Dale Jorgenson

Samuel W. Morris University Professor

Investment: Volume Two -- Preface

This is the second of two volumes containing my empirical studies of investment behavior and is devoted to the cost of capital approach to tax policy. The first volume, entitled Capital Theory and Investment Behavior , focuses on the cost of capital as a determinant of investment expenditures. The unifying framework for the two volumes is a model of capital as a factor of production that I introduced in 1963.

The cost of capital approach has supplied an important intellectual impetus for reforms of capital income taxation in the United States and around the world. The early 1980s continued the shift from income to consumption as a base for taxation that had characterized the postwar period. Earlier, three landmark reports in Sweden , the United Kingdom , and the United States had proposed taking these developments to their logical conclusion by completely replacing income by consumption as a tax base. (1)

Initially, incentives for investment were enhanced by allowing more rapid write offs of investment outlays. An alternative policy was to offset tax liabilities by subsidies for investment. In the United States an investment tax credit was introduced for this purpose in 1962. The ultimate incentive would be to treat investment expenditures symmetrically with outlays on current account, thereby removing investment from the tax base and shifting the base to consumption. (2)

To stimulate saving, individuals were allowed to establish tax favored accounts. In the United States these took the form of pension funds for corporate and noncorporate businesses and Individual Retirement Accounts. Contributions were excluded from income for tax purposes and earnings were not taxed until withdrawn, so that saving through these accounts was removed from the tax base. By allowing unlimited contributions to the accounts the tax base could be shifted from income to consumption.

When the Administration of President Ronald Reagan took office in January 1981, there was widespread concern about the slowdown of U.S. economic growth. Tax reform received overwhelming support from the Congress with the enactment of the Economic Recovery Tax Act of 1981. The 1981 Tax Act combined substantial reductions in tax rates with sizable enhancements in incentives for saving and investment.

Beginning with the introduction of accelerated depreciation in 1954 and the investment tax credit in 1962, U.S. tax policy had incorporated progressively more elaborate tax preferences for specific forms of capital income. The 1981 Tax Act brought these developments to their highest point with adoption of the Accelerated Cost Recovery System and a 10 percent investment tax credit. These tax provisions severed the connection between capital cost recovery and the economic concept of income.

The tax reforms of the early 1980s substantially reduced the burden of taxation on capital income. However, these policies also heightened discrepancies among burdens on different types of capital. This gave rise to concerns in the Congress about distortions in the allocation of capital. In the State of the Union Address in January 1984 President Reagan announced that he had requested a plan for further reform from the Treasury, initiating a lengthy debate that eventuated in the Tax Reform Act of 1986. (3)

The 1986 Tax Act abruptly reversed the direction of U.S. tax policy. The tax base was broadened by wholesale elimination of tax preferences for both individuals and corporations. The investment tax credit was repealed for property placed in service after December 31, 1985. Capital consumption allowances were brought into line with economic depreciation. Revenues generated by base broadening were used to finance sharp reductions in tax rates at corporate and individual levels.

The 1986 Tax Act reflected a new conceptual framework for the analysis of capital income taxation. This framework had its origins in two concepts introduced in the 1960s -- the effective tax rate, pioneered by Arnold C. Harberger (1962, 1966), and the cost of capital, originated in my papers of 1963 and 1965. These papers are reprinted in the companion volume, Capital Theory and Investment Behavior . The cost of capital and the effective tax rate were combined in the concept of the marginal effective tax rate introduced in my 1980 paper with Alan J. Auerbach , reprinted as chapter 8 below.

Marginal effective tax rates must be carefully distinguished from the average effective tax rates introduced by Harberger (1962, 1966). Marginal and average tax rates differ substantially, since changes in tax laws usually apply only to new assets. In the model of capital as a factor of production that I introduced in 1963, new and existing assets are perfect substitutes in production, so that marginal rather than average tax rates are relevant for measuring distortions in the allocation of capital.

The average effective tax rates presented by Harberger (1962) included corporate income taxes and property taxes, but not individual taxes on corporate distributions. Harberger (1966) incorporated individual taxes on dividends as well as taxes on capital gains realized on corporate equity. (4) Martin S. Feldstein and Lawrence H. Summers (1979) presented average effective tax rates for the U.S. corporate sector that included individual as well as corporate tax liabilities.

Widespread applications of the cost of capital and the marginal effective tax rate are due to the fact that these concepts facilitate the representation of economically relevant features of complex tax statutes in a highly succinct form. The cost of capital summarizes the information about the future consequences of investment essential for current decisions. The marginal effective tax rate characterizes the tax consequences of these decisions and greatly enhances the transparency of tax rules.

My testimony before the Committee on Finance of the U.S. Senate on October 22, 1979, and the Committee on Ways and Means of the U.S. House of Representatives on November 14, 1979, provided the initial impetus for using marginal effective tax rates in reforming the taxation of income from capital. This testimony applied marginal effective tax rates based on my work with Auerbach to quantify differences in the tax treatment of different types of capital. Although the cost of capital approach had no effect on the 1981 Tax Act, this approach spread rapidly among tax policy analysts, both inside and outside the government.

An important milestone in diffusion of the cost of capital approach was the Conference on Depreciation, Inflation, and the Taxation of Income from Capital, held at the Urban Institute in Washington , D.C. , on December 1, 1980. The participants in this conference included tax analysts from universities, research institutions, the U.S. Department of the Treasury, and the staff of the U.S. Congress. I presented marginal effective tax rates for all types of assets and all industries in my paper with Martin A. Sullivan, reprinted as chapter 9 below.

The publication of the conference proceedings in 1981 was followed quickly by the first official estimates of marginal effective tax rates by the President's Council of Economic Advisers in early 1982. Subsequently, marginal effective tax rates helped to frame the alternative proposals that led to the Tax Reform Act of 1986. An important objective of this legislation was to "level the playing field" by equalizing marginal effective tax rates on different types of capital income. (5)

The literature on the cost of capital approach developed at an explosive pace during the 1980s, leading to the presentation of the Treasury proposal, requested by President Reagan, in November 1984. (6) This proposal as accompanied by marginal effective tax rates for different types of assets. An important objective was to equalize these rates. A second objective was to insulate the definition of capital income from the impact of inflation. The cost of capital provided the analytical framework for achieving both of these objectives.

Many of the important issues in implementing marginal effective tax rates have been debated for nearly three decades following introduction of the cost of capital in my 1963 paper, reprinted in the companion volume, Capital Theory and Investment Behavior . Statutory tax rates and the definition of taxable income provide only part of the information required. In addition, estimates of economic depreciation are necessary to assess the need for capital cost recovery. Since the income tax base is not insulated from inflation, rates of inflation must also be taken into account.

The first empirical issue in measuring the cost of capital is the description of capital cost recovery. The cost of capital formula introduced in my 1963 paper allowed for differences between tax and economic depreciation. The modeling of provisions for capital cost recovery as the present value of deductions from income was the crucial innovation in my 1967 and 1971 papers with Robert E. Hall, reprinted as chapters 1 and 2 below. This formulation of capital cost recovery has been adopted in almost all subsequent studies.

My 1967 paper with Hall was the first application of the cost of capital approach to the analysis of tax policy. We modeled the introduction of accelerated depreciation in 1954 and new guidelines for asset lifetimes in 1962. We considered the impact of the investment tax credit, also introduced in 1962. Finally, we analyzed the impact of a hypothetical tax reform, treating investment expenditures in the same way as expenditures on current account, thereby shifting the tax base from income to consumption.

My 1971 paper with Hall analyzed all of the tax policies considered in our 1967 paper. In addition to adoption of the investment tax credit in 1962 we analyzed the 1964 repeal of the Long Amendment, which had reduced the base for depreciation by the amount of the credit. We also considered reduction in the corporate income tax rate in 1964 and suspension of the investment tax credit in 1966.

The model for taxation of corporate capital income I developed with Hall was also used in implementing the concept of capital as a factor of production in my 1973 paper on postwar U.S. economic growth with Laurits R. Christensen, reprinted in my 1995 book, Productivity, volume 1: Postwar U.S. Economic Growth . Christensen and I extended this model to noncorporate and household capital incomes. This enabled us to include differences in returns due to taxation in our measure of the growth of capital input.

Christensen and I imbedded estimates of the cost of capital into a complete system of U.S. national accounts. The critical innovation in this accounting system was the construction of internally consistent accounts for income, product, and wealth. We distinguished two approaches to the analysis of economic growth. We used data on inputs and outputs from the production account to allocate the sources of economic growth between investment and productivity. We divided the uses of economic growth between consumption and savings by means of data from the income and expenditure account.

In the Christensen-Jorgenson accounting system saving was linked to the asset side of the wealth account through capital accumulation equations. These equations provided a perpetual inventory of assets of different vintages. Prices for these different vintages were linked to rental prices of capital inputs through a parallel set of capital asset pricing equations. The complete system of vintage accounts contained stocks of assets of each vintage and their prices. Stocks were cumulated to obtain asset quantities, while prices were used to derive the cost of capital for each asset.

My initial modeling of tax provisions for capital cost recovery was based on the plausible assumption that taxpayers choose among alternative provisions to minimize their tax liabilities. This assumption was used, for example, in my papers with Hall and my paper with Christensen. Sullivan and I presented a detailed study of actual practices for capital cost recovery in our 1981 paper, chapter 9 below. This description has been employed in many subsequent studies, including my 1986 paper with Kun-Young Yun , reprinted as chapter 11, and our 1991 book, Tax Reform and The Cost of Capital.

The second empirical issue in measuring the cost of capital is inflation in asset prices. A comparison of alternative treatments of inflation in measures of the cost of capital was included in my 1968 papers with Calvin D. Sabered , reprinted in the companion volume, Capital Theory and Investment Behavior. The assumption of perfect foresight or rational expectations of inflation emerged as the most appropriate formulation and has been used in most subsequent studies.

The third empirical issue in implementing the cost of capital is measuring economic depreciation. This was the focus of my 1974 paper, "The Economic Theory of Replacement and Depreciation," reprinted as chapter 5 below. This paper presented a theory of replacement investment and a dual theory of economic depreciation. The theory of depreciation was originally introduced by Harold S. Hotelling (1925) and subsequently developed by Kenneth J. Arrow (1964) and Hall (1968). My paper also surveyed the empirical literature and concluded that depreciation at a constant geometric rate provides a satisfactory approximation for measuring the cost of capital.

Charles R. Hulten and Frank C. Wykoff (1981a) developed an econometric methodology for measuring economic depreciation. Hall (1971a) had modeled prices of assets as functions of age and asset characteristics. The important innovation by Hulten and Wykoff was to allow for "censoring" asset prices by retirements. They demonstrated that a geometric decline in efficiency of assets with age provides a satisfactory approximation to the actual decline. (7)

Hulten , James W. Robertson, and Wykoff (1989, p. 255) carefully documented the stability of efficiency decline in the face of changes in tax policy and sharp increases in energy prices during the 1970s. They concluded that "the use of a single number of characterize the process of economic depreciation (of a given type of capital asset) seems justified in light of the results of this chapter." Constant depreciation rates greatly simplify the model of capital as a factor of production in chapter 5. Replacement is proportional to the stock of capital, while depreciation is proportional to the asset price.

Constant geometric depreciation rates based on those of Hulten and Wykoff (1981a) were incorporated into estimates of marginal effective tax rates in my papers with Sullivan and Yun , reprinted as chapters 9 and 11, my 1991 book with Yun , and the international comparisons I give in my 1993 book, edited with Ralph Landau. The Hulten-Wykoff depreciation rates were also employed in measuring capital input my 1987 book with Frank Gollop and Barbara Fraumeni , Productivity and U.S. Economic Growth . The results are summarized in my 1988 paper on growth of the U.S. economy, reprinted in my 1995 volume, Productivity, volume 1: Postwar U.S. Economic Growth .

The marginal effective tax rates introduced in my 1980 paper with Auerbach included corporate taxes. Marginal effective tax rates for corporate sources income including both corporate and personal taxes provided the basis for detailed comparisons of taxes in Germany , Sweden , the U.K. , and the U.S. for 1980 by Mervyn A. King and Don Fullerton (1984). Fullerton (1987) and Fullerton , Robert Gillette, and James Mackie (1987) provided comparisons among tax rates for corporate, noncorporate , and housing sectors of the U.S.

The tax base for corporate income depends on provisions for capital cost recovery, while the tax base for personal income depends on the treatment of corporate distributions -- dividends, interest, and capital gains. To analyze the impact of tax incentives for investment the corporate income tax is the focus. Incentives to save are reflected in the personal income tax. Of course, both levels of taxation are required in assessing the impact of the corporate income tax.

My 1993 paper presented international comparisons of tax reforms for capital income over the period 1980-1990 in the G7 countries -- Canada , France , Germany , Italy , Japan , the United Kingdom , and the United States -- together with Australia and Sweden . These comparisons were based on marginal effective tax rates for different types of capital income in all nine countries for the years 1980, 1985, and 1990. These tax rates were constructed by nine teams, one from each country, using a common methodology incorporating that of King and Fullerton (1984).

My international comparison of marginal effective tax rates revealed widespread changes in the taxation of income from capital, similar to those in the U.S. Base broadening through elimination of investment and saving incentives and reductions in tax rates were nearly universal. This resulted in considerable "leveling of the playing field" among different assets. However, wide gaps among effective tax rates remained in all nine countries, so that important opportunities for tax reform still exist.

The incorporation of personal taxes into the corporate cost of capital raised a host of new issues. (8) In the "new" view proposed by King (1977) the corporation retinas earnings sufficient to finance the equity portion of investment and dividends are determined by the residual cash flow. The marginal source of equity funds is retained earnings, so that the tax rate on dividends does not affect the price of capital services or the effective tax rate on corporate income.

Under the new view of corporate finance and taxation, the most attractive investment available to the corporation is to liquidate its assets and repurchase its outstanding shares. Each dollar of assets liquidated enhances the value of the remaining shares. Repurchasing the firm's outstanding shares is ruled out by assumption, so that equity is "trapped" in the firm. Accordingly, this view of corporate taxation has been characterized as the "trapped equity" approach.

In chapter 11 Yun and I presented an alternative model of the cost of capital in the corporate sector. This is the "traditional" view of corporate finance and taxation employed, for example, by Harberger (1966), Feldstein and Summers (1979), and James Poterba and Summers (1983). In the traditional view the marginal source of funds for the equity portion of the firm's investments is new share issues, since dividends are fixed by assumption. An additional dollar of new share issues adds precisely one dollar to the value of the firm's assets.

It is important to underline the critical role of the assumption that dividends are a fixed proportion of corporate income. If the firm were to reduce dividends by one dollar to finance an additional dollar of investment, stockholders would avoid personal taxes on the dividends. The addition to investment would produce a capital gain that is taxed at a lower rate and shareholders would experience an increase in wealth. It is always in the interest of shareholders for the firm to finance investment from retained earnings rather than new issues of equity.

As Sinn (1991) has emphasized, both the traditional and the new view of corporate taxation depend crucially on assumptions about financial policy of the firm. The traditional view depends on the assumption that dividends are a fixed proportion of corporate income, so that the marginal source of funds for financing investment is new issues. The new view depends on the assumption that new issues of equity (or repurchases) are fixed, so that the marginal source of funds is retained earnings.

In fact, firms use both sources of equity finance, sometimes simultaneously. The King-Fullerton framework employed in my international comparisons of 1993 is based on the actual distribution of new equity finance from new issues and retained earnings. Since retained earnings greatly predominate over new issues, this approach turns out to be empirically equivalent to adopting the new view. Sinn (1991) suggests choosing new issues and retained earnings to minimize the cost of equity finance. This is also equivalent empirically to the new view for most countries.

The second set of issues raised by the introduction of personal taxes into the corporate cost of capital relates to the treatment of debt and equity in the corporate tax structure. In chapter 11 Yun and I assumed that debt-equity ratios are the same for all assets within the corporate sector. This assumption was also employed by King and Fullerton (1984). However, Roger H. Gordon, James R. Hines, and Summers (1987) argued that different types of assets should be associated with different debt-equity ratios.

The inclusion of personal taxes on corporate distributions to equity holders also raises more specific issues on the impact of inflation in asset prices. A comprehensive treatment of these issues is provided by Feldstein (1983). Since nominal interest expenses are deductible at the corporate level, while nominal interest payments are taxable at the individual level, an important issue is the impact of inflation is reflected point for point in changes in nominal interest rates. Summers (1983) provided empirical support for this assumption.

The cost of capital has become an indispensable analytical tool for studies of the economic impact of tax policies. These studies have taken two forms. First, the cost of capital has been incorporated into investment functions in macroeconometric models. These models are useful primarily in modeling the short-run dynamics of responses to changes in tax policy. More recently, the cost of capital has been incorporated into applied general equilibrium models that focus on the impact of tax policy on the allocation of capital. These models are essential for capturing the long-run effects of tax reforms.

Investment functions incorporating the cost of capital were proposed for the Brookings quarterly macroeconometric model of the U.S. in my 1965 paper, reprinted in the companion volume, Capital Theory and Investment Behavior . I also constructed more detailed investment functions in papers I published with James A. Stephenson in 1967 and 1969 and Sidney S. Handel in 1971, also reprinted in that volume. This work is summarized in my 1971 survey paper, reprinted there.

At the beginning of the debate over the Economic Recovery Tax Act of 1981 the investment equations for all major forecasting models for the U.S. economy had incorporated the cost of capital. (9) Simulations of alternative tax policies had become the staple fare of debates over the economic impacts of specific tax proposals. Illustrations of this type of simulation study are provided by my 1971 paper, reprinted as chapter 4 below, and my 1976 paper with Gordon, reprinted as chapter 7. Both papers employed modifications of the DRI quarterly macroeconometric model of the U.S.

My 1971 paper replaced the investment functions in the DRI quarterly model with alternative functions like those in my 1971 paper with Hall, chapter 2. The specific focus of this paper was the economic impact of a new system of depreciation tax rules -- the Assets Depreciation Range (ADR) System announced by President Richard M. Nixon on January 11, 1971. I simulated the U.S. economy for the five year period 1971-5 with and without this policy change. I also compared the impact of the ADR System with that of re-institution of the investment tax credit, which has been suspended in 1969.

The economic impacts of the tax policy changes analyzed in my papers with Hall, chapters 1 and 2, were limited to simulations of investment expenditures and capital stocks. We modeled the short-run dynamics of investment by holding prices and interest rates as well as the level of economic activity constant. By incorporating our investment functions into the DRI quarterly model, I was able to project impacts on employment and economic activity, prices and interest rates, and government deficits. All of these variables were determined endogenously by the DRI model.

The focus of chapter 7 with Gordon was a proposal to employ the investment tax credit as an instrument of counter-cyclical policy. Under this proposal a tax credit would apply during recessions, but would be eliminated during booms. We simulated the impact of the credit, including its introduction in 1962, its enhancement with repeal of the Long Amendment in 1964, the suspension of the credit in 1966-7, repeal in 1969, and, finally, re-institution in 1971. We again employed the DRI model in these simulations.

The conclusion of my work with Gordon was that the use of the investment tax credit as an instrument of counter-cyclical policy had been highly detrimental to economic stability. Of the five major changes in the tax credit since its introduction in 1962, three were badly mistimed and two were in the wrong direction. We concluded that growth, rather than stabilization, should be the primary criterion for selecting the tax credit rate.

One of the important innovations in the econometric model of investment that I introduced in 1963 was the use of an explicit model of production. This generated a sizable literature that is summarized and evaluated in chapter 6. In this chapter I focus on the assumptions that the production function is Cobb-Douglas with elasticity of substitution equal to unity and constant returns to scale. I showed that both of these assumptions were consistent with a broad range of empirical evidence from both cross section and time series studies of productions. They have become the standard assumptions for much subsequent work on investment behavior, including that reported in chapters 1 ,2,4 and 7 and the work of Andrew B. Abel (1981).

Another important issue in this type of application, emphasized by Robert E. Lucas (1976) in his critique of econometric methods for policy evaluation, is modeling expectations about future prices of investment goods. This is required in measuring the cost of capital and simulating the impact of changes in tax policy. Since my 1968 papers with Sabered , future prices have been modeled by means of perfect foresight or rational expectations. However, macroeconometric models, such as the DRI model, did not incorporate rational expectations into simulations of alternative policies.

The model of capital as a factor of production, introduced in my 1963 paper and described in much greater detail in chapter 5, contains two dynamic relationships. The first is an accumulation equation, expressing capital stock as a weighted sum of past investments. The second is a capital asset pricing equation, expressing the price of investment goods as the present value of future rentals of capital services. Both relationships should be incorporated into simulations of the effects of changes in tax policy. Macroeconometric models have incorporated the backward-looking equation for capital stock, but have omitted the forward-looking equation for the price of investment of goods.

The omission of the capital asset pricing equation from macroeconometric models was due to the lack of simulation techniques appropriate for perfect foresight or rational expectations. (10) To evaluate the economic impact of the 1981 tax reforms, I constructed a dynamic general equilibrium model with Yun that incorporated both the backward-looking equation for capital stock and the forward-looking equation for asset pricing. This model is presented in our 1986 paper, "The Efficiency of Capital Allocation," reprinted as chapter 10.

In the model presented in chapter 10 equilibrium is characterized by an intertemporal price system that clears markets for labor and capital services and consumption and investment goods. This equilibrium links the past and the future through markets for investment goods and capital services. Assets are accumulated through investments, while asset prices equal the present values of future services. Consumption must satisfy conditions for intertemporal optimality of the household sector under perfect foresight. Similarly, investment must satisfy requirements for asset accumulation.

In collaboration with Lawrence Lau, Christensen and I constructed an econometric model of producer behavior based on the translog production possibility frontier. This is presented in our 1973 paper, reprinted in the companion volume, Econometrics and Producer Behavior . We estimated this model from data on inputs and outputs in the production account of the Christensen-Jorgenson accounting system. Yun and I incorporated this production model into the model of U.S. economic growth presented in chapter 10.

An important feature of the model of chapter 10 is the representation costs of adjustment. The production possibility frontier captures the demand for capital services and costs of adjusting this demand through the supply of investment goods. The production of investment goods entails foregoing the opportunity to produce consumption goods. The costs of adjusting capital services through investment are external rather than internal and are reflected in the market price of investment goods.

By contrast my 1973 paper, reprinted in chapter 3, presented a model of investment behavior with internal adjustment costs and irreversibility. (11) Internal adjustment costs are reflected in the loss of capital services that must be devoted to the installation of capital rather than the production of marketable output. The cost of capital in these models is a shadow price that reflects both the market price of investment and the shadow value of installation. The model of external adjustment costs in chapter 10 has the decisive advantage that the cost of capital depends only on the market price of investment.

In 1975 Christensen, Lau, and I constructed an econometric model of consumer behavior based on the translog indirect utility function. Our 1975 Christensen, Lau, and I constructed an econometric model of consumer behavior based on the translog indirect utility function. Our 1975 paper is reprinted in the companion volume, Modeling Consumer Behavior . We estimated this model from data in the income and expenditure account of Christensen-Jorgenson accounting system. Yun and I incorporated this model of consumer behavior into the model presented in chapter 10.

My paper with Yun , "Tax Policy and Capital Allocation," reprinted as chapter 11 below, employed our model to simulate the economic impact of the 1981 Tax Act. We found that this tax reform increased potential U.S. economic welfare by 3.5 to 4 percent of the U.S. private national wealth in 1980. We also considered alternative tax reforms, including a shift from income to consumption as the base for taxation. We found that the replacement of corporate and personal income taxes by a consumption tax would have produced dramatic gains in welfare, amounting to 26 to 27 percent of private national wealth.

Yun and I also evaluated the economic impact of the 1986 tax reform, using a new version of our dynamic general equilibrium model of the U.S. economy in our 1990 paper, reprinted as chapter 12. We summarized the 1986 reform in terms of changes in tax rates, the treatment of deductions from income, the availability of tax credits, and provisions for indexing. We also summarized reform proposals that figured prominently in the debate leading up to the 1986 Tax Act. For this purpose we utilized marginal effective tax rates and tax wedges, defined in terms of differences in tax burdens imposed on different types of capital.

My 1990 paper with Yun evaluated the effects of changes in tax policy on economic efficiency by measuring the corresponding changes in potential economic welfare. The reference level of welfare, which served as the basis of comparison among alternative tax policies, was the level attainable under U.S. tax law prior to 1986. Finally, we analyzed losses in efficiency associated with different tax wedges.

We found that much of the potential gain in welfare from the 1986 reform was dissipated through failure to index the income tax base for inflation. At rates of inflation near zero the loss is not substantial. However, at moderate rates of inflation, like those prevailing in the mid-1980s, the loss is significant. Second, the greatest welfare gains would have resulted from equalizing effective tax rates on household and business assets. The potential welfare gains from an income-based tax system, reconstructed along these lines, would have exceeded those from an consumption-based system.

Effective tax rates or tax wedges do not complete the analysis of distortionary effects of capital income taxes. These effects also depend on substitutability among assets. As an example, consider the allocation of capital between short-lived and long-lived depreciable assets in the corporate sector. Even if the interasset difference in tax treatment were large, the distortion of capital allocation could be small if services of the two types of assets were not substitutable. Similarly, the distortion in resource allocation over time could be small if intertemporal substitutability in consumption were small.

In my 1991 article with Kun-Young Yun , reprinted as chapter 13, efficiency losses due to taxation were assessed by means of the concept of the excess burden of taxation. We measured this burden by comparing the growth of the U.S. economy under the actual tax system with growth under a purely hypothetical "lump sum" system. Under the alternative system taxes are levied as a lump sum deduction from wealth, rather than as a proportion of transactions in outputs and factor services. Taxes on these transactions insert tax wedges between demand and supply prices and produce losses in efficiency.

In chapter 13 we evaluated alternative tax policies in terms of the welfare of a representative consumer. For this purpose we employed an intertemporal expenditure function. This gives the wealth required to achieve the level of welfare produced by each tax policy at prices associated with a reference tax policy. We measured the excess burden for the U.S. tax system before and after the Tax Reform Act of 1986.

We summarized our comparisons of alternative tax policies in terms of average and marginal excess burdens per dollar of tax revenue raised. The average excess burden is the cost per dollar if the tax is wholly replaced by a lump sum tax. The marginal excess burden is the cost of replacing only the first dollar of revenue raised by the tax. Marginal excess burdens are relevant for reform. We find that the marginal excess burden of taxation after the 1986 Tax Act was 39.1 cents per dollar of revenue raised, while this burden was 47.2 cents per dollar before the reform.

While the 1986 Tax Act reduced the excess burden of taxation, this legislation did not successfully address the imbalances of the U.S. tax system. After the reform the marginal excess burden of sales taxes was only 26.2 cents per dollar of revenue raised, while the cost of property taxes was even lower at 17.6 cents per dollar. By contrast the cost of income taxes was 49.7 cents per dollar. Substantial imbalances among different income tax programs also remained. The marginal efficiency cost of corporate income taxes was 44.8 cents per dollar. The cost of labor income taxes was 37.6 cents, while the cost of individual taxes on capital income was $1.02 for every dollar of revenue raised!

The analysis of the economic impact of tax policy required the integration of the cost of capital into macroeconometric models and applied general equilibrium models. The economic impact of the Tax Reform Act of 1986 has been analyzed by means of models of both types. In the simulations of tax policies presented in chapters 11 and 12, each policy was associated with an intertemporal equilibrium, including markets for different types of capital. The disaggregation of capital exposed all the margins for substitution affected by the 1981 and 1986 changes in tax policy.

Shortly after the passage of the Tax Reform Act of 1986, the Department of the Treasury (1987) published a study by Fullerton , Gillette, and Mackie (1987) of the effect of the new legislation on marginal effective tax rates. The results were incorporated into an applied general equilibrium model by Fullerton , Yolanda K. Henderson, and Mackie (1987) and used to estimate the economic impact. Fullerton (1987) presented a closely related study of marginal effective tax rates, while Fullerton and Henderson (1989a, 1989b) conducted a parallel simulation study. (12)

My final objective is to evaluate the cost of capital as a practical guide to tax reform. The primary focus is U.S. tax policy, since the cost of capital has been used much more extensively in the U.S. than other countries. Auerbach and I introduced the key concept, the marginal effective tax rate, early in the debate over the U.S. Economic Recovery Tax Act of 1981. We showed that the tax policy changes of the early 1980s, especially the 1981 Tax Act, would increase barriers to efficient allocation of capital.

By contrast the Tax Reform Act of 1986 substantially reduced barriers to efficiency. The erosion of the income tax base to provide incentives for investment and saving was arrested through vigorous and far-reaching reforms. Incentive were sharply curtailed and efforts were made to equalize marginal effective tax rates among assets. The shift toward expenditure and away from income as a tax base was reversed. My international comparisons of 1993 showed that these reforms had important parallels in other industrialized countries.

The cost of capital approach has also proved its usefulness in pointing the direction for future tax reforms. The initial focus was the allocation of capital within the corporate sector, as in my papers with Auerbach and Sullivan, chapters 8 and 9, and the important extensions by Fullerton and King (1984). My 1990 paper with Yun , chapter 12, revealed that important discrepancies remain between effective tax rates on owner-occupied housing and business capital, so that opportunities still exist for improvements in efficiency.

My international comparisons of 1993 showed that equalizing the tax burdens on housing and business capital has proved to be extraordinarily difficult within the framework of the income tax. An alternative approach would be to revive proposals to convert the tax base from income to consumption. However, taxation of owner-occupied housing as an expenditure involves precisely the same difficulties as taxation on the basis of income, since expenditure and income on this housing are identical.

In the United States proposals to replace income by consumption as a tax base include the "flat tax" of Hall and Rabushka (1983), a European-style consumption-based value added tax, and a comprehensive retail sales tax on consumption. My papers of 1986 and 1990 with Yun , reprinted as chapters 11 and 12, demonstrated that a consumption tax would have been markedly superior to the tax reforms that took place in 1981 and 1986. However, so also showed that an income-based tax that "levels the playing field" among all assets is superior to a consumption tax.

My overall conclusion is that the cost of capital and the closely related concept of the marginal effective tax rate have provided an important intellectual impetus for tax reform. Effective tax rates at both corporate and personal levels are now available for many countries around the world. International comparisons of tax reforms have provided extensive illustrations of successful applications of the cost of capital approach. My hope is that these illustrations will serve as an inspiration and a guide for policy makers who share my goal of making the allocation of capital with a market economy more efficient.

Endnotes:

1. See Sven-Olaf Lodin (1978), James E. Meade, et. al. ( 1978), and U.S. Treasury (1977).

2. Robert E. Hall and Alvin Rabushka (1983) and David F. Bradford (1986) presented detailed proposals for a consumption-based tax in the United States . These were rejected in favor of an income-based approach by the U.S. Treasury (1984).

3. An illuminating account for the tax debate preceding the 1986 Tax Act has been given by Jeffrey H. Birnbaum and Alan S. Murray (1987).

4. Harberger's (1966) estimates were subsequently revised and corrected by John Shoven (1976).

5. The objectives of the 1986 tax reform are discussed by Charles E. McLure and George R. Zodrow (1987).

6. See U.S. Treasury (1984).

7. See, especially, Hulten and Wykoff (1981a), p. 387.

8. Summaries of the alternative views of taxation and corporate finance are given by Anthony B. Atkinson and Joseph E. Stiglitz (1980), esp. Pp. 128-159, Auerbach (1983), and Hens-Werner Sinn (1991).

9. See for example, Robert S. Chirinko and Robert Eisner (1983) and Jane G. Gravelle (1984).

10. These techniques were introduced by David Lipton, Poterba , Jeffrey Sachs, and Summers (1982) and Ray C. Fair and John B. Taylor (1983) long after the methodology for constructing and simulating macroeconometric forecasting models had crystallized.

11. This model combined features of models originated by Arrow (1968) and Lucas (1967a).

12. Henderson (1991) surveys six studies of the economic impact of the 1986 Tax Act by means of applied general equilibrium models. Except for my model with Yun , these models did not include the capital asset pricing equation and are subject to the "Lucas critique."