Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
The Effects of information on Energy-Efficient investment Lack of adequate information is frequently blamed for the slowness of consumers' responses to the rapid price increases for energy over the past decade. The assump- tion of economic rationality generally followed in formal energy demand analysis emphasizes the role of information: under that assumption, rational action requires full information about the available alternatives and their relative costs and benefits. Energy prices provide some of this information, since they show the economic effect of using different amounts of different fuels. But prices do not provide information about the effects of particular actions on energy use. If consumers do not know how much energy they can save with attic insulation, or that check- ing for leaks in industrial boilers can cut a business's energy use by 10 percent, or that two refrigerators of the same size may take vastly different amounts of electricity to run, they will fail to take actions that would save them money. In a period of rapidly changing energy prices and of threats to the availability of fuels, it is important for energy users to know how their choices will affect their energy consumption. To help consumers make choices in their own interest, government has developed and dissemi- nated huge amounts of information about the effects of actions on energy use. It has calculated the fuel economy of automobiles and distributed millions of copies of the ratings to prospective automobile purchasers; it has sponsored programs involving energy audits of homes and nonresidential buildings; it has supported research on the effects of new energy-saving technologies on energy use; and it has published reports of the findings of such research. This information can make a difference. The automobile fuel economy labels required by the Environ- 61
62 mental Protection Agency, for example, were rated the most useful source of gasoline mileage information by new car purchasers in 1978 and 1979, a period in which more than two-thirds of purchasers rated fuel economy as a "very important" or "extremely important" factor in their pur- chase decisions (McNutt and Rucker, 1981). Because information is necessary for an economically rational response to energy price signals, because govern- ment has relied heavily on information to promote energy efficiency, and because information can be an effective policy tool, energy demand analysis that cannot assess the effects of information is seriously incomplete. This chapter summarizes existing knowledge about what informa- tion energy users have and about how policy makers can provide better information for consumers; it also dis- cusses two conceptual approaches for analyzing energy demand based on a recognition that energy information may be incomplete. HOW COMPLETE IS CONSUMER:; ' INFORMATION? In formal demand models, the simplest assumptions about energy information are that full information is available to consumers and that they use it: rational economic action follows. The converse is also true: models that assume rational economic action also assume, implicitly or explicitly, that energy users are acting with full information. But full information is not always available. The economic cost of future energy is unknown, and even the effects of action on energy use are often uncertain. In commercial buildings, for example, actual energy savings from particular sets of technical improvements are often 50 percent more than predicted or 80 percent less than predicted from the best available technical information (Office of Technology Assessment, 1982). Furthermore, the information that is available is not necessarily used when energy users act. Evidence shows that people are often unaware of available information; when they are aware of it, they often fail to act on it because of inertia, lack of comprehension, mistrust, or some other reason (for a discussion of the evidence on this point, see Stern and Aronson, 1984). To cite one of many examples, home energy audits have been offered to residential customers at low cost or even for free, but typically fewer than 5 percent of the eligible consumers
63 request the audits. And of those who have had energy audits, only a minority take energy-saving action as a result. There are several reasons for this, mostly apart from the accuracy of the information: people may not trust the source of the information or the ability of con- tractors to do the work well; they may find it too diffi- cult to gather all the information necessary to decide which investments to make; or they may not have the nec- essary capital to invest in energy-efficient technology. Such problems with energy information constitute a major barrier to the penetration of energy-efficient technologies. Although it is impossible to estimate the size of the barrier with precision, two kinds of findings offer a rough gauge of its magnitude. Numerous experi- mental studies of one informational procedure, feedback on energy use, find that it yields average reductions in energy use of around 10 percent--and up to almost 20 per- cent when energy costs are high--without any investment in technology (Winkler and Winett, 1982; for a review of the literature, see Geller, Winett, and Everett, 1982: Chapter 5). And studies of residential energy conserva- tion programs that act largely by offering information show that under some conditions participants invest twice as much in energy efficiency as nonparticipants during the same period (Stern, Black, and Elworth, 1982a) or save twice as much energy (Hirst, White, and Goeltz, 1983b). Thus, it is safe to assert that improved delivery of information could make up a significant part of the dif- ference between the level of adoption of energy-efficient technology under present conditions and the level of adoption that is economic for energy users. Information could certainly speed the rate of adoption, and it might also bring the final level of adoption closer to the eco- nomic optimum. Treatment of Information in Formal Models Formal demand models have little to say about what people do with incomplete information; how they respond to dif- ferent kinds of information; or how they deal with an environment in which information is confusing, conflict- iIt is difficult to conclusively evaluate what propor- tion of this effect is attributable to the self-selection of program participants.
64 ing, and often untrustworthy. Instead, formal demand models ordinarily subsume the effects of information under other explanatory concepts. For example, lag coefficients and other indices of investment dynamics can be inter- preted in part as mathematical expressions of the slow spread of full information among energy users; whether or not they are so interpreted, such a process shows up as a dynamic effect in models. If information speeds adoption, it changes such lag coefficients. Another example is the concept of the discount rate (discussed in Chapter 1). When estimated from data on purchases of energy-efficient equipment, the discount rate reflects not only a preference for present value over future value, but energy consumers' responses to imper- fections in the available information and failure to believe and act on information that may in fact be accu- rate. Improvements in information or in consumers' use of it would then appear to change the discount rate in models that estimate that variable. Knowledge About Information From Problem-Oriented Research Surveys of energy users' beliefs and knowledge, small- scale experimental studies involving energy information, and formal evaluation studies of information-based energy programs offer some insights about what information con- sumers have and how informational policies and programs can change what energy users know and what they do. The evidence is that energy users possess incomplete information for making energy choices. Furthermore, sev- eral studies have shown systematic misconceptions among energy users about the amounts of energy used by various household appliances (Becker, Seligman, and Darley, 1979; Kempton, Harris, Keith, and Weihl, 1982; Mettler-Meibom and Wichmann, 1982). People tend to overestimate the amount of energy consumed by household lighting and tele- visions and to underestimate the energy used by water heaters and some other appliances. Parallel misconcep- tions exist about what actions can save energy in the home: a survey of 400 Michigan families found that the average householder believed reduced lighting could save twice as much energy as reduced use of hot water (Kempton, Harris, Keith, and Weihl, 1982). If people act on such beliefs, the ultimate penetration of energy-efficient water heaters is likely to fall far short of what would be in homeowners' self-interest.
65 There is a pattern to misconceptions about home energy use. Generally, overestimation occurs for energy uses that are visible or that must be activated by hand each time they are used; underestimation occurs for energy uses that do not have these characteristics. This pattern follows what cognitive psychologists call the availability heuristic (Tversky and Kahneman, 1974): people tend to overestimate the frequency or importance of events that are easily called to mind. It is worth emphasizing that people's misconceptions persist despite the availability of other, presumably more accurate, information. The misconceptions can persist even when accurate information is delivered directly to people who are misinformed. In the Michigan study, for example, householders who received computerized energy audits of their homes with specific recommendations for energy-saving activity had virtually the same patterns of belief about what saves energy as people who had not received the energy audits. Improving Consumers' Information At least one readily quantifiable factor is known to mediate the impact of information on action--the cost of energy (Stern and Aronson, 1984). This is best demon- strated by the research on the effect of frequent feedback about household energy use on future energy use (reviewed by Winkler and Winett, 1982). As noted above, energy savings achieved in feedback experiments are a function of the cost of energy to the household. When households have low energy costs, feedback has no effect, but with high energy costs, the savings have been as high as 20 percent. Corroboration of the relationship between price and information is available from research using an entirely different methodology. Using simple econometric models to estimate the effect on national energy consump- tion of expenditures by the U.S. Department of Energy on conservation programs, a group at Oak Ridge National Laboratory found that federal expenditures were signifi- cantly related to energy use only in interaction with energy price (Greene, Hirst, Soderstrom, and Trimble, 1982). One interpretation of these findings is that information matters only when prices are relatively high. The other side of this is more to the point at present, now that prices are high: the higher energy prices rise, the more important is the quality of the energy informa- tion that consumers have.
66 But to know that well-delivered information makes a difference does not tell how to deliver it well. The evidence from numerous program evaluation studies and controlled experiments shows that merely making energy information available to people is not enough: offering printed advice on how to save energy, for example, usually has little or no effect on behavior (for reviews of the experimental research, see Ester and Winett, 1982; Geller, Winett, and Everett, 1982; Shippee, 1980; Winett and Neale, 1979). The evidence suggests, however, that energy information can make a difference if it is presented in an attractive format and an easily understandable style, if it is vivid and personalized, if it is clearly relevant to the particular energy user's situation, if it is avail- able through several media (prominently including word- of-mouth), and if it comes from a trusted source (for a discussion of the evidence on these points and the analy- sis leading to these conclusions, see Stern and Aronson, 1984). Unfortunately for the purposes of formal policy analy- sis, none of these factors is readily quantifiable. Yet some of them can have a sizable effect on energy use. For example, people who viewed a videotaped information pro- gram constructed on many of the above principles saved 10 to 20 percent of household energy in comparison with con- trol households (Winett et al., 1982). With additional information in the form of energy use feedback, energy savings increased to around 15 to 25 percent. These savings did not involve any investments in improved energy efficiency. In energy audit programs, which use information to promote energy-efficient investments, the evidence suggests that a program's effectiveness depends on such qualitative variables as the program's promotional efforts, convenience and consumer protection features, and ability to gain trust (as discussed in Chapter 3). Other important qualitative variables include the effectiveness of energy auditors and other program personnel as communi- cators and various features of the ways recommended investments are described to the program's clients (for further discussion of these issues, see Stern and Aronson, 1984). The net effect of programs for residential energy efficiency that rely largely on information can sometimes be substantial. One example illustrates the potential. A residential conservation program operating in the North- east in 1979 offered free energy audits, access to
67 approved contractors, inspection of all installed improve- ments, and access to reduced-interest loans. Of home- owners requesting the energy audits, about 20 percent had work done under the program, and they did virtually all the work the auditors recommended as economically justi- fied (Stern, Black, and Elworth, 1981). According to the reports of those who requested audits but did not sign contracts with the program, they made major investments projected to save more than 50 percent of what the program participants could be expected to save. By contrast, a comparison group of eligible homeowners who had not requested energy audits reported investments that would save only about 25 percent of what the program's full participants could be expected to save (Stern, Black, and Elworth, 1982a). If these estimates are accurate, the high energy prices of the late 1970s had motivated home- owners to make about one-quarter of the economically justifiable investments in energy efficiency by mid-1980, while people who requested audits from the conservation program were making, on the average, more than half of the economically justifiable investments.2 The results of this study suggest that a well- constructed informational effort aimed at energy-efficient investments can go a long way toward bringing about the investment that full information would induce in rational actors. The data as a whole suggest that the qualitative aspects of information, more than the quantity of infor- mation available, is crucial to the effect of information 2 This finding may be an overestimate of the effect of information because the program offered financial incen- tives as well as information and because program partici- pants were self-selected. But financial incentives prob- ably account for only a small part of the program's effect because less than one-third of the 20 percent who signed contracts with the program used its low-interest loans (Stern, Black, and Elworth, 1981). And self-selection may also have been a minor influence. The sample of nonpar- ticipanta differed little from a statewide sample in an adjacent state, either in socioeconomic characteristics or in investments in energy efficiency. Furthermore, the factors that influenced investments in the state that did not offer a program accounted for only a small amount of investment compared with what program participation seems to have produced (Stern, Black, and Elworth, 1982a).
68 on action. But the data do not tell which qualitative factors matter most nor how much difference each makes alone or in combination with others. HOW DOES INFORMATION AFFECT BEHAVIOR? It may be reasonable to imagine the effect of information on energy use in the same terms as as economic stimuli are imagined to affect economic behavior: by producing an eventual steady-state response after a period of dynamic change. Thus, it makes sense to ask if information affects the final penetration of energy-saving technolo- gies and practices, the rate at which the changes take place, or both. Although researchers have not framed the questions in quite this way, the available knowledge sug- gests answers. To the extent that purchase decisions are influenced by erroneous impressions about how much energy a technology uses, information must affect the final penetration of energy-efficient technology. And to the extent that accurate information penetrates the society slowly, the responses to changing economic conditions are likely to be slowed. There are reasons to expect that information can make a difference, then, in both the statics and dynamics of energy demand. To understand these effects, one must understand the process by which energy information comes to affect behav- ior. There are at least two possible descriptions of the process, each with its own implications for what knowledge is needed: that people act rationally on the basis of whatever information they possess; or that information diffuses through informal networks with people acting by example or on trust, rather than as a result of rational choice. Rational Action Based on Imperfect Information As a first approximation of the role of information, for- mal demand analyses might suppose that consumers act rationally on the basis of the information that is in their awareness, as they understand, interpret, and trust it. In this view, available information is transformed by social and behavioral processes into effective infor- mation (that is, the information consumers act on); behavior follows from effective information by principles
69 of rational choice. 3 Thus, the key analytical problem is to learn what leads consumers to understand and inter- pret the available information as they do. Energy infor- mation may have static and dynamic components; conse- quently, the rate and final penetration of accurate energy information set limits on the rate and final penetration of economically rational responses to conditions in energy markets. To understand the effects of informational policies and programs, an analyst must determine how much and how quickly policies and programs, operating in the existing informational context, change effective informa- tion for energy users. Given this "effective information" model, an obvious first step is to gather information about what consumers believe to be true about energy use and energy-efficient investments. ThiS step can be carried out by conducting detailed surveys of consumers' beliefs about the costs and benefits of various investments in energy efficiency (e.g., Kempton, Harris, Keith, and Weihl, 1982). Data from such surveys could be incorporated into existing formal models by substituting estimates of perceived costs and benefits for the terms that define the costs and bene- fits of alternative actions. One might compare an exist- ing model that estimates energy-efficiency investments under full information with a parallel model that holds parameters constant but substitutes beliefs about capital cost and operating cost for the best available information about actual costs. This procedure could estimate an approximate upper bound for the effect improved informa- tion might have on the rate and final penetration of energy-efficient technologies. Data on consumers' beliefs could also be combined with measures of actual energy use and investments in energy efficiency to calculate discount rates based on effective information. (This could be done, for example, with discrete choice models of the type 3 The assumptions of rationality may not hold, however. By most definitions, a rational consumer is one who con- verts beliefs about costs and benefits into expected value (i.e., multiplies by probability), corrects for inflation, compares the considered action with the costs and benefits of alternative actions, and takes the effects of any tax benefits into account in all these calculations. Some of these assumptions are questionable, and others are improbable for some consumers (see Chapter 2; Kempton and Montgomery, 1982; Stern and Aronson, 1984).
70 presented in Appendix A.) A plausible hypothesis is that when investment choices are modeled on the basis of what people believe to be true about costs and benefits, rather than on the assumption of full and accurate information, estimated discount rates will come closer to those used by most investors in financial markets. Research to assess consumers' beliefs should be disag- gregate because beliefs about energy may vary geographi- cally, by fuel, or as a function of sectors of the econ- omy, the purposes for which energy is used, the consumer's economic status, or other factors. Such disaggregated research could identify for what groups misinformation is a serious barrier to energy efficiency and for whom special informational programs might be developed. Surveys to assess effective information would be use- ful, but they do not address the key analytical question of how available information gets transformed into energy users' beliefs (effective information) or the key policy question of how to improve effective information. The answers to these questions are difficult to get and then difficult to incorporate into formal models. The follow- ing questions must be addressed: Which information do people notice? Which information do they understand? Which information do they believe? What makes information credible? How do people establish their beliefs in the face of conflicting information, and what information, if any, can change such beliefs? How do people account for the uncertainty of information? These questions have rarely been asked in the context of formal energy demand models; indeed, they call to mind processes of attention, cognition, and judgment that are not readily modeled as economically rational action. Yet just such processes may make the difference between a successful informational effort and a worthless one (Stern and Aronson, 1984). What does it take to make available information effec- tive? Parts of the answer have already been suggested. Available information is likely to become effective infor- mation when it is designed to attract attention, when it is vivid and personalized, when it comes from trusted sources, and so forth. The evidence of these relation- ships comes from general research in cognitive and social psychology and from problem-oriented experiments and pro- gram evaluation studies (Stern and Aronson, 1984:Chapter 4). More detailed answers can be expected from continued research along the same lines. Careful experiments and program evaluations can assess the effects on consumers' beliefs of different mixtures of procedures for presenting information.
71 The same methods could also be used to identify indices, similar to the miles-per-gallon index for auto- mobiles, that express a building's energy efficiency in familiar units and so improve people's understanding of energy efficiency in buildings. Careful field research on suggested indices and modes of presentation would be necessary because the most successful methods for making full information more effective probably vary with energy uses, types of consumer, and possibly also in other ways. While it is possible to study the features of informa- tion and information delivery that affect people's beliefs and action, it would be difficult to incorporate these features into formal energy demand models. First of all, the most effective informational efforts involve a com- bination of elements that probably act synergistically. It is possible in principle to separate the effects by experimentation or by careful analysis of a diverse enough group of informational programs, but it would be diffi- cult. Furthermore, the policy questions concern the effect of information on action rather than on the beliefs that lead to action. Thus, the simpler approach from a policy viewpoint would be to leave out the measurement of beliefs in the effort to identify workable means of making information effective. This notion leads to a second model of how information affects behavior--one that depends less on assumptions of rational action. Diffusion of Innovation An alternative view of the role of information comes from the view of social change as diffusion. In this view, change spreads through society along social communication networks and is subject to a range of social and economic variables that mediate the process. When information is fed into these communications networks, it may be trans- formed as it is transmitted. And through a process of social influence, behavioral change may occur. From this perspective, the key to understanding how energy-efficient technologies "penetrate a market" is understanding com- munication and social influence, especially word-of-mouth communication within identifiable groups or networks: homeowners in a city, firms in an industry, municipal governments, and so forth. It has been suggested that innovations are adopted as a function of their relative advantage over previous practices, their compatibility with the adopter's values,
72 their apparent simplicity, the ease with which they can be adopted on a trial basis, and the observability of their outcome (Rogers with Shoemaker, 1971). Thus, the diffusion model incorporates economic cost (in the concept of relative advantage), but only after a potential adopter hears of an innovation. Information also fits in the model at that stage: it can demonstrate the advantage of adopting an innovation, make the results of adoption more readily observable, or show an adopter the results early to minimize the costs of error. In contrast to the effective information model, the diffusion model does not hold that it is necessary for every consumer to know all the facts needed for rational action before an energy-efficient innovation will be adopted or that tangible personal benefits are the only motives for improving energy efficiency. It may be enough to get relatively full information to sources that con- sumers trust. Some people will assimilate all the infor- mation from these sources, but others will act on trust, without rational calculation; conceivably, trust will lead some people to do what a rational actor would do with full information. Diffusion can also impede action. News can travel fast about the failure of an innovation to offer the advantages claimed for it or about unanticipated negative conse- quences of adoption. News of unsavory experiences with urea-formaldehyde foam insulation may have spread by word of mouth even before the problem was mentioned in the mass media--and the news undoubtedly affected many people who made no effort to estimate the size of the risk. The diffusion perspective has some general validity for describing transitions to improved energy efficiency (Darley and Beniger, 1981; Stern and Aronson, 1984). It emphasizes several variables as important in determining whether information about energy efficiency will affect action: the consumer's trust in the source of informa- tion; a program's ability to get information into working communication networks; the use of vivid, simple, and personalized images or phrases of the kind people are likely to repeat; and so forth. These variables are very different from those considered important under an assump- tion of rational economic action, though the diffusion model also has room for variables related to costs and benefits under the concept of relative advantage. The importance of communication and influence factors is not easily quantified, but it can be roughly estimated by f ield experiments that assesses the effect of extra
73 efforts to get information into communication networks or otherwise to assist the diffusion process. Process eval- uations of energy information programs that assess com- munication variables can also give an indication of the magnitude of their effects. These research methods can operate in the same ways we suggested for assessing the effects of nonincentive aspects of financial incentive programs (see Chapter 3). While it may be possible to estimate bounds for the importance of communication and social influence in infor- mation programs, it is not now possible to model the dif- fusion process itself. Only limited research has been done to determine the processes by which energy-efficient innovations diffuse in the society and none, to our knowl- edge, has used knowledge about communication in social networks to forecast the final penetration or the rate of adoption of particular energy-efficient technologies. Many research questions need further attention if the diffusion model is to be useful for those purposes. In particular, it would be useful to know more about which sources are trusted by different types of consumers; which networks spread information about energy technologies among individuals, firms, professional groups, and public agencies; how energy information spreads through these social networks; and what kinds of communication most convincingly demonstrate to consumers the relative advan- tage of energy-efficient innovations. For this analysis, it makes sense to use a market seg- mentation approach: an approach aimed at identifying groups of people who think in the same way about a pos- sible innovation or who trust common sources of informa- tion. This approach is valuable because different kinds of consumers communicate in different networks and find different kinds of information relevant and persuasive. In addition, people use systematically different ways of combining the elements of information that they have, and systematically different aspects of the information affect their decisions. To tell how people use information, it would be useful to observe the actual decision making, particularly among people contemplating purchase of high- capital-cost innovations. Laboratory experiments can also aid understanding of the kinds of information that matter in these decisions. Survey methods may be useful for finding out which information soures are trusted and heeded by particular groups of consumers. Segmentation analyses, conducted by survey or laboratory experimental methods, can be further refined in comparison group
74 studies that test preliminary findings by targeting information from existing energy information programs to particular segments of the public. Despite these options for improving understanding of how diffusion affects energy-related decisions, much con- ceptual work and research is required before such pro- cesses could be formally modeled. It may prove useful to integrate empirical work on the span and scope of acquain- tance networks (e.g., Garevitch, 1961; Travers and Milgram, 1969) with mathematical simulations of the pro- cesses of influence diffusion within those networks (e.g. Pool and Kochen, no date). Such research holds promise in the long run for useful insight into the dynamics of the influence of energy information. CONCLUSIONS People do not act always on full information about the costs and benefits of energy-efficient investment, even when full information is available. Analysis of energy demand, then, depends on understanding the conditions under which information influences action. One assumption is that people take rational economic action based on what they believe to be true (effective information). This assumption is worth developing and testing. It calls for research to assess what consumers believe~about the costs and effects of investments in energy efficiency and on the ways these beliefs change. Surveys can determine what people believe, but the effective information perspective offers no guidance for understanding the conditions under which information changes beliefs. An alternative assumption is that innovations diffuse through communication networks, with the quality of avail- able information acting as one link in the process. The diffusion perspective identifies communication and influ- ence variables, such as trust, vividness and simplicity of format, and face-to-face communication in informal social networks, as important in determining whether information is effective. Because of evidence that such variables are important in energy choices, further research into them is a high priority. The magnitude of their effects can be roughly estimated for inclusion in demand analyses by field experimentation and process evaluation of energy information programs, but formal models of the process by which energy information produces behavioral change are still a long way off.
