Journal of Banking & Finance 44 (2014) 114–129Contents lists.docx

Journal of Banking & Finance 44 (2014) 114–129
Contents lists available at ScienceDirect
Journal of Banking & Finance
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c
a t e / j b f
Macro-financial determinants of the great financial crisis:
Implications
for financial regulation q
http://dx.doi.org/10.1016/j.jbankfin.2014.03.001
0378-4266/� 2014 Elsevier B.V. All rights reserved.
q We would like to thank the Editor, an anonymous referee, Luc
Laeven, Ross
Levine, Marco Pagano, Andrea Sironi, Randy Stevenson,
Gianfranco Torriero,
Giuseppe Zadra and seminar participants at IFABS Conference
and ISTEIN seminar
for helpful comments. This paper’s findings, interpretations,
and conclusions are
entirely those of the authors and do not necessarily represent
the views of the
World Bank and the Italian Banking Association.
⇑ Corresponding author. Tel.: +39 02 58362725.
E-mail addresses: [email protected] (G. Caprio Jr.),
[email protected]
(V. D’Apice), [email protected] (G. Ferri), [email protected]
(G.W. Puopolo).
Gerard Caprio Jr. a, Vincenzo D’Apice b,c, Giovanni Ferri d,e,
Giovanni Walter Puopolo f,⇑

a Williams College, United States
b Economic Research Department of Italian Banking
Association, Italy
c Istituto Einaudi (IstEin), Italy
d LUMSA University of Rome, Italy
e Center for Relationship Banking & Economics – CERBE, Italy
f Bocconi University, CSEF and P. Baffi Center, Italy
a r t i c l e i n f o
Article history:
Received 15 April 2012
Accepted 4 March 2014
Available online 29 March 2014
JEL classification:
G01
G15
G18
G21
Keywords:
Banking crisis
Government intervention
Regulation
a b s t r a c t
We provide a cross-country and cross-bank analysis of the
financial determinants of the Great Financial
Crisis using data on 83 countries from the period 1998 to 2006.
First, our cross-country results show that
the probability of suffering the crisis in 2008 was larger for
countries having higher levels of credit
deposit ratio whereas it was lower for countries characterized
by higher levels of: (i) net interest margin,
(ii) concentration in the banking sector, (iii) restrictions to bank
activities, (iv) private monitoring. The

bank-level analysis reinforces these results and shows that the
latter factors are also key determinants
across banks, thus explaining the probability of bank crisis. Our
findings contribute to extend the analyt-
ical toolkit available for macro and micro-prudential regulation.
� 2014 Elsevier B.V. All rights reserved.
1. Introduction ment (BCBS, 2010a), has focused more on the
stability of the finan-
As much as it was known that the Great Depression of the 1930s
was the acid test for any reputable macroeconomic theory, the
out-
break of the Great Financial crisis in 2008 has shaken not only
financial institutions, but also long-held beliefs and theories on
how the regulation of the financial system should be structured,
with renewed emphasis on macro-prudential supervision and
reforming micro-prudential regulation.
In turn, the financial regulatory reforms have sparked a vibrant
debate among institutions, academics and practitioners. On the
one hand, the Basel Committee, starting with its consultative
docu-
cial system, arguing that the costs of the new regulation will be
much lower than the relative benefits (see BCBS, 2010b; MAG,
2010). On the other hand, the banking industry argues that the
new measures could put economic growth at risk imposing high
costs on the financial intermediaries and, in turn, on economic
sys-
tems (IIF, 2010). In the middle, some academics argue that the
prin-
ciples implicitly or explicitly subscribed by the Basel
Committee
may be questionable to secure more resilient financial systems
(see among others Ferri, 2001; Barth et al., 2004, 2006; Caprio,
2010).

In this debate, we study whether a wide set of banking indica-
tors, such as business model, funding strategy, market structure,
efficiency, stability, profitability, regulation, the quality of
gover-
nance and a measure of financial globalization could explain the
ex-post incidence of the crisis both across countries (that is, at
the macro-level) and across banks (that is, at the micro-level),
and be added to the analytical toolkit available for prudential
supervision. Specifically, in the cross-country analysis we
investi-
gate the (macro) financial determinants of the probability that a
country experienced the crisis in 2008, as reported by Laeven
and Valencia (2010), using data on 83 countries from 1998 to
2006. In the cross-bank analysis, by contrast, we pursue a
twofold
http://crossmark.crossref.org/dialog/?doi=10.1016/j.jbankfin.20
14.03.001&domain=pdf
mailto:[email protected]
http://www.sciencedirect.com/science/journal/03784266
http://www.elsevier.com/locate/jbf
G. Caprio Jr. et al. / Journal of Banking & Finance 44 (2014)
114–129 115
objective: first, using the information on the 10 largest banks by
average total asset during 1998–2006 for all the countries
consid-
ered in our sample we focus on the determinants of the
probability
that a bank experienced some form of distress during the Great

Financial crisis, to understand whether the main results obtained
in the cross-country analysis hold also with bank-level data.
Sec-
ond, we can address the potential problem of omitted variables
arising from the cross-country analysis.
A novel feature of our approach with respect to the related lit-
erature1 consists in measuring the financial indicators used as
explanatory variables taking into account all the information
relative
to the 9 years that preceded the Great Financial crisis and not
just to
the most recent years before its outbreak. In fact, we firmly
believe
that the early signals of what happened, starting from 2008,
were
already embedded in the financial characteristics of the
countries
and their banks several years before the crisis erupted. Then,
the
use of such ‘‘back-in-time variables’’ is justified by the fact
that these
contain information about (i) the health of the financial system
in
the past and (ii) how this evolves over time. As a consequence,
they
may be useful in understanding the genesis of the crisis.
Our cross-country analysis shows that, first, countries with a
higher credit/deposit ratio had higher probability to be in crisis
in 2008. Next, a few determinants negatively impinged on the
probability of crisis. Specifically, such probability was lower
for
countries with a higher level of net interest margin, higher level
of concentration in the banking sector, higher level of private
mon-

itoring, and more restrictions on bank activities. Moving to the
cross-bank analysis, it is important to underline that our micro-
level evidence contributes to reinforce these results by showing
that they hold not only across countries, but also across banks.
In
other words, we find that the financial factors found at the
coun-
try-level are also key determinants at the bank-level, thus
explain-
ing the probability of bank crisis as well.
In particular, among the various determinants of the crisis, a
crucial role is played by the net interest margin indicator. This
factor tends to be more significant the greater the importance of
deposits. In fact, banks that had a large and stable deposit base
likely paid less for funds (thus reaching a higher level of net
inter-
est margin) than the ones who had to rely on wholesale markets,
which proved to be more volatile. At the same time, net interest
margin also tends to be lower in banking systems more
extensively
engaged in securitization, both directly as securitization fees
dis-
place interest earnings (and interest on the securities is accruing
to off-balance sheet entities), and indirectly as securitization
boosts the supply of credit from non-bank entities which leads,
other things equal, to a decrease in the lending rates.
The rest of the paper is structured as follows. Section 2
provides
a review of the literature. Section 3 describes the data and the
dif-
ferent models employed in the econometric specifications,
focus-
ing first on the analysis across countries and then across banks.
Section 4 looks at the empirical results, while Section 5 offers

some
robustness checks. Finally, Section 6 concludes discussing some
lessons and policy implications.
2. Literature review
Over the years, several scholars have studied financial crises,
focusing on their possible causes and above all on predicting
their
time of occurrence. Historically, however, the economic
analysis
showed more success at identifying the incidence of the crises
across firms, banks or countries (i.e. cross-sectional) rather than
at forecasting the timing of crises (i.e. in time-series analysis).
For
instance, focusing on the financial crisis of 2008, Rose and
Spiegel
1 See among the others Rose and Spiegel (2009), Claessens et
al. (2010), Barth et al
(2004) and Beck et al. (2006).
.
(2009) use a latent variable approach to investigate whether a
wide
number of factors could have predicted the incidence and the
sever-
ity of the crisis for many countries. They find few clear reliable
indi-
cators of the incidence of the great recession in the pre-crisis
data:
more precisely, only the natural logarithm of 2006 real GDP per
capita and the size of the equity market run-up prior to the
crisis
result in a significant causality with the severity of the crisis.
In this regard, the closest paper to our cross-country analysis is
Barth et al. (2004). Using their database on bank regulation and

supervision in 107 countries to assess the relationship between
specific regulatory and supervisory practices and banking-sector
development, efficiency, and stability, they show that the likeli-
hood of suffering a major crisis is greater the more countries:
(1)
restrict bank activities (or prevent or discourage diversification
of
income through non-traditional activities); (2) put limits on for-
eign bank entry/ownership; (3) exacerbate moral hazard via a
more generous deposit insurance scheme. On the other hand,
nei-
ther capital stringency nor official supervisory powers – which
approximate respectively pillars one and two of Basel II – are
robustly linked to banking crises when controlling for other
super-
visory/regulatory policies. Similarly, there is no significant
associa-
tion between private-sector monitoring and the likelihood of a
banking crisis and only a weak positive relationship between
gov-
ernment ownership and the likelihood of a crisis.
Our macro-level analysis differs from theirs along several
dimensions. First, we focus on a different crisis episode, that is
the Great Financial crisis started in 2008. Second, we use a
different
set of macro-financial indicators as possible explanatory
variables
of the probability for a country to be in crisis in 2008. Third,
we do
not restrict the observations to a precise year (for example 1999
as
done by the already cited authors), but rather we take the annual
mean of these financial factors from 1998 to 2006, to take into
account the long-term evolution of the financial sector before
the

crisis broke out internationally. Finally, we reinforce our cross-
country results by also investigating the determinants of the
crisis
at the bank-level.
Before the great financial crisis broke out in 2008, Demirguc-
Kunt and Detragiache (1998) investigate the relationship
between
banking crises and measures aimed at increasing the level of
finan-
cial liberalization in 53 countries during the period 1980–1995.
They find that banking crises are more likely to occur in
liberalized
financial systems. However, they do not consider data on
regula-
tion and supervision. Mehrez and Kaufman (2000) employ a
mul-
tivariate probit model for 56 countries from 1977 to 1997 to
examine how the level of corruption (i.e. transparency) affects
the likelihood of financial crises. They report that, in countries
where the government policy is characterized by lack of
transpar-
ency, banks have incentives to raise credit above the optimal
level,
thus increasing the probability of a banking crisis.
Using data on 69 countries from 1980 to 1997, Beck et al.
(2006)
study the impact of bank concentration, bank regulation, and
national institutions on the probability that a country can
experi-
ence a systemic banking crisis. They also examine the
international
differences in bank capital regulations, rules restricting bank
entry,
regulatory restrictions on bank activities and the overall institu-

tional environment. They show that crises are less likely to
occur
in economies characterized by: (1) more concentrated banking
sys-
tems; (2) fewer regulatory restrictions on banks (i.e. lower
barriers
to bank entry and fewer restrictions on bank activities); (3)
national institutions that facilitate competition. In addition,
Shehzad and De Hann (2009) analyze the impact of financial
reform on systemic and non systemic banking crises in 85 coun-
tries, from 1973 to 2002, finding that certain types of financial
reform reduce the likelihood of crisis.
Focusing on the Great Financial crisis, Giannone et al. (2011)
study cross-country differences in output loss between 2008 and
116 G. Caprio Jr. et al. / Journal of Banking & Finance 44
(2014) 114–129
2009 using indices of country risk for more than one hundred
countries. They find that the set of policies that favour
liberaliza-
tion in credit markets are negatively correlated with the output
growth in 2008 and 2009.
Moreover, Claessens et al. (2010) investigate the causes of a
broader set of financial crises finding that the recent crisis has
four
major features similar to earlier episodes. First, in most
countries,
asset prices rapidly increased before the crisis. Second, several
key
economies experienced episodes of credit booms ahead of the
cri-
sis. Third, there was a dramatic expansion in a variety of

marginal
loans. Fourth, the regulation and the supervision of financial
insti-
tutions failed to keep up with developments. They also find that
the recent crisis was different from the previous ones in, at
least,
four new aspects. First, there was a widespread use of complex
and opaque financial instruments. Second, the
interconnectedness
among financial markets, nationally and internationally, with
the
United States at the core, had increased in a short time period.
Third, the degree of leverage of financial institutions
accelerated
sharply. Fourth, the household sector played a central role.
Moving to bank-level data, Beltratti and Stulz (2012) analyze
98
large banks from 20 countries and investigate whether bank per-
formance is related to bank-level governance, country-level
gover-
nance, bank balance sheet and profitability characteristics
before
the crisis, and country-level regulation. Their key results are:
(1)
banks that the market favored in 2006 showed especially poor
returns during the crisis; (2) banks with more shareholder-
friendly
boards performed worse during the crisis; (3) banks in countries
with stricter capital requirement regulations and with more
inde-
pendent supervisors performed better; (4) banks in countries
with
more powerful supervisors experienced worse stock returns; (5)
large banks with more Tier 1 capital and more deposit financing
at the end of 2006 showed significantly higher returns during

the
crisis.
Moreover, also focusing on micro-level data, De Jonghe (2010)
analyzes the impact of revenue diversity of financial
corporations
on the banking system stability and investigates why some
banks
are better able to shelter themselves from the storm. He shows
that
the shift to non-traditional banking activities, which generate
com-
missions, trading and other non-interest income, increases
individ-
ual banks risks and thus reduces the stability of the financial
system. Finally, DeYoung and Torna (2013) investigate whether
and how banks’ shift from traditional to nontraditional income
sources contributed to the failure of hundreds of US depository
institutions between 2008 and 2010.
3. Data and methodology
In this section we describe the data and the econometric models
we employ in the analysis (i) across countries and (ii) across
banks,
whereas in Section 4 we show the corresponding results. We
start
from the cross-country analysis.
4 For four countries, and more precisely Macau, Malta, Bahrein
and Oman, we had
to determine the status crisis/no crisis on our own by analyzing
the financial stability
review provided by their central banks. In fact, all the
information corresponding to
the independent variables is available whereas the information
about the dependent
variable is not available in Laeven and Valencia (2010).

5 We give here the synthetic description of the variables. For
more details, see
Appendix B.
6 This variable only includes customer deposits and does not
include interbank
3.1. Cross-country analysis
In the macro-level analysis we employ aggregate (that is, coun-
try-level) information on the country’s financial system from
the
period 1998 to 2008 for 83 countries, including OECD as well
as
non-OECD and developing countries.2,3
2 See Appendix A for the list of countries included in the
sample.
3 We select 1998 as starting year because the measurement of
our regulation
variables begins in that year. In fact, we borrow these variables
from the three surveys
conducted by Barth et al. (2004). Specifically, survey I was
conducted to assess the
state of regulation in 1998, survey II to assess the state of
regulation in 2002, and
finally, survey III to assess the state of regulation in 2005.
In order to identify the macro and financial structure factors
contributing to the Great Financial crisis, we run cross-country
regressions on the determinants of the probability that a country
experienced the crisis in 2008. Specifically, for 83 countries,
we
estimate several probit models in which the dependent variable,
that is CRISIS, is a dummy equal to one if the country is
classified
as either borderline crisis or systemic crisis according to the

definition introduced by Laeven and Valencia (2010),4 and zero
otherwise. Moreover, we investigate the impact of the degree of
financial globalization on the probability of crisis by estimating
an
instrumental-variables probit model accounting for this
variable’s
potential endogeneity. Finally, we also test the role of the
country’s
governance quality as explanatory variable of the probability of
crisis
through the indices provided by Kaufman et al. (2010).
Here are the variables employed in the cross-country
specifications.
As independent variables we employ a wide set of country
indicators to take into account the various characteristics of the
national financial systems, such as, e.g., banking efficiency,
stability, profitability, market structure, quality of governance
and regulation. In particular, we use the following banking
indica-
tors, all measured as means over the period 1998–20065:
(i) NET_INTEREST_MARGIN, measuring the country bank’s
net
interest revenue, as a share of its interest-bearing assets, to
proxy the banking system orientation towards traditional
activity;
(ii) ROA, return on assets (net income to total assets);
(iii) ROE, the country return on equity (net income to total
equity);
(iv) COST_INCOME, total costs as a share of total income of
all the

country’s banks, proxying bank efficiency;
(v) Z-SCORE, the aggregate bank z-score;
(vi) CREDIT_DEPOSIT, the country’s loan/deposit ratio.6
While a
high ratio indicates high intermediation efficiency, a ratio sig-
nificantly above one suggests reliance on possibly unstable
non-deposit funding (see Beck et al., 2000; Beck et al., 2010;
Merrouche and Nier, 2010)7;
(vii) CONCENTRATION, the share of the country’s three
largest
banks in all country’s banks assets. While big banks could
reduce risk via enhanced asset diversification, they could
raise risk if managers and shareholders anticipate ‘‘too-big-
to-fail’’ policies by regulators;
(viii) BANK_ASSETS_GDP, the ratio between the country’s
deposit
money bank assets and its GDP.
We also take into account the degree of financial globalization
with the following variable:
INT_DEBT_ISS_GROSS_GDP, the gross flow of international
bond
issues by the country scaled by its GDP, proxying the degree to
deposits.
7 The ratio of credit to deposit measures how much non-deposit
funding are used
to increase domestic credit. These alternative sources of funding
include short-term
debt (e.g. commercial paper and asset-backed commercial
paper) and long-term debt
(e.g. bonds). Though desirable, a breakdown of funding into

short-term and long-term
instruments is not available either from International Monetary
Fund, such as
International Financial Statistics, or from international bank-
level databases, such as
Bankscope (see Huang and Ratnovski, 2009).
Table 1
Summary statistics.
Mean Std. Dev. Min. Max. Observations Annual Average
Change
Dependent variables
CRISIS
(Probit Models) 0.23 0.42 0 1 83
CRISIS ORDERED
(Ordered Probit Model) 0.36 0.71 0 2 83
CRISIS_COST_GDP
(Tobit Model) 1.11 3.2 0 18.5 83
Independent variables
Banking variables
NET_INTEREST_MARGIN 4.31 2.35 0.88 12.78 83 �0.07
ROA 1.03 1.13 �1.82 4.38 83 0.13
ROE 10.73 10.09 �26.57 50.08 83 1.17
COST_INCOME 67.74 14.91 29.78 108.45 83 �1.28
Z-SCORE 11.25 6.63 4.2 39.17 81 �0.8
CREDIT_DEPOSIT 102.19 38.6 35.83 248.78 83 0.51
CONCENTRATION 67.42 18.95 24.8 100 83 �0.63
BANK_ASSETS_GDP 70.25 43.15 15.99 185.11 83 1.63
INT_DEBT_ISS_GROSS_GDP 21.44 25.62 0.04 145.98 73 2.26
Banking regulation variables

RESTRICTION 7.43 1.74 3.33 12 83
PRIVATE MONITORING 8.23 1.29 5.5 11 83
CAPITAL 6.15 1.49 3 10 83
ENTRY 7.41 0.87 3.67 8 83
SUPERVISION 10.94 2.31 5 14.25 82
Governance quality variables
KKZ_MEAN 0.45 0.86 �1.4 1.93 83
KKZ_RULELAW 0.45 0.93 �1.39 1.95 83
KKZ_REGQUAL 0.57 0.81 �1.3 1.92 83
KKZ_VOICE 0.42 0.86 �1.56 1.63 83
KKZ_POLSTAB 0.2 0.9 �2.1 1.58 83
KKZ_GOVEFF 0.55 0.94 �1.4 2.16 83
KKZ_CONCORR 0.48 1.04 �1.17 2.48 83
Table describes the summary statistics of the variables used in
our cross-country analysis. For each country, the independent
variables are computed as the annual average
over the period 1998–2006. For the Independent variables
‘‘Banking variables’’ we also computed the Annual average
change, defined as the average change of the variables
between two consecutive years.
Detailed definitions of the variables are given in Appendix B.
8 All the models are estimated with heteroschedasticity-robust
standard errors.
114–129 117
which a country’s financial system is interlinked with
international
financial markets.
We also use five measures of bank regulation, taken from Barth
et al. (2004), and computed as mean of their three surveys:

(i) RESTRICTION, the value of the ‘‘Overall Restrictions’’
index,
measuring the extent of regulatory restrictions on bank
activities in securities markets, insurance, real-estate, and
owning shares in non-financial firms;
(ii) PRIVATE MONITORING, the ‘‘Private Monitoring’’ index,
mea-
suring the degree to which regulations empower, facilitate,
and encourage the private sector to monitor banks;
(iii) CAPITAL, the ‘‘Capital Regulation’’ index, which can be
con-
sidered as a proxy of Basel Pillar 1;
(iv) ENTRY, the ‘‘Entry Requirements’’ index, proxying the
hur-
dles for entrants to get a bank license;
(v) SUPERVISION, the ‘‘Official Supervisory’’ index,
measuring
the degree to which the country’s bank supervisor has the
authority to take specific actions. It can be seen as a measure
of Basel Pillar 2.
Regarding the institutional quality of the country we use the
indices of governance quality provided by Kaufman et al.
(2010):
‘‘Voice and Accountability’’, ‘‘Political Stability and Absence
of Vio-
lence’’, ‘‘Government Effectiveness’’, ‘‘Regulatory Quality’’,
‘‘Rule of
Law’’ and ‘‘Control of Corruption’’.
Table 1 reports the summary statistics (together with their clas-
sification group) of all the variables we just described. Only in

the
case of the Independent Variables ‘‘Banking Variables’’ we also
compute the annual average change, defined as the average
change
of the variables between two consecutive years. In particular,
the
latter statistics provides some information about the (average)
evolution of the variables over 1998–2006, i.e. if they have
increased or diminished and the extent of this change.
We start our investigation of the determinants of the probabil-
ity that a country experienced the crisis in 2008 by estimating
the
following probit model8:
ProbðCRISISc ¼ 1jXÞ¼ Uða þ b1NET INTEREST MARGINc
þ b2CREDIT DEPOSIT c þ b3CONCENTRATIONc
þ b4RESTRICTIONc þ b5PRIVATE MONITORINGcÞ; ð1Þ
where the dependent variable, CRISISc, is a dummy equal to 1
if the
country c is classified as crisis and 0 otherwise, U is the
standard
normal cumulative distribution function, and X is the set of
explan-
atory variables.
The choice of the explanatory variables is motivated by several
reasons. First, we believe that one of the most important
banking
causes of the recent crisis is the shift from the ‘‘originate to
hold
(OTH)’’ model to the ‘‘originate to distribute (OTD)’’ model
(see
for example Berndt and Gupta, 2009; Mian and Sufi, 2009;
Stiglitz, 2010; FSA, 2009; D’Apice and Ferri, 2010; Trichet,

2009).
Therefore, the variable NET_INTEREST_MARGIN allows
testing
whether higher incentives to perform traditional banking
activities
could be a deterrent against the crisis. In fact, a lower net
interest
margin implies higher incentives for traditional banks to look
for
other income sources (that is, ‘‘searching for yield’’) and to
shift
(2014) 114–129
to new business models (Beck et al., 2000, 2010; Gambacorta
and
Marques-Ibanez, 2011).
Moreover, one of the most striking features of the great finan-
cial crisis was the impact on the money markets and global
liquid-
ity (Cecchetti, 2009; Allen and Carletti, 2008; Brunnermeier,
2009).
Thus, we use the variable CREDIT_DEPOSIT to account for the
role
of maturity mismatching. In fact, a ratio significantly above one
suggests that private sector lending is funded with non-deposit
sources, which could result in funding instability (Beck et al.,
2000, 2010).
Another important aspect of the country’s resilience to the crisis
concerns the structure of the banking system, and in particular
the
degree of bank concentration (Carletti, 2010). In truth,

economic
theory provides two conflicting predictions on the relationship
between concentration and stability. On the one hand, for the
char-
ter value hypothesis (Allen and Gale, 2000a, 2000b, 2004)
concen-
tration enhances stability, whereas, on the other hand, the
optimal
contracting hypothesis (Boyd et al., 2005) argues exactly the
oppo-
site. In our empirical work, the use of the variable
CONCENTRA-
TION is meant to capture the structure of the banking system.
Finally, many analyses argue that the flaws in the regulatory
framework played a very important role in leading to the crisis
(Demirguc-Kunt and Serven, 2009; Coval et al., 2009; Buiter,
2007; De Michelis, 2009). To account for the country’s
regulatory
regime, we use the corresponding indices provided by Barth
et al. (2004). In our base model (1), RESTRICTION proxies the
regulatory restrictions on banking activities, whereas PRIVATE
MONITORING proxies Basel Pillar 3.
We present the results of cross-country model (1) in Section
4.1.
3.1.1. The link with the financial globalization
In the previous analysis of the macro-financial determinants of
the crisis we disregarded the role played by the degree to which
a
country’s financial system is interlinked with international
financial markets. Of course, this could be a primary factor,
especially considering that international contagion was one of
the chief channels to spread the crisis worldwide. Thus, here,
we

extend our cross-country analysis and include this indicator
together with the other determinants of the crisis. Specifically,
we follow the related literature and measure the country’s
degree
of financial globalization using the gross flow of international
bond issues as percentage of its GDP, that is the variable
INT_DEBT_ISS_GROSS_GDP.
In any case, the inclusion of this independent variable may raise
problems of endogeneity. On the one hand, indeed, an extensive
literature underlined the possibility that the extent of financial
globalization of a country depends not only on its level of
develop-
ment and its policies, but is also influenced by deeper
fundamental
factors.9 On the other hand, however, it is also true that, in our
mod-
els, the explanatory variables are measured as the average of the
annual observations from 1998 to 2006, a circumstance which
could
attenuate (or even eliminate) the inverse causality effect of the
prob-
ability of crisis on the independent variables (measured much
before
2008). To better understand this issue, we compute the cross-
coun-
try correlation between the variable
INT_DEBT_ISS_GROSS_GDP
observed in 2008 and its annual mean from 1998 to 2006. We
find
that the resulting correlation is very high (about 0.95),
indicating a
strong persistence of these fundamental factors cited by the
litera-
ture,10 and thus suggesting that the inclusion of this variable
could

indeed raise problems of endogeneity.
9 See for example Acemoglu and Johnson (2005), Collins
(2005), Faria and Mauro
(2005), Kose et al. (2006), Spiegel (2008), Tytell and Wei
(2005), and Wei (2006).
10 Actually, the variable INT_DEBT_ISS_GROSS_GDP is quite
persistent over time. In
fact, all cross-country correlations between any two years of
this variable within the
period 1998–2006 are higher than 0.7. Results are available
upon request.
In order to avoid biased estimates when measuring the proba-
bility of crisis in 2008, we control for the endogeneity issue by
introducing several instrumental variables denoting the various
characteristics of the countries. In our setting, this translates
into
using an instrumental variables probit model, which is
practically
equivalent to estimating simultaneously the following two
equations:
INT DEBT ISS GROSS GDPc ¼ a þ c1 NET INTEREST
MARGINc
þ c2CREDIT DEPOSIT c
þ c3CONCENTRATIONc
þ c4RESTRICTIONc
þ c5PRIVATE MONITORINGc
þ c6LEGAL ORIGIN � SOCIALIST c
þ c7ETHNIC FRACTIONALIZATIONc
þ c8SCALED CAPITAL DISTANCEc þ ec;
ð2Þ
ProbðCRISISc ¼ 1jXÞ
¼ Uða þ b1 INT DEBT ISS GROSS GDPc

þ b2NET INTEREST MARGINc
þ b3CREDIT DEPOSIT c þ b4 CONCENTRATIONc
þ b5RESTRICTIONc
þ b6PRIVATE MONITORINGcÞ: ð3Þ
In Eq. (2), we linearly estimate INT_DEBT_ISS_GROSS_GDP
using as
control variables all the regressors of the base model (1) plus
the
variables: (i) LEGAL ORIGINS – SOCIALIST taken from La
Porta
et al. (1997 and following updates), capturing the legal
characteris-
tics of former-socialist countries, (ii) ETHNIC
FRACTIONALIZATION
taken from Alesina et al. (2003), measuring the degree of ethnic
het-
erogeneity in the countries, and (iii)
SCALED_CAPITAL_DISTANCE.
The latter variable is computed as the distance between the
capital
of each country and the USA capital divided by the highest
distance
from Washington available in our sample, to guarantee the
homo-
geneity of this measure with the other variables of our setting.
In
our framework, this ratio captures the distance of the country
from
the origin of the recent crisis (scaled by the maximal distance
from
the USA).
In Eq. (3), the probit model uses the value of INT_DEBT_ISS_
GROSS_GDP from Eq. (2) together with the major cross-
country

determinants of the crisis found in the previous section, whereas
the dependent variable does not change.
The aforementioned variables prove good instruments in our
regression.11 In fact, to avoid endogeneity, it is crucial that the
instruments are at the same time: (1) VALID (i.e., correlated
with
the financial globalization proxy), and (2) EXOGENOUS (i.e.,
uncorre-
lated with the probability of the crisis).
We control that our instrumental variables satisfy both condi-
tions, and find that they are: (1) related to financial
globalization
(that is, they are valid), and (2) uncorrelated with the error term
(that is, they are exogenous),12 thus supporting the view that,
in
our framework, they are indeed good instruments.
We report the cross-country results of the link with the degree
of financial globalization in Section 4.1.1.
3.1.2. The link with the quality of governance
Another important macro-variable in determining the probabil-
ity that a country experienced the crisis in 2008 could be the
11 They are also among the most widely used in the related
literature (see Beck et al.,
2006; Ponczek and Mattos, 2009; Glaeser et al., 2004).
12 Results are not reported and are available upon request.
15 In fact, it is not suitable to assign an identical value to all
banks belonging to the

same country independently of the financial situation associated
to the individual
banks (that is, independently of whether each bank is in distress
or not).
16 In other words, the dependent variable would exactly be a
linear combination of
such explanatory variables.
17 The ‘‘random-intercept model’’, also called ‘‘mixed-effects
model’’, contains both
114–129 119
quality of governance. In this section we precisely address this
issue by investigating the role of this factor as possible cross-
country determinant of the crisis, using as a proxy the
governance
indices provided by Kaufman et al. (2010).
We start by testing the explanatory power of the variable
KKZ_MEAN (which is the simple average of the six indicators
pro-
vided by Kaufman et al. (2010)13 when added to our cross-
country
model (Eq. (1)). However, since the correlation between this
indica-
tor and the variable NET_INTEREST_MARGIN is very high
(i.e. �0.60),
we follow a standard two-step procedure to deal with the issue
of
multicollinearity among the two variables. Namely, we first
regress
KKZ_MEAN on NET_INTEREST_MARGIN and a constant,
that is:

KKZ MEANc ¼ c0 þ c1 NET INTEREST MARGINc þ nc; ð4Þ
then, we use the estimated residual nc from (4) as explanatory
var-
iable in the following probit model:
ProbðCRISISc ¼ 1jXÞ
¼ Uða þ b1NET INTEREST MARGINc
þ b2 CREDIT DEPOSITc
þ b3 CONCENTRATIONc þ b4RESTRICTIONc
þ b5 PRIVATE MONITORINGc
þ b6 KKZ MEAN RESIDUALcÞ; ð5Þ
where the variable KKZ_MEAN_RESIDUAL is indeed the
estimated
residual nc from (4). In this way, KKZ_MEAN_RESIDUAL
captures
the effect of governance quality that is not explained by the
variable
NET_INTEREST_MARGIN.
After controlling for multicollinearity, we can properly investi-
gate the explanatory power of the quality of governance as
possi-
ble cross-country determinant of the crisis. Section 4.1.2 reports
the results.
3.2. Cross-bank analysis
The analysis across banks has a twofold objective: first, by
investigating the determinants of the probability that a bank
expe-
rienced some form of distress during the crisis, it allows us to
understand whether the results obtained in the cross-country
anal-
ysis hold also with bank-level data, that is whether the main

deter-
minants identified above are also ‘‘micro-founded’’; second, it
also
allows us to address the potential problem of omitted variables
arising from the cross-country analysis.
In fact, one of the main drawbacks of the cross-country analysis
developed so far is exactly the potential problem of omitted
vari-
ables, since our macro-approach does not allow the inclusion of
dummy variables identifying each country. Therefore, the
effects
attributed to the variables explicitly considered in the previous
sections could instead be originated by other country
characteristics.
In this section we follow Laeven and Levine (2009) and collect
information on the 10 largest banks by average total asset
during
1998–2006 for all the countries considered in our sample.
Because
some countries have data on fewer than 10 banks, our final
sample
consists of 755 banks across 83 countries.14 Using the same
defini-
tions provided in the cross-country analysis (see Section 3.1 and
Appendix B) but obviously referring to each bank in the sample,
we compute the variables ROA_BL, ROE_BL,
COST_INCOME_BL,
Z-SCORE_BL, NET_INTEREST_MARGIN_BL and
CREDIT_DEPOSIT_BL.
13 See Appendix B for further details.
14 Focusing on the 10 largest banks enhances comparability
among countries, as in
Laeven and Levine (2009), and at the same time avoids an

unbalanced distribution of
tail events. Moreover, our sample accounts for a huge portion of
the total banking
system assets in each country.
On the contrary, the variables CONCENTRATION,
RESTRICTION, PRI-
VATE MONITORING, CAPITAL, ENTRY and SUPERVISION
cannot be
computed for each bank, since they are defined only at the
country
level. Thus, in the latter case, we assign the same value to all
banks
belonging to the same country. Consistently with the
methodology
described in the cross-country analysis, all the bank-level
variables
are computed as the annual mean over the period 1998–2006.
In the cross-bank analysis, however, we cannot use the same
dependent variable employed so far, i.e. CRISIS, because first,
that
variable is defined only at the country level,15 and second,
regress-
ing CRISIS on regressors like country dummies, RESTRICTION
and
MONITORING is not statistically feasible since the
corresponding
model would generate perfect (or deterministic) dependence
between CRISIS and such regressors.16
For these reasons, here we use as dependent variable
CRISIS_BL,
a dummy variable being 1 if the bank failed or received a
govern-
ment recapitalization during the crisis and 0 otherwise. In fact,
such variable not only is defined at the bank-level but,

certifying
that the bank experienced some form of distress during the
crisis,
it is very much in line with the dependent variable used in the
cross-country analysis.
In order to investigate whether the results obtained in the
macro-level analysis hold also with bank-level data, we estimate
the following cross-bank probit model:
ProbðCRISIS BLi ¼1jXÞ¼Uðaþb1NET INTEREST MARGIN
BLi
þb2 CONCENTRATIONi þb3CREDIT DEPOSIT BLi
þb4 RESTRICTIONi þb5PRIVATE MONITORINGiÞ; ð6Þ
presenting the corresponding results in Section 4.2.
Next, we address the omitted variable problem and run a bank-
level random-intercept model17 that allows controlling for all
the
unmeasured factors associated with each country. In our setting,
by allowing each country to have a different intercept, such
model
is essentially equivalent to the approaches based on the
introduction
of country dummies.
In practice, we estimate the following (random-intercept) pro-
bit model:
ProbðYb;c ¼ 1jXb;c; ucÞ¼ Uða þ bXb;c þ zb;c ucÞ; ð7Þ
where c is the index identifying the country, b is the index
identify-
ing the bank, Yb,c corresponds to the dummy variable
CRISIS_BL, Xb,c

is the set of bank-level variables described above together with
the
country-level regulatory variables defined in Section 3.1, and
zb,c are
the covariates corresponding to the random effects and can be
used
to represent the random intercept in each country. Specifically,
in
random-intercept models, zb,c is simply the scalar 1.
In our bank-level setting, the term uc � N(0, r2u) is the error
related to the specific country, it has zero mean and is common
to all banks from the same country. In other words, it identifies
all the unmeasured factors associated with country c that affect
the probability for a bank of that country to be in crisis. On the
con-
trary, eb,c � N(0, r2e ) concerns the error related to bank b in
country
fixed effects and random effects. The fixed effects are
analogous to the standard
regression coefficients and are estimated directly, whereas the
random effects are not
directly estimated but are summarized according to their
estimated variances.
Specifically, random effects may take the form of random
intercepts. In a cross-
section model, random effects are useful for modelling intra-
group correlation; that
is, observations in the same panel are correlated because they
share common panel-
level random effects.
19 The reader should bear in mind that our results pertain to the
financial crisis

reaching its climax in 2008 and generalizing them to less
extreme crises cases may be
inappropriate.
20 We also try to interact this variable with CONCENTRATION
to test whether bigger
(2014) 114–129
c, it has zero mean and identifies all the unmeasured bank-
specific
factors which affect the probability of bank crisis.18
Moreover, compared to a standard probit model, in random-
intercept models there is just one additional parameter to esti-
mate, that is r2u. Specifically, if the estimation of r
2
u turns out to
be significant, it implies that each country would have a
different
intercept (more precisely, the intercept of country c would be
a + uc), meaning that all the unmeasured factors associated with
each country, thus captured by its intercept, would indeed affect
the probability for a bank to be in crisis. On the contrary, if the
esti-
mation of r2u turns out to be insignificant, then all unmeasured
factors associated with each country do not influence the
probabil-
ity of bank crisis.
All the results regarding the cross-bank analysis are described
in Section 4.2.
4. Results

In this section we report the results of the various models
described in the previous sections. We start from the cross-
country
results.
4.1. Cross-country determinants of the crisis
The evidence corresponding to the probit model (1) is reported
in column Probit 1 of Table 2. First, the coefficient of
NET_INTER-
EST_MARGIN is negative and statistically significant. This
indicates
that countries with a higher level of net interest margin had a
lower probability to be in crisis in 2008. A higher net interest
income is associated with less securitization, and might also be
picking up the impact of competition (e.g. less competitive
systems
such as Australia and Canada survived the crisis quite well,
even
though this evidence might be related to other features of their
regulatory or institutional settings as well). Indeed, a higher
level
of net interest margin represents a strong incentive for banks to
undertake traditional activities, such as loans, instead of riskier
non-traditional activities such as securities trading (IMF, 2012;
Gambacorta and Marques-Ibanez, 2011).
Second, the coefficient of CREDIT_DEPOSIT is positive and
statis-
tically significant, meaning that countries with a higher credit/
deposit ratio had a higher probability to be in crisis in 2008. As
a
matter of fact, a ratio significantly above one suggests reliance
on
possibly unstable non-deposit funding. While some of these

alter-
native sources may be as stable as customer deposits, i.e. retail
bonds funding, many other can prove highly volatile, such as in
the case of interbank or money market loans. Thus, considering
that the crisis featured a dramatic drop in the availability of
whole-
sale funding, our evidence suggests that in most countries the
alternative funding sources were of the volatile type.
Third, the coefficient of CONCENTRATION is negative and
statisti-
cally significant, indicating that countries with a higher level of
con-
centration in the banking sector had a lower probability to be in
crisis in 2008, a finding consistent with the absence of crisis,
for
example, in Australia or Canada. This result seems in line with
the
empirical evidence provided by Beck et al. (2006) who find that
cri-
ses are less likely to occur in economies with more concentrated
banking systems. Indeed, a more concentrated banking system
implies that the bank’s charter value is higher and, as a conse-
quence, the incentives for bank owners and managers to take
exces-
sive risk are lower. Apparently, the beneficial bank’s charter
value
effect on risk taking seems to have prevailed – in our sample of
countries – on the possible additional detrimental effect passing
18 In fact, the random-intercept probit model (7) may also be
stated in terms of the
latent linear response Y⁄b,c = a + bXb,c + uc + eb,c, with Yb,c
= 1 if Y
⁄
b,c > 0 and Yb,c = 0

otherwise, where Y⁄b,c denotes the latent variable.
through the impact of the level of competition on manager
compen-
sation schemes. Compensation policy seemed, in fact, to be a
crucial
factor during the crisis: countries with a lot of merger activity
in
banking tended to have compensation systems that favored
growth
of banks’ balance sheets, i.e. rewarding return without much
atten-
tion, if any, to the risk, and these countries seemed to have the
most
severe problems during the crisis (see Barth et al., 2012).
Fourth, the coefficient of RESTRICTION is negative and
statisti-
cally significant. This suggests that more restrictions on bank
activ-
ities lowered the probability for the country to be in crisis in
2008.
In other words, the regulatory-induced specialization in the
bank-
ing sector enhances financial stability.19 However, it is
important to
notice that this result contrasts with earlier studies (Barth et al.,
2006), and might in fact be a proxy for enforcement, that is
countries
that cared about imposing activity restrictions might have been
enforcing other regulations, and thus it might be the
enforcement
rather than the restrictions per se that matters. Indeed, Barth et
al.
(2012) cite numerous breakdowns in enforcement of regulation
as

major causes of the recent crisis. Unfortunately, we do not have
good
direct measures of such enforcement to test this interpretation.
Fifth, the coefficient of PRIVATE MONITORING is negative
and
statistically significant, meaning that countries with a higher
level
of private monitoring had a lower probability to be in crisis in
2008. This result appears in contrast with the previous
literature,
which found private monitoring important in a variety of areas
but practically with no influence on the stability of financial
sys-
tems (see Barth et al., 2006). However, we should recall that
earlier
research did not pertain to the evidence related to the recent
crisis.
Yet, this finding is in line with the goal of the third pillar of the
Basel II Capital Accord.
In addition to the baseline model (1) discussed above, we esti-
mate 9 cross-country probit models to investigate whether the
other variables described in Section 3.1 have a role in
explaining
the probability of country crisis. Specifically, in each model we
add to the five determinants of Eq. (1), in turn and one per time,
one of the following variables: COST_INCOME, Z_SCORE,
ROA,
ROE, BANK_ASSET_GDP, CAPITAL, ENTRY,
SUPERVISION, GDP_L
and POP_L. The latter macro-variables GDP_L and POP_L are
respectively the logarithm of country’s GDP and the logarithm
of
country’s population. Columns Probit 2 to Probit 10 of Table 2
show interesting evidence. First, the results of the baseline

model
are robust to the inclusion of these other regressors. Second,
macro-variables like measures of bank efficiency (COST-
INCOME),
stability (Z-SCORE), profitability (ROA and ROE), bank size
(BANK_ASSETS_GDP),20 the logarithm of GDP and the
logarithm of
population are never significant at the conventional levels.
More-
over, controlling for the five determinants characterizing the
probit
model (1), the coefficients of the variables CAPITAL and
ENTRY turn
out to be positive and significant.21 In particular, with regard to
CAP-
ITAL, this suggests that higher levels of initial capital
restriction
increased the probability for the country to be in crisis in 2008.
Indeed, one lesson we learnt from the great financial crisis is
that capital adequacy was almost irrelevant in determining the
occurrence of the crisis. For example, Northern Rock got in
trouble
just few weeks after the bank had announced plans to return
excess capital to shareholders. On the other hand, the
importance
of capital in lowering the risk-taking approach followed by
banks
and more concentrated banking systems can explain the
probability of the crisis.
However, also the interaction variable is not significant. The
result is not reported in
the tables but it is available upon request.
21 However, none of these two variables showed a significant
coefficient if regressed

individually against the dependent variable.
Table 2
Cross-country analysis.
Variables Probit 1 Probit 2 Probit 3 Probit 4 Probit 5
NET_INTEREST_MARGIN �0.2414** �0.2316** �0.2506**
�0.2704** �0.2402**
(0.1117) (0.1175) (0.1145) (0.116) (0.1107)
CREDIT_DEPOSIT 0.0174*** 0.0174*** 0.0167*** 0.0180***
0.0173***
(0.0045) (0.0046) (0.005) (0.0047) (0.0046)
CONCENTRATION �0.0235** �0.0231** �0.0239**
�0.0244** �0.0232*
(0.0115) (0.0112) (0.0114) (0.0119) (0.0119)
RESTRICTION �0.4298*** �0.4343*** �0.4611***
�0.4125*** �0.4355***
(0.1252) (0.1264) (0.1239) (0.1247) (0.1275)
PRIVATE MONITORING �0.3858** �0.3766** �0.4079**
�0.3605** �0.3886**
(0.1734) (0.1662) (0.1759) (0.1649) (0.1807)
COST-INCOME �0.004
(0.0109)
Z-SCORE �0.0193
(0.0238)
ROA 0.2215

(0.1851)
ROE �0.0037
(0.0203)
Constant 5.9711*** 6.1322*** 6.7117*** 5.5357*** 6.0496***
(2.0933) (2.2314) (2.0606) (1.9792) (2.2514)
Observations 83 83 81 83 83
Pseudo R-squared 0.4176 0.4184 0.4023 0.4258 0.418
Probit 6 Probit 7 Probit 8 Probit 9 Probit 10
NET_INTEREST_MARGIN �0.2465* �0.2567** �0.2476**
�0.2543** �0.2434**
(0.1311) (0.1199) (0.1195) (0.1203) (0.1059)
CREDIT_DEPOSIT 0.0174*** 0.0184*** 0.0164*** 0.0208***
0.0172***
(0.0048) (0.0052) (0.0045) (0.0051) (0.0053)
CONCENTRATION �0.0237** �0.0251** �0.0225*
�0.0235** �0.0217*
(0.0118) (0.0122) (0.0118) (0.0119) (0.0111)
RESTRICTION �0.4338*** �0.4658*** �0.4575***
�0.4983*** �0.4480***
(0.1255) (0.1211) (0.1191) (0.1304) (0.1344)
PRIVATE MONITORING �0.3867** �0.3898** �0.3899**
�0.3871** �0.4189**
(0.1724) (0.1838) (0.1779) (0.1709) (0.1731)
BANK ASSETS_GDP �0.0003
(0.0052)

CAPITAL 0.2319*
(0.129)
ENTRY 0.3645*
(0.2059)
SUPERVISION 0.1576
(0.1007)
GDP_L 0.0213
(0.1085)
POP_L 0.0504
(0.1802)
Constant 6.0551*** 4.8571** 3.5369 4.3758* 5.3179*
(2.1986) (2.3979) (2.5927) (2.5379) (2.9828)
Pseudo R-squared 0.4176 0.4536 0.4376 0.4407 0.418
Table shows the estimation of several cross-country probit
models, starting from our baseline model (Eq. (1)) shown in
Column Probit 1. In all these models, the dependent
variable is CRISIS (which is a dummy variable equal to 1 if the
country is classified as either borderline crisis or systemic crisis
by Laeven and Valencia and 0 otherwise).
Robust standard errors are reported in parenthesis. Summary
statistics are given in Table 1 and the definitions of explanatory
variables are provided in Appendix B.
* Statistical significance of the parameter at 10% significance
level.
** Statistical significance of the parameter at 5% significance
level.
*** Statistical significance of the parameter at 1% significance
level.

114–129 121
works only in specific circumstances. In fact, Laeven and
Levine
(2009), combining the Bankscope data with the regulatory data-
base, show that higher capital does lead to safer banks when
there
is a strategic owner, but when ownership is highly diversified,
higher capital is associated with more risk taking. In other
words,
when there is no strategic (that is large) owner, everyone free
rides
and looks for higher returns. On the contrary, banks keep a
more
conservative approach when a strategic owner has control
rights,
which is like having a high franchise value, just as some
concen-
trated-ownership banks adopted the practice of plowing more
bonuses into deferred equity to keep managers more risk averse.
The evidence relative to ENTRY indicates that a higher level of
entry restriction increased the probability for the country to be
in crisis in 2008. This result is in accordance with the
mainstream
view that entry restrictions build inefficiencies and, hence,
contrib-
ute to instability. It should also be noticed that our result is not
in
contrast with the role of concentration. In fact, as shown by
Beck
et al. (2006), competition reduces fragility when controlling for
concentration. Finally, the regulatory variable SUPERVISION
is
not significant at the conventional levels.

Before closing this section, it is worthwhile assessing the quan-
titative importance of the cross-country determinants of the
crisis
within our base model (1). In this regard, we computed the mar-
ginal effect, that is the change in the probability of country’s
crisis
for an infinitesimal change of the independent variables,
obtaining
that NET_INTEREST_MARGIN and CREDIT_DEPOSIT are the
two
most important regressors among the banking variables, whereas
RESTRICTION is the most important one among the regulatory
Table 3
Cross-country analysis: financial globalization and governance
quality.
Variables Probit IV Probit All Probit 11 Probit 12 Probit 13
NT_DEBT_ISS_GROSS_GDP �0.0163 0.017
(0.023) (0.0667)
�0.2309** �0.2660**
(0.0789) (0.1934) (0.1075) (0.1051) (0.122)
CREDIT_DEPOSIT 0.0175*** 0.0207* 0.0140** 0.0147**
0.0153***
(0.0051) (0.0106) (0.006) (0.0059) (0.0056)
CONCENTRATION �0.0189 �0.0336** �0.0308** �0.0297**
�0.0277**

(0.0119) (0.0165) (0.0125) (0.012) (0.0119)
RESTRICTION �0.3730*** �0.8098*** �0.3096* �0.3448**
�0.2896*
(0.123) (0.2629) (0.1636) (0.1573) (0.1691)
PRIVATE MONITORING �0.3491** �0.5210** �0.5440***
�0.5494*** �0.5171***
(0.1396) (0.2115) (0.1629) (0.1674) (0.1615)
COST-INCOME 0.0083
(0.0222)
Z-SCORE �0.0192
(0.0404)
ROA 0.4895
(0.35)
ROE 0.0135
(0.0338)
BANK_ASSETS_GDP �0.0115
(0.0163)
CAPITAL 0.2643
(0.2017)
ENTRY 0.1152
(0.2516)
SUPERVISION 0.2147
(0.2681)
GDP_L 0.081
(0.1488)

POP_L 0.1306
(0.2943)
KKZ_MEAN_RESIDUAL 1.0833**
(0.5033)
KKZ_RULELAW_RESIDUAL 0.8967**
(0.4453)
KKZ_REGQUAL_RESIDUAL 0.9282*
(0.5123)
Constant 5.5077** 2.7754 6.9808*** 7.1977*** 6.3805**
(2.3141) (7.8998) (2.3426) (2.3833) (2.3569)
Pseudo R-squared 0.4821 0.4728 0.4673
Column Probit IV shows the results of the instrumental
variables probit model (3) taking also into account the
endogeneity issue due to the inclusion of the variable
INT_DEBT_ISS_GROSS_GDP among the main determinants of
the crisis, i.e. Eq. (2). In columns Probit 11–13 we report the
results corresponding to the role of the quality of
governance as possible cross-country determinant of the crisis.
Finally, in column Probit All we test whether all the macro-
variables belonging to the categories ‘‘Banking
Variables’’ and ‘‘Banking Regulation Variables’’ shown in
Table 1 have a role in determining the probability for a country
to be in crisis in 2008 when considered jointly, that is
all together as explanatory variables.
In all these probit models the dependent variable is a dummy
equals to 1 if the country is classified as crisis and 0 otherwise.
Robust standard errors are reported in
parenthesis. Summary statistics are given in Table 1 and the
definitions of the explanatory variables are provided in

Appendix B.
level.
level.
level.
(2014) 114–129
variables. Column D Probit of Table 4 shows the results and
high-
lights that a marginal increase of NET_INTEREST_MARGIN
deter-
mines a 4.4% reduction in the probability of crisis, whereas a
marginal increase of CREDIT_DEPOSIT determines a 0.3%
increase
in such probability. Moreover, a change of
NET_INTEREST_MAR-
GIN, CREDIT_DEPOSIT and RESTRICTION in the range of
�0.5 to
+0.5 standard deviations from the mean implies, respectively, a
change of 10.5%, 12.4% and 13.9% in the probability of crisis.
4.1.1. The degree of financial globalization: empirical evidence
In this section we show the empirical evidence corresponding
to the instrumental variable probit model, i.e. Eqs. (2) and (3),
employed in Section 3.1.1 to investigate the degree to which a
country’s financial system is interlinked with international
finan-
cial markets.
As highlighted in column Probit IV of Table 3, we find that the
results of the baseline model (1) remain almost unchanged since
the variable INT_DEBT_ISS_GROSS_GDP is insignificant.

This sug-
gests that the probability for a country to be in crisis in 2008 is
not influenced by the degree to which the country’s financial
sys-
tem is interlinked with international financial markets, after
con-
trolling for the relevant macro-determinants of the crisis. The
only difference with the model outlined in Eq. (1) is that the
vari-
able CONCENTRATION becomes insignificant in this
specification.
For what concerns the instrumental variable approach, we find
that the correlation coefficient between the Eqs. (2) and (3), i.e.
rho, is equal to 0.85 while the likelihood ratio test of whether
such
coefficient is significantly different from zero gives a p-value
of
0.09, thus confirming our suspect that the variable
INT_DEBT_ISS_
GROSS_GDP is indeed endogenous.
Before closing this paragraph, and in view of the cross-country
results highlighted in the previous section, we also test whether
all
the macro-variables discussed so far (and more precisely, those
belonging to the categories ‘‘Banking Variables’’ and ‘‘Banking
Reg-
ulation Variables’’ shown in Table 1), when considered jointly,
that
is all together as explanatory variables, have a role in
determining
Table 4
Marginal effects, ordered probit and tobit specifications.

Variables D Probit Ord. Probit Tobit M_Coll
(0.0196) (0.1033) (0.7019)
CREDIT_DEPOSIT .0031*** 0.0185*** 0.1076*** 0.0174***
(0.001) (0.0048) (0.0369) (0.0045)
CONCENTRATION �.0043** �0.0320*** �0.1958**
�0.0235*
(0.0022) (0.0121) (0.096) (0.0115)
RESTRICTION �.0786*** �0.4720*** �2.7722***
�0.556***
114–129 123
the probability for a country to be in crisis in 2008. Therefore,
we
estimate a cross-country probit model which includes all the
afore-
mentioned variables at the same time (including as well the
instru-
mented measure of financial globalization) and show the results
in
column Probit All of Table 3. We notice that, in such a
specification,
the implications of our base model (Eq. (1)) still hold. In fact,
the
five variables characterizing our cross-country analysis are
confirmed as major determinants of the crisis, whereas on the
con-
trary, all the other variables, as suspected, are insignificant.
(0.0269) (0.1243) (0.9715) (0.133)
PRIVATE MONITORING �.0705** �0.3248** �1.565

�0.386*
(0.0287) (0.1569) (0.9675) (0.1734)
NIM_RESIDUAL �0.241*
(0.1117)
Constant 32.0186*** 5.8697**
(10.3615) (2.0716)
Observations 83 83 83 83
Pseudo R-squared 0.4176 0.3682 0.1893 0.4176
Column D Probit shows the marginal effects of the cross-
country variables char-
acterizing the baseline model (1), that is the change in the
probability of country
crisis for an infinitesimal change of the independent variables.
Column Ord. Probit
shows the results of the ordered probit model in which the
dependent variable is
ORDERED_CRISIS, i.e. a dummy equal to 1 if the country is
classified as borderline
crisis, 2 if the country is classified as systemic crisis by Laeven
and Valencia (2010)
and 0 otherwise. Column Tobit reports the results of the cross-
country Tobit model
with left-censoring limit equal to 0 and dependent variable
corresponding to
CRISIS_COST_GDP (the variable is provided by Laeven and
Valencia (2010), and
measures the country’s cost of public support to the financial
system in terms of its
GDP). See section 5.2 for more information about this variable.
Finally, Column
M_Coll reports the results of the probit model (9) employed to
resolve the potential

multicollinearity among the explanatory variables of our
baseline model.
Robust standard errors are reported in parenthesis. Summary
statistics are given in
Table 1 and the definitions of the explanatory variables are
provided in Appendix B.
level.
level.
level.
4.1.2. The quality of governance: empirical evidence
In this section we report the results characterizing the use of
the quality of governance as possible cross-country determinant
of the crisis, as discussed in Section 3.1.2. We start by showing
the empirical evidence corresponding to Eq. (5).
As highlighted in column Probit 11 of Table 3, we find that the
coefficient of the variable KKZ_MEAN_RESIDUAL is positive
and
significant, implying that countries with stronger governance
had
a higher probability to be in crisis. This result is in line with the
nature of the recent financial crisis, as it was mostly a
developed
countries crisis. For what concerns the other macro-variables,
most
importantly, the results of our base specification do not change.
We also investigate whether a more appropriate index with
respect to KKZ_MEAN provides different results from the ones
just
discussed. In particular, given the financial nature of our study,
we

believe that the most appropriate indices among the ones
provided
by Kaufman et al. (2010) are the variables KKZ_REGQUAL and
KKZ_RULELAW.22 Thus, we test their role in explaining the
probabil-
ity of country crisis using the same methodology employed in
the
case of the variable KKZ_MEAN. More precisely, since the
correlation
between these two indices and NET_INTEREST_MARGIN is,
respec-
tively, �0.56 and �0.66, we apply the same two-step procedure
described in Eqs. (4) and (5). Columns Probit 12 and 13 of
Table 3
show the corresponding results. We notice that, first, these two
indi-
ces exhibit positive and significant coefficients, in line with the
results characterizing the variable KKZ_MEAN, and second, the
main
implications of our baseline model (1) do not change.
23 In light of this result, we also tried to identify four possible
country characteristics
and investigated whether they had an impact on the probability
for a bank to be in
crisis. More precisely: (i) We selected four possible country
characteristics based on:
(1) the legal origin, (2) the degree of economic freedom, (3) the
degree of political
freedom and (4) our cluster analysis. (ii) For each
characteristic, we formed groups of
countries and allocated each country to a specific group
according to the value of the
characteristics chosen. (iii) Finally, we included dummies
representing the groups of
countries (and thus the common characteristics) in our bank-
level base model (Eq.

4.2. Cross-bank results
In this section we present the empirical evidence corresponding
to the cross-bank analysis described in Section 3.2.
We start by showing the results of the bank-level model (6).
Interestingly, in column BL1 of Table 5 we notice that all the
vari-
ables used in the cross-country model (1), except
CONCENTRATION,
are also significant when measured using individual bank data.
It is
worth mentioning that CONCENTRATION was already weakly
sig-
nificant or insignificant at all in some of the cross-country
estimates
seen in the previous sections. In addition, consistently with the
analysis provided by the cross-country models 2–9 (see Table
2),
we estimate other seven bank-level probit models in which, in
turn
and one per time, we add to the five explanatory variables of
Eq. (6)
one of the following variables: COST_INCOME_BL, ROA_BL,
ROE_BL,
Z_SCORE_BL, CAPITAL, ENTRY, and SUPERVISION.
Columns BL2-BL8
of Table 5 highlight that our micro-results are robust to the
inclu-
sion of such bank-level regressors as control variables,
confirming
also that the macro-financial factors we have identified above
are
indeed key determinants at micro-level as well.
The results of the random-intercept model described in Eq. (7)

are instead shown in Table 6. We notice that the estimated stan-
dard deviation ru (shown in the second part of the table labeled
‘‘Random-effects’’) is significant, being equal to 1.93 with a
stan-
dard error of 0.49, and represents the estimated standard
deviation
22 See the definition of the two variables in Appendix B.
in the intercept. This result, together with the evidence provided
by the likelihood-ratio test, highlights the presence of random
effects at the country-level and suggests that all the unmeasured
factors associated with each country affect the probability of
bank
crisis. Moreover, the first part of the table shows the estimated
coefficients of the cross-bank variables we identified. Except
CON-
CENTRATION, they are all significant, meaning that our results
are
robust when considering unmeasured effects at the country
level.
In other words, on the one hand we showed that there exist
unmeasured factors associated to each country that affect the
probability for a bank to be in crisis. Nevertheless, when we
take
into account such factors, thus allowing each country to have a
dif-
ferent intercept, our results do not change qualitatively.23
5. Robustness checks
In this section we investigate whether our macro-results are
sensitive to different definitions of the country crisis.
Therefore,
we perform two model specifications as robustness checks: an
ordered probit model (in which the dependent variable is a
dummy
valued zero in case of no country crisis, one in case of
borderline

crisis, and two in case of systematic crisis), and a tobit model
(in
(6)). Unfortunately, none of the proposed country-
characteristics turned out to be
significant. The corresponding analysis and results are not
reported in the paper and
are available upon request.
Table 5
Bank-level probit.
Variables Probit BL1 Probit BL2 Probit BL3 Probit BL4 Probit
BL5 Probit BL6 Probit BL7 Probit BL8
NET_INTEREST_MARGIN_BL �0.1644*** �0.1582***
�0.1667*** �0.1691*** �0.1718*** �0.1627*** �0.1707***
�0.1658***
(0.0369) (0.0366) (0.0427) (0.0379) (0.0384) (0.0379) (0.0369)
(0.037)
CREDIT_DEPOSIT_BL 0.0047*** 0.0039*** 0.0047***
0.0047*** 0.0044*** 0.0046*** 0.0047*** 0.0045***
(0.0012) (0.0013) (0.0012) (0.0012) (0.0012) (0.0012) (0.0012)
(0.0012)
CONCENTRATION 0.0056 0.0054 0.0055 0.0051 0.0087*
0.0057 0.0059 0.0045
(0.0046) (0.0046) (0.0047) (0.0046) (0.0048) (0.0045) (0.0046)
(0.0042)
RESTRICTION �0.2064*** �0.2014*** �0.2067***
�0.2116*** �0.2290*** �0.2046*** �0.2106*** �0.2000***
(0.0431) (0.0442) (0.0429) (0.0433) (0.0482) (0.0432) (0.045)

(0.042)
PRIVATE MONITORING �0.1871*** �0.2037***
�0.1871*** �0.1901*** �0.1661*** �0.1801*** �0.1827***
�0.2114***
(0.0619) (0.0628) (0.0619) (0.0614) (0.0629) (0.0651) (0.0621)
(0.0689)
COST/INCOME_BL �0.0021
(0.0035)
ROA_BL 0.0118
(0.0724)
ROE_BL 0.0069
(0.007)
ZSCORE_BL �0.0018
(0.0013)
CAPITAL 0.0234
(0.0471)
ENTRY 0.1726*
(0.093)
SUPERVISION �0.0438
(0.0328)
Constant 0.9948* 1.3006** 1.0006* 1.0142* 0.9886 0.7705
�0.3115 1.7138**
(0.5951) (0.6258) (0.5898) (0.6035) (0.6555) (0.7426) (0.9375)
(0.799)
Observations 755 744 755 754 684 755 755 755
Pseudo R-squared 0.2108 0.1911 0.2108 0.2148 0.2287 0.2113
0.2178 0.2147

Table shows, using information on individual data of the banks
(micro-level analysis), the estimation of several probit models,
starting from the baseline bank-level model
(Eq. (6)) shown in Column Probit BL1. In all these models, the
dependent variable is CRISIS_BL which is a dummy variable
equal to 1 if the bank failed or received a
recapitalization by the government during the crisis and 0
otherwise.
Robust standard errors are reported in parenthesis. The
corresponding methodology and variables are described in
Section 3.2.
level.
level.
level.
24 As reported in Laeven and Valencia (2010), there are three
types of support: (i)
liquidity support, (ii) gross restructuring costs, and (iii) asset
purchases and
guarantees. We disregard liquidity support for three reasons.
First, strictly speaking,
it is provided by the Central Bank and not by the Government.
Second, as such, it is
impossible to break it down by country across the Euro zone.
Third, more importantly,
liquidity support may be a really transient measure whose cost
could be inherently
limited. Thus, we use the conservative measure of Government
support given by the
sum of restructuring costs and asset purchases (we also
disregard the entity of the
guarantees as it is difficult to quantify what their real ex-post

cost would actually be).
(2014) 114–129
which the proxy of country crisis is the ratio of financial
support
offered by the government to the country’s GDP).
5.1. Borderline versus systemic country crisis
As with most situations in life, there is black and white but
there
is also grey. In fact, some countries crossed the Great Financial
crisis
quite well, while, among those bowing down, some suffered
more.
In this section we account for the different degree of intensity
of the
country crisis and distinguish between countries experiencing a
borderline crisis and countries truly subject to a systemic crisis,
as reported by Laeven and Valencia (2010). To accomplish this
task,
we estimate the same cross-country probit model (1) using a
differ-
ent dependent variable, i.e. ORDERED_CRISIS, which is a
dummy
equal to 1 if the country is classified as borderline crisis, 2 if
the
country is classified as systemic crisis and 0 otherwise.
Column Ord. Probit of Table 4 shows that the main results char-
acterizing our cross-country analysis of the crisis are not
sensitive
to the distinction between borderline versus systemic crisis,
thus
reinforcing the role of our financial determinants. In other

words,
the important characteristics of the financial systems outlined in
Sections 3.1 and 4.1 (maturity mismatch, business model, etc.)
keep playing a crucial role also on countries not so severely
affected by the crisis.
5.2. The cost of the crisis
As second robustness check, we consider an alternative proxy of
the country crisis based on the extent to which the Government
had to employ public finances to avoid the meltdown of the
national banking system. This indicator reflects a certain type
of
costs associated with the crisis and captures the willingness of
the governments to help the private banking system overcoming
the financial turmoil. We measure the country’s cost of the
crisis
as the sum of the various forms of support provided by the
Govern-
ment to the national banking system expressed in terms of the
country GDP.24 By definition, this variable has a lower bound
of zero
in case of countries either not in crisis, or, alternatively, giving
no
financial support.
From an econometric standpoint, the lower bound on the range
of the dependent variable suggests using a Tobit rather than an
OLS
specification. We follow this approach and estimate a Tobit
model
in which the dependent variable is CRISIS_COST_GDP, i.e. the
coun-
try’s cost of public support to the financial system as a ratio to
its
GDP, whereas the explanatory variables are the five

determinants
of the crisis shown in Eq. (1).
Column Tobit of Table 4 highlights that the results obtained
with the baseline model (1) remain almost unchanged. The only
difference is that PRIVATE MONITORING is no longer
significant at
the conventional levels (however, it has a p-value of 11%).
5.3. Multicollinearity
Our independent variables characterizing the cross-country
model (1) draw on a quite heterogeneous set of financial
Table 6
Cross-bank analysis: random-intercept probit model.
Mixed-
effects 1
Mixed-
effects 2
Mixed-
effects 3
Mixed-
effects 4
Mixed-
effects 5
Mixed-
effects 6

Mixed-
effects 7
Mixed-
effects 8
Fixed-effects variables
NET_INTEREST_MARGIN_BL �0.3813** �0.3631**
�0.3793** �0.3983** �0.3994** �0.3781** �0.3969**
�0.3840**
(0.1636) (0.163) (0.1664) (0.1643) (0.1662) (0.165) (0.1641)
(0.1634)
CREDIT_DEPOSIT_BL 0.0084*** 0.0076** 0.0084***
0.0085*** 0.0078** 0.0084*** 0.0085*** 0.0084***
(0.003) (0.0033) (0.0031) (0.003) (0.0031) (0.0031) (0.003)
(0.003)
CONCENTRATION 0.0036 0.003 0.0037 0.0022 0.009 0.0043
0.0052 0.0013
(0.0203) (0.02) (0.0204) (0.0203) (0.0196) (0.0205) (0.0202)
(0.0206)
RESTRICTION �0.7790*** �0.7565*** �0.7791***
�0.7874*** �0.7758*** �0.7709*** �0.7851*** �0.7548***
(0.2907) (0.2865) (0.2909) (0.2895) (0.2797) (0.2933) (0.293)
(0.2881)
PRIVATE MONITORING �0.5702* �0.5873* �0.5710*
�0.5710* �0.5279* �0.5288* �0.5595* �0.6012*
(0.3096) (0.3048) (0.3102) (0.3072) (0.2979) (0.3162) (0.3066)
(0.3139)
COST/INCOME_BL �0.0002
(0.0115)

ROA_BL �0.0123
(0.1914)
ROE_BL 0.0231
(0.0217)
ZSCORE_BL �0.0022
(0.0029)
CAPITAL 0.195
(0.2403)
ENTRY 0.7781
(0.564)
SUPERVISION �0.0836
(0.1616)
Constant 5.6642 5.7775 5.6683 5.582 5.4415 3.9669 �0.259
6.8461
(4.0077) (3.9826) (4.0123) (3.9857) (3.8831) (4.4391) (5.5221)
(4.6632)
Random-effects
std. dev. ru 1.948869
*** 1.899072*** 1.950832*** 1.937124*** 1.841586***
1.969257*** 1.901419*** 1.920985***
(0.4933981) (0.4834884) (0.4948426) (0.4905849) (0.4739024)
(0.4976826) (0.4789264) (0.4910517)
LR test vs. base regression
(Prob>=chibar2)
0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

Observations 755 744 755 754 684 755 755 755
Table shows the estimation of several random-intercept probit
models based on the following: Prob(Yb,c = 1|Xb,c,uc) = H (a +
bXb,c + zb,cuc), where c is the index identifying the
country, b identifies the bank, Yb,c corresponds to the dummy
variable CRISIS_BL, Xb,c is the set of bank-level variables
described in Section 3.2 together with the country-level
regulatory variables defined in Section 3.1, zb,c are the
covariates corresponding to the random effects. Finally, the
country error term is uc � N(0, r2u).
Robust standard errors are reported in parenthesis. The
corresponding methodology and variables are described in
Section 3.2.
level.
level.
level.
114–129 125
indicators. For example, on one side we have country-level
banking
variables like NET_INTEREST_MARGIN, COST_INCOME,
ROA and ROE,
observed annually and related to the aggregate balance sheet of
the banks, so belonging to the category of financial performance
indicators. On the contrary, the regulatory variables
RESTRICTION,
PRIVATE MONITORING, CAPITAL, ENTRY and
SUPERVISION are the
indices reported in the three surveys by Barth et al. (2004), and
somehow reflect the countries’ financial background (structure).
Given the different nature of our independent variables, one

may
suspect that, at the country level, the latter variables (‘‘i.e.
struc-
ture variables’’) could not only have a direct impact on the
proba-
bility for a country to be in crisis in 2008, but might also affect
some of the ‘‘performance variables’’, giving rise to a problem
of
multicollinearity.
In this section, we investigate the issue of potentially-correlated
regressors at the macro-level and show that, once
multicollinearity
is taken into account, our cross-country results still hold.
In order to detect the presence of multicollinearity, we start by
computing the cross-country correlation between the
independent
variables used in the probit model (1). We find that all the
correla-
tion coefficients are smaller than |0.3| except that between the
variables RESTRICTION and NET_INTEREST_MARGIN
which is
0.38.25 This evidence suggests that, in general, the linear
relationship
between any pair of these macro-variables is quite weak, and
most
25 Results are not reported in the paper but are available upon
request.
importantly, that the ‘‘structure variables’’ do not have a big
influ-
ence on the ‘‘performance variables’’. However, the presence of
a cor-
relation coefficient close to 0.4 might induce a minimum
suspect of
multicollinearity between the variables

NET_INTEREST_MARGIN and
RESTRICTION.
To check whether this suspect is justified, we estimate a
slightly
different cross-country model based on a two-step procedure.
We
first regress NET_INTEREST_MARGIN on RESTRICTION and
a con-
stant, that is:
NET INTEREST MARGINc ¼ c0 þ c1 RESTRICTIONc þ nc;
ð8Þ
then, we use the estimated residual n from (8) as explanatory
vari-
able in the following cross-country probit model:
ProbðCRISISc ¼ 1jXÞ¼Uðaþb1NIM RESIDUALc
þb2 CREDIT DEPOSIT c þb3CONCENTRATIONc
þb4 RESTRICTIONc þb5 PRIVATE MONITORINGcÞ; ð9Þ
where the variable NIM_RESIDUAL is indeed the estimated
residual n
from regression (8).26
The rationale of this two-step procedure is quite intuitive. By
regressing NET_INTEREST_MARGIN on RESTRICTION, in
fact, we
are isolating the effects of regulatory restrictions on the interest
26 Both Eqs. (8) and (9) are estimated with heteroskedasticity
robust standard
errors.

(2014) 114–129
margin, and let the residual capture all the other factors
affecting
the dependent variable.
Column M_Coll of Table 4 shows that substituting the variable
NIM_RESIDUAL in lieu of NET_INTEREST_MARGIN in Eq.
(9) does
not alter the validity of our cross-country base specification. In
fact,
the estimated coefficient of NIM_RESIDUAL is negative and
significant, while all the other regressors remain qualitatively
unchanged.
6. Conclusions
Although there is a growing consensus on the principle that
regaining stability – today as much as it did in the 1930s –
requires
better regulation of the marketplace (see D’Apice and Ferri,
2010),
the agreement as to what ‘better’ might entail is far from being
established. More of the same seems hardly a satisfactory
response. In previous decades, banking systems underwent a
deep
transformation progressively moving away from a type of
banking
centered on personal relationships to hinge on more
standardized
and impersonal approaches. The change, theorized by some con-
sultants and academics, and designed to reach unprecedented
high
returns, prescribed gearing the banks with financial markets and
modifying the bank business model. This had a notable impact
on the transformation of some banking systems, in particular

those
that shifted toward a new business model (that is originate-to-
dis-
tribute, OTD), while others remained fastened to the traditional
business model (that is originate-to-hold, OTH). Our evidence
sug-
gests that a more traditional banking system had a lower
probabil-
ity to be in crisis in 2008. Thus, a return to an old style
banking, like
the one prevailing during the Quiet Period in the US after the
Great
Depression (Gorton, 2009), is being considered in some
quarters.
Furthermore, the results stated in this paper, both at the coun-
try-level and at the bank-level, provide additional fuel to that
dis-
cussion and can help policymakers calibrate new regulations, by
achieving a reasonable trade-off between financial stability and
economic growth, and contribute to extend the analytical toolkit
available for macro-prudential supervision and reforming micro-
prudential regulation. In fact, the traditional banking business
model, that is banks with higher net interest margin, has proved
resilient through the crisis. At the same time, the higher capital
levels prescribed by Basel III will penalize commercial banks.
The
need for traditional banks to increase their own funds will have
two consequences. First, since capital is costly and there is a
race
for deposits, banks will have to increase the price of their loans,
making credit more expensive with negative consequences on
growth and no additional positive effects on stability. Second,
there
will be a tendency to reduce lending in order to shrink the
denom-

inator of capital ratios (de Larosière, 2011). Also, higher capital
requirements, in response to the vulnerabilities highlighted in
the paper, require some care. In recent years, both countries
featur-
ing traditional approaches to banking, such as in Ireland, Spain,
and
CRISIS dummy equal to 1 if the country is clas
Valencia (2010) and 0 otherwise.
NET_INTEREST_MARGIN annual mean from 1998 to 2006 of
the
its interest-bearing assets (source: Beck
ROA and ROE ROA = annual mean of return on assets
mean of return on equity (net income
COST_INCOME annual mean value of total costs as a s
(Source: Beck et al. (2010))
Z-SCORE annual mean of aggregate bank z-score
ratio of return on assets plus capital-as
CREDIT_DEPOSIT annual mean from 1998 to 2006 of priva
saving deposits in deposit money bank
Eastern Europe, and those which moved to a more securitized
approach, as the US, plunged into serious crises. The fact that
cap-
ital was mostly not significant should give regulators pause:
like
doctors who used leeches (and thought that they were helping
patients), or econometricians who go into a dark room looking
for a black cat that is not there and scream ‘I’ve got it’, they
may
be settling for a popular cure in lieu of the one that actually
works.
And as Laeven and Levine (2009) find, the impact of capital

requirements might vary with the ownership structure of banks.
Finally, low interest rate contexts, for instance after the bust of
the dotcom and subprime bubbles, could lead banks to search
for
yields in more complex non-traditional activities that increase
their exposure to new risks. Regulators should also consider
that,
compared to traditional credit risks, evaluating these new risks
is
more difficult, especially those related to the complex financial
contracts so deeply entrenched in the OTD model, which played
a major role in the recent crisis. This suggests ending or
reversing
prolonged periods of low interest rates, the typical background
of
the broader regulatory framework artificially boosting
securitiza-
tion, and that measures to divulge and verify more information
about risk taking in banking (so that market monitoring might
finally work) are desiderable.
To conclude, the Authorities should carefully ponder the finan-
cial determinants of the Great Crisis, especially now that a
major
increase in minimum bank capital is being enforced within the
framework of Basel 3 and micro-and-macro-prudential
regulation
is to be implemented in most countries.
Appendix A. Sample information
Countries that experienced a Systemic Crisis are: Austria, Bel-
gium, Denmark, Germany, Iceland, Ireland, Latvia,
Luxembourg,
Netherlands, United Kingdom, USA.

Countries that experienced a Borderline Crisis: Greece, Hun-
gary, Kazakhstan, Portugal, Slovenia, Spain, Sweden,
Switzerland.
Other countries in the sample are: Argentina, Australia,
Bahrain,
Bangladesh, Belize, Bolivia, Botswana, Brazil, Bulgaria,
Burundi,
Canada, Chile, Colombia, Costa Rica, Croatia, Czech Republic,
Egypt,
El Salvador, Estonia, Finland, Guatemala, Guyana, Honduras,
Hong
Kong, India, Indonesia, Israel, Italy, Japan, Jordan, Kenya,
Korea,
Kuwait, Lithuania, Macau, Macedonia, Malaysia, Mali, Malta,
Mau-
ritius, Moldova, Morocco, New Zealand, Norway, Oman,
Pakistan,
Panama, Papua New Guinea, Peru, Philippines, Poland, Saudi
Arabia, Senegal, Singapore, Slovakia, South Africa, Sri Lanka,
Swazi-
land, Thailand, Togo, Trinidad and Tobago, Tunisia, Turkey,
Uruguay.
Appendix B
Definition of the variables used in the cross-country analysis.
sified as either borderline crisis or systemic crisis by Laeven
and
accounting value of the bank’s net interest revenue, as a share
of
et al., 2010)
(net income to total assets) from 1998 to 2006. ROE = annual
to total equity) from 1998 to 2006 (Source: Beck et al. (2010))
hare of total income of all country’s banks from 1998 to 2006

from 1998 to 2006 (Source: Beck et al. (2010)). The z-score is
the
set-ratio to the 5-years standard deviation of return on assets
te credit by deposit money banks as a share of demand, time and
s (Source: Beck et al. (2010))
CONCENTRATION annual mean from 1998 to 2006 of the
share of the country’s three largest banks in all country’s bank
assets (Source: Beck et al. (2010))
BANK_ASSETS_GDP annual mean of the ratio between the
country’s deposit money bank assets and its GDP from 1998 to
2006 (Source: Beck et al. (2010))
INT_DEBT_ISS_GROSS_GDP annual mean from 1998 to 2006
of the gross flow of international bond issues by the country
scaled by its
GDP (Source: Beck et al. (2010))
RESTRICTION mean value of the ‘‘Overall Restrictions’’ index
reported in the three surveys by Barth et al. (2004). This
index measures the degree to which banks face regulatory
restrictions on their activities in: (a) securities
markets, (b) insurance, (c) real-estate, and (d) owning shares in
non-financial firms. The index can take
values from 0 to 4 for each of these four sub-categories, where
4 indicates the most restrictive
regulations on this sub-category of bank activity. Thus, the
index of overall restrictions can potentially
range from 0 to 16
PRIVATE MONITORING Mean value of the ‘‘Private

Monitoring’’ index reported in the three surveys by Barth et al.
(2004). The
index measures the degree to which regulations empower,
facilitate, and encourage the private sector to
monitor banks. It reflects the information on whether: (1) bank
directors and officials are legally liable
for the accuracy of information disclosed to the public, (2)
banks must publish consolidated accounts, (3)
banks must be audited by certified international auditors, (4)
100% of the largest 10 banks are rated by
international rating agencies, (5) off-balance sheet items are
disclosed to the public, (6) banks must
disclose their risk management procedures to the public, (7)
accrued interest/principal, though unpaid,
enter the income statement while the loan is still non-
performing, (8) subordinated debt is allowable as
part of capital, and (9) there is no explicit deposit insurance
system and no insurance was paid the last
time a bank failed. The private monitoring index has a minimum
value of 0 and a maximum value of 9,
where larger numbers are associated with a greater regulatory
empowerment of private monitoring of
banks
CA PITAL Mean value of the ‘‘Capital Regulation’’ index
reported in the three surveys by Barth et al. (2004). This
index includes information on (1) the extent of regulatory
requirements regarding the amount of capital
banks must hold and (2) the stringency of regulations on the
extent to which the source of funds that
count as regulatory capital can include assets other than cash or
government securities, borrowed funds,
and on whether the regulatory/supervisory authorities verify the
sources of capital. Large values indicate
more stringent capital regulations

ENTRY Mean value of the ‘‘Entry Requirements’’ index
reported in the three surveys by Barth et al. (2004). The
index essentially counts the number of requirements for
obtaining a banking license: (1) draft by-laws;
(2) intended organizational chart; (3) financial projections for
first 3 years; (4) financial information on
main potential shareholders; (5) background/experience of
future directors; (6) background/experience
of future managers; (7) sources of funds to be used to capitalize
the new bank; and (8) market
differentiation intended for the new bank
SUPERVISION Mean value of the ‘‘Official Supervisory’’
index reported in the three surveys by Barth et al. (2004). This
index measures the degree to which the country’s commercial
bank supervisory agency has the authority
to take specific actions. It is determined by the information
provided on the following features of official
supervision: (1) does the supervisory agency have the right to
meet with external auditors about banks?
(2) are auditors required to communicate directly to the
supervisory agency about elicit activities, fraud,
or insider abuse? (3) can supervisors take legal action against
external auditors for negligence? (4) can
the supervisory authority force a bank to change its internal
organizational structure? (5) are off-balance
sheet items disclosed to supervisors? (6) can the supervisory
agency order the bank’s directors or
management to constitute provisions to cover actual or potential
losses? (7) can the supervisory agency
suspend the directors’ decision to distribute: a) dividends? b)
bonuses? c) management fees? (8) can the
supervisory agency supersede the rights of bank shareholders-
and declare a bank insolvent? (9) can the
supervisory agency suspend some or all ownership rights? (10)
can the supervisory agency: a) supersede

shareholder rights? b) remove and replace management? c)
remove and replace directors? The official
supervisory index has a minimum value of 0 and a maximum
value of 14, where larger numbers indicate
a greater power
KKZ_VOICE Mean value of ‘‘Voice and Accountability’’ index
from 1998 to 2006. This index reflects perceptions of the
extent to which a country’s citizens are able to participate in
selecting their government, as well as
freedom of expression, freedom of association, and a free
media. This index ranges from �2.5 (weak) to
2.5 (strong) governance performance.
KKZ_POLSTAB mean value of ‘‘Political Stability and
Absence of Violence’’ index from 1998 to 2006. This index
reflects
perceptions of the likelihood that the government will be
destabilized or overthrown by unconstitutional
or violent means, including politically-motivated violence and
terrorism. This index ranges from �2.5
(weak) to 2.5 (strong) governance performance
KKZ_GOVEFF mean value of ‘‘Government Effectiveness’’
index from 1998 to 2006. This index reflects perceptions of
the quality of public services, the quality of the civil service
and the degree of its independence from
political pressures, the quality of policy formulation and
implementation, and the credibility of the
(continued on next page)
114–129 127

government’s commitment to such policies. This index ranges
from �2.5 (weak) to 2.5 (strong)
governance performance.
KKZ_REGQUAL Mean value of ‘‘Regulatory Quality’’ index
ability of the government to formulate and implement sound
policies and regulations that permit and
promote private sector development. This index ranges from
�2.5 (weak) to 2.5 (strong) governance
performance
KKZ_RULELAW Mean value of ‘‘Rule of Law’’ index from
1998 to 2006. This index reflects perceptions of the extent to
which agents have confidence in and abide by the rules of
society, and in particular the quality of
contract enforcement, property rights, the police, and the courts,
as well as the likelihood of crime and
violence. This index ranges from �2.5 (weak) to 2.5 (strong)
governance performance
KKZ_CONCORR mean value of ‘‘Control of Corruption’’ index
extent to which public power is exercised for private gain,
including both petty and grand forms of
corruption, as well as ‘‘capture’’ of the state by elites and
private interests. This index ranges from �2.5
(weak) to 2.5 (strong) governance performance
KKZ_MEAN Mean value of the six measures provided by
Kaufman et al. (2010) from 1998 to 2006
(2014) 114–129
References

Acemoglu, D., Johnson, S., 2005. Unbundling institutions.
Journal of Political
Economy 113 (5), 949–995.
Alesina, A., Devleeschauwer, A., Easterly, W., Kurlat, S.,
Wacziarg, R., 2003.
Fractionalization. Journal of Economic Growth 8, 155–194.
Allen, F., Gale, D., 2000a. Comparing Financial System. MIT
Press, Cambridge, MA.
Allen, F., Gale, D., 2000b. Financial contagion. Journal of
Political Economy 108, 1–
33.
Allen, F., Gale, D., 2004. Asset price bubbles and monetary
policy. In: Desai,
Meghnad, Said, Yahia (Eds.), Global Governance and Financial
Crisis. Routledge,
London, pp. 19–42.
Allen, F., Carletti, E., 2008. The Role of Liquidity in Financial
Crises. Jackson Hole
Conference Proceedings, Federal Reserve Bank of Kansas City,
pp. 379–412.
Barth, J., Caprio Jr., G., Levine, R., 2004. Bank regulation and
supervision: what works
best? Journal of Financial Intermediation 13, 205–248.
Barth, J., Caprio Jr., G., Levine, R., 2006. Rethinking Bank
Regulation: Till Angels
Govern. Cambridge University Press.
Barth, J., Caprio Jr., G., Levine, R., 2012. The Guardians of

Finance: Getting Them to
Work for Us. MIT Press.
Basel Committee on Banking Supervision (BCBS), 2010a. Basel
III: A Global
Regulatory Framework for More Resilient Banks and Banking
Systems.
Basel Committee on Banking Supervision (BCBS), 2010b. An
Assessment of the
Long-Term Economic Impact of the New Regulatory
Framework.
Beck, T., Demirgüç-Kunt, A., Levine, R., 2000. A new database
on financial
development and structure. World Bank Economic Review 14,
597–605.
Beck, T., Demirgüç-Kunt, A., Levine, R., 2010. Financial
institutions and markets
across countries and over time: the updated financial
development and
structure database. World Bank Economic Review 24 (1), 77–
92.
Beck, T., Demirgüç-Kunt, A., Levine, R., 2006. Bank
concentration, competition and
crises: first results. Journal of Banking and Finance 30, 1581–
1603.
Beltratti, A., Stulz, R.M., 2012. Why did some banks perform
better during the credit
crisis? A cross-country study of the impact of governance and
regulation.
Journal of Financial Economics 105, 1–17.

Berndt, A., Gupta, A., 2009. Moral hazard and adverse selection
in the originate-to-
distribute model of bank credit. Journal of Monetary Economics
56, 725–743.
Boyd, J., De Nicolo, G., Jalal, A., 2005. The Theory of Bank
Risk Taking and
Competition Revisited. IMF WP.
Brunnermeier, M., 2009. Deciphering the liquidity and credit
crunch 2007–08.
Journal of Economic Perspectives 23 (1), 77–100.
Buiter, V., 2007. Lessons from the 2007 Financial Crisis.
Background Paper
Submitted to the UK Treasury Select Committee, December 11,
2007.
Caprio, G. Jr., 2010. Financial Regulation in a Changing World:
Lessons from the
Recent Crisis. In: Paolo Savona (Ed.), Financial Collapse: How
are the Biggest
Nations and Organizations Managing the Crisis? Fondazione
Cesifin ‘‘Alberto
Predieri’’.
Carletti, E., 2010. Competition, Concentration and Stability in
the Banking Sector.
IstEin WP No. 8.
Cecchetti, S.G., 2009. Crisis and responses: the federal reserve
in the early stages of
the financial crisis. Journal of Economic Perspectives 23 (1),
51–76.
Claessens, S., Köse, M.A., Terrones, M.E., 2010. The global

financial crisis: how
similar? How different? How costly? Journal of Asian
Economics 21, 247–264.
Collins, S.M., 2005. Comments on financial globalization,
growth, and volatility in
developing countries, by Eswar Prasad, Kenneth Rogoff, Shang-
Jin Wei, and M.
Ayhan Kose. In: Ann Harrison (Ed.), Globalization and Poverty.
Coval, J., Jurek, J., Stafford, E., 2009. The economics of
structured finance. Journal of
Economic Perspectives 23 (1), 3–25.
D’Apice, V., Ferri, G., 2010. Financial Instability: Toolkit for
Interpreting Boom and
Bust Cycle. Palgrave Macmillan Studies in Banking and
Financial Institutions,
London.
De Jonghe, O., 2010. Back to the basics in banking? A micro-
analysis of banking
system stability. Journal of Financial Intermediation 19, 387–
417.
de Larosière, J., 2011. Don’t Punish the Banks That Performed
Best. Financial Times,
March 3.
De Michelis, A., 2009. Overcoming the Financial Crisis in the
United States. OECD
Economics Department W P No. 669.
Demirguc-Kunt, A., Detragiache, E., 1998. Financial
Liberalization and Financial
Fragility. World Bank Policy Research Paper No. 1917.

Demirguc-Kunt, A., Serven, L., 2009. Are the sacred cows
dead? Implications of the
financial crisis for macro and financial policies. World Bank
Research Observer
25, 91–124.
DeYoung, R., Torna, G., 2013. Nontraditional banking activities
and bank failures
during the financial crisis. Journal of Financial Intermediation
22, 397–421.
Faria, A., Mauro, P., 2005. Institutions and the External Capital
Structure of
Countries. IMF W P No. 04/236.
Ferri, G., 2001. Capital Adequacy Requirements: Impact and
Evolution. Opening
Remarks, As Guest Editor, of the Special Issue (no. 3/2001)
Economic Notes, pp.
319–326.
Financial Services Authority (FSA), 2009. The Turner Review a
Regulatory Response
to the Global Banking Crisis.
Gambacorta, L., Marques-Ibanez, D., 2011. The bank lending
channel: lessons from
the crisis. Economic Policy 26 (66), 135–182.
Giannone, D., Lenza, M., Reichlin, L., 2011. Market freedom
and the global recession.
IMF Economic Review 59, 111–135.
Glaeser, E.L., La Porta, R., Lopez-De-Silanes, F., Shleifer, A.,
2004. Do institutions
cause growth? Journal of Economic Growth 9, 271–303.

Gorton, G.B., 2009. Slapped in the face by the invisible hand:
banking and the panic
of 2007. Prepared for the Federal Reserve Bank of Atlanta’s
2009 Financial
Markets Conference: Financial Innovation and Crisis, May 11–
13, 2009.
Huang, R., Ratnovski, L., 2009. Why are Canadian Banks More
Resilient? IMF WP/09/
152.
Institute of International Finance (IIF), 2010. Interim Report on
the Cumulative
Impact on the Global Economy of Proposed Changes in the
Banking Regulatory
Framework.
International Monetary Fund (IMF), 2012. Global Financial
Stability Report
(October).
Kaufman, D., Kraay, A., Mastruzzi, M., 2010. The Worldwide
Governance Indicators:
A Summary of Methodology, Data and Analytical Issues. World
Bank Policy
Research Paper No. 5430.
Kose, M.A., Prasad, E., Rogoff, R., Wei, S.-J., 2006. Financial
Globalization: A
Reappraisal. IMF WP 06/189.
Laeven, L., Levine, R., 2009. Bank governance, regulation and
risk taking. Journal of
Financial Economics 93, 259–275.

Journal of Banking & Finance 44 (2014) 114–129Contents lists.docx

Journal of Banking & Finance 44 (2014) 114–129Contents lists.docx

Recommended

Recommended

More Related Content

Similar to Journal of Banking & Finance 44 (2014) 114–129Contents lists.docx

Similar to Journal of Banking & Finance 44 (2014) 114–129Contents lists.docx (20)

More from donnajames55

More from donnajames55 (20)

Recently uploaded

Recently uploaded (20)

Journal of Banking & Finance 44 (2014) 114–129Contents lists.docx