# The influence of firms' financial policy on tax reform.

1. IntroductionThe tax system may significantly reduce real capital investment and hence productivity and real incomes. Yet there is no consensus as to the mechanisms involved, as the study of investment behaviour has not yielded clear guidance to policy-makers. The disagreement centres on whether and to what extent taxes on capital income change the cost of finance. While it is agreed that only taxes on returns to the asset that is used as the source of finance at the margin can affect the cost of finance, theoretical and empirical work has failed to identify a unique marginal source of finance. In these circumstances any policy recommendation concerning corporate tax policy may be sensitive to the 'view' adopted by the analyst. We explore the degree of sensitivity in this paper.

Bond, Devereux and Gammie (1996), for example, have recently analysed a specific proposal for reform of the UK system and estimate that in the long-run the capital stock could be increased by around 5% at a cost to tax revenue of [pounds]4 billion (about 2% of general government expenditure) by removing the taxes on retained earnings which produce the bias against investment: they make an explicit assumption that investments are financed by retained earnings at the margin.(1) In the UK, Corporation Tax has been reduced from 45% in 1984 to 31% in 1998, as part of a process of reform in which taxation has been shifted from income (corporate and individual) to expenditure, but the stimulation of investment through further tax reform is a declared objective of government. The US 1986 Tax Reform Act was designed to be broadly revenue-neutral, shifting the burden of tax from individual to corporation. Auerbach and Slemrod (1997) cite estimates of the welfare effects of the Act, ranging from 0.2% of real income to 0.04%, depending on the view taken of the impact of capital taxation. These differences in estimates can arise since, as shown below, one theoretical view predicts that reducing dividend taxation would increase investment, while another - widely held - predicts zero effect on investment and positive capital gains for shareholders.

This paper is concerned with one aspect of this issue: how the financial structure and financial policy of the firm affect the incidence, growth, and distributive effects of tax reform. Financial structure and behaviour vary widely within the G7 group of major economies, to some extent because of national institutional and legal differences, but also possibly in response to tax differences. Browne (1994) evaluates the evidence for and against a variety of explanations for these structural differences. He points out that the usage of financial instruments is correlated across countries with the size of the tax wedge, indicating that financial structure may be endogenous to tax policy. He also points out the contrasts between the USA, the UK, and Canada on the one hand, and Japan, Germany, and France on the other, illustrating the different relationship between the banking system and corporate firms in these countries. In the first group, the more distant relationship suggests higher agency costs of debt, which may help explain their lower debt-equity ratios; Browne (op. cit.) also reviews the evidence that dividend smoothing occurs, consistent with the agency theory.

To represent a span of views, we incorporate three alternative submodels in a numerical dynamic general equilibrium model of the UK: in two submodels the financial structure is fixed, in the third it is endogenous. Four illustrative but simple tax policy experiments are considered to investigate the submodels' quantitative implications for investment and capital stock, for welfare, for equity capital gains, and for tax revenue.

Until recently, models of the linkages between financial behaviour and investment have endogenized either the investment or the financing decisions. Their focus has been on determining financial structure when an investment plan is given, or relating aggregate investment to aggregate market value of the firm when financial structure is indeterminate. These approaches are typical of finance (see, e.g., Bradley et al., 1984, and Titman and Wessels, 1988) and macroeconomics (see, e.g., Abel and Blanchard, 1983), respectively.

Recent researchers (see, e.g., Osterberg, 1989, and Mauer and Triantis, 1994) have developed models of the joint determination of financial structure and investment. Osterberg (1989) adopts the q theory of investment and modifies q to take account of financial structure in a deterministic dynamic general equilibrium model where financial structure affects firm value. The trade-off between agency costs of borrowing and tax advantages of bond finance results in an interior solution for the endogenous debt-equity ratio. Osterberg's model has the desirable feature that, although q is still a 'sufficient statistic' for investment, the endogeneous adjustment of financial structure has real effects. This is our most general model, and we attempt to quantify these real effects.

1.1 Three views of corporate income taxation

The lack of a widely accepted theory of corporate finance and disagreement on the interaction of taxation and corporate financial policy are the main reasons for the conflicting views on the corporate income tax. based on a specific assumption regarding the effects of taxation on corporate financial policy one may distinguish at least three views of the corporate income tax.

There are primarily three views of the impact of the corporate income tax.(2) The traditional (or old) view of dividend taxation implies that dividend taxation at the personal level, when coupled with business taxation at the corporate level, results in double taxation of the income attributable to investments financed at the margin with new share issues. The second, the new (or trapped equity or tax capitalisation) view of corporate taxation, with the basic assumption that retained earnings are the marginal source of finance, argues that the dividend taxation is capitalised in share prices and therefore cannot affect the firm's investment decisions. Finally, the third view of corporate income taxation assumes that new issues of bonds are the marginal source of finance and this in turn implies that the tax advantages of debt finance may remove the distortionary effects of corporate taxation.

Under the old view, taxes on dividend payments will affect corporate investment through changes in the cost of finance, more precisely through changes in the cost of equity. Shareholders are assumed to value dividends more highly than equivalent capital gains on their shares: dividends are seen as a signal of financial health and future earnings prospects of the corporation to shareholders, and the fact that shareholders and managers have conflicting interests results in an enforcement of dividend payments on managers as a monitoring device. In equilibrium, the marginal benefit from extra dividends will be traded off against the extra tax imposed on dividends relative to capital gains. Coupled with a policy of maintaining a fixed debt-capital ratio, a high dividend pay-out ratio makes retentions insufficient to fully fund investment, so that marginal investments are effectively financed by new share issues.

This traditional view implies that the returns to investment financed by share issue and retained earnings are subject to both business and individual level taxes, with the effective individual tax rate equal to a weighted average of the tax rates on dividends and capital gains. This double taxation of dividend income generates potentially important allocative effects. In particular, an increase in the dividend tax rate will raise the effective tax rate on investment income and thus tend to discourage investment.

By contrast, the basic rationale underlying the new view is the assumption that earnings on equity-financed investments can be ultimately distributed to shareholders only in the form of taxable dividends. Without any alternative distributions, such as share repurchases, the assumption implies that equity is 'trapped' within the corporation in the sense that shareholders can receive distributions of returns to their equity investments only by paying an individual level tax on such distributions. As in the old view, the debt-capital ratio is kept fixed by the firm through bond issue; but as no new shares are issued, investment is funded at the margin by retentions.

The allocative effects of the new view are as follows. Under this view, the effective rate of taxation of investment financed with retained earnings is independent of the individual tax rate on dividends. If capital gains are untaxed, or taxed at a lower effective annual rate than are dividends, investment financed with retained earnings is subject to a lower tax burden than is investment financed with new share issues. Thus, dividend taxation does not generate significant effects on investment financed with retained earnings. However, future changes in dividend taxes will be capitalised in share prices, implying windfall gains/losses to existing shareholders.(3)

Finally, as an alternative to both 'old' and 'new', we follow the work of Jensen and Meckling (1976) and Osterberg (1989) in assuming that managers seek to maximize the market value of debt plus equity rather than equity alone. Jensen and Meckling viewed the firm as a 'contracting arena' in which the conflicting interests of bondholders, stockholders, and managers are negotiated. In modelling firms' behaviour based on the work of Jensen and Meckling, Osterberg simplified by considering the conflict between bondholders and stockholders only, using agency costs to model the contractual restrictions intended to control this conflict. Agency costs of debt combine with taxes favouring debt (for instance, corporate income tax allowing interest payments to be deductible from the corporate tax base) to yield an interior solution for the endogenous debt-equity ratio.

1.2 Relationship to other studies and evidence

The dynamic general equilibrium tax model of this paper was developed by Goulder and Summers (1989) and extended by others.(4) A feature of the model is its asset price approach to taxation. The approach synthesizes the q theory of investment in Tobin (1969) and the adjustment cost investment framework developed in Lucas (1967) and Treadway (1969). The use of the q theory of investment links the real sector with the financial sector, and this linkage permits us to estimate the effects of tax on investment by assessing their impact on firms' values and to capture the capitalization effects of tax changes through asset valuations. The model therefore permits us to make assessments of short-run effects of tax policy on asset values as well as long-term impacts on capital accumulation.

Empirical evidence is inconclusive concerning the impact of capital income taxation. Poterba and Summers (1985), Poterba (1987), Nadeau (1988), Gordon and Mackie-Mason (1990), and Bond, Chennells and Devereux (1996) found negative relationships between dividend taxes and payout using either US or UK data. While these findings support the traditional view, other evidence rejects the negative relationship: Bolster and Janjigian (1991) found no support for the tax influence on dividend payout in a study of US companies. Poterba and Summers (1985) also examined the effects of dividend taxes on the level of investment decisions and again concluded that the traditional view explained the interactions of taxes and investment decisions better than the new view. But the findings of Auerbach (1984), Bond and Meghir (1994), and Schiantarelli (1996) can be regarded as being consistent with the new view. Auerbach (1984) observes a high correlation between new share issues and higher earnings; he argues that this implies that new share issues are perceived to be a higher cost source of funds than retained earnings, which contradicts one of the implications of the old view that at the margin firms should be indifferent between new share issues and retentions. Bond and Meghir (1994) and Schiantarelli (1996) find investment sensitive to the availability of internal funds.

1.3 Structure of the paper

The remainder of this paper is organized as follows. Section 2 outlines the main elements of the intertemporal equilibrium model, including its parameter values; Section 3 reports and analyses results from the policy simulations; and the final section provides a brief conclusion.

2. Structure of the dynamic applied general equilibrium tax model

The model incorporates production, household and government sectors, and considers real as well as financial decisions of agents. Physical commodities are of three types: (i) scarce factors, i.e. labour, leisure, and capital stock; (ii) final goods; and (iii) investment goods. Financial assets are of two types: equities and bonds. Table 1 introduces some useful notation and sets out certain key differences in the cost of capital and its components between the models to be developed below.

2.1 The production sector

We assume that managers seek to maximise the value of the firm and that a no-arbitrage equation governs the relationship between returns on debt and returns to equities. The first assumption establishes the basis for both the firm's investment [TABULAR DATA FOR TABLE 1 OMITTED] behaviour and its financial structure. The second defines how the firm's market value is determined by asset holders. In model 1 and model 2, the market value of the firm is the market value of its shares, and in model 3 the market value is the sum of equity and bonds. These models differ, however, in dividend policy and sources of finance. All models share the same constant returns to scale CES production technology, using homogeneous labour and capital to produce value-added.

2.1.1 Model 1: tax capitalisation (the 'new' view) Model 1 assumes no new equity issue to finance investment: the sources of investment finance are retained earnings and borrowings. At the margin the source of finance is retained earnings since firms borrow so as to maintain a constant debt-capital ratio. This implies that dividends are determined as the residual of earnings (net of interest and taxes) plus borrowing minus investment expenditure.

To derive a closed form expression for the market value of the firm in model 1 we invoke the no-arbitrage condition. This states that the expected returns from holding equity (the sum of dividends D and capital gains [Mathematical Expression Omitted] net of tax) must be in line with those from holding bonds (the nominal interest rate [r.sub.B] net of tax). Thus the no-arbitrage condition requires that at any point in time

[Mathematical Expression Omitted] (1)

Here, S is the value of equity, such that [Mathematical Expression Omitted] with [P.sub.E] the share price of equity and [Mathematical Expression Omitted] the number of equity shares.

The differential equation above can be solved to find the time path of S. In order to ensure a unique solution, it is necessary to impose a condition that rules out explosive behaviour. With the transversality condition satisfied and the assumption of perfect foresight about future dividends, the solution to eq. (1) becomes

S(t) = [integral of] [Theta]D(s)[Pi](s, t) ds between limits [infinity] and t (2)

where

[Pi](s, t) = exp [[integral of] -(1 - [[Tau].sub.r])/(1 - [[Tau].sub.g]) [r.sub.B] (u)du between limits s and t].

Dividend payments are residually determined as follows

D = R + BN - IE (3)

Revenues after tax and dividend payments are given by

R = [PY - [P.sub.L]L - [r.sub.B]B](1 - [[Tau].sub.c]) + [[Tau].sub.c]A (4)

Net of tax investment expenditure IE is

IE = (1 - [[Tau].sub.k])[P.sub.K]I (5)

The stock of corporate bonds B and the rate of debt finance BN are

B = K[P.sub.K]K, [Mathematical Expression Omitted]. (6)

Expression (6) states that firms issue debt BN to finance new investment so as to maintain a constant debt-capital ratio. The notation is as follows: Y = output, P = price of output, L = labour demand by the firm, [P.sub.L] = wage rate, I = investment, [P.sub.K] = replacement price of capital goods, A = value of depreciation allowances.

2.1.2 Model 2: the traditional or 'old' view Model 2 allows for new equity issue to finance investment. Therefore sources of investment finance include new equity issue as well as retained earnings and borrowings. At the margin the residual source of finance is new shares. This results from the assumption that firms borrow so as to maintain a constant debt-capital ratio. Firms pay dividends equal to a constant fraction, [Mathematical Expression Omitted], of earnings.

The key equation in this submodel defines dividend payments

[Mathematical Expression Omitted] (7)

With new share issues SN the no-arbitrage condition in (1) becomes

[Mathematical Expression Omitted] (8)

The market value of the firm in expression (2) is accordingly modified as

S(t) = [integral of] [[Theta]D(s) - SN][Pi](s, t) ds between limits [infinity] and t (9)

Substituting eq. (7) into the cash flow identity (3) yields an expression equating funds from new share issue to the cash flow

[Mathematical Expression Omitted] (10)

In words, the new share issue equals investment less new bond sales and less earnings, plus dividends.

2.1.3 Model 3: endogenous financial policy In accordance with the firm's behaviour of maximising the market value of debt plus equity so as to control the conflict between bondholder and stockholders, the cost of debt finance is the sum of interest payments and agency costs. Investment decreases with the debt-equity ratio for tax reasons. We assume, however, that there are bond covenants that are negotiated to restrict the level of debt for a given value of equity. The higher the debt-equity ratio, the more likely the covenant will be violated, resulting in restrictions on investment activities and a decrease in firm value. Thus, the marginal cost of issuing bonds increases with the debt-equity ratio, b. Accordingly, we postulate that the agency cost function is quadratic in the debt-equity ratio and takes the following functional form

[[Phi].sub.g](b) = ([Zeta]/2)[[b - [Gamma]].sup.2]/b. (11)

[Gamma] and [Zeta] are agency cost parameters, and formally b is equal to the ratio of the value of debt stocks (B) to the value of equity (S). Hence, the total cost of debt finance comprises the agency cost of debt [[Phi].sub.g](b)B as well as interest payments to existing bond stocks [r.sub.B]B.

The dividend policy of the firm is represented by a constant fraction, [Mathematical Expression Omitted], of the market value of shares

[Mathematical Expression Omitted] (12)

[Mathematical Expression Omitted] is derived from the no-arbitrage equation to solve for the steady-state dividend-share ratio. Therefore any change in tax parameters will also change [Mathematical Expression Omitted].

Model 3 assumes no new share issues and therefore the sources of investment finance are retained earnings and borrowings. At the margin the source of finance is bond issues since firms distribute a constant share of the market value of the shares. In order to obtain a differential equation for the market value of the firm, we exploit the profit accounting identity which states that earnings before interest payments and taxes [Mathematical Expression Omitted] equal the sum of the total cost of debt finance (interest payments ([r.sub.B]B) plus agency costs of debt [[Phi].sub.g](b)B), tax liabilities (TC = [[Tau].sub.L]L + [[Tau].sub.c][PY - (1 - [[Tau].sub.L])L - [r.sub.B]B] - [r.sub.c]A), retained earnings (RE), and dividends [Mathematical Expression Omitted]. Formally, this profit-accounting identity can be written as

[Mathematical Expression Omitted] (13)

Substitute RE from the investment-finance identity, [Mathematical Expression Omitted], and add [Mathematical Expression Omitted] to both sides of the equation above. Using eqs (12) and (11) one can can then derive the following relation

[Mathematical Expression Omitted] (14)

where NCF is the net cash flow of the firm, defined as

NCF = [PY - [P.sub.L]L](1 - [[Tau].sub.c]) + [[Tau].sub.c]A - IE (15)

Note that the market value of the firm relation, V = B + S, implies

[Mathematical Expression Omitted] (16)

Finally, substituting this last relation into eq. (14), with [Mathematical Expression Omitted], and using the no-arbitrage condition, gives us the following differential equation

[Mathematical Expression Omitted]. (17)

[Gamma] is the cost of finance, the weighted average of cost of debt finance, and cost of equity finance, defined as

[Mathematical Expression Omitted] (18)

Expression (17) is a linear differential equation in V and integrating the expression yields the following expression for V

V(t) = [integral of] NCF(s)[Pi](s, t)ds between limits [infinity] and t (19) (19)

where

[Pi](s, t) = exp [- [integral of] [Gamma](u) du between s and t].

[Gamma] is the discount factor which maintains the equality between the integral of V(t) and B(t) + S(t). The revenue of the firm is therefore the present value of net cash flow, discounted by the cost of finance, which in turn depends on the capital structure.

2.2 Depreciation allowances

We distinguish between tax effects on old and new capital stocks. The two terms reflecting the value of depreciation allowances on existing capital, [A.sup.E], and future acquisitions, [A.sup.N], are

[Mathematical Expression Omitted] (20)

[Mathematical Expression Omitted] (21)

Here, [Mathematical Expression Omitted] is the rate of depreciation allowances applied to investment expenditures, the first year depreciation allowance rate, and [Mathematical Expression Omitted] is the standard rate of depreciation allowances applied to accumulated capital stock.

2.3 Transitional dynamics

We assume that investment generates real adjustment costs, and that output is defined net of these costs. This implies that investment decisions will balance the costs of new capital (acquisition costs plus adjustment costs) against the higher cash flows made possible by a large capital stock. The maximization of the market value of the firm subject to the adjustment technology gives rise to investment functions typical of q-theory (see Tobin, 1969)(5) so that the firms invest when the stock market value of their assets exceeds their replacement cost.

Although q-theory links the real sector with the financial sector as explained above, most q models assume that neither the market value of a firm nor its cost of finance is affected by the decision as to how investment is financed. Debt and equity are perfect substitutes in models 1 and 2. However, Model 3 assumes that managers are forced to maximise the market value of debt plus equity by the existence of financial constraints, and this gives rise to an optimal determination of debt-equity ratios.

Now, financial structure affects q, but only indirectly through the discount rate (cost of finance) varying with the debt-equity ratio, since q is the present discounted value of after-tax marginal products of capital. However, this result is based on two more restrictions: first, that the number of equity shares is assumed fixed, i.e., no new equity issue; second, that the firm pays out a constant fraction of the market value of its shares as dividends.

The firm's problem under each of the three models is to choose the sequence {L(t), I(t)} plus {b(t)} so as to maximise expression (2), (9), or (19), respectively, subject to the real and financial constraints written above. Solving this problem we obtain the following expression for the investment function

I/K = h(Q) (22)

where h([center dot]) = [[[[Phi].sub.d] + (I/K)[[Phi][prime].sub.d]].sup.-1] and [[Phi].sub.d](I/K) is the adjustment cost function. Assuming that the adjustment costs function is quadratic in I/K, the investment function is linear

I/K = [Alpha] + 1/[Beta] Q (23)

For each model we consider, the appropriate tax-adjusted Tobin's q is, respectively

[Mathematical Expression Omitted] (24)

[Mathematical Expression Omitted] (25)

[Q.sub.3] = [V = [A.sup.E]/[P.sub.K]K - 1 + b + [[Tau].sub.k] + [A.sup.N]] [P/(1 - [[Tau].sub.c])[P.sub.K] (26)

where [Mathematical Expression Omitted]. The difference between first two equations is that the tax capitalization hypothesis implies the presence of 1/[Theta] multiplying the market value of equity, correcting for the effects of tax capitalization in affecting the firm's investment decisions. [Theta] is referred to as 'the dividend retention preference ratio' and is defined as the amount of after-tax profits which shareholders receive when a firm uses one pound of after-tax profits to increase its dividend payout (see King, 1977, and Poterba and Summers, 1985). The third equation differs from the first two in allowing for maximising the market value of equity plus bonds instead of maximising only the market value of equity.

2.4 The model of household behaviour

This section of the model is quite conventional. The representative household is treated as forward-looking, having perfect foresight and an infinite time horizon. Thus the determination of current optimal savings and labour supply is a fundamentally intertemporal problem. Formally, at each point in time t the household chooses a path of consumption C and leisure H to maximise

U(t) = [integral of] 1/(1 - (1/[Sigma])) between limits [infinity] and t u[(C(s), H(s))].sup.1/(1 - (1/[Sigma]))] [exp (-[Rho](s - t))]ds (27)

In the above expression, [Rho] is the pure rate of time preference - the rate at which the household discounts future utilities - [Sigma] is the intertemporal elasticity of substitution, and u(., .) is the instantaneous (intratemporal) utility function, defined as

[Mathematical Expression Omitted] (28)

where [[Xi].sub.0] is the share parameter and [[Xi].sub.1] is the elasticity of intratemporal substitution.

The household's behaviour is subject to a dynamic budget constraint

[Mathematical Expression Omitted] (29)

where, since C is a composite good, P is a composite price index inclusive of indirect taxes, namely value-added taxes at the rate [[Tau].sub.v]. In the model the aggregate labour time endowment grows at a constant rate, g, which determines the steady-state growth rate of the economy. This rate reflects both population growth and labour productivity growth.

2.5 Government behaviour

The government's budget constraint is

[summation over i] (1 + [[Tau].sub.vi]) [P.sub.i][G.sub.i] + Tr = T (30)

where [P.sub.i] is the price of good i, [[Tau].sub.vi] denotes the value-added tax rate levied on good i, [G.sub.i] represents the exogenously given government purchase of good i, Tr is transfer payments, and T is tax collections summed over goods of type i. Total lump-sum redistributive transfer payments, i.e. transfers to households are exogenously given as Tr. Transfers formally account for income tax allowances and exemptions, and are permitted to vary to maintain an equal yield equilibrium throughout model simulations. In the absence of government borrowing, (30) is static.

2.6 Summary of submodels

We now have the following models to compare predictions for the effects of selected fiscal policies:

(i) Model 1: eqs (1), (2), (3), (4), (5), (6), and (24).

(ii) Model 2: eqs (4), (5), (6), (7), (8), (9), (10), and (25).

(iii) Model 3: eqs (1), (4), (5), (11), (12), (14), (16), (17), (19), and (26).

3. Results of fiscal policy experiments

This section reports results from simulations of the effects of reductions in four different tax rates: taxes on dividend income, corporate income, interest income and capital gains. We first (i) eliminate the tax on dividend income at the personal level by giving a full credit to taxes paid at the corporate level, i.e. the dividend tax credit rate is equalized with the corporate tax rate. We then (ii) halve the corporate tax rate (a zero rate would not yield an interior solution in model 3). We also (iii) eliminate taxes on capital gains and (iv) on interest income i.e. their rates are set to zero. These revised equilibria are generated based on calibrated values and the parameters reported in Table 2. Among the parameters, we use calibrated values for [Rho] and [[Xi].sub.0] such that they satisfy the balanced growth path relation and capital market equilibrium condition. A value for [Mathematical Expression Omitted] in model 3 is obtained consistent with the no-arbitrage condition. Values for agency cost parameters are guesstimated as [Zeta] = 0.5 and [Gamma] = 0.53 to be consistent with the debt-equity ratio. Production parameters are estimated from national income aggregates, conditional on an elasticity of factor substitution of 0.8. In 1990 there was no special first year tax depreciation rate, so [Mathematical Expression Omitted].

We gauge the allocative and distributional effects of the unanticipated permanent policy changes through their impacts on earnings (net cash-flow), equity value, firm value, capital stock, cost of finance, and the debt-equity ratio. All tax changes are equal-yield, personal tax allowances being adjusted (via Tr in (30)). An important caveat is that we consider only the individual application of each tax policy: if these policies were applied in combination the effects cannot be inferred by adding up the effects of the individual components, since there are many important interactions between these policies.

Table 4 sets out the long run effects on investment I, equity values S, tax revenue, and welfare. Note that the capital stock effect is identical to the I effect in the steady [TABULAR DATA FOR TABLE 2 OMITTED] [TABULAR DATA FOR TABLE 3 OMITTED] state. Our aim here is to examine whether modelled differences in firms' financial policies matter in designing tax reform, and so we set out the growth, welfare, revenue (i.e. revenue loss for the specific tax instrument) and wealth effects of the four tax policies, for each of our three models. We use a revenue cost-effectiveness approach when comparing the relative effects of tax policies across models. Further detail on long-run as well as other effects are shown in Tables 5, 6, and 7 below.

Table 4 Long run effects by policy and model Model 1 Model 2 Model 3 tax capitalisation old view endogenous model (i) elimination of dividend taxation I -0.00 5.20 4.78 S 13.64 2.67 2.65 revenue effect -0.84 -1.10 -1.00 welfare effect 0.00 1.10 1.16 (ii) halving corporate tax I 5.58 6.02 4.98 S 16.37 16.08 13.01 revenue effect -2.00 -2.10 -1.92 welfare effect 1.13 1.24 1.07 (iii) elimination of capital gains tax I 3.83 1.53 1.48 S -4.23 0.66 0.83 revenue effect -0.43 -0.32 -0.32 welfare effect 0.81 0.35 0.46 (iv) elimination of interest income tax I 5.92 5.87 6.71 S 3.07 3.02 2.34 revenue effect -1.25 1.24 -1.43 welfare effect 1.20 1.21 1.73 Key: S = value of equity, I = investment. All figures express percentage changes from the base case. Welfare effects are denoted by change in equivalent variations as percentage of initial total wealth.

3.1 Eliminating dividend taxation

Table 4 shows that with model 1 (the tax capitalization approach), eliminating taxes on dividend income at the personal level creates distributional effects only. Raising the 25% dividend tax credit rate up to the level of corporation tax rate (34%) generates a 13.6% rise in equity value (i.e. share prices) with no real effect whatsoever beyond increasing the net income stream and wealth of shareholders at the expense of non-shareholders. This reflects the fact that the source of finance at the margin is retained earnings and therefore dividend taxation does not enter into the firm's cost of capital in this model. Robust policy implications follow: no proposal for tax integration of corporate and personal income tax could be recommended on welfare grounds, as it would simply amount to an income transfer to shareholders at the expense of taxpayers.

[TABULAR DATA FOR TABLE 5 OMITTED]

In model 2, with new equity issues as the marginal source of finance, the elimination of double taxation of dividend income reduces the cost of new equity finance, and therefore lowers the overall cost of capital through its impact on the cost of finance. This policy change thus generates completely different outcomes from those in model 1: capital stock and the investment rate rise by 5.2%. [TABULAR DATA FOR TABLE 6 OMITTED] The policy also raises share values, with the effect that existing shareholders derive the most benefit from this policy. The more investment is financed by retentions, the greater the benefits to existing shareholders.

In model 3, with endogenous debt-equity ratio and no new share issue, the elimination of double taxation of dividend income brings about a shift from [TABULAR DATA FOR TABLE 7 OMITTED] debt finance to retentions by adjusting the debt-equity ratio. Table 4 reveals that in the long-run, the capital stock rises by 4.8%, less than that obtained in model 2. In this case, because the firm does not take advantage of cheaper new equity finance, the decline in the cost of finance is lower than in model 2. Considering the balance of welfare and revenue effects, model 1 indicates no support for this policy, while results from model 3 seem preferable to those from model 2 with lower revenue cost and greater welfare gain.

3.2 Halving the corporation tax rate

As can be deduced from the cost of capital formulae in Table 1, changes in the corporate income tax rate generate three possible effects: (i) change in the after-tax rate of return on capital; (ii) change in the value of the depreciation allowance on new investment; and (iii) change in the cost of finance. Halving the corporate income tax increases the after-tax rate of return on capital by directly reducing the cost of capital. On the other hand, this policy raises the cost of finance by making bond finance more expensive relative to equity finance, and also reduces the relative importance of capital allowances. Table 4 reveals that the effect on the after-tax rate of return on capital outweighs the other effects on the cost of finance and on capital allowances, resulting in capital accumulation in all three models. However, cutting the corporate tax rate has greater impacts on both real variables (e.g. capital stock) and financial variables (e.g. equity value and firm value) in model 3 than in either of the other two models. This is the result of endogenous adjustment of the debt-equity ratio: the rate cut in corporation tax reduces the tax advantage of debt, reduces the debt-equity ratio and raises the cost of finance. As can be seen in Table 5, the debt-equity ratio declines by 2.7% while the cost of finance rises by 7.7% (of its benchmark value). It is noteworthy that even a small change in the cost of finance or the debt-equity ratio can make an impact on capital accumulation.

This policy has major distributional as well as real effects in all models. Capital gains to existing shareholders result from the policy affecting both old and new capital, and these effects also differ across our submodels: the S effects shown in Table 4 are much greater for this policy than for the others. Finally, looking at welfare gains relative to revenue cost, model 2 predicts both greatest welfare gain and greatest revenue loss; but for both models 2 and 3, this policy yields the poorest returns in terms of welfare gains per unit of revenue loss.

3.3 Eliminating capital gains taxation

Table 4 reveals that such a policy has a significantly greater impact on capital accumulation in model 1, in which marginal investment is financed by retained earnings. With capital gains tax in place, reducing dividend payments to fund an investment saves the tax that would have been paid on the dividend, but incurs a tax liability on the increase in the share value of the firm. So removing the capital gains tax makes retained earnings a more attractive source of finance and this is reflected in a lower discount rate applied to retained earnings.

It is interesting to note that a capital loss arises from this policy according to model 1. The reason should be clear: with equity trapped, firms can divert dividend distributions into retained profits to fund the additional investment predicted by this model, and shareholders are the losers. The other models both predict modest capital gains for existing shareholders with some dividend growth.

In terms of welfare gains and revenue costs, all three models show quite high returns for this policy. Not surprisingly, model 1 shows the highest gain, but it is interesting to note that model 3 shows a higher welfare gain than model 2 in spite of predicting smaller capital stock growth. The explanation lies in the higher value of dividends achieved in model 3, in which firms have a flexible capital structure.

3.4 Eliminating taxation of interest income

The last policy change involves the elimination of taxes on interest income. Since at the margin investment is financed by non-borrowing sources in the first two models, we observe no significant difference between the results obtained in these two models. It is worth noting that although in all models the cut in interest income tax lowers the cost of bond finance, managers in model 3 are also able to change their financial structure by raising the debt-equity ratio by 2.47%. As a result, the cost of finance in model 3 is reduced (by 7.28% of benchmark, see Table 6) through increases in the relative share of bond finance in the total cost of finance. This decline in the cost of finance generates a greater rise in capital accumulation than in models 1 and 2 (6.7% as against c.5.9%). This highlights the importance of endogenous adjustment in financial structure. Notice, however, that there are two distinct mechanisms at work: first, the debt-equity ratio is adjusted as the trade-off between agency and other costs change; and second, taking on more debt to maximise the value of the firm can be achieved by increasing bond-financed investment. The small rise in the value of equity indicates that the latter mechanism is important. We further investigate this issue below under Sensitivity Analysis.

3.5 Sensitivity analysis

We have carried out sensitivity analysis only with respect to the adjustment cost parameter [Beta] and the agency cost parameter [Zeta], since these parameters affect both the short- and long-run results. The long-run results do not change with respect to the elasticity of intertemporal substitution, [Sigma]. Other parameters for which the short- and long-run results might appear to be sensitive are the elasticity of production substitution, [Epsilon] and the elasticity of intratemporal substitution, [[Xi].sub.1]; but these are not in fact sources of intrinsic dynamics in the model.

Varying the adjustment cost parameter, [Beta], with low values half the central case, and high values double, has the following effects (see Tables 5 and 6):

(i) high adjustment cost induces high short-term capital gains, with investors enjoying long-term growth prospects combined with low short-term investment expenditure;

(ii) the short-term differences between high and low adjustment cost cases persist into the steady-state, though the magnitudes of the differences become relatively small;

(iii) the qualitative predictions discussed above in Section 3 remain unchanged across these sensitivity experiments.

Overall, adjustment costs mainly affect the short-run dynamics of the models and hence the distributional impact of policy change.

Varying the agency cost parameter, [Zeta], in the same way (see Table 7) concerns model 3 only. The main effects to be noted are:

(i) high values of [Zeta] reduce the variability of the debt-equity ratio, b; this is as expected since the agency cost function penalises deviations from b;

(ii) more surprisingly, varying the agency cost parameter has very little effect on investment I, i.e., has very little real effect on the economy. This indicates that the significant investment impact in model 3 from reducing the interest income tax is due to bonds being the marginal source of funds and to the maximand being the value of equity plus debt;

(iii) the cost of finance and the redistributive effects of capital gains are slightly more sensitive to [Zeta], but in general the model is insensitive to the agency cost parameter.

4. Conclusion

The main purpose of this paper has been to examine the importance of firms' financial policies for the government's design of tax policies. The results imply that financial structure is important enough that the government must match its tax policies to firms' financial policies. This largely reflects the fact that each financial policy implies a different marginal source of funds and the impacts are determined by the way in which the tax policy experiments interact with the assumptions being made about the marginal sources of finance.

Given uncertainty about the best model, policymakers might be interested in knowing which tax policy is least affected by the choice of model. In all three models the elimination of the interest income tax leads to significant investment and welfare effects with moderate revenue and distribution impacts. Or, if the focus of tax reform was to mitigate the double taxation of capital income one may compare the halving of the corporate rate with the elimination of dividend taxation. We find that the former has the greater effect in promoting investment but also has a higher revenue cost. However, the results of the latter on growth and distribution vary greatly across models.

Perhaps not surprisingly, no single tax policy dominates the others across models. Eliminating the interest income tax and cutting the corporation tax clearly have the greatest impact on investment, regardless of the way in which financial structure is modelled. The welfare effects of these two policies almost dominate the others. Of these two policies, however, cutting the corporate tax has higher revenue losses for all models.

Finally, it is important to note that the tax policy effects reported are not additive and that tax policy properly requires the analysis of combinations of instruments as in Ballard et al. (1985) and Goulder and Summers (1989). The results here emphasise that such analyses should consider the impact of the assumed financial policies of firms for both the short- and long-term impact of tax reforms.

Acknowledgements

We are grateful to Charles Ballard, William Perraudin and two anonymous referees for helpful comments and advice on earlier drafts of this paper: the usual disclaimer applies. Hutton also wishes to acknowledge financial support from the Human Capital and Mobility Programme of the EU (Grant No. ERBCHRX-CT94-0493).

1 The same argument appears to be behind the UK Treasury argument for the 1997 Finance Act reduction in Corporation Tax, designed to raise investment by raising retention. See Treasury (1997).

2 As pointed out by Bond, Chennells, and Devereux (1996) there is a fourth view, the tax irrelevance view. This view predicts that divident taxation should have no impact on dividend payout. The prediction relies on the assumption that the marginal shareholders are tax-exempt.

3 See, Zodrow (1991) for a review of the arguments separating the old and new views.

4 For instance, Kenc (1992) extends these models to capture the endogenous financial adjustments suggested by Osterberg (1989). Navin (1992) also develops a model with endogenous financial behaviour to analyze tax policy issues: this model differs in a number of ways from Osterberg's, for example relying on the existence of bankruptcy costs in obtaining an interior solution.

5 In fact. what Tobin suggested is that the rate of investment is a function of q. However, we can only observe average q, namely the ratio of the market value of existing capital to its replacement cost. Hayashi (1982) showed that if the firm is a price-taker with constant returns to scale in both installation and production, then marginal q is equal to average q. Dixon et al. (1992) point out the existence of another condition: dividends should be a function of capital stock, investment, and a vector of short-run variables and it must be homogeneous of degree one in capital and investment.

References

Abel, A.B. and Blanchard, O.J. (1983). 'An Intertemporal Model of Savings and Investment', Econometrica, 51, 675-92.

Auerbach, A. and Slemrod, J. (1997). 'The Economic Effects of the Tax Reform Act of 1986', Journal of Economic Literature, XXXV, 589-632.

Auerbach, A.J. (1984). 'Taxes, Firm Financial Policy, and the Cost of Capital: an Empirical Analysis', Journal of Public Economics, 23, 27-57.

Ballard, C.L., Fullerton, D., Shoven, J.B. and Whalley, J. (1985). A General Equilibrium Model for Tax Policy Evaluation, University of Chicago Press, Chicago, IL.

Bolster, P.J. and Janjigian, V. (1991). 'Divided Policy and Valuation Effects of the Tax Reform Act of 1986', National Tax Journal, 44(4), 511-18.

Bond, S., Chennells, L. and Devereux, M.P. (1996). 'Taxes and Company Dividends: a Microeconometric Investigation Exploiting Cross-section Variation in Taxes', Economic Journal, 106, 320-33.

Bond, S. and Devereux, M. (1988). 'Financial Volatility, the Stock Market Crash and Corporate Investment', Fiscal Studies, 9, 72-81.

Bond, S., Devereux, M.P. and Gammie, M. J. (1996). 'Tax Reform To Promote Investment', Oxford Review of Economic Policy, 12(2), 109-117.

Bond, S. and Meghir, C. (1994). 'Dynamic Investment Models and the Firms' Financial Policy', Review of Economic Studies, 61(2), 197-222.

Bradley, M., Jarrell, G. and Kim, E. (1984). 'On the Existence of Optimal Capital Structure: Theory and Evidence', Journal of Finance, 34, 857-77.

Browne, F.X. (1994). 'Corporate Finance: Stylized Facts and Tentative Explanations', Applied Economics, 26, 485-508.

Dixon, P.B., Parmenter, B.R., Powell, A. A. and Wilcoxen, P.J. (1992). Notes and Problems in Applied General Equilibrium Economics, North-Holland, Amsterdam, London, and Tokyo.

Gordon, R.H. and Mackie-Mason, J.K. (1990). 'Effects of the Tax Reform Act of 1986 on Corporate Financial Policy and Organisational Form', in J. Slemrod (ed.), Do Taxes Matter? The Impact of the Tax Reform Act of 1986, MIT Press, Cambridge, MA, 91-131.

Goulder, L.H. and Summers, L.H. (1989). 'Tax Policy Asset Prices, and Growth: a general Equilibrium Analysis', Journal of Public Economics, 38, 265-96.

Hayashi, F. (1982). 'Tobin's Marginal and Average q: a Neoclassical Interpretation', Econometrica, 50, 213-24.

Jenkinson, N. (1981). 'Investment, profitability, and the Valuation Ratio', Discussion paper, Bank of England, London.

Jensen, M.C. and Meckling, W.H. (1976). 'Theory of the Firm: Managerial Behaviour, Agency Costs and Ownership Structure', Journal of Financial Economics, 3, 305-60.

Kenc, T. (1992). 'Dynamic General Equilibrium Modelling: a Study of the UK in the 1980s', Ph.D. thesis, University of York.

King, M.A. (1977). Public Policy and the Corporation. Chapman Hall, London.

Lucas, R.E. (1967). 'Adjustment Costs and Theory of Supply', Journal of Political Economy, 75, 321-34.

Mauer, D.C. and Triantis, A.J. (1994). 'Interactions of Corporate Financing and Investment Decisions: a Dynamic Framework', Journal of Finance, 49(4), 1253-79.

Nadeau, S. (1988). 'A Model to Measure the Effects of Taxes on the Real and Financial Decisions of the Firm', National Tax Journal, 41, 467-81.

Navin, J.C. (1992). 'An Assessment of Tax Policies in Models with Endogenous Financial Behavior', Ph.D. thesis, Michigan State University, East Lansing, IL.

OECD (1991). Taxing Profits in a Global Economy: Domestic and International Issues, OECD, Paris.

ONS (1993a). Financial Statistics, Office for National Statistics, London.

ONS (1993b). National Income and Expenditure, Office for National Statistics, London.

Osterberg, W.P. (1989). 'Tobin's q, Investment and the Endogenous Adjustment of Financial Structure', Journal of Public Economics, 40(3), 293-319.

Patterson, K.D. and Pesaran, B., (1992). 'The Intertemporal Elasticity of Substitution in Consumption in the United States and the United Kingdom', Review of Economic Studies, 124, 573-84.

Poterba, J. and Summers, L. (1983). 'Dividend Taxes, Corporate Investment, and Q', Journal of Public Economics, 22, 135-67.

Poterba, J.M. (1987). 'Tax Policy and Corporate Savings', Brookings Paper on Economic Activity, 2, 455-515.

Poterba, J.M. and Summers, L.H. (1985). 'The Economic Effects of Dividend Taxation', in E. Altman, and M. Subrahmanyam (eds), Recent Advance in Corporate Finance, Irwin, Homewood, Il.

Schiantarelli, F. (1996). 'Tax Reform to Promote Investment', Oxford Review of Economic Policy, 12(2), 70-89.

Titman, S. and Wessels, R. (1988). 'The Determinants of Capital Structure Choice', Journal of Finance, 49(4), 1-19.

Tobin, J. (1969). 'A General Equilibrium Approach to Monetary Theory', Journal of Money, Credit and Banking, 1, 15-29.

Treadway, A.B. (1969). 'On Rational Enterpreneurial Behavior and the Demand for Investment', Review of Economic Studies, 36(1), 227-39.

Treasury, HM (1997). The Budget Brief, Her Majesty's Stationery Office, London.

Zodrow, G.R. (1991). 'On the "Traditional" and "New" Views of Dividend Taxation', National Tax Journal, 44(4), 497-511.

Printer friendly Cite/link Email Feedback | |

Author: | Hutton, John P.; Kenc, Turalay |
---|---|

Publication: | Oxford Economic Papers |

Date: | Oct 1, 1998 |

Words: | 8001 |

Previous Article: | Credibility and monetary policy in a model with growth. |

Next Article: | Active labour market policies and job tenure. |

Topics: |