In light of this inefficiency with standards, economists have long argued for market-based approaches to regulate pollution, using such policy tools as emission taxes and tradable permits. U.S. policymakers use tradable permits to control sulfur dioxide emissions from power plants and nitrogen oxide emissions from industrial point sources. These policies permit firms to trade allowances on a ton-for-ton basis. As a result, pollution sources that can reduce their emissions easily will be financially rewarded for doing so by other sources that face much higher costs.

Such policies are effective at reducing abatement costs; however, they currently overlook something that may be even more important: the damages from pollution emissions such as sulfur dioxide and nitrogen oxides depend upon where they are released. If the emissions occur near a large population center that already experiences bad air quality, the damages will often be much higher than if they occur in a sparsely populated, pristine area.

A number of policy instruments have been proposed to remedy this. Emission taxes could be calibrated to the marginal damage of emissions at each location. Trading regimes could also be adjusted to reflect spatially-variant damages. For example, markets for a cap-and-trade program could be subdivided into homogenous submarkets, allowing ton-for-ton trading within each submarket. The problem with these approaches is that they create very thin markets that may be vulnerable to manipulation. Alternatively, regulators could establish fixed exchange rates between sources that are inversely proportional to the ratio of the marginal damage of emissions. Permit trading programs that feature such fixed exchange rates effectively capture the heterogeneity in marginal damages, achieving pollution reductions at minimum social cost.

In a 2009 paper, we developed a fixed exchange rate policy design for air pollution. We summarize our findings in this article. Instead of allowing firms to exchange permits on a ton-for-ton basis, the permits in our system reflect the relative harm caused by emissions from different regulated sources. The exchange rates are equal to the inverse of the ratio of the firms’ marginal damages per ton of emissions.

Of course, this efficient system requires that regulators know source-specific marginal damages in order to determine the exchange rates between sources. Regulators historically lacked these data, and so they could not implement policies featuring trading ratios. However, given recent advances in computational power and model design, the marginal damages from all of the point and aggregated nonpoint air pollution sources in the continental United States can now be calculated. We can use the Air Pollution Emission Experiments and Policy (APEEP) model to track the consequences of emissions through air quality modeling, exposure, dose-response, and valuation for many individual sources and pollutants.

With these data, we computed the marginal damages for all 10,000 sources measured by the U.S. Environmental Protection Agency for six air pollutants: coarse particulate matter (particles up to 10 micrometers in size), fine particulate matter (particles up to 2.5 micrometers in size), nitrogen oxides, sulfur dioxide, volatile organic compounds, and ammonia. Using these marginal damages, we then calculated trading ratios between each pair of sources. The analysis highlights how the marginal damages vary across space. Damages are greatest, per ton of emissions, for sources in urban locations. For example, emissions of sulfur dioxide in large eastern cities cause damages that are 50 times larger than equivalent emissions produced in rural western locations. By employing the trading ratios discussed above, efficient policies would reflect these differences by encouraging considerably more abatement in high-damage urban areas relative to rural sources.

THE EMPIRICAL MODEL

APEEP is a traditional integrated assessment model in its basic structure; it connects emissions, through air quality modeling, to exposures, physical impacts, and valuation. The 10,000 pollution sources in the model consist of all the individual and grouped sources identified by the EPA. Sources of emissions encompass both ground-level sources as well as point sources. Ground level sources include smaller sources such as vehicles, residences, and small industrial or commercial facilities without a smokestack. The discharges from ground-level sources are reported in aggregate for each county by the EPA. As a result, APEEP documents the county location of each ground-source emission; however, the model cannot determine where within the county the emission occurs. In contrast, point sources, such as power plants and other large industrial or commercial establishments with a tall smokestack, are modeled individually. Further, point sources are differentiated by smokestack height because height can affect the dispersion of emissions and, ultimately, the damages generated from these sources.

APEEP permits a characterization of the variation in marginal damages across sources and across space. These damage estimates allow policymakers to calibrate the efficient trading ratios discussed above: the inverse ratio of firms’ marginal damages. To calculate marginal damages, the model first estimates total damages from all reported emissions in the model (which uses data from 2002). APEEP then adds one ton of one pollutant at one source to reported baseline emissions and recomputes national damages. The difference between the two calculations is the marginal damages. Note that no other source has its emissions change from the baseline levels. By holding all emissions (except the additional ton) at baseline levels, this approach isolates the source and pollutant-specific damage per ton. This algorithm also captures secondary pollutants, such as certain components of fine particulate matter and tropospheric ozone, that are formed from emissions of the six pollutants. APEEP attributes the damage due to such secondary pollutants back to the source of emissions.

This marginal damage algorithm is repeated for each of the six pollutants and the approximately 10,000 sources covered in APEEP. This involves 60,000 repetitions of the simulation design described above. Further, the 10,000 sources in APEEP include all of the anthropogenic emissions of these six pollutants in the continental United States.

Although APEEP computes damages due to health effects as well as reduced crop and timber yields, materials depreciation, lost recreation services, and reduced visibility, previous research has shown that the largest share of damages stems from impacts on human health, especially premature mortality rates. Therefore, the following discussion highlights the methodology used in APEEP to model mortality impacts. APEEP includes the relationship between long-term exposures to fine particulate matter and adult mortality rates reported by C. A. Pope et al. in a 2002 paper, and fine particles and infant mortality rates reported in a 2006 paper by T. J. Woodruff et al. Additionally, APEEP models the effect of exposures to ozone on mortality rates among all populations by using the findings from a 2004 paper by M. L. Bell et al.

The final stage of the APEEP model attributes a dollar value to the impact on mortality rates due to exposures to fine particles and ozone. The value affixed to premature mortality risks in APEEP is derived from J. R. Mrozek and L. O. Taylor’s 2002 meta-analysis of studies that estimate hedonic wage models. They report that an additional 1:10,000 chance of accidental death is worth a wage premium of roughly $200 per year. This literature commonly refers not to the wage premium directly, but rather to the Value of a Statistical Life (VSL). Specifically, dividing the risk premium by the change in the probability of death yields a VSL of $2 million. In this analysis, the VSL is tailored to the age of the exposed population. This involves the use of detailed mortality rate data to calculate the expected years of life remaining for each census-defined age group. Next, we infer the value of a year of life so that the present value of life for a middle-aged male worker is equal to the VSL reported by Mrozek and Taylor. We then assume the value of a life-year is the same for all people regardless of age. (While the value of an expected year of life might vary by income, age, gender, race, and other demographic characteristics, those differences have historically not been used in environmental policy.) This method places a relatively lower value on mortality risks faced by the elderly, relative to younger populations, since the elderly have fewer life-years remaining. The modeling choice regarding whether (and how) to tailor the value of mortality risks to populations of different ages is especially important because the elderly constitute a large proportion of the population affected by air pollution.

RESULTS

The marginal damages from the nearly 10,000 U.S. sources of the six air pollutants covered in APEEP are reported in Table 1. The median marginal damages of emissions for ammonia, sulfur dioxide, and fine particles are $900, $970, and $1,170 per ton, respectively. The median marginal damages of emissions for coarse particles, nitrogen oxides, and volatile organic compounds are in the neighborhood of $200 per ton.

Table 1 also shows the significant degree of spatial variation in the marginal damages of emissions across the United States. The difference between the marginal damages caused by low-damage and high-damage sources is largest in magnitude for ammonia and fine particulate matter. The range for fine particles between the lowest-damage sources (those in the first percentile of the distribution) and the highest-damage sources is $41,000 per ton and the equivalent range for ammonia is $59,000. In contrast, this range for nitrogen oxides is just $2,000.

The maps in Figures 1 and 2 show the distribution of marginal damages for fine particles and sulfur dioxide across the United States for emissions from ground-level sources. The substantial difference in the impact of a ton of fine particulate matter, for example, that is evident in Table 1 is clearly shown in Figure 1. Sources in the lower percentiles of the distribution are in rural areas in the western United States. Sources with marginal damages of emissions near the 50th percentile are found in suburban locations and small urban areas. Sources whose emissions produce the largest marginal damages are located in the largest metropolitan areas. A similar spatial pattern exists for sources of sulfur dioxide. This can be observed by examining Table 1 and Figure 2.

Table 2 reports the ratios of marginal damages for sulfur dioxide between six source locations that reflect the first, 25th, 50th, 75th, 99th and 99.9th percentiles of the distribution of sulfur dioxide sources. This table provides an example of the trading ratio table that would be necessary to implement an efficient trading regime. For example, let MD01 represent the damage caused by one ton of sulfur dioxide emitted for the first percentile source, and MD25 represent the damage caused by one ton of sulfur dioxide emitted for the 25th percentile source. An example of a source location at the first percentile is Klamath County, OR, and an example of a source at the 25th percentile is Tyler County, TX. The trading ratios shown in Table 2 reflect the number of tons of abatement at each column source that are equal (in terms of avoided damages) to one ton abated at each row source. The trading ratio is simply MD01 ÷ MD25, which equals 0.4. This ratio implies that it takes 2.5 tons of sulfur dioxide abated from Klamath County to equal one ton from Tyler County — a relatively small difference. But consider the more stark example of Hudson County, NJ (which is located just upwind of New York City); it would take 50 tons of Klamath County abatement to offset one ton of Hudson County sulfur dioxide emissions.

Figures 1 and 2 show that trades between sources in the first and 99.9th percentiles correspond to trades between rural sources and sources located in the largest cities such as New York and Los Angeles. These large exchange rates (between approximately 50:1 for sulfur dioxide and 170:1 for fine particles) reflect the higher benefits of abatement in large cities, due to the large exposed populations, relative to the lower benefits of abatement in rural (low population) areas.

Under the United States’ current system of tradable permits, a facility in an urban area looking to buy an allowance to emit an additional ton of pollution can purchase that allowance from any source at one market-determined price. That allowance then legally permits the emission, even though the additional emission may cause more damage than the emission it replaces if the source purchases the allowance from a rural facility. Such a trade should be discouraged because it increases the overall damages from emissions by moving emissions to places where they cause greater harm. Discouraging such exchanges is the essence of our trading ratio policy.

Our policy proposal raises the relative price of permits in high-damage (urban) locations, discouraging more emissions in those places. Since buying permits and continuing to emit pollution becomes more expensive for urban sources, they are encouraged to do more abatement rather than purchase permits. The opposite holds for firms in rural locations; since it is relatively cheaper for them to buy permits, such facilities are encouraged to buy permits and generate relatively greater quantities of emissions. The degree to which firms face different prices for permits is dictated by the trading ratios. Since these are calibrated to the marginal damages of each firm’s emissions, firms trade on the basis of damages, not tons. The greater the difference in damages caused by emissions from any pair of firms, the higher the exchange rate and the greater the spread between the prices that they face for pollution allowances.

Since damages from the air pollutants studied in this paper tend to follow a rural-urban gradient, a policy employing these efficient trading ratios would encourage substantially more abatement in the metropolitan areas. Whether emissions in rural areas increase relative to current levels would depend on the aggregate emission limits. If aggregate abatement increases, rural emissions are not likely to increase by very much. Even if aggregate emissions increase, rural concentrations of pollution will likely remain far below urban concentrations because emissions of most pollutants are so highly concentrated in metropolitan areas. However, extant ambient air standards could also be used to prevent emissions from relocating into areas that exceed the standards.

Conclusion

Our proposal shows how regulators of pollution can design efficient air pollution control policies that reflect the damages caused by emissions from regulated sources. The next step in the progression of market-based environmental regulations requires estimates of the source-specific marginal damages of emissions. Our 2009 paper reports such estimates and shows that, in the context of cap-and-trade policies, achieving efficiency depends on policymakers setting trading ratios or exchange rates between firms to govern trades of pollution allowances between and among firms. The trading ratios should reflect the marginal damage of emissions from each source. The trading ratios get the prices right by reflecting the damage of emissions.

The marginal damages of pollution emissions reported in our 2009 paper are highest in large metropolitan areas and much lower in sparsely populated rural areas. In light of this pattern, the policy design proposed in this paper argues that a substantial increase in abatement should occur in cities and especially in large metropolitan areas. The policy also implies that emissions will shift to more rural locations where damages per ton are considerably lower. On net, this reduces the damage caused by pollution.

Abatement yields greater benefits when conducted in densely populated cities relative to rural regions. As such, this policy argues that an optimal policy must reflect this difference in benefits, with the goal of maximizing the return to society’s investments of scarce resources in environmental quality.

Readings

• “Efficient Pollution Regulation: Getting the Prices Right,” by N. Z. Muller and R. O. Mendelsohn. American Economic Review, Vol. 99 (2009).
• “Fine Particulate Matter (PM2.5) Air Pollution and Selected Causes of Postneonatal Infant Mortality in California,” by T. J. Woodruff, J. D. Parker, and K. C. Schoendorf. Environmental Health Perspectives, Vol. 114, No. 5 (2006).
• “Lung Cancer, Cardiopulmonary Mortality, and Long-Term Exposure to Fine Particulate Air Pollution,” by C. A. Pope, R. T. Burnett, M. J. Thun, E. E. Calle, D. Krewski, K. Ito, and G. D. Thurston. Journal of the American Medical Association, Vol. 287, No. 9 (2002).
• “Markets in Licenses and Efficient Pollution Control Programs,” by W. D. Montgomery. Journal of Economic Theory, Vol. 5 (1972).
• “Measuring the Damages from Air Pollution in the United States,” by N. Z. Muller and R. O. Mendelsohn. Journal of Environmental Economics and Management, Vol. 54, No. 1 (2007).
• “NOx Emissions from Large Point Sources: Variability in Ozone Production, Resulting Health Damages, and Economic Costs,” by D. Mauzerall, B. Sultan, N. Kim, and D. F. Bradford. Atmospheric Environment, Vol. 39 (2005).
• “On Marketable Air Pollution Permits: The Case for a System of Pollution Offsets,” by A. J. Krupnick, W. E. Oates, and E. Van de Verg. Journal of Environmental Economics and Management, Vol. 10 (1983).
• “Ozone and Short-Term Mortality in 95 U.S. Urban Communities, 1987–2000,” by M. L. Bell, A. McDermott, S. L. Zeger, J. M. Samet, and F. Domenici. Journal of the American Medical Association, Vol. 292, No. 19 (2004).
• “The Empirical Properties of Two Classes of Designs for Transferable Discharge Permit Markets,” by S. E. Atkinson and T. H. Tietenberg. Journal of Environmental Economics and Management, Vol. 9 (1982).
• “Transferable Discharge Permits and the Control of Stationary Source Air Pollution: A Survey and Synthesis,” by T. H. Tietenberg. Land Economics, Vol. 56, No. 4 (1980).
• “What Determines the Value of Life? A Meta-Analysis,” by J. R. Mrozek and L. O. Taylor. Journal of Policy Analysis and Management,