Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 43
The Effects of Financial incentives on Energy-Efficient investments in Residential Buildings A number of studies have concluded that rates of investment in energy efficiency are far from economically optimal levels. Substantial investments are not made that would, by substituting technology for energy, lower overall costs for energy users without changing their levels of services (Hirst, Fulkerson, Carlsmith, and Wilbanks, 1982; Office of Technology Assessment, 1980, 1982; Ross and Williams, 1981; Sant and Carhart with Bakke and Mulherkar, 1981; Solar Energy Research Institute, 1981; Stobaugh and Yergin, 1979). A study by a group at the Mellon Institute (Sent, 1979) concluded, for example, that the actual mix of fuels and capital characterizing energy consumption was quite different from what it would have been if society had been minimizing long-run costs: in 1978, the nation actually consumed 79 quadrillion Btus (quads) of energy, of which 45 percent was oil; in con- trast, the Mellon study estimated that if consumers had been minimizing life-cycle costs, total consumption would have been about 59 quads and oil consumption would have been as much as 30 percent less. In addition, the total cost to society of energy services (fuel plus capital) would have been 17 percent lower in the model's least-cost case than in the actual 1978 situation. Uneconomically optimal" has been variously defined, but the essence of the concept is minimization of cost. The concept usually implies comparing the net present values of alternatives under stated assumptions about the oper- ating costs of equipment, the discount rate for future energy costs, and the useful life of the equipment. Cal- culations of optimality should also discount future values according to the uncertainty that they will be achieved, 43

OCR for page 43
44 In the residential sector of the economy, rates of investment in energy efficiency are probably even further from optimal levels than in other sectors. Most home- owners are less likely than industrial or commercial decision makers to spend money on energy efficiency in exchange for cost savings over the long term (Hausman, 1979). It has been estimated that economically justified energy savings of 50 percent and more are technically possible in the average residence (Solar Energy Research Institute, 1981). Yet national tax return and Energy Information Administration (EIA) data show that only about 5 percent of all households made conservation investments in 1977 and 1978. Most households in the EIA survey invested about $720--about one-half of the level of $1,500 suggested by some analyses as optimal (California Public Utilities Commission, 1980; Energy Information Administra- tion, 1980; Hirst, Goeltz, and Manning, 1982). A number of explanations have been offered for why households invest less in energy efficiency than is economically justified. Some of the major barriers to these investments include: . lack of information concerning conservation tech- nologies and the magnitude of the costs and benefits of using them; · the existence of confusing and conflicting infor- mation about the effects of energy-saving technologies and practices; · limited choices because intermediaries (such as appliance manufacturers, building owners, or home builders), who are not concerned with operating costs, make many decisions; · the relatively time-consuming and complicated processes necessary for one to become informed about appropriate conservation measures and to actually make purchases; · consumer distrust of the suppliers of conservation services and information, particularly small contractors in the home improvement field; . but most economic analyses of energy efficiency have dis- counted these values at the market rate of interest, with- out considering that returns on investments in energy efficiency may be less certain that returns on investment in organized markets (Chernoff, 1983).

OCR for page 43
45 · the ~invisibility" of energy flows; that is, the difficulty of observing and assessing the effects of one's investments or behavior on one's energy use; and · lack of readily available cash to pay the high first costs of installation, coupled with an inability to find financing or a preference not to use it. Because of the variety of recognized barriers to eco- nomically justified investment in household energy effici- ency, programs aimed at reducing those barriers might be expected to succeed in shortening the time scale for replacement of inefficient capital stocks and in raising the final level of penetration for energy-efficient tech- nologies. Governments and utilities are spending large sums on incentive-based programs to achieve these goals. The Tennessee Valley Authority, for example, has provided more than S250 million in interest-free loans to its cus- tomers (Berry, 1982), and federal energy tax credits are expected to cost $2.5 billion between 1981 and 1986 (Hirst, Goeltz, and Manning, 1982). A variety of smaller grant, loan, and rebate programs exist to encourage more investment by offering financial incentives. Formal energy demand models should, ideally, provide a means of comparing the effects of alternative policy actions on investments in residential energy efficiency. Such models begin with a baseline estimate of the time scale and penetration level for new energy-efficient tech- nologies without incentive policies (see Figure 1). Then, expected diffusion curves for different types and levels of incentives can be estimated and compared with the base- line. Analyses of this type have been made in formal models. As we noted in Chapter 1, the methods used to define the baseline diffusion curves and the curves resulting from incentive policies inspire little confidence. There is usually no basis for validating the correspondence between a model and actual behavior. Little is known about either the dynamics of the adoption of energy- efficient technologies or about the final level of their penetration. There is also little behavioral knowledge about the way incentives operate on which to estimate how incentive policies would change consumer response. In estimating · . · ~ . . ~ . . ~ , ~ _ _ ~ ~ incentive effects, analysts typically Interpret eacn con- templated incentive as a change in the cost of energy or the cost of equipment. Such conversions involve assump- tions that are often left implicit. For example, the

OCR for page 43
46 100 In us co I U] Z 50 o ~3 z A: ILL o WITH INCENTIVE '' ~~ ''''~ '/ W ITH O UT I N CENTIVE TIME (years) FIGURE 1 Hypothetical curves showing penetration of new energy-efficient technologies as a function of time, with and without a financial incentive. Hirst-Carney (1978) model of residential energy use, in analyzing the impact of federal tax credits, assumed that the sole effect of the tax credit was to reduce the initial cost of the equipment: that is, the 15 percent tax credit was treated as a 15 percent reduction in the cost of eligible energy-efficient devices. No considera- tion was given to other plausible effects of the tax credit. For example, the fact of government support may have bolstered some consumers' confidence that the covered conservation investments would be effective. Other con- sumers may have largely disregarded the credit because repayment would occur 4 to 16 months after purchase, because they have no tax liability against which to claim a credit, or because they do not habitually keep the records or do the extra paperwork necessary to claim a tax credit. To raise such possibilities is to suggest that the economic meaning of a financial incentive is not always obvious and to note the empirical questions that must be answered if formal models are to offer reliable interpretations of the effect of financial incentives. In exploring the relationship of financial incentives to investment in the energy-efficiency of residential

OCR for page 43
47 capital stock, this chapter focuses on three issues: (1) the relationship between the size of an incentive and con- sumers' responses; (2) the possibility that people may respond differently to different types of incentive, even when the incentives have the same financial value; and (3) the importance of nonfinancial variables to the effective- ness of financial incentive programs. We discuss evidence that the effect of incentive size may not be what is usu- ally modeled and that incentive type and other nonfinan- cial variables may be critical to the success of incentive programs. Finally, we present approaches for improving understanding of how financial incentives work--or fail to work--in the context of residential energy con- servation. HOW DOES THE SIZE OF AN INCENTIVE AFFECT CONSUMERS ' RESPONSES ? The simplest interpretation of a financial incentive is that it lowers or defers the cost of energy efficiency relative to the alternatives. Larger incentives make investment economic for more people, so more people ultimately make investments. Incentives may also increase the pace of investment by making capital more available. In this interpretation, the result is that both the pace and the final level of investment will appear as smooth functions of the size of the incentive. But incentives may operate in other ways. They may, for example, func- tion mainly as attention-getting devices; or they may mainly speed action by people who would have taken the same action without the incentive. Incentives as Attention-Getting Devices There is a rationale in psychology for thinking of an incentive as a way to attract attention. If energy users are problem avoiders who do not normally pay attention to energy (Stern and Aronson, 1984), an incentive will matter only if its size or the accompanying marketing effort is great enough to command a consumer's attention. After the incentive passes this threshold, further increases will make little difference. In this interpretation, behavior is a step function of the size of the incentive. The aggregation of the individual step functions may produce a smooth curve, but if the underlying process is atten-

OCR for page 43
48 tional, thinking of the size of the incentive as crucial misses the point: efforts to attract attention may make more difference than the size of the incentive. There is some evidence that threshold effects do exist with energy use. Analysis of data from the Wisconsin time-of-use electricity pricing experiment (Heberlein and Warriner, 1982) showed that, over a range of price differ- entials from 2:1 to 8:1, the effect of price was not as great as that of personal commitment or other nonprice variables. Most of the effect of time-of-use prices on behavior seems to have been achieved by the 2:1 price differential, and additional price doublings made little difference. These findings, however, may not generalize to incentives aimed at major investments rather than at changes in the way consumers operate household equipment. The only data on threshold effects for such incentives come from studies of consumer response to real or proposed loan subsidies. Those data suggest that there is little consumer interest in loan subsidies, except for interest- free loans. At least 70 percent of all home energy improvements are undertaken without the aid of loans `(Berry, 1982), and offers of reduced-interest loans have not attracted many additional householders. In one study of participants in a conservation program that offered bank loans at reduced rates, 70 percent paid cash for retrofits averaging over $1,100 in cost (Stern, Black, and Elworth, 1981). A survey by Northeast Utilities (1981) reported that the amount of money people intended to bor- row did not change as fast as the loan rate. But the availability of interest-free loans alters the picture. Programs that offer interest-free loans generally have a higher proportion of participants using loans than pro- grams that offer only low-interest loans (Berry, 1982). In the study by Stern, Black, and Elworth (1981), 49 per- cent of people who did not have retrofit work done under a conservation program that offered reduced-interest loans said they would have taken advantage of interest-free loans (payable on resale of the house). These findings are suggestive, but not conclusive. If large-scale loan subsidies are being considered, it would be worthwhile to be more certain about the possibility that a threshold effect exists. Research should specifi- cally address the possibility of a discontinuity in con- sumers' responses to loan rates, such as might result from a process of attracting attention. ., . This possibility can be studied initially in small-scale experimental studies that assess consumers' interest in loans at various rates.

OCR for page 43
49 Some such studies have been conducted (see Berry, 1982 for a review), but they have tended to be larger than neces- sary in their sampling effort and inadequate in the detail in which they explore variations in the incentives. Bet- ter methods for evaluating the effect of loan rates exist and are discussed in a somewhat broader context in the next section. Incentives as an Impetus to Speed Changes Already Planned It is also possible that incentives affect only those energy users who are already paying attention to the costs of energy efficiency--that is, those who have already decided to invest. If so, incentives may increase the pace of investment, but not its final level. They might do even less: an incentive program may simply shift the financing of planned investments from savings or other sources of loans to the incentive program. The limited evidence that is available is consistent with the hypothesis that incentives affect mainly the com- mitted investors: 60 percent or more of the people who claim conservation and solar tax credits report that they would have invested as much without the incentives (Berry, 1982). People who take advantage of loan subsidies have similar characteristics to people who invest without special incentives: people with higher incomes are more likely to participate in subsidized loan programs and apply for larger loans (Berry, 1981; Northeast Utilities, 1981); people who live in energy-inefficient houses, and therefore have more need to invest, are more likely to take advantage of incentives for investment (Pacific Gas and Electric Company, 1982). Thus, it is likely that most people who take advantage of financial incentives would have invested anyway, sooner or later. Incentives do appear, however, to increase the pace of investment: 78 percent of participants in a Pacific Gas and Electric Company interest-free loan program stated that they would not have installed their conservation equipment during the time period in which they actually purchased it if the loan had not been available (Pacific Gas and Electric Company, 1982). Similarly, only 29 per- cent of householders who accepted low-interest conserva- tion loans from Minnesota's Northern States Power company said they would have made the same investments without the loan (Hirst, Goeltz, Thornsjo, and Sundin, 1983). The

OCR for page 43
so clearest evidence comes from a 2-year study of partici- pants in the Bonneville Power Administration's home weath- erization program (Hirst, White, and Goeltz, 1983b): households that took advantage of the program's interest- free loans reduced their energy use by 12 percent in the first year and by 14 percent in the second year as com- pared with nonparticipants; households that used the pro- gram's energy audits but did not take the loans showed no savings in the first year but 8 percent savings in the second year, as compared with those who did not use the audits. The incentive seems clearly to have speeded investment; only further follow-up research can determine if there was also an effect on the final level of invest- ment achieved. The idea that incentives work only for people who are already paying attention to energy efficiency is plausible because of the kind of behavior involved. Consider the difference between a low-interest car loan and a low- interest energy loan. The purpose of the former is usually to influence a choice among alternatives: most people who are considering buying a car are planning to borrow, so are likely to shop around for a low loan rate. A loan subsidy may get such people to purchase a different model or to make a planned purchase sooner; it is unlikely to increase the number of cars in use. But the primary purpose of a conservation loan program is to spread an innovation--to increase the number of people who decide to make a purchase. Thus, energy loans probably are not perceived and do not function like car loans. Someone who has not yet decided on energy efficiency is not likely to be shopping for low loan rates and will not notice them. Furthermore, a conservation incentive program that says one must go into debt to conserve energy is not making an effective sales pitch, regardless of the interest rate offered. The idea that incentives only affect people who would have invested anyway remains only a hypothesis. But because of its policy implications if true, it seems advisable to conduct some small experiments to test it. Research designs that include comparison groups not offered incentives are the best way to control for self- selection--the possibility that people choose to partici- pate in an incentive program only after making the deci- sion to invest. A comparison group makes it possible to separate differences due to an incentive from those due to other, simultaneous events. It also makes it possible to tell whether people offered the incentive differed in

OCR for page 43
51 any potentially important ways from those not offered the incentive. Still more reliable interpretations can be made from randomized experiments--those in which house- holds are randomly assigned to treatment groups. This approach puts statistical bounds on the possibility that treatment groups differ systematically in ways (including unmeasured ones) that might affect their response to an incentive. Comparison groups not offered incentives can be obtained by studying programs in nearby areas (e.g., different utility service areas) that do and do not offer incentives. But because the programs (and areas) may differ in other ways than the incentives they offer (see below), random assignment of eligible consumers to differ- ent types of incentive is the surest way to obtain reli- able data. Small-scale experiments with incentive programs can be designed to test simultaneously the effects of incentives and other program components. For example, such experi- ments could explore the possibility that for encouraging new investments, a conservation program might get more out of its resources by simplifying the process of investment than by offering a financial incentive. HOW DOES THE TYPE OF INCENTIVE AFFECT CONSUMERS' RESPONSES? Formal demand models usually compare types of incentives in terms of their economic value, and, given the present state of knowledge, the comparison requires making tenta- tive assumptions. For example, a simple economic model may compare a loan and a rebate in terms of net present value: rebates decrease present costs while loans defer them, so, given knowledge of consumers' discount rates, the two incentives can be compared by calculating the present value of the loan. By this method it becomes possible to tell which loan rates are equal in economic value to which rebate rates. Thus, to equate loans or rebates in terms of economic value, the relevant discount rate must be postulated or estimated. The available data on consumer preference for incen- tives, if interpreted in the language of discount rates, suggest an anomaly. Preference for loans or rebates appears to be a function of income, with higher-income consumers more likely than lower-income consumers to pre- fer loans (Berry, 1982). If this difference in preference is interpreted in terms of net present value, the implica-

OCR for page 43
52 tion is that high-income people, who express a preference to defer some present cost with a loan, have higher dis- count rates than low-income people. This finding appears to contradict other evidence that income is inversely proportional to discount rates (Hausman, 1979; McFadden and Dubin, 1982). The contradiction can probably be resolved by focusing on attributes of the types of incen- tive other than those related to time discounting. Per- haps rebates are attractive to low-income people because they reduce the perceived size of the capital commitment while loans do not. Or it may be that low-income con- sumers are simply averse to indebtedness. Incentives vary qualitatively in several ways that are not captured by the usual calculations of economic value. The kinds of incentives offered for energy-efficiency investments include grants, price discounts, rebates, and loans. A qualitative classification of types of incen- tives may prove useful for understanding energy users' responses. Grants decrease the cost of an investment before it is paid and are usually offered by a government agency. Some of them, such as those offered by low-income home weath- erization programs, cover labor as well as capital costs. Including labor may attract energy users because it solves the problem some people have of finding someone to do the work, but it may repel people if they do not trust the program or the workers it provides. Discounts on the price of energy-efficient equipment are usually offered by manufacturers or retailers. They are like grants in terms of their immediacy and may be even easier to obtain because they do not involve paper- work. Rebates are delayed discounts. They usually involve a process of filing a claim, which can vary from trivial to tedious. Their effectiveness is likely to depend on whether claiming the rebate is seen as worth the effort. Tax credits, like rebates, are delayed discounts, but they are different from rebates in that they have longer delays and require record-keeping as well as filing of forms. They require a change in the routine of filing tax returns and are not available to energy users who have no tax liability. Loans must be approved in advance, and the possibility of refusal may deter some consumers from applying. In addition, people may think about loans differently from the way they think about other incentives: for a person with little economics education, discounts, rebates, and

OCR for page 43
53 even tax credits are probably easier than loans to imagine as price decreases. If so, a smooth curve relating the size of an incentive to behavior will be observed more regularly for nonloan incentives than for loans. Further- more, there is evidence that many householders simply do not consider going into debt for energy efficiency. Householders consistently say that the availability of low-interest loans makes little difference in their deci- sions to invest in home energy efficiency (e.g., Olsen and Cluett, 1979; Stern, Black, and Elworth, 1981). Interest- free loans, as already noted, may be qualitatively dif- ferent from low-interest loans if they do not seem like borrowing. Generally, procedures that make loan repayment seem painless--by deferring it until a house is sold or by ensuring that loan payments are never greater than energy savings--may make a large difference in response to loan programs. The above descriptions emphasize certain qualitative characteristics of financial incentives that seem likely to make a difference in consumer response: delay in receipt of the incentive, effort involved in obtaining it, uncertainty that it will be received, the incentive's effect on household budgets and cash flows, and trust in the offering organization. We identify these particular variables as important on the basis of examination of behavioral research on energy conservation programs, on energy use more generally, and on other actions of indi- viduals and households (for a detailed discussion, see Stern and Aronson, 1984:Chapters 3, 4). Some of the above variables seem capable of quantifi- cation and inclusion in formal demand models. Delay, for example, differentiates tax credits from more immediate incentives and may differentiate among programs offering the same type of incentive (e.g., rebates). The concept of discounting can be used to estimate the effects of delay, although the lack of an independent empirical basis for estimating discount rates makes this a questionable approach until more data are in. Uncertainty about receiving the incentive differentiates among tax credits, loans, and rebates--and especially within the class of loan programs. This uncertainty can be estimated either from program data or from the perceptions of potential clients; the latter may be the more relevant index. Effect on budgets and cash flows distinguishes primarily among loan programs. For example, programs that defer repayment until a house is sold may be more attractive than those that defer payment for 7 years, even in an area where the average house is sold in 7 years.

OCR for page 43
54 We do not believe that sufficient data yet exist for quantifying even the most quantitative of these variables. But methods exist for gathering such data, and it is already possible to offer some hypotheses for testing: (1) Delay in receipt of an incentive is an especially important barrier for low-income consumers; (2) Programs that keep energy investments from draining household bud- gets are significantly more attractive than other programs offering incentives of similar size, but special marketing efforts may be necessary to emphasize this feature; (3) The red tape, record-keeping, and other effort needed to take advantage of some incentives is a significant deter- rent to consumers, especially low-income consumers. These and related hypotheses can be investigated. While some relevant data might come from evaluations of existing incentive programs and pilot studies of new pro- grams under consideration, a more systematic and promising method can be used: small, laboratory studies that com- pare people's initial interest in a great variety of pos- sible incentive packages. Methods of focused group dis- cussion can provide some initial insights, but more quan- titative estimates can be made by using methodology adapted from psychometric studies of decision making under uncertainty (e.g., Kahneman and Tversky, 1979; Tversky and Kahneman, 1981) or from research on multiattribute deci- sion analysis (e.g., Klein, 1983; Svenson, 1979; Tversky and Sattath, 1979). In these approaches, people are pre- sented with a series of hypothetical choices and decision problems that vary in terms of the attributes, trade-offs, steps, contextual factors, and rules used in making deci- sions. With these methods, a large number of theoretical questions about choice can be investigated in a few experiments. Of course, such laboratory studies may not be general- izable to actual choices. Self-reported interest and intention are highly imperfect indicators of behavior in a new situation. But the purpose of laboratory studies is not to draw conclusions but rather to narrow the range of likely hypotheses for more rigorous test under field conditions. Small field experiments that systematically vary the factors that seem important would be a logical next step. In anon to or aster initial field trials, the most promising incentive packages can be implemented experimentally to compare their effects with those of existing incentive packages or other comparison con- ditions.

OCR for page 43
55 Some qualities of financial incentive programs are less easily quantified, even with psychometric methods. Some of them, however, are equally likely to be important. One of these, trust in the sponsoring organization, has been mentioned above. It and others are discussed in the next section. HOW DO CONSUMERS RESPOND TO NONFINANCIAL FEATURES OF INCENTIVE PROGRAMS? Nonfinancial factors clearly have important effects in incentive programs. At times, they may make more differ- ence than the incentives themselves. Conservation pro- grams that are identical in terms of the incentives they offer are received very differently by their prospective clients. For example, New York State requires its regu- lated utilities to offer financing of energy efficiency at 9-11 percent interest; the proportion of households receiving energy audits who take advantage of the financ- ing ranges from less than 1 to more than 40 percent across the utilities (Scherer, 1981). The Bonneville Power Administration offered interest-free loans through 11 utilities participating in its pilot weatherization pro- gram. The proportion of households given energy audits under the program ranged from 10 to 58 percent, and the percentage of audited households accepting the interest- free loans ranged from 8 to nearly 90 percent (Lerman, Bronfman, and Tonn, 1983). Most of the available evidence for explaining this variation comes from studies of conservation loan pro- grams. It suggests that the most important influence on a program's effectiveness is not the loan rate or the term but implementation: promotion to the clientele, success in simplifying the decision process and in alleviating fears of shoddy work, and the ability of the sponsoring organization to gain trust. The same variables seem to affect conservation programs that rely primarily on infor- mation, such as the Residential Conservation Service and other energy audit programs. Promotion begins with getting a consumer's attention. Utility bill enclosures generally fail in this respect, while direct mail and media advertising are somewhat more effective; news items including statements from prominent public officials attract still more attention (Rosenberg, 1980). Personal communication from friends or trusted local organizations is probably the most effective form

OCR for page 43
\ 56 of promotion (e.g., Leonard-Barton, 1980, 1981; Rogers with Shoemaker, 1971), and this approach has been used with success in energy conservation programs (Olsen and Cluett, 1979; Fitchburg Office of the Planning Coordina- tor, 1980). Promotion also involves an effort to "sell" conserva- tion to households. A recent study of the pilot weather ization program of the Bonneville Power Administration - (Lerman, Bronfman, and Tonn, 1983) concluded that differ- ential success among the participating utilities was due in considerable degree to the level of a utility's effort and commitment to the program. In particular, utilities succeeded in getting more homes weatherized when they had larger staffs for conservation, when management and per- sonnel were dedicated to the program, and when their staff members promoted the program rather than contracting the job to outside organizations that were paid a fee for each audit they conducted. Simplification of the decision process is often iden- tified as a major benefit energy conservation programs can offer households. Even though incentives directly address only financial issues, the success of an incentive program may hinge on how well it assists clients in resolving other problems. A home retrofit investment depends on choices about how to get information, what information to believe, what to do, how to get it done, and how to pay for it--and each of those choices can be difficult. The convenience some programs offer in terms of "one-stop shopping" for energy conservation has sometimes proved more important to potential clients than the availability of a financial incentive (Stern, Black, and Elworth, 1981). Residential energy consumers are concerned about the reliability of the work done by providers of energy- efficiency home improvements. In one study of a statewide residential conservation program that offered low-interest loans, free audits, a pool of certified contractors, and inspection of all retrofit work (Stern, Black, and Elworth, 1981), the two reasons most frequently cited by participants for their involvement were: n I trusted the work because it would be inspected" (98 percent) and n I didn't have to worry about finding a reliable contractor n (96 percent). In contrast, only 14 percent of partici- pants cited the availability of low-interest loans as a reason for their participation. Although consumer pro- tection guarantees are not a part of financial incentives, they seem to make a major difference in the attractiveness

OCR for page 43
57 of a program that offers incentives. If participants' self-reports are a valid guide, improving consumer protection may be a more effective way to spend scarce program funds than offering stronger financial incentives. Finally, trust in a conservation program is obviously Trust probably makes more dif- ference for programs that offer energy information or installation services than for those that offer only financial incentives, but since these features are so often combined, trust can be a major issue in many incen- tive programs. Even in programs that offer nothing but an incentive, trust in the sponsoring organization may affect its ability to get consumers' attention. Some suggestions for building trust in programs that involve energy information are offered in Stern and Aronson (1984). Most of the above variables are not how readily quan- tifiable for inclusion in formal energy models--and they may not be even after additional research. Existing models generally do not use concepts that correspond to those qualitative variables, which therefore usually show up as variability in estimates of the rate and ultimate penetration of energy efficiency. Given the extreme range of variation among similar programs in these respects, the qualitative factors are obviously candidates for further analysis. Such analysis is best carried out by problem-oriented research. The importance of such factors as consumer protection guarantees, simplicity of decisions, and the identity of the organization offering the incentive might initially be addressed by detailed small surveys, psycho- metric experiments, or even focused group discussions of the tone used in market research. - crucial to its success. _ But the limitations of these methods are more severe here than for analyzing loan packages, because it may be impossible to describe the alternatives well enough for participants to understand them. An alternative first step would be to address the issues of promotion, implementation, trust, and so forth _ . In program evaluation studies. Particularly valuable are studies of groups of programs (e.g., Stern, Black, and Elworth, 1981): better yet, groups of programs that are identical in their formal requirements but are implemented by various organizations (for example, the Bonneville pilot program studied by Lerman, Bronfman, and Tonn (1983) and residential conservation programs operating under a common set of state guidelines). Such research can set approximate bounds on the size of effect a given incentive

OCR for page 43
58 may have depending on other conditions of its implementa- tion. It can also generate hypotheses to test in small- scale field experiments aimed at finding the most effec- tive ways to use the available resources. Similar research on programs that do not feature financial incen- tives (e.g., energy audit programs) can help identify non- financial interventions that might greatly improve the return on an expense for financial incentives (see Chapter 4). CONCLUSIONS Consumer response to different types and levels of finan- cial incentives raises important policy questions both for governments and for the many gas and electric utilities that offer such incentives. The overall cost of federal and state residential energy tax credits, loan subsidies for residential retrofit, and rebates for retrofit and purchase of energy-efficient appliances is probably sev- eral hundred million dollars a year. Unfortunately, very little is known about the absolute and relative effective- ness of these financial incentives in stimulating instal- lation of energy-efficient technology. Underlying this ignorance is a lack of knowledge about how incentives affect consumer behavior. Financial incentives for residential energy efficiency seem to have positive net effects, at least under some conditions. They can accelerate the rate of investment, but their effect on the final level of penetration or investment has not yet been established. The magnitude of this effect is difficult to establish because there are no reliable estimates of the final level of investment that would be observed in the absence of incentives. Also, the level of investment depends simultaneously on incentives and on other factors that are influenced by policy. Data can be collected to improve the basis for esti- mating the penetration of energy-efficient technology in the absence of incentives. Such data could come from large-scale surveys of existing capital stock and of rates of replacement of the equipment. Large panel surveys could determine rates of replacement, but to assess the nature of the new stock with respect to energy efficiency, smaller, more detailed surveys might be a more efficient research method. The existing literature on patterns of diffusion of innovation can generate hypotheses to test

OCR for page 43
59 against survey data, and the data could be used to esti mate the baseline for penetration of new technology. It seems clear that the effect of incentives is very much dependent on factors not usually incorporated in formal demand models. Some of these, such as threshold effects, the income level of potential participants, length of delay in realizing the incentive, and the effect of an incentive on household cash flow, are fairly easy to quantify and include in formal models. But promotion and management of incentive programs, two variables that seem among the most critical (Lerman, Bronfman, and Tonn, 1983; Scherer, 1981), are not so readily quantifiable. Furthermore, the effect of an incentive seems to depend on the availability in or around the program of other aids to consumers, such as credible energy information, assis- tance in simplifying the decision process, and consumer protection. Trust in the organizations responsible for an incentive program is an important qualitative factor, and its importance probably increases if the incentive program also offers information or installation services. Because so little is known about the magnitude of the effect the above variables have on consumer behavior, one cannot now predict or explain consumer response to dif- ferent kinds of incentives (interest-free versus low- interest loans, rebates versus loans, and so forth). But as this chapter notes, it is possible to conduct problem- oriented research in order to produce estimates, or at least upper and lower bounds, for the effects of many of the important quantitative and qualitative variables. The existing data, inadequate as they are, raise several ques- tions that should be the focus of research: · Are partial loan subsidies of any value at all in encouraging new investment in conservation or in acceler- ating planned investment? · Are rebates, grants, or other immediate incentives especially valuable attention-getting devices? Are they the only forms of incentive likely to affect low-income people? · Is some of the money spent on financial incentives better spent on nonfinancial modes of attracting invest- ment, such as marketing of conservation programs, guaran- tees of consumer protection, simplification of investment and payment procedures, and liaison between organizations providing incentives and others that may have higher credibility as information sources?

OCR for page 43
60 Our analysis suggests that a small proportion of the resources devoted to offering energy tax credits, utility loan subsidies, and other financial incentives could use- fully be redirected to a relatively inexpensive effort to conduct studies addressed to the above questions. Such studies hold the potential not only for evaluating exist- ing incentive programs, but for advising decision makers on how best to spend moneys allocated to promoting invest- ment in residential energy efficiency. The available formal models do not now encompass many of the important variables discussed in this chapter. These variables are not easily quantified, and they inter- act in unknown ways, making the analytic problems very complex. Some models are particularly inappropriate for analyzing the effect of financial incentives. Among these are the simple (e.g., reduced-form econometric) models of household energy demand that do not even include capital costs within the model, but estimate demand only from fuel prices and incomes. Other, more disaggregate, models explicitly include household choices regarding purchase of new equipment (Dubin, 1982; Goett and McFadden, 1982). These equipment choice models are explicitly sensitive to operating cost (which is a function of energy efficiency) and capital cost (which is affected by financial incen- tives). Such models are better able to incorporate some of the variables discussed here. A brief discussion of how one might do this appears as Appendix A. However, because of the importance of qualitative factors, it makes sense to be modest in the use of models for analyzing incentives for energy efficiency. High priority should be given to investing analytic resources in studies of the organizational and behavioral factors that seem often to spell success or failure in incentive programs. When one interest-free loan program has only 8 percent of its eligible households using the loans while a formally identical program has 90 percent participation, a research program restricted to analyzing the readily quantifiable financial variables--in formal models or by other methods--is ignoring the factors that may prove to be the most important.