Financial Development and Economic Growth: The Role of Stock Markets
Author(s): Philip Arestis, Panicos O. Demetriades and Kul B. Luintel
Source: Journal of Money, Credit and Banking, Vol. 33, No. 1 (Feb., 2001), pp. 16-41
Published by: Ohio State University Press
Stable URL: http://www.jstor.org/stable/2673870 .
Accessed: 16/07/2014 01:16
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 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
PHILIP ARESTIS
PANICOS O. DEMETRIADES
KUL B. LUINTEL
FinancialDevelopmentandEconomicGrowth:
TheRole of StockMarkets
Utilizing time series methods and data from five developed
economies, we examine the relationshipbetween stock marketdevelopment and economic growth, controllingfor the effects of the
bankingsystem and stock marketvolatility.Ourresults supportthe
view that, althoughboth banks and stock marketsmay be able to
promoteeconomic growth,the effects of the formerare more powerful. They also suggest that the contributionof stock marketson
economic growthmay have been exaggeratedby studies thatutilize
cross-countrygrowthregressions.
THE GROWINGIMPORTANCE
of stock markets arourldthe
worldhas recentlyopened a new avenueof researchinto the relationshipbetween financial developmentand economic growth, which focuses on the effects of stock
marketdevelopmerlt.lIn this context, various stock marketdevelopmentirldicators
have been found to explain partof the variationof growthrates across countries,in
some cases over arldabove the effects of the banking system (Atje and Jovanovic
1993; Levine and Zervos 1998). Since these results have been obtainedfrom crosscountry growth regressions, tlley can at best provide only a broad-brushpicture of
the relationshipbetween stock marketsand growth,the details of which may reasonably be expected to vary considerablyacross countries, depending on institutional
characteristicsand circumstances. Furthermore,given the widespread skepticism
concerningthe robustnessof econometricresults derivedfrom cross-countrygrowth
regressions, these results must be viewed with some caution (Levine and Renelt
The authorsare grateful to two anonymousreferees for constructivecomments. They also thank the
participantsof the 1997 DevelopmentEconomics Study Group annual conference (Universityof Birmingham)and the 1999 Royal Economic Society (RES) annualconference (Universityof Nottingham)for
useful comments.They acknowledgefinancialsupportfrom the ESRC (GrantNo. R000236463).
1. Worldstock marketcapitalizationgrew from $2 trillionin 1982 to $4.7 trillionin 1986, $10 trillion
in 1993 and $15.2 trillion in 1996, implying an averageannualgrowthrate of 15 percent;emergingmarket capitalizationgrew from less than 4 to 13 percentof total world capitalization(Demirgu-Kunt and
Levine 1996; Singh 1997).
PHILIPARESTISis
a professorof economicsat SouthBankUniversity,
London.E-mail:
p.arestis@sbu.ac.ukPANICOS0. DEMETRIADES
is a professorof economicsat LeicesterUni-
versity.KULB.
LUINTELis a readerin economicsat BrunelUniversity.
Journalof Money,Credit,andBanking,Vol. 33, No. 1 (February2001)
Copyright2001 by The Ohio State University
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
PHILIPARESTIS,PANICOSO. DEMETRIADES,AND KUL B. LUINTEL : 17
1992; Arestis and Demetriades 1997; Luintel and Khan 1999).2 In the specific context of the cross-country relationship between stock market development arld
growth, for example, the presence of endogeneity has been shown to considerably
weaken the estimated effect of stock market indicators (Harris 1997). There are,
therefore,importanteconometricadvantagesin examining the role of stock markets
irl the relationship between financial developmerltand growth using time series
methods. Besides being better able to address issues of causality and endogeneity,
they are also less likely to suffer from other limitationsof cross-countrygrowthregressiorls.3Even setting aside econometricissues, time series methods can provide
useful insights irltodifferencesof this relationshipacross countriesarldmay illuminate importantdetails that are hiddenin averaged-outresults.
This paperutilizes time series methodsto reexaminethe relationshipbetween economic growthand stock marketdevelopment,while controllingfor the effects of the
commercial banking sector and stock market volatility. Inevitably, a time series
analysishas its own limitations.Specifically,the need to obtain a long time series of
stock marketdevelopmentindicatorsnarrowsdown the focus of the empiricalanalysis to five developed economies, namely, Germany,the United States, Japan, the
United Kingdom, and France.While the absence of less-developed economies from
our sample means that no direct inferences can be made about the contributionof
stock markets at early stages of economic development, our findings nevertheless
have implications for the debate on bank-basedversus capital-market-basedfinancial systems (see, for example, Rajan and Zingales 1995 and 1996; Horiuchi and
Okazaki 1994; Edwardsand Fischer 1994; Corbettand Jenkinson 1994). Thus, our
results could be indirectlyvaluable for less-developed economies, in that they may
irlformpolicy decisions relating to the adoptionor otherwise of specific types of financial system.
The rest of the paperis structuredas follows. In sectiorl1 we provide a discussion
of the role of stock marketsand banks in the process of economic growth and summarise the existing empiricalliterature.In section 2 we outline our data and econometric methodology. In section 3 we present our findings and discuss their
implicationsfor the debate on financial systems. Finally in section 4 we provide a
summaryand some concludingremarks.
1. STOCKMARKETS,BANKS, AND ECONOMICGROWTH
Positive EMfects
of StockMarkets
Recent theoreticalcontributionssuggest that stock marketsmay promotelong-run
growth. Stock marketsencourage specializationas well as acquisitionand dissemi2. Quah (1993) emphasizes the nonexistence of balancedgrowth paths, Caselli, Esquivel, and Lefort
(1996) andLevine and Renelt (1992) focus on omittedvariablebias or misspecification,Evans (1995) and
Pesaranand Smith (1995) dwell on the heterogeneityof slope coefficients across countries, while problems of causalityand endogeneityare exploredby Demetriadesand Hussein (1996) and Harris(1997).
3. The view that time series studies of economic growthoffer importantadvantagesover cross-country growthregressionsis gaining acceptance.See, for example, Jones (1995), Evans (1997), Kocherlakota
andYi (1997), and Klenow and Rodriguez-Clare(1997).
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
18 : MONEY,CREDIT,AND BANKING
nationof information(Diamond 1984; Greenwoodand Jovanovic 1990; Williamson
1986) andmay reducethe cost of mobilizing savings, therebyfacilitatinginvestment
(Greenwoodand Smith 1997). Well-developedstock marketsmay enhancecorporate
control by mitigatingthe principal-agentproblem throughaligning the interests of
managersand owners, in which case managerswould strive to maximize firm value
(DiamondandVerrecchia1982; Jensen and Murphy1990).
StockMarketLiquidity
Levine (1991) and Bencivenga,Smith, and Starr(1996) suggest thatstock markets
make financialassets tradedin them less risky because they allow savers to buy and
sell quickly and cheaply when they wish to alter their portfolios. Companies at the
same time enjoy easy access to capital throughequity issues. Less-risky assets and
easy access to capital marketsimprove the allocation of capital, an importantchannel of economic growth.More savings and investmenttherebymay also ensue, further enhancing long-term economic growth. It is conceded, though, that increased
liquiditycan also influencegrowthnegatively(see, for example,Levine 1997). There
are three channels throughwhich this may take place (Demirguoc-Kunt
and Levine
1996). The firstis thatgreaterstock marketliquidity,by increasingthe returnsto investment,may reduce savings rates. The second is that, given the ambiguouseffect
of uncertaintyon savings, greaterstock marketliquiditymight in fact reduce savings
rates throughits negative impact on uncertaintysince less uncertaintymay decrease
the demand for precautionarysavings. The third channel operates throughthe euphoriaand myopia thatmay be encouragedby highly liquid stock markets.Dissatisfied participantsfind it easy to sell quickly which can lead to disincentivesto exert
corporatecontrol, thus affecting adversely corporategovernance and hurting economic growthin the process (see, however,Jensen and Murphy1990).
TheRole of Volatility
Anotherimportantcharacteristicof stock marketsis that of price volatility,as this
may underminethe ability of stock marketsto promote an efficient allocation of investment.The undesirableside effects of volatility were recognized early on in the
literature,notablyby Keynes (1936), who was particularlysharp-tongued:
As the organisationof investmentmarketsimproves, the risk of the predominanceof
speculationdoes . . . increase . . . Speculatorsmay do no harm as bubbles on a steady
streamof enterprise. . . a serious situationcan develop . . . when enterprisebecomes the
bubble on a whirlpool of speculation.When the capital developmentof a countrybecomes a by-productof the activities of a casino, the job is likely to be ill-done.... It is
usually agreedthatcasinos should, in the public interest,be inaccessible and expensive.
And perhapsthe same is trueof Stock Exchanges. (pp. 158-59)
More recentliteratureis less conclusive on this issue. While a certaindegree of price
volatility in the stock marketis clearly desirable, since it may reflect the effects of
new informationflows in an efficient stock market,some evidence suggests that the
observed levels of volatility may be "excessive."This may reflect independenceof
stock-market-assetvalues from underlyingfundamentals(Shiller 1981 and 1989),
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
PHILIPARESTIS,PANICOSO. DEMETRIADES,AND KUL B. LUINTEL : 19
even though the debate on the presence of excess volatility in stock returnsis far
from settled. If present,excessive volatility is likely to result in an inefficientallocation of resources, upwardpressures on interest rates in view of the higher uncertainty,hamperingboth the volume and the productivityof investmentand, therefore,
reducinggrowth (Federer1993; DeLong et al. 1989). Furthermore,excessive stock
tradingmay very well induce "noise"into the marketto the detrimentof efficient resource allocation(DeLong et al. 1989).
BANKS
The relationshipbetween stock marketsand growthmay also be influencedby the
link between stock marketsand financialintermediaries,which is not unambiguous.
Stock marketsand banks are clearly substitutesources for corporatefinance since
when a firmissues new equity its borrowingneeds from the bankingsystem decline.
Assuming that banks and financialintermediariesare in a betterposition than stock
marketsto addressagency problems (for example, Diamond 1984; Stiglitz 1985), it
is then possible that stock marketdevelopmentmay hampereconomic growth if it
happensat the expense of bankingsystem development.
Similar views are expressed by the literatureon capital-market-basedfinancial
systems that predicts a very weak relationshipbetween stock markets and growth
since corporateinvestment is not financed throughissues of equity (Mayer 1988;
see, also, Fry 1997). Corbettand Jenkinson(1994) when discussing the contribution
of stock marketto corporateinvestmentfinancing,suggest thatit was negativein the
United Kingdom and only small positive overall in the United States during the
1970s and 1980s. Akyuz (1993) and Singh (1997) arguethat unfavorableeconomic
shocks produce macroeconomicinstability throughthe interactionsbetween stock
marketsand foreign exchangemarkets,which affect economic growthadversely.
On the other hand, at the aggregate level increased stock market capitalization
may be accompaniedby an increase in the volume of bank business, if not an increase in new lending, as financialintermediariesmay provide complementaryservices to issuers of new equity such as underwriting.Thus, it is likely that at the
aggregatelevel the developmentof the stock marketgoes hand in hand with the developmentof the bankingsystem.
EmpiricalEvidence
Existing evidence points to stock marketdevelopmenttakingplace in tandemwith
other aspects of financialdevelopment.Using data for forty-fourindustrialand developing countries for the period 1986-1993, Demirguoc-Kunt
and Levine (1996)
conclude thatcountrieswith well-developed stock marketsalso have well-developed
banks and nonbankfinancial intermediaries,while countries with weak stock markets tend to have weak banksand financialintermediaries.Demirguoc-Kunt
andMaksimovic (1996), in their investigationof the effect of stock marketdevelopmenton
firms'financingchoices in thirtyindustrialand developing economies from 1980 to
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
20
: MONEY,CREDIT,AND BANKING
1991, find thatinitial improvementsin the functioningof a developing stock market
producehigher debt/equityratios for large firms, with small firmsnot being significantly affected. In already developed stock markets,furtherdevelopmentleads to
substitutionof equity for debt financing, especially for long-term debt. Boyd and
Smith (1996) arguethat as an economy develops the ratio of debt to equity tends to
increase,with the two sources of financebeing complementary.
Levine and Zervos (1998), utilizing cross-countryregressions for a number of
countriescovering the period 1976-1993, demonstratethat variousmeasuresof equity marketactivity are positively correlatedwith measuresof real activity and that
the association is particularlystrong for developing countries. Conditioning on a
number of variables, including indicators of banking development, they conclude
that stock marketsprovide differentfinancial services from banks. They argue that
stock marketsmay enhance growth throughliquidity,which makes investmentless
risky, therebyenabling companies to enjoy permanentaccess to capital throughliquid equity issues. Atje and Jovanovic (1993), using a similar approach,also find a
significantcorrelationbetween economic growthand the value of stock markettrading relative to GDP for forty countries over the period 1980-88. However, Harris
(1997) shows thatthis relationshipis at best weak. Reestimatingthe same model for
forty-nine countries over the period 1980-91, but using currentinvestmentrather
thanlagged, andutilizing two-stage least squares,he demonstratesthatin the case of
the full sample (which includesboth developedand developingcountries),and of the
subsampleof developingcountries,the stock marketvariabledoes not offer much incrementalexplanatorypower.In the subsampleof developed countries,althoughthe
level of stock marketactivityhas some explanatorypower,its statisticalsignificance
is weak.
The volatility characteristicof stock marketshas been investigatedfrom the point
of view of its relationshipto the size of stock marketsand capital control liberalization. Demirguoc-Kunt
and Levine (1996) findthatin a sampleof forty-fourdeveloped
and emerging markets from 1986 to 1993, large markets tend to be less volatile.
Also, thatinternationallyintegratedmarketstend to be less volatile. Levine and Zervos (1995) explore the effect of liberalizing capital controls in sixteen countries
which reduced substantiallybarriersto internationalcapital and dividend flows in
the 1980s. They conclude that stock marketvolatility increases significantlyimmediately following capitalcontrolliberalizationin approximatelyhalf of the countries
considered and does not decrease significantly in any of them. This result may be
combinedwith thatof Demirguoc-Kunt
and Levine (1996) to suggest that,in the long
run, stock returnvolatility is lower in countries with more open capital markets.
However,more recently Levine and Zervos (1998) have examinedvolatility in a relationship that concentrateson stock marketliquidity and economic performance.
They conclude that in a cross-section approach,this link is not statisticallyrobust
(and it is unexpectedlypositive in most cases). We investigatethis link furtherin this
paper and are able to improve substantiallyon this result and show that the link between volatility and growth is significantly strong and negative in our time series
analysis as shown below.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
PHILIPARESTIS,PANICOSO. DEMETRIADES,AND KUL B. LUINTEL : 21
2. DATAAND ECONOMETRICMETHODOLOGY
We are motivatedby two primaryobjectives: First, to explore the long-runrelationshipbetween stock marketvolatility,stock marketdevelopment,bankingsystem
developmentand the level of output. In so doing, the magnitudeof the estimated
long-runoutputelasticities with respect to the measuresof bankingsystem development and the stock marketdevelopmentis likely to shed light on the relativeimportance of the two componentsof the financial system for output growth. Second, to
investigatethe causal flows in this relationship,that is, between outputand banking
system developmenton one hand and output and stock marketdevelopmenton the
other.
Data and Measurement
We employ quarterlydata on output and indicatorsof banking system development, stock market development and stock market volatility for Germany during
1973:1-1997:4, the United States for 1972:2-1998:1, Japanfor 1974:2-1998:1, the
United Kingdom for 1968:2-1997:4, and France for 1974:1-1998:1.40ur variables
are measuredas follows. Outputis measuredby the logarithmof real GDP (LY);
stock marketdevelopmentby the logarithmof the stock marketcapitalizationratio
(LMC), defined as the ratio of stock marketvalue to GDP; bankingsystem development by the logarithmof the ratio of domestic bank credit to nominal GDP (LBY);
stock marketvolatility (SMV) is measuredby an eight-quartermoving standarddeviation of the end-of-quarterchange of stock marketprices.5
There are, of course, otherpossible indicatorsof financialdevelopment.As far as
banking sector development is concerned a deposit-based measure could also be
used. Earlierwork by Arestis and Demetriades(1996), however,suggests that in the
case of developedeconomies credit-basedindicatorsare more likely to exhibit a stable long-runrelationshipwith outputthandeposit-basedones. As far as stock market
development is concerned, cross-section studies have found liquidity-based measures, such as the ratio of the value of traded shares to GDP, to be more closely
linked with economic growththanmarketcapitalizationindicators.Nevertheless,in
a time series context the capitalizationindicatorhas a numberof advantagesover
transaction-basedmeasures.Firstly,it is a stock variableratherthan a flow variable
that makes comparisonwith the bank development-basedindicator,which is also a
stock, more meaningful.Secondly, primarilyfor the same reason it is more likely to
have time series propertiesthat make it suitable for cointegrationanalysis. Furthermore, the discussion of section 1 suggests thatthereare also conceptualreasons as to
why stock market capitalizationmay be more closely linked to economic growth
than transactions-basedmeasures. Despite this we test the sensitivity of our results
4. All data series were extractedfrom the online informationservice DatastreamInternational.Stock
marketvariablesare end-of-quarterprice indices and marketvalues.
5. We first calculatedthe logarithmicfirst differencesof the end-of-quarterstock marketprice index.
We then computeda moving eight-quarterstandarddeviationas a measureof stock marketvolatility.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
22
: MONEY,CREDIT,AND BANKING
using alternativemeasuresof stock marketdevelopment;however,this is only possible for the United Kingdom and the United States since data on these variablesare
not availablefor the othercountriesin our sample for a sufficientlylong period.
Methods
We carryout our empiricalinvestigationin a vector autoregression(VAR) framework. Recent literature(for example, Gonzalo 1994; Hargreaves1994; Haug 1996)
suggests thatfor sample sizes of aroundone hundreddatapoints, the maximumlikelihood approachof Johansen(1988) performsin generalbetterthan a range of other
estimators of long-run relationships (cointegrating vectors). Further, Toda and
Phillips (1993), Hall and Wickens (1993), and Hall and Milne (1994) show it to be
an efficient method of testing causality.We thereforefollow this method to identify
the numberof cointegratingvectors amongst the variablesspecified in the VAR and
then examine the directionof causality.
The Johansen(1988) method is based on a vector errorcorrection(VECM) representationof a VAR(p)model which can be writtenas
aXt = rlaxt l + r2aXt 2 + ..............
+ rp laXt p+l + rIXt-p+ vDt + Ut
(1)
where x is an nxl vector of the first order integrated [that is, I(1)] variables,
rl,r2,...,rp are nxnmatricesof unknownparameters,D is a set of I(0) determimistic
variablessuch as constant,trend,and dummies, and u is a vector of normallyand independentlydistributederrorswith zero mean and constant variance. The steadystate (equilibrium)propertiesof equation (1) are characterizedby the rank of [I, a
squarematrixof size n. In our case n = 4. The existence of a cointegratingvector implies that [I is rank deficient. Johansen(1988) derives the maximal eigenvalue and
trace statistic for testing the rank of [I. Appropriatecritical values are tabulatedin
Osterwald-Lenum(1992). If [I is of rankr (0<r<n) then it can be decomposedinto
two matricesot (nxr) and ,3 (nxr)such that
[I = ot,X'.
(2)
The rows of ,3 are interpretedas the distinct cointegratingvectors whereby ,A'xform
stationaryprocesses. The ots are the error correction coefficients that indicate the
speeds of adjustmenttowardequilibrium.Substituting(2) into (1) we get
AXt= rlAxt_l + r2AXt_2+ .- - - + rp_lAxt_p+l + ot(l3'Xt-p)+ VDt + Ut (3)
This is a basic specificationfor the test of long-runcausality.A test of zero restrictions on the otsis a test of weak exogeneity when the parametersof interestare long
run (Johansen and Juselius 1992). Hall and Wickens (1993) and Hall and Milne
(1994) interpretweak exogeneity in a cointegratedsystem as a notion of long-run
causality.We employ weak exogeneity tests to examine the issue of long-runcausal-
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
PHILIPARESTIS,PANICOSO. DEMETRIADES,AND KUL B. LUINTEL : 23
ity between the variablesin the system. The null of ot=O can be tested by the standardlikelihood ratio test.
A numberof issues are importantin the estimation and interpretationof cointegrating vectors. First, in view of the various (financial) policy changes that have
taken place during the sample period, it is plausible to allow for the possibility of
structuralbreakin the cointegratingrelationships.We addressthis issue directly by
testing the null of parameterand rank constancy in the cointegratingrelationships
following Quintos(1995) and Hansenand Johansen(1993, 1998). The hypothesisof
interesthere is that [I and the rankof [I, p(TI),or the numberof cointegratingrelationships remainsstable overtime.The null of parameterand rankconstancy can be
statedas
Ho p(H) = r, or p(H) = p(H) for all t .
(4)
The alternativehypothesisthatallows for both the parametersand numberof cointegratingranksto change is
Ha:p(TI) + p(TI),for all or some of the t .
(5)
Quintos (1995, p. 412) provides a likelihood ratio (LR) statistic which tests the null
of no structuralbreakundera single breakdate. Implementationof Quintos'stest requires splittingthe sample at the breakdate; estimatingseparatemodels for the preand postbreakdates; and testing whether subsample eigenvalues are sigrlificantly
differentfrom those of the full sample.In view of the fairly small sample we have we
do not follow this approach.Instead,we implementthe rankstabilitytests in a recursive frameworkas suggested by Hansen and Johansen(1993, 1998).6 The relevant
LR test can be shown as
q
q,
LR=TE ln(1-Xi)-TjE
i=l
(1-Xli)
i=l
where X and Bl are the full and resursivesample estimates of the eigenvalues of matrix 1I; subscriptj indicates the starting date of recursion such that Tj= T1+ 1,
Tl + 2,...,T. Thus, our approachessentially involves estimating the cointegrating
vector(s) using full sample and then testing whetherthe full sample results (that is,
cointegratingparametersand ranks)remainstable when the model is estimatedover
the recursivesubsample.The recursiveLR test is X2(2)distributed.
Second, it is shown that cointegratingrelationshipsare sensitive to the treatment
of deterministic terms in the cointegrating space (Baillie and Bollerslev 1994;
Diebold, Gardeazabal,andYilmaz 1994). To resolve this, Johansen(1992) suggests
6. It should be noted that Quintos's test is based on Hansen and Johansen(1993). One of the advantages of recursivetests of structuralbreakis that we do not requireto identify break date endogenously
which is importantin view of our sample size.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
24
: MONEY,CREDIT,AND BANKING
identifying appropriatedeterministicterms in the cointegratingspace throughrank
tests following the so-called Pantulaprinciple. Crowderand Hoffman (1996) and
LuintelandPaudyal(1998), among others,implementcointegrationtests along these
lines and we follow this approach.Third,it is well known that the results are sensitive to lag length selection in the VAR. Generallylag lengths are specified following
some informationcriteria(for example,Akaike 1973; Schwarz 1978). However,Hall
(1989) and Johansen(1992) suggest thatthe lag length should be specified such that
the VAR residuals are empirically Gaussian. Cheung and Lai (1993) show that the
lag length selection based on informationcriteriamay not be adequatewhen errors
contain moving average terms. We specify lag lengths as the minimum length for
which thereis no significantautocorrelationin the estimatedVARresiduals.
Finally, the identification and interpretationof the cointegrating vectors, ,X,as
long-run economic relationships is another pertinentissue. Given r cointegrating
vectors, Johansen(1991) suggests identificationthroughthe test of r2just-identiJ9ing
restrictions.Pesaranand Shin (1994), however, show that Johansen'sidentification
scheme is deficient in that the maximized log-likelihood values associatedwith any
set of just-identifying
restrictionswould be identical, which makes it impossible to
distinguishbetween the competing set of restrictions.Instead,they suggest identification of long-run economic relationshipthroughthe tests of over-identifyingrestrictions.This requirestesting for r2+k (where k ' 1) restrictionsin a cointegrating
space spannedby r stationaryrelationshipswhich gives k over-identifyingrestrictions. It should be noted that when a unique cointegratingvector is found then the
problemof interpretationdoes not arise, and one may simply test for just identifying
restrictionsin the form of a normalizationrestriction.
3. EMPIRICALRESULTSAND ANALYSIS
We begin by carryingout unit root tests which suggest that the all variables are
I(1).7We then performcointegrationanalysis for each of the five countries, the results of which arereportedin Tables 1-5. In orderto allow for any deterministicseasonality, centered quarterlydummies are included in unrestrictedform throughout
the estimation.Part(a) of each table contains the results from recursiveestimation.
In view of the sample size, the startingyear for the recursiveestimationis chosen as
1990(4) for all countriesexcept for the United Kingdom, in which case because of
the longer sample we begin recursiveestimationat 1987(4). The last column of each
table reportsthe LR tests underthe null thatthe full samplecointegratingrankis stable over the recursivesubsample.The rejectionof the null indicates structuralbreaks
in the cointegratingrankthatforms the basis for the introductionof appropriateshift
dummies.
Once the structuralbreak is identified,we then reestimatethe cointegrationrank
7. Unit root tests for all variables(which are not reportedhere) can be obtainedfrom the authorsupon
request.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
PHILIPARESTIS, PANICOSO. DEMETRIADES,AND KUL B. LUINTEL : 25
for the full sample by allowing for structuralshifts through shift dummies.8Since
tracestatisticsandmaximaleigenvaluestatisticsprovidequalitativelysimilarresults,
only the formerare reportedfor the sake of brevity.In the event of multiple cointegratingvectors, identificationof each vector to an economically interpretablerelationship is achieved throughtests of over-identifyingrestrictions(see Pesaranand
Shin 1994). Each vector is normalizedon the variablefor which we could find evidence of errorcorrection(thatis, negative and significantloading factors, ot).These
results are reportedin part(b) of each table. Finally in part(c) we reportthe results
of weak exogeneity tests, which are expected to shed light on the patternsof longrun causalityin each system.
The resultson Germany,reportedin Table 1, show evidence of a breakin the cointegratingrank during 1991-92, which coincides with the period of the Germanreunification.Once this structuralbreakis takeninto accountthroughthe introduction
of an interceptdummyfor the periodof 1991(1)-1992(4) in the cointegratingspace,
we continue to find evidence of a single cointegratingvector.Tests of normalityand
serial correlationsuggest thatthe VAR residualsare empiricallyGaussian.9
The cointegratingvector is normalizedon output,given the correctly signed and
stronglysignificanterrorcorrectionterm (ot).The cointegratingvector for this country shows a positive relationshipbetween the level of real GDP and bankingsystem
development,as well as a positive stock marketcapitalizationeffect. It also shows
that stock marketvolatility,a variabletreatedweakly exogenous to the system, has a
positive but insignificanteffect.l° Banking system development is endogeneous to
the output vector whereas stock market capitalizationis not. Hence, in Germany,
thereis bidirectionalcausalitybetween bankingsystem developmentand the level of
outputwhile stock marketcapitalizationis weakly exogeneous to the outputvector
in the long run. Stock marketcapitalization,however,affects GDP throughthe positive and significantcointegratingparameter.The coefficients on LBY and LMC are
significantat the 1 percentlevel, with the formerbeing more than three times larger
thanthe latter.These results are clearly not surprisinggiven the close relationshipof
the bankingsystem with industryin Germanyand the relativelyminorrole played by
the stock marketthere (see, for example,Arestis and Demetriades1996).
In the case of the United States (see Table2) the pictureis ratherdifferentin view
of the endogeneity of stock marketcapitalizationand the weak exogeneity of real
GDP, once the structuralshift in 1990:1-1994:4 is accountedfor. This period coincides with a downturnin the U.S. economy, a substantialfall in bond marketyields
and a serious numberof defaults of "junk"bonds. There is only one cointegrating
vector for this country,which is normalizedon LMC.Accordingto this vector,LMC
8. We tested for both slope and interceptdummiesbut the formerwere insignificantin all cases.
9. In all cases recursive estimation is conducted without the introductionof any dummy variable.
However,few dummieswere introducedin the otherestimationsto captureblips in the data.Exclusion of
these dummiesdoes not change the resultsqualitatively,except for failureof the normalitytest. These are
not reportedbut can be obtainedfrom the authors.
10. The likelihoodratiotests could not rejectthe weak exogeneity of SMV.The test statisticis distributed as chi-square(1)which gives ap-value of 0.136. In Tables 1-7, the coefficients for SMV, unlike the
othercoefficients,are not elasticities.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
TABLE 1
FINANCIAL
DEVELOPMENT
ANDECONOMIC
GROWTH
IN GERMANY
1A: RECURS1VE
ESTIMAT10N
OFCO1NTEGRATING
VECTORS(LAG=5)
TraceStatisticHo rank=p
Sample:
1973(1)-
p=O
1990(4)
1991(4)
1992(4)
1993(4)
1994(4)
1995(4)
1996(4)
1997(4)
p' 1
48.18***
52.89***
49.16''''''?
45.19***
46.46***
46.80'''5''''
46.03***
47.90
17.31
16.10
15.06
14.68
15.79
16.22
17.12
17.58
Eigenvalues
p'2
2.71
3.05
2.24
1.92
2.07
1.92
1.64
1.71
k1
k2
0.369
0.404
0.365
0.320
0.309
0.296
0.272
0.273
0.196
0.168
0.157
0.149
0.152
0.152
0.156
0.154
Rankstabilitytest
4
0.039
0.042
0.029
0.024
0.025
0.022
0.018
0.018
0.54
6.54**
5.98''''''
0.81
0.34
0.26
1.42
Full sample
1B: ESTIMATED COINTEGRATINGVECTOR AFTER ALLOWING FOR STRUCTURAL BREAK (LAG=5)
p=0
TraceStatisticHo rank=p
p' 1
53.67***
Normalizedon LY
Constant
Coefficient
p-value
5.893
0.0002
18.03
Dummy: 1991(1)-1992(4)
0.0633
0.0623
p'2
2.59
LMC
0.1316
0.0079
NorEs: 1. Treatedas weakly exogenous.
p-values are that of the likelihood ratiotests underthe null that the parameteris zero
Vectorautocorrelationtests: F(45, 158) = 1.1959 [0.2110]
Vectornormalitytest: chi-square(6): 4.082 [0.6652]
c: WEAK EXOGENEITY TESTS
LY
Loading
p-value
(0C)
-0.1450
0.000
LMC
0.2550
0.3678
LBY
0.2148
0.000
NorEs:p-values are thatof the likelihood ratiotests underthe null that the loading factoris zero.
***, ** and * indicate statisticalsignificanceat 1 percent,5 percent,and 10 percent,respectively.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
LBY
SMVi
0.4405
0.0000
1.239
0.1243
oos
TABLE
......
2
FINANCIAL DEVELOPMENT AND ECONOMIC GROWTH IN THE UNITED STATES
2A: RECURSIVE ESTIMATIONOF COINTEGRATINGVECTORS (LAG=4)
Sample:
1972(2)-
TraceStatisticH0:rank=p
p=O
P' I
p'2
990(4)
991(4)
992(4)
993(4)
994(4)
995 (4)
996(4)
998(1)
78.0***
71.29
67.04***
67.14**t
69.004'*$
69.23
70.39***
71.58
43.23*'**
43.28'
42.28***
41.21i*'
42.89***
42.73
43.76***
42.24
Eigenvalues
X,
19.14,,,
20.08'
19.70
19.00
19.42
19.41
19.9l
19.62
0.400
0.317
0.278
0.277
0.267
0.260
0.251
0.261
k
L
0.298
0.282
0.257
0.242
0.249
0.233
0.228
0.208
Rankstabilitytest
0.176
0.162
0.137
0.135
0.132
0.131
0.130
0.131
6.9**,,,
12.01'
4.61
3.82
2.39
2.31
1.48
Full-sample
2B: ESTIMATED COINTEGRATINGVECTOR AFTER ALLOWING FOR STRUCTURAL BREAK (LAG=4)
p=0
TraceStatisticHo rank=p
p' 1
35.07***
Normalizedon LMC
Coefficient
p-value
Constant
N/a
p'2
10.32
Dummy: 1990(1)-1991(4)
-0.1939
0.465
2.5
LY
3.208
0.000
NorEs: 1. Treatedas weakly exogenous.
p-values are that of the likelihood ratiotests underthe null that the parameteris zero.
VectorautocolTelationtests: F(45,170) = 1.219 [0.184]
Vectornormalitytest: chi-square(6): 8.092 [0.231]
2C: WEAK EXOGENEITY TESTS
LY
Loading
p-value
(0C)
-0.0013
0.617
LMC
-0.109
0.0001
LBY
0.003
0.269
NorEs:p-values are thatof the likelihood ratiotests underthe null thatthe loading factoris zero.
***, ** and * indicatestatisticalsignificance at 1 percent,5 percent,and 10 percent,respectively.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
LBY
0.447
0.779
SMVI
-6.331
0.181
28
: MONEY,CREDIT,AND BANKING
is positively relatedto real GDP and to bankingsector development(LBY) and negatively relatedto stock marketvolatility. However,only the GDP coefficient is significant in this vector. The weak exogeneity of real GDP and banking system
developmentsuggests that capital marketdevelopment,which in the long run is influencedby these two weakly exogenous variables,has no long-runcausal influence
on either of these variables.This result shows that there is clear evidence to suggest
thatin the U.S. financialdevelopmentdoes not cause real GDP in the long run. Consequently,these results are in sharpcontrastto the ones obtainedfor Germanyand
may, to some extent, reflect the internationalcharacterand the natureof the banking
system in the United States which is a capital market-basedsystem (as opposed to
the bank-basedsystem in Germany).
The results for Japanare reportedin Table 3. The recursiveestimates show two
cointegrating vectors throughout 1990-1995 and one cointegrating vector after
1996. Such a shift in the numberof significanteigenvalues is indicative of a structural break in the long-run relationship. We therefore treat the data points up to
1995:4 as the full sample and computeLR tests of rank stabilityfor the other years
reportedin the table. Tests reveal that in Japanstructuralbreaksin the cointegrating
ranksappearduring1991-1992 and 1996-1998. The formerperiodcoincides with a
steep decline in bank profits, reflecting worsening in the scale of nonperforming
loans, due to a sharpfall in asset values, especially in the propertymarket,and a tight
monetarypolicy. The structuralbreakduringthe period 1996:1-1998:1 is clearly not
surprisinggiven that this period coincides with one of the worst economic crises in
Japan'spostwarhistory.Importantly,the Japanesefinancial system has been at the
centerof these problems,with many financialinstitutionsbecoming insolvent. Once
this break is taken into account throughthe introductionof interceptdummies for
1996:1-1998:1 two cointegrating vectors emerge. These vectors are respectively
normalizedon real GDP and banking system development,which show clear evidence of errorcorrection.The over-identifyingrestrictionsthat are accepted by the
data include two normalizationrestrictions,linear homogeneity of LBY and a positive coefficient on LMC in the firstvector and exclusion of LY from the second vector. The first vector shows that both the stock market and the banking system
development indicators influence output positively; however, the influence of the
formeris aboutone-sixth of the latter.Stock marketvolatility,on the otherhand, exerts a negativeand significantinfluenceon the developmentof the real economy.The
second vector is essentially a positive and significantrelationshipbetween the banking system developmentand stock marketcapitalization.Once again, stock market
volatility enterswith a negative and significantcoefficient,suggesting thatincreased
stock marketvolatility a weakly exogenous variableto the systemll impactsnegatively on the developmentof the banking system. The estimatedbanking development vector displays a downwardshift during1996:1-1998:4, reflectinga significant
autonomousshift of 1.13 percent.This effect is not significantin the outputvector.
11. Weakexogeneity test of SMV from the system assumes a p-value of 0.972 which is chi-square(2)
distributed.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
4.s
TABLE
...
s
3
FINANCIAL DEVELOPMENT AND ECONOMIC GROWTH IN JAPAN
3A: RECURSIVE ESTIMATION OF COINTEGRATINGVECTORS (LAG=5)
Sample:
1973(1)-
TraceStatisticHo rank=p
p=0
pSl
ps2
990(4)
991(4)
992(4)
993(4)
994(4)
995(4)
996(4)
1998(1)
68.46***
69.468*t
68.56***
65.73***
61.42***
59.83-'
51.80*
53.40**
41.27***
41.54**t
35.53**
35.12**
33.97*
36.09v'
27.99
28.32
Xl
18.91
16.56
15.35
14.58
12.38
13.39
13.71
14.71
0.333
0.325
0.356
0.321
0.282
0.239
0.230
0.230
Eigenvalues
2
0.284
0.297
0.236
0.229
0.229
0.230
0.145
0.132
4
Rankstabilitytest
0.167
0.135
0.160
0.132
0.112
0.119
0.114
0.118
3.1 1
6.47**
6.77**
4.72
2.61
#
22.63t**
21.37***
# 1973(1)-1995(4) treatedas full sample (see text for details).
3B: ESTIMATED COINTEGRATINGVECTORS AFTER ALLOWING FOR STRUCTURAL BREAK (LAG=5)
TraceStatisticHo rank=p
ps 1
p=0
80.91***
Vector
ps2
23.01***
7.78
1
Normalizedon LY
Constant
Coefficient
p-value
Dummy: 1996(1)-1998(4)
12.92
0.000
-0.0111
0.3866
LBY
LMC
1.000
0.000
0.130
0.008
Vector 2
Normalizedon LBY
Constant
Coefficient
p-value
0.5939
0.0002
Dummy: 1996(1)-1998(4)
-0. 1130
0.0381
LMC
SMVl
0.304
0.000
-1 .755
0.000
NorEs: 1. Treatedas weakly exogenous.
p-values are thatof the likelihood ratiotests underthe null that the parameteris zero.
Test of over-identifyingrestrictions:chi-square(1)= 1.8842 [0.1699]
Vectorautocorrelationtests:F(45,161)=1.1371[0.2782]
Vectornormalitytest: chi-square(6): 8.7299 [0.1890]
3C: WEAK EXOGENEITY TESTS
LY
Loading
p-value
Loading
p-value
(0C) of vector
1
(0C) of vector 2
-0. 1471
0.0000
0.0763
0.0000
LMC
0. 1661
0.0119
-0.1410
0.011 1
LBY
-0.901
1
0.0000
0.9005
0.0002
NorEs:p-values are thatof the likelihood ratio tests underthe null that the loading factoris zero.
***, ** and * indicate statisticalsignificanceat 1 percent,5 percent,and 10 percent,respectively.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
SMV
-1.336
0.000
30
: MONEY,CREDIT,AND BANKING
Finally,the weak exogeneity tests suggest a feedbackrelationshipbetween real GDP
and both partsof the financialsystem since all threevariablesare endogenous to the
system. Withperhapsthe exception of the negativeand significantinfluenceof stock
marketvolatility,these results should not be surprisingin view of the relativeimportance of the bankingsectorin Japan(CorbettandJenkinson1994;Arestis andDemetriades 1996).l2
The results pertainingto the United Kingdom, presentedin Table 4, display significantdifferenceswith those of the other countries,reflectingperhapsthe uniqueness of its financialsystem. To startwith, we find evidence of a structuralshift in the
estimatedrelationshipsduring 1987:1-1991:4. As 1987 was the year of one of the
most importantderegulationsof the U.K. financialsystem in recenthistory the Big
Bang this is once again not a surprisingresult.There were also statisticalredefinitions in the mid-1980s, pertainingto the inclusion of Building Societies in the banking system statistics. Once these structuralshifts are taken into account, we find
evidence of two cointegratingvectors. They are normalizedon stock marketcapitalization andbankingdevelopment.The data-identifyingrestrictionsare the following
five: exclusion of real GDP from the firstcointegratingvector,exclusion of LMC and
linearhomogeneity between LBY and LY in the second vector, and two normalization restrictions.The firstvector is a simple and straightforwardpositive relationship
between stock marketcapitalizationand banking sector development:the two parts
of the financial system exhibit a stable long-run positive association subject to a
shift in the 1987-91 period. It also shows that stock marketvolatility impacts negatively on stock marketcapitalization.The second vector suggests thatbankingsector
developmentis explainedby real GDP growth,while stock marketvolatility exerts a
significant negative influence. The weak exogeneity tests show that real GDP is
weakly exogeneous with respectto the LBY vector, and marginallyso in the case of
the LMC. Thus, the LMC vector appearsto cause LY, although marginally,in the
long run. The long-run banking development vector has no relationship with the
level of output and stock marketdevelopment.In the long run, causality runs from
LBY to LMC and from LMC to GDP.There is no directlong-runcausal relationship
between bankingsystem developmentand real GDP for this country.
In conclusion, the evidence on the United Kingdom suggests that in the long run
causalityflows from bankingsystem developmentto stock marketdevelopment.It is
also evident, however,that the flow of causality from financialsystem development
to real GDP is, at best, weak. On the other hand, banking system developmentand
stock marketdevelopment are both negatively affected by stock marketvolatility.
This evidence could also be interpretedas suggesting that the U.K. financialsystem
is not a strongpromoterof domestic economic growth,which to some extent reflects
its weak links with industry,in thatit is a typical capitalmarket-basedsystem, andits
internationalcharacter.
Turningfinally to France,we find evidence of instabilityin the cointegratingrank
12. Results including shift dummies for the period 1991(1)-1992(4) are qualitativelysimilar.Hence
they are not reportedfor the sake of brevitybut are availableon request.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
TABLE
4
IN THE UNITED KINGDOM
FINANCIAL DEVELOPMENT AND ECONOMIC GROWTH
4A: RECU,;tSIVEESTIMATIONOF COINTEGRATINGVECTOI;LS(LAG=5)
Eigenvalues
TraceStatisticHo rank=p
p'2
p=0
pcl
Sample
1976(1)-
67.40***
59.47***
49.66*
51.45*
51.85*
53.32**
54.65**
53.89***
55.52**
56.60**
56.75
987(4)
988(4)
989(4)
990(4)
991(4)
992(4)
993(4)
994(4)
995(4)
996(4)
1997(4)
14.78
15.94
10.44
10.62
9.68
9.52
10.54
10.96
11.46
12.01
12.59
35.85***
36.67***
25.44
26.11
25.62
25.93
26.39
24.79
25.13
25.16
24.34
,
4
4
Rankstabilitytest
0.329
0.240
0.243
0.243
0.241
0.242
0.240
0.238
0.240
0.240
0.240
0.234
0.221
0.158
0.156
0.154
0.153
0.143
0.121
0.116
0.108
0.094
0.148
0.153
0.090
0.086
0.071
0.065
0.067
0.069
0.069
0.068
0.067
20.171***
11.099**@
8.532**
7.062**
6.174**
4.017
4.429
2.610
2.019
0.969
Full-sample
4B: ESTIMATED COINTEGRATINGVECTORS AFTER ALLOWING FOR STRUCTURAL BREAK (LAG=5)
TraceStatistic Ho rank=p
15.63
47.72**
105.50**
Vector
pc2
pc 1
p=o
1
NormalizedonLMC
Constant
Coefficient
p-value
0.791
0.0004
Dummy: 1986(1)-1991(4)
0.490
0.026
LBY
SMV
0.499
0.0527
-33.57
0.000
LY
SMV
Vector 2
Normalizedon LBY
Constant
Coefficient
p-value
-5.355
0.0011
Dummy: 1986(1)-1991(4)
-31.080
0.000
1.000
0.000
0.8173
0.0000
NarEs:p-values are thatof the likelihood ratiotests underthe null that the parameteris zero.
Test of over-identifyingrestrictions:chi-square(l) = 1.0176 [0.3131]
Vectorautocorrelationtests: F(80,262) = 1.001 [0.485]
Vectornormalitytest: chi-square(8): 11.933 [0.1542]
4C: WEAK EXOGENEITY TESTS
LMC
Loading
p-value
Loading
p-value
(0C) of vector
1
(0C) of vector 2
-0.0210
0.0303
-0.0012
0.6011
LY
LBY
0.081
0.000
-0.082
0.0001
0.0130
0.0531
-0.006
0.3247
NorEs:p-values are that of the likelihood ratiotests underthe null thatthe loading factoris zero.
***, ** and * indicate statisticalsignificanceat l percent,5 percent,and l0 percent,respectively
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
SMV
-0.014
0.000
0.003
0.000
32
: MONEY,CREDIT,AND BANKING
during 1990:1-1993:4. This period follows the completion of financial reforms,
which commenced in 1984 with the comprehensivereformof Frenchbankinglegislation, the official policy of deregulation and the harmonization of institutional
arrangementswithin the EuropeanUnion. By 1992 capitalaccountliberalizationhad
been completed.Also, 1992-93 saw the ERM turmoilwith the Frenchfranccoming
understrongspeculativeattack.Furthermore,given the importantderegulationof the
Frenchfinancialsystem that took place duringthe mid-1980s, we allowed for additional structural shifts by incorporating an intercept dummy for the period of
1985:1-1985:4 in our estimation.l3Thus, for France,we allow for two sets of shift
dummies,for 1985:1-1985:4 and 1990:1-1993:4.
Cointegrationresults for France, which take into account the above structural
shifts, are reportedin TableSb. The trace statisticshows evidence of two cointegrating vectors. On the basis of the sign and significance of the associated loading factors we normalize them on real GDP and stock marketcapitalizationrespectively.
The identifying restrictionsthat are data acceptable comprise exclusions of LMC
and the 1985 dummy from the first vector, exclusion of LY from the second vector,
and the two normalizationrestrictions.The first vector shows a positive impact of
bankingsystem developmentand capitalmarketon real GDP and a negativeand significanteffect of stock marketvolatility.The second vector, portraysa positive relationship between the banking system and stock marketcapitalizationwhile stock
marketvolatility is found to have a negative influence on LMC. This vector appears
to have been affectedby the financialreformsof the mid-1980s, which seem to have
boosted stock marketcapitalization.In this sense this provides supportto the argument put forwardby Arestis and Demetriades(1996) that the French financial system shifted towardbecoming a capital-market-basedone, following those reforms
(see, also, Bertero 1994). Althoughboth the stock marketcapitalizationand banking
system developmentappearwith positive and significantcointegratingparametersin
the outputvector,it is importantto note that the magnitudeof the latter'scoefficient
is almost seven times thatof the former.Thus, the effect of bankingdevelopmenton
real GDP appearsto be much strongerthan that of the stock market.The weak exogeneity tests reveal that no variableis weakly exogenous with respect to any of the
vectors, suggesting an abundanceof feedback relationshipsbetween the variablesin
the system. In the long run there is feedback between LY and LMC as all loadings
are significant.Both real outputand stock marketdevelopmentvectors enterinto the
banking system development suggesting that in the long run both LY and LMC
cause LBY.
In conclusion, the resultson Francesuggest thatwhile the Frenchfinancialsystem
has directly contributedto long-termoutputgrowth, the role of the banking system
has been much more substantialthanthatof the stock market;this was in spite of the
growing importanceof the latter as a result of the financial reforms of the 1980s.
Furthermore,even though the long-run developmentof the stock markethas gone
13. In view of the sample size we did not compute recursive tests of stability for 1985:1-1985:4
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
5
TABLE
IN FRANCE
F1NANCIAL DEVELOPMENT AND ECONOMIC GROWTH
5A: RECURSIVE ESTIMATION OF COINTEGRATINGVECTORS (LAG=4)
rank=p
TraceStatisticHO:
Sample:
1973(1)-
p=0
990(4)
991(4)
992(4)
993(4)
994(4)
995(4)
996(4)
1998(1)
52.22**
53.40**
55.11***
55.198**
56.22***
58.42***
60.21***
60.43
Eigenvalues
0.204
0.190
0.180
0.178
0.165
0.167
0.154
0.149
0.333
0.322
0.316
0.295
0.307
0.298
0.304
0.288
9.23
10.20
11.18
1 1.58
10.26
1 1.28
11.53
11.89
24.72
25.40
26.25
27.26
25.41
27.31
26.89
27.53
4
Rankstabilitytest
0.127
0.131
0.133
0.132
0.115
0.118
0.117
0.115
5.40*
4.90*
4.97*
4.98*
2.08
1.79
0.42
2
kl
ps2
pc 1
Full-sample
5B: ESTIMATED COINTEGRATINGVECTORS AP liER ALLOWING FOR STRUCTURAL BREAK (LAG=4)
TraceStatistic H rank=p
p' 1 °
p=0
10.69
41.48***
1 16.4***
Vector
p52
1
-3.358
0.0023
0.3730
0.0023
-0.0383
0.0003
0.0561
0.0017
Coefficient
p-value
SMV
LBY
Dummy 1990(1)-1993(4)
LMC
Normalizedon LY
Vector 2
Dummy: 1985(1)-1986(4)
Normalizedon LMC
5.164
0.000
0.738
0.0076
Coefficient
p-value
SMV
LBY
-56.060
0.0036
NarEs:I7-valuesare that of the likelihood ratiotests underthe null thatthe parameteris zero
Test of over-identifyingrestrictions:chi-square(1)= 0.0068 [0.9342]
Vectorautocorrelationtests: F(80,187) = 1.1606 [0.206]
Vectornormalitytest: chi-square(8): 12.454 [0.1321]
5C: WEAK EXOGENEITY TESTS
LY
Loading
p-value
Loading
p-value
(0C) of vector
1
(0C) of vector 2
-0.0160
0.0059
-0.002
0.005
LMC
LBY
SMV
1.0730
0.0031
-0.057
0.0025
0.0790
0.0003
0.0060
0.0039
-0.1040
0.0000
-0.0006
0.0068
NOTES:
p-values are thatof the likelihood ratiotests underthe null thatthe loading factor is zero.
***, ** and * indicatestatisticalsignificance at 1 percent,5 percent,and l0 percent,respectively.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
34
: MONEY,CREDIT,AND BANKING
hand in hand with the development of the banking system, the former appearsto
have been much more of a follower in the process of economic development, respondingpositively to both outputgrowth and bankingsystem development.On the
otherhand, stock marketvolatility seems to have been detrimentalto both long-term
outputgrowthand stock marketdevelopment.
Alternative
Measuresof StockMarketDevelopment:
SensitivityChecks
In orderto check the sensitivityof our resultsvis-a-vis the alternativemeasuresof
stock market development, we repeated the empirical analysis using two further
measures, namely, the logarithmsof the ratio of total stock markettransactionsto
GDP (TRY) and the ratio of total stock market transactionsto market valuation
(TRMV) for the United States and the United Kingdom.The stock markettransaction datafor the United Kingdomare availablefor a periodof 1976:1-1998:1 andfor
the US 1973:1-1998:1. In view of the shortersamples for the United Kingdom, this
sensitivity analysis should be taken as indicative only. For other sample countries
lack of data on stock markettransactions,or the insufficientlength of this time series, preventedus from carryingout sensitivity analysis.
The results,obtainedfrom alternativemeasures of financial development are reportedin Tables 6 and 7. They are, in qualitativeterms,broadly similarto those obtained using the stock marketcapitalizationvariable.Therefore,in the interests of
brevity we report only a summary of these findings. Specifically, for the United
Stateswe continueto find one cointegratingvector,whicheverof the above measures
is used. In either case the cointegratingvector is normalized on the stock market
variable in view of the correct sign and significance of the correspondingloading
factor.The normalizedvector depicts a positive association between stock market
development,banking system development, and real GDP. Real GDP and banking
system developmentappearweakly exogenous when using the first variablebut become endogenous when using the second. Stock marketvolatility has negative effects in both vectors, albeit significant only in the TRMV vector. Further,weak
exogeneity tests reveal qualitativelysimilarcausal patternsto those reportedin Table
2c with respect to TRY vector whereas LY and LBY become endogenous with respect to TRMV vector.
In the case of the United Kingdom, the results reportedin Table 7 still show two
cointegratingvectors with the firstalternativemeasureof stock marketdevelopment,
TRY.The first cointegratingvector, normalizedon TRY, is very similar to the one
normalizedon LMC (see Table 4b) and shows a positive association between stock
marketdevelopmentand bankingsector development;stock marketvolatility exerts
a significantnegative effect on TRY.The second vector, normalizedon LBY, now
shows an insignificantreal outputeffect but the volatility effect is negativeand much
stronger.Weak exogeneity tests suggest that there exists a feedback effect between
stock market development and banking system development;however, the causal
flows from financialvariablesto outputaremore significantnow. Withthe second alternativemeasure(TRMV), we find only one cointegratingvector, which is normalized on LBY and shows a positive and significant relationshin between LBY and
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
TABLE
6
FINANCIAL DEVELOPMENT AND ECONOMIC GROWTH IN THE UNITED STATES
TRANSACTIONS-BASED MEASURES OF STOCK MARKET DEVELOPMENT SAMPLE:
1973:1-1998: 1
A. RATIO OF STOCK MARKET TRANSACTIONS TO GDP RATIO (TRY)
ESTIMATED COINTEGRATINGVECTOR AFTER ALLOWING FOR STRUCTURAL BREAK
Trace StatisticHo rank=p
p' 1
p=0
31.32***
Coilltegratillg
12.52
p'2
2.99
Vector
NormalizedonTRY
Coefficient
p-value
Constant
Dtlmmy: 1987(1)-1991(4)
-0.128
N/a
0.532
LBY
LY
SMV
1.464
0.235
5.299
0.003
-1.775
0.641
NorEs: 1. Treatedas weakly exogenous.
Vectorautocorrelationtests: F(80,116) = 1.281 [0.135]
Vectornormalitytest: chi-square(8): 9.323 [0.156]
WEAK EXOGENEITY TESTS
LY
Loading
p-value
TRY
0.0024
0.5881
(0C)
-0.2528
0.0181
LBY
-0.005
0.3026
B. RATIO OF STOCK MARKET TRANSACTIONS TO STOCK MARKET VALUATION (TRMV)
ESTIMATED COINTEGRATINGVECTOR AFTER ALLOWING FOR STRUCTURAL BREAK; SAMPLE:
TraceStatisticHo rank=p
ps 1
p=0
36.80***
Coilltegratillg
ps2
2.477
Vector
NormalizedonTRMV
Coefficient
p-value
12.28
Constant
Dummy: 1987(1)-1991(4)
LBY
LY
SMV
0.181
0.426
3.372
0.044
2.527
0.009
-7.144
0.054
N/a
Vectorautocorrelationtests: F(80,116) = 1.225 [0.178]
Vectornormalitytest: chi-square(8): 8.24 [0.201]
WEAK EXOGENEITY TESTS
LY
Loading
p-value
* k*,
**
(0C)
1973:1-1998: 1
0.0068
0.0440
TRMV
-0.0815
0.078
LBY
-0.0063
0.0642
and * indicate statisticalsignificanceat 1 percent,5 percent,and 10 percent,respectively.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
7
TABLE
FINANCIAL DEVELOPMENT AND ECONOMIC GROWTH IN THE UNITED KINGDOM
TRANSACTIONS-BASED MEASURES OF STOCK MARKET DEVELOPMENT
A. RATIO OF STOCK MARKET TRANSACTIONS TO GDP RATIO (TRY); SAMPLE: 1976:1-1997:2
ESTIMATED COINTEGRATINGVECTOR AFTER ALLOWING FOR STRUCTURAL BREAK
TraceStatistic Ho rank=p
ps 1
p=0
84.51
Vector
ps2
35.24
18.75
1
NormalizedonTRY
Constant
Dummy: 1987(1)-1991(4)
LBY
Coefficient
p-value
-0.932
0.762
0.553
0.020
0.630
0.000
Normalizedon LBY
Constant
Dummy: 1987(1)-1991(4)
LY
Coefficient
p-value
-0.697
0.006
2.122
0.003
- 8.506
0.120
Vector
SMV
-34.74
0.006
2
SMV
- 131.40
0.0054
NorEs:p-values are that of the likelihood ratiotests underthe null that the parameteris zero.
Vectorautocorrelationtests: F(80, 116) = 1.023 [0.452]
Vectornormalitytest: chi-square(8): 11.839 [0.156]
WEAK EXOGENEITY TESTS
LY
Loading
p-value
Loading
p-value
(0C) of vector
1
(0C) of vector
2
TRY
0.0069
0.0045
-0.002
0.005
LBY
-0.154
0 0004
0.030
0.0002
SMV
0.1096
°°°°
-0.0363
0.000
B. RATIO OF STOCK MARKET TRANSACTIONS TO STOCK MARKET VALUATION (TRMV)
ESTIMATED COINTEGRATINGVECTOR AFTER ALLOWING FOR STRUCTURAL BREAK; SAMPLE:
TraceStatistic Ho rank=p
ps 1
p=0
76.57***
CoilltegratiTlg
33.720
0.1340
0-0004
-0.0065
0.0006
1976:1-1992:2
ps2
15.19
Vector
Normalizedon LBY
Constant
Dummy: 1987(1)-1991(4)
Coefficient
p-value
-8.607
0.452
0.003
0.001
TRMV
LY
0.612
0.0004
1.598
0.258
SMV
- 15.85
0.000
Vectorautocorrelationtests: F(80,116) = 1.072 [0.365]
Vectornormalitytest: chi-square(8): 9.920 [0.271]
WEAK EXOGENEITY TESTS
LY
Loading
p-value
(0C)
0.009
0.038
TRMV
-0.006
0.939
LBY
-0.072
0.001
***, ** and * indicate statisticalsignificance at 1 percent,5 percent,and 10 percent,respectively.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
SMV
-0.012
0.0004
PHILIPARESTIS, PANICOSO. DEMETRIADES,AND KUL B. LUINTEL : 37
TRMV.The real GDP effect is insignificant and stock marketvolatility is significantly negative.Weakexogeneity tests indicatecausalityfrom LBY to LY and no associationbetween LBY and TRMV.
Implicationsfor the Debclteon FinanciczlSystems
Even though our results vary across countries, they accord reasonablywell with
widely acceptedviews regardingthe comparativeability of varioustypes of financial
system to stimulate investment and growth. Specifically, our findings are broadly
consistent with the view thatbank-basedfinancialsystems are more capable of promoting long-rungrowththanAnglo-Saxon type systems because they are betterable
to addressagency problems and short-termism(for example, Stiglitz 1985; see also
Singh 1997). Specifically, both (i) the positive influence of the banking system on
real GDP in Germany,Japan,and Franceand (ii) the absence or weakness of a positive causal link from financialdevelopmentto real GDP in the United Kingdom and
the United States, accordwell with this view. Whatis also interestingin the cases of
GermanyandJapanis thatboth bankingsystem and stock marketdevelopmentseem
to have played a positive role in promotinglong-rungrowth, even though in quantitativetermsthe contributionof the stock marketwas substantiallysmaller.
4. SUMMARYAND CONCLUSIONS
Our empiricalanalysis shows that while stock marketsmay be able to contribute
to long-termoutputgrowth,their influence is, at best, a small fractionof that of the
banking system. Specifically,both stock marketsand banks seem to have made importantcontributionsto outputgrowth in France, Germanyand Japan,even though
the latter's contributionhas ranged from about one-seventh to aroundone-thirdof
the latter.Finally, the link between financialdevelopmentand growth in the United
Kingdom and the United States was found to be statisticallyweak and, if anything,
to run from growthto financialdevelopment.Thus, our findings are consistent with
the view thatbank-basedfinancialsystems may be more able to promote long-term
ones.
growththancapital-market-based
Our findings also suggest that stock marketvolatility had negative real effects in
Japanand France.In the case of the United Kingdom stockmarket volatility seems
to have exerted negative effects both on financialdevelopmentand output. Finally,
the effects of stock market volatility in Germany were found to be insignificant.
While in principlethe presenceof volatility in stock prices may reflectefficientfunctioning of stock markets,our findings do not supportthis hypothesis. Furthermore,
they are consistent with the findings of Aizenman and Marion (1996), who found
thatothermeasuresof volatility fiscal, monetaryand external also have negative
real effects. This, of course, may suggest that volatility of any kind reflects general
economic uncertaintyand is, therefore,negativelycorrelatedto real economic activity. Clearly furtherresearch is needed before more definitive conclusions can be
drawnon this issue, especially on the channelsthroughwhich stock marketvolatility
may affect economic activity.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
38
: MONEY,CREDIT,AND BANKING
The rich diversity in our results complements the findings obtained f1 omcrosscountrygrowth regressions concerning the relationshipbetween stock marketsand
growth.It also confirmsthe view that cross-countrygrowthregressionsat best only
providea broad-brushpictureof the relationshipbetween financialdevelopmentand
growth,which misses out many importantdetails.There aregood theoreticalreasons
why the relationshipbetween the financialsystem and growthmay vary substantially
across countries. This paper's findings suggest that these reasons are empirically
sound. Thus, the broad-brushconclusion that stocl marketdevelopmenthelps promote economic growthmust now be viewed with some caution.14
LITERATURECITED
Aizenman,Joshua,and Nancy P. Marion."Volatilityand the InvestmentResponse."National
Bureauof Economic Research,Workingpaper5841, November 1996.
Akaike, Hirotuku."InformationTheory and an Extensionof the MaximumLikelihoodPrinciple."In SecondInternationalSymposiumon InformationTheory,edited by B. PetrovandF.
Csaki, pp. 267-81. Budapest:AkademiaiKiado, 1973.
Akyuz, Yilmaz. "FinancialLiberalization:The Key Issues." In Finance and the Real Economy, edited by Yilmaz Akyuz and GuntherHeld. Santiago: United Nations University,
1993.
Arestis, Philip, and Panicos O. Demetriades."Financeand Growth:InstitutionalConsiderations and Causality."Paperpresentedat the Royal Economic Society Annual Conference,
Universityof Wales Swansea, 1-4 April, 1996.
. "FinancialDevelopmentand Economic Growth:Assessing the Evidence."Economic
Journal 107 (May 1997), 783-99.
Atje, Raymond, and Boyan Jovanovic. "Stocks Marketsand Development."EuropeanEconomic Review37 (April 1993), 632-40.
Baillie, RichardT., and Tim Bollerslev. "Cointegration,Fractional Cointegration,and Exchange Rate Dynamics."Journal of Finance 49 (June 1994), 737-45.
Bencivenga, Valerie R., Bruce D. Smith, and Ross M. Starr."EquityMarkets,Transactions
Costs, and CapitalAccumulation:An Illustration."WorldBank EconomicReview 10 (May
1996), 241-65.
Bertero, Elisabetta. "The Banking System, Financial Markets, and Capital Structure:Some
New Evidence from France."OxfordReviewof EconomicPolicy 10 (Winter1994), 68-78.
Boyd, JohnH., and Bruce D. Smith. "TheCoevolutionof the Real andFinancialSectorsin the
GrowthProcess."WorldBank EconomicReview 10 (May 1996), 371-96.
Caselli, F., G. Esquivel, and F. Lefort. "Reopeningthe ConvergenceDebate:A New Look at
Cross-CountryGrowthEmpirics."Journal of Economic Growth1 (1996), 363-90.
14. One of the referees suggests that the comparisonof our results with those of cross-sectionalstudies is somewhatunfair,in thatour study is not really aboutthe long-runeffects of financialintermediation
on economic growth in the spirit of cross-section studies that include both developed and developing
countries.Whilst we accept the premise of the referee's argument,we would make two comments. The
first is that mixing developed and developing countriesmay not be so fruitfulin view of the institutional
differences between them. The second comment is that a comparisonbetween time series results and
those from cross-section studies is still useful per se, althoughthe qualificationby the anonymousreferee
should be borne in mind when attemptingsuch comparisons.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
PHILIPARESTIS, PANICOSO. DEMETRIADES,AND KUL B. LUINTEL : 39
Cheung, Yin-Wong, and Kon-S. Lai. "Finite Sample Sizes of Johansen's Likelihood Ratio
Tests for Cointegration."OxfordBulletin of Economics and Statistics 55 (August 1993),
313-28.
Corbett,Jenny,andTim Jenkinson."TheFinancingof Industry,1970-1989: An International
Comparison."Discussion Paper no. 948, London: Centre for Economic Policy Research,
1994.
Crowder,WilliamJ., and Dennis L. Hoffman."TheLong-RunRelationshipbetween Nominal
InterestandInflation:The FisherEquationRevisited."Journalof Money,Credit,and Banking 28 (February1996), 103-18.
Davis, E. Philip. Pension Funds, Retirement-IncomeSecurity,and CapitalMarkets:An International Perspective.Oxford:ClarendonPress, 1995.
DeLong, J.Bradford,AndreiSchleifer,LawrenceH. Summers,and RobertJ. Waldmann."The
Size and Incidence of the Losses from Noise Trading."Journal of Finance 44 (September
1989), 681-96.
Demetriades, Panicos O., and Khaled Hussein. "Does Financial Development Cause Economic Growth?"Journalof DevelopmentEconomics 51 (December 1996), 387-411.
Demetriades,Panicos O., and Kul B. Luintel."FinancialDevelopment,Economic Growthand
Banking Sector Controls: Evidence from India."Economic Journal 106 (March 1996),
359-74.
Demirguoc-Kunt,
Aslie, and Ross Levine. "StockMarkets,CorporateFinance, and Economic
Growth:An Overview."WorldBankEconomicReview 10 (May 1996), 223-39.
Demirguoc-Kunt,
Aslie, andVojislavMaksimovic. "StockMarketDevelopmentand Financing
Choices of Firms."WorldBankEconomicReview 10 (May 1996), 341-69.
Diamond, Douglas W. "FinancialIntermediationand Delegated Monitoring."Review of Economic Studies 51 (July 1984), 393-414.
Diamond,Douglas W., and R. E. Verrecchia."OptimalManagerialContractsand Equilibrium
SecurityPrices."Journal of Finance 37 (May 1982), 275-87.
Diebold, Francis X., Javier Gardeazabal,and Kamil Yilmaz. "Cointegrationand Exchange
Rate Dynamics."Journal of Finance 49 (June 1994), 727-35.
Edwards, Jeremy, and Klaus Fischer. Banks, Finance, and Investment in Germany, Cambridge:CambridgeUniversityPress, 1994.
Evans, Paul. "How to EstimateGrowthEquationsConsistently."Mimeo, Ohio State University; presentedat the 7th WorldCongressof the EconometricSociety, Tokyo, 1995.
. "GovernmentConsumptionandGrowth."EconomicInquiry35 (April 1997), 209-17.
Federer,J. Peter."TheImpactof Uncertaintyon AggregateInvestmentSpending."Journal of
Money,Credit,and Banking25 (February1993), 30-45.
Fry,Maxwell J. "InDefence of FinancialLiberalisation."EconomicJournal 107 (May 1997),
754-70.
Gonzalo, Jesus. "Five AlternativeMethods of Estimating Long-Run EquilibriumRelationships."Journalof Econometrics60 (January-February
1994), 203-33.
Greenwood,Jeremy,and Boyan Jovanovic."FinancialDevelopment,Growth,and the Distributionof Income."Journal of Political Economy98 (October 1990), 1076-1107.
Greenwood,Jeremy,and Bruce Smith. "FinancialMarketsin Developmentand the Development of Financial Markets."Journal of Economic Dynamics and Control 21 (January
1997), 145-82.
Hall, StephenG. "MaximumLikelihoodEstimationof CointegratingVectors:An Example of
Johansen's Procedure."Oxford Bulletin of Economics and Statistics 51 (March 1989),
213-18.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
40
: MONEY,CREDIT,AND BANKING
Hall, StephenG., andAlistairMilne. "TheRelevanceof P-StarAnalysis to UK MonetaryPolicy."EconomicJournal 104 (May 1994), 597-604.
Hall, Stephen,andMike Wickens. "Causalityin IntegratedSystems."LondonBusiness School
Discussion Papers 27-93 (November 1993).
Hansen, Henrik, and Soren Johansen."RecursiveEstimationin CointegratedVAR Models."
Preprint1, Instituteof MathematicalStatistics, Universityof Copenhagen,1993.
. "SomeTests of ParameterConstancyin CointegratedVARModels."Instituteof Mathematical Statistics, Universityof Copenhagen,1998.
Hansen, Henrik, and KatarinaJuselius. Manual of CointegrationAnalysis of Time Series
CATSin RATS.Instituteof Economics:Universityof Copenhagen,1994.
Hargreaves,Colin. "A Review of Methods of Estimating CointegratingRelationships."In
Nonstationary Time Series Analysis and Cointegration,edited by Colin Hargreaves,pp.
87- 131. New York:OxfordUniversityPress, 1994.
Harris, Richard D.F. "Stock Markets and Development: A Reassessment."European Economic Review41 (1997), 139-46.
Haug,AlfredA. "Testsfor Cointegration:A MontecarloComparision."Journal of Econometrics 71 (March1996), 89-1 15.
Horiuchi,Akiyoshi, and Ryoko Okazaki."CapitalMarketsand the BankingSector:Efficiency
of JapaneseBanks in ReducingAgency Costs."In Japan, Europeand InternationalFinancial Markets:Analytical and Empirical Perspectives, edited by Ryuzo Sato, RichardM.
Levich, and Rama V. Ramachandran,pp. 197-219. Cambridge: Cambridge University
Press, 1994.
Jensen, Michael C., and Kevin J. Murphy."PerformancePay and Top ManagementIncentives."Journal of Political Economy98 (April 1990), 225-64.
Johansen,Soren. "StatisticalAnalysis of Co-IntegratingVectors."Journal of Economic Dynamics and Control12 (June-September1988), 231-254.
. "Estimationand HypothesisTesting of CointegratingVectorsin GaussianVectorAutoregressiveModels."Econometrica59 (November 1991), 1551-80.
. "Determinationof CointergationRankin the Presenceof a LinearTrend."OxfordBulletin of Economicsand Statistics 54 (August 1992), 383-97.
Johansen,Soren, and KatarinaJuselius. "Some StructuralHypotheses in a MultivariateCointegrationAnalysis of PurchasingPower Parityand Uncovered InterestParityfor the UK."
Journalof Econometrics53 (July-September1992), 211-44.
Jones, CharlesI. "Time Series Tests of Endogenous GrowthModels." QuarterlyJournal of
Economics 110 (May 1995), 495-525.
Klenow,PeterJ., andAndresRodriguez-Clare."EconomicGrowth:A Review Essay."Journal
of MonetaryEconomics40 (December 1997), 597-618.
Kocherlakota,NarayanaR., and Kei-MuYi. "Is There EndogenousLong-Run Growth?Evidence from the United States and the United Kingdom."Journal of Money, Credit, and
Banking29 (May 1997), 235-60.
Keynes, John Maynard.The General Theoryof Employment,Interest, and Money. London:
Macmillan, 1936.
Levine, Ross. "Stock Markets,Growth, and Tax Policy."Journal of Finance 46 (September
1991),
1445-65.
. "FinancialDevelopmentand Economic Growth:Views andAgenda."Journal of Economic Literature35 (1997), 688-726.
Levine, Ross, and David Renelt. "A Sensitivity Analysis of Cross-CountryGrowthRegressions."AmericanEconomicReview82 (September1992), 942-63.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions
PHILIPARESTIS, PANICOSO. DEMETRIADES,AND KUL B. LUINTEL : 41
Levine, Ross, and Sara Zervos. "CapitalControl Liberalizationand Stock MarketDevelopment."Policy ResearchWorkingPaper 1622, WashingtonD.C.: WorldBank, 1995.
. "Stock Markets, Banks, and Economic Growth."American Economic Review 88
(June 1998), 537-58.
Luintel,Kul B., andMosahidKhan."AQuantitativeReassessmentof Finance-GrowthNexus:
Evidence from Multivariate VAR." Journal of Development Economics 60 (1999),
381-405.
Luintel, Kul B., and KrishnaPaudyal. "CommonStochastic Trends between Forwardand
Spot ExchangeRates."Journal of InternationalMoneyand Finance 17 (1998), 279-97.
Mayer, Colin. "New Issues in CorporateFinance." European Economic Review 32 (June
1988), 1167-88.
Osterwald-Lenum,Michael. "ANote with Quartilesof the AsymptoticDistributionof the ML
CointegrationRank Test Statistics."OxfordBulletin of Economics and Statistics 54 (August 1992), 461-72.
Pesaran,M. Hashem, and Yongcheol Shin. "Long-RunStructuralModelling."University of
Cambridge:Departmentof Applied lEconomics,WorkingPaperno. 9419, September1994.
Pesaran,M. Hashem,and R. Smith. "EstimatingLong-RunRelationshipsfrom Dynamic HeterogeneousPanels."Journalof Econometrics68 (1995), 79-113.
Quah, Danny. "EmpiricalCross-Section Dynamics in Economic Growth."European Economic Review37 (April 1993), 426-34.
QuintosE. Carmela."Sustainabilityof the Deficit Process with StructuralShifts."Journal of
Business and EconomicStatistics 13 (1995), 409 - 16.
Rajan,RaghuramG., and Luigi Zingales. "WhatDo We Know aboutCapitalStructure?Some
Evidence FromInternationalData."Journal of Finance 50 (1995), 1421-60.
Rajan,RaghuramG., and Luigi Zingales. "FinancialDependence and Growth."Mimeo, University of Chicago, 1996.
Schwarz, Gideon. "Estimatingthe Dimension of a Model." Annals of Statistics 6 (1978),
461-64.
Shiller,RobertJ. "Do Stock Prices Move Too Much to Be Justifiedby SubsequentChangesin
Dividends?"AmericanEconomicReview71 (1981), 421-36.
. MarketVolatility.Cambridge,Mass.: MIT Press, 1989.
Singh,Ajit. "StockMarkets,FinancialLiberalisationandEconomic Development."Economic
Journal 107 (May 1997), 771-82.
Stiglitz, JosephE. "CreditMarketsand the Controlof Capital."Journalof Money,Credit,and
Banking 17 (February1985), 133-52.
Toda,HiroY., and Peter C.B. Phillips. "VectorAutoregressionsand Causality."Econometrica
61 (November 1993), 1367-93.
Williamson, Stephen D. "Costly Monitoring, Financial Intermediation, and Equilibrium
CreditRationing."Journalof MonetaryEconomics 18 (September1986), 159-79.
This content downloaded from 129.20.73.3 on Wed, 16 Jul 2014 01:16:39 AM
All use subject to JSTOR Terms and Conditions