mc project

Hendrik Hakenes – Iftekhar Hasan – Phil Molyneux – Ru Xie
Small banks and local economic
development
Bank of Finland Research
Discussion Papers
5 • 2014

1
Small Banks and Local Economic Development*
Hendrik Hakenes†
, Iftekhar Hasan‡
, Phil Molyneux§
, and Ru Xie**
Abstract
This paper discusses the effects of small banks on economic growth. We first theoretically
show that small banks operating at a regional level can spur local economic growth. As
compared with big interregional banks, small regional banks are more effective in promoting
local economic growth, especially in regions with lower initial endowments and severe credit
rationing. We then test the model predictions using a sample of German banks and
corresponding regional statistics. We find that small regional banks are more important
funding providers in regions with low access to finance. The empirical results support the
theoretical hypotheses.
JEL Classification: G21, O16, R11.
Keywords: small banks, regional economic growth.
*
We thank the German Savings and Loan Association (DSGV) for the provision of data. We would like to
thank Franklin Allen, Thomas Bloch, Santiago Carbó-Valverde, Daniel Foos, Giorgio Gobbi, Andre Guettler,
Andreas Hackethal, Richard J. Herring, Michael Koetter, Laetitia Lepetit, Céline Meslier, Emmanuelle Nys,
Klaus Schaeck, Reinhard H. Schmidt, Amine Tarazi, Anjan Thakor, and Mark Wahrenburg for their helpful
comments. We sincerely thank the referee for his/her detailed comments. Errors and omissions remain the
responsibility of the authors.
†
University of Bonn & Max Planck Institute for Research on Collective Goods & Centre for Economic Policy
Research (CEPR)
‡
Fordham University and Bank of Finland
§
Bangor Business School, Bangor University !!!
**
Bangor Business School, Bangor University

2
1. INTRODUCTION
Banking systems have experienced a trend toward greater market concentration and a decline
in the number of banks in recent decades. Driven by market power (Claessens and Laeven,
2004, 2005; Carbó-Valverde et al., 2009; Koetter et al., 2012), scale economy (Berger, 1995;
DeYoung et al., 2009; DeYoung, 2012; Wheelock and Wilson, 2012) and too-big-to-fail
motives (Mishkin, 2006; Kane, 2000; Stolz and Wedow, 2010; Laeven and Valencia, 2010;
Carow et al. 2011) banks have become larger.1
However, these big, complex, and
increasingly opaque banks exploited financial safety nets and took risks that endangered
economies (Blinder, 2010; Feldman and Stern, 2010; Liikanen, 2012; IMF, 2013). The
widespread bailout of big banks during the recent financial crisis introduced the possibility of
break-ups post-crisis, and this has turned analysts’ attentions to the role of small banks in
local economies (Buiter, 2009; Rosenblum, 2011). Ring-fencing proposals in Europe and a de
facto fence around foreign banks in the U.S. have addressed some policy concerns regarding
big banks. In contrast, small local and regional banks often have limited access to
government safety nets and mainly focus on financing individuals/households, as well as
small to medium-sized enterprises (SMEs).2
A major concern regarding small local and
1
Higher credit ratings and lower funding costs may also be related to access to government safety nets (Brewer
and Jagtiani, 2013). The rapid growth in investment banking and the “originate-to-distribute” securitization
model also encouraged banking sector consolidation as well as moral hazard, thereby resulting in excessive risk-
taking (Blinder, 2010; Feldman and Stern, 2010). There is some evidence also that big banks have more volatile
returns (Goddard et al., 2004).
2
In Germany, however, some small banks are protected by insurance funds run by the Bundesverband der
Volks- und Raiffeisenbanken (cooperative banks) and the Savings Bank Association (DSGV) (savings banks).
These offer institution protection schemes to their members. Also, the German savings bank sector had a

3
regional banks is that they are believed to be weak at loan diversification, as highlighted in
the 1980s S&L crisis in the U.S., when more than 700 primarily small S&Ls failed. In
Europe, however, certain small banks, such as savings banks and co-operative banks, can
diversify their loan portfolios as they are organized under central entities that pool risk
management and risk-sharing strategies. Multi-bank loan pools are applied as an instrument
that allows participating banks to share regional credit risk. Loan pools exploit the
diversification benefits available to large banks by spreading risk over regional investments,
while remaining independent and benefiting from local knowledge (Gintschel and Hackethal,
2004). These pooling arrangements, coupled with local knowledge and lower loan screening
and debtor monitoring costs, can result in small banks being effective at promoting local and
regional economic growth (Petersen and Rajan, 2002; Bofondi and Gobbi, 2003).
There is no clear empirical evidence of the differing effects between small local and regional
banks and big interregional banks on local economic growth. Some empirical studies suggest
that small banks with limited market power reduce SME financing constraints (Carbó-
Valverde et al., 2009; Degryse and Ongena, 2007). Based on data from Italy, Guiso et al.
(2004) find that local financial development is positively associated with entry of new firms,
increased competition, and economic growth. Berger et al. (2004) also argue that healthy
community banks improve SME financing. In a recent paper, DeYoung et al. (2012) find that
SMEs are especially reliant on community bank finance. However, they also argue that small
banks can exacerbate economic downturns during recessions due to low diversification. An
alternative empirical finding is by Beck et al. (2010), who show that an increase in intra-state
bank M&As mitigated regional income inequality by affecting labor market conditions.
government guarantee until 2001 (Gropp et al., 2013; Fischer et al., 2012).

4
These conflicting views raise questions about the role of bank size on firm-financing and
economic development.
Since previous studies have not addressed the question whether small banks or big banks are
more efficient at promoting local economic growth, our work aims at providing evidence on
this issue. In particular, we try to answer the question whether small regional banks are more
efficient in promoting local economic growth as compared to big interregional banks. To the
best of our knowledge, our work is the first to explicitly account for the different roles of
regional versus interregional banks in assessing the effectiveness of financial development on
economic growth. Our results are in line with Guiso et al. (2004), in that local financial
development matters for economic growth. We further demonstrate that regional financial
development is especially important for relatively poor regions in an integrated economy,
mainly due to problems associated with attracting interregional financing.
In our paper, we first develop an overlapping generations model that builds on the work by
Bernanke and Gertler (1989) and includes small regional banks in a setting of financial
development and economic growth. There is some theoretical literature on the impact of
financial development on regional growth. In Greenwood and Jovanovic (1990), for instance,
banks produce information, which improves the allocation of resources and fosters growth.
As a larger number of firms use bank services, banks become more efficient, which feeds into
development. Levine (1991) has a growth model based on Diamond and Dybvig (1983) that
focuses on liquidity shocks, which also shows that financial development entails higher
growth. Morales (2003) develops a Schumpeterian model that focuses on creative destruction
and the role of banks. The paper shows that a policy that incentivizes financial intermediary

5
monitoring activity can improve economic growth due to reduced moral hazard. The
literature on the role of regional financial institutions on growth is, however, relatively
sparse. The paper closest to ours is probably Boyd and Smith (1997). In their model,
entrepreneurs from two regions suffer from a state verification problem. In the absence of
interregional capital flows, both regions converge to the same output level. With capital
flows, the richer region finds it easier to attract new capital, and there is no convergence.
Hakenes and Schnabel (2010) theoretically examine the role of bank size on real economic
development, arguing that small regional institutions can give local entrepreneurs a
competitive edge and promote local economic growth in underdeveloped regions. However,
their model is static. Our theoretical model extends the aforementioned literature in three
main ways. First, we explicitly combine elements of both regional financial development and
economic growth in a comparative static equilibrium model. Second, we show that the effect
of regional bank development on economic growth is conditional on the state of economy,
with small bank development becoming more important for regions with lower initial
endowments and severe credit rationing. Finally, unlike the earlier literature, we differentiate
between the roles of small regional banks and big interregional banks on local economic
growth. We find that small regional banks are more effective in promoting local economic
growth, especially in underdeveloped regions.
We then empirically test these predictions on a sample of 457 German savings banks and
corresponding regional statistics over 1995 to 2004. We use German data for two reasons.
First, focusing on a single country limits the unobserved heterogeneity in the sample. Second,
Germany has a strong local banking sector. German savings banks (Sparkasse) follow a
regional principle; they operate predominantly within their own region and typically do not

6
compete with other regions’ savings banks. The regions in which the savings banks operate
are easily identifiable and are closely linked to German administrative districts, called
counties (Kreise). Counties are the second smallest administrative units that provide
macroeconomic data on a regular basis. Therefore, a focus on Germany yields financial and
macroeconomic data on a disaggregated level. Our empirical results reveal that small regional
institutions, namely savings banks, play a prominent role in enhancing local economic
development in under-developed regions. The results remain robust if we control for a variety
of bank and region-specific effects, as well as alternative economic development measures.
Based on these findings we argue that small local and regional banks can spur local economic
growth.
Our paper is organized as follows. Section 2 develops the model and derives various
predictions. In section 3 we outline the empirical methodology and the sample and then
present the results, including robustness tests. Section 4 concludes.
2. THE THEORETICAL MODEL
2.1. THE MODEL ENVIRONMENT
In this section, we construct a growth model that stresses the role of small regional banks.
The growth model is based on Bernanke and Gertler (1989) and Kiyotaki and Moore (1997)
and hence we use an overlapping generations framework. In addition, we allow capital to
flow out of the region. Our economy is small and open so the interest rate is fixed. As a
deviation from perfect financial markets, we assume that entrepreneurs suffer from a moral

7
hazard problem and that the scope of moral hazard can be reduced by bank monitoring.
Consider an economy with an infinite sequence of two-period overlapping generations. At
each date t = 0, 1, 2, …, a continuum of mass L = 1 of young agents is born. All agents live
for two periods. They are risk-neutral, and care only about old-age consumption. In his first
period, an agent works to earn money, which he invests. In the second period, an agent
disinvests and consumes. There is a single final good that can either be consumed or used to
build capital for the following period. Capital is produced by entrepreneurs using an
indivisible capital production technology. This technology is modelled exactly as in Tirole
(2006, chapter 9.2, based on Holmstrom and Tirole, 1997): each project requires I units of the
final good and produces an expected return of R units of capital with probability p. The
entrepreneur can decide whether to work or shirk. If he works, the probability of success is
pH. If he shirks, the probability is pL, and the entrepreneur gets a private benefit B. The
successes or failures of different entrepreneurs are stochastically independent, so there is no
aggregate risk. The private benefit can be reduced to b < B if the entrepreneur is monitored by
a bank. Bank monitoring costs are c per entrepreneur. The net present value of the project is
positive if the entrepreneur behaves, and negative if he shirks. Assume that parameters are
such that the entrepreneur can attract funds only if he is monitored (for example because B is
large). The bank monitoring cost c is also a measure for regional bank efficiency in our
model, with a higher c indicating lower bank efficiency. For now, this c is assumed to be
exogenous. We will discuss endogenous bank efficiency below.
If the entrepreneur is promised a sum Re from the return R, he will work if and only if pH Re ≥
pL Re + b, which is equivalent to

8
Re ≥ b/Δp (IC)
with Δp = pH – pL. Consequently, the bank and investor cannot demand more than R –
Re = R – b/Δp from the entrepreneur, otherwise they would induce him to shirk. Thus the
entrepreneur cannot credibly promise to pay back more than an expected Rp = pH (R – b/Δp),
namely, the pledgeable part of the entrepreneur’s income.
Capital is used to produce consumption goods for the next period. Let Kt be the capital stock
at date t. Then the quantity of the final good output is Yt = F(Kt, L), following a neoclassical
production function. For concreteness, assume that Yt = β Kt
α
L1−α
= β Kt
α
. Hence, the
complete production process takes two steps. First, final goods must be turned into capital by
entrepreneurs. Second, capital in combination with labor is used to produce new final goods
for consumption.
In summary, the economy works as follows. A generation of agents earns wages while
young. Agents need this income to save for future consumption. There are two different ways
to save. An agent can either invest (that is, deposit) money at a bank, which then hands out
the money to entrepreneurs in the form of loans and monitors these entrepreneurs. Otherwise,
an agent can take out a loan, become an entrepreneur, and apply the capital production
technology. Capital can then be used in the future to produce new final consumption goods.
These consumption goods are used to repay the loan to the bank; the rest can be consumed by
the now-old entrepreneur. However, production requires not only capital but also labor.
Hence, entrepreneurs must pay out wages to the next generation; these wages will then be

9
saved for the next period production process. At this point, the model starts over again.
For a bank, there are two alternative classes of investment. It can either hand out loans, as
described above, or it can invest in an interregional financial market. For exposition, we
assume that the region is considered to be a small open economy, and hence the interest rate r
is exogenous. Investment in the financial market works as follows. One unit of final
consumption goods from one period turns into 1 + r units of the final good in the following
period. Alternatively, the region can also borrow from financial markets instead of investing.
Of course, interest rates are endogenous and volatile in reality. Adding volatility would not
change the model, because all agents in the model are risk-neutral. Having an exogenous
interest rate is equivalent to assuming a small open economy, where the other regions are not
modelled explicitly. In a larger model, one could have heterogeneous regions, some that
borrow on the market, and some that lend. The aggregate interest rate r would be the same,
and would be determined by the aggregate market clearing condition. In terms of German
counties, the assumption of small open economies seems fair, as no single county can likely
influence the world interest rate, or even the German interest rate level.
Both capital and labor are assumed to be immobile between regions. As a consequence of
capital immobility, interregional investment and borrowing is possible only with respect to
the final consumption good. The assumptions that the final consumption good (money) is
perfectly mobile and that labor is perfectly immobile are not crucial. However, labor needs to
be less mobile than money; this assumption seems innocuous. Let us also assume that small
regional banks invest only within their respective region, and call interregional banks those
banks that also invest outside their respective region. Without loss of generality, one can

10
assume that interregional banks only invest outside their respective region.
The structure of the regional economy, therefore, can be summarized as follows.
Entrepreneurs use their capital to produce consumption goods. For capital production, they
require financing, which they obtain from the regional banking system. Regional banks, in
turn, refinance from regional investors (that is, depositors). Any remaining regional savings
are invested at interregional banks, which then invest outside their region3
.
Importantly, there are two kinds of equilibrium. If the incentive constraint (IC) is binding,
then entrepreneurs can earn a higher return than investors. In this case, there is credit
rationing: investors would prefer to become entrepreneurs, but they cannot get a loan. To
attract a loan, they would have to offer a higher interest rate, which would violate the
incentive constraint. In the second kind of equilibrium, the incentive constraint is not binding,
hence entrepreneurs and investors must earn the same expected return. Otherwise, some
investors would prefer to start their own firms. We first discuss the equilibrium with a
binding constraint (section 2.2). In section 2.3, we provide sufficient and necessary
conditions for a binding constraint, and then discuss the second case.
2.2. EQUILIBRIUM WITH BINDING INCENTIVE CONSTRAINT
3
Regional banks are likely to be under-diversified. In our model, there is no aggregate risk, so diversification
plays no role. However, in a model with regional risk factors, diversification would matter. There are two issues.
Regional banks exposed to regional risk may suffer from regional shocks. This would dampen their growth path.
Furthermore, undiversified banks may hold more equity (which is not in the current model). This would raise
the financing costs of regional banks, indirectly leading to higher costs of lending c.

11
Consider a region in which agents are initially endowed with consumption goods At. Agents
can now try to borrow more from a bank to generate capital to produce more consumption
goods. Assume for now that pledgeable income is small such that the incentive constraint
(IC) is binding. Consequently, to obtain a loan, entrepreneurs must invest their own complete
endowment At, and thus, they borrow the difference I −At from a regional bank.
If banks grant nt loans altogether, the aggregate loan volume in the region is Ft = nt (I − At).
The lending volume from regional banks Ft depends on regional bank efficiency, Ft = f(c).
More efficient banks lead to better capital allocation and thus decreasing c has a positive
effect on bank lending. From each loan, a bank receives the pledgeable portion Rp = pH (R –
b/Δp) and the zero-profit condition implies that it repays Rp − c to investors (that is,
depositors). The entrepreneur retains the non-pledgeable part pH R − Rp. After repayment,
investors and entrepreneurs have an aggregate capital of
𝐾𝑖 = 𝑛 𝑡�𝑅 𝑝 − 𝑐� and 𝐾𝑒 = 𝑛 𝑡�𝑝 𝐻 𝑅 − 𝑅 𝑝�. (1)
Aggregate capital in the region is thus
𝐾𝑡 = 𝐾𝑖 + 𝐾𝑒 = 𝑛 𝑡(𝑝 𝐻 𝑅 − 𝑐). (2)
This aggregate capital is now used to produce consumption goods by employing next
period’s labor force. The labor market is competitive within each region, thus the equilibrium
wage is

12
𝑊𝑡 = 𝛽(1 − 𝛼)𝐾𝑡
𝛼
. (3)
The net output of final consumption goods is
𝑌𝑡 − 𝑊𝑡 = 𝛽𝐾𝑡
𝛼
− 𝛽(1 − 𝛼)𝐾𝑡
𝛼
= 𝛽𝛼𝐾𝑡
𝛼
= 𝛽𝛼𝑛 𝑡
𝛼(𝑝 𝐻 𝑅 − 𝑐) 𝛼
. (4)
Of this aggregate output, a fraction Ki/(Ki + Ke) belongs to investors, and a fraction Ke/(Ki
+Ke) to entrepreneurs. Now recall that as an alternative, banks could have invested abroad, in
which case they may have offered an interest rate of r. Given that the financial system is
competitive, an investment abroad must earn the same rate as an investment in the bank’s
home region. Because investors have invested an aggregate Ft = nt (I − At) in the region,
𝑛 𝑡(1 + 𝑟)(𝐼 − 𝐴 𝑡) =
𝑅 𝑝−𝑐
𝑝 𝐻 𝑅−𝑐
𝛽𝛼𝑛 𝑡(𝑝 𝐻 𝑅 − 𝑐) 𝛼
. (5)
Solving for nt,
𝑛 𝑡 =
1
�
𝛽𝛼(𝑅 𝑝−𝑐)
(1∓𝑟)(𝐼−𝐴 𝑡)
1−𝛼
. (6)
Now if nt projects are financed, the aggregate amount of capital is given by (2), and the wage
Wt is given by (3), thus
𝑊𝑡 = 𝛽(1 − 𝛼)𝐾𝑡
𝛼
= 𝛽(1 − 𝛼)𝑛 𝑡
𝛼(𝑝 𝐻 𝑅 − 𝑐) 𝛼

13
= 𝛽(1 − 𝛼) �
𝛽𝛼�𝑅 𝑝 − 𝑐�
(1 + 𝑟)(𝐼 − 𝐴 𝑡)
�
𝛼
1−𝛼
. (7)
This wage Wt serves as assets for new projects for the young generation, with At+1 = Wt. Then
a new period starts and the cycle begins anew. If Wt > At, the economy is growing; if Wt < At,
the economy is shrinking.
Let us discuss the financial system in more detail. Given that nt projects are financed within
the region, there is an aggregate amount nt At of inside finance that is not channelled through
the financial system. Each entrepreneur needs an additional I−At, and hence the aggregate
amount of regional lending is Ft = nt (I −At). Now the region is inhabited by L = 1 agents
each endowed with At, thus the difference At − nt At − nt (I − At) = At − nt I is invested
outside the region by interregional banks. Importantly, there can be a capital outflow from
less developed regions. Consider a poor region with low At. Then an entrepreneur requires a
large amount of outside finance. The amount of pledgeable capital in relation to the amount
of finance is then rather low. This low amount of pledgeable capital can only be compensated
by high marginal productivity, which implies that it must be scarce. Hence, in equilibrium,
the number of firms nt must be small. Thus, only a few firms receive financing, while part of
the region’s savings is drained away from the bank’s respective region.
2.3. EQUILIBRIUM WITH NON-BINDING INCENTIVE CONSTRAINT
Before discussing the equilibrium, let us derive a condition for when the incentive constraint
binds. If it does, returns are higher for an entrepreneur. An entrepreneur then invests At and

14
receives pH R − Rp; investors pay I − At to obtain Rp − c in return. Consequently, the
incentive constraint binds if and only if
𝑝 𝐻 𝑅−𝑅 𝑝
𝐴 𝑡
>
𝑅 𝑝−𝑐
𝐼−𝐴 𝑡
, or equivalently
𝐴 𝑡 < 𝐼
𝑝 𝐻 𝑅−𝑅 𝑝
(8)
Now consider an economy with larger endowments, such that the incentive constraint is not
binding. If there are nt firm in the region, the according investment is nt I. The return is an
amount of capital Kt = nt (pH R – c), already taking into account the monitoring costs c.
Because of the production function Yt = β Kt
α
Lt
1–α
, the wage equals marginal productivity,
Wt = β (1–α) Kt
α
Lt
–α
= β (1–α) Kt
α
, which implies a return of Yt – Lt Wt = β α Kt
α
. Now in
equilibrium, the return must equal that from a risk-free investment, nt (1+r) I. As a
consequence,
𝑛 𝑡(1 + 𝑟)𝐼 = 𝛽𝛼𝑛 𝑡
𝛼
(𝑝 𝐻 𝑅 − 𝑐) 𝛼
, or equivalently,
𝑛 𝑡 = �
𝛽𝛼(𝑝 𝐻 𝑅−𝑐) 𝛼
(1+𝑟)𝐼
1−𝛼
. (9)
Because the incentive constraint does not bind, it becomes irrelevant. The friction on the
capital market vanishes, and consequently, investment in the region is independent from its
initial endowment At. Finally, calculate the wage level as
𝑊𝑡 =
1−𝛼
𝛼
(1 + 𝑟) 𝑛 𝑡 =
1−𝛼
𝛼
�
𝛽𝛼(𝑅−𝑐) 𝛼
(1+𝑟) 𝛼 𝐼
1−𝛼
. (10)

15
2.4. COMPARATIVE STATIC RESULTS AND HYPOTHESES
When the incentive Constraint (8) is not binding, Equation (9) implies that bank efficiency
has a positive impact on the number of firms, and thus also on output (GDP), and (10)
implies that bank efficiency increases the wage level. However, all these are level effects:
more bank efficiency leads to a higher GDP, but not to higher GDP growth. Because the
incentive constraint does not bind, the financial market works without friction, and output is
only influenced by current bank efficiency.
When the incentive constraint is binding, an increase in bank efficiency has three effects.
First, less capital is consumed by the bank and more is left for the production of final goods.
Second, more capital can be attracted from the global financial market. And third, the wage
level increases. Consequently, in the next period, wealth in this region will be higher and
entrepreneurs will find it easier to attract more funds. Hence, bank efficiency has a positive
growth effect. Potentially, the region can grow until the incentive constraint ceases to bind.
Putting both cases together, bank efficiency weakly increases growth.
Proposition 1: A more efficient regional banking system spurs regional GDP growth, ∂(Yt −
Yt−1)/∂c ≤ 0.
Proofs are in the appendix. Now the second proposition follows the first one. If the initial
endowment At is large, entrepreneurs are rich enough that the incentive constraint does not
bind. In that case, bank efficiency has a positive effect on the level of output, but not on

16
growth. The smaller At, the higher the shadow price of the binding constraint. As a result,
bank efficiency then has a larger effect on growth. Putting both cases together, the positive
effect of bank efficiency on growth is smaller for richer regions (large At).
If the incentive constraint were non-binding for all firms in all regions, then no firm would be
credit constrained. Consequently, as argued above, there would be no relation between bank
efficiency and growth. Realistically, some firms within each region suffer from borrowing
constraints. The poorer the region, the more severe these constraints will be. This is exactly
the precondition to test Proposition 2, which states that in a region with more initial assets At,
an improvement of bank efficiency has a smaller impact on growth due to decreasing returns.
Proposition 2: In a less developed region with lower At, the impact of an improvement in
bank efficiency on economic growth is larger, ∂/∂At ∂(Yt − Yt−1)/∂c ≤ 0.
In our model, we have treated bank efficiency as exogenous. The parameter c is a measure for
the cost of banks, and hence implicitly also a measure of their comparative (dis)advantage. In
reality, bank costs consist of many components, like screening costs and cost of capital.4
In
our model, as in many others, c is treated as exogenous. Let us thus discuss what factors
influence c. In principle, the past decisions of banks may have influenced c. For example,
there may be “learning by doing”: larger banks may have learned more about how to monitor
their borrowers, and thus have a lower c. Consequently, they will increase lending, fostering
growth. There may be further consequences. Some banks, taking into account future cost
4
In Germany, a country in which there is an implicit government guarantee, the cost of capital is lower for small
regional savings banks.

17
reductions, will offer lower loan rates in order to attract volume. Both of these effects would
be beneficial for growth in the region. As a result, the relaxation of the assumption of
exogeneity of c does not adversely influence our results.
3. EMPIRICAL EVIDENCE
3.1. DATA AND VARIABLES
In order to test the findings of our model, we focus on the German regional banking market,
in which three bank types operate, namely, branches of big commercial banks, savings
banks, and cooperative banks (Altunbas, et al. 2001; Schmidt and Tyrell, 2004; Hackethal,
2004). Major commercial banks operate nationally, whereas local savings banks and credit
cooperatives operate in regions where they are headquartered. The small size and
independence of savings and credit cooperatives can give rise to a competitive advantage,
specifically because of customer proximity and the capacity for quick decision-making in the
local retail banking market (Slotty, 2009). Furthermore, because of the regional footing of
shareholders or members, they are more important funding providers to individuals and
SMEs, especially in relatively poor regions with less of a big bank presence.
Our unique data set combines information from three main resources. Financial and other
information on savings banks are obtained from the German Savings Bank Association;5
regional macroeconomic statistics are provided by the German Federal Statistic Ministry; and
information on local banking market share of big banks and savings banks is provided by the
5
We exclude Landesbanken from our sample, given their national/interregional role.

18
German Bundesbank. In total, the bank-specific data covers 457 savings banks and the
regional and bank market share information of 440 administrative districts from 1995 to
2004.6
Matching of the data sets is not always one-to-one. In some administrative districts,
there is more than one savings bank, and some can also operate in more than one district. In
the first case, we simply aggregate data from the balance sheets of savings banks within the
given district. In the second case, we only consider the region where the headquarters of the
savings bank is located. After this procedure, we are left with 395 regions over a time span of
10 years. Table I reports the definitions and main characteristics of the sample.
[Insert Table I here]
In-line with the classic literature on finance and growth (King and Levine, 1993), regional
economic growth Yit is measured by the growth rate of regional GDP per capita
(GDP_Capita_Growth). As a robustness test, following Guiso (2004) and Shaffer, Hasan, and
Zhou (2014), we also use the growth rate of new business registrations
(Business_Reg_Growth) to approximate regional industrial growth. The two measures are not
significantly related, the correlation ratio between the two being -5.29%.
In the theoretical model we employ bank monitoring costs as an indicator of the level of local
financial development. In the empirical analysis, we use three savings bank performance
indicators to proxy for efficiency. First, we use two indirect indicators, ROA and ROE, which
6
The timeframe is mainly due to data availability issues and a ten year period we believe sufficient to test our
model. Also in recent years, savings banks have started to offer direct banking to their customers, leading to
some softening in the regional principle, so this also supports the use of an earlier time frame.

19
indicate that efficient banks should be more profitable. Then, from Koetter and Wedow
(2010), who find that bank quality (proxied by cost efficiency) explains economic growth, we
use a direct cost efficiency measure derived from stochastic cost frontier estimation. Cost
efficiency measures show how close bank i is to the estimated industry-wide best-practice
cost frontier in year t. As shown in Table I, the average cost efficiency for regional savings
bank is 82.8% over the sample period (similar to earlier findings of Altunbas et al., 2001).7
Details on our cost efficiency estimation methodology are provided in Appendix A2. As a
robustness check, we employ bank Z-score as further bank efficiency proxy so as to take
bank risk taking into consideration. Banks with higher risk-taking are less efficient in capital
allocation and project financing. We use the bank Z-score to measure bank insolvency risk.
The Z-Score is estimated as the ratio between the sum of a bank’s average return on assets
and capitalization (equity/total assets) and the standard deviation of the return on assets. The
Z-score indicates the number of standard deviations that a bank’s return on assets has to drop
below its expected value before equity is depleted and the bank becomes insolvent. Banks
with a low Z-Score indicate that the institution is exposed to higher risk.
To compare the role of big versus small banks on local economic development, we present
the market shares of two banking groups in the respective local market. Large commercial
banks with branches all over the country belong to the big bank group (Market_Share_Big) in
7
Over the study period from 1995 to 2004 there was consolidation in the savings bank sector with the number
of banks falling from 607 to 457 (Bloch, 2008). Simple correlation of our direct cost efficiency measure with
M&A activity per year (the latter from Table 4, p37 in Bloch, 2008) reveals an inverse relationship (- 0.4138).
Similar relationships are found for our direct measures of efficiency. The level of M&A activity in the savings
bank sector appears to be inversely linked to bank cost efficiency (however measured).

20
our sample, whereas savings banks are small banks that only operate in the region in which
they are based (MS_Savings_Bank). The local market shares of the large commercial banks
and the savings banks are calculated on the basis of the number of branches of any one bank
in each administrative district over the total number of bank branches in the same district.
GDP growth can be influenced by factors other than financial sector development, and as
such, these need to be controlled for. First we take account of competition faced by the
savings bank sector using the Lerner index (Fernandez de Guevara et al., 2005; Carbó-
Valverde et al., 2009). The Lerner index is a non-structural measure of competition defined
as:
𝐿 =
𝑃−𝑀𝐶
𝑃
(11)
where P denotes the output price of savings bank i, and MC is their marginal cost obtained by
differentiation of the total cost function shown in Appendix A3. The index ranges from a high
of 1 to a low of 0, with higher values indicating greater market power. As shown in Table 1,
the average value for the Lerner index is 0.236 over the entire sample period.8
We also control for a variety of other macro indicators including: inflation, the local labor
8
We do not have cooperative bank data for our individual localities, so the Lerner index is calculated just for
our sample of savings banks. We do not believe that competition between these two types of banks affects
heterogeneity in terms of composition of the regional competition indicators, as the Lerner index of 0.236 is
similar to that found in previous studies on German banking (see Koetter and Poghosyan, 2009; Buch et al.
2010).

21
supply, and the absolute value of GDP per capita in a given administrative region. High
inflation rates can result in a loss of confidence for future investments and suspend economic
growth. We apply the annual CPI index in Germany to control for this macroeconomic shock.
In the classical production function, output depends on two inputs, namely, capital and labor.
Consequently, one concern of our analysis is that the results may be driven by features of the
local labor supply not captured by a region’s financial development. We therefore include
local labor supply (Labor_Supply) measured by the size of the regional labor force to isolate
this effect. Rich or poor regions might have different growth patterns. We first include
regional GDP per capita (GDP_Capita) to control for the effect of a region’s initial
endowments. We also include a dummy variable (Group Dummy (west)) to determine
whether a region belongs to the former West Germany before the reunification of 1990 to
control for any potential regional bias.
Furthermore, we include bank size, the equity ratio, and the percentage of non-interest
income as bank-specific controls. Bank size (ln_Totalasset) is defined as the natural
logarithm of total assets and is used to control for different characteristics across relatively
large and small savings banks, as well as economies of scale. Banks with a lower equity ratio
(Equity_Ratio) might enjoy a relatively higher return on equity, and this could generate
biased results. Besides, this ratio reflects management’s attitude to risk, and a control for this
variable may also adjust for this possible bias. Finally, some savings banks might focus on
potentially higher marginal commission business, which could have a positive influence on
bank profitability and efficiency measures. Consequently, we control for bank non-interest
income (Share_of_Fee_Inc) to isolate this effect. Table 1 reports the summary statistics.

22
3.2. METHODOLOGY
The following empirical analysis seeks to test the two propositions derived from our
theoretical model. Namely, does a more efficient local banking system promote regional GDP
growth and is this effect larger in less developed regions? The link between finance and
growth as first advocated by Schumpeter (1934) has spawned a significant empirical
literature (King and Levine, 1993; Arestis and Demetriades, 1997; Levine and Zervos, 1998;
Rajan and Zingales, 1998; Claessens and Laeven. 2003; Bekaert et al., 2005; Beck et al.,
2008). Following on from this, we test the propositions presented above using the following
model, which examines the effect of local financial development on economic growth:
itl
J
j
j
itjitmitit DummyYearXFinLocalYaY εβββδ +++++= ∑=
− __
1
1
(12)
where subscript i indicates regions and t indicates years. In line with the theoretical model, Yit
stands for local economic growth. We use two proxies for Yit: the growth rate of regional
GDP per capita and the growth rate of business registration in a certain region. Local_Finit is
an indicator of local financial development proxied by one of three variables: savings bank
cost efficiency, return on total assets, and return on total equity. As a robustness check, we
employ bank Z-score as a further bank efficiency proxy. We also control for a variety of
other variables, Xit, covering indicators of: local banking market competition; inflation;
regional labor supply; bank size; share of fee income; capital strength (proxied using an
equity ratio); and a binary variable indicating whether a region belongs to the former West
Germany. εit is the disturbance, with vi the unobserved regional specific effect and uit the
idiosyncratic error.
A major technical challenge in the empirical finance and growth literature is how to address

23
potential endogeneity problems, as the use of endogenous regressors will lead to inconsistent
and biased estimations. Following Levine et al. (2000), Beck et al. (2000), and Beck and
Levine (2004), we employ dynamic panel estimators to address potential endogeneity
concerns relating to the direction of causality between financial development and economic
growth. Specifically, we use the dynamic system GMM panel estimators proposed by
Arellano and Bover (1995) and Blundell and Bond (1998) with Windmeijer (2005) robust
errors. The dynamic system GMM panel estimators use lagged first-differences of the
variables as additional instruments for equations in levels, which is more efficient than
difference GMM estimators.
3.3. EMPIRICAL RESULTS
BASELINE RESULTS
We first test proposition 1 presented above by estimating the direct link between regional
bank performance and local economic growth. We estimate the baseline model in Equation
(12) using the two-step system GMM estimator with Windmeijer (2005) small sample robust
correction (Carbó-Valverde et al., 2009). The dependent variable is the growth rate of GDP
per capita in a certain region. Savings bank efficiency is proxied by return on equity, return
on total assets, and cost efficiency. All regressions include macroeconomic and bank specific
controls. The GMM estimation results, shown in Table II, indicate a significantly positive
coefficient for the bank performance variables. In all the three regression models, we treat the
lagged dependent variable (L.GDP_Capita_Growth) and lagged bank efficiency (L.ROA,
L.ROE, L.Cost Efficiency) as endogenous GMM style variables. The result remains robust if

24
we only apply the lagged dependent variable (L.GDP_Capita_Growth) as an endogenous
variable. Two tests support our model specification: the null of second-order autocorrelation
(AR(2)) is rejected, and heteroscedasticity-consistent Hansen J-tests do not reject the validity
of our instrument set at least at the 5% level. This illustrates that small bank performance
improvements are effective in promoting regional economic growth. The first two regression
models that use indirect bank efficiency indicators—ROA and ROE—indicate that efficiency
is positively associated with regional economic development: namely, a higher level of bank
profitability spurs region’s economic development in the following period. The final
regression model contains the results with respect to the direct cost efficiency indicator
derived from frontier estimation. Here the results again confirm that savings bank cost
efficiency is associated with an increase in regional economic growth, consistent with the
findings of Koetter and Wedow (2010). All three indicators are highly correlated with
economic growth. As a robustness check, we employ the bank Z-score to proxy for the
degree of bank risk taking. The results show a significantly negative relationship between
bank risk and regional economic growth. Savings banks with higher Z_Score are positively
correlated with regional economic growth. The results illustrate that regional savings banks
with a lower degree of risk taking are associated with higher local economic growth rates.9
Overall, these results confirm our theoretical findings and show that small regional banks are
important fund providers in promoting local economic development.10
In addition, in
regression models (4) to (6), we employ clustered standard errors to control for potential
unobserved factors that may cause correlation of error-terms across regions. The results
remain robust.
9
The results for robustness checks are not reported here. However, they are available upon request.
10
Our estimations suggest divergence of growth across regions due to different initial conditions.

25
Turning to the various control variables, we see that inflation is negatively related to regional
economic growth. This result is consistent with the notion that a higher inflation rate leads to
uncertainty about the future profitability of investment projects and, as a result, suspends
economic development. The Lerner index and local labor supply are not significantly
associated with regional economic growth. Bank size, the amount of non-interest income, and
capital strength (equity ratio) do not appear to be linked to regional economic growth in most
model specifications.
[Insert Table II here]
INITIAL ENDOWMENTS AND REGIONAL ECONOMIC DEVELOPMENT
We further want to test proposition 2 from the theoretical model. We assess whether the
impact of financial development varies in regions with different initial endowments. If small
regional banks can prevent financial resources flowing from poor to rich regions, then such
banks will help cover local capital demand and the impact of regional bank efficiency should
be stronger in regions with low initial endowments. We create an interaction term and add
this variable to all the regression models. The interaction term is the product of bank
efficiency and the initial regional GDP per capita in 1995. The introduction of this interaction
term enables us to differentiate the relationship between savings bank efficiency and regional
economic growth in less versus more developed regions.

26
The results presented in Table III show a significantly positive coefficient for the bank
efficiency variables. This illustrates that savings bank efficiency improvements are effective
in promoting regional economic growth. The coefficients of the interaction term are negative
and highly significant, indicating that the effect is stronger in less developed regions, rather
than in highly developed regions. This confirms our theoretical findings and shows that
regional banks are important in promoting local economic development in poor regions. The
results also support the theoretical findings of Lucas (1990) and Boyd and Smith (1997), who
discuss the danger of capital flows from the poor to rich regions.
Overall, our findings suggest that the presence of savings banks as regional funding providers
is positively associated with local economic growth and that this effect is stronger in
relatively poor regions. Small banks appear to be important funding providers in such regions
and as such they may limit capital flows from poor to rich regions. The results are consistent
with our theoretical model.
[Insert Table III here]
THE ROLE OF ACCESS TO FINANCE
Without access to external finance, firms cannot get enough funding for their projects, which
results in slower economic growth for a region. In our theory, we have shown that financing
constraints are negatively correlated with regional economic growth, especially in regions
with weak financial development. In this subsection, we assess whether the impact of
regional saving banks on economic development varies in regions with different credit

27
market conditions. We employ two measures for regional access to finance: commercial bank
branches per 100,000 persons (Branch) and private lending per person in a certain region
(Private_credit). We subdivide the sample according to the medians of the variable Branch.
Regions with average commercial bank branches per 100,000 persons greater than the sample
median are classified as markets with high access to finance. The regions with average
commercial bank branches per 100,000 persons lower than the sample median are classified
as markets with low access to finance. As a robustness check we also subdivide the sample
according to the median of the variable Private_credit.
Table IV contains the results. Regression models (1), (3), and (5) report results in regions
with relatively high access to finance. Regression models (2), (4), and (6) report results in
regions with relatively low access to finance. The significantly positive coefficients for the
bank efficiency variables, especially in regions with low access to finance, indicate that
regional savings banks are more important external finance providers in regions with higher
financing constraints. This confirms our theoretical findings and shows that regional banks
are important in promoting local economic development in regions with binding incentive
constraint.
[Insert Table IV here]
SMALL BANKS vs. BIG BANKS
We have shown that local financial development is essential for regional economic growth,
especially in poor regions. A further interesting question is: do small banks or big banks have
a differential impact on a region’s economic development? In order to answer this question

28
we proxy for small and big bank presence using branch market shares of regional savings
banks, regional cooperative banks, and big commercial banks based on the number of
branches in the region. The regional savings banks and regional cooperative banks represent
small banks, while big interregional commercial banks are big banks. So far we have used
the growth rate of GDP per capita as a dependent variable. In the following analysis, we
approximate regional economic development using the growth rate of new business
registrations (Business_Reg_Growth) in a certain region as a robustness check. A further
advantage of this approach is that we can mitigate concerns over causality between regional
economic growth and bank performance. Again, we use the interactive variable to
differentiate the effect of financial development in poor versus rich regions. The interaction
term is defined as the product of market share of big or small banks and regional GDP per
capita. A negative sign on the interaction term will indicate that the effect is stronger in poor
regions.
[Insert Table V here]
Estimates in Table V reveal that regions in which there is a larger market share of regional
savings banks have higher growth in new business registrations, suggesting more prosperous
economic conditions. As the growth rate of regional business registration mainly represents
regional SME growth, these results confirm the findings of Carbó-Valverde et al. (2009) and
Degryse and Ongena (2007), who show that savings banks are more effective at promoting
SME development. We therefore find that savings banks have a positive and significant effect
on regional development proxied by the growth rate of business registrations, and the highly
significant negative coefficient on the interaction term indicates that the effect is particularly
strong in relatively poor regions. The cooperative banks, another representative from the

29
regional banking group, exhibit similar effect on promoting regional economic development.
The coefficient is positive but insignificant for the market share of cooperative banks, and the
coefficient of the interaction term is negative and insignificant, indicating a higher market
share of credit cooperatives is weakly associated with higher growth rates of new business
registration.
Our findings for large commercial banks, however, are in the opposite direction; the
coefficients for the market share of large banks have a negative and insignificant sign and the
coefficient for the interaction term is positive and insignificant. The results suggests that a
higher proportion of large commercial bank branches in a given region might not improve
regional development. A possible reason is that the presence of large commercial banks may
crowd out small banks that have a stronger focus on local SME finance. Also, our results
using the growth rate of new business registrations demonstrates that the effect of big
interregional banks on local economic development diverges between poor and prosperous
regions. This confirms the findings of our theoretical model, which shows that more
developed regions are attractive for big banks due to higher initial endowments. However, in
less developed regions with lower initial endowments, big banks only have a limited
presence. In this context, big banks have a more positive role on local economic development
for rich regions, as compared to poor regions. In contrast, small banks that do not have the
opportunity for mobility are engines for local growth, especially in poor regions that face
potential capital outflows from interregional banks. Our findings, therefore, remain robust to
alternative measures of regional economic development (growth in GDP per capita and new
business registrations).

30
4. CONCLUSION
The potential role of small banks in a financially-integrated market is not obvious. Earlier
empirical studies suggest that local financial development remains important and that there is
a causal link between local financial development and real economic growth. In this paper,
we investigate the influence of small banks and big banks on real economic development.
We propose a theoretical model of financial development and economic growth. The model
suggests that under credit rationing, rich regions are more attractive for external investments
because of higher initial endowments and because poor regions tend to have fewer projects
that need financing. Thus, in a financially-integrated market, capital will flow from poor to
rich regions. Accordingly, due to the lower level of economic activity, an expansion in
investment will have a rather high marginal productivity in poor regions. Thus, small banks
can improve local economic development and will have a higher impact on economic growth
in such regions. We then empirically test the roles of small banks and big banks on economic
growth. The GMM estimations show that small bank development improves local economic
growth, with the effect being stronger in less developed regions. Overall, both the theoretical
and empirical parts of our analysis suggest that small banks may be effective in preventing
capital outflows from poor to rich regions.
Our analysis illustrates the important role of small regional banks in a financially-integrated
market. Interregional bank consolidation may destroy the regional rootedness of these banks,
turning small regional banks into big banks and thus losing the potential benefits. Small and
regional banks play a role in many financial systems, not only Germany. For example, local
and regional cooperative and savings banks play an important role in many European banking

31
systems, particularly in Austria, Belgium, France, Italy, Norway, Portugal, Spain and
Sweden. In the US, banks that fall under the community reinvestment act (CRA) from 1977
also have a regional development focus. Also in both US and UK, governments have created
Community Development Financial Institutions. In this context, the findings have relevance
beyond the German setting as many other European countries as well as the U.S. (community
banks) host a variety of banks with important local or/and regional focus.
Overall, our paper discusses the optimal architecture of a regional and interregional banking
system. Financial policies that supports regional granularity may prevent a capital drain and
thus foster development, especially in less developed regions.

32
Table I: Variable definitions and summary statistics
Variable Definition N Mean Std.Dev. Min. Max.
Panel A: Key variables of interest
Regional Economic Development
GDP_Capita_Growth (%) Annual growth rate of regional gross domestic
product per capita.
3554 1.995 3.142 -15.983 25.484
Business_Reg_Growth (%) Annual growth rate of regional new business
registrations.
2765 1.957 11.913 -30.120 83.260
Bank Efficiency and Risk Taking
ROA (%) Return on total assets of savings banks. 3575 0.234 0.170 -3.967 2.038
ROE (%) Return on total equity of savings banks. 3575 5.556 4.330 -4.045 69.300
Cost Efficiency
Cost efficiencies of savings banks estimated using
stochastic frontier analysis. It represents a relative
performance measure of savings banks.
2776 0.828 0.039 0.519 0.919
Z_Score Z-score is estimated as
ROA+(equity/assets))/Sd(ROA).
2870 3.839 21.095 0.210 37.503
Panel B: Control variables
Regional Specific Controls
Lerner Lerner index measure of regional bank
competition.
2776 0.236 0.057 0.049 0.520
Branch
Bank branches per 100,000 residents. 3935 49.076 20.271 12.918 116.788

33
Private_Credit (Tds. Euro) Private credit per capita in thousand Euro. 3935 27.603 35.558 2.393 361.177
GDP_Capita (Tds. Euro) Absolute value of regional gross domestic product
per capita in thousand Euro.
3949 23.226 7.782 10.118 74.118
Inflation (%) Annual German Consumer Price Index (CPI). 3950 0.619 0.673 -0.678 1.874
Labor_Supply (Mio.) Total population in a region in Million. 3950 0.233 0.188 0.040 1.736
Savings_rate (%) Regional annual savings rate in percentage. 3950 10.193 1.005 6.800 12.300
MS_Savings_Bank(%) Market share of regional savings bank branches in
a region as a proportion of total bank branches.
3555 34.822 6.164 8.475 61.333
MS_Coop_Bank (%)
Market share of regional cooperative bank
branches in a region as a proportion of total bank
branches.
3555 27.881 9,696 4.950 58.879
MS_Big_Commercial (%) Market share of large commercial bank branches
divided by the total number of bank branches in a
region.
3555 37.297 9.179 12.500 74.510
Group Dummy (West) A dummy variable equals to one if a region
belongs to the former West Germany.
3950 0.833 0.373 0.000 1.000
Bank Specific Controls
In_Totalasset Natural logarithm of savings bank total assets. 3575 6.081 0.400 4.993 7.529
Share_of_Fee_Inc. (%) Percentage of non-interest income relative to
interest income.
3575 21.311 5.000 6.227 53.042
Equity_Ratio (%) Equity to total assets ratios for savings banks. 3575 4.338 0.923 2.032 8.720

34
Table II: Impact of regional savings bank efficiency on local economic growth
This table reports results from GMM estimations of the effects of small savings banks on regional economic development. The dependent variable is the growth rate of
regional GDP per capita. The measures of savings bank efficiency include two indirect indicators—(1) ROA and (2) ROE—and a direct measure derived from stochastic
frontier estimation, namely (3) Cost Efficiency. Regression models (1) to (3) contain results with Windmeijer (2005) small-sample two-step standard errors. Regression
models (4) to (6) contain results with clustered standard errors. The period covers the years 1995 to 2004. *** denotes significance at the 1%-level; ** denotes significance at
the 5%-level; * denotes significance at the 10%-level.
Dependent Variable: GDP_Capita_Growth
(1) (2) (3) (4) (5) (6)
Coeff. p-Val. Coeff. p-Val. Coeff. p-Val. Coeff. p-Val. Coeff. p-Val. Coeff. p-Val.
L. GDP_Capita_Growth -0.034 0.755 -0.047 0.667 -0.055 0.470 -0.034 0.851 0.010 0.951 -0.263* 0.077
L.ROA 2.850* 0.055 2.850 0.105
L.ROE 0.116* 0.068 0.042* 0.098
L.Cost_Eff. 3.883* 0.100 -19.289** 0.013
Lerner -4.614 0.717 -7.142 0.524 1.913 0.738 -4.614 0.767 -0.892 0.941 -16.609* 0.071
Inflation -10.453*** 0.000 -10.323*** 0.000 -10.389*** 0.000 -10.453*** 0.000 -10.268*** 0.000 -10.176*** 0.000
Labor_Supply -0.363 0.543 -0.302 0.586 -0.596 0.148 -0.363 0.674 -0.390 0.608 -0.183 0.804
In_Totalasset 0.129 0.862 0.337 0.656 -0.252 0.571 0.129 0.894 0.569 0.539 0.689 0.525
Share_of_Fee_Inc. -0.009 0.907 -0.015 0.834 0.074 0.236 -0.009 0.927 -0.021 0.813 0.157 0.176
Equity_Ratio -0.111 0.757 0.187 0.676 0.124 0.371 -0.111 0.797 0.058 0.898 -1.356*** 0.008
Group Dummy (west) -0.552 0.316 -0.868 0.162 0.134 0.750 -0.552 0.468 -0.616 0.404 -0.152 0.855
_cons 12.610 0.150 10.838 0.148 7.155 0.128 12.610 0.238 8.740 0.291 29.576*** 0.001
Year_Dummies Yes Yes Yes Yes Yes Yes
Number of Obs. 2873 2873 2862 2873 2873 2862
Number of Banks 367 367 366 367 367 366
Hansen test (p-value) 0.159 0.244 0.093 0.159 0.220 0.089
AB test AR(1) (p-value) 0.000 0.000 0.000 0.005 0.001 0.019
AB test AR(2) (p-value) 0.766 0.844 0.906 0.839 0.690 0.224
Clustered SE No No No Yes Yes Yes

35
Table III: Initial Endowments and Regional Economic Development
This table reports the impact of small savings banks on regional economic development with different initial
conditions. The dependent variable is the growth rate of regional GDP per capita. The measures of savings bank
efficiency include two indirect indicators—(1) ROA and (2) ROE—and a direct measure derived from
stochastic frontier estimation, namely (3) Cost Efficiency. The proxy for regional initial condition is regional
GDP per Capita in 1995. The period covers the years 1995 to 2004. *** denotes significance at the 1%-level;
** denotes significance at the 5%-level; * denotes significance at the 10%-level.
(1) (2) (3)
Coeff. p-Val. Coeff. p-Val. Coeff. p-Val.
L. GDP_Capita_Growth -0.229** 0.021 -0.215** 0.031 -0.318** 0.010
L.ROA 15.923*** 0.000
L.ROE 0.493*** 0.000
L.Cost_Eff. 18.024*** 0.003
ROA×(Initial GDP_Capita) -0.683*** 0.000
ROE×(Initial GDP_Capita) -0.022*** 0.001
Cost_Eff.×(Initial GDP_Capita) -1.108*** 0.000
Lerner -9.150 0.492 -3.002 0.806 -23.589 0.159
GDP_Capita 0.182*** 0.002 0.161*** 0.007 0.812*** 0.000
Inflation -9.010*** 0.000 -8.682*** 0.000 -3.566** 0.040
Labor_Supply -0.687 0.280 -0.878 0.165 0.360 0.709
In_Totalasset 0.223 0.764 0.543 0.488 -1.721** 0.048
Share_of_Fee_Inc. -0.016 0.854 -0.002 0.976 0.410*** 0.002
Equity_Ratio -0.205 0.627 0.180 0.726 -0.314 0.610
Group Dummy (west) -1.195 0.116 -1.193 0.137 1.311 0.324
_cons 8.895 0.311 3.869 0.629 -3.788 0.748
Year_Dummies yes yes yes
Number of Obs. 2866 2866 2855
Number of Banks 366 366 365
Hansen test (p-value) 0.330 0.299 0.946
AB test AR(1) (p-value) 0.000 0.000 0.023
AB test AR(2) (p-value) 0.256 0.309 0.087

36
Table IV: The Role of Access to Finance
This table reports results from GMM estimations of the effects of small savings banks on regional economic development under low or high access to finance. The dependent
variable is the growth rate of regional GDP per capita. The measures of savings bank efficiency include two indirect indicators—(1) ROA and (2) ROE—and a direct
measure derived from stochastic frontier estimation, namely (3) Cost Efficiency. Regression models (1), (3), and (5) with H contain results in regions of relative high access
to finance. Regression models (2), (4), and (6) with L contain results in regions of relative low access to finance. The period covers the years 1995 to 2004. *** denotes
significance at the 1%-level; ** denotes significance at the 5%-level; * denotes significance at the 10%-level.
(1) (2) (3) (4) (5) (6)
H L H L H L
Coeff. p-Val. Coeff. p-Val. Coeff. p-Val. Coeff. p-Val. Coeff. p-Val. Coeff. p-Val.
L. GDP_Capita_Growth -0.083 0.495 0.071 0.621 -0.243** 0.033 0.016 0.917 -0.184** 0.032 -0.140** 0.046
L.ROA -0.612 0.866 1.542* 0.075
L.ROE 0.034 0.780 0.053 0.280
L.Cost_Eff. -3.535 0.489 -13.230*** 0.007
Lerner 23.051 0.230 -27.635* 0.063 14.296 0.481 -24.171* 0.069 9.503 0.199 -15.411*** 0.002
Inflation -8.700*** 0.000 -12.737*** 0.000 -7.488*** 0.000 -12.344*** 0.000 -9.344*** 0.000 -12.722*** 0.000
Labor_Supply -2.777 0.154 -0.158 0.859 -1.319 0.412 -0.408 0.637 8.026** 0.022 -1.902* 0.086
In_Totalasset -1.390 0.334 0.530 0.619 -1.017 0.506 0.907 0.386 2.088** 0.046 -0.016 0.973
Share_of_Fee_Inc. -0.014 0.910 -0.170 0.164 0.087 0.456 -0.126 0.284 -0.128 0.183 -0.016 0.796
Equity_Ratio -0.527 0.420 0.039 0.953 -0.685 0.323 -0.152 0.828 0.817*** 0.000 -0.103 0.704
Group Dummy (west) 1.916* 0.061 -1.504* 0.090 1.462 0.121 -1.047 0.287 -3.912*** 0.000 -1.075** 0.011
_cons 15.124 0.317 21.358* 0.085 12.058 0.386 17.534* 0.099 0.131 0.991 30.498*** 0.000
Year_Dummies Yes Yes Yes Yes Yes Yes
Number of Obs. 1249 1624 1249 1624 1249 1623
Number of Banks 157 210 157 210 157 367
Hansen test (p-value) 0.763 0.850 0.505 0.820 0.267 0.129
AB test AR(1) (p-value) 0.002 0.000 0.006 0.000 0.000 0.000
AB test AR(2) (p-value) 0.548 0.290 0.073 0.433 0.150 0.791

37
Table V: Small vs. Big Banks
This table compares the impact of small vs. big bank presence on local economic development. The dependent
variable is growth rate of regional business registration. The proxies for small and big bank presence in a certain
region are the market shares of the regional savings banks and big commercial banks based on the number of
branches. The period covers the years 1995 to 2004. *** denotes significance at the 1%-level; ** denotes
significance at the 5%-level; * denotes significance at the 10%-level.
Dependent Variable: Growth Rate of Business Registration
Coeff. p-Val. Coeff. p-Val. Coeff. p-Val.
(1) (2) (3)
L.Business_Registration 0.235 0.171 -0.270 0.125 -0.040 0.820
MS_Savings_Bank 1.289** 0.024
MS_Saving × GDP_Capita -0.041** 0.029
MS_Coop_Bank 0.065 0.799
MS_Coop × GDP_Capita 0.001 0.916
MS_Big_Commercial -0.263 0.367
MS_Big × GDP_Capita 0.005 0.588
GDP_Capita 1.421** 0.029 0.030 0.894 -0.131 0.715
Inflation -3.969 0.787 -47.318*** 0.001 -26.565* 0.072
Labor_supply -2.452 0.142 2.821 0.170 2.364 0.228
Savings_rate 0.364 0.548 0.827 0.137 0.788 0.106
Group Dummy (west) 0.458 0.850 -6.617** 0.044 -6.894** 0.024
Year_Dummies yes yes yes
Constant -29.914 0.315 56.043*** 0.001 47.143*** 0.004
Number of observations 1975 1975 1975
Hansen test (p-value) 0.234 0.352 0.297
AB test AR(1) (p-value) 0.000 0.009 0.001
AB test AR(2) (p-value) 0.154 0.469 0.681

38
A. Appendix
A.1. Proofs
Proof of Proposition 1: We need to differentiate between the two cases, with and without
binding incentive constraint (IC). If the constraint does not bind, Equation (9) applies. In each
period, the number of firms is independent from last period’s level. As a consequence, output
is identical, thus there is no growth. If (IC) does bind, Equation (6) applies, and nt depends on
last period’s wage level Wt–1, and hence from the initial endowment At = Wt–1. Output at date t
is Yt = β Kt
α
= β nt
α
(pHR – c)α
. Now the previous period’s output Yt−1 is independent of
today’s cost c, and hence, we only need to consider ∂Yt/∂c instead of ∂(Yt − Yt−1)/∂c.
An increase in regional lending Ft now directly increases the number of projects in the region.
Ft = nt (I − At), and hence, nt = Ft/(I − At). More projects produce more capital such that Kt =
nt (pH R − c). Hence, output Yt increases. Because c is negatively associated with Ft, the
partial derivative with respect to c is negative, ∂(Yt − Yt−1)/∂c < 0. This proposition also holds
true in the absence of credit rationing. Only the multiplier is smaller because outside finance
crowds out inside finance, and hence, nt = Ft/I.
Proof of Proposition 2: According to (6), nt is an increasing function in At. According to (7),
Wt increases in nt, and hence, it also increases in At. If (8) fails to hold, there is no credit
rationing. In that case, Wt does not depend on At.
Take the derivative dWt/dFt, which is positive. Then consider this derivative at different
values of At. If the derivative is smaller for higher values of At, the first part of the proof is

39
complete. Noting that (2), (3), and nt = Ft/(I − At) yields
𝑊𝑡 = 𝛽(1 − 𝛼)(
𝐹𝑡(𝑝 𝐻 𝑅−𝑐)
𝐼−𝐴 𝑡
) 𝛼
. (13)
The derivative with respect to Ft is
𝜕𝑊𝑡
𝜕𝐹𝑡
=
𝛽(1−𝛼)𝛼(𝑝 𝐻 𝑅−𝑐)
𝐼−𝐴 𝑡
(
𝐹𝑡(𝑝 𝐻 𝑅−𝑐)
𝐼−𝐴 𝑡
) 𝛼−1
(14)
According to this term, ∂Wt/∂Ft is larger for larger At. However, one needs to take into
account that the natural (or equilibrium) number of projects (and thus the amount of regional
finance) also depends on At. Considering Ft = nt (I – At) and (5) yields
𝜕𝑊𝑡
𝜕𝐹𝑡
= (1 + 𝑟)(1 − 𝛼)
𝑅 𝑝−𝑐
. (15)
Hence, the two effects cancel out exactly. However, this result holds true only if (8) holds and
the incentive constraint thus binds. In more developed regions, nt = Ft/I, and hence, the first
effect is weaker. In this case ∂Wt/∂Ft is smaller for larger At. Since F depends on bank
efficiency c ∂Wt/∂(–c) is smaller for larger At, which completes the proof.
A.2. Bank Cost Efficiency
Following Berger and Mester (1997), the cost frontier is specified as a Fourier-flexible
functional form for each year t in our analysis:

40
[ ]
( )[ ]
[ ] ε
θϕρ
γδ
βα
+++++++
++++
++
+++
+++
+++=
∑∑ ∑
∑∑
∑
∑∑ ∑ ∑
∑∑ ∑∑
∑ ∑
= = ≠≥
= =
=
= = = =
= = = =
= =
3
1
3
1
3
,
3
1
3
1
3
1
3
1
3
1
3
1
3
1
3
1
3
1
3
1
3
1
2
11
1
3
1
3
1
0
)sin()cos(
)sin(cos
)sin()cos(
lnlnlnln
)lnlnlnln(
2
1
lnlnln
i j ikjk
kjiijkkjiijk
i j
jiijjiij
i
iiii
i m i l
lliimiim
i j l m
mllmjiij
i l
llii
ZZZbZZZa
ZZbZZa
ZbZa
PTQTPQ
TtPPQQ
TtPQaTC
(16)
where
ln TC = the natural logarithm of total costs (operating plus financial cost);
ln Qi = the natural logarithm of bank outputs including bank loans, securities, and off-balance
sheet items;
ln Pl = the natural logarithm of input prices including wage rate, interest rate and physical
capital price;
T = time trend;
Zi= the adjusted values of the log output ln Qi such that they span the interval [0, 2π];
a, β, δi, γ, φ, θ, ρ, a, b and t are coefficients to be estimated.
We use the stochastic cost frontier approach to generate annual estimate of cost efficiency for
each bank. The sample covers 457 regional savings banks from 1994 to 2005. As shown in
Table 1, the average cost efficiency for regional savings bank in Germany is 82.8% over the
entire sample period.

41
References
Altunbas, Y., Evans. L. and Molyneux, P. (2001) Ownership and Efficiency in Banking,
Journal of Money, Credit & Banking, 33, 926-954.
Arellano, M. and Bover, O. (1995) Another look at the instrumental-variable estimation of
error-components models, Journal of Econometrics 68 , 29–52.
Arestis, P. and Demetriades, P. (1997) Financial Development and Economic Growth:
Assessing the Evidence, Economic Journal, 107 , 783-799.
Beck, T., Levine, R. and Loayza, N. (2000) Finance and the Source of Growth, Journal of
Financial Economics, 58, 261–300.
Beck, T. and Levine, R. (2004) Stock Markets, Banks, and Growth: Panel Evidence, Journal
of Banking & Finance, 28, 423-442.
Beck, T., Demirgüc-Kunt, A., Laeven L. and Levine, R. (2008) Finance, Firm Size, and
Growth, Journal of Money, Credit and Banking, Blackwell Publishing, 40, 1379-1405..
Beck, T., Levine, R. and Levkov, A. (2010) Big Bad Banks? The Winners and Losers from
Bank Deregulation in the United States, Journal of Finance, 65, 1637–1667.
Bekaert, G., Harvey, C.R. and Lundblad, C. (2005) Does Financial Liberalization Spur
Growth?, Journal of Financial Economics, 7, 3-55.

42
Berger, A. N. (1995) The Profit-Structure Relationship in Banking--Tests of Market-Power
and Efficient-Structure Hypotheses, Journal of Money, Credit and Banking, 27, 404-31.
Berger, A. N., Hasan, I. and Klapper, L.F. (2004) Further Evidence on the Link between
Finance and Growth: An International Analysis of Community Banking and Economic
Performance, Journal of Financial Services Research, 25, 169–202.
Bernanke, B. S. and Gertler, M.L. (1989) Agency Costs, Net Worth, and Business
Fluctuations, American Economic Review, 79, 14–31.
Blinder, A.S. (2010) The Squam Lake Report: Fifteen Economists in Search of Financial
Reform, Journal of Monetary Economics 57, 892-902.
Bloch, T. (2008) The Effects of Bank Mergers on Small Business Lending in Germany,
Goethe University Accounting & Finance Working Paper, No 193, Frankfurt. November.
Blundell, R. and Bond, S. (1995) Initial Conditions and Moment Restrictions in Dynamic
Panel Data Models, Journal of Econometrics, 87, 115-143.
Bofondi, M. and Gobbi, G. (2003) Bad Loans and Entry in Local Credit Markets,
unpublished working paper, Bank of Italy.
Boyd, J. H. and Smith, B. D. (1997) Capital Market Imperfections, International Credit
Markets, and Nonconvergence, Journal of Economic Theory, 73, 335–364.

43
Brewer, III E. and Jagtiani, J. (2013) How Much Did Banks Pay to Become Too-Big-to-Fail
and to Become Systemically Important?, Journal of Financial Services Research, 43, 1-35.
Buch, C. M., Koch, C. T. and Koetter, M. (2010) Do banks benefit from internationalization?
Revisiting the market power-risk nexus, Deutsche Bundesbank Discussion Paper, Series 2:
Banking and Financial Studies, No 09/2010, Frankfurt, Deutsche Bundesbank.
Buiter, W. (2009) Too Big to Fail is Too Big, The Financial Times. June 24th
, Retrieved
2012-06-21.
Carbó-Valverde, S., Rodríguez-Fernández, F. and Udell, G. F. (2009), Bank Market Power
and SME Financing Constraints, Review of Finance, 13, 309-340.
Carow, K.A., Kane, E. and Narayanan, RP. (2011) “Safety-Net Losses from Abandoning
Glass-Steagall Restrictions, Journal of Money, Credit and Banking, 43, 1371-1398.
Claessens, S. and Laeven, L. (2003) Financial Development, Property Rights, and Growth,
Journal of Finance, 58, 2401-2436.
Claessens, S. and Laeven, L. (2004) What Drives Bank Competition? Some International
Evidence, Journal of Money, Credit and Banking, 36, 563-583.
Claessens, S. and Laeven, L. (2005) Financial Dependence, Banking Sector Competition and
Economic Growth, Journal of the European Economic Association, 3, 179-207.

44
Degryse, H. and Ongena, S. (2007) The Impact of Competition on Bank Orientation, Journal
of Financial Intermediation, 16, 399–424.
DeYoung, R., Evanoff, D. and Molyneux, P. (2009) Mergers and Acquisitions of Financial
Institutions: A Review of the Post-2000 Literature, Journal of Financial Services Research
36, 87-110.
DeYoung, R. (2012) Economies of Scale in Banking, in: Efficiency and Productivity Growth:
Modelling in the Financial Services Industry, edited by Fotios Pasiouras, London, John Wiley
& Sons.
DeYoung, R., Gron, A., Torna, G. and Winton, A. (2012) Risk Overhang and Loan Portfolio
Decisions: Small Business Loan Supply Before and During the Financial Crisis, working
paper, available at SSRN: http://ssrn.com/abstract=2140952.
Diamond, D.W.and Dybvig, P. H. (1983) Bank Runs, Deposit Insurance, and Liquidity.
Journal of Political Economy, 91, 401–19.
Diamond, D. W. (1984) Financial Intermediation and Delegated Monitoring, Review of
Economic Studies, 51, 393–414.
Feldman, R. J. and Stern, G.H. (2010) The Squam Lake Report: Observations from Two
Policy Professionals, Journal of Monetary Economics, 57, 903-912.

45
Fernandez de Guevara, J., Maudos, J. and Perez, F. (2005) Market Power in European
Banking Sectors, Journal of Financial Services Research, 27, 109-137.
Fischer, M J., Hainz, C., Rocholl, J. and Steffen, S. (2012) Government Guarantees and Bank
Risk Taking Incentives, AFA 2012 Chicago Meetings Paper,
http://ssrn.com/abstract=1786923 or http://dx.doi.org/10.2139/ssrn.1786923
Gintschel, A. and Hackethal, A. (2003) Multi-Bank Loan Pool Contracts, working paper,
http://ssrn.com/abstract=493362 or http://dx.doi.org/10.2139/ssrn.493362.
Goddard, J., Molyneux, P. and Wilson, J.O.S. (2004) Dynamics of Growth and Profitability
in Banking, Journal of Money, Credit and Banking, 36, 1069-90.
Greenwood, J. and B. Jovanovic (1990) Financial Development, Growth, and the Distribution
of Income, Journal of Political Economy, 98, 1076-1107.
Gropp, R., Gründl, C. and Guettler, A. (2013) The Impact of Public Guarantees on Bank Risk
Taking: Evidence from a Natural Experiment, Review of Finance, doi: 10.1093/rof/rft014.
Guiso, L., Sapienza, P. and Zingales, L. (2004) Does Local Financial Development Matter?,
Quarterly Journal of Economics, 119, 929 – 969.
Hackethal, A. (2004) German Banks and Banking Structure, in: J. P. Krahnen, and R. H.
Schmidt (eds.), The German Financial System, 71–105. Oxford, Oxford University Press.

46
Hakenes, H. and Schnabel, I. (2010) The Threat of Capital Drain: A Rationale for Regional
Banks?, Journal of Institutional and Theoretical Economics, 166, 662 – 689.
Shaffer, S., Hasan, I. and Zhou, M., 2014. New Small Firms and Dimensions of Economic
Performance, Economic Development Quarterly, (forthcoming).
Holmstrom, B. and Tirole, J. (1997) Financial Intermediation, Loanable Funds, and the Real
Sector, Quarterly Journal of Economics 112, 663 – 692.
International Monetary Fund (IMF) (2013) Global Financial Stability Report, Transitional
Challenges to Stability, October, Washington DC, IMF.
King, R. G. and Levine, R. (1993) Finance and Growth: Schumpeter Might be Right,
Quarterly Journal of Economics, 108, 717–737.
Kiyotaki, N. and Moore, J. (1997) Credit Cycles, Journal of Political Economy, 105, 211–
248.
Koetter, M. and Wedow, M. (2010) Finance and growth in a bank-based economy: Is it
quantity or quality that matters?, Journal of International Money and Finance, 29, 1529-
1545.

47
Koetter, M., Kolari, J.W. and Spierdijk, L. (2012) Enjoying the Quiet Life under
Deregulation? Evidence from Adjusted Lerner Indices for U.S. Banks, Review of Economics
and Statistics, 94, 462–480.
Koetter, M. and T. Poghosyan (2009) The identification of technology regimes in banking:
Implications for the market power-financial fragility nexus, Journal of Banking and Finance
33, 1413-1422.
Levine, R. (1991) Stock Markets, Growth, and Tax Policy, Journal of Finance, 46, 1445-
1465.
Levine, R. (2005) Finance and Growth: Theory and Evidence, in: P. Aghion and S. Durlauf,
(eds.), Handbook of Economic Growth, Vol. 1, North Holland, Amsterdam, pp. 865-934.
Levine, R. and Zervos, S. (1998) Stock Markets, Banks, and Economic Growth, American
Economic Review, 88 , 559-586.
Levine, R., Loayza, N. and Beck, T. (2000) Financial Intermediation and Growth: Causality
and Causes, Journal of Monetary Economics, 46, 31-77.
Laeven, L. and Valencia, F. (2010) Resolution of Banking Crises: The Good, the Bad, and
the Ugly, IMF Working Papers 10/146, Washington DC, IMF.
Liikanen, E (2012) High-level Expert Group on Reforming the Structure of the EU Banking
Sector, , 2 October 2012, Brussels, EU.

48
Lucas, Robert E, Jr, (1990) Why Doesn't Capital Flow from Rich to Poor Countries?,
American Economic Review, 80, 92-96.
Morales, M.F. (2003) Financial Intermediation in a Model of Growth through Creative
Destruction, Macroeconomic Dynamics, 7,:363-93.
Petersen, M., and Rajan, R. (2002) Does Distance Still Matter: The Information Revolution in
Small Business Lending, Journal of Finance, 57, 2533–2570.
Rajan, R. G. and Zingales, L. (1998) Financial Dependence and Growth, American Economic
Review, 88, 559-586.
Rosenblum, H. (2011) Choosing the Road to Prosperity: Why We Must End Too Big to
Fail—Now? Federal Reserve Bank of Dallas Annual Report, 3-23.
Schmidt, R. H. and Tyrell, M. (2004) What Constitutes a Financial System in General and the
German Financial System in Particular? in: J. P. Krahnen, and R. H. Schmidt (eds.), The
German Financial System,. Oxford, Oxford University Press, pp. 19 – 68
Schumpeter, J. (1934) The Theory of Economic Development, Cambridge MA, Harvard
University Press.

49
Slotty, C. F. (2009) Does Relationship Lending Promote Growth? Savings Banks and SME
Financing, working paper, http://ssrn.com/abstract=1376251 or
http://dx.doi.org/10.2139/ssrn.1376251.
Stolz, S. M. and Wedow, M. (2010) Extraordinary Measures in Extraordinary Time- Public
Measures in Support of the Financial Sector in the EU and the United States, ECB Occasional
Paper Series, No. 117, Frankfurt: European Central Bank.
Tirole, J. (2006) The Theory of Corporate Finance, MIT Press, Cambridge, Massachusetts.
Wheelock, D. and Wilson, P. (2012). Do large banks have lower costs? New estimates of
returns to scale for US banks, Journal of Money, Credit, and Banking, 44, 171-99.
Windmeijer, F. (2005) A finite sample correction for the variance of linear efficient two-step
GMM Estimators, Journal of Econometrics, 126, 25–51.

BANK OF FINLAND RESEARCH DISCUSSION PAPERS
ISSN 1456-6184, online
1/2014 Bill Francis – Iftekhar Hasan – Jong Chool Park – Qiang Wu Gender
differences in financial reporting decision-making: Evidence from
accounting conservatism. 2014. 58 p. ISBN 978-952-6699-63-9, online.
2/2014 Esa Jokivuolle – Jussi Keppo Bankers' compensation: Sprint swimming in
short bonus pools? 2014. 40 p. ISBN 978-952-6699-64-6, online.
3/2014 Iftekhar Hasan – Chun-Keung (Stan) Hoi – Qiang Wu – Hao Zhang Beauty is
in the eye of the beholder: The effect of corporate tax avoidance on the cost
of bank loans. 2014. 67 p. ISBN 978-952-6699-65-3, online.
4/2014 Kaushik Mitra – Seppo Honkapohja Targeting nominal GDP or prices:
Guidance and expectation dynamics. 2014. 47 p. ISBN 978-952-6699-66-0,
online.
5/2014 Hendrik Hakenes – Iftekhar Hasan – Phil Molyneux – Ru Xie Small banks and
local economic development. 2014. 49 p. ISBN 978-952-6699-69-1, online.

http://www.suomenpankki.fi/en • email: Research@bof.fi
ISBN 978-952-6699-69-1, ISSN 1456-6184, online

mc project

More Related Content

What's hot (20)

Similar to mc project (20)

Recently uploaded (20)

mc project