Do equity investments affect banks profitability?. Evidence from
OECD countries. Francisco Gonzlez* University of Oviedo Abstract
This paper analyzes the influence of equity investments on banks
profitability in a panel data of 24 OECD countries. The results
suggest a positive influence
ofbanksequityinvestmentsonbanksinterestratemarginandbanksnet
incomethatisnotoutweighedbyadditionalrequirementsofprovisionsand
capital that supervisory authorities establish to control bank
risk. The positive
effectequityinvestmentshaveonbanksinterestmarginisconsistentwith
thebanksabilityasshareholderstoobtainbenefitsinthelending
relationship they also keep with firms.
Keywords:equityinvestments,bankprofitability,regulation,bankingand
commerce, panel data. JEL Classification: G21, G24, G28. * The
author acknowledges the financial support provided by the Spanish
Ministry of
ScienceandTechnology,projectBEC2000-0982.Correspondenceto:Francisco
GonzlezRodrguez,DepartmentofBusinessAdministration,UniversityofOviedo.
AvenidadelCristoS/N,33071.Oviedo.Spain.Tel.:+34985103698.e-mail:
[email protected]. 1 Do equity investments affect banks
profitability?. Evidence from OECD countries. Abstract This paper
analyzes the influence of equity investments on banks profitability
in a panel data of 24 OECD countries. The results suggest a
positive influence
ofbanksequityinvestmentsonbanksinterestratemarginandbanksnet
incomethatisnotoutweighedbyadditionalrequirementsofprovisionsand
capital that supervisory authorities establish to control bank
risk. The positive
effectequityinvestmentshaveonbanksinterestmarginisconsistentwith
thebanksabilityasshareholderstoobtainbenefitsinthelending
relationship they also keep with firms.
Keywords:equityinvestments,bankprofitability,regulation,bankingand
commerce, panel data. JEL Classification: G21, G24, G28.
2Introduction
Oneofthesubjectsarousinggreatinterestanddebateinfinancialsystem
designisthedegreeofseparationthatshouldexistbetweenbankingand
commerce(Saunders,1994).Themainbenefitsindicatedfortheaffiliation
betweenbankingandcommerceincludereducingtheconflictsofinterests and
information asymmetries between the shareholders of the borrowing
firm
andtheirdebtholders.Offsettingthesebenefits,defendersofseparating
bankingandcommercearguethatenablingbankstoholdsharesin
borrowingfirmswouldincreasetheinstabilityofthebankingsystemupon
increasingthelevelofbankrisk.Bank'srisk-shiftingincentivescausedby
depositinsurance,whentheinsurancepremiumdoesnotreflectabanks
risk,havelongbeenrecognized1.Theriseofbankingriskwhenequity
investmentsareallowedobligesauthoritiestomonitorbanksandincreases
the supervision costs and the probability of bank
runs.Followingtheschemeofbenefitsandcostsassociatedwiththeaffiliation
betweenbankingandcommerce,thepreviousliteraturecontainsbasically
theoretical models focused on analyzing the socially optimal equity
stake as a trade-off between the increase in the risk of the bank's
asset portfolio and the
increaseinefficiencyofthefirmsinvestmentsuponloweringtheconflict
betweendebtholdersandshareholders(Boydetal.,1998;Johnetal.,1994;
Park, 2000).
Atthesametime,thedifferentpointofviewthatnationalregulatorshold
abouttheimportanceofcostsandbenefitshashinderedaninternational
coordinationofnationalregulationsdealingwiththeaffiliationbetween
bankingandcommerce.However,therearesomecommonpatternsbetween
countriesontheregulationsaboutbanksownershipofnon-financialfirms.
These regulations usually limit a banks investment in the equity of
a firm to a
certainpercentageofthebankscapitaland/orimposealimitonabanks
investment inthe equity ofafirmtoa certain percentage of either the
firms
capitaloritsvotingrights.Forexample,bankequityinvestmentsmaynot
exceed45%ofbankscapitalinBelgiumorItaly,60%inFinland,France,
Germany,Greece,PortugalorSpain,the25%inCzechRepublicorPoland.
Examples of the limits in the percentage of firms capital owned by
a bank are 50%inNorway,25%inPortugal,10%inCanadaandFinland,and5%in
Belgium,Japan,theNetherlands,andSweden.Germany,Spainand 1 See
Kareken and Wallace (1978), Merton (1977,1978) and Dothan and
Williams (1980).
3Switzerlandareexamplesofcountrieswherebanksarenotsubjecttothis
latter type of regulation. As a result, banks in these countries
can be the sole
ownerofnon-financialfirms.Ontheotherhand,intheUSbankingand
commercehavebeentraditionallyseparatedsincetheGlassSteagallActof
1933 which meant that equity investments can only be made by bank
holding companiesprovidedthattheydonotrepresentmorethan5%ofafirm's
votingshares.Recently,theGramm-Leach-BlileyFinancialServices
ModernizationActof1999hasloosenedrestrictionsonbankownershipof
equityinnon-financialfirms,althoughthelawcarefullymaintainsthe
separationofbankingandcommercebylimitingthetimethatbanksmay hold
such equity stakes and the amount of such holding relative to the
banks capital2.
Despitethedifferencesinlegallimitsonbankequityinvestments,countries
havetwobasicmechanismstokeeptheadditionalriskderivedfromequity
investmentsundercontrol,no matterwhatthemaximumlegallimitofthem
is.These two basic mechanisms are provisions and the capital
adjusted-risk
ratio.Regulatorsimposefundprovisioningonbankstocompensatefuture
capital loss. Moreover, the adoption of the capital adjusted-risk
ratio for most
ofthecountriesafter1993followingthe1987BasleAccordonCapital
Standards3alsorepresentsacontinuousmechanismofkeepingtheeffectof
equityinvestmentsonbankriskundercontrol.Asequityinvestmentsare
consideredriskyinvestments,bankswithgreaterequityinvestmentsare
required to have a higher capital. Although theoretical literature
has agreed that the affiliation between banking and commerce has
consequences for both the bank-owned firm and the owner
bank,mostoftheempiricalliteraturehasfocusedonanalyzingthe
consequences for the efficiency of borrowing firms. However, the
ability of the
ownerbanktoextractafirmssurplusortransferringfundsfromthefirm
forces us to complete the analysis of the effect on the firms
profitability with
thatoftheeffectonbanksprofitabilityinordertoevaluateallofthe
consequencesoftheaffiliationbetweenbankingandcommerce4.However,
2SeePecchioli(1987),Schuijer(1992)orBarthetal.(2002)foradescriptionofthe
regulationsontheassociationbetweenbankingandcommerceinseveralcountries.A
historical perspective on changes in the US system is given in
Berger et al. (1995).
3Anexampleofthewideestablishmentoftheminimumrequiredcapital-to-assetratio
according to the Basle guidelines is the fact that a hundred out of
the one hundred and seven countries analyzed by Barth et al.
(2001b) have adopted it.
4Inspiteofagreementofthetheoreticalworksonforecastinganimprovementonthe
firmsefficiencywiththebankshareholding,theavailableevidenceisnotconclusive.
4only very recently Barth et al. (2001a, 2002) have analyzed in a
sample of 60
and107countriesrespectively,theeffectsthatdifferentnationalregulations
ontheabilityofbankstoownandcontrolnonfinancialfirmshavebothon
financial sector performance and banking system stability in these
countries. Using country-level data they do not find beneficial
effects from restricting the
mixingofbankingandcommerce.Despitenotfindingareliablerelationship
between the restrictions on mixing banking and commerce and the
level of the
bankingsectordevelopment,theyconcludethatthosecountriesthatrestrict
banksfromowningnonfinancialfirmshaveamuchhigherprobabilityof
suffering a major banking crisis.In this paper, we provide
additional empirical evidence analyzing the influence of the banks
investments in the equity of firms on banks interest rate margin
andonbanksnetincomeusingacountry-levelpaneldataof24OECD
countries.IncontrasttoBarthetal.(2001a,2002),weusetherealratioof
banks equity investments to total banks assets in each country
instead of an index, ranging from 1 to 4, of the national
restrictions on the ability of banks
toownandcontrolnonfinancialfirms.Wealsotakeintoaccountthe
unobserved country heterogeneity using a panel database from 1987
to 1997.
Intheregressionswealsocontrolforotherbanksbalance-sheetvariables,
macrovariablesandvariablesofthebankingandfinancialmarket
development in the country.
Evenafterconsideringbanksprovisionsandcapitalrequirementswefinda
positiveinfluenceofbanks'equityinvestmentsonbanks'profitability.This
resultsuggeststhatthecontrolofbankriskwhichiscarriedoutbythe
authorities under provision and bank's capital regulations does not
outweigh
thehigherreturnsobtainedfrominvestmentsintheequityofnonfinancial
firms. The positive influence of bank equity investments in banks
interest rate
marginalsosuggeststhatbankshareholdingallowsthebanktotake
advantageofitslendingrelationshipwiththefirm.Infact,thisisthemain
benefitofthebankequityinvestmentbecausewedonotobservedifferences
AlthoughthereareworkscarriedoutinJapan,GermanyandSpain,theresultsare
contradictoryevenwithinthosedoneinthesamecountry.Amongtheworkswhichfind
evidenceofapositiveeffectofbankshareholdingonthefirmsefficiencywehaveKim
(1991),Hoshi et al. (1991) and Pushner (1995) all in Japan, Cable
(1985) and Gorton and
Schmidt(2000)inGermanyinthe1970sandZoido(1998)inSpain.Nevertheless,the
positiverelationshipbetweenbankshareholdingandthefirmsefficiencyisneither
observed by Weinstein and Yafeh (1998) in Japan, nor by Gorton and
Schmidt (2000) in Germany in the 1980s nor by Bergs and Snchez del
Villar (1991) in
Spain.5inbanksprofitabilitycausedbycapitalgainsorlossesderivedfromequity
transactions. The rest of the paper is structured as follows:
Section 2 describes the theory
behindourempiricalstudyinmoredetail.Section3dealswiththe
characteristicsofthedatabaseandthemethodologyusedandtheempirical
results are analyzed in Section 4. Finally, Section 5 presents the
conclusions. 1.Theoretical Background The financial literature does
not predict a clear effect of equity investments on
banksprofitability.First,theportfolio
theorysuggeststhatsinceinvestment
inequityisriskierthaninvestmentindebt,thosebanksincreasingthe
proportionoftheirinvestmentsinequitysecuritiescanexpecttheirportfolio
to bring higher profits and with it higher risk. The mere reference
to the high
varianceofreturnonequityinvestmenthasbeentakenasevidenceofits
incrementaleffectuponthebanksportfolio.Basedonthisidea,regulators
fromcertaincountriesjustifytheregulationthatseparatesbankingfrom
commerceasatooltoreducetheinstabilityinthebankingsystemandthe
probability of a banking crisis. However, economists have long
recognized that the introduction of any assets into the allowable
investment set may improve
theportfoliosrisk-returnefficiency.Aslongasthenewassetislessthan
perfectlypositivelycorrelatedwiththeexistingportfoliocomponents,its
addition may prove mean-variance improving (Langohr and Santomero,
1985).
Apartfromasimplesubstitutionofdebtformoreriskyassetsinthebank's
asset portfolio, the presence of a bank in the ownership of the
borrowing firm can modify the firms investment decision and give
rise to additional increases on the expected return and bank's risk
on the originally suggested by portfolio
theory.Park(2000)takesintoaccounttheagencyconflictswhichtakeplace
between the bank, the firm and the regulator in order to prove that
the bank's ownership of a commercial firm may also increase the
bank's profitability and
risk.Theconsequenceofanimprovementinthefirmsinvestmentefficiency
once the problems of under-investment and over-investment in the
firm have beenreducedwouldbeanincreaseinthebank'sprofitability.The
consequence of a higher banks incentive to allow the firm to
undertake risky
projectswouldmeananincreaseinthebank'srisk.Theparticipationinthe
surplusexpectedfromriskyprojectscanchangetheinitialpositionthata
6bank would have to avoid funds transfers when it is an only a
firm's creditor
(SmithandWarner,1979).Sincethefirmisallowedtoundertakeriskier
projects,boththeequityanddebtofthefirmmaybecomeriskierand
consequently the banks asset portfolio, too. On the other hand,
Boyd et al. (1998) also suggest an additional reason for a
positiveinfluenceofequityinvestmentsonbanksprofitabilitybasedonthe
ideathatbankequitypositionsonnonfinancialfirmsstrengthenthebanks
abilitytoextractsurplusfromborrowers5.Thus,ownerbankscanalso
increasetheirprofitabilitynotonlybythehigherexpectedreturnsofriskier
investments but also by the fact that a larger portion of the
surplus generated by externally financed investments accrues to
banks, and less accrues to the originating investor. Although, Boyd
et al. (1998) agree with Park (2000) when
pointingoutthepositiveinfluenceofequityinvestmentsonbank's
profitability, they do not agree on the consequences on the firm's
efficiency. In Boyd et al.'s (1998) model, nonbanking shareholders
have fewer incentives to strive for due to the funds' expropriation
carried out by the shareholder bank and the consequence is a
reduction of the firms efficiency.
Contrarytothepreviousarguments,Santos(1999)statesthatbankequity
investmentsdonotincreasebanksprofitabilityandrisk. Hisview is
thatby limiting the banks ability to use equity, the regulator
forces the bank to use
moredebtinordertochannelthenecessaryfundsintothefirm.This outweighs
the effects of the reduction of the banks stakes in the firms
capital and, insome cases,itmight evencreatethenegative effect
ofincreasingthe
banksriskoffailurebecauseheassumesthatdebtisthebank'spreferred
financialinstrumenttoencouragethefirmtoincreasetheriskofits
investment projects. This effect is more significant when firms
depend largely on banks to raise external funds or on bank oriented
systems, and when the
regulationlimitsbanksequityinvestmentstoacertainpercentageofthe
firms capital. Therefore, Santos (1999) suggests that there is no
relationship
betweenbanksequitypositionsincommercialfirmsandthebanks
profitabilitybecausethebankoutweighsanydifferenceintheequity
investment in order to reach the banks mean-variance target by
means of the lending decision.
5Thebanksabilitytoobtainsurplusfromfirmswhenthebankonlyhasalending
relationshipwiththefirmhasalsobeenanalyzedbyRajan(1992)andVonThadden
(1995).
7Nevertheless,allthepreviouspredictionsaboutbanksprofitabilityandrisk
donottakeintoaccountthesupervisoryactivitydevelopedbyauthoritiesto
controlbankrisk.Thus,differencesinrisk-returnbankportfolioscausedby
differencesinthelevelofequityinvestmentsbeforeconsideringsupervisory
activity could disappear after this is considered. In this sense,
two of the main
mechanismsofbankssupervisionaretheminimumrequirementsofbanks
capital and provisions. On the one hand, the highest risk of the
asset portfolio
inbankswithagreaterproportionoftheirassetsinvestedinequityof
nonfinancial firms could be compensated for by a higher percentage
of capital
entailedbytherisk-adjustedbank'scapitalrequirement.Thehighercostof
equity compared to cost of debt for banks will have negative
consequences for
bank'sprofitabilitythatcanoutweighthehigherexpectedreturnsofequity
investments.Ontheotherhand,similartocapitalregulation,ifbankswith
higherequityinvestmentsareobligedtosupplyhigherprovisions,theymay
loseafterprovisionstheexpectedhigherprofitabilitythatequityinvestments
originatebeforeprovisions.Therefore,thedifferencesinbanksprofitability
beforeconsideringprovisionsandcapitalrequirementscoulddisappearafter
considering the effect of both types of legal
requirements.Usingprovisionsandcapitalrequirementsasvariablestocontroltheriskof
bankassetportfolioimplicitlyinvolvesusingtheriskmeasuredevelopedby
thenationalsupervisorwhenfixingtheminimumlevelsforprovisionsand
capital.Ifweassumethatnationalsupervisoryauthoritieshavemore
informationaboutbanksthananyotheroutsideinvestor,measuringriskin
this way would be better than using risk measures defined on the
basis of the banks account statements.
Thedifferenthypothesesdiscussedinthissectiondonotclearlysuggest
whicheffectofequityinvestmentsmadebybanksshouldprevail.Hence,we
shallnotformulateahypothesisonaprevailingeffect.Instead,weexamine
the data to ascertain which effect prevails. We have estimated the
relationship
inOECDcountriesbetweentheproportionoftotalbanksassetinvestedin
equityofnonfinancialfirmsandbanksnetinterestincomeandbanksnet
income.Wehavealsotakenintoaccounttheconsequencesofequity
investments on banks risk adopting the method of risk measurement
used by
supervisoryauthorities.Thus,wehaveincludedintheregressionsas
independentvariablesthetwobasicinstrumentsthatauthoritiesuseto
controlbank'srisk:provisionsandcapitalratios.Additionally,wehavealso
included control variables related to the financial development of
the country, macro variables and other banks variables. 8 2.The
dataWe have used time series and cross-country data derived from
balance sheets and income statements of commercial banks in OECD
countries, as available from the Bank Profitability database
published by the OECD. Although some
OECDcountrieshavestate-ownedandcooperativebankswehaveonly
considered commercial banks in each country to avoid confusing
effects from thetype ofbank ownership.Initially, the datasetcovers
all OECDcountries
(29)from1987to1997.Togetherwiththeinformationobtainedfrombanks'
financialstatementswehaveobtainedadditionalinformationaboutthelevel
ofinflationandgrowthofeachcountryfromtheOECDHistoricalStatistics.
Informationaboutthefinancialdevelopmentandstructureofeachcountry
wasobtainedfromtheFinancialStructureDatabasecompiledbyBecketal.
(2001). Table 1 summarizes the variables used in this paper and
their source. Insert Table 1`
Thefinalstudywascarriedoutin24OECDcountriessinceinformationon banks
equity investments (EQUINV) for Canada, the Czech Republic, Poland,
theUKandtheUSwasnotavailable.Asinformationonthevariablesused was
not available in each country for all years from 1987 to 1997 we
have an unbalanced
panel.Wehaveusednetinterestincome(NETINTER)andbanksnetincome
(NETINC) to total bank assets as measures of banks profitability.
Net interest
incomeorbanksinterestmarginistheinterestincomeminusinterest expense
over total assets and it captures the profitability of the
intermediation activity of banks. Its potential relationship with
banks equity investments will show whether banks use their
shareholders position to increase their benefits on the lending
relationship through higher interest margins Net income is net
interest income plus non-interest earnings minus overhead
costandprovidesameasureofbankprofitsbeforeprovisionsandtaxes.As the
capital gains or losses and dividends of firms' shares are included
in the net income but not in the net interest income, the
relationship between banks 9equity investments and banks net income
would depict not only the effect on thebanks
lendingrelationshipbutalsoonthedirectyields(capitalgainsor
losses,dividends)thatbanksgetfromtheinvestmentinthefirmsequity.
When comparing both estimations we can separate both effects.
Inordertoanalyze whetherthebank's risk control mechanisms
designedby
theregulators(provisioningandbankcapitalrequirements)decreasethe
differencesthatequityinvestmentscouldcauseinbank'sprofitabilitywe
includetheprovisions(PROVIS)andbankcapitalratios(CAPITAL)inthe
regressions.Weincludetotalbankprovisions(provisionsonloansand
securities)insteadofonlyprovisionsonsecuritiesbecausethepotential
effects of the banks equity investments on the lending relationship
that banks
alsokeepwiththefirmscanalsoproducesomeinfluenceonprovisionsfor
loans. Moreover, separate information of provisions for loans and
securities is not available for most of the countries, which
prevents us from including both variables independently of each
other in regressions.
FollowingDemirgc-KuntandHuizinga(2001)wealsoincludeotherbank
characteristics that along with equity investment could give rise
to differences in banks profitability, such as LOANS, non-interest
earnings (NONINTER) and
banksoverheadcost(OVERHEADC).Totalbanksassetsdivideallthese
previous banks variables in each year. We include the annual growth
rate of real GDP per capita (GROWTH) and the inflation rate
(INFLATION) since they are macro variables potentially affecting
banks'profitabilityineachcountry.Additionally,Demirgc-Kuntand
Huizinga(2001)haveshowntherelevanceoffinancialdevelopmentoverthe
financialstructureofthecountryinorderto
explainthebanksprofitability. In a sample of developed and
developing countries they found that banks have both higher profits
and interest margins in underdeveloped financial systems
andoncetheycontrolfortheleveloffinancialdevelopment,financial
structure,i.e.therelativedevelopmentofbanksversusmarkets,doesnot
haveanindependenteffectontheirprofitabilityorinterestmargins.To
controlthedevelopmentofthefinancialsystemweincludethevariables
proposedbyBecketal(2001)formeasuringthevolumeandactivityofthe
bankingsectorandthestockmarket.Tomeasurethesizeofthebanking
sectorweusetheratioofthetotaldomesticassetsofdepositmoneybanks
divided by GDP (BANKASSET). The size of the stock market is proxied
by the ratioofthestockmarketcapitalizationdividedbyGDP(MAKTCAP).To
measureactivity,weusethecredittotheprivatesectorbydepositmoney
10banksdividedbyGDP(PRIBC)toproxythecreditactivityofthebanking
sector while the total value of stocks traded divided by GDP
(SMTVT) is of the stock market activity. To measure national
regulatory restrictions we use the index of restrictions on
banksowningnon-financialfirmselaboratedbyBarthetal.(2001b)
(RESTRICT).Thisindexrangesfrom1to4withahighervalueindicating
higherrestrictionsintheabilityofbankstoownandcontrolnon-financial
firms. The grading scale of the index is the following: (1)
Unrestricted- a bank may own 100% of the equity in any
non-financial firm. (2) Permitted- a bank may own 100% of the
equity in a non-financial firm, but ownership is limited based on
banks equity capital. (3)
Restricted-abankcanonlyacquirelessthan100%oftheequityina
non-financial firm. (4)
Prohibited-abankmaynotacquireanyequityinvestmentinanon-financial
firm. Inoursamplethereisnocountrywherebanksarenotallowedtoacquire
equityinnon-financialfirms,whichiswhytheindexinOECDcountries ranges
from 1 to 3.
Table2providesthelistofcountriesincludedinthisstudyandthemean
values of the country-level variables used in the empirical work.
Insert Table 2` The correlations matrix of variables used in the
empirical analysis is shown in
Table3.FocusingonthecorrelationsofEQUINVweobservethatitis
positivelycorrelatedwiththetwovariablesofbanksprofitability(NETINTER
andNETINC).EQUINVandLOANScorrelatenegatively,whichisconsistent with
the fact that equity and debt securities are alternative
investments, and
thereforethehighertheequityinvestment,thelowertheproportionofbank
total assets invested in loans. Moreover, as the equity investments
are riskier
thanotherassets,bankswithhigherequityinvestments(andlowerloans)
11also have higher provisions and higher capital requirements.
Thus, EQUINV is positively correlated with PROVIS and CAPITAL.
MacrovariablesalsocorrelatewithEQUINV.Theannualgrowthrateofreal GDP
per capita has a positive correlation coefficient while the annual
inflation
hasanegativeone.Regardingfinancialdevelopmentvariables,EQUINV
correlatesnegativelywiththesizeofthestockmarket(MAKTCAP)but
positively with their activity level
(SMTVT).However,thereisnocorrelationbetweentheindexofrestrictionsonbanks
owningnon-financialfirmselaboratedbyBarthetal.(2001)andthe
percentage of equity investments undertaken by banks to total bank
assets in
eachcountry.Theabsenceofcorrelationbetweenthesetwovariables
indicates that regulatory restrictions on banking and commerce
affiliation are not binding and that other variables are the main
determinants of the equity
investmentactivityofbanksineachcountry.Inthiscase,usingthereal
values of banks' equity investments can be more suitable than using
an index of the range of legal restrictions in each country when we
aim to analyze the
consequencesonbanks'profitabilityduetotheaffiliationbetweenbanking
and commerce. Insert Table 3` 3.Multivariate
analysis3.1.Methodology
Theavailabilityofapaneldataenablesustocorrectcountry-specificand
time-specificeffects.TheBreuschandPagan(1980)Lagrangemultipliertest
(LMtest)rejectsthenullhypothesisthaterrorsareindependentwithin
countries, i.e., the country effects are relevant and OLS
estimations would be biased. The country effect can be fixed or at
random. It is clear that the fixed
effectsmodelisaparticularcaseoftherandomeffectsmodelwhenthe
variablerepresentingthecountryeffectsisnon-stochastic.However,as
pointedoutbyHsiao(1986),whentheindividualeffectsarecorrelatedwith
theregressors,therandomeffects'modelproducesbiasedestimationsof
coefficients. In such cases, considering these effects as fixed
leads to the same 12results as when such a correlation is
explicitly included in the model. In this case, we use the Hausman
_2 to test the null hypothesis that the random and
fixedeffectsmodelsarenotsignificantlydifferent.Inalltheestimationsthe
Hausmans test rejects the null hypothesis and only the results
obtained with
thefixedeffectsGLSestimationarereported.Weestimateatwo-wayfixed
effectmodelbecausewealsoincludetimedummyvariablesforeachyearto
captureanyunobservedmacroeconomictimeeffectnotincludedinthe
regression that is country invariant. The equations estimated to
analyze the influence of banks equity investments on banks
profitability are as follows6: | | 1FD M B it i it it it q o | o +
+ + + + = it INwhere INit is the dependent variable (either
NETINTER or NETINC) for country
iinyeart,Bitarebanksvariablesforcountryiinyeart,Mitrepresentsthe
macro variables, FDit are the financial development variables,i is
a country-specific effect and qit is a white-noise error term. As
theabove estimationsassume exogeneityofthe explanatoryvariables, we
alsouseinstrumentvariablesmodelstocontrolforpotentialbiasesdueto
endogeneityandtochecktherobustnessoftheresults.Weconstruct
instrumentsforthebalancesheetandincomestatementvariablesofthe
right-hand-side.Inparticular,weuselaggedvaluesofthesevariablesas
instruments.Weworkwithtwolagstoavoidcasesinwhichtheremightbe
first-order autocorrelation of the residuals. This technique
assumes that past
valuesoftheexplanatoryvariablesarenotcorrelatedwiththe
contemporaneous values of the explanatory variables. 6 Demirgc-Kunt
and Huizinga (2001) use a similar specification to analyze the
impact of financial development and financial structure on bank
profits and margins in a sample of
banksfromdevelopedandunderdevelopedcountries.However,theydonotconsider
the
influenceofbankequityinvestmentsandtheydonotcontrolforcountry-specificand
time-specific effects as they use mean values for each variable
over the sample period or
betweenestimations.However,wetakeadvantageofourpaneldatabasetocontrolfor
unobserved country heterogeneity using within estimations.
133.2.Results
Theresultsoftwo-wayfixedeffectsorwithinestimationsontheinfluenceof
equityinvestmentsonthenetinterestincomeareshownintable4.Table5
showsthetwo-wayfixedeffectsestimationswithinstrumentsforbanks'
variables.
EQUINVhasapositiveandstatisticallysignificantcoefficientinallthe
models.Thisresultisconsistentwithbanks'usageoftheirshareholder
positiontoincreasetheirinterestmarginsandtoobtainbenefitsinthe
lendingrelationshipstheyalsokeepwithfirms.Thispositiveinfluenceof
equity investments on banks' net interest income remains after
controlling for
theprovisionsandcapitalratiosrequiredbythenationalsupervisory
authoritiestocontrolbankrisk.Thus,thepositiveeffectofbankequity
investmentsonnetinterestincomeisnotoutweighedbytheadditional
requirements(provisionsandcapitalratios)thatsupervisorsestablishto
controlthepotentialhigherriskofbankequityinvestments.Moreover,this
positiveinfluenceisobtainedbothinthewithin(table4)asininstrumental
variables (table 5) estimations.
ThecoefficientsofPROVISarepositiveandstatisticallysignificantinall
specifications. CAPITAL also has positive coefficients in the
within estimations while it does not have statistically significant
coefficients in the instrumental
variablemodels.ThepositivecoefficientsofPROVISandCAPITALmay
indicatethathighernetinterestincomesareassociatedwithriskier
investment portfolios that also oblige banks to have higher capital
ratios and
provisions.Nevertheless,ashasjustbeenshown,thepositivecoefficientof
EQUINV remains statistically significant after including PROVIS and
CAPITAL in the regressions.
Thepercentageofloansontotalbankassetsdoesnotappeartohavea
statistically significant effect on banks interest margin.
Non-interest earnings
arenegativelyrelatedtonetinterestincomesinallestimationssuggestinga
possiblesubstitutioneffectbetweenthesetwotypesofbankincomes.
However, the overhead cost variable has positive coefficients in
the fixed effect
estimations.ThiseffectisconsistentwiththeresultsobtainedbyDemirgc-KuntandHuizinga(2001)forasampleofcountrieswithdevelopedand
underdevelopedfinancialsystemsindicatingthatbankscanpassonthese
coststotheircustomers.However,whenweuseinstrumentalvariablesin
Table5tocorrectpotentialproblemsof endogeneity,thenegative
effectdoes 14not remain and we do not obtain significant
coefficients for OVERHEADC. The
macrovariables(GROWTHandINFLATION)arepositiveandstatistically
significant indicating that banks have greater interest margins in
inflationary and growing
environments.UnlikeDemirgc-KuntandHuizinga(2001),thefinancialdevelopment
variablesusedasacontrolarenotstaticallysignificantinanyspecification.
Although not reported in the paper, we have included the variables
measuring thesizeandtheactivityofthebankingsectorandthestockmarket
simultaneously.Wehavealsoincludedalternativevariablesoffinancial
developmentsuchasthestockmarketturnoverratio.Nevertheless,the
resultsdidnotchange.Thisnon-significanteffectoffinancialdevelopment
variablesinOECDcountries,contrarytotheresultsobtainedbyDemirgc-kuntandHuizinga(2001)forasampleofdevelopedandundeveloped
countries, may be the result of the higher similarity of financial
development among OECD countries. Insert Table 4` Insert Table 5`
In addition to the effect on banks' interest margin we also look at
the effect of equity investments on bank's profitability when
analyzing banks' net income.
Asthisbankprofitabilityvariableincludesnetinterestincomeandthenon-interestearningswearealsocapturingothereffectsofequityinvestments
additionaltothoseinthelendingrelationship.Theresultsofthese
estimationsareshownintables6and7.Resultsarebasicallyanalogousto
those shown in tables 4 and 5. Insert Table 6` Insert Table 7` 15
EQUINVhaspositivecoefficientsinallspecificationssuggestingthatthe
positiveinfluenceofequityinvestmentsonbanksinterestmarginremains
whenweincludethecapitalgainsorlossesofequityinvestmentsandother
non-interest earnings in the measure of bank profitability
analyzed. CAPITAL
andPROVISalsokeeptheirpositivecoefficientsinthewithinestimations
whereasPROVISdoesnothavestatisticallysignificantcoefficientswhenwe
usetwolagsasinstrumentsofbanks'variables.Thus,equityinvestments
have a positive effect on overall bank profitability after taking
into account the
requirementsthatsupervisoryauthoritiesimposetocontroltheadditional
bank risk that can derive from higher bank equity investments.
Non-interest earning variable, NONINTER, has positive coefficients
consistent with its contribution to bank profits. Likewise, as
overhead costs are expenses
includedinthebanknetincome,theyhaveanegativeinfluence.Macro
variables (GROWTH and INFLATION) have the same positive influence
shown in tables 4 and 5, and the variables of the development of
the financial system
donothavestatisticallysignificantcoefficientsasinthenetinterestincome
regressions. Although not shown in the paper we have also tested
additional specifications to check the robustness of the results.
Thus, we have also analyzed the effect
ofequityinvestmentsonbanks'profitbeforetaxes.Inthiscase,thehigher
provisionsforwhichahigherpercentageofequityinvestmentsareneeded
have been directly deduced from the dependent variable and PROVIS
has not
beenincludedasanindependentvariable.Wealsousedependentvariables
adjusted by risk. The net interest income and the net income in
each year are
divided,respectively,bythestandarddeviationofNETINTERandNETINC
overthe1987-1997period.TheresultsdidnotchangeandEQUINVshows
positive and statistically significant coefficients in all these
specifications. 4.Conclusions
Thispaperusesapanelcountry-dataof24OECDcountriesinorderto analyze
the influence of equity investments on banks' profitability. The
results
showthatbankequityinvestmentshaveapositiveeffectonnetinterest
16incomeandonnetincome.Thispositiveinfluenceremainsthesameafter
controllingforthepotentialincreaseofbankriskthathigherequity
investments can originate. Thus, the highest profitability that
portfolio theory
suggestsforbankswithhigherequityinvestmentsdoesnotdisappearafter
consideringthehighestprovisionsandcapitalratiosthatthesebanksare
obliged to keep. The positive influence on net interest income is
consistent with the view that
bankscanusetheirshareholderpositioninnon-financialfirmstoobtain
benefits in the lending relationship that they usually keep with
firms in which they also take equity. In fact, the positive
influence on banks interest margin is the main benefit of the bank
equity investments because we do not observe
differencesinbanks'profitabilitycausedbycapitallossesorgainsderived
from equity transactions.
Thebank'sabilitytotakeadvantageofthebenefitsthatbankshareholding
canhaveonthefirm'sefficiencyalsopreventsempiricalworkswhichonly
analyze the effect on the firm's performance from being a complete
measure of
theconsequencesonbusinessefficiencythatthetheoreticalliteraturehas
suggested for the affiliation between banking and commerce.
5.ReferencesBarth, J.R., G. Caprio Jr. and R. Levine, 2001a,
Banking systems around the
globe:doregulationsandownershipaffectperformanceandstability?,in
FredericS.Mishkin,Editor:Prudentialsupervision:Whatworksandwhat
doesnt, University of Chicago Press.
Barth,J.R.,G.CaprioJr.andR.Levine,2001b,Theregulationand
supervisionofbanksaroundtheworld:anewdatabaseinRobertE.Litan
andRichardHerring,Editors,Integratingemergingmarketcountriesintothe
globalfinancialsystem,Brookings-WhartonPapersonFinancialServices,
Brookings Institution Press.
Barth,J.R.,G.CaprioJr.andR.Levine,2002,Bankregulationand
supervision: What works best? World Bank Research.
Beck,T.,A.Demirgc-Kunt,andR.Levine,2001,Thefinancialstructure
database,inFinancialStructureandEconomicGrowth,editedbyDemirgc-Kunt,A.andR.Levine,TheMITPress,Cambridge,Massachusetts,London,
2001.
17Berger,A.;A.KashyapandJ.Scalise,1995,ThetransformationoftheU.S.
banking industry: what a long, strange trip it has been, Brookings
Papers on Economic Activity 2, 55-218. Bergs, A. and E. Snchez del
Villar, 1991, Relaciones banca-industria desde la perspectiva del
mercado burstil espaol, in A. Torrero (editor), Relaciones
banca-industria. La experiencia espaola. Espasa Calpe. Boyd, J., C.
Chang and B. Smith, 1998, Moral hazard under commercial and
universal banking, Journal of Money, Credit and Banking 30,
426-471. Cable,J., 1985,Capital
marketinformationandindustrialperformance:The role of West German
Banks, The Economic Journal 95, 118-132. Demirgc-Kunt, A. and H.
Huizinga, 1999, Determinants of commercial bank
interestmarginsandprofitability:someinternationalevidence,Worldbank
economic review 13, 379-408.
Demirgc-Kunt,A.andH.Huizinga,2001,Financialstructureandbank
profitability, in Financial Structure and Economic Growth, edited
by Demirgc-Kunt, A. and R. Levine, The MIT Press, Cambridge,
Massachusetts, London. Dothan, U. and J. Williams, 1980, Banks,
bankruptcy, and public regulation. Journal of banking and Finance
4, 65-88. Gorton, G. and F.A. Schmid, 2000, Universal banking and
the performance of German firms, Journal of Financial Economics 58,
29-80. Hoshi, T., A. Kashyap and D. Schafstein, 1991, Corporate
structure, liquidity and investment: Evidence from Japanese
industrial groups, Quarterly Journal of Economics, February, 33-59.
Hsiao,C.,1986,Analysisofpaneldata.EconometricSocietyMonograph11,
Cambridge University Press, Cambridge. John, K., T.A. John and A.
Saunders, 1994, Universal banking and firm risk-taking, Journal of
Banking and Finance 18, 307-323. Kareken, J.H. and N. Wallace,
1978, Deposit insurance and bank regulation: A partial-equilibrium
exposition, Journal of Business 51, 413-438. Kim, S., 1991, The use
of equity positions by banks: The Japanese evidence, Economic
Review of the Federal Reserve Bank of San Francisco, Fall, 41-55.
Langohr,H.andA.M.Santomero,1985,Theextentofequityinvestmentby
European banks, Journal of Money, Credit, and Banking 17, 243-252.
Merton,R.C.,1977,Ananalyticderivationofthecostofdepositinsurance
and loans guarantees, Journal of Banking and Finance 1, 512-520.
18Merton,R.C.,1978,Onthecostofdepositinsurancewhenthereare
surveillance costs, Journal of Business 51, 439-452. Park, S. 2000,
Effects of the affiliation of banking and commerce on the firms
investmentandthebanksrisk,JournalofBankingandFinance24,1629-1650.
Pushner,G.1995,Equityownershipstructure,leverageandproductivity:
Empirical evidence from Japan, Pacific-Basin Finance Journal 3,
241-256. Pecchioli, R.M., 1987, Prudential supervision in banking.
OECD, Paris.
Rajan,R.G.,1992,Insidersandoutsiders:thechoicebetweeninformedand
arm`s length debt, Journal of Finance 47, 1367-1400. Santos, J.,
1999, Bank capital and equity investment regulations, Journal of
Banking and Finance 23, 1095-1120. Saunders, A., 1994, Banking and
commerce: An overview of the public policy issues, Journal of
Banking and Finance 18, 231-254.
Saunders,A.andL.Schumacher,2000,Thedeterminantsofbankinterest
ratemargin:aninternationalstudy,Journalofinternationalmoneyand
finance 19, 813-832. Schuijer, J., 1992, Banks under stress. OECD,
Paris. Smith, C.W. and J.B. Warner, 1979, On financial contracting:
an analysis of bond covenants, Journal of Financial Economics 7,
117-161.
VonThadden,E.L.,1995,Long-termcontracts,short-terminvestmentand
monitoring, Review of Economic Studies 62, 557-575. Weinstein, D.E.
and Y. Yafeh, 1998, On the cost of a bank-centered financial
system:evidencefromchangingmainbankrelationsinJapan,Journalof
Finance 53, 635-672.
Zoido,M.E.1998,Unestudiodelasparticipacionesaccionarialesdelos
bancosenlasempresasespaolas,InvestigacionesEconmicasXXII,427-467.
19Table 1 The variables This table describes the variables
collected for the 29 OECD countries. We present the description and
the sources from which each variable is collected.
VariableDefinition Banks Characteristics
EQUINVBanksequityinvestmentsovertotalbanks'assets.Bank
Profitability (2000), OECD Publications.
NETINTERInterestincomeminusinterestexpenseovertotalassets.Bank
Profitability (2000), OECD Publications.
NETINCGrossincome(netinterestincomeplusnon-interestearnings)
minusoperatingexpensesovertotalassets.BankProfitability (2000),
OECD Publications.
PROVISProvisionsovertotalassets.BankProfitability(2000),OECD
Publications.
CAPITALBookvalueofequityovertotalassets.BankProfitability(2000),
OECD Publications.
LOANSTotalloansovertotalassets.BankProfitability(2000),OECD
Publications.
OVERHEADCPersonnelexpensesandsomeothernon-interestexpensesover
total assets NONINTERNet profit or loss on financial operations,
net fees and commissions and other non-interest earnings over total
assets. Bank Profitability (2000), OECD Publications. Macro
Indicators GROWTHAnnualgrowthrateofrealGDPpercapita.OECDHistorical
Statistics (2001), OECD Publications. INFLATIONThe annual inflation
from consumer price indices. OECD Historical Statistics (2001),
OECD Publications. Financial Development
BANKASSETTotalassetsofdepositmoneybanksdividedbyGDP.Becketal.
(2001): Financial structure database.
PRIBCPrivatecreditbydepositmoneybankstoGDP.Becketal.(2001):
Financial structure database.
MAKTCAPStockmarketcapitalizationtoGDP.Becketal.(2001):Financial
structure database.
SMTVTStockmarkettotalvaluetradedtoGDP.Becketal.(2001): Financial
structure database.
RESTRICTIndexofcountryrestrictionsonbanksowningnonfinancialfirms
constructedbyBarthetal.(2001).Thisindexrangesfrom1to4 with a higher
value of the index indicating higher restrictions in the ability of
banks to own and control nonfinancial firms. 20 Table 2 Bank
country-level characteristics This table shows the mean values of
the country-level variables used in the empirical work over the
1987-1997 period. EQUINVNETINTERNETINCPROVISCAPITAL
LOANSNONINTEROVERHEADC
Australia3.9252.5561.5010.4669.08757.3991.8042.858
Austria3,0251.7470.8670.4804.52350.0870.8661.744
Belgium1,0891.4300.6260.2853.03133.6550.4931.298
CanadaN.A2.6961.530.5565.18170.8621.3412.508 Czech
RepublicN.A2.8680.971.44710.18747.15010.27312.178
Denmark3.5972.9011.4931.2307.34143.5160.8892.297
Finland2.371.8810.0500.1755.87558.7321.8073.636
France3.0131.3640.7190.4283.97238.4970.7691.414
Germany2.7442.0450.9720.3933.88254.8680.6121.683
Greece4.1951.9341.5960.4784.85329.9702.3052.643
Hungary3.5124.8080.303-0.3128.97341.112-0.6733.832
Iceland2.2374.4491.9531.3837.48572.5052.1694.663
Ireland0.4472.3901.5250.1356.27054.3301.1702.032
Italy2.1812.8051.3330.6496.36843.0960.9692.443
Japan3.9281.2470.4440.2373.02959.1080.1560.959
Korea2.3722.1281.1021.1066.98051.3811.2662.292
Luxembourg0.5770.8410.7560.3452.95923.7190.3860.472
Mexico1.5255.1712.6371.5746.72061.4842.0754.610
Netherlands1.0191.9800.9440.2874.13958.5520.8911.927 New
Zealand0.7362.7241.3840.2074.92371.6681.6132.955
Norway1.9073.1061.3340.9425.24475.2240.9822.752
PolandN.A5.1753.2310.7728.40638.1362.0333.978
Portugal2.9062.9531.6830.9269.77242.0640.9552.223
Spain3.1723.4391.6470.6758.70444.9300.9062.699
Sweden2.2982.1530.744-0.0645.76147.5811.3632.773
Switzerland2.6761.3811.3020.6956.03256.5771.5471.591
Turkey4.8908.1144.3291.2265.00039.9860.4494.235
UKN.A2.6821.5960.6934.47657.1641,7182.804 United
statesN.A3.5741.9150.6347.18863.2951.8603.519 21 Table 2
(continuation) Bank country-level characteristics This table shows
the mean values of the country-level variables used in the
empirical work over the 1987-1997 period. All variables are divided
by total bank assets.
GROWTHINFLATIONBANKASSETPRIBCMAKTCAPSMTVTRESTRICT
Australia2.2923.7230.6470.5660.5840.2522
Austria2.0662.4251.1990.8780.0970.0601
Belgium2.1382.1080.9240.4390.3190.0463
Canada1.4922.7230.5770.5090.5270.2263 Czech
Republic-0.14414.7300.8200.5560.2670.0973
Denmark1.4312.6460.5460.4150.2870.1072
Finland1.8423.0330.7350.7150.2390.0851
France1.7542.2150.9780.8470.2640.1252
Germany1.5612.2921.1600.9020.2280.2641
Greece1.33312.6750.4300.2060.1040.0351
Hungary2.32520.3150.3730.3180.0390.009N.A
Iceland1.4008.2690.4460.4200.1100.0073
Ireland5.8502.5500.3560.2730.2650.1411
Italy1.7544.4380.7060.5070.1530.0603
Japan2.2541.1691.2571.1080.8300.4303
Korea5.8545.7300.5190.4930.3030.3483
Luxembourg4.5851.9770.3130.2972.3330.0211
Mexico1.13936.1690.1990.1570.2010.0883
Netherlands2.3611.9771.0190.8110.5650.3251 New
Zealand2.1083.7690.6260.5520.4610.1052
Norway2.2463.4000.6840.5450.2030.1072
Poland3.53313.4250.3530.0710.0280.0292
Portugal3.7547.0610.7770.5810.1040.0322
Spain2.9314.5380.9190.6580.2470.1601
Sweden1.4084.0000.5340.4460.5300.2363
Switzerland0.7002.3001.6671.5530.8591.0303
Turkey1.94672.6150.1950.1350.1010.0973
UK2.1614,1460.9630.9310.9640.4681 United
states2.1853.2770.7570.6620.6900.5013 22Table 3 Correlations This
table shows the correlation between the country-level variables
used in the empirical work over the 1987-1997 period
EQUINVNETINTERNETINCPROVISCAPITAL LOANSOVERHEADCNONINTERGROWTH
INFLATIONBANKASSETPRIBCMAKTCAPSMTVT NETINTER0.321**____
NETINC0.287**0.783**____ PROVIS0.170**0.379**0.476**____
CAPITAL0.203**0.337**0.295**0.237**____ LOANS-0.154**
0.068-0.0080.0840.073____ OVERHEADC0.210**
0.481**0.221**0.264**0.436**0.147**____
NONINTER-0.052-0.0670.0390.198**0.323**0.0880.760**____ GROWTH
-0.199** -0.0240.113*-0.154**0.067-0.137*-0.147**-0.065____
INFLATION0.388**0.829**0.710**0.263**0.065-0.190**0.307**-0.128**-0.160**____
BANKASSET0.027-0.503**-0.402**-0.170**-0.189**0.181**-0.342**-0.041-0.154**-0.443**____
PRIBCIT0.026-0.471**-0.379**-0.147**-0.164**0.333**-0.296**-0.011-0.144*-0.428**0.943**____
MAKTCAP-0.203**-0.306**-0.153**-0.116**-0.232**-0.132*-0.313**-0.0660.155**-0.224**0.0780.179**____
SMTVT0.143*-0.144*-0.019-0.088-0.0140.228**-0.0980.0650.117-0.172**0.423**0.519**0.304**____
RESTRICT0.1070.253**0.218**0.188**0.124*0.172**0.223**0.121*-0.130*0.246**-0.089-0.053-0.0910.155**
** Significant at 1 % level* Significant at 5 % level 23Table 4
Effect of equity investments on bank net interest income (Fixed
effects)
Thistableshowstheresultsofatwo-wayfixedeffectmodel.Thedependentvariableisthebanksnetinterest
income. As independent variables we include banks, macro and
country financial development variables. Banks
variablesaretheproportionofequityinvestments(EQUINV),thecapitalratio(CAPITAL),theprovisions
(PROVIS),theproportionofLOANS,thenon-interestearnings(NONINTER)andtheoverheadcosts
(OVERHEADC). All these variables are divided by the total bank
assets in the country. The annual growth rate of real GDP per
capita (GROWTH) and the annual inflation for consumer prices
indices (INFLATION) are the macro variables. Total assets of
deposit money bank assets to GDP (BANKASSET) and the stock market
capitalization
(MAKCAP)areintroducedtomeasurethesizeofthebankingsectorandthestockmarketrespectivelyinthe
country.Finally,theprivatecreditbydepositmoneybankstoGDP(PRIBC)andthestockmarkettotalvalue
tradedtoGDP(SMTVT)areincludedtomeasuretheactivityofthebankingsectorandthestockmarket
respectively. The standard errors are corrected for
autocorrelation, and the corresponding t-statistics are given in
parentheses below. Dependent variable: Net Interest Income
(NETINTER) (1)(2)(3)(4)(5)(6)(7)(8) INTERCEPT2.210*** (4.37)
2.040*** (4.05) 1.770*** (3.52) 1.413*** (2.87) 2.081*** (3.96)
1.931*** (3.68) 1.592*** (3.07) 1.254** (2.47) EQUINV0.145***
(3.10) 0.144*** (3.16) 0.114** (2.49) 0.106** (2.44) 0.117** (2.50)
0.120*** (2.64) 0.087* (1.92) 0.086** (2.00) CAPITAL 0.127***
(3.62) 0.156*** (4.61) 0.137*** (3.93) 0.164*** (4.88) PROVIS
0.205*** (3.29) 0.262*** (4.38) 0.209*** (3.31) 0.265*** (4.43)
LOANS0.000 (0.00) -0.003 (-0.37) -0.004 (-0.51) -0.008 (-1.11)
0.007 (0.75) 0.003 (0.33) 0.002 (0.25) -0.003 (-0.41)
NONINTER-0.874*** (-14.60) -0.901*** (-15.28) -0.915*** (-15.57)
-0.956*** (-16.85) -0.883*** (-14.50) -0.911*** (-15.21) -0.929***
(-15.62) -0.969*** (-16.96) OVERHEADC0.442*** (6.86) 0.511***
(7.73) 0.461*** (7.40) 0.555*** (8.81) 0.449*** (6.85) 0.518***
(7.74) 0.468*** (7.45) 0.560*** (8.87) GROWTH0.067*** (4.15)
0.073*** (4.60) 0.065*** (4.19) 0.073*** (4.88) 0.063*** (3.70)
0.069*** (4.10) 0.061*** (3.75) 0.068*** (4.37) INFLATION0.023***
(2.91) 0.021*** (2.76) 0.020** (2.54) 0.017** (2.26) 0.025***
(3.20) 0.023*** (2.98) 0.021*** (2.76) 0.017** (2.39) BANKASSET
-0.279 (-0.98) -0.193 (-0.70) -0.237 (-0.87) -0.118 (-0.45)
MAKTCAP-0.005 (-0.06) -0.014 (-0.16) -0.015 (-0.18) -0.027 (-0.34)
PRIBC -0.642 (-1.51) -0.488 (-1.17) -0.536 (-1.32) -0.310 (-0.79)
SMTVT 0.295 (1.15) 0.278 (1.12) 0.309 (1.26) 0.291 (1.25) R2
overall75.23%79.76%76.54%79.80%77.96%81.81%78.77%80.99%
F24.83***25.43***26.02***28.38***24.74***25.48***26.42***29.00***
LM _2202.35***83.05***
82.21***56.37***206.01***85.41***76.31***55.13*** Hausman
_2182.54***166.54***55.45***42.77***49.49***48.57***744.73***48.09***
# observations200198200198198196198196 # countries2424242424242424
*** Significant at 1 % level** Significant at 5 % level*Significant
at 10% level 24Table 5 Effect of equity investments on bank net
interest income (Fixed effects with instrumental variables)
Thistableshowstheresultsofatwo-wayfixedeffectmodelwithinstrumentsforbanks'variables.Thedependent
variable is the banks net interest income. As independent variables
we include banks, macro and country financial
developmentvariables.Bankvariablesaretheproportionofequityinvestments(EQUINV),thecapitalratio
(CAPITAL),theprovisions(PROVIS),theproportionofLOANS,thenon-interestearnings(NONINTER)andthe
overheadcosts(OVERHEADC).Allthesevariablesaredividedbythetotalbankassetsinthecountry.Weuseas
instruments two lags of each one of these banks' variables. The
annual growth rate of real GDP per capita (GROWTH)
andtheannualinflationforconsumerpricesindices(INFLATION)arethemacrovariables.Totalassetsofdeposit
money bank assets to GDP (BANKASSET) and the stock market
capitalization (MAKCAP) are introduced to measure the size of the
banking sector and the stock market respectively in the country.
Finally, the private credit by deposit money banks to GDP (PRIBC)
and the stock market total value traded to GDP (SMTVT) are included
to measure the
activityofthebankingsectorandthestockmarketrespectively.Thestandarderrorsarecorrectedfor
autocorrelation, and the corresponding t-statistics are given in
parentheses below. Dependent variable: Net Interest Income
(NETINTER) (1)(2)(3)(4)(5)(6)(7)(8) INTERCEPT2.398*** (3.12)
2.165*** (2.83) 2.413*** (3.09) 2.148*** (2.75) 2.595*** (3.13)
2.262*** (2.72) 2.595*** (3.07) 2.214** (2.60) EQUINV0.254***
(4.60) 0.220*** (3.87) 0.254*** 84.57) 0.220*** (3.86) 0.245***
(4.42) 0.209*** (3.68) 0.245*** (4.40) 0.210*** (3.68) CAPITAL
-0.007 (-0.12) 0.008 (0.13) 0.000 (0.00) 0.017 (0.27) PROVIS
0.198** (2.19) 0.200** (2.19) 0.207** (2.24) 0.210** (2.25)
LOANS-0.001 (-0.11) 0.000 (0.00) -0.001 (-0.08) -0.000 (-0.03)
-0.008 (-0.56) -0.004 (-0.30) -0.008 (-0.55) -0.005 (-0.35)
NONINTER-0.308*** (-3.29) -0.316*** (-3.41) -0.306*** (-3.22)
-0.318*** (-3.38) -0.308*** (-3.24) -0.314*** (-335) -0.308***
(-3.19) -0.318*** (-3.34) OVERHEADC-0.131 (-1.29) -0.111 (-1.10)
-0.130 (-1.28) -0.0112 (-1.11) -0.124 (-1.18) -0.102 (-0.99) -0.124
(-1.17) -0.103 (-0.99) GROWTH0.074*** (3.01) 0.077*** (3.15)
0.075*** (2.89) 0.076*** (2.95) 0.078*** (3.05) 0.080*** (3.17)
0.078*** (2.85) 0.077*** (2.86) INFLATION0.043*** (3.33) 0.047***
(3.60) 0.044*** (3.21) 0.046*** (3.41) 0.046*** (3.52) 0.049***
(3.79) 0.046*** (3.33) 0.048*** 83.51) BANKASSET -0.274 (-0.66)
-0.265 (-0.65) -0.281 (-0.67) -0.257 (-0.62) MAKTCAP-0.019 (-0.15)
-0.004 (-0.03) -0.019 (-0.15) -0.004 (-0.03) PRIBC -0.142 (-0.23)
-0.178 (-0.29) -0.142 (-0.23) -0.164 (-0.27) SMTVT -0.154 (-0.40)
-0.040 (-0.10) -0.154 (-0.39) -0.021 (-0.05) R2
overall58.18%64.66%58.20%64.70%53.65%61.91%.53.65%62.06%
F7.44***7.48***7.01***7.06***7.42***7.48***6.98***7.07*** LM
_295.73***51.88***22.43***15.39***88.45***52.16***18.37***13.70***
Hausman
_2129.58***54.06***216.90***59.6341.58***20.3862.96***36.14*** #
observations194194194194192192192192 # countries2424242424242424
*** Significant at 1 % level** Significant at 5 % level*Significant
at 10% level 25Table 6 Effect of equity investments on bank net
income (Fixed effects) This table shows the results of a two-way
fixed effect model. The dependent variable is the banks net income.
As independent variables we include banks, macro and country
financial development variables. Banks variables are the proportion
of equity investments (EQUINV), the capital ratio (CAPITAL), the
provisions (PROVIS), the proportion
ofLOANS,thenon-interestearnings(NONINTER)andtheoverheadcosts(OVERHEADC).Allthesevariablesare
divided by the total bank assets in the country. The annual growth
rate of real GDP per capita (GROWTH) and the
annualinflationforconsumerpricesindices(INFLATION)
arethemacrovariables.Totalassetsofdepositmoney
bankassetstoGDP(BANKASSET)andthestockmarketcapitalization(MAKCAP)areintroducedtomeasurethe
size of the bankingsector and the stock market respectivelyin the
country. Finally,the privatecredit by deposit money banks to GDP
(PRIBC) and the stock market total value traded to GDP (SMTVT) are
included to measure the
activityofthebankingsectorandthestockmarketrespectively.Thestandarderrorsarecorrectedfor
autocorrelation, and the corresponding t-statistics are given in
parentheses below. Dependent variable: Net Income (NETINC)
(1)(2)(3)(4)(5)(6)(7)(8) INTERCEPT2.242*** (4.41) 2.069*** (4.09)
1.798*** (3.57) 1.435*** (2.90) 2.067*** (3.94) 1.915*** (3.66)
1.573*** (3.04) 1.234** (2.43) EQUINV0.150*** (3.21) 0.149***
(3.27) 0.119*** (2.59) 0.111** (2.55) 0.120** (2.55) 0.123***
(2.69) 0.089* (1.97) 0.088** (2.05) CAPITAL 0.128*** (3.65)
0.158*** (4.65) 0.138*** (3.96) 0.165*** (4.92) PROVIS 0.209***
(3.34) 0.267*** (4.45) 0.210*** (3.33) 0.266*** (4.46) LOANS-0.001
(-0.10) -0.004 (-0.48) -0.005 (-0.62) -0.009 (-1.24) 0.007 (0.75)
0.003 (0.33) 0.002 (0.25) -0.003 (-0.41) NONINTER0.125** (2.08)
0.097* (1.65) 0.084 (1.42) 0.042 (0.74) 0.116* (1.90) 0.088 (1.47)
0.070 (1.18) 0.029 (0.52) OVERHEADC-0.553*** (-8.55) -0.482***
(-7.26) -0.533*** (-8.53) -0.438*** (-6.94) -0.550*** (-8.39)
-0.481*** (-7.18) -0.531*** (-8.45) -0.439*** (-6.98)
GROWTH0.066*** (4.08) 0.072*** (4.54) 0.065*** (4.13) 0.072***
(4.82) 0.064*** (3.73) 0.069*** (4.13) 0.062*** (3.78) 0.069***
(4.41) INFLATION0.022*** (2.80) 0.020*** (2.64) 0.019** (2.43)
0.016** (2.14) 0.025*** (3.17) 0.023*** (2.95) 0.021*** (2.72)
0.017** (2.35) BANKASSET -0.286 (-1.00) -0.200 (-0.72) -0.244
(-0.89) -0.123 (-0.47) MAKTCAP-0.002 (-0.02) -0.011 (-0.13) -0.012
(-0.14) -0.024 (-0.30) PRIBC -0.638 (-1.51) -0.484 (-1.17) -0.532
(-1.31) -0.304 (-0.78) SMTVT 0.320 (1.25) 0.303 (1.22) 0.334 (1.36)
0.316 (1.36) R2
overall12.76%22.77%15.69%26.51%20.59%30.12%22.58%30.67%
F7.32***7.94***8.18***9.63***7.51***8.17***8.61***10.15*** LM
_2197.91***81.68***81.71***56.66***206.19***86.02***76.96***55.98***
Hausman
_278.08***163.11***55.46***231.23***50.46***31.69***1277.36***62.51***
# observations200198200198198196198196 # countries2424242424242424
*** Significant at 1 % level** Significant at 5 % level *
Significant at 10% level 26Table 7 Effect of equity investments on
bank net income (Fixed effects with instrumental variables) This
table shows the results of a two-way fixed effects model with
instruments for banks' variables. The dependent
variableisthebanks'netincome.Asindependentvariablesweincludebanks,macroandcountryfinancial
developmentvariables.Banksvariablesaretheproportionofequityinvestments(EQUINV),thecapitalratio
(CAPITAL),theprovisions(PROVIS),theproportionofLOANS,thenon-interestearnings(NONINTER)andthe
overheadcosts(OVERHEADC).Allthesevariablesaredividedbytotalbankassetsinthecountry.Weuseas
instrumentstwolagsofeachoneofthesebanks'variables.TheannualgrowthrateofrealGDPpercapita
(GROWTH) and the annual inflation for consumer prices indices
(INFLATION) are the macro variables. Total assets of deposit money
bank assets to GDP (BANKASSET) and the stock market capitalization
(MAKCAP) are introduced
tomeasurethesizeofthebankingsectorandthestockmarketrespectivelyinthecountry.Finally,theprivate
creditbydepositmoneybankstoGDP(PRIBC)andthestockmarkettotalvaluetradedtoGDP(SMTVT)are
included to measure the activity of the banking sector and the
stock market respectively. The standard errors are corrected for
autocorrelation, and the corresponding t-statistics are given in
parentheses below. Dependent variable: Net Income (NETINC)
(1)(2)(3)(4)(5)(6)(7)(8) INTERCEPT0.327 (0.52) 0.298 (0.47) 0.167
(0.26) 0.116 (0.18) 0.083 (0.12) -0.017 (-0.03) -0.139 (-0.21)
-0.283 (-0.41) EQUINV0.122*** (2.71) 0.117** (2.51) 0.125*** (2.80)
0.119** (2.5) 0.091** (2.03) 0.080* (1.72) 0.096** (2.15) 0.082*
(1.78) CAPITAL 0.078 (1.58) 0.080 (1.62) 0.089* (1.79) 0.095*
(1.90) PROVIS 0.025 (0.33) 0.039 (0.52) 0.062 (0.82) 0.080 (1.05)
LOANS0.023** (2.27) 0.023** (2.28) 0.018* (1.78) 0.019* (1.79)
0.030*** (2.69) 0.031*** (2.77) 0.026** (2.31) 0.027** (2.40)
NONINTER0.067 (0.88) 0.066 (0.87) 0.049 (0.64) 0.047 (0.61) 0.081
81.06) 0.079 (1.03) 0.060 (0.78) 0.056 (0.73) OVERHEADC-0.169**
(-2.04) -0.166** (-2.00) -0.176** (-2.14) -0.173** (-2.09) -0.147*
(-1.73) -0.140* (-1.65) -0.153* (-1.82) -0.145* (-1.72)
GROWTH0.056*** (2.78) 0.056*** (2.79) 0.45** (2.16) 0.045** (2.16)
0.057*** (2.78) 0.058*** 82.80) 0.044** (1.99) 0.043** (1.98)
INFLATION0.013 (1.26) 0.014 (1.29) 0.008 (0.76) 0.009 (0.79) 0.019*
(1.76) 0.020* 81.84) 0.012 (1.13) 0.013 (1.19) BANKASSET -0.192
(-0.57) -0.191 (-0.56) -0.111 (-0.33) -0.106 (-0.31) MAKTCAP-0.034
(-0.32) -0.032 (-0.30) -0.033 (-0.32) -0.030 (-0.29) PRIBC -0.602
(-1.21) -0.613 (-1.23) -0.524 (-1.06) -0.533 (-1.08) SMTVT 0.385
(1.23) 0.419 (1.33) 0.475 (1.51) 0.525* (1.65) R2
overall25.22%27.41%20.14%23.41%36.02%38.95%32.12%35.83%
F2.65***2.50***2.67***2.54***2.85***2.73***2.91****2.82*** LM
_2124.34***51.14***38.70***16.51***129***58.34***39.93***18.38***
Hausman _219.6821.17750.32***31.89***14.0914.0953.19***47.65*** #
observations194194194194192192192192 # countries2424242424242424
*** Significant at 1 % level** Significant at 5 % level *
Significant at 10% level