Effect of Capital Structure on Firms' Product Market Performance: Empirical Evidence from

Indian Manufacturing
Author(s): Arindam Bandyopadhyay
Source: Economic and Political Weekly, Vol. 40, No. 9 (Feb. 26 - Mar. 4, 2005), pp. 866-876
Published by: Economic and Political Weekly
Stable URL: http://www.jstor.org/stable/4416277 .
Accessed: 14/07/2014 06:07
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Economic and Political Weekly is collaborating with JSTOR to digitize, preserve and extend access to
Economic and Political Weekly.

http://www.jstor.org

This content downloaded from 111.68.103.104 on Mon, 14 Jul 2014 06:07:32 AM

All use subject to JSTOR Terms and Conditions

Effect
of Capital Structure on
Firms'
Product Market Performance
Empirical Evidence

from

Indian

Manufacturing

This paper empirically examines the effect of a firmn'scapital structureon its product market
outcome. The strategic consideration in the product marketmay induce Indian corporates to
take on higher debt in order to gain strategic advantage. Using a balanced panel of 538
Indian manufacturingfirmnsover the period 1989 to 2000, the paper studies the relationship
between short- and long-term debt and sales performance(export as well as total sales). In the
case of long-term debt, firms take time to build their infrastructure.Hence, considering a
longer time horizon of two years and seven years of taking the loan, the paper finds that
long-term debt boosts sales growth of firms belonging to the top 50 and large business
houses. However, long-term debt is inconsequentialfor total growth of sales for smaller group
and unaffiliatedfirms. The studyfinds empirical evidence on the interplay between the
financial structure and product marketperformance in the Indian corporate sector.
ARINDAMBANDYOPADHYAY

Hypotheses

ntil the late 1980s, the corporatefinanceliteraturehas

ignored the interactionbetween capital structureand
firm'sproductmarketdecisions.The lack of interestin
decisions is due to the 'irrelevancepropothese output-related
sition' of finance theory. This proposition, known as the
(1958) theorem,postulatesthatthe choice of
Modigliani-Miller
a firm'scapitalstructure(i e, its financingdecisionof holding
debtorequity)is irrelevantfor its value.In manycircumstances,
wherethe productmarketis imperfectlycompetitiveand firms
have some marketpower,the irrelevancewouldbe brokenand
financialstructureand outputmarketdecisionswouldbe interrelated.Currentresearchin corporatefinancehasbegunpointing
betweenthe financialandrealdecisions
to a directinterrelation
of firms. That interrelationcomes from the role of financial
in conveyinginformationto investorsas well as the
instruments
productmarketrivalsandconsumers.The literaturestressesthat
a firm's mode of financinginfluencesboth the firm's conduct
in the productmarketas well as the conductof other market
participants,therebyinfluencingcompetitiveoutcomes.Harris
andRaviv(1991) makethis pointanddiscussrecenttheoretical
work which model productmarketand capitalstructureinterontheinteraction
actions.Ravid(1988)alsosurveystheliterature
betweencapitalstructureand productmarketdecisions. Both
these surveysidentifytwo types of interactions:the effect on
a firm'sproductmarketstrategyandtheeffecton productchoice.
Titman(1984) and Maksimovicand Titman(1991) show how
thecapitalstructurecan affecta firm'schoiceof productquality
and the viabilityof its product'swarranties.Thus, the capital
structurecan alter a firm's ability to compete in the product
market.Recentempiricalevidenceon how financingdecisions
affect productmarketcompetition among firms has further
stimulatedinterestin the area.
866

In theory,therearetwo schoolsof thoughton the interaction

betweenfirmproductmarketstrategyandits financingchoices.
One believes that a firm's debt issue can lead to stiffercompetitionin the productmarketby raisingits outputin a strategic
way.BranderandLewis(1986and1988)andMaksimovic(1988)
in theirpioneeringwork analysehow debt financingcommits
a firmto a moreaggressiveoutputstancein the productmarket.
Rotembergand Scharfstein(1990) and Bolton and Scharfstein
(1990) also predictthat increaseddebt will lead to increased
outputatthefirmlevel andattheindustrylevel andtherebymake
the competitionstiffer.
Anotherline of argument,put forth by Telser (1966) and
extendedby Bolton and Scharfstein(1990) suggests that too
muchdependenceon outsidefinancinghindersa firm's ability
to compete,promptingindustryrivalsto pursuepredatorymarket
strategies.Chavelierand Scharfstein(1996) proposethatexternallyfinancedfirmsinvestless in marketsharebuildingduring
recessions,raisingpricecost marginsto boostshort-termprofits
at the expense of locked-incustomers.The othermore recent
papersby Showalter(1995), Dasguptaand Titman(1998) and
Grimaud(2000) have shown thatdebt leads to weakercompetitionin the outputmarketby helpingfirmscolludeandincrease
their prices while cuttingoutput.
The limitedamountof empiricalliteratureinvestigatingthe
links between capital structureand productmarketdecisions
consists of the following papers:Showalter(1999) studiesthe
industries.Basedon
strategicuse of debt in US manufacturing
his own theoreticalworkShowalter(1995) andthatof Brander
andLewis (1986), he regressesthe debtratiosof manufacturing
firmson variablesapproximating
demandand cost uncertainty
as well as a set of controlvariables.He finds significantrelationshipsbetweenthe uncertaintymeasuresandfirmsleverage,
and can supportthe hypothesisof strategicuse of debt.
Chevalier(1995)considersthesupermarket
industryin theUS
duringthe late 1980s. In an event studyanalysis,she finds the
Economicand PoliticalWeekly Febraury26, 2005

This content downloaded from 111.68.103.104 on Mon, 14 Jul 2014 06:07:32 AM

All use subject to JSTOR Terms and Conditions

announcement of a leveraged buyout within the industry to

increasethe expected profitof rivals. However, leveraged buyouts
encourage local entryand expansion by rivals. Her results suggest
a higher leverage to soften product market competition. These
findings contrast with the qualitative results found in the theoretical literature following Brander and Lewis (1986).
Opler and Titman (1994) find that during industry downturns
highly levered firms are the most vulnerable. They find that firms
with higher levels of debt lose more sales and market share than
their more conservatively financed competitors. Thus, these
empirical results seem to be at odds with the predictions of the
theoretical models. Philips (1995) also provides evidence that
financial leverage interacts with product market competition. In
his intra-industryanalysis about US markets, he tests whether
industryoutput is affected by changes in firms' capital structure.
He finds a change in the firms' marketshare following an increase
in financial leverage. Depending on certain industry characteristics, however, this effect goes in opposite directions. This study
shows the importance of including industry specific characteristics both on the supply and demand side to understandthe firms'
capital structuredecisions. Hellmann and Puri (1999) investigate
the relationshipbetween the type of capital that new firms choose
to finance their projects and their product market strategies, and
the corresponding market outcomes. Their work is based on a
unique data set of 173 start-upcompanies in California's Silicon
Valley. They find firms pursuing innovator strategies being more
likely to use venturecapital financing. Also, they are much faster
in bringing new products on the market than imitator firms.
As one can see, the relevant empirical works focusing on capital
structureand product market interactions are limited to US firms
operating in various industries in various cities of US. The
informationalproblemis much more severe for developing country
firms (like Indian firms) competing in the product market. Das
and Bandyopadhyay (2003) and Bandyopadhyay and Das
(forthcoming) provide empirical evidence about the linkage
between the firm's financing decisions and real market performance for Indianfirms operatingin foreign anddomestic markets.
They find that the issuance of commercial papers or debentures
by firms leads to betterperformance in the product market;while
it directly impact domestic sales of the firms, it also acts as a
signal and stimulates its foreign sales.
The empirical literatureon interactionsbetween firm financing
and product market performance generally seeks to determine
whether debt financing either hurts or boosts competitive performance. Short-termand long-term debt contracts involve tradeoffs for entrepreneurs.Short-termdebt comes with lower interest
chargesattached.However, anentrepreneurmustgenerallyproduce
positive results within a year or two. If these results don't
materialise, the entrepreneurmay default on the loan and there
may be a shift in control to the investors. In this case, shortterm debt is a powerful disciplining device for good firms,
because it gives control back to the investor. Therefore, the shortterm forces an entrepreneurto abandon his or her dream if the
business is realistically doomed to failure. It is a way of committing the entrepreneurto a realistic viewpoint. Thus, short-term
leverage provides a firm with an incentive to perform better in
the product market [Poitevin 1989]. In contrast, long-term debt
gives the entrepreneurmore time to make his or her company
successful and pay back the debt, with the trade-off being higher
interest payments. Therefore, firms taking long-term debt take

and distributionnetworksto gain strategicadvantageover the

longer periodof time.
Myiworkproposesthatdebt financingmust have a positive
influence on firm's competitiveperformancein the product
market.To motivatethis claim, I empiricallyexaminethe relationshipbetweenshort-and long-termcorporatedebtandsales
performance
(exportas well as totalsales)usingIndianfirmlevel
datafroma panelof 533 firmsover 12 years.However,a study
of competitiveperformancefollowingcapitalstructurechanges
may suffer from a potentialendogeneityproblemwhich first
needs to be taken care of. Anotherconcern is there may be
unobservedfactorsarisingfromthe marketenvironmentwhich
mayjointly influenceboth a firm's financialstructureand its
competitiveperformance.One way to mitigateconcernsabout
the endogenousnatureof the relationbetweencapitalstructure
and productmarketperformanceis to look at the real market
performancechangesfollowing the changes in financialdecisions.Accordingly,we performsome univariateparametric
and
teststo examinetheconsequencesof takingboth
non-parametric
short-and long-termdebton variousreal marketvariableslike
exports,total sales, advertisingintensity,marketingintensity,
R&D intensityanddistributionintensity.These univariatetests
will tell us how firms(if they are actually)can use theirshortand long-termloans to gain strategicadvantagein the product
market.Next, in my multivariateanalysis of productmarket
impactof long-termdebt, I use lag structuresto mitigatethe
simultaneityproblems.The unobservedfirm specific factors
whichmaydisturbtheerrorstructure
(withinfirmautocorrelation)
are being correctedin my panel Tobit model by incorporating
21 industrydummiesand three group dummies.The yearly
changesare also being capturedby taking12 yearlydummies.
Similarly,in studyingthe statisticalsignificanceof interactions
betweenfirms'short-term
debtfinancingandexportperformance
I applya two-stepGMMestimationmethodto takecareof the
possible endogeneityof short-termleverage.
This paperis organisedas follows. In Section II, I briefly
discuss the corporatefinancingpatternin Indiaduringthepostreformperiod.Here,I also talkaboutdevelopmentsin theIndian
financialmarket,which providesa backgroundto my study.
Section III discussesthe data,summarystatisticsand variable
construction
methods.There,I alsolookatthetrendsin financing
of
pattern mysamplefirmsovertheperiod1989-2000.SectionIV
containsthe methodologyof differenteconometrictestsandthe
mainempiricalresultsof thechapter.Usingunivariate
parametric
andnon-parametric
testsandmultivariate
panelTobitandGMM
regressions,I have madean attemptto find empiricalevidence
of theeffectsof capitalstructure
onproductmarketswithparticular
referenceto the Indiancorporatesector.In SectionV, I discuss
the majorfindingsof my empiricaltests and conclude.

II
IndianFinancialSector:Some RelevantIssues
Corporate Financing Pattern in India

Indiahas historicallyfolloweda financialintermediary

based
system, where banks and financial intermediariesplayed a
dominantrole.Thecorporatefinancingpatternin Indiaindicates
that,on average,internalsourcesconstituteaboutone-thirdof
total sourcesof funds, while externalsourcesaccountfor the
time to buildup theirinfrastructure,
R&D, marketingchannels rest.As far as the RBI Report,Trendand Progressof Banking

Economicand PoliticalWeekly February26, 2005

This content downloaded from 111.68.103.104 on Mon, 14 Jul 2014 06:07:32 AM

All use subject to JSTOR Terms and Conditions

867

in India2001-02 is concerned,the shareof borrowingsin total

sources(allsourcesof financefromthecapitalmarket)hasmoved
inverselywithequityfinancingin thepost-reformperiod(1991).
The mainsourcesof long-termdebtaredevelopmentfinanceby
the all-Indiaand state financial institutions(DFIs), who lend
mainly for investmentin the prioritysector, and debentures
(cumulativeand non-cumulative),which is a capital market
instrument.
Short-termfinancing,mainlyfor workingcapitalpurposes,is
usuallyprovidedby thecommercialbanksas a mixof cashcredit
andbillsdiscountingfacilities.Commercialpaper(CP)cameinto
existencefollowinga RBI notificationin 1990 as a new shorttermdebtinstrument.
CPsusuallyhasa maturityof 90 days.CPs
can also be issuedfor maturityperiodsof 180 andone yearbut
the most active marketis for 90 day CPs.
Duringthe early 1990s, the Indianfinancialsector was undergoingimportantchanges.The bankingsectorreforms,with
the publicationof the NarasimhamCommitteeReport,aimedat
increasingthe profitabilityandefficiency of the then27 public
sectorbanks.Entryderegulation
wasaccompanied
byprogressive
of
rates
on
interest
deregulation
depositsandadvances,reduction
of reserverequirements
andremovalof creditallocation.Strengtheningfinancialsystemshasbeenoneof thecentralissuesbecause
soundfinancialsystemsdrivecompetitiveefficiencyin the real
sectorsof theeconomy.Thus,theprincipalobjectiveof financial
sectorreformwas to improvemobilisationof financialsavings,
puttingthemto productiveuse, transformingvariousrisks and
acceleratingthe growthprocessof the real sectorby removing
structuraldeficiencies affecting the performanceof financial
institutionsand financialmarkets.
FromOctober1994,interestrateswerederegulatedin a phased
mannerandby October1997,bankswereallowedto set interest
rateson all termdepositsof maturityof morethan30 days and
on all advancesexceedingRs 2,00,000.Threemajorcreditrating
agencieshadbeen set up by the early 1990s.1The Securityand
ExchangeBoardof India(SEBI) was given regulatorypowers
in 1992to overseethefinancialmarketsanda newstockexchange
was set up in 1994 (NationalStockExchange).Withthis background,I will examinewhether,in the post-reformperiod,both
the short-and long-termdebt can contributeto promotingcorporateperformancein the productmarkets.

general.Similarly,intheforeignmarket,a morepracticalmeasure
wouldbe to look at the changesin its exportshareas proportion
of totalsales.I use a firm'stotalsalesgrowthatthecurrentperiod
to gauge its performancein the productmarketin general,and
in theexport
exportintensityto salesto determineitsperformance
market.
Proxies for Capital Structure
Capitalstructureis definedby short-and long-termleverage.
Short-termleverageis the ratioof short-termdebtto totalassets.
Long-termleverageis the ratioof long-termdebtto totalassets.
Total debt is total borrowingsof firms. In Prowess,total borrowingsincludeall formsof debt-interestbearingor otherwise.
All securedand unsecureddebt is includedunderborrowings.
Thus,borrowingsincludedebtfrombanks(bothshort-andlongloans,fixeddeposits
term)andfinancialinstitutions,
inter-corporate
frompublicanddirectors,foreignloans,loansfromgovernment,
etc. Fundsfrom the capital marketthroughthe issue of debt
instrumentssuch as debentures(bothconvertibleand non-convertible)and commercialpaperare also includedhere.
I defineshort-termdebtas the loans of shortmaturityof less
thanone year. AccordinglyI take short-termbankborrowings
since theyhave a maturityof less thana year.I havealso added
commercialpaper,whichis a relativelynew typeof debtinstrument throughwhich corporatessource their short-termfund
requirements.The currentportion of long-termdebt is also
includedin generatingthe short-termdebt variable.This is the
amountof long-termdebtdue for repaymentwithin12months.
It measuresthe fundsneededfor repaymentof debtin the near
future.
Long-termdebtsarethoseloanshavinga maturityof morethan
one year.I subtractshort-term
debtfromtotaldebtto derivelongtermdebt. In AppendixA, I discuss in detail the construction
of these financialvariables.
Control Variables

Profitabi
of salesgrowth
lityandinvestmentcanbedeterminants
andexportgrowthandmay be correlatedwith leverage.Therefore,one shouldcontrolfor profitabilityin anyempiricalmodel
designedto measuretheeffect of debton salesor exportgrowth.
III
Similarly,leveragecoefficientsmaybe biasedif the modelfails
to control for investmentspending,which might have been
ResearchDesign
financedwith debt. Throughoutthis chapter,firmprofitability
(proxiedby cash profitover total assets), investment(proxied
Proxies for Product Market Performance
by growthin fixed assets,net of revaluedreservesover assets),
In examiningthe link betweena firm's productmarketper- andsize (proxiedbynaturallog of totalassets)areusedascontrols
formanceandcapitalstructure,previousempiricalresearchhas in regressionof sales growthor exportgrowthon both shortoften linkedprice-settingbahaviourwith some aspect of debt and long-termleverage.
financingto reflect how a firm's financial status affects its
competitivebahaviour[e g, Chavelier 1995; Phillips 1995; Data
Variable Construction
ChavelierandScharfstein1996].However,firmscanimplement and Description,
Statistics
Summary
a numberof alternativepoliciesthatsignificantlyaffectproduct
marketoutcomesbutthatmaynotbe reflectedinhowtheychoose
Thedataareretrievedfromthe Prowessdatabaseprovidedby
to pricetheirproducts.Examplesof such policies aredecisions the Centrefor Monitoringthe IndianEconomy(CMIE).Firms,
aboutfixedinvestments,researchanddevelopmentexpenditure, whichweredropped,includefirmswithoutthe basicdatafrom
advertising,promotionand distributionactivities.One way to 1989 to 2000 and without any industrycategory.Further,I
build a practicalmeasureof performancethat summarisesin- droppedmany firms with zero wages and salaries but with
formationfromthecombinedeffectsof pricingandotherproduct positive sales. All these correctionsresultedin a final sample
marketstrategiesis to look at the firm's total sales growthin of 538 firms.This final sampleincludes242 independentand

868

Economic and Political Weekly

This content downloaded from 111.68.103.104 on Mon, 14 Jul 2014 06:07:32 AM

All use subject to JSTOR Terms and Conditions

Febraury 26, 2005

small group firms and 296 top 50 and large business groupaffiliatedfirms.
Table 1 shows the comparisonbetween top 50 and large
businesshouseswiththeirprivatestand-alonecounterparts.
The
in
firms
on
and
business
are
average bigger size
top
large
group
(intermsof totalsalesandtotalassets)incomparisonwithsmaller
firms.Boththeparametric
univariate
groupandprivatestand-alone
ranksumtestconfirmthatthedifference
t-testandnon-parametric
in size is significant.Thereis alsosignificantdifferencein average
sales growthdistributionbetween two categoriesof firms.
FromTable1, one can also see thattherearesomeremarkable
differencesin the compositionof corporatedebtin capitalstructurebetweengroupaffiliationsand stand-alonefirms.The top
groupfirmson averageare moredependenton long-termloan
comparedto the smallergroupfirms.On theotherhand,smaller
groupandprivatefirmshave moreleveragetowardsshort-term
Table 1: Descriptive Statistics: Average Comparison
between Top 50 Business Group vs Non-top 50 and Private
Stand-alone Firms
(Unitsin Rs Million,others in numbers)
Allfirms Top 50 Non-top50 t statistics
Business Business
for
Group
Group Difference
Firms
Firms
Panel A: mean difference
Exportsales ratio(in per cent)
Annualsales growthat time t
SHORTLEVa
LTLEVb
INVEST$
PROFITABILITY@
LNASSETS#
LNSALES#

7.3
0.19
0.15
0.26
0.1
0.06
6.65
6.64

6.52
0.19
0.13
0.28
0.11
0.07
7.28
7.24

8.26
0.19
0.17
0.24
0.95
0.05
5.88
5.88

-4.7'**
0.07
-12.42**
7.92***
3.21***
5.73***
42.49***
42.17***
Wilcoxon
z-statistics
fordifference
in
distribution

Panel B: mediandifference
Exportsales ratio
(in per cent)
Annualsales growthat time t
a
SHORTLEV
LT_LEVb
INVEST$
PROFITABILITY
LNASSETS#
LNSALES#

loans. The averageprofitabilityand investmentare higherfor

top groupcomparedto privategroupfirms.
Table 2 shows that the most remarkabledifferencebetween
thetopgroupandstand-alonefirmsis themuchhighercoefficient
of variationof exportintensityand growthof sales for smaller
groupandprivatestand-alonefirms.Therefore,I needto control
for groupaffiliationin assessing firm's performance.
Table 3 gives a correlationmatrixof the main variablesfor
the differentgroupsof firms.For both the top groupand nontop groupfirms,short-termleverageis positivelyrelatedto the
export-salesratio. Similarly,long-termleverage is positively
relatedto growthof sales. This is the firstindicationthatfirms'
capital structureis related to real marketperformance.The
profitabilityis inverselyrelatedwith bothshort-and long-term
leverageimplyingthat firms with good growthprospectswill
exhausttheirinternalsourcesof fundsbeforesolicitingoutside
financing.However,in all the cases the correlationcoefficients
betweenindependentvariablesare not high enoughwhichmay
causemulticollinearity
whenI takethemtogetheras regressors.
In every regression,I have checked the correlationsamong
andthetestsreject
independentvariablesalongwithinstruments
the presenceof multicollinearityproblem.
In Table 4, my samplefirms have been classified under21
industrycategoriesaccordingto theirbusinessactivities.Here,
I have harmonisedCMIEindustrycategorieswith NIC 2 digit
Thetablealsodisplays
industry
categoryforindustryclassification.
thenumberof firmsundereachindustrycategory.LaterI control
for industryeffects in assessing the role of long-termdebt on
firm performance.
Trends in Financing during 1989-2000
InTables5 and6, I recordcertaintrendsin corporatefinancing
over the sample period. In Table 5, I look at the long-term
financingtrendsbetweentheperiods1989to 1995andthen1995
Table 2: Sample Descriptive Statistics

1.53
0.135
0.126
0.23
0.06
0.07
6.59
6.59

2.06
0.14
0.114
0.24
0.07
0.07
7.24
7.27

0.78
0.13
0.147
0.2
0.05
0.06
5.81
5.93

8.67***
2.34***
-11.2**
9.27***
8.27***
5.7***
38.63***
33.9***

(Unitsin Rs Million,others in numbers)

Mean

Std Dev

CVa

Allfirms
EXPSLRP
7.3
14.76
2.02
GRSALES
0.2
0.79
3.95
SHORTLEV
0.15
0.14
0.93
LT_LEV
0.26
0.22
0.85
INVEST
0.1
0.19
1.9
Notes: z-statisticdenotestheoutcomeofa Wilcoxonrank-sum(Mann-Whitney) PROFITABILITY
0.06
0.11
1.83
test of equalityof distributionbetween series.
LNASSETS
6.65
1.48
0.22
a Short-termleverage measures the short-termdebtof a companyas
LNSALES
6.64
1.44
0.22
a fractionof itstotalassets. Short-term
debtconsistsofshort-term
bank Top 50 and large business group
+commercialpaperloan+currentportionof long-termdebt.
borrowing
EXPSLRP
6.52
1.65
10.78
b Long-termleverage measures a firm'sindebtness towards longGRSALES
0.19
0.54
2.84
termdebt as proportionof its total assets. Long-termdebt is total
SHORTLEV
0.13
0.11
0.85
borrowing- short-termdebt.
LT_LEV
0.28
0.24
0.86
$ AnnualInvestmentas proportionof Assets: change in gross fixed
INVEST
0.11
0.19
1.73
assets (one periodlag) net of revaluedreserves over totalassets.
PROFITABILITY
0.07
0.11
1.57
@AnnualProfitability:
Cash Profit/Total
Assets. Cashprofitis measured
LNASSETS
7.28
1.43
0.2
as net profit+ depreciation+amortisation.
LNSALES
7.24
1.36
0.19
# LNSALESis naturallog of totalsales; LNASSETSis naturallog of Non-top50 and privatestand-alonefirms
total assets. We have taken either of thermas proxyfor firmsize
EXPSLRP
8.26
18.49
2.24
(FSIZE).TotalAssets includefixedassets, investments,andcurrent GRSALES
0.19
1.01
5.32
assets.
SHORTLEV
0.17
0.16
0.94
c The observations are separated into top 50 business and large
0.24
LT_LEV
0.2
0.83
business groupaffiliatedand theirnon-top50 groupcounterparts. INVEST
0.09
0.19
2.11
The WilcoxonRank-sumtest is a non-parametrictest. The null
PROFITABILITY
0.05
0.12
2.4
hypothesisis thatvariablesinbothgroupsare frompopulationswith
LNASSETS
5.88
1.15
0.2
the same distributionand the same medians.
LNSALES
5.89
1.17
0.2
***,**denote significanceat the 5 per cent or better,5-10 per centlevel, respectively.
Note:a CVis coefficientof variation= Std dev/Mean

Economic and Political Weekly

February 26. 2005

This content downloaded from 111.68.103.104 on Mon, 14 Jul 2014 06:07:32 AM

All use subject to JSTOR Terms and Conditions

Min

Max

0
-1
0
-0.89
-4.61
-2.51
1.82
0

100
30.53
2.02
3.57
3.57
1.53
12.59
12.22

0
-1
0
-0.51
-1.1
-1.65
3.14
0

100
15.71
1.71
3.57
3.57
1.53
12.59
12.22

0
-1
0
-0.89
-4.61
-2.51
1.82
0

100
30.53
2.02
2.14
1.82
0.52
11.4
10.28

869

to 2000. FromTable 5, we see thatfor my samplefirms,bank

debtanddebentureswere almostas significantas DFI loans as
of 1989.However,the largedifferencein the figuresfor thefirst
andsecondcolumnsfor debenturesuggestthatdebentureissue
was dominatedby some of the largestfirms,a patternthathad
not changedeven in 1995 and in 2000. This is also confirmed
by columns3, 4 and 5 thatgive an idea aboutthe distribution
of firms.By 1995, bankdebt as a sourceof long-termfinance
had diminisheddrasticallyin importance.However,it slightly
recovereduntil2000 butwasstillmuchlowerthanthe 1989level.
In contrast,DFI lending and debentureshave played a more
importantrole duringthis period.DFI lendinggrew rapidlyby
1995,andthensloweddownslightlyby2000.Debenturessharply
increasedall over.Bankdebtrecoveredafter1995.Thesetwo are
responsiblefor offsettingthe slightdecreasein DFI borrowing.
FromTable6, it is clearthatshort-termbankborrowingconstitutes the majorsourceof short-termdebt requirementof firms
in all theyears1991, 1995andtill theyear2000. However,there
is a growingdemandforcommercialpaperinthelateryearsmainly
by the largefirms.Thereis also growingimportanceof funds
neededfor repaymentof debtin the nearfuture(reflectedby the
ratioof current
debt).
portionof long-termdebtovertotalshort-term

Table 4: Industry Categories of Sample Companies

IND1
IND2
IND3
IND4
IND5
IND6
IND7
IND8
IND9
IND10

IND11
IND12
IND13
IND14

IV

ModelsandResults
Econometric

IND15

Effect of Short-Term Debt on Export Performance

First,I wantto addressthe questionas to how short-termdebt
affectsexportperformance.
Forthis,I analysetheeffectof shorttermleverageon firm performancein the foreignmarket,controllingfor the otherfirm characteristicsof exportingfirms.In
orderto makea statementon the short-termleverageposition
of the firmI need to take into accountall the sourcesof shorttermloansavailableto the firm.I defineshort-termleverageas
theratioof short-term
debtto totalassets.Short-term
debtincludes
bankborrowingsplus commercialpapersplus a currentportion
of long-termdebt.Also leverageis endogenouslydeterminedby

Number
of Firms

Industry IndustryType
Dummy

IND16

IND17
IND18
IND19
IND20
IND21

16
Hotel,banking, insuranceand financialservices
Manufactureof dairyproducts,sugar, tea, coffee,
39
vegetable oils and fats, bakeryand food products
Manufactureof beverages, breweries,tobacco and
relatedproducts
3
46
Manufactureof cottontextiles
Manufactureof wool, silkand man-madefibretextiles
18
Manufactureof jute and othervegetable fibretextiles
2
(exceptcotton)
of textileproducts(including
2
Manufacture
wearingapparel)
Manufactureof wood and wood products,plywood,
furnitureand fixtures
3
Manufactureof paperand paperproducts,newsprint
and printing,publishingand allied
17
Manufactureof organicand inorganicchemicalsand
chemicalproducts,fertilisers,pesticides,drugs,medicines
and alliedproducts,matches, explosives, paints,dyes
and pigments,photographicand cinematographicgoods
89
Manufactureof rubber,solid rubbertyres, tube, plastic,
26
petroleumand coal products
Manufacture
of non-metallic
mineralproductslikecement,mica
etc
36
stone, glass andglass products,ceramicand refractory,
Basic metaland alloys industries:ironand steel,
ferroalloys, aluminium,casting of metals, copper,
steel tubes, transmissiontowers, etc
52
Manufactureof metalproductsand parts,except
7
machineryand equipment
Manufactureof machineryand equipmentotherthan
transport
equipment:electronics,electrical,equipment,
computers,hydraulics,engineering,insulatedwiresand
cables, fireprotectionequipment,industrialmachinery
forfood and textileindustries,etc
61
Manufacture
of transportequipmentandparts:shipsand
boatsbuilding,railwayandtramwayequipment,commercial
vehicles,passengercars andjeeps, automobilesancillaries
and transportequipment,two and threewheelers,bicycles,
81
cycle rickshaws,aircrafts,bullockcarts, etc
3
Jewelleryand relatedarticles
Powergenerationandelectricitygenerationandtransmission 6
Diversified(miscellaneous)
27
Watchesand clocks
1
Othermanufacturing:
medical,surgical,scientificand
measuringequipment,opticalgoods, stationeryarticles,
3
sportsand athleticgoods, etc
Totalnumberof firms
538

Table 3: Correlation Matrix

Allfirms
EXPSLRP
GRSALES
LNASSETS
INVEST
PROF
SHORTLEV
LT_LEV
Top 50 and largebusiness groupfirms
EXPSLRP
GRSALES
LNASSETS
INVEST
PROF
SHORTLEV
LT_LEV
Non-top50 and privatestand-alonefirms
EXPSLRP
GRSALES
LNASSETS
NVEST
PROF
SHORTLEV
LT_LEV

870

EXPSLRP

GRSALES

1.00
0.02
0.07
-0.03
0.04
0.1
-0.1

1.00
-0.02
0.06
0.13
-0.05
0.06

1.00
-0.01
0.11
-0.03
-0.01
0.11
-0.1
1.00
0.03
0.13
-0.03
0.08
0.09
-0.1

LNASSETS

INVEST

PROF

1.00
-0.01
0.03
-0.15
0.016

1.00
0.28
-0.14
0.06

1.00
-0.33
-0.07

1.00
-0.18

1.00

1.00
-0.03
0.05
0.18
-0.04
0.08

1.00
-0.05
-0.02
-0.08
-0.04

1.00
0.2
-0.1
0.03

1.00
-0.24
-0.02

1.00
-0.16

1.00

1.00
-0.01
0.08
0.11
-0.06
0.06

1.00
-0.03
0.01
-0.11
-0.02

1.00
0.36
-0.16
0.11

1.00
-0.39
-0.14

1.00
-0.19

1.00

Economic

and Political Weekly

This content downloaded from 111.68.103.104 on Mon, 14 Jul 2014 06:07:32 AM

All use subject to JSTOR Terms and Conditions

SHORTLEV

LT_LEV

Febraury 26, 2005

the firmat a point in time. The possibilityof expandingto the

exportmarketin thefuturemaycausethe firmto increaseshortterm leverage to ease capacity constraints,build distribution
networks,increasemarketingefforts.or promotetheirproduct
throughincreasedadvertising.Thus, in orderto answerhow
short-termleverage affects export performance,the possible
endogeneityof leverageshould be taken into account.
To test whethershort-termdebt affects firm's exportperformance,I estimatethe following equation:
+ y2FSIZE+ y3NRFATAit
EXPSLRPit=Yo+ y1EXPSLRPit-I
+ y6T+ai +Ui
+ 4CPROFTAit+ySHORTLEVit
.. (1)

Here, Y represents the dependent and X represents the independent variables.

Short term leverage changes are likely to reflect changes in
expectations about future product market outcomes. Note that
as we see from the descriptive statistics that the top 50 and large
business group firms are typically bigger than stand-alone or
smallergroupfirms. One may arguethatthese differences between
the two sub-samples drive my results. In order to correct for this
possibility, I introduce the variable FSIZE as additional control
variable.2 The variable FSIZE is the natural log of total assets.
OLS estimates or even panel Tobit estimates are biased and
inconsistent due to endogeneity problems. Therefore, I estimate
my models using an instrumentalvariable approach.The instrumental variable estimation technique controls for the fact that
the explanatory variables are likely to be correlatedwith the error
term and the firm-specific effect, and deals with possible
endogeneity problems. Equation (2) is therefore estimated with

whereEXPSLRPis thepercentageof totalsalesexportedby firm

i at time t, NRFATAis the gross fixed assets net of revalued
SHORTLEV
reservesovertotalassets,CPROFTAis profitability,
is short-term
debtovertotalassetsandT a timetrendto controls
for timely changes.Lags of exportare includedto controlfor
Table 5: Relative Importance of Different Sources
thatmaycontributeto performance
firmspecificcharacteristics
Long-Term Debt in 1989,1995 and 2000
over time.
To test whetherexport performanceparametersare signi(1)
(2)
(3)
(4)
(5)
Median
3rd
1st
ficantlydifferentbetweenthe top 50 and largebusinessgroup
_(Num)/,(Denom) Sample
i
i
Mean
Quartile
Quartile
firmsandsmallergroupor privatestand-alonefirms,I introduce
two dummyvariables,DTOP50andDPVT.DTIP50dummyis Year= 1989
0.31
0.31
0.01
0.2
0.54
equalto 1 whenthe firmis ownedby top 50 andlargebusiness DFI/LTD
LTBNKD/LTD
0.21
0.25
0
0.17
0.43
firm
not
DPVT
is
to
I
if
either
does
belong DEBEN/LTD
equal
groups.Similarly
0.13
0.23
0
0.2
0.00
to any businessgroupor it belongs to smallergroup.I derive Year= 1995
0.35
0.47
0.1
0.43
0.7
differentsets of resultsfor the two types of firms.I also derive DFI/LTD
0.1
0.1
0.1
0
0.00
resultsforall firmstakentogetherandthereI comparetheeffects LTBNKD/LTD
DEBEN/LTD
0.34
0.2
0
0.1
0.34
of DTOP50andDPVTwith respectto firmsbelongingto other Year= 2000
business houses representedby anotherdummy DOTHGRP. DFI/LTD
0.42
0.34
0.31
0
0.63
0.17
0.14
0
0.01
0.22
Takingfirstdifferencesof equation(1) eliminatesthe a i, which LTBNKD/LTD
DEBEN/LTD
0.32
0.21
0
0.02
0.29
were the source of the bias in the OLS estimator.
This gives:
Notes:The numbersin the firstcolumnare the ratiosof sum (overall sample
firms)of a particulartype of debt, to the sum (overall sample firms)of
= y1AEXPSLRPit_
+ y2AFSIZEi+ y3ANRFATAit
long-termdebt.Thenumbersinthe nextfourcolumnsaresamplemeans
AEXPSLRPit
+ Y4ACPROFTAit+ y5ASHORTLEVit+ Auit

i = 1...... N t = 1,....,
...(2)
ArellanoandBond( 1991) arguethata moreefficientestimator
resultsfromthe use of additionalinstrumentswhose validityis
basedon orthogonalitybetweenlaggedvaluesof the dependent
variableyit and the errorsuit.The first two observationsare
lost to lags and differencing.The first differencesof the exogenousvariableswill serveas its own instrumentsin estimating
the first differencedequations.Now I have to instrumentfor
whichis clearly
AEXPSLRPtli = (EXPSLRPiti-EXPSLRPt_2),
correlatedwith the errorAuit= (uit- uit_). Assumingthat uit
arenotautocorrelated,
foreachi at t=3, EXPSLRPiactsas valid
instrumentfor AEXPSLRPitJ.Similarly,at t=4, EXPSLRPil,
are valid instruments.Continuingin this fashion,
EXPSLTRPi2
I obtainan instrumentmatrixwithone row for each timeperiod
that I am instrumenting.
The basic instrumentset used in my resultsin Table7 is of the
form:

Zi =

Yi,

0.....

0..........0....

Axi3

o0

Y,,

Yi2

0.........0....

Axi4

O............

o0

Economic and Political Weekly

Yi.....

Yii~~

1991
1992
199
1993

0l2.. 0..

and quartilevalues (N=538).

DFIis loan fromDevelopmentFinancialInstitutions.LTBNKD
is longtermbankdebt. DEBENis debenture.LTDis totallong-termdebt.
Table 6: Relative Importance of Different Sources
Short-Term Debt in 1991, 1995 and 2000
(1)
(2)
Z(Num)/Y(Denom)Sample
i
i
/
Mean

Year= 1991
STBNKBOR/STD
CP/STD
CURLTD/STD
Year= 1995
STBNKBOR/STD
CP/STD
CURLTD/STD
Year= 2000
STBNKBOR/STD
CP/STD
CURLTD/STD

(3)
(5)
(4)
1st
Median
3rd
Quartile
Quartile

0.9
0.01
0.09

0.92
0.003
0.08

0.93
0
0

1
0
0

1
0
0.06

0.85
0.02
0.13

0.87
0.01
0.12

0.83
0
0

1
0
0

1
0
0.16

0.7
0.07
0.23

0.82
0.03
0.15

0.73
0
0

0.99
0
0

1
0
0.22

Notes:The numbersin the firstcolumnare the ratiosof sum (overall sample

firms)of a particulartype of debt, to the sum (overall sample firms)of
short-termdebt. The numbers in the next four columns are sample
means and quartilevalues (N=538)
STBNKBORis short-termbank borrowing.CP is commercialpaper
borrowing.CURLTDis currentportionof long-termdebt. STD is total
short-termdebt.,
ThoughCP/STDhave "0"value at 3rdQuartile,they are positiveat 99
percentileindicatingthe presence of extremevalues.

February 26, 2005

This content downloaded from 111.68.103.104 on Mon, 14 Jul 2014 06:07:32 AM

All use subject to JSTOR Terms and Conditions

871

the system of generalisedmethodsof moments(GMM)3estimatorsas proposedby Arellanoand Bond (1991).4For details

aboutthe techniqueI refer to the appendixB.
The ArellanoandBond two-stepGMMresultsas reportedin
Table 7 show GMM estimatesof equation2 (based on first
difference).Exportin the currentperiodis significantlyrelated
to the previousyear's changes.We clearly see thatexportrespondspositively to the currentincreasein firm's short-term
leverage.This is truefor all firmstakentogether(column2) as
well as for two sub groups (columns 3 and 4). The control
andinvestment.
Ifcurrentprofit
variablesarefirmsize,profitability
rises, the smallergroupor stand-alonefirm'sexportdecreases.
This may be due to the fact that if profitrises, firms are more
willingto sale in thedomesticmarketinsteadof theriskyforeign
market.5However,the rise in investmentboostsexport.Forthe
top groups,I find a change in profit or investmentdoes not
significantlyinfluenceexport.If the firm gets bigger,it experiencesan increasein foreignsales.Thussize mattersforexport.
Effect of Long-Term Debt on Firm Performance
The pooled cross-sectional-timeseries Tobit regressionsI
estimatebelow resemblethose of Oplerand Titman(1994). I
test the significanceof lagged values of long-termleverage
to examinethe effect of long-termdebt on
(LT_LEVERAGE)
firm'sgrowthof salesinthecurrentyear.Theyhavethefollowing
generalform:
Sales Growthit= 0 + PILNSALESi. + REINVESTi,
_ + 03CPROFRAi,_I
21
+ P4LT_LEVERAGEi _2(or t7) + yj (Industry Dummies).
j=1
12
3
+X t, (YearDummies),+y k(GroupDummies)+eit ...(3)
t=1

k='

The lagged structureused in the above equationis meantto

mitigatesimultaneityproblems.As far as the errorstructureis
concerned,the MaximumLikelihoodEstimationof panelTobit
assume that eit is uncorrelatedwith Eit' and ct t, = 0, when t ? t';

Table 7: Export Equations

GMM Estimates (All Variables in First Differences)
DependentVariable:EXPSLRPit
Sample Period:1991-2000
AllFirmsCombinedTop 50 and Large SmallerGroup
Independent
and Private
Variables
Business Group
Stand-alone
0.5***
(37.27)
0.61***
(2.41)
-0.65
(-0.95)
0.14
(0.2)
2.53***
(2.52)
2953
296
0.0006
0.5730
0.13
1665.5 (15)

0.39***
(Y1)
(19.19)
EXPSLRPit_1
0.78***
(Y2)
(2.23)
FSIZEt
1.34
(Y3)
(0.89)
NRFATAit
-2.72
(Y4)
(-1.44)
CPROFTAit
2.42***
(Y5)
(2.33)
SHORTLEVi,
No of observations
5339
No of firms
538
AR1
0.0001
AR2
0.1865
0.39
Sargantest
Waldtest
500.14 (15)

0.22***
(15.41)
1.36***
(3.86)
2.46***
(2.06)
-5.26***
(-3.05)
2.19***
(2.33)
2386
242
0.0168
0.1081
0.1
656.93(14)

Notes: z values are inthe parentheses.Timedummiesandgroupdummiesare

includedbutnot reported.
GMMresultsaretwostep estimateswithone periodlagofthe dependent
variable.
AR1and AR2are tests forthe GMMestimators,the P-values reported
referto the two-stepGMMestimators.
restrictions
fortheGMMestimators,
Sarganis a test oftheoveridentifying
theP-valuesareonlyreportedandnumberofinstruments
is inthebrackets.
FSIZEis proxiedby naturallog of totalassets.
***:Significantat 5 per cent or better;**:Significantat 5-10 per cent.
Table 8: Capital Structure and Product Market Performance:
The Effect of Long-Term Leverage on Firm's Growth of Sales Panel Tobit Regressions
Dep Var:
Group
Sales
Growth
at t

AllFirms

Top 50 Business

Non-top50
Business Group
2-Year
7-Year 2-Year 7-Year 2-Year 7-Year
Lagged
Lagged Lagged Lagged Lagged Lagged
LT_LEV LT_LEV LT_LEVLT_LEV LT_LEVLT_LEV

-0.44
-1.04
-0.55*** -0.72***
(-1.56)
(-1.02)
(-2.96) (-2.71)
LNSALES 0.07*** 0.11*** 0.05*** 0.07***
(6.43)
(5.12)
(5.79) (4.28)
INVEST
0.92***
1.84*** 0.89*** 1.68***
(13.11)
(10.76) (15.97) (13.67)
CPROFTA -0.29***
-0.2
-0.27*** -0.13
(-2.16)
(-0.8)
(-2.46) (-0.69)
0.34*** 0.38*** 0.39*** 0.46***
LT_LEV
(5.88)
(3.1)
(8.55)
(4.7)
Observations 5347
2662
2955
1475
Observation 1314 Left 889 left 643 left 437 left
Summary censored censored censored censored
and 4033 and 1773 and 2312 and 1038
ununununcensored censored censored censored
obserobserobser- observations
vations vations vations
LRChi2
statistic
391.47
247.64
455.54 308.89
d.f
36
34
31
26
Prob.>Chi2 0.00
0.00
0.00
0.00
Pseudo R2
0.03
0.04
0.1
0.1
Intercept

-0.54
-1.21
(-1.41) (-0.78)
0.09*** 0.17***
(3.99) (3.41)
0.95*** 2.02***
(5.91) (4.86)
-0.26
-0.21
(-0.92) (-0.37)
0.25** 0.26
(1.81) (0.97)
2392
1187
642 left 452 left
censored censored
and 1690 and 735
ununcensored censored
obser- observations vations

wheret and t' are indexesfor time periodswhen observations

of thesamefirmarecollected.Similarly,IassumeEitisuncorrelated
with ej, Vi ji at same t.
However,one may articulatethe argumentfor unobservedor
unmodelledfirmspecificvariablewhichmayactuallyinfluence
the dependentvariableand are thus capableof introducinga
simultaneitybias in my empiricalspecification.It is a very
difficultcase to arguesince there is no such endogeneitytest
in a panel structure(keeping in mind the Hausmantwo step
estimationmethodof endogeneitytest). In sucha case, I control
these unobservedfactorsby takingthreebusinessgroupdummies. 12 yearlydummiesand 21 industrydummies.I assume
133.02 81.86
that firms within each business group category and industry
33
31
0.00
0.00
buttheirbahaviourvaries
categoryhavecommoncharacteristics
0.02
0.02
across groups and industriesand also across various years.
Iinclude21 industrydummiesand12yearlydummies Notes: ***:Significantat 5 per cent or better.**:Significantat 5-10 per cent.
Accordingly,
The dependentvariableis firmannualsales growthat timet, given by
to controlfirm specific fixed effects.
(Sale, -Salet,_l)/Sale,t_.The dependent variable is left censored at
It is obviousthatI shoulddrop 1 dummyeach from 12 year
zero. LSALESis the contemporaneousnaturallogarithmoftotalassets.
INVESTis Investmentwhich is the growth in fixed assets minus
dummiesand21 industrydummiesand3 groupdummiesto avoid
revaluedreserves over total assets at t-1. CPROFTAis Profitability,
dummytrap,whichwill arisedue to the multicollinearity
probwhichis the cash profitoverassets at t-1. LT_LEV
is the long-termdebt
lem. Herelong-termleveragechanges are likely to reflectthe
overtotalassets, and is measuredeitherat t-2ort-7.Thesampleperiod
is 1989-2000.Theregressionsinclude21 industryand 12 yeardummies
cumulativeeffect of past decisions.
and DTOP50,DOTHGRPand DPVTthree groupaffiliationdummies
Six regressionoutputsfrom equation3 reportedin Table 8.
(not reported).The numbersare the coefficientsof the Tobitmodel.
I canmakefollowinginferencesfromtheresultsof Table8. First,
Figuresinside bracketsare the t values.
872

Economic and Political Weekly

This content downloaded from 111.68.103.104 on Mon, 14 Jul 2014 06:07:32 AM

All use subject to JSTOR Terms and Conditions

Febraury 26, 2005

firmsthatinvestedmorein the previousyear seem to do better

in theproductmarketnextyear.This is supportedby thepositive
andsignificantestimatedcoefficientof INVESTin all six cases.
Second,firmsthatweremoreprofitablein thepreviousyeartend
to observelowersales growth.This is reflectedby negativeand
significantestimatedcoefficient of CPROFTAfor all firms
combinedand for top 50 businessgroupcategory.
Moreimportantly,the resultsshow that for both two period
and seven periodlag values of long-termleverage(LT_LEV),
thereexists a remarkablyconsistentpatternin the way capital
in theproductmarketsales
structureinfluencesfirmperformance
for all firmstakentogether.The resultsindicatethatan increase
in the use of long-termdebtfinancingsignificantlyboostssales
growthmainlyfor the top 50 andlargebusinessgroupaffiliated
firmsaftertwoyearsorsevenyearsof takingtheloan.Thuslongtermdebtcan committhe firmto competein the productmarket
andmayinfactactivatethefirmtotakeaggressiveanoutputstance.
Atthesametime,however,long-termleverageis inconsequential
to performancefor smallergroupand privatestandalonefirms.

univariate results are displayed in Table 10. The results from

Panel A and Panel B confirm that both the top group and private
stand-alone firms, taking long-term debt on average (both mean
and median), increase advertising expenses, marketing efforts,
build distributionnetworks and improve R&D infrastructureover
time. All of these may help the firm to expand market share in
the product market in the long run.
Table 9: Univariate Tests: Effect of Short-Term Debt on
Product Market Variables
(1)
Real MarketVariables

Panel A: means
Exportsales ratio(percent)
Allfirms
Top 50 business group
Non-top50 firms
Parametric and Non-parametric Univariate Results
R&Dintensity
Allfirms
I lookattheresultsof univariateparametric
andnon-parametric Top50 business group
tests on real marketvariables.In Table 9, I see the effects of Non-top50 firms
Advertisingintensity
market- All
short-term
debton the exportsales ratio,advertisement,
firms
ing, R&D and distributionintensities.I comparethe average Top50 business group
values betweenbefore and after the short-termloan has been Non-top50 firms
Marketing
intensity
taken.The parametrict-testshows (in panelA of Table9) that All
firms
mainlytop groupfirmson averagespendmoreon advertising, Top50 business group
and researchand developmentsubsequentto takingshort-term Non-top50 firms
Distribution
intensity
debtin orderto gain strategicadvantagein the productmarket. All
firms
Both the top groupand privatestand-alonefirms' averagedis- Top50 business group
tributionintensityis higherafterthe loan has been takencom- Non-top50 firms

paredto the previousloan. Firmsalso export more following

short-termdebt in comparisonto the previousyearof the loan.
InpanelB of Table9, we reporttheresultsof Wilcoxonsigned
ranktests to find the importanceof short-termcapitalstructure
to firmexportandotherstrategicrealmarketvariables.Wilcoxon
tests are conductedto evaluatethe
signed-ranknon-parametric
of
in
significance changes these measures.Observationsare
separatedintosituationsbeforeandaftertheloanhavebeentaken.
The null hypothesisis thatthe beforeand aftershort-termdebt
are frompopulationswith the same distributionsand the same
medians.Wilcoxontests generallyindicatethatthe beforeand
aftershort-term
debtforthesefirmsarenotdrawnfromthesame
distribution.I see thatfirmexports,expenditureon advertising,
marketing,distributionand researchand developmentincrease
withshort-term
debt.Thus,I findtherelationshipbetweenshorttermdebtandfirmbahaviourin therealmarket.Thisrelationship
is evidentfor both types of firms.
Similarly,in establishingthe importanceof long-termdebt
financingon the real economy, my objectiveis to identifyits
impacton variousproductmarketstrategies(e g, advertising,
R&D,etc) whichthe firmsmightimplemarketing,distribution,
mentgiven theirown as well as theirrivals'choice of financing
instruments.
Accordingly,I look at the long-termimpactof debt
financing on these real market variablesthroughunivariate
andnon-parametric
tests.I presenta timeseriestable
parametric
of R&D expense, advertising, marketing and distribution
expensesfor all firmsthattook long-termdebtfromthe current
year (t) till seven years (t+7) after the loan was taken. The

Panel B: medians
Exportsales ratio(percent)
Allfirms
Top 50 business group
Non-top50 firms
R&Dintensity
Allfirms
Top 50 business group
Non-top50 firms
Advertisingintensity
Allfirms
Top 50 business group
Non-top50 firms
Marketingintensity
Allfirms
Top 50 business group
Non-top50 firms
Distribution
intensity
Allfirms
Top 50 business group
Non-top50 firms

(2)
t-1

(3)
t

(4)
t+1

(5)
t-stat for
Difference
Between
Beforeand
After
(Col2 and
Col 4)

7.23
6.45
8.21

7.64
6.79
8.71

7.93
6.99
9.11

8.7**
6.32***
6.05"**

0.0005
0.0005
0.0005

0.0006
0.0006
0.0006

0.0006
0.0006
0.0006

1.87***
2.87***
0.62

0.0058
0.0065
0.0048

0.006
0.0068
0.0049

0.006
0.007
0.0049

4.27***
4.3***
1.28

0.016
0.0165
0.015

0.016
0.017
0.015

0.017
0.0172
0.016

5.72***
4.86***
3.18***

0.02
0.023
0.015

0.02
0.024
0.016

0.02
0.024
0.016

4.47***
3.11***
3.56***
z-stat for
Difference
Between
Beforeand
After
(Col2 and
Col 4)

1.65
2.24
0.88

1.88
2.46
1.05

2.06
2.72
1.31

11.32***
9.26***
6.64***

0.00
0.00
0.00

0.00
0.00
0.00

0.00
0.00
0.00

4.81***
4.16***
2.43***

0.0005
0.0006
0.0005

0.0005
0.0006
0.0005

0.0005
0.0006
0.0005

1.43
2.71***
-0.89

0.0092
0.01
0.008

0.0095
0.011
0.0081

0.01
0.011
0.0082

8.1***
7***
4.21***

0.012
0.013
0.009

0.012
0.014
0.01

0.0123
0.014
0.01

8.71***
7.94**"
4.1***

Notes:Thistablecomparestheeffectivenessofshort-term
debton realvariables.
***denotes significantat 5 per cent or betterand **denotes
significant
at 5-10 percent;z-statisticfordifferencebetweenpairedseries denotes
the outcome of a "Wilcoxonsigned-ranktest" for difference in the
distributions.Both the "t-test"and "WilcoxonSigned-ranktests" are
pairedunivariatetests that compare the average values of common
sample between the two series.
Yeart is the year of the issuance of short-termdebt and this is takenas
the controlperiod.

Economicand PoliticalWeekly February26, 2005

This content downloaded from 111.68.103.104 on Mon, 14 Jul 2014 06:07:32 AM

All use subject to JSTOR Terms and Conditions

873

V
ConcludingDiscussions

debt influencesa firm's R&D, advertising,marketing,anddistributionstrategies.In case of long-termdebt, firms take time
to buildinfrastructure
throughincreasedmarketingandpromoThe strategic use of debt models shows that, under imperfect tions.R&Dwhichhavelong-termimpacton theirproductmarket
competition, firms may have strategic incentives to take debt. performance.
Consideringa longertimehorizon,I findthatlongor
use
their
firms
could
term
debt
boosts
totalsalesgrowthfortop50 andlargebusinesses
pockets
deeper
may
Financially healthy
strategically spend on building distribution networks, increase groupaffiliatedfirms.However,for the unaffiliatedfirms,it is
marketingefforts and advertising for product promotion to prey inconsequentialon total growthof sales. Thus,debt can shape
on rivals or deter potential entrants. My results suggest that the industrycompetition.Consequently,I find empiricalevidence
strategic consideration in the output market induce firms to take on the existenceof a linkagebetweenfirm's choice of capital
higher debt in order to gain strategic advantage. This establishes structuraland its productmarketperformance.
a link between debt and firm competition in the product market.
Basedon my empiricalfindings,I proposethatdevelopments
I distinguish between short-term debt and long-term debt to in the debt marketcould be an importantdeterminant
for corexamine their impact on a firm's product market outcomes. I porateperformance.
In this context,creditratingagencieshave
roletoplayinthedebtmarkets.Centralbanksshould
compare the top group affiliated firms with their smaller group animportant
or unaffiliated counter parts. I find that short-term debt induces bemoreresponsible
formaintaining
financialstabilityandintegrity
the firms to do well in exports. I also discover that short-term of thefinancialmarket.Therecentlyinitiatedmajorlegalreforms
Table 10: Effect of Long-Term Debt on Real Market Variables
(Unitsare in Rupees Million)
t

t+1

t+2

t+3

t+4

t+5

AllFirms

1.63

Top 50

2.68

Non-top50

0.31

AllFirms

15.97

Top 50

26.1

Non-top50

3.26

AllFirms

35.11

Top 50

53.76

Non-top50

11.72

AllFirms

61.59

Top 50

102.49

Non-top50

10.29

1.73**
(1.56)
2.82**
(1.4)
0.36
(1.23)
16.95'**
(9.09)
27.81***
(8.65)
3.43**
(4.28)
37.44***
(11.7)
57.22***
(10.42)
12.78**
(7.47)
65.46***
(4.3)
109.29*"
(3.99)
11.01**
(6.08)

1.92**
(1.64)
3.15**
(1.52)
0.4
(0.88)
18.14'*
(9.09)
29.83***
(8.65)
3.61***
(4.28)
40.1***
(11.26)
61.47***
(10.1)
13.52"**
(6.92)
69.7***
(4.02)
116.35***
(3.72)
11.7**
(5.6)

2.12**
(1.69)
3.48**
(1.58)
0.44
(0.84)
19.56***
(8.92)
32.27***
(8.51)
3.82***
(4.13)
42.58***
(10.49)
65.45***
(9.41)
14.25**
(6.39)
74.3***
(3.82)
124.31 **
(3.56)
12.36***
(5.2)

2.32**
(1.49)
3.8**
(1.38)
0.49
(0.82)
21.03**
(8.65)
34.78***
(8.26)
4.03***
(3.79)
45.28***
(10.01)
69.73***
(8.99)
15.04"**
(6.02)
79.1***
(3.51)
132.46**
(3.26)
13.1**
(4.89)

2.53
(1.27)
4.13
(1.22)
0.52
(0.42)
22.64**
(8.13)
37.52"**
(7.81)
4.24***
(3.34)
48.37***
(9.35)
74.63***
(8.42)
15.9***
(5.5)
84.52**
(3.28)
141.6***
(3.05)
13.93**
(4.56)

2.68
(1.01)
4.4
(0.95)
0.56
(0.45)
24.37***
(7.54)
40.48***
(7.27)
4.45**
(2.92)
51.43**
(8.13)
79.44***
(7.29)
16.8***
(4.93)
90.87***
(3.02)
152.33***
(2.81)
14.89***
(4.2)

2.9
(0.94)
4.76
(0.88)
0.61
(0.49)
26.02*"
(6.58)
43.35***
(6.37)
4.62***
(2.31)
54.68***
(7.08)
84.71***
(6.4)
17.58**
(4.26)
97.56***
(2.56)
163.76***
(2.38)
15.8***
(3.46)

AllFirms

0.00

Top 50

0.00

Non-top50

0.00

AllFirms

0.2

Top 50

0.3

Non-top50

0.1

AllFirms

6.4

Top50

13.8

Non-top50

2.6

AllFirms

6.8

Top 50

16.8

Non-top50

2.8

0.00***
(4.18)
0.00**
(3.55)
0.00**
(2.2)
0.2***
(12.86)
0.3***
(12.45)
0.1***
(5.11)
7.1**
(26.1)
15.25**
(21.54)
2.9"*
(14.5)
7.8*"
(27.9)
18.85***
(23.5)
3.1***
(15.05)

0.00"*
(4.1)
0.00***
(3.47)
0.00'**
(2.19)
0.2***
(12.86)
0.4***
(12.45)
0.1**
(5.11)
7.7'**
(24.12)
16.7**
(20.03)
3.1"**
(13.28)
8.4***
(25.37)
20.7'**
(21.67)
3.4***
(13.04)

0.00**
(4.1)
0.00'**
(3.55)
0.00'**
(2.02)
0.2***
(11.75)
0.4'**
(11.56)
0.1"**
(4.41)
8.3**
(22)
17.9***
(18.43)
3.3**
(11.81)
9***
(23.01)
22.4***
(19.73)
3.8***
(11.84)

0.00***
(3.01)
0.00**
(2.5)
0.00**
(1.72)
0.2***
(11.25)
0.4'**
(11.21)
0.1***
(4.1)
8.9**
(20.14)
19.7***
(16.92)
3.6***
(10.74)
9.7**
(20.76)
24.3***
(17.83)
3.95***
(10.7)

0.00
(1.07)
0.00
(1.05)
0.00
(0.37)
0.2***
(9.52)
0.5"*
(9.91)
0.1**
(2.91)
9.6**
(17.98)
21.3"*
(15.04)
3.95**
(9.76)
10.4***
(18.91)
25.7***
(16.4)
4.1***
(9.53)

0.00
(0.8)
0.00
(0.85)
0.00
(0.14)
0.2**
(7.8)
0.5**
(8.5)
0.1 **
(1.98)
10.3***
(15.47)
23.4***
(12.77)
4.15**
(8.62)
11.2***
(16.94)
27.5*
(14.96)
4.3'"*
(8.19)

0.00
(0.55)
0.00
(0.3)
0.00
(0.62
0.2***
(5.06)
0.5***
(6.32)
0.1
(0.21)
11***
(12.93)
24.7***
(10.97)
4.3***
(6.71)
12.2***
(14.38)
28.5***
(13.18)
4.4**
(6.18)

Real MarketVariables
Panel B: means
R&Dexpense

Advertisingexpense

expense
Marketing

Distribution
expense

Panel B:medians
R&Dexpense

Advertisingexpense

expense
Marketing

Distribution
expense

t+6

t+7

Notes: Sign ranktests the equalityof matched pairs of observationsusing the Wilcoxonmatched-pairssigned-rankstest. The null hypothesis is that both
distributionsare the same. Inpanel A, t values are in the parentheses and in panel B, z-values are in the parentheses.

874

Economicand PoliticalWeekly Febraury26, 2005

This content downloaded from 111.68.103.104 on Mon, 14 Jul 2014 06:07:32 AM

All use subject to JSTOR Terms and Conditions

in thefinancialsectorrelatingto securitieslaws,fraudsin banks,

regulatoryframework,assetsecuritisation,andpaymentsystems
may facilitatethe developmentof debt marketsand benefitthe
borrowingfirms. Banking developmentis also necessaryto
improveaccessof fundsforboththetopgroupandprivatestandalone firms.
Appendix A: Variable Description
Financial Variables
Total Debt = Short-termDebt + Long-termDebt.
Short-termDebt = Short-termbank borrowing
+ Commercialpaper+ Currentportionof long-termloans.
Long-termDebt = Long-termborrowing= Total debt
- short-term bank borrowing - commercial paper - current

portion of long-termdebt.
LT_LEVrefersto long-termleverage.
Long-termleverage= Long-termborrowing/totalassets.
SHORT_LEVrefersto short-termleverage.
Short-termleverage= short-termdebt/totalassets.
Real Market Variables

estimationperiodis 1991-2000.In orderto avoiddummytrap,

oneyeardummyin everythreecases arebeingdroppedandother
years are being compared.These time dummiesare used as
additionalinstruments.
The reliabilityof the GMMestimationproceduredependsvery
muchon the validityof instruments.I considerthe validityof
the instrumentsby presentinga Sargantest. The Sargantest is
a teston overidentifyingrestrictions.It is asymptoticallydistributed as X2 and tests the null hypothesis of validity of the
instruments.P-valuesreportthe probabilityof
(overidentifying)
incorrectlyrejectingthe nullhypothesis,so thata P valueabove
0.05 impliesthatthe probabilityof incorrectlyrejectingthe null
is above0.05. In thiscase, a higherP-valuemakesit morelikely
that the instrumentsare valid.
The consistencyof the estimatesalso dependson the absence
of serialcorrelationin the errorterms.This will be the case if
the differencedresidualsdisplaysignificantnegativefirstorder
serialcorrelation
andnon-secondorderserialcorrelation.
I present
tests for first orderand second orderserial correlationrelated
to the estimatedresidualsin firstdifferences.The test statistics
are asymptoticallydistributedas standardnormalvariables.
The nullhypothesishererelatesto 'insignificance'so thata low
P-valuefor the test on first orderserialcorrelationand a high
P-valuefor the test on the secondorderserialcorrelationssuggests thatthe disturbancesarenot seriallycorrelated.The serial
correlationtests (AR1 and AR2 in the table) referto the twostep estimates.
I also presentWald tests. These test statistics are also asymptoticallydistributedas X2variables.The Waldtest testsjoint
Thenullhypothesis
significanceof all, ora subsetof parameters.
refersto 'insignificance',implyingthat low P-valuessuggest
joint significance.This can termedas goodness of fit test for
my model. ilG3

GRSALESrefersto growthrateof sales in the currentperiod.

LN SALESrefersto the logarithmof sales.Thisvariablereflects
theunobservedfactorsthatarerelatedto thesize of thecompany.
LNASSETSis naturallog of total assets.
EXPINTrefersto the exportintensityof firms in percentage.
It capturesthe effect of exposureto internationalcompetition.
Advertisingintensityis measuredas the ratioof advertisement
expenditureto total sales. It capturesthe effect of intangible
assets.
Distributionintensityis the distributionexpenditureover total
sales.
Email:arind@hotmail.com
Marketingintensityis proportionof total sales that the firm
Notes
spendson marketingexpenditure.
R&Dintensitymeasurewhatproportion
of totalsalesfirmsspend
on researchand development.
[The authorwould like to thank Sandwip KumarDas, Sudipto Dasgupta,
CPROFTAis firm'sprofitabilityandmeasuredbydividingfirm's KennethJ Kopecky, RobbertTagart and Sugato Dasgupta for extremely
cash profitby its total assets. Cash profitis derivedby adding helpful comments. All remainingerrors are of course my own.]
the non-cashchargessuch as depreciationand amortisationto 1 They areCreditRatingInformationServices of IndiaLimited(CRISIL),
the profitaftertax.
InvestmentInformationand CreditRating Agency (ICRA), and Credit
INVESTis how mucha firminvestsin a period.It is measuredas
Analysis & Research (CARE).
changeingrossfixedassetsnetof revaluedreservesovertotalassets. 2 Controlshave also been carriedout by interactingall variableswith size
and industrydummies. Results are in general robust. However, when
INDi is an industrydummyfor the ith industry.It takesa value
all potential interactions are introduced, the number of coefficients
of 1 forcompaniesbelongingto the ithindustryand0 otherwise.
increasesremarkablyand the interpretationbecomes harder.Moreover,
Thenumbersof firmsin variousindustrycategoriesarepresented
we are not interested in the effect of size on export as we are more
in Table 4.
interestedto see the effect of group affiliation. For this reason, these
Appendix B: Details of the System GMM
Estimation Method
Arellanoand Bond (1991) derived a generalisedmethodof
moments estimator for estimating the coefficients using
laggedlevel of the dependentvariableand differencesof the
strictlyexogenousvariables.In STATA7,xtabondimplements
this estimatorknown as the Arellano-Bonddynamic panel
estimator.In all estimatorsI controlfor time effects by adding
time dummiesfor 1989-2000. In constructinglags and taking
first differences,two cross-sections are lost. Therefore,the
Economic and Political Weekly

results are not reported.

3 GMMis a robustestimatorin that,unlikemaximumlikelihoodestimation,
it does notrequireinformationof theexactdistributionof thedisturbances.
In fact, many common estimators in econometrics can be considered
as special cases of GMM. The theoreticalrelation that the parameters
shouldsatisfy is usuallyorthogonalityconditionsbetweensome (possibly
nonlinear) function of the parametersf(O) and z, set of instrumental
variables.The GMM estimatorselects parameterestimates so that the
sample correlationsbetween the instrumentsand the functionf are as
close to zero as possible.
4 Arellanoand Bond (1991) developed a GeneralisedMethodof Moments
estimatorthat treats the model as a system of equations, one for each
time period. The equations differ only in their instrument/moment
condition sets. The predeterminedand endogenous variables in first

February 26, 2005

This content downloaded from 111.68.103.104 on Mon, 14 Jul 2014 06:07:32 AM

All use subject to JSTOR Terms and Conditions

875

Grimaud,A (2000): 'ProductMarketCompetitionandOptimalDebtContracts:

The LimitedLiabilityEffect Revisited', EuropeanEconomicReview44
(10): 1823-40.
Harris, M and A Raviv (1991): 'Capital Structureand the Informational
Role of Debt', Journal of Finance, 45: 321-49.
Hellman, T and M Puri (1999): 'The Interactionbetween ProductMarket
andFinancingStrategy:TheRole of VentureCapital',StanfordUniversity,
Working Paper.
Maksimovic,V (1988): 'CapitalStructurein RepeatedOligopolies', RAND
Journal of Economics. 19 (3): 389-407.
Arellano, M and S Bond (1991): 'Some Tests of Specification for Panel Maksimovic,V and S Titman(1991): 'FinancialReputationand Reputation
Data: Monte Carlo Evidence and an Application to Employment
for ProductQtjiality',Review of Financial Studies, 2: 175-200.
Equations', Review of Economic Studies, 58: 277-97.
Modigliani, F and M Miller (1958): 'The Cost of Capital, Corporation
Financeand the Theoryof Investment',AmericanEconomicReview,48:
Bandyopadhyay,A and S K Das (forthcoming):'The Linkage between the
Firm'sFinancingDecisions andReal MarketPerformance:A PanelStudy
261-97.
of Indian CorporateSector', Journal of Economics and Business.
Opler, T C and S Titman (1994): 'Financial Distress and Corporate
Bolton. P and D S Scharfstein (1990): 'A Theory of PredationBased on
Performance',Journal of Finance, 49 (3): 1015-40.
Agency Problemsin FinancialContracting',AmericanEconomicReview Phillips, G M (1995): 'IncreasedDebt and IndustryProductMarkets:An
80: 93-106.
Empirical Analysis', Journal of Financial Economics, 37: 189-238.
Brander,J A and T R Lewis (1986): 'Oligopoly and Financial Structure: Poitevin, M (1989): 'Financial Signalling and Deep Pocket Argument',
The Limited-LiabilityEffect', American Economic Review 76: 956-70.
RAND Journal of Economics, 20: 26-40.
- (1988): 'BankruptcyCosts and the Theory of Oligopoly', Canadian Ravid, S A (1988): 'On Interactionsof Productionand FinancialDecisions',
Journal of Economics 21: 221-43.
Financial Management: 87-99.
Chevalier,J A (1995): 'CapitalStructureand Product-MarketCompetition: Rotemberg,J J ard D S Scharfstein(1990): 'ShareholderValue Maximisation
and ProductMarket Competition', Review of Financial Studies 3 (3):
EmpiricalEvidencefromthe SupermarketIndustry',AmericanEconomic
367-91.
Review, 85 (3): 415-35.
Chevalier,J A and D Scharfstein (1996): 'Capital-MarketsImperfections Showalter,D M (1995): 'Oligopoly and Financial Structure:A Comment',
American Economic Review, 85: 647-53.
andCountercyclicalMarkups:TheoryandEvidence',AmericanEconomic
Review 86 (4): 703-25.
-(1999): 'StrategicDebt:Evidencein Manufacturing',InternationalJournal
Das, S K and A Bandyopadhyay (2003): 'Quality Signals and Export
of Industrial Organisation, 17: 319-33.
Performance:A Micro-Level Study. 1989-97', Economic and Political Telser, L G (1966): 'CutthroatCompetitionand the Long Purse', Journal
Weekly48 (39): 4135-43.
of Law and Economics, 9: 259-77.
Dasgupta,S and S Titman (1998): 'Pricing Strategyand$inancial Policy', Titman,S (1984), 'The Effect of CapitalStructureon the Firm's Liquidation
Review of Financial Studies 11 (4): 705-37.
Decision', Journal of Financial Economics 26: 137-52.

differencesareinstrumentedwith suitablelags of theirown levels. Strictly

exogenous regressors, as well as any other instruments,can enter the
instrumentmatrix in the conventional instrumentalvariables fashion:
in first differences, with one column per instrument.
5 Ina separateregression,we observedthatincreasein profitabilitypositively
affects the firm's growth in domestic sales.

References

Green

Left

Weekly

Australia's most widely read left newspaper

i g

Subscription rates:

D6 months(22 issues)$A82.50
($A148.50outside Asia-Pacific)

($A198 outside Asia-Pacific)

rq~*

by cheque/moneyorder/creditcard

i.
nol~~~~~~~
;Pay
dleprtai n!
--iM

876

612 9690 1230

PyaPhone:

Fax: 612 9690 1381

Mail: PO Box 394, Broadway NSWAustralia 2007

|

t msubscriptions@greenleft.org.au
EnEmail:
}

Economicand PoliticalWeekly Febraury26, 2005

This content downloaded from 111.68.103.104 on Mon, 14 Jul 2014 06:07:32 AM

All use subject to JSTOR Terms and Conditions