiency measures, lack of staff time, priority setting within organizations, and – in particular – the investor/user dilemma are all relevant barriers. Allowing for sectorspecific differences in the relevance of these individual barriers yields a more heterogeneous picture. The numbers and types of relevant barriers vary across sub-sectors, and the majority of sub-sectors are subject to relatively few barriers. The statistically most significant barriers are found for the sub-sector of public administrations. These findings are robust, independent of whether the definition of an organization's energy efficiency performance includes only measures that have actually been realized or also those that are being planned. For planned projects, however, organizations appear to underestimate internal priority setting as a barrier to energy efficiency. © 2009 Elsevier B.V. All rights reserved.
1. Introduction Improving energy efficiency is often seen as the fastest and most cost-effective way to achieve global greenhouse gas emission targets (e.g. IEA, 2008). Consequently, strategies for obtaining more energy services such as heat, light or mobility with the same or less energy input have recently attracted increased attention from policymakers and academics alike. Other reasons for this renaissance of energy efficiency include associated cost savings for companies and households in the light of expected high future energy prices, improved security of energy services, and other co-benefits such as employment or productivity gains, or health benefits due to lower emissions of local pollutants (e.g. nitrogen oxides and sulphur). On the policy side, the Spring European Council Presidency Conclusions stress the need “to increase energy efficiency in the EU so as to achieve the objective of saving 20% of the EU's energy consumption compared to projections for 2020, as estimated by the Commission in its Green Paper on Energy Efficiency” (European Council, 2007, p. 20). In view of that, the Action Plan (European Commission, 2006) outlines a framework of policies and measures for all end-use sectors (residential, tertiary, industry and transportation) ⁎ Fraunhofer Institute for Systems and Innovation Research (Fraunhofer ISI), Breslauer Strasse 48, 76139 Karlsruhe, Germany. Tel.: +49 721 6809 203; fax: +49 721 6809 272. E-mail address:
[email protected]. 0921-8009/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.ecolecon.2009.02.008
and the transformation sector to improve energy efficiency. Accordingly, “additional investment expenditure in more efficient and innovative technologies will be more than compensated by the more than € 100 billions annual fuel savings” in the EU (European Commission, 2006, p. 3). The commercial buildings (tertiary) sector is estimated to exhibit the highest relative potential for energy savings of 30%. The measures proposed to realize these potentials include implementing energy management systems, promoting public– private energy efficiency funds or financing packages and energy audits in small and medium-sized companies and in the public sector. In particular, such policy measures are supposed to help overcome the so-called barriers to energy efficiency which are preventing energy efficiency measures from being realized. A thorough understanding of the nature of these barriers is crucial when designing cost-efficient policy measures. Most empirical analyses of barriers to energy efficiency are in the form of case studies, where theory-based hypotheses are derived from various (partially overlapping) concepts grounded in neo-classical economics, institutional economics, organizational theory, sociology, and psychology. Case studies are then carried out in selected organizations either for a specific technology, such as electric motors (de Almeida, 1998; Ostertag, 2003), or for specific sectors such as manufacturing, the public sector, or the food industry (Ramesohl, 1998; Wuppertal Institute et al., 1998; Schleich et al., 2001; Sorrell, 2003; Sorrell et al., 2004), which are typically characterized by a low energy-cost share. Such case studies are well
J. Schleich / Ecological Economics 68 (2009) 2150–2159
suited to gaining insights into complex decision-making processes and structures within organizations. Their findings, though, are usually limited to an analytical generalization; that is, the observed outcomes of decision-making processes are explained by identifying relevant causal mechanisms (Yin, 1994). The basis for generalizing in a statistical sense, however, is weak. Complementary to the case study approach, a few analyses do exist which rely on surveys exploring the empirical relevance of barriers to energy efficiency. An early paper by Brechling and Smith (1994) analyses the take-up of wall insulation, loft insulation and double glazing in the UK household sector. In a similar study, Scott (1997) considers attic insulation, hot water cylinder insulation, and low energy light bulbs in Irish households. In these studies, information costs, transaction costs, restricted access to capital, the investor/user dilemma, and small potential savings are found to impede the diffusion of energyefficient measures. Using company-level survey data from the United States' Environmental Protection Agency's Green Lights Program, DeCanio (1998) and DeCanio and Watkins (1998) find that industry investment behaviour is not only driven by economic factors, but also by firms' characteristics such as organizational and bureaucratic factors, location, product portfolio and profitability. de Groot et al. (2001) explored whether barriers to energy efficiency vary across sectors and across firms' characteristics in the Dutch industry, but the relatively few observations did not allow for sector-specific analyses. Finally, Schleich and Gruber (2008) econometrically assess the relevance of several types of barriers to energy efficiency for the German commercial and services sector (small commercial businesses and private and public service organizations). By running individual regressions for 19 sub-sectors, they assess several types of barriers to organizations' energy efficiency performance within those sub-sectors. Lack of information about energy consumption patterns and the investor/user dilemma were found to be the most frequent barriers. This paper assesses the relevance of various types of barriers to energy efficiency in the commercial and services sector at the sectoral level and across sub-sectors. The barriers explored include transaction costs, hidden costs, lack of capital, risk and uncertainty, and the investor/ user dilemma. Relying on the same data set and applying a similar methodology as Schleich and Gruber (2008), two types of models are estimated econometrically. First, in the “Sector model”, regressions are run to test the relevance of the barriers for organizations' energy efficiency performance at the level of the entire German commercial and services sector. Second, the “Sub-sector model” allows for interaction effects between the sub-sectors and barriers in order to explore differences in the relevance of barriers across sub-sectors. For both types of models, two distinct indicators are applied to measure organizations' energy efficiency performance. One definition considers those measures which have actually been realized or are planned in the organization. The other definition only includes realized measures. In summary, the paper relates to the existing literature as follows. First, econometric analyses of barriers to energy efficiency are carried out for energy consumers which have – apart from Schleich and Gruber (2008) – been neglected so far. Second, the analyses focus on organizations' general energy performance and not on specific energyefficient technologies as was the case in Brechling and Smith (1994), Scott (1997), DeCanio (1998) or DeCanio and Watkins (1998). Third, distinguishing between different energy efficiency measures of organizations' makes it possible to explore whether organizations under- or overestimate some barriers when it comes to implementing planned energy efficiency measures. Fourth, and arguably the most relevant contribution is that this paper permits a general assessment of the barriers by running regressions on the (almost) entire commercial and services sector at once. In addition to Schleich and Gruber (2008), it allows differences in the relevance of barriers to be explored across subsectors. Since the findings help to identify sub-sectors where certain barriers are more relevant than in other sub-sectors, they can also be used to design more cost-effective policy measures.
2151
The paper is organized as follows. Section 2 provides an overview of the barriers to energy efficiency. A description of the sectors and the data is given in Section 3. Section 4 presents the models and Section 5 the results. The concluding Section 6 also discusses implications for policy making to address the main barriers. 2. Nature of barriers to energy efficiency In the commercial and services sector in Germany, the share of energy costs is usually well under three percent of total costs (e.g. Schmid et al., 2003), and thus rather low, and investments in energy efficiency do not affect the core production processes. Therefore, energy efficiency tends to receive less attention in these organizations than in energy-intensive industry sectors such as the power, the mineral processing, or the iron and steel industries. For example, in a recent empirical study conducted among Swiss companies, Cooremans (2007) finds that energy efficiency projects are rejected primarily because they are not perceived as “strategic”. There is a large body of literature on the nature of barriers to energy efficiency, which draws on partly overlapping concepts from neo-classical economics, institutional economics (principal-agency theory and transaction cost economics), behavioural economics, sociology and psychology (Stern, 1986; Howarth and Andersson, 1993; Jaffe and Stavins, 1994a; Howarth and Sanstad, 1995; Brown, 2001; Sorrell et al., 2004). This literature tries to explain why organizations fail to invest in energy efficiency even though it is profitable under current economic conditions to do so — a “phenomenon” that has also been referred to as the “energy efficiency gap” (Jaffe and Stavins, 1994b). Relying on the taxonomy developed in Sorrell et al. (2004), this section provides a brief summary of the main concepts of barriers to energy efficiency. 2.1. Imperfect information If individuals lack adequate information on either energy efficiency opportunities or the energy performance of technologies, they may invest too little in energy efficiency. First, there might be inadequate information about the levels and patterns of current energy consumption. The availability of such information depends on the level of sub-metering, the information content of utility bills, the use of computerized information systems, the time devoted to analysing consumption information, etc. Gathering and analysing information on current energy consumption is associated with investment, operational, and staff costs, which can also be interpreted as a particular category of transaction costs. Second, organizations may lack information on specific energysaving opportunities either because they have failed to evaluate energy efficiency opportunities, or because there is no information on the costs and performance of specific energy-saving technologies. For example, the performance of technologies such as control systems, motors, and variable-speed drives may be difficult to evaluate even after purchase because detailed metering is not feasible. Therefore, the performance of the energy-efficient technology remains unknown. Similarly, details about the performance of new energy-efficient technologies are only known to the investor, but would be of value to others, too (public-goods character of information). In this case, markets undersupply such information (market failure). Third, information on the energy consumption of new and refurbished buildings, process plants, and equipment and machinery could be asymmetric, resulting in adverse selection and thus inefficient outcomes. For example, the market value of a house should, among many other characteristics, also reflect its energy performance. While this information may be available to the seller, potential buyers have difficulty in recognising and evaluating energy performance. As a consequence, their bids for the house may be too low. Eventually, unless the buyer can adequately assess the energy
2152
J. Schleich / Ecological Economics 68 (2009) 2150–2159
performance of a building, or unless the seller is able to credibly disclose this information (e.g. through demonstrations, publication of technical materials), only energy-inefficient buildings could be offered on the market. Thus, the logic originally developed by Akerlof (1970) for the second-hand car market, may also hold for energyefficient technologies. 2.2. Hidden costs Organizations may fail to invest in seemingly profitable energyefficient technologies because there are additional costs associated with their use which are hidden to the observer, but not to the organization. That is, decision makers may be very well aware of those hidden costs, but they cannot easily be observed or adequately quantified via, say, an engineering–economic investment appraisal carried out by external energy auditing companies. First, there are the costs (utility losses) that result from the inferior performance of energy-efficient technologies with respect to dimensions other than energy services. Examples include energy-efficient production processes leading to increased noise, the installation of cavity wall insulation in an old building encouraging damp (because the moisture which previously escaped through the walls is then trapped in the building), energy-efficient light bulbs providing a different lighting quality, or a variable-speed drive requiring extra maintenance and training. Second, hidden costs may be part of the production costs associated with individual technologies such as the costs for identifying opportunities, detailed investigation and design, formal investment appraisal, procedures for seeking credit approval, additional maintenance, replacement, early retirement, retaining or hiring staff, or for production interruptions during equipment installation. Third, there may be hidden costs related to the general, overhead costs of energy management, which correspond to the search costs discussed earlier in the context of the “imperfect information” barrier. On the one hand, these costs partly depend on factors beyond the control of the organization, such as the existence of standardized labelling schemes for energy-using technologies. On the other hand, these search costs also depend on factors within the organization such as organizational procedures for purchasing and procurement. The broader category of transaction costs (Coase, 1991) includes all the organizational costs associated with establishing and maintaining an energy management scheme, investing in specific energy-saving technologies, and implementing specific energy-efficient options within broader investment programmes (for example, choosing an energy-efficient motor rather than a standard one). In contrast to the costs related to inferior performance and production costs, transaction costs are heavily dependent on organizational and contractual structures, procedures, incentives, and routines (Ostertag, 2003). If these hidden costs are real and substantial, firms' observed failure to adopt seemingly “profitable” energy-efficient technologies would still be consistent with standard neoclassical optimization. However, if these costs are unobservable or difficult to measure, attributing the failure to adopt these technologies to unobservable costs would be next to tautological. Exploring firms' voluntary participation in the US Environmental Protection Agency's Green Lights Program, DeCanio and Watkins (1998) find that firms' individual characteristics (like size, location, or profitability) affect the adoption of energy-efficient lighting. Since standard neoclassical theory would not allow for participation to be systematically associated with firms' characteristics, these findings cast doubt on the validity of the standard profit-maximization assumption. 2.3. Risk and uncertainty High (implied) discount rates are often observed for investments in energy efficiency. In essence, however, high discount rates are a
restatement rather than the source of the energy efficiency gap. Instead, stringent investment criteria and the rejection of particular energy-efficient technologies may represent a rational response to perceived risk. They may result from financial risks such as businessspecific risk, regulatory risk, or general economic risk caused by the business cycle, fluctuation of exchange rates and energy prices, etc. In particular, stochastic future energy prices or stochastic prices for allowances under emission trading systems render returns on investments in energy-efficient technologies uncertain. From this perspective, risk-averse company managers can be expected to invest less.1 However, investments in energy efficiency also lower energy bills or the level of carbon emissions and thus reduce the financial risks resulting from uncertainty about the prices for energy or emissions allowances (Howarth and Sanstad, 1995; Ben-David et al., 2000). Thus, if investors took into account the effects on total company costs and profits, they might actually invest more. In that sense, uncertainty should encourage energy efficiency rather than act as a barrier (Sutherland, 1996). The relative magnitude of both effects is company-specific and generally ambiguous. New energy-efficient technologies may also be associated with technical risks. If energy-efficient technologies are unreliable, the risk of breakdowns and disruptions might outweigh any potential benefits from reduced energy costs. Ignoring these technologies is not only perfectly rational, but also avoids inefficient outcomes. Since there is no empirical evidence, however, that energy-efficient technologies in general actually carry higher risks than standard technologies, this would have to be assessed on a case by case basis. Finally, waiting to invest in irreversible energy efficiency technologies may be optimal if future economic conditions are uncertain (Hassett and Metcalf, 1993; van Soest and Bulte, 2001). For example, investing in a more energy-efficient technology may turn out to be unprofitable if energy prices fall after the new technology has been implemented. Similarly, government grants for investments in energy efficiency may be introduced, or technology may improve dramatically. Thus, there is an option value associated with postponing investments (McDonald and Siegel, 1986; Dixit and Pindyck, 1994). Critics of the option value approach argue that it fails to account for the potential costs of delaying energy efficiency investments (Howarth and Sanstad, 1995). For example, including a heat recovery system in the design of a new building or plant is cheaper than retrofitting one afterwards. 2.4. Access to capital If organizations' external access to capital is limited so that they can only borrow at high interest rates, this may prevent energy efficiency projects from being undertaken even if they exhibit a high expected rate of return. If small and medium-sized companies are involved, lenders may demand a high risk-adjusted rate of return because the economic risk is higher (Sutherland, 1996). For example, the product portfolio of smaller companies tends to be less diversified, so they are more vulnerable to negative economic shocks. Likewise, they may only have a limited ability to offer collateral. From the perspective of transaction costs economics though, it may be that the costs necessary to investigate the creditworthiness of small companies render such loans unprofitable (Golove and Eto, 1996). In addition to this external “access to capital” problem, which relates to all types of investments, there may also be internal capital budgeting procedures which tend to discriminate against energy efficiency projects. Since energy efficiency investments are often 1 Differences in attitudes towards risk between company managers and owners (shareholders) may result in diverging incentives, and hence give rise to a principalagent problem (see Holmstrom, 1979). Here, company share holders may be riskneutral because they are able to diversify their portfolio, while managers tend to be risk-averse because their salary and job security are linked exclusively to their company. Ideally, the managers' contract should be designed so that it is also in the managers' best interest to invest in a riskier project.
J. Schleich / Ecological Economics 68 (2009) 2150–2159
classified as discretionary maintenance projects, they are usually assigned lower priority than essential maintenance projects or strategic investments (Sorrell et al., 2004). Likewise, the application of (short) payback periods rather than discounted cash flow analyses as an investment criterion neglects the (expected) positive cash flow from energy-cost savings in the long run. Similarly, relatively high hurdle rates may be required for small projects in general and energy efficiency projects in particular, because the transaction costs of determining the profitability of such investments represent a larger portion of the expected savings. Finally, because of constraints on their time and attention, top management may not consider energy-cost savings a strategic priority, or may favour large, strategic projects which are more prestigious than energy management activities. More specifically, findings from management literature suggest that the strategic character of an investment is the primary reason for its approval, even more important than profitability (e.g. Butler et al., 1993; Carr and Tomkins, 1998)2. Following Teece et al. (1997), organizations, in particular upper management, do not tend to value energy very highly because it is seen as being part of the organization's material resources, in contrast, for example, to information, which belongs to the highly valued non-material resources. Thus, the internal “access to capital” problem may not only result from hard investment criteria such as the rate of return or payback time of an investment project, but also from soft factors such as strategic priorities, the status of energy efficiency, reputation, or the relative power of those responsible for energy management within the organization (Morgan, 1985; DeCanio, 1994; Sorrell et al., 2004). 2.5. Split incentives and appropriability In the context of energy use, arguably the most well-known example for split incentives is the landlord/tenant or user/investor dilemma (e.g. IEA, 2007). Neither the landlord nor the tenant may have the incentive to invest in energy efficiency in a building if the investor is not able to appropriate the benefits from the resulting energy-cost savings. On the one hand, the landlord may not invest in energy efficiency if the investment costs cannot be passed on to the tenant who stands to benefit from the lower energy costs. On the other hand, tenants may not invest if they are likely to move out before fully benefiting from the savings in energy costs. In principle, this dilemma could be avoided if the investor were able to credibly transmit the information about the benefits arising from the investment and to enter into a contract with those benefiting from the investment. Such a contract would have to secure the appropriation of cost savings so that the investor would be able to recover the initial investment costs. However, the costs of verifying energy-cost savings and the costs for the contractual arrangements are often prohibitive. As Jaffe and Stavins (1994b) point out, asymmetric information and transaction costs are the sources for the investor/user dilemma. Besides the landlord/tenant problem, there are also other situations where split incentives may prevent adequate investments in energy efficiency. For example, if managers – because of job rotation – remain in their post only for a short time, they may have limited incentives to invest in energy-efficient projects, which have a longer payback time. Similarly, depending on their compensation scheme, managers may not be indifferent to the equality of making or saving the same amount of money. Further, if departments (in larger organisations) are not accountable for their own energy costs, department managers may have no incentive to invest in energy efficiency because the benefits in terms of cost savings accrue elsewhere. Similarly, the person responsible for purchasing equipment may have a strong incentive to minimize capital costs, but may 2 See also the brief survey on the organizational decision-making literature in Cooremans (2007).
2153
not be accountable for operating costs (including energy costs). Likewise, if the energy performance of a product or a system cannot easily be observed, developers may ‘get away with’ cheaply installed systems that the purchaser discovers to be sub-optimal afterwards. 2.6. Bounded rationality Orthodox neoclassical economics assumes a rational decisionmaker choosing the optimal solution given the available information on alternatives. In particular, the decision is not tainted by cognitive limits or biases. In contrast, organizational theory suggests that lack of time, attention, or cognitive limits on the ability to adequately process information may prevent optimizing behaviour. Instead, bounded rationality may result in satisfying behaviour, using routines, or rules of thumb (Simon, 1957, 1959). Because of bounded rationality, some opportunities for improving energy efficiency are neglected — even if there is access to perfect information and the incentive structure is appropriate. For example, small motor end-users tend to consider only delivery time or price instead of life-cycle costs when buying a new motor to replace an old one (de Almeida, 1998). Similarly, when making decisions about investment priorities, companies are likely to focus on the core production process rather than on ways to save energy costs and making money. Likewise, in cases where investments in energy-efficient technologies are being considered, the same profitability or payback criteria may be required as are applied to the core production technologies, even though the economic risks associated with the former are much lower. In conclusion, the different concepts underlying the barriers to energy efficiency each provide different insights into the nature of these barriers. In particular, neoclassical economics distinguishes between barriers which root in market failures and those which do not (Jaffe and Stavins, 1994b). Market failures result, for example, from the public good attributes of information, or from asymmetric information in energy service markets. In contrast, barriers such as uncertain energy prices or hidden costs from production disruption or from lower product performance are not market failures. Other barriers to energy efficiency which are market failures, but which do not explain the energy efficiency gap, include distortions in energy pricing. That is, energy prices may be lower than socially desired because they do not adequately reflect the environmental costs associated with energy production and consumption, or because fuel or final energy uses are subsidized. From the perspective of neoclassical economics, policy intervention would only be justified for market failures provided that the benefits arising from intervention exceed its cost (Jaffe and Stavins, 1994b). Hence, improved energy efficiency would be a “by product” of improved economic efficiency. Concepts from institutional economics provide additional insights into barriers which are internal to the organisations, such as information costs, overhead costs of energy management, incentive structures and appropriability. Transaction cost economics, in particular, maintains that policy intervention and different institutional structures may lower transaction costs. Finally, departing from the presumption of individual rationality, concepts from behavioural economics, organizational theory, sociology and psychology have contributed to a better understanding of actual decision-making in organizations — also in terms of energy efficiency. Nevertheless, these concepts overlap to a certain extent and more than one factor may explain the observed (lack of) adoption of energy-efficient measures. The relative contribution of these factors may also vary between technologies as well as within organizations, across organizations within the same sector and across sectors or sub-sectors in the economy. From a policy perspective, regulation – if it can be justified – would depend on the nature of the barrier. Using an existing data set for the German commercial and services sector, the subsequent sections explore the prevalence of various barriers within this sector as well as across sub-sectors.
2154
J. Schleich / Ecological Economics 68 (2009) 2150–2159
3. Description of sectors and data In Germany, official energy balances partition final energy consumption into four end-use sectors: industry, private households, transportation, and the combined commercial and services sector. The latter consists of small industrial enterprises (with no more than 20 employees), all public and private services as well as the agricultural and construction sectors. Hence, the German commercial and services sector differs from the categories ‘service sector’, ‘public sector’ or ‘tertiary sector’ usually found in United Nations or Eurostat energy statistics. As defined above, the German commercial and services sector accounts for about 15% of final energy consumption and 6.5% of direct CO2-emissions in Germany. If the indirect CO2-emissions resulting from the generation of power and heat used in the commercial and services sector are also taken into account, this share rises to 18% (Arbeitsgemeinschaft Energiebilanzen, 2006; BMU, 2007). The analyses in this paper are based on a representative, crosssectional survey of 2848 companies and public institutions in the commercial and services sector in Germany (Geiger et al., 1999).3 These data were gathered in personal interviews and comprise information on energy use patterns, energy management, measures to improve energy performance and also on perceived barriers to energy efficiency. Because of missing data, about 2000 observations are suitable for the analyses presented in this paper. In the original dataset, the observations were broken down into 23 fairly homogeneous sub-sectors which reflect the structure in official statistics. To capture heterogeneity within sub-sectors, some were further broken down into sub-splits. Since, in this paper, single regressions are run on observations from numerous sub-sectors, only those subsectors are included in the analysis where all the variables can be formed in a consistent and meaningful way. For example, “energy consumption per employee” would not be a meaningful explanatory variable in sub-sectors like agriculture, horticulture, hospitals or schools, which are also assigned to the combined commercial and services sector. Table 1 shows the fifteen sub-sectors used in this paper in greater detail. Interview partners in the surveyed organizations were given a set of possible energy-saving measures and asked which of those measures had already been realized in their organization or were planned to be implemented. The measures considered differed across sub-sectors and generally included both technical and organizational measures. For the commercial businesses and trade organizations, these measures typically referred to specific production technologies. Similarly, for the public and private services organizations, which tend to be dominated by space heating, the measures primarily included technological and management options to reduce heat demand and to improve the energy performance of the heat supply system. From a technology perspective, cross-sectional and sector-specific measures can be distinguished. Cross-cutting measures can be applied in all subsectors and include, for example, thermal insulation of outside walls, roofs and basement ceilings, thermally insulated windows and glazing, heat recovery in refrigerators and freezers, and using outdoor sensors to control indoor temperature. In comparison, sector-specific measures vary across sub-sectors. For laundries, for example, the list of measures considered contains, among others, heat recovery from flue gas or condensate return to preheat the boiler, heat recovery from the air discharged by the drier, or heat recovery from warm washing water to heat the rinsing or fresh water thereby reducing the liquor ratio, i.e. the ratio of water volume in the drum (in l) to the dry load (in kg). In the survey, interviewees could indicate whether some of the suggested measures were not feasible for their organization, for example for technical reasons. While all the suggested measures are
3
The questionnaires are available upon request from the author.
Table 1 Description of sub-sectors. Sector
NACE description
Small commercial businesses and trade Bakeries Bread, fresh pastry goods and cakes Butchers Processing and preserving of meat and meat products Car repair industry Maintenance and repair of motor vehicles Construction Construction Laundries and dry Washing and dry-cleaning cleaners Metal industry Basic metals and fabricated metal products Retail trade Retail trade Wholesale trade Wholesale trade Wood working and Wood and wood products processing Public and private services organizations Banks & insurance Financial intermediation companies Gastronomy Restaurants, bars, canteens and catering Hotel industry Non-commercial organizations Public administrations Services⁎
Hotels Non-commercial organizations, activities of membership organizations Public organizations Services: lawyers, architects, small private health services, private agencies etc.
NACE Nr. 15.81 15.1 50.20 45 93.01 27, 28 52 51 20
65, 66, 67 55.3, 55.4, 55.5 55.1 91 75, 80, 90 74.11,74.20, 93.05
⁎For this study, the split for lawyers, architects, small private health services and private agencies was used.
energy-efficient, no information is available on whether they are truly cost-efficient for individual organizations.4 Hence, differences in the take-up of energy-efficient measures across organizations may – at least to some extent – be rationalized by cost heterogeneity. In addition, interviewees were also asked to evaluate the relevance of some potential barriers to energy efficiency within their organization. Thus, the survey data allow an assessment of the determinants for, and the barriers to, energy efficiency. However, since the prime objective of the survey was to gather detailed information on energy use, rather than to gain a deeper understanding of the nature of barriers to energy efficiency, the survey data does not capture all the barriers discussed in Section 2. Similarly, the variables used are proxies rather than perfect measures for those barriers. Finally, unlike in Brechling and Smith (1994), Scott (1997), DeCanio (1998), or DeCanio and Watkins (1998), the survey questions on perceived barriers to energy efficiency were not related to specific technologies, but rather to energy-efficient investments in general. Consequently, the analyses here focus on organizations' energy performance in general rather than on specific technologies.
4. The models Following Schleich and Gruber (2008), organizations were split into two types, “active” and “inactive” adopters of energy efficiency measures. Thereby, two definitions of “active/inactive” were used. In the first definition, an organization was termed “Active A&P” if it had adopted, or was planning to adopt, at least 50% of the set of energy efficiency measures which were deemed feasible for the individual organization.5 For example, if a particular organization had adopted, or was planning to adopt 5 of 9 feasible measures, this organization was termed “active”. The sets of feasible measures differ by sub-sector and 4 Nevertheless, if considered in economic-engineering type net present value (NPV) analyses, these measures are typically identified as “no-regret” measures (i.e. investments yielding a positive net present value) (see, e.g. chapters 6 and 7 in IPCC, 2007). 5 Of course, using a 50 % threshold is arbitrary. However, additional estimations using 40 % and 60 % as the threshold showed that the results are robust to variations in this share.
J. Schleich / Ecological Economics 68 (2009) 2150–2159
2155
Table 2 Overview of sub-sectors. Number of observations
Average number of employees
Average annual energy use [MWh]
Active in energy efficiency (adopted & planned measures) [%]
Active in energy efficiency (adopted measures) [%]
Small commercial businesses and trade Bakeries 88 Butcheries 76 Car repair industry 78 Construction 205 Laundries and dry cleaners 67 Metal industry 116 Retail trade 291 Wholesale trade 164 Wood working and processing 94
9 8 8 34 21 9 28 36 8
378 151 339 187 2148 125 455 808 209
51.14 81.58 53.85 51.71 58.21 37.07 40.21 37.80 53.19
37.50 76.32 39.74 33.66 34.33 30.17 26.46 28.66 36.17
Public and private services organizations Banks & insurance companies 144 Gastronomy 102 Hotel industry 128 Non-commercial organizations 125 Public administrations 93 Services 257 Average 135
146 10 17 30 70 30 34
1455 217 593 515 960 69 508
45.83 38.24 68.75 43.20 43.01 31.52 46.06
29.17 26.47 52.34 28.00 29.03 35.40 34.32
are explained in detail in Geiger et al. (1999). The second definition, “Active A”, only includes measures which organizations had actually adopted. Whether “Active A” or “Active A&P” more adequately reflects active adopters is a matter of subjective judgement. For comparison, Schleich and Gruber (2008) only used “Active A” as an indicator. Comparing the results for both indicators makes it possible to judge whether organizations under- or overestimate some barriers when it comes to implementing planned energy efficiency measures. The shares of “Active A&P” and “Active A” organizations for each sub-sector are also displayed in Table 2. Since no information on the profitability of the energy efficiency measures is available, however, the data do not allow the observed share of adopted energy efficiency measures to be compared with a profit-maximizing share. Hence, unlike DeCanio and Watkins (1998), the empirical results cannot be used to test the profitmaximization assumption of neoclassical economic theory. The dependent variable in the econometric models is dichotomous and takes the value of one if the organization is “active”. For “inactive” organizations, the dependent variable is zero. The explanatory variables available from the survey are: ENERKNOW: Split of final energy consumption into thermal energy and electricity consumption is unknown INFO: Lack of information about energy efficiency measures TIME: Lack of time to analyse potentials for energy efficiency UNCERTAIN: Energy costs may vary in the future PRIORITY: Other investment priorities RENTED: Organization space is rented PURCH: Organization automatically considers energy efficiency when purchasing new equipment ENERGY: Total annual specific energy consumption in kWh per employee SIZE: Number of employees in the organization ENERKNOW, INFO, TIME, UNCERTAIN, PRIORITY and RENTED are the dummy variables for the barriers. They assume the value of one if the associated statement is judged to be true for the organization, and zero otherwise. There is no one-to-one relationship between the barriers discussed in the conceptual Section 2 and the variables used as proxies in the empirical part. Instead, several variables reflect more than one type of barrier. ENERKNOW and INFO are primarily proxies for the barrier of imperfect information, TIME reflects hidden staff costs for energy management, UNCERTAIN captures financial risks, PRIORITY serves as a proxy for lack of capital resulting from internal priority setting within organizations, and RENTED stands for the splitincentives barrier. Procedures and management systems which automatically require energy efficiency to be considered for purchas-
ing are assumed to decrease information and other transaction costs. Further, such procedures may also serve as indicators of whether energy efficiency is integrated into the cultural and organizational procedures of an organization. Hence, PURCH is expected to have a positive effect on organizations' take-up of energy-efficient measures. The importance of energy consumption and energy costs to the organization are captured by the variable ENERGY. Thus, the expected sign for ENERGY is also positive. Finally, to control for size, specific energy consumption is used, that is, the total annual energy consumption is divided by the number of employees. Since larger organizations may cope better with certain barriers to energy efficiency such as imperfect information, transaction costs and bounded rationality, SIZE is also included as an explanatory variable and should increase the propensity of being “active”. In the actual specification, the natural log of specific energy consumption and number of employees is used.6 Two types of models are estimated. The “Sector model” explores the impact of the “explanatory” variables on a rather aggregated level. Effects which are specific to the N individual sub-sectors are captured through sub-sector dummy variables Di. The index function for the “Sector model” is specified as CONST + β 1 ⁎ENERKNOW + β2 ⁎INFO + β 3 ⁎TIME + β4 ⁎UNCERTAIN + β 5 ⁎PRIORITY + β 6 ⁎RENTED + β7 ⁎PURCH + β 8 ⁎ENERGY N X β 10i *Di + e + β 9 ⁎SIZE + ð1Þ i=1
The “Sub-sector model” allows for (sub) sector-specific differences in the relevance of individual barriers across organizations. In this way it is possible to identify sectors in which particular barriers are statistically significant. To model these possible interaction effects, sector dummies Di are multiplied by the barrier dummies. In a similar way, sector-specific responses to energy use are captured. Including these additional variables, however, comes at the cost of
6 Following Scott (1997), the explanatory variables may be grouped into two sets, one containing rather subjective information (INFO, TIME, UNCERTAIN, PRIORITY) and the other more objective information (ENERKNOW, RENTED, PURCH, ENERGY, SIZE). Including energy prices as an additional (objective) explanatory variable would make it possible to capture economic incentives to reduce energy use. The survey, however, did not ask for electricity or heat prices. Besides, since these prices are unlikely to vary significantly in the cross-sectional data set used, the explanatory power of including energy prices is expected to be rather small.
2156
J. Schleich / Ecological Economics 68 (2009) 2150–2159
lower degrees of freedom. The index function for the “Sub-sector model” is then
CONST + + +
N X i=1 N X
N X
β1i ⁎Di ⁎ENERKNOW +
i=1
N X
β4i ⁎Di ⁎UNCERTAIN + β7i ⁎Di ⁎PURCH +
i=1
N X
N X
β2i ⁎i Di ⁎INFO +
i=1
β5i ⁎Di ⁎PRIORITY +
i=1
N X
β3i ⁎Di ⁎TIME
INFO
β6i ⁎Di ⁎RENTED
β8i ⁎Di ⁎ENERGY + β9 SIZE +
i=1
ENERKNOW
i=1 N X i=1
Table 3 Logit estimation results for the sector model.
N X
β10i ⁎Di + e:
TIME UNCERTAIN
i=1
ð2Þ Eqs. (1) and (2) were estimated as standard Logit and Probit models using STATA 9. 5. Estimation results Estimation results are shown in Table 3 for the “Sector model” and in Table 4 for the “Sub-sector model”.7 The tables list the parameter estimates together with their standard errors (in parentheses). First, the findings for the “Sector model” are analyzed. 5.1. Results for the Sector model Table 3 implies that the estimation results for both types of independent variable “Active A” and “Active A&P” are quite similar for most parameters in terms of magnitude and statistical significance. The estimated models explain about 15% of the variation in the dependent variable (goodness of fit) as captured by the coefficient of determination (Pseudo R2). Considering that the data is crosssectional and that organizations in the commercial and services sector tend to be heterogeneous, the estimated regression equations account for a relatively high percentage of the variation. With regard to the individual explanatory variables, the parameter estimate associated with ENERKNOW exhibits a negative sign and is found to be statistically significant regardless of whether adopted and planned measures (“Active A&P”) or only adopted measures (“Active A”) are used to construct the independent variable.8 Hence, poor information on energy consumption patterns prevents energy-saving measures being identified and evaluated properly. In contrast, since INFO is far from being statistically different from zero for both specifications, the findings at the sectoral level do not confirm that lack of information about energy-efficient measures acts as a barrier. The parameter estimates associated with TIME show a negative and statistically significant impact on the take-up of energy efficiency, corroborating the hypothesis that staff's lack of time to analyse energy efficiency potentials is a barrier.9 In contrast, the finding for UNCERTAIN, which is not statistically significant, does not support the hypothesis that the financial risk associated with uncertain future energy prices acts as a barrier to organizations' take-up of energy-efficient technologies. Next, PRIORITY exhibits a negative sign for “Active A” and “Active A&P”, but is only statistically significant for the former. As pointed out in Section 2, investments in energy efficiency may be crowded out by other investments because of internal capital 7 Since the results for the Probit and the Logit procedures are almost identical, only those for the Logit procedure are shown. To further save space and because the estimation results for both types of independent variable “Active A” and “Active A&P” are quite similar, only the findings for “Active A&P” are displayed for the “Sub-sector model”. Finally, the parameter estimates for the sub-sector dummies β10i are not reported in Tables 3 and 1. All results not reported here are available from the author upon request. 8 In this paper, “statistically significant” means statistically significant at the 10% level or lower, i.e. the P-values (not reported to save space) are no greater than 0.1. 9 Arguably, TIME and INFO reflect similar concepts such as hidden staff costs of energy management. Consequently, collinearity may result in high standard errors rendering parameter estimates insignificant. The correlation coefficient between TIME and INFO of 0.36 indicates that both variables are indeed positively correlated, but the relatively small value together with the large sample size do not indicate that collinearity is a problem.
PRIORITY RENTED PURCH ENERGY SIZE Sample size Log likelihood Pseudo R2
Active A&P
Active A
− 0.453⁎⁎ (0.102) 0.054 (0.116) − 0.421⁎⁎ (0.113) 0.064 (0.104) − 0.087 (0.106) − 1.246⁎⁎ (0.124) 0.226 (0.176) 0.220⁎⁎ (0.056) 0.340⁎⁎ (0.045) 2028 − 1198 0.144
− 0.449⁎⁎ (0.046) − 0.073 (0.126) − 0.250⁎ (0.122) 0.034 (0.110) − 0.273⁎ (0.114) − 1.150⁎⁎ (0.148) 0.147 (0.200) 0.188⁎⁎ (0.060) 0.372⁎⁎ (0.046) 2028 − 1077 0.150
⁎Individually statistically significant at least at 10% level. ⁎⁎Individually statistically significant at least at 1% level.
allocation procedures, or because of “soft” criteria such as strategic priorities, or the status of energy management. The finding that PRIORITY is only statistically significant for “Active A” but not for “Active A&P” suggests that respondents may underestimate the relevance of internal priority setting for energy efficiency projects which have been planned, but not yet realized.10 The dummy variable RENTED is also statistically significant and negative, supporting the hypothesis that split incentives are a barrier to energy efficiency. Further, PURCH exhibits the hypothesized positive sign, but is not statistically significant. Hence, at the sectoral level, the findings do not support the notion that integrating energy efficiency as a criterion into purchasing decisions reduces barriers to energy efficiency.11 ENERGY is significant and exhibits the expected positive sign: the economic incentives to save energy are stronger for higher consumption levels. Finally, SIZE is also found to be statistically significant. The positive sign supports the view that larger organizations are more able to cope with imperfect information, credit constraints, risk and uncertainty, or bounded rationality. In sum, the results are robust to both estimation procedures and to the choice of independent variables. To the extent that the variables included in the model reflect the barriers presented in Section 2, the findings generally support the hypotheses developed therein. 5.2. Results for the Sub-sector model The “Sub-sector model” allows for (sub)sector-specific differences in the relevance of individual barriers and energy use across organizations through the interaction of sector-specific dummies with these variables. The parameter associated with SIZE is assumed to be identical for all sub-sectors. The first column in Table 4 contains results which relate to all sectors, while estimates for the interaction parameters are shown in the matrix part of Table 4. For example, the first row displays the results for the interaction of ENERKNOW with the sub-sector dummies, i.e. −0.630 is the estimated value for the parameter associated with ENERKNOW in the sub-sector retail trade.
10 Also note that the point estimate for the parameter is about three times larger (in absolute terms) for “Active A” compared to “Active A&P”. 11 A possible explanation is that, since PURCH takes the value of one for more than 90 % of the observations, there is only little variation to be utilized by the regression analyses and the statistical relation between PURCH and the dependent variables is weak.
Table 4 Logit estimation results for “Active A&P” for the sub-sector model. Small commercial businesses and trade
Public and private services organizations
Bakeries Butchers Car repair Construction Laundries Metal ENERKNOW
TIME UNCERTAIN PRIORITY RENTED PURCH ENERGY
− 0.806 (0.683) − 0.524 (0.764) 0.130 (0.883) 1.018 (0.834) − 1.490⁎ (0.877) − 0.750 (0.695 − 0.451 (1.231) − 0.401 (0.483)
− 1.244⁎ (0.545 0.821 (0.625 0.047 (0.571 − 1.191⁎ (0.551 − 0.128 (0.548 − 1.244⁎ (0.656 0.119 (1.027 0.179 (0.311
SIZE
0.355⁎⁎ (0.049) CONSTANT 5.516 (5.134) Sample size 2028 Log likelihood − 1143 2 Pseudo R 0.1834
− 0.240 (0.304) − 0.025 (0.354) − 0.437 (0.341) − 0.078 (0.309) 0.226 (0.325) − 0.917⁎ (0.447) − 0.505 (0.510) 0.092 (0.134)
− 0.536 (0.626) − 0.191 (0.759) 0.074 (0.679) − 0.683 (0.659) − 0.202 (0.622) − 0.902 (0.757)
0.781⁎ (0.423)
− 0.716 (0.481) 0.521 (0.545) − 0.236 (0.543) 0.739 (0.485) − 0.604 (0.475) − 1.506⁎⁎ (0.517) 0.632 (0.759) 0.284 (0.257)
Non-commerc. org Public admin. Services
− 0.630⁎ (0.289) 0.305 (0.343) − 0.168 (0.331) − 0.159 (0.296) − 0.131 (0.311) − 1.783⁎⁎ (0.319) 1.070⁎ (0.578) 0.380⁎ (0.169)
− 0.526 (0.403) − 0.246 (0.547) − 0.935⁎ (0.463) − 0.497 (0.404) − 0.092 (0.408) − 1.256⁎ (0.747)
− 0.857⁎ (0.402) 1.030⁎ (0.494) − 0.652 (0.494) − 0.400 (0.434) 0.537 (0.456) − 2.092⁎⁎ (0.567) 0.496 (0.794) 0.238 (0.195)
− 0.618 (0.470) 0.185 (0.548) − 0.714 (0.492) 0.598 (0.472) − 0.544 (0.506) − 0.416 (0.622) 1.042 (0.778) 0.090 (0.254)
0.299 (0.400) − 0.452 (0.431) − 0.565 (0.419) 1.021⁎⁎ (0.398) − 0.265 (0.453) − 0.800⁎ (0.437) − 0.377 (0.568) 0.115 (0.213)
− 0.928⁎ (0.499) − 0.629 (0.578) 0.017 (0.434) − 0.547 (0.483) 0.237 (0.481) − 1.083⁎ (0.487) 0.679 (1.237) 0.104 (0.293)
− 0.305 (0.420) 0.055 (0.488) − 0.768 (0.470) 0.906⁎ (0.457) − 0.010 (0.447) − 0.970⁎ (0.567) 1.045 (0.750) − 0.013 (0.303)
0.281 (0.236)
− 0.867⁎ (0.511) − 1.457⁎ (0.718) 0.479 (0.610) 0.285 (0.495) − 1.099⁎ (0.503) − 1.748⁎ (0.851) 0.205 (0.792) 0.319 (0.289)
− 0.127 (0.311) 0.231 (0.341) − 0.880⁎⁎ (0.331) − 0.216 (0.312) 0.552 (0.321) − 1.730⁎⁎ (0.330) 0.115 (0.412) 0.355⁎⁎ (0.135)
J. Schleich / Ecological Economics 68 (2009) 2150–2159
INFO
0.431 (0.541) − 0.172 (0.528) − 1.461⁎ (0.583) 0.364 (0.539) 0.199 (0.548) − 0.719 (0.604) − 0.585 (1.024) 0.290 (0.313)
Retail trade Wholesale trade Wood working Banks & insur. comp. Gastronomy Hotels
⁎Individually statistically significant at least at 10% level. ⁎⁎Individually statistically significant at least at 1% level.
2157
2158
J. Schleich / Ecological Economics 68 (2009) 2150–2159
Hence, lack of knowledge about energy consumption patterns exhibits the expected negative sign for the vast majority of the sub-sectors and is statistically significant for car repair, retail trade, wholesale trade, gastronomy and public administrations. Results for the second row imply that lack of information about energy efficiency measures (INFO) is only a statistically significant barrier to energy efficiency in the public administrations sub-sector. Lack of time to analyse energysaving potentials (TIME) is found to be statistically significant in two of the five sub-sectors falling under public and private services (i.e. non-commercial organizations and services), but only in one of the nine sub-sectors belonging to small commercial businesses and trade (i.e. bakeries). UNCERTAIN turns out to be statistically significant in three sub-sectors, twice in the services sectors with a positive sign (banks & insurance companies, hotels) and once with a negative sign (car repair). This finding is consistent with the theoretical concepts presented in Section 2.3, which present arguments explaining why the impact of uncertainty stemming from stochastic energy prices is generally ambiguous. Such countervailing effects would also rationalize the finding for the “Sector model”, where UNCERTAIN turned out to be not statistically significant. The findings for PRIORITY imply that internal priority setting slows down the diffusion of energy efficiency measures in the sub-sectors butcheries and public administration. The split-incentives problem arising from rented office space (RENTED) is the barrier most frequently found to be statistically significant. Subsectors where RENTED is not significant are primarily those where renting buildings or office space is not very common. Table 4 further indicates that integrating energy consumption into purchasing procedures (PURCH) positively affects the diffusion of energy efficiency only in the retail trade. For all other sub-sectors, PURCH is not statistically significant.12 Specific energy use (ENERGY) shows the expected positive sign in all but two sub-sectors and is found to be statistically significant in three sub-sectors. Finally, the result for SIZE is almost identical to that in the “Sector model”. In sum, only a few of the variables identified as statistically significant in the “Sectoral model” are also statistically significant for many sub-sectors. Hence, the strength of the barriers is likely to be different across sectors. When comparing the results between models, however, it has to be kept in mind that there are considerably fewer degrees of freedom in the “Sub-sector model” than in the “Sector model”, leading to larger standard errors in the former. Further, all parameters which are statistically significant exhibit the expected signs — except for INFO for the sub-sector wholesale trade. Thus, the results for the “Sub-sector model” are generally consistent with those of the “Sector model”. The results for the “Sub-sector model” further indicate that the types and numbers of barriers to energy efficiency vary across sub-sectors. For most sub-sectors, one or two barriers are statistically significant, but no clear pattern appears to exist regarding combinations of barriers. As expected, the more energy-intensive sectors like laundries and bakeries tend to exhibit the least number of barriers.13 In comparison, the largest number of barriers is found in the public administrations sector.14
12 For the sub-sectors laundries and non-commercial services, all or almost all companies had integrated energy efficiency into purchasing procedures. Since a constant is included, no interaction term for PURCH was included for these sub-sectors to prevent the regressor matrix from becoming (near) singular. 13 An appropriate indicator for energy intensity would be “energy use per employee”. This measure can be derived by dividing “Average annual energy use” by the “Number of employees” in Table 2. Accordingly, the five most energy-intensive sectors in the sample are “Laundries and dry cleaners”, “Car repair industry”, “Bakeries”, “Hotel industry” and “Wood working and processing”. 14 The findings from case studies by Sorrell et al. (2004, Chapter 3) rationalize this outcome. Several types of split-incentives problems were found to inhibit energy efficiency in public administrations in Germany. Usually, there are no arrangements for the decentralized accountability of energy costs. Instead, these costs are typically paid out of the administrations' general budget for operational costs. Further, since public budgeting laws limit the transfer of energy cost savings to other areas, energy-saving units are not necessarily able to also appropriate the associated cost savings.
6. Conclusions Exploring the relevance of various barriers to energy efficiency based on a large sample for the German commercial and services sector indicates that most of the barriers considered turned out to be statistically significant at an aggregate level (“Sector model”). In general, therefore, these results complement the findings from case study analyses and the existing survey-based studies (e.g. DeCanio, 1998; DeCanio and Watkins, 1998; de Groot et al., 2001; Schleich and Gruber, 2008). In the light of the discussion of the various barriers in Section 2, it has to be kept in mind that the data available only allowed a subset of these barriers to be captured. Results from the “Sub-sector model” yield a more heterogeneous picture. The numbers and types of relevant barriers vary across subsectors, and most sub-sectors are subject to relatively few of the barriers explored. Allowing for sector-specific responses, these results help to identify sub-sectors where a policy measure is likely to be effective. Since individual measures are usually not able to address several targets, a mix of different policies would be required to tackle the multiple types of barriers (see also Jochem and Gruber, 1990; Gruber and Brand, 1991). While split incentives (i.e. the landlord/tenant problem) have previously been identified as a barrier to energy efficiency in the private housing sector (e.g. Scott, 1997), the findings in this paper imply that this is also the case for the commercial and services sector in Germany. Thus, the implementation of the EU Directive on the Energy Performance of Buildings (European Parliament and Council, 2003), which requires an energy performance certificate when renting or selling existing properties in the household, commercial and public sectors from 2008 on, can be expected to improve energy performance in the commercial and services sector. In particular, since this energy certificate includes key data on energy consumption and must be made available to tenants and clients, it lowers the transaction costs for assessing the energy performance of properties, and thus directly addresses the source of the landlord/tenant problem as discussed in Section 2.5. Lack of information about energy consumption patterns is found to be a barrier to energy efficiency in about one third of the sub-sectors. Installing metering devices and implementing energy management systems would help overcome this barrier. Further findings of the “Sub-sector model” indicate that relatively few sub-sectors suffer from insufficient information about energy efficiency measures, or lack of time to analyse potential energy savings. In principle, contract energy service management via energy service companies is designed to overcome these barriers (e.g., Sorrell et al., 2004; Mills, 2003; Ostertag, 2003). With the exception of priority setting, the empirical findings presented in this paper do not depend on whether the indicator to determine organizations' energy efficiency performance only includes measures which have already been realized, or planned ones as well. The findings of the “Sector model” suggest that respondents underestimate internal priority setting as a barrier to planned projects. Following the argument by Cooremans (2007) or Teece et al. (1997), adequate measures would have to change upper management's priorities and get them to stop regarding energy services as “non strategic” or as belonging to the organization's lowly valued material resources. At the sectoral level, lower specific energy consumption is found to slow down the take-up of energy efficiency in the German commercial and services sector. Similar to the findings of DeCanio (1998), the “Sub-sector model” results imply that increasing energy costs via energy taxes or via the European Union Emissions Trading System for greenhouse gas emissions may accelerate the diffusion of energy-efficient technologies only in the more energyintensive sub-sectors. In addition, the results by Newell et al. (1999) imply that price policies will be more effective when applied in combination with other policies such as standards, labelling or information campaigns. While the findings of this paper generally support the view that barriers help to explain the lack of investment in energy-efficient (and arguably also cost-efficient) technologies, it is beyond its scope, however, to assess whether these barriers should be overcome because they inhibit economic efficiency. For example, the costs for installing
J. Schleich / Ecological Economics 68 (2009) 2150–2159
meters may be prohibitively high in sub-sectors with low energy costs. Further, from the perspective of neoclassical economics, the inability to access capital may well constitute a barrier, but it need not imply a failure in capital markets that should be corrected. If small companies are considered high-risk borrowers, potential lenders may demand a high risk-adjusted rate of return (Sutherland, 1996). In this case, the market outcome is efficient and policy interventions are not justified. In addition, if energy-efficient technologies have an inferior technical and economic performance compared to alternatives, it would be warranted to impose strict investment criteria or restrict capital budgets for energy efficiency investments within organizations. Finally, there could be hidden costs which are true economic costs associated with investments in new energy-efficient technologies such as production interruptions, or costs for training staff. A rational and economically efficient technology choice would have to take these types of costs into account, and they do not justify policy interventions. Ultimately, even if the barriers identified are also found to inhibit economic efficiency, policies to overcome those barriers would still have to be cost-efficient. Thus, ex ante cost–benefit analyses would have to be conducted for any proposed energy efficiency policies. Alternatively, policy makers could also draw on experiences gained from existing programmes. Increased data availability and the continuing diffusion of micro-econometric evaluation techniques are expected to reduce the uncertainty about how many kilowatt hours and CO2-emissions can actually be saved due to a particular policy intervention. Acknowledgements The author is particularly grateful for insightful suggestions made by three anonymous reviewers. The paper also benefited from discussions at the 31st Annual Conference of the International Association of Energy Economists (IAEE) in 2008 in Istanbul, Turkey. Final thanks go to Gillian Bowman-Koehler for proof-reading the paper. The usual disclaimer applies. References Akerlof, G., 1970. The market for lemons. Quarterly Journal of Economics 84, 488–500. Arbeitsgemeinschaft Energiebilanzen e. V., 2006. Auswertungstabellen zur Energiebilanz für Deutschland 1990 bis 2005. BMU (Bundesministerium für Umwelt, Naturschutz und Reaktorsicherheit) (Federal Ministry for Environment, Nature Conservation and Nuclear Safety), 2007. Revidierter Nationaler Allokationsplan 2008–2012 für die Bundesrepublik Deutschland, Berlin. Ben-David, S., Brookshire, D., Burness, S., McKee, M., Schmidt, C., 2000. Attitudes toward risk and compliance emission permit markets. Land Economics 76, 590–600. Brechling, V., Smith, S., 1994. Household energy efficiency in the UK. Fiscal Studies 15, 44–56. Brown, M.A., 2001. Market failures and barriers as a basis for clean energy policies. Energy Policy 29, 1197–1207. Butler, R., Davies, L., Pike, R., Sharp, J. (Eds.), 1993. Strategic Investment Decisions. Routledge, London. Carr, C., Tomkins, C., 1998. Context, culture and the role of the finance function in strategic decisions. A comparative analysis of Britain, Germany, the U.S.A. and Japan. Journal of Management Accounting Research 9, 213–239. Coase, R., 1991. The nature of the firm. In: Williamson, O.E., Winter, S. (Eds.), The Nature of the Firm. Origins, Evolution, and Development. Oxford University Press, New York, pp. 18–33. Cooremans, C., 2007. Strategic fit of energy efficiency. European Council for EnergyEfficient Economy (Paris): Proceedings of the 2007 ECEEE Summer Study. Saving Energy — Just Do It! La Colle sur Loup, Côte d'Azur, France. de Almeida, E.L.F., 1998. Energy efficiency and the limits of market forces: the example of the electric motor market in France. Energy Policy 26, 643–653. DeCanio, S.J., 1994. Agency and control problems in US corporations: the case of energy efficient investment projects. Journal of the Economics of Business 1, 105–123. DeCanio, S.J., 1998. The efficiency paradox: bureaucratic and organizational barriers to profitable energy saving investments. Energy Policy 26, 441–454. DeCanio, Watkins, W.E., 1998. Investment in energy efficiency: do the characteristics of firms matter? Review of Economics and Statistics 80, 95–107. de Groot, H.L.F., Verhoef, E.T., Nijkamp, P., 2001. Energy savings by firms: decisionmaking, barriers and policies. Energy Economics 23, 717–740. Dixit, A.K., Pindyck, R.S., 1994. Investment Under Uncertainty. Princeton University Press, Princeton. European Commission, 2006. Action plan for energy efficiency: realising the potential. COM(2006)545 final.
2159
European Council, 2007. Brussels European Council 8/9 March 2007 Presidency Conclusions, 7224/1/07 REV 1, Brussels. European Parliament and Council, 2003. Directive 2002/91/EC of the European Parliament and of the Council of 16 December 2002 on the energy performance of buildings. Official Journal of the European Communities L1/65–71. Geiger, B., Gruber, E., Megele, W., 1999. Energieverbrauch und Energieeinsparung in Gewerbe, Handel und Dienstleistung. Physica, Heidelberg. Golove, W.H., Eto, J.H., 1996. Market barriers to energy efficiency: a critical reappraisal of the rationale for public policies to promote energy efficiency. University of California, Lawrence Berkeley Laboratory report LBL-38059. Gruber, E., Brand, M., 1991. Promoting energy conservation in small and medium-sized companies. Energy Policy 19, 279–287. Hassett, K.A., Metcalf, G.E., 1993. Energy conservation investment: do consumers discount the future correctly? Energy Policy 21, 710–716. Holmstrom, B., 1979. Moral hazard and observability. Bell Journal of Economics 19, 74–91. Howarth, R.B., Andersson, B., 1993. Market barriers to energy efficiency. Energy Economics 15, 262–272. Howarth, R.B., Sanstad, A.H., 1995. Discount rates and energy efficiency. Contemporary Economic Policy 13, 101–109. Intergovernmental Panel on Climate Change (IPCC), 2007. Climate change 2007: mitigation. In: Metz, B., Davidson, O.R., Bosch, P.R., Dave, R., Meyer, L.A. (Eds.), Contribution of Working Group III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom. International Energy Agency (IEA), 2007. Mind the Gap — Quantifying Principal-Agent Problems in Energy Efficiency. OECD/IEA, Paris. International Energy Agency (IEA), 2008. World Energy Outlook 20088. OECD/IEA, Paris. Jaffe, A.B., Stavins, R.N., 1994a. Energy-efficiency investments and public policy. The Energy Journal 15, 43–65. Jaffe, A.B., Stavins, R.N., 1994b. The energy-efficiency gap: what does it mean? Energy Policy 22, 804–810. Jochem, E., Gruber, E., 1990. Obstacles to rational electricity use and measures to alleviate them. Energy Policy 18, 340–350. McDonald, R., Siegel, D., 1986. The value of waiting to invest. Quarterly Journal of Economics 101, 707–727. Mills, E., 2003. Risk transfer via energy-savings insurance. Energy Policy 31, 273–281. Morgan, G., 1985. Images of Organisation. Sage, London. Newell, R.G., Jaffe, A.B., Stavins, R.N., 1999. The induced innovation hypothesis and energy-saving technological change. Quarterly Journal of Economics 114, 941–975. Ostertag, K., 2003. No-Regret Potentials in Energy Conservation: An Analysis of their Relevance, Size and Determinants. Physica, Heidelberg. Ramesohl, S., 1998. Successful implementation of energy efficiency in light industry. In: Lawrence Berkeley National Laboratory (Ed.), Proceedings of the International Workshop on Industrial Energy Efficiency Policies: Understanding Success and Failure, International Network for Energy Demand Analysis in the Industrial Sector (INEDIS). Schleich, J., Gruber, E., 2008. Beyond case studies: barriers to energy efficiency in commerce and the services sectors. Energy Economics 30, 449–464. Schleich, J., Böde, U., Köwener, D., Radgen, P., 2001. Chances and barriers for energy services companies. In: European Council for an Energy-Efficient Economy (Ed.), A Comparative Analysis for the German Brewery and University Sectors. Proceedings of the 2001 ECEEE Summer Study. ADEME, Paris. Schmid, C., Brakhage, A., Radgen, P., Layer, G., Arndt, U., Carter, J., Duschl, A., Lilleike, J., Nebelung, U., 2003. Möglichkeiten, Potenziale, Hemmnisse und Instrumente zur Senkung des Energieverbrauchs branchenübergreifender Techniken in den Bereichen Industrie und Kleinverbrauch. Report by the Fraunhofer Institute for Systems and Innovation Research, and Forschungsstelle für Energiewirtschaft, Karlsruhe and München. Scott, S., 1997. Household energy efficiency in Ireland: a replication study of ownership of energy saving items. Energy Economics 19, 187–208. Simon, H.A., 1957. Models of Man. Wiley, London. Simon, H.A., 1959. Theories of decision-making in economics and behavioural science. The American Economic Review 49, 253–283. Sorrell, S., 2003. Making the link: climate policy and the reform of the UK construction industry. Energy Policy 31, 865–878. Sorrell, S., O'Malley, E., Schleich, J., Scott, S., 2004. The Economics of Energy Efficiency — Barriers to Cost-effective Investment. Edward Elgar, Cheltenham. Stern, P.C., 1986. Blind spots in policy analysis: what economics doesn't say about energy use. Journal of Policy Analysis and Management 5, 200–227. Sutherland, R.J., 1996. The economics of energy conservation policy. Energy Policy 24, 361–370. Teece, D., Pisano, G., Shuen, A., 1997. Dynamic capabilities and strategic management. Strategic Management Journal 18, 509–533. van Soest, D.P., Bulte, E.H., 2001. Does the energy-efficiency paradox exist? Technological progress and uncertainty. Environmental and Resource Economics 18, 101–112. Wuppertal Institute, AKF-Institute for Local Government Studies, Energie-verwertungsagentur, Fraunhofer Institute for Systems and Innovation Research, Institute for Psychology at the University of Kiel, Amstein & Walthert, Bush Energie, 1998. InterSEE (Interdisciplinary Analysis of Successful Implementation of Energy Efficiency in Industry, Service, and Commerce), Final Report in 5 Volumes. Wuppertal, Copenhagen, Vienna, Karlsruhe, Kiel, Zürich, Felsberg. Yin, R.K., 1994. Case Study Research: Design and Methods. Sage Publications, London.
