Depending on pollutant mobility, air pollution may be local (factory odors, smoke), regional (acid rain) or even global (greenhouse gas build-up, ozone depletion by CFC's). Obviously localized problems are easier to deal with than larger-scale problems.

Under the Clean Air Act Amendments of 1970, the EPA assumed principal authority from the states for controlling air pollution, and adopted the usual "command-and-control" (CAC) regulatory approach. The EPA distinguished conventional from hazardous pollutants. It defined uniform ambient air quality standards for each conventional pollutant (short- and long-term averages).

All conventional air pollutants have primary National Ambient Air Quality Standards (NAAQS) designed to protect human health; some also have secondary standards to protect aesthetics, vegetation, etc. The stricter standard applies. These standards were defined by the EPA without regard to cost of compliance. Enforcing these standards was left to the state governments. Not surprisingly, many states missed the federal deadlines for achieving primary standards. Areas out of compliance with these standards were designated as non-attainment areas.

States are required to file State Implementation Plans (SIP's) with the EPA. The EPA is authorized to deny federal sewage and transportation funds for states that don't file SIP's. (New Castle County has been threatened with this for non-attainment of ozone NAAQS).  All major new or modified pollution sources in non-attainment areas must achieve lowest achievable emissions rates (LAER) regardless of cost or control practices at other sources.

The EPA's prevention of significant deterioration (PSD) policies are intended to keep regions with high air quality from letting their air quality deteriorate toward the primary standards. PSD regions are designated as Class I, II or III depending on allowable deterioration.

The EPA's air quality standards are clearly inefficient. Most are based on poorly-documented health thresholds and target zero adverse health effects. They confuse ambient concentrations with actual human exposures. Costs of compliance are explicitly excluded from consideration. Uniform standards ignore critical locational differences and time factors. Even if the chosen standards were efficient, the CAC approach is still not cost-effective. The divergence between CAC and least-cost compliance is particularly high for standards of intermediate toughness.

Emissions trading policies can provide substantial efficiency gains. A system of emission reduction credits allows firms achieving more than the required level of control to either bank or sell credits for emissions reductions. Offset policies require new sources to offset their emissions by acquiring emission reduction credits from existing sources. Bubble policies define standards for multi-firm areas which can be satisfied by actual emission reductions and/or acquisition of emission credits. Netting policies treat individual firms as bubbles, so that emissions increases from one process can be offset by emissions reductions in other processes within the same firm.

Emissions trades are becoming increasingly common although they generally involve high transactions costs because they must be approved by regulators.

Title IV of the 1990 Clean Air Act Amendments established the Acid Rain Program, .  The primary goal was to reduce annual dulfur dioxide emissions by 10 million tons below 1980 levels via a two-phase tightening of the regulations on fossil fuel-fired power plants.  Phase I began in 1995 and affects 445 units at 110 utilities in 21 eastern and midwestern states.  Emissions data indicate that 1995 SO2 emissions at these units nationwide were reduced by almost 40% below their required level.  Phase II starts in 2000, and tightens emissions limits for large plants and sets restrictions on 2,000+ smaller plants fired by coal, oil, and gas.  The Amendments also targeted a 2 million ton reduction in NOx emissions by the year 2000.

The most innovative aspect of the Acid Rain Program created a market for tradeable sulfur dioxide emissions allowances.  Utility plants regulated under the Program were allocated sulfur dioxide emission allowances based on their historic fuel consumption. Each allowance authorizes a unit to emit 1 ton of SO₂ during or after a specified year.  Allowances may be bought, sold, or banked.  At the end of each year, each SO₂ source must hold an amount of allowances at least equal to its annual emissions.  Phase II of the Acid Rain Program will cap total allowances at 8.95 million.

One interesting aspect of this program is that anyone may acquire allowances and participate in the trading system.  Want to do your part to improve air quality?  Environmental groups and ordinary citizen can currently buy up SO₂ allowances for about $200 each.

Are such markets vulnerable or price manipulation? A diverse initial allocation of permits can reduce the likelihood of cartelization.

Hazardous air pollutants are controlled separately under the Clean Air Act. EPA is supposed to to list hazardous pollutants, quickly define ambient standards, and enforce them. Standards for hazardous pollutants are supposed to protect human health "with an adequate margin of safety." Caught between environmentalists' demands for zero-emissions standards and industrial necessity, the EPA has actually listed very few hazardous pollutants, and has tried to incorporate risk assessments and benefit-cost analyses into their regulatory policies.