Asia Pacic Management Review 22 (2017) 159e165

H O S T E D BY Contents lists available at ScienceDirect

Asia Pacic Management Review

journal homepage: www.elsevier.com/locate/apmrv

Non-linear dynamics of size, capital structure and protability: Empirical

evidence from Indian manufacturing sector
Dinesh Jaisinghani*, Kakali Kanjilal
International Management Institute, B-10, Qutab Institutional Area, New Delhi, India

a r t i c l e i n f o a b s t r a c t

Article history: The identication of optimal level of capital structure has been a topic of research for many years now.
Received 19 December 2015 Yet, none of the theories on capital structure has been able to provide a convincing answer to this op-
Accepted 26 December 2016 timum debt problem. The current study aims at advancing the research on non-linear relationship be-
Available online 14 March 2017
tween capital structure and rm performance for manufacturing sector in India. This has been
accomplished by analyzing the non-linear associations among rm size, capital structure and prot-
Keywords:
ability. The study deploys panel threshold regression methodology as proposed by Hansen (1999) to nd
Threshold panel
out the different regimes in which capital structure differentially impacts protability of rms based on
Capital structure
Indian manufacturing sector
their respective sizes. The study is based on a sample of 1194 publicly traded manufacturing rms in
Firm size India. The time frame considered is from 2005 to 2014. The results conrm the signicance of a single
Asymmetric information theory threshold for size, thereby indicating the presence of two separate regimes in which capital structure
Trade-off theory differentially impacts protability. This threshold or cut-off size level is estimated to be around 148
million rupees. It is found that rms which exceed the threshold size are positively impacted by the use
of debt in their capital structure and vice versa. The ndings have useful implications for small size rms
as they can reduce their overall costs of doing business by reducing the debt in their total capital. The
empirical evidence supports both trade-off and asymmetric information theories of capital structure.
2017 College of Management, National Cheng Kung University. Production and hosting by Elsevier
Taiwan LLC. All rights reserved.

1. Introduction
One of the most debatable areas in corporate nance pertains to
nding the manner of optimum utilization of debt. The optimal
debt level helps to minimize the overall cost of capital while
enhancing the protability of a rm. The identication of most
appropriate level of debt is vastly researched in corporate nance.
However, the ndings are mixed. It is often found that debt has
positive, negative, or no impact on the protability of a rm. The
identication of optimal debt level has often been debated and
various authors have proposed different solutions for the optimal-
debt problem. Some authors suggest that rms should try to weigh
the cost of debt vis-a -vis its advantage and then decide on the
optimal level of debt (Ferri & Jones, 1979; Kester, 1986; Rajan &
Zingales, 1995). Some other authors argue that there is no partic-
ular target level of debt and capital structure choices are made on
* Corresponding author.
E-mail addresses: dinesh200384@gmail.com (D. Jaisinghani), kakali@imi.edu
(K. Kanjilal).
Peer review under responsibility of College of Management, National Cheng
Kung University.
the basis of requirements and availability of funds (Arosa, Richie, &
Schuhmann, 2014; Hovakimian, 2004; Lee, Su, & Lin, 2012). A host
of other theories also try to explain the key determinants of the
optimal level of debt for different rms. The two most prominent
theories pertaining to optimal capital structure include the trade-
off theory and the asymmetric information theory.
The trade-off theory proposes that the optimal capital structure
is determined by weighing the cost and benets driven from the
use of debt. Kraus and Litzenberger (1973) argue that the major
benet of debt is the tax shield advantage associated with it. This is
because interest paid to debt holders, in most countries, is
deductible from the gross prots before calculating the tax liability
of a rm. On the other hand, there are certain costs associated with
debt such as cost of bankruptcy and cost of liquidation. These
benets and costs of debt are evaluated in order to arrive at the
optimal capital structure. Miller (1977) adds to this theory and
asserts that the capital structure of rms is determined by the in-
teractions between corporate tax and differential rates of personal
tax on interest income and dividend income. Thus, the trade-off
theory proposes an optimal level of debt at which point the over-
all cost of capital is minimized. There have been a host of studies

http://dx.doi.org/10.1016/j.apmrv.2016.12.003
1029-3132/ 2017 College of Management, National Cheng Kung University. Production and hosting by Elsevier Taiwan LLC. All rights reserved.
160 D. Jaisinghani, K. Kanjilal / Asia Pacic Management Review 22 (2017) 159e165
that have empirically tried to validate the trade-off theory
(Antoniou, Guney, & Paudyal, 2008; Dudley, 2012; Guha-Khasnobis
& Bhaduri, 2002; Handoo & Sharma, 2014).
The asymmetric information theory of capital structure claims
that rms use their capital structure as a signaling instrument to
the market. Ross (1977) proposes that rms that carry higher levels
of debt are perceived to be of better quality as compared to rms
that have low levels of debts. This is because rms having a stable
stream of earnings can easily meet their interest obligations on
time. Firms with more volatile earnings, on the other hand, are
more likely to default and hence carry higher bankruptcy risks. This
signaling effect helps the rms in managing future capital re-
quirements. Myers and Majluf (1984) propose that there is an
asymmetric information problem between managers and investors.
Managers being insiders know more than the investors who only
invest money but do not actively participate in the managerial
process. The investors would prefer to factor this asymmetric in-
formation problem before lending money to the rm. This problem
is more severe for equity issues than for debt issues. Hence, in-
vestors discount equity more than debt. In order to avoid this dis-
counting problem, rms rely on equity as a last resort. Thus, rms
prefer to follow a hierarchical path for issuing capital. Initially, rms
prefer to utilize the retained earnings, and then they issue debt due
to its lower cost. Finally, when the debt capacity is exhausted, rms
issue equity. There have been various studies that have empirically
validated the asymmetric information theory in different contexts
(Bayrakdaroglu, Ege, & Yazici, 2013; De Jong, Verbeek, &
Verwijmeren, 2010; Paligorova & Xu, 2012; Psillaki & Daskalakis,
2009).
The above discussion highlights that the identication of
optimal capital structure is a critical issue for every rm. However,
the issue pertaining to the identication of the non-linear dynamics
of optimal capital structure is yet to be investigated in depth. The
current study aims at nding the optimal level of size that helps in
identifying the right combination of debt and equity for various
categories of rms operating in the manufacturing sector in India.
Although, India is the third largest economy in terms of aggre-
gate GDP as scaled by purchasing power parity, the sustainability of
India's high growth trajectory depends on the growth of the
manufacturing sector. The sector generates a meagre 16 percent of
the nation's GDP as compared to the services sector that contrib-
utes 55 percent.1 Massive workforce and abundance in certain
natural resources such as iron ore, cotton and coal can possibly
make India as the most viable manufacturing alternative to China.
On the other hand, poor transport infrastructure, high cost of po-
wer, rising cost of capital, and labour issues are some of the factors
that are plaguing the competitive edge of the sector. Besides, the
cost of power is approximately 50 percent higher in India than in
China. The cost of capital hovers around 10 to 12 percent as against
the international average of 6e8 percent. The stringent labour laws
in India have been a major concern confronting the growth of
manufacturing sector (Fallon & Lucas, 1993; Gupta, Hasan, &
Kumar, 2009).
The preceding discussion highlights that high capital intensity,
accompanied by very high cost of capital, has been one of the major
deterrents to the growth of manufacturing sector in India. The In-
dian economy in general and the manufacturing sector in particular
can benet greatly by any reduction in the overall cost of capital.
One of the most important modes of accomplishing this is through
1
The share of India's manufacturing output to overall GDP has been only 15.8% in
2010e11, as compared to 30% in China, 31% in Korea, 36% in Thailand, 26% in
Malaysia, 25% in Indonesia and 22% in Singapore. Manufacturing sector employs
12% of the Indian workforce or 53 million people (CMIE Business Beacon).
establishing the right balance between various sources of funds.
More precisely, the overall cost of capital can be reduced by
achieving the right combination of debt and equity in the total
capital. In this backdrop, it is very important to identify the most
appropriate combination of debt to equity ratio for different types
of rms in India. This shows that the current study is highly rele-
vant in the current context.
The present study is different from previous studies in three
important aspects. First, it deploys a methodology which can take
into account the non-linear relationship between capital structure
and protability. Second, the current study explains the complex
nature of relationships among size, capital structure, and prot-
ability. This is achieved by considering size as the basis for grouping
rms and then nding the differential impact of capital structure on
protability across rms belonging to different size groups. Finally,
the study deals with an exhaustive datasets which encompass lis-
ted rms from multiple industries and a long time frame. To the
best of our knowledge such a study has never been conducted for
the Indian markets before.
2. Data, methodology and results
2.1. Data collection
The current dataset consists of records on several rm-wide
variables for publicly listed manufacturing rms in India. The
data has been collected from Prowess database. The study period is
from 2005 to 2014.2 The original data was obtained for 3501 pub-
licly listed companies. This dataset has been subject to certain l-
ters. All companies with missing data for any time period have been
deleted.3 Companies with negative total net-worth have also been
deleted. Finally, companies with negative sales and negative total
assets have been deleted. The ltering process yields a nal sample
of 1194 rms.
2.2. Measures
2.2.1. Dependent variable
The dependent variable for the current study is Return on Assets
(ROA). ROA has been taken as prot prior to interest, tax, depreci-
ation, and amortization as a percentage of total assets. This de-
nition of ROA avoids earning manipulations by rms to a great
extent and is also independent of the manner in which protability
is shared among shareholders, debtholders and governments.
2.2.2. Explanatory and control variables
In the current study capital structure (CS) is the major variable of
interest. CS has been dened as long-term borrowings as a per-
centage total assets. CS is also the regime-dependent explanatory
variable in the panel threshold estimation. Size of the rm (SIZE) is
another important variable of interest. SIZE represents the natural
log of total assets. Total assets are originally measured in million of
Rupees. Many previous studies have pointed out that rms with
different sizes have different capital structure (Deesomsak,
Paudyal, & Pescetto, 2004; Wald, 1999). This suggests that there
is a differential impact of debt on protability based on the size of
the rm. This relationship can be exploited by considering size as
the threshold variable. Further, marketing intensity (MI), intangible
2
Indian rms mostly observe April to March scal year in order to match the tax
year.
3
This has been primarily done to avoid the dataset becoming an unbalanced
panel. The threshold panel model applied in this study is applicable to the balanced
panel.
 D. Jaisinghani, K. Kanjilal / Asia Pacic Management Review 22 (2017) 159e165 161

Table 1
Descriptive statistics and correlations.

ROA MI CS SIZE INTI CR INVESTI

Panel A: Descriptive statistics

Mean 13.631 2.596 29.853 7.701 0.486 399.455 6.5916
Median 12.800 1.300 30.300 7.573 0.001 244.100 1.500
Maximum 76.100 79.800 94.700 15.118 57.500 552326.90 94.500
Minimum 26.900 0.000 0.000 1.902 0.000 6.700 0.000
Std. Dev. 7.750 3.827 18.734 1.811 2.424 5086.38 11.374
Panel B: Correlations
ROA e
MI 0.064 e
CS 0.273 0.165 e
SIZE 0.080 0.031 0.095 e
INTI 0.013 0.122 0.039 0.055 e
CR 0.022 0.009 0.033 0.0003 0.003 e
INVESTI 0.025 0.073 0.252 0.209 0.006 0.005 e

intensity (INTI), current ratio (CR), and investment intensity

(INVESTI), have been selected as control variables. MI is measured
ROAit
mi b01 CSit xit ; SIZEit  l (1A)
as sum total of marketing and advertisement expenses as a per- mi b02 CSit xit ; SIZEit > l
centage of net sales, INTI is measured as sum of all intangible assets
as a percentage of total assets, CR is measured as current assets as a In Equation (1) SIZE is considered as the threshold variable. l
percentage of current liabilities and INVESTI is measured as total represents the threshold parameter. CSit is the regime-dependent
investments as a percentage of total assets. explanatory variable. mi s represent cross section specic xed ef-
fect parameter. Zit represents a set of regime-independent
2.3. Descriptive statistics explanatory variables including MI, INTI, CR, INVESTI and SIZE. xit
is the error term having zero mean and a constant variance s2 .
Table 1 reports the descriptive statistics of the variables Equation (1) implies that the observations are segregated into two
considered. The mean value of the dependent variable (ROA) is discrete classes depending upon the conditions of the threshold
13.63 percent and the median is 12.8 percent. Similarly, the mean variable SIZEit being larger or smaller than the threshold value l.
(median) value of CS stands at 29.85 percent (30.3 percent). This b1 and b2 are two slope coefcients in two different regimes.4 In
implies that almost one-third of total assets are nanced through order to estimate these two slope coefcients one has to ensure
long-term borrowings. This suggests that overall cost of capital of a that the elements in CSit are time variant. Also, the threshold var-
rm is greatly impacted by the proportion of long-term debt in the iable SIZEit has to be time variant. The analysis holds good in situ-
overall capital. Such high proportion of long-term borrowings in ations where T is nite and N tends to innity.
the overall capital structures warrants the identication of the In order to estimate the threshold Equation (1), certain modi-
optimal level of debt. Table 1 also reports the correlations among cations have to be made. First, Equation (1) has to be averaged
different variables. The table clearly highlights that none of the over T as shown in the following equation.
correlations, among the explanatory variables, are above 0.3.
Hence, the dataset does not suffer from the problem of acute ROAi mi b0 CSi l g0 Z i xi (2)
multicollinearity.
Fig. 1 (A) and (B) display size and ROA for the years 2005 and where,
2009 respectively for 10 categories of rms which are classied on
the basis of their debt level. The companies in category 1 represent
X
T
rms having the least proportion of debt and the companies in ROAi T 1 ROAit ;
category 10 represent rms having the highest proportion of debt. t1
Fig. 1(A) and (B) show that SIZE and ROA vary widely across the
different categories of rms. Also, there appears to be a prima facie
evidence of non-linear relationship between debt and protability. X
T
CSi l T 1 CSit l;
t1
2.4. Estimation methodology

The article applies Hansen (1999) threshold panel methodology X

T
Z i T 1 Zit ;
for the identication of the optimal level of size which would
t1
impact the protability in a non-linear way as per the rm's capital
structure. The threshold model is developed for balanced panels
with xed effect estimations. The estimation methodology is briey X
T
explained next. xi T 1 xit ; (3)
t1
The equation to be estimated in this case can be written as:
Then the difference between Equations (1) and (2) has to be
ROAit mi b01 CSit JSIZEit  l b02 CSit JSIZEit > l g0 Zit found, which yields the following equation.
xit
(1)
where J() represents the indicator function. The indicator func- 4
The computation of multiple regimes (three or more regimes) is explained in
tion can be alternatively written in the following form. Hansen (1999).
162 D. Jaisinghani, K. Kanjilal / Asia Pacic Management Review 22 (2017) 159e165

Fig. 1. (A): Size and ROA across rms with different debt levels - 2005. The gure presents the average size and protability for rms across different levels of debt. All the rms are
segregated into ten equal groups on the basis of their debt levels. The companies in group 1 represent the rms having the least proportion of debt in their capital structure and
companies in group 10 represent the rms having the highest proportion of debt in their capital structure. Size is measured as natural log of total assets and ROA is measured as
prot before depreciation, interest, tax and appropriations scaled by total assets. The analysis has been conducted for the year 2005. (B): Size and ROA across rms with different
debt levels - 2009. The gure presents the average size and protability for rms across different levels of debt. All the rms are segregated into ten equal groups on the basis of
their debt levels. The companies in group 1 represent the rms having the least proportion of debt in their capital structure and companies in group 10 represent the rms having
the highest proportion of debt in their capital structure. Size is measured as natural log of total assets and ROA is measured as prot before depreciation, interest, tax and ap-
propriations scaled by total assets. The analysis has been conducted for the year 2009.

2 3 2 * 03 2 * 3
ROA*it b0 CS*it l g0 Zit* x*it : (4) ROA*i2 CSi2 l Zi2
ROA*i 4 5 *
; CSi l 4 5; Zi 4 5 and x*i
*
0
where, ROA*iT CS*iT l *
ZiT
2 * 3
xi2
ROA*it ROAit  ROAi 4 5;
x*iT
(5)
CS*it l CSit l  CSi l;
Similarly, all cross-sectional observations can be stacked over
the time periods. Let us represent these matrices by F*, P* l, * ,
Zit* Zit  Z i and
and * respectively. Applying this setting to Equation (4) the
following functional form is obtained.
x*it xit  xi :
F* P* lb * g * (6)
The data as well as the error term can now be stacked for one
cross-section and simultaneously for the rst time period. This will The above equation can be solved for a particular value of l by
yield the following representations, using ordinary least squares (OLS) method. In order to avoid too
 D. Jaisinghani, K. Kanjilal / Asia Pacic Management Review 22 (2017) 159e165
much observations becoming part of one regime, one can specify a
certain minimum number of observations to fall into each regime.
The procedure of parameter estimation and its hypothesis testing
are explained in Hansen (1999). The critical values are derived
using the bootstrap technique.
2.5. Empirical results
The rst step in estimating any panel regression requires
testing the stationarity of variables. In case the data is not sta-
tionary, suitable modications have to be applied to make the data
stationary. Three different tests of panel stationarity have been
conducted to nd out stationarity of the variables. The results
show that all the variables are stationary in their base form and
hence are conducive for threshold panel regression analysis.5 The
next step in threshold model requires dening a threshold vari-
able, a regime-dependent explanatory variable, and regime-
independent explanatory variables. In the current analysis, SIZE
is considered as the threshold variable and CS as the regime-
dependent explanatory variable. MI, INTI, CR, INVESTI and SIZE
are regime-independent explanatory variables. The rationale for
considering SIZE as the threshold variable is that it effectively
captures the ability of a rm to borrow capital from the open
market. Besides, many authors have considered size as one of the
primary variables impacting the capital structure of a rm (Huang
& Song, 2006; Ozkan, 2001; Wald, 1999). There are two important
factors to be considered in a threshold panel regression analysis.
The rst is the p-value of the threshold parameter that is obtained
through the bootstrap procedure. The second is the signicance of
various explanatory variables in the regression equation. The F-
statistics and their associated p-values for different threshold
parameters are estimated by repeating the bootstrap procedure
300 times. The signicance levels of the individual explanatory
variables are estimated by analyzing their coefcients and stan-
dard errors.
Tables 2e4 present the results of the threshold panel analysis.
Table 2 demonstrates the statistical ndings of the threshold
variable SIZE. The F-statistics and associated p-values for single
and double threshold parameters, presented in Table 2, clearly
show that only the single threshold parameter is signicant at the
1 percent level. This implies that there exists a single threshold
level of SIZE beyond which CS may have differential impact on the
protability of rms. The estimated threshold parameter of SIZE is
around 4.9965 in case of the single threshold model. The signi-
cance of single threshold model indicates the existence of two
different regimes. SIZE is calculated as natural log of total assets.
Thus, converting the threshold value of 4.9965 into its base form
yields an estimated size of approximately 148 million Rupees.
Hence, it is clear that the regime dependent variable e CS e can
differentially impact protability of rms with size greater and
less than 148 million Rupees. In the remaining of the paper the
two regimes are dened as large rms and small rms
respectively.
Table 3 reports the results for regime-dependent explanatory
variable CS for two regimes. The table clearly provides evidence
that the magnitude and the sign of the coefcients vary widely
across these two regimes. For the rst regime, which represents
small rms, it is observed that the coefcient of CS is negative and
signicant at the 1 percent level. However, the coefcient of CS for
the second regime, representing large rms, is positive and sig-
nicant at the 10 percent level. The estimated values of CS in two
regimes are 0.0626 and 0.0132 respectively. The analysis of
5
The results of stationarity tests are included in Appendix 1.
163
White's (heteroscedasticity corrected) error terms also yields
similar results. Thus, it is clear that capital structure has differential
impact on protability of rms depending on their size group. For
large rms, higher debt in the capital structure is associated with
improved protability whereas for small rms higher debt would
lead to reduced protability. The results indicate that large
rms are better able to utilize the debt in their capital structure.
The ndings of the study do recommend that there exists a
strong non-linear relationship between debt and protability for
manufacturing rms in India. Moreover, the analysis of two regime-
dependent slope coefcients of CS shows that the marginal impact
of increase in debt on ROA is much higher for the rst regime
(0.0626) than that for the second regime (0.0132). This signies
that debt has higher impact for small rms than for large rms.
Thus, the benets driven from rebalancing the capital structure
would be higher for small rms than for large rms.
Table 4 reports the estimation results of regime dependent
explanatory variables that have been considered as control vari-
ables in the model. The table demonstrates that MI and INVESTI are
having negative and signicant impact on ROA whereas the co-
efcients of INTI and CR are not statistically signicant. The nega-
tive impact of MI on ROA implies that rms spending heavily on
marketing efforts are not able to generate excess protability. The
result that INVESTI is negatively associated with protability im-
plies that manufacturing rms in India are not properly managing
their investments and hence they should focus more on their core
business.
3. Conclusion
3.1. Discussion
The current study analyzes the non-linear relationships among
size, capital structure and protability for the Indian manufa-
cturing rms. Threshold panel regression method has been
deployed for estimating the results. The results show that there
exists a single threshold level of SIZE, which determines the
non-linear impact of capital structure on protability for
manufacturing rms in India. Specically, it is observed that rms
with assets more than (less than) 148 million rupees are positively
(negatively) impacted by increasing the debt in their overall
capital. One possible reason for the differential impact of capital
structure on the protability of a rm based on its size is that the
increased level of debt is associated with increased monitoring
expenses and imposing of certain covenants by the lenders. These
expenses may arise due to the appointment of additional inde-
pendent directors and professional auditors. Another possible
reason for these ndings can be that the small rms are generally
the young rms. The projects undertaken by these rms require
continuous investments before they can reach the break-even
stage and start generating positive cash ows. These small and
young rms often nd it difcult to invest more funds in these
projects and simultaneously pay their interest obligations on
time. This may lead to a negative association between capital
structure and protability for the small rms.
The results of this study contradict the ndings of Lin and Chang
(2011); Cheng, Liu, and Chien (2010) and Cuong and Canh (2012)
who have established the existence of multiple threshold levels
of debt for rms operating in Taiwan, China and Vietnam respec-
tively. However, the results are similar to the ndings of Ahmad and
Abdullah (2013) who demonstrate the presence of a single
threshold debt level for rms operating in Malaysia. The results also
support the ndings of Jaggi and Gul (1999) that size acts as the
moderating variable for the relationship between capital structures
and free cash ows for US rms. The overall ndings of the present
164 D. Jaisinghani, K. Kanjilal / Asia Pacic Management Review 22 (2017) 159e165

Table 2
Test for threshold effect between size, capital structure and ROA.

Single-threshold effect test Double-threshold effect test

Threshold value F p-value Threshold values F p-value

ROA 4.9965 30.4225*** 0.0100 4.3758 4.9965 8.4546 0.7400

The above table presents the results of the tests of signicance of the threshold values. The F-statistics and their corresponding p-values are obtained by repeating the
bootstrap process 300 times for each case. *** indicates signicance at the 1% level.

Table 3
Threshold panel regression results.

Variable (CS) Coefcients OLS-SE t-stats p-value White-SE t-stats p-value

Regime 1 0.0626 0.0154 4.0653*** 0.0000 0.0195 3.2044*** 0.0014

Regime 2 0.0132 0.0070 1.8713* 0.0613 0.0084 1.5722 0.1160

The above table presents the result of the panel threshold regression analysis. Capital structure (CS) is the regime-dependent explanatory variable. The impact of CS is
measured for two separate regimes based on the threshold value of the threshold parameter. ***, ** and * indicate signicance at the 1%, 5% and 10% levels respectively.

Table 4
Threshold panel regression results.

Variable Coefcients OLS-SE t-stats p-value White-SE t-stats p-value

MI 0.0877 0.0331 2.6511*** 0.0080 0.0359 2.4415** 0.0147

SIZE 2.2726 0.1245 18.2481*** 0.0000 0.1311 17.3299*** 0.0000
INTI 0.0073 0.0441 0.1648 0.8692 0.0431 0.1686 0.8661
CR 0.0001 0.0001 0.7142 0.4752 0.0002 0.4210 0.6745
INVESTI 0.0484 0.0106 4.5635*** 0.0000 0.0117 4.1355*** 0.0000

The above table presents the result of the panel threshold regression analysis. The explanatory variables are regime independent and hence their impact is uniform across all
the regimes. ***, ** and * indicate signicance at the 1%, 5% and 10% levels respectively.

study go in favour of the trade-off and asymmetric information
theories of capital structure.
3.2. Research implications
The results obtained in the current study bear certain useful
insights for managers of rms. The optimal level of capital struc-
ture conveys that large sized rms are benetted by issuing more
debt. This nding can be deployed by very large manufacturing
rms that currently have very low debt levels. The rms can in-
crease their protability by having more debt in their total capital.
Also, these rms can convey positive signal to the market by
issuing more debt. Another useful implication is related to the
capital budgeting decisions of large rms. The capital budgeting
decisions are based primarily on comparison between the cost of
capital and the expected returns from a project. The evidence
presented in this study implies that, in addition to the cost of
capital analysis, managers of large rms should also consider the
divergence from the optimal capital structure while making the
capital budgeting decisions.
The results also bear certain critical implications for the
managers of small rms. The empirical results highlight the fact
that small sized rms are negatively impacted by the presence of
high levels of debt. This implies that small rms are having
higher implicit cost of debt capital compared to large rms.
These ndings, combined with the fact that manufacturing rms
in India are highly capital intensive, convey that small rms can
greatly benet by reducing the debt in their overall capital. This
can be done by focusing on equity capital as the major source of
funds. Also, small rms can opt for nancing their capital re-
quirements through hedge funds and venture capital funds.
These nancing strategies will help small rms to reduce their
overall cost of doing business and thereby generating positive
economic value.
Acknowledgement
The authors gratefully acknowledge valuable comments offered
by the anonymous referees and the editor of the journal. The au-
thors express sincere gratitude for the review, which have helped to
improve the quality of the paper signicantly.
Appendix 1

Table A1
Results of Tests of Stationarity.

Variable IPS p-value ADF - Fisher p-value PP - Fisher p-value

ROA 15.763*** 0.000 3649.150*** 0.000 4330.800*** 0.000

MI 18.935*** 0.000 3741.220*** 0.000 4448.740*** 0.000
CS 1722.130*** 0.000 3248.210*** 0.000 3990.370*** 0.000
SIZE 8.898*** 0.000 3654.760*** 0.000 6064.170*** 0.000
INTI 23.163*** 0.000 1927.980*** 0.000 2210.330*** 0.000
CR 24.221*** 0.000 3948.820*** 0.000 4642.330*** 0.000
INVESTI 31.779*** 0.000 3315.150*** 0.000 3896.020*** 0.000

The above table presents the results of three alternative tests of stationarity in panel datasets. The null hypothesis is each of the three tests is that the underlying series is non-
stationary. The rejection of the null hypothesis indicates that the underlying series is stationary. *** indicates signicance at 1% level.
 D. Jaisinghani, K. Kanjilal / Asia Pacic Management Review 22 (2017) 159e165 165

References free cash ows and size on corporate debt policy. Review of Quantitative Finance
and Accounting, 12(4), 371e381.
Kester, W. C. (1986). Capital and ownership structure: A comparison of United
Ahmad, A. H., & Abdullah, N. A. H. (2013). Investigation of optimal capital structure
States and Japanese manufacturing corporations. Financial Management, 15,
in Malaysia: A panel threshold estimation. Studies in Economics and Finance,
5e16.
30(2), 108e117.
Kraus, A., & Litzenberger, R. H. (1973). A state-preference model of optimal nancial
Antoniou, A., Guney, Y., & Paudyal, K. (2008). The determinants of capital structure:
leverage. Journal of Finance, 28(4), 911e922.
Capital market-oriented versus bank-oriented institutions. Journal of Financial
Lee, Y., Su, S., & Lin, W. (2012). Capital structure timing in markets with different
and Quantitative Analysis, 43(01), 59e92.
characteristics. International Journal of Business and Finance Research, 6(3),
Arosa, C. M. V., Richie, N., & Schuhmann, P. W. (2014). The impact of culture on
53e67.
market timing in capital structure choices. Research in International Business and
Lin, F. L., & Chang, T. (2011). Does debt affect rm value in Taiwan? A panel
Finance, 31, 178e192.
threshold regression analysis. Applied Economics, 43(1), 117e128.
Bayrakdaroglu, A., Ege, I., & Yazici, N. (2013). A panel data analysis of capital
Miller, M. H. (1977). Debt and taxes. Journal of Finance, 32(2), 261e275.
structure determinants: Empirical results from Turkish capital market. Inter-
Myers, S. C., & Majluf, N. S. (1984). Corporate nancing and investment decisions
national Journal of Economics and Finance, 5(4), 131.
when rms have information that investors do not have. Journal of Financial
Cheng, Y. S., Liu, Y. P., & Chien, C. Y. (2010). Capital structure and rm value in China:
Economics, 13(2), 187e221.
A panel threshold regression analysis. African Journal of Business Management,
Ozkan, A. (2001). Determinants of capital structure and adjustment to long run
4(12), 2500e2507.
target: Evidence from UK company panel data. Journal of Business Finance &
Cuong, N. T., & Canh, N. T. (2012). The effect of capital structure on rm value for
Accounting, 28(1e2), 175e198.
Vietnam's seafood processing enterprises. International Research Journal of
Paligorova, T., & Xu, Z. (2012). Complex ownership and capital structure. Journal of
Finance and Economics, 89, 221e233.
Corporate Finance, 18(4), 701e716.
De Jong, A., Verbeek, M., & Verwijmeren, P. (2010). The impact of nancing sur-
Psillaki, M., & Daskalakis, N. (2009). Are the determinants of capital structure
pluses and large nancing decits on tests of the pecking order theory. Financial
country or rm specic? Small Business Economics, 33(3), 319e333.
Management, 39(2), 733e756.
Rajan, R. G., & Zingales, L. (1995). What do we know about capital structure - some
Deesomsak, R., Paudyal, K., & Pescetto, G. (2004). The determinants of capital
evidence from international data. Journal of Finance, 50(5), 1421e1460.
structure: Evidence from the Asia pacic region. Journal of Multinational
Ross, S. A. (1977). The determination of nancial structure: The incentive-signalling
Financial Management, 14(4), 387e405.
approach. Bell Journal of Economics, 8(1), 23e40.
Dudley, E. (2012). Capital structure and large investment projects. Journal of
Wald, J. K. (1999). How rm characteristics affect capital Structure: An international
Corporate Finance, 18(5), 1168e1192.
comparison. Journal of Financial Research, 12(2), 161e187.
Fallon, P. R., & Lucas, R. E. (1993). Job security regulation and the dynamic demand
for labor in India and Zimbabwe. Journal of Development Economics, 40(2),
241e275.
Ferri, M. G., & Jones, W. H. (1979). Determinants of nancial structure: A new Dinesh Jaisinghani is currently working as a Doctoral Scholar at International Man-
methodological approach. Journal of Finance, 34(3), 631e644. agement Institute (IMI), New Delhi, India. He has completed his graduation from
Guha-Khasnobis, B., & Bhaduri, S. N. (2002). Determinants of capital structure in University of Mumbai and his post graduation (MBA) from FORE School of Manage-
India (1990e1998): A dynamic panel data approach. Journal of Economic Inte- ment, New Delhi. His major areas of interest are Corporate Finance, Financial Econo-
gration, 17(4), 761e776. metrics, Market Microstructure and Financial Economics. He has published papers in
Gupta, P., Hasan, R., & Kumar, U. (2009). Big reforms but small payoffs: Explaining national and international journals including Journal of Indian Business Research, In-
the weak record of growth in Indian manufacturing. India Policy Forum, 5, ternational Journal of Business and Emerging Markets, International Journal of Busi-
59e108. ness Competition and Growth, Emerald Emerging Markets Case Studies among others.
Handoo, A., & Sharma, K. (2014). A study on determinants of capital structure in
India. IIMB Management Review, 26(3), 170e182.
Hansen, B. E. (1999). Threshold effects in non-dynamic panels: Estimation, testing, Dr. Kakali Kanjilal is currently working as Associate Professor with International
and inference. Journal of Econometrics, 93(2), 345e368. Management Institute (IMI), New Delhi, India. She has more than 15 years of industry,
Hovakimian, A. (2004). The role of target leverage in security issues and teaching and research experience. She has published research articles in many
repurchases. The Journal of Business, 77(4), 1041e1072. renowned international journals including Economic Modelling, Macroeconomics and
Huang, G., & Song, F. (2006). The determinants of capital structure: evidence from Finance in Emerging Market Economies, Energy Policy, Energy Economics, Pacic Asian
China. China Economic Review, 17(1), 14e36. Journal of Energy among others.
Jaggi, B., & Gul, F. (1999). An analysis of joint effects of investment opportunity set,