As pointed out by Grubb (1997), there are several key assumptions imbedded in the energy-economy models that influence the shape of the least-cost mitigation pathway. For a pre-determined target, these relate to the determinants of technical change; capital stock turnover and the inertia in the energy system; discounting; and, the carbon cycle. When the target is uncertain, they include in addition the probability attached to each target and risk aversion (see Chapter 10) which tend to favour a more aggressive departure from current trends.
The discount rate will not be discussed because it is less important in cost-efficiency frameworks (when the target is pre-determined) than in a cost-benefit one (when the discount rate reduces the weight of future environmental impacts, see Chapter 10). Neither are the very few studies discussed which try to assess different benefits in terms of environmental co-benefits of reducing GHG emissions presented. Wigley et al. (1996), for example, show pathway-related differentials up to 0.2°C in global mean temperature and 4cm in global mean sea-level (by 2100) for the WGI and WRE 550 stabilization pathways. See Chapter 10 for an elaboration of these timing issues. This part will rather insist on the key features of technical change that are numerically of utmost importance.
To the extent that the cost of reducing emissions is lower in the future than at present, the overall cost of stabilizing the CO2 concentration is less if emissions mitigation is shifted towards the future. This shift occurs in all models. The extent of the shift that minimizes the cost of limiting the concentration of atmospheric CO2 depends, at least in part, on the treatment of technological change. Without technological change, the problem is simple and the results of Hotelling (1939) apply. With endogenous technological change, the problem becomes more complex.
This discussion of the determinants of technological change must begin with the acknowledgement that no adequate theory of endogenous technological change exists at present. Many researchers have contributed to the field, but the present state of understanding is such that present knowledge is partial and not necessarily fully consistent. Although no complete theory of technological change exists, two elements have been identified and explored in the literature: induced technological change (ITC) and learning-by-doing (LBD). Work by Ha-Duong et al. (1997), Grubb et al. (1995), Grubb (1997), and Kypreos and Barreto (1999) examined the implication of ITC, LBD, and inertia within the context of uncertainty and an imperative to preserve the option of concentration ceilings such as 450ppmv. They conclude that emissions mitigation can be shifted from the future towards the present under appropriate circumstances.
Goulder and Mathai (1998) also explore how the effect on timing depends on the source of technological change. When the channel for technological change is R&D, ITC makes it preferable to concentrate more abatement efforts in the future. The reason is that technological change lowers the costs of future abatement relative to current abatement, making it more cost-effective to place more emphasis on future abatement. However, when the channel for technological change is LBD, the presence of ITC acts in two opposite directions. On the one hand, ITC makes future abatement less costly but, on the other hand, there is an added value to current abatement because such abatement contributes to experience or learning and helps reduce the costs of future abatement. Which of these two effects dominates depends on the particular nature of assumptions and firms. In recent years, there has been a good deal of discussion about the potential for ITC (e.g., Anderson et al., 1999). Proponents argue that such changes might substantially lower, and perhaps even eliminate, the costs of CO2 abatement policies. These discussions have exposed very divergent views as to whether technological change can be induced at no cost, or whether a resource cost is involved. For example, in a 1995 article, Porter and van der Linde (1995) contend that properly designed regulation can trigger innovation that may partially or more than fully offset the costs of compliance. Indeed, they argue that firms can actually benefit from more stringent regulation than that faced by their competitors in other countries. However, in an accompanying article, a strongly contrary view is put forward by Palmer et al. (1995). Examining available data, they found that such offsets pale in comparison to expenditures for pollution abatement and control.
Including R&D driven ITC in climate mitigation models leads to ambiguous results in terms of time profile and tax level in a cost-benefit framework In a cost-effectiveness framework, the optimal tax is lower in the case with R&D driven ITC and has to be set up early even if the effective resulting abatement shifts from the near-term to the more distant future. If there are market failures in the R&D market (e.g., knowledge spillover), then subsidies for R&D are justified as it enhances social welfare and raises the abatement level (Goulder and Schneider, 1999; Weyant and Olavson, 1999; Goulder and Mathai, 2000).
However, R&D driven-ITC can reduce the gross costs of a carbon tax under special circumstances. Specifically, if R&D has been substantially over-allocated towards the fossil fuel industries prior to the imposition of a carbon tax, the carbon tax can reduce this allocative inefficiency and, as a result, its costs can be quite low or even negative. A substantial prior misallocation towards carbon-intensive industries could occur if there were prior subsidies towards R&D in the fossil fuel industries (with no comparable subsidies in other industries), or if there were substantial positive spillovers from R&D in non-carbon industries (with no comparable spillovers in the fossil fuel industries). Under other plausible initial conditions, however, R&D driven-ITC raises, rather than lowers, the net social costs of a given carbon tax because of the crowding out of R&D from other sectors; to put it clearly the tax level for a given abatement is lower than under the hypothesis of exogenous technical change and part of this decrease is offset when all the general equilibrium effects are accounted.
The same model has been employed to compare the costs of achieving a given
abatement target through carbon taxes and R&D subsidies (Schneider and Goulder,
1997; Goulder and Schneider, 1999). If there are no spillovers to R&D, the
least-cost way to reach a given abatement target is through a carbon tax alone.
The carbon tax best targets the externality from the combustion of fossil fuels
related to climate change, and thus is the most cost-effective. However, if
there are spillovers to R&D, the least-cost way to achieve a given abatement
target is through the combination of a carbon tax and R&D subsidy. If spillovers
are present, there is a market failure in the R&D market as well as a (climate
change related) market failure associated with the use of carbon. Two instruments
(the R&D subsidy and the carbon tax) are needed to address the two distinct
market failures most efficiently. In general, a R&D subsidy by itself does
not offer the least-cost approach to reducing carbon emissions. Results from
this model are highly sensitive to assumptions about the nature and extent of
knowledge spillovers. Further empirical work that sheds light on these spillovers
would have considerable value.
Other reports in this collection