World Development Vol. 97, pp.

1–13, 2017
0305-750X/Ó 2017 Elsevier Ltd. All rights reserved.

www.elsevier.com/locate/worlddev
http://dx.doi.org/10.1016/j.worlddev.2017.03.018

Financial Inclusion, Bank Concentration, and Firm Performance

LISA CHAUVET a,b,c,d and LUC JACOLIN e,*
a
IRD, Paris, France
b
PSL Research University, Paris, France
c
Université Paris-Dauphine, France
d
DIAL, Paris, France
e
Banque de France, Paris, France
Summary. — This study focuses on the impact of financial inclusion and bank concentration on the performance of firms in developing
and emerging countries. Using firm-level data for a sample of 55,596 firms in 79 countries, we find that financial inclusion, i.e., the dis-
tribution of financial services across firms, has a positive impact on firm growth. This positive impact is magnified when bank markets
are less concentrated, a proxy for more competition among banks. We also find that more competitive banks favor firm growth only at
high levels of financial inclusion, while bank concentration is particularly favorable to foreign and state-owned firms and increases firm
growth at low levels of financial inclusion. In countries with limited financial deepening, the quality of the banking system (financial
inclusion and bank competition) may be as important in promoting firm performance as its overall size.
Ó 2017 Elsevier Ltd. All rights reserved.

Key words — financial inclusion, bank concentration, firm performance

1. INTRODUCTION credit in GDP, the positive finance/growth link disappears

and a case for ‘‘too much finance” may be made (Arcand,
Access to credit has been identified as one of the main obsta- Berkes, & Panizza, 2015). Turning to developing countries,
cles to the development of private sector in developing coun- beyond caveats stemming from the large size of the informal
tries. Since the early 1990s, a string of theoretical and sector (Guerineau & Jacolin, 2014), the question has been,
empirical research studies have shown the existence of a posi- conversely, to determine whether there is a case for ‘‘not
tive relationship between financial development, firm perfor- enough finance” where undersized financial sectors, usually
mance, and economic growth. Financial development has bank-led with little or no financial market development, play
been shown to improve the proportion of innovative and virtually no role in boosting economic growth, let alone corpo-
productivity-enhancing investment projects, to reduce transac- rate growth or productivity (Henderson, Papageorgiou, &
tion costs, and more generally to improve the allocation of Parmeter, 2013; Méon & Weill, 2010; Deidda & Fattouh,
capital and risk management. The literature has so far mostly 2002). Rioja and Valev (2004) find that in countries with a
focused on financial deepening effects, i.e., the volume of credit share of private credit in GDP lower than 14%, financial devel-
available to the private sector. opment has little effect on economic growth. Some specific
This article builds on the existing literature to try and deter- weaknesses of developing countries, such as poor institutions
mine whether the ‘‘quality” of financial development affects (Demetriades & Hook Law, 2006), insufficient financial com-
firm performance. We examine two dimensions of the quality petition due to political deadlock (Rajan & Zingales, 2003),
of financial development. The first is the financial inclusion of and high inflation (Rousseau & Wachtel, 2002), have been
firms. The idea here is to distinguish between volume (financial highlighted in the literature as dampening or suppressing the
depth) and the distribution of credit across firms (the share of finance–growth relationship.
firms in an industry with access to credit). The second dimen- The second strand of the literature to which our paper
sion that we examine is the banking market structure, which relates deals more specifically with how financial inclusion
may affect the volume as well as the price of credit. Examining affects economic outcomes. The international agenda has
aspects of financial development other than financial deepen- recently brought to light the significant role played by financial
ing – namely financial inclusion and bank concentration – inclusion, that is, the extent to which households or firms have
may provide additional insight into the mechanisms through access to financial products and services. In addition to
which financial development affects developing countries. supply-side indicators of access to finance such as branch
This paper relates to three strands of the literature on finan- density, the number of ATMs or, more recently, market
cial development and economic performance. The first exami-
nes how financial deepening affects economic growth. Since
the 2009 financial and economic crisis, the literature has * We wish to thank Agnès Dufour and Emilie Debels for excellent research
focused on the possibility of a non-linear relationship between assistance. We also thank Florian Leon, and the participants in the Ferdi-
economic activity and financial deepening, especially in devel- Banque de France seminar and of the AEL German Economic
oped countries, where large financial sectors may face dimin- Association conference for useful comments. This paper is a product of
ishing returns (Philippon & Reshef, 2013), divert resources the Franc Zone and Development Financing Studies Division
from other productive sectors (Deidda, 2006), or increase the (COMOZOF) of the Banque de France. It reflects the opinions of the
volatility of economic activity (Easterly, Islam, & Stiglitz, authors and does not necessarily express the views of the Banque de
2001; Loayza & Rancière, 2006). Empirical estimates show France. The usual disclaimers apply. Final revision accepted: March 9,
that, above thresholds ranging from 80% to 110% of private 2017.
1
2 WORLD DEVELOPMENT

penetration of mobile phones as a proxy for mobile banking, a crises than concentrated ones, where higher profits provide
consensus has been reached to measure financial inclusion by buffers during crises and reduce incentives for excessive
the share of households or firms that have access to financial risk-taking. The authors also point out that a lower number
services (GPFI, 2013) and is now regularly surveyed by inter- of banks may facilitate bank supervision especially in coun-
national organizations (IMF Financial Access Survey, Findex tries with low administrative capacity. Finally, entering for-
database). The literature suggests that the impact of financial eign banks may adopt cream-skimming strategies, i.e.,
deepening on growth may depend on the degree of financial targeting large- and low-risk exporting companies and
inclusion. Abdmoulah and Jelili (2013) show for example that sovereign debt. Such strategies may force domestic banks
non-linearities between growth and financial development can out of the market, thereby constraining credit to riskier
be explained by access to finance, measured by the density of and more opaque firms, and leading to a decline of aggre-
branches, which acts as a regime switching-trigger. Some stud- gate credit (Detragaiche, Tressel, & Gupta, 2008). As shown
ies find that the impact of financial inclusion on growth by Claessens and Horen (2014), the effect of foreign bank
depends on firms’ access to credit rather than households’ entry thus yields different outcomes depending on the char-
(Beck, Büyükkarabacak, Rioja, & Valev, 2012), most notably acteristics of the host country, in particular the quality of its
by reducing the ‘‘financing gap” faced by small- and medium- regulatory framework, the importance of information asym-
sized firms and industries (GPFI, 2011). 1 Financial inclusion metries and the nationality of ownership (international
reduces liquidity constraints and encourages investment. The banks from the OECD as opposed to pan-African groups
distribution of credit across firms at the sectoral level therefore for instance). These authors find that foreign bank entry
has important effects on the industrial structure, competition, may have a negative impact on credit in low-income coun-
or the degree of informality in the sector, particularly in low- tries, where they have a limited market share and where
income countries (Beck, Demirguc-Kunt, & Maksimovic, large information asymmetries limit the ability of banks to
2005). In some cases, the biggest or foreign-owned firms lend.
may reap most of the benefits of financial development, while Our paper aims to disentangle the impact of financial deep-
smaller and locally owned firms do not and in some cases are ening, financial inclusion, and the price and efficiency effects
even crowded out from financing, as shown by Harrison and induced by changes in market structure in the banking sector.
McMillan (2003) in Côte d’Ivoire. We use stacked World Bank Enterprise Surveys (WBES) for a
Finally, the last strand of the literature to which this work set of 79 developing countries over the period 2006-14. Our
can be related examines how competition in banking market dependent variable is firm sales growth. We measure financial
(bank concentration, market power) influences the diversity, depth at the country level using the share of private credit in
profitability of banking services, and firm performance GDP. Bank concentration is also measured at the country
(Beck et al., 2012; Northcott, 2004). According to the tradi- level using the market share of the three largest banks in terms
tional market power view, bank competition can affect the of balance sheet size. Both variables are available from the
efficiency of the financial sector, the quality of the products World Bank Global Financial Development Database. Finan-
and the degree of innovation in that sector. Love and cial inclusion is measured for each country, at the industry
Martı́nez Perı́a (2015) and Leon (2015) find that more com- level, as the share of firms with access to credit. In our baseline
petition 2 increases firms’ access to credit. Claessens and estimates, access to credit is measured as having a loan from a
Laeven (2005) also show that bank competition favors the financial institution, but we test the robustness of the results to
growth of sectors that are highly dependent on bank financ- alternative measures such as having an overdraft facility or
ing. The literature also emphasizes in particular the role financing part of the working capital with bank credit. Along
played by foreign banks in lowering the cost of banking ser- with the three financial development variables – depth, inclu-
vices and credit (Demirguc-Kunt, Levine, & Min, 1998; De sion, and bank concentration – we also include the interaction
la Torre, Martı́nez Perı́a, & Schmukler, 2010). Market struc- term of financial inclusion and bank concentration in order to
ture effects may be particularly large in developing and understand to what extent the impact on growth of the entry
emerging countries where lower efficiency and the high cost of new banks into the market depends on wider access to
of credit magnify the gains expected from the diffusion of credit among firms, i.e., whether these two dimensions are
international best practices (Hermes & Lensink, 2003). complementary or not.
Entering banking markets dominated for instance by large We show that, in developing and emerging countries, where
public banks may also yield important efficiency gains by financial deepening may be too limited to have a significant
lowering prevailing interest margins (Demirguc-Kunt, macroeconomic impact on growth, access to credit by a larger
Laeven, & Levine, 2004). proportion of firms at the sectoral level enhances firm growth,
However, the literature also highlights that bank competi- except when the banking sector is very concentrated. We also
tion may not necessarily yield the expected efficiency gains find that bank concentration may help firm performance, but
or favor financial inclusion. The information hypothesis this effect turns negative as the level of economic development
argues that in a less competitive market, banks are more and financial inclusion rise. This suggests that policies that
inclined to invest in information acquisition (Hauswald & increase the access of firms to financial products and services
Marquez, 2006). In that case, competition may decrease may be as instrumental in boosting the firm performance
access to credit by reducing incentives for banks to invest and economic development as the promotion of financial mar-
in assessing credit worthiness of opaque borrowers. kets or financial deepening alone.
Cetorelli and Gambera (2001) show that, while exercising The paper is structured as follows. Section 2 presents the
an overall depressing effect on growth, bank concentration model, while the data are presented in Section 3. Then in
has a positive effect on the growth of sectors that are most Section 4 we present the baseline results along with some
dependent on external financing by facilitating access to robustness checks. In Section 5 we discuss the heterogeneity
credit of younger firms. Furthermore, Beck, De Jonghe, of the impact given specific firm characteristics in order to
and Schepens (2013) show that less concentrated banking explore the potential crowding-out effects. Finally in Section 6
markets may also be less stable and more vulnerable to we conclude.
 FINANCIAL INCLUSION, BANK CONCENTRATION, AND FIRM PERFORMANCE 3

2. MODEL of firms with extreme growth rates. Our specification draws on

Beck et al. (2005) and Harrison, Lin, and Xu (2014). We
In this article, we explore the role of financial inclusion in include the size of the firm, SIZEi;k;j;t , which can take three val-
firm performance, assuming that financial inclusion affects ues: one when the firm is small (less than 20 employees), two
firm performance differently according to the degree of bank when it employs between 20 and 100 employees, and three
competition, approximated by bank concentration. We thus when it is large (more than 100 employees). We also control
estimate an econometric model of the following form: for the kind of ownership by including a dummy variable
which is equal to one when part (or all) of the firm is owned
GROWTH i;k;j;ðt;t3Þ ¼ a þ bX i;k;j;t þ cY j;ðt;t3Þ by a foreign entity, FOREIGNi;k;j;t , as well as a dummy vari-
þ dCONCENTRATION j;ðt;t3Þ able which is equal to one when part (or all) of the firm is
owned by the State, STATEi;k;j;t . Our specification also
þ kINCLUSION k;j;t includes a dummy variable, EXPORTi;k;j;t , which is equal to
þ hCONCENTRATION j;ðt;t3Þ one when the firm exports part of its production either directly
or indirectly (as a supplier of an exporting firm). Finally, we
 INCLUSION k;j;t þ lj þ sk;t þ ei;k;j;t ð1Þ include the logarithm of lagged sales (deflated and converted
into US dollars) to account for catching-up effects.
where GROWTH i;k;j;ðt;t3Þ is sales growth of firm i, in industry k,
In all specifications we also control for firms’ access to
country j. The growth rates are computed over three years,
credit. Our main measure is a dummy variable which is equal
between t and t-3. Xi;k;j;t is a set of time-varying firm-level char-
to one if the firm currently has a loan from a financial institu-
acteristics, including the initial value of sales. Yj;ðt;t3Þ is a set of tion, LOANi;k;j;t . As an alternative measure, we also use a
country-level variables averaged over the three years for which dummy variable which is equal to one if the firm has an over-
firm growth is computed. Yj;ðt;t3Þ includes country size, draft facility, OVERDRAFTi;k;j;t . Finally, the last alternative
income per capita (lagged i.e., averaged over the period span- measure of firms’ access to finance is a dummy variable which
ning from t-3 to t-6), as well as an indicator of control of cor- is equal to one if the firm finances part of its working capital
ruption. 3 Yj;ðt;t3Þ also includes the degree of financial with bank credit, WC-BANKi;k;j;t .
development as proxied by the share of private credit in Table 1 presents the basic descriptive statistics of these firm-
GDP. Eq. 1 accounts for country fixed effects, lj , as well as level data. Our sample of firms is mostly composed of large
industry x year dummies, sk;t , which are included to account formal firms. One quarter is outward-looking, exporting either
for time-varying heterogeneity within industries. Following directly or indirectly. Ten and one percent are foreign or state-
Moulton (1990), the standard errors are clustered at the coun- owned respectively. Among our sample of firms, 40% currently
try level. 4 have a loan, 47% have an overdraft facility, and 37% finance
Our model includes the interaction term of bank concentra- part of their working capital using a bank credit. This confirms
tion (measured at the country level) with the industry-level that access to overdraft facilities represents the first main step
measure of financial inclusion, CONCENTRATIONj;ðt;t3Þ x toward financial inclusion for firms in developing and emerg-
INCLUSIONk;j;t . This interaction term is intended to capture ing countries.
whether and to what extent banking market structure affects
the impact of financial inclusion on firm performance. (b) Country-industry-level variables

We measure financial inclusion at the country-industry level.

3. DATA Using standard international measures of financial inclusion,
we construct financial inclusion, INCLUSION-LOANk;j;t , as
In order to estimate Eqn. (1), we combine country-level the share of firms in industry k of country j which have a bank
financial characteristics with firm-level characteristics for a loan. Our intention is to capture the more or less even distri-
set of 79 developing countries. We use the stacked World Bank bution of credit among firms at the country-sector level.
Enterprise Surveys (WBES) which cover the 2006-14 period Unlike a firm’s individual access to credit, which depends on
(repeated cross-section). Our sample is composed of more than its own risk characteristics (and the bank’s choice to grant
55,000 firms. The sample of countries, years, and firms is pre- access to credit to an individual client), country-sectoral finan-
sented in Table 6 in Appendix A. The baseline estimations cial inclusion is mainly a function both of the characteristics of
only include developing and emerging countries. We exclude sectoral financing needs, which are likely to be similar across
the high-income countries from the baseline sample. 5 Conflict countries (Rajan & Zingales, 1998), and each country’s finan-
and post-conflict countries are also excluded from the main cial development. The WBES data cover 23 sectors and the
sample. Including both the high-income and conflict and distribution of firms across sectors is presented in Table 7 in
post-conflict countries does not alter the results (see below). 6 Appendix A.
Table 1 presents some descriptive statistics of
(a) Firm-level variables INCLUSION-LOANk;j;t . On average, the share of firms with
a loan in a specific sector is 40%, but the variance is quite
Firm-level data in local currencies have been deflated using large, with some country-sectors having no loans at all, while
the same base year (100 = 2005) and converted into US dol- in others all firms have a loan. We also test the robustness of
lars. GDP deflators and exchange rates are obtained from the results using INCLUSION-OVERDRAFTk;j;t , which
the IMF’s International Financial Statistics (IFS). Each sur- reflects the share of firms in the industry that have an over-
vey of the WBES includes information on the sales in the year draft facility, which is slightly higher than the share of firms
preceding the survey, as well as three years before. This allows with a loan. We also test the robustness of our results using
us to compute the annual average growth rate of sales over a third definition of inclusion, namely the share of firms in
three years for each available survey, GROWTHi;k;j;ðt;t3Þ . As the industry-country that finance part or all of their working
it is commonly done in the literature, we drop the one percent capital with bank credit, INCLUSION-WC-BANKk;j;t .
4 WORLD DEVELOPMENT

Table 1. Summary statistics

Variables N Mean sd Min Max
Firm-level variables
GROWTHi;k;j;ðt;t3Þ 55,596 0.05 0.35 1.00 4.71
SALESi;k;j;t3 Logarithm 55,596 13.08 2.48 2.62 28.62
STATEi;k;j;t Dummy 55,596 0.01 0.12 0.00 1.00
FOREIGNi;k;j;t Dummy 55,596 0.10 0.30 0.00 1.00
EXPORTi;k;j;t Dummy 55,596 0.24 0.43 0.00 1.00
SIZEi;k;j;t 55,596 1.76 0.80 0.00 3.00
LOANi;k;j;t Dummy 55,596 0.40 0.49 0.00 1.00
OVERDRAFTi;k;j;t Dummy 55,120 0.47 0.50 0.00 1.00
WC-BANKi;k;j;t Dummy 53,480 0.37 0.48 0.00 1.00
Country-industry-level variables
INCLUSION-LOANk;j;t 2,221 0.40 0.30 0.00 1.00
INCLUSION-OVERDRAFTk;j;t 2,203 0.45 0.33 0.00 1.00
INCLUSION-WC-BANKk;j;t 1,943 0.33 0.28 0.00 1.00
Country-level variables
FINDEVj;ðt;t3Þ %GDP 120 0.32 0.23 0.00 1.30
INCOMEj;ðt3;t6Þ Logarithm 120 7.36 1.04 5.17 9.42
POPULATIONj;ðt;t3Þ Logarithm 120 16.05 1.67 11.35 21.01
CORRUPTIONj;ðt;t3Þ 120 0.39 0.58 1.42 1.37
CONCENTRATIONj;ðt;t3Þ 120 0.74 0.20 0.25 1.00
CONCENTRATION-5j;ðt;t3Þ 102 0.82 0.16 0.33 1.00
CONCENTRATION-Lernerj;ðt;t3Þ 97 0.27 0.11 0.02 0.60
Note on the number of observations (N):
– 55,596 is the total number of firms in the baseline sample.
– 2221 is the total number of country-sector-years in the baseline sample.
– 120 is the total number of country-years (79 countries with one to three years for each), see Table 6 in Appendix A.

(c) Country-level variables Financial Development Data (GFDD) web page. In Appendix
B we present the specific formula used by the World Bank to
The estimations control for a large array of country-level compute the Lerner index.
time-varying factors. We include the logarithm of income Table 1 presents some descriptive statistics of the country-
per capita lagged one period to avoid endogeneity problems level variables used in our model. On average, financial devel-
as much as possible, INCOMEj;ðt3;t6Þ . We also control for opment represents 32% of GDP. This average hides a large
the size of the market using the logarithm of the population, heterogeneity across countries. For example, 10% of the firms
POPULATIONj;ðt;t3Þ . Finally, we include a proxy for the in the baseline sample are located in countries with financial
quality of economic institutions using an indicator of the con- development lower than 12% of GDP, while another 10% of
trol of corruption, CORRUPTIONj;ðt;t3Þ . This indicator is the firms are located in countries with financial development
computed by Kaufmann, Kraay, and Mastruzzi (2011) and higher than 70% of GDP. Bank concentration is rather high
ranges from weak (2.5) to strong (+2.5) control of corrup- in our sample, with the three largest banks representing on
tion. We also include a measure of financial development, average 74% of the bank market. Overall, the developing
FINDEVj;ðt;t3Þ , which is defined as the share of private credit and emerging countries in our sample display lower financial
from financial banks and other financial institutions in GDP, development and higher bank concentration than developed
and averaged over three years. countries.
Our main variable of interest is bank concentration, which is
measured at the country level using three different variables.
The first two are market structure indicators: the market share 4. THE IMPACT OF FINANCIAL INCLUSION AND
of the three largest banks in terms of balance sheet size – BANK CONCENTRATION ON FIRM GROWTH
CONCENTRATIONj;ðt;t3Þ –, the market share of the five lar-
gest banks – CONCENTRATION-5j;ðt;t3Þ . The third one, the (a) Baseline results
Lerner index – CONCENTRATION-Lernerj;ðt;t3Þ – is a non-
structural indicator which proxies bank market power. This The estimation of Eqn. (1) is reported in Table 2. Column
latter variable is defined as the difference between output (1) of Table 2 is estimated using the OLS estimator, with coun-
prices and marginal costs (relative to prices). Prices are calcu- try and sector-year dummies and standard errors clustered at
lated as total bank revenue over assets, whereas marginal costs the country level. The firm-level control variables suggest that
are obtained by taking the first derivative from an estimated there is a catch-up effect – firms with low levels of sales in t – 3
translogarithmic cost function with respect to output. Higher have better growth prospects in t. State-owned firms display
values of the Lerner index indicate less bank competition. It lower growth rates, while foreign-owned companies and larger
is calculated following Demirguc-Kunt and Martinez Peria and exporting firms perform better. Turning to the country-
(2010)’s methodology using underlying bank-by-bank data level control variables, column (1) suggests that firm perfor-
from Bankscope. Financial development and bank concentra- mance is negatively associated with the level of development,
tion variables are all obtained from the World Bank Global in line with the idea of a catching-up effect. Country size
 FINANCIAL INCLUSION, BANK CONCENTRATION, AND FIRM PERFORMANCE 5

Table 2. Baseline estimates of the impact of financial inclusion and bank concentration on firm growth
Baseline Aggregation of firm-level controls Firm fixed effect
(1) (2) (3)
INCLUSION-LOANk;j;t 0.366*** 0.390*** 0.942**
(2.86) (3.07) (2.70)
INCLUSION-LOANk;j;t x CONCENTRATIONj;ðt;t3Þ 0.511*** 0.498*** 1.066**
(3.03) (2.85) (2.11)
CONCENTRATIONj;ðt;t3Þ 0.433*** 0.404*** 0.993***
(3.86) (3.85) (2.85)
FINDEVj;ðt;t3Þ 0.339 0.373 0.628**
(1.45) (1.59) (2.19)
LOANi;k;j;t 0.0344*** 0.00350 0.0555***
(3.14) (0.27) (4.30)
LNSALESi;k;j;t3 0.0826*** 0.0783*** 0.123***
(4.97) (5.91) (5.72)
STATEi;k;j;t 0.0357*** 0.136 0.103
(3.24) (1.30) (1.03)
FOREIGNi;k;j;t 0.0354*** 0.0814 0.0111
(3.01) (1.64) (0.47)
EXPORTSi;k;j;t 0.0664** 0.0292 0.0525***
(2.01) (0.62) (2.99)
SIZEi;k;j;t 0.110*** 0.0793** 0.0628**
(4.21) (2.29) (3.30)
LNINCOMEj;ðt3;t6Þ 0.802*** 0.775*** 0.227***
(2.84) (2.86) (3.14)
LNPOPULATIONj;ðt;t3Þ 0.599 0.535 0.677
(1.05) (0.95) (0.78)
CORRUPTIONj;ðt;t3Þ 0.101 0.101 0.0217
(0.64) (0.67) (0.11)
Observations 55,596 55,596 7,490
Countries 79 79 25
Country dummies yes yes no
Firm fixed effect no no yes
Sector-year dummies yes yes no
Year dummies no no yes
ðaÞ
Threshold in CONCENTRATIONj;ðt;t3Þ 0.716 0.783 0.884
ðbÞ
Threshold in INCLUSION-LOANk;j;t 0.847 0.811 0.932
Robust standard errors clustered at the country level. t-Student in parenthesis. ***p<0.01, **p<0.05, *p<0.1.
(a) Level of CONCENTRATIONj;ðt;t3Þ above which INCLUSION-LOANk;j;t has a negative impact.
(b) Level of INCLUSION-LOANk;j;t above which CONCENTRATIONj;ðt;t3Þ has a negative impact.

and the control of corruption are not significantly correlated Bank concentration, measured at the country level,
with firm growth for our sample of countries. CONCENTRATIONj;ðt;t3Þ , has a positive effect on firm
Regarding our variables of interest, column (1) in Table 2 growth. Combined with the negative coefficient of the interac-
suggests that firm growth is positively correlated with firms’ tion term between financial inclusion and bank concentration,
access to credit, LOANi;k;j;t . It is also positively correlated with the last row of Table 2 shows that bank concentration has a
financial inclusion measured at the sector level, positive effect on firm growth for levels of financial inclusion
INCLUSION-LOANk;j;t . The impact of financial inclusion lower than 85% of firms with a bank loan. Only 1,093 firms
on firm growth is however dampened by bank concentration, out of the 55,596 belong to a sector with such a high level of
which can be interpreted as a bank market structure effect: less financial inclusion. This suggests that only for a highly inclu-
concentrated bank markets are associated with wider access to sive financial system does bank concentration have a negative
financial services by a larger spectrum of enterprises. Bank effect on firm performance (the market power view). For most
competition also has various efficiency effects such as driving of the firms of the sample bank concentration has a positive
the price of credit down and improving financial infrastruc- effect on their growth performance. This positive effect of bank
ture. We interpret the negative effect of the interaction term concentration (for levels of financial inclusion lower than 85%)
of financial inclusion with bank concentration as showing may appear counter-intuitive at first sight, since bank compe-
the positive effects of bank competition on both firms’ access tition has various efficiency effects on credit markets (lower
to credit and the price of credit. As indicated in the second prices, better financial infrastructure, more financial inclu-
row from the bottom of Table 2, our estimates suggest that sion). This result is consistent with the information hypothesis
when bank concentration is higher than 72% financial inclu- which suggests that banks are more inclined to invest in infor-
sion at the sector level has a negative effect on firm perfor- mation acquisition in a more concentrated market. This result
mance. In our sample, 19,288 firms (34%) are located in may also be explained by suboptimal banking market condi-
countries that display bank concentration above this thresh- tions in developing and emerging countries. In infant bank
old. markets there may only be market space for a limited number
6 WORLD DEVELOPMENT

of financial institutions and new entrants may impair the sol- developing and emerging countries, with two points in time
vency and stability of the banking sector, which in turn may for each firm. The countries for which we have the information
constrain the supply of credit. Second, as shown by as to whether the firm was interviewed twice are marked with a
Detragiache et al. (2008), entering banks may in some coun- star in Table 6 in Appendix A. Column (3) of Table 2 suggests
tries adopt cream-skimming strategies by focusing on large that accounting for firm-level time-unvarying unobservable
private and public borrowers instead of developing a wider cli- heterogeneity does not alter the conclusions of our baseline
ent base, hence driving profits and credit down. Competition estimate (even at the cost of restricting the sample of firms
effects may depend on the nationality of entering banks (local, and countries).
regional, or OECD-based), which determines their knowledge
of local markets and risk-taking. Finally, new bank entry may
impede credit in countries with poor regulatory frameworks 5. ROBUSTNESS CHECKS
and where information asymmetries are large.
Thus our results favor the pragmatic view that bank compe- In what follows we provide further robustness checks of the
tition in itself is not necessarily conducive to higher firm baseline results presented in Table 2 along two dimensions: (1)
growth (market power view), in part because the sources of the measures of inclusion and concentration; and (2) sample
imperfect competition prevailing in poorer countries may dif- dependence.
fer from one country to the other. Our results do not validate Table 3 re-estimates our baseline estimation (regression (1)
any particular view of what kind of market imperfections may in Table 2) using alternative measures of financial inclusion
prevent bank competition from affecting firm growth posi- (one based on overdrafts and the other based on the share
tively: different national bank markets may be affected by dif- of working capital financed with bank credit) and bank con-
ferent combinations of market imperfections (large centration (one based on the concentration of the five main
information asymmetries, lack of investment in information banks and the other based on a Lerner index). Panel A of
technologies, insufficient international financial integration), Table 3 uses the same measure of bank concentration as in
or bank strategies (market cream-skimming, credit rationing the baseline specification (market share of the three main
in unfavorable business climates). However, our results sug- banks). The first column, where inclusion is defined by having
gest that what matters most is not only the level of competi- a bank loan, merely reproduces our baseline result. When in
tion but also its modalities, in particular to what extent it is columns (2) and (3) we switch to the two alternative measures
associated with more financial inclusion: an inclusive banking of inclusion the results are unaltered. Then in Panel B we
system, one which reaches a wider spectrum of firms and switch to measuring bank concentration using the market
maybe also encourages firms to enter the formal sector, share of the five main banks (instead of the three main banks),
appears to be more efficient in spurring economic growth. but again the results are unaltered whatever the measure of
Finally, financial deepening, measured at the country level financial inclusion used.
by the share of private credit in the economy, Finally in Panel C, we use a non-structural measure of bank
FINDEVj;ðt;t3Þ , is not found to significantly affect firm perfor- concentration based on a Lerner index. This drastically
mance, which, combined with our previous results, suggests reduces the sample of firms, which drops to around 48,000.
that what matters most is not the aggregate level of credit from However, the results are very similar to those obtained in
the banking sector but how it is distributed among firms. This Panels A and B, but are less robust, as illustrated by the loss
also suggests that the price effect of bank competition may be in significance of the coefficient of financial inclusion in
larger than its quantity effect: in developing countries, an column (1).
increase in the volume of private credit is not necessarily asso- Table 4 presents the results when the baseline estimates
ciated with more firms having access to credit, which may is tested for sample dependence. In this Table we measure
explain why FINDEVj;ðt;t3Þ has no significant correlation with financial inclusion using the share of firms with
firm growth in our analysis. a loan, INCLUSION-LOANk;j;t , and bank concentration
In columns (2) and (3) of Table 2 we test the robustness of as the market share of the three main banks,
the baseline estimate with respect to the potential endogeneity CONCENTRATIONj;ðt;t3Þ .
of the firm-level controls. We do this in two ways. First, we re- We first add to the sample the post-conflict countries
aggregate the firm-level controls at the country-sector-year (column (1)) and the high-income countries (column (2)).
level, as is usually performed in the literature (Harrison The results are very similar to the baseline estimate presented
et al., 2014). We thus replace the firm-level variables with in Table 2. Then in columns (3) to (5) we sequentially drop the
the average value of initial sales and size of firms belonging countries from Sub-Saharan Africa, Asia, and Latin America.
to the same industry, and with the share of firms in the indus- Again the results are unaltered suggesting that they are not
try which are foreign-owned, state-owned, and exporting. We driven by one of the sub-regions in the sample. Finally, in
do not re-aggregate LOANi;k;j;t (whether the firms have a loan the last column, we only keep manufacturing firms, but again
from a financial institution) at the industry level because it is the results are very similar, both in terms of significance and
already captured in INCLUSION-LOANk;j;t . Except for magnitude. The last two rows of Table 4 suggest that the
SIZEk;j;t and SALESk;j;t all the firm-level characteristics, once threshold in CONCENTRATIONj;ðt;t3Þ for which financial
re-aggregated at the industry level, lose their significance. inclusion has a negative effect on firm performance is higher
More important, our variables of interests, namely on samples of countries which are less developed (adding con-
INCLUSION-LOANk;j;t ; CONCENTRATIONj;ðt;t3Þ , and flict and post-conflict countries and dropping Latin America).
their interaction term, are not altered by this change, either When services are excluded, the degree of bank concentration
in terms of significance or in magnitude. must be very high for financial inclusion to affect firm perfor-
Second, we address the potential endogeneity of firm-level mance negatively (above 98.9%), which is case of only 2,909
variables that may stem from time-invariant confounding fac- firms in our sample of manufacturing firms. These results sug-
tors. To do this, we account for a firm fixed effect. We stacked gest that financial inclusion is favorable for firm growth even
firm-level panel data from the World Bank Enterprise Surveys for high levels of bank concentration when the firm is located
(WBES). Our sample is then composed of 3,745 firms in 25 in the poorest countries, as well as for manufacturing firms.
 FINANCIAL INCLUSION, BANK CONCENTRATION, AND FIRM PERFORMANCE 7

Table 3. Using an alternative measure of financial inclusion and bank concentration

Measure of INCLUSION
LOAN OVERDRAFT WC-BANK
(1) (2) (3)
Panel A – Market share of 3 main banks
INCLUSIONk;j;t 0.366*** 0.306*** 0.522**
(2.86) (2.93) (2.59)
INCLUSIONk;j;t x CONCENTRATIONj;ðt;t3Þ 0.511*** 0.401*** 0.609**
(3.03) (2.71) (2.47)
CONCENTRATIONj;ðt;t3Þ 0.433*** 0.408*** 0.531***
(3.86) (3.26) (3.47)
FINDEVj;ðt;t3Þ 0.339 0.426* 0.0680
(1.45) (1.76) (0.22)
ACCESSi;k;j;t 0.0344*** 0.0356 0.0226***
(3.14) (1.61) (3.01)
Observations (Country) 55,596 (79) 55,120 (79) 53,480 (78)
Panel B – Market share of 5 main banks
INCLUSIONk;j;t 0.402** 0.323** 0.614**
(2.14) (2.54) (2.13)
INCLUSIONk;j;t x CONCENTRATION-5j;ðt;t3Þ 0.493** 0.366** 0.655**
(2.26) (2.18) (2.03)
CONCENTRATION-5j;ðt;t3Þ 0.305** 0.257** 0.344*
(2.34) (2.02) (1.73)
FINDEVj;ðt;t3Þ 0.340 0.397 0.131
(1.21) (1.44) (0.32)
ACCESSi;k;j;t 0.0342*** 0.0355 0.0223***
(3.09) (1.59) (2.96)
Observations (Country) 52,794 (65) 52,386 (65) 50,890 (63)
Panel C – Lerner Index
INCLUSIONk;j;t 0.163 0.212*** 0.412**
(1.35) (3.23) (2.12)
INCLUSIONk;j;t x CONCENTRATION-Lernerj;ðt;t3Þ 0.606* 0.721*** 1.213**
(1.74) (3.13) (2.39)
CONCENTRATION-Lernerj;ðt;t3Þ 0.151 0.0252 0.914
(0.35) (0.06) (1.15)
FINDEVj;ðt;t3Þ 0.381 0.388 0.174
(1.32) (1.35) (0.40)
ACCESSi;k;j;t 0.0363*** 0.0391 0.0236***
(3.29) (1.64) (2.96)
Observations (Country) 48,806 (64) 48,512 (64) 46,806 (63)
Country dummies yes yes yes
Sector-year dummies yes yes yes
Firm and country-level controls yes yes yes
*** **
Robust standard errors clustered at the country level. t-Student in parenthesis. p < 0.01, p < 0.05, *p < 0.1.

(a) Do some firms benefit more from financial inclusion than We now examine whether firm characteristics – firm size, type
others? of ownership, and exporting activity – affect the relationship we
just highlighted. To do so, we estimate a model in which
The last step in our analysis consists in examining whether inclusion and concentration are both interacted with firm-
the impact of financial inclusion and bank concentration level characteristics. In column (1) of Table 5 we find that the
depends on some specific country and firm characteristics. impact of both financial inclusion and bank concentration on
We have found so far that financial inclusion has a positive firm performance does not depend on the size of the firm,
effect on firm growth, which is dampened by the degree of con- SIZEi;k;j;t . Controlling for the firm’s access to credit (with the
centration of the banking sector. This non-linear relationship LOANi;k;j;t variable) whether the firm is small or large does
may stem from the fact that a more concentrated banking sec- not alter the way it benefits from financial inclusion or bank
tor tends to charge higher credit prices and to concentrate its concentration. A similar conclusion emerges when comparing
lending on a small number of large firms. We also find that the exporting and non-exporting firms in Column (2) of Table 5.
bank concentration in developing and emerging countries Controlling for their access to credit, and whether they are
affects firm performance positively, notably when financial outward-looking or not, the impact of financial inclusion and
inclusion is low. For higher levels of financial inclusion, a cer- bank concentration does not depend on their exporting activity
tain degree of bank competition is more favorable to firm (neither of the interaction terms of INCLUSIONk;j;t and
growth. CONCENTRATIONj;ðt;t3Þ with EXPORTi;k;j;t are significant).
8 WORLD DEVELOPMENT

Table 4. Estimations on sub-samples

CONFLICT HIC SSA ASIA LAC SERVICES
ADDED ADDED DROPPED DROPPED DROPPED DROPPED
(1) (2) (3) (4) (5) (6)
INCLUSION-LOANk;j;t 0.312** 0.217** 0.389*** 0.328*** 0.244*** 0.426**
(2.52) (2.06) (3.05) (2.76) (4.15) (2.40)
INCLUSION-LOANk;j;t x CONCENTRATIONj;ðt;t3Þ 0.428** 0.348** 0.575*** 0.486** 0.320*** 0.431**
(2.49) (2.37) (3.30) (2.55) (3.46) (2.48)
CONCENTRATIONj;ðt;t3Þ 0.310*** 0.282** 0.328*** 0.371*** 0.470* 0.335*
(2.66) (2.50) (2.87) (3.00) (1.72) (1.77)
FINDEVj;ðt;t3Þ 0.153 0.0776 0.0939 0.191 0.768** 0.304
(0.69) (0.36) (0.59) (0.86) (2.33) (0.94)
LOANi;k;j;t 0.0364*** 0.0374*** 0.0367*** 0.0135 0.0362*** 0.0231*
(3.44) (4.09) (3.23) (0.80) (3.53) (1.88)
Observations 63,967 71,851 44,453 38,370 38,857 36,159
Countries 92 107 53 69 56 79
Country dummies yes yes yes yes yes yes
Sector-year dummies yes yes yes yes yes yes
Firm and country-level controls yes yes yes yes yes yes
ðaÞ
Threshold in CONCENTRATIONj;ðt;t3Þ 0.729 0.624 0.677 0.675 0.763 0.989
ðbÞ
Threshold in INCLUSIONj;ðt;t3Þ 0.724 0.810 0.570 0.763 1.468 0.777
Robust standard errors clustered at the country level. t-Student in parenthesis. ***p < 0.01, **p < 0.05, *p < 0.1.
(a) Level of CONCENTRATIONj;ðt;t3Þ above which INCLUSION-LOANk;j;t has a negative impact.
(b) Level of INCLUSION-LOANk;j;t above which CONCENTRATIONj;ðt;t3Þ has a negative impact.

Table 5. The effect of financial inclusion and development depending on firm and country characteristics
Variable: Firm-level characteristics Country-level characteristics
SIZE EXPORTS FOREIGN STATE FINDEV LNINCOME
(1) (2) (3) (4) (5) (6)
INCLUSIONk;j;t 0.314*** 0.341*** 0.368*** 0.366*** 0.358*** 1.113***
(3.04) (2.91) (2.84) (2.86) (2.96) (5.44)
CONCENTRATIONj;ðt;t3Þ 0.387*** 0.422*** 0.429*** 0.432*** 0.519** 4.672***
(3.16) (3.77) (3.82) (3.85) (2.03) (3.44)
INCLUSIONk;j;t x CONCENT:j;ðt;t3Þ 0.522*** 0.545*** 0.529*** 0.512*** 0.515*** 0.483***
(3.26) (3.13) (3.06) (3.05) (2.88) (3.12)
INCLUSIONk;j;t x VARIABLE 0.0340 0.177 0.0834* 0.00469 0.0243 0.0969***
(0.87) (1.25) (1.83) (0.07) (0.30) (3.92)
CONCENT:j;ðt;t3Þ x VARIABLE 0.0321 0.0455 0.0524** 0.0896* 0.322 0.525***
(1.63) (1.32) (2.21) (1.76) (0.47) (3.27)
VARIABLE 0.084*** 0.019 0.026 0.070*** 0.143 0.218
(3.97) (0.48) (1.22) (3.16) (0.35) (0.74)
LOANi;k;j;t 0.035*** 0.036*** 0.035*** 0.034*** 0.034*** 0.035***
(3.00) (3.37) (3.17) (3.14) (3.14) (3.16)
Observations 55,596 55,596 55,596 55,596 55,596 55,596
Country 79 79 79 79 79 79
Country dummies yes yes yes yes yes yes
Sector-year dummies yes yes yes yes yes yes
Firm and country-level controls yes yes yes yes yes yes
*** **
Robust standard errors clustered at the country level. t-Student in parenthesis. p < 0.01, p < 0.05, *p < 0.1.

However, we find that the kind of ownership as shown in col- inclusion up to higher levels of bank concentration compared to
umns (3) and (4) of Table 5 may affect how financial inclusion and locally owned firms. Column (3) of Table 5 also suggests that
bank concentration boost firm growth. Indeed, both interaction when FOREIGNi;k;j;t equals one the level of financial inclusion
terms of FOREIGNi;k;j;t (whether the firm is foreign-owned) with for which bank concentration starts having a negative impact
inclusion and concentration are significantly positive. Comput- on firm performance is around 91%, while it is around 81% when
ing the turning points, we find that when FOREIGNi;k;j;t equals FOREIGNi;k;j;t equals zero. This suggests that foreign-owned
one, INCLUSIONk;j;t has a positive effect on firm growth for firms benefit from bank concentration up to higher levels of
levels of CONCENTRATIONj;ðt;t3Þ below 85%, while when financial inclusion compared with other firms.
FOREIGNi;k;j;t equals zero, this threshold is lower, at around State-owned enterprises also seem to benefit more from
70%. This suggests that foreign-owned firm benefit from financial bank concentration than other kinds of firms, as illustrated
 FINANCIAL INCLUSION, BANK CONCENTRATION, AND FIRM PERFORMANCE 9

in column (4) of Table 5. When STATEi;k;j;t equals one, bank We also find that bank concentration may favor firm per-
concentration always has a positive impact on firm growth formance in developing and emerging countries. This result is
(the turning point in INCLUSIONk;j;t is above 100%), while in line with a slew of recent literature which underlines how
when STATEi;k;j;t equals zero bank concentration has a posi- suboptimal business environments and cream-skimming
tive effect on firm growth until financial inclusion reaches strategies by entering banks may undermine the expected
85%; only for higher levels of financial inclusion does the firm benefits of bank competition. We also find that such effects
benefit from more bank competitiveness. are gradually eliminated by economic development and
These results confirm that foreign-owned firms tend to ben- financial inclusion.
efit more than other firms from financial inclusion and bank We test the robustness of these results using alternative
concentration, whereas state-owned firms benefit only from definitions of financial inclusion and bank competition. We
bank concentration. These results provide additional support also test our model on sub-samples of countries from differ-
to our view that, in our sample of developing and emerging ent continents and by restricting analysis to the manufactur-
countries, bank concentration mostly favors foreign and ing sector. We not only find similar results as in the baseline
state-owned firms and that bank competition only has a posi- estimate, but also infer that financial inclusion is particularly
tive impact on overall firm growth if credit is widely supplied important in the less developed countries, where financial
to firms, i.e., if financial inclusion is broad enough to include inclusions is lower, bank concentration higher, and the rela-
not only lower risk public and foreign firms but also locally tive gains of financial inclusion and financial reform more
owned private firms. significant.
The impact of financial inclusion and bank concentration Finally, we use the firm- and country-level heterogeneity in
may not only depend on firm-level characteristics but also the data to refine our analysis of the effect of financial inclu-
on country-level characteristics. In columns (5) and (6) of sion and bank concentration on firm performance. While
Table 5 we thus examine whether the impact of financial financial inclusion is found to benefit firms of all sizes, irre-
inclusion and bank concentration depends on the level of spective of whether they export or not, we find that foreign
financial and economic development of countries. We find firms benefit most from financial inclusion and bank concen-
that the level of financial development, FINDEVj;ðt;t3Þ , does tration and that state-owned firms benefit most from bank
not alter the relationships between inclusion, concentration concentration. This may reflect privileged access to credit that
and growth: neither of interaction variables of financial inclu- these types of firms may enjoy in some developing and emerg-
sion and bank concentration with financial development are ing countries.
significant. These conclusions suggest that insufficient financial inclu-
When interacting our variables of interest with income per sion represents one channel which, along with weak gover-
capita, we find that both interaction terms have a significantly nance, institutions, and large information asymmetries,
negative effect on firm growth. It seems that for more devel- reduces the expected benefits from the development of bank-
oped countries the relationships presented in the baseline ing systems in developing countries. A narrow client base
results of Table 2 are dampened. More specifically, assuming and credit portfolios highly concentrated on government
the level of bank concentration is the average of the sample debt and a minority of large, often international, private
(74%), then financial inclusion has a positive impact on firm firms seem to largely dampen the benefits of financial devel-
growth until the country reaches a level of income per capita opment, notably in the poorest countries. These results sug-
higher than 2500 USD. Then financial inclusion tends to gest that the expected benefits of financial deepening and
impede firm growth. Column (6) also suggests that for a level competition for firm performance can only be seen when
of financial inclusion around the average (40%) bank concen- the bank client base and bank credit portfolios become more
tration has a positive impact on firm performance in countries inclusive.
with income per capita below 5000 USD. We interpret this as Financial development policy should therefore include
yet another example of the non-linearity of the relationship specific plans to enhance financial inclusion, alongside finan-
between financial and economic development: in low- and cial market development and credit growth. Such strategic
lower middle-income countries, banking systems are not large plans involve reducing information asymmetries, which may
enough to impact firm growth and what matters is financial induce credit rationing from credit institutions, increasing con-
inclusion and how banks choose to compete with each other. sumer protection, and improving the business climate of the
As the level of economic development rises in emerging coun- banking sector. If overdraft facilities remain the cornerstone
tries, the expected positive relationships between firm perfor- of access to credit in these countries, diversifying credit access
mance and bank competition take hold and financial toward long-term credit is also paramount.
inclusion may not be as relevant. Further research is clearly needed to bring light to the
impact of financial inclusion on economic performance in
developing and emerging countries. Beyond firm growth,
6. CONCLUSION the relationship between financial inclusion and productivity
and export performance might be of interest, as well as
In this article, we examine whether financial deepening con- studies focusing on price effects, credit rationing, and the
tributes to the growth of the private sector in developing and business environment of the banking sector to explain and
emerging countries. More specifically, we investigate how the reduce financial exclusion. We believe that focusing further
more or less even distribution of credit across firms in a sector on variables that measure the quality of financial develop-
affects firm performance, as well as the role played by the ment, such as financial inclusion and the modalities of bank
banking market structure. We find that financial inclusion competition, brings an exciting new dimension to the analy-
has a positive impact on the growth of firms and that bank sis of financial development, especially in developing and
competition (lower bank concentration) magnifies this effect. emerging countries where the size alone of the financial sec-
The positive interaction between financial inclusion and bank tor may not be the most significant variable of interest or
competition suggests they may be complementary. policy focus.
10 WORLD DEVELOPMENT

NOTES

1. Households’ access to credit may also increase economic growth, 5. We exclude the following high-income countries: The Bahamas,
particularly if it finances durable goods rather than consumption (GPFI, Barbados, Bulgaria, Croatia, Czech Republic, Estonia, Germany, Greece,
2013). Hungary, Ireland, Israel, Republic of Korea, Latvia, Lithuania, Oman,
Poland, Portugal, Romania, Russian Federation, Serbia and Montenegro,
2. Bank competition is measured by non-structural indicators of com- Slovakia, Slovenia, Spain, St. Kitts and Nevis, Sweden, and Trinidad and
petition such as Lerner, Boone, or H indicators. Tobago.

3. It ranges from weak (2.5) to strong (2.5) control of corruption 6. Conflict and post-conflict countries are Burundi, Afghanistan,
(Worldwide Governance Indicators). Angola, Algeria, Chad, Congo, Dem. Rep., Côte d’Ivoire, Eritrea, Iraq,
Israel, Liberia, Nepal, Nigeria, Sierra Leone, South Sudan, Sudan, Sri
4. The results in Tables 2–5 are virtually the same when the standard Lanka, West Bank and Gaza, Yemen, Myanmar, and Ukraine.
errors are clustered at the sector level or at the country and year levels.

REFERENCES

Abdmoulah, W., & Jelili, R. B. (2013). Access to finance thresholds and GPFI. (2013). The 20 basic set of financial inclusion indicators. Global
the finance-growth nexus. Economic Papers: A Journal of Applied Partnership for Financial Inclusion April.
Economics and Policy, 32(4), 522–534. Guérineau, S., & Jacolin, L. (2014). L’inclusion financiére en Afrique sub-
Arcand, J. L., Berkes, E., & Panizza, U. (2015). Too much finance?. saharienne: Faits stylisés et déterminants. Revue d’économie finan-
Journal of Economic Growth, 20(2), 105–148. ciére, 116(4), 57–80.
Beck, T., Demirguc-Kunt, A., & Maksimovic, V. (2005). Financial and Harrison, A. E., Lin, J. Y., & Xu, L. C. (2014). Explaining Africa’s (dis)
legal constraints to growth: Does firm size matter?. Journal of Finance, advantage. World Development, 63, 59–77.
60(1), 137–177. Harrison, A. E., & McMillan, M. S. (2003). Does direct foreign investment
Beck, T., Büyükkarabacak, B., Rioja, F. K., Valev, N. T., et al. (2012). affect domestic credit constraints?. Journal of International Economics,
Who gets the credit? And does it matter? Household vs firm lending 61(1), 73–100.
across countries. The BE Journal of Macroeconomics, 12(1), 1–46. Hauswald, R., & Marquez, R. (2006). Competition and strategic infor-
Beck, T., De Jonghe, O., & Schepens, G. (2013). Bank competition and mation acquisition in credit markets. Review of Financial Studies, 19(3)
stability: Cross-country heterogeneity. Journal of Financial Intermedi- , 967–1000.
ation, 22(2), 218–244. Henderson, D. J., Papageorgiou, C., & Parmeter, C. F. (2013). Who
Cetorelli, N., & Gambera, M. (2001). Banking market structure, financial benefits from financial development? New methods, new evidence.
dependence and growth: International evidence from industry data. European Economic Review, 63, 47–67.
The Journal of Finance, 56(2), 617–648. Hermes, N., & Lensink, R. (2003). Foreign direct investment, financial
Claessens, S., & Laeven, L. (2005). Financial dependence, banking sector development and economic growth. The Journal of Development
competition, and economic growth. Journal of the European Economic Studies, 40(1), 142–163.
Association, 3(1), 179–207. Kaufmann, D., Kraay, A., & Mastruzzi, M. (2011). The worldwide
Claessens, S., & Horen, N. (2014). Foreign banks: Trends and impact. governance indicators: Methodology and analytical issues. Hague
Journal of Money, Credit and Banking, 46(s1), 295–326. Journal on the Rule of Law, 3(2), 220–246.
Deidda, L. G. (2006). Interaction between economic and financial Leon, F. (2015). Does bank competition alleviate credit constraints in
development. Journal of Monetary Economics, 53(2), 233–248. developing countries?. Journal of Banking & Finance, 57, 130–142.
Deidda, L., & Fattouh, B. (2002). Non-linearity between finance and Loayza, N. V., & Rancière, R. (2006). Financial development, financial
growth. Economics Letters, 74(3), 339–345. fragility, and growth. Journal of Money, Credit and Banking,
De la Torre, Augusto, Martı́nez Perı́a, M. S., & Schmukler, S. L. (2010). 1051–1076.
Bank involvement with SMEs: Beyond relationship lending. Journal of Love, I., & Martı́nez Perı́a, M. S. (2015). How bank competition affects
Banking & Finance, 34(9), 2280–2293. firms’ access to finance. The World Bank Economic Review, 29(3),
Demetriades, P., & Hook Law, S. (2006). Finance, institutions and 413–448.
economic development. International Journal of Finance & Economics, Méon, P.-G., & Weill, L. (2010). Does financial intermediation matter for
11(3), 245–260. macroeconomic performance?. Economic Modelling, 27(1), 296–303.
Demirguc-Kunt, A. & Martinez Peria, M. S. (2010). A framework for Moulton, B. R. (1990). An illustration of a pitfall in estimating the effects
analyzing competition in the banking sector : An application to the of aggregate variables on micro units. The Review of Economics and
case of Jordan. Policy Research Working Paper Series 5499, The World Statistics, 72(2), 334–338.
Bank. Northcott, C. A. (2004). Competition in banking: A review of the
Demirguc-Kunt, A., Levine, R., & Min, H.-M. (1998). Opening to foreign literature. In Staff Working Papers. Bank of Canada.
banks: Issues of stability, efficiency, and growth. In S. Lee (Ed.). The Philippon, T., & Reshef, A. (2013). An international look at the growth of
Implications of Globalization of World Financial Markets (pp. 1–46). modern finance. The Journal of Economic Perspectives, 73–96.
Seoul, Korea: The Bank of Korea. Rajan, R. G., & Zingales, L. (1998). Financial dependence and growth.
Demirguc-Kunt, A., Laeven, L., & Levine, R. (2004). Regulations, market American Economic Review, 559–586.
structure, institutions, and the cost of financial intermediation. Journal Rajan, R. G., & Zingales, L. (2003). The great reversals: The politics of
of Money, Credit and Banking, 36(3), 593–622. financial development in the twentieth century. Journal of Financial
Detragaiche, E., Tressel, T., & Gupta, P. (2008). Foreign banks in poor Economics, 69(1), 5–50.
countries: Theory and evidence. The Journal of Finance, 63(5), Rioja, F., & Valev, N. (2004). Does one size fit all? A reexamination of the
2123–2160. finance and growth relationship. Journal of Development Economics, 74
Easterly, W., Islam, R., & Stiglitz, J. E. (2001). Shaken and stirred: (2), 429–447.
Explaining growth volatility. Annual World Bank Conference on Rousseau, P. L., & Wachtel, P. (2002). Inflation thresholds and the
Development Economics (Vol. 191, pp. 211). . finance–growth nexus. Journal of International Money and Finance, 21
GPFI, Global Partnership for financial inclusion. (2011). SME Finance (6), 777–793.
Policy Guide. IFC, Washington, DC.
 FINANCIAL INCLUSION, BANK CONCENTRATION, AND FIRM PERFORMANCE 11

APPENDIX A. ADDITIONAL TABLES

Table 6. List of countries, survey years, and number of observations

Country 2006 2007 2008 2009 2010 2011 2012 2013 2014 Total
Albania 0 121 0 0 0 0 0 137 0 258
Antigua and Barbuda 0 0 0 0 119 0 0 0 0 119
Argentina* 757 0 0 0 818 0 0 0 0 1,575
Armenia 0 0 0 192 0 0 0 205 0 397
Azerbaijan 0 0 0 316 0 0 0 142 0 458
Bangladesh* 0 0 0 0 0 0 0 1,306 0 1,306
Belarus 0 0 174 0 0 0 0 209 0 383
Belize 0 0 0 0 144 0 0 0 0 144
Benin 0 0 0 95 0 0 0 0 0 95
Bolivia* 392 0 0 0 140 0 0 0 0 532
Bosnia 0 0 0 237 0 0 0 280 0 517
Botswana* 252 0 0 0 191 0 0 0 0 443
Brazil* 0 0 0 988 0 0 0 0 0 988
Burkina Faso* 0 0 0 304 0 0 0 0 0 304
Cameroon* 0 0 0 314 0 0 0 0 0 314
Cape Verde* 0 0 0 77 0 0 0 0 0 77
Chile* 737 0 0 0 855 0 0 0 0 1,592
China 0 0 0 0 0 0 2,441 0 0 2,441
Colombia* 790 0 0 0 824 0 0 0 0 1,614
Costa Rica 0 0 0 0 326 0 0 0 0 326
Djibouti 0 0 0 0 0 0 0 63 0 63
Dominican Republic 0 0 0 0 275 0 0 0 0 275
Ecuador* 441 0 0 0 296 0 0 0 0 737
Egypt 0 0 0 0 0 0 0 1,948 0 1,948
El Salvador* 521 0 0 0 262 0 0 0 0 783
Ethiopia 0 0 0 0 0 328 0 0 0 328
Gabon 0 0 0 83 0 0 0 0 0 83
Gambia 120 0 0 0 0 0 0 0 0 120
Georgia 0 0 211 0 0 0 0 193 0 404
Ghana 0 436 0 0 0 0 0 459 0 895
Guatemala* 439 0 0 0 373 0 0 0 0 812
Guyana 0 0 0 0 121 0 0 0 0 121
Honduras* 348 0 0 0 215 0 0 0 0 563
India 0 0 0 0 0 0 0 0 8,168 8,168
Indonesia 0 0 0 1,089 0 0 0 0 0 1,089
Jordan 0 0 0 0 0 0 0 429 0 429
Kazakhstan 0 0 0 350 0 0 0 281 0 631
Kenya 0 578 0 0 0 0 0 591 0 1,169
Kyrgyzstan 0 0 0 162 0 0 0 0 0 162
Laos 0 0 0 346 0 0 177 0 0 523
Lebanon 0 0 0 0 0 0 0 381 0 381
Lesotho 0 0 0 112 0 0 0 0 0 112
Macedonia 0 0 0 258 0 0 0 327 0 585
Madagascar 0 0 0 288 0 0 0 280 0 568
Malawi* 0 0 0 104 0 0 0 0 293 397
Mali 0 423 0 0 80 0 0 0 0 503
Mauritius 0 0 0 295 0 0 0 0 0 295
Mexico* 1,097 0 0 0 1,220 0 0 0 0 2,317
Moldova 0 0 0 328 0 0 0 265 0 593
Mongolia 0 0 0 350 0 0 0 274 0 624
Montenegro 0 0 0 67 0 0 0 96 0 163
Morocco 0 0 0 0 0 0 0 322 0 322
Mozambique 0 431 0 0 0 0 0 0 0 431
Namibia 236 0 0 0 0 0 0 0 144 380
Nicaragua* 390 0 0 0 248 0 0 0 0 638
Niger* 0 0 0 97 0 0 0 0 0 97
Pakistan* 0 736 0 0 0 0 0 505 0 1,241
(continued on next page)
12 WORLD DEVELOPMENT

Table 6 (continued)
Country 2006 2007 2008 2009 2010 2011 2012 2013 2014 Total
*
Panama 333 0 0 0 129 0 0 0 0 462
Paraguay* 334 0 0 0 257 0 0 0 0 591
Peru* 545 0 0 0 815 0 0 0 0 1,36
Philippines 0 0 0 910 0 0 0 0 0 910
Rwanda 157 0 0 0 0 0 0 0 0 157
Samoa 0 0 0 66 0 0 0 0 0 66
Senegal* 0 411 0 0 0 0 0 0 343 754
Serbia 0 0 0 314 0 0 0 283 0 597
South Africa 0 812 0 0 0 0 0 0 0 812
St. Lucia 0 0 0 0 130 0 0 0 0 130
Suriname 0 0 0 0 152 0 0 0 0 152
Swaziland 206 0 0 0 0 0 0 0 0 206
Tajikistan 0 0 218 0 0 0 0 170 0 388
Tanzania 367 0 0 0 0 0 0 344 0 711
Togo 0 0 0 89 0 0 0 0 0 89
Tunisia 0 0 0 0 0 0 0 557 0 557
Turkey 0 0 594 0 0 0 0 658 0 1,252
Uganda 516 0 0 0 0 0 0 355 0 871
Uruguay* 376 0 0 0 397 0 0 0 0 773
Venezuela* 0 0 0 0 135 0 0 0 0 135
Vietnam 0 0 0 858 0 0 0 0 0 858
Zambia 0 406 0 0 0 0 0 526 0 932
Total 9,354 4,354 1,197 8,689 8,522 328 2,618 11,586 8,948 55,596
Countries with a star are those for which we can include a firm fixed-effect.

Table 7. List of sectors

Sector Frequency Percent
Food 6,568 11.81
Tobacco 154 0.28
Textile 3,093 5.56
Garments 3,858 6.94
Leather 789 1.42
Wood 970 1.74
Paper 561 1.01
Publishing 1,136 2.04
Chemicals and refined petroleum 3,369 6.06
Rubber and plastics 2,187 3.93
Metallic and non-metallic mineral products 3,568 6.42
Fabricated metal products 2,885 5.19
Machinery and equipment 1,892 3.40
Electronics 1,403 2.52
Motor vehicles 987 1.78
Furniture 1,702 3.06
Other manufacturing 1,037 1.87
Retail trade 2,496 4.49
Wholesale trade 5,521 9.93
IT 820 1.47
Transport and construction 6,013 10.82
Hotel 2,132 3.83
Other services 2,455 4.42
Total 55,596 100.00

APPENDIX B. COMPUTATION OF THE LERNER computed as the difference between the price of bank
INDEX output, P, and marginal costs, MC, divided by the price P:
(P  MC)/P.
We use the computation of the Lerner Index provided by the P is calculated as total bank revenue over assets and MC is
World Bank Global Financial Development Data (GFDD) calculated by taking the first derivative from a following
which uses Demirguc-Kunt and Martinez Peria (2010) translogarithmic cost function with respect to Qi;t
methodology and data from Bankscope. The Lerner Index is (Demirguc-Kunt & Martinez Peria, 2010):
 FINANCIAL INCLUSION, BANK CONCENTRATION, AND FIRM PERFORMANCE 13

1 X3 X3
1  W 1;i;t : ratio of interest expenses to total deposits and
LnC i;t ¼ a1 lnQi;t þ a2 lnQ2i;t þ bk lnW k;i;t þ dk lnW 2k;i;t money market funding (proxy for input price of deposits)
2 2
k¼1 k¼1  W 2;i;t : ratio of personnel expenses to total assets (proxy
X
3
1 for input price of labor)
þ ck lnQi;t  lnW k;i;t þ k1 lnW 1;i;t  lnW 2;i;t  W 3;i;t : ratio of other operating and administrative
2
k¼1 expenses to total assets (proxy for input price of equip-
þ k2 lnW 1;i;t  lnW 3;i;t þ k3 lnW 2;i;t  lnW 3;i;t ment/fixed capital).
The translog cost function also includes bank fixed effects,
þ s1 Trend t þ s2 Trend 2t þ s3 Trend t  lnQi;t
li , as well as a trend to capture the influence of technical
X
3 change leading to shifts in the cost function over time
þ nk Trend t lnW k;i;t þ li þ ei;t ð2Þ (Demirguc-Kunt & Martinez Peria, 2010). In line with the lit-
k¼1 erature, the estimation is done under the restrictions of sym-
metry and degree one homogeneity in the price of inputs.
 i denotes banks and t denotes years However, as noted by Demirguc-Kunt and Martinez Peria
 C i;t : bank total operating plus financial costs (2010) the results are not altered when this constraint is
 Qi;t : bank total assets dropped.

Available online at www.sciencedirect.com

ScienceDirect