The Effect of Dividend Tax Relief on Investment Incentives

The Effect of Dividend Tax Relief on Investment Incentives

629

National Tax JournalVol. LVI, No. 3September 2003

Abstract - Reducing the double tax on corporate income has anambiguous effect on marginal effective tax rates. It depends on thespecifics of the policy, the extent of debt finance, whether one adoptsthe new or the old view of dividend taxes, the identity of the mar-ginal investor, the importance of international capital flows, andthe replacement tax regime. We illustrate or discuss each of thesesources of ambiguity. We also model the excludable dividendamount (EDA), a feature of the President’s dividend tax proposal,and we calculate the marginal effective tax rate for the 5/15 divi-dend and capital gains tax relief proposal that has become law.

INTRODUCTION

The recent proposal by the Bush Administration to elimi-nate shareholder–level taxes on dividends and retained

earnings revived interest in reducing the double taxation ofincome earned on corporate equity. The Bush Admin-istration’s proposal ultimately led to enactment of a reduc-tion in the tax rate on dividends and on capital gains, the so–called 5–15 proposal.

Reducing the double tax—“integration” in the parlance ofpublic finance economics—has long been advocated as a de-sirable tax reform, but there is significant uncertainty con-cerning the size of the economic benefits associated withlower dividend taxes. Modeling dividend tax changes is com-plex and there is disagreement in the academic literatureabout the size of integration’s likely effect on the incentive toinvest. We illustrate several sources of ambiguity using theHall–Jorgenson user cost of capital—marginal effective taxrate framework. First, we show that integration’s benefits de-pend on the extent to which investment is financed with eq-uity rather than with debt. Second, integration’s benefits de-pend on the extent to which corporate investment is burdenedby dividend taxes, as under the “old view” of dividend taxes,rather than by capital gain taxes, as under the “new or trappedequity view” of dividend taxes. Third, integration’s benefitsdepend on the marginal investor, for example, on whetherthe marginal investor is a taxable individual or instead isappropriately thought of as an average of all investors, in-cluding tax exempt entities such as pension funds. Fourth,integration’s benefits may depend on the extent to which the

The Effect of Dividend Tax Reliefon Investment Incentives

Robert CarrollCouncil of EconomicAdvisers, Washington,DC 20502

Kevin A. HassettAmerican EnterpriseInstitute, Washington,DC 20036

James B. Mackie IIIOffice of Tax Analysis,U.S. Treasury,Washington, DC 20220

NATIONAL TAX JOURNAL

630

economy is open to international capitalflows. Finally, the benefits of integrationmay depend on specific firm–level char-acteristics such as the asset life of a firm’smachines.

The Bush Administration’s original pro-posal for eliminating the double tax ondividends intended to implement theprinciple that all corporate income shouldbe taxed once. The mechanism for ensur-ing that corporate income was taxed oncewas an Excludable Dividend Amount(EDA) that restricted the shareholders’excludable amount to income that hadbeen previously taxed under the corpo-ration income tax at a 35 percent tax rate.Without such a mechanism, a shareholderexclusion of corporate income could re-sult in a zero tax imposed on income thatbenefits from corporate tax preferencessuch as exclusions, deductions, and cred-its.

Previous analyses of the effect of inte-gration on the marginal effective tax rateshave not included the effects of an EDA,even though EDAs were a feature of pastintegration proposals, such as those dis-cussed in U.S. Treasury (1992). In this pa-per, we modify the traditional user costframework in order to model the EDA. Itturns out that the EDA can have surpris-ing effects on investment incentives, com-pared to shareholder exclusions that donot incorporate an EDA. As expected, inmany cases the EDA raises the tax cost ofcorporate investment since the benefits ofdividend tax reduction can be offset by areduction in the value of corporate–leveltax shields. In other cases, however, theEDA increases marginal investment incen-tives, even though its intention is to limittax benefits. In addition, the EDA may notpromote tax neutrality to as great an ex-tent as full exclusion without EDAs,which provides more aggregate tax reliefto the corporate sector. However, an EDAcosts less than a straight exclusion, thusbiasing the comparison somewhat in fa-vor of an exclusion without an EDA. Our

analysis of EDAs is supported by the rig-orous dynamic optimization analysis ofEDAs given in Auerbach and Hassett(2003b) and builds on and complementsthe analyses of Gale and Orszag (2003)and Esenwein and Gravelle (2003).

THE DOUBLE TAX ON CORPORATEPROFITS: AN OVERVIEW OF ITSEFFECTS

Current law “double taxes” corporateprofits (U.S. Treasury, 1992). This is seenmost easily if we assume that the tax codemeasures and taxes economic income sothat statutory tax rates equal marginal ef-fective tax rates. In this case, corporateincome is taxed once under the corpora-tion income tax at rate u. It is taxed againunder the individual income tax whendistributed as a dividend or realized as acapital gain upon sale of shares, say at rateTe. The marginal effective total tax rate onincome from an equity financed invest-ment then is u + (1 – u)Te, reflecting thesum of the corporate tax rate and theshareholder tax rate. Double taxation po-tentially distorts a number of economicchoices.

The double tax adds to the overall taxburden on a typical investment in the U.S.economy, and so may discourage savingand investing in the aggregate. This po-tentially reduces capital formation andsaving and slows economic growth.

Because of double taxation, corporateequity financed investments typically aretaxed more heavily than similar invest-ments undertaken by pass–through enti-ties such as S corporations, partnershipsor sole proprietorships. In addition, thedouble tax adds to the tax burden on busi-ness investment relative to essentiallyuntaxed owner–occupied housing. Con-sequently, double taxation inefficientlydiscourages the use of the corporate formof organization or investment in corpora-tions, as well as investment in businessesas opposed to owner–occupied housing.


631

It thus contributes to an unproductive useor allocation of our nation’s stock of realcapital.

The double tax on corporate equity alsomay create financial distortions. It mightdiscourage corporations from financingwith equity in favor of using debt, onwhich the company may deduct interestpayments, to finance their activities. Thismay make corporate capital structurestoo rigid and too vulnerable to bank-ruptcy and financial distress. In addition,by distinguishing between dividends,taxed as ordinary income, and retainedearnings and share repurchases, taxed ascapital gains, the personal level tax oncorporate earnings may discourage com-panies from paying dividends. Undersome theories of the firms, this may im-pose additional costs on investors, whoreceive a smaller fraction of their earn-ings as dividends than, but for taxes, theywould prefer.

The effect of double taxation on debt–equity ratios has been recognized to be animportant theoretical concern for decades.Remarking on the theory in their famoustextbook, Atkinson and Stiglitz comparea classical system like our own to one thatintegrates corporate and personal taxes(an “imputation” system). They remarkthat “the switch from a classical systemto imputation may make a substantial dif-ference” and equity finance may be muchmore likely (Atkinson and Stiglitz, 1980,p. 141). Early empirical work failed to finda significant effect of marginal tax rateson finance, but recent studies have beenmore successful in finding a link. In a re-cent review article, more recent studieshave found that higher marginal tax ratestend to increase debt levels—the effectpredicted by theory (Graham, 2003).

Studies of taxes and dividend payouthave found that there is a significant link.Poterba (1987) found that payout tends torespond sharply to swings in marginal taxrates. To the extent that firms have an in-centive to allow cash to pile up within the

firm, or to undertake investment projectswith negative net present values, theseincentives may exacerbate problems ofasymmetric information and separation ofownership from control.

As these issues are quite difficult toquantify, we focus henceforth on the link-age between dividend taxation and realinvestment. Recent empirical studies havegenerally found a significant link betweenthe user cost of capital and real invest-ment, thus dividend tax policy can havea significant impact on the economy if itaffects the user cost of capital in a mate-rial manner. This paper discusses the ef-fect of integration on the incentives thatguide investors in making real investmentdecisions using a cost of capital/effectivetax rate framework.

THE TRADITIONAL COST OFCAPITAL/MARGINAL EFFECTIVETAX RATE MODEL

The Conceptual Framework

We analyze the investment incentiveeffects of integration using a model of in-vestment incentives closely related to thecost of capital approach associated withMervyn King, Don Fullerton, and theircolleagues (e.g., King and Fullerton, 1984;Fullerton, 1987; Mackie, 2002). The con-ceptual framework is that of the neoclas-sical theory of investment pioneered byDale Jorgenson (1963) and Robert Halland Dale Jorgenson (1967).

According to the neoclassical theory ofinvestment, the firm will continue to in-vest until, at the margin, the after–tax cashflow from the last dollar invested equals$1. So, in equilibrium

[1] 1 – k = ∫(1 – u)ce–(rc–π+δ)tdt + u(1 – k)z,

where k is the investment tax credit rate,u is the statutory corporate income taxrate, c is the asset’s pre–tax rental rate, δis the economic depreciation rate, rc is the


632

firm’s nominal after–tax discount rate, πis the inflation rate and z is the presentvalue of tax depreciation allowances on aone dollar investment.

Integrating to determine c, and thensubtracting δ gives the cost of capital, thepre–tax after–depreciation rate of returnrequired to cover the investment’s tax costand the real after–tax opportunity cost offunds. If we denote by ρc the cost of capi-tal for a corporate investment, then

[2] ρc = c – δ = (1 – k){rc – π + δ}(1 – uz)

/(1 – u) – δ.

The cost of capital is the real rate of re-turn, net of depreciation, that is just suffi-cient to cover the investment’s tax cost andits real opportunity cost of funds, rc – π.

Shareholder Taxes: The Debatebetween the Old and the NewViews of Dividend Taxes

Investor level taxes enter through thediscount rate, rc. If we assume that share-holders require an after–all–tax real rateof return of s, then the discount rate foran equity financed investment is

[3] rc = re = (s + π)/(1 – Te).

Determining the appropriate value forTe reflects the view one takes on the de-bate between the new view and the oldview of dividend taxes (Auerbach, 2001;Auerbach and Hassett, 2003a; Bradford,1981; King, 1977; Poterba and Summers,1985; and Zodrow, 1991). Each alternativeis based on a theory designed to explainwhy a firm would pay dividends even

though they are taxed more heavily thanare capital gains on reinvested earningsor share repurchases.

The old view holds that dividends of-fer a non–tax benefit that offsets their taxdisadvantage. Corporations set dividendpayments so that, for the last dollar ofdividends paid, the extra non–tax benefitsof dividends equal their extra tax cost.Under this theory, marginal investment isfinanced by new share issues, implyingthat the value of a marginal dollar in thecompany is one dollar, q = 1. The reason qis always one is that this simple model hasno adjustment costs and investment re-sponds quickly to dividend tax payments.For example, if dividend taxes are low-ered, the marginal after–tax product ofcapital increases instantaneously, therebypushing q above 1. Investment then oc-curs driving q back to one. The dividendtax raises the effective tax rate to the ex-tent that firms payout current earnings asdividends, while capital gains taxes raisethe effective tax rate to the extent thatfirms retain and reinvest current earningsor distribute earnings to stockholders viashare repurchases. Under the old view, Te= Pm + (1 – P)ω, where P is the dividendpayout ratio, m is the dividend tax rateand ω is the effective accrual tax rate oncapital gains, which reduces the statutorytax rate on gain to account for deferral andsometimes for the tax free–step–up in ba-sis at death (King and Fullerton, 1984).1

Under the new view of dividend taxes,dividends offer no non–tax benefits, butare assumed to be the only means of dis-tributing funds to shareholders. Becausedividend taxes must be paid on corporatedistributions, they reduce the marginalvalue of corporate shares, and for this rea-

1 Our approach to including shareholder level taxes differs slightly from that in the King–Fullerton model. TheKing–Fullerton model allows for differences between the capital gains tax rate and the dividends tax rate bycalculating a nominal discount rate appropriate for each, and then taking a weighted average of the two as thediscount rate for corporate equity. In contrast, we weight the tax rates on gains and dividends in calculatinga weighted average tax rate on equity, as in Auerbach (1983a and 2001), Gravelle (1994); and U.S. Treasury(1992). We adopt this approach because it seems slightly more consistent with the theoretical development ofthe old view of dividend taxes.


633

son the new view is sometimes called thetax capitalization view. Specifically, if thefirm is paying dividends, the investor isindifferent between receiving a dividendwith an after–tax value of (1 – m) and hav-ing the firm reinvest the earnings withinthe company, thereby raising share valueby q, which generates an after–tax capitalgain of q(1 – ω). So, the increase in thevalue of the company when it retains adollar is q = (1 – m)/(1 – ω) < 1.

Under the new view, because of tax sav-ings for shareholders, firms would gen-erally prefer to finance investment out ofretained earnings rather than by issuingnew shares. For investment financed byretained earnings, the dividend tax has noeffect on the incentive to invest, since itreduces proportionately the investment’safter–tax cost and its after–tax return, leav-ing the rate of return unaffected. In con-trast, the capital gains tax on share appre-ciation acts as a deterrent to investmentfinanced through retained earnings. Un-der the new view, Te = ω.

The empirical support for either viewis mixed. Numerous studies show a sta-tistical relationship between the dividendpayout ratio and the tax penalty on divi-dends relative to capital gains, seemingto offer support for the old view. How-ever, the theory of the new view only pre-dicts that dividend taxes are irrelevantwhen they are constant through time.Since variation of tax rates over time isrequired if one is to estimate an equation,one cannot form a sharp conclusion aboutthe relevant merit of either view from thisevidence. Firms rarely issue new shares,offering some support for the new view.However, marginal investment might stillbe financed by new shares even if shareissues occur infrequently. Some papershave found evidence that some firms maybe on the new share issue margin, whileothers may be on the retained earningsmargin, suggesting that each theory maybe appropriate for some firms at sometimes.

One piece of evidence against the newview, that share repurchases occur fre-quently, while the new view assumes thatfirms must distribute via dividends, iswidely interpreted as conclusive proofthat the new view is flawed. It is not clear,however, that the ability of firms to dis-tribute via share repurchases invalidatesthe new view’s implication that the divi-dend tax does not affect the cost of capital(Auerbach, 1989a and 2001; Auerbach andHassett, 2003a). If earnings are distributedvia share repurchase, then the dividendtax would seem to have no potential toaffect investment incentives because divi-dends are not paid. Furthermore, undernew view logic, the tax on distributions,which in this case would be a capital gainstax on repurchases, would be irrelevant,since it affects proportionately both aninvestment’s cost and its return, leavingthe rate of return unaffected. The capitalgains tax on share appreciation wouldcontinue to burden the investment.

Following the U.S. Treasury (1992), weadopt the old view in most calculations,but also test the sensitivity of our resultsto this assumption by performing an al-ternative set of calculations under the newview. Auerbach and Hassett (2003a) sug-gest that about half of firms in the U.S.likely fall under each view, so analyzingthe change in investment incentives un-der each view is important for under-standing the impact of any policy.

Debt Financed Investment andTaxes on Lenders

If a bondholder is taxed at rate θ, thenthe firm’s discount rate for a debt financedinvestment is

[4] rc = rd = i(1 – u) = (s + π)(1 – u)/(1 – θ)

which accounts for the firm’s ability todeduct the return paid to bondholders,i.e., interest, and for tax the lender payson interest income.


634

A Weighted Average Discount Rate

For mixed debt and equity financing,the firm’s discount rate, r, is a weightedaverage of re and rd .

[5] rc = we re + wd rd .

The appropriate weights, we and wd arebased on the assumed share of the invest-ment that is financed by debt. Under thenew view, however, the weights are ad-justed to reflect the undervaluation of cor-porate equity (Auerbach, 1983a).2

The Marginal Effective Tax Rate

The marginal effective tax rate is theproportion of the investment’s pre–tax re-turn that is needed to cover the tax cost.The marginal effective total tax rate in-cludes taxes at the company level and atthe investor level and is computed as

[6] METTR = (ρc – s)/ρc.

The METTR can be interpreted as thehypothetical tax rate that, if applied toeconomic income, would have the sameincentive effects as those implied by theactual tax system.3

The Parameter Assumptions

In the calculations presented below, theafter–tax required rate of return, s, is set to4 percent,4 and the inflation rate to 3 per-cent.5 Statutory tax rates reflect federal in-come taxes only.6 In most calculations, thetax rates are a weighted average across allinvestors,7 including both taxable investorsand tax exempt investors, but sensitivityanalysis is performed using a higher set oftax rates. In most calculations, financing isassumed to be 35 percent and 65 percentequity,8 but sensitivity results are presentedbased on a higher leverage ratio. In calcu-lations based on the old view of dividendtaxes, we assume that companies payout50 percent of their earnings as dividends.9

2 The market value of equity is dividend by its price, q, in order to arrive at the amount of new capital that isequity financed: wd = B/(B + E/q), where B is the bonds issued to financed the marginal investment and E isthe market value of retained earnings used to finance the marginal investment.

3 We adopt a similar approach in measuring the cost of capital and marginal effective tax rate for an investmentin unincorporated business and in owner occupied housing. See, e.g., Fullerton (1987) and Mackie (2002).

4 It is common to assume a value of s in this range. See, e.g., Auerbach (1983b and 1996), and King and Fullerton(1984).

5 This rate of inflation is somewhat higher than that suggested by very recent historical experience, but appearsreasonable when compared against historical experience over a longer time frame. It also is consistent withinflation rates that have been assumed in other fairly recent calculations of effective tax rates, e.g., Auerbach(1996).

6 The capital gains tax rate is an effective accrual equivalent tax rate, computed by reducing the statutory rateby 1/2

to reflect the benefits of deferral.7 The tax rates are constructed based on the methodology in King and Fullerton (1984), beginning with statu-

tory tax rates on individuals calculated from the Treasury’s Individual Tax Model and reflect fully phased inEGTRRA. The corporate tax rate is 35 percent, the tax rate on dividends is 17.1 percent, the effective accrualtax rate on capital gains is 6.6 percent, the tax rate on interest income is 18.8 percent, the tax rate on homeownersis 23.3 percent, the tax rate on noncorporate business income is 27.2 percent, and itemizing homeownersdeduct interest at a 23.3 percent tax rate, but only 60 percent of homeowners itemize. Tax rates on dividends,interest, and capital gains are weighted averages across taxable and tax exempt investors, using ownershipweights from the Flow of Funds Accounts.

8 This assumption is consistent with recent historical data in the Flow of Funds Accounts, and also with aleverage ratio in the range of 30 percent – 40 percent that is typical in effective tax rate calculations (Auerbach,1996; Fullerton, 1987; Gravelle, 1994).

9 Recent data from the National Income and Products Accounts (Table 1.16) suggest that once S corporationsare removed, dividends account for about 50 percent of corporate after–tax profits (with the IVA and CCA),while dividends account for about 60 percent of after–tax profits neglecting the IVA and CCA. Recent data oncommon stock yields of S&P 500 companies (Economic Report of the President, 2002, Table B–95), suggestthat dividends have accounted for between 32 and 71 percent of earnings after taxes between 1990 and 2000,


635

In our calculations, we do not alter fin-ancial policy in response to changes intaxes.

We note in passing that there is littleon which to base these assumptions in theexisting literature. We really don’t knowhow marginal investments are financed,nor do we know the tax status of marginalinvestors, if there is such a thing. None-theless, the particular assumption chosencan have a large effect on the results. Someof these issues we deal with through sen-sitivity calculations, but others we do notaddress. For example, even seemingly in-consequential assumptions, such as usingan average tax rate to compute the usercost, rather than computing an averageuser cost across different potential taxrates, can affect the results.

THE INVESTMENT INCENTIVEBENEFITS OF SHAREHOLDEREXCLUSIONS

The integration plan proposed by theBush administration is a modified versionof a shareholder exclusion applied to bothdividends and future capital gains. Toanalyze the Bush plan, it appears sensibleto move by steps from current law to thefull plan. In this section, we begin this bypresenting calculations that proceed fromcurrent law to a 50 percent dividend ex-clusion, and then to a 100 percent exclu-sion of both dividends and capital gainson corporate stock, which would extendthe benefits of integration to corporateearnings that are paid out as dividends as

well as to those that are retained and rein-vested within the company. Two variationsof the capital gains exclusion are consid-ered: an explicit exemption and exclusionfor dividend reinvestment plans (DRIPs).After considering these shareholder exclu-sions, the President’s proposal, which asdescribed earlier limits tax relief to theEDA, is examined in the following section.The paper also includes calculations forthe 5/15 plan that was enacted during thecompletion of this paper.

Modeling Shareholder Exclusions

Shareholder exclusions are modeled byadjusting the shareholder’s tax rate onequity financed corporate investment.With a 50 percent dividend exclusion, Te= P(1/2)m + (1 – P)ω, while with an exclu-sion of both dividends and capital gainson corporate stock, Te = 0.

Determinants of the Incentive Benefitsof Shareholder Exclusions

The old–view calculations in Table 1show that under current law, corporateinvestment faces a substantially highereffective tax rate, 33.5 percent, than doesinvestment in the noncorporate sector, 20.0percent, and in owner–occupied housing,3.5 percent.10 The overall economy wideeffective tax rate is 19.1 percent.

Integration in the form of either a 50percent dividend exclusion or a 100 per-cent exclusion of both dividends and capi-

and averaged about 50 percent over the period. (If we add inflation on the equity value of the company toearnings, then dividends would account for between 19 percent and 45 percent of nominal earnings over thisperiod, and would average 33 percent.) Our assumption is consistent with Fullerton, Gillette, and Mackie(1987) who assumed a 50 percent payout ratio in calculating investment incentives under the old view. It islarger than, but in the same ballpark as, the 43 percent nominal payout ratio assumed in Treasury’s 1992Integration Report and the 42 percent nominal payout ratio that would be consistent with Gravelle’s (1994)assumption of a 67 percent payout of real earnings (re – π). It is substantially larger than the 30 percent nomi-nal payout ratio assumed in Auerbach (1996).

10 Owner–occupied housing has a positive, rather than zero, effective tax rate because the calculations assumethat only 60 percent of investment in housing is by taxpayers who itemize, and so can deduct interest. Becauseof this, the weighted average tax rate at which homeowners deduct interest is less than the tax rate on interestincome.


636

Alt

erna

tive

Mod

elin

g of

Infl

atio

n,O

ld V

iew

Old

Vie

w o

f Div

iden

d T

axes

Hig

h D

ebt,

Old

Vie

wN

ew V

iew

of

Div

iden

d T

axes

Hig

her

Stat

utor

y Ta

xR

ates

, Old

Vie

w

Cur

rent

Law

50%

Div

iden

dE

xclu

sion

100%

Div

iden

dan

dC

apit

alG

ains

Exc

lusi

on

Cor

pora

te S

ecto

rN

onco

rpor

ate

Bus

ines

s Se

ctor

Ow

ner–

Occ

upie

d H

ousi

ngE

cono

my

wid

e

Stan

dar

d d

evia

tion

in th

eco

st o

f cap

ital

33.5

%20

.0%

3.5%

19.1

%

0.00

869

29.6

%20

.0%

3.5%

17.4

%

0.00

741

22.5

%20

.0%

3.5%

14.5

%

0.00

563

20.6

%15

.2%

6.3%

13.5

%

0.00

447

12.8

%15

.2%

6.3%

10.7

%

0.00

339

29.6

%20

.0%

3.5%

17.4

%

0.00

739

22.5

%20

.0%

3.5%

14.5

%

0.00

563

40.1

%22

.6%

6.9%

24.1

%

0.01

097

24.8

%22

.6%

6.9%

17.2

%

0.00

571

25.1

%20

.0%

3.5%

15.5

%

0.00

619

Cur

rent

Law

100%

Div

iden

dan

dC

apit

alG

ains

Exc

lusi

onC

urre

ntL

aw

100%

Div

iden

dan

dC

apit

alG

ains

Exc

lusi

onC

urre

ntL

aw

100%

Div

iden

dan

dC

apit

alG

ains

Exc

lusi

on

100%

Div

iden

dE

xclu

sion

and

DR

IP

Mar

gina

l Eff

ecti

ve T

otal

Tax

Rat

e

TA

BL

E 1

SHA

RE

HO

LD

ER

TA

X R

EL

IEF

AN

D IN

VE

STM

EN

T IN

CE

NT

IVE

S: S

EN

SIT

IVIT

Y T

O K

EY

PA

RA

ME

TE

R A

SSU

MPT

ION

S


637

tal gains would potentially offer substan-tially improved investment incentives.11

Under the old view, the proposals wouldreduce the tax burden on corporate invest-ment which would translate into a reduc-tion in the economy wide marginal effec-tive total tax rate of 9 percent and 24 per-cent, respectively. The lower corporate taxcost also levels the playing field by reduc-ing differences in the taxation of invest-ment across sectors (i.e., the corporatebusiness sector, the noncorporate businesssector, and owner–occupied housing).This improved tax neutrality is reflectedin the calculations by a reduction in thestandard deviation in the cost of capital,12

compared to current law.13 Not surpris-ingly, 100 percent exclusion of both divi-dends and capital gains does substantiallymore to reduce effective tax rates and im-prove tax neutrality than does a 50 per-cent exclusion of dividends.

Debt vs. Equity Financing

The investment incentive benefits ofintegration are sensitive to underlyingparameter assumptions. One importantassumption is the extent to which corpo-rate investment is financed with debtrather than with equity. With a high debt/asset ratio, the ability of the company todeduct interest means that a large fractionof corporate investment escapes currentlaw’s double tax on corporate equity, im-plying that current law does little to dis-courage corporate investment (Stiglitz,1973), although the double tax itself may

encourage a high degree of leverage ifborrowing decisions are affected by taxes.Consequently, because the existing distor-tion of the double tax is smaller, integra-tion has less potential for improving taxneutrality when corporate investment islargely debt financed than it does whencorporate investment is largely equity fi-nanced.

These effects are illustrated in Table 1in the calculations labeled high debt,which modify the old view calculationsby raising the leverage ratio from 35 per-cent to 65 percent. With high debt, a 100percent exclusion would lead to a smallerreduction in the effective tax rate for in-vestment in the corporate sector and to asmaller reduction in the standard devia-tion in the cost of capital, compared to theold view calculations.

Perhaps surprisingly, in the high debtcalculations, 100 percent exclusion re-verses the existing tax bias against corpo-rate investment; the effective tax rate oncorporate investment falls below the ef-fective tax rate on noncorporate businessinvestment. This reversal occurs in partbecause corporations receive a subsidy ondebt financed investment that is muchlarger than that granted to investment bynoncorporate businesses. The subsidy ondebt financed investment arises becausecorporations deduct interest at a tax ratesubstantially higher than the tax rate gen-erally imposed on interest income (see,e.g., Mackie, 2002). One aspect of this sub-sidy arises because interest is not indexed

11 The calculations in the table are capital stock weighted averages across multiple types of equipment andstructures, as described in Mackie (2002). One limitation of the stocks used in Mackie (2002) is that they arebased on NIPA data and so include in the corporate sector S corporations that are not subject to double taxa-tion. The capital stocks used here, however, have to be adjusted to move assets owned by S corporations intothe noncorporate sector, based on tax return data on asset holdings of S corporations by industry.

12 The standard deviation is the square root of the (capital stock weighted) sum across all investments of thesquared deviation of each investment’s cost of capital from the mean cost of capital. If all investments weretaxed equally, the standard deviation would be zero because all investments would have a cost of capitalequal to the mean. Differential taxation creates differences in the cost of capital across investments (i.e.,nonneutralities) and raises the standard deviation above zero. The greater are the tax differences across in-vestments, the larger is the standard deviation in the cost of capital.

13 The standard deviation is not a substitute for an explicit welfare cost calculation that includes both changes intax costs and elasticities that reflect how responsive investors are to changes in tax costs. The standard devia-tion is meant to be suggestive of the relative tax neutrality of current law and each proposal.


638

for inflation. The corporation can deductthe inflation component of the nominalinterest rate, an amount that correspondsto a repayment of principal rather thanincome, at a higher tax rate than that im-posed on interest recipients. 14 In addition,the tax rate differential leads to a tax sub-sidy for investments entitled to such pref-erences as accelerated depreciation. Thecorporation takes tax deductions for suchpreferences at a tax rate higher than thatimposed on the income that the invest-ment earns.

Old View vs. New View

Another important assumption is thatthe old view of dividend taxes informs thecalculations. Because the capital gains taxrate is smaller than the dividend tax rate,the tax cost of corporate equity financedinvestment under current law is smallerunder the new view than it is under theold view of dividend taxes. In addition,under the new view integration’s reduc-tion in the tax rate on dividends does notreduce the tax cost of corporate invest-ment. Consequently, to the extent that onefavors the new view, the benefits of share-holder tax relief would be smaller thanunder the old view. Nonetheless, the cal-culations in Table 1 suggest that the in-vestment incentive effects may continueto be substantial as long as tax relief isprovided to retained earnings as well asdividends. Even under the new view,shareholder tax relief would reduce thecorporate marginal effective total tax rateby over 7 percentage points. In evaluat-ing integration, however, it is importantto recall that under the new view a sub-stantial part of the revenue cost of inte-

gration goes to providing a windfallbenefit to shareholders, in the form of anincrease in share values caused by elimi-nating previously capitalized dividendtaxes, rather than to reducing the marginaltax cost of investment in the corporatesector. Without shareholder level taxes, qwould rise to one. Even under the oldview, windfall benefits to shareholderscan occur in the short run if the stock ofcorporate capital cannot expand immedi-ately to its new equilibrium level. The taxreductions that generate these windfallsraise the revenue cost of providing relieffrom double–taxation.15

High Tax Rate Marginal Investor

The benefits of integration also dependon the statutory tax rates on capital incomethat underlie the calculation of the effec-tive tax rates. Statutory tax rates in turnreflect assumptions about the marginal in-vestor, a subject on which there is littleagreement in the economics profession(Auerbach, 2001). The tax rates used in theprevious calculations are economy wideaverages, reflecting the marginal rate ontaxable investors as well as a zero rate fortax exempts. In such calculations, all cur-rent investors are marginal investors, andare included with weights reflecting theirownership of the existing capital stock.

A competing view is that high tax costinvestors are the marginal investors. Inaddition to theoretical arguments that thehigh cost investor might be the last in,there is a body of empirical work suggest-ing that the marginal corporate investormight face a high tax rate (e.g., Harris andKemsley, 1999; Gentry, Kemsley, andMayer, 2003). This view is reflected in

14 To illustrate this subsidy, consider a debt financed investment that receives economic depreciation, so the z =δ/(r – π + δ). In this case, the cost of capital is (r – π)/(1 – u), where r = i(1 – u). If interest income is taxed at rateθ, lender must charge an interest rate of i = (s + π)/(1 – θ) in order to earn his required after–tax rate of return.Hence, the cost of capital is s/(1 – θ) – π(u – θ)/((1 – θ)(1 – u), which implies a subsidy for debt when u > θ. Thecost of capital is less than s/(1 – θ), so the marginal effective tax rate is less than θ.

15 Concentrating tax relief on new equity financed investment would eliminate these windfalls and lower therevenue cost of integration, but in practice this can be difficult to do and offers uncertain net benefits (Auerbach,1989b).


639

Table 1 by calculations that exclude taxexempts and so include only taxable in-dividual investors in the construction ofthe statutory tax rates.16

At the higher statutory tax rates, cur-rent law’s tax penalty on corporate invest-ment is considerably greater than it is inthe earlier calculations. Moreover, thehigher tax penalty is caused by highershareholder taxes. Consequently, by elimi-nating shareholder level taxes, integrationdoes much more to reduce the corporateeffective tax rate and to improve tax neu-trality than it does when lower tax ratesunderlie the calculations.

Exclusion via a DRIP andthe Taxation of Inflation

Rather than eliminating or reducing thecapital gains tax on the sale of corporateshares, the benefits of integration could beextended to retained earnings by allowing(or requiring) companies to have dividendreinvestment plans (DRIPs), or in someother way to adjust the basis of shares inorder to eliminate the capital gains tax onshare price rises caused by retained earn-ings.17 DRIPs were discussed in U.S. Trea-sury (1992) and a DRIP variant is a featureof the Bush Administration’s dividend taxrelief proposal.

It is unclear how to account for a DRIPin the model outlined above. Followingthe basic King–Fulleton approach,adopted by Auerbach (1996), the firmwould split the nominal equity return (thereal return plus inflation) between divi-

dends and retentions. In such a model, aDRIP might eliminate the tax on the partof the nominal rate of return that is re-tained and reinvested in the company (or,formally, distributed as a dividend andthen reinvested), in the same way that adividend exclusion eliminates the tax onthe part of the nominal return that is dis-tributed as a cash dividend. Yet this treat-ment would seem to combine a DRIP withinflation indexing of capital gains on stockand would not distinguish between aDRIP and a capital gains exclusion.

An alternative modeling strategy as-sumes the inflation component of thenominal return accrues as an increase inshare values. In this approach, the firm di-vides real earnings, earnings (re – π), be-tween dividends and retentions and in-flation is taxed to the shareholder as acapital gain on the appreciation of hisshares. This alternative approach is devel-oped and used in Auerbach (1983a and b),Gravelle (1994), and U.S. Treasury (1992).It allows one to distinguish between aDRIP, capital gains indexing, and a capi-tal gains exclusion while avoiding theseeming inconsistency between the taxa-tion of dividends that are distributed ascash and those that are retained and rein-vested. It also seems consistent with theassumption of the cost of capital modeloutlined above that the firm holds the as-set forever.

The final column of Table 1 presents thecalculations based on this alternativemodeling of the taxation of the inflationreturn to equity18 for a combination of a

16 In these calculations, the corporate tax rate is 35 percent, the tax rate on dividends is 26.4 percent, the effectiveaccrual tax rate on capital gains is 9.6 percent, the tax rate on interest income is 23.2 percent, the tax rate onhomeowners is 23.3 percent, and the tax rate on noncorporate business income is 27.2 percent.

17 A dividend tax relief proposal would have to explicitly disallow DRIPs if such relief were not intended, sinceshareholders may voluntarily participate in DRIPs under current law. There may be little tax incentive toparticipate in voluntary DRIPs under a partial (e.g., 50 percent) dividend exclusion, and our calculations fora 50 percent exclusion do not include an effect from DRIPs.

18 Given our assumed tax rates, after–tax real return, and inflation rate, our 50 percent nominal dividend payoutratio implies that about 80 percent of the real return on corporate equity (re – π) is paid out as a dividend,which we use as the real payout ratio in these calculations. This payout ratio is higher than the historicalaverage real dividend payout ratio of about 2/3s reported by Gravelle (1994), but is closer to the 72 percentreal payout ratio assumed in the Treasury’s 1992 Integration Report.


640

dividend exclusion plus a mandatoryDRIP. Calculations for current law19 andfor a 100 percent shareholder exclusionwould be unaffected by the change in themodeling of inflation. Under this alterna-tive modeling of inflation, the combina-tion of a dividend exclusion and a DRIPdoes less to lower the tax cost imposedon income from corporate investmentthan does a dividend exclusion plus acapital gains exclusion, because the DRIPleaves in place a tax on inflationary in-creases in asset value.

Other Issues

The model used to calculate effectivetax rates assumes that the U.S. economyis closed to international capital flows. Theextent to which capital is internationallymobile is an unsettled issue (Feldstein andHorioka, 1980; and Harberger, 1980), butit is clear that the effects of capital incometaxes can be substantially different in aneconomy open to international capitalflows than in an economy closed to suchflows.

International capital flows break thelink between investment in the U.S. andsaving by U.S. residents (Slemrod, 1988),and hence break the equivalence betweentaxes on investment and taxes on saving.One implication is that U.S. personal in-come taxes may do less to discourage in-vestment in the U.S. economy. A corollaryis that U.S. shareholder level taxes maydo less to discourage investment in thecorporate sector of the U.S. economy,relative to investment in the U.S. non-corporate business sector or in owner–oc-cupied housing, because the double taxdoes not translate into a higher cost ofcapital for the U.S. corporate sector. Thus,reducing or eliminating taxes on U.S.shareholders may do little to promote a

more efficient allocation of capital withinthe U.S. Nonetheless, reducing personaltaxes may still encourage more saving byU.S. citizens, even when investment in-centives for U.S. companies are not af-fected by taxes on their U.S. sharehold-ers. Analyzing integration in an openeconomy raises a number of additionalissues (U.S. Treasury, 1992; and Grubertand Mutti, 1994) and is beyond the scopeof this paper.

The calculations reported in Table 1 takeno account of replacement taxes; theyimplicitly assume that nondistortinglump–sum taxes are used to finance anyrevenue shortfall. By ignoring the distor-tions that would be caused by real worldtaxes that are needed to make–up the lostrevenue, the calculations overstate anyimplied benefits to the economy for inte-gration. This overstatement is larger, thelarger is the tax cut afforded corporateinvestment. Consequently, taking accountof differing revenue costs would likelymake less generous plans, such as a 50percent dividend exclusion, comparemore favorably with more generousplans, such as a full shareholder level ex-clusion.

Although our cost of capital calcula-tions assume that either all firms operateaccording to the old view of dividends orall operate according to the new view ofdividends, this assumption may not beappropriate. As emphasized by Auerbachand Hassett (2003a), some firms may beon each margin, in addition, firms maychange margins according to their circum-stances. Under current law, the tax costfaced by old view firms is higher than thatfaced by new view firms, thereby creat-ing a potential distortion in the allocationof investment. Integration can reduce oreliminate this distortion, thereby provid-ing additional economic benefits.20

19 For current law, this is a consequence of using a real dividend payout ratio that is equivalent to the nominalpayout ratio used in the earlier calculations.

20 We have benefited from discussion about this with Randy Mariger.


641

THE PRESIDENT’S PROPOSAL

We turn now to consideration of thePresident’s proposal, including its EDAaccounts. The President’s proposal re-sembles a 100 percent shareholder dividendexclusion combined with a DRIP that ex-tends the benefits of integration to earningsretained by the company. The President’sproposal, however, limits shareholder taxrelief to income that has been fully taxed atthe company level. It does this through theuse of an EDA. Only earnings out of thecompany’s EDA are excludable by share-holders. Earnings in excess of thecompany’s EDA are taxable to sharehold-ers. EDA is calculated as Tax/u – Tax.

Tax Preferences, Investment Incentives,and Integration

Corporate tax preferences are special de-ductions, credits, and exclusions that actto reduce the corporate level tax on invest-ment income below the statutory tax rate.21

Accelerated depreciation deductions are acommon example of a tax preference. Cur-rent U.S. tax law does not allow corporatetax preferences to be passed through toshareholders. Thus, corporate tax prefer-ences can reduce the tax paid by the com-pany but do not directly reduce the taxpaid by the shareholder.

Although perhaps not immediately ob-vious, the K–F model conforms to this le-gal structure. To see this quickly, recall thegeneral result that expensing reduces themarginal effective tax rate to zero. The K–F model obtains this result for the mar-ginal effective corporate tax rate (MECTR),that is, the difference between the cost ofcapital and the firm’s real discount rate(rc – π), expressed as a percentage of thecost of capital, as is apparent by setting z= 1 equation [2].

Expensing, however, does not reducethe marginal effective total tax rate to zero.For example, for an equity financed in-vestment and with no inflation, with ex-pensing the METTR = Te. The shareholdertax still applies. This occurs because thecompany’s tax deduction is too small tooffset the shareholder’s future tax liabil-ity on the cash flow. To offset both com-pany level and shareholder level taxes,both the shareholder and the companymust deduct the cost of the investment;the preference would have to be passedthrough. The same result occurs if thepreference comes as a tax exemption forthe cash flow. With u = 0, the MECTR = 0,but the METTR = Te; the preference is notpassed through to the shareholder.

Tax Preferences and Integration

Because of corporate tax preferences,simply exempting from shareholder leveltaxation all income from corporate equi-ties can be inconsistent with the policy goalof taxing corporate income once. Pure pref-erence income, for example, is taxed onlyat the shareholder level. A simple share-holder exemption would eliminate thesingle level of tax, making the income fromsuch an investment tax exempt in total, i.e.,METTR = 0. It is this problem that the EDAis intended to address by insuring thatpreference income continues to be taxed atthe shareholder level so that a single levelof tax is collected on all corporate profits.

Modeling the President’s Proposal withits EDA

We model the EDA as an additionalcash flow earned by the firm and paid outto the shareholder in the form of tax ex-emption certificates.22 With the EDA, capi-tal market equilibrium is given by

21 We are defining preferences relative to an idealized income tax. Not all such items would be consideredpreferences relative to a consumption tax baseline.

22 Following the general approach taken in the cost of capital literature based on the neoclassical theory ofinvestment, the effect of debt finance on the investment’s cost of capital, including its effect through the EDA,


642

[1.1] 1 = ∫(1 – u)ce–(r–π+δ)t dt + uz

+ x∫EDAte–rtdt

where,

EDA = (Tax)/u – Tax = (u(taxable income)/u) – u(taxable income) = (1 – u){ce(π–δ)t –(tax depreciation)t },

x = a factor reflecting taxes reduced bythe EDA,

and the other terms are as usually defined(the investment tax credit has been ig-nored).

This equilibrium condition differs fromthat in equation [1] by the addition of thepresent value of the tax savings from theEDA account.

Integrating, solving for c, and subtract-ing economic depreciation gives

[2.1] ρc= (r – π + δ){1 – χz}/(1 – χ) – δ,

as an expression for the cost of capital,where χ = u – x(1 – u) is the “EDA ad-justed” corporate tax rate.

In this formulation, one important ef-fect of the EDA is to act as a reduction inthe corporate tax rate from u to χ. To com-plete the modeling of the EDA, we mustspecify the firm’s nominal discount rateby relating it to the investor’s after–taxopportunity cost of funds, s + π, andspecify x, the EDA tax adjustment factor.

To facilitate our analysis, we considertwo cases, one in which a firm’s dividends

permanently exceed its EDA and the otherin which its dividends permanently areless than its EDA. Following our assump-tion that financial policy is not affected bytaxes, we assume that the relationshipbetween dividends and the EDA is exog-enous. A complete modeling would allowfirms to choose the relationship betweendividends and EDA, taking account ofrestrictions imposed by the degree of pref-erence income earned by the firm, as wellas allowing for such other factors as varia-tion overtime between a company’s per-manent income and its taxable income,and the severity of agency costs. Suchcomplications are beyond the scope of thispaper, but serve to limit the generality ofour results.23 We will first examine an eq-uity financed investment and then con-sider debt finance.

D > EDA

In this case, the firm always pays divi-dends in excess of its EDA. According tothe rules specified in the President’s pro-posal, all of the EDA would apply to divi-dends. Hence an additional dollar of EDAwould entitle the shareholder to a tax re-duction of $m. Consequently, the marginalinvestment’s after–tax rate of return is

D(1 – m)/q + RE(1 – ω) + mEDA/q

where D is dividends, RE is retained earn-ings, q is the marginal value of a dollar’sworth of capital in the corporation (i.e.,

is captured through the discount rate, as explained below. Hence, the EDA measure included in the cash flowin equation [1.1] is not reduced directly for interest deductions, in the same way that the tax payments on theinvestment are not reduced to account for interest in equation [1].

23 We do not consider the case in which the firm sets dividends equal to EDA and finances new equity invest-ment by issuing new shares. While such a policy might help to maximize the tax saving from the President’sproposal, it nonetheless might not be universally observed. It would be easier to implement to the extent thatthe firm did not earn preference income. Even then, however, it might be difficult to implement to the extentthat the firm experiences differences between its permanent income and its transitory income. For example,a firm in a highly cyclical industry or that otherwise has a large difference between permanent and transitoryincome might not wish to let its dividend policy vary step for step with its taxable income. In addition, theseparation of ownership from control may causes dividends to be less than EDA for some firms. The strategyof setting dividends equal to EDA would be more difficult for a firm with significant tax preferences, sincedoing so would not let it distribute all of its earnings as a dividend. If the firm must distribute earnings viadividends, then it must be the case that dividends exceed EDA to the extent that there is preference income.


643

the price of the investment). Note that theinvestor’s return includes the tax savingsfrom the EDA account. The investor re-quires that this after–tax rate of returnequal his opportunity cost, so that in equi-librium

s + π = D(1 – m)/q + RE(1 – ω) + mEDA/q.

For an old–view firm on the new shareissue margin q = 1. Rearranging the right–hand side and defining P as the dividendpayout ratio (P = D/(D + RE)), gives

[3.1] (s + π)/(1 – Pm – (1 – P)ω) = D + RE

+ mEDA/(1 – Pm – (1 – P)ω)

so the firm’s discount rate is given by

[3.2] re = (s + π)/(1 – Pm – (1 – P)ω).

The value of x is apparent from equa-tion [3.1]. A one dollar addition to EDAgenerates a tax reduction of m. But thisis an after–tax amount. Recall that thecash flows from the EDA in [1.1] are dis-counted at the pre–tax discount rate (theyare on a pre–tax basis); hence we mustexpress the benefit from the EDA in pre–tax terms by grossing up by shareholdertaxes. So

x = m/(1 – Pm – (1 – P)ω).

We also can now interpret the “EDAadjusted” tax rate, χ. The EDA adjustedtax rate shows the net corporate level taxthat is paid, after accounting for the share-holder level tax savings from the EDA.That is, corporate tax is paid at rate u, buteach dollar of after–tax corporate incomegenerates shareholder level tax savings ofx, giving χ as the net tax rate on corporateincome.

Consider now a new view firm that fi-nances marginal investment by retainingearnings. Because the EDA is exhausted,at the margin both the dividend tax andthe capital gains tax continue to apply.Since we assume that the firm is payingdividends, shareholders must be indiffer-ent between dividends and retentions so1 – m = q(1 – ω), or q = (1 – m)/(1 – ω). Inequilibrium, the firm’s after–tax rate ofreturn is s + π = D(1 – ω) + RE(1 – ω) + m(1– ω)EDA/(1 – m), so the required pre–taxrate of return, the firm’s discount rate, is

[3.3] re = (s + π)/(1 – ω),

and the EDA tax adjustment factor is24

x = m/(1 – m).

D < EDA

If the firm pays dividends that are lessthan its EDA, then according to the rulesof the President’s proposal the EDA isused first to reduce taxes on dividendsand then the remaining EDA is used toincrease shareholder basis to relieve taxon earnings that are retained within thecompany. Consequently, the share-holder’s after–tax rate of return is D(1 –m)/q + RE(1 – ω) + mD/q + ω(EDA – D)/q = D(1 – ω)/q + RE(1 – ω) + ωEDA/q. Inequilibrium the investor requires that

s + π = D(1 – ω)/q + RE(1 – ω) + ωEDA/q.

Under the old view q = 1, so the firm’sdiscount rate is

[3.4] re = (s + π)/(1 – ω).

This differs from the discount rateunder current law as well as from the dis-count rate that would apply under a

24 The adjustment factor is not m/ (1 – ω) because of tax capitalization. The tax savings of m costs q, and q = (1 –m)/(1 – ω), so x = (m/q)/(1 – ω) = m/ (1 – m). The “gross up” intuition still works: tax savings with an after–tax value of m/q have to be grossed up by capital gains taxes on the investment because the flows in [1.1] areon a pre–shareholder tax basis.


644

straight shareholder level exclusion. TheEDA eliminates the dividend tax rate, butat the margin an additional dollar of EDAtranslates into an additional dollar of ba-sis adjustment. So the capital gains taxburdens the investment’s nominal return.

The EDA tax adjustment factor is

x = ω/(1 – ω).

Under the new view (retained earningsmargin), q = 1 because dividends and capi-tal gains are both taxed at the rate ω. So

re = (s + π)/(1 – ω)

and

x = ω/(1 – ω).

For a firm paying dividends that are lessthan its EDA, there is no difference be-tween the old view and the new view ofdividend taxes.

Debt Finance

Debt finance can be included by com-puting the discount rate as

rd = i(1 – χ).

The discount rate, the after–tax interestrate, depends on which equity regime thefirm is in, since that determines the taxrate χ. The appropriate tax rate is χ ratherthan u because the interest deduction re-duces taxable income and hence the EDA.That is

rd = i(1 – u) + xi(1 – u) = i(1 – (u – x(1 – u))

= i(1 – χ).

With a mix of debt and equity, the corpo-rate discount rate (rc) is a weighted aver-age of that on debt and that on equity, withthe weights determined by the proportionof financing that is debt, taking account

of the undervaluation of equity that canoccur under the new view.

EDA Intuition

Before proceeding to calculations thatcompare the effects of the EDA on invest-ment incentives with those under otherforms of shareholder tax relief, it is worth-while working a bit with the EDA analy-sis in order to get a feel for how it affectsincentives. A shareholder exclusion withan EDA is substantially different from astraight shareholder exclusion or from adividend exclusion plus a DRIP.

1. Compared to a straight exclusion, orthe combination of a dividend exclu-sion and a DRIP, an EDA generallyprovides less tax relief on invest-ments that receive tax preferences.This is a primary purpose of thepolicy. This can be illustrated easilyif we ignore inflation and consideran equity financed investment thatis expensed (z = 1) and whose in-come is entirely paid out as a divi-dend by a firm for whom D > EDA.In such a case, the cost of capitalwould be rc, the firm’s after–tax dis-count rate. Without an EDA, rc = s,so that the METTR = 0. With an EDA,however, rc = s/(1 – m) and theMETTR = m, the shareholder’s taxrate on dividend income. This pointhas been emphasized elsewhere,e.g., Esenwein and Gravelle (2003).

2. Because of the effects of the EDA, un-der the President’s Proposal it is pos-sible (though not necessarily likely)that some investments would face ahigher effective tax rate than they dounder current law. This occurs whenan investment is subsidized in thesense of facing a negative marginaleffective corporate tax rate. The intu-ition of this result can be appreciatedeasily in equation [1.1] by positing taxbreaks (e.g., accelerated deductions


645

and credits) sufficient to give the EDAa negative present value, so that theEDA reduces the present value of theinvestment’s cash flow.

3. The EDA can provide a larger incen-tive to invest than would either asimple shareholder exclusion ofdividends and capital gains, or adividend exclusion plus a DRIP. Al-though the EDA is generally thoughtof as limiting the tax benefits of in-tegration compared with a straightexclusion, this does not always oc-cur at the margin. Suppose that theinflation rate is zero and consider thecase of an investment that receivesno preferences25 undertaken by afirm for which dividends exceed theEDA. For this firm the cost of capi-tal is (rc)/(1 – χ). Substituting rc = s/(1 – Te), χ = u – m(1 – u)/(1 – Te), andTe = Pm + (1 – p)c, shows that the costof capital is s/{(1 – u)(1 + m – Te)}.This is less than the cost of capitalwith a straight exclusion of income,s/(1 – u), because m > Te. That is, theEDA provides shareholder tax reliefvalued at m on an investment whichis subject to shareholder tax at rateTe. For an old–view firm, only if P =1 would the EDA give the same mar-ginal tax cost as obtained under astraight shareholder exclusion. Fora new view firm with D > EDA, theEDA never would provide the sameincentives as a straight exclusion be-cause for such a firm the cost of capi-tal always will be lower with an EDAthan without an EDA.

4. Although a straight shareholder ex-clusion, or a dividend exclusion plusa DRIP, would have no direct effecton the tax cost of a debt financed in-vestment, by reducing the tax rateat which firms evaluate interest de-

ductions, an EDA potentially affectsincentives on both debt financed andequity financed investments. As dis-cussed above, corporate debt can besubsidized when the tax codemismeasures income and the corpo-rate tax rate exceeds the tax rate paidby lenders on their interest income.By lowering the tax rate at which thefirm evaluates interest deductions,the EDA reduces this benefit,26

thereby potentially improving neu-trality between debt and equity fi-nanced corporate investment.

5. With a straight shareholder exclu-sion, or a dividend exclusion plus aDRIP, a firm’s dividend policy, or therelation of its dividends to its taxableincome, has no effect on investmentincentives. In contrast, under thePresident’s proposal, for a similarmarginal investment, the cost ofcapital can be much different for afirm who expects D < EDA than fora firm who expects D > EDA. Gen-eral conclusions are not possible. Forexample, for an equity financed in-vestment, the cost of capital can dif-fer depending on whether the firmis at the new share issue margin (oldview) or the retained earnings mar-gin (new view), on the firm’s pay-out ratio, and on the degree of pref-erence income earned on the mar-ginal investment. As noted above,when there is no preference income,both old view and new view firmsface a lower tax cost of equity fi-nanced investment if D > EDA thanthey do if D < EDA.27 Sufficientamounts of preference income, how-ever, can reverse this result for bothnew view and old view firms. Forold view firms, all else equal, thelower the dividend payout ratio,

25 No preference means that the marginal effective corporate tax rate equals the statutory corporate tax rate.26 In some cases the EDA can turn the current tax subsidy for debt into tax penalty, since it is possible that χ < θ.

Such could be the case for an old view firm for whom D > EDA that was owned by a high bracket investor.27 For an old view firm, the tax cost is the same if the payout ratio is one.


646

the more likely it is that the cost ofcapital will be higher when D < EDAthan when D > EDA.

6. With a straight shareholder exclu-sion, or a dividend exclusion plus aDRIP, the cost of capital is the samefor new view and as it is for old viewfirms. In contrast, under thePresident’s proposal, for firms ex-pecting that D > EDA the cost ofcapital for a firm at the new shareissue margin (old view) can differsubstantially from that for a firm onthe retained earnings margin (newview). On a similar investment, it isoften (but not always) the case thatthe old view firm will face at leastthe same tax cost as that faced by anew view firm because of differencesin the discount rate and in the EDAtax adjustment factor, x.

7. A straight exclusion would imposeno shareholder tax on the inflationcomponent of the return on corpo-rate equity financed investment. Incontrast, integration with an EDAwould continue to tax inflation. Thisis immediately apparent from thedifference in the discount rates. Withan exclusion, the discount rate for acorporate equity financed invest-ment is s + π, while with an EDA itis (s + π) grossed up by the appro-priate tax, e.g., by Pm + (1 – P)ω foran old view firm with D > EDA. Thecombination of a dividend exclusionplus a DRIP, however, could tax theinflationary component of the nomi-nal return, even if the proposal didnot contain an EDA.

8. Relative to a straight dividend ex-clusion, or to a dividend exclusionplus a DRIP, an EDA has an uncer-tain effect on tax neutrality. Byraising the tax cost of corporate in-vestments that receive some tax pref-erences, it reduces neutrality as be-tween corporate and noncorporate

investments. By creating tax differ-ences between old view and newview firms, it reduces neutrality atthat margin. Relative to a straightexclusion, but not necessarily to aDRIP, by taxing the inflation returnon corporate equity financed invest-ment, the President’s Proposal re-duces neutrality between the corpo-rate and noncorporate sectors. Incontrast, by reducing the tax advan-tage on corporate debt financed in-vestment, the President’s proposalimproves neutrality as between debtand equity financed corporate in-vestments. By raising the tax cost ofcorporate preference investmentsrelative to corporate non–preferenceinvestments, it acts to improve taxneutrality within the corporate sec-tor. Finally, by lowering the tax costof some corporate fully taxed invest-ments, it helps to promote neutral-ity between the corporate and thenoncorporate sectors.

Further complicating the comparisonis the difference between the revenuecost of an EDA and, say, a dividend ex-clusion plus a DRIP. On neutralitygrounds, an EDA is likely to comparemore favorably with a revenue neutralpartial dividend exclusion plus (partial)DRIP than it does with the more costlypolicy of full dividend exclusion plus(full) DRIP.

Rationales for the EDA

The discussion above suggests that anEDA can have incentive effects that differsubstantially from those obtained undersimpler forms of shareholder tax relief.These differences moreover do not neces-sarily lead to more neutral taxation of in-vestment income, nor do they necessarilylower taxes on capital income.


647

An EDA, however, offers several im-portant potential benefits. First, as men-tioned above, an EDA helps to ensure thatcorporate income is taxed once, a policygoal that is attractive to some in the taxpolicy community. Second, the EDA helpshold down the tax revenue cost of inte-gration. It does this by targeting share-holder level tax relief in two ways: (a) itfocuses the benefits of integration on cor-porate income that has been fully taxed,and (b) it reduces windfalls by limitingthe benefits of integration to income thathas been earned since the date of enact-ment (i.e., generally provides prospectivetax relief).28 Third, by limiting shareholderlevel tax relief to income that has beentaxed at the corporate level, an EDA helpsto reduce the incentive to engage in ag-gressive tax planning. It thus reduces theincentive for corporate tax shelters, aproblem that many worry has been grow-ing significantly in recent years (U.S. Trea-sury, 1999). Fourth, by giving sharehold-ers clearer information about corporatetax payments than is available form cor-porate annual reports, an EDA may helpto improve informational transparencyand promote informed investmentchoices and responsible corporate gover-nance.

The Effect of the President’s Proposalon Investment Incentives

Table 2 shows the effects of thePresident’s proposal on the level and dis-persion of marginal effective tax rates. Forpurposes of these calculations, we needto make some assumption about the pro-portion of investment/capital that is madeor held by firms for whom D > EDA andfor whom D < EDA. Based on historicaltax and financial data, we assume that 40percent of corporate investment is fromfirms for whom D > EDA and 60 percentfrom firms for which D < EDA.

Not surprisingly, comparing the calcu-lations in Table 2 with those in Table 1 sug-gest that the President’s proposal may doless to reduce the tax burden on corpo-rate investment than would a full share-holder exclusion. It also would do slightlyless to reduce the marginal corporate taxburden than would a dividend exclusionplus a DRIP, but more than would a 50percent dividend exclusion. The neutral-ity results follow in step.

Three factors account for the higher cor-porate effective tax rate under thePresident’s proposal compared to astraight shareholder exclusion. One is thatthe EDA stops preferences from being

TABLE 2THE EFFECTS OF THE PRESIDENT'S DIVIDEND TAX RELIEF PROPOSAL ON THE MARGINAL

EFFECTIVE TOTAL TAX RATE

President’s Proposal “Revenue Neutral”Partial Exclusion (61%)

Old View New ViewAltern. Taxation ofInflation Old View Old View

Corporate SectorNoncorporate Business SectorOwner–Occupied Housing

Economy wide

Standard deviation in thecost of capital

27.5%20.0%

3.5%16.5%

0.00653

25.5%20.0%

3.5%15.6%

0.00602

27.4%20.0%3.5%

16.4%

0.00650

27.5%20.0%3.5%

16.5%

0.00680

28 This is a looser restriction than limiting the tax reductions to income on investments that have been madesince the date of enactment. The EDA account stops the company from distributing tax free previously accu-mulated but retained corporate profits. It does not stop the company from distributing tax free new incomeearned from investments made prior to the date of enactment.


648

passed through. Another is that thePresident’s Proposal continues to tax theinflation return on corporate equity,whereas a full shareholder exclusionwould not tax inflation. Finally, by lower-ing the tax rate at which corporationsevaluate their interest deductions, theEDA raises the tax cost of a debt financedcorporate investment.

Because a dividend exclusion with anEDA does not lower the tax cost of a cor-porate investment relative to the tax coston income from investment in other sec-tors as much as does a full exclusion (withor without a DRIP, but without an EDA),in these calculations adding an EDA actsto reduce tax neutrality. Our calculationssuggest, however, that within the corpo-rate sector the EDA acts to promote neu-trality by reducing the benefit of tax pref-erences and hence equalizing effective taxrates across investments.29 Unfortunately,this effect is not enough to offset the neu-trality reducing effects of the EDA on cor-porate vs. other investment.

In terms of “bang for the buck” thePresident’s proposal might compare morefavorably to a full shareholder exclusionbecause EDAs reduce the revenue cost ofproviding shareholder level tax relief.Stated equivalently, an EDA is likely tocompare more favorably to a straight per-centage exclusion that had the same rev-enue cost. While we don’t know the pre-

cise equal revenue partial exclusion, weimplement the idea of a revenue con-straint by calculating the percentage ex-clusion of dividends and capital gains oncorporate shares that would give the samemarginal corporate effective tax rate as thePresident’s proposal. In the calculationsreported in Table 2, this turns out to be a61 percent exclusion under old view as-sumptions. By construction, a 61 percentexclusion has the same effect on the cor-porate and economy wide weighted av-erage marginal effect tax rate as does thePresident’s proposal.30 The exclusion alsodoes about the same as the President’sproposal to promote neutrality, as shownby a standard deviation in the cost of capi-tal that is about the same under the twotax plans.

THE 5/15 PLAN

As this paper is being completed, a pro-posal to reduce the tax rate on dividendsand capital gains to 15 percent, or 5 per-cent for low income taxpayers (falling tozero in 2008), has been enacted and signedinto law as part of the Jobs and Growthbill. Table 3 shows the effect of this pro-posal on the level and dispersion of mar-ginal effective total tax rates.31

Compared to current law, 5/15 looksbest in calculations based on the old viewof dividend taxes, where it lowers the

29 These calculations are not reported in Table 2.30 The exclusion applies to both the real and the inflation component of the return on an equity financed invest-

ment in the corporate sector. To the extent that the exclusion was limited to the real part of the return, e.g., aswith a DRIP, the equivalent exclusion percentage would be higher.

31 These calculations assume a 14.5 percent statutory tax rate on dividends and on capital gains. The 5/15proposal also would reduce taxes on capital gains earned on investment in the noncorporate business sector,but our model does not capture these effects.

TABLE 3THE EFFECTS OF THE 5/15 PROPOSAL ON THE MARGINAL EFFECTIVE TOTAL TAX RATE

Old View New View High Debt, Old View

Corporate SectorNoncorporate Business SectorOwner–Occupied Housing

Economy wide

Standard deviation in the cost of capital

29.4%20.0%

3.5%17.3%

0.00734

27.7%20.0%3.5%

16.6%

0.00685

17.6%15.2%

6.3%12.4%

0.00393


649

marginal effective total tax rate on incomefrom corporate investment by over 4 per-centage points. Under new view and highdebt (65 percent leverage) assumptions,5/15 continues to improve investmentincentives, but it does less to lower thelevel and dispersion of effective tax ratesthan in calculations based on the old view.

The 5/15 proposal looks about the sameas a 50 percent dividend exclusion in cal-culations based on the old view of divi-dend taxes, as show by nearly equivalenteffects on the marginal effective total taxrate on corporate income and on the stan-dard deviation in the cost of capital. The5/15 proposal also looks about the sameas a 50 percent dividend exclusion in cal-culations that assume 65 percent debt fi-nancing.32 In contrast, by providing sometax relief to capital gains on corporateshares, in calculations based on the newview, 5/15 would reduce the level anddispersion of level marginal effective taxrates, while a 50 percent dividend exclu-sion would have no effect.

Of course, because it offers only partialrelief from double taxation, 5/15 does farless to reduce the level and dispersion ofmarginal effective tax rates than wouldthe President’s proposal, but at a muchlower revenue cost.

CONCLUSION

This paper has considered the effect ofintegration on real investment incentives.We found that integration’s effects varyconsiderably, depending on such factorsas the specifics of the tax policy, assump-tions about financing, and the identity ofthe marginal investor. In addition, includ-ing an EDA, while offering some clear ben-efits, significantly complicates the analy-sis, and can act to reduce the size of someof the expected benefits from integration.Nonetheless, over a wide range of policyand parameter assumptions, we find that

integration is likely to offer the possibilityof a lower overall tax cost of investmentand a more uniform distribution of taxcosts across different investments.

Acknowledgments

We are grateful to Andrew Lyon, RandyMariger, and Eileen Mauskopf for helpfulcomments and discussions. Any optionsor conclusions reflect the authors’ viewsand should not be attributed to the orga-nizations with which they are affiliated.

REFERENCES

Atkinson, Anthony B., and Joseph E. Stiglitz.Lectures on Public Economics. 1980.

Auerbach, Alan J.“Corporate Taxation in the United States.”Brookings Papers on Economic Activity 2(1983a): 451–513.

Auerbach, Alan J.“Taxation, Corporate Financial Policy andthe Cost of Capital.” Journal of Economic Lit-erature 21 No. 3 (September, 1983b): 905–40.

Auerbach, Alan J.“Tax Policy and Corporate Borrowing.” InAre the Distinctions between Debt and EquityDisappearing?, edited by Richard W. Kopckeand Eric S. Rosengren, 136–62. Boston, MA:The Federal Reserve Bank of Boston, 1989a.

Auerbach, Alan J.“Capital Gain Taxation and Tax Reform.”National Tax Journal 42 No. 3 (September,1989b): 451–513.

Auerbach, Alan J.“Tax Reform, Capital Allocation. Efficiencyand Growth.” In Economic Effects of Funda-mental Tax Reform, edited by Henry J. Aaronand William G. Gale, 29–81. Washington,D.C.: Brookings Institution Press, 1996.

Auerbach, Alan J.“Taxation and Corporate Financial Policy.”NBER Working Paper No. 8203. Cambridge,MA: The National Bureau of Economic Re-search, April, 2001.

32 We do not show a high debt calculation for a 50 percent dividend exclusion.


650

Auerbach, Alan J., and Kevin A. Hassett.“On the Marginal Source of InvestmentFunds.” Journal of Public Economics 87 No. 1(January, 2003a): 205–32.

Auerbach, Alan J., and Kevin A. Hassett.“A Note on the User Cost of Capital WhenDividend Tax Relief is Tied To The EDA.”American Enterprise Institute Mimeo, Feb-ruary 2003b.

Bradford, David F.“The Incidence and Allocation Effects of aTax on Corporate Distribution.” Journal ofPublic Economics 15 No. 1 (February, 1981):1–22.

Esenwein, Gregg A., and Jane G. Gravelle.“The Taxation of Dividend Income: An Over-view and Economic Analysis of the Issues.”CRS Report RL31597, Congressional Re-search Service, The Library of Congress, 2003.

Feldstein, Martin, and Charles Horioka.“Domestic Saving and International Capi-tal Flows.” Economic Journal (1980): 314–29.

Fullerton, Don.“The Indexation of Interest, Depreciation,and Capital Gains and Tax Reform in theUnited States.” Journal of Public Economics32 No.1 (February, 1987): 25–51.

Fullerton, Don, Robert Gillette, and James B.Mackie III.

“Investment Incentives Under the Tax Re-form Act of 1986.” In Compendium of Tax Re-search 1987, 131–71. Washington, D.C.: De-partment of the Treasury, Office of TaxAnalysis, 1987.

Gale, William G., and Peter R. Orszag.“The Administration’s Proposal to Cut Divi-dend and Capital Gains Taxes.” TheBrookings Institution, Mimeo, January, 2003.

Gentry, William, Deen Kemsley, andChristopher J. Mayer.

“Are Dividend Taxes Capitalized Into SharePrices: Evidence From Real Estate Invest-ment Trusts.” Journal of Finance 58 No. 1(February, 2003): 261–82.

Graham, John R.“Taxes and Corporate Finance: A Review.”Duke University, 2003. Available online athttp://papers.ssrn.com/5013/papers.cfm?abstract_id=264516.

Gravelle, Jane G.The Economic Effects of Taxing Capital Income.Cambridge, MA: The MIT Press, 1994.

Grubert, Harry, and John Mutti.“International Aspects of Corporate Tax In-tegration: the Contrasting Role of Debt andEquity Flows.” The National Tax Journal 47No. 1 (March, 1994): 111–33.

Hall, Robert, and Dale W. Jorgenson.“Tax Policy and Investment Behavior.”American Economic Review 57 No.3 (June,1967): 391–414.

Harberger, Arnold C.“Vignettes on the World Capital Market (inEmpirical Studies of the Rate of ReturnCapital).” American Economic Review 70 No.2. Papers and Proceedings of the Ninety–Second Annual Meeting of the AmericanEconomic Association (May, 1980): 331–7.

Harris, Trevor, and Deen Kemsley.“Dividend Taxation in Firm Valuation: NewEvidence.” Journal of Accounting Research 37No. 2 (September, 1999): 275–91.

Jorgenson, Dale W.“Capital Theory and Investment Behavior.”American Economic Review 53 No. 2 (May,1963): 247–59.

King, Mervyn A.Public Policy and The Corporation. London:Chapman and Hall. 1977.

King, Mervyn A., and Don Fullerton, eds.The Taxation of Income from Capital. Chicago:The University of Chicago Press. 1984.

Mackie, James B. III.“Unfinished Business of the 1986 Tax ReformAct: An Effective Tax Rate Analysis of Cur-rent Issues in the Taxation of Capital In-come.” National Tax Journal 55 No. 2 (June,2002): 293–337.

Poterba, James M.“Tax Policy and Corporate Saving.” InBrookings Papers on Economic Activity No. 2(1987): 455–503.

Poterba, James M., and Lawrence H. Summers.“The Economic Effects of Dividend Taxes.”In Recent Advances in Corporate Finance, ed-ited by Edward Altman and MartiSubrahmanyam, 227–84. Homewood, IL: Ri-chard D. Irwin, 1985.


651

Slemrod, Joel.“Effects of Taxation with International Capi-tal Mobility.” In Uneasy Compromise: Prob-lems of a Hybrid Income–Consumption Tax,edited by Henry J. Aaron, Harvey V. Galperand Joseph A. Pechman, 115–48. Washing-ton, D.C.: The Brookings Institution, 1988.

Stiglitz, Joseph E.“Taxation, Corporate Financial Policy andthe Cost of Capital.” Journal of Public Eco-nomics 2 No.1 (February, 1973): 1–34.

U.S. Department of the Treasury.Integration of the Individual and CorporateTax Systems: Taxing Business Income Once.

Washington, D.C.: U.S. Government Print-ing Office, January, 1992.

U.S. Department of the Treasury.The Problem of Corporate Tax Shelters: Discus-sion, Analysis and Legislative Proposals. Wash-ington, D.C.: U.S. Department of the Trea-sury, July, 1999.

Zodrow, George R.“On the ‘Traditional’ and ‘New’ Views ofDividend Taxation.” National Tax Journal 44No. 4 (December, 1991): 491–510.

The Effect of Dividend Tax Relief on Investment Incentives

Documents

Transcript of The Effect of Dividend Tax Relief on Investment Incentives