Dale Jorgenson

Samuel W. Morris University Professor

GROWTH: Volume Two -- Preface

This is the second of two volumes containing my empirical studies of economic growth and presents an econometric model of the United States that captures the dynamic mechanisms underlying long-run trends. The first volume, Econometric General Equilibrium Modeling, introduces the econometric approach to modeling economic growth, using econometric representations of technology and preferences as basic building blocks. Earlier approaches to empirical modeling had "calibrated" the behavioral responses of producers and consumers to a single data point.

Calibration economizes radically on the use of data, but requires highly restrictive assumptions about technology and preferences, such as fixed input-output coefficients. This assumption is contradicted by the massive evidence of energy conservation in response to changes in world energy prices beginning in 1973. As a consequence of these changes and new environmental policies, a wealth of historical experience has accumulated on the price responsiveness of producers and consumers. This experience provides the empirical basis for estimating the impact of alternative economic policies.

This volume includes econometric studies of the impacts of energy, environmental, trade, and tax policies. The concept of an intertemporal price system provides the unifying framework. Such a price system balances demands and supplies for products and factors of production at each point of time. A forward-looking feature of the price system is that asset prices are linked to the present values of future capital services. This is combined with backward linkages among current capital services, the stock of capital, and past investments in modeling the long-run dynamics of economic growth.

The natural starting point for modeling economic growth is the neo-classical theory of economic growth originated by Robert M. Solow (1956). In modeling the interrelationships between economic policies and economic growth we employ the form of this theory developed by David Cass (1965) and Tjalling C. Koopmans (1967). In the neo-classical model wage rates grow at the same rate as productivity in the long run, while rates of return depend on productivity growth and parameters describing saving behavior. However, these long-run characteristics of economic growth are independent of economic policies.

The neo-classical theory of economic growth also provides a basis for interpreting intermediate-run growth trends. In this context the intermediate run refers to the time needed for the capital-output ratio to converge to its long-run stationary value. Since this often requires several decades, intermediate-run trends are critical for policy evaluation. These have the same determinants as long-run trends, but also depend on economic policies through their effects on capital accumulation and rates of productivity growth over shorter periods.

In Chapter 1 Peter J. Wilcoxen and I present an econometric model for analyzing the impacts of energy and environmental policies on the U.S. economy. We first introduce a distinction among industries and commodities. This makes it possible to model differences in the responses of producers to changes in energy prices and the imposition of pollution controls. We also distinguish among households by level of wealth and demographic characteristics. This facilitates modeling differences in responses of consumers to price changes and controls on pollution.

It is important to recognize at the outset that the dominant tradition in general equilibrium modeling does not employ econometric methods. This tradition originated with the seminal work of Wassily W. Leontief (1951, 1953) on input-output analysis. One important advantage of the "fixed-coefficients" assumption underlying input-output analysis is that the resulting general equilibrium model can be solved as a system of linear equations. In addition, the unknown parameters describing technology and preferences can be "calibrated" to match the data from a single inter-industry transactions table.

The obvious disadvantage of the assumption of fixed input-output coefficients is that energy and environmental policies induce changes in these coefficients. In fact, the objective of pollution control regulations is to induce producers and consumers to substitute less polluting inputs for more polluting ones. In addition, the fixed-coefficients assumption is directly contradicted by massive evidence of price-induced energy conservation in response to higher world energy prices beginning in 1973. Reductions in energy-output ratios induced by the successive energy crises of the 1970s and 1980s have averaged 15-20 percent.

In Chapter 1 Wilcoxen and I have analyzed a "natural experiment" provided by variations in energy prices in the 1970s and 1980s. Over the period 1972-1987 U.S. emissions of carbon dioxide, an important by-product of the combustion of fossil fuels, were stabilized by price-induced energy conservation. We attempt to separate the long-run impact of higher energy prices from the intermediate-run impact of sharp increases in energy prices associated the energy crisis periods of 1973-1975 and 1978-1980. We find that "price shocks" account for almost two-thirds of the slowdown in economic growth during the period 1974-1985. Long-run increases in energy prices account for the remaining one-third.

The first successful implementation of an applied general equilibrium model without the fixed-coefficients assumption of input-output analysis is due to Leif Johansen (1960). Johansen retained the fixed-coefficients assumption in modeling demand for intermediate goods, including energy, but employed linear logarithmic or Cobb-Douglas production functions in modeling substitution between capital and labor services and technical change. He replaced the fixed-coefficients assumption for consumers by a system of demand functions originated by Ragnar Frisch (1959). These representations of technology and preferences make only minimal use of econometrics.

The essential features of Johansen's approach to applied general equilibrium modeling have been preserved in the models surveyed by Lars Bergman (1985) and John Shoven and John Whalley (1992). The unknown parameters describing technology and preferences in these models are determined by "calibration" to a Social Accounting Matrix (SAM) for a single point of time, supplemented by a small number of parameters estimated econometrically. Almost all general equilibrium models retain the fixed-coefficients assumption of Leontief and Johansen for energy and other intermediate goods, even in the face of widespread contradictory evidence.

Applied general equilibrium models calibrated to data for a single point of time have usually retained the assumption of constant returns to scale in production. As a consequence of this assumption, commodity prices can be expressed as a function of factor prices, using the nonsubstitution theorem of Paul Samuelson (1951). This greatly facilitates the solution of general equilibrium models by permitting a reduction in the dimensionality of the space of prices determined by the model. This feature of nonlinear general equilibrium models has been exploited in applications of "fixed point" methods for solution of these models pioneered by Herbert Scarf (1973, 1984).

To overcome the limitations of the Johansen approach, it is essential to employ econometric methods, like those I present in Chapter 2. In order to implement econometric models of producer behavior I generate complete systems of demand functions, giving inputs of capital, labor, energy, and materials inputs (KLEM) in each industrial sector as functions of input prices and the level of output. Ernst R. Berndt and I (1973) originated this approach to modeling producer behavior and Edward A. Hudson and I (1974) employed the results in modeling energy policy and U.S. economic growth. Applications of this approach are presented in the accompanying volume, Econometric General Equilibrium Modeling.

Chapter 2 presents a more detailed econometric model of producer behavior that I have constructed with Barbara M. Fraumeni (1981). This includes separate econometric models for each of thirty-five industries that make up the U.S. economy. For each industry the model consists of a complete system of input demand functions, together with an equation that determines the rate of productivity growth. The rate of productivity growth and the input-output coefficients for capital, labor, energy, and materials inputs are functions of the prices of all four inputs as well as time trends. Estimation of these models requires data from a time series of interindustry transactions tables in current and constant prices. More details are provided our paper in the volume, Econometrics and Producer Behavior.

Similarly, econometric models of consumer behavior can be used to overcome the limitations of the Frisch model employed by Johansen. Hudson and I employed a complete system of demand functions, giving quantities demanded as functions of prices and total expenditure. However, this model was based on the notion of a representative consumer. In Chapter 2 I describe an alternative approach based on Lawrence Lau's (1982) theory of exact aggregation. One of the most remarkable features of models based on exact aggregation is that systems of demand functions for individuals can be recovered uniquely from the system of aggregate demand functions. This makes it possible to incorporate the implications of the theory of the individual consumer into a model of aggregate consumer behavior.

Chapter 2 presents an econometric model of aggregate consumer behavior based on the theory of exact aggregation that I have constructed with Lau and Thomas M. Stoker (1982). This approach requires time-series data on prices and aggregate quantities consumed, as well as cross-section data on individual quantities consumed, individual total expenditures, and attributes of individual households, such as demographic characteristics. Systems of aggregate demand functions depend on statistics of the joint distribution of individual total expenditures and attributes of individuals associated with differences in preferences. Additional details are given in our paper in the volume, Aggregate Consumer Behavior.

The starting point for construction of the econometric models presented in Chapter 2 is a system of national accounts for the U.S. presented in my 1980 paper and implemented in my paper with Fraumeni (1980). This accounting system integrates production and income and expenditure accounts for the U.S. economy with a wealth account. The production account includes a complete interindustry transactions table for the thirty-five industrial sectors of the model in current and constant prices and is described in greater detail in Chapter 1 in the volume, International Comparisons of Economic Growth. The income and expenditure and wealth accounts are also given in current and constant prices. These update and extend the accounts presented in my paper with Laurits Christensen, Chapter 5 in the volume, Postwar U.S. Economic Growth.

In Chapter 3 Wilcoxen and I present an intertemporal general equilibrium model for analyzing energy and environmental policies. This model incorporates the econometric representations of technology and preferences given in Chapter 2 as basic building blocks. The production model includes systems of demand functions for capital, labor, energy, and materials inputs and a model of endogenous productivity growth for each of thirty-five sectors of the U.S. economy. The model of consumer behavior is based on exact aggregation and includes a systems of demand functions for five commodity groups - energy, food, nondurable goods, capital services, and other services for each of 672 household types.

In Chapter 3 Wilcoxen and I analyze the impact of environmental regulations on U.S. economic growth. We have utilized detailed data on costs of compliance imposed on individual industries by these regulations. We first simulate future U.S. economic growth with the existing regulations in place in order to provide a base case for comparison with growth under alternative environmental policies. These policies correspond to different costs of pollution control for the industries and generate different time paths for U.S. economic growth. Removing environmental regulations produces an alternative growth path and we refer to this as the alternative case.

An intertemporal submodel incorporates backward-looking and forward-looking equations that determine the time paths of capital stock, consumption, and asset prices. Given the values of these variables in any time period, an intratemporal submodel determines prices that balance demands and supplies for the thirty-five commodity groups included in the model, capital and labor services, and non-competing imports. The two models must be solved simultaneously to obtain a complete intertemporal equilibrium. Additional details are given in Chapter 1. Wilcoxen (1992) surveys alternative computational approaches for solving intertemporal general equilibrium models.

We decompose the overall effect of environmental regulations into components associated with pollution abatement in industry and controls on motor vehicle emissions. We measure the impact of the regulations by eliminating each type of control separately and then eliminating both. We compare growth in the base case with growth in these alternative cases. The growth path with pollution controls differs from the base case at the initial equlibrium, at steady state growth, and on the transition path that traces out the U.S. economy's adjustment to the alternative environmental policy.

Pollution controls have led to a reduction of 2.6 percent in the level of the national product, resulting from an even greater decline in capital accumulation. For many industries the most important impact of environmental regulation is through mandatory investment in costly pollution abatement equipment. The transition to long-run equilibrium of the capital-output ratio requires more than two decades. This illustrates the critical importance of the dynamics of adjustment to the long-run steady state. The industries most affected by these regulations are the motor vehicles and coal mining industries. Primary metals and petroleum refining have followed closely behind.

In Chapter 5 Wilcoxen and I analyze the impact of the Clean Air Act Amendments of 1990. For this purpose we project the growth of the U.S. economy with and without this legislation. We begin with estimates of the cost of compliance with the legislation in the year 2005 prepared by the Environmental Protection Agency (1991), since the provisions of this legislation will be phased in gradually over a fifteen year period. The sectors affected by the new pollution controls are electric utilities and primary metals. There is a short-run surge of investment to take advantage of lower prices of investment goods before the full impact of the legislation works its way through the economy, but the long-run impact is to reduce capital accumulation and economic growth.

The Clean Air Act Amendments of 1990 included a provision, Section 812, that required the Environmental Protection Agency to conduct periodic studies of the impact of the Clean Air Act. The Agency submitted the first of these studies to the Congress in 1997. This study compared benefits and costs of the Clean Air Act during the period 1970 to 1990, prior to the Clean Air Act Amendments. The methodology for analyzing the impact of the Clean Air Act on the growth and structure of the U.S. economy is similar to that presented in Chapters 3 and 5. For this purpose new data on costs of pollution controls have been constructed and inserted into the model that Wilcoxen and I have present in Chapter 3.

The 1997 study of the Clean Air Act by the Environmental Protection Agency combines costs of pollution controls with detailed estimates of environmental benefits. These estimates are based on emissions of pollutants covered by the Clean Air Act, effects of these emissions on air quality, and impacts of air quality on human health, natural ecosystems, and agriculture. Each of these impacts is evaluated economically to obtain the benefits of reductions in emissions and increases in air quality resulting from environmental regulations.

Emissions of greenhouse gases, such as carbon dioxide, gradually increase atmospheric concentrations. These gases trap heat reflected from the earth in the form of light and warm the atmosphere. The possibility that carbon dioxide emissions from fossil fuel combustion might lead to warming of the global climate through the "greenhouse effect" has emerged as a leading international concern. On December 10, 1997, 160 nations signed the Kyoto Agreement, calling for reductions in emissions of carbon dioxide and other greenhouse gases.

A policy often proposed for reducing emissions of carbon dioxide is a tax on fossil fuels in proportion to carbon content. This is known as a "carbon" tax and could to reductions in energy use, substitution of other forms of energy for fossil fuels, and substitution among fossil fuels to reduce carbon dioxide emissions. For example, substitution of natural gas for coal in electricity generation, holding the output of electricity constant, would reduce carbon dioxide emissions. In Chapter 4 Wilcoxen and I employ the model presented in Chapter 3 to analyze the economic impact of a carbon tax.

We assume that carbon dioxide emissions are proportional to fossil fuel use. We calculate the carbon content of each fuel by combining data from the Department of Energy on the heat content of the fuels with data from the Environmental Protection Agency on emissions of carbon dioxide per unit of heat produced by combustion. We consider three alternative goals:

1. Stabilizing carbon dioxide emissions at 1990 levels.

2. Decreasing carbon emissions gradually over the period 1990-2005 until they are twenty percent below 1990 levels.

3. Doing nothing until 2000, then gradually increasing the carbon tax over the period 2000-2010 to stabilize emissions at 2000 levels.

To simulate the impact of a carbon tax we constrain total carbon dioxide emissions and calculate the required level of the carbon tax. We hold the real value of government spending constant and allow the average tax on labor income to adjust in order to leave the government deficit unchanged. We hold the marginal tax rate on labor income constant, so that adjustments in the average tax rate reflect changes in the implicit zero-tax threshold. The principal impact of a carbon tax in all three simulations is to increase purchasers' prices of coal and crude oil. The rising price of fossil fuels results in a decline in primary energy use, as well as substitutions among fuels to reduce emissions.

The impact of a carbon tax is most severe for the coal mining industry. Even with the least stringent restrictions on emissions, coal production would fall by sixteen percent, relative to the base case. The most extreme restrictions could lead to a decline of more that fifty percent in coal mining and a fifteen percent reduction in electricity production. While the economy-wide effects of a carbon tax would be limited, the costs rise very rapidly with emission reductions. Since the benefits of a carbon tax depend on policies adopted by other countries, a cost-benefit comparison can be carried out only at the world level, as in the pioneering studies of Alan S. Manne and Richard Richels (1992) and William Nordhaus (1994), summarized in Chapter 1.

Daniel T. Slesnick, Wilcoxen, and I have evaluated the impact of a carbon tax on economic welfare in Chapter 9 of the volume, Measuring Social Welfare.

To estimate the impact of a carbon tax on individual welfare we consider a population of infinite-lived consumers or "dynasties." Households are classified by demographic characteristics, but each type is linked to similar types in the future through intergenerational altruism. We define social welfare on the distribution of individual welfare over the population of households.

The welfare cost of a carbon tax is dominated by a loss in efficiency; the equity impact of the tax can be positive or negative, depending on the degree of aversion to inequality.

In Chapter 7 Wilcoxen and I compare the effects of alternative tax instruments for reducing carbon dioxide emissions. We consider a Btu tax with tax rates proportional to the heat content of each fuel and an ad valorem tax with rates proportional to the value of the fuel. We constrain emissions at 1990 levels and calculate the rates required for each of the taxes. As in Chapter 4, tax revenues are used to reduce the average tax on labor income, while holding marginal tax rates constant. As expected, the carbon tax achieves reductions with minimum impact on the U.S. economy and has the greatest impact on coal production. The energy tax is similar in its impact to a carbon tax.

The ad valorem tax produces the most severe distortions in the economy, but has the least impact on coal mining.

In the tax simulations reported in Chapters 4 and 7 the principal macroeconomic mechanism for adjusting to changes in tax policy is the alteration of rates of capital formation. A second mechanism is the pricing of capital assets through forward-looking expectations of future prices and discount rates. Both these mechanisms are captured by the intertemporal general equilibrium that Wilcoxen and I have presented in Chapter 3. The most important impacts of the alternative tax policies are on the energy sector. The overall impact on the U.S. economy reflects the tax distortions introduced into energy markets.

In Chapter 6 Wilcoxen and I summarize our contribution to Energy Modeling Forum 12, a comparison of the costs of limiting emissions of carbon dioxide organized by the Energy Modeling Forum at Stanford University. More than a dozen different models were compared in this study. Three features of our model are important in these comparisons. First, our model is highly disaggregated for both producers and consumers. Second, the model incorporates econometric models that reflect historical changes in energy prices. Third, productivity growth in each sector is an endogenous function of relative prices.

Even a modest carbon tax would raise substantial revenue. We first consider a lump-sum rebate of this revenue to households. This is, however, not the most likely use of the tax revenue, so that we also consider lower marginal and average taxes on labor income. Finally, we consider lower average and marginal taxes on capital income. We find that methods for "recycling" the revenue have significant effects on the overall impact of a carbon tax. With a lump-sum rebate to households the aggregate output of the U.S. economy would decline by 1.70 percent in the long run, relative to the base case. When marginal as well as average tax rates on labor income are reduced, the decline in output is only 0.69 percent or less than half. Finally, a reduction in capital income taxes produces a gain in output of 1.10 percent.

The recycling of revenues from a carbon tax in order to reduce capital income illustrations the notion of a "double dividend." The first dividend is the improvement in environmental quality that results from lower carbon dioxide emission. The second is the stimulus to economic growth from greater investment and more rapid capital accumulation as a consequence of lower capital income taxes. This double dividend is the result of trading an existing tax distortion, namely, the distortion that results from capital income taxes, against a new distortion - the distortion of energy markets resulting from a carbon tax. Lawrence Goulder (1995) has surveyed the literature on the double dividend.

In Chapter 8 Mun S. Ho and I present an intertemporal general equilibrium model for analyzing the impact of changes in trade policy on U.S. economic growth. This model preserves all of the features of the model summarized in Chapter 1, including disaggregation into 35 industries and the corresponding commodities. The domestic supply of each commodity is the sum of the output of domestic industries and "competitive" imports, defined as imported commodities that are also produced in the U.S. At this level of disaggregation imports and domestically produced commodities are imperfect substitutes. We take the price of imports in foreign currency to be exogenous; the price in domestic currency depends on the terms of trade.

Domestic prices of imports include tariffs levied by the U.S. government. We model non-tariff barriers, such as quotas and "voluntary" export restrictions on foreign supplies in terms of tariff equivalent increases in prices of U.S. imports. Prices of imports reflect both tariff and non-tariff barriers. For each commodity group we model the share of imports in domestic supply econometrically. This share is a function of the ratio of the price of imports to the price of the domestically produced commodity, as well as a time trend. We model noncompetitive imports as inputs into the importing industries. Prices of these imports also reflect barriers to trade.

We express the demand for U.S. exports as a function of rest of the world output and the price of U.S. exports. The price of U.S. exports in foreign currency depends on the U.S. domestic price, the terms of trade, and foreign tariffs. To complete the current account balance we include exogenous components, such as U.S. foreign aid. The foreign balance is exogenous and the terms of trade endogenous in our model. We provide a base case projection of U.S. economic growth with existing trade policies in place. We then simulate the impact of alternative trade policies, beginning with a multilateral reduction in tariffs with no changes in quantitative restrictions on trade.

The impact of a multilateral elimination of tariffs is to raise U.S. consumption of goods and services by 0.16 percent in the first year of the policy change. However, the impact on consumption rises over time to 0.82 percent in the long-run. The mechanism underlying the dynamics of the adjustment to tariff reductions is that a decline in the price of imports of capital goods stimulates investment and results in more rapid growth of the U.S. economy. Since the current account balance is exogenous, the terms of trade must fall in order to accommodate the higher level of imports, implying an increase in U.S. international competitiveness.

Commodities with the highest U.S. tariff levels - textiles, apparel, rubber, leather, and glass - have the largest gains in imports; chemicals, electrical machinery (including computers), and instruments face the highest tariff barriers in the rest of the world and benefit most from the increase in exports. The output and employment effects of tariff reductions largely parallel the shifts in imports and exports. Important penetration is so high in food, furniture, and leather industries that output falls, relative to the base case. Capital and labor shift to the U.S. industries that are the most competitive internationally.

We also consider elimination of quantitative restrictions on U.S. imports, as well as multilateral tariff reductions. The gains to the U.S. economy from removing these restrictions is considerably greater than the effects of tariff cuts alone. We compare the results of our simulations with those of Alan Deardorff and Robert Stern (1986) and Whalley (1985), using static multilateral general equilibrium models. Our estimates of the impact of multilateral tariff reductions for the first year are comparable with the results of these static simulations. The effects of dynamic adjustments of businesses and households to changes in trade policy, excluded from static models, greatly predominate in the long run.

In Chapter 9 Ho and I consider the impact of environmental regulation on U.S. trade. Specifically, we consider the effects of the environmental regulations described in Chapter 3 on the competitiveness of U.S. industries and patterns of imports and exports. In Chapter 3 Wilcoxen and I have found that the imposition of environmental regulations has a substantial cost in foregone domestic output. We have decomposed the impact of environmental regulation between pollution controls on industry and controls on motor vehicle emissions.

The domestic output of coal mining and motor vehicles, the industries most affected by these controls, has been adversely affected by the imposition of these controls.

We construct a base case for analysis of the impact of environmental policy on U.S. trade by projecting the growth of output, imports, and exports with environmental regulations in place. We then consider the impact of alternative policies that eliminate these regulations. U.S. aggregate exports would rise by 0.27 percent in the long run, while U.S. imports would fall by 0.15 percent. Elimination of environmental regulations would produce a modest fall in the U.S. terms of trade, resulting in an increase in the competitiveness of U.S. industries.

Although the aggregate impact of pollution controls is relatively small, these controls have a substantial impact on the commodity composition of trade.

Elimination of controls on motor vehicle emissions would increase imports and exports of vehicles by about ten percent. Exports of chemicals, petroleum products, and primary metals are adversely affected by pollution controls on industry, since controls increase the prices of U.S. exports of these commodity groups. These qualitative results are relatively insensitive to the magnitudes of our estimates of the elasticities of demand for imports and supply of exports.

In Chapter 10 Ho and I consider the impact of restrictions on carbon dioxide emissions on U.S. trade. We first consider the imposition of carbon taxes that would achieve the three alternative goals Wilcoxen and I have examined in Chapter 4. These taxes are levied on imports of crude oil and natural gas, as well as imports of refined petroleum products. Although coal exports are relatively modest, we assume that this coal is also subject to taxation. The main effect of stabilization of carbon dioxide emissions at 1990 levels would be to reduce oil imports by 3.6 percent, leading to a slight appreciation of the terms of trade and a fall in competitiveness of U.S. industries.

Coal exports would fall by 37 percent, while exports of refined petroleum products would decline by 10 percent. Other substantial losses of exports would be in the energy-intensive manufacturing industries - primary metals, rubber, and stone, clay, and glass. The slight deterioration in the terms of trade leads to increased imports of products other than crude petroleum and natural gas.

All of these effects would be greatly magnified by adoption of a more stringent goal for reduction of carbon dioxide emissions, while a less stringent goal would diminish the policy impacts. Finally, we consider the substitution of an energy tax for a carbon tax; this provides a less efficient policy instrument for achieving the goal of reduced emissions.

Carbon taxes affect U.S. trade flows through three different channels. First, these taxes stimulate energy conservation and lead to reduced imports of fossil fuels. The composition of net exports is shifted away from energy-intensive commodities. Finally, capital accumulation and productivity growth are adversely affected by carbon taxes, reducing economic growth and changing the pattern of trade. Our model captures the difference between short-run and long-run impacts of carbon taxes through the dynamic mechanisms that affect economic growth. An important limitation of these results is that we do not consider international coordination of policies to reduce carbon dioxide emissions.

In Chapter 11 Wilcoxen and I consider the impact of fundamental tax reform on energy markets. For this purpose we consider the effects of substituting a tax on consumption for corporate and individual income taxes at both federal and state and local levels. For this purpose we introduce models of the demand for different types of capital services for each of the thirty-five industrial sectors of the U.S. economy and the household sector. These models depend on tax policies through detailed measures of the cost of capital for each type of capital services presented in my book with Kun-Young Yun (1991), Tax Reform and the Cost of Capital.

Measures of the cost of capital incorporate the characteristic features of U.S. tax law and summarize information about the future consequences of investment decisions required for current decisions about the allocation of capital. The concept of the cost of capital makes it possible to represent the economically relevant features of tax statutes in a very succinct form. The model we present in Chapter 11 is the ninth version of the model we originally presented in Chapter 3. Successive version of the model have incorporated more efficient solution algorithms and additional features, such as the disaggregation of capital services in Chapter 11, that enhance the flexibility and usefulness of the model.

We consider substitution of a consumption tax for existing income taxes that would leave the government deficit unchanged. This substitution would have an immediate and powerful impact on the level of economic activity. Individuals would sharply curtail consumption of both goods and leisure and shift the composition of output toward investment. Real consumption would initially decline by five percent, but would grow rapidly and overtake the level of consumption under the current tax system within five years. Investment would jump sharply, generating much more rapid capital accumulation and a higher level of the national product.

Changes in relative prices would stimulate energy conservation, but this would be outweighed by the impact of increased economic growth, leading to an increase in energy consumption. This consumption would be somewhat more carbon intensive and emissions of carbon dioxide would increase. A recent report by the Alliance to Save Energy (1998), Price It Right, shows how fundamental tax reform could be combined with energy taxes that would reflect the costs of environmental damages. This study uses the same modeling framework as the one Wilcoxen and I present in Chapter 11.

I conclude that an intertemporal price system provides the appropriate conceptual framework for modeling the impacts of energy, environmental, trade, and tax policies. The econometric approach presented in this volume makes it possible to preserve the features of aggregate growth models, while disaggregating the policy impacts. The studies presented in the volume distinguish among thirty-five sectors of the U.S. economy and also identify thirty-five commodity groups. In modeling consumer behavior we distinguish among 672 different household types, broken down by demographic characteristics. Aggregate demand functions are obtained by summing over individual demand functions.

The econometric method for modeling technology and preferences can be contrasted with the calibration approach employed in earlier general equilibrium models. The overwhelming advantage of the econometric method is that responses of production and consumption patterns to changes in energy prices, environmental controls, trade restrictions, and tax policies are derived from historical experience. The implementation of the econometric approach requires a system of national accounts that successfully integrates capital accounts with income and production accounts. The new accounting system incorporates an accumulation equation relating capital to past investments with an asset-pricing equation linking the price of assets to future prices and rates of return.

An important feature of the modeling framework for consumer behavior presented in this volume is that systems of individual demand functions can be recovered from the system of aggregate demand functions. The representation of consumer preferences underlying these individual demand functions can be used to generate measures of individual welfare that are useful in evaluating the distributional consequences of economic policies. These measures are cardinal and interpersonally comparable and can be combined into a measure of social welfare that captures both efficiency and equity impacts of policy changes.

Intertemporal general equilibrium modeling provides a very worthwhile addition to methodologies for evaluating the impact of energy, environmental, trade, and tax policies. The neo-classical theory of economic growth is essential for understanding the dynamic mechanisms that underly long-run and intermediate-run growth trends. The econometric implementation of this theory is critical for exploiting the wealth of historical experience that has accumulated over the past several decades. This experience, interpreted within an intertemporal framework, provides valuable guidance in the formulation of future economic policies.

I would like to thank June Wynn of the Department of Economics at Harvard University for her excellent work in assembling the manuscripts for this volume in machine-readable form. Renate d'Arcangelo of the Editorial Office of the Division of Engineering and Applied Sciences at Harvard edited the manuscripts, proofread the machine-readable versions and prepared them for typesetting. Warren Hrung, then a senior at Harvard College, checked the references and proofread successive versions of the typescript. William Richardson and his associates provided the index. Gary Bisbee of Chiron Incorporated typeset the manuscript ands provide camera-ready copy for publication. The staff of The MIT Press, especially Terry Vaughn, Victoria Richardson, Andrea Werblin, and Michael Sims, has been very helpful at every stage of the project. Financial support was provided by the Program on Technology and Economic Policy of the Kennedy School of Government at Harvard. As always, the author retains sole responsibility for any remaining deficiencies in the volume.