The Legal Environment, Banks, and Long-Run Economic Growth

Author(s): Ross Levine

Source: Journal of Money, Credit and Banking, Vol. 30, No. 3, Part 2: Comparative Financial
Systems (Aug., 1998), pp. 596-613
Published by: Ohio State University Press
Stable URL: http://www.jstor.org/stable/2601259 .
Accessed: 14/01/2015 03:31

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.

Ohio State University Press is collaborating with JSTOR to digitize, preserve and extend access to Journal of
Money, Credit and Banking.

http://www.jstor.org

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
 ROSS LEVINE

TheLegalEnvironment,
Banks,andLong-Run
EconomicGrowth
This paperexamines the relationshipbetween the legal system and banldingdevelop-
ment and traces this connection throughto long-runrates of per capita GDP growth,
capital stock growth, and productivitygrowth. The data indicate that countries where
the legal system (1) emphasizes creditorrights and (2) rigorously enforces contracts
have better-developedbanks than countries where laws do not give a high priorityto
creditors and where enforcement is lax. Furtherrnore,the exogenous component of
bankingdevelopment-the componentdefinedby the legal environment-is positively
and robustlyassociatedwith percapitagrowth,physical capitalaccumulation,andpro-
ductivitygrowth.

THISPAPERADDRESSES TWOQUESTIONS. First, do cross-

countrydifferences in the legal rights of creditors,the efficiency of contractenforce-
ment, and the origin of the legal system explain cross-countrydifferencesin the level
of bankingdevelopment?Second, do better-developedbanks cause fastereconomic
development;that is, is the componentof bankingdevelopmentdefined by the legal
environmentpositively associatedwith long-runratesof economic growth,capitalac-
cumulation,andproductivitygrowth?
Examiningthe relationshipbetween the legal system and bankingdevelopmentis
valuable irrespectiveof issues associatedwith long-rungrowth.First,banksmay in-
fluence the level of income per capita and the magnitude of cyclical fluctuations
(BernankeandGertler1989, 1990). Second, manyeconomists stressthatunderstand-
ing the evolution of legal and financial systems is essential for understandingeco-
nomic development (North 1981; Engerman and Sokoloff 1996). Consequently,
quantitativeinformationon the relationshipbetweenthe legal environmentandbanks
will improve our understandingof business cycles and the process of economic de-
velopment.
Furthermore,examiningthe causal links between banksand economic growthhas
bothconceptualandpolicy implications.On the conceptualfront,a long literaturede-
batesthe importanceof banksin economic development.Startingas early as Bagehot
(1873), economistshave arguedthatbetterbanks banksthatarebetterat identifying
creditworthyfirms,mobilizing savings, pooling risks, and facilitatingtransactions-

The authorthanksShaghil Ahmed,CraigBurnside,Bill Easterly,JaneIhrig,Phil Keefer,LantPritchett,

and AndreiShleifer for very helpful comments.

ROSSLEVINE
is a professor of economicsat the Universityof Virginia.
Journalof Money, Credit,and Banking,Vol. 30, No. 3 (August1998,Part2)

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
 ROSS LEVINE : 597

accelerateeconomic growth. Others,however, questionthe importanceof the finan-

cial system in the developmentprocess or disagree with the causal interpretationas
discussedin Levine (1997a). Robinson(1952), for example,arguesthateconomic de-
velopmentcreatesdemandsfor financialservices andthe financialsystemrespondsto
provide these services. Evidence on whetherbankscause growthwill help reconcile
these conflicting views. There are also potentialpolicy implicationsassociated with
clarifyingthe causalrelationshipbetweenbanksandgrowth.Forexample,if evidence
suggests that greaterbankingdevelopment induces faster economic growth and we
can identifythe legal determinantsof bankingdevelopment,then this supportsgrant-
ing a higherpriorityto those reformsthatimprovethe functioningof the bankingsec-
tor. Alternatively,if the component of banking development associated with legal
factorsis unrelatedto economic development,this lowers the prioritygiven to these
legal factorsin any reformpackage.
By studyingthe connection between the legal environmentand bankingdevelop-
ment andthentracingthis link throughto long-runeconomic development,this paper
fills two gaps in the literature.First,considerresearchon the relationshipbetweenthe
legal system and financialdevelopment.Througharduousdatacollection and careful
analysis, LaPorta,Lopez-de-Silanes, Shleifer, and Vishny (1998, 1997; henceforth
LLSV) have substantiallyadvancedresearchinto the legal determinantsof financial
development.LLSV (1998) collect and summarizeinformationon the legal systems
of forty-ninecountries.LLSV (1997) then use these data to show that legal systems
that rigorouslyprotectcreditorsand enforce contractsencouragebetterfunctioning
debtandequitymarketsthanlegal systems thataremore lax in safeguardingcreditors
and enforcingcontracts.This papercomplementsLLSV (1997) by examiningthe re-
lationshipbetween legal systems andbanking-sectordevelopment.
Second, this papercontributesto researchon the causalrelationshipbetweenbank-
ing developmentand long-runeconomic growth.AlthoughKing and Levine (1993a)
show that the level of financialdevelopmentin 1960 is a good predictorof growth
over the next thirtyyears,RajanandZingales(1998) note that"financialdevelopment
may simplybe a leadingindicatorratherthana causalfactor."A majorimpedimentto
circumventingthis critiquehas been the absence of valid instrumentalvariablesfor
extractingthe exogenous componentof bankingdevelopment.Given the recentwork
by LLSV (1998), however,this paperexamineswhetherthe exogenous componentof
bankingdevelopment the componentdefinedby the legal system is positively as-
sociated with economic development in a cross-section of countries over the
1976-1993 period.LLSV (1998) note that(a) the forty-ninecountriesin theirsample
can be dividedinto countrieswith predominantlyEnglish,French,German,andScan-
dinavianlegal originsand (b) countriestypicallyobtainedtheirlegal systems through
occupationor colonization. Consequently,I view legal origin as an exogenous "en-
dowment"in studyingbankingdevelopmentand growthover the 1976-1993 period.
LLSV (1998) go on to show thatcross-countrydifferencesin legal origin accountfor
a significant amount of cross-countryvariationin the legal codes defining creditor
rightsand the efficiency of the legal system in enforcingthose codes. In turn,this pa-
peruses cross-countrydifferencesin legal origin,creditorrights,andthe efficiency of

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
598 : MONEY, CREDIT,AND BANKING

contractenforcementto identify the exogenous componentof bankingdevelopment.

Besides examining the determinantsof bankingdevelopmentand tracing this rela-
tionshipthroughto per capitaGDP growth,this paperalso examines the relationship
betweenthe exogenous componentof bankingdevelopmentandbothcapitalstock ac-
cumulationandproductivitygrowth.
The paperfindsthatcross-countrydifferencesin the legal rightsof creditorsandthe
efficiency with which legal systemsenforcethose rightsexplainover half of the cross-
countryvariationin banking-sectordevelopment.Empirically,I define bankingde-
velopment as credit allocated by commercial and other deposit-takingbanks to the
privatesector divided by GDP. The data show thatcountrieswith legal systems that
give a high priorityto banks receiving the full presentvalue of their claims against
firmshave better-developedbanks than countrieswhere the legal codes do not em-
phasize the rightsof creditors.Furthermore,enforcinglegal codes is as importantas
the legal codes themselves. The data indicate that countriesthat effectively enforce
compliancewith laws tend to have better-developedbanksthan countrieswhere en-
forcementis lax. Finally,the dataalso indicatethatcountrieswith a Germanlegal sys-
tem tend to have better-developedbanks, even after controlling for the level of
economic development.Thus, the legal system materiallyinfluencesbankingdevel-
opment.
Next, the paperfindsthatthe componentof bankingdevelopmentdefinedby the le-
gal environmentis positively associatedwith long-runratesof economic growth,cap-
ital accumulation,and productivitygrowth. The paper uses the LLSV (1998) legal
indicatorsas instrumentalvariablesto extractthe exogenous componentof banking
development. The GeneralizedMethod of Moments (GMM) results are robust to
changes in the instrumentalvariablesand to changes in the conditioninginformation
set. More specifically, using either (a) the LLSV (1998) measuresof creditorrights
and contractenforcementor (b) the LLSV (1998) measuresof legal origin produces
the same conclusions. Also, tests of the overidentifyingrestrictionsindicatethat the
datado not rejectthe hypothesisthatthe instrumentalvariablesareuncorrelatedwith
the errorterm,which strengthensthe confidenceone has in the instruments.Further-
more, after controllingfor a very wide arrayof indicatorsdesigned to measureeco-
nomic, political, and cultural characteristics,the data still indicate a statistically
significantandeconomicallylargerelationshipbetweenthe exogenous componentof
bankingdevelopmentandthe rateof economic growth.
This papermust be viewed as complementinga numberof recentefforts aimed at
reconciling whetherfinancialdevelopmentis simply a good predictorof economic
growth.Takinga microeconomicapproach,Rajanand Zingales (1998) show that,in
countrieswith well-developed financial systems, industriesthat are naturallyheavy
users of externalfinancinggrow relativelyfasterthanotherindustries.Alternatively,
in countries with poorly developed financial systems, industriesthat are naturally
heavy users of external financing grow more slowly than other industries.Further-
more, Demirguc-KuntandMaksimovic (1996) show thatfirmsin countrieswith bet-
ter-developedfinancialsystems grow fasterthanthey could have grown withoutthis
access. While these microeconomicstudies must respectivelyidentify (a) the "natur-

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
 ROSS LEVINE : 599

al" tendency of industriesto use externalfundingand (b) how fast firmswould have
grown in differentfinancialenvironments,the resultssupportthe conclusion thatbet-
ter financialsystems facilitateeconomic development.Furthermore,this paperis con-
sistent with Levine (1997b), where I examine the interactions among the legal
environment,financial intermediaries,and growth. In contrastto the currentpaper
thatfocuses on banks,Levine (1997b) uses measuresof financialintermediationfrom
King andLevine (1993a, b) thatincludethe activitiesof nonbankfinancialintermedi-
aries.Finally,this paperalso complementsan innovativeevent studyby Jayaratneand
Strahan(1996). They show thatwhen individualstatesof the UnitedStatesrelaxedin-
trastatebranchingrestrictionsthe quality of bank loans rose and per capita GDP
growthaccelerated.
The remainderof the paperis organizedas follows. Section 1 discusses the dataand
presentsevidence on the legal determinantsof bankingdevelopment.Section 2 inves-
tigates whetherthe componentof the banking system defined by the legal environ-
ment is linked with per capita GDP growth, per capita capital stock growth, and
productivityimprovements.Section 3 summarizesthe results and discusses policy
implications.

1. THE LEGALDETERMINANTSOF BANKING DEVELOPMENT

This section examines whetherdifferences in the legal rights of creditors,the effi-

ciency with which legal systems enforce those rights,and legal origin explain differ-
ences in bankingdevelopment.First, the section describesthe legal variablesused to
characterizenationallegal systems. Second, the section empiricallydefines the mea-
sureof bankingdevelopment.Finally,I presentevidence regardingthe legal determi-
nantsof bankingdevelopment.

A. Legal Determinants
1. CreditorRights.The ability of banksto persuadefirmsto pay theirloans differs
across nationallegal systems. Legal systems differ in terms of the rights of banks to
repossess collateralor liquidatefirms in the case of default. Legal systems differ in
termsof the rightsof banksto removemanagersin corporatereorganizations.Finally,
legal systems differ in termsof the prioritygiven to securedcreditorsrelativeto other
claimantsin corporatebankruptcy.
More specifically,this paperuses four measuresof the legal rightsof banks.
AUTOSTAYequalsone if a country'slaws impose an automaticstay on the assets
of the firmuponfiling a reorganizationpetition.AUTOSTAYequalsOif this restriction
does not appearin the legal code. The restrictionwould preventbankersfrom gaining
possession of collateralor liquidatinga firmto meet a loan obligation.
MANAGESequal one if the firmcontinuesto manageits propertypendingthe res-
olutionof the reorganizationprocess, andzero otherwise.In some countries,manage-
ment stays in place until a final decision is made about the resolution of claims. In
othercountries,managementis replacedby a team selected by the courtsor the credi-

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
600 : MONEY,CREDIT,
ANDBANKING

tors.If managementstayspendingresolution,this reducespressureon managementto

pay bankloans.
SECURED1equalsone if securedcreditorsarerankedfirstin the distributionof the
proceedsthatresultfrom the dispositionof the assets of a bankruptfirm.SECURED1
equals zero if nonsecuredcreditors,such as the governmentor workers,get paid be-
fore securedcreditors.In cases where SECURED1equals zero, this certainlyreduces
the attractivenessof lending securedcredit.
CREDITORis a conglomerateindex of these three individualcreditorrights indi-
cators that is designed to be positively associated with creditorrights. Specifically,
CREDITOR= SECURED1-AUTOSTAY-MANAGES and takes on values be-
tween 1 (best) and-2 (worst).AlthoughI am not able to controlfor cross-countrydif-
ferences in bankruptcy procedures, I expect countries with higher values of
CREDITORto have better-developedbanks,all else equal.
The individual country values for AUTOSTAY,MANAGES,SECURED1, and
CREDITORare providedin Table 2. Summarystatisticson CREDITORare given in
Table 1. As shown thereis substantialcross-countryvariationin CREDITOR,where
the maximumvalue is 1, the minimumvalue is-2, andthe standarddeviationis about
1. Brazil, Colombia, France,Mexico, Peru, and the Philippines(all countrieswith a
Frenchlegal origin)arecountrieswhere CREDITOR=-2, indicatingthattheirlegal
systems do not stress the rights of creditors.In contrast,the legal codes of Egypt,
Hong Kong, India,Indonesia,Israel, Korea,Malaysia,Nigeria, Pakistan,Singapore,
Thailand,United Kingdom, and Zimbabwe stress the rights of creditors, such that
CREDITOR= 1. These are measuresof the laws on the books, however, and do not
incorporateinformationregardingenforcement.
2. Enforcement.The laws governingsecuredcreditorswill affect securedcreditors
only to the extentthatthe laws areenforced.Consequently,measuresof the efficiency
of the legal system in enforcingcontractsareincludedfrom LLSV (1998).
RULELAWis an assessment of the law-and-ordertraditionof the country that
ranges from 10, strong law-and-ordertradition,to 1, weak law-and-ordertradition.
This measurewas constructedby InternationalCountryRisk Guide (ICRG)and is an
averageover the period 1982-1995. Given the contractualnatureof banking,higher
values of the RULELAWarelikely to positively influencebankingdevelopment.

TABLE 1
SUMMARY
STATISTICS:
ANNUALAVERAGES,
1976-1993
Standard
Variable Mean Median Maximum Minimum Deviation

OUTPUTGROWTH 0.023 0.018 0.097 -0.014 0.022

CAPITALSTOCK GROWTH 0.028 0.024 0.095 -0.019 0.025
PRODUCTIVITY GROWTH 0.017 0.014 0.079 -0.019 0.017
BANK 0.844 0.760 2.678 0.122 0.536
CREDITOR -0.310 0.000 1 -2 1.070
ENFORCE 7.451 8.310 9.990 3.545 2.154
Observations:42
NOTES:OUTPUT GROWTEI= per capita GDPgrowth. CAPITALSTOCKGROWTH = per capita capital growth. PRODUCTIVITY
GROWTH= OUTPUTGROWTH-(0.3)*CAPITAL STOCKGROWTH.BANK= depositmoney bankcreditto the privatesectordivided by
GDP.
CREDlTOR= index of securedcreditorrights.ENFORCE= index of law andcontractenforcement.

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
 ROSSLEVINE : 601

TABLE 2
LEGAL
VARIABLES
ANDBANKING
DEVELOPMENT
Country CREDlTOR ENFORCE BANK

Australia -1 9.36 0.77

Canada -1 9.48 0.83
Hong Kong 1 8.52 1.19
India 1 5.14 0.46
Israel 1 6.18 0.96
Malyasia 1 7.11 1.00
New Zealand 0 9.65 0.58
Nigeria 1 3.55 0.23
Pakistan 1 3.95 0.45
Singapore 1 8.72 1.50
South Africa 0 5.85 0.62
Thailand 1 6.91 0.75
United Kingdom 1 9.10 1.22
United States -1 9.50 0.77
Zimbabwe 1 4.36 0.14
Average-English 0.47 7.16 0.77
Argentina - 1 5.13 0.29
Belgium 0 9-74 0-53
Brazil -2 6.31 0.23
Chile -1 6.91 0.75
Colombia -2 4.55 0.25
Egypt 1 5.11 0.44
France -2 9.09 1.51
Greece -1 6.40 0.50
Indonesia 1 5.04 0.48
Italy -1 8.75 0.69
Mexico -2 5.95 0.24
Netherlands -1 9.68 1.31
Peru -2 3.59 0.12
Philippines -2 3.77 0.45
Portugal - 1 8.63 0.96
Spain 0 8.10 1.31
Turkey -1 5.57 0.35
Average-French -1.00 6.61 0.61
Austria 0 9.80 1.36
Germany 0 9.50 1.64
Japan 0 9.34 1.96
Korea 1 6.97 0.82
Switzerland -1 9.99 2.68
Taiwan 0 8.84 1.38
Average-German 0.00 9.07 1.64
Denmark 0 9.66 O.69
Finland -1 9.58 1.22
Norway -1 9.86 0.93
Sweden -1 9.79 0.87
Average-Scandinavian -0.75 9.72 0.93
NOTES: Countriesare divided into those with English, French,German,and Scandinavianlegal origins. CREDITORis an index of the legal
rightsof securedcreditors,with values between - 2 and I, andwherelargervalues indicategreatercreditorrights.ENFORCEis an index of the
efficiency of the legal system in enforcingcontracts,with values potentiallybetween Oand 10, andwhere largervalues indicategreatercontract
enforcementefficiency. BANKequals bankcreditto the privatesectordivided by GDP.

CONRISKis an assessmentof the riskthata governmentwill andthereforecan-

modify a contractafterit has been signed. CONRISKrangesfrom 10, low risk of con-
tract modification,to 1, high risk of contractmodification.Specifically, "modifica-
tion"meanseitherrepudiation,postponement,or reducingthe government'sfinancial
obligation.This measurewas constructedby ICRGand is an averageover the period

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
602 : MONEY,CREDIT,
ANDBANKING

1982-1995. Legal systems thateffectively enforcecontractswill tendto supportbank-

ing activities.
ENFORCEequals the averageof RULELAWand CONRISK.The empiricalanaly-
ses focus on this aggregate index of the efficiency of the legal system in enforcing
contracts,ENFORCE,and the aggregateindex of creditorrights, CREDITOR.
Again, the individual country values for RULELAW,CONRISK,and ENFORCE
are providedin Table 2. Summarystatistics on ENFORCEare given in Table 1. As
shown thereis substantialcross-countryvariationin ENFORCE,wherethe maximum
value is 9.99, the minimumvalue is 3.55, andthe standarddeviationis 2.2. The coun-
tries with very high values of enforce, values of ENFORCEgreaterthan 9, are Aus-
tralia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Japan,
Netherlands,New Zealand,Norway, Sweden, and Switzerland.In contrast,countries
wherecontractenforcementis poor, values of ENFORCEless than5, include Colom-
bia, Nigeria,Pakistan,Philippines,Peru,andZimbabwe.
3. Origin.LLSV (1998) collect and summarizeinformationon the legal systems of
forty-ninecountries.Based on the work of comparativelegal scholars,LLSV (1998)
place these forty-ninecountriesinto four legal families, eitherEnglish, French,Ger-
man,or Scandinavian.The English legal system is common law, wherethe laws were
primarilyformedby judges tryingto resolve particularcases. In contrast,French,Ger-
man, and Scandinavianlegal systems are based on the civil law tradition,where the
laws were generallyformedby legal scholars.
As describedin Glendon,Gordon,and Osakwe (1982), Roman law was compiled
underthe directionof ByzantineEmperorJustinianin the sixth century.Over subse-
quentcenturies,the Glossatorsand CommentatorsinterpretedRomanlaw and adapt-
ed the law to the problemsof theirown day. The initialJustiniantexts plus the workof
the Glossatorsand Commentatorsbecame known as thejus commune,the common
law, of Europe.Eventually,individualcountriesdecided to formalizetheirown legal
codes. The Scandinaviancountries did this in the seventeenth and eighteenth cen-
turies.These countrieshave remainedrelativelyunaffectedfrom the far-reachingin-
fluences of the Germanandespecially the Frenchcivil codes.
The FrenchCivil Code of 1804 is concise, meant to be accessible to the general
population,and relies heavily on thejus commune.Napoleon saw the permanenceof
the Code as moreimportantthanthe fleetingnatureof his militaryconquests.He made
it a priorityto secure its adoptionin Franceand all conqueredterritories,including
Italy,Poland,the low countries,andthe HabsburgEmpire.Also, Franceextendedher
legal influenceto partsof the NearEast,NorthernandSub-SaharanAfrica,Indochina,
Oceania, FrenchGuiana, and the FrenchCaribbeanislands duringthe colonial era.
Furthermore,the French Civil Code was a major influence on the Portugueseand
Spanishlegal systems, which helped spreadthe Frenchlegal traditionto Centraland
South America.
The Germancivil code appearedalmost a centurylater.Following the unification
of GermanyunderBismarckin 1871, it took over twenty years to complete the code
(BurgerlichesGesetzbuch)in 1896. In termsof consistency, technicalprecision, and
detail, the Germancivil code was unprecedented.The Germancode exerteda big in-

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
 ROSS LEVINE : 603

fluence on Austriaand Switzerland,as well as China(andhence Taiwan),Czechoslo-

vakia, Greece, Hungary,Italy, and Yugoslavia. Also, the Germancivil code heavily
influencedthe Japanesecivil code, which helped spreadthe Germanlegal traditionto
Korea.
Since the English, French, and German systems were spread primarilythrough
conquestand imperialism,this papertakesnationallegal origin as an exogenous "en-
dowment"in examiningthe relationshipbetweenthe legal system andbankingdevel-
opmentover the period 1976-1993.
Furthermore,LLSV (1998) tracedifferencesin legal origin throughto differences
in the legal rules covering securedcreditorsandthe qualityof contractenforcementin
forty-ninecountries.They show thatlaws andenforcementqualityvary systematical-
ly with legal origin. More specifically, LLSV (1998) show that common law coun-
tries-countries based on the English tradition have laws thatemphasizethe rights
of creditorsto a greaterdegreethanthe French,German,and Scandinaviancountries.
Frenchcivil law countriesprotectcreditorsthe least, with Germanand Scandinavian
civil law countries falling in the middle. LLSV (1998) also examine enforcement
quality.Countrieswith a Frenchlegal heritagehave the lowest qualityof law enforce-
ment, while countrieswith Germanand Scandinavianlegal traditionstend to be the
best at enforcingcontracts.Table 2 summarizessome of these findings.

B. BankingDevelopment
A large literatureexamines the ties between banks and economic activity. Ideally,
researcherswould constructcross-countrymeasuresof how well banksidentifyprof-
itableactivities,exertcorporategovernance,mobilize resources,managerisk, andfa-
cilitatetransactions.In practice,however, economistshave been unableto accurately
measurethese financialservices for a broadcross-sectionof countries.Consequently,
researcherstraditionallyuse measures of the overall size of the banking sector to
proxy for "financialdepth."Financialdepth,however, does not measurewherethe fi-
nancial system allocates capital.To minimize this problem,I use a measureof bank-
ing development,BANK,constructedby Levine andZervos (1998).
BANK equals the value of loans made by commercial banks and other deposit-
takingbanks to the privatesector divided by GDP. BANKimprovesupon traditional
financialdepthmeasuresof bankingdevelopmentby isolating creditissued by banks
to the privatesector,as opposedto creditissued to governmentsor public enterprises.
As Table 1 and Table 2 illustrate,thereis a wide variationin the value of BANK.The
values presentedin Tables 1 and 2 are averagevalues of BANKover the 197S1993
period.The maximumvalue is 2.7 (Switzerland),the minimumis 0.1 (Peru,Zimbab-
we), and the standarddeviationis 0.5.

C. Evidenceon the Legal Determinantsof BankDevelopment

The datasuggest a strong,positive relationshipbetween bankingdevelopmentand
both the rights of creditorsand the efficiency of contractenforcement.Table 3 pre-
sents the resultsof regressingBANKon differentcombinationsof the legal indicators.

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
604 : MONEY, CREDIT,AND BANKING

TABLE 3
BANKING DEVELOPMENT AND LEGAL SYSTEMS, 1976-1993
1 2 3 4 5 6

Constant -0.32 -2.23 0.52 -0.15 0.93 -- 1.45

[0.028] [0.001] [0.248] [0.807] [0.000] [0.014]
INITIALOUTPUT 0.37 - 0.13 0.03 0.26
[0.001] [0.083] [0.766] [0.O00]
CREDITOR 0.09 0.17 0.09
[0.020] [0.001] [0.103]
ENFORCE 0.16 0.19 0.16
[0.000] [0.000] [0.000]

German 0.51 0.66

[0.017] [0.001]

E.nglish - 0. 19 0.08
[0.199] [0.639]
French - 0.30 - 0.06
[0.038] [0.714]
R-squge 0.56 0.43 0.51 0.56 0.30 0.50
NOTES: INITIALOUTPUT= Initiallogarithmof real percapitaGDP. CREDITOR= index of securedcreditorrights.ENFORCE= index of
law andcontractenforcement.Gez7nan,English,French = Dummy variablesfor correspondinglegal origin.
The null hypothesisthatthe regressorsexcplainnone of the cross-countryvariationin bankingdevelopmentis rejectedat the 0.001 level in all
of the regressions.
[Heteroskedasticity-consistentP-values in squarebrackets]

I computethe value of BANKasthe averageover the 197S1993 period.Thus,thereis

one observationpercountry.In regression(1), the regressorsare CREDITORandEN-
FORCE.Eachlegal indicatorenterspositively and significantlyat the S percentlevel.
Regressions (2), (3), and (4) also include the logarithmof per capita GDP (INITIAL
OUTPUT)to control for the overall level of economic development.When INITIAL
OUTPUT is included along with either CREDITOR[equation (2)] or ENFORCE
[equation(3)], the legal variablesentersignificantlyat the 1 percentlevel. WhenINI-
TIALOUTPUT,CREDITOR,and ENFORCEare simultaneouslyincluded [equation
(4)], the P-value on CREDITORrises to about0.1 while the resultson ENFORCEdo
not change.
Moreover,the empiricalrelationshipbetweenthe legal systemindicatorsandbank-
ing developmentareeconomicallymeaningful.Forexample,a one standarddeviation
increasein CREDITOR(1.1) would increaseBANKby0.1 (using the smallestvalue of
the estimatedcoefficients), which is about 12 percent of the mean value of BANK.
More impressively, a one standarddeviation increase in ENFORCE(2.2) would in-
creaseBANKby 0.35, which is over 40 percentof the mean value of BANK.Thus, the
legal rights of creditorsand the ability to enforce those rights are stronglytied to the
ratioof bankcreditto the privatesectoras a shareof GDP. As shownby the regression
R2s, the creditorrights and enforcementindicatorsaccountfor over half of the cross-
countryvariationin banking-sectordevelopment.

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
 ROSS LEVINE : 605

The dataalso suggest thatlegal originhas a profoundimpacton bankdevelopment.

The regressorsin regression(5) of Table 3 aredummyvariablesfor German,English,
andFrenchlegal origin as definedby LLSV (1998) and shown in Table 2. Regression
(6) also includes INITIALOUTPUT.The majormessage that emerges from regres-
sions (5) and (6) is that countries with a Germanlegal origin have better-developed
banks.While countrieswith a Frenchlegal traditiontend to have less well developed
banksthanothercountrieson average [regression(5)], this resultdoes not hold when
controllingfor the overall level of economic development [regression(6)]. In con-
trast,the dummyvariablefor a Germanlegal traditionenters with a positive and sig-
nificantcoefficient even aftercontrollingfor INITIALOUTPUT.
In sum, the legal system mattersfor bankingdevelopment.Differences in banking
developmentcan be tracedback to the legal origin of the country.Even aftercontrol-
ling for the level of economic development, countries with a Germanlegal system
tend to have better-developedbanks.Moreover,the dataidentifyparticularaspectsof
the legal system thatare importantfor banking-sectordevelopment.Countrieswhere
the legal system gives a high priorityto banks getting the full presentvalue of their
loans to firms have better-developedbanks. Furthermore,countries where the legal
system effectively enforces contractstend to have better-developedbanksthancoun-
tries thatless efficiently enforce contracts.

2. BANKS AND GROWTH

This section examines whether the exogenous component of banking develop-

ment the component associated with national legal characteristics is related to
rates of per capita GDP growth, capital stock growth, and productivity growth.
Specifically, this section uses the legal determinantsof bankingdevelopmentexam-
ined in section 1 as instrumentalvariables.The data suggest that the component of
bankingdevelopmentdefinedby the legal environmentis positively and robustlyas-
sociatedwith long-runratesof economic development.

A. Methodology
In the traditionof recentcross-countryempiricalstudies of economic growth,this
paper uses data averaged over long periods, such that there is one observationper
country.The basic regressiontakes the form:

G(i) = a(i) + :(i)BANK + y(i)X + £(i), (l)

where the dependentvariable,G(i), is either real per capita GDP growth, per capita
capital stock growth, or productivitygrowth (index i distinguishesamong these de-
pendentvariables),BANKequals credit to the privatesector divided by GDP, and X
representsa matrixof conditioninginformationthatcontrolsfor otherfactorsassoci-
ated with economic growth(for example, income per capita,education,political sta-

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
606 : MONEY,CREDIT,
ANDBANKING

bility, ethnic diversity,civil rights,bureaucraticefficiency, andindicatorsof trade,fis-

cal, and monetary policy). Given data availability, the data cover the period
197S1993.
In contrastto traditionalcross-countryinvestigations,however, this paperseeks to
examine whethercross-countryvariationsin the exogenous componentof banking-
sector developmentexplain cross-countryvariationsin the rateof economic develop-
ment. Thus, I use the legal determinantsof banking development as instrumental
vanables for BANK.Specifically, I select a vector of instrumentalvariablesZ for the
equations specified by equation (1). Assuming that E[£] = 0 and that E[££'] = Q,
where Q is unrestricted,implies a set of orthogonalityconditions,E[Z'£] = 0. This
produces a nonlinearinstrumentalvariableestimatorof the coefficients in equation
(1). After computingthese GMM estimates,I use a standardLagrangemultipliertest
of the overidentifyingrestrictionsto see whetherthe instrumentalvariablesare asso-
ciated with growthbeyond theirability to explain cross-countryvariationin banking-
sectordevelopment.

B. The GrowthIndicators
As indicatedby equation (1), this paper examines three dependentvariables:per
capitaGDP growth(OUTPUTGROWTHS), per capitacapitalstock growth(CAPITAL
STOCKGROWTH),andproductivitygrowth(PRODUCTIVITYGROWTH). Specifi-
cally, define PRODUCTIVITY GROWTHas follows:

PRODUCTIVITYGROWTH
= OUTPUTGROWTH-

K8(CAPITAL
STOCKGROWTH).

To obtain empirical estimates, I (a) obtain OUTPUT GROWTHfrom national ac-

counts data, (b) use CAPITALSTOCKGROWTHfrom King and Levine (1994), (c)
select a standard value for v (v = 0.3), and then compute PRODUCTIVITY
GROWTH.

C. ConditioningInformation,X
This paperexamines whetherthe exogenous componentof bankingdevelopment
has a robust,independentrelationshipwith the growthindicators.To enhanceconfi-
dence in the analysis,it is importantto controlfor "otherfactors."Thatis, I wantto re-
duce the chances thatregression(1) omits an importantexplanatoryvariable(Levine
and Renelt 1992). Given thatthe maximumsample size is only forty-threecountries,
there are limits on the numberof variablesthat can be included in X in any one re-
gression. Consequently?the analysis includes three differentconditioning informa-
tion sets in all of its analyses.
Conditioning information set 1 included a constant, the logarithm of initial per
capita GDP, and initial secondaryschool enrollment.The initial income variableis
used to capturethe convergenceeffect highlightedby BarroandSala-i-Martin(1995).

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
 ROSSLEVINE: 607

As in manycross-countryanalyses,initialsecondaryschool enrollmentis used to con-

trol for investmentin humancapitalaccumulation.I sometimesreferto conditioning
informationset 1 as the "simple"conditioninginformationset.
Conditioninginformationset 2 includes conditioninginformationset 1 plus the ra-
tio of governmentconsumptionto GDP, the inflationrate, and the black marketex-
change ratepremium.All of these variablesare averagedover the 1976-1993 period.
The black marketexchange rate premiumis frequentlyused as an overall index of
trade,exchange rate, and price distortions(Easterly 1994; Levine and Zervos 1998).
The inflationrate and size of the governmentserve as indicatorsof macroeconomic
stability (Easterlyand Rebelo 1993; Fischer 1993). Thus, conditioninginformation
set 2 is designed to controlfor policy distortionsin studyingthe relationshipbetween
bankingdevelopmentand economic development.I sometimes referto conditioning
informationset 2 as the "policy"conditioninginformationset.
Conditioninginformationset 3 includes conditioning informationset 1 plus the
number of revolutions and coups (Banks 1994), the number of assassinationsper
thousandinhabitants(Banks 1994), an index of politicalrights,an index of civil liber-
ties (Gastil 1990), an index of bureaucraticred tape (Mauro1995), and the degree of
ethnicdiversity(EasterlyandLevine 1997). This groupof conditioninginformationis
designedto controlfor badpolicies, bad bureaucracies,bad institutions,andbad poli-
tics, as emphasizedby KnackandKeefer(1995). I sometimesreferto conditioningin-
formationset 3 as the "political"conditioninginformationset.

D. GMMResultswith CREDITORand ENFORCEas Instruments

Table 4 summarizesthe instrumentalvariableresults from nine regressions. The
dependentvariableis either OUTPUTGROWTH,CAPITALSTOCKGROWTH,or
PRODUCTIVITY GROWTH.For each of these three dependentvariables, Table 4
presentsthe regressionsresultsfor the threeconditioninginformationsets, thatis, the
simple, policy, and political conditioninginformationsets. In the Table 4 results,the
instrumentalvariablesare CREDITORand ENFORCE,which measurethe degree to
which legal codes emphasizethe rightsof creditorand the efficiency of the legal sys-
tem in enforcinglaws and contractsrespectively.Furthermore,I treatthe X matrixas
exogenous because I am focussing on examining whetherthe exogenous component
of bankingdevelopmentas defined by the legal environmentis associated with eco-
nomic development.For conciseness, I presentonly the statisticson the BANKcoeffi-
cients as well as the tests of the overidentifyingrestrictions.
The resultsindicatea very strongconnectionbetweenthe exogenous componentof
bankingdevelopmentand the growthindicators.BANKenters all nine of the regres-
sions positively and significantlyat the 1 percentlevel. The componentof BANKde-
fined by creditorrights and the efficiency of contractenforcementis closely tied to
long-run rates of per capita GDP growth, capital stock growth, and productivity
growth.Furthermore,the datado not rejectthe orthogonalityconditionsat the 10 per-
cent level in any of the nine regressions.Thus, the resultsareconsistentwith the state-
ment that the creditor rights indicator and the contract enforcement indicator

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
608 : MONEY,CREDIT,AND BANKING

TABLE 4
BANKS, GROWTH,
AND THESOURCES
OFGROWTH
DependentVariable
Independent Conditioning OUTPUT CAPITALSTOCK PRODUCTIVITY
Variable Information GROWTH GROWTH GROWTH
BANK 1 0.056 0.058 0.046
(0.011) (0.012) (0.009)
[0.001] [0.001] [0.001]
{0.01 } {0.40} {0.01 }
BANK 2 0.056 0.061 0.042
(0.017) (0.018) (0.014)
[0.003] [0.002] [0.004]
{0.21} {0.06} {0.05}
BANK 3 0.043 0.035 0.038
(0.009) (0.010) (0.008)
[0.001] [0.001] [0.001]
{0.38} {2.13} {0.21 }
NOTES: (Heteroskedasticity-consistentstandarderrorsin parentheses)
[P-values in squarebrackets]
{LM-testof ovelidentifying restrictionsin braces}
Clitical values for LM-Test( I d.f.): 10%= 2.71; 5% = 3.84.
Conditioninginformation1: Otherregressorsinclude a constant,logalithm of initial per capitaGDP, and initial secondaryschool enrollment
rate.Observations= 40.
Conditioninformation2: OtherregressorsincludeconditioninginformationI, plus the ratioof governmentconsumptionspendingto GDP, in-
flationrate,and the black marketexchange ratepremium.Observations= 40.
Conditioninformation3: Otherregressorsinclude conditioninginformationI, plus the numberof revolutionsandcoups, the numberof assas-
sinationsperthousandinhabitants,index of political lights, index of civil liberties,index of bureaucraticred tape, anddegree of ethnicdiversi-
ty. Obsevations = 38.
Instruments:conditioninginformationset plus CREDITORand ENFORCE,which are indexes of securedcreditorslegal nghts and the effi-
ciency of law enforcement.

influence the growth indicatorsonly throughtheir impact on bankingdevelopment.

Also, the resultsare economically meaningful.As noted earlier(Table 3), a one stan-
darddeviationincreasein both CREDITORandENFORCEwould increaseBANKby
0.45 (which is about one-half of the mean value of BANK).According to Table 4, a
rise in the exogenous componentof BANKby 0.45 would increasethe rateof percapi-
ta GDP growth by almost 2 percentper year (0.45*0..043) over the sample period.
(This example uses the political conditioninginformationset results, which have the
smallestBANKcoefficients.) Accumulatingover the eighteen years of the sample,the
resultssuggestedthatrealpercapitaGDP would have been about40 percenthigherin
1993 given this 0.45 increase in the exogenous componentof banking-sectordevel-
opment. The estimates imply an economically large relationshipbetween banking-
sector developmentand economic performance.Legal reformsthatenhancecreditor
rights and contractenforcement may substantiallyaccelerate long-run rates of per
capitaGDP growth,capitalaccumulation,and productivitygrowth.

E. GMMResults withLegal OriginDummiesas Instruments

Table 5 presents additionalresults regardingthe relationshipbetween the three
growth indicatorsand the exogenous componentof banking-sectordevelopmentus-
ing an alternativeinstrumentset. Specifically, the instrumentalvariablesare dummy
variablesfor legal origin,eitherEnglish, French,or German.Some view these instru-
ments as betterthan CREDITORand ENFORCEbecause legal origin is less proneto

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
 ROSS LEVINE : 609

TABLE 5

BANKS, GROWTH, AND THE SOURCES OF GROWTH: ALTERNATIVE INSTRUMENTS

DependentVariable
Independent Conditioning OUTPUT CAPITALSTOCK PRODUCTIVITY
Variable Information GROWTH GROWTH GROWTH
BANK 1 0.052 0.037 0.046
(0.020) (0.015) (0.017)
[0.013] [0.020] [0.010]
{0.56} {0.37} {2.12}
BANK 2 0.043 0.034 0.017
(0.018) (0.015) (0.012)
[0.019] [0.032] [0.175]
{0.80} {0.19} {5.46}
BANK 3 0.019 0.020 0.016
(0.009) (0.010) (0.007)
[0.033] [0.052] [0.027]
{3-32} {0.03 } {5.57 }
NOTES: (Heteroskedasticity-consistentstandarderrorsin parentheses)
[P-values in squarebrackets]
{ LM-testof overidentifyingrestrictionsin braces}
Criticalvalues for LM-Test(2 d.f.): 10% = 4.61; 5% = 5.99.
Conditioninginformation1: Otherregressorsinclude a constant,logarithmof initialper capitaGDP, and initial secondaryschool enrollment
rate.Observations= 43.
Conditioninformation2: Otherregressorsincludeconditioninginformationl, plus the ratioof governmentconsumptionspendingto GDP, in-
flationrate,andthe black marketexchange ratepremium.Observations= 43.
Conditioninformation3: Otherregressorsincludeconditioninginformationl, plus the numberof revolutionsandcoups, the numberof assas-
sinationsper thousandinhabitants,index of political rights,index of civil liberties,index of bureaucraticredtape, anddegree of ethnicdiversi-
ty. Observations= 41.
Instruments:conditioninginformationset plus dummy variablesfor legal origin.

endogeneityproblemsthanmeasuresof the legal rightsof creditorsandthe efficiency

of thejudicial system. Onthe negativeside, dummyvariablesfor legal origingive less
guidanceregardingthe particularcharacteristicsof the legal system thatareimportant
for banking-sectordevelopment.For this section's purposes,the relevantpoint is that
the two sets of instrumentsgive similarresults.
Table 5 summarizesthe results from nine regressions,where the dependentvari-
able is eitherOUTPUTGROWTH,CAPITALSTOCKGROWTH,,or PRODUCTIVI-
TY GROWTH.For each of these three dependent variables, Table 5 presents the
regressionresults for the simple, policy, and political conditioninginformationsets.
In all of the OUTPUTGROWTHregressions,BANKenterspositively andsignificant-
ly at the S percentlevel. Also, BANKenterspositively and significantlyin two out of
the three CAPITALSTOCKGROWTHregressionsand in two out of the threePRO-
DUCTIVITYGROWTHregressions.With the political conditioninginformationset,
BANKenters the CAPITALSTOCKGROWTHregression with a P-value of 0.052.
With the policy conditioning information set, BANK enters the PRODUCTIVIItY
GROWTHregression with a P-value of 0.18. With these two exceptions noted, the
dataindicatea strong,positive relationshipbetween the growthindicatorsandthe ex-
ogenous componentof bankingdevelopment.While the legal origininstrumentstend
to producesmallercoefficients thanthe CREDITORandENFORCEinstruments,the
coefficients still indicate an economically large relationshipbetween bankingsector
developmentand long-rungrowth.Using the same example as above, a rise in the ex-
ogenous component of BANK by 0.45 would increase the rate of per capita GDP

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
610 : MONEY,CREDIT,
ANDBANKING

growthby 0.9 of 1 percentper year (0.45*0.019) over the sampleperiod.Accumulat-

ing over the eighteenyearsof the sample,the resultssuggest thatrealpercapitawould
have been about 18 percenthigherin 1993 with a 0.45 increasein the exogenous com-
ponentof banking-sectordevelopment.
The hardy links between banking development and long-run rates of per capita
GDP growth,capital stock growth,and productivitygrowthpass some additionaldi-
agnostictests. The datado not rejectthe orthogonalityconditionsat the 5 percentlev-
el in the nine regressions. At the 10 percent level, the PRODUCTIVITYGROWTH
regressionswith the policy and political conditioninginformationsets rejectthe null
hypothesis that the instrumentalvariables are uncorrelatedwith the errorterm. In
these regressionsand at this higherP-value, the datasuggest thatlegal originmay not
be specific enough;thatis, legal originmay be associatedwith growthbeyond its link
to banking-sectordevelopment.In general, however, the results are consistent with
the statementthatthe legal environmentaffects the rateof economic developmentby
influencingbanking-sectordevelopment.

F. CautionaryNote
It is importantto be clear aboutwhat these resultsdo not show. The paperdoes not
show thateconomic growthdoes not influencethe bankingsystem. The resultsdo not
contradicttheoriesby Patrick(1966), Greenwoodand Jovanovic (1990), and Green-
wood and Smith(1997), which suggest thatcausalityrunsin both directions;banking
developmentinfluenceseconomic growth,andeconomic growthinfluencesbanking-
sector development.This paperprovides evidence for the hypothesisthat the exoge-
nous componentof bankingdevelopmentpromoteseconomic growth.Furthermore,
this paperexaminesneitherthe determinantsnor the effects of variousfinancialregu-
lations. Thus, I do not consider the determinantsor effects of deposit insurance
(Calomiris 1989; Demirguc-Kuntand Detragiache 1997; Kane 1985,1989), restric-
tions on bankingactivities(KrosznerandRajan1994;Calomiris1995), or a wide array
of supervisoryand regulatoryissues that may affect bank stabilityand performance
(Barth,Nolle, and Rice 1996; BIS 1997; Calomirisand Gorton 1991; Krosznerand
Strahan1996). Rather,this papermakes a more limited point: the legal environment
influences the banking sector and this componentof banking-sectordevelopmentis
stronglylinked with long-runratesof economic growth.

3. SUMMARYAND CONCLUSIONS

This paperfirstexaminedthe connectionbetween the legal environmentandbank-

ing developmentand then studiedthe link between thatpartof banking-sectordevel-
opmentassociatedwith the legal environmentand rates of economic growth,capital
accumulation,and productivityimprovements.The dataindicatea close relationship
between the legal system andbankingdevelopment.Countrieswherelegal codes em-
phasize the rights of creditorshave better-developedbanks, as measuredby bank
credit to the privatesector divided by GDP, thancountrieswhere laws do not give a

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
 ROSS LEVINE : 611

high priorityto creditorsin the case of corporatebankruptcyor reorganization.Fur-

thermore,enforcementmatters.Countrieswith legal systems thatrigorouslyenforce
laws and contractshave better-developedbanksthancountrieswhere enforcementis
more lax. Moreover, these differences can be tracedback to the legal origin of the
country. As noted by LLSV (1998), English common law countrieshave laws that
emphasize the rights of creditorsto a greaterdegree than the French, German,and
Scandinaviancountries.On average,Frenchcivil law countriesprotectcreditorsthe
least. In termsof law enforcement,countrieswith a Frenchlegal heritagehave, on av-
erage, the lowest qualityof law enforcement,while countrieswith Germanand Scan-
dinavianlegal traditionstend to be the best at enforcinglaws andcontracts.
The paperalso finds thatthe exogenous componentof bankingdevelopment the
componentdefinedby the legal environment is positively associatedwith econom-
ic growth.This findingwas robustto changes in the conditioninginformationset and
to alterationsin the instrumentalvariables. The policy implications are clear. Al-
though changing legal codes and improvingthe efficiency with which legal systems
enforce laws andcontractsis difficult,the economic returnsto improvingthe legal en-
vironmentappearvery large. For many countries,these reforms could begin at the
level of regulationand implementation.For example, it may be prohibitivelydiffi-
cultr undesirable to change a country's law that imposes an automaticstay on
the assets of a firmupon filing a reorganizationpetition.Nevertheless,corporatereor-
ganizationprocedurescould be improved to reduce delays and uncertainty,so that
bankersfeel greaterconfidence aboutreceiving the full presentvalue of their loans.
Thus, this paper's results emphasize the prominentrole that legal reformslefined
broadlyan have in stimulatingeconomic developmentby improvingthe function-
ing of the bankingsystem.

LITERATURECITED

Bagehot,Walter.LombardStreet.Homewood,Ill.: RichardD. Irwin,1873 (1962 edition).

Banks, ArthurS. "Cross-NationalTime Series Data Archive."Binghamton:Centerfor Social
Analysis, StateUniversityof New York at Binghamton,1994.
Barth,JamesR., Daniel E. Nolle, and TaraN. Rice. "CommercialBanking Structure,Regula-
tion, and Performance:An InternationalComparison."Economic and Policy Analysis
WorkingPaper,Office of the Comptrollerof the Currency,United States, September1996.
Barro,Robert,and Xavier Sala-i-Martin.Economic Growth.New York:McGraw-Hill, 1995.
Bernanke,Ben, and Mark Gertler. "Agency Costs, Net Worth, and Business Fluctuations."
AmericanEconomicReview79 (March1989),14-38.
. "Financial-Fragilityand Economic Performance."QuarterlyJournal of Economics
105 (February1990), 87-114.
BIS. "CorePrinciplesfor Effective Banking Supervision."No. 30, Basle Committeeon Bank
Supervision,1997.
Calomiris,CharlesW. "DepositInsurance:Lessons fromthe Record."FederalReserveBankof
Chicago, EconomicPerspectives(May/June1989), 1>30.
. "TheCosts of RejectingUniversalBanking:AmericanFinancein the GermanMirror,
187>1914." In Coordinationand Information:Historical Perspectiveson the Organization

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
612 : MONEY,CREDIT,AND BANKING

ofEnterprise,
editedby Naomi R. LamoreauxandDaniel M. G. Raff, pp.257-321. Chicago:
NationalBureauof Economic Research,Universityof Chicago, 1995.
Calomiris,CharlesW., and GaryGorton."TheOriginsof BankingPanics:Models, Facts, and
Bank Regulation."In FinancialMarketsand FinancialCrises,edited by R. Glenn, pp.
109-74. Hubbard,Chicago:Universityof Chicago Press,1991.
Demirguc-Kunt,Asli, and Vojislav Maksimovic. "FinancialConstraints,Uses of Funds, and
FirmGrowth:An InternationalComparison."Unpublished,WorldBank,1996.
Demirguc-Kunt,Asli, and Enrica Detragiache. "The Determinantsof Banking Crises: Evi-
dence from Developed and Developing Countries."World Bank Policy Research Group
WorkingPaperno. 1828,1997.
Easterly,William. "EconomicStagnation,Fixed Factors,and Policy Thresholds."Journalof
Monetary Economics33 (1994),525-57.
Easterly,William, and Sergio Rebelo. "FiscalPolicy and Economic Growth:An EmpiricalIn-
vestigation."JournalofMonetary Economics32 (December 1993),417-57.
Easterly, William, and Ross Levine. "Africa's Growth Tragedy: Policies and Ethnic Divi-
sions." Quarterly Journalof Economics(November 1997), 1203-50.
Engerman,StanleyL., andKennethL. Sokoloff. "FactorEndowments,Institutions,andDiffer-
entialPathsof GrowthamongNew WorldEconomies:A View fromEconomicHistoriansof
the United States." In How LutinAmericaFell Behind,edited by Stephen Haber, pp.
260-304. Stanford,Calif.: StanfordUniversityPress, 1996.
Fischer,Stanley."TheRole of MacroeconomicFactorsin Growth."Journalof Monetary Eco-
nomics32 (December 1993),485-511.
Gastil, R. D. "TheComparisonSurveyof Freedom:Experiencesand Suggestions."Studiesin
Comparative InternationalDevelopment 25 (1990), 25-50.
Glendon,MaryAnn, MichaelGordon,andChristopherOsakwe. Comparative LegalTradition
ina Nutshell.St. Paul, Minn.:West PublishingCo.,1982.
Goldsmith,RaymondW. FinancialStructure andDevelopment. New Haven,Conn.:Yale Uni-
versityPress, 1969.
Greenwood,Jeremy,andBoyan Jovanovic."FinancialDevelopment,Growth,andthe Distrib-
ution of Income."Journalof PoliticalEconomy98 (October1990, Pt. 1), 1076-1107.
Greenwood,Jeremy,and Bruce Smith. "FinancialMarketsin Development, and the Develop-
ment of FinancialMarkets."Journalof Economic DynamicsandControl21 (January1997),
145-86.
Hansson,Pontus, and LarsJonung."Financeand Growth:The Case of Sweden, 1834-1991."
Unpublished,StockholmSchool of Economics.
Hicks, John.A Theoryof EconomicHistory.Oxford:ClarendonPress, 1969.
Jayaratne,Jith, and Philip E. Strahan. "The Finance-GrowthNexus: Evidence from Bank
BranchDeregulation."Quarterly Journalof Economics111 (August 1996), 639-70.
Kane, EdwardJ. TheGathering Crisisin FederalDepositInsurance.Cambridge:MIT Press,
1985.
. "How Incentive-IncompatibleDeposit-InsuranceFunds Fail." National Bureau of
EconomicResearch,WorkingPaperno.2836, 1989.
King, RobertG., andRoss Levine. "FinanceandGrowth:SchumpeterMight Be Right."Quar-
terlyJournalof Economics108 (August 1993),717-38. (a)
andGrowth:TheoryandEvidence."JournalofMonetary
. "Finance,Entrepreneurship,
Economics32 (December 1993), 513-42. (b)
. "Capital Fundamentalism, Economic Development, and Economic Growth."
Carnegie-RochesterSeries onPublicPolicy40 (June 1994),259-92.

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions
 ROSS LEVINE : 613

Knack, Stephen, and Philip Keefer. "Institutionsand Economic Performance:Cross-Country

Tests Using AlternativeInstitutionalMeasures."Economic and Politics 7 (1995), 207-27.
Kroszner,RandallS., andRaghuramG. Rajan."Isthe Glass-SteagallAct Justified?A Studyof
the U.S. Experiencewith Universal Bankingbefore 1933."AmericanEconomicReview 84
(September1994), 810-32.
Kroszner,RandallS., andPhilip E. Strahan."RegulatoryIncentivesandthe ThriftCrisis:Div-
idends,Mutual-to-StockConversions,and FinancialDistress."Journal of Finance 51 (Sep-
tember1996), 1285-1320.
Laporta,Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert W. Vishny. "Law
andFinance."Journalof Political Economy,forthcoming,1998.
. "Legal Determinantsof External Finance." Journal of Finance 52 (July 1997),
1131-50.
Levine, Ross. "FinancialDevelopmentandEconomic Growth:Views andAgenda."Journalof
EconomicLiterature35 (June 1997), 688-726. (a)
. "Law,Finance,andEconomicGrowth."Unpublished,Universityof Virginia,1997. (b)
Levine, Ross, and David Renelt. "A Sensitivity Analysis of Cross-CountryGrowth Regres-
sions."AmericanEconomicReview 82 (September1992),942-63.
Levine, Ross, and Sara Zervos. "Stock Markets,Banks, and Economic Growth."American
EconomicReview(June 1998).
Mauro,Paolo. "CorruptionandEconomic Growth."QuarterlyJournalof Economics 110 (Au-
gust 1995), 681-712.
McKinnon, Ronald I. Money and Capital in Economic Development. Washington, D.C.:
BrookingsInstitution,1973.
North,Douglas C. Structureand Change in EconomicHistory.New York:Norton,1981.
Patrick,Hugh."FinancialDevelopmentandEconomic Growthin UnderdevelopedCountries."
EconomicDevelopmentCulturalChange 14 (January1966), 174-89.
Rajan,RaghuramG., andLuigi Zingales."FinancialDependenceandGrowth."AmericanEco-
nomicReview(June 1998).
Robinson,Joan."TheGeneralizationof the GeneralTheory."In TheRate of Interestand Oth-
erEssays. London:MacMillan,1952.
Schumpeter,Joseph A. The Theoryof Economic Development, 1912, translatedby Redvers
Opie. Cambridge,Mass.:HarvardUniversityPress, 1934.
Shleifer,Andrei,andRobertW. Vishny. "A Survey of CorporateGovernance."Journalof Fi-
nance 52 (June 1997),737-84.
. "Corruption."QuarterlyJournalof Economics 109 (1993),599-617.
Wachtel,Paul, and Peter Rousseau. "FinancialIntermediationand Economic Growth:A His-
toricalComparisonof the United States,the United Kingdom,andCanada."InAnglo-Amer-
ican Financial Systems,edited by M. D. Bordo andR. Sylla. Homewood, Ill.: Business One
Irwin, 1995.

This content downloaded from 193.255.248.150 on Wed, 14 Jan 2015 03:31:40 AM

All use subject to JSTOR Terms and Conditions