See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/277305221

Capital Structure:Theory and Evidence

Research · May 2015

DOI: 10.13140/RG.2.1.3670.4160

CITATIONS READS
2 257

2 authors:

Nalinaksha Bhattacharyya Cameron K.J. Morrill

University of Alaska Anchorage University of Manitoba
26 PUBLICATIONS   190 CITATIONS    26 PUBLICATIONS   148 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Corporate governance View project

Financial Services Marketing View project

All content following this page was uploaded by Nalinaksha Bhattacharyya on 27 May 2015.

The user has requested enhancement of the downloaded file.

 Capital Structure:Theory and Evidence
Nalinaksha Bhattacharyya∗
Cameron Morrill†

This draft: May 27, 2015

Preliminary and Incomplete

Please Do Not Quote Without Permission

JEL Classification: C21; C23; G32.

Keywords: Capital Structure.

∗
University of Alaska Anchorage, College of Business and Public Policy, 3211 Providence Drive, Anchorage, AK
99508, USA. Tel:(907)786-1949, Fax:(907)786-4115, E-Mail: [email protected]
†
Asper School of Business 181 Freedman Crescent University of Manitoba, Winnipeg, MB R3T 2N2 Canada. Tel:
(204) 474-8435, Fax: (204) 474-7545, E-mail: [email protected]

1
 Capital Structure:Theory and Evidence
Abstract

In this paper, we develop and test empirically a model of capital structure. We show analyt-
ically that the square of 1 minus the debt ratio (total debt divided by total assets) is positively
associated with the inverse of total assets and negatively associated with the log of total assets
divided by the square of total assets. Empirical results using U.S. company data from Compu-
stat are consistent with the predictions of our model, even after controlling for other potential
confounding variables.

2
Introduction
Understanding capital structure has long been one of the important research issues in Finance.
In 1984, Myers (1984) wrote:
“ I will start by asking, ”How do firms choose their capital structures?” Again,
the answer is, “We don’t know.””(Page 575)
Some twenty four years later Antoniou et al. (2008) found that :
“In spite of extensive research, Myers’ (1984). p. 575 classic question “How do
firms choose their capital structures?” remains unanswered.” (Page 59)
In this paper we develop a model of capital structure as the outcome of optimization of expected
return by risk neutral shareholder. We also test our model with data
Rest of the paper is organized as follows. The next section reviews the literature to provide
the context for this paper. In the section titled A Simple Model Of Capital Structure we
develop the model for capital structure. We develop the econometric model that we use to test
our theory in the section titled Developing the Econometric Model. Sections titled Data and
Results describe the data and present our results respectively. The last section concludes.

Literature Review
The starting point for our review would be the capital structure irrelevance propositions of
Modigliani and Miller (1958). They argue that given an investment program in a perfect capital
market, capital structure is irrelevant to firm value. Therefore, there is no such thing as an
optimal capital structure. However, DeAngelo and DeAngelo (2006) raise questions about the
relevance of Modigliani and Miller (1958).
Subsequent work in the capital structure literature has identified more clearly many of
the benefits and costs of leverage. Costs include financial distress and stockholder-bondholder
agency costs, while often-cited benefits include tax savings and mitigated manager-shareholder
agency costs. The trade-off model suggests firms identify a leverage target that optimally bal-
ances the various costs and benefits of debt. Therefore, the optimal level of debt-to-equity is
likely to vary with variables like the marginal tax rate, level of financial distress and composition
of firm assets.
The pecking order framework (Myers (1984)) largely rejects the idea of an optimal debt-to-
equity. According to pecking order proponents, information asymmetry and financial distress
costs discourage managers from issuing common stock or other risky securities. Instead, man-
agers will prefer to finance real investment by internal funding, debt (especially default-risk free
debt), hybrid debt-equity securities, and then equity, in that order. As such, there is no optimal
level of debt-to-equity.
In their review of the capital structure literature, Graham and Leary (2011) identify several
limitations in published work in this area. One of the problems, they suggest, is that there does
not yet appear to be a one-size-fits-all capital structure theory (p. 311). The Capital Structure
Question is therefore still an open one. Leverage is also a slow moving persistent variable ( see

3
Mueller (2012)) and this indicates that capital structure is best understood as the outcome of a
long run equilibrium process.

Motivation
The motivation for this paper comes from Modigliani and Miller (1958). They, after ennunciating
their well known Propositions 1 and 2, went on to consider the case where the cost of debt will
be rising with leverage. They recognized that in that case the cost of equity will be a concave
curve. They nevertheless maintained that the cost of capital for the firm will nevertheless remain
invariant with respect to the capital structure because “Should the demand by the risk-lovers
prove insufficient to keep the market to the peculiar yield-curve MD, this demand would be
reinforced by the action of arbitrage operators.” [Page 276 Modigliani and Miller (1958)]. In
the last part of their paper (section E of Modigliani and Miller (1958) pages 281-287) they also do
some regression analysis to demostrate the empirical validity of their propositions. In contrast,
we recognize that the shareholders will settle for the capital structure which will maximize their
expected return and we demonstrate the validity of our model through regression results.

[Figure 1 about here.]

A Simple Model of Capital Structure

Let us consider a one period model. The starting point is at time 0 and the end point is at time
1.
Time 0 1 Remarks
Investment I ≥ 0 Not a decision variable. Determined exogenously.
Debt D≥0 Chosen by the shareholders. A Decision Variable.
Stock S≥0 D+S=I
Output Observe X is a realization of the stochastic variable x̃. x̃ is
output given by x̃ = θf (I) + ε̃ where ε̃ is i.i.d and E (ε̃) =
X 0. f (I) is an increasing and concave function of
I(∴ f 0 (I) > 0, f 00 (I) < 0). The increasing and concave na-
ture of f (I) captures the stylized fact that with increasing
investment the expected outout will increase at a dimin-
ishing rate. In other words f (I) captures the diminiishing
marginal productivity of capital.
Debt Raise Pay K(D) is the amount you are required to pay the debt hold-
Debt D K(D) ers as per agreed contract.K(D) incorporated both the re-
payment of Principal amount borrowed along with interest
costs. The payment to debt holders will be Min(X, K (D))

4
Therefore Return to Shareholders will be a stochastic variable r̃ given by

x̃ − K (D)
r̃ =
I −D
θf (I) + ε̃ − K (D)
= (1)
I −D

Taking expectation of both sides the Expected Return to Shareholders will be E(r̃) given by

θf (I) − K (D)
E (r̃) = ∵ E (ε̃) = 0 (2)
I −D
The shareholder’s objective function can be written as

θf (I) − K (D)
M aximize E (r̃) = (3)
D I −D
The First Order Condition (FOC) for optimization is

dE (r̃) 1 0
dD
= 2 [− (I − D) K (D) + (θf (I) − K (D))] = 0 (4)
(I − D)

The Second Order Condition for optimization is

d2 E (r̃) 1 h
2 00 0 0
i
= 4 (I − D) {− (I − D) K (D) + K (D) − K (D)}
dD2 (I − D)
1 0
− 4 [{− (I − D) K (D) + (θf (I) − K (D))} {−2 (I − D))}]
(I − D)
1 h
2 00 0 0
i
= 4 (I − D) {− (I − D) K (D) + K (D) − K (D)}
(I − D)
∵ − (I − D) K 0 (D) + (θf (I) − K (D)) = 0 by FOC
K 00 (D)
=− < 0 ∀D 6 I, K 00 (D) > 0 (5)
I −D

Therefore the First Order Condition defines the Optimum Debt Level, since E(r̃) is a globally
concave function of D.
Suppose K (D) = (1 + r) D + τ D2 , r > 0, τ > 0, then K 0 (D) = (1 + r) + 2τ D.
Substituting for K (D) and K 0 (D) in (4), we get

5
 − (I − D) {(1 + r) + 2τ D} + θf (I) − (1 + r) D − τ D2 = 0


⇒ − (I − D) (1 + r) − (I − D) 2τ D + θf (I) − (1 + r) D − τ D2 = 0
⇒ − I (1 + r) + (1 + r) D − 2τ ID + 2τ D2 + θf (I) − (1 + r) D − τ D2 = 0
⇒τ D2 − 2τ ID + {θf (I) − I (1 + r)} = 0
θf (I) − I (1 + r)
⇒D2 − 2ID + =0
" τ #
r
1 θf (I) − I (1 + r)
⇒D∗ = 2I ± 4I 2 − 4
2 τ
r
θf (I) − I (1 + r)
⇒D∗ = I ± I 1 − (6)
τ I2
r
D∗ (1 + r) θf (I)
⇒ =1± 1+ − (7)
I τI τ I2
∗ 2
 
D (1 + r) θf (I)
⇒ 1− =1+ − (8)
I τI τ I2

Developing The Econometric Model

Optimal Debt to Total Asset Ratio will be given by (7). The regression equation is suggested
by (8). The regression equation will be of the form 1
 2
DEBT 1 f (T OT AL ASSET )
1− = β0 + β1 + β2 (9)
T OT AL ASSET T OT AL ASSET T OT AL ASSET 2

The predicted signs for estimated coefficients in (9) are:2

β1 > 0
β2 < 0
β1 + β2 < 0

We estimated the coefficients and tested our hypothesis for two different functional forms for
√
f (ASSET S). These are f (ASSET S) = ln (ASSET S) and f (ASSET S) = ASSET S. The
1
The econometric relationship tested is not tautological. We can demonstrate it as follows. Let us write IN V T A =
1
T OT AL ASSET
. Further let us represent the left hand side of (9) as y. Then

dy
y = (1 − DEBT x IN V T A)2 . ∴ = 2 (1 − DEBT x IN V T A) (−DEBT ) < 0.
dIN V T A
So if our econometric equation was tautological then we would have had β1 < 0. But our prediction is β1 > 0.
2
We can hypothesize β1 > 0 and β2 < 0 by looking at the signs of the coefficients in (8). We can see that
β1 = (1+r) and β2 = − τθ . On economic grounds we can conclude that θf (I) > (1 + r)I. The functional form we use
τ √
are f (I) = ln (I) and f (I) = I. For both these functional forms f (I) < I over the relevant range. That implies
that θ > (1 + r) . So we can predict that for our model β1 + β2 < 0.

6
results presented in this paper are those for f (ASSET S) = ln (ASSET S). The results are
√
similar for f (ASSET S) = ASSET S.

Data
Our data is from Compustat and consists of annual financial data for all non-financial U.S.
companies with fiscal years ending between January 1998 and December 2007. Cutting off the
sample at December 2007 prevents contamination of our data by the 2008 financial crisis. In
addition, we included in our study only companies with sales, total assets and shareholders’
equity greater than zero in all nine years between 1998 and 2007. This left us with a sample of
3,067 firms and 27,603 firm-years. Sample descriptive statistics are presented in Table 1.

<<Table 1 about here>>

The mean (median) firm-year had total assets of almost $4.566 billion ($397.87 million) and
total debt of $1.09 billion ($31.21 million). A correlation matrix of the variables in the sample
is presented in Table 2. By virtue of the large sample size, many of the correlations are
statistically significant even though the value of Pearson r is often relatively low. The absolute
values of the correlations do not exceed 0.5, with one exception. There is a very high
correlation (0.844) between Var1 and Var2, the two independent variables in our model. This
is not surprising given that both variables are functions of only total assets.

<<Table 2 about here>>

Regression Results
The results of pooled linear regressions testing our model are shown in Table 3. We use Tobit
to estimate our coefficients. Model 1 is the model as we derived it without any control
variables. For the sake of ready reference the models we estimate are:
Model 1: Y = β0 + β1 V AR1 + β2 V AR2 + ε̃
Model 2: Y = β0 + β1 V AR1 + β2 V AR2 + β3 P ROF IT + β4 IN CV OL + β5 T AXSHLD+
β6 CHSHP R + β7 M KT BOOK + β8 F IXED + β9 SIZE + β10 EF F T AX+
β11 DIV P AY + β12 ZSCORE + β13 BET A + β14 RDSALE + β15 M AT U RIT Y +
β11 CAP XDEF + Industry and Year Dummy Variables +ε̃
where

Y is the dependent variable in our regression. Y originates from (8) and is defined as

2
Y = 1 − DEBT
AT . AT is Total Assets.
1
VAR1 is one of the explanatory variables suggested by (8). It is defined as V AR1 = AT . AT
is Total Assets.
ln(AT )
VAR2 is the other explanatory variables suggested by (8)l. It is defined as V AR2 = AT 2 .
AT is Total Assets.
DEBT is Total Debt.

7
PROFIT is Operating Profit divided by Total Asset (AT)
INCVOL is Change in Earning minus Average Change in Earnings.
TAXSHLD is Non debt Tax Shield measured by the ratio of Depreciation Expenses to Total
Assets (AT).
CHSHPR is Change in Share Price rcorded as the ratio of price change in a period to the
price at the beginning of the period.
MKTBOOK is ratio of Market Value to Book Value.
FIXED is the ratio of Net Tangible Assets to Total Assets (AT).
SIZE is the Natural Logarithm of 1 plus Sales.
EFFTAX is the ratio of Income Tax Expense to Total Taxable Income.
DIVPAY is the ratio of Ordinary Dividends to Net Income.
ZSCORE is the Altman Z Score. It is given by the formula ZSCORE = 3.3 Operating Income
Assets +
Sales Retained Earnings Net Working Capital Market Equity
Assets + 1.4 Assets + 1.2 Assets + 0.6 (Current liabilities + Long Term Debt)
BETA is Monthly Beta from Compustat for month ending in fiscal year end.
RDSALE is the ratio of research and development expense to sales
MATURITY is the proportion of long-term debts due in three years or more
CAPXDEF is Capital expenditures divided by total assets.

We also used dummy variables to control for 2-digit SIC and the years. Those results are not
reported here. Again, for a ready reference, the predicted signs for estimated coefficients are:

β1 > 0
β2 < 0
β1 + β2 < 0

The coefficients of Var1 and Var2 are of the predicted sign and are statistically significant (p
<0.01). The sum of the coefficients of Var1 and Var2 is negative as per our predictions and are
statistically significant (p <0.01). This suggests strong support for our model.

<<Table 3 about here>>

As a further test, we include in the regression control variables that have been found to be
associated with capital structure, drawn largely from Antoniou et al. (2008) and Graham and
Leary (2011). These variables include: PROFIT, operating profit divided by total assets;
INCVOL, a measure of income volatility defined as the difference between year t and year t-1
earnings, less the mean change in earnings over the entire eight-year period for the firm;
CHSPHR, change in share price over the fiscal year, adjusted for stock splits and dividends;
MKTBOOK, the book value of total assets less the book value of equity plus the market value
of equity, divided by the book value of total assets; FIXED, net tangible assets divided by
total assets; SIZE, measured as the natural log of total sales plus one; EFFTAX, calculated as

8
income tax expense divided by income before tax; DIVPAY, the dividend payout ratio
calculated as ordinary dividends declared divided by net income; ZSCORE, Altman’s Z-Score;
RDSALE, research and development expense divided by total sales; and the market model
BETA, computed over the sixty monthly data points ending in the firm-year’s fiscal year end.
In addition, we include dummy variables to control for the effects of 59 two-digit SIC
industries and eight fiscal years. The results of this enlarged model are presented as Model 2
in Table 3. The coefficients of Var1 and Var2 and their sum are significant (p<0.01) and are in
the predicted direction.

Robustness Check
The robustness of our result was investigated in two different ways. First we estimate a the
coefficients for a different functional relationship. Recall that our general result is valid for any
production function which is an increasing and concave function of investment. We have used
ln (ASSET S) and reported the results in this paper. Additionally we also used another
√
increasing and concave function, ASSET S to estimate our coefficients and check our
hypotheses. Secondly we use Panel regressons to check our reults. Our model passes both
these robustness checks.

Conclusion
We see from the results that the predictions of our model are validated in the data even when
we control for other plausible explanatory variables.

References
Antoniou, A., Guney, Y., and Paudyal, K. (2008). The determinants of capital structure:
Capital market-oriented versus bank-oriented institutions. Journal of Financial and
Quantitative Analysis, 43(1):59–92.

DeAngelo, H. and DeAngelo, L. (2006). The irrelevance of the mm dividend irrelevance

theorem. Journal of Financial Economics, 79(2):293.

Graham, J. R. and Leary, M. T. (2011). A review of empirical capital structure research and
directions for the future. Annual Review of Financial Economics, 3(1):309–345.

Modigliani, F. and Miller, M. H. (1958). The cost of capital, corporation finance and the
theory of investment. The American Economic Review, 48(3):261.

Mueller, M. (2012). Essays on the persistence of leverage in residual-based portfolio sorts.

Ph.D Dissertation, The University of British Columbia.

Myers, S. C. (1984). The capital structure puzzle. The Journal of Finance, 39(3):575.

9
 Table 1: Descriptive Statistics
Variable N Mean Median Std. Deviation Minimum Maximum
Y 27603 0.71 0.74 0.26 0.01 1
−7 −8
VAR1 27603 0.35x10 0.25x10 0.36x10−6 0.13x10−11 0.4x10−4
VAR2 27603 015x10−11 0.13x10−15 0.10x10−9 0.43x10−22 0.16x10−7
AT 27603 4566.81 397.87 19054.91 0.03 795337
DEBT 27603 1090.22 31.21 5568.84 0 376870
PROFIT 27603 0.036551 0.67 1.5427 -6.023 251
INCVOL 27603 -0.0068682 0.01 26.445 -1907.5 1486.4
TAXSHLD 27603 0.043733 0.03 0.11837 -9.0087 10.929
CHSHPR 27603 9.9159 0.11 62.951 -0.99545 4993.8
MKTBOOK 27603 2.0415 1.44 3.1582 -2.7232 203.93
FIXED 27603 0.29052 0.21 0.24975 -0.70905 2.8953
SIZE 27603 5.7498 5.87 2.4472 -0.70905 12.836
EFFTAX 27603 0.37216 0.32 24.461 -133 4042
DIVPAY 27603 0.16183 0 10.873 -1654.7 340.65
ZSCORE 27603 91.943 1.85 634.6 -53.966 22814
BETA 27603 481.08 1.01 1537.1 -106.32 9997
RDSALE 27603 33.633 0.19 278.16 -37.705 13779
MATURITY 27603 1.926 0.78 52.624 -37.705 7098
CAPXDEF 27603 0.051395 0.04 1.5505 -52.304 211.93

Y is the dependent variable in our regression. VAR1 and VAR2 are the two explanatory
variables suggested by our model. Y, VAR1 and VAR 2 are defined below. AT is Total Assets.
DEBT is Total Debt. PROFIT is Operating Profit divided by Total Asset (AT) INCVOL is
Change in Earning minus Average Change in Earnings. TAXSHLD is Non debt Tax Shield
measured by the ratio of Depreciation Expenses to Total Assets (AT). CHSHPR is Change in
Share Price rcorded as the ratio of price change in a period to the price at the beginning of the
period. MKTBOOK is ratio of Market Value to Book Value. FIXED is the ratio of Net
Tangible Assets to Total Assets (AT). SIZE is the Natural Logarithm of 1 plus Sales. EFFTAX
is the ratio of Income Tax Expense to Total Taxable Income. DIVPAY is the ratio of Ordinary
Dividends to Net Income. ZSCORE is the Altman Z Score. It is given by the formula

Operating Income Sales Retained Earnings

ZSCORE = 3.3 + + 1.4
Assets Assets Assets
Net Working Capital Market Equity
+1.2 + 0.6
Assets (Current liabilities + Long Term Debt)

BETA is Monthly Beta from Compustat for month ending in fiscal year end. RDSALE is the
ratio of research and development expense to sales MATURITY is the proportion of long-term
debts due in three years or more CAPXDEF is Capital expenditures divided by total assets.
The dependent variable Y and the explanatory variables VAR1 and VAR2 originates from our

2
model and are defined as Y = 1 − DEBT
AT
1
, V AR1 = AT , V AR2 = ln(AT
AT 2
)

10
 Table 2: Correlation Coefficients
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17)
Y (1) 1
VAR1 (2) .215** 1
VAR2 (3) .182** .948** 1
PROFIT (4) -.148** -.352** -.328** 1
INCVOL (5) 0.004 -0.011 -0.011 -0.002 1
TAXSHLD (6) -.077** .014* .019** -.152** 0.003 1
CHSHPR (7) .013* 0.007 0.007 0.007 -0.002 -.019** 1
MKTBOOK (8) .174** .172** .148** -.131** 0.001 -.014* .029** 1
FIXED (9) -.439** -.121** -.094** .146** 0.001 .329** -0.008 -.116** 1
SIZE (10) -.368** -.499** -.431** .455** 0.003 .023** -0.01 -.127** .220** 1
EFFTAX (11) 0.005 -0.004 -0.004 0.001 -0.002 -0.006 0 -0.001 -0.004 -0.002 1
DIVPAY (12) -.016** -.034** -.042** 0.008 0.001 0.005 0 -0.001 .016** .026** 0 1
ZSCORE (13) -.071** -.285** -.271** .638** -0.005 -.149** .014* -.065** .087** .445** 0.002 0.011 1
BETA (14) -0.001 0.009 0.007 0.002 0.009 .023** -0.002 0.001 -0.007 -.014* 0 0 -0.011 1
RDSALE (15) .023** 0.007 0.002 -.086** 0.005 -.021** -0.01 0.011 -.033** -.080** -0.001 -0.001 -.064** .024** 1
MATURITY (16) -.336** -.190** -.130** .155** -0.001 -.086** -.038** -.084** .187** .264** .020** .020** .111** -.024** -0.015 1
CAPXDEF (17) -.125** -.074** -.062** .083** 0.005 .372** -0.002 .013* .544** .067** -0.004 0.006 .065** -0.008 -.024** .052** 1

Y is the dependent variable in our regression. VAR1 and VAR2 are the two explanatory variables suggested by our model. Y,

11
VAR1 and VAR 2 are defined below. AT is Total Assets. DEBT is Total Debt. PROFIT is Operating Profit divided by Total
Asset (AT) INCVOL is Change in Earning minus Average Change in Earnings. TAXSHLD is Non debt Tax Shield measured by
the ratio of Depreciation Expenses to Total Assets (AT). CHSHPR is Change in Share Price rcorded as the ratio of price change
in a period to the price at the beginning of the period. MKTBOOK is ratio of Market Value to Book Value. FIXED is the ratio
of Net Tangible Assets to Total Assets (AT). SIZE is the Natural Logarithm of 1 plus Sales. EFFTAX is the ratio of Income Tax
Expense to Total Taxable Income. DIVPAY is the ratio of Ordinary Dividends to Net Income. ZSCORE is the Altman Z Score.
It is given by the formula
Income Sales Earnings Capital Market Equity
ZSCORE = 3.3 Operating Assets + Assets + 1.4 Retained
Assets + 1.2 Net Working
Assets + 0.6 (Current liabilities + Long Term Debt) BETA is
Monthly Beta from Compustat for month ending in fiscal year end. RDSALE is the ratio of research and development expense to
sales MATURITY is the proportion of long-term debts due in three years or more CAPXDEF is Capital expenditures divided by
total assets. The dependent variable Y and the explanatory variables VAR1 and VAR2 originates from our model and are defined

2 1 )
as Y = 1 − DEBT AT , V AR1 = AT , V AR2 = ln(AT
AT 2
Correlation with AT and Debt not computed because these variables are used in computing Y, VAR! and VAR2.
 Table 3: TOBIT Regression Results.

Coefficients
( Asymptotic Normal Statistics )

Independent Variables Predicted Sign Model 1 Model 2

Intercept ? 0.73***(372.88) 0.88***(36.47)
VAR1 + 0.91x106 ***(33.05) 0.17x106 ***(8.34)
VAR2 - -0.22x1010 ***(-28.7) -0.37x109 ***(-3.83)
PROFIT ? 0.21*** (3.80)
INCVOL ? 0.00005 (0.954)
TAXSHLD ? 0.85*** (5.32)
CHSHPR ? 0.00004* (1.79)
MKTBOOK ? 0.014*** (20.08)
FIXED ? -0.366*** (-40.47)
SIZE ? -0.036*** (-44.29)
EFFTAX ? 0.0007 (1.18)
DIVPAY ? -0.0004 (-1.23)
ZSCORE ? 0.00002*** (5.95)
RDSALE ? 0.000006 (0.92)
MATURITY ? 0.00017*** (3.024)
CAPXDEF ? 0.0023 (1.28)
N 27603 27603
VAR1+VAR2 - -0.22x1010 ***(-28.69) -0.37x109 ***(-3.83)
Wald χ2 1095.12*** 14651.71***
Pseudo R2 5.82% 36.32%

Y is the dependent variable in our regression. VAR1 and VAR2 are the two explanatory variables suggested by
our model. Y, VAR1 and VAR 2 are defined below. AT is Total Assets. DEBT is Total Debt. PROFIT is
Operating Profit divided by Total Asset (AT) INCVOL is Change in Earning minus Average Change in
Earnings. TAXSHLD is Non debt Tax Shield measured by the ratio of Depreciation Expenses to Total Assets
(AT). CHSHPR is Change in Share Price rcorded as the ratio of price change in a period to the price at the
beginning of the period. MKTBOOK is ratio of Market Value to Book Value. FIXED is the ratio of Net
Tangible Assets to Total Assets (AT). SIZE is the Natural Logarithm of 1 plus Sales. EFFTAX is the ratio of
Income Tax Expense to Total Taxable Income. DIVPAY is the ratio of Ordinary Dividends to Net Income.
ZSCORE is the Altman Z Score. It is given by the formula ZSCORE =
3.3 Operating Income
Assets
Sales
+ Assets +1.4 Retained Earnings
Assets
+1.2 Net Working
Assets
Capital Market Equity
+0.6 (Current liabilities + Long Term Debt)
BETA is Monthly Beta from Compustat for month ending in fiscal year end. RDSALE is the ratio of research
and development expense to sales MATURITY is the proportion of long-term debts due in three years or more
CAPXDEF is Capital expenditures divided by total assets. The dependent variable Y and the explanatory
variables VAR1 and VAR2 originates from our model and are defined as
“ ”2
ln(AT )
Y = 1 − DEBT AT
, V AR1 = AT1
, V AR2 = AT 2 We also used dummy variables to control for 2-digit SIC
and the years. Those results are not reported here.

12
 Figure 1: Reproduced Figure 2 from Modigliani and Miller (1958)

The figure is reproduced from Modigliani and Miller (1958) where it appears as Figure 2 on page 275.
According to Modigliani and Miller (1958), the curve MD is the cost of equity. They said :
“Should the demand by the risk-lovers prove insufficient to keep the market to the peculiar
yield-curve MD, this demand would be reinforced by the action of arbitrage operators.” [Page
276]
Our contention is that the shareholders will stop at the maximum point of MD and this will give us the
optimal capital structure. In this paper we have analytically determined the structural form for such an
optimal capital structure and econometrically tested the predictions

13

View publication stats