The bottom-up approach and the energy efficiency gap

C.C. Koopmans, D.W. te Velde r Energy Economics 23 2001 57]75 59

2. The bottom-up approach and the energy efficiency gap

In this section we review the literature about the bottom-up approach and the energy efficiency gap, with particular emphasis on the gap that exists between top-down and bottom-up modelers. Analysis of bottom-up information usually involves drawing energy conservation Ž . Ž supply curves CSCs on the basis of individual energy saving techniques CSCs . were introduced by Meier et al., 1983 . Fig. 1 depicts an example of a CSC. Each step in the figure represents an energy saving technique that improves the energy efficiency without changing the level of output. Each technique saves an amount of energy per year against certain investment and maintenance costs. As costs and benefits are spread over more than 1 year, a discount rate is required to discount future costs and benefits. Bottom-up scientists often propose a real discount rate of 5]8, based on market rates for loans. The figure shows a clear price sensitiveness of energy conservation and hence, of implied energy efficiency. Fig. 1. Energy conservation supply curve. To compare bottom-up information to top-down analysis, we need to link CSCs Ž . to economic production functions Huntington, 1994; Blumstein and Stoft, 1995 . An isoquant derived from a production function essentially displays the same thing Ž . as a CSC. Stoft 1995 derives a CSC from a production function and shows that this CSC is a conditional factor demand function, with the output or utility held constant. This implies that CSCs can only be used to analyze the substitution effects of factor price changes and not the output effect. In Fig. 2 the isoquant or best-practice frontier depicts combinations of energy and other inputs that provide the same energy service. Point T represents the technical optimum for energy saving, but achieving this requires a lot of other inputs, mainly capital. Given the input prices, micro-economic theory can derive Ž . the cost-minimizing combination of the inputs point E . C.C. Koopmans, D.W. te Velde r Energy Economics 23 2001 57]75 60 Fig. 2. Isoquant representation. The position of point E depends on the ratio of the energy price to the price of other inputs P rP . If the other input is capital, using more capital implies saving e i energy over a number of years in the future. In this situation, P is the mean e discounted price of future energy use. For low discount rates, as often used by bottom-up modelers, point E is optimal. If a higher discount rate is used, P is e lower and the optimal choice becomes point E U . Implicit discount rates of 25 or Ž . more are often observed in reality Koomey and Sanstad, 1994 , although this does not tell us whether investors really utilize such high discount rates, or whether they use normal discount rates in a more complicated context than the simple cost- benefit approach used by bottom-up scientists. Ž . The observation that the actual situation point A in Fig. 2 seems sub-optimal lies at the heart of the debate between bottom-up and top-down modelers Ž . Scheraga, 1994 . There is a rich literature on possible explanations for this energy Ž . Ž . efficiency gap see introduction . In this paper we follow Jaffe and Stavins 1994a,b who distinguish between market failure and non-market failure explanations of the energy efficiency gap. Market failures relate to limited information surrounding investments in energy efficiency. Information about energy efficiency is underprovided by ordinary mar- ket activity, due to the public goods nature of information. Once a firm creates information, it cannot prevent others from using the same information without paying for it. Similarly, the act of implementation itself creates a positive external- ity by providing useful information to other potential adopters. For instance, it may turn out that realized returns of investments fall short of the returns promised by engineers, so that the case for an energy efficiency gap may be weaker than Ž previously believed see Hassett and Metcalf, 1997 for this point in the case of attic . insulation . Finally, if the potential adopter does not share in the energy costs, the C.C. Koopmans, D.W. te Velde r Energy Economics 23 2001 57]75 61 useful information in his hands remains unexploited. Such principal]agent prob- lems may arise for instance if tenants, and not the landlord, pay the energy bill. Non-market failure explanations reflect why observed behavior is optimal, ex- plaining why it is rational to use a high discount rate or to wait and see. For instance, irreversible investments in energy efficiency whose benefits depend on uncertain, future energy prices require the use of high discount rates by risk-averse investors. Uncertainty may also explain why a wait-and-see policy can be optimal. Ž . Metcalf 1994 argues that the ability not to invest in energy efficiency until more Ž information is available increases the net present value of non-investment option . value . Another non-market failure explanation is that technologists’ estimates of costs of techniques often fail to incorporate transaction costs. Yet another non- market failure explanation is that most bottom-up calculations are based on the average firm. In fact, a technique may be profitable for an average firm in the sector, but not for all firms facing heterogenous cost functions. Finally, imperfec- tions in capital markets may impede investments. Some firms must pay substantial premia to obtain loans. If these firms are taking higher risks than others, then this is optimal from the lender’s point of view. Fig. 3 summarizes the discussion, by pointing out which hurdles need to be overcome when incorporating bottom-up information on energy saving techniques Ž . in top-down models a move from E to A . Bottom-up analysts argue that a certain Ž . number of investments in energy efficiency are technically possible level T , a part Ž . of which is profitable level E . However, adjusting for non-market failures Ž . rational use of high discount rates , means that fewer investments in energy Ž U . efficiency are profitable level E . Finally, after accounting for market failures, e.g. limited information, fewer investments will be achieved than are economically efficient. We have thus reached point A, representing the level of efficiency actually achieved. Fig. 3. Concepts of energy efficiency. C.C. Koopmans, D.W. te Velde r Energy Economics 23 2001 57]75 62 One final notion about the energy efficiency gap is worth mentioning. Some argue that neither point A nor points E and E U could be socially efficient. Such a social optimum can be attained if energy prices fully internalize the environmental damages of energy use, and after the abolishment of subsidies on energy use. However, in this paper we abstract from the socially optimal level of energy efficiency. We do not state which of the explanations of the energy efficiency gap call for government intervention. In this paper, we are merely interested in the behavioral modeling of investing in energy efficiency.

3. ICARUS