16

"bottom-up" approaches. The top-down method involves creating a model of the entire economy, with equations describing the paths over time of key variables such as the GDP, population, energy prices, and the rate of technological progress. The model is then run under a base case scenario in which no action is taken to control greenhouse-gas emissions. The result of the base case is compared to cases in which alternative policy actions are taken to limit emissions, such as a carbon tax or creation of tradable emissions permits.

All of the commonly used top-down models are constructed in such a way as to include the assumption that reductions in greenhouse-gas emissions can only be purchased at the expense of a reduction in the output of other goods and services. In all the top-down models, the various sectors and agents in the economy are presumed to be operating in a perfectly efficient manner, so that if an additional constraint is placed on their activities (such as being required to reduce emissions of greenhouse gases), the amount of ordinary goods and services that can be produced must fall. This assumption that is central to top-down models is appropriate in some applications," but it has serious drawbacks if the analysis covers decades of time.

The bottom-up method takes a different approach. Instead of assuming that existing patterns of production are optimal, this method recognizes that a variety of economic, institutional, organizational, cultural, and political barriers prevent firms and individuals from taking advantage of best-practice techniques. In particular, the bottom-up studies have focused on how much greater energy efficiency could be achieved if the barriers to cost-effective investments in energy efficiency were eliminated. Unlike the top-down studies, the bottom-up studies admit the possibility that some energy savings (and hence greenhouse-gas reductions) could be achieved without loss to the larger economy.

9. Many of these models had their origins in attempts to estimate the short-term consequences of oil price stocks such as those that occurred in the 1970s. The problem of modeling long-term effects of a gradually-phased-in climate protection policy is quite different.

A. "Top-down" Cost Estimates

17

A number of leading top-down model estimates were reviewed by the IPCC (1996c). Many of the results and model runs presented in the literature differ with respect to the time periods they cover, as well as other assumptions. To achieve comparability, the IPCC relied on an exercise carried out by the Energy Modeling Forum at Stanford University in which several of the main models were run under a common set of assumptions. A typical scenario estimates the consequences of a 20% reduction in emissions from 1990 levels, implemented by means of a carbon tax. The eight models for which the comparison was carried out showed GDP losses in 2010 ranging from 0.9% to 1.7%, with an average of 1.2%. The carbon tax required to achieve the 20% reduction from the baseline ranged from $50 per metric ton of carbon to $260 per metric ton, averaging $170 per metric ton.

A similar modeling exercise was conducted by an Interagency Analytical Team (IAT) of the U.S. government during 1997. In this review, three models—the Data Resources, Inc. (DRI) macroeconomic model, the Markal-Macro model, and the Second Generation Model (SGM)—were used to estimate the path of GDP under the scenario of emissions stabilized at 1990 levels in 2010 with a 10-year phase-in." The DRI model showed a decrease in GDP from the baseline through 2012 followed by an increase beginning in 2013;" in 2010 the decrease of GDP from the baseline was about 0.4%. Markal-Macro showed a GDP loss of about 0.6% in 2010, and SGM a loss of about 0.1%. The implicit carbon prices that would bring about the 1990 emissions levels in 2010 were $95 per ton in the DRI model, $81 per ton in the SGM model, and $145 per ton in the Markal-Macro model."

10. Other assumptions made in the IAT analysis were that revenues from auction of the carbon emission permits were recycled into the economy through delict reduction, that the energy/GOP ratio decreased at a rate of 1.25% per year, and that there was no international trading of the carbon emission permits.

11. This later expansion in GDP predicted by the DRI model is a peculiar consequence of the model's original design as a short-run forecasting tool. In this model, the projected deficit reduction from the permit revenues leads to greater investment, which eventually raises the GOP. This type of naïve Keynesian structure is not shared by most other top-down models.

12. One metric ton (tonne) is 1,000 kilograms, or 2,205 pounds. Thus, the IAT estimates of the implicit carbon price are $105/tonne for the DRI model, $89лonne for the SGM model, and $160-tonne for the Markal-Macro model.

18

According to the IAT analysis, a $100 per ton carbon tax translates into a price increase of 26 cents per gallon of refined petroleum product (e.g., gasoline), $1.49 per thousand cubic feet of natural gas, $52.52 per ton of coal, and 2 cents per kilowatt hour of electricity. Estimated GDP losses were smaller under scenarios allowing international trading of the carbon permits (Interagency Analytical Team 1997).

It should be noted that none of these scenarios is directly comparable to the "doubling of atmospheric concentration of CO," case that is the basis for the direct climate change damage estimates reported in Section IV-B. It is not a simple matter to go from changes in rates of emission of greenhouse gases to changes in atmospheric concentrations. The relationship depends, among other things, on the amount of GHGs already in the atmosphere, the ability of the various carbon sinks to absorb some of the annual emissions, and the rate at which carbon dioxide from the atmosphere is mixed into the ocean. The fraction of CO2 emissions that will be reabsorbed by terrestrial and oceanic sinks is one of the areas of greatest scientific uncertainty.

Even if the top-down modeling results are taken at face value, it is clear that the GDP losses projected by these studies are hardly disastrous. A loss of 1% of GDP is not insignificant—it amounts to about $70 billion (1992 dollars) per year at the current GDP level—yet it amounts to less than six months of normal economic growth. That is, a permanent loss of 1% of GDP means only about a six months' delay in achieving any particular aggregate standard of living that would be reached in the ordinary course of economic growth. Even under the conservative assumption of the top-down models that reduced emissions necessarily lead to a GDP loss, normal economic growth swamps the effects of the GHG reduction policy.

Figure 2 (see page 19) illustrates growth paths through 2050 with and without GHG reductions, under the assumption that the GHG control measures would reduce GDP by 1%. The effect of the 1% reduction is barely visible on a chart of this size; the changes in GDP are entirely dominated by the increases due to economic growth on either of the two paths. This illustrates quite dramatically the fact

that it is the growth performance of an economy: more than anything else, that determines the evolution of the standard of living in the long run.

Other considerations point to the conclusion that a proactive policy to reduce greenhouse-gas emissions would not seriously disrupt the economy. Job reallocations caused by a reduction in fossil fuel use would be small relative to the average pace of job turnover. For example, the entire coal mining industry in the United States employed only 106,000 workers in 1995, down from 246,000 in 1980. Thus, this industry has been losing jobs at an average rate of just over 9,000 per year over the period 1980-95 without any GHG control measures in place (data from U.S. Bureau of the Census 1997, Table 654). Yet the U.S. economy creates about one and a half to two million net new jobs per year, and the gross number of jobs created and destroyed through the normal process of economic change is larger (Worsham 1996; U.S. Department of Labor 1996-97). If the rate of job decline in coal were to double it would still be less than 1.5% of the normal annual rate of total net job creation. Without minimizing the hardships of adjustment to displaced coal workers, this sort of incremental change in the sectoral distribution of jobs would not be difficult for the economy to absorb, and it would be sensible to include transitional support for displaced workers (such as retraining expenses) as an integral part of any national greenhouse-gas reduction policy.

It is worth bearing in mind that the aggregate number of jobs in the United States is determined in the short run primarily by the Federal Reserve's monetary policies (and the Fed's response to current and expected changes in the economy), and in the long run by structural factors such as the size and age composition of the population. Neither the short-run nor the long-run determinants of the total number of jobs have much to do with the relative prices of different types of energy. Unexpected price changes always lead to changes in the value of capital. goods, and to reallocations of both capital and labor across sectors. Policy-driven fossil fuel price increases designed to reduce CO2 emissions could be phased in gradually (to give workers and managers time to adjust), and should begin to be factored into current investment decisions (because of the eventual necessity of