These estimated economic savings from reduced air pollution damages may be sufficiently large to offset from 30 to 100 percent of carbon abatement costs (IPCC, 1996b).

Large as these sums are, this approach underestimates damages over time because it ignores the fact that in baseline projections, in the absence of carbon taxes, rising population and economic growth would increase exposure to air pollution while increased fuel consumption would reduce air quality. Current environmental standards typically limit emissions on a per-BTU basis for stationary sources and on a per-mile basis for vehicles. As the economy grows, trying to maintain environmental quality by making each mile traveled or BTU generated less polluting can become increasingly costly. At some point, fuel taxes that provide incentives to conserve energy and limit the amount of driving become cost-effective adjuncts to such regulations (Eskeland and Devarajan, 1996).

One of the few models to incorporate the savings from reducing pollution damages estimates how much reduction in CO2 emissions could be had for "free" under a "no regrets" policy-a policy under which net abatement costs are exactly offset by non-climate benefits. When firms are assumed to be able to substitute labor and capital for energy, a reduction of nearly 50 percent of baseline emissions can be attained with no loss in economic welfare (Boyd, Krutilla and Viscusi, 1995).

Despite these impressive magnitudes, most models of carbon abatement policies have ignored the potential savings in pollution damages, and the remaining few have incorporated them rather crudely. More recent and comprehensive research on the benefits of air pollution control is now available to be incorporated into climate models (U.S. EPA, 1996). Other benefits from lower energy consumption, such as reduced road congestion and fewer environmental impacts from fuel mining, refining, and transport, could also be taken into account.

4

A. TRADE ISSUES

Concerns about adverse international trade consequences are strong among businessmen who fear that if the United States unilaterally adopted a carbon tax its energy-intensive industries almost certainly would become less competitive in international trade. International shifts in the location of production, so-called 'leakage' effects, would mean that cuts in domestic emissions would be partially offset by increases in emissions abroad as energy-intensive industries expanded overseas (Barrett, 1994; Gaskins and Weyant, 1993). For these reasons, a carbon tax adopted unilaterally would be more costly to the U.S. economy and less effective in reducing global emissions than one adopted by many countries acting together.

Empirical research into the magnitude of leakage effects is somewhat inconclusive. The penalty for acting unilaterally would be greater for small economies than for large economies like that of the United States. Studies of unilateral emissions reduction policies in OECD countries predict leakage rates of between 3.5 and 70 percent (Manne, 1993; Oliveira-Martins et al., 1992; Pezzey, 1992; Rutherford, 1992). Studies in the United States suggest that cost differences created by international differences in environmental policies have minor impacts on trade and investment flows (Repetto, 1995).

12

There would also be international trade gains, however. Given the size of the U.S. market, a tax-induced fall in U.S.

energy consumption would help restrain world fuel prices, improving our terms of trade as a net oil-importing economy. Baseline projections suggest that in the absence of higher energy taxes, fossil fuel consumption in the United States and the rest of the world will continue to grow substantially in coming decades. One implication, given the already declining oil production in this country, is that OPEC will supply an increasing share of world petroleum markets, and the United States will become increasingly dependent on imports from OPEC producers. OPEC's world market share might reach 72 percent by 2015, more than in 1974 when its supply restrictions precipitated the first "oil shock" (U.S. EIA, 1996). A carbon tax would mitigate these energy security and trade risks.

If the United States adopted a carbon tax unilaterally, the risks of climate change faced by other countries would fall, reducing their incentive to act (Barrett, 1994). However, increasing international coordination within the Framework Convention on Climate Change means that unilateral action is unlikely. European countries have stated their willingness to adopt a carbon tax if major trading partners—the United States in particular-do the same (Commission of the European Communities, 1992). Sweden and Norway already have signaled willingness by adopting small taxes. Developing countries are clearly unwilling to commit themselves to action unless industrial countries that have emitted most of the carbon dioxide to date take the lead. Nonetheless, substantial economic

1 The leakage rate is measured as the increase in emissions outside the tax region over the emissions reduction within the region.

reforms and restructuring of energy markets have taken place in developing and transitional economies, for purely economic reasons. Energy subsidies have been reduced, subsidies to energy-intensive industries have fallen, and institutional changes have opened the energy sector to improved technology and greater efficiency. These significant changes have already reduced CO2 emissions in these countries well below their projected levels (Reid and Goldemberg, 1997). Though undertaken for strictly economic reasons, these policy changes in other countries have greatly reduced the risks of unilateral U.S. action.

B. DISTRIBUTIONAL EFFECTS

Even if the macroeconomic economic impacts of a carbon tax were small or positive, some household, regional and sectoral groups might be adversely affected, if they are highly dependent on fossil fuels and have limited ability to adjust to price changes. These distributional considerations are important policy issues: a carbon tax program may be considered undesirable if costs are felt disproportionately by low-income groups or specific industry sectors, even if predicted overall economic burdens are small.

1. DISTRIBUTIONAL EFFECTS ON

HOUSEHOLDS

It is commonly reckoned that a carbon tax would be regressive, hitting lowerincome households proportionately harder than those better off. To illustrate, a $100 per ton carbon tax instituted in 1990 would have created a

burden amounting to 10.1 percent of income for the lowest decile of households arrayed by household income but only 1.5 percent for the top decile (Poterba, 1991). However, when households are arrayed by expenditure instead of income the carbon tax is much less regressive, accounting for 3.7 percent to 2.3 percent of total expenditures for the bottom and top deciles respectively. Household expenditure is a better indicator of longterm ability to pay--and hence a better base against which to judge distributional effects--because it is more stable over time than household income. Short-term fluctuations in income are typically cushioned by borrowing or adding to savings. At any time, there will be many households whose incomes are temporarily and abnormally inflated or depressed.

At best, such calculations provide imperfect guides: they reflect only the direct effect of higher energy prices on expenditures, omitting the fact that goods whose production relies on fuel-everything from cars to foodstuffs-will exhibit price rises too. A Canadian study that accounted for direct and indirect effects by assessing the carbon-intensity of the complete set of purchases made by different income or expenditure groups found that a $100 carbon tax would only be mildly regressive even for households grouped by income. The burden was 2.9 percent for the bottom income quintile and 1.8 percent for the top quintile (Hamilton and Cameron, 1994).

However, since the long-run distributional impact will also depend on

changes in employment, wages. and asset prices, only overall economic models can make comprehensive estimates. An analysis using a general equilibrium model traced the welfare impacts of a carbon tax on 16,128 separate household groups categorized by family size, age and sex of household head, race, region, type of residence, and wealth. The model predicted that, on the whole, the distributional effects of a carbon tax would be modest, ranging from mildly progressive to mildly regressive, depending on underlying assumptions (Jorgenson et al., 1992).

Of course, the overall distributional effect of a carbon tax would depend on the use of the revenues. Expanding the earned-income tax credit would benefit low-income households; reducing capital gains taxes would benefit the rich. There are trade-offs between equity and efficiency concerns. Efficiency gains seem greatest if taxes on investment income are reduced; equity objectives are better served if taxes on labor earnings are reduced. A combination of the two-using the revenues to cut both capital and labor taxes--would balance these objectives.

Moreover, the distributional impacts of a carbon tax should be judged against the impacts of a business-as-usual policy. A number of studies predict that global warming would lead to higher food prices. Since food is as much a necessity as energy is, a business-asusual policy might also be regressive (Scheraga et al., 1993). Similarly, since poorer households typically inhabit areas of lower environmental quality

and are generally in poorer health, they benefit disproportionately from an improvement in air quality (Yin, 1993). However, they may value these relatively large benefits absolutely less than the smaller benefits felt by better off households (Harrison and Rubinfeld, 1978). All in all, the effects of a carbon tax on the distribution of income and expenditures among households would probably be small; and effects on vulnerable groups could rather readily be offset by cost-of-living adjustments in government transfer programs and by other policy

instruments.

2. DISTRIBUTIONAL EFFECTS BY SECTOR AND REGION

The energy industry generally, and the coal mining industry particularly, will experience the greatest impacts from a carbon tax, under which the tax rate on coal, a carbon-intensive fuel, will be relatively high. Compared to the business-as-usual scenario, coal output may decrease by 25 percent and prices may rise by 35 percent by 2020 under a tax that stabilizes emissions at 1990 levels (Jorgenson et al., 1992). Electric utilities, the next most affected sector, face output losses and price increases of just over 5 percent compared with the projected baseline, mainly because they use coal. But these reductions do not imply absolute declines because substantial growth in output is projected for both sectors in the baseline. Some railroads would suffer adverse impacts since coal makes up a substantial fraction of their freight traffic. Of course, other industries, such as the

natural gas and renewable energy industries, would probably do better if a carbon tax were enacted.

Regional impacts of a carbon tax inevitably correspond to sectoral impacts. Coal mining states, such as West Virginia and Wyoming, would be disproportionately affected. But, according to one study, the Pacific Northwest would fare relatively well because of the availability of cheap hydropower-assuming such power continues to be heavily subsidized (DeWitt, Dowlatabadi, and Kopp, 1991). Global warming, like a carbon tax, would also affect some regions and industries disproportionately. Along with coastal areas, regions dependent on agriculture, water sup

ply, and forestry would suffer most (OECD, 1996).

To summarize, studies predict that a carbon tax will be only mildly regressive for households, and those effects could be offset by revenue recycling and cost-of-living adjustments in government transfer programs. The distribution impacts of a carbon tax might well be no worse than those of global warming itself. However, coal mining and coal mining regions would be significantly affected, since coal output would fall relative to a business-asusual scenario and, assuming a sufficiently high tax, would fall absolutely as well.