Worki ng PaPer Seri eS

no 1568 / j uLY 2013

Bank Lending and MonetarY
tranSMiSSion in the euro area

Roberto A. De Santis and Paolo Surico
In 2013 all ECB
publications
feature a motif
taken from
the 5 banknote.
note: This Working Paper should not be reported as representing
the views of the European Central Bank (ECB). The views expressed are
those of the authors and do not necessarily refect those of the ECB.
European Central Bank, 2013
Address Kaiserstrasse 29, 60311 Frankfurt am Main, Germany
Postal address Postfach 16 03 19, 60066 Frankfurt am Main, Germany
Telephone +49 69 1344 0
Internet http://www.ecb.europa.eu
Fax +49 69 1344 6000
All rights reserved.
ISSN 1725-2806 (online)
EU Catalogue No QB-AR-13-065-EN-N (online)
Any reproduction, publication and reprint in the form of a different publication, whether printed or produced electronically, in whole
or in part, is permitted only with the explicit written authorisation of the ECB or the authors.
This paper can be downloaded without charge from http://www.ecb.europa.eu or from the Social Science Research Network electronic
library at http://ssrn.com/abstract_id=2294854.
Information on all of the papers published in the ECB Working Paper Series can be found on the ECBs website, http://www.ecb.
europa.eu/pub/scientifc/wps/date/html/index.en.html
Acknowledgements
We are grateful to David Marques-Ibanez, Frederic Malherbe, Elias Papaioannou, Diego Rodriguez-Palenzuela and three anonymous
referees for very useful comments and conversations. The views expressed in this paper are those of the authors, and do not necessarily
refect those of the European Central Bank. Financial support from the European Research Council (Starting Grant 263429) is gratefully
acknowledged.
Roberto A. De Santis
European Central Bank; e-mail: roberto.de_santis@ecb.europa.eu
Paolo Surico
London Business School and CEPR; e-mail: psurico@london.edu

Abstract: To what extent does the availability of credit depend on monetary policy?
And, does this relationship vary with bank characteristics? Based on a common
source of balance sheet data for the four largest economies of the euro area over the
period 1999-2011, we uncover three main regularities. First, the effect of monetary
policy on bank lending is significant and heterogeneous in Germany and Italy, which
are characterised by a large number of banks; but it is very weak in Spain and more
homogeneous in France, where the banking industry has a higher degree of market
concentration. Second, there is some evidence that monetary policy exerts larger
effects on cooperative and savings banks with lower liquidity and less capital in
Germany and savings banks with smaller size in Italy. Third, heterogeneity across
groups of banks belonging to the same category in any particular country is found to
be less pronounced.
Keywords: credit availability, monetary policy, heterogeneous effects; commercial,
cooperative, and savings banks.

JEL classification: C33, E44, E52, G21
1

Non-technical summary
The transmission of monetary policy in the euro area has been the focus of a
comprehensive set of studies on pre-1999 samples undertaken by the Eurosystem
Monetary Transmission Network, jointly by the European Central Bank (ECB) and
the euro area National Central Banks (Angeloni, et al., 2003). This coordinated
research effort has documented that pre-1999 an increase in interest rates tended to
reduce loan growth in France, Germany, Italy and Spain. Moreover, they found that
banks in these four countries with more liquid asset holdings showed weaker loan
adjustments.

This paper studies a similar question using data post-1999, when the ECB was in
charge of monetary policy in the euro area. Our findings suggest that heterogeneity
in the transmission of monetary policy to banks lending activities appears associated
with heterogeneity across countries and across typologies of banks in the same
country, while it is broadly homogenous within the bank typology group in each
country. The effects of interest rate changes engineered by decisions at the European
Central Bank are in Germany and Italy far stronger than in Spain and France.
Interestingly, the German and Italian banking industries are populated by a relatively
larger number of saving and cooperative banks. This contrasts with the French and
Spanish markets, which appear dominated by fewer commercial banks. Moreover,
heterogeneity across countries and across typologies of banks within each country
seems far more pronounced and significant than heterogeneity across groups of
banks belonging to the same category in any particular country.

Our findings also suggest that changes in the cost of funding engineered by monetary
policy actions exert their maximum impact on cooperative and saving banks in
Germany, especially those with lesser liquidity and lower capital, and saving banks
in Italy, especially those with smaller size. Large commercial banks, on the other
hand, appear more capable to isolate their lending activities from changes in
monetary policy conditions. Small banks are best placed to refinance the real
economy, in particular small- and medium-sized firms, which are the biggest
generator of employment in the economy. The analysis suggests that the increase in
the number of cooperative and savings banks that have access to the ECB standard
and non-standard measures during the recent financial crisis is likely to improve the
transmission mechanism of monetary policy in the euro area.

Another important policy issue is the evidence that unexpected monetary policy
changes had limited implications for the bank loans in Spain over the 1999-2011
period. The limited impact of interest rates on bank lending in Spain suggests that in
a monetary union country-specific excessive growth of credit should be counteracted
with instruments that limit the fall in lending standards during boom times.


2
1. INTRODUCTION
A key requirement for an optimal currency area is that the business cycle of the
joining countries be sufficiently synchronized and the structure of the economy be
sufficiently similar for the transmission of monetary policy to be homogeneous
within the currency union. While a number of empirical studies have provided
tentative evidence that euro area countries may not be too far from such requirement
along a number of important dimensions, including price changes in product markets
(see De Grauwe and Mongelli, 2005 and the reference therein), empirical and
anecdotal evidence suggest that the banking industry may be a significant source of
heterogeneity in the transmission of monetary policy In particular, the response of
bank lending to monetary conditions may vary across countries and within the
banking sector, thereby making endogenously heterogeneous a common monetary
policy.

A similar question was addressed out using data pre-1999 by the Eurosystem
Monetary Transmission Network (MTN) in the euro area (Angeloni, et al., 2003).
2

The results from the Eurosystem MTN conducted using data pre-1999 indicated that
the impact of interest rate changes was not highly heterogeneous across countries,
given that an increase in interest rates tended to reduce loan growth in France,
Germany, Italy and Spain, particularly of banks with less liquid asset holdings (see
e.g. Ehrmann, et al., 2003). In this paper, we review the analysis using post-1999
observations.

A main challenge to investigating these issues empirically is that disentangling credit
demand versus credit supply is not straightforward and it is the main identification
problem of the literature pioneered by Kashyap and Stein (1995 and 2000).
Following their contributions, a popular approach to study how lending activities
vary with monetary conditions is to focus on micro data at bank level and project a
measure of credit conditions (typically loans) on a measure of monetary policy
(typically a short-term interest rate), bank-specific characteristics, business cycle
indicators and their interactions. Particular emphasis is given to the so-called bank
lending channel, namely the impact of the interaction between monetary policy and
banks individual characteristics on lending activities.

Most of the available evidence on the euro area, however, is typically based on
individual economies, with dataset that are not readily comparable in terms of data
source, level of disaggregation and sample period. More importantly, the literature is
scant of comparative studies that try to quantify after the introduction of the euro the
extent to which the transmission of monetary policy through bank lending is
heterogeneous across euro area countries, in a way that may depend on banks and

2
The MTN is probably the most comprehensive exercise to date to analyse the transmission
mechanism of monetary policy. It was an extensive three-year joint effort by the European Central
Bank and the other Eurosystem central banks.
3
borrowers characteristics, regulatory environment, financial development and
institutions characteristics.

The paper aims at filling this important gap in three steps:

1. Using a common data source on banks balance sheet data over the sample
1999-2011, we present some distinctive characteristics of bank lending across
the four largest economies of the euro area. Particular emphasis is given to
differences in bank-specific characteristics meant to capture supply
conditions (such as capital, liquidity, size, profitability) as well as to bank
typologies. A main finding, highlighted in Section 2, is that France, Germany,
Italy and Spain are very different in their composition of commercial, saving,
cooperative and real estate banks as well as in terms of individual
characteristics within each typology of banks.
2. Then, we turn to the possible sources of such heterogeneity. A non-
exhaustive list of candidates, which is discussed in Section 3, includes size,
liquidity and capital. Moreover, the transmission of the monetary policy can
be highly affected by a specific relationship of banks with their customers,
the network if banks, the concentration of the bank industry, the
characteristics of the borrowers, the structure and the development of the
nonfinancial sector. We will analyse this by looking at various typologies of
banks.
3. Finally, in Section 4 we present an empirical model of the bank lending
channel and in Section 5 we provide formal econometric evidence in favour
of the hypothesis that heterogeneity in banks characteristics lead to
heterogeneity in the monetary transmission across countries and typology of
banks. Two approaches are used: (i) the linear ordinary least square (OLS)
approach, which consists of interacting the policy instrument with the
candidate source of heterogeneity (i.e. size, liquidity and capital); (ii) the
nonlinear quantile regression approach, which splits the sample around the
exogenous policy instrument and the threshold values. The first more
traditional approach, which identifies heterogeneity with observed
characteristics such as size, capital and liquidity, allows us to study
heterogeneity of the monetary transmission mechanism across countries and
across banks. The second approach is used to study heterogeneity between
typologies of banks, recognising that banks are inherently different not only
in observed characteristics, but also in unobserved dimensions such as
business model, risk propensity, managerial ability and borrowers
characteristics.

Section 6 compares the results obtained with the post-1999 sample with the pre-1999
evidence and the final section provides some policy conclusions.

4
2. BANK LENDING ACROSS THE EURO AREA: SOME FACTS
The euro area banking system is the largest in the world. Total on-balance sheet
assets of the euro area banks were EUR 31.1 trillion at the end of 2009, totalling
almost 345% of euro area nominal GDP and about 3.75 times the size of the US
banking system. The credit intermediation process in the euro area is dominated by
banks which account for about three-quarters of the market (as opposed to one
quarter in the United States). Furthermore, over and above the high overall level of
bank dependence, there are notable differences across euro area countries and across
typologies of banks.

For example, the Bank Lending Survey (BLS),
3
which is designed to provide
information on supply and demand conditions in the euro area credit markets and the
lending policies of euro area banks, indicate that credit access is heterogeneous
across countries with the degree of dispersion increasing since the beginning of 2007,
before the financial crisis actually started. (see Figure 1). This is not inconsistent
with the notion that the bank lending channel is highly heterogeneous across coun
triesand possiblyacross banks typologies (see also Ciccarelli, Maddaloni and Peydro,
2013).

In this section, we present banks balance sheet descriptive statistics grouped by
country and typology of banks for the four largest economies of the euro area:
France, Germany, Italy and Spain. We use proprietary data from Bankscope, which
are detailed in Appendix A. The focus is on key banks characteristics such as size,
liquidity, capital and profitability to assess differences and similarities across
countries. Furthermore, we use data from a common source to make sure that the
differences in the data set are not responsible for differences in the empirical results
available in the literature. This appears an advantage relative to earlier studies, which
have typically focused on a single country (Angeloni, et al, 2003; Chatelain, et al,
2003; Ehrmann and Worms, 2004; Gambacorta, 2005 and 2008; Jimenez et al.,
2012), or have used synthetic aggregate data for the euro area (Altunbas et al., 2004
and 2009), Europe and the US as a whole (Gambacorta and Marques-Ibanez, 2011).

The relevant specialization categories in the Bankscope database are: (i) commercial
banks, (ii) savings banks and (iii) cooperative banks. Because of the low number of
observations for each country, we exclude real estate and mortgage banks as well as
medium- and long-term credit banks from the econometric analysis in section 5, but
we report descriptive statistics for them in this section. The selected categories (i) to
(iii) represent more than eighty percent of the euro area corporate and household
credit markets. Commercial banks are defined in Bankscope as mainly active in a
combination of retail banking individuals, Small and Medium Enterprises (SMEs),

3
The BLS addresses issues such as credit standards for approving loans as well as credit terms and
conditions applied to enterprises and households. It also asks for an assessment of the conditions
affecting credit demand. The survey is addressed to senior loan officers of a representative sample of
euro area banks and is conducted four times a year starting from the first quarter of 2003. The sample
group participating in the survey comprises around one hundred banks from all euro area countries
and takes into account the characteristics of their respective national banking structures.

5
wholesale banking (large corporates) and private banking (not belonging to groups of
saving banks, co-operative banks). Saving banks refer to banks mainly active in retail
banking (individuals, SMEs) and belonging to a group of savings banks which,
unlike commercial banks, are characterized by broadly decentralized distribution
network, providing local and regional outreach.

Net percentages for the euro area
-20
-10
0
10
20
30
40
50
60
70
80
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
12 EA countries due to a bank's capital position
due to a bank's market financing due to a bank's liquidty position

Standard deviation across the initial twelve euro area countries
0
5
10
15
20
25
30
35
40
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
12 EA countries due to a bank's capital position
due to a bank's market financing due to a bank's liquidty position

Figure 1.
Changes in credit standards on loans to corporations
(net percentages of banks reporting tightening standards and standard deviation across countries)
Source: European Central Bank.
Note: The net percentage refers to the difference between the sum of the percentages for tightened considerably
and tightened somewhat and the sum of the percentages for eased considerably and eased somewhat. See
http://www.ecb.int/stats/money/lend/html/index.en.html for further information. Sample period: 2003Q1 2012Q1.



6
Cooperative banks have a cooperative ownership structure and are mainly active in
retail banking (individuals, SMEs).
4
Real estate and mortgage banks are mainly
active in mortgage financing and project development. Medium and long-term credit
banks provide loans finance such as export finance, finance of projects in less
developed countries, environmental programs and loans to small and medium sized
firms.

To give a sense for the large extent of heterogeneity in the euro area data, Figure 2
and Tables 1 and 2 report descriptive statistics from banks balance sheets in France,
Germany, Italy and Spain, grouped by typology of banks. A few facts about banks
heterogeneity are worth noting:
- commercial banks have a large market share in all countries, which is particularly
sizeable in Italy; savings banks are very important in Spain and Germany;
cooperative banks have a sizable market share in France and Italy; medium and
long term credit bank play only a marginal role; and real estate and mortgage
banks are mainly present in Germany (see Figure 2).
- cooperative banks have the smallest size in Germany and Italy; savings banks are
particularly large in France and Spain (see Table 1).
- French savings banks are the most liquid; while German savings banks are the
least liquid. Overall, Italian banks and the commercial banks in all four countries
are relatively liquid (see Table 1).
- Italian (German) banks are the most (least) capitalised. Savings banks are the
least capitalised banks in all countries except Italy (see Table 1).
5

- French banks have the lowest loan loss provision, which is particularly low
among savings banks (see Table 2).
- German, Spanish and French banks finance their activity largely by means of
deposits (about 80%); Italian banks instead rely only for about 54% to deposits,
one third of banks activity is financed by other liability means, such as debt
securities. This financing structure is homogenous among banks within the same
country (see Table 2).
- French and Spanish banks are the most profitable, having the largest return on
average assets (ROAA = net profit / average assets) and on average equity
(ROAE = net profit / average shareholders equity). Spanish banks appear the
most efficient as measured by the lowest cost/income ratio (= operating expenses
/ operating income) (see Table 2).
- Spanish (German) banks recorded the fastest (slowest) growth rate of credit (see
Table 1).


4
Cooperative banks are owned by the depositors and often offer rates more favourable than for-profit
banks. Typically, membership is restricted to employees of a particular company, residents of a
defined neighbourhood, members of a certain labour union or religious organizations, and their
immediate families. This specialization category includes also Banche Popolari in Italy,
Volksbank in Germany, Caja rural in Spain or Banque populaire in France.
5
Commercial banks and cooperative banks are more capitalised than savings banks most likely
because they can find capital on the market, the former issuing equities, and the latter finding
members, while savings banks can increase capital only through retained earnings and through the
intervention of municipalities.
7
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Germany Italy Spain France
Real estate and mortage,
medium and long-term credit
Cooperative
Savings
Commercial
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Germany Italy Spain France
Real estate and mortage,
medium and long-term credit
Cooperative
Savings
Commercial
Loans

Assets

Figure 2.
Countries Market Share for Loans and Assets by Bank Type
Source: Bankscope and authors calculations. Sample period: 1999-2011.





8
All Commercial Savings Cooperative
Germany
Liquidity / Assets (%) 14.5 22.5 12.6 15.0
Capital / Assets (%) 5.5 6.5 5.0 5.8
Size (log of assets, EUR mil.) 6.3 6.9 7.1 5.8
Loan growth (%) 1.8 3.4 1.7 1.8
Number of observations 18669 1364 5848 10645
Italy
Liquidity / Assets (%) 20.0 20.4 19.1 20.3
Capital / Assets (%) 10.6 7.4 8.7 11.5
Size (log of assets, EUR mil.) 5.9 8.0 7.4 5.4
Loan growth (%) 9.8 10.0 8.7 10.2
Number of observations 6536 1055 592 4680
Spain
Liquidity / Assets (%) 14.1 19.3 13.5 11.8
Capital / Assets (%) 7.4 7.5 6.9 8.6
Size (log of assets, EUR mil.) 8.3 7.6 9.1 7.0
Loan growth (%) 12.4 12.4 14.4 3.4
Number of observations 1332 540 518 248
France
Liquidity / Assets (%) 19.3 24.2 46.5 12.4
Capital / Assets (%) 7.6 6.7 5.7 9.6
Size (log of assets, EUR mil.) 8.1 7.2 9.0 8.7
Loan growth (%) 7.0 6.6 7.4 8.2
Number of observations 2671 1300 260 829
Table 1.
Descriptive statistics by typologies of banks
(Median, sample period: 1999-2011)
Source: Bankscope and authors calculations.
9
Table 2.
Banks Balance Sheet and Financial ratios in the Largest euro area Countries
(Median, sample period: 1999-2011)

Germany
Al l Commerci al Savi ngs Cooperati ve
Assets (%)
Liquid assets 14.5 22.5 12.6 15.0
Loans 60.9 54.6 60.8 61.1
Fixed assets 1.4 0.4 1.3 1.5
Other assets 23.3 22.5 25.3 22.4
Total assets 100.0 100.0 100.0 100.0
Li abi l i ty and Equi ty (%)
Deposits 86.1 84.1 87.6 85.7
Other liabilities 8.4 9.4 7.4 8.5
Equity capital 5.5 6.5 5.0 5.8
Total liabilities and equity 100.0 100.0 100.0 100.0
Profi tabi l i ty and effi ci ency (%)
ROAA 0.2 0.3 0.2 0.3
ROAE 4.1 4.7 3.5 4.3
Net interest margin 2.6 2.1 2.5 2.7
Cost to income ratio 70.4 69.1 68.1 71.8
Loan loss provision to asset ratio (%) 0.4 0.3 0.5 0.4
Size (log of assets, EUR) 6.3 6.9 7.1 5.8

Italy
Al l Commerci al Savi ngs Cooperati ve
Assets (%)
Liquid assets 20.0 20.4 19.1 20.3
Loans 66.5 66.0 71.2 65.4
Fixed assets 1.4 1.2 1.5 1.5
Other assets 12.2 12.3 8.2 12.8
Total assets 100.0 100.0 100.0 100.0
Li abi l i ty and Equi ty (%)
Deposits 53.9 58.5 55.5 53.1
Other liabilities 35.5 34.1 35.8 35.4
Equity capital 10.6 7.4 8.7 11.5
Total liabilities and equity 100.0 100.0 100.0 100.0
Profi tabi l i ty and effi ci ency (%)
ROAA 0.7 0.6 0.7 0.7
ROAE 6.0 6.7 6.9 5.7
Net interest margin 3.1 2.7 3.3 3.2
Cost to income ratio 69.6 67.6 66.8 70.8
Loan loss provision to asset ratio (%) 0.3 0.4 0.4 0.3
Size (log of assets, EUR) 5.9 8.0 7.4 5.4



10
continued
Spain
Al l Commerci al Savi ngs Cooperati ve
Assets (%)
Liquid assets 14.1 19.3 13.5 11.8
Loans 70.9 66.9 70.7 74.7
Fixed assets 1.8 1.1 2.2 2.1
Other assets 13.3 12.8 13.7 11.4
Total assets 100.0 100.0 100.0 100.0
Li abi l i ty and Equi ty (%)
Deposits 77.7 80.2 73.5 81.3
Other liabilities 14.9 12.3 19.7 10.1
Equity capital 7.4 7.5 6.9 8.6
Total liabilities and equity 100.0 100.0 100.0 100.0
Profi tabi l i ty and effi ci ency (%)
ROAA 0.7 0.7 0.8 0.5
ROAE 9.4 8.6 10.7 5.9
Net interest margin 2.3 2.3 2.5 2.2
Cost to income ratio 58.7 56.5 58.7 61.8
Loan loss provision to asset ratio (%) 0.3 0.3 0.3 0.3
Size (log of assets, EUR) 8.3 7.6 9.1 7.0


France
Al l Commerci al Savi ngs Cooperati ve
Assets (%)
Liquid assets 19.3 24.2 46.5 12.4
Loans 66.9 59.3 39.3 75.1
Fixed assets 0.6 0.6 0.6 0.7
Other assets 13.2 15.8 13.6 11.8
Total assets 100.0 100.0 100.0 100.0
Li abi l i ty and Equi ty (%)
Deposits 73.0 71.9 90.1 77.6
Other liabilities 19.4 21.4 4.2 12.8
Equity capital 7.6 6.7 5.7 9.6
Total liabilities and equity 100.0 100.0 100.0 100.0
Profi tabi l i ty and effi ci ency (%)
ROAA 0.7 0.7 0.4 0.8
ROAE 8.4 10.3 7.9 7.9
Net interest margin 2.1 2.5 1.5 2.2
Cost to income ratio 65.8 66.9 70.6 63.3
Loan loss provision to asset ratio (%) 0.2 0.2 0.1 0.3
Size (log of assets, EUR) 8.1 7.2 9.0 8.7

Source: Bankscope and authors calculations.
Note: Net interest margin = net interest income / total earnings assets. Return on average assets (ROAA) = net profits /
average assets. Return on Average Equity (ROAE) = net profits / average shareholders' equity. The cost/income ratio =
operating expenses / operating income.

This set of descriptive statistics exemplifies the significant extent of heterogeneity in
euro area data across countries and typology of banks. At the same time, they call for
a deeper understanding of the relevant source(s) of heterogeneity and the
11

consequences for the conduct of monetary policy. The rest of the paper aims at
addressing these issues.

3. SOURCES OF HETEROGENEITY IN BANK LENDING
The starting point of the bank lending channel is the recognition of the imperfect
functioning of capital markets and the existence of incomplete contracts in a
violation of the Modigliani-Miller (1958) theorem. The imperfections are brought
about by the pervasiveness of asymmetric information and the associated agency
problems that can result in large gaps between the lenders' expected returns and the
borrowers' costs of funds.
6
In this context, banks are not able to substitute freely
across different sources of finance as well as firms may not easily replace bank loans
with other form of financing (e.g. market debt and trade credit). However, there are
many factors that can influence credit supply, the most important being: the specific
characteristics of banks, a particular relationship with customers, the networks
among banks, and the degree of market competition. Moreover, bank lending is
influenced by borrowers characteristics as well as the structure and the development
of the nonfinancial sector.

Banks characteristics
Given that credit provision requires the evaluation of both projects and borrowers as
well as the monitoring of the borrowers performance before and during the loan, the
theoretical literature attributes a prominent role to net worth (banks capital) in
reducing the agency costs of borrowing (Bernanke and Gertler, 1995; Holmstrom
and Tirole, 1997 and 1998; Bernanke, Gertler and Gilchrist, 1999; Gertler and
Kiyotaki, 2010), while the empirical literature suggests using size (banks total
assets) as a proxy for informational asymmetries. It is generally argued that
following a monetary tightening, the banks less likely to contract loan supply are: (i)
the larger banks as they can raise external funds more easily (Kashyap and Stein,
1995); (ii) the more capitalized banks as they have more equity securities to absorb
future losses (Kishan and Opiela, 2000; Van den Heuvel, 2002) and less moral
hazard problems at the bank level given that relatively more money is at stake
(Bernanke, 2007); but also (iii) the more liquid banks as they can use their liquidity
to satisfy the demand for loans (Kashyap and Stein, 2000, and Chatelain et al., 2003).

Relationship lending
In several euro area countries, the market for intermediated finance is characterised
by relationship rather than arms length lending. It is very common that bank
customers establish long lasting relationships with banks, with a prominent example
being the German system of house banks, in which firms conduct most of their

6
The role of credit in the economy becomes important once the assumption of perfect information is
relaxed. Indeed, the essence of the credit creation process is the gathering and transmission of
information, which are needed to evaluate projects and borrowers and to monitor borrowers'
performance after the loan. In particular, banks have the expertise to channel savings to small business
projects that are information-intensive and particularly hard to evaluate.
12

financial business with one bank only. With most German banks operating as
universal banks, and therefore supplying their customers with the full range of
financial services, this implies a much closer linkage to a single bank than in many
other countries. For the creditor, this could also imply an implicit guarantee to have
access to (additional) funds even if the central bank follows a restrictive monetary
policy. In such a case, the reaction of bank loan supply to monetary policy should be
at least muted. Italian banks seem to be characterized by a similar business model,
according to which many small banks entertain close relationships with their
customers (especially small firms). This is true also for France as most small firms
have business relationships with one bank only. However, French small firms do not
account for a large share of GDP. Typically, house bank relationships exist between
relatively small banks for which the loan business with non-banks is still a central
activity and their customers. However, also the typology of banks is very
important, given the personal contact typical of credit cooperatives.

Bank networks
Banks have set up networks of various kinds. This is particularly true for two sectors:
savings banks and credit cooperatives in Germany. Both sectors consist of an upper
tier of large banks serving as head institutions and a lower tier of smaller banks
that entertain very close relationships to the head institutions, leading to an internal
liquidity management. On average, the lower tier banks deposit short-term funds
with the upper tier banks, and receive long-term loans in turn. Therefore, these
types of banks might be less affected by a monetary policy shock even if they have a
relatively lower liquidity ratio.

Banks concentration
Since the inception of the single currency, the banking industry in the euro area has
continued to experience a gentle trend in market concentration, mostly driven by the
steadily increase in the number of mergers and acquisitions. While the evolution of
market concentration over time appears rather homogeneous across the four
economies,
7
we note that the extent of concentration for the entire sample period
stands at different levels in the various countries and, using various measures of
concentration indices, it is higher in Germany and Italy. These two countries are also
characterised by a banking system with many more banks per unit of GDP or
working population (see Figure 3).

Therefore, the transmission mechanism can differentiate across countries due to the
different degree of market competition, because the sensitivity of the lending rates to
monetary policy rates may also depend upon the intensity of market competition
(Klein, 1972; Monti, 1972). As the market become more competitive, the degree of
pass through rises. This hypothesis has been tested favourably. Hannan and Berger
(1991), for example, show that lending rates are sticky and that this thickness
increases with market concentration in accordance with the prediction of the Klein-
Monti model. The survey by Berger et al. (2004) on the impact of bank concentration

7
More specifically, the ranking of the four countries in terms of market concentration has not changed
over the sample period.
13

and competition provides additional reviews of the existing literature. Clearly, the
assumption of perfect competition may not seem appropriate for the banking sector,
given the large barriers to entry.

0.0
0.5
1.0
1.5
2.0
2.5
Germany Italy Spain France
N.BanksperEURbillionGDP


0
5
10
15
20
25
30
35
Germany Italy Spain France
N.BankspermillionWorker

Figure 3.
Number of banks per GDP and per working population
Source: Bankscope, European Central Bank.and own calculations.
Note: based on the average 1999-2011 period.



Balance sheet channel
The transmission of monetary policy through credit markets is also affected by
borrowers' net worth, cash flow and liquid assets, namely the balance sheet channel.
It is widely recognised that firms' balance sheets deteriorate with a monetary
contraction, through a reduction of both net worth (and thus collateral) and cash
flows. As pointed out by Jimnez, et al (2011), tighter monetary conditions may
14

reduce supply through increased agency costs of banks; but it also may influence
demand because of reductions in net worth and expected investment. They suggest
controlling for borrowers characteristics with firm-month fixed effects, which
capture time-varying observed and unobserved firm heterogeneity. They use the
credit register of Spain, which is collected by the Banco de Espaa acting in its
capacity as bank supervisor, to attain identification. Unfortunately, we do not have
information on the borrower side, which prevents us from investigating this
additional channel.

Nonfinancial sectors
Heterogeneity in bank lending across countries and typology of banks may also
reflect differences in the real economy. If firms in different sectors are run using
different business models, have diverse managerial practice or face different risks,
for instance, then heterogeneity in lending activities may be rooted outside the
banking industry. And if the euro area is not a fully optimal currency area, then
perhaps such an asymmetry would not necessarily be undesirable as heterogeneity in
bank lending may compensate for heterogeneity in the structure of the economy.
While controlling for this effect would require access to borrowers characteristics
(something that unfortunately we do not observed in our data), the empirical
specification below allows for GDP growth and its interaction with bank
characteristics to influence loan growth heterogeneously. This suggests that in our
model any evidence of heterogeneity in the lending response to monetary conditions
would be occurring over and above the heterogeneous impact captured by GDP
growth and the associated interaction terms, which are likely to capture some of the
effects stemming from the nonfinancial sectors.
4. MODELLING THE EFFECTS OF MONETARY POLICY ON LENDING
This section presents an empirical specification to study the impact of monetary
policy on bank lending. Our estimation strategy follows the contributions by
Kashyap and Stein (1995 and 2000), Kishan and Opiela (2000), Ehrmann, et al.
(2003), Altunbas, et al. (2009) and Gambacorta and Marques (2011), among many
others. These studies share the emphasis that some form of heterogeneity matters for
the transmission of monetary policy and this is explored by introducing an
interaction term between the policy instrument and the candidate source of
heterogeneity. We carry out the same exercise.

In addition, one should recognise that banks are inherently different not only in
observed characteristics such as size, capital and liquidity, etc., but also in
unobserved dimensions such as business model, risk propensity and managerial
ability. While the main focus of this paper is on cross-country heterogeneity,
monetary policy may also have diverse effects within typology of banks for each
country. To investigate this possibility, we will complement the results by splitting
the sample around some exogenous variables and threshold values using an
estimation strategy that allows heterogeneity in the transmission of monetary policy
across groups of banks within the same country and bank typology.
15

4.1 The empirical specification

The transmission of monetary policy through the bank lending channel requires the
identification of the monetary policy shock as well as controlling for loan demand
determinants. A vast empirical literature has proposed alternative identification
strategies to decompose changes in the short-term interest rate into the systematic
and the non-systematic component of monetary policy. Among those, one of the
most popular set of restrictions assumes that the short-term interest rate responds
contemporaneously to inflation, real activity and possibly a measure of credit
conditions. Accordingly, we label monetary policy shocks the residuals of a Taylor-
type rule in which changes in the 3-month overnight index swap are orthogonalised
vis--vis euro area real GDP growth, inflation and the growth of loans to non-
financial corporations and households.
8
Previous studies have typically used the
realised short-term interest rate as monetary indicator (see for instance Kashyap and
Stein, 1995 and 2000, and the literature they have pioneered).

To control for loan demand determinants, we use nominal GDP growth at time t and
time t-1, the past value of loan growth and the interaction of current and past nominal
GDP growth with individual bank characteristics at time t-1, such as banks'
profitability measures (such as the return on equity), loan loss provisions, size,
capital and liquidity. The aggregate variables and their interaction with bank
characteristics are used to isolate loan demand effects. In addition, the bank-specific
variables aim at capturing individual characteristics associated with the cyclical
effects not captured by GDP growth (through returns on equity) and borrowers' risk
(through loan loss provisions), as banks exposed to financially stronger borrowers
are expected to make relatively smaller non-performing loans.
9


Finally, we allow banks to respond differently to the monetary policy shock in a way
that may also depend on individual characteristics. By doing so, we hope to
disentangle the policy effect that is common across all banks from the policy effect
that may vary according to the size, liquidity or capital position of each bank.

4.2 The Ordinary Least Squares (OLS) specification

A separate model is estimated for each country and each bank typology. The
econometric specification takes the following form:



(1)

8
The finding of heterogeneity is robust to compute the monetary policy shock using the first lag of
inflation, output and credit aggregate as instruments for their contemporaneous values. The correlation
of the policy shocks obtained with OLS and IV is 0.85.
9
Loan loss provisions are non-cash expenses for banks to account for future losses on loan defaults.
This guarantees a bank's solvency and capitalisation if and when the defaults occur. Given that the
loan loss provisions increase with the riskiness of the loans, it is often used as a proxy for bank risk
(see also Altunbas, et al., 2009).
t i
j
j t t i j
j
j t t i j t i
j
j t t i j t i t i
j
j t j
j
j t j t i t i
u NGDP z a MP z a z a
NGDP w a LLP a ROE a
NGDP a MP a L a a L
,
1
0
1 , 9
1
0
1 , 8 1 , 7
1
0
1 , 6 1 , 5 1 , 4
1
0
3
1
0
2 1 , 1 0 ,
+ A + + +
A + + +
A + + A + = A



=

=

=

=

=

16
where L
i,t
is the first difference of the logarithm of loans of bank i in period t to
private non-banks in deviation from the average loan growth for that bank. Following
Altunbas, Gambacorta and David-Marques (2009), we exclude interbank positions.
10

The variable MP
t
refers to the monetary policy shock, NGDP
t
stands for the log
difference of nominal GDP (in deviation from the average GDP growth), ROE
i,t
is
the return on equity of bank i, LLP
i,t
describes loan loss provisions to asset ratio of
bank i, w
i,t
= [ROE
i,t
, LLP
i,t
] and z
i,t
represents a vector of bank characteristics such
as size, liquidity and capital. w
i,t
and z
i,t
are standardised value for each bank. The
term
i,t
represents the source(s) of unobserved heterogeneity across banks (See
Appendix A for detailed information on the construction of the variables).

Negative coefficients on the monetary policy shock (i.e.
2
< 0) but insignificant
coefficients on its interaction with bank-specific variables would indicate that the
transmission of monetary policy on lending activities is homogeneous across banks.
Positive coefficients on the interaction terms between the MP variable and the bank
characteristics (i.e.
8
> 0) would indicate that the effects of monetary policy vary
with the size, liquidity and capital of individual banks.

4.3 The Quantile Regression (QR) specification

The OLS specification allows studying the extent of heterogeneity in the effects of
monetary policy on lending both across countries and across typologies of banks
within each country. In this section, we move one step further and ask whether there
exists heterogeneity within typology of banks for each country. As it seems arbitrary
to take a stand a priori on the relevant source of heterogeneity across groups of banks
within the same category (intra-group heterogeneity), our empirical strategy will be
based on quantile regressions (QR), which allow us to deal with unobserved
heterogeneity. To provide intuition for the way quantile regressions work it is useful
to draw an analogy between Ordinary Least Squares (OLS) and Least Absolute
Deviations (LAD). As OLS provides an estimate of the average effect, LAD provides
an estimate of the median effect. Quantile regressions generalize the idea behind
LAD in a way that allows the econometrician to characterize the entire distribution of
lending responses to unanticipated movements in the policy rate across financial
institutions. To the extent that policy institutions, including the central bank, is
concerned about risk management in the credit market, looking at the average effect
is likely to be inappropriate for the purpose of identifying the characteristics that are
likely to make a bank more sensitive to changes in its cost of funding and business
cycle conditions.

In the heterogeneous response model, bank lending is treated as a potential latent
outcome. The potential outcome L
i,t
at date t is latent because, given the monetary
policy shock, MP, and other observable individual covariates, x and z, and macro

10
See Giannone, Lenza, Pill and Reichlin (2012) for a study of the effects of the ECB monetary
policy, especially the nonstandard tools, on the Euro area interbank market.
17

covariates, NGDP, the observed outcome for each bank i is only one of the possible
realizations in the admissible space of outcomes. The quantiles, Q
r
, of the potential
outcome distributions conditional on covariates are denoted by:
Q
r
(L
i,t
|MP, NGDP, w
i
, z
i
) with (0,1) (2)

and the effect of the treatment, here the unanticipated interest rate change, on
different points of the marginal distribution of the potential outcome is defined as:

(3)

The quantile treatment model can then be written as:

L
i,t
= q(MP
t
, MP
t-1
, NGDP
t
, NGDP
t-1
, w
i,t-1
, z
i,t-1
, u
i,t
)
with u
i,t
|MP, NGDP, W
i
, Z
i
~U(0,1) (4)

where q(.) = Q
r
(MP|NGDP, w
i
, z
i,
u
i
) and u
i,t
captures unobserved heterogeneity
across banks i with the same observed characteristics w
i
, z
i
and "treatment" MP. The
variable u
i,t
is usually referred to as the rank variable as it determines the relative
ranking of unit of observations in terms of potential outcomes. To the extent that the
unanticipated component of monetary policy is independent from bank-specific
characteristics, QTE

measures the causal effect of monetary policy on loan growth,

holding the unobserved characteristics driving heterogeneity fixed at u
i,t
= .

To estimate quantile effects, we can use methods outlined by Koenker and Bassett
(1978), which are based on the following conditional moment restrictions:

Prob[L q(MP,NGDP,x,)|MP,NGDP,w,z]
= Prob[u |MP,NGDP,w,z] =

for each (0,1). The estimated parameters, os, are the results of the following
optimization problem:


where
t
(u) = u(t I(u<0)) and the indicator function I() takes value of one for
negative values of u and zero otherwise. The penalty function above is asymmetric
and piecewise linear. The asymmetry is introduced by the tilting term (t I(u<0))
which weights differently the absolute residuals associated with the different parts of
the conditional distribution of the endogenous variable. By varying the weights in the
tilting term, quantile regressions yield a set of estimates for the slope coefficient over
the conditional distribution of the latent variable, which in the present context is loan
growth.

The empirical specification of the conditional -th quantile distribution of loan
growth then takes the following form:


( )
MP
z w NGDP MP L Q
QTE
i i i
c
A A c
=
, , ,
t
t
( ) o
t
o
'
1
min
i i
n
i
x y

=
18

(5)


for each (0,1), where the variables have been defined in the previous section.

Specification allows unlike earlier studies- the impact of monetary policy shocks,

2
, and its interaction with bank characteristics,
8
, to vary across endogenously
determined groups of banks within each country. Furthermore, we also allow the
growth rate of GDP and its interaction with all individual characteristics to have
heterogeneous effects across the conditional distribution of loan growth. Given that
the financial sector typically represents a small fraction of GDP in each of the four
countries we consider, allowing for heterogeneity in
3
,
6
and
9
suggests that any
evidence of heterogeneity in
2
and
8
is most likely to capture heterogeneity in the
bank lending response to monetary conditions, over and above any possible
heterogeneous influence coming from the nonfinancial sectors.
5. EMPIRICAL EVIDENCE
In this section, we present the main results of the paper, namely the effects of non-
systematic changes in the short-term interest rate on lending activity. We distinguish
between the direct effect of monetary policy on loans, as exemplified by the
estimated coefficients o
2
, and the indirect effect coming from the interaction between
the short-term interest rate and bank-specific variables (i.e. the bank lending channel)
such as size, liquidity and capital, as exemplified by the coefficients o
8
. In each table
and panel, we report the sum of the coefficients on the variable of interest at time t
and t-1. Given our focus on the effects of monetary policy in the euro area, we
investigate the impact of the monetary policy shock using data for the period 1999-
2011. We estimate a separate specification for each country and then within each
country for each typology of banks. We consider neither real estate and mortgage
banks nor medium and long term credit banks separately because of the insufficient
number of observations for each country.

The first column of Figure 4 reports the direct effect of the monetary policy shock on
loan growth in Germany (first row), Italy (second row), Spain (third row) and France
(fourth row). Non-systematic changes in the cost of short-term financing have a
negative impact in all countries with the largest extent of heterogeneity in Germany
and Italy. The impact is milder and statistically insignificant in Spain.

The interaction between monetary policy shocks and bank's size (liquidity) in the
second (third) column reveals that this form of the bank lending channel is active in
Italy, Spain and France (Germany and Italy). Finally, there is little evidence in the
last column of Figure 4 that banks' capital may significantly dampen the effect of
monetary policy on lending activities.


( ) ( ) ( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )



=

=

=

=

=

A + + +
A + + +
A + + A + = A
1
0
1 , 9
1
0
1 , 8 1 , 7
1
0
1 , 6 1 , 5 1 , 4
1
0
3
1
0
2 1 , 1 0 ,
j
j t t i j
j
j t t i j t i
j
j t t i j t i t i
j
j t j
j
j t j t i t i
NGDP z MP z z
NGDP w LLP ROE
NGDP MP L L Q
t o t o t o
t o t o t o
t o t o t o t o
t
19




Figure 4.
Impact of monetary policy shock and its interaction with bank characteristics on loan growth
The econometric specification takes the form of equations (1) and (5). Prior to estimation the interest change is
orthogonalized by taking the residuals on a regression of interest rate change on euro area real GDP growth,
euro area GDP deflator inflation, euro area loan growth and lagged interest rate change. The panels report the
sum of the coefficient of the variables at time t and t-1. Source: Bankscope, annual data. QR (LS) estimates in
black (blue) refer to quantile (least squares) regressions. Shaded areas (dotted lines) are 95% confidence bands
estimated using robust standard errors. Estimates are reported for [.05, .95] at .05 unit intervals. Sample: see
Table 1.


Figure 5 decomposes the estimates of the effect of the monetary policy shock by
banks' categories with the effects of monetary policy on lending for commercial,
cooperative and saving banks reported in the first, second and third column,
respectively. We find that the impact of non-systematic changes in the short-term
interest rate is significant and larger for cooperative banks in Italy and France and for
saving banks in Italy. A fraction of commercial banks in Germany, Italy and France
also appear negatively affected by monetary policy shocks, although the coefficients
seem imprecisely estimated.




20


Figure 5.
Impact of monetary policy shock on loan growth across typologies of banks
The econometric specification takes the form of equations (1) and (5). Prior to estimation the interest change is
orthogonalized by taking the residuals on a regression of interest rate change on euro area real GDP growth,
euro area GDP deflator inflation, euro area loan growth and lagged interest rate change. The panels report the
sum of the coefficient of the variables at time t and t-1. Source: Bankscope, annual data. QR (LS) estimates in
black (blue) refer to quantile (least squares) regressions. Shaded areas (dotted lines) are 95% confidence bands
estimated using robust standard errors. Estimates are reported for [.05, .95] at .05 unit intervals. Sample: see
Table 1.


As the interaction between monetary policy and size does not seem to have
significant effects on lending activities (with the possible exception of German
cooperative banks and Italian saving banks), we move to the impact of the interaction
between the policy shock and size, liquidity and capital in Figures 6, 7 and 8
respectively, focusing on the countries and typologies of banks for which there is
evidence of significant effects or heterogeneous behaviour.

Figure 6 reveals that the bank lending channel working through size is significant
and heterogeneous for German cooperative banks and Italian saving banks.

Figure 7 reveals that the bank lending channel working through liquidity is
significant and heterogeneous for German cooperative and saving banks, Italian
cooperative banks.
21



Figure 6.
Impact of monetary policy interacted with size on loan growth across typologies of banks
The econometric specification takes the form of equations (1) and (5). Prior to estimation the interest change is
orthogonalized by taking the residuals on a regression of interest rate change on euro area real GDP growth,
euro area GDP deflator inflation, euro area loan growth and lagged interest rate change. The channel is
calculated using the interaction between the monetary policy shocks at time t and t-1 and liquidity at time t-1. The
panels report the sum of the coefficient of the variables at time t and t-1. Source: Bankscope, annual data. QR (LS)
estimates in black (blue) refer to quantile (least squares) regressions. Shaded areas (dotted lines) are 95%
confidence bands estimated using robust standard errors. Estimates are reported for [.05, .95] at .05 unit
intervals. Sample: see Table 1.


Figure 7.
Impact of monetary policy interacted with liquidity on loan growth across typologies of banks
The econometric specification takes the form of equations (1) and (5). Prior to estimation the interest change is
orthogonalized by taking the residuals on a regression of interest rate change on euro area real GDP growth,
euro area GDP deflator inflation, euro area loan growth and lagged interest rate change. The channel is
calculated using the interaction between the monetary policy shocks at time t and t-1 and liquidity at time t-1. The
panels report the sum of the coefficient of the variables at time t and t-1. Source: Bankscope, annual data. QR (LS)
estimates in black (blue) refer to quantile (least squares) regressions. Shaded areas (dotted lines) are 95%
confidence bands estimated using robust standard errors. Estimates are reported for [.05, .95] at .05 unit
intervals. Sample: see Table 1.
22
In line with the aggregate results in Figure 4, we detect little evidence of a significant
bank lending channel operating in Spain and France. Finally, as for the interaction
between monetary policy and capital, Figure 8 reports significant effects only for
German cooperative and saving banks.















Figure 8.
Impact of monetary policy interacted with capital on loan growth across typologies of banks
The econometric specification takes the form of equations (1) and (5). Prior to estimation the interest change is
orthogonalized by taking the residuals on a regression of interest rate change on euro area real GDP growth,
euro area GDP deflator inflation, euro area loan growth and lagged interest rate change. The channel is
calculated using the interaction between the monetary policy shocks at time t and t-1 and liquidity at time t-1. The
panels report the sum of the coefficient of the variables at time t and t-1. Source: Bankscope, annual data. QR (LS)
estimates in black (blue) refer to quantile (least squares) regressions. Shaded areas (dotted lines) are 95%
confidence bands estimated using robust standard errors. Estimates are reported for [.05, .95] at .05 unit
intervals. Sample: see Table 1.

Cooperative banks, which are typically present in the financial, crafts and
agricultural sectors, differ from stockholder banks by their organization, their goals,
their values and their governance. Cooperative banks are often created by persons
belonging to the same local or professional community or sharing a common interest
with the ownership, which is widely shared. The first aim of cooperative banks is not
to maximise profit but to provide the best possible products and services to its
members. They provide access to financial services to individuals who would
otherwise be excluded from such offers. Also saving banks, unlike commercial
banks, are characterized by broadly decentralized distribution network, providing
local and regional outreach. Given that the core business of the credit cooperatives
and savings banks is local, the heterogeneous impact of the monetary policy shocks
on local income and deposits might explain the findings.

In summary, the direct effect of policy rate changes on lending activities is highly
heterogeneous in Germany and Italy, but it is far more muted in Spain and France.
As for the bank lending channel, we find evidence that it is mostly operating through
liquidity in Germany and Italy, with the significant and heterogeneous effects mostly
driven by cooperative banks. There is some evidence that the effects of monetary
23

policy on bank lending may change with the size of banks in Italy, Spain and France
but these findings do not seem to correlate with the classification in commercial,
cooperative and saving banks, thereby suggesting that other forms of (unobserved)
heterogeneity may be at play. Next section discusses some possible interpretations
for these findings.
6. COMPARISON WITH EARLIER MICRO STUDIES FOR INDIVIDUAL
EURO AREA COUNTRIES
At this point, it is useful to compare our estimates with those obtained by selected
studies using balance sheet data on individual banks for single euro area countries.
Most of these earlier studies on single countries have typically focussed on pre-1999
samples. The comparison is meant to highlight any possible time variation in the
interest rate and bank lending channel in the euro area.

We already mentioned that the impact of interest rate changes were not so
heterogeneous across countries as an increase in interest rates tended to reduce loan
growth in France, Germany, Italy and Spain, particularly of banks with less liquid
asset holdings (see e.g. Ehrmann, et al., 2003). Conversely, in reviewing the analysis
using post-1999 observations, we find clear evidence that the impact of monetary
policy shocks is heterogeneous across countries and across typologies of banks.
Moreover, we find that bank lending supply amplifies the effect of monetary policy
changes of savings and cooperative banks in Germany and savings banks in Italy,
which mainly serve individuals and small firms.

This summary study by Ehrmann, et al. (2003) was then followed by country specific
studies that are here highlighted.

Using quarterly data for Germany over a pre-1999 sample, Worms (2003) argues that
a bank's share of short-term interbank deposits relative to total assets (as opposed to
its size) is paramount to explain the bank lending reaction to monetary policy. This is
because of the existence of long-term lending relationships between the majority of
German banks and their loan customers ("house-bank relationships"). Our results
corroborate this view that the bank lending channel in Germany works through
liquidity and not the size of the bank.

The estimates we obtain for Italy resemble those reported by Gambacorta (2003)
using quarterly data over a pre-1999 sample with regard to the role of liquidity. In
addition, we document a significant interaction between size and changes in the
short-term interest rate since 1999.

Based on annual data over the period 1991-1998, Hernando and Martiniz-Pages
(2003) find evidence for a bank lending channel working through liquidity in Spain.
On the other hand, using monthly data on loan applications, Jimnez et al. (2011)
conclude that the effects of monetary policy are stronger for banks of smaller size.
24

Our results suggest that on average over the 1999-2011 period policy interest rates
have not affected credit growth in Spain and the bank lending channel was impaired.

As for France, Loupias, et al. (2003) employ annual data over the period 1993-2000
and find that, following a change in the interest rate, the reduction of loan growth in
French banks with lesser liquid assets (relative to total assets) tends to be larger than
the reduction in more liquid banks. We instead cannot find a bank lending channel
for France.
7. POLICY CONCLUSIONS
The transmission of monetary policy in the euro area has been the focus of a
comprehensive set of studies on pre-1999 samples undertaken by the Eurosystem
Monetary Transmission Network, jointly by the European Central Bank (ECB) and
the euro area National Central Banks (Angeloni, et al., 2003). This coordinated
research effort has documented that pre-1999 an increase in interest rates tended to
reduce loan growth in France, Germany, Italy and Spain. Moreover, they found that
banks in these four countries with more liquid asset holdings showed weaker loan
adjustments.

This paper studies a similar question using data post-1999, when the ECB was in
charge of monetary policy in the euro area. Our findings suggest that heterogeneity
in the transmission of monetary policy to banks lending activities appears associated
with heterogeneity across countries and across typologies of banks in the same
country, while it is broadly homogenous within the bank typology group in each
country.

Our findings also suggest that changes in the cost of funding engineered by monetary
policy actions exert their maximum impact on cooperative and saving banks in
Germany, especially those with lesser liquidity and lower capital, and saving banks
in Italy, especially those with smaller size. Large commercial banks, on the other
hand, appear more capable to isolate their lending activities from changes in
monetary policy conditions. Small banks are best placed to refinance the real
economy, in particular small- and medium-sized firms, which are the biggest
generator of employment in the economy. The analysis suggests that the increase in
the number of cooperative and savings banks that have access to the ECB standard
and non-standard measures during the recent financial crisis is likely to improve the
transmission mechanism of monetary policy in the euro area.

Another important policy issue is the evidence that monetary policy shocks had
limited implications for the bank loans in Spain over the 1999-2011 period. The
limited impact of interest rates on bank lending in Spain suggests that in a monetary
union country-specific excessive growth of credit should be counteracted with
instruments that limit the fall in lending standards during boom times.

25

REFERENCES

Angeloni, I., A. Kashyap and B. Mojon (2003), Monetary Policy Transmission in
the Euro Area, Cambridge: Cambridge University Press.
Altunbas, Y., G. de Bondt and D. Marques-Ibanez (2004), Bank Capital, Bank
Lending and Monetary Policy in the euro area, Kredit und Kapital, Vol.4, pp.
443-465.
Altunbas, Y., L. Gambacorta and D. Marques-Ibanez (2009), Securitisation and
the Bank Lending Channel, European Economic Review, 53, pp. 996-1009.
Berger, A.N., A. Demirguc-Kunt, R. Levine, and J.G. Haubrich (2004), Bank
Concentration and Competition: An Evolution in the Making, Journal of Money,
Credit and Banking, Vol. 36, No. 3, pp. 433-451.
Bernanke, B.S. and M. Gertler (1995), Inside the Black Box: The Credit Channel
of Monetary Policy Transmission, Journal of Economic Perspective, Vol. 9, No.
4, pp. 27-48.
Bernanke, B.S., Gertler, M. and Gilchrist, S. (1999), The Financial Accelerator in
a Quantitative Business Cycle Framework, in Handbook of Macroeconomics,
ed. J. Taylor and M. Woodford, Amsterdam: Elsevier, pp. 1341-93.
Bernanke, B.S. (2007), The Financial Accelerator and the Credit Channel,
speech on 15 June at the The Credit Channel of Monetary Policy in the Twenty-
first Century Conference, Federal Reserve Bank of Atlanta.
Chatelain, J.B., Generale, A., Vermeulen, P., Ehrmann, M. Martinez-Pages, J. and
A. Worms (2003), Monetary Policy Transmission in the euro area: New
Evidence from Micro data on Firms and Banks, Journal of the European
Economic Association, Vol. 1, pp. 731-742.
Ciccarelli, M., A. Maddaloni and J.L. Peydro (2013), Heterogeneous
Transmission Mechanism: Monetary Policy and Financial Fragility in the Euro
Area, ECB Working Papers, n. 1527.
De Grauwe, P and Mongelli, F.P. (2005). Endogeneities of Optimum Currency
Areas: What Brings Countries Sharing a Single Currency. ECB Working Paper
Series, No. 1091.
Ehrmann, M., L. Gambacorta, J. Martinez-Pages, P. Sevestre and A. Worms,
(2003), Financial Systems and the Role of Banks in Monetary Policy
Transmission in the Euro Area, eds. I. Angeloni, A. Kashyap and B. Mojon
(eds.), Monetary Policy Transmission in the Euro Area, Cambridge: Cambridge
University Press, pp. 235-269.
Ehrmann, M. and A. Worms (2004), Bank Networks and Monetary Policy
Transmission, Journal of the European Economic Association, Vol. 2(6), pp.
1148-1171.
Gambacorta, L. (2003), The Italian Banking System and Monetary Policy
Transmission: Evidence from Bank-Level Data, eds. I. Angeloni, A. Kashyap
26

and B. Mojon, Monetary Policy Transmission in the Euro Area, Cambridge:
Cambridge University Press, pp. 323-346.
Gambacorta L. (2005), Inside the Bank Lending Channel, European Economic
Review, Vol. 49, pp. 1737-1759.
Gambacorta L. (2008), How Do Banks Set Interest Rates?, European Economic
Review, Vol. 52, pp. 792-819.
Gambacorta, L. and D. Marques-Ibanez, 2011, The Bank Lending Channel:
Lessons from the Crisis, Economic Policy, pp. 137-168.
Gertler, M. and Kiyotaki, N. (2010), Financial Intermediation and Credit Policy in
Business Cycle Analysis, in Handbook of Monetary Economics, eds. B.M.
Friedman and M. Woodford, Amsterdam: Elsevier, pp. 547-99.
Giannone, D., M. Lenza, H. Pill and L. Reichlin (2012), The ECB and the
Interbank Market, CEPR Discussion Paper, No. 8844.
Hannan, T. H., and A. Berger (1991), "The Rigidity of Prices: Evidence for the
Banking Industry, American Economic Review, Vol. 81, No. 4, pp. 938-945.
Hernando, I. and J. Martinez-Pages (2003), Is There a Bank-Lending Channel of
Monetary Policy in Spain, eds. I. Angeloni, A. Kashyap and B. Mojon,
Monetary Policy Transmission in the Euro Area, Cambridge: Cambridge
University Press, pp. 284-296.
Holmstrom, B. and Tirole, J. (1997), Financial Intermediation, Loanable Funds,
and the Real Sector, Quarterly Journal of Economics, Vol. 112, pp. 663-91.
Holmstrom, B. and Tirole, J. (1998), Private and Public Supply of Liquidity,
Journal of Political Economy, Vol. 106, No. 1, pp. 1-40.
Jimenez, G., S. Ongena, J.L. Peydr Alcalde, and J. Saurina, (2012), Identifying
Loan Supply and Balance-sheet Channels with Loan Applications, American
Economic Review, Vol. 102, No. 5, 2301-2326.
Kashyap, A., and J. Stein (1995), The Impact of Monetary Policy on Bank
Balance Sheets, Carnegie-Rochester Conference Series on Public Policy, Vol.
42, pp. 151-95.
Kashyap, A., and J. Stein (2000), What Do a Million Observations on Banks Say
About the Transmission of Monetary Policy?, American Economic Review, Vol.
90, pp. 407-28.
Kishan, R.P. and T. P. Opiela (2000), Bank Size, Bank Capital, and the Bank
Lending Channel, Journal of Money, Credit, and Banking, Vol. 32 (1), pp. 121-
141.
Klein, M. A. (1971), A Theory of the Banking Firm, Journal of Money, Credit
and Banking, Vol. 3, pp. 205218.
Koenker, Roger and Gilbert W. Bassett (1978), Regression Quantiles,
Econometrica 46, pp. 33-50.
27

Loupias, C., Savignac, F. and P. Sevestre (2003), Is There a Bank-Lending
Channel in France? Evidence from Bank Panel Data, eds. I. Angeloni, A.
Kashyap and B. Mojon, Monetary Policy Transmission in the Euro Area,
Cambridge: Cambridge University Press, pp. 297-308.
Modigliani, F. and M. Miller (1958), The Cost of Capital, Corporation Finance
and the Theory of Investment, American Economic Review, Vol. 48, pp. 261-
297.
Monti, M. (1972), Deposit, Credit and Interest Rate Determination under
Alternative Bank Objective functions, eds. K. Shell and G.P. Szeg,
Mathematical methods in investment and finance, Amsterdam: North-Holland,
pp. 431454.
Van den Heuvel, S. (2002), Does Bank Capital Matter for Monetary
Transmission?, Economic Policy Review, Federal Reserve Bank of New York,
pp. 1-7.
Worms, A. (2003), The Reaction of Bank Lending to Monetary Policy Measures
in Germany, eds. I. Angeloni, Kashyap, A. and B. Mojon, Monetary Policy
Transmission in the Euro Area, Cambridge University Press, pp 270-283.
28

APPENDIX A: THE DATA

The data we use are commercially distributed through Bankscope, a comprehensive
database provided by the rating agency Fitch Ibca containing detailed annual
information on about 8,000 European banks, 14,000 North American banks and more
than 6,000 banks from other areas around the world. The data cover the universe of
banks in the four euro area economies we focus on. The global disclosure format
provides consistent financial criteria standardised across countries and accounting
standards. Each bank report contains detailed unconsolidated and/or consolidated
balance sheets and income statements. Our results below are robust to using
unconsolidated or consolidated data, so we only report estimates based on
consolidated balance sheets. Our data are annual. Other databases, such as the SNL
Financials or Thomson Reuters, provide quarterly data on banks balance sheet, but
only for the largest banks and they are very weak on historical information and
coverage for non-listed banks. These alternative databases, therefore, are not really
suitable to study the bank lending channel.

The specialization categories in the Bankscope database of relevance for our analysis
are: (i) commercial banks, (ii) savings banks and (iii) cooperative banks. Thereby, we
excluded (also because of the low number of observations for each country) real
estate and mortgage banks, medium- and long-term credit banks, investment banks
and securities houses, islamic banks, non-banking credit institutions, specialised
government credit institutions, bank holdings and holding companies, central banks,
and multilateral government banks.

Bankscope contains data from 1989 onwards. However, the coverage during the first
half of 1990s is very limited. Furthermore, given our focus on the effects of
monetary policy in the euro area, it makes sense to select a sample characterized by a
homogeneous policy regime, as represented, for instance, by the transfer of
responsibility for setting the area wide short-term interest rate from national
monetary authorities to the European Central Bank. With this goal in mind, our
sample begins in 1999 and ends in 2011, with the last observation reflecting data
availability at time of writing.

More specifically, we employ four releases of data in electronic format so as to keep
track of the balance sheets of merged and failed banks which are no longer reported
in the new releases. By doing so we reduce the errors related to survivorship bias and
to a spurious burst of credit growth that only reflect take-overs between banks.

To quantify the bank lending channel, we consider (i) size measured by the logarithm
of banks' assets, (ii) liquidity measured by cash, interbank lending, reserves at central
banks plus government securities divided by total assets and (iii) capital measured by
banks equity capital and retained earnings divided by total assets. To ensure that the
size of the coefficients is comparable across variables, we standardise individual
characteristics across banks in each year. This implies that the parameters a
2j
in
equation (1) and
2j
in equation (5) below can be interpreted as a direct measure of
29

the overall monetary policy effect on loans, given that the average of the interaction
terms is zero. Also, bank-specific variables are normalised relative to the cross-
sectional average in each year, in an effort to remove low-frequency components.
Loan loss provisions divided by total assets as well as return on equity, defined as net
profits over a fiscal year divided by shareholders' equity, are also standardised.

30