More recent analysis has increasingly taken stock of the administrative costs involved in actual utility demand-side management programmes. These costs are estimated to range from negligible (Krause, 1994; Eto et al., 1994) to substantial (Joskow and Marron, 1993), depending primarily on the type of incentive programme involved;

policy effectiveness: the results of bottom-up studies vary as a function of the way they account for transaction costs and transition costs. At one end of the spectrum, it is assumed that a complete shift to efficient equipment will be achieved within one cycle of capital stock turnover (10 to 20 years); at the other end, political and institutional constraints inhibit the use of the most energyefficient technology over periods of 40 years or more.

In addition to differences in their approach to technology, bottom-up and topdown models typically contain important differences in their assumptions about consumer surpluses, so-called intangible costs, and the role of market barriers (Sutherland, 1991).

An example of a consumer surplus would be the extra satisfaction or value that a consumer derives from a particular automobile that may not be reflected in its capital or operating costs. Common examples of intangible costs are (1) the cost of becoming sufficiently informed about a new technology in order to consider it seriously as an option, (2) the perceived risks associated with the capital or operating costs of a technology, (3) the various transaction costs associated with finding, ordering, shipping, installing, operating, and maintaining a technology, and (4) externally or internally imposed restrictive investment criteria that differ from the social time preference of consumers and the opportunity cost of capital of firms. A difference in one or more of these intangible cost factors or consumer surpluses between two goods could be sufficient to offset differences in their tangible costs.

Top-down modellers, who tend to be economists, are generally reluctant to doubt the economic efficiency of business and household consumption choices, although they recognize the classical imperfections in markets (oligopoly, natural monopoly, subsidies, etc.). They therefore tend to assume that if a technology does not penetrate the market to the extent that a bottom-up engineering/economic analysis (which focuses only on tangible costs) suggests that it could, it is probably at least in part because the intangible cost and/or consumer surplus differences are at least large enough to make the investment unattractive. As a consequence of this general assumption, many economists are more likely to put their faith in values for AEEI and elasticity of substitution that have emerged from studies of the economy using statistical regression of historical aggregate data sets. Since the historical data are based on actual behaviour, behavioural parameters estimated from them may represent a synthesis of all tangible and intangible costs, including differences in consumer surplus.

Sweeping generalizations about bottom-up modelling are as hazardous as those

about the top-down aproach. However, bottom-up analysts in the past were more likely to be engineers and physicists. Their detailed knowledge of thermodynamic potentials and the energy-using characteristics of new and emerging technologies has led them to be aware of the apparent economic potential for society to adopt new technologies more quickly than in the past, at rates that, in aggregate, would imply values for AEEI and elasticity of substitution far higher than the values emerging from the historical data relied on originally by top-down modellers. This issue is particularly important when faced by the very long timeframe required by the climate change issue.

In addition, bottom-up modellers point out that econometrically derived relationships may incorporate market imperfections. While many top-down modellers assume that these imperfections are either negligible or very costly to correct, bottom-up modellers have noted the significant successes in fostering technological innovation and greater energy efficiency resulting from government initiatives during the energy crises of the 1970s and utility and government initiatives in the 1980s, and have pointed to these experiences as evidence that well-designed research and development and policy programmes could significantly affect the evolution of the AEEI term and even the elasticity of substitution. Thus, insofar as institutional reforms could lower transaction costs and remove barriers to the adoption of technologies whose life cycle costs are advantageous, there would be room for no-regrets improvements.

The empirical foundations of this position come from evaluation studies of the impacts of demand-side management programmes on end-use markets, product choices, consumers, manufacturers, and trade allies (DeCanio, 1993; Howarth and Anderson, 1993; Howarth and Winslow, 1994; Koomey and Sanstad, 1994; Krause, Vine, and Gandhi, 1989; Levine and Sonnenblick, 1994; Nadel, 1992). Though the argument has been challenged by various critics (Sutherland, 1991; Joskow and Marron, 1993), the authors of these studies suggest that there exists significant potential to capture the "negative cost potential" that can be realized by well-designed demand-side management programmes.

Given these results, bottom-up modellers tend to argue that appropriate policy will cause energy-efficient products to have lower transaction costs and risks than are generally assumed by firms and households, and perhaps even higher consumer surpluses. From this perspective the market as currently configured simply isn't delivering levels of energy efficiency that would be economically advantageous, not just at the microlevel of direct financial costs but also at more aggregate and inclusive levels of cost. They suggest that significant potentials for emission reductions can be realized at net economic savings through a combination of mandatory energy efficiency standards, labelling and auditing programmes, least-cost planning-oriented utility regulatory reforms, profit incentives for utility demand-side management programmes, market-pull incentives provided through innovative utility or government procurement programmes, and increased research and development and commercialization efforts for small-scale modular technologies with fast gestation periods, notably energy efficiency improvements

and renewables.

The differences in results between top-down and bottom-up modelling analyses are thus rooted in a complex interplay among differences in purpose, model structure, and input assumptions. The growing tendency for the development of hybrid modelling approaches means that differences of model structure are becoming a less important factor in many climate modelling studies, but the overall difference in perspective and thinking about energy markets represented by these two modelling approaches remains. Hybrid modelling approaches allow the exploration of the relative importance and implications of different input assumptions and are likely to narrow the difference in results between bottom-up and top-down studies. They will not by themselves, however, resolve the underlying question of whether energy markets are efficient with respect to the delivery of energy efficiency gains or not. Moreover, it remains to be seen whether these new approaches will be equally fruitful in fulfilling the range of purposes discussed above in Section 8.4.2.1. In particular, it seems clear that we need to improve our ability to explore long-term development paths or configurations of technology that are very different from those typical of experience in past decades (i.e., the alternative future scenarios that are often the subject of backcasting analyses). Much work remains to be done to address these larger issues in a satisfactory way.

8.4.4 Beyond Energy: Carbon Sinks and Nonenergy Greenhouse Gas Emissions

The focus of the literature on the potential for controlling CO2 from energy sources has developed in part because of the ready availability and adaptability of models that were designed to analyze energy markets. However, because of the lack of similar ready-made models that could be adapted to analyses of carbon sequestration and reductions in emissions of methane, nitrous oxide, halogenated substances, and other greenhouse gases, debates in these areas have not been structured around the bottom-up and topdown modelling approaches. Nevertheless, controversies in these fields are, in fact, related to fundamental issues similar to the ones underlying the bottom-up versus topdown division in the energy field, namely, the reasons for a wedge between the direct cost of technical alternatives from an engineering viewpoint and the overall costs of their adoption and implementation if transaction costs and economic general equilibrium effects are included in the accounting.

84.4.1 Carbon Sequestration Studies

Carbon sequestration cost studies fall into four general categories: (1) studies of the cost of removal and storage of carbon dioxide from emission sources such as power plants, (2) studies of biomass energy technologies that allow displaced fossil carbon to remain undisturbed, (3) studies of practices to maintain and expand the biological carbon sink, particularly in forests, and (4) studies of technologies to expand the storage of carbon in wood products. Technologies for the removal and storage of carbon dioxide are prohibitively expensive at this time (see e.g., Riemer, 1993). Biomass energy is treated

elsewhere in this report as a renewable energy technology that lowers net emissions of carbon dioxide. The analysis of the costs of increasing the use of long-lived wood products is not well developed. Consequently, this review of carbon sequestration cost studies will concentrate on the expansion of forest and agricultural carbon sinks.

measures.

28

Most carbon sink cost analyses have examined the direct costs of specific technical Under this approach, forestry or agricultural practices are matched with appropriate geographic regions. The pairs are defined as unique technologies with specific production functions. Then, in a very simple analysis, three key variables are identified for each region/practice combination:

1.

the suitable land area for that practice (e.g., hectares);

2.

3.

the treatment cost and land cost for the practice (e.g., annualized costs per
hectare per year);

the annual carbon yield (e.g., tonnes of carbon per hectare per year).

Two critical results can be derived from these three pieces of data. The potential yield of carbon in tonnes per year, from a given region/practice can be derived by multiplying the first and third factors. The unit cost of carbon sequestration, in dollars per tonne, can be derived by dividing the second factor by the third. These results can then be combined across practices within a region to develop a supply curve for carbon sequestration.

The simple analysis described above has been employed by some studies (see, e.g., Moulton and Richards, 1990). However, estimation of the three variables listed above land area, land and treatment costs, and carbon yield - is not simple in practice. As discussed in Box 8.1 [BOX 8.1 DEALS WITH BASIC PRINCIPLES FOR ASSESSMENT OF THE WELFARE COST OF A TAX. THERE IS NO OTHER BOX WITH THIS CHAPTER. PLEASE CHECK REFERENCE.] above, there is no consensus on the definition of the summary statistic "dollars per tonne of carbon sequestration."

As with energy studies, differences among the studies that give rise to a wedge between direct and social costs occur at each stage of a carbon sequestration cost analysis. These stem from different assumptions about the suitability of forestry and agricultural practices, different treatments of land and forestry practice implementation costs, and different estimates of the expected accomplishments of the forestry practices and valuation of those accomplishments in terms of greenhouse gas emission benefits as well as double dividend GDP and natural resource conservation benefits.

As Table 8.4 indicates, a variety of forestry practices may contribute to increasing the size of forest and agricultural carbon sinks. Just as in the case of the best available technologies in the energy field, the appropriate forestry and agricultural practice for

carbon sequestration is, in theory, the one that provides promise as a cost-effective way to capture and store carbon (see, e.g., Parks and Hardie, 1995). The analysis of "intangible" factor costs or of "double dividends" has been hampered by the fact that many studies have imposed constraints on the forestry or agricultural practices or types of land that they consider. These have included requirements that the adopted practices decrease soil erosion (Moulton and Richards, 1990), meet local needs (Andrasko et al., 1991), or provide other environmental benefits such as habitat preservation. Ideally, these factors should be accounted for as beneficial secondary effects, which in many cases appear to be` higher than for energy efficiency.

[Table 8.4]

Significant differences in the cost assessments of sink studies can be traced back to disagreements about the assumptions and data related to the many factors that determine total costs: land costs, first treatment and maintenance costs, transaction costs, the valuation of wood and agricultural products, and the discount rate applied to expenditures.

With respect to land costs, studies of carbon sequestration costs in the United States have, for example, employed several methods to identify the social costs associated with conversion of land to forest. These have included the use of land rental rates derived from surveys (Moulton and Richards, 1990), the use of market prices adjusted for the elasticity of demand for agricultural land (Richards et al., 1993), the use of the estimated lost profits from removing the land from agricultural production (Parks and Hardie, 1995), and the use of consumer surplus loss from increases in food prices due to the constriction of agricultural land availability (Adams et al., 1993). In the absence of extensive experience, estimating land costs has proven difficult because of the low reliability of data on the elasticities of demand for agricultural land. Also, uncertainties about the future of government subsidies for agriculture raise questions about land costs. These subsidies tend to drive a wedge between the market prices and the social cost of land.

These accounting difficulties are obviously far more difficult in countries that do not have well-established land markets, that have land tenure laws that do not allow permanent transfer of land at all, or where the government owns a significant portion of the land. Further, even where land markets do function, the lack of data about market activities often renders land cost estimates speculative at best.

The transaction costs associated with establishing forestry and agricultural programmes can be significant. Even in well-developed market economies, the costs of programme administration can rise to as much as 15% of the total costs of land rental, establishment, and maintenance (Richards et al., 1993). It is reasonable to expect that