An important part of climate policy debates is underpinned by a lasting controversy between believers and non-believers in the existence of a large untapped efficiency potential in the economy. If there, this potential could be realized at such a small societal cost that it would be more than compensated by cost savings that accrue from the efficiency improvements. Options that have a negative net social cost add up to an overall negative cost potential that may be quite large. Figure 8.1 is a sketch of the successive marginal costs of abatement, as a function of GHG reduction relative to some baseline point A. The total cost is simply the area between the curve and the horizontal axis. From A to B, marginal abatement costs are negative, and from B onwards, they are positive. The debate revolves around the size of the total (negative) cost from A to B. The studies discussed in this subsection argue that the negative cost area is potentially quite significant, and compensates to a large extent for the positive costs incurred after point B. Most other B-U studies analyzed in the next subsection do not even attempt to evaluate the relative positions of points A and B, since they optimize the system even in the absence of carbon constraint, and thus compute only the points beyond point B.
Figure 8.1: A typical cost curve.
Krause (1995, 1996) identifies two main reasons why the negative cost area may be quite large: untapped potential for efficiency gains mainly in end-use technologies, both on the demand and on supply sides. Several major studies concretize this view in Europe as well as in North America (USA and Canada). For Europe, the monumental IPSEP reports (summarized in Krause et al. (1999)) conclude that emissions could be reduced by up to 50% below the 1990 level by 2030, at a negative overall cost. This involves the judicious implementation of technologies and practices in all sectors of the economy, and the application of a large number of government policies (incentives, efficiency standards, and educational). In the US, some of the 5-Lab studies (Brown et al., 1997a, 2000, particularly the HE/LC scenario) indicate that the Kyoto reduction target could be reached at negative overall cost ranging from US$7 billion to US$34 billion. Another study based on the NEMS model (Koomey et al., 1998) indicates that 60% of the Kyoto gap could be bridged with an overall increase in the US GDP. The latter study contrasts with another NEMS study (Energy Information Administration, 1998) that indicates GDP losses from 1.7% to 4.2% (depending on the extent of permit trading and sink options) for the USA to reach the Kyoto targets. Laitner (1997, 1999) further stresses the impact of efficient technologies on the aggregate cost of mitigation in the USA. In Canada, the MARKAL model was used with and without certain efficiency measures in various sectors (Loulou and Kanudia, 1998; Loulou et al., 2000): the results show costs of Kyoto equal to US$20 billion without the additional efficiency measures, versus US$26 billion when efficiency measures are included in the database. Again in Canada, the ISTUM model was used (Jaccard et al., 1996, Bailie et al., 1998) considering a set of pro-active options. For example, in the residential sector large emissions reductions of 17% to 25% relative to 1990 could be achieved as early as 2008 with many negative costs options, and beyond that level of reduction, the marginal costs is ranging from US$25 to US$89/tC.
As extensively discussed in SAR, many economists argue that the real magnitude
of negative cost options is not so large if account is taken of:
These arguments should not be used to refute the very existence of negative cost potentials. They indicate that the applicability of non-price policy measures apt to overcome barriers to the exploitation of these potentials must be given serious attention. Some empirical observations do confirm that active sectoral policies can result in significant efficiency gains, in demand-side management for electricity end-uses for example. However, the many sources of gaps between technical costs and economic costs cannot be ignored (see the taxonomy of Jaffe and Stavins, 1994). The few existing observations (Ostertag, 1999) suggest that the transaction costs may represent, in many cases, a large fraction of the costs of new technology, and there is always an uncertainty about the efficiency and the political acceptability of the policies suggested in the above studies. This issue is clearly exemplified by the set of studies carried out in the USA and collected under the name 5-LAB studies. In these, some scenarios produce positive incremental costs and others negative costs, depending on the aggressiveness with which efficiency measures are implemented (Interlaboratory Working Group, 1997; Brown et al., 1998).
Contrary to the studies discussed above, the partial equilibrium studies reviewed in this section do not report negative costs. This is because the least-cost algorithms employed, which are powerful to compute the incremental cost of the system with and without a carbon constraint (i.e., point B in Figure 8.1), demand a set of somewhat arbitrary parameters to be calibrated in such a way that they calculate a suboptimal baseline; but such an operation demands resorting to a set of somewhat arbitary parameters and the results are less easy to interpret. This is why the B-U studies reported hereafter explore only the section of the cost curve with positive carbon prices (section CD in Figure 8.1).
It is very hard to encapsulate in a short presentation the many studies carried out with a B-U approach using a crosscutting, carbon-pricing instrument. Figure 8.1 summarizes a number of these results, obtained with a variety of B-U models applied to a single Annex I country or region, ignoring the trade effects. Included are those studies that contain enough information to present the marginal abatement cost along with the level of GHG emission variation from 1990 (other studies that reported only the total abatement cost are discussed separately). In Figure 8.2, each point represents one particular reduction level (relative to 1990) and the corresponding marginal cost of reduction. Points that are linked together by a line correspond to a multi-run study effected with the same model, but in which the amount of reduction was varied.
Evidently, Figure 8.2 shows considerably discrepancies from study to study. These large variations are explained by a number of factors, some of which reflect the widely differing conditions that prevail in the countries studied, while others result from the modelling and scenario assumptions. These variations are discussed next, illustrated by examples from Figure 8.2.
Several studies are not represented in Figure 8.2,
since only incremental or average costs were reported. For instance, a German
study (Jochem, 1998) indicates reductions of 30% to 40% in 2010 at average costs
ranging from US$12 to US$ 68/tCO2eq. In Canada (Loulou and Lavigne,
1996), a measure of the impact of demand reduction is obtained by running MARKAL
with and without elastic demands for energy services: the total cost is US$52
billion with fixed demands, andUS$42 billion with elastic demands. Chung et
al. (1997) arrive at much higher total costs for Canada, using a North American
equilibrium level (the higher cost apparently results from fewer technological
options than in MARKAL) A Swedish MARKAL study (Nystrom and Wene, 1999) find
total cost of 210 billion Swedish krona for a stabilization scenario, against
640 billion Swedish krona for a 50% emissions reduction in 2010. This same study
investigates the opportunity cost of a nuclear phase out, and evaluates a rebound
effect on the demands of a 9% emissions reduction for Sweden.
Other reports in this collection