NBER WOEXING PAPER
SERIES
DO TAXES AFFECT CORPORATE FINANCING DECISIONS?
Jeffrey
K. MacKie-Mason
Working Paper No. 2632
NATIONAL
BUREAU OF ECONOMIC RESEARCH
1050 Massachusetts Avenue
Cambridge, MA 02138
June 1988
This research is part of NEERs research program in Taxation. Any opinions
expressed are those of the author not those of the National Bureau of
Economic Research.
NEER Working
Paper #2632
June 1988
DO
TAXES AFFECT CORPORATE FINANCING DECISIONS?
AB
A
new empirical method and data set are used to study the effects of tax
policy on corporate financing choices. Clear evidence emerges that non-debt
tax shields "crowd out" interest deductibility, thus decreasing the desir-
ability of debt issues at the margin. Previous studies which failed to find
tax effects examined debt-equity ratios rather than individual, well-speci-
fied financing choices. This paper also demonstrates the importance of con-
trolling for confounding effects which other papers ignored. Results on
other (asymmetric information) effects on financing decisions are also
presented.
Jeffrey K. l4acKie-Mason
Department of Economics
University of Michigan
Ann Arbor,
MI
48109
Do Taxes Affect Corporate Financing Decisions?
-
Jeffrey
K. MacKie-Mason
1. INTRODUCTION
Nearly everyone believes taxes must be important to financing decisions, but no substantial support
has ever been found in empirical analyses. Myers (1984] wrote that "I know of no study clearly
demonstrating that a firm's tax status has predictable, material effects on its debt policy. I think the
wait for such a study will be protracted" (p.588). A similar conclusion is reached by Poterba [1986].
Recent studies which fail to find plausible or significant tax effects include Ang and Peterson (1986],
Long and Malitz (1985], Bradley, .Jarrell and Kim [1984],
Marsh (1982], and Williamson 11981].'
This paper disentangles several confounding interactions to provide clear and robust evidence of
substantial tax effects. The probability that a firm will issue debt to raise new funds decreases as
the expected value of interest deductibility declines.
The paper uses a different empirical method and data set from previous studies. I study the
observable, incremental financing choices made by firms; previous researchers look as a firm's debt-
equity ratio. The leverage ratio approach is subject to simultaneous equations bias, misspecification
of the choice model, and dynamic misspecification. The incremental choice approach overcomes
most of these problems.
I construct a sample of financing decisions from a data source not previously utilized to study
capital structure decisions: new public-issue security registrations with the SEC. The sample con-
stains 1418 observations from 1977—1984, covering new registrations for 613 different firms. The
large size of the data set and the focus on well-defined individual decisiosn allows me to control for
many more factors than has been possible in previous studies.
The results provide some of the first clear evidence that tax policy does significantly affect
financing decisions. The higher are a firm's non-debt tax shields (e.g.,
investment
tax credits, tax
loss carryforwards), the less likely it is to issue debt at the margin, because expected tax shield
I would
like to thank Alan Auerbach, Rob Geriner, Roger
Gordon, Jerry ilausman. Paul Healy, Jeff Muon,
Myron Scholes, Greg Niehaus, Bob Pindydc, Jim Poterba, Jay Ritter, Nejat.Scyhun, Glen Sueyoshi, and seminar
audiences at MIT, Michigan and NBER for helpful comments and advice. Donna Lawson and Sue Majewski provided
superb research assistance, With
support
from a Rackham Research Grant. The first version of this paper was
prepared with
ñnancial
support from the Alfred P. Sloan Foundation.
One exception is Bartholdy, Fisher and Mints (1987], which does not ftnd convincing non-debt tax shield effects,
but uses variation in Canadian corporate tax rates to identify significant interest deductititity effects.
1
"crowding out" towers the value of interest deductibility. Evidence is provided which indicates that
previous studies may have failed to confirm the tax shield effect because they did not control for
confounding effects.
The next section describes the main themes of tax-related and other theories of financing de-
cisions. Section 3 discusses the empirical method, and contrasts it with prior research. Section 4
presents the empirical analysis. This begins with a description of the testable implications of theory
and the data employed, followed by the results, specification and robustness tests, prediction tests,
and calculations of the economic magnitudes of the results. Sample selection and data construction
details are provided in an appendix.
2. GENERAL HYPOTHESES
In recent years there have been many new theoretical treatments of capital structure decisions.
I first summarize tax.based theories of financing decisions, and present testable implications. In
order to test the tax stories, however, it is necessary to control for any other behavioral effects
which may interact or be correlated with the tax effects. Other theories are presented in Section
2.2.
2.2 Taxes and Financing
Taxes are hypothesized to enter a firm's leverage decision calculus.2 An optimal leverage theory
describes supply and demand functions for debt in a competitive capital market which clears in
equilibrium. The benefits and costs of outstanding debt determine the firm's debt supply function;
the benefits and costs of holding debt determine the market demand. The firm's optimal debt level
is determined as the market-clearing equilibrium quantity. Taxes may provide benefits or costs for
both suppliers and demanders of finandal instruments.
Miller [1977)
inspired
the most recent theories of optimal leverage by setting the firm's debt-
level decision in an equilibrium supply and demand framework to investigate the effects of taxes.
Letting the personal tax rate on equity be zero, individuals will be indifferent between holding debt
and equity if r =
rd(l — r4, where
the subscripts (d,e) distinguish
between the risk-adjusted
returns to equity and debt, and t,4 is the personal tax rate on debt.3 In terms of bond prices,
Pd =
, personal
taxes imply a downward sloping demand curve; to induce investors to hold larger
2
use
the tenn leverage" because ol its common see is the literature. With
more than two securities, an optimal
capital structure Is a set of optimal security/assets ratio..
The results go through as long as the personal tax rate on debt is greater
than that on equally.
2
quantities of bonds, price must fall because marginal investors will be those with increasingly high
tax rates on the return to debt (see Figure 1).
On the supply side the tax deductibility of interest payments, but not payments to equity,
subsidizes corporate borrowi g Payments to security holders cost either r or
—
where
is the corporate tax rate. Indifference between selling bonds ad issuing equity requires Pd =
P(1
—
i-s). If
all firms face the same corporate tax rate, the supply curve for bonds is perfectly
horizontal (curve SM in Figure 1). As with Modigliani and Miller's (1958) Proposition I, leverage is
irrelevant to firm value, although in aggregate an optimal debt-equity ratio exists for the economy,
determined by bond market equilibrium.
More recent papers have discussed other tax-related supply-side costs of borrowing. Interest
deductibility has current value only if the firm is paying positive taxes. In an uncertain world, the
higher are other tax shields, the greater is the probability that the firm will find itself in a non-tax
status; hence the lower the expected value of the interest deduction (DeAngelo and Masulis (1980];
Ross
(1985]).
In general, the marginal tax rate on interest deductions may be decreasing in the level of debt
outstanding: r
= r(B),
r
< 0,
where B represents bonds outstanding. The margin between
borrowing and issuing equity is determined by r = r
[1 —
rc(B)]4
The resulting supply curve is
given by d =
P
[1 —
r0(B)].
The supply price of debt is increasing in the debt level of the firm,
so the firm in general has an optimal interior leverage ratio.
The supply curve slopes upward since leverage increases the expected after-tax cost of borrowing
(r,(B) <
0).
One implication of the model is that the firm's propensity to issue debt (rather than
equity) will increase as its marginal tax rate increases. However, statutory marginal tax rates
on U.S. corporations have varied little in the modern economy, and effective marginal tax rates
are nearly impossible to measure without access to confidential corporate tax returns.5 Another
implication is that the higher are a firm's available non-debt tax shields, the less likely it will be
to issue debt. This prediction will be tested in the empirical analysis using a number of different
publicly available measures of expected tax shield capacity.
The leverage theory also predicts that the firm's optimal debt ratio depends on the location of
the market demand curve for bonds. If arbitrage ensures that a single risk-adjusted interest rate
The notation is heuristic; I am ignoring, for instance, the proper treatment of
mathematical expectation when
functions
are nonlinear in the stochastic variables.
Bartholdy, Fisher and Minta (lflT) test this implication for arms in Canada, where statutory tax rates have
fluctuated.
3