Almost all the studies analyze the effects of a GHG reduction policy through a tax on carbon. The ranges of the tax are from modest levels (RMB Y9/tC7) in 2010 for Garbaccio et al. (2000), US$10/tC for Burtraw et al. (1999); up to high levels (US$254/tC for Dessus and OConnor (1999), and US$840/tC for Brendemoen and Vennemo (1994). The US studies employ relatively modest taxes, between US$10/tC and US$67/tC. Only two studies consider alternative programmes: Aunan et al. (2000) considers a National Efficiency Programme, and Cifuentes et al. (2000) considers energy efficiency improvements. The level of abatement considered by these two studies is relatively modest, however.
An analysis of ancillary benefits requires a time line and a clear definition of the key constituents of the baseline against which the prospective scenario can be measured, including the economic, demographic, regulatory8, environmental9, and technological conditions, and their implications for emissions or other inputs to an ancillary benefit calculation.
The importance of the baseline is evident in a review of previous studies for the USA in Burtraw et al. (1999). Assessments varied in their estimates of ancillary benefits, chiefly because they employed different assumptions regarding the regulatory baselines, that is the 1990 US Clean Air Act Amendments and, especially, the tradable permit programme for SO2. Among these baseline parameters, the most critical are the spatial location of emissions relevant to potentially exposed populations, regulatory conditions, and available technologies (Morgenstern, 2000). The importance of the location of emission reductions and exposed populations means that highly disaggregated models are the preferred tools of analysis. This may conflict with other goals for the analysis of GHG mitigation strategies. For example, large CGE models, which are used for cost estimation, operate at a different scale than the more localized models relevant to estimating ancillary benefits.
Most of the studies in Table 8.5 use static or dynamic CGE models (one uses an econometric model) that provide T-D and sectorally aggregate estimates of ancillary benefits and/or costs. The Burtraw et al. (1999) model stands out for the location specificity of its economic model (although only for the electricity sector), which permits more credible modelling of population exposure reductions than that from spatially aggregate models. Another specific feature is its detailed representation of investment choices and their dependence on other factors covered in the model. Finally, several studies do not use an economic model. Instead, they follow a B-U approach, positing some increase in energy efficiency or reduction in carbon and estimating the ancillary benefits that would result, at a reasonably detailed spatial level. Such studies suffer from not accounting for behavioural adjustments, such as energy substitutions, which could alter their estimates of ancillary benefits considerably. The high ratio of ancillary benefits to the carbon tax for Garbaccio et al. (1999) appears to arise from very optimistic assumptions about energy substitution elasticities.
Emissions and Environmental Media Modelling
All the studies in Table 8.5 account for the most important pollutant affecting public health particulates. Most, however, do not consider secondary particulate formation from SO2 and NOx, or do so in a very simplistic manner. In a developing country, direct particulate emissions are likely to be a large fraction of particulate mass, making the lack of attention to secondary products less important. In developed countries, however, secondary products are likely to be far more important than primary particulates. Omitting these products could bias ancillary benefit estimates downwards; using proportionality assumptions or other simple approaches raises uncertainties and may carry biases. Only one study considered lead emissions (Dessus and OConnoer, 1999); few address ozone.
The Abt study (Abt Associates and Pechan-Avanti Group, 1999) is the most comprehensive in its modelling of secondary particulate formation and dispersion. It found that 12 urban areas in the USA would come into compliance with the recently promulgated standard for particulate matter less than 2.5 microns (PM2.5)10 for a carbon tax of US$67 (US$1996). Without this tax, these areas would not be able to meet the new standard. With there being at best sparse information on the actual PM2.5 concentrations in US urban areas, these estimates should be viewed as highly speculative.
Health Effects Modelling
Three recent studies (Hagler-Bailly 1995; Lee et al., 1995; European Commission, 1999) developed methods that set the stage for much of the recent estimates of ancillary benefits. However, studies that draw on this literature, but reduce its information to coefficients that link emissions directly to health effects (or values) ignore spatial and demographic heterogeneity. This is particularly so when such coefficients are generated for one country or region and then directly applied to another, without taking into account local conditions. In the absence of country-specific information, transfer of risk information may be made between countries, with appropriate caveats to take into account underlying differences in health status, access to care, and other important factors (see Box 8.2).
|Box 8.2. The Impact of Air Pollution on Health Differs
For any society, deaths at earlier ages result in more productive years of life lost than for those that occur at later ages. One study in Delhi, India, found that children under 5 and adults over 65 years of age are not at risk from air pollution, because other causes of death (notably infectious diseases) predominated in those who survive to reach these age groups (Cropper et al., 1997). However, people between 15 and 45 years of age are at increased risk of death from air pollution relative to those in developed countries. Since the population distribution in India includes many more people in these middle age groups, the net impact on the country from air pollution measured in terms of years of life lost is similar to that of a developed country.
Most of the studies rely on concentrationresponse functions from the literature on health, and apply them using a standard methodology (Ostro, 1996; EPA, 1999). The most important health effects are premature mortality and chronic respiratory effects.
Aside from differences in the base rates of the effects11, due to local characteristics such as the age distribution of the population and health care services, other factors help explain the different outcomes of the studies. First, some use PM10, while others use fine particles (PM2.5), or serveral components of them (sulphates and nitrates). When the individual components of PM2.5 are used, the implicit assumption is that their risk is similar to that of PM2.5. To date, this has not been verified (especially for nitrates, the secondary particulate product from NOx emissions). Second, studies that look at age groups separately generally report higher impacts (Aunan et al. (2000), for example, used a steeper doseresponse coefficient for people older than 65 years of age than that used by other studies). Very few consider the chronic effects on mortality, derived from cohort studies (e.g., Pope et al., 1995) (Abt Associates and Pechan-Avanti Group, 1999 is one, while others consider it for their high estimate only). Use of the latter results in estimates of death three times larger than use of the time series studies. Also, few studies consider effects on child mortality. Finally, different studies consider different health endpoints, which is important for reconciling morbidity estimates.
Valuation of Effects
The most important monetary benefit is related to mortality risk reductions, which can be expressed in terms of the VSL (see Chapter 7). The VSL should ideally be indigenously estimated (Krupnick et al., 2000)12 but almost of the studies build on a consensus on the appropriate values to use (Davis et al., 2000), given the state of research on valuation (albeit concentrated in the UK and USA).
A major difference in the treatment of values across the studies is whether these values are adjusted for different income levels and increased for future income growth. Adjustments that assume an income elasticity of willingness to pay (WTP) of 1.0 are inconsistent with the admittedly thin literature. A number of studies found elasticities in the 0.2-0.6 range based on income differentials within a country. Such elasticities, when applied to transfers among countries, yield quite high values. Most of the developing country ancillary benefit studies reported in Table 8.6 use an income elasticity of 1.0. The US Science Advisory Board has endorsed the idea of making adjustments for future income growth within a country.
The state of the art of the valuation of air pollution-related mortality effects is currently in ferment, with serious questions being raised about the inappropriateness of basing such valuation on labour market studies. Ad hoc adjustments for the shorter life span of those thought to be most affected by air pollution (the elderly and ill) have been made but more credible estimates of willingness to pay await new research. Such efforts are more likely to lower such estimates relative to current estimates than raise them (see Davis et al., 2000 and Krupnick et al., 2000).
All the studies, except those in the USA, assume that improvements in public health count as externalities and, hence, as ancillary benefits. As noted in Krupnick et al. (2000), this assumption may not always hold. Burtraw et al. (1999) and Abt Associates and Pechan-Avanti Group (1999) count the abatement cost savings from reducing SO2 emissions in response to a carbon tax because SO2 emissions are capped in the USA. Similar adjustments are not made for SO2 and other pollutant taxation in Europe. Moreover, not all ancillary benefits are necessarily externalities. In some cases, these effects may be already internalized in the price of goods and services: for example, where accident insurance against road fatalities exists, much of this effect is already accounted for through purchasing insurance and the penalties for failure to obtain it.
Treatment of Uncertainty
The uncertainty that surrounds the estimates of benefits is no less than that associated with mitigation costs, extending from physical modelling, through valuation, to modelling choices. Several of the studies use Monte Carlo simulation, but others use less sophisticated sensitivity analyses to characterize uncertainties.
Allowance for Ancillary Costs
None of the studies reviewed in this assessment reported estimates of ancillary costs. Some studies, such as Burtraw et al. (1999), discuss the bounce-back effect associated with energy substitution to natural gas and other less carbon-intensive fuels. However, even these studies, not surprisingly, estimate positive net ancillary benefits from GHG mitigation policies. The issue is whether the models were designed to capture ancillary costs. In general, our conclusion is no, except for fossil fuel substitution in the power and transport sectors. From an energy substitution perspective, substitution to nuclear power or hydropower does not generate reported ancillary costs because these ancillary effects are not present in the studies. Other sources of ancillary costs were also left out of the modelling exercise, either because of model boundaries or through making some standard modelling choices. All the studies examined effects on one country or region, and therefore do not consider the leakage effect. None of the studies considered health linkages that might result from slower income and employment growth following the implementation of a GHG mitigation policy.
Other reports in this collection