Convergence in Euro-zone Retail Banking- What Interest Rate Pass-through Tells Us About Monetary Policy Transmission, Competition and Integration

Journal of International Money and Finance

23 (2004) 461492

www.elsevier.com/locate/econbase

ding author. T .ress: gh.sander CorresponE-mail add0261-5606/$ - see front matt

doi:10.1016/j.jimonn.2004.0el.: +49-221-82753419; fax: +49-221-82753131

@t-online.de (H. Sander).er# 2004 Elsevier Ltd. All rights reserved.2.001Convergence in euro-zone retail banking?What interest rate pass-through tells us aboutmonetary policy transmission, competition

and integration

Harald Sander a,b,, Stefanie Kleimeier b,ca Faculty of Economics and Business Administration, Claudiusstr.1, University of Applied Sciences

Cologne, 50678 Koln, Germanyb METEOR Fellow, Tongersestraat 53, Maastricht University, 6211 LM Maastricht, The Netherlands

c Limburg Institute of Financial Economics, Tongersestraat 53, Maastricht University,

6211 LM Maastricht, The Netherlands

Abstract

This study aims at unifying the empirical research on interest-rate pass-through in theeuro zone. After endogenously determining structural breaks we select optimal pass-throughmodels, which allow for thresholds and asymmetric adjustment. By applying these models tomonetary policy shocks as well as cost-of-funds changes, we show that in post-break periodsmonetary policy transmission has become faster, that heterogeneity across the euro zone hasdecreased in some banking markets, and that more competition improves the pass-throughpredominantly in deposit markets. As national characteristics are still important pass-through determinants, convergence remains incomplete and monetary policy will continue tooperate in a heterogeneous euro zone.# 2004 Elsevier Ltd. All rights reserved.

JEL classication: E43; E52; E58; F36

Keywords: Interest rates; Monetary policy; European Monetary Union; European banking; Competition

in banking; European nancial integration; Banking structure; Asymmetric adjustment; Cointegration

analysis; Threshold cointegration; Structural breaks

1. Introduction

How uniform is the monetary transmission process in the euro zone? Given thedominant role of bank nance in the euro zone, banks are important conveyers ofmonetary policy impulses.1 However, banking markets are often considered to be

H. Sander, S. Kleimeier / Journal of International Money and Finance 23 (2004) 461492462more resistant to convergence than other parts of the monetary transmission mech-anism. As such, divergences in national banking market structures and competitionas well as a lack of euro-zone banking market integration can be expected to leadto heterogeneous eects of monetary policy across the euro-zone economies.Recent literature has therefore focused on empirical analyses of the pass-through

of monetary policy impulses to retail banking interest rates in the euro zone.2

Overall, these studies agree that there is a substantial degree of short-run bankinterest rate stickiness. Furthermore, all studies nd considerable dierences in thepass-through not only across dierent bank lending and deposit rates but alsoacross countries. These dierences are typically attributed to the divergent struc-tures of national nancial systems. However, the single currency is often perceivedto be a unifying force by making the pass-through faster, more complete and morehomogeneous over the recent years. Nevertheless, the dierences in the results ofpass-through studies remain large and can be attributed mainly to four factors: (1)the choice of the exogenous market interest rate, (2) the length and timing of thesample periods, particularly with respect to the treatment of possible structuralbreaks, (3) the chosen methodology for the pass-through analysis, and (4) thedesign of the analysis of pass-through determinants.In this study we provide a unifying analysis of the euro-zone pass-through mech-

anism by addressing these four issues: First, the pass-through is investigated byusing both proxies for monetary policy rates as well as proxies for the banks costof funds. The rst approach focuses on the transmission of monetary policy impul-ses into the nancial sector while the second approach highlights the role of com-petition and market structures. Both approaches can be found in the literature andshould therefore be viewed as complementary. Our unifying analysis allows for adirect comparison. Second, we investigate if and when the pass-through has chan-ged between 1993 and 2002 by not postulating, but endogenously searching forstructural breaks. Third, we estimate a large variety of pass-through models,including threshold and asymmetric adjustment models. The model nally used foreach retail rate in each country is automatically selected according to statisticalcriteria set a priori. Finally, we investigate the determinants of the size, speed andconvergence of the pass-through process.The results of our study can be summarized as follows: First, the euro-zone pass-

through mechanisms have undergone considerable structural changes in the past

1 See Bernanke and Gertler (1995) and Kashyap and Stein (1993) for a discussion of the dierent

transmission channels of monetary policy.2 This literature includes BIS (1994), Cottarelli et al. (1995), Borio and Fritz (1995), Mojon (2000), de

Bondt (2002), de Bondt et al. (2002); Kleimeier and Sander (2002, 2003), Sander and Kleimeier (2002),

and Toolsema, Sturm and de Haan (2002), Heinemann and Schuler (2003).

decade. However, these structural breaks do not necessarily coincide with theintroduction of the single currency but have often occurred much earlier. Thisresult contests exogenously setting the break point in January 1999. One wouldthen eventually attribute the observed changes in the pass-through process to theintroduction of the single currency, while it may in fact reect the impact of earlierchanges in EU banking market regulation, or expectational eects in the run up toEMU, or the impact of lower money market rate volatility prior to 1999. A secondresult is that during the post-break period the pass-through of monetary policy

features of national nancial markets as well as macroeconomic factors such as

463H. Sander, S. Kleimeier / Journal of International Money and Finance 23 (2004) 461492interest-rate volatility, structural ination and growth can explain a considerablepart of the pass-through heterogeneity. However, legal and cultural dierencesremain statistically signicant determinants. We therefore conclude that neitherstructural convergence of nancial systems across countries nor a single monetarypolicy regime can be expected to fully homogenize the euro-zone pass-through inthe near future.

2. Data and methodology

2.1. Data selection

We investigate the pass-through process for ten dierent loan and deposit ratesin ten euro-zone countries over the period from January 1993 until October 2002.These rates are available from the ECB with a monthly frequency.3 We comparethe monetary policy approach with the cost-of-funds approach. For the for-mer we use the overnight money market rate as a proxy for the monetary policystance. For the industrial organization inspired cost-of-funds approach we follow

3 The ECB provides data for the following retail interest rate: overdrafts on cash accounts (N1), mort-

gage loans to households (N2), consumer loans to households (N3), short-term loans to enterprises

(N4), medium and long-term loans to enterprises (N5), and other lending rates (N6), current account

deposits (N7), time deposits (N8), savings accounts (N9), and other deposit rates (N10). Whereas some

national series start as early as 1980, data for a larger number of EMU member countries are available

only since the mid 1990s. Considering potential disturbing eects of the EMS crisis on our results, we

decided to focus on the period after 1992. We include Austria, Belgium, Finland, France, Germany,

Ireland, Italy, Netherlands, Portugal, and Spain in our sample.impulses has improved with respect to lending but not to deposit rates. We alsond that there is no improvement over time in the pass-through of cost-of-fundschanges. Furthermore, and in contrast to some earlier studies, we nd an incom-plete long-run pass-through for most retail rates. Interesting also, the size of thepass-through is typically higher the shorter maturity of the lending rate. However,the grip that monetary policy now has on long-term lending rates, such as mort-gage rates, has also improved. Whilst the pass-through mechanism has generallyremained heterogeneous across euro-zone countries, the market for short-term cor-porate lending has become more homogeneous, thus conveying the statisticalpicture of a more integrated market. Finally, we nd that the distinct structural

de Bondt (2002) by selecting the market interest rate with the highest correlation

with the respective retail lending or deposit rate as a proxy for the cost of funds. In

our study, this leads to the choice of the 10-year rate as the cost of funds rate for

mortgages, the 12-months rate for consumer loans, the 1-month rate for short-term

This specication avoids spurious regression problems but leads to a loss of

H. Sander, S. Kleimeier / Journal of International Money and Finance 23 (2004) 461492464information about long-run relationships. Fortunately, this information can be

recovered if BR and M are cointegrated. The VAR then needs to be augmented by

an (lagged) error correction term (ECT):

DBRt Xki1

bBR;iDBRti b1DMt Xni1

bM;iDMti bECTECTt1 et: 3

4 The main data source is Datastream. More details are given in Sander and Kleimeier (2005) avail-

able as LIFE Working Paper WP04-005 at http://www.fdewb.unimaas.nl/nance/workingpapers/.5 Whenever an optimal lag length has to be determined, the minimum AIC criterion is used allowing

for a maximum of four lags.corporate loans, the 6-months rate for medium- and long-term corporate loans, the

1-month rate for current account deposits, and the 3-months rate for time deposits

and savings accounts. For the analysis of the structural determinants of the pass-

through process, we collect a large number of banking market descriptors from

recent publications of the ECB (2000, 2002) and the OECD. Moreover, the usual

macro-economic and nancial development control variables are collected.4

2.2. The empirical pass-through model

Our empirical pass-through analysis employs a unifying approach that utilizes

VAR and cointegration methodologies allowing for asymmetric and threshold

adjustment. Traditionally, the pass-through process has simply been modeled as a

VAR process (Cottarelli and Kourelis, 1994):

BRt b0 Xki1

bBR;iBRti b1Mt Xni1

bM;iMti et; 1

where BRt and Mt are lending and market rates, respectively, and k and n indi-

cate the optimal lag lengths.5 However, it is important to recognize that the time

series for interest rates typically exhibit an I(1) property. In this case, the empirical

pass-through model is best estimated using rst dierences:

DBRt Xki1

bBR;iDBRti b1DMt Xni1

bM;iDMti et: 2

The ECT measures the deviation from the long-run equilibrium, which can beobtained from the estimated error of the cointegration regression:

BRt h0 hMt ut: 4We estimate the appropriate version of the pass-through model as either Eqs. (1)

and (2), or (3) depending on the time series and cointegration properties of theinterest rate series.6 In all specications, the impact multiplier is estimated by thecoecient b1. A value of less than 1 indicates sluggish adjustment, also known as

465H. Sander, S. Kleimeier / Journal of International Money and Finance 23 (2004) 461492lending rate stickiness. The long-run relationship between market rates and retailrates is given by Eq. (4) and can be interpreted either as a cointegration relation-ship or as the long-run solution of the VAR. The long-term multiplier h can bedirectly obtained from estimating Eq. (4) if the rates are cointegrated. Otherwise,the long-term multiplier has to be calculated from (1) or (2) as:

h b1 Pn

i1 bM;i1Pki1 bBR;i : 5

A full pass-through in the long run is reected by h 1. An imperfect pass-through h < 1 could be caused by a less than perfect elasticity of demand forbanking products, the existence of market power, a lack of market contestability,switching costs, or information asymmetries. If the long-run pass-through is foundto be overshooting h > 1 in lending markets, this can be interpreted as a situationwhere banks increase lending rates to compensate for higher risks instead ofrationing credit.7

Given the major developments in the euro zone since 1992, the long-run relation-ship may be subject to structural changes. However, unlike other pass-throughstudies we do not exogenously postulate a break point and then test for its pres-ence. Instead, we determine the presence and timing of the break endogenously byestimating a supremum F (supF) test for Eq. (4). This test can be interpreted as arolling test where standard Chow tests are conducted for a series of dierent breakpoints, which move through the mid-80% of the sample period.8 On the base ofthese tests we constructwhen appropriatepre- and post-break periods for everynational retail interest rate. This allows us to obtain additional information on thetiming of structural changes and to estimate pass-through models for break-freesample periods.

6 We employ various tests to establish whether or not the interest rate series exhibit unit roots. Given

the likely presence of a structural break, we conduct standard unit root tests for the pre- and post-break

periods. For the full period we additionally estimate unit root tests, which are valid in the presence of a

structural break. Details are available in Sander and Kleimeier (2004).7 De Bondt (2002) discusses a model where banks price higher default probabilities into lending rates.

His perfect-competition model assumes that banks are able to distinguish between risky and non-risky

borrowers.8 For details on this test see Andrews (1993), Diebold and Chen (1996), Hansen (1992). SupF equals

the largest Chow F-statistic and is compared to critical values as reported by Hansen (1992).

H. Sander, S. Kleimeier / Journal of International Money and Finance 23 (2004) 461492466While most pass-through studies focus on symmetric adjustment towardthe long-run equilibrium, we have advocated in a previous study (Sander andKleimeier, 2002) that threshold and asymmetric adjustment mechanisms shouldboth be considered for two main reasons: First, retail rate adjustment patterns inthe euro zone are indeed frequently either asymmetric or occur only beyond a cer-tain threshold. Thus, they should not be ignored. Second, using models with asym-metries allows us to detect cointegration in cases where there are asymmetries andwhere other methods would thus fail to detect cointegration and wrongly re-directthe researcher to the pass-through model of Eq. (2).We include ve asymmetric specications for the adjustment of interest rates.

Consider rst the symmetric pass-through model. Here the ECT is dened as

ECTt1 ut1 6

and cointegration testing is based on the DurbinWatson (DW), DickeyFuller(DF) and augmented DickeyFuller (ADF) tests. As the rst asymmetric modelwe consider the threshold autoregressive model (TAR0) developed by Tong(1983). The model distinguishes whether the explained interest rate is above orbelow its equilibrium level. Thus, the TAR0 allows for asymmetric adjustmentdepending on the sign of the equilibrium deviation. For example, if the moneymarket rate decreases without an immediate adjustment of the lending rate, weobtain a positive realization of the error term ut. When in this case the auto-regressive decay is faster than in the case of money market rate increases, the lend-ing rate adjustment is faster downward than upward. For this TAR0 model, theECT is dened as

ECTt1 It ut1 1 It ut1 7

where It represents a Heaviside indicator for dierent states of ut1 such that

It 1 if ut1 00 if ut1 < 0

: 8

Using this denition we estimate Eq. (9):

Dut Itq1ut1 1 Itq2ut1 Xmi1

q2iDuti et: 9

Cointegration testing takes the form of a modied ADF test. The null of no coin-tegration is rejected if the estimated F-statistic for H0: q1 q2 0 is statisticallysignicant based on critical values provided by Enders and Siklos (2000). If coin-tegration is established, an F-test for H0: q1 q2 indicates the presence of asym-metry.The second asymmetric model (TAR) is a modication of the TAR0 in the

sense that the threshold is now allowed to deviate from zero. The rationale is thatretail rates may adjust dierently to a disequilibrium once a certain minimum devi-

ation in one direction is exceeded. For the TAR model, the Heaviside indicator inconjunction with Eq. (7),9 is dened as

467H. Sander, S. Kleimeier / Journal of International Money and Finance 23 (2004) 461492It 1 if ut1 a00 if ut1 < a0: 10

Following Chan (1993), the optimal threshold a0 is found by searching over themid-80% of the distribution of ut and selecting the model for which the residualsum of squares is minimized. Cointegration and asymmetry testing proceeds withthe above-described F-tests.The third variation is a Band-TAR model (B-TAR), which can reect both

interest rate stickiness, driven by menu-cost behavior of banks, as well as interestrate smoothing. For example, menu-cost behavior could be relevant if we nd coin-tegration only outside a band bordered by a0 and a0. For the B-TAR model, theHeaviside indicator in conjunction with Eq. (7) is now dened as

Ijt I1t 1 if ut1 a0 and 0 otherwiseI2t 1 if jut1j < a0 and 0 otherwiseI3t 1 if ut1 a0 and 0 otherwise

8 0 leading to theM-TAR specication.The objective of the methodology employed in this study is to obtain the optimal

October 1996 for the monetary policy approach and December 1996 for the cost-of-funds approach. Note that dierent banking market segments in dierent coun-

H. Sander, S. Kleimeier / Journal of International Money and Finance 23 (2004) 461492468tries show dierent breakpoints. In particular, for Spain, Portugal, and Italy break-points are as early as 1995 and 1996, possibly showing the impact of the run-up toEMU with reduced money market rate volatility and ination convergence.Another early starter is Ireland where some breakpoints are already found inDecember 1993.

3.2. Size and speed of the euro-zone pass-through

The optimal pass-through model is estimated for all break-free periods for themonetary policy and cost-of-funds approach, respectively. Table 2 illustrates ourresults by reporting the unweighted averages of the obtained multipliers10. In thelong run, with the possible exception of short-term corporate loans, the size of thepass-through process is still far from complete. For the monetary policy approach

10 The details of this analysis including all individual country and rate multipliers can be found in

Sander and Kleimeier (2004).pass-through model rather than arbitrarily selecting one. As such, we identifybreak-free sub-periods. We then proceed with unit root testing. If the rates are I(0),we estimate the pass-through model as in Eq. (1). If the rates are I(1), we rst esti-mate all ve asymmetric TAR-type models, select the best asymmetric model basedon the AIC criterion, and test this best model for asymmetric cointegration. Ifasymmetric cointegration is conrmed, we estimate the pass-through model as inEq. (3) with the appropriate asymmetric ECT. If asymmetric cointegration is rejec-ted, we test for symmetric cointegration andif conrmedinclude a symmetricECT in the pass-through model of Eq. (3). If symmetric cointegration is also rejec-ted, the pass-through model is estimated according to Eq. (2) without any ECT.Finally, based on the selected pass-through model multipliers are obtained for avariety of positive and negative interest rate shocks.

3. Pass-through and monetary transmission

3.1. Structural changes in euro-zone banking

The euro-zone banking system has undergone dramatic structural changes in thepast decade driven by not only the introduction of the single currency but also the1992/93 ERM crisis and EU regulatory changes, including the 2nd Banking Direc-tive. Our analysis indicates that the endogenously determined structural breaks inthe long-run relationship between market and retail rates already occur beforeJanuary 1999. The results reported in Table 1 reveal that the average breakpoint is

Table1

Structuralbreaksinthelong-runrelationship

Country

Bankrate

Monetarypolicy

approach

Cost-of-fundsapproach

supFa

Breakpoint

supFa

Breakpoint

Austria

N2mortgage

loansto

households

196.25

July-97

14.41

February-99

N3consumerloansto

households

253.10

September-98

188.86

August-97

N4short-termloansto

enterprises

196.26

August-97

236.56

August-97

N7currentaccountdeposits

221.39

Novem

ber-99

200.23

Novem

ber-99

N8timedeposits

199.14

March-97

220.36

March-97

Belgium

N2mortgage

loansto

households

89.26

August-95

60.78

May-98

N3consumerloansto

households

319.18

Decem

ber-95

182.81

Decem

ber-95

N4.1short-termloansto

enterprises

48.19

April-95

54.69

March-95


enterprises

21.25

January-94

24.45

Decem

ber-93

N5mediumandlong-termloansto

enterprises

65.56

October-95

38.12

August-96

N8timedeposits

23.91

Decem

ber-93

26.79

Decem

ber-93

N9savingsaccounts

226.14

Decem

ber-95

221.73

Decem

ber-95

Finland

N2mortgage

loansto

households

105.93

September-96

99.64

March-94

N3consumerloansto

households

101.80

September-96

86.25

September-97


enterprises

56.00

January-96

47.41

April-98


49.11

February-97

46.62

February-98

N8timedeposits

170.27

August-97

193.92

Novem

ber-99

France

N4short-termloansto

enterprises

132.42

June-97

150.03

June-97


enterprises

222.29

March-97

169.78

April-97

N8timedeposits

11.38

January-00

insignicant

8.24

January-00

insignicant

N9savingsaccounts

112.11

May-98

104.56

May-98

GermanyN2mortgage

loansto

households

56.62

October-96

36.16

June-95

N3consumerloansto

households

442.05

February-97

480.06

March-97

N4short-termloansto

enterprises

71.72

July-00

81.82

February-03


enterprises

11.22

January-00

insignicant

9.99

January-00

insignicant

N8.1timedeposits

40.30

September-99

22.97

September-99

N8.2timedeposits

22.35

September-99

9.06

January-00

insignicant

N9.1savingsaccounts

935.80

September-99

732.66

September-99

N9.2savingsaccounts

36.67

October-95

33.47

Novem

ber-95

Ireland

N1overdraftsoncash

accounts

128.80

Decem

ber-98

145.26

Decem

ber-98

(continued

onnextpage)

469H. Sander, S. Kleimeier / Journal of International Money and Finance 23 (2004) 461492

Table1(continued)

Country

Bankrate

Monetarypolicy

approach

Cost-of-fundsapproach

supFa

Breakpoint

supFa

Breakpoint

Ireland

N2mortgage

loansto

households

61.51

August-99

49.53

March-94

N4short-termloansto

enterprises

33.00

Novem

ber-95

35.15

Decem

ber-93


enterprises

40.68

Decem

ber-93

70.41

Decem

ber-93

N6otherlendingrates

45.07

January-00

4.66

January-00

insignicant

N9.1savingsaccounts

149.76

Decem

ber-93

152.03

Decem

ber-93

N9.2savingsaccounts

129.03

Decem

ber-93

205.33

Decem

ber-93

Italy

N2mortgage

loansto

households

69.42

Decem

ber-97

131.97

May-98


enterprises

31.26

February-95

30.09

July-99


enterprises

39.91

February-95

21.20

June-94


enterprises

48.08

Novem

ber-97

43.01

Decem

ber-96


70.21

February-95

44.16

February-95

N8timedeposits

110.97

September-97

102.79

January-97

Nether-

lands

N2mortgage

loansto

households

61.82

September-96

41.20

June-95

N4short-termloansto

enterprises

30.02

August-97

66.16

August-98


466.47

Decem

ber-98

466.47

Decem

ber-98

N8.1timedeposits

64.49

Novem

ber-95

51.29

Novem

ber-95

N8.2timedeposits

65.67

Decem

ber-95

56.09

Decem

ber-95

Portugal

N2mortgage

loansto

households

98.82

September-97

58.86

Decem

ber-95

N3consumerloansto

households

178.60

April-95

73.79

April-98


enterprises

100.14

July-94

235.71

October-99


enterprises

173.12

February-95

74.02

Novem

ber-99

N8.1timedeposits

43.22

January-96

13.83

Novem

ber-00

N8.2timedeposits

47.58

February-96

32.09

July-96

Spain

N2mortgage

loansto

households

69.89

September-96

69.89

September-96

N3consumerloansto

households

111.60

Novem

ber-96

103.25

March-94

N4short-termloansto

enterprises

9.52

September-96

12.18

Novem

ber-96


enterprises

59.41

March-96

60.49

September-94


48.46

February-95

99.52

January-95

N8timedeposits

80.42

March-96

113.56

Decem

ber-93

aThesupFtestisbasedontheestimatedcoe

cientforEq.(4)usingmonthlydata

forthefullsampleperiodofJanuary

1993to

October2002.Statistical

signicance

ofthebreakpointisestablished

basedoncriticalvaluesreported

byHansen(1992).

H. Sander, S. Kleimeier / Journal of International Money and Finance 23 (2004) 461492470

Table2

Theaveragepass-throughprocessanditsasymmetries

Retailrates

Period

Statis-

tica

Multipliersfora+1%

shock

Asymmetriesinmultipliersb

+1%versus1

%shock

+1%versus+0.25%shock

impact

1 mth

3 mths

6 mths

12

mths

long-

run

1 mth

3 mths

6 mths

12

mths

1 mth

3 mths

6 mths

12

mths

PanelA:The

Monetary

Policy

Approach

All

pre

average

0.20

0.31

0.42

0.49

0.53

0.56

1.00

0.98

0.98

0.98

1.00

1.00

1.00

0.99

stddev

0.17

0.23

0.29

0.31

0.31

0.32

0.00

0.10

0.08

0.06

0.00

0.08

0.07

0.06

post

average

0.20

0.37

0.48

0.53

0.54

0.57

1.00

1.01

0.99

0.99

1.00

1.00

0.99

0.99

stddev

0.17

0.28

0.31

0.33

0.36

0.38

0.00

0.13

0.11

0.08

0.00

0.07

0.07

0.06

Alllending

pre

average

0.20

0.33

0.46

0.54

0.58

0.62

1.00

0.98

0.99

0.99

1.00

1.00

1.01

1.00

stddev

0.15

0.23

0.31

0.33

0.34

0.35

0.00

0.10

0.07

0.03

0.00

0.06

0.03

0.01

post

average

0.22

0.43

0.56

0.62

0.65

0.68

1.00

1.02

1.00

1.00

1.00

1.00

1.00

1.00

stddev

0.15

0.26

0.30

0.31

0.34

0.37

0.00

0.16

0.13

0.07

0.00

0.06

0.05

0.01

Alldeposit

pre

average

0.20

0.28

0.38

0.43

0.46

0.47

1.00

0.97

0.97

0.97

1.00

1.00

0.99

0.99

stddev

0.20

0.24

0.25

0.26

0.27

0.26

0.00

0.10

0.09

0.09

0.00

0.09

0.10

0.09

post

average

0.17

0.27

0.35

0.38

0.38

0.40

1.00

0.98

0.97

0.97

1.00

0.99

0.99

0.99

stddev

0.20

0.27

0.30

0.31

0.32

0.34

0.00

0.06

0.08

0.09

0.00

0.08

0.09

0.09

N2mortgageloans

tohouseholds

pre

average

0.14

0.22

0.32

0.43

0.52

0.54

1.00

1.02

1.01

1.00

1.00

1.02

1.01

1.00

stddev

0.12

0.19

0.26

0.32

0.39

0.34

0.00

0.05

0.04

0.01

0.00

0.05

0.04

0.01

post

average

0.21

0.45

0.55

0.57

0.57

0.62

1.00

0.97

0.95

0.96

1.00

0.99

0.98

1.00

stddev

0.18

0.28

0.29

0.29

0.29

0.32

0.00

0.10

0.10

0.08

0.00

0.03

0.04

0.01

N3consumerloans

tohouseholds

pre

average

0.16

0.27

0.36

0.42

0.46

0.63

1.00

0.98

0.98

0.98

1.00

1.00

1.00

1.00

stddev

0.13

0.24

0.30

0.34

0.35

0.51

0.00

0.05

0.05

0.05

0.00

0.00

0.00

0.00

post

average

0.17

0.37

0.49

0.55

0.56

0.60

1.00

1.03

1.01

1.00

1.00

1.01

1.01

1.00

stddev

0.12

0.33

0.43

0.51

0.54

0.53

0.00

0.05

0.02

0.00

0.00

0.03

0.02

0.00

N4short-termloans

toenterprises

pre

average

0.24

0.43

0.58

0.68

0.71

0.74

1.00

0.98

0.99

1.00

1.00

0.98

1.00

1.00

stddev

0.16

0.22

0.31

0.31

0.30

0.29

0.00

0.08

0.02

0.00

0.00

0.09

0.02

0.00

post

average

0.24

0.46

0.67

0.77

0.84

0.87

1.00

1.08

1.05

1.03

1.00

0.99

1.00

1.00

stddev

0.15

0.22

0.23

0.21

0.27

0.36

0.00

0.24

0.19

0.09

0.00

0.06

0.07

0.02

N5mediumandlong-

term

loansto

enterprises

pre

average

0.22

0.32

0.44

0.49

0.50

0.51

1.00

0.94

0.96

0.97

1.00

1.01

1.01

0.99

stddev

0.19

0.25

0.36

0.36

0.36

0.36

0.00

0.19

0.15

0.05

0.00

0.03

0.05

0.05

(continued

onnextpage)


Table2(continued)

Retailrates

Period

Statis-

tica

Multipliersfora+1%

shock

Asymmetriesinmultipliersb

+1%versus1

%shock

+1%versus+0.25%shock

impact

1 mth

3 mths

6 mths

12

mths

long-

run

1 mth

3 mths

6 mths

12

mths

1 mth

3 mths

6 mths

12

mths

PanelA:The

Monetary

Policy

Approach

post

average

0.24

0.38

0.42

0.46

0.49

0.50

1.00

0.96

0.97

0.99

1.00

0.97

0.98

1.00

stddev

0.14

0.28

0.24

0.22

0.22

0.23

0.00

0.06

0.03

0.02

0.00

0.06

0.03

0.01

N7currentaccount

deposits

pre

average

0.06

0.10

0.15

0.19

0.21

0.23

1.00

0.91

0.92

0.96

1.00

1.05

1.03

1.02

stddev

0.08

0.13

0.18

0.21

0.22

0.22

0.00

0.21

0.14

0.07

0.00

0.11

0.07

0.05

post

average

0.04

0.09

0.16

0.21

0.23

0.22

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.00

stddev

0.08

0.14

0.22

0.29

0.32

0.32

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

N8timedeposits

pre

average

0.28

0.39

0.52

0.60

0.64

0.64

1.00

0.99

0.97

0.96

1.00

0.97

0.96

0.96

stddev

0.23

0.25

0.21

0.16

0.15

0.16

0.00

0.05

0.09

0.12

0.00

0.10

0.12

0.12

post

average

0.26

0.40

0.50

0.53

0.53

0.57

1.00

0.96

0.94

0.94

1.00

0.98

0.97

0.97

stddev

0.22

0.28

0.27

0.27

0.28

0.31

0.00

0.08

0.10

0.12

0.00

0.11

0.13

0.13

N9savings

accounts

pre

average

0.14

0.20

0.27

0.30

0.31

0.33

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.00

stddev

0.10

0.16

0.22

0.25

0.25

0.23

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

post

average

0.11

0.18

0.22

0.21

0.21

0.20

1.00

1.01

1.01

1.01

1.00

1.01

1.01

1.01

stddev

0.13

0.24

0.28

0.28

0.27

0.25

0.00

0.03

0.03

0.02

0.00

0.03

0.03

0.02

PanelB:The

CostofFundsApproach

All

pre

average

0.33

0.43

0.55

0.61

0.63

0.65

1.00

1.00

0.98

1.00

1.00

1.01

1.01

1.01

stddev

0.29

0.25

0.31

0.32

0.32

0.34

0.00

0.07

0.26

0.20

0.00

0.06

0.04

0.05

post

average

0.33

0.42

0.53

0.58

0.60

0.60

1.00

1.03

1.03

1.03

1.00

1.04

1.05

1.06

stddev

0.31

0.27

0.30

0.30

0.34

0.34

0.00

0.12

0.17

0.20

0.00

0.17

0.30

0.37

Alllending

pre

average

0.31

0.46

0.60

0.69

0.72

0.73

1.00

1.00

1.02

1.03

1.00

1.00

1.01

1.01

stddev

0.19

0.25

0.31

0.31

0.31

0.33

0.00

0.09

0.19

0.24

0.00

0.03

0.03

0.05

post

average

0.33

0.45

0.60

0.66

0.71

0.69

1.00

1.05

1.06

1.06

1.00

1.04

1.08

1.10

stddev

0.22

0.22

0.25

0.26

0.31

0.31

0.00

0.13

0.21

0.26

0.00

0.21

0.39

0.47

Alldeposit

pre

average

0.35

0.39

0.49

0.50

0.51

0.54

1.00

1.00

0.93

0.95

1.00

1.02

1.01

1.00

stddev

0.39

0.26

0.31

0.30

0.29

0.31

0.00

0.04

0.33

0.13

0.00

0.08

0.05

0.04

post

average

0.34

0.39

0.44

0.46

0.45

0.47

1.00

1.00

0.99

0.98

1.00

1.03

1.01

1.00

stddev

0.40

0.33

0.34

0.33

0.32

0.35

0.00

0.08

0.07

0.06

0.00

0.10

0.07

0.05


N2mortgageloansto

households

pre

average

0.24

0.37

0.54

0.59

0.66

0.60

1.00

1.04

1.10

1.13

1.00

0.98

1.02

1.03

stddev

0.21

0.31

0.41

0.37

0.39

0.37

0.00

0.15

0.35

0.45

0.00

0.06

0.05

0.09

post

average

0.22

0.31

0.43

0.55

0.67

0.65

1.00

1.06

1.12

1.15

1.00

1.00

1.03

1.04

stddev

0.20

0.23

0.23

0.29

0.44

0.44

0.00

0.15

0.34

0.44

0.00

0.08

0.06

0.09

N3consumerloansto

households

pre

average

0.32

0.47

0.43

0.58

0.61

0.63

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.00

stddev

0.21

0.26

0.26

0.31

0.36

0.37

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

post

average

0.20

0.32

0.46

0.54

0.56

0.56

1.00

0.99

1.00

1.00

1.00

1.00

1.00

1.00

stddev

0.09

0.15

0.22

0.30

0.34

0.35

0.00

0.02

0.00

0.00

0.00

0.00

0.00

0.00

N4short-termloansto

enterprises

pre

average

0.33

0.52

0.71

0.81

0.85

0.91

1.00

0.98

0.97

0.97

1.00

1.01

1.00

1.01

stddev

0.17

0.23

0.28

0.29

0.28

0.28

0.00

0.05

0.07

0.06

0.00

0.01

0.01

0.01

post

average

0.42

0.55

0.73

0.75

0.77

0.72

1.00

0.11

1.08

1.07

1.00

1.11

1.20

1.24

stddev

0.26

0.21

0.21

0.21

0.24

0.20

0.00

0.17

0.20

0.21

0.00

0.34

0.65

0.80

N5mediumandlong-

term

loansto

enterprises

pre

average

0.39

0.45

0.62

0.71

0.69

0.67

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.00

stddev

0.22

0.18

0.20

0.23

0.20

0.24

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

post

average

0.44

0.53

0.71

0.75

0.77

0.76

1.00

1.06

1.01

0.99

1.00

1.02

1.02

1.01

stddev

0.19

0.21

0.20

0.22

0.24

0.28

0.00

0.10

0.03

0.03

0.00

0.05

0.06

0.03

N7currentaccount

deposits

pre

average

0.10

0.17

0.22

0.25

0.25

0.25

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.00

stddev

0.09

0.18

0.24

0.25

0.25

0.24

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

post

average

0.10

0.18

0.23

0.27

0.28

0.28

1.00

0.97

0.98

0.99

1.00

1.01

1.00

1.00

stddev

0.10

0.16

0.20

0.25

0.28

0.28

0.00

0.07

0.07

0.02

0.00

0.03

0.02

0.10

N8timedeposits

pre

average

0.51

0.50

0.67

0.67

0.68

0.70

1.00

1.01

1.00

0.98

1.00

1.04

1.01

1.01

stddev

0.44

0.20

0.27

0.22

0.20

0.21

0.00

0.05

0.04

0.05

0.00

0.11

0.06

0.06

post

average

0.50

0.54

0.64

0.65

0.64

0.66

1.00

1.01

1.00

0.97

1.00

1.05

1.03

1.00

stddev

0.47

0.32

0.31

0.27

0.25

0.27

0.00

0.11

0.09

0.05

0.00

0.15

0.10

0.07

N9savings

accounts

pre

average

0.17

0.25

0.32

0.33

0.35

0.37

1.00

0.99

0.73

0.90

1.00

1.00

1.00

1.00

stddev

0.16

0.21

0.21

0.28

0.29

0.26

0.00

0.05

0.66

0.25

0.00

0.01

0.00

0.00

post

average

0.12

0.15

0.17

0.17

0.17

0.17

1.00

1.01

1.00

1.00

1.00

1.00

1.00

1.00

stddev

0.13

0.17

0.19

0.20

0.20

0.20

0.00

0.02

0.01

0.01

0.00

0.01

0.00

0.00

aThereported

statisticsaretheunweightedaverage(average)andthestandard

deviation(std

dev)oftheestimatedmultipliersbasedontheoptimalpass-

through

model.

bAsymmetriesinmultipliersaredened

asthemultiplierfora+1%

changedivided

bythemultiplierforthe1

%or+0.25%change,respectively.


we nd long-run multipliers for loans on average around 0.6 to 0.7. For deposits,the average even lies below 0.5. For the cost-of-funds approach the obtained long-run multipliers are somewhat higher but also fall short of a full pass-through.11

Viewed from an industrial organization perspective, the latter result indicates thateuro-zone banking markets may exhibit some form of imperfect competition, suchas market power, lack of contestability, switching costs, or informational asymme-tries. Turning to the short run, our impact and intermediate multipliers indicate thepresence of severe price rigidities for both approaches. Nevertheless, the results

H. Sander, S. Kleimeier / Journal of International Money and Finance 23 (2004) 461492474also show remarkable increases in the intermediate multipliers for the post-breakperiod. This reects faster price adjustments for some banking products, parti-cularly mortgages, consumer loans, and short-term loans to enterprises. Regardingmortgages, it is striking that the eciency of the pass-through process hasincreased with respect to monetary policy impulses while the role of cost of fundshas diminished in the short-run adjustment. Possibly the increasing use of exiblerate mortgages is reected in these results. Consequently, and as argued by Sellon(2002) in the context of the US, monetary policy targeted at short-term marketrates has increased its impact on the cost of mortgages. In consumer lending,though some improvements have been taking place, the pass-through remainsamong the least perfect. For corporate loans the picture is mixed. Regarding short-term corporate loans, the already fast and almost complete monetary policy pass-through has improved over time while the cost-of-funds 6-months, 12-months, andlong-run multipliers have decreased. The opposite picture emerges for longer-termcorporate loans. Given the nature of these loans a market rate with a matchedmaturity might be a better explanatory variable.12

In summary, the multipliers seem to indicate that the size and speed of the pass-through have improved in the post-break period. However, this observation is onlyvalid for the lending rates reaction to monetary policy innovations. For the cost-of-funds approach the results are less clear-cut. To prove these points statisticallywe regress the size and speed of the pass-through on post-break, country, and ratedummies.13 Size is dened as the value of the long-run multiplier (h). Speed isdened as the impact and intermediate multipliers relative to the long-run multi-plier. The results are shown in Table 3. They conrm that the size of the pass-through has not improved signicantly in the post-break period. However, a stat-istically signicant increase in the speed of the pass-through process in the periodfrom 1 to 6 months is clearly identiable for the monetary policy approach but not

11 This result does not depend on the choice of the market rate proxy and is thus standing in contrast

to the studies by de Bondt (2002) and de Bondt et al. (2002). Given the partly dierent approaches and

timing of the structural breaks, reconciling these dierences remains an important task for future

research.12 It could be argued that monetary policy targeted at short-term interest rates has only an improved

inuence on the short-term rather than long-term lending rates to enterprises. If, for example, the cen-

tral bank wants to inuence the cost of investment borrowing of small and medium size enterprises it

appears that she should particularly consider her policys impact on longer-term market rates.13 We are grateful to Robert DeYoung for suggesting this regression framework.

Table3

Countryandmarketdeterminantsoftheinterest-ratepass-through

Independentvariableb

Dependentvariablea

Size

Speed

Convergence

long-run

impact

1mth

3mth

6mth

12mths

rlong-run

blong-run

PanelA:The

Monetary

Policy

Approach

Constant

0.561

0.464

0.753

0.696

0.752

0.840

0.357

3.454

4.269

4.164

5.974

5.906

8.504

10.603

2.162

1.426

Pre-break-m

ultiplier

4.226

1.877

Austria

0.054

0.206

0.182

0.194

0.188

0.156

0.057

0.672

0.372

1.664

1.301

1.483

1.911

1.765

0.309

0.258

Belgium

0.222

0.019

0.200

0.005

0.071

0.100

0.171

1.993

1.724

0.178

1.620

0.046

0.824

1.285

1.056

0.906

Finland

0.051

0.235

0.470

0.247

0.151

0.085

0.069

1.446

0.361

1.941

3.433

1.933

1.569

0.986

0.385

0.602

Germany

0.026

0.055

0.150

0.280

0.243

0.163

0.075

0.500

0.207

0.521

1.245

2.496

2.882

2.156

0.479

0.240

Ireland

0.144

0.167

0.091

0.246

0.212

0.146

0.057

0.569

1.045

1.426

0.688

1.993

2.280

1.753

0.327

0.246

Italy

0.218

0.117

0.173

0.111

0.159

0.105

0.220

1.013

1.619

1.024

1.341

0.919

1.755

1.292

1.303

0.421

Netherlands

0.072

0.093

0.013

0.222

0.222

0.114

0.005

1.019

0.505

0.771

0.098

1.744

2.329

1.330

0.027

0.433

Portugal

0.289

0.353

0.370

0.155

0.054

0.033

0.197

1.336

2.116

3.054

2.825

1.266

0.590

0.397

1.152

0.573

Spain

0.134

0.297

0.403

0.089

0.082

0.184

0.088

0.448

0.983

2.574

3.095

0.732

0.894

2.242

0.515

0.197

N3-consumer-loans-to-households

0.018

0.069

0.010

0.019

0.030

0.090

0.114

3.974

0.150

0.694

0.090

0.179

0.378

1.267

0.769

1.988

N4-short-term-oans-to-enterprises

0.250

0.126

0.051

0.016

0.026

0.036

0.249

0.287

2.622

1.563

0.554

0.187

0.407

0.632

2.086

0.165

(continued

onnextpage)


Table3(continued)

Independentvariableb

Dependentvariablea

Size

Speed

Convergence

long-run

impact

1mth

3mth

6mth

12mths

rlong-run

blong-run

N5-m

edium-andlong-termloansto

enterprises

0.021

0.065

0.016

0.014

0.033

0.015

0.026

1.302

0.213

0.772

0.172

0.161

0.503

0.245

0.213

0.781

N7-currentaccountdeposits

0.392

0.053

0.021

0.009

0.010

0.029

0.593

2.740

3.318

0.525

0.182

0.082

0.120

0.410

4.001

1.210

N8-timedeposits

0.006

0.072

0.016

0.031

0.050

0.028

0.151

0.795

0.062

0.916

0.185

0.374

0.812

0.511

1.302

0.504

N9-savingsaccounts

0.201

0.073

0.207

0.079

0.039

0.032

0.264

2.423

1.725

0.736

1.856

0.755

0.502

0.449

1.806

1.234

N6/N10-otherlendingordepositratesc

0.364

0.272

0.115

0.034

0.019

0.001

0.358

0.306

1.567

1.381

0.515

0.162

0.124

0.011

1.227

0.078

Post-breakdummy

0.001

0.071

0.164

0.157

0.092

0.036

0.116

0.017

1.563

3.188

3.269

2.540

1.108

1.721

Adjusted

R2

34.3%

24.3%

29.9%

27.7%

24.2%

8.8%

26.7%

10.7%

Numberofobservations

114

114

114

114

114

114

114

56

PanelB:The

CostofFundsApproach

Constant

0.642

0.735

0.567

0.786

0.749

0.891

0.380

5.425

5.349

5.571

5.248

7.371

9.439

14.668

2.718

2.619

Pre-break-m

ultiplier

5.954

3.363

Austria

0.034

0.400

0.051

0.046

0.214

0.170

0.092

0.753

0.254

2.735

0.424

0.388

2.432

2.526

0.592

0.353

Belgium

0.024

0.139

0.094

0.086

0.215

0.156

0.014

0.395

0.196

1.035

0.854

0.790

2.657

2.531

0.101

0.198

Finland

0.097

0.396

0.095

0.136

0.120

0.158

0.093

1.478

0.735

2.727

0.801

1.156

1.377

2.362

0.601

0.687

Germany

0.066

0.159

0.201

0.108

0.213

0.140

0.110

1.396

0.547

1.195

1.845

1.002

2.660

2.283

0.781

0.709

Ireland

0.199

0.067

0.277

0.285

0.334

0.203

0.026

2.154

1.597

0.493

2.467

2.577

4.053

3.215

0.182

1.073


Italy

0.131

0.339

0.000

0.001

0.182

0.128

0.077

1.361

1.044

2.462

0.003

0.013

2.199

2.014

0.530

0.665

Netherlands

0.048

0.335

0.054

0.053

0.174

0.133

0.058

0.261

0.365

2.302

0.450

0.450

1.9860

1.986

0.374

0.124

Portugal

0.145

0.328

0.110

0.031

0.158

0.121

0.091

5.531

1.146

2.354

0.9640

0.279

1.881

1.881

0.618

2.699

Spain

0.282

0.389

0.055

0.067

0.138

0.147

0.025

1.469

2.235

2.807

0.484

0.601

1.657

2.304

0.173

0.706

N3-consumerloansto

households

0.111

0.069

0.116

0.008

0.024

0.062

0.087

3.498

1.113

0.622

1.281

0.089

0.364

1.228

0.745

2.187

N4-shortterm

loansto

enterprises

0.154

0.002

0.053

0.050

0.011

0.047

0.208

1.063

1.864

0.017

0.709

0.6850

0.203

1.1210

2.166

0.751

N5-m

ediumandlongterm

loans

0.049

0.065

0.102

0.160

0.131

0.013

0.112

2.141

toenterprises

0.469

0.571

1.094

1.732

1.914

0.255

0.923

1.247

N7-currentaccountdeposits

0.439

0.060

0.186

0.131

0.053

0.041

0.408

5.063

4.088

0.510

1.914

1.367

0.748

0.751

3.260

2.687

N8-timedeposits

0.002

0.185

0.113

0.120

0.032

0.060

0.153

2.688

0.025

2.021

1.512

1.619

0.589

1.420

1.578

2.023

N9-savingsaccounts

0.299

0.108

0.054

0.065

0.093

0.123

0.271

2.456

2.876

0.948

0.577

0.706

1.357

2.339

2.237

1.465

N6/N10otherlendingordepositratesc

0.640

0.159

0.008

0.463

0.360

0.455

0.427

3.024

3.050

0.691

0.041

2.482

2.590

4.277

1.746

0.843

postbreakdummy

0.052

0.030

0.049

0.069

0.052

0.039

0.042

1.039

0.556

1.103

1.552

1.586

1.561

0.723

Adjusted-R

237.5%

8.6%

7.4%

10.2%

10.8%

15.6%

23.0%

18.9%

Numberofobservations

115

115

115

115

115

115

115

57

aThedependentvariablesoftheseOLSregressionsarethemultipliersfora+1%

shock

inthemonetary

policy

orcostoffundsrate.

bForeach

independentvariabletheestimatedcoe

cientisreported

inthetoprowandthetstatisticisreported

initalicsinthebottomrow.

cTheother

ratesreferto

N10other

depositratesforthemonetary

policy

approach

andN6other

lendingratesforthecostoffundsapproach.In

each

case,theseratesaccountforonly2observationsinthesample.


for the cost-of-funds approach. Broadly speaking, monetary policy actions thattarget overnight money market rates have increased in relevance as compared tothe role of cost of funds.With respect to country specics, the monetary policy approach indicates a

somewhat larger pass-through size in Portugal and possibly Italy but a smaller onefor Belgium. Speed is signicantly lower for Portugal, Finland, and Spain for the 1-month horizon but higher for Germany, Ireland, and the Netherlands within the 3-to 6-months horizon. For the cost-of-funds approach we nd a signicant larger

H. Sander, S. Kleimeier / Journal of International Money and Finance 23 (2004) 461492478pass-through size only for Spain, while in the speed regressions the dummies forthe various countries reveal a very heterogeneous picture. One conclusion thatmight emerge from this comparative analysis is that the country-specic responsesare more uniform to monetary policy rate changes than to measures of cost offunds.14

With respect to specic markets, the monetary policy approach indicates asignicantly larger pass-through size for short-term corporate loans and a smallersize for current account and saving deposits. These results are in line with theMonti-Klein model of the monopolistic or oligopolistic banking rm, whichpredicts that smaller elasticities imply higher intermediation margins.15 Sincethe case can be made that interest changes may have a larger impact on theshort-run funding choice of borrowers than on the wealth of depositors, the latterssupply of deposits may be comparatively less elastic, hence leading to a smallerand/or slower pass-through for deposits. Regarding speed we do not identifysignicant market-specic changes but this result might be a consequence ofthe denition of the speed variable in the presence of size changes in the samedirection.

3.3. Asymmetries in the euro-zone pass-through

For the majority of the national retail interest rates, the pass-through mechan-isms are most accurately described by asymmetric models. We select them in 51%of all cases for the cost-of-funds approach and in 46% of all cases for the monetarypolicy approach. From the pre- to the post-break period, the share of cases wherethe asymmetric model is selected increases from 42% to 60% and from 29% to 62%,respectively. This strengthens our case to utilize all proposed asymmetric modelsfor empirically determining the optimal pass-through model. Furthermore, thisresult also implies that the majority of interim multipliers are now dependent onthe direction and size of the market interest rate shock.16

14 This may, however, also indicate intrinsic problems with the cost-of-funds approach, which applies

eventually not appropriatelya uniform cost-of-funds variable for each type of retail rate for all coun-

tries.15 For a discussion of various versions of the Monti-Klein model see e.g. Freixas and Rochet (1997).16 Note that the multiplier asymmetries reported in Table 2 are qualitatively dierent from the notion

of asymmetry in the TAR modelling. However, given that most of our interim multipliers are smaller

than unity, we can associate a positive interest rate shock with a negative ECT and thus a below-equilib-

rium state. Consequently, the two types of asymmetry are somewhat comparable.

In Table 2, asymmetries regarding the direction of the shocks are illustrated bydividing the multipliers for the positive 1% shock by the multiplier for the negative1% shock. Deviations from unity indicate asymmetries. For example, a ratio of 1.1implies that a positive shock has a 10% higher impact than a negative shock.17 In

can be made for the monetary policy approach. On average, mortgages reactalmost symmetrically to monetary policy rate shocks. However, our individual

479H. Sander, S. Kleimeier / Journal of International Money and Finance 23 (2004) 461492country multipliers reveal a large number of cases where a relatively faster down-ward adjustment takes place.18 Regarding deposits, on average asymmetries showthe expected pattern in both approaches but diminish over timepossibly due torecent developments in nancial markets. Nevertheless, faster downward adjust-ments of time deposit rates are still present.Another type of asymmetry is reected in the impact of large versus small shocks

on retail interest rates. The last four columns in each Panel of Table 2 givethe relative multiplier for a large +1% versus a small +0.25% interest rate shock.A ratio larger than unity implies that the retail rate reacts more strongly tolarge shocks. While the overall picture in the monetary policy approach isone of symmetry, this is not true for the cost-of-funds approach. In line withthe menu cost argument, mortgages, short-term corporate loans, andto alesser extenttime deposits react faster to large than to small shocks in the cost offunds.

4. Pass-through and retail banking market integration

Pass-through studies are increasingly regarded important for assessing the degreeof nancial integration in the euro-zone retail banking market. Although retailinterest rates have been somewhat converging, this is not necessarily an indicationof an integrated market. In Kleimeier and Sander (2002, 2003) we have shown thateuro-zone retail banking markets are still not integrated when cointegration is con-sidered as an integration indicator and have arrived at a No, No, and Maybeproposition with respect to the integration of mortgage, consumer lending, and

17 It should, however, be recalled that these data are averages and could easily be misinterpreted. If

some countries are faster in upward adjustments and others in downward adjustments, the average

would still be 1. In such cases, however, a high standard deviation can reveal the underlying asymmetry.18 Individual country multipliers can be obtained from Sander and Kleimeier (2004).imperfectly competitive banking markets one would expect a faster upward adjust-ment for loans rates because the degree of asymmetry is negatively correlated withthe elasticity of the respective loan demand. For deposits, one would expect fasterdownward adjustment again depending on the respective elasticities. This theoreti-cal reasoning is largely conrmed by the average ratios for lending and depositrates. This result is strongest for loans in the cost-of-funds approach where mort-gages and short-term corporate loans exhibit the highest degree of asymmetry.Long-term corporate loans, for which a higher demand elasticity can be reasoned,exhibit a more symmetric behavior. Similar observations, though to a lesser degree,

short-term corporate lending markets.19 However, with a single monetary policyand well-integrated wholesale nancial markets, a fast and homogeneous pass-through would create a uniform behavior of euro-zone retail interest rates.Our regression results reported in Table 3 have already shown an increased

speed of the pass-through in the post-break period once we control for countryand market characteristics. In order to test whether or not the pass-through pro-

loans indicating more convergence in this market. For current account and savingsaccount rates the positive coecient indicates higher heterogeneity. The b-conver-

H. Sander, S. Kleimeier / Journal of International Money and Finance 23 (2004) 461492480gence measure is borrowed from the empirical growth literature.21 According tothis concept, growth rates are negatively related to initial levels. Thus, countrieswith initially low multipliers for a given retail rate should see them growing fasterover time, while countries with initially higher multipliers see them growing lessfast or even declining. We nd some evidence for b-convergence for the monetarypolicy approach but not for the cost-of-funds approach. The dummy for short-term corporate loans is again positive. In sum, the evidence for a more homo-geneous pass-through processwith the eventual exception of short-run corporatelendingremains very limited and lends some support to our No-No-Maybehypothesis.

5. Pass-through and banking market structure

5.1. Pass-through determinants

In this section we investigate potential determinants of the pass-through processby making use of the estimated multipliers. We particularly analyze the role ofnancial market structures after controlling for macroeconomic determinant fac-tors. Regarding macroeconomic variables, it has been advocated that money mar-ket rate volatility is positively correlated with bank interest rate margins (Saunders

19 This argument has also been made by the European Commissions Economic and Financial Com-

mittee in a special report (EFC, 2002) and by Cabral et al. (2002).20 Note that the subscripts indicate country j (Austria to Spain) and period t (pre-break, post-break).

Thus, each long-run multiplier is compared to the cross-country average long-run multiplier for its

respective period.21 See Durlauf and Quah (1999).cess has become more homogenous in the euro zone, we regress two types of con-vergence measuresb and r convergenceon post-break, country, and ratedummies. r-convergence measures the variation of the long-run multiplier and isdened as jhj,t hMean,tj/hMean,t.20 Therefore, a negative coecient for this dummyindicates less variation and consequently more convergence. The results are alsoshown in Table 3. The r-convergence regression reveals a slight increase in hetero-geneity in the post-break period for the monetary policy approach. No signicantchanges are detectable for the cost-of-funds approach. Whereas country dummiesdo not play a role, we nd a signicantly negative dummy for short-term corporate

and Schumacher, 2000) and negatively correlated with the pass-through (Cottarelliand Kourelis, 1994; Mojon, 2000; de Bondt et al., 2002). Other relevant macro-economic control variables are structural ination, economic growth, and nancialdevelopment.22 Secondly, we collect four sets of variables describing the nancialstructure of the euro zone: (1) market structure concerning size and concentration,(2) bank protability and bank health, (3) availability of alternative nance, and(4) foreign bank activities. Following the tradition in the literature, we regress thepass-though determinants directly on the multipliers.23 Our analysis concentrateson the multipliers obtained from the monetary policy approach for three reasons:First, retail interest rates collected by the ECB are very heterogeneous acrossEurope, e.g. with respect to the maturity structure of the loans or deposits. Thus,selecting a cost-of-funds rate with a common maturity for all countries is rather

481H. Sander, S. Kleimeier / Journal of International Money and Finance 23 (2004) 461492arbitrary. Second, the monetary policy rate has gained importance relative to cost-of-funds rate. Finally, as monetary policy aects both, the cost-of-funds and theretail rate, the monetary policy approach already covers an important part of thecost-of-funds channel, particular when taking into account forward looking beha-vior by market participants.

5.2. The role of competition

In the spirit of the industrial organization approach to banking, higher concen-tration and lower competition are expected to lead to a faster and larger pass-through. It is theoretically not clear whether a concentration ratio or a Herndahlindex is the most appropriate measure for market concentration (see Berger andHannan, 1989). Therefore, we opt for an internal competition index that avera-ges the ve-rm concentration ratio (CR5) and the Herndahl index.24 To accountfor dierences across markets these indicators are obtained for both, loan anddeposits markets. In a similar way we construct a foreign competition indexwhich is composed of the number of foreign bank branches and subsidiaries andthe share of non-resident intermediated liabilities (loans) or non-resident inter-mediated assets (deposits), respectively.Table 4 presents the results. We nd that more internal competition leads to a

signicant reduction in price rigidities as indicated by the internal competition

22 Financial development is typically measured by a ratio of nancial assets or liabilities to GDP with

the view that the higher the ratio, the higher the degree of nancial system development and the faster

the pass-through. The two most common measures are broad money to GDP, reecting nancial dee-

pening on the asset side, and private credit to GDP, the most comprehensive indicator of nancial

activities of intermediaries. We have employed both measures but report only the results for credit to

GDP as this indicator performs better in the regressions.23 In contrast to the speed regression reported in Table 3, we now focus directly on the multipliers. This

choice is driven by the fact that increased competition could lead to an increase in both, the short- and

long-run multiplier, so that speed may not change at all.24 Each variable is transformed into an index number ranging from 0 to 1 with 1 indicating the highest

expected impact on the pass-through multipliers. For example, a high concentration ratio results in a

low index number, which enters our internal competition variable. Consequently, we expect a positive

coecient for this variable in the panel regression.

coecient in the regressions for the impact multiplier and for the 1-month multi-

pliers. This coecient does, however, become insignicant from three months

onwards and no long-run impact can be established. Surprisingly, the coecient

H. Sander, S. Kleimeier / Journal of International Money and Finance 23 (2004) 461492482for foreign competition is signicant but has the wrong sign in all regressions. This

result as well as the missing long-run eect of internal competition could be caused

by dierences between the loan and deposit markets. Theoretically, the competition

eect should be more pronounced for deposit rates as these are less aected by

informational imperfections than loans, which are more prone to moral hazard and

adverse selection problems. To capture these dierences we introduce deposit slope

dummies for both competition indicators. Our modied regressions now show a

signicant positive impact of more internal and foreign competition in the deposit

market in the short- and in the long run. Furthermore, the results indicate that less

competition leads to faster downward than upward adjustment of deposit rates.

This type of asymmetry is in line with our theoretical priors. Regarding loans, a

signicant impact of internal competition can no longer be established. This result

probably points to the more important role of other market imperfections such as

the lack of contestability, switching cost, informational imperfections, and thus

credit rationing in loan markets.25

With respect to the macroeconomic variables our results conrm the positive

role of reduced volatility in the money market. This is, however, only true for loan

rates and even there the eect diminishes over time as a reduction in volatility does

not aect the long-run pass-through. In deposit markets, lower volatility decreases

both the short- and long-term pass-through. High ination typically leads to a

lower pass-through in deposit markets but not in lending markets, possibly reect-

ing the role of market power. High growth is uniformly found to increase the pass-

through in the long- but not in the short-run. Financial development plays only a

marginally positive role.After controlling for nancial market structure and macroeconomic dierences,

some retail rate dummies still remain statistically signicant. In line with industrial

organization reasoning, short-term corporate loans show a higher pass-through

while current and savings accounts markets exhibits more stickiness. Country char-

acteristics are also persistent. Cecchetti (1999) hypothesizes that cultural and legal

dierences may obstruct the convergence process in the euro zone. To test this

hypothesis we include Cecchettis legal family dummies, in particular a dummy for

the German legal system (used for Austria and Germany), for the Scandinavian

legal system (used for Finland), and for the English legal system (used for Ireland).

The results suggest that in particular in the German legal system the pass-through

is signicantly lower.

25 A remaining puzzle is the statistically signicant and negative coecient for foreign competition in

the loan market. This could possibly indicate that foreign banks prefer to enter markets with low pass-

through.

Table4

Structuraldeterminantsoftheinterestr

atepassthroughforthemonetary

policy

approach:Theroleofcompetition

Independentvariablesa

Dependentvariable

impactmultiplier

1month

multiplier

3monthsmultiplier

+1%

shock

+1%

shock

1%

shock

1%

shock

+1%

shock

+1%

shock

1%

shock

1%

shock

Constant

0.492

0.601

0.435

0.573

0.435

0.573

0.509

0.711

0.430

0.631

2.871

3.587

1.753

2.307

1.753

2.307

1.812

2.603

1.460

2.197

internalcompetition

0.213

0.146

0.300

0.178

0.300

0.178

0.227

0.064

0.224

0.076

2.223

1.407

2.161

1.155

2.161

1.155

1.443

0.378

1.356

0.426

internalcompetition

deposit

0.196

0.303

0.303

0.420

0.397

1.963

2.045

2.045

2.577

2.316

foreigncompetition

0.289

0.488

0.419

0.625

0.419

0.625

0.563

0.885

0.536

0.875

2.102

3.370

2.107

2.910

2.107

2.910

2.503

3.751

2.270

3.523

foreigncompetition

deposit

0.346

0.313

0.313

0.513

0.561

2.838

1.729

1.729

2.583

2.680

money

marketvolatility

0.083

0.094

0.126

0.137

0.126

0.137

0.122

0.140

0.096

0.115

2.262

2.685

2.372

2.630

2.372

2.630

2.042

2.452

1.525

1.910

money

market

volatilityd

eposit

0.162

0.165

0.243

0.240

0.243

0.240

0.267

0.266

0.249

0.250

3.435

3.642

3.563

3.580

3.563

3.580

3.465

3.610

3.070

3.220

ination

0.026

0.016

0.006

0.016

0.006

0.016

0.029

0.044

0.012

0.029

1.119

0.698

0.185

0.466

0.185

0.466

0.739

1.187

0.296

0.740

inationd

eposit

0.063

0.114

0.110

0.162

0.110

0.162

0.122

0.203

0.111

0.197

2.433

3.878

2.913

3.717

2.913

3.717

2.863

4.256

2.496

3.923

growth

0.036

0.046

0.014

0.024

0.014

0.024

0.016

0.000

0.045

0.028

1.381

1.840

0.365

0.646

0.365

0.646

0.367

0.010

1.001

0.648

creditto

GDP

0.051

0.024

0.232

0.193

0.232

0.193

0.198

0.144

0.241

0.189

0.508

0.251

1.584

1.346

1.584

1.346

1.197

0.911

1.391

1.137

N4short-termloans

toenterprises

0.029

0.047

0.138

0.155

0.138

0.155

0.211

0.239

0.187

0.217

0.678

1.145

2.246

2.563

2.246

2.563

3.030

3.581

2.558

3.088

N7currentaccount

deposits

0.192

0.183

0.285

0.276

0.285

0.276

0.375

0.360

0.399

0.384

3.358

3.353

3.445

3.415

3.445

3.415

3.998

4.052

4.059

4.101

(continued

onnextpage)


Table4(continued)


Dependentvariable

impactmultiplier

1month

multiplier

3monthsmultiplier

+1%

shock

+1%

shock

1%

shock

1%

shock

+1%

shock

+1%

shock

1%

shock

1%

shock

N9savings

accounts

0.175

0.300

0.244

0.371

0.244

0.371

0.255

0.454

0.271

0.482

3.136

4.536

3.023

3.783

3.023

3.783

2.785

4.215

2.823

4.249

Germanlegalsystem

0.173

0.211

0.227

0.271

0.227

0.271

0.163

0.228

0.164

0.231

2.367

2.989

2.145

2.583

2.145

2.583

1.357

1.983

1.303

1.908

Scandinavianlegal

system

0.197

0.228

0.291

0.334

0.291

0.334

0.350

0.411

0.396

0.456

2.392

2.864

2.441

2.836

2.441

2.836

2.597

3.178

2.800

3.343

English

legalsystem

0.461

0.611

0.418

0.582

0.418

0.582

0.274

0.524

0.060

0.319

2.352

3.173

1.473

2.039

1.473

2.039

0.853

1.673

0.178

0.968

Adjusted

R2

29.5%

35.9%

37.0%

40.1%

37.0%

40.1%

39.7%

45.9%

38.2%

44.1%

independentvariablesa

dependentvariable

6monthsmultiplier

12monthsmultiplier

longr

unmultiplier

+1%

shock

+1%

shock

1%

shock

1%

shock

+1%

shock

+1%

shock

1%

shock

1%

shock

Constant

0.484

0.696

0.345

0.561

0.350

0.577

0.242

0.493

0.213

0.485

1.688

2.498

1.153

1.936

1.149

1.948

0.748

1.587

0.608

1.442

internalcompetition

0.185

0.006

0.180

0.015

0.154

0.069

0.125

0.078

0.068

0.151

1.155

0.032

1.077

0.082

0.901

0.375

0.692

0.407

0.348

0.723

internalcompetition

deposit

0.469

0.437

0.529

0.523

0.565

2.825

2.525

2.994

2.822

2.814

foreigncompetition

0.605

0.918

0.468

0.824

0.539

0.851

0.352

0.751

0.441

0.876

2.635

3.815

1.951

3.289

2.207

3.325

1.361

2.801

1.570

3.012

foreigncompetition

deposit

0.472

0.581

0.445

0.636

0.696

2.330

2.752

2.063

2.813

2.838

money

marketvolatility

0.094

0.110

0.080

0.100

0.057

0.073

0.060

0.082

0.010

0.014

1.532

1.895

1.254

1.646

0.874

1.178

0.876

1.267

0.128

0.203


money

market

volatilityd

eposit

0.287

0.282

0.283

0.283

0.285

0.276

0.291

0.289

0.231

0.229

3.643

3.750

3.436

3.611

3.401

3.453

3.279

3.452

2.392

2.518

ination

0.051

0.065

0.047

0.064

0.074

0.087

0.080

0.099

0.050

0.071

1.286

1.716

1.143

1.639

1.771

2.176

1.800

2.351

1.030

1.545

inationd

eposit

0.137

0.216

0.134

0.224

0.145

0.224

0.146

0.247

0.129

0.239

3.147

4.430

2.951

4.416

3.149

4.329

2.986

4.552

2.426

4.058

growth

0.052

0.036

0.079

0.061

0.086

0.070

0.093

0.073

0.125

0.104

1.191

0.874

1.728

1.403

1.848

1.587

1.899

1.582

2.350

2.059

creditto

GDP

0.118

0.059

0.181

0.123

0.082

0.016

0.140

0.073

0.149

0.075

0.701

0.368

1.025

0.736

0.455

0.095

0.736

0.404

0.719

0.387

N4shorttermloans

toenterprises

0.269

0.295

0.227

0.258

0.296

0.321

0.261

0.295

0.215

0.252

3.790

4.347

3.060

3.645

3.919

4.444

3.259

3.898

2.472

3.072

N7currentaccount

deposits

0.412

0.399

0.441

0.425

0.430

0.417

0.462

0.444

0.483

0.463

4.312

4.395

4.421

4.505

4.234

4.322

4.281

4.390

4.124

4.223

N9savingsaccounts

0.249

0.442

0.287

0.509

0.230

0.421

0.277

0.524

0.248

0.518

2.675

4.024

2.946

4.443

2.322

3.602

2.630

4.277

2.172

3.898

Germanlegalsystem

0.110

0.176

0.091

0.163

0.056

0.125

0.033

0.115

0.002

0.087

0.899

1.501

0.714

1.332

0.431

0.998

0.242

0.878

0.013

0.612

Scandinavianlegal

system

0.384

0.451

0.392

0.457

0.358

0.431

0.319

0.395

0.282

0.364

2.793

3.414

2.730

3.323

2.449

3.071

2.058

2.686

1.673

2.281

English

legalsystem

0.027

0.276

0.233

0.042

0.271

0.017

0.449

0.139

0.567

0.229

0.082

0.863

0.679

0.126

0.778

0.051

1.215

0.389

1.413

0.592

adjusted

R2

45.0%

50.7%

42.7%

48.9%

45.2%

50.9%

40.9%

48.2%

35.4%

43.4%

Note:Thesamplesize

foreach

regressionis102observationspooledacrossperiods(pre-break,post-break),countries(Austriato

Spain),andrates(N

1to

N10).

aForeach

independentvariabletheestimatedcoe

cientisreported

inthetoprowandthet-statisticisreported

initalics

inthebottom

row.Theesti-

matesarebasedonanOLSregression.


5.3. The role of banking market structure and monetary policy eectiveness

The pass-through analysis can also be employed to investigate the role of nan-cial markets in the eectiveness of monetary policy transmission. Kashyap andStein (1997) and Cecchetti (1999) have argued that a composite measure of mon-etary policy eectiveness, consisting of measures of bank health, number ofbanks, and availability of alternative nance, can explain the high level of monet-ary policy eectiveness in Europe. They argue that small and unhealthy banks aremore heavily aected by shocks and that the transmission to the real economy willbe stronger the less alternative nance is available. We employ a similar eective-

H. Sander, S. Kleimeier / Journal of International Money and Finance 23 (2004) 461492486ness indicator. As we concentrate on the nancial market side of the transmissionprocess only, we do make some adjustments, particularly by adding a measure forbanking market competition. Our eectiveness indicator is thus composed of fourdimensions: Internal competition, alternative nance, bank health, and importanceof small banks.26 The rationale might be as follows: A monetary tightening mightshift the loan supply curve especially of small and unhealthy banks. Whether thisleads to a fast increase in lending rates depends on the elasticity of the loandemand curve and the degree of lending rate stickiness or credit rationing in thecredit market. For any given monetary shock, the less competitive the market andthe less elastic the demand for loans, i.e. the less alternative nance is available, thelarger will be the increase in lending rates.27 Consequently, we expect a positiveimpact of our eectiveness indicator on the pass-through.In Table 5 the results of our regression analyses are reported. Overall, our eec-

tiveness indicator has the expected positive sign and is signicant for all multipliersexcept the impact multiplier. Moreover, the eectiveness indicator becomes moreimportant the longer the time-horizon of the multiplier. As far as foreign compe-tition is concerned, the coecient has the wrong sign but is not statistically signi-cant. However, when introducing a deposit slope dummy, foreign competition hasa positive eect in the deposit markets. With respect to the macroeconomic vari-ables, we can again conrm the positive role of reduced money market rate vola-tility particularly over the rst six months. In a similar manner, both, higherination and less nancial development lead to a slower pass-through, but theeects are only statistically signicant in the rst few months. During this timeeconomic growth seems unimportant. However, in the longer term, higher growth

26 More specically, we dene eectiveness as the equally weighted average of internal competition,

alternative nance, and bank size and health. The denitions for the 3 elements in this indicator are:

Internal competition CR5Herfindahl=2 with both variables based on either loans or deposits,respectively; alternative finance publicly traded firms stock market capitalization intermediatedliabilities=3; bank size and health loan provisions operating cost number of banks=3. Again,for building the index each included variable was transformed into an index number ranging from 0 to 1

with 1 indicating the highest expected impact on the pass-through multipliers.27 This contradicts the view that more alternative nancesuch as high stock market capitalization

will lead to a more competitive banking market and thus faster pass-through. In fact, it appears that our

indicators for the availability of alternative nance are negatively correlated with the pass-through mul-

tipliers thus supporting the loan demand-side view.

Table5

Structuraldeterminants

oftheinterest-rate

pass-through

forthemonetary

policy

approach:Therole

ofbankingmarket

structure

andmonetary

policy

eectiveness


Dependentvariable

impactmultiplier

1month

multiplier

3monthsmultiplier

+1%

shock

+1%

shock

1%

shock

1%

shock

+1%

shock

+1%

shock

Documents

Convergence in Euro-zone Retail Banking- What Interest Rate Pass-through Tells Us About Monetary Policy Transmission, Competition and Integration