Review of Other Studies

Analyses of business fixed investment in the United States may be divided into several categories. One group of studies treats investment in terms of its fundamental determinants, such as fiscal and monetary policies, using complete macroeconomic models of the U.S. economy (for example, Brayton and Clark (1985)). A second approach—the most common one in the literature and the one into which category this study falls—is essentially partial equilibrium and relates investment to its proximate macroeconomic determinants. A third approach dismisses conventional empirical techniques and argues that changes in tax incentives in recent years have been so powerful as to invalidate the conventional macroeconomic approach to the behavior of business investment (for example, Roberts (1982)). But whatever the analytical approach taken, the dominant issue in recent empirical studies of fixed investment has been the importance of fiscal incentives Excellent surveys of the literature on the impact of tax policy on business investment have been provided by Bosworth (1984) and Chirinko (1986). These surveys note that overall agreement does not appear to exist on the role of tax policy: some studies ascribe very little influence to tax variables in affecting the course of business fixed investment, whereas others indicate a more substantial role. To the extent that a consensus exists, it appears to be that the predominant determinants of business fixed investment are major macroeconomic variables, such as the rate of growth of output, the level of interest rates, and the state of inflationary expectations. Tax policy, although found generally significant, appears to play a subsidiary role.

One substantial strand of the literature involves assessing the impact on business fixed investment of the ERTA of 1981 and the Tax Equity and Fiscal Responsibility Act (TEFRA) of 1982, the relevant provisions of which are outlined in the Appendix. Bosworth (1985) examined the behavior of business investment from 1981 to 1984 and suggested that tax policy may not have been the major factor stimulating the growth of business investment. Although he found that aggregate equations under-predicted investment growth, the areas of strongest increase were not those that had received the greatest stimulus from tax policy. Brayton and Clark (1985) performed simulations using the Federal Reserve/MPS model of the U.S. economy. They found quite substantial ERTA/ TEFRA effects on business investment, but the bulk of the impact stemmed from the multiplier/accelerator mechanism on output, with the magnitude of the pure tax policy effect—operating through the cost of capital—somewhat smaller. Sahling and Akhtar (1985) estimated two standard investment models—one based on the specification in the MPS model and the other on that of the Bureau of Economic Analysis model. They found that conventional aggregative models tracked behavior quite well, and that the fast growth in business investment in plant and equipment in the early phase of the current expansion stemmed primarily from the rapid increase in output and the steep decline in the cost of funds. The role attributed to changes in tax policy was significant but secondary. Meyer (1984), also employing an approach that relied on the cost of capital, found that changes in tax policy helped to explain changes in investment behavior, but by no means fully. By contrast, Feldstein and Jun (1986) found more substantial tax policy effects than most other studies, but their methodology appeared to confound the conceptually distinct influences of changes in statutory tax policy and changes in effective tax rates attributable to variations in inflation and the associated taxation of artificial inventory profits.

A second major strand of the literature examines the potential impact of the TRA of 1986 on business fixed investment. This body of research is discussed in Evans and Kenward (1988) and will thus not be reviewed at length here. The essential conclusion from that literature is that the TRA raised the cost of capital for both machinery and equipment and nonresidential structures through the elimination of the investment tax credit and the shift to less generous depreciation allowances. These factors are generally judged likely to outweigh a possible enhancement of the efficiency of investment resulting from the shift to a more neutral tax system.

A distinguishing feature of the empirical results presented later in the present paper is that the effects of all major tax policy changes undertaken thus far in the 1980s—ERTA/TEFRA and TRA—are treated in a common framework permitting an overall evaluation to be made. Most other studies have analyzed one episode or the other but not both.

Investment Equations

The equations used in this study are based on the standard neoclassical theory of capital accumulation, according to which the optimal combination of factor inputs used by firms depends on the relative prices of those inputs. If output is produced under competitive conditions and if the production function is of the Cobb-Douglas type, the desired capital stock at each point in time will be given by the expression:⁶

where

K^d = desired net capital stock

Y = output

C = real user cost of capital

α = a constant.

The flow of net investment is the change in the actual capital stock as it adjusts toward a new desired level. Because the actual capital stock can be changed only gradually over time, net investment (that is, the change in the net capital stock) will depend on current and lagged values of the desired capital stock:

where

$l_t^N$ = net investment

and where

is a polynomial in the lag operator, L, such that (L^jX_t =X_t-1 Combining equations (1) and (2) and assuming that replacement investment is proportional to the lagged capital stock, we can write gross investment, l^g, as

where γ(L) = αβ(L), and 8 is the rate of economic depreciation.

Fiscal Incentives and the Cost of Capital

The cost of capital (C plays a central role in the theory of investment outlined in the previous subsection. This concept stems from the neoclassical theory of optimal capital accumulation (Hall and Jorgenson (1967)) in which the price of a capital asset is equated to the present value of after-tax services expected to accrue from that good. According to this approach, the cost of capital in real terms may be expressed as follows:⁷

View Expanded

where

• u = the marginal corporate tax rate

• i = the average cost of funds⁸

• δ = the rate of economic depreciation

• ṗ^e = the expected rate of change of the price of fixed assets⁹

• τ = the tax rate on capital gains¹⁰

• k = the investment tax credit per dollar of new investment

• z = the present value of a dollar of depreciation deductions

• q = the price of capital assets relative to the GNP deflator.

An important implicit assumption in the traditional cost of capital formulation is that each asset is depreciated for tax purposes only once. However, as observed by Gordon, Hines, and Summers (1986), if a liquid secondary market exists for a particular asset category (as, for example, with several types of transportation equipment and preeminently with commercial structures), it may be possible for an asset to be depreciated for tax purposes several times by different owners. Such a practice—described as “asset churning”—would likely magnify the effects of depreciation concessions for such assets and, concomitantly, would magnify the negative effects stemming from withdrawal of previous concessions. Gordon, Hines, and Summers suggest that such a process may have been at work in the case of hotels and office buildings, for which investment expenditure rose sharply in the wake of the ERTA; subsequently, a sharp decline occurred in such expenditures when it became clear that the TRA would curtail previously generous depreciation allowances.

Capital gains taxation raises the level of the cost of capital—via the expected inflation term—and affects the time profile when either tax rates or inflation are changing (see equation (5)). In constructing the cost of capital in real terms, the variable τ is approximated in the empirical work by the (legislated) maximum rate on capital gains. Note, however, that this construction may not be representative of the effective capital gains tax rate because capital gains are often deferred or excluded from tax at the end of the lifetime of the asset. To illustrate the implications of alternative assumptions about the tax rate on capital gains. two measures of the cost of capital variables are provided in Table 4. The first measure sets the capital gains tax rate to its maximum, whereas the second sets the tax rate to zero. The effect of setting the capital gains tax rate to zero is to lower the level of the cost of capital, especially in the early part of the decade when inflation was high.

Table 4.

The Cost of Capital Under Alternative Views of Capital Gains Taxation

Table 4.

The Cost of Capital Under Alternative Views of Capital Gains Taxation

		Item	1982	1983	1984	1985	1986
Equipment
	Real cost of capital
		With capital gains tax rate set to
		maximum rate	9.7	8.6	9.4	9.3	8.3
		With capital gains tax rate set to zero	8.1	7.3	8.2	8.3	7.3
Structures
	Real cost of capital
		With capital gains tax rate set to
		maximum rate	7.2	5.1	5.9	5.8	4.5
		With capital gains tax rate set to zero	5.2	3.4	4.3	4.5	3.2

View Table

Table 4.

The Cost of Capital Under Alternative Views of Capital Gains Taxation

		Item	1982	1983	1984	1985	1986
Equipment
	Real cost of capital
		With capital gains tax rate set to
		maximum rate	9.7	8.6	9.4	9.3	8.3
		With capital gains tax rate set to zero	8.1	7.3	8.2	8.3	7.3
Structures
	Real cost of capital
		With capital gains tax rate set to
		maximum rate	7.2	5.1	5.9	5.8	4.5
		With capital gains tax rate set to zero	5.2	3.4	4.3	4.5	3.2

Estimation

Equations corresponding to equation (4) above were estimated by ordinary least squares (with a correction for serial correlation) over sample periods beginning in the first quarter of 1964. All variables were normalized with respect to middle expansion path GNP, to reduce potential heteroscedasticity, and the restrictions γ_1i, = -γ_2i were imposed, as noted earlier.¹¹

The changes in the investment tax laws during the 1980s were substantial and. in the case of the TRA, involved a fundamental shift in philosophy away from the previous approach of using tax concessions to encourage investment. Some commentators (for example, Roberts (1982)) have argued that economic behavior shifted fundamentally after the tax legislation of the early 1980s, in ways that would not be captured by a traditional approach. Under these circumstances, a conventional investment function such as equation (4) could well fall victim to the so-called Lucas critique.¹² In particular, after a change in tax regime, the behavioral responses of economic agents—in this instance, businesses—might change, with the result that empirical parameters, such as the Cobb-Douglas coefficient (a) or the adjustment coefficients (7), might not be constant over time. Such a result would undermine the simulations reported in the next section.

With this in mind, Chow-Fisher tests to examine parameter stability were conducted. The equations were first estimated over a period to the third quarter of 1981, before the introduction of the ERTA. The estimation period was then extended to the fourth quarter of 1985, covering the implementation of the ERTA and TEFRA. Finally, the equations were estimated over a period extending to the second quarter of 1987. The producers’ durable equipment equation passed the tests at the 5 percent significance level for both of the subperiods. The structures equation exhibited stabilih in the first subperiod but marginally failed the Chow-Fisher test at the 5 percent significance level—although it passed at the 1 percent significance level—in the second subperiod. Reflecting these results, the out-of-sample dynamic forecast performance of the structures equation from the first quarter of 1986 to the second quarter of 1987 was poor, whereas the machinery and equipment equation tracked developments reasonably well. A large part of the forecast error pertaining to the structures equation stemmed from the collapse of investment in the oil sector following the oil price fall at the end of 1985. This factor is not captured in the structures equation.

Passing a Chow-Fisher test is only a necessary condition—it is not sufficient—for parameter stability; nevertheless, the test results provide some confidence that tax legislation of the 1980s did not alter fundamental investment behavior so that the effects of the tax changes can be adequately analyzed via their effects on the cost of capital using the empirically estimated equations. For the purpose of the counterfactual simulations discussed in Section III, the fourth quarter of 1985 was chosen as the estimation end period for both equations, reflecting a desire to utilize as much information in the data sample as possible while retaining some residual doubts about parameter stability in the structures equation after the introduction of the TRA.

Producers’ Durable Equipment

where

• w₀ = 0.0066 (4.8)

• w₁ = 0.0096 (8.6)

• w₂ = 0.0118 (10.5)

• w₃ = 0.0132 (11.0)

• w₄ = 0.0139 (11.1)

• w₅ = 0.0140 (11.0)

• w₆ = 0.0136 (10.6)

• w₇ = 0.0127 (10.0)

• w₈ =0.0115 (8.9)

• w₉ =0.0100 (7.6)

• w₁₀ = 0.0083 (6.3)

• w₁₁ = 0.0066 (5.1)

• w₁₂ = 0.0047 (4.0)

• w₁₃ = 0.0030 (3.2)

• w₁₄ = 0.0014 (2.4)

and where Σw_i = 0.1410; R² = 0.967; DW = 2.15; ρ = 0.82 (13.0); and the sample period was 1964:1-1985:4.

Nonresidential Structures

where

• v₀ = 0.00121 (2.6)

• v₁ = 0.00162 (3.9)

• v₂ = 0.00189 (4.2)

• v₃ = 0.00203 (4.1)

• v₄ = 0.00207 (4.0)

• v₅ = 0.00201 (3.9)

• v₆ = 0.00188 (3.6)

• v₇ = 0.00168 (3.2)

• v₈ = 0.00143 (2.8)

• v₉ = 0.00115 (2.3)

• v₁₀ = 0.00085 (1.8)

• v₁₁ = 0.00055 (1.4)

• v₁₂ = 0.00026 (1.1)

and where Σv_i = 0.01863; R² = 0.944; DW= 1.44; p = 0.92 (20.0); and the sample period was 1964:1-1985:4.

The notation for the foregoing is as follows:

• IE = investment in producers’ durable equipment, 1982 dollars

• IS = investment in nonresidential structures, 1982 dollars

• Y = GNP, 1982 dollars

• Y* = middle expansion path GNP, 1982 dollars¹³

• CE =the real cost of capital for producers’ durables

• CS = the real cost of capital for nonresidential structures

• KS and KE = the stock of nonresidential structures, 1982 dollars, and the stock of producers’ durable equipment, 1982 dollars, respectively¹⁴

• D = a dummy variable to allow for credit controls in 1980:2

• R² = the adjusted coefficient of determination

• DW = the Durbin-Watson statistic

• ρ = the estimated serial correlation coefficient.

The coefficients W_i and v_i were estimated using third-degree Almon polynomial distributed lags with a zero end-point constraint; figures in parentheses beneath coefficients are t-statistics.

The Effects of Tax Policy on the Cost of Capital

The ERTA of 1981 reflected the underlying view that had prevailed through much of the 1960s and 1970s—that the tax code should be used to encourage some activities and discourage others. In this instance, the activity to be encouraged was capital formation, and to this end incentives were provided to promote business fixed investment. The investment tax credit (ITC) was liberalized to cover a broader range of shortlived equipment, and the Accelerated Cost Recovery System (ACRS) for asset depreciation was introduced. Both these measures reduced the after-tax cost of capital for new equipment and nonresidential structures; the extension of the ITC to a broader range of equipment raised the value of the variable k in equation (5), while the ACRS increased the present value of depreciation allowances—the variable z in equation (5).

Under the TEFRA of 1982. tax incentives for investment were reduced, offsetting in part the effects of the ERTA.¹⁶ The rationale for the changes was essentially twofold: first, it was judged necessary to take measures aimed to reduce fiscal deficits, and second, the reduction in taxation of corporate income resulting from the ERTA had led to concerns about tax equity. The planned extension of the ACRS toward even more favorable depreciation methods (175 percent declining balance compared with 150 percent declining balance under the ERTA] was withdrawn. In addition, some of the benefits of the ITC were withdrawn by reducing the depreciable base of assets by 50 percent of the ITC. Notwithstanding the changes introduced in the TEFRA. the underlying approach of using the tax code to encourage business investment remained in place.

The TRA of 1986 made further substantial changes in tax laws affecting business fixed investment.¹⁷ In particular, the underlying philosophy of tax policy with regard to business investment became substantially less interventionist; henceforth, the tax code was to be as neutral as possible, rather than actively encouraging investment. The new view was that the additional investments encouraged by tax preferences were likely to be relatively inefficient. Consequently, it was better for investment decisions to be made with reduced attention to tax consequences. Such a result could be achieved by eliminating investment incentives and at the same time lowering tax rates. In this vein, the ITC was abolished, effective the beginning of 1986, although the legislation was not passed until late in the year; depreciation schedules for both equipment and structures were made substantially less generous; and the maximum tax rate on capital gains was raised to 28 percent, while the maximum corporate tax rale was reduced from 46 percent to 34 percent, effective 1988.¹⁸

Figure 6 and Table 5 show the estimated effects of tax policy changes on the real after-tax cost of capital since 1981. In the case of equipment, the effect of the ERTA was to raise the present value of a dollar of depreciation allowances from 78 cents to 86 cents. The changes in depreciation allowances, when combined with the more generous ITC, lowered the real cost of capital for equipment by 1 percentage point in 1982. As regards structures, the present value per dollar of depreciation allowances increased, owing to the ERTA. from 46 cents to 65 cents. As a result, the cost of capital for structures was l¼ percentage points lower than it would have been without the ERTA.

U.S. Real User Cost of Capital

Citation: IMF Staff Papers 1989, 001; 10.5089/9781451956825.024.A002

• Download Figure
• Download figure as PowerPoint slide

View Full Size

U.S. Real User Cost of Capital

Citation: IMF Staff Papers 1989, 001; 10.5089/9781451956825.024.A002

• Download Figure
• Download figure as PowerPoint slide

U.S. Real User Cost of Capital

Citation: IMF Staff Papers 1989, 001; 10.5089/9781451956825.024.A002

• Download Figure
• Download figure as PowerPoint slide

Table 5.

Effect of Tax Changes on the Real Cost of Capital

Note: Figures for the present value of depreciation allowances are in dollars; those for the cost of capital are in percent. Figures exclude relative price effects.

^aHistorical values for variable z in equation (5).

^bHistorical values that would have prevailed without tax policy changes.

Table 5.

Effect of Tax Changes on the Real Cost of Capital

		Item	1982	1983	1984	1985	1986	First Half ot 1987
Equipment
	Present value of depreciation
		allowancesa	0.86	0.88	0.88	0.89	0.91	0.89
		Excluding tax policy changesb	0.78	0.82	0.82	0.83	0.86	0.86
	Real cost of capital		9.7	8.6	9.4	9.3	8.3	9.6
		Excluding tax policy changesb	10.6	9.3	10.2	10.0	8.8	8.3
Structures
	Present value of depreciation
		allowancesb	0.65	0.70	0.67	0.66	0.71	0.53
		Excluding tax policy changesb	0.46	0.52	0.50	0.53	0.59	0.61
	Real cost of capital		7.2	5.1	5.9	5.8	4.5	5.0
		Excluding tax policy changesa	8.5	6.3	7.2	6.9	5.2	4.5

Note: Figures for the present value of depreciation allowances are in dollars; those for the cost of capital are in percent. Figures exclude relative price effects.

^aHistorical values for variable z in equation (5).

^bHistorical values that would have prevailed without tax policy changes.

View Table

Table 5.

Effect of Tax Changes on the Real Cost of Capital

		Item	1982	1983	1984	1985	1986	First Half ot 1987
Equipment
	Present value of depreciation
		allowancesa	0.86	0.88	0.88	0.89	0.91	0.89
		Excluding tax policy changesb	0.78	0.82	0.82	0.83	0.86	0.86
	Real cost of capital		9.7	8.6	9.4	9.3	8.3	9.6
		Excluding tax policy changesb	10.6	9.3	10.2	10.0	8.8	8.3
Structures
	Present value of depreciation
		allowancesb	0.65	0.70	0.67	0.66	0.71	0.53
		Excluding tax policy changesb	0.46	0.52	0.50	0.53	0.59	0.61
	Real cost of capital		7.2	5.1	5.9	5.8	4.5	5.0
		Excluding tax policy changesa	8.5	6.3	7.2	6.9	5.2	4.5

Note: Figures for the present value of depreciation allowances are in dollars; those for the cost of capital are in percent. Figures exclude relative price effects.

^aHistorical values for variable z in equation (5).

^bHistorical values that would have prevailed without tax policy changes.

The benefit of the ITC on producers’ durable equipment was reduced by the TEFRA in 1983, offsetting in part the effects of the ERTA on the cost of capital (see Table 5). In 1983, the combined effects of ERTA/TEFRA lowered the cost of capital for equipment by ¾ percentage point relative to what it would have been without both pieces of legislation. In the ease of structures, both the TEFRA and the 1984 Deficit Reduction Act (DEFRA) had little impact on the cost of capital. In 1984. the combined effect of all three pieces of legislation was to lower the cost of capital for structures by 1¼ percentage points compared with what it would have been otherwise.

As noted above, the TRA of 1986 brought about further major changes in business taxation. The net effect was to increase substantially the cost of capital for both equipment and structures. The combined effect of ERTA/TEFRA/DEFRA and TRA was to push up the cost of capital for equipment and structures by 1¼ and ½ percentage points, respectively, relative to what they would have been without all this legislation.

Tax Policy Simulations

The first tax policy simulation was performed by assuming no change in the tax treatment of investment over the period 1981-87, and by comparing the path of investment simulated under that assumption with the path of investment predicted by the equation.¹⁹ Actual values of output and interest rates were used in the simulation; no attempt was made to determine the secondary effects of tax policy on these variables or any repercussions of these effects on investment. The results of the simulation are presented in Table 6. On this basis, the stimulative effect of the legislative changes on investment amounted to roughly 2 percent of the level of both producers’ durable equipment and nonresidential structures by 1985.

Table 6.

Sources of Change in U.S. Business Fixed Investment

(As percentage of investment in the first quarter of 1983)

^aNormal growth was assumed to be the average of output growth that took place in the economic expansions that began after troughs in real GNP in 1960:4 and 1975:1. The subsequent expansions were the only two of comparable length to the present one. The calculation was made by comparing results from a simulation with normal output growth thus defined with one with real GNP held constant at its value of 1982:4.

^bCalculated by comparing a simulation with actual output growth with one with “normal” output growth.

^cCalculated using a simulation in which the cost of funds was held at its 1982:3 value.

^dCalculated using a simulation in which the relative prices of investment goods were held at their 1982:3 values.

^eCalculated by comparing a simulation in which none of the changes in tax policy from 1981 onward took place with one in which all the changes took place.

^fActual decline in petroleum drilling investment from 1985:4 to 1987:2.

^gCalculated multiplicatively. A positive residual indicates growth in investment in excess of that explained by the contributing factors.

Table 6.

Sources of Change in U.S. Business Fixed Investment

(As percentage of investment in the first quarter of 1983)

Year	Actual Charge	Normal Output Recoverya	Above Average Output Recoveryb	Lower Cost of Fundsc	Changes in Relative Pricesd	Tax Policye	Petroleum Drilling Collapsef	Residualg
			Nonresidential fixed investment
1983:4	14.3	8.5	0.8	1.7	0.3	0.6	—	1.8
1984:4	30.1	17.5	5.0	4.2	1.1	1.8	—	-1.7
1985:4	36.2	23.9	1.6	4.7	1.7	2.1	—	-0.5
1987:2	28.2	31.6	-9.0	5.1	2.5	1.4	-3.6	1.7
			Producers’ durable equipment
1983:4	21.3	10.7	1.0	1.9	0.3	0.5	—	5.6
1984:4	39.3	22.3	6.3	4.8	1.3	1.6	—	-0.7
1985:4	49.0	31.2	2.1	5.5	2.3	2.0	—	1.0
1987:2	48.8	41.9	-11.5	6.2	3.6	1.5	—	6.1
			Nonresidential structures
1983:4	2.6	4.5	0.5	1.4	0.2	0.9	—	-4.7
1984:4	14.8	8.7	2.6	3.2	0.6	2.2	—	-3.0
1985:4	14.8	10.5	0.7	3.1	0.7	2.1	—	-2.7
1987:2	-6.1	12.5	-4.3	3.0	0.5	1.3	-9.7	-7.9

^aNormal growth was assumed to be the average of output growth that took place in the economic expansions that began after troughs in real GNP in 1960:4 and 1975:1. The subsequent expansions were the only two of comparable length to the present one. The calculation was made by comparing results from a simulation with normal output growth thus defined with one with real GNP held constant at its value of 1982:4.

^bCalculated by comparing a simulation with actual output growth with one with “normal” output growth.

^cCalculated using a simulation in which the cost of funds was held at its 1982:3 value.

^dCalculated using a simulation in which the relative prices of investment goods were held at their 1982:3 values.

^eCalculated by comparing a simulation in which none of the changes in tax policy from 1981 onward took place with one in which all the changes took place.

^fActual decline in petroleum drilling investment from 1985:4 to 1987:2.

^gCalculated multiplicatively. A positive residual indicates growth in investment in excess of that explained by the contributing factors.

View Table

Table 6.

Sources of Change in U.S. Business Fixed Investment

(As percentage of investment in the first quarter of 1983)

Year	Actual Charge	Normal Output Recoverya	Above Average Output Recoveryb	Lower Cost of Fundsc	Changes in Relative Pricesd	Tax Policye	Petroleum Drilling Collapsef	Residualg
			Nonresidential fixed investment
1983:4	14.3	8.5	0.8	1.7	0.3	0.6	—	1.8
1984:4	30.1	17.5	5.0	4.2	1.1	1.8	—	-1.7
1985:4	36.2	23.9	1.6	4.7	1.7	2.1	—	-0.5
1987:2	28.2	31.6	-9.0	5.1	2.5	1.4	-3.6	1.7
			Producers’ durable equipment
1983:4	21.3	10.7	1.0	1.9	0.3	0.5	—	5.6
1984:4	39.3	22.3	6.3	4.8	1.3	1.6	—	-0.7
1985:4	49.0	31.2	2.1	5.5	2.3	2.0	—	1.0
1987:2	48.8	41.9	-11.5	6.2	3.6	1.5	—	6.1
			Nonresidential structures
1983:4	2.6	4.5	0.5	1.4	0.2	0.9	—	-4.7
1984:4	14.8	8.7	2.6	3.2	0.6	2.2	—	-3.0
1985:4	14.8	10.5	0.7	3.1	0.7	2.1	—	-2.7
1987:2	-6.1	12.5	-4.3	3.0	0.5	1.3	-9.7	-7.9

^aNormal growth was assumed to be the average of output growth that took place in the economic expansions that began after troughs in real GNP in 1960:4 and 1975:1. The subsequent expansions were the only two of comparable length to the present one. The calculation was made by comparing results from a simulation with normal output growth thus defined with one with real GNP held constant at its value of 1982:4.

^bCalculated by comparing a simulation with actual output growth with one with “normal” output growth.

^cCalculated using a simulation in which the cost of funds was held at its 1982:3 value.

^dCalculated using a simulation in which the relative prices of investment goods were held at their 1982:3 values.

^eCalculated by comparing a simulation in which none of the changes in tax policy from 1981 onward took place with one in which all the changes took place.

^fActual decline in petroleum drilling investment from 1985:4 to 1987:2.

^gCalculated multiplicatively. A positive residual indicates growth in investment in excess of that explained by the contributing factors.

By the second quarter of 1987. with the negative effects of the TRA beginning to feed through, the joint impact of tax policy on both components of business fixed investment was estimated to have been positive by 1 percent—relative to a baseline simulation in which none of the packages was enacted. Alternatively stated, these tax policy changes accounted for 1.4 percentage points of cumulative investment growth from the first quarter of 1983 to the second quarter of 1987.

To assess the long-run implications of the various tax packages, simulations were conducted from 1981 to 1991, with baseline values for the period 1987:3-1991:4 for GNP. interest rates, and inflation in line with the August 1987 version of the Fund’s World Economic Outlook exercise. This approach was made necessary because the full impact of the TRA (which took effect only in late 1986) would not be fully evident in simulations terminating in mid-1987. The results suggested that, by 1991, the effects of the TRA on business fixed investment would dominate those from the ERTA/TEFRA; by the fourth quarter of 1991, total business fixed investment would be 1¾ percent lower than if the tax code had remained entirely as it was before 1981.

Other Simulations

Three other sets of simulations were carried out to indicate the contributions of different explanatory variables to the behavior of business fixed investment in the current expansion. In the first set of simulations, the average cost of funds was maintained at its exceptionally high level of the third quarter of 1982, whereas all other variables—including inflation—followed their actual historical paths,²⁰ As a supplementary exercise, an additional experiment was run in which both the cost of funds and the inflation rate were held at their levels in the third quarter of 1982. In the second simulation, the growth of output was constrained to equal the average growth of output in the only two postwar expansions of comparable length to the present one (see footnote a to Table 6). In the final simulation, the deflators for the two categories of investment goods—which declined relative to the GNP deflator during the recovery—were constrained to increase at the same rate as the GNP deflator during the current expansion.

Summary of Simulation Results

The simulation results can be summarized according to factors affecting the growth of business investment in the period 1981-85 and in the period 1986-87.

Factors Underlying the Growth

of Business Investment, 1981-85

The most important factor accounting for the rapid rise in business fixed investment in the first phase of the expansion was the robust growth of output, which accounted for almost three fourths of the total rise in fixed investment (Table 6). Among the other factors, the large drop in the cost of funds appears to have been the most important. Roughly three fourths of the effects attributed to the decline in the cost of funds stemmed from the reduction in real intertest rates; the remainder reflected positive effects from the simultaneous decline in both the nominal cost of funds and the rate of inflation. When both the inflation rate and the nominal borrowing cost decline, the present value of depreciation allowances is raised, which lowers the cost of capital, while the real after-tax borrowing cost increases because of the tax deductibility of interest payments. In the simulation conducted, the former effect outweighed the latter. The effect of the changing relative price of capital goods had a somewhat smaller impact, as did changes in tax policy. Each of these factors explained about 2 percentage points of the 36 percent growth in business fixed investment in this period.²¹

The simulation results account almost fully for the actual rise in business fixed investment in the period. However, this result reflects in part offsetting errors in the equations for producers’ durable equipment and for nonresidential structures. The inability of each equation to explain fully the recent change in the composition of investment spending may reflect a shift in demand away from more traditional investment goods (such as industrial structures) toward high-technology items (such as electronic computing equipment) in which dramatic technological breakthroughs have occurred in recent years.