Description
Financial Liberalization refers to reduction of any sort of regulations on the financial industry of a given country.
FINANCIAL STUDY FOR FINANCIAL
LIBERALIZATION, BANK CRISES AND GROWTH
Abstract
This paper studies the efects of financial liberalization and bank-
ing crises on growth. It shows that financial liberalization spurs on average economic growth.
Banking crises are harmful for growth, but to a lesser extent in countries with open financial
systems and good institutions. The positive efect of financial liberalization is robust to diferent
definitions. While the removal of capital account restrictions is efective by increasing
financial depth, equity market liberalization afects growth directly. The empirical analysis is
performed through GMM dynamic panel data estimations on a panel of 90 countries ob- served
in the period 1975-1999.
JEL classification: C23, F02, G15, O11.
Keywords: Capital account liberalization, equity market liberaliza-
tion, financial development, institutions, dynamic panel data.
-
1
1 Introduction
In the last two decades an increasing number of countries have eliminated
controls on international capital movements. However, the global economic
crises of recent years have led many economists to reconsider the beneficial
efects of financial liberalization on economic performance. Although the
issue has been widely debated, there are no conclusive results on the efects of
financial integration on growth
1
.
In theory, international financial liberalization softens financing con-
straints and improves risk-sharing, thereby fostering investments. It may also
have a positive impact on the functioning and development of finan- cial
systems, and on corporate governance
2
. These arguments suggest that we
should expect a positive relation between international financial liber-
alization and economic growth
3
. However, the presence of distortions may
reduce the positive efects of liberalization. In fact, information asymmetries may
lead to a bad allocation of capital, and weak financial and legal sys- tems
could induce capital?ights towards countries with better institutions. Moreover,
banking crises may come along with financial liberalization, as it is well
documented in the literature
4
.
Table 1 shows mean equality tests for growth, financial development
and the occurrence of banking crises across diferent treatment (open, bank
crises) and control (closed, no crises) groups of countries, observed annu- ally
between 1975 and 1999. The results suggest that countries without
restrictions on capital account or equity market transactions had, on aver-
age, higher growth rates and financial development (as measured by credit to
the private sector as a ratio of GDP). The occurrence of banking crises is
associated with lower growth rates and financial development. However, it is not
clear whether there is correlation between openness and the occurrence of bank
crises. When we consider an overall index there is no significant diference
in the frequency of crises between countries with and without re- strictions on
capital account transactions. The picture becomes clearer once we split the index
between "systemic" and "non-systemic" banking crises.
1
See Edison et al. (2003) for a review of the empirical literature.
2
See Klein and Olivei (2000) and Levine (2000) for empirical evidence on the positive
impact of financial liberalization on growth, .
3
Evidence on the positive relation between financial development and growth is pro-
vided by a large literature (see Demirguc-Kunt and Levine, 2001 for a survey). The results in
La Porta et al. (1999) suggest that good corporate governance spurs growth.
4
See Kaminsky and Reinhart (1999) and Aizenmann (2002) for a survey.
2
Open countries experienced a lower number of systemic crises but a higher
number of non-systemic crises. The higher frequency of non-systemic crises may
be a reason for the concern of economists and governments on the efects of
financial liberalization on economic performance.
In this paper we assess empirically the efects of international financial
liberalization and banking crises on growth. We admit the possibility that
banking crises come along with financial liberalization, as shown by previous
works and by row 5 of Table 1, and investigate their joint impact on growth
5
.
Kaminsky and Schmukler (2002) and Tornell et al. (2004) suggest in difer- ent
ways that institutional quality may matter at shaping the relationship between
financial liberalization, crises and long-run growth
6
. Therefore, we control for
institutions and their interactions with financial openness and crises. To
have a better understanding of the mechanism that links the variables of our
interest, we assess whether liberalization and crises afect growth through
financial depth. Also in this case, we control for institu- tional quality.
Inspired by the results in Acemoglu and Johnson (2003)
7
, we distinguish
between institutions aimed at contractual as opposed to property rights
protection.
The empirical analysis is performed on a panel dataset that covers 90
countries over the period 1975-1999. We adopt the Dynamic Panel Data
approach proposed by Arellano and Bover (1995) and Blundell and Bond
(1998). We use two indicators of financial liberalization, that distinguish
between capital account and equity market liberalization.
Our results show that capital account liberalization has a positive efect on
growth, once we control for banking crises, whose impact is negative. The
absence of capital account controls is good for growth because it fosters
financial development and mitigates the harmful efects of banking crises.
Capital account liberalization allows firms to raise funds more easily on the
5
Causality between financial liberalization and banking crises is left aside from our
empirical analysis.
6
Kaminsky and Schmukler (2002) show evidence that equity market liberalization
brings about financial chaos in the short-run, but has positive long-run efects, since it in-
duces changes in institutions supporting the functioning of the domestic financial market.
Tornell et al. (2004) suggest that liberalization and crises afect growth through financial
development; given financial openness, good institutions make bank crises less likely, and
foster capital in?ows.
7
Acemoglu and Johnson (2003) show that contractual protection afects financial struc-
ture more than property rights protection, but has limited efects on economic perfor-
mance. Vice versa, property right protection afects GDP growth, productivity and in-
vestments, but not the financial structure.
3
international financial markets, and thus sufer less from domestic crises.
Moreover, banking crises turn out to be less harmful for growth in countries
where property and contractual rights are better protected. Equity mar- ket
liberalization instead has a strong direct efect on growth and does not interact
with banking crises.
There are many contributions in the literature on the efects of financial
liberalization on long-run growth. Bekaert, Harvey and Lundblad (2003) is the
closest work to this paper. These authors as well consider both capital account
and equity market liberalization, and control for bank crises. They also allow
for heterogeneity in the efects of liberalization depending on cross-country
diferences in institutional quality. The main elements that distinguish our
contribution are the attention to the interaction between financial
liberalization and bank crises, the analysis of the mechanism that links them to
growth through financial development and the use of a diferent dynamic panel
data technique.
The remainder of the paper is organized as follows. Section 2 describes the
econometric model and the variables we used. Section 3 reports the
estimation results and comments on them. Section 4 councludes.
2 Data and empirical strategies
2.1 The econometric model
We assess the growth efects of financial liberalization and banking crises by
adding these variables to a dynamic version of the standard growth regres- sion
8
. We follow the dynamic panel data approach suggested by Arellano and
Bover (1995) and Bond and Blundell (1998)
9
. This methodology is pre- ferred to
the cross-sectional regressions because it allows to account for the impact of the
policy changes, imbedded in the indexes of financial liberaliza- tion and of crisis
episodes, on growth. This dynamic panel technique is also helpful to amend the
bias induced by omitted variables in cross-sectional es- timates, and the
inconsistency caused by endogeneity both in cross-sectional and static panel
(fixed and random efects) regressions.
We formulate the standard neoclassical growth model in a dynamic panel
8
See among others, Barro (1997) and Barro and Sala-i-Martin(1995).
9
The system-DPD methodology dominates the diference-DPD proposed by Arellano
and Bond (1991) because it amends problems of measurement error bias and weak in-
struments, arising from the persistence of the regressors (as pointed out by Bond et al.,
2001).
4
data form, and estimate the following dynamic system:
?y
it
= o?y
it
÷
1
+ |
0
?X
it
+ o?F lib
it
+ ¸?Bcr
it
+ ?v
t
+ ?
i,t
(1)
y
it
= oy
it
÷
1
+ |
0
X
it
+ oF lib
it
+ ¸Bcr
it
+ q
i
+ v
t
+ i,t, (2)
where time indexes refer to non-overlapping five-year periods. ?y
it
is the
average annual growth rate of real per capita GDP over five years. y
it
is
the logaritm of real per capita GDP, and the coefcient on its lag, o = e
5
ì
,
supports conditional convergence if it implies ì< 0. Variables indexed by
t ÷ 1 are observed at the beginning of the five-year period, and covariates
are expressed in period averages. Matrix X
it
contains determinants of GDP
growth, such as human capital, population growth and other factors that
account for diferent long-run per capita output across countries. F lib
i
(t+k,t)
and Bcr
i
(t+k,t
)
are indicators of financial liberalization and banking crises.
q
i
, v
t
and
it
are respectively the unobservable country- and time-specific
efects, and the error term. The presence of country efect in equation (2)
corrects the omitted variable bias. The diferences in equation (1) and the
instrumental variables estimation of the system are aimed at amending in-
consistency problems
10
. We instrument diferences of the endogenous and
predetermined variables with lagged levels in equation (1) and levels with
diferenced variables in equation (2). For instance, we take y
it
÷
3
as instru-
ment for ?y
it
÷
1
and F lib
it
÷
2
for ?F lib
it
in (1) and ?y
it
÷
2
as instrument for y
it
÷
1
and ?F lib
it
÷
1
for F lib
it
in (2). We estimate the system by General-
ized Method of Moments with moment conditions E[?y
it
÷
s
(
it
÷ it÷
1
)] =
0 for s > 2, and E[?z
it
÷
s
(
it
÷ it÷
1
)] = 0 for s > 2 on the predeter-
mined variables z, for equation (1); E[?y
i,t
÷
s
(q
i
+ c
i,t
)] = 0 and E[?z
i,t
÷s
(q
i
+ c
i,t
)] = 0 for s = 1 for equation (2). We treat all regressors as predeter-
mined. The validity of the instruments is guaranteed under the hypothesis
that
it
are not second order serially correlated. Coefcient estimates are
consistent and efcient if both the moment conditions and the no-serial cor-
relation are satisfied. We can validate the estimated model through a Sargan
test of overidentifying restrictions, and a test of second-order serial corre-
lation of the residuals. As pointed out by Arellano and Bond (1991), the
estimates from the first step are more efcient, while the test statistics from the
second step are more robust. Therefore, we will report coefcients and statistics
from the first and second step respectively.
1
0
See
Temple (1999) for a survey on the methodologies used in growth regressions.
5
2.2 Financial liberalization and financial fragility: the data
To explore the impact of financial liberalization and banking crises on growth we
need to measure these variables. The literature on financial liberalization has
proposed diferent indicators that difer along several directions. The major
distinctions are based on the de iure vs de facto definition criterion, the
characterization on a zero-one vs continuous scale, and the market they refer to.
In our analysis we construct an index of liberalization of both capital
account and equity market based on two diferent sources
11
. The first one is a
dummy variable provided by the IMF in its Annual Report on Exchange
Arrangements and Exchange Restrictions (AREAER), that is available for a
maximum of 212 countries starting from 1967
12
. This is the most com- monly
used measure of restrictions on international financial transactions. It takes
value 1 if a country has experienced restrictions on capital account transactions
during the year, and zero otherwise. Our yearly measure of financial
liberalization, opIM F, equals 1 and 0 when the IMF dummy is re- spectively 0
and 1. The second indicator is based on Bekaert et al.'s (2003) chronology of
ofcial equity market liberalization, that is available for 95 countries from 1980.
Our variable opBHL difers from opIM F because it only accounts for equity
market liberalization, but not for globalization of the credit market for
instance. Moreover, diferently from the AREAER, Bekaert et al.'s measure
does not contemplate policy reversals, so that a country is labeled as open
ever since its first year of liberalization. As the IMF-based indicators, it takes
value 1 and zero in case of internationally open and closed country-years,
respectively. Both opIM F and opBHL are
expressed as five-year averages, thereby taking values in the [0,1] interval.
There are alternative measures that are able to account for diferent de-
grees of liberalization instead of just the presence or absence thereof. Quinn's (1997)
index scores the intensity of capital account controls on a scale from 0 to 4 with
steps of 0.5. However, it is hardly suited for panel studies since it is available
for a significant number of countries only for four years, 1958, 1973, 1982 and
1988. Other contributions have used de facto measures, as
1
1
We
focus on de iure zero-one measures, that classify a country as financially liberalized
if there are no legal restrictions to international trade of financial instruments.
1
2
Classification methods have changed in 1996, so that there are 13 separate indexes
now, that can hardly be compared to the previous single indicator. Miniane (2000) har-
monized the classifications, though for a limited number of countries, and over a short
time span. Therefore, the last observation for opIMF in our dataset dates back to 1996.
6
data on international capital?ows as a ratio of GDP. The idea is that ac-
tual international capital?ows are a good proxy for the degree of financial
openness. A more comprehensive discussion on the available indicators can be
found in Edison et al. (2002).
Banking crises are subject to various classifications as well. As for liber-
alization, we adopt a zero-one anecdotal indicator of bank crises, proposed by
Caprio and Klingebiel (2001). The authors keep record of 117 systemic and 51
non-systemic crises occurred in 93 and 45 countries respectively, from the late
Seventies on. On a yearly base, our variable Bcr takes value 2 if the country
has experienced a systemic banking crisis, meaning that much or all of a bank's
capital has been exhausted; 1 if the banking crisis involved less severe losses;
and 0 otherwise. We use two alternative data reductions for robustness analysis:
Bcr012 takes value 2 if Bcr equals 2 at least once over the period, 1 if at least a
1 is scored, zero otherwise. Bcr012av instead accounts also for the duration of
crisis episodes, since it equals the period average of Bcr.
The other covariates in our growth regressions are variables commonly
accounted for in the empirical growth literature (see Barro, 1997), such as
secondary school attainment, the growth rate of population, government ex-
penditure and investments as a ratio of GDP. Other factors that we want to take
into account are financial development, proxied by the ratio of credit to the
private sector over GDP, and, at a further stage of the analysis, insti- tutional
quality, as measured by the government anti-diversion policy index (Hall and
Jones, 1999) and by the indicator of efciency of the judiciary system (see La
Porta et al. 2003). The first indicator mainly accounts for property rights
protection, while the other refers more to contractual rights.
The sample consists of data for a maximum of 90 countries over the
period 1975-1999 or 1980-1999 depending on the indicator of financial lib-
eralization adopted. Since keeping the larger sample gives us a longer time-
series in the panel analysis, we will go on reporting results from the 1975-99
sample for opIM F and from 1980-99 for opBHL. Since we average over non-
overlapping five-year periods, either four or five observations for each
country are available. More detail on the countries in our sample and on all
variables is given in the appendix.
7
3 Empirical evidence
3.1 Liberalization, banking crises and growth
Table 2 reports results from dynamic-panel estimations of the augmented
growth regression, which includes the usual control variables (initial GDP,
secondary school attainment, population growth, government spending and
investments over GDP) plus indicators of financial liberalization, financial
development, and banking crises. Consistently with the previous cross-
country growth studies (see Barro, 1997 and Barro and Sala-i-Martin, 1995), we
find significant evidence that countries with lower initial real per capita GDP
have grown faster than the initially richer ones, conditional on the other
variables. Our estimates imply a convergence rate of about 1.5% per year
13
.
Population growth and investments have the signs predicted by growth theory
(respectively negative and positive) in most of the estimates, though not always
significant.
Capital account openness has zero-efect on growth. Equity market lib-
eralization instead exhibits a significant positive coefcient (columns 1 and 5).
These results are in line with Bekeart et al.'s (2003) findings. Using the same
measure of financial liberalization, they show that equity market lib-
eralization significantly afects growth, while the relation between the IMF
measure and growth is fragile.
As a wide strand of literature (see Aizenmann, 2002 for a survey) points out,
the removal of restrictions on capital?ows may expose financial sys- tems to
turmoil and possibly crises
14
. If that is the case, the costly impact of
financial crises
15
, brought about by liberalization, could be responsible for the
coefcient estimates for opIM F in column 1. To control for this hy- pothesis, we
include the bank crisis indicator in the regression of columns 2. Once we control
for the occurrence of bank crises, the positive coefcient for opIM F becomes
significant. As expected, banking crises strongly restrain growth. Moreover,
the interaction between capital account openness and crises in column 3 is
positive. This suggests that, irrespective of whether
1
3
T he
convergence rate is computed as ˆ =
ln(
5ˆ
)
. o
ì
1
4
Among others, Kaminsky and Reinhart (1999) show that financial liberalization of-
ten precedes banking crises, Glick and Hutchison (1999) find that financially
liberalized emerging market economies are more likely to experience twin crises,
Demirguc-Kund and Detragiache (1998) show that banking crises occur more often in
liberalized financial systems.
1
5
A number of papers try to quantify the output costs of financial crises. See among
others Edwards (1999), Honohan and Klingebiel (2001), and De Gregorio and Lee (2004).
8
financial liberalization triggers instability in the banking sector, countries
without capital account restrictions are less prone to the negative efects of
banking crises than financially closed economies. Thus, capital account
liberalization has no strong direct efects on growth, but it is important to
mitigate the negative efects of banking crises.
The results are slightly diferent if we restrict the focus on equity market
liberalization. As in Bekeart et al., equity market openness and banking crises
have indeed strong opposite efects, respectively positive and negative, but the
introduction of the crises variable does not afect the efectiveness of equity
market liberalization on growth. Moreover, we find no interaction between the
two variables (see column 7). In fact, it is not so surprising that free
international equity trade alone can be less of help in case domestic banks get
into troubles. Firms that rely on credit may be severely hurt by banking crises,
and find it difcult to shift abruptly to equity financing, even if they can sell
shares on the international market. If instead they have free access to
international credit markets, they might raise funds more easily there, and thus
sufer less from domestic crises.
Opposite results are obtained by Eichengreen and Leblang (2002). They
show that the negative efects of domestic crises are neutralized by the pres- ence
of controls on capital controls. One reason could be that they use a diferent
indicator of crises (by Bordo et al., 2001) that encompasses both exchange and
banking crises.
As a robustness check, we replicate the estimations in Table 2 using an
indicator of Banking crises that accounts also for the duration of banking
crises, Bcr012av. Table 5 reports coefcients only for liberalization, bank crises
and their interaction. The results are not remarkably diferent from the ones
we obtained using the discrete crisis indicator.
3.2 Institutions, Financial Liberalization and Growth
After Hall and Jones' (1999) seminal paper, a wide strand of growth lit-
erature has focused on institutions as a primary determinant of economic
performance. Alfaro et al. (2004) have shown that institutions are an im-
portant determinant of capital in?ows. Tornell et al. (2004), in line with this
argument, suggest that in financially open countries institutional qual- ity
afects both the occurrence of banking crises and the extent of capital in?ows.
Banking crises may occur as a by-product of openness, as credit markets get
thicker, especially if there is a poor legal environment. In open
9
economies, the presence of good institutions facilitates capital in?ows from
abroad, when domestic banking crises reduce the amount of credit available to
firms
16
. As a result, banking crises are expected to be less harmful for
growth in countries where property and contractual rights are better pro-
tected. Symmetrically, financial liberalization might turn out to be growth-
restraining in countries with worse institutions. In order to assess empiri- cally
these implication we include interactive terms in our dynamic growth
regressions.
Table 3 shows results from system-GMM estimations that include the
same regressors in columns 1-3 of Table 2, plus the interactions of capital
account liberalization with indicators of institutional quality. We also inves-
tigate the relation between liberalization, financial development and
overalleconomic development
17
. Institutional quality is proxied here by the
gov- ernment antidiversion policy index constructed by Hall and Jones (1999).
This measure varies between [0,1] and takes higher values for governments with
more efective policies for supporting production
18
.
Growth is positively afected by financial liberalization and negatively by
bank crises under every specification of the model. As reported in column 4,
the efect of bank crises is indeed diferent across countries with good and bad
institutions. The term that controls for bank crises in institutionally developed
countries is strongly positive. Thus, the cost of banking crises in terms of
growth is reduced by good institutions. The interaction with capital account
openness, in column 3, is negligible.
As the interaction with credit market development in Column 1 shows,
financial liberalization restrains economic growth in countries with small
credit markets. Thus, studying the efects of capital account openness on
financial development might be of help in understanding the transmission to
economic growth. Column 2 shows that banking crises have a bigger impact
in countries with high levels of credit market development. In fact, if firms
rely more heavily on credit financing, they are more severely hurt by banking
crises.
Table 3b replicates the exercise of Table 3 using the equity market liber-
1
6
In
Tornell et al. this mechanism works to a diferent extent across tradables and
nontradables sectors. We leave this aspect aside of the analysis.
1
7
Financial development is measured by credit market depth, while the index of overall
economic development is taken from the classification in World Development Indicators.
1
8
The index is an equal-weighetd avarage of 5 variables: (i) law and order (ii) bu-
reaucratic quality (iii) corruption (iv) risk of expropriation (v)government repudiation of
contracts.
10
alization index. The most significant result, in column 5, points in the same
direction as column 5 in Table 3. Good institutions reduce the destructive
efects of bank crises.
Hall and Jones' (1999) indicator of institutional quality accounts mainly for
property right protection, i.e. the degree of private property protections against
government and elite expropriations. Inspired by Acemoglu and Johnson
(2003), we assess the role of institutions aimed at protecting private contracts.
Thus, we replicate the exercise in Tables 5 and 6 using the degree of efciency of
the judiciary as a diferent measure of institutional quality. This variable, built
by La Porta et al. (2003), captures the legal costs of contract enforcement and
takes values in [0,7].
The evidence in columns 1 and 2 of Table 6 shows that contractual pro-
tection does not bring heterogeneity in the efects of financial liberalization and
crises on growth, which remain respectively positive and negative.
3.3 Liberalization, crises and financial development
The evidence in the previous sections suggests that bank crises tend to
restrain growth, but to a lesser extent if good institutions and financial
openness help channelling funds into the economy. Moreover, column 4 in
Table 2 indicates that capital account liberalization becomes unin?uential for
growth, once we control for financial depth. These results suggest that the
efect of capital account liberalization on growth is generally positive, and is
possibly transmitted through the credit market. In this section, we assess how
financial development (FD) is afected by international liberal- ization and bank
crises. To this end, we estimate the following dynamic
system
?F D
it
= a?F D
it
÷
1
+ b?F lib
it
+ c?Bcr
it
+ g?interaction
it
+ ?u
t
+ ?e
it
F D
it
= a (F D
it
÷
1
) + b (F lib
it
) + c (Bcr
it
) + g (interaction
it
) + h
i
+ u
t
+ e
it
with two-step GMM. The coefcients in column 1 of Table 4 strongly support
the hypothesis that capital account liberalization boosts financial depth
19
.
The estimates in columns 2 and 4 show that financial liberalization has the
same efects across countries with diferent institutional and economic
development. Column 1 does not support the view that bank crises slow
1
9
This result is consistent with previous evidence by Levine (2001) and Klein and Olivei
(2000).
11
down the process of financial development
20
. However, column 3 suggests
that feedback from banking crises to credit market depth may indeed take
place, with the expected positive and negative signs, respectively in countries with
high and low degrees of property rights protection.
Columns 3 and 4 of Table 6 instead suggest that contractual protection
plays a role in shaping the efect of openness and bank crises on financial
depth. A good legal environment for business turns bank crises into expan-
sions of the credit markets, consistent with the "bumpy path" proposed by
Tornell et al. (2004). Vice-versa, where contractual rights are weak, credit
markets are restrained by both openness and banking crises.
4 Conclusion
This paper provides an enpirical evaluation of the efects of financial lib-
eralization and banking crises on growth. Our analysis accounts for the
interaction between liberalization and crises, and allows for unequal efects
across countries with diferent degrees of institutional and economic develop-
ment. We also investigate the transmission of these efects through financial
depth.
The overall lesson we draw from the results in section 3 is that the re-
moval of capital account restrictions boosts growth mainly through indirect
efects. In fact, financial liberalization has not only a beneficial impact on
financial development but also allows to smooth the destructive efects of
financial distress. Banking crises are indeed extremely harmful for economic
performance. The cost of crises is higher in countries with bad institutions, as
well as in the closed ones, while they have less impact in liberalized
economies and in countries with higher quality of institutions. The efect of
banking crises on growth is mainly a direct one, even though we show that
feedbacks on credit market development are also possible.
The positive efects of financial liberalization are robust to diferent def-
inition. In fact, we also show a positive relation between equity market
liberalization and growth. Our results, consistent with Bekaert et al.(2004),
point towards a direct efect of equity market integration. However, eq- uity
market openness and banking crises have strong opposite efects but do not
interact. This evidence can be partly reconciled with the mechanism
2
0
Demirguc-Kunt and Detragiache (1998) also show that financial liberalization tend to
push financial development while financial fragility slows down the process.
12
proposed by Tornell et al. (2004). In fact, firms that rely on credit may
be severely hurt by banking crises, and find it difcult to shift abruptly to
equity financing, even if they can sell shares on the international market. If
instead they have free access to international credit markets, they might raise
funds more easily there, and thus sufer less from domestic crises.
13
References
[1] Acemoglu, Daron and Simon Johnson (2003) "Unbundling institu-
tions", MIT Mimeo
[2] Acemoglu, Daron and Fabrizio Zilibotti (1997) "Was Prometeus Un-
bound by Chance? Risk, Diversification and Growth", Journal of Po-
litical Economy, 105, 709-752.
[3] Aizenmann, Joshua (2002) "Financial Opening: Evidence and Policy
Options", NBEP, wp 8900.
[4] Arellano, Manuel and Stephen Bond (1991) "Some Test of Specification
fo Panel Data: Monte-Carlo Evidence and Application to Employment
Equations", Review of Economic Studies, 58(2), 277-297.
[5] Arellano, Manuel and Olivia Bover (1995) "Another Look at The In-
strumental Variable Estimation of Errorr-Component Models", Journal of
Econometrics, 68, 29-52.
[6] Arteta, Carlos, Barry Eichengreen and Charles Wyplosz (2001) "When
Does Capital Account Liberalization Help More than Hurts?", Mimeo.
[7] Bacchetta, Philippe and Eric Van Wincoop (1998) "Capital Flow
to Emerging Markets: Liberalization, Overshooting, and Volatility." NBER
wp 6530.
[8] Barro, Robert J. (1991) "Economic Growth in a cross-section of coun-
tries", Quarterly Journal of Economics, 106 407-43.
[9] Barro, Robert J. (2001). "Determinants of Economic Growth: A Cross-
Country Empirical Study", Harvard Institute for International Devel-
opment, Development Discussion Paper No. 579.
[10] Barro, Robert J. (2001). "Economic Growth in East Asia Before and
After the Financial Crisis." NBER wp 8330.
[11] Barro, Robert, N. Gregory Mankiw and Xavier Sala-i-Martin (1995)
"Capital Mobility in Neoclassical Models of Growth." The American
Economic Review 85(1) 103-115
[12] Barro, Robert and Xavier Sala-i-Martin (1995) Economic Growth. MIT
Press.
14
[13] Barth, James R., Gerard Jr Caprio and Ross Levine (2000) "Bank-
ing Systems Around the Globe: Do regulation and Ownership Afect
Performance and Stability?", Mimeo.
[14] Barth, James R., Gerard Jr Caprio and Ross Levine (2000) "The Reg-
ulation and Supervision on Banks Around the World: a new Dataset.",
Mimeo.
[15] Beck, Thorsten, Asli Demirgüç-Kunt and Ross Levine (2000) "A new
database on financial development and structure", World Bank Eco-
nomic Review, September 2000, 597-605.
[16] Beck, Thorsten, Ross Levine and Norman Loayza (1999) "Financial
Intermediation and Growth: Causality and Causes." Mimeo.
[17] Beck, Thorsten, Ross Levine and Norman Loayza (2000) "Finance and
the Sources of Growth." Journal of Financial Economics 58.
[18] Beck, Thorsten and Ross Levine (2001) "Stock Markets, Banks, and
Growth: Correlation and Causality." Mimeo.
[19] Bekaert, Geert, Campbell R. Harvey and Christian Lundblad (2003)
"Does Financial Liberalization Spur Growth?" NBER wp 8245.
[20] Berglof, Erik and Patrick Bolton (2000) "The Great Divide and Beyond-
Financial Architecture in Transition." Mimeo.
[21] Blundell, Richard and Stephen Bond (1998) "Initial Conditions and Mo-
ment Restrictions in Dynamic Panel Data Models". Journal of Econo-
metrics, 87, 115-143.
[22] Bond, Stephen, Clive Bowsher and Frank Windmeijer (2001)
"Criterion-based inference for GMM in autoregressive panel data mod-
els", Economics Letters, 73, 379-388
[23] Caprio, Gerard and Daniela Klingebiel (2003) "Episodes of Systemic
and Borderline Financial Crises", Mimeo
[24] Caselli, Francesco, Gerardo Esquivel and Fernando Lefort (1996) "Re-
opening the Convergence Debate: a New Look at Cross-Country
Growth Empirics", Journal of Economic Growth, 1, 363-389.
15
[25] Chanda, Areendam (2001) "The In?uence of Capital Controls on Long
Run Growth: Where and How Much?". Brown University, Mimeo .
[26] Cull, Robert, Lemma W. Senbet and Marco Sorge (2000) "Deposit
Insurance and Financial Development." World Bank, Mimeo.
[27] De Gregorio, José and Jong-Wha Lee (2004) "Growth and Adjustment
in East Asia and Latin America", ADB Institute Research Paper.
[28] Demirguc-Kunt, Asli and Erica Detragiache (1998) "The Determinants
of Banking Crises in Developing and Developed Countries." IMF Staf
Papers 45.
[29] Demirguc-Kunt, Asli and Erica Detragiache (2000) "Does Deposit In-
surance Increase Banking System Stability?." World Bank, Mimeo.
[30] Demirguc-Kunt, Asli and Ross Levine (1999) "Bank_Based and
Market-Based Financial Systems: Cross-Country Comparisons." World
Bank, Mimeo.
[31] Dornbusch, Rudiger and Alejandro Reynoso (1989) "Financial Factors
in Economic Development." The American Economic Review, 79.
[32] Edison, Hali J., Michael W. Klein, Luca Ricci and Torsten Sloek (2002)
"Capital account liberalization and economic performance: survey and
synthesis", NBER Working Paper 9100.
[33] Edwards, Sebastian (1993) "Openness, Trade Liberalization, and
Growth in Developing Countries." Journal of Economic Literature 31.
[34] Edwards, Sebastian (1999) "The Length and Cost of Banking Crises",
IMF Working Paper 99/30.
[35] Edwards, Sebastian (2001) "Capital Mobility and Economic Perfor-
mance: Are Emerging Economies Diferent?" NBER wp 8076.
[36] Eichengreen, Barry and David Leblang (2003) "Capital Account Liber-
alization and Growth: Was Mr. Mahathir Right?" International Jour- nal of
Finance and Economics, 8: 205-224.
[37] Eichengreen, Barry and Michael Mussa (1998) "Capital Account Lib-
eralization and IMF". Financial and Development 35, IMF.
16
[38] Eichengreen, Barry, Michael Mussa, Giovanni Dell'Ariccia, Erica Detra-
giache, Gian Maria Milesi_Ferretti and Andrew Tweedie (1999) "Lib-
eralizing Capital Movements: Some Analytical Issues." IMF, Economic
Issues 17.
[39] Eichengreen, Barry (2001) "Capital Account Liberalization: What Do
the Cross-Country Studies Tell Us?" University of California, Berkeley,
Mimeo.
[40] Greenwood, Jeremy and Boyan Jovanovic (1990) "Financial Develop-
ment, Growth, and the Distribution Income." Journal of Political Econ- omy,
98.
[41] Grilli, Vittorio and Gian Maria Milesi-Ferretti (1995) "Economic Efect
and Structural Determinants of Capital Controls", IMF Staf Papers, Vol.
42, No. 3.
[42] Hall, R. and C. Jones (1999) "Why Do Some Countries Produce So
Much More Output per Worke Than Others?", Quarterly Journal of
Economics, 114, 83-116.
[43] Honohan, Patrick and Daniela Klingebiel, 2001 "Controlling Fiscal
Costs of Banking Crises", World Bank Policy Research Working Pa- per
2441.
[44] Islam, Nazrul, 1995 "Growth Empirics: A Panel Data Approach".
Quarterly Journal of Economics, 110, 1127-1170.
[45] Kaminsky, Graciela and Carmen Reinhart (1999) "The Twin Crises:
Causes of Banking and Balance-of-Payments Problems", American Eco-
nomic Review, 89(3) 473-500
[46] Kaminsky, Graciela and Sergio Schmukler, (2002) "Short-Run Pain,
Long-Run Gain: The Efects of Financial Liberalization", Mimeo.
[47] King, Robert G. and Ross Levine (1993) "Finance and Growth: Schum-
peter Might be Right." Quarterly Journal of Economics 108.
[48] Klein,Michael and Giovanni Olivei (1999) "Capital Account Liberaliza-
tion, Financial Depth, and Economic Growth." NBER wp 7384.
[49] Kraay, Aart (1998) "In Search of the Macroeconomic Efects of Capital
Account Liberalization" World Bank, Mimeo.
17
[50] La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer and
Robert W. Vishny (1997) "Law and finance", Journal of Political Econ- omy
106, 1113-1155.
[51] La Porta, Rafael, Florencio F. Lopez-de-Silanes and Andrei Shleifer I
(2003) "What Works in Securities Laws?" Harvard, Mimeo.
[52] Levine, Ross (1997) "Financial Development and Economic Growth:
Views and Agenda." Journal of Economic Literature 35.
[53] Levine, Ross (2000)."International Financial Liberalization and Eco-
nomic Growth." Review of International Economics
[54] Martin, Philippe and Helene Rey (2002). "Financial Globalization and
Emerging Markets: With or Without Crash?," NBER Working Papers 9288.
[55] McKinnon, Ronald I. (1988) "Monetary and Exchange Rate Policies
for International Financial Stability: A Proposal". The Journal of
Economic Perspectives, 2.
[56] Miniane, Jacques (2000) "A New Set of Measures on Capital Account
Restrictions", Johns Hopkins University, Mimeo.
[57] Prasad, Eswar, Kenneth Rogof, Shang-Jin Wei and M. Ayhan Kose
(2003) "Efects of Financial Globalization on Developing Countries:
some Empirical Evidence", IMF
[58] Quinn, Dennis (1997) "The Correlates of Change in International Fi-
nancial Regulation", The American Political Science Review 91.
[59] Rodrik, Dani (1998) "Who Needs Capital-Account Convertibility?"
Harvard University, Mimeo.
[60] Temple, Jonathan (1999) "The New Growth Evidence", Journal of Eco-
nomic Literature 37, 112-156.
[61] Tornell, Aaron, Frank Westermann and Lorenza Martinez (2004) "The
Positive Link Between Financial Liberalization Growth and Crises",
NBER wp 10293
[62] Wyplosz, Charles (2002) "How Risky is Financial Liberalization in the
Developing Countries?" Comparative Economic Studies, June, forth-
coming
18
Table A. Countries, Financial Liberalizationand Growth
Country #opIMF #opBHL #bc2 # bc1 Growth Country #opIMF #opBHL #bc2 # bc1 Growth
Algeria 0 0 2 0 1.192 Kenia 0 1 9 4 0.377
Argentina 3 1 10 0 0.497 Korea 0 1 3 0 5.828
Australia 12 0 0 4 2.021 Lesotho 0 0 0 12 1.545
Austria 5 0 0 0 2.296 Malawi 0 0 0 0 1.237
Bangladesh 0 1 10 0 2.174 Malaysia 21 1 3 4 4.020
Barbados 0 0 0 0 2.664 Mali 0 0 3 0 0.539
Belgium 21 0 0 0 2.039 Mauritius 0 1 0 1 4.257
Benin 0 0 3 0 0.768 Mexico 7 1 15 0 0.887
Bolivia 16 0 9 0 -0.295 Mozambique 0 0 9 0 -2.361
Botswana 0 1 0 2 5.102 Nepal 0 0 1 0 1.959
Brasil 0 1 7 0 1.200 Netherlands 19 0 0 0 1.960
Cameroon 0 0 11 0 0.132 NewZealand 12 1 0 4 0.802
Canada 21 0 3 0 1.844 Nicaragua 3 0 11 0 -4.073
Central Africa 0 0 24 0 -2.805 Niger 1 0 17 0 -1.321
Chile 0 1 7 0 3.459 Norway 1 0 7 0 2.704
Colombia 0 1 6 0 1.543 Pakistan 0 1 0 0 2.729
Congo 0 0 8 0 1.531 Panama 21 0 2 0 1.435
CostaRica 3 0 1 6 0.805 Papua NGuinea 0 0 0 11 -0.851
Cyprus 0 0 0 0 5.968 Paraguay 2 0 5 0 1.673
Denmark 8 0 0 6 1.838 Peru 9 1 8 0 -0.697
DominicanRep 0 0 0 0 2.722 Philippines 0 1 9 0 0.736
Ecuador 17 1 9 0 -0.045 Portugal 3 1 0 0 3.042
Egypt 0 1 5 5 3.661 Rwanda 0 0 0 9 0.084
19
Table A(cont'd). Countries, Financial LiberalizationandGrowth
Country #opIMF #opBHL #bc2 #bc1 Growth Country #opIMF #opBHL #bc2 #bc1 Growth
El Salvador 0 0 1 0 -0.036 Senegal 0 0 4 0 0.003
Fiji 0 0 0 0 1.216 Sierra Leone 0 0 10 0 -2.047
Finland 5 0 4 0 2.007 Singapore 18 0 0 1 5.486
France 6 0 0 2 1.843 SouthAfrica 0 1 0 12 -0.053
Gambia 5 0 0 7 -0.310 Spain 2 1 9 0 1.852
Germany 21 0 0 3 2.095 Sri Lanka 0 1 5 0 2.677
Ghana 0 1 8 3 0.212 Sweden 3 0 1 0 1.404
Greece 0 1 0 5 1.253 Switzerland 4 0 0 0 0.968
Guatemala 12 0 0 4 0.485 Syria 0 0 0 0 1.892
Haiti 0 0 0 0 4.066 Thailand 0 1 8 0 4.765
Honduras 8 0 0 0 0.098 Togo 0 0 3 0 -0.967
HongKong 21 0 0 6 4.622 Trinidad&Tobago 2 1 0 12 1.620
Iceland 0 1 0 3 2.158 Tunisia 0 1 0 5 2.483
India 0 1 0 7 3.298 Turkey 0 1 4 1 1.688
Indonesia 21 1 3 0 3.801 Uganda 0 0 6 0 1.719
Iran 3 0 0 0 0.504 United Kingdom 17 0 0 22 2.073
Ireland 4 0 0 0 4.324 United States 21 0 0 8 2.404
Israel 0 1 7 0 1.676 Uruguay 15 0 4 0 1.723
Italy 6 0 0 6 2.273 Venezuela 9 1 2 5 -1.046
Jamaica 0 1 6 0 -0.268 Zaire 0 0 0 0 -5.585
Japan 16 1 9 0 2.528 Zambia 0 0 1 0 -1.818
Jordan 0 1 0 2 2.141 Zimbabwe 0 1 5 0 0.200
20
Table B. Variables: definitions and sources
Variable Definition Availability Sources
y Beginning of period real per capita GDP yearly, 1975-99 Penn World Tables 6.1
sec 25 Percentage of population aged 25 or above 5-year, 1975-99 Barro and Lee (2001)
with some secondary education
grpop average yearly population growth rate yearly, 1975-99 Penn World Tables 6.1
gov government share of y yearly, 1975-99 Penn World Tables 6.1
inv investment share of y yearly, 1975-99 Penn World Tables 6.1
privo Private credit by deposit money banks yearly, 1975-99 Beck et al. (2003)
and other financial institutions to GDP
opIM F
opBHL
Bcr
Bcr012
Bcr012av
GADP
LDC
ef f _jud
Equals 0 if restrictions on capital account transactions
are in place, 1 otherwise. n-year period average
Equals 1 ever since the year of ofcial equity market
liberalization, 0 elsewhere. n-year period average
Equals 2 if systemic banking crises, 1 if non-systemic
crises, 0 if no crises have occurred in the year.
Equals 2 if systemic banking crises, 1 if non-systemic
crises, 0 if no crises have occurred in the period
Average of Bcr over the period
Government anti-diversion policy index. Accounts for: law
and order, burocratic quality, risk of expropriation, corruption,
government repudiation of contracts. Values in [0,1]
Dummy for developing countries
Assessment of the efciency and integrity of the legal
environment as it afects business, particularly foreign
firms. Values in [0,10]
yearly, 1975-99
yearly, 1980-99
yearly, 1975-99
average 1986-95
average 1980-83
AREAER, IMF
Bekaert et al. (2003)
Caprio and Klingebiel
(2003)
CK (2003)
CK (2003)
Hall and Jones
(1999)
WDI
La Porta et al
(2003), from ICR
21
Table 1. Financial Liberalization, Banking Crises, Financial Development and Growth
Mean equality tests - 90 countries
Open vs Open vs BC vs Systemic BC Non-systemic
Closed CA Closed SM No BC vs No BC BC vs No BC
Growth .008
---
(.002)
.016
---
(.002)
÷.019
---
÷.023
---
Financial Development .034
---
.039
---
(.003) (.003)
÷.05
(.005)
Bank Crises
(.009)
.009
(.007)
÷.039
---
(.
011)
÷.057
---
(.
017)
÷.001
(.011)
(.025)
÷.057
--
(.025)
Systemic BC
Non-Systemic BC
Period
÷.098
---
(.
018)
.107
---
(.021)
1975-99
÷.099
---
(.
021)
.043
---
(.017)
1980-99
1975-99
1975-99
1975-99
This table reports the diferences in mean between treated (open, bank crisis) and control (closed, no bank crisis)
groups, and their standard errors (in parenthesis).
---
and
--
indicate rejection of the null of zero-diference at 1
and 5 % significance level. The test is performed on annual data for the countries in Table A. The variables of
interest are the growth rate of real per capita GDP, the growth rate of credit to the private sector, and the 0-1
indicators of occurrence of bank crises.
22
Table 2. Financial Liberalization, Bank Crises and Growth
Dynamic Panel Data - System GMM
GMM GMM GMM GMM GMM GMM GMM GMM
y
t
÷5
.
(
954
.036)
.
(
92
9
)
.034
.
(
933
.033)
.
(
906
.034)
.
(
972
.035)
.
(
94
6
)
.032
.
(
94
6
)
.032
.
(
923
.037)
sec 25
grpop
÷.044
(.025)
.317
÷.016
(.028)
÷.021
(.028)
÷.015
(.027)
÷.045
(.032)
÷.004
(.029)
÷.003
(.029)
÷.002
(.029)
(.2.311)
÷1.284
(2.178)
÷1.401
(2.101)
(2.014)
÷.998
1.
.
563
)
(2 464
(2 168)
1.
.
156
1.142
(2.08)
(2 201)
1.
.
136
gov
÷.002
(.002)
.001
(.002)
.001
(.002)
.001
(.002) ÷.001
(.002)
.002
(.002)
.002
(.002)
.002
(.002)
inv
.
(
015
.002)
.
(
01
4
)
.002
.
(
014
.002)
.
(
013
.002)
.
(
015
.002)
.
(
01
3
)
.002
.
(
01
4
)
.002
.
(
013
.002)
privo
opIM F
opBHL
.037
(.046)
.
(
08
7
)
.041
.027
(.044)
.
(
086
.050)
.055
(.040)
.052
(.057)
Bcr012
.
(
012
.006)
.
(
01
3
)
.006
.011
(.007)
.
(
012
.006)
opIM F - Bcr012
opBHL - Bcr012
Countries
Period
m
2
Sargan
90
1975-99
.246
.418
÷.041
(.018)
89
1975-99
.119
.635
÷.044
(.017)
.064
(.039)
89
1975-99
.105
.677
÷.043
(.016)
89
1975-99
.100
.713
82
1980-99
.264 .160
÷.051 (.016)
81
1980-99
.201 .381
÷.054
(.019)
.002
(.006)
81
1980-99
.200
.541
÷
.
0
5
3
(
.
0
1
5
)
8
1
1
9
8
0
-
9
9
.
1
9
7
.
6
6
0
System-GMM estimates. Dependent variables: log and log-diference of real per capita GDP. Regressors are log and
log-diferences of: lagged real per capita GDP, secondary attainment, government and investments share of GDP,
indicators of financial liberalization and bank crises. Instruments: lagged levels for diferences, lagged diferences for
levels. Two-steps estimations. Coefcients and standard errors (in parenthesis) are from the first step.
5 and 10 per cent significance coefcients in bold and italics. P-values for Sargan overidentification test and m
2
test for second-order serial correlation of residuals are from the second step.
23
Table 3. Capital Account Liberalization, Bank Crises and Growth
Dynamic Panel Data - System GMM - Interactions
GMM GMM GMM GMM GMM GMM
opIM F
.
(
099
.045)
.
(
082
.041)
.
(
127
.052)
.07
4
(.043)
.
(
114
.054)
.078
(.044)
Bcr012
÷.049
(.017)
÷.049
(.020)
÷.047
(.018)
÷.144
(.052)
÷.049
(.018)
÷.060
(.019)
opIM F - privo
l
Bcr012 - privo
h
opIM F - GADP
l
Bcr012 - GADP
h
opIM F - LDC
Bcr012 - (1 ÷ LDC)
Countries
Period
m
2
Sargan
÷.115
(.06)
89
1975-99
.108
.670
.001
(.031)
89
1975-99
.091
.623
÷.214
(.149)
88
1975-99
.08
.568
.
(
13
4
)
.068
88
1975-99
.237
.300
÷.062
(.100)
89
1975-99
.091
.487
.058
(.043)
88
1975-99
.09
.405
System-GMM estimates. Dependent variables: log and log-diference of real per capita GDP.
Regressors are log and log-diferences of: lagged real per capita GDP, secondary attainment,
government and investments share of GDP, capital account liberalization, bank crises and
interactions with financial development, insitutional quality, economic development. Subscritps
l and h indicate that the variable is below and above cross-sectional average. Instruments: lagged
levels for diferences, lagged diferences. Two-steps estimations. Coefcients and standard errors
(in parenthesis) are from the first step. 5 and 10 per cent significant coefcients in bold and
italics. P-values for Sargan overidentification test and m
2
test for second-order serial
correlation of residuals are from the second step.
24
Table 3b. Equity Market Liberalization, Bank Crises and Growth
Dynamic Panel Data - System GMM - Interactions
GMM GMM GMM GMM GMM GMM
opBHL
.
(
012
.006)
.
(
010
.005)
÷.005
(.009)
(.006
.008
)
.006
(.011)
.
(
010
.005)
Bcr012
÷.048
(.015)
÷.067
(.020)
÷.056
(.016)
÷.195
(.052)
÷.053
(.016)
÷.063
(.015)
opBHL - privo
l
Bcr012 - privo
h
opBHL - GADP
l
Bcr012 - GADP
h
opBHL - LDC
Bcr012 - (1 ÷ LDC)
Countries
Period
m
2
Sargan
÷.006
(.007)
81
1980-99
.151
.486
.042
(.026)
81
1980-99
.208
.458
.057
(.037)
80
1980-99
.218
.245
.
(
14
8
)
.064
80
1980-99
.527
.342
.005
(.012)
81
1980-99
.194
.285
.056
(.035)
81
1980-99
.218
.306
System-GMM estimates. Dependent variables: log and log-diference of real per capita GDP.
Regressors are log and log-diferences of: lagged real per capita GDP, secondary attainment,
government and investments share of GDP, capital account liberalization, bank crises and
interactions with financial development, insitutional quality, economic development. Subscritps
l and h indicate that the variable is below and above cross-sectional average. Instruments: lagged
levels for diferences, lagged diferences. Two-steps estimations. Coefcients and standard errors
(in parenthesis) are from the first step. 5 and 10 per cent significant coefcients in bold and
italics. P-values for Sargan overidentification test and m
2
test for second-order serial
correlation of residuals are from the second step.
25
Table 4. Capital Account Liberalization, Bank Crises and
Financial Development - Dynamic Panel Data - System GMM
GMM GMM GMM GMM GMM
privo
t
÷1
opIM F
Bcr012
.725
(.078)
.516
(.134)
.02
.709
(.078)
.654
(.199)
.025
.731
(.080)
.469
(.158)
.719
(.086)
.584
(.217)
.021
.734
(.077)
.399
(.154)
(.063) (.063)
÷.390
(.224)
(.063)
÷.073
(.069)
opIM F - GADP
l
Bcr012 - GADP
h
opIM F - LDC
Bcr012 - (1 ÷ LDC)
Countries
Period
m
2
Sargan
79
75-99
.216
.394
÷.618
(.439)
78
75-99
.275
.501
.602
(.313)
78
75-99
.384
.451
÷.165
(.342)
79
75-99
.276
.432
.528
(.249)
79
75-99
.185
.411
System-GMM estimates. Dependent variables: log and log-diference of
private credit to GDP. Regressors are log and log-diferences of: lagged
private credit to GDP, capital account liberalization, bank crises and
interactions with financial development, insitutional quality,
economic
development. Subscritps l and h indicate that the variable is below and
above cross-section average. Instruments: lagged levels for diferences,
lagged diferences for levels. Two-steps estimations. Coefcients and
standard errors (in parenthesis) are from the first step. 5 and 10 per cent
significant coefcients in bold and italics. P-values for Sargan test and
m
2
test for second-order serial correlation of residuals from the second.
26
Table 5. Financial Liberalization, Bank Crises and Growth
Robustness analysis
opIM F Bcr012av opIM F - opBHL Bcr012av
Bcr012av
opBHL-
Bcr012av
sys ÷ GM M .
(
084
.039)
÷.037
(.024)
.
(
012
.006)
÷.049
(.022)
sys ÷ GM M
.022
(.043) ÷.048
(.025)
.13
3
(.072)
.014
(.0086)
÷.046
(.007)
÷.002
(.009)
OLS rows replicate Table 1 (columns 2-3, 6-7) with Bcr012av instead of Bcr012, FE
rows Table 2 (columns 2-3, 6-7), GLS Table 2b (columns 2-3, 6-7), dif-GMM Table 3
(columns 2-3, 6-7), sys-GMM Table 4 (colunms 2-3, 6-7). Standard errors within
parenthesis, 5% and 10%significant coefcients respectively in bold and italics.
27
Table 6. Capital Account Liberalization, Bank Crises
Financial Development and Growth
Growth Growth FD FD
opIM F .025 .061 .873 .444
(.040) (.027) (.118) (.118)
Bcr012
÷.034
(.013)
÷.040
(.026)
.109
(.093)
÷.126
(.125)
opIM F - eff _jud
l
Bcr012 - ef f _jud
h
.093
(.083)
.019
(.043)
÷1.001
(.448)
.500
(.184)
Estimates in column 1-2 replicate columns 4-5 of Table 5, with
eff _jud instead of GADP ; column 3-4 replicate columns 2-3
of Table 6. Standard errors in parenthesis, 5 and 10 per cent
significant coefcients respectively in bold and italics.
28
doc_419456733.docx
Financial Liberalization refers to reduction of any sort of regulations on the financial industry of a given country.
FINANCIAL STUDY FOR FINANCIAL
LIBERALIZATION, BANK CRISES AND GROWTH
Abstract
This paper studies the efects of financial liberalization and bank-
ing crises on growth. It shows that financial liberalization spurs on average economic growth.
Banking crises are harmful for growth, but to a lesser extent in countries with open financial
systems and good institutions. The positive efect of financial liberalization is robust to diferent
definitions. While the removal of capital account restrictions is efective by increasing
financial depth, equity market liberalization afects growth directly. The empirical analysis is
performed through GMM dynamic panel data estimations on a panel of 90 countries ob- served
in the period 1975-1999.
JEL classification: C23, F02, G15, O11.
Keywords: Capital account liberalization, equity market liberaliza-
tion, financial development, institutions, dynamic panel data.
-
1
1 Introduction
In the last two decades an increasing number of countries have eliminated
controls on international capital movements. However, the global economic
crises of recent years have led many economists to reconsider the beneficial
efects of financial liberalization on economic performance. Although the
issue has been widely debated, there are no conclusive results on the efects of
financial integration on growth
1
.
In theory, international financial liberalization softens financing con-
straints and improves risk-sharing, thereby fostering investments. It may also
have a positive impact on the functioning and development of finan- cial
systems, and on corporate governance
2
. These arguments suggest that we
should expect a positive relation between international financial liber-
alization and economic growth
3
. However, the presence of distortions may
reduce the positive efects of liberalization. In fact, information asymmetries may
lead to a bad allocation of capital, and weak financial and legal sys- tems
could induce capital?ights towards countries with better institutions. Moreover,
banking crises may come along with financial liberalization, as it is well
documented in the literature
4
.
Table 1 shows mean equality tests for growth, financial development
and the occurrence of banking crises across diferent treatment (open, bank
crises) and control (closed, no crises) groups of countries, observed annu- ally
between 1975 and 1999. The results suggest that countries without
restrictions on capital account or equity market transactions had, on aver-
age, higher growth rates and financial development (as measured by credit to
the private sector as a ratio of GDP). The occurrence of banking crises is
associated with lower growth rates and financial development. However, it is not
clear whether there is correlation between openness and the occurrence of bank
crises. When we consider an overall index there is no significant diference
in the frequency of crises between countries with and without re- strictions on
capital account transactions. The picture becomes clearer once we split the index
between "systemic" and "non-systemic" banking crises.
1
See Edison et al. (2003) for a review of the empirical literature.
2
See Klein and Olivei (2000) and Levine (2000) for empirical evidence on the positive
impact of financial liberalization on growth, .
3
Evidence on the positive relation between financial development and growth is pro-
vided by a large literature (see Demirguc-Kunt and Levine, 2001 for a survey). The results in
La Porta et al. (1999) suggest that good corporate governance spurs growth.
4
See Kaminsky and Reinhart (1999) and Aizenmann (2002) for a survey.
2
Open countries experienced a lower number of systemic crises but a higher
number of non-systemic crises. The higher frequency of non-systemic crises may
be a reason for the concern of economists and governments on the efects of
financial liberalization on economic performance.
In this paper we assess empirically the efects of international financial
liberalization and banking crises on growth. We admit the possibility that
banking crises come along with financial liberalization, as shown by previous
works and by row 5 of Table 1, and investigate their joint impact on growth
5
.
Kaminsky and Schmukler (2002) and Tornell et al. (2004) suggest in difer- ent
ways that institutional quality may matter at shaping the relationship between
financial liberalization, crises and long-run growth
6
. Therefore, we control for
institutions and their interactions with financial openness and crises. To
have a better understanding of the mechanism that links the variables of our
interest, we assess whether liberalization and crises afect growth through
financial depth. Also in this case, we control for institu- tional quality.
Inspired by the results in Acemoglu and Johnson (2003)
7
, we distinguish
between institutions aimed at contractual as opposed to property rights
protection.
The empirical analysis is performed on a panel dataset that covers 90
countries over the period 1975-1999. We adopt the Dynamic Panel Data
approach proposed by Arellano and Bover (1995) and Blundell and Bond
(1998). We use two indicators of financial liberalization, that distinguish
between capital account and equity market liberalization.
Our results show that capital account liberalization has a positive efect on
growth, once we control for banking crises, whose impact is negative. The
absence of capital account controls is good for growth because it fosters
financial development and mitigates the harmful efects of banking crises.
Capital account liberalization allows firms to raise funds more easily on the
5
Causality between financial liberalization and banking crises is left aside from our
empirical analysis.
6
Kaminsky and Schmukler (2002) show evidence that equity market liberalization
brings about financial chaos in the short-run, but has positive long-run efects, since it in-
duces changes in institutions supporting the functioning of the domestic financial market.
Tornell et al. (2004) suggest that liberalization and crises afect growth through financial
development; given financial openness, good institutions make bank crises less likely, and
foster capital in?ows.
7
Acemoglu and Johnson (2003) show that contractual protection afects financial struc-
ture more than property rights protection, but has limited efects on economic perfor-
mance. Vice versa, property right protection afects GDP growth, productivity and in-
vestments, but not the financial structure.
3
international financial markets, and thus sufer less from domestic crises.
Moreover, banking crises turn out to be less harmful for growth in countries
where property and contractual rights are better protected. Equity mar- ket
liberalization instead has a strong direct efect on growth and does not interact
with banking crises.
There are many contributions in the literature on the efects of financial
liberalization on long-run growth. Bekaert, Harvey and Lundblad (2003) is the
closest work to this paper. These authors as well consider both capital account
and equity market liberalization, and control for bank crises. They also allow
for heterogeneity in the efects of liberalization depending on cross-country
diferences in institutional quality. The main elements that distinguish our
contribution are the attention to the interaction between financial
liberalization and bank crises, the analysis of the mechanism that links them to
growth through financial development and the use of a diferent dynamic panel
data technique.
The remainder of the paper is organized as follows. Section 2 describes the
econometric model and the variables we used. Section 3 reports the
estimation results and comments on them. Section 4 councludes.
2 Data and empirical strategies
2.1 The econometric model
We assess the growth efects of financial liberalization and banking crises by
adding these variables to a dynamic version of the standard growth regres- sion
8
. We follow the dynamic panel data approach suggested by Arellano and
Bover (1995) and Bond and Blundell (1998)
9
. This methodology is pre- ferred to
the cross-sectional regressions because it allows to account for the impact of the
policy changes, imbedded in the indexes of financial liberaliza- tion and of crisis
episodes, on growth. This dynamic panel technique is also helpful to amend the
bias induced by omitted variables in cross-sectional es- timates, and the
inconsistency caused by endogeneity both in cross-sectional and static panel
(fixed and random efects) regressions.
We formulate the standard neoclassical growth model in a dynamic panel
8
See among others, Barro (1997) and Barro and Sala-i-Martin(1995).
9
The system-DPD methodology dominates the diference-DPD proposed by Arellano
and Bond (1991) because it amends problems of measurement error bias and weak in-
struments, arising from the persistence of the regressors (as pointed out by Bond et al.,
2001).
4
data form, and estimate the following dynamic system:
?y
it
= o?y
it
÷
1
+ |
0
?X
it
+ o?F lib
it
+ ¸?Bcr
it
+ ?v
t
+ ?
i,t
(1)
y
it
= oy
it
÷
1
+ |
0
X
it
+ oF lib
it
+ ¸Bcr
it
+ q
i
+ v
t
+ i,t, (2)
where time indexes refer to non-overlapping five-year periods. ?y
it
is the
average annual growth rate of real per capita GDP over five years. y
it
is
the logaritm of real per capita GDP, and the coefcient on its lag, o = e
5
ì
,
supports conditional convergence if it implies ì< 0. Variables indexed by
t ÷ 1 are observed at the beginning of the five-year period, and covariates
are expressed in period averages. Matrix X
it
contains determinants of GDP
growth, such as human capital, population growth and other factors that
account for diferent long-run per capita output across countries. F lib
i
(t+k,t)
and Bcr
i
(t+k,t
)
are indicators of financial liberalization and banking crises.
q
i
, v
t
and
it
are respectively the unobservable country- and time-specific
efects, and the error term. The presence of country efect in equation (2)
corrects the omitted variable bias. The diferences in equation (1) and the
instrumental variables estimation of the system are aimed at amending in-
consistency problems
10
. We instrument diferences of the endogenous and
predetermined variables with lagged levels in equation (1) and levels with
diferenced variables in equation (2). For instance, we take y
it
÷
3
as instru-
ment for ?y
it
÷
1
and F lib
it
÷
2
for ?F lib
it
in (1) and ?y
it
÷
2
as instrument for y
it
÷
1
and ?F lib
it
÷
1
for F lib
it
in (2). We estimate the system by General-
ized Method of Moments with moment conditions E[?y
it
÷
s
(
it
÷ it÷
1
)] =
0 for s > 2, and E[?z
it
÷
s
(
it
÷ it÷
1
)] = 0 for s > 2 on the predeter-
mined variables z, for equation (1); E[?y
i,t
÷
s
(q
i
+ c
i,t
)] = 0 and E[?z
i,t
÷s
(q
i
+ c
i,t
)] = 0 for s = 1 for equation (2). We treat all regressors as predeter-
mined. The validity of the instruments is guaranteed under the hypothesis
that
it
are not second order serially correlated. Coefcient estimates are
consistent and efcient if both the moment conditions and the no-serial cor-
relation are satisfied. We can validate the estimated model through a Sargan
test of overidentifying restrictions, and a test of second-order serial corre-
lation of the residuals. As pointed out by Arellano and Bond (1991), the
estimates from the first step are more efcient, while the test statistics from the
second step are more robust. Therefore, we will report coefcients and statistics
from the first and second step respectively.
1
0
See
Temple (1999) for a survey on the methodologies used in growth regressions.
5
2.2 Financial liberalization and financial fragility: the data
To explore the impact of financial liberalization and banking crises on growth we
need to measure these variables. The literature on financial liberalization has
proposed diferent indicators that difer along several directions. The major
distinctions are based on the de iure vs de facto definition criterion, the
characterization on a zero-one vs continuous scale, and the market they refer to.
In our analysis we construct an index of liberalization of both capital
account and equity market based on two diferent sources
11
. The first one is a
dummy variable provided by the IMF in its Annual Report on Exchange
Arrangements and Exchange Restrictions (AREAER), that is available for a
maximum of 212 countries starting from 1967
12
. This is the most com- monly
used measure of restrictions on international financial transactions. It takes
value 1 if a country has experienced restrictions on capital account transactions
during the year, and zero otherwise. Our yearly measure of financial
liberalization, opIM F, equals 1 and 0 when the IMF dummy is re- spectively 0
and 1. The second indicator is based on Bekaert et al.'s (2003) chronology of
ofcial equity market liberalization, that is available for 95 countries from 1980.
Our variable opBHL difers from opIM F because it only accounts for equity
market liberalization, but not for globalization of the credit market for
instance. Moreover, diferently from the AREAER, Bekaert et al.'s measure
does not contemplate policy reversals, so that a country is labeled as open
ever since its first year of liberalization. As the IMF-based indicators, it takes
value 1 and zero in case of internationally open and closed country-years,
respectively. Both opIM F and opBHL are
expressed as five-year averages, thereby taking values in the [0,1] interval.
There are alternative measures that are able to account for diferent de-
grees of liberalization instead of just the presence or absence thereof. Quinn's (1997)
index scores the intensity of capital account controls on a scale from 0 to 4 with
steps of 0.5. However, it is hardly suited for panel studies since it is available
for a significant number of countries only for four years, 1958, 1973, 1982 and
1988. Other contributions have used de facto measures, as
1
1
We
focus on de iure zero-one measures, that classify a country as financially liberalized
if there are no legal restrictions to international trade of financial instruments.
1
2
Classification methods have changed in 1996, so that there are 13 separate indexes
now, that can hardly be compared to the previous single indicator. Miniane (2000) har-
monized the classifications, though for a limited number of countries, and over a short
time span. Therefore, the last observation for opIMF in our dataset dates back to 1996.
6
data on international capital?ows as a ratio of GDP. The idea is that ac-
tual international capital?ows are a good proxy for the degree of financial
openness. A more comprehensive discussion on the available indicators can be
found in Edison et al. (2002).
Banking crises are subject to various classifications as well. As for liber-
alization, we adopt a zero-one anecdotal indicator of bank crises, proposed by
Caprio and Klingebiel (2001). The authors keep record of 117 systemic and 51
non-systemic crises occurred in 93 and 45 countries respectively, from the late
Seventies on. On a yearly base, our variable Bcr takes value 2 if the country
has experienced a systemic banking crisis, meaning that much or all of a bank's
capital has been exhausted; 1 if the banking crisis involved less severe losses;
and 0 otherwise. We use two alternative data reductions for robustness analysis:
Bcr012 takes value 2 if Bcr equals 2 at least once over the period, 1 if at least a
1 is scored, zero otherwise. Bcr012av instead accounts also for the duration of
crisis episodes, since it equals the period average of Bcr.
The other covariates in our growth regressions are variables commonly
accounted for in the empirical growth literature (see Barro, 1997), such as
secondary school attainment, the growth rate of population, government ex-
penditure and investments as a ratio of GDP. Other factors that we want to take
into account are financial development, proxied by the ratio of credit to the
private sector over GDP, and, at a further stage of the analysis, insti- tutional
quality, as measured by the government anti-diversion policy index (Hall and
Jones, 1999) and by the indicator of efciency of the judiciary system (see La
Porta et al. 2003). The first indicator mainly accounts for property rights
protection, while the other refers more to contractual rights.
The sample consists of data for a maximum of 90 countries over the
period 1975-1999 or 1980-1999 depending on the indicator of financial lib-
eralization adopted. Since keeping the larger sample gives us a longer time-
series in the panel analysis, we will go on reporting results from the 1975-99
sample for opIM F and from 1980-99 for opBHL. Since we average over non-
overlapping five-year periods, either four or five observations for each
country are available. More detail on the countries in our sample and on all
variables is given in the appendix.
7
3 Empirical evidence
3.1 Liberalization, banking crises and growth
Table 2 reports results from dynamic-panel estimations of the augmented
growth regression, which includes the usual control variables (initial GDP,
secondary school attainment, population growth, government spending and
investments over GDP) plus indicators of financial liberalization, financial
development, and banking crises. Consistently with the previous cross-
country growth studies (see Barro, 1997 and Barro and Sala-i-Martin, 1995), we
find significant evidence that countries with lower initial real per capita GDP
have grown faster than the initially richer ones, conditional on the other
variables. Our estimates imply a convergence rate of about 1.5% per year
13
.
Population growth and investments have the signs predicted by growth theory
(respectively negative and positive) in most of the estimates, though not always
significant.
Capital account openness has zero-efect on growth. Equity market lib-
eralization instead exhibits a significant positive coefcient (columns 1 and 5).
These results are in line with Bekeart et al.'s (2003) findings. Using the same
measure of financial liberalization, they show that equity market lib-
eralization significantly afects growth, while the relation between the IMF
measure and growth is fragile.
As a wide strand of literature (see Aizenmann, 2002 for a survey) points out,
the removal of restrictions on capital?ows may expose financial sys- tems to
turmoil and possibly crises
14
. If that is the case, the costly impact of
financial crises
15
, brought about by liberalization, could be responsible for the
coefcient estimates for opIM F in column 1. To control for this hy- pothesis, we
include the bank crisis indicator in the regression of columns 2. Once we control
for the occurrence of bank crises, the positive coefcient for opIM F becomes
significant. As expected, banking crises strongly restrain growth. Moreover,
the interaction between capital account openness and crises in column 3 is
positive. This suggests that, irrespective of whether
1
3
T he
convergence rate is computed as ˆ =
ln(
5ˆ
)
. o
ì
1
4
Among others, Kaminsky and Reinhart (1999) show that financial liberalization of-
ten precedes banking crises, Glick and Hutchison (1999) find that financially
liberalized emerging market economies are more likely to experience twin crises,
Demirguc-Kund and Detragiache (1998) show that banking crises occur more often in
liberalized financial systems.
1
5
A number of papers try to quantify the output costs of financial crises. See among
others Edwards (1999), Honohan and Klingebiel (2001), and De Gregorio and Lee (2004).
8
financial liberalization triggers instability in the banking sector, countries
without capital account restrictions are less prone to the negative efects of
banking crises than financially closed economies. Thus, capital account
liberalization has no strong direct efects on growth, but it is important to
mitigate the negative efects of banking crises.
The results are slightly diferent if we restrict the focus on equity market
liberalization. As in Bekeart et al., equity market openness and banking crises
have indeed strong opposite efects, respectively positive and negative, but the
introduction of the crises variable does not afect the efectiveness of equity
market liberalization on growth. Moreover, we find no interaction between the
two variables (see column 7). In fact, it is not so surprising that free
international equity trade alone can be less of help in case domestic banks get
into troubles. Firms that rely on credit may be severely hurt by banking crises,
and find it difcult to shift abruptly to equity financing, even if they can sell
shares on the international market. If instead they have free access to
international credit markets, they might raise funds more easily there, and thus
sufer less from domestic crises.
Opposite results are obtained by Eichengreen and Leblang (2002). They
show that the negative efects of domestic crises are neutralized by the pres- ence
of controls on capital controls. One reason could be that they use a diferent
indicator of crises (by Bordo et al., 2001) that encompasses both exchange and
banking crises.
As a robustness check, we replicate the estimations in Table 2 using an
indicator of Banking crises that accounts also for the duration of banking
crises, Bcr012av. Table 5 reports coefcients only for liberalization, bank crises
and their interaction. The results are not remarkably diferent from the ones
we obtained using the discrete crisis indicator.
3.2 Institutions, Financial Liberalization and Growth
After Hall and Jones' (1999) seminal paper, a wide strand of growth lit-
erature has focused on institutions as a primary determinant of economic
performance. Alfaro et al. (2004) have shown that institutions are an im-
portant determinant of capital in?ows. Tornell et al. (2004), in line with this
argument, suggest that in financially open countries institutional qual- ity
afects both the occurrence of banking crises and the extent of capital in?ows.
Banking crises may occur as a by-product of openness, as credit markets get
thicker, especially if there is a poor legal environment. In open
9
economies, the presence of good institutions facilitates capital in?ows from
abroad, when domestic banking crises reduce the amount of credit available to
firms
16
. As a result, banking crises are expected to be less harmful for
growth in countries where property and contractual rights are better pro-
tected. Symmetrically, financial liberalization might turn out to be growth-
restraining in countries with worse institutions. In order to assess empiri- cally
these implication we include interactive terms in our dynamic growth
regressions.
Table 3 shows results from system-GMM estimations that include the
same regressors in columns 1-3 of Table 2, plus the interactions of capital
account liberalization with indicators of institutional quality. We also inves-
tigate the relation between liberalization, financial development and
overalleconomic development
17
. Institutional quality is proxied here by the
gov- ernment antidiversion policy index constructed by Hall and Jones (1999).
This measure varies between [0,1] and takes higher values for governments with
more efective policies for supporting production
18
.
Growth is positively afected by financial liberalization and negatively by
bank crises under every specification of the model. As reported in column 4,
the efect of bank crises is indeed diferent across countries with good and bad
institutions. The term that controls for bank crises in institutionally developed
countries is strongly positive. Thus, the cost of banking crises in terms of
growth is reduced by good institutions. The interaction with capital account
openness, in column 3, is negligible.
As the interaction with credit market development in Column 1 shows,
financial liberalization restrains economic growth in countries with small
credit markets. Thus, studying the efects of capital account openness on
financial development might be of help in understanding the transmission to
economic growth. Column 2 shows that banking crises have a bigger impact
in countries with high levels of credit market development. In fact, if firms
rely more heavily on credit financing, they are more severely hurt by banking
crises.
Table 3b replicates the exercise of Table 3 using the equity market liber-
1
6
In
Tornell et al. this mechanism works to a diferent extent across tradables and
nontradables sectors. We leave this aspect aside of the analysis.
1
7
Financial development is measured by credit market depth, while the index of overall
economic development is taken from the classification in World Development Indicators.
1
8
The index is an equal-weighetd avarage of 5 variables: (i) law and order (ii) bu-
reaucratic quality (iii) corruption (iv) risk of expropriation (v)government repudiation of
contracts.
10
alization index. The most significant result, in column 5, points in the same
direction as column 5 in Table 3. Good institutions reduce the destructive
efects of bank crises.
Hall and Jones' (1999) indicator of institutional quality accounts mainly for
property right protection, i.e. the degree of private property protections against
government and elite expropriations. Inspired by Acemoglu and Johnson
(2003), we assess the role of institutions aimed at protecting private contracts.
Thus, we replicate the exercise in Tables 5 and 6 using the degree of efciency of
the judiciary as a diferent measure of institutional quality. This variable, built
by La Porta et al. (2003), captures the legal costs of contract enforcement and
takes values in [0,7].
The evidence in columns 1 and 2 of Table 6 shows that contractual pro-
tection does not bring heterogeneity in the efects of financial liberalization and
crises on growth, which remain respectively positive and negative.
3.3 Liberalization, crises and financial development
The evidence in the previous sections suggests that bank crises tend to
restrain growth, but to a lesser extent if good institutions and financial
openness help channelling funds into the economy. Moreover, column 4 in
Table 2 indicates that capital account liberalization becomes unin?uential for
growth, once we control for financial depth. These results suggest that the
efect of capital account liberalization on growth is generally positive, and is
possibly transmitted through the credit market. In this section, we assess how
financial development (FD) is afected by international liberal- ization and bank
crises. To this end, we estimate the following dynamic
system
?F D
it
= a?F D
it
÷
1
+ b?F lib
it
+ c?Bcr
it
+ g?interaction
it
+ ?u
t
+ ?e
it
F D
it
= a (F D
it
÷
1
) + b (F lib
it
) + c (Bcr
it
) + g (interaction
it
) + h
i
+ u
t
+ e
it
with two-step GMM. The coefcients in column 1 of Table 4 strongly support
the hypothesis that capital account liberalization boosts financial depth
19
.
The estimates in columns 2 and 4 show that financial liberalization has the
same efects across countries with diferent institutional and economic
development. Column 1 does not support the view that bank crises slow
1
9
This result is consistent with previous evidence by Levine (2001) and Klein and Olivei
(2000).
11
down the process of financial development
20
. However, column 3 suggests
that feedback from banking crises to credit market depth may indeed take
place, with the expected positive and negative signs, respectively in countries with
high and low degrees of property rights protection.
Columns 3 and 4 of Table 6 instead suggest that contractual protection
plays a role in shaping the efect of openness and bank crises on financial
depth. A good legal environment for business turns bank crises into expan-
sions of the credit markets, consistent with the "bumpy path" proposed by
Tornell et al. (2004). Vice-versa, where contractual rights are weak, credit
markets are restrained by both openness and banking crises.
4 Conclusion
This paper provides an enpirical evaluation of the efects of financial lib-
eralization and banking crises on growth. Our analysis accounts for the
interaction between liberalization and crises, and allows for unequal efects
across countries with diferent degrees of institutional and economic develop-
ment. We also investigate the transmission of these efects through financial
depth.
The overall lesson we draw from the results in section 3 is that the re-
moval of capital account restrictions boosts growth mainly through indirect
efects. In fact, financial liberalization has not only a beneficial impact on
financial development but also allows to smooth the destructive efects of
financial distress. Banking crises are indeed extremely harmful for economic
performance. The cost of crises is higher in countries with bad institutions, as
well as in the closed ones, while they have less impact in liberalized
economies and in countries with higher quality of institutions. The efect of
banking crises on growth is mainly a direct one, even though we show that
feedbacks on credit market development are also possible.
The positive efects of financial liberalization are robust to diferent def-
inition. In fact, we also show a positive relation between equity market
liberalization and growth. Our results, consistent with Bekaert et al.(2004),
point towards a direct efect of equity market integration. However, eq- uity
market openness and banking crises have strong opposite efects but do not
interact. This evidence can be partly reconciled with the mechanism
2
0
Demirguc-Kunt and Detragiache (1998) also show that financial liberalization tend to
push financial development while financial fragility slows down the process.
12
proposed by Tornell et al. (2004). In fact, firms that rely on credit may
be severely hurt by banking crises, and find it difcult to shift abruptly to
equity financing, even if they can sell shares on the international market. If
instead they have free access to international credit markets, they might raise
funds more easily there, and thus sufer less from domestic crises.
13
References
[1] Acemoglu, Daron and Simon Johnson (2003) "Unbundling institu-
tions", MIT Mimeo
[2] Acemoglu, Daron and Fabrizio Zilibotti (1997) "Was Prometeus Un-
bound by Chance? Risk, Diversification and Growth", Journal of Po-
litical Economy, 105, 709-752.
[3] Aizenmann, Joshua (2002) "Financial Opening: Evidence and Policy
Options", NBEP, wp 8900.
[4] Arellano, Manuel and Stephen Bond (1991) "Some Test of Specification
fo Panel Data: Monte-Carlo Evidence and Application to Employment
Equations", Review of Economic Studies, 58(2), 277-297.
[5] Arellano, Manuel and Olivia Bover (1995) "Another Look at The In-
strumental Variable Estimation of Errorr-Component Models", Journal of
Econometrics, 68, 29-52.
[6] Arteta, Carlos, Barry Eichengreen and Charles Wyplosz (2001) "When
Does Capital Account Liberalization Help More than Hurts?", Mimeo.
[7] Bacchetta, Philippe and Eric Van Wincoop (1998) "Capital Flow
to Emerging Markets: Liberalization, Overshooting, and Volatility." NBER
wp 6530.
[8] Barro, Robert J. (1991) "Economic Growth in a cross-section of coun-
tries", Quarterly Journal of Economics, 106 407-43.
[9] Barro, Robert J. (2001). "Determinants of Economic Growth: A Cross-
Country Empirical Study", Harvard Institute for International Devel-
opment, Development Discussion Paper No. 579.
[10] Barro, Robert J. (2001). "Economic Growth in East Asia Before and
After the Financial Crisis." NBER wp 8330.
[11] Barro, Robert, N. Gregory Mankiw and Xavier Sala-i-Martin (1995)
"Capital Mobility in Neoclassical Models of Growth." The American
Economic Review 85(1) 103-115
[12] Barro, Robert and Xavier Sala-i-Martin (1995) Economic Growth. MIT
Press.
14
[13] Barth, James R., Gerard Jr Caprio and Ross Levine (2000) "Bank-
ing Systems Around the Globe: Do regulation and Ownership Afect
Performance and Stability?", Mimeo.
[14] Barth, James R., Gerard Jr Caprio and Ross Levine (2000) "The Reg-
ulation and Supervision on Banks Around the World: a new Dataset.",
Mimeo.
[15] Beck, Thorsten, Asli Demirgüç-Kunt and Ross Levine (2000) "A new
database on financial development and structure", World Bank Eco-
nomic Review, September 2000, 597-605.
[16] Beck, Thorsten, Ross Levine and Norman Loayza (1999) "Financial
Intermediation and Growth: Causality and Causes." Mimeo.
[17] Beck, Thorsten, Ross Levine and Norman Loayza (2000) "Finance and
the Sources of Growth." Journal of Financial Economics 58.
[18] Beck, Thorsten and Ross Levine (2001) "Stock Markets, Banks, and
Growth: Correlation and Causality." Mimeo.
[19] Bekaert, Geert, Campbell R. Harvey and Christian Lundblad (2003)
"Does Financial Liberalization Spur Growth?" NBER wp 8245.
[20] Berglof, Erik and Patrick Bolton (2000) "The Great Divide and Beyond-
Financial Architecture in Transition." Mimeo.
[21] Blundell, Richard and Stephen Bond (1998) "Initial Conditions and Mo-
ment Restrictions in Dynamic Panel Data Models". Journal of Econo-
metrics, 87, 115-143.
[22] Bond, Stephen, Clive Bowsher and Frank Windmeijer (2001)
"Criterion-based inference for GMM in autoregressive panel data mod-
els", Economics Letters, 73, 379-388
[23] Caprio, Gerard and Daniela Klingebiel (2003) "Episodes of Systemic
and Borderline Financial Crises", Mimeo
[24] Caselli, Francesco, Gerardo Esquivel and Fernando Lefort (1996) "Re-
opening the Convergence Debate: a New Look at Cross-Country
Growth Empirics", Journal of Economic Growth, 1, 363-389.
15
[25] Chanda, Areendam (2001) "The In?uence of Capital Controls on Long
Run Growth: Where and How Much?". Brown University, Mimeo .
[26] Cull, Robert, Lemma W. Senbet and Marco Sorge (2000) "Deposit
Insurance and Financial Development." World Bank, Mimeo.
[27] De Gregorio, José and Jong-Wha Lee (2004) "Growth and Adjustment
in East Asia and Latin America", ADB Institute Research Paper.
[28] Demirguc-Kunt, Asli and Erica Detragiache (1998) "The Determinants
of Banking Crises in Developing and Developed Countries." IMF Staf
Papers 45.
[29] Demirguc-Kunt, Asli and Erica Detragiache (2000) "Does Deposit In-
surance Increase Banking System Stability?." World Bank, Mimeo.
[30] Demirguc-Kunt, Asli and Ross Levine (1999) "Bank_Based and
Market-Based Financial Systems: Cross-Country Comparisons." World
Bank, Mimeo.
[31] Dornbusch, Rudiger and Alejandro Reynoso (1989) "Financial Factors
in Economic Development." The American Economic Review, 79.
[32] Edison, Hali J., Michael W. Klein, Luca Ricci and Torsten Sloek (2002)
"Capital account liberalization and economic performance: survey and
synthesis", NBER Working Paper 9100.
[33] Edwards, Sebastian (1993) "Openness, Trade Liberalization, and
Growth in Developing Countries." Journal of Economic Literature 31.
[34] Edwards, Sebastian (1999) "The Length and Cost of Banking Crises",
IMF Working Paper 99/30.
[35] Edwards, Sebastian (2001) "Capital Mobility and Economic Perfor-
mance: Are Emerging Economies Diferent?" NBER wp 8076.
[36] Eichengreen, Barry and David Leblang (2003) "Capital Account Liber-
alization and Growth: Was Mr. Mahathir Right?" International Jour- nal of
Finance and Economics, 8: 205-224.
[37] Eichengreen, Barry and Michael Mussa (1998) "Capital Account Lib-
eralization and IMF". Financial and Development 35, IMF.
16
[38] Eichengreen, Barry, Michael Mussa, Giovanni Dell'Ariccia, Erica Detra-
giache, Gian Maria Milesi_Ferretti and Andrew Tweedie (1999) "Lib-
eralizing Capital Movements: Some Analytical Issues." IMF, Economic
Issues 17.
[39] Eichengreen, Barry (2001) "Capital Account Liberalization: What Do
the Cross-Country Studies Tell Us?" University of California, Berkeley,
Mimeo.
[40] Greenwood, Jeremy and Boyan Jovanovic (1990) "Financial Develop-
ment, Growth, and the Distribution Income." Journal of Political Econ- omy,
98.
[41] Grilli, Vittorio and Gian Maria Milesi-Ferretti (1995) "Economic Efect
and Structural Determinants of Capital Controls", IMF Staf Papers, Vol.
42, No. 3.
[42] Hall, R. and C. Jones (1999) "Why Do Some Countries Produce So
Much More Output per Worke Than Others?", Quarterly Journal of
Economics, 114, 83-116.
[43] Honohan, Patrick and Daniela Klingebiel, 2001 "Controlling Fiscal
Costs of Banking Crises", World Bank Policy Research Working Pa- per
2441.
[44] Islam, Nazrul, 1995 "Growth Empirics: A Panel Data Approach".
Quarterly Journal of Economics, 110, 1127-1170.
[45] Kaminsky, Graciela and Carmen Reinhart (1999) "The Twin Crises:
Causes of Banking and Balance-of-Payments Problems", American Eco-
nomic Review, 89(3) 473-500
[46] Kaminsky, Graciela and Sergio Schmukler, (2002) "Short-Run Pain,
Long-Run Gain: The Efects of Financial Liberalization", Mimeo.
[47] King, Robert G. and Ross Levine (1993) "Finance and Growth: Schum-
peter Might be Right." Quarterly Journal of Economics 108.
[48] Klein,Michael and Giovanni Olivei (1999) "Capital Account Liberaliza-
tion, Financial Depth, and Economic Growth." NBER wp 7384.
[49] Kraay, Aart (1998) "In Search of the Macroeconomic Efects of Capital
Account Liberalization" World Bank, Mimeo.
17
[50] La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer and
Robert W. Vishny (1997) "Law and finance", Journal of Political Econ- omy
106, 1113-1155.
[51] La Porta, Rafael, Florencio F. Lopez-de-Silanes and Andrei Shleifer I
(2003) "What Works in Securities Laws?" Harvard, Mimeo.
[52] Levine, Ross (1997) "Financial Development and Economic Growth:
Views and Agenda." Journal of Economic Literature 35.
[53] Levine, Ross (2000)."International Financial Liberalization and Eco-
nomic Growth." Review of International Economics
[54] Martin, Philippe and Helene Rey (2002). "Financial Globalization and
Emerging Markets: With or Without Crash?," NBER Working Papers 9288.
[55] McKinnon, Ronald I. (1988) "Monetary and Exchange Rate Policies
for International Financial Stability: A Proposal". The Journal of
Economic Perspectives, 2.
[56] Miniane, Jacques (2000) "A New Set of Measures on Capital Account
Restrictions", Johns Hopkins University, Mimeo.
[57] Prasad, Eswar, Kenneth Rogof, Shang-Jin Wei and M. Ayhan Kose
(2003) "Efects of Financial Globalization on Developing Countries:
some Empirical Evidence", IMF
[58] Quinn, Dennis (1997) "The Correlates of Change in International Fi-
nancial Regulation", The American Political Science Review 91.
[59] Rodrik, Dani (1998) "Who Needs Capital-Account Convertibility?"
Harvard University, Mimeo.
[60] Temple, Jonathan (1999) "The New Growth Evidence", Journal of Eco-
nomic Literature 37, 112-156.
[61] Tornell, Aaron, Frank Westermann and Lorenza Martinez (2004) "The
Positive Link Between Financial Liberalization Growth and Crises",
NBER wp 10293
[62] Wyplosz, Charles (2002) "How Risky is Financial Liberalization in the
Developing Countries?" Comparative Economic Studies, June, forth-
coming
18
Table A. Countries, Financial Liberalizationand Growth
Country #opIMF #opBHL #bc2 # bc1 Growth Country #opIMF #opBHL #bc2 # bc1 Growth
Algeria 0 0 2 0 1.192 Kenia 0 1 9 4 0.377
Argentina 3 1 10 0 0.497 Korea 0 1 3 0 5.828
Australia 12 0 0 4 2.021 Lesotho 0 0 0 12 1.545
Austria 5 0 0 0 2.296 Malawi 0 0 0 0 1.237
Bangladesh 0 1 10 0 2.174 Malaysia 21 1 3 4 4.020
Barbados 0 0 0 0 2.664 Mali 0 0 3 0 0.539
Belgium 21 0 0 0 2.039 Mauritius 0 1 0 1 4.257
Benin 0 0 3 0 0.768 Mexico 7 1 15 0 0.887
Bolivia 16 0 9 0 -0.295 Mozambique 0 0 9 0 -2.361
Botswana 0 1 0 2 5.102 Nepal 0 0 1 0 1.959
Brasil 0 1 7 0 1.200 Netherlands 19 0 0 0 1.960
Cameroon 0 0 11 0 0.132 NewZealand 12 1 0 4 0.802
Canada 21 0 3 0 1.844 Nicaragua 3 0 11 0 -4.073
Central Africa 0 0 24 0 -2.805 Niger 1 0 17 0 -1.321
Chile 0 1 7 0 3.459 Norway 1 0 7 0 2.704
Colombia 0 1 6 0 1.543 Pakistan 0 1 0 0 2.729
Congo 0 0 8 0 1.531 Panama 21 0 2 0 1.435
CostaRica 3 0 1 6 0.805 Papua NGuinea 0 0 0 11 -0.851
Cyprus 0 0 0 0 5.968 Paraguay 2 0 5 0 1.673
Denmark 8 0 0 6 1.838 Peru 9 1 8 0 -0.697
DominicanRep 0 0 0 0 2.722 Philippines 0 1 9 0 0.736
Ecuador 17 1 9 0 -0.045 Portugal 3 1 0 0 3.042
Egypt 0 1 5 5 3.661 Rwanda 0 0 0 9 0.084
19
Table A(cont'd). Countries, Financial LiberalizationandGrowth
Country #opIMF #opBHL #bc2 #bc1 Growth Country #opIMF #opBHL #bc2 #bc1 Growth
El Salvador 0 0 1 0 -0.036 Senegal 0 0 4 0 0.003
Fiji 0 0 0 0 1.216 Sierra Leone 0 0 10 0 -2.047
Finland 5 0 4 0 2.007 Singapore 18 0 0 1 5.486
France 6 0 0 2 1.843 SouthAfrica 0 1 0 12 -0.053
Gambia 5 0 0 7 -0.310 Spain 2 1 9 0 1.852
Germany 21 0 0 3 2.095 Sri Lanka 0 1 5 0 2.677
Ghana 0 1 8 3 0.212 Sweden 3 0 1 0 1.404
Greece 0 1 0 5 1.253 Switzerland 4 0 0 0 0.968
Guatemala 12 0 0 4 0.485 Syria 0 0 0 0 1.892
Haiti 0 0 0 0 4.066 Thailand 0 1 8 0 4.765
Honduras 8 0 0 0 0.098 Togo 0 0 3 0 -0.967
HongKong 21 0 0 6 4.622 Trinidad&Tobago 2 1 0 12 1.620
Iceland 0 1 0 3 2.158 Tunisia 0 1 0 5 2.483
India 0 1 0 7 3.298 Turkey 0 1 4 1 1.688
Indonesia 21 1 3 0 3.801 Uganda 0 0 6 0 1.719
Iran 3 0 0 0 0.504 United Kingdom 17 0 0 22 2.073
Ireland 4 0 0 0 4.324 United States 21 0 0 8 2.404
Israel 0 1 7 0 1.676 Uruguay 15 0 4 0 1.723
Italy 6 0 0 6 2.273 Venezuela 9 1 2 5 -1.046
Jamaica 0 1 6 0 -0.268 Zaire 0 0 0 0 -5.585
Japan 16 1 9 0 2.528 Zambia 0 0 1 0 -1.818
Jordan 0 1 0 2 2.141 Zimbabwe 0 1 5 0 0.200
20
Table B. Variables: definitions and sources
Variable Definition Availability Sources
y Beginning of period real per capita GDP yearly, 1975-99 Penn World Tables 6.1
sec 25 Percentage of population aged 25 or above 5-year, 1975-99 Barro and Lee (2001)
with some secondary education
grpop average yearly population growth rate yearly, 1975-99 Penn World Tables 6.1
gov government share of y yearly, 1975-99 Penn World Tables 6.1
inv investment share of y yearly, 1975-99 Penn World Tables 6.1
privo Private credit by deposit money banks yearly, 1975-99 Beck et al. (2003)
and other financial institutions to GDP
opIM F
opBHL
Bcr
Bcr012
Bcr012av
GADP
LDC
ef f _jud
Equals 0 if restrictions on capital account transactions
are in place, 1 otherwise. n-year period average
Equals 1 ever since the year of ofcial equity market
liberalization, 0 elsewhere. n-year period average
Equals 2 if systemic banking crises, 1 if non-systemic
crises, 0 if no crises have occurred in the year.
Equals 2 if systemic banking crises, 1 if non-systemic
crises, 0 if no crises have occurred in the period
Average of Bcr over the period
Government anti-diversion policy index. Accounts for: law
and order, burocratic quality, risk of expropriation, corruption,
government repudiation of contracts. Values in [0,1]
Dummy for developing countries
Assessment of the efciency and integrity of the legal
environment as it afects business, particularly foreign
firms. Values in [0,10]
yearly, 1975-99
yearly, 1980-99
yearly, 1975-99
average 1986-95
average 1980-83
AREAER, IMF
Bekaert et al. (2003)
Caprio and Klingebiel
(2003)
CK (2003)
CK (2003)
Hall and Jones
(1999)
WDI
La Porta et al
(2003), from ICR
21
Table 1. Financial Liberalization, Banking Crises, Financial Development and Growth
Mean equality tests - 90 countries
Open vs Open vs BC vs Systemic BC Non-systemic
Closed CA Closed SM No BC vs No BC BC vs No BC
Growth .008
---
(.002)
.016
---
(.002)
÷.019
---
÷.023
---
Financial Development .034
---
.039
---
(.003) (.003)
÷.05
(.005)
Bank Crises
(.009)
.009
(.007)
÷.039
---
(.
011)
÷.057
---
(.
017)
÷.001
(.011)
(.025)
÷.057
--
(.025)
Systemic BC
Non-Systemic BC
Period
÷.098
---
(.
018)
.107
---
(.021)
1975-99
÷.099
---
(.
021)
.043
---
(.017)
1980-99
1975-99
1975-99
1975-99
This table reports the diferences in mean between treated (open, bank crisis) and control (closed, no bank crisis)
groups, and their standard errors (in parenthesis).
---
and
--
indicate rejection of the null of zero-diference at 1
and 5 % significance level. The test is performed on annual data for the countries in Table A. The variables of
interest are the growth rate of real per capita GDP, the growth rate of credit to the private sector, and the 0-1
indicators of occurrence of bank crises.
22
Table 2. Financial Liberalization, Bank Crises and Growth
Dynamic Panel Data - System GMM
GMM GMM GMM GMM GMM GMM GMM GMM
y
t
÷5
.
(
954
.036)
.
(
92
9
)
.034
.
(
933
.033)
.
(
906
.034)
.
(
972
.035)
.
(
94
6
)
.032
.
(
94
6
)
.032
.
(
923
.037)
sec 25
grpop
÷.044
(.025)
.317
÷.016
(.028)
÷.021
(.028)
÷.015
(.027)
÷.045
(.032)
÷.004
(.029)
÷.003
(.029)
÷.002
(.029)
(.2.311)
÷1.284
(2.178)
÷1.401
(2.101)
(2.014)
÷.998
1.
.
563
)
(2 464
(2 168)
1.
.
156
1.142
(2.08)
(2 201)
1.
.
136
gov
÷.002
(.002)
.001
(.002)
.001
(.002)
.001
(.002) ÷.001
(.002)
.002
(.002)
.002
(.002)
.002
(.002)
inv
.
(
015
.002)
.
(
01
4
)
.002
.
(
014
.002)
.
(
013
.002)
.
(
015
.002)
.
(
01
3
)
.002
.
(
01
4
)
.002
.
(
013
.002)
privo
opIM F
opBHL
.037
(.046)
.
(
08
7
)
.041
.027
(.044)
.
(
086
.050)
.055
(.040)
.052
(.057)
Bcr012
.
(
012
.006)
.
(
01
3
)
.006
.011
(.007)
.
(
012
.006)
opIM F - Bcr012
opBHL - Bcr012
Countries
Period
m
2
Sargan
90
1975-99
.246
.418
÷.041
(.018)
89
1975-99
.119
.635
÷.044
(.017)
.064
(.039)
89
1975-99
.105
.677
÷.043
(.016)
89
1975-99
.100
.713
82
1980-99
.264 .160
÷.051 (.016)
81
1980-99
.201 .381
÷.054
(.019)
.002
(.006)
81
1980-99
.200
.541
÷
.
0
5
3
(
.
0
1
5
)
8
1
1
9
8
0
-
9
9
.
1
9
7
.
6
6
0
System-GMM estimates. Dependent variables: log and log-diference of real per capita GDP. Regressors are log and
log-diferences of: lagged real per capita GDP, secondary attainment, government and investments share of GDP,
indicators of financial liberalization and bank crises. Instruments: lagged levels for diferences, lagged diferences for
levels. Two-steps estimations. Coefcients and standard errors (in parenthesis) are from the first step.
5 and 10 per cent significance coefcients in bold and italics. P-values for Sargan overidentification test and m
2
test for second-order serial correlation of residuals are from the second step.
23
Table 3. Capital Account Liberalization, Bank Crises and Growth
Dynamic Panel Data - System GMM - Interactions
GMM GMM GMM GMM GMM GMM
opIM F
.
(
099
.045)
.
(
082
.041)
.
(
127
.052)
.07
4
(.043)
.
(
114
.054)
.078
(.044)
Bcr012
÷.049
(.017)
÷.049
(.020)
÷.047
(.018)
÷.144
(.052)
÷.049
(.018)
÷.060
(.019)
opIM F - privo
l
Bcr012 - privo
h
opIM F - GADP
l
Bcr012 - GADP
h
opIM F - LDC
Bcr012 - (1 ÷ LDC)
Countries
Period
m
2
Sargan
÷.115
(.06)
89
1975-99
.108
.670
.001
(.031)
89
1975-99
.091
.623
÷.214
(.149)
88
1975-99
.08
.568
.
(
13
4
)
.068
88
1975-99
.237
.300
÷.062
(.100)
89
1975-99
.091
.487
.058
(.043)
88
1975-99
.09
.405
System-GMM estimates. Dependent variables: log and log-diference of real per capita GDP.
Regressors are log and log-diferences of: lagged real per capita GDP, secondary attainment,
government and investments share of GDP, capital account liberalization, bank crises and
interactions with financial development, insitutional quality, economic development. Subscritps
l and h indicate that the variable is below and above cross-sectional average. Instruments: lagged
levels for diferences, lagged diferences. Two-steps estimations. Coefcients and standard errors
(in parenthesis) are from the first step. 5 and 10 per cent significant coefcients in bold and
italics. P-values for Sargan overidentification test and m
2
test for second-order serial
correlation of residuals are from the second step.
24
Table 3b. Equity Market Liberalization, Bank Crises and Growth
Dynamic Panel Data - System GMM - Interactions
GMM GMM GMM GMM GMM GMM
opBHL
.
(
012
.006)
.
(
010
.005)
÷.005
(.009)
(.006
.008
)
.006
(.011)
.
(
010
.005)
Bcr012
÷.048
(.015)
÷.067
(.020)
÷.056
(.016)
÷.195
(.052)
÷.053
(.016)
÷.063
(.015)
opBHL - privo
l
Bcr012 - privo
h
opBHL - GADP
l
Bcr012 - GADP
h
opBHL - LDC
Bcr012 - (1 ÷ LDC)
Countries
Period
m
2
Sargan
÷.006
(.007)
81
1980-99
.151
.486
.042
(.026)
81
1980-99
.208
.458
.057
(.037)
80
1980-99
.218
.245
.
(
14
8
)
.064
80
1980-99
.527
.342
.005
(.012)
81
1980-99
.194
.285
.056
(.035)
81
1980-99
.218
.306
System-GMM estimates. Dependent variables: log and log-diference of real per capita GDP.
Regressors are log and log-diferences of: lagged real per capita GDP, secondary attainment,
government and investments share of GDP, capital account liberalization, bank crises and
interactions with financial development, insitutional quality, economic development. Subscritps
l and h indicate that the variable is below and above cross-sectional average. Instruments: lagged
levels for diferences, lagged diferences. Two-steps estimations. Coefcients and standard errors
(in parenthesis) are from the first step. 5 and 10 per cent significant coefcients in bold and
italics. P-values for Sargan overidentification test and m
2
test for second-order serial
correlation of residuals are from the second step.
25
Table 4. Capital Account Liberalization, Bank Crises and
Financial Development - Dynamic Panel Data - System GMM
GMM GMM GMM GMM GMM
privo
t
÷1
opIM F
Bcr012
.725
(.078)
.516
(.134)
.02
.709
(.078)
.654
(.199)
.025
.731
(.080)
.469
(.158)
.719
(.086)
.584
(.217)
.021
.734
(.077)
.399
(.154)
(.063) (.063)
÷.390
(.224)
(.063)
÷.073
(.069)
opIM F - GADP
l
Bcr012 - GADP
h
opIM F - LDC
Bcr012 - (1 ÷ LDC)
Countries
Period
m
2
Sargan
79
75-99
.216
.394
÷.618
(.439)
78
75-99
.275
.501
.602
(.313)
78
75-99
.384
.451
÷.165
(.342)
79
75-99
.276
.432
.528
(.249)
79
75-99
.185
.411
System-GMM estimates. Dependent variables: log and log-diference of
private credit to GDP. Regressors are log and log-diferences of: lagged
private credit to GDP, capital account liberalization, bank crises and
interactions with financial development, insitutional quality,
economic
development. Subscritps l and h indicate that the variable is below and
above cross-section average. Instruments: lagged levels for diferences,
lagged diferences for levels. Two-steps estimations. Coefcients and
standard errors (in parenthesis) are from the first step. 5 and 10 per cent
significant coefcients in bold and italics. P-values for Sargan test and
m
2
test for second-order serial correlation of residuals from the second.
26
Table 5. Financial Liberalization, Bank Crises and Growth
Robustness analysis
opIM F Bcr012av opIM F - opBHL Bcr012av
Bcr012av
opBHL-
Bcr012av
sys ÷ GM M .
(
084
.039)
÷.037
(.024)
.
(
012
.006)
÷.049
(.022)
sys ÷ GM M
.022
(.043) ÷.048
(.025)
.13
3
(.072)
.014
(.0086)
÷.046
(.007)
÷.002
(.009)
OLS rows replicate Table 1 (columns 2-3, 6-7) with Bcr012av instead of Bcr012, FE
rows Table 2 (columns 2-3, 6-7), GLS Table 2b (columns 2-3, 6-7), dif-GMM Table 3
(columns 2-3, 6-7), sys-GMM Table 4 (colunms 2-3, 6-7). Standard errors within
parenthesis, 5% and 10%significant coefcients respectively in bold and italics.
27
Table 6. Capital Account Liberalization, Bank Crises
Financial Development and Growth
Growth Growth FD FD
opIM F .025 .061 .873 .444
(.040) (.027) (.118) (.118)
Bcr012
÷.034
(.013)
÷.040
(.026)
.109
(.093)
÷.126
(.125)
opIM F - eff _jud
l
Bcr012 - ef f _jud
h
.093
(.083)
.019
(.043)
÷1.001
(.448)
.500
(.184)
Estimates in column 1-2 replicate columns 4-5 of Table 5, with
eff _jud instead of GADP ; column 3-4 replicate columns 2-3
of Table 6. Standard errors in parenthesis, 5 and 10 per cent
significant coefcients respectively in bold and italics.
28
doc_419456733.docx