1. R ev ista M ex ica n a d e E co n o m ¶ y F in a n za s, V o l. 5 , N o . 1 (2 0 0 6 ), p p . 1 3 -2 7
³a 13
F IN A N C IA L S Y S T E M S A N D B A N K IN G
C R IS E S : A N A S S E S S M E N T
A n to n io R u iz -P o r ra s ¤
A cco u n tin g a n d F in a n ce D ep a rtm en t
T ecn o l¶ g ico d e M o n terrey, C a m p u s C iu d a d d e M ¶ x ico
o e
(R eceived 5 J a n u a ry 2 0 0 6 , a ccep ted 1 4 M a rch 2 0 0 6 )
A b str a c t
T ra d itio n a lly a n o ld co n cern a m o n g eco n o m ists h a s referred to th e e® ects th a t sp eci¯ c ¯ n a n cia l
sy stem s m a y h a v e o n eco n o m ic p erfo rm a n ce. H ere w e in v estig a te th e sty lised fa cts" a m o n g
¯ n a n cia l sy stem s a n d b a n k in g crises b y u sin g in d iv id u a l a n d p rin cip a l-co m p o n en ts in d ica to rs
a n d sets o f O L S reg ressio n s. T h e stu d y relies o n a set o f b a n k in g fra g ility, ¯ n a n cia l stru ctu re
a n d d ev elo p m en t in d ica to rs fo r a sa m p le o f 4 7 eco n o m ies b etw een 1 9 9 0 a n d 1 9 9 7 . T h e sty lised
fa cts su g g est th a t ¯ n a n cia l d ev elo p m en t is a sso cia ted to ¯ n a n cia l sy stem s lea d ed b y sto ck
a n d secu rities m a rk ets. F u rth erm o re th e ev id en ce su g g ests th a t su ch a sso cia tio n is m a g n i¯ ed
d u rin g ep iso d es o f b o rd erlin e o r sy stem ic b a n k in g crises. T h u s w h a t o u r ¯ n d in g s m ig h t
su g g est is th a t b a n k in g crises m a y en co u ra g e ¯ n a n cia l d ev elo p m en t a n d th e tra n sfo rm a tio n
o f ¯ n a n cia l sy stem s in to m a rk et-b a sed o n es.
R e su m e n
T ra d icio n a lm en te u n a v ieja p reo cu p a ci¶ n en tre lo s eco n o m ista s se re¯ ere a lo s efecto s q u e
o
lo s sistem a s ¯ n a n ciero s p u d iera n ten er en el d esem p e n o eco n ¶ m ico . A q u¶ in v estig a m o s lo s
~ o ³
" h ech o s estiliza d o s" en tre lo s sistem a s ¯ n a n ciero s y la s crisis b a n ca ria s m ed ia n te el u so d e
in d ica d o res sim p les y d e co m p o n en tes-p rin cip a les, y m ed ia n te el u so d e reg resio n es d e M C O .
E l estu d io se fu n d a m en ta en u n co n ju n to d e in d ica d o res d e fra g ilid a d b a n ca ria , estru ctu ra
¯ n a n ciera y d esa rro llo ¯ n a n ciero p a ra u n a m u estra d e 4 7 eco n o m ¶ s d u ra n te el p erio d o 1 9 9 0 -
³a
1 9 9 7 . L o s h ech o s estiliza d o s su g ieren q u e el d esa rro llo ¯ n a n ciero se a so cia a a q u ello s sistem a s
¯ n a n ciero s en d o n d e p red o m in a n lo s m erca d o s d e v a lo res y d e a ctiv o s. A d em ¶ s, la ev id en cia
a
su g iere q u e d ich a v in cu la ci¶ n es m a g n i¯ ca d a d u ra n te ep iso d io s d e crisis b a n ca ria s sist¶ m ica s y
o e
n o sist¶ m ica s. E n co n secu en cia , n u estro s h a lla zg o s su g ieren q u e la s crisis b a n ca ria s p u d iera n
e
fo m en ta r el d esa rro llo ¯ n a n ciero y la tra n sfo rm a ci¶ n d e lo s sistem a s ¯ n a n ciero s en sistem a s
o
b a sa d o s en lo s m erca d o s.
J E L c la ssi¯ c a tio n : G 1 , F 4 , C 2
K ey w o rd s: F in a n cia l sy stem s, b a n k in g crises, ¯ n a n cia l stru ctu re, ¯ n a n cia l d ev elo p m en t
I a m g ra tefu l to S p iro s B o u g ea s, R ich a rd D isn ey, A la n S tew a rt D u n ca n , P h ilip M o ly -
n eu x , J o se A n to n io N u n ez-M o ra , a n d H u m b erto V a len cia -H errera fo r v a lu a b le su g g estio n s o n
~
ea rlier d ra fts. H o w ev er, th e u su a l d iscla im er a p p lies.
¤
A cco u n tin g a n d F in a n ce D ep a rtm en t. T ecn o l¶ g ico d e M o n terrey, C a m p u s C iu d a d d e
o
M ¶ x ico . C a lle d el P u en te 2 2 2 , C o l. E jid o s d e H u ip u lco , C . P . 1 4 3 8 0 , T la lp a n , M ¶ x ico , D .F .
e e
T elep h o n e: + 5 2 (5 5 ) 5 4 8 3 -2 2 3 8 . E -m a il: ru iz.a n to n io @ itesm .m x

2. 14 A . R u iz-P o rra s / F in a n cia l S y stem s a n d B a n k in g C rises: A n a ssessm en t
1 . In tr o d u c tio n
Traditionally an old concern among economists has referred to the e®ects that ¯-
nancial systems may have on economic performance. 1 This concern has encour-
aged the development of theories and empirical research to assess the relative
merits of di®erent ¯nancial systems for policy purposes. 2 Here we investigate
the stylised facts" among ¯nancial systems and banking crises by using in-
ternational evidence. The study relies on a set of banking fragility, ¯nancial
structure and development indicators for a sample of 47 economies between
1990 and 1997.
The investigation is motivated by the idea that banking crises may carry
repercussions of social nature in addition to the ones of private one. 3 Financial
and non ¯nancial ¯rms and even the security of the payment system may be
a®ected by banking crises [Freixas and Rochet (1997) , Goodhart et a l. (1998) ] .
Particularly we focus on the relationship among ¯nancial systems and banking
crises because such empirical studies are scarce in spite of the recognition that
they may be useful to improve our understanding of the likelihood of banking
fragility [Demirguc-Kunt and Detragiache (1998) ] . 4
The study of the stylised facts may be interesting for theoretical purposes.
According to some economists, di®erent ¯nancial systems provide di®erent in-
centives and opportunities to share risks and to encourage speci¯c performance
goals due to the existence of competition among ¯nancial markets and banks
[Allen and Gale (2000) , (2004) ] . However, with exception of the studies of
Levine (2002) , and Lopez and Spiegel (2002) , we do not know about other
empirical assessments related to such claim. Thus the characterisation of the
stylised facts may provide evidence to develop and support theoretical claims.
Methodologically, we characterise the stylised facts with assortments of
indicators. 5 Categorical banking fragility indicators refer to episodes of systemic
1
S u ch co n cern s ca n b e tra ced b a ck to B a g eh o t (1 8 7 3 ) a n d F ish er (1 9 3 3 ). A cco rd in g
to th e fo rm er, G erm a n y h a d o v erco m e th e U n ited K in g d o m a s a n in d u stria l p o w er in th e
n in eteen cen tu ry d u e to th e rela tiv e su p erio rity o f th e G erm a n b a n k -b a sed ¯ n a n cia l sy stem
w ith resp ect to th e m a rk et-b a sed B ritish o n e. A cco rd in g to th e la tter, th e sev erity o f th e
A m erica n G rea t D ep ressio n w a s m a in ly a resu lt o f p o o rly p erfo rm in g ¯ n a n cia l m a rk ets.
2
S ee G etler (1 9 8 8 ), S a n to m ero (1 9 8 9 ), L ev in e (2 0 0 2 ), B eck (2 0 0 3 ), A llen a n d G a le (2 0 0 0 )
a n d (2 0 0 4 ) fo r so m e su rv ey s a n d rev iew s o n th e th eo ries reg a rd in g ¯ n a n cia l stru ctu re a n d
eco n o m ic p erfo rm a n ce. C o m p a ra tiv e stu d ies o f th e ex istin g ¯ n a n cia l sy stem s a lo n g th e w o rld
a re G o ld sm ith (1 9 6 9 ), F ra n k el a n d M o n tg o m ery (1 9 9 1 ), D em irg u c-K u n t a n d L ev in e (1 9 9 9 ),
a n d A llen a n d G a le (2 0 0 0 ).
3
C a p rio a n d K lin g eb iel (1 9 9 6 a ) a n d (1 9 9 6 b ) sh o w th a t th e co sts th a t th e eco n o m ies p a y
d u e to b a n k in g in so lv en cy ep iso d es u su a lly a re a b o v e 1 5 % o f th eir G D P . M o reo v er, th ese co sts
d o n o t in clu d e th e o n es a sso cia ted to th e ex ch a n g e-ra te crises a n d eco n o m ic d o w n tu rn s th a t
u su a lly a cco m p a n y su ch ep iso d es.
4
E x istin g stu d ies a re m a in ly d escrip tiv e a n d rela te to sp eci¯ c ca ses o f ¯ n a n cia l fra g ility,
lik e th e A sia n C risis a n d th e U S ex p erien ce in th e eig h ties a n d n in eties [S ee A llen (2 0 0 1 ) a n d
H o en ig (2 0 0 1 ), resp ectiv ely ]. In a n in tern a tio n a l co n tex t stu d ies th a t in clu d e so m e ¯ n a n cia l
stru ctu re d eterm in a n ts a re th e o n es o f D em irg u c-K u n t a n d H u izin g a (2 0 0 0 ) o n b a n k in g p ro f-
ita b ility a n d th e o n e o f B eck , D em irg u c-K u n t, a n d L ev in e (2 0 0 3 ) o n b a n k in g co n cen tra tio n
a n d crises.
5
T h e a b sen ce o f a ccep ted em p irica l d e¯ n itio n s fo r ¯ n a n cia l stru ctu re a n d ¯ n a n cia l d e-

3. R ev ista M ex ica n a d e E co n o m ¶ y F in a n za s, V o l. 5 , N o . 1 (2 0 0 6 ), p p . 1 3 -2 7
³a 15
and borderline banking crises compiled by Caprio and Klingebiel (2002) . Data
extracted from the database of Beck, Demirguc-Kunt and Levine (2000) , are
used to build ¯nancial structure and development assortments of indicators
according to the guidelines of Levine (2002) . The assortments include measures
of activity, size and e±ciency of the intermediaries. Individual and principal-
component indicators are used for the empirical assessments.
Our research aims to identify patterns or " stylised facts" that may shed
light on the normative properties of di®erent ¯nancial systems. We do this by
comparing relationships between indicators under di®erent banking conditions.
The idea underlying the identi¯cation of such patterns is to suggest answers
to some of the following questions: What are the main empirical relationships
among ¯nancial systems and banking crises? How does banking fragility a®ect
the relationships between ¯nancial structure and ¯nancial development? Which
type of implications may be derived from these ¯ndings?
Our ¯ndings have implications for academic and policy purposes. Specif-
ically the stylised facts suggest that ¯nancial development is associated to ¯-
nancial systems leaded by stock and securities markets. Furthermore the evi-
dence suggests that such association is magni¯ed during episodes of borderline
or systemic banking crises. These ¯ndings are consistent with the idea that
crises have had a signi¯cant impact on the historical development of ¯nancial
systems. Thus what our ¯ndings might suggest is that banking crises may en-
courage ¯nancial development and the transformation of ¯nancial systems into
market-based ones.
The paper is divided in ¯ve sections. Section 2 describes the data used in
the econometric analyses. Section 3 discusses methodological issues to assess
the limits and scope of our ¯ndings. Section 4 characterises the stylised facts
among ¯nancial structure and development with banking fragility. Section 5
summarises and discusses the main ¯ndings. Finally, the appendix focuses on
the principal-component methodology.
2 . F in a n c ia l a n d b a n k in g in d ic a to r s
Here we describe the ¯nancial and banking indicators used in the analysis.
We believe this task particularly relevant because of the absence of empirical
de¯nitions for ¯nancial system features and banking fragility. Thus, before
proceeding, we de¯ne certain de¯nitions for operative purposes. Speci¯cally, in
the following we will refer to ¯nancial development as the level of development
of both intermediaries and markets, while by ¯nancial structure we will mean
the degree to which a ¯nancial system is based on intermediaries or markets.
Banking fragility will mean a situation in which borderline or systemic banking
crises are present in an economy.
We build the indicators by extracting ¯nancial and banking data from two
databases. We use panel-data extracted from the cross-country database on
¯nancial development and structure [Beck, Demirguc-Kunt and Levine (2000) ] ,
to build the ¯nancial system indicators. Furthermore, we use data from the
v elo p m en t m a k es n ecessa ry to u se a sso rtm en ts o f in d ica to rs in th e a n a ly sis. S u ch a p p ro a ch
h a s b een u sed b y L ev in e (2 0 0 2 ) a n d B eck (2 0 0 3 ) to a n a ly se th e d eterm in a n ts o f lo n g -ru n
eco n o m ic p erfo rm a n ce.

4. 16 A . R u iz-P o rra s / F in a n cia l S y stem s a n d B a n k in g C rises: A n a ssessm en t
database on episodes of borderline and systemic banking crises [Caprio and
Klingebiel (2002) ] , to build qualitative indicators of banking fragility. The main
advantage of using these databases is that they allow us to treat consistently
the ¯nancial and banking system data across economies and across time. 6
The ¯nancial and banking data and their main features are summarised in
the table:
Table 1. Financial and Banking Data.
D e¯ n itio n V a ria b le T im e sp a n C o u n tries O b serv a tio n s
B a n kin g fra gility va ria bles
D u m m y v a ria b le o n sy stem ic
ep iso d es o f b a n k in g fra g ility yS 1 9 7 5 -1 9 9 9 93 113
(crisis= 1 , n o n crisis= 0 )
D u m m y v a ria b le o n b o rd erlin e
ep iso d es o f b a n k in g fra g ility yB 1 9 7 5 -1 9 9 9 44 50
(crisis= 1 , n o n crisis= 0 )
F in a n cia l stru ctu re a n d d evelo p m en t va ria bles
O v erh ea d co sts o f th e b a n k in g
sy stem rela tiv e to b a n k in g B O H C 1 9 9 0 -1 9 9 7 129 719
sy stem a ssets
P riv a te cred it b y d ep o sit m o n ey
b a n k s to G D P D B P C Y 1 9 6 0 -1 9 9 7 160 3901
(B a n k cred it ra tio )
P riv a te cred it b y d ep o sit m o n ey
b a n k s a n d o th er ¯ n a n cia l T IP C Y 1 9 6 0 -1 9 9 7 161 3923
in stitu tio n s to G D P
(P riv a te cred it ra tio )
S to ck m a rk et ca p ita lisa tio n
to G D P SM C Y 1 9 7 6 -1 9 9 7 93 1171
(M a rk et ca p ita lisa tio n ra tio )
S to ck m a rk et to ta l v a lu e tra d ed
to G D P SM V Y 1 9 7 5 -1 9 9 7 93 1264
(T o ta l v a lu e tra d ed ra tio )
N o tes: T h e co m p lete ¯ n a n cia l d ev elo p m en t a n d stru ctu re d a ta b a se in clu d es sta tistics o n
th e size, a ctiv ity a n d e± cien cy o f v a rio u s in term ed ia ries (co m m ercia l b a n k s, in su ra n ce
co m p a n ies, p en sio n fu n d s a n d n o n -d ep o sit m o n ey b a n k s) a n d m a rk ets (p rim a ry eq u ity
a n d p rim a ry a n d seco n d a ry b o n d m a rk ets). R eg a rd in g th e d a ta b a se o n b a n k in g crises,
it co m p rises o f th e tw o v a ria b les in clu d ed h ere.
The sample was built according to data availability. It includes data for Ar-
gentina, Australia, Bolivia, Brazil, Barbados, Canada, Switzerland, Chile, Co-
lombia, Costa Rica, Cyprus, Germany, Ecuador, Egypt, Spain, Ghana, Greece,
6
T h e d a ta b a ses a n d d eta ils o n th eir co n stru ctio n a re a v a ila b le a t th e w eb site o f th e W o rld
B a n k . T h e a d d ress is th e fo llo w in g : h ttp :/ / eco n .w o rld b a n k .o rg [T itles: A n ew d a ta b a se
o n ¯ n a n cia l d ev elo p m en t a n d stru ctu re" a n d E p iso d es o f sy stem ic a n d b o rd erlin e ¯ n a n cia l
crises" ].

5. R ev ista M ex ica n a d e E co n o m ¶ y F in a n za s, V o l. 5 , N o . 1 (2 0 0 6 ), p p . 1 3 -2 7
³a 17
Guatemala, Honduras, Ireland, Jamaica, Jordan, Japan, Kenya, Korea, Mo-
rocco, Mexico, Malaysia, Namibia, Nigeria, Netherlands, Norway, New Zealand,
Peru, Philippines, Paraguay, El Salvador, Sweden, Thailand, Trinidad and To-
bago, Tunisia, Turkey, Taiwan, United States, Venezuela, South Africa and
Zimbabwe. Hence, the sample comprises data for 47 economies and 115 bank-
ing crises over the period 1990-97.
We de¯ne seven individual indicators. We follow Demirguc-Kunt and
Levine (1999) and Levine (2002) to build two assortments of indicators to anal-
yse the features of ¯nancial systems. The assortment of structural indicators
contains individual measures of the activity, size and e±ciency of stock markets
relative to that of banks. The assortment of development indicators contains
measures of the activity, size and e±ciency of stock markets and banks. The
banking fragility indicator is a qualitative variable for borderline and systemic
crises that follows the standard convention of the fragility literature [Demirguc-
Kunt and Detragiache (1998) , and Hardy and Pazarbasioglu (1999) ] .
The six ¯nancial system indicators use di®erent measures to assess the
structure and the degree of ¯nancial development. The structural assortment
is integrated by the Structure-Activity, Structure-Size and Structure-E±ciency
indicators. In this assortment market-based ¯nancial systems are associated
to large values of the indicators while bank-based ones are associated to small
values. The ¯nancial development assortment is integrated by the Finance-
Activity, Finance-Size and Finance-E±ciency indicators. In this assortment,
¯nancial development is associated to large values of the indicators while un-
derdevelopment is associated to small values. 7
Furthermore we build two aggregate indicators to summarise the informa-
tion content of the assorted indicators. We follow the methodological approach
of Levine (2002) to de¯ne and construct them by using principal-component
multivariate methods. More speci¯cally, each aggregate indicator is de¯ned as
the ¯rst linear combination of the three individual indicators that integrate each
assortment. The way in which the aggregate indicators summarise the relevant
information is based on proportions of the total variance that are accounted
from the individual indicators. [See Appendix for further details]
We use the principal-components methodology to simplify the task to un-
derstand the explanatory multivariate data in terms of a smaller number of
uncorrelated variables. Intuitively, given that a ¯rst principal-component is
positively correlated to the each of the individual indicators, it can be inter-
preted as a measure of what is common to all the variables. Given the lack of
empirical de¯nitions for ¯nancial development and ¯nancial structure, we can
interpret the aggregate indicators as indexes of scale for the degree of develop-
ment and of the relative prominence of markets in the ¯nancial system.
7
A cco rd in g to D em irg u c-K u n t a n d L ev in e (1 9 9 9 ), la rg e a n d sm a ll v a lu es d ep en d o n th e
m ed ia n o f ea ch ty p e o f in d ica to r. H en ce, b y d e¯ n itio n , th is criterio n a ssu m es th a t h a lf o f th e
o b serv a tio n s b elo n g to a certa in ca teg o ry w h ile th e o th er h a lf to a n o th er. T h is criterio n , in
sp ite o f b ein g a rg u a b le, a llo w s u s to a v o id p o ten tia l ex trem e-v a lu e p ro b lem s.

6. 18 A . R u iz-P o rra s / F in a n cia l S y stem s a n d B a n k in g C rises: A n a ssessm en t
The set of ¯nancial and banking indicators is summarised in the following
table:
Table 2. Financial and Banking Indicators.
N am e D e¯ n itio n M ea su rem en t
B a n kin g fra gility in d ica to rs
B in a ry v a ria b le fo r fra g ility E p iso d es o f sy stem ic a n d / o r
C risis B a n k in g crisis = 1 b o d erlin e b a n k in g crises
N o n b a n k in g crisis = 0
F in a n cia l stru ctu re in d ica to rs
A ctiv ity o f sto ck m a rk ets
SM V Y
S tru ctu re A ctiv ity S T C A C T = ln D B P C Y rela tiv e to th a t o f b a n k s
S ize o f sto ck m a rk ets
SM C Y
S tru ctu re S ize S T C S I Z = ln D B P C Y rela tiv e to th a t o f b a n k s
E ± cen cy o f sto ck m a rk ets
S tru ctu re E ± cien cy S T C E F F = ln (S M V Y £ B O H C ) rela tiv e to th a t o f b a n k s
F irst p rin cip a l co m p o n en t o f S ca le In d ex o f ¯ n a n cia l
S tru ctu re A g g reg a te th e set o f in d iv id u a l ¯ n a n cia l stru ctu re
stru ctu re in d ica to rs
F in a n cia l d evelo p m en t in d ica to rs
A ctiv ity o f sto ck m a rk ets a n d
F in a n ce A ctiv ity F I N A C T = ln (S M V Y £ T I P C Y ) in term ed ia ries
S ize o f sto ck m a rk ets a n d
F in a n ce S ize F I N S I Z = ln (S M C Y £ T I P C Y ) in term ed ia ries
SM V Y
F in a n ce E ± cien cy F I N E F F = ln B O H C F in a n cia l secto r e± cien cy
F irst p rin cip a l co m p o n en t o f S ca le In d ex o f ¯ n a n cia l
F in a n ce A g g reg a te th e set o f in d iv id u a l ¯ n a n cia l d ev elo p m en t
d ev elo p m en t in d ica to rs
N o tes: L a rg e v a lu es o f th e ¯ n a n cia l stru ctu re in d ica to rs a re a sso cia ted to m a rk et-b a sed
¯ n a n cia l sy stem s; sm a ll o n es to b a n k -b a sed o n es. L a rg e v a lu es o f th e ¯ n a n cia l d ev elo p -
m en t in d ica to rs rela te to h ig h lev els o f ¯ n a n cia l d ev elo p m en t.
3 .M e th o d o lo g ic a l issu e s o n in d ic a to r s a n d e c o n o m e tr ic m e th o d o lo g y
Here we discuss some methodological issues. We regard this discussion as crucial
because it allows to asses the limits and scope of our ¯ndings. Such discussion
is necessary in spite that we use the most extensive data publicly available

7. R ev ista M ex ica n a d e E co n o m ¶ y F in a n za s, V o l. 5 , N o . 1 (2 0 0 6 ), p p . 1 3 -2 7
³a 19
and that the sample allows us to characterise most ¯nancial systems in the
world. Thus we will begin the discussion by examining the indicators used
in the empirical assessments. Latter, we will continue by explaining the OLS
regression approach used to assess the stylised facts among ¯nancial systems
and banking crises.
We begin by examining the ¯nancial structure and development indicators.
These indicators are useful to assess the stylised facts because the data used
to build them are consistent across economies and across time. However, the
indicators have certain limitations. The ¯rst one relates to the fact that the
information that indicators provide is relative to the sample. 8 Thus it would
not be surprising if any classi¯cation for an economy changes when the sample
changes. Furthermore, a second one is that the interpretation of the aggregate
principal-component indicators is somewhat subj ective.
The referred subj ectivity argument can be extended to include the banking
fragility indicator. The characterisation of banking crises periods is not as direct
as it seems [Caprio and Klingebiel (2002) ] . The time span of banking crises is
not easy to determine. Financial distress periods, where the banking system
has negative worth, can occur over a period of time, before and after being
detected. Also it is not always clear when a crisis is over. Thus, even at a
mere qualitative level, the characterisation of banking crises with a categorical
variable requires certain judgement.
However, in spite of the above limitations, we use the best and most exten-
sive ¯nancial and banking data publicly available. It allows us to quantitatively
characterise the main features of most ¯nancial systems by using econometric
techniques like OLS and panel-data ones. Particularly the possibility to use
di®erent econometric techniques to study the data suggest us a concrete ¯rst
approach to study such stylised facts. Such assessment approach is based on
comparisons among indicator relationships under di®erent situations of banking
performance. Here the chosen approach is based on OLS regression techniques.
Methodologically, the relationships among ¯nancial systems and banking
crises are assessed with four regression sets (One for the aggregate indicators
and the other three for the individual ones) . Each set is built by subsets of three
single-variable regressions that describe the associations between a speci¯c pair
of indicators under di®erent data samples. In each subset, the ¯rst regression
estimates an association using all the sampled data. The second and third
regressions re-estimate the same association using two data sub-samples that
are di®erentiated according to the banking fragility indicator.
Each regression set analyses one speci¯c relationship among the indicators.
The ¯rst set analyses the relationships among the ¯nancial development indi-
cators with respect to the Structure-Activity one. The second set analyses the
relationships with respect to the Structure-Size one. The third analyses the re-
lationships with respect to the Structure-E±ciency one. It may be argued that
the underlying assumption behind these regressions is that ¯nancial structure
8
F o r ex a m p le, ¯ n a n cia l stru ctu re in d ica to rs ca n in d ica te th a t certa in eco n o m ies m a y h a v e
b a n k -b a sed ¯ n a n cia l sy stem s b eca u se th eir sto ck m a rk ets a re v ery u n d erd ev elo p ed b y in tern a -
tio n a l sta n d a rd s. C o n v ersely, ¯ n a n cia l sy stem s o f eco n o m ies w ith sm a ll a n d u n d erd ev elo p ed
b a n k in g sy stem s m a y b e a ssu m ed a s m a rk et-b a sed o n es [D em irg u c-K u n t a n d L ev in e (1 9 9 9 )].

8. 20 A . R u iz-P o rra s / F in a n cia l S y stem s a n d B a n k in g C rises: A n a ssessm en t
causes ¯nancial development. However, this is not the case. The assessment of
stylised facts does not imply any causality.
4 . E c o n o m e tr ic a sse ssm e n t o f sty lise d fa c ts
Here we report the econometric results associated to the four regression sets used
to assess the stylised facts among ¯nancial systems and banking crises. First
we report the results associated to the three sets used to investigate the rela-
tionships among individual indicators. Later we report the results associated to
the fourth set of aggregate indicators. The regression subsets between pairs of
indicators are estimated with three data samples according to the econometric
procedure described above. In all the regressions we have included a constant
term to eliminate constant e®ects. 9
The ¯rst regression set analyses the relationship between ¯nancial devel-
opment and the relative activity of stock markets with respect to that of banks.
We summarise the results of the regression set of individual indicators in the
following table:
Table 3. Financial Systems and Banking Crises
(Financial Development and Structure-Activity Indicators) .
2 2 2
¯ R ¯ R ¯ R
(t) (t) (t)
R eg resso r A ll O b serv a tio n s S ta b le B a n k in g S y stem s F ra g ile B a n k in g S y stem s
In d ica to r (1 ) (2 ) (3 )
R egressed In d ica to r: F in a n ce-A ctivity
S tru ctu re 0.51 ¤ ¤ ¤ 0.69 0.47¤ ¤ ¤ 0.64 0.54 ¤ ¤ ¤ 0.72
A ctiv ity (2 7 .1 2 6 ) (1 9 .9 6 1 ) (1 6 .2 6 6 )
R egressed In d ica to r: F in a n ce-S ize
S tru ctu re 0.55 ¤ ¤ ¤ 0.35 0.44 ¤ ¤ ¤ 0.27 0.64 ¤ ¤ ¤ 0.43
A ctiv ity (1 2 .8 5 3 ) (8 .8 9 9 ) (8 .5 4 8 )
R egressed In d ica to r: F in a n ce-E ± cien cy
S tru ctu re 0.59 ¤ ¤ ¤ 0.73 0.52 ¤ ¤ ¤ 0.67 0.65 ¤ ¤ ¤ 0.77
A ctiv ity (2 7 .2 3 5 ) (1 9 .4 3 3 ) (1 7 .0 7 4 )
N o tes: T h e reg ressio n s u se O L S to estim a te eq u a tio n s o f th e fo rm : y = ® + ¯ x , w h ere
y a n d x a re th e reg ressed a n d reg resso r in d ica to rs, resp ectiv ely. T h e reg ressio n s u se
d i® eren t o b serv a tio n s fo r co m p a riso n p u rp o ses. S p eci¯ ca lly, th e ¯ rst co lu m n refers to
reg ressio n s th a t in clu d e a ll th e o b serv a tio n s. T h e seco n d co lu m n refers to reg ressio n s
th a t in clu d e o b serv a tio n s fo r w h ich th e b a n k in g fra g ility v a ria b le is eq u a l to zero . T h e
th ird co lu m n refers to th e o n es fo r w h ich th e b a n k in g fra g ility v a ria b le is eq u a l to o n e.
E a ch co lu m n co n ta in s th e estim a te o f ¯ , th e t-sta tistic o f th is estim a te (in p a ren th eses)
2
a n d th e R v a lu e o f th e reg ressio n . O n e, tw o a n d th ree a sterisk s in d ica te sig n i¯ ca n ce
lev els o f 1 0 , 5 a n d 1 p ercen t resp ectiv ely. T h e estim a ted co e± cien ts fo r co n sta n ts a re
n o t rep o rted .
9
T h e co e± cien ts a n d t-sta tistics a sso cia ted to th ese co n sta n t v a lu es a re n o t rep o rted in
th e reg ressio n resu lts in d ica ted b elo w . W e d o th is fo r sim p licity p u rp o ses a n d to fo cu s o n th e
rela tio n sh ip s a m o n g th e in d ica to rs.

9. R ev ista M ex ica n a d e E co n o m ¶ y F in a n za s, V o l. 5 , N o . 1 (2 0 0 6 ), p p . 1 3 -2 7
³a 21
Table 3 shows that ¯ n a n cia l d ev elo p m en t is a sso cia ted to a rela tiv e in crea se in
th e a ctiv ity o f sto ck m a rk ets w ith resp ect to th a t o f b a n k s. All the associa-
tions are positive and statistically signi¯cant (1 percent signi¯cance level) . The
consistency and robustness of these associations hold independently of bank-
ing stability considerations. Interestingly, the comparisons among data samples
suggest that su ch a sso cia tio n s a re m a g n i¯ ed d u rin g ep iso d es o f b o rd erlin e o r
sy stem ic b a n k in g crises. The regression coe±cients, ¯ , and coe±cients of de-
termination, R 2 , are higher for samples involving fragile banking systems.
The second regression set analyses the relationship between ¯nancial de-
velopment and the relative size of stock markets with respect to that of banks.
We summarise the results of the regression set of individual indicators in the
following table:
Table 4. Financial Systems and Banking Crises
(Financial Development and Structure-Size Indicators) .
2 2 2
¯ R ¯ R ¯ R
(t) (t) (t)
R eg resso r A ll O b serv a tio n s S ta b le B a n k in g S y stem s F ra g ile B a n k in g S y stem s
In d ica to r (1 ) (2 ) (3 )
R egressed In d ica to r: F in a n ce-A ctivity
S tru ctu re 0.10 ¤ ¤ ¤ 0.12 0.07¤ ¤ ¤ 0.06 0.14 ¤ ¤ ¤ 0.21
S ize (6 .5 9 1 ) (3 .7 4 1 ) (5 .0 6 8 )
R egressed In d ica to r: F in a n ce-S ize
S tru ctu re 0.18 ¤ ¤ ¤ 0.16 0.15 ¤ ¤ ¤ 0.12 0.21 ¤ ¤ ¤ 0.20
S ize (7 .6 2 8 ) (5 .3 5 5 ) (5 .0 4 2 )
R egressed In d ica to r: F in a n ce-E ± cien cy
S tru ctu re 0.14 ¤ ¤ ¤ 0.16 0.11 ¤ ¤ ¤ 0.10 0.18 ¤ ¤ ¤ 0.24
S ize (7 .1 6 5 ) (4 .6 4 2 ) (5 .1 3 4 )
N o tes: T h e reg ressio n s u se O L S to estim a te eq u a tio n s o f th e fo rm : y = ® + ¯ x , w h ere
y a n d x a re th e reg ressed a n d reg resso r in d ica to rs, resp ectiv ely. T h e reg ressio n s u se
d i® eren t o b serv a tio n s fo r co m p a riso n p u rp o ses. S p eci¯ ca lly, th e ¯ rst co lu m n refers to
reg ressio n s th a t in clu d e a ll th e o b serv a tio n s. T h e seco n d co lu m n refers to reg ressio n s
th a t in clu d e o b serv a tio n s fo r w h ich th e b a n k in g fra g ility v a ria b le is eq u a l to zero . T h e
th ird co lu m n refers to th e o n es fo r w h ich th e b a n k in g fra g ility v a ria b le is eq u a l to o n e.
E a ch co lu m n co n ta in s th e estim a te o f ¯ , th e t-sta tistic o f th is estim a te (in p a ren th eses)
2
a n d th e R v a lu e o f th e reg ressio n . O n e, tw o a n d th ree a sterisk s in d ica te sig n i¯ ca n ce
lev els o f 1 0 , 5 a n d 1 p ercen t resp ectiv ely. T h e estim a ted co e± cien ts fo r co n sta n ts a re
n o t rep o rted .
Table 4 shows that ¯ n a n cia l d ev elo p m en t is a sso cia ted to a rela tiv e in crea se
in th e size o f sto ck m a rk ets w ith resp ect to th a t o f b a n k s. Again the associa-
tions are positive and statistically signi¯cant. The consistency and robustness
of these associations hold independently of banking stability considerations.
Again the comparisons among data samples suggest that such associations are

10. 22 A . R u iz-P o rra s / F in a n cia l S y stem s a n d B a n k in g C rises: A n a ssessm en t
2
magni¯ed during banking crisis periods. The coe±cients ¯ and R are notably
higher for samples involving periods of banking crises.
The third regression set analyses the relationship between ¯nancial de-
velopment and the relative e±ciency of stock markets with respect to that of
banks. We summarise the results of the regression set of individual indicators
in the following table:
Table 5. Financial Systems and Banking Crises
(Financial Development and Structure-E±ciency Indicators) .
2 2 2
¯ R ¯ R ¯ R
(t) (t) (t)
R eg resso r A ll O b serv a tio n s S ta b le B a n k in g S y stem s F ra g ile B a n k in g S y stem s
In d ica to r (1 ) (2 ) (3 )
R egressed In d ica to r: F in a n ce-A ctivity
S tru ctu re 0.61 ¤ ¤ ¤ 0.77 0.57¤ ¤ ¤ 0.73 0.64 ¤ ¤ ¤ 0.81
E ± cien cy (3 1 .0 2 4 ) (2 2 .5 1 4 ) (1 9 .4 7 1 )
R egressed In d ica to r: F in a n ce-S ize
S tru ctu re 0.70 ¤ ¤ ¤ 0.48 0.60 ¤ ¤ ¤ 0.44 0.82 ¤ ¤ ¤ 0.55
E ± cien cy (1 5 .6 3 9 ) (1 1 .8 6 1 ) (1 0 .2 9 4 )
R egressed In d ica to r: F in a n ce-E ± cien cy
S tru ctu re 0.67¤ ¤ ¤ 0.72 0.61 ¤ ¤ ¤ 0.64 0.74 ¤ ¤ ¤ 0.80
E ± cien cy (2 7 .0 8 7 ) (1 8 .4 1 5 ) (1 9 .2 1 0 )
N o tes: T h e reg ressio n s u se O L S to estim a te eq u a tio n s o f th e fo rm : y = ® + ¯ x , w h ere
y a n d x a re th e reg ressed a n d reg resso r in d ica to rs, resp ectiv ely. T h e reg ressio n s u se
d i® eren t o b serv a tio n s fo r co m p a riso n p u rp o ses. S p eci¯ ca lly, th e ¯ rst co lu m n refers to
reg ressio n s th a t in clu d e a ll th e o b serv a tio n s. T h e seco n d co lu m n refers to reg ressio n s
th a t in clu d e o b serv a tio n s fo r w h ich th e b a n k in g fra g ility v a ria b le is eq u a l to zero . T h e
th ird co lu m n refers to th e o n es fo r w h ich th e b a n k in g fra g ility v a ria b le is eq u a l to o n e.
E a ch co lu m n co n ta in s th e estim a te o f ¯ , th e t-sta tistic o f th is estim a te (in p a ren th eses)
2
a n d th e R v a lu e o f th e reg ressio n . O n e, tw o a n d th ree a sterisk s in d ica te sig n i¯ ca n ce
lev els o f 1 0 , 5 a n d 1 p ercen t resp ectiv ely. T h e estim a ted co e± cien ts fo r co n sta n ts a re
n o t rep o rted .
Table 5 shows that ¯ n a n cia l d ev elo p m en t is a sso cia ted to a rela tiv e in crea se in
th e e± cien cy o f sto ck m a rk ets w ith resp ect to th a t o f b a n k s. Once again the
associations are positive and statistically signi¯cant. Once more the compar-
isons among data samples suggest that such associations are magni¯ed during
banking crisis periods.
The fourth regression set analyses the relationship between the ¯nancial
development and ¯nancial structure aggregate indexes. We summarise the re-
sults of the regression set of indicators in the following table:

11. R ev ista M ex ica n a d e E co n o m ¶ y F in a n za s, V o l. 5 , N o . 1 (2 0 0 6 ), p p . 1 3 -2 7
³a 23
Table 6. Relationships Between Financial and Banking Aggregate
Indicators (Regression Analysis) .
2 2 2
¯ R ¯ R ¯ R
(t) (t) (t)
R eg resso r A ll O b serv a tio n s S ta b le B a n k in g S y stem s F ra g ile B a n k in g S y stem s
In d ica to r (1 ) (2 ) (3 )
R egressed In d ica to r: F in a n ce-A ggrega te
S tru ctu re 0.69 ¤ ¤ ¤ 0.56 0.59 ¤ ¤ ¤ 0.51 0.81 ¤ ¤ ¤ 0.60
A g g reg a te (1 8 .3 4 5 ) (1 3 .5 2 3 ) (1 1 .2 5 9 )
N o tes: T h e reg ressio n s u se O L S to estim a te eq u a tio n s o f th e fo rm : y = ® + ¯ x , w h ere
y a n d x a re th e reg ressed a n d reg resso r in d ica to rs, resp ectiv ely. T h e reg ressio n s u se
d i® eren t o b serv a tio n s fo r co m p a riso n p u rp o ses. S p eci¯ ca lly, th e ¯ rst co lu m n refers to
reg ressio n s th a t in clu d e a ll th e o b serv a tio n s. T h e seco n d co lu m n refers to reg ressio n s
th a t in clu d e o b serv a tio n s fo r w h ich th e b a n k in g fra g ility v a ria b le is eq u a l to zero . T h e
th ird co lu m n refers to th e o n es fo r w h ich th e b a n k in g fra g ility v a ria b le is eq u a l to o n e.
E a ch co lu m n co n ta in s th e estim a te o f ¯ , th e t-sta tistic o f th is estim a te (in p a ren th eses)
2
a n d th e R v a lu e o f th e reg ressio n . O n e, tw o a n d th ree a sterisk s in d ica te sig n i¯ ca n ce
lev els o f 1 0 , 5 a n d 1 p ercen t resp ectiv ely. T h e estim a ted co e± cien ts fo r co n sta n ts a re
n o t rep o rted .
Table 6 con¯rms that ¯ n a n cia l d ev elo p m en t is a sso cia ted to m a rk et-b a sed ¯ -
n a n cia l sy stem s. Not surprisingly, the associations are positive and statistically
signi¯cant and their consistency and robustness hold independently of banking
performance. Once again, the coe±cients ¯ and R 2 are higher when banking
distress is present. Thus this regression set provides an overview of the stylised
facts among ¯nancial systems and banking crises.
We summarise our ¯ndings by indicating that the evidence suggests that
d ev elo p ed ¯ n a n cia l sy stem s a re lea d ed b y sto ck a n d o th er secu rities m a rk ets.
Moreover it also suggests that su ch a sso cia tio n is m a g n i¯ ed d u rin g ep iso d es o f
b o rd erlin e o r sy stem ic b a n k in g crises. These ¯ndings are consistent with the
idea that crises have had a signi¯cant impact on the historical development
of ¯nancial systems [Allen and Gale (2000) ] . Thus what our ¯ndings might
suggest is that b a n k in g crises m a y en co u ra g e ¯ n a n cia l d ev elo p m en t a n d th e
tra n sfo rm a tio n o f ¯ n a n cia l sy stem s in to m a rk et-b a sed o n es.
5 . C o n c lu sio n s a n d d isc u ssio n
This paper has shown the results of an investigation regarding the clari¯cation
of the stylised facts among ¯nancial systems and banking crises. The motivation
of such study relies on theoretical and practical concerns. The former relates
to how the ¯nancial contracting process and the functioning of intermediaries
and markets may depend on the ¯nancial structure prevailing in an economy.
While the latter relates to the consideration that banking crises may carry
repercussions of private and social nature. Overall, this investigation aims to
provide some elements regarding the discussions around the e®ects that ¯nancial
systems may have on economic performance.

12. 24 A . R u iz-P o rra s / F in a n cia l S y stem s a n d B a n k in g C rises: A n a ssessm en t
Research results may be summarised by indicating that the evidence sug-
gests that d ev elo p ed ¯ n a n cia l sy stem s a re lea d ed b y sto ck a n d secu rities m a r-
k ets. This result was obtained after estimating regressions between pairs of
individual and principal-components indicators. In all cases the conclusion
was supported by consistent and statistically signi¯cant regression coe±cients.
Furthermore other results suggested that su ch rela tio n sh ip is m a g n i¯ ed d u rin g
ep iso d es o f b a n k in g fra g ility . In all the regressions the coe±cients ¯ and R 2
were higher when banking distress was present.
These ¯ndings are consistent with the idea that crises have had a signi¯-
cant impact on the historical development of ¯nancial systems [Allen and Gale
(2000) ] . Speci¯cally, our ¯ndings suggest that b a n k in g crises m ig h t en co u ra g e
¯ n a n cia l d ev elo p m en t a n d th e tra n sfo rm a tio n o f ¯ n a n cia l sy stem s in to m a rk et-
b a sed o n es. We consider this conclusion interesting under the basis that the
question What drives the evolution of the ¯nancial system?" is one of the
main research issues in the literature of comparative ¯nancial systems [Allen
and Gale (2004: 701) ] .
Academically, what follows up from this study is that further research can
be carried along the lines of the literature of comparative ¯nancial systems
[Allen and Gale (2000) and (2004) ] . The relevance of such research is justi¯ed
under theoretical and practical purposes. Well designed ¯nancial and banking
systems are essential to guarantee the smooth allocation of resources within and
across economies. Further issues regarding globalization, ¯nancial fragility, risk
management and regulation practices may be analysed under the guidelines of
this literature. 1 0 We hope to encourage further research in such directions.
A p p e n d ix
The p rin cip a l co m p o n en ts a n a ly sis (PCA) is a method for re-expressing mul-
tivariate data. It allows to reorient the data so that the ¯rst few dimensions
account for much as information as possible. The central idea is based on the
concept of the proportion of the to ta l v a ria n ce (the sum of the variance of the
p original variables) that is accounted for by each of the new variables. PCA
transforms the set of correlated variables (x 1 ;:::;x p ) to a set of uncorrelated
variables (y 1 ;:::;y p ) called principal components in such a way that y 1 explains
the maximum possible of the total variance, y 2 the maximum possible of the
remaining variance, and so on.
The aim of PCA is to interpret the underlying structure of the data in
terms of the most important principal components. Usually, the ¯rst principal
component may be interpreted as a measure of what is common to the set of
correlated variables (x 1 ;:::;x p ) . Such interpretation relies on the fact that the
¯rst principal component is the best one-dimensional summary of the data.
Particularly, for the aims of the analysis developed here, the ¯rst principal
component may be interpreted as a sca le in d ex that summarises the information
contained on a particular set of variables.
Mathematically, the PCA problem is to determine the coe±cients a ij for
the following linear system:
10
S ee R u iz-P o rra s, V ¶ sq u ez-Q u ev ed o a n d N u n ez-M o ra (2 0 0 6 ) fo r a n ex a m p le o f su ch a n a l-
a ~
y sis in th e co n tex t o f th e M ex ica n ex p erien ce.

13. R ev ista M ex ica n a d e E co n o m ¶ y F in a n za s, V o l. 5 , N o . 1 (2 0 0 6 ), p p . 1 3 -2 7
³a 25
y 1 = a 1 1 x 1 + a 2 1 x 2 + ::: + a p 1 x p
y 2 = a 1 2 x 1 + a 2 2 x 2 + ::: + a p 2 x p
..
.
y p = a 1 p x 1 + a 2 p x 2 + ::: + a p p x p
where (x 1 ;:::;x p ) is the former set of correlated variables and (y 1 ;:::;y p ) is
the set of principal component variables.
Each principal component is de¯ned by the variables with which it is most
highly correlated. The ¯rst principal component, denoted by y 1 , is given by the
linear combination of the original variables X = (x 1 ;:::;x p ) with the largest
possible variance (where the variance is interpretable as the information con-
tained in the data) . The second principal component denoted by y 2 , is given
by the linear combination of X that accounts for the most information (highest
variance) not already captured by y 1 ; that is y 2 is chosen to be uncorrelated
with y 1 . All the subsequent principal components are chosen to be uncorrelated
with all previous principal components.
Because principal components analysis seeks to maximise variance, it can
be highly sensitive to scale di®erences across variables. Particularly, because
the data analysed here involve arbitrary units of measurement, our approach to
construct indexes based on principal component may be meaningless without
introducing further conditions. We standardise the data to avoid this problem.
Such data are denoted X S . This standardisation is achieved by introducing an
orthogonality condition for the coe±cients. Algebraically, such condition can
be written as:
Xp
a 2 = 1 ; (j = 1;2;:::;p )
ij
i= 1
Xp
a ij a ik = 0 ; (j 6 k ; j = 1;2;:::;p ; k = 1;2;:::;p )
=
i= 1
Analytically, the solution to the principal components problem stated above is
obtained by performing an eigenvalue decomposition of the correlation matrix
(i.e., the covariance matrix of the standardised data) . Thus, ¯nding the eigen-
value vector ¸ = (¸ 1 ;:::;¸ p ) and eigenvectors of the correlation matrix solves
the problem.
The variance of each principal component can be obtained by listing the
eigenvalues from the largest to the smallest ¸ 1 ¸ ¸ 2 ¸ ::: ¸ ¸ p ¸ 0. The
variance of ¯rst principal-component will be the eigenvalue ¸ 1 . The propor-
tion of total variance explained by the ¯rst principal-component will be then
¸1
¸ 1 + ¸ 2 + :::+ ¸ p . Bartholomew et. a l. (2002) , and Lattin, Carroll and Green (2003)
provide detailed explanations on the principal-components methodology.
R e fe r e n c e s
A llen , F . (2 0 0 1 ). F in a n cia l stru ctu re a n d ¯ n a n cia l crisis. In tern a tio n a l R eview o f F in a n ce,
2 (1 / 2 ), p p . 1 -1 9 .

14. 26 A . R u iz-P o rra s / F in a n cia l S y stem s a n d B a n k in g C rises: A n a ssessm en t
A llen , F . a n d D . G a le (2 0 0 4 ). C o m p a ra tiv e ¯ n a n cia l sy stem s: A d iscu ssio n " in S . B h a t-
ta ch a ry a , A . B o o t a n d A . T h a k o r, ed s., C red it In term ed ia tio n a n d th e M a croeco n o m y :
M od els a n d P erspectives, O x fo rd U n iv ersity P ress, O x fo rd , p p . 6 9 9 -7 7 0 .
A llen , F . a n d D . G a le (2 0 0 0 ). C o m pa rin g F in a n cia l S y stem s, M IT P ress, C a m b rid g e.
B a g eh o t, W . (1 8 7 3 ). L o m ba rd S treet: A D escrip tio n o f th e M o n ey M a rket, H en ry S . K in g
& C o ., L o n d o n .
B a rth o lo m ew , D .J ., F . S teele, I. M o u sta k i a n d J .I. G a lb ra ith (2 0 0 2 ). T h e A n a ly sis a n d
In terp reta tio n o f M u ltiva ria te D a ta fo r S ocia l S cien tists, C h a p m a n & H a ll/ C R C , B o ca
R a to n .
B eck , T . (2 0 0 3 ). S to ck m a rk ets, b a n k s a n d eco n o m ic d ev elo p m en t: T h eo ry a n d ev id en ce"
in B IS ,E u ro pe's C h a n gin g F in a n cia l L a n d sca pe: R ecen t D evelo p m en ts a n d P ro spects,
B IS p o licy p a p ers V o l. 8 , p p . 3 6 -5 4 . B a n k fo r In tern a tio n a l S ettlem en ts, B a sel.
B eck , T ., A . D em irg u c-K u n t a n d R . L ev in e (2 0 0 3 ). B a n k co n cen tra tio n a n d crises. N B E R
W o rkin g P a per 9 9 2 1 , N a tio n a l B u rea u o f E co n o m ic R esea rch , M a ssa ch u setts.
B eck , T ., A . D em irg u c-K u n t a n d R . L ev in e (2 0 0 0 ). A n ew d a ta b a se o n th e stru ctu re a n d
d ev elo p m en t o f th e ¯ n a n cia l secto r. T h e W o rld B a n k E co n o m ic R eview , 1 4 (3 ), p p .
5 9 7 -6 0 5 .
B ern a n k e, B . a n d M . G ertler (1 9 9 0 ). F in a n cia l fra g ility a n d eco n o m ic p erfo rm a n ce. Q u a r-
terly J o u rn a l o f E co n o m ics, 1 0 5 (1 ), p p . 8 7 -1 1 4 .
C a p rio , G . a n d D . K lin g eb iel (2 0 0 2 ). E p iso d es o f sy stem ic a n d b o rd erlin e ¯ n a n cia l crises.
W o rld B a n k R esea rch D o m estic F in a n ce D a ta S ets, O n lin e, J u ly 2 0 0 2 .
C a p rio , G . a n d D . K lin g eb iel (1 9 9 6 a ). B a n k in so lv en cies: C ro ss-co u n try ev id en ce. P o licy
R esea rch P a per 1 6 2 0 , W o rld B a n k , W a sh in g to n .
C a p rio , G . a n d D . K lin g eb iel (1 9 9 6 b ). B a n k in so lv en cy : B a d lu ck , b a d p o licy o r b a d b a n k -
in g ? . D o cu m en t p rep a red fo r th e A n n u a l B a n k C o n feren ce o n D evelo p m en t E co n o m ics,
A p ril 2 5 -2 6 , 1 9 9 6 , W o rld B a n k , W a sh in g to n .
D em irg u c-K u n t, A . a n d E . D etra g ia ch e (1 9 9 8 ). T h e d eterm in a n ts o f b a n k in g crises in d ev el-
o p in g a n d d ev elo p ed co u n tries. In tern a tio n a l M o n eta ry F u n d S ta ® P a pers, 4 5 (1 ), p p .
8 1 -1 0 9 .
D em irg u c-K u n t, A . a n d H . H u izin g a (2 0 0 0 ). F in a n cia l stru ctu re a n d b a n k p ro ¯ ta b ility.
W o rld B a n k W o rkin g P a per 2 4 3 0 , W o rld B a n k , W a sh in g to n .
D em irg u c-K u n t, A . a n d R . L ev in e (1 9 9 9 ). B a n k -b a sed a n d m a rk et-b a sed ¯ n a n cia l sy stem s:
C ro ss co u n try co m p a riso n s. W o rld B a n k W o rkin g P a per 2 1 4 3 , W o rld B a n k , W a sh in g -
to n .
F isch er, I. (1 9 3 3 ). T h e d eb t-d e° a tio n th eo ry o f g rea t d ep ressio n s. E co n o m etrica , 1 (4 ), p p .
3 3 7 -3 5 7 .
F ra n k el, A . a n d J .D . M o n tg o m ery (1 9 9 1 ). F in a n cia l stru ctu re: A n in tern a tio n a l p ersp ectiv e.
B roo kin g P a pers o n E co n o m ic A ctivity , 1 9 9 1 (1 ), p p . 2 5 7 -2 9 7 .
G ertler, M . (1 9 8 8 ). F in a n cia l stru ctu re a n d a g g reg a te eco n o m ic a ctiv ity : A n o v erv iew .
J o u rn a l o f M o n ey , C red it a n d B a n kin g, 2 0 (3 ), p p . 5 5 9 -5 8 8 .
G o ld sm ith , R .W . (1 9 6 9 ). F in a n cia l S tru ctu re a n d D evelo p m en t, Y a le U n iv ersity P ress, N ew
H a v en .
G o o d h a rt, C ., P . H a rtm a n n , D . L lew elly n , L . R o ja s-S u ¶ rez a n d S . W eisb ro d (1 9 9 8 ). F in a n -
a
cia l R egu la tio n . W h y , H o w a n d W h ere N o w ? , R o u tled g e, L o n d o n .
H o en ig , T .R . (2 0 0 1 ). P ersp ectiv es o n ¯ n a n cia l crises: W h a t h a v e w e lea rn ed fro m th e ev en ts
o f recen t y ea rs? . P a p er p resen ted to th e E leven th A n n u a l H y m a n P . M in sky C o n fer-
en ce o n F in a n cia l S tru ctu re, A p ril 2 6 , 2 0 0 1 , T h e J ero m e L ev i E co n o m ics In stitu te,
B lith ew o o d .
L a ttin , J ., D .J . C a rro ll, a n d P .E . G reen (2 0 0 3 ). A n a ly zin g M u ltiva ria te D a ta , T h o m so n
B ro o k s/ C o le, T o ro n to .
L ev in e, R . (2 0 0 2 ). B a n k -b a sed o r m a rk et-b a sed ¯ n a n cia l sy stem s: W h ich is b etter? . J o u rn a l
o f F in a n cia l In term ed ia tio n , 1 1 (4 ), p p . 3 9 8 -4 2 8 .
L o p ez, J .A . a n d M .M . S p ieg el (2 0 0 2 ). F in a n cia l stru ctu re a n d m a cro eco n o m ic p erfo rm a n ce
in th e sh o rt a n d lo n g ru n . W o rkin g P a per P B 0 2 -0 5 , C en ter fo r P a ci¯ c b a sin M o n eta ry

15. R ev ista M ex ica n a d e E co n o m ¶ y F in a n za s, V o l. 5 , N o . 1 (2 0 0 6 ), p p . 1 3 -2 7
³a 27
a n d E co n o m ic S tu d ies-E co n o m ic R esea rch D ep a rtm en t-F ed era l R eserv e B a n k o f S a n
F ra n cisco , S a n F ra n cisco .
M ish k in , F .S . (1 9 9 9 ). G lo b a l ¯ n a n cia l in sta b ility : F ra m ew o rk s, ev en ts, issu es. J o u rn a l o f
E co n o m ic P erspectives, 1 3 (4 ), p p . 3 -2 0 .
R u iz-P o rra s, A ., N . V ¶ sq u ez-Q u ev ed o a n d J .A . N u n ez-M o ra (2 0 0 6 ). E fecto s d e la g lo b -
a ~
a liza ci¶ n ¯ n a n ciera en la a d m in istra ci¶ n y reg u la ci¶ n d e riesg o s b a n ca rio s en M ¶ x ico .
o o o e
C o n ta d u r¶ y A d m in istra ci¶ n , fo rth co m in g .
³a o
S a n to m ero , A .M . (1 9 8 9 ). T h e ch a n g in g stru ctu re o f ¯ n a n cia l in stitu tio n s: A rev iew essa y.
J o u rn a l o f M o n eta ry E co n o m ics, 2 4 (2 ), p p . 3 2 1 -3 2 8 .