Ekonomi Koperasi (Softskill) : Interest Rate Chanel

MONETARY TRANSMISSION IN THE EURO AREA: DOES THE INTEREST RATE CHANNEL EXPLAIN ALL

Nama Kelompok:

Anis Pratiwi D

Hilma Noor F

Shelvy Septiani

Kelas:

2EB28

Universitas Gunadarma

PTA 2015/2016

NBER WORKING PAPER SERIES

MONETARY TRANSMISSION IN THE EURO AREA:

Ignazio Angeloni

Anil K. Kashyap

Benoît Mojon

Daniele Terlizzese

Working Paper 9984

http://www.nber.org/papers/w9984

NATIONAL BUREAU OF ECONOMIC RESEARCH

1050 Massachusetts Avenue

Cambridge, MA 02138

September 2003

An earlier version of this paper was circulated as ECB Working Paper 114, titled “Monetary Transmission

in the Euro Area: Where Do We Stand?". Kashyap thanks the ECB, the Houblon Norman Fund and the

National Science Foundation (through a grant administered by the National Bureau of Economic Research)

for financial support. The views expressed in this paper are not necessarily shared by the institutions to which

the authors are affiliated. We are grateful to all members of the Eurosystem Monetary Transmission Network

for their continuous co-operation and to the participants in the conference “Monetary Policy Transmission

in the Euro Area”, held at the ECB on 18 and 19 December 2001, for many helpful comments. We

particularly thank A. Dieppe, M. Ehrmann and L. Monteforte for helping us assemble some of the data and

S. Sommaggio for editing the text. All errors and shortcomings are our responsibility alone. The views

expressed herein are those of the authors and are not necessarily those of the National Bureau of Economic

Research.

©2003 by Ignazio Angeloni, Anil K. Kashyap, Benoît Mojon, and Daniele Terlizzese.  All rights reserved.

Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided that

full credit, including © notice, is given to the source.

Monetary Transmission in the Euro Area: Does the Interest Rate Channel Explain All?

Ignazio Angeloni, Anil K. Kashyap, Benoît Mojon, and Daniele Terlizzese

NBER Working Paper No. 9984

September 2003

JEL No. E52, E20

ABSTRACT

Drawing on recent Eurosystem research that uses a range of econometric techniques and a number

of new data sets, we propose a comprehensive description of how monetary policy affects the euro

area economy. We focus mainly on three questions: (1) what are the stylized facts concerning the

transmission of monetary policy for the area as a whole and for individual countries? (2) can the

“classic” interest rate channel (IRC) alone, without capital market imperfections, explain these facts?

(3) if not, is the bank lending channel a likely candidate to complete the story? We find plausible

euro-area wide monetary policy responses for prices and output that are similar to those generally

reported for the U.S. However, investment (relative to consumption) seems to play a larger role in

euro area monetary policy transmission than in the U.S. We cannot reject the hypothesis that the IRC

completely characterizes transmission in a few countries, and estimate it to be substantial in almost

all. Where the IRC is not dominant, there is normally some direct evidence supporting the presence

of a bank lending channel (or other financial transmission channel). The cases where financial

effects appear important can be further split according to whether they primarily relate to

consumption or investment.

Ignazio Angeloni

European Central Bank,

Kaiserstrasse 29, D-60311

Frankfurt am Main

Germany

ignazio.angeloni@ecb.int

Anil Kashyap

University of Chicago

Graduate School of Business

1101 E. 58th Street

Chicago, IL 60637

and NBER

anil.kashyap@gsb.uchicago.edu

Benoît Mojon

European Central Bank

Kaisersrasse 29, D-60 311

Frankfurt am Main

Germany

benoit.mojon@ecb.int

Daniele Terlizzese

Banca d'Italia

Servizio Studi

Via Nazionale 91, 00184

Roma

daniele.terlizzese@bancaditalia.it

3

1.

Introduction

In late 1999, the European Central Bank (ECB) launched, together with the National

Central Banks (NCBs) of the countries that had just adopted the euro, a major research initiative

to study the transmission of monetary policy. The objective was to put together, in a reasonably

short time, a comprehensive set of information on how the monetary policy of the newly created

central bank would affect the economy of the so-called “euro area” . In this paper we provide a

1

selective summary of the main findings contained in the 23 papers that were produced during this

project – the full results, including a more complete version of this paper, are forthcoming in

Angeloni, Kashyap, and Mojon (2003).

In doing so, we draw on three types of analyses that were conducted during the project.

The first consists of (six) macroeconomic studies of both the entire euro area economy and the

constituent countries. The second of (seven) studies of firm-level investment behavior. The third

of (ten) studies that analyze bank-level balance sheets and income statements. The goal of this

paper is to bring together disparate pieces of evidence to produce a unified picture. We put the

emphasis on cross-checking these various pieces of evidence, with the belief that when many

indicators point in a similar direction this is unlikely to be due to chance, while a finding that

different indicators yield different conclusions suggests more cautious conclusions should be

drawn. The stress on the need for cross-checking was indeed a central theme of the whole project.

The reader interested in more details regarding the results should consult Angeloni, Kashyap, and

Mojon (2003).

One important caveat to this exercise is that almost all of the empirical work relies on pre-

1999 data, before the introduction of the euro and the structural break associated with the creation

of the ECB. As a result, the Lucas critique no doubt applies forcefully. Still, we believe that the

evidence collected is relevant for several reasons.

First, the European Economic and Monetary Union (EMU) was a gradual process. It

included a prior convergence in policies and economic performance during which agents had time

to prepare and adjust. Much of this adjustment is likely to have occurred before 1999, some to be

ongoing after the new currency is introduced. Data prior to 1999 are likely to contain early

information about the new regime. Second, the use of panel data on firms and banks is intended

1

The euro area is composed of all the countries that have adopted the euro as their currency. Currently, the area

includes 12 of the 15 countries belonging to the European Union (the exceptions being Denmark, Sweden and the

United Kingdom). The Eurosystem includes the ECB and the euro area National Central Banks.

4

inter alia to reduce the need for a long time-series dimension. Moreover, panel data might help

identifying structural parameters that can be viewed as approximately constant during the

transition. Finally, the aforementioned emphasis on cross-checking different sources of evidence

using alternative methodologies and data should be of help in uncovering unreliable results.

To organize the discussion in an interesting fashion (as opposed to offering a laundry list

of results), we ask whether the evidence is consistent with the view that monetary policy

transmission in the euro area can broadly be described as taking place through the classical

interest rate channel (IRC). By IRC we mean the response of aggregate demand components,

GDP and prices to the change in the policy controlled interest rate that would take place if there

were no capital (and insurance) market imperfections . Equivalently, the question we ask is

2

whether such imperfections are crucial to understand the main features of monetary transmission

in the euro area.

We focus on this specific “null hypothesis” for several reasons. First, since the IRC is the

conventional way in which monetary policy is presumed to operate in a large, fairly closed

economy with a developed financial system, it is logical to ask whether it can explain the facts

before looking at alternatives. Second, the commonly accepted — if not always fully accurate —

picture of the euro area financial market is one in which banks play a dominant role. To check

whether this prominence has implications for the transmission mechanism it is methodologically

sound to start with a working assumption that denies any such assumption. Third, much of the

concern that has been voiced about the potentially asymmetric effects of the single monetary

policy appears to be grounded in observed asymmetries in the financial structure of firms or

banks, as well as in their vulnerability to informational problems. These features would be clearly

of little relevance if the IRC were to account for the bulk of the transmission mechanism. Finally,

taking IRC dominance as the null hypothesis provides a disciplined way to look for alternative

explanations. Highlighting the places where the interest rate effects do not appear to be the whole

story helps identify where other channels (such as balance sheet, liquidity constraints, bank credit

supply, and the like) may be important. This in turn can help guide the measurement and

monitoring of the most relevant information for policymakers.

2

The IRC would be the only channel through which monetary policy would affect aggregate spending in a closed

economy where: (1) the central bank was able to influence the term structure of market real interest rates; (2) all

agents were able to borrow and lend at those rates and, (3) all insurance possibilities with respect to credit risks were

available. In this world, no agent’s expenditure would be affected by the availability of liquidity or collateral, nor

would precautionary motives to hold wealth arise.

5

In reaching our assessment about the channels of transmission, we place relatively little

weight on the exchange rate channel. One reason for doing so is that we expect domestic channels

of monetary policy transmission to become more important for the euro area economy (which is

relatively closed to international trade) than was previously true for the member countries. In

addition, estimates of the exchange rate channel for the period prior to the euro should be more

affected by the regime shift than estimates of domestic channels. Finally, the empirical evidence

on the effects of monetary policy on exchange rates is very mixed: the response of exchange rates

to monetary policy is notoriously hard to predict. For all these reasons, we will mainly focus on

domestic channels in our analysis.

The remainder of the paper is organised into six parts. In section 2 we briefly explain the

logic underlying our analysis, focusing in particular on how we intend to bring the various pieces

of empirical evidence to bear on the issue.

Section 3 summarizes, as a background, the overall response to a monetary policy shift of

both prices and quantities in the euro area. This section draws upon evidence from both Vector

Autoregressive (VAR) models and large scale euro area models and includes some comparisons

between the euro area and both euro area countries the U.S.

We then check in section 4 whether the interest sensitive components of GDP – those

components of aggregate demand that are traditionally taken to be most subject to inter-temporal

substitution effects (fixed investment and, in principle, durable consumption) – appear to be able

to account for the bulk of the GDP response.

Section 5 introduces the microeconomic evidence on non-financial firms’ behavior. These

data are used to determine whether a prominent role in the transmission of interest sensitive

components is also matched by a prominent role of the interest rate (or of the cost of capital) in

driving the movements of those components.

Section 6 looks for evidence on banks' lending that would be consistent with non-interest

channels of monetary transmission, to complement the assessment so far reached concerning the

prominence of the IRC.

Section 7 contains our summary of the analysis. We first summarize our understanding of

how policy effects appear to be transmitted in each euro area country, before reviewing the main

findings for the overall euro area.

6

2.

Combining disparate pieces of evidence

Ideally, one would like to test for the dominance of the IRC in a sharp statistical sense,

using a single encompassing model with the appropriate data. Unfortunately, a spate of structural

changes (most recently, the introduction of the euro) and well-known data limitations preclude

this for the euro area. Hence we need to draw upon rather disparate pieces of evidence, trying to

make them comparable in order to discover similarities and to corroborate findings, but eventually

we will need to use our judgement in deciding what they tell us. Our “testing strategy” thus yields

a suggested interpretation of the empirical results; alternative interpretations may be possible, and

we will flag them when appropriate.

The intuition for our approach is relatively simple. We will try to build up the case in favor

of our “null hypothesis” that the IRC is the dominant channel of monetary transmission by

checking whether we find evidence of conditions that should be true, or at least probable, if our

null were true; the more of these conditions we are able to verify, and the more robust is the

evidence in their favor — being confirmed by different approaches and data sets — the more

confident we will be about our “null”.

3

In particular, we will start checking whether interest sensitive spending categories (which

we will take to be represented by investment) account for the bulk of the spending changes that

4

occur after a shift in monetary policy. This is a natural question to ask when assessing the role of

the IRC. Plainly, if the IRC is important we should see important movements of those components

of the expenditure that are traditionally taken to be more interest sensitive. However, it is also a

question whose discriminatory power is rather weak. First, interest sensitive spending might move

responding to channels different from the IRC. Second, even non-durable consumption might

move in reaction to changes of the interest sensitive components. Finally, the dichotomy between

interest sensitive and non-interest sensitive is not as sharp as economic textbooks would suggest.

For all these reasons we see the analysis of the composition of the output response to a monetary

policy shift as a suggestive step in analysing the role that is played by the IRC.

3

This corresponds to one of the main “patterns of plausible inference” expounded by Polya (1954): “..if a certain

circumstance is more credible with a certain conjecture than without it, the proof of that circumstance can only

enhance the credibility of that conjecture”. This is essentially an application of Bayes’ theorem.

4

Theory suggests that interest rates should influence other spending categories besides investment, notably durable

consumption and inventory investment. Unfortunately, the lack of homogeneous data for many euro area countries

prevents a fully satisfactory statistical measurement of these categories. Durable consumption is measured separately

only in Italy, France and Finland. Likewise, in all countries but France and the Netherlands inventory investment is

computed as a residual in the national income and product accounts. Thus, there is little one can do to systematically

study how interest rate changes influence these variables. See Angeloni et al. (2003) for details.

7

We will then move to ask a few more specific questions, using micro data on firms and

banks. The first of these tries to address one of the difficulties in interpreting the composition of

the output response mentioned above. We will ask whether there is evidence that the spending

shifts appear to be due directly to changes in interest rates, not just indirectly because other

determinants of such spending categories are affected by policy. We check whether the

investment demand of firms adjust through the effect of policy on the user cost of capital or by

changing the tightness of firms’ liquidity constraints. We will also ask, in connection with the

possible role of financial factors, whether any subsets of firms where one might think liquidity

problems should be more acute show stronger sensitivities of investment to liquidity. A further

kind of question moves from the idea that if financial frictions are important, it can be expected

that banks play a role in amplifying and transmitting their effect throughout the economy. We will

therefore also ask whether there is evidence of a strong role of bank loan supply shifts in

amplifying the effects of interest rates.

The evidence we compile — first, by asking whether the response of interest sensitive

components of spending account for the bulk of observed changes in aggregate demand; second,

by checking whether such response can be explained, to a sufficiently large extent, by the direct

effect of interest rate changes on investment; third, by checking for the presence of cash-flow or

liquidity effects on investment and for independent evidence regarding banks’ behavior following

changes in monetary policy — allows us to pass a judgment about the nature of the transmission

mechanism in the various countries.

3.

The overall effects of monetary policy in the euro area

As a first step we present, in this section, some results concerning the effects of monetary

policy on GDP and prices in the euro area. We gather four types of measures of these effects,

using either country or euro area aggregate models and two econometric approaches: VARs and

5

large scale models. All estimates are assembled from different chapters of Angeloni, Kashyap,

and Mojon (2003), and these primary sources should be consulted for more detail.

5

The euro area aggregate data that extend back to 1970 are an artificial construct and in principle pose several

problems to the estimation of models for the area. It is however somewhat reassuring that euro area macroeconomic

time series have been found to exhibit business cycle properties quite similar to the one observed in the U.S. (see

Agresti and Mojon, 2003).

8

3.1

VAR estimates of the effects of monetary policy

To start with, a set of basic VAR models for the euro area are estimated by Peersman and

Smets (2003). In their models, an average short-term interest rate for the euro area is used as a

proxy of the area-wide stance of monetary policy over the period. The main results can be seen in

Figure 1, which displays a comparison of the impulse responses from two different models of the

euro area (one without and one with M3 as an endogenous variable) and one model for the U.S.

(that uses an identification scheme quite similar to the one of Christiano, Eichenbaum and Evans,

1999).

In both economies, the interest rate increase reduces output for a few quarters; the

recovery starts within roughly one and a half years. Both in the U.S. and in the euro area, the price

level gradually falls after a monetary tightening. The decline is not significant in either region for

several quarters, but eventually the effect becomes strongly significant and permanent. The

magnitudes of the responses are comparable if one corrects for the differences in the size of the

initial interest rate shock. The similarity between the impulse responses in the euro area and the

U.S. increases if one focuses on the euro area model with M3. The latter model may be more

appropriate for the euro area over our sample period, in light of the prominence assigned to

monetary aggregates by a number of central banks.

The effects of monetary policy on the GDP and prices of the euro area aggregate are also

broadly consistent with the effects of monetary policy shocks identified within each country.

Figure 2 shows results obtained using VAR models as in Mojon and Peersman (2003).

We must stress that, given the monetary policy regime differences between the countries

during the period over which these VARs were estimated, comparisons of the quantitative results

across countries are not warranted. But qualitative comparisons are still valid and several results

follow from a direct comparison of the findings at the country level and for the euro aggregate

VARs. First, a monetary tightening leads to a reduction in output and inflation in virtually all

countries and in the euro area as a whole. While this is not surprising, since this prediction was

undoubtedly one of the things that the model builders considered in settling on their preferred

specifications, it is nevertheless reassuring to observe that this presumption is confirmed by the

data. A second observation, probably more informative, is that the peak response of inflation

comes after the peak response in output for essentially all the estimates.

A third observation is that a close examination of the country level VARs reveals a few

counter-intuitive results. The VAR point estimates for Austria, Greece, Ireland and the

9

Netherlands imply that a monetary tightening is expected to raise either output (see Table 3) or

investment (see Mojon and Peersman, 2003) or both at some point in the first three years.

However, once uncertainty is accounted for, often the estimates are not significantly different

from zero. We therefore put more weight on structural models for these countries in forming our

judgments.

3.2

Evidence from large-scale econometric models

Alternative estimates of the impact of changes in the short-term interest rate can be

calculated using large-scale econometric models that are maintained by central banks for use in

policy analysis.  The criticisms of these models are well-known, but we view them as providing a

useful an independent robustness check on the VARs.

We rely on a host of these models.  The first is the Area-Wide Model (AWM) of Fagan,

Henry and Mestre (2001).  This model treats the entire euro-area as an integrated economy and is

used for policy analysis at the ECB.  A second set of models are the individual country-level

models that have been developed over the years by each the euro area national central banks

(NCBs). Harmonized simulations (imposing common interest rate and exchange rate assumptions)

by van Els et al (2003b) have been used to construct a “bottom up” aggregated set of results for

the euro area.  As a comparison we also report simulations from the FRB/US model, the large

scale model used by the staff of the Board of Governors of the Federal Reserve System.

6

These results are presented in Table 1 for the euro area aggregates and the U.S., and in

upper panel of Table 3 for country level evidence. The tables show the percentage deviations from

the baseline following an exogenous change in the short-term rate (by 100 basis points for eight

quarters, returning to baseline afterwards) and associated paths for the exchange rate and the long-

term interest rate, consistent with arbitrage relationships (see van Els et al. (2003b) for a

discussion).

The qualitative pattern of the response of output and prices already seen in Figure 1 is

broadly confirmed by the structural model simulations, despite the differences in the models and

the profiles of the shocks. Specifically, on both sides of the Atlantic, one observes a relatively

quick and strong output response whereas the price response is muted.

7

6

We are grateful to the Flint Brayton for providing us with these simulations.

7

See also van Els et al. (2003a) for a more thorough comparison of monetary policy shocks simulations with

structural models and VARs.

10

3.3

The exchange rate channel

Before further describing the domestic channels of transmission, we briefly compare the

effects of an exchange rate shock in the euro area and the U.S.  In van Els et al. (2003b), the

decomposition of the effects of monetary policy shocks showed that, up to one year, the exchange

rate channel of monetary policy is the dominant mover of euro area GDP and prices. Table 2

shows the percentage changes relative to the baseline following a permanent exogenous change in

the nominal effective exchange rate (by 5 percent).  For comparison’s sake similar results for U.S.

are also presented.

This “pure” exchange rate shock shows that, again most markedly in the short run, the

euro area GDP and prices are more sensitive to changes in the exchange rate than their U.S.

counterparts. While these responses to the exchange rate obviously need to be factored in the

evaluation of the monetary policy stance, it is not obvious how to do this appropriately nor that, in

practice, the exchange rate channel is large in the transmission of monetary policy. The path of the

exchange rate in the van Els et al. (2003b) simulation and Table 1 imposes an uncovered interest

rate parity condition. However, the strong link between interest rates and exchange rates that is so

assumed, and that are central to establish the importance of the exchange rate channel in the

transmission mechanism, cannot be taken for granted. In fact, there is considerable evidence

showing that the uncovered interest parity assumption is empirically falsified. Moreover, with the

change in the monetary policy regime that accompanied the introduction of the euro the link

between short-term interest rates and exchange rates is particularly uncertain.

3.4

Overall summary of the euro aggregate evidence

We read the evidence in this section as suggesting two broad conclusions. First, VAR and

large scale model analyses for the euro area confirm plausible monetary policy effects on output

and prices. In the VARs, an unexpected increase in the short-term interest rate temporarily

reduces output, with the peak effects occurring after roughly one year. Prices respond more

slowly, hardly moving during the first year and then falling gradually over the next few years.

Again, these VAR properties are similar to those reported for the U.S. The structural models of

the U.S. and the euro area broadly confirm this picture.

Second, there is one interesting difference between the euro area and the U.S.. Changes in

the exchange rates have, in the short-run, larger effects on GDP and prices in the euro area. As we

11

will see in the next section, there is a further interesting difference between the effects of

monetary policy in the U.S. and the euro area: the difference in the composition of the GDP

response to a monetary policy shock. We find the composition of the output response interesting,

as it is suggestive of the relative importance of interest sensitive spending shifts.

4.

The composition of the output response following a monetary shock

To study the composition of the output response we study the ratios between the

derivatives with respect to the interest rate of investment and consumption, on the one hand, and

GDP on the other. To put it in plain English, we measure the total change in investment (or

consumption) following a shift in monetary policy relative to the total change in GDP. We refer to

this measure as the contribution of investment (or consumption) to the overall GDP change . The

8

idea is that these contributions can more easily be compared across models and countries, thus

alleviating the problems of comparability among VARs and structural models. This is because,

since we are comparing how much one variable moves relative to another following a policy shift,

the nature of the shock moving both should be less relevant .

9

While providing a convenient summary of the composition of the output response, the

investment (or consumption) contribution suffer from a few shortcomings. One problem (alluded

to earlier) is that we cannot independently measure all the components of spending that we expect

to be interest sensitive. At this point, all that we can consistently study is private consumption and

total investment. This means that we cannot parse out the effects of changes in durable

consumption expenditure from the rest of consumption. Similarly, the investment responses that

we measure include the movements in government investment, but exclude shifts in inventories.

A second problem is what to do about exchange rate channel effects.  As mentioned

earlier, we expect this channel in the future to be less powerful than before the creation of the

currency area. But we cannot deny the fact that this channel could have been important in some

episodes of the last two decades. To abstract from the role of the exchange rate channel we

8

In the case of VAR simulations, where the interest rate shocks only last one period, the contribution is based on the

cumulated responses of GDP and investment (or consumption). This smoothes out some of the short-run noise. Also,

note that cumulating up to time t the responses to a one-off shock occurring in t-k is approximately equivalent to

observe, at time t, the response to a shock sustained from t-k to t. This is in fact the measure we will adopt when

looking at large scale models, where the shock is sustained for several quarters. In all cases the interpretation is the

same given in the text: the total (dollar or euro) change in investment (or consumption) following a shift in monetary

policy relative to the total (dollar or euro) change in GDP (or domestic demand).

9

One exception to this is when the persistence of the policy shift is significantly altered. However, this is unlikely to

be the case for the kind of shifts that are considered throughout the paper. (See Angeloni et al. (2003) for further

discussion on and comparison of these contribution statistics across countries.)

12

compute the investment contribution (IC) not only comparing the investment response to the full

response of GDP, but also comparing it to the changes in domestic demand (consumption plus

investment).

In sum, the magnitudes of the IC must be interpreted recognizing that the investment

response we record is a conservative estimate of the IRC effects (since durable consumption and

inventory investment are omitted, and government investment is included). This is obviously

more of problem for the comparisons where investment movements are compared to investment

plus consumption, since the misattribution of the durables responses is magnified in this case.

Moving to the results (Table 4) we note that, in the FRB-US model, consumption is

responsible for the bulk of the GDP adjustment following an interest rate shock, both after one

and three years, whereas in the AWM and in the NCB models, the contributions of consumption

and investment are initially balanced, and at three years the latter dominates. In other words, it

seems that in the transmission of monetary policy a relatively greater role is played by

consumption in the U.S., and by investment in the euro area. This is consistent with the strength of

wealth effects on consumer behavior in the U.S., as embodied in the FRB model

10

(see Table 4 for

the composition of the GDP response in VARs,  and, Angeloni et al. (2003) and Angeloni,

Kashyap and Mojon (2003) for deeper analyses of these results, together with robustness checks).

Overall, there is a large degree of correspondence between the estimates of the IC coming

from the VARs (neglecting Greece, Ireland and Austria, which produce the previously mentioned

counter-intuitive results) and those coming from structural models. With the exception of France,

in all the countries where meaningful comparisons can be made (Belgium, Germany, Finland,

France, Italy, and Spain), accounting for over 85% of euro area GDP, the VARs and structural

models yield broadly similar conclusions. However, it should be noted that for a few cases, the

two measures of the contribution are rather different. When looking at structural model, in

Belgium and Germany IC is much larger when defined in terms of “consumption plus

investment”, while the opposite is true in Portugal. When looking at VARs, in Italy and Spain the

contributions are much larger when defined in terms of GDP.

Turning to the specific countries, we judge aggregate demand shifts to be by and large

dominated by investment in Austria, Finland, Ireland, Italy, Luxembourg, the Netherlands, and

Spain. Investment contributions are somewhat smaller in Germany and much smaller in Greece.

Finally, the French, Belgian, and Portuguese evidence is ambiguous. This (obviously subjective)

13

assessment is reached by neglecting suspicious VAR results for Greece, Ireland, Austria and the

Netherlands (see Figure 2) and by checking whether the majority of the reliable measures of the

IC are above 60%. Indeed, in Ireland, Italy, Luxembourg and Spain, the contribution of

investment often accounts for more than 100% of the response of GDP. This is actually because

net trade provides an offsetting contribution due to a sharp decline of imports.

11

For Germany,

model based simulations show a discrepancy between the contribution of investment to GDP

(which is small) and to domestic demand (which is large). VAR based contributions are roughly

consistent and point to a relatively small IC. For France, Belgium and Portugal the available

measures of the IC are evenly split, some suggesting a large value and some a small one. We do

not see any clear basis for deciding which assessment is more reliable.

Putting all this together, the evidence seems supportive of a dominant role of the

investment in the transmission of monetary policy in 7 of the 12 euro area countries. The share of

investment in the GDP response to monetary policy shocks is somewhat smaller in Germany,

although it dominant in the structural models when seen relative to that of consumption. The

situation in France, Belgium and Portugal is unclear. As mentioned in section 2, the key question

becomes whether the strong observed investment responses are attributable to interest rates or are

reflecting transmission via balance sheet, bank lending or other channels. We will address this

question in the next section.

5.

Firm level estimates of the effect of interest rates on investment

The aim of this section is to link the movements in investment to changes in interest rates

– the direct test of the critical link in the IRC. The simultaneity problems that plague

macroeconomic estimates of the interest sensitivity of investment can be partially overcome by

relying instead on firm-level data. In particular, since the identification of the elasticity investment

with respect to the user cost of capital in the micro data can be achieved through cross-firm

variation in taxes and relative prices, the tendency of central banks to react to fluctuations in

aggregate demand may not cause as much trouble for firm-level studies. That said, there are still

plenty of challenges to consistently estimating this effect with micro data; see the thorough

discussion on this point in Chatelain et al. (2003b).

10

See, for example, Reifschneider, Tetlow and Williams (1999).

11

The strong reaction of net trade results in surprisingly large investment contributions not only for very small and

open economies such as Luxembourg and Ireland, but also in the VAR simulation for Spain and Italy (in the latter

case, the structural model simulation confirms for year 3 after the shock also yields a very large IC)

14

These challenges notwithstanding, we do not have any a priori reasons to suspect that

these potential biases differ much across countries. Yet, there are countries where liquidity and

cash-flow effects appear to be very important and countries where they appear hardly to matter. It

is these differences, rather than the exact size of any coefficient estimates, that drive our

assessment. In the next section we will attempt to cross-validate and to further qualify the

conclusions that emerge from this analysis.

5.1

Linking the policy rate to the determinants of investment

In order to identify the full effects of monetary policy on investment, it is necessary to map

the instrument controlled by the monetary authority into the determinants of investment. The

investment equations that we rely upon, described in detail in Chatelain et al. (2003b), are

specified so that firms’ investment rate is determined by current and lagged values of sales

growth, the growth of the user cost of capital, and the ratio of cash-flow to capital. Thus, our

assessment depends on a set of estimates that relate the policy interest rate to sales, the user cost

of capital, and cash-flow.

12

We use aggregate data for Germany, France, Italy, and Spain to estimate these linkages.

We computed the three determinants of investment in a similar way for each country, and then

sought to correlate them with the relevant policy rate. We experimented with simple, single

equation estimates and with VAR estimates obtained appending those variables to the VARs

described in section 4. Fortunately, the three main conclusions presented below are relatively

robust to these different approaches.

First, the user cost was most strongly affected by the policy rate in Spain, and its influence

was fairly similar in the other three countries. Second, sales seemed to be well-described as a near

random walk in each of the countries and there was a weak connection between the policy rate

and sales. We view this conclusion as clearly unsatisfactory, most likely driven by the

simultaneity between sales and interest rates, prominent in the aggregate data. However, while

this might bias towards zero the measure of the overall interest rate elasticity of investment, it is

unlikely to significantly affect our assessment of the relative role of cash-flow, which is our main

focus of interest. Finally, the policy rate was more strongly related to cash-flow in Italy and to a

12

There are several ways to establish the linkages between the policy rate and these variables, see Chatelain et al.

(2003b) and Gaiotti and Generale (2003) two alternative strategies.

15

lesser extent in France than in Spain or Germany. Changing the details of the regression

specifications never changed the general properties of the linkages that we estimated.

For the other countries for which we have micro estimates of the investment equations

(Belgium, Luxembourg, Austria and Finland), due to data shortages we calibrated the coefficients

of the linkage equations to be the average values estimated for the other four countries, but also

cross checked the results with other estimates of the linkages.

The elasticities of user cost, sales and cash-flow to the policy interest rate at years 1, 2 and

3 (shown in Table 5) are used to combine the elasticity of investment to the user cost, to sales and

to cash-flow, estimated on micro data, to get the overall elasticity of investment to the policy rate:

we simulate the micro economic regression equations together with the linkage equations,

assuming that the economies start in a steady state (that is consistent with the sample properties of

the micro data sets) and are hit by an increase in the policy interest rate. We then compute the

implied elasticity of investment.

Our assessment turns on the importance of cash-flow in the estimated investment

elasticities. In particular, we compare the estimate of the overall elasticity when all the linkages

between the policy rate and the determinants of investment are permitted and the estimate

obtained when the cash-flow effects are suppressed.  If the cash-flow effects are important in

explaining the overall elasticity then we argue that the IRC does not provide a full explanation of

the monetary transmission.

Importantly, for cash-flow effects to matter two conditions must hold. First, the cash-flow

coefficients in the investment equation must be substantial. Second, the link between the policy

rate and cash-flow must be significant. If either of these conditions fail then liquidity effects

cannot play an important role in how monetary policy influences investment.

5.2

Parsing the interest rate effects

The elasticities of investment with respect to the policy rate (with and without cash-flow

effects) are shown, for the 8 countries where we have micro-data based estimates, in Table 6.

In Finland and Spain the pair of investment elasticities (overall and with cash-flow effects

blocked out) are virtually identical. Also in Luxembourg, the role of the liquidity variable appears

rather limited. This leads us to conclude that interest rate effects alone are responsible for the

16

influence of the policy rate. Combined with our prior evidence, these countries appear to be cases

where the IRC might be sufficient to explain monetary transmission.

Germany is a case, along with Greece, where the investment contributions to GDP

responses were somewhat lower than for other countries. The data in Table 6 show that in

Germany the cash-flow effects on investment are minimal. Similar conclusions are reached by

von Kalckreuth (2003). While he finds that firms with a low credit score (as assigned by the

Bundesbank) have a relatively high investment sensitivity to cash-flow, his calibrations suggest

that these effects are not quantitatively large.

Overall, we conclude that for the purposes of modeling investment in Germany, the IRC is

a satisfactory characterization, but that some non-interest channels could be operative for other

components of spending. We will look for confirmation of this conjecture in the banking data. In

particular, we will see whether bank lending to households is more importantly affected by

monetary policy than elsewhere.

In Belgium and France the aggregate evidence previously examined was ambiguous

regarding the role of investment. When looking at micro data, we see that in Belgium a sizeable

part of the investment elasticity to the policy rate appears to operate through cash-flow effects. As

the linkage equations were arbitrarily set equal to the average of the four main countries, we

cross-checked this conclusion picking different sets of linkage coefficients. The result appears to

be somewhat sensitive to the selection of the linkage equation, and in particular the role of cash-

flow appears substantially diminished assuming that the links are those estimated for Germany or

Spain (the two countries where the interest rate elasticity of cash-flow is estimated to be the

smallest). While this casts some doubts about the role of financial factors in Belgium, Butzen et

al. (2003) find some evidence of a stronger effect of monetary policy on small firms, a finding

usually interpreted as supportive of financial factors being at work in the transmission mechanism.

Overall, we conclude that financial factors seem to matter for investment, and possibly for other

components of GDP, given our agnostic conclusion about the aggregate role of investment in

explaining aggregate demand movements.

As for France, Table 6 shows that the cash-flow effects are large, accounting for roughly

half of the total investment response. Indeed, Chatelain and Tiomo (2003) find that adding cash-

flow to their equation eliminates the statistical significance of the user cost for their full sample of

firms. These cash-flow effects, however, are not uniformly strong across all firms, with equipment

producers showing the highest sensitivity. Collectively these findings suggest that interest rate

17

induced changes in investment are unlikely to be the whole story in France. However, as in the

case of Belgium, the prior ambiguity means that we cannot say whether any financial effects

should be expected only for investment or for both investment and consumption.

Finally, Austria and Italy are cases where the IRC dominance was broadly consistent with

the aggregate evidence on the composition of the output response but appears doubtful in light of

the data in Table 6. In both these countries the cash-flow effects appear to be relatively large,

possibly more important than the interest rate effects. As with Belgium, the results for Austria

were cross-checked using alternative choices for the linkage equations. The relative importance of

the liquidity measure remains sizeable even when picking the links estimated for Italy or Spain

(where the interest elasticity of user cost is highest). In addition, Valderrama (2003) finds for

Austria stronger effects of monetary policy on small and young firms, and smaller effects for

firms that have a tighter credit relationship with a bank (hausbank). Both findings seem supportive

of an important role of the credit channel in shaping the transmission of monetary policy. A

similar supportive evidence for the role of financial factors is reported for Italy by Gaiotti and

Generale (2003): they find that the effect of cash-flow on investment is stronger for small firms

and for firms with a larger share of intangible assets. Overall, we provisionally conclude that

financial factors, by influencing investment, appear to play, both in Italy and in Austria, a

noticeable role in monetary transmission, and we will look for more evidence supporting this

conclusion in the next section.

6.

The role of banks in the transmission of monetary policy

To complete our assessment, we now examine the evidence on banks. While we have

identified in a number of countries a role for financial factors, this does not necessarily imply a

role for banks. Balance sheet or broad credit channel effects could be quite important. These

balance sheet effects could operate in addition to any effects attributable to banks, or even in

absence of effects generated by bank loan supply. The evidence that follows therefore can be

viewed as a cross check to see whether the provisional assignments made in the last section

should be refined to include a role for banks in the transmission mechanism.

In considering the refinement, we draw on three types of evidence. Whenever possible we

rely on the findings of the country-specific, individual bank level analyses, abridged versions of

which appear in Angeloni, Kashyap, and Mojon (2003). To succinctly summarise these findings,

Table 7 reports a verbal description of their results. In cases where the country level data are

18

missing or inconclusive we turn to two other sources. One piece, reported in Table 8, comes from

expanding the VAR models estimated by Mojon and Peersman (2001) to include data on

monetary aggregates, retail bank interest rates and bank loans. In some cases we also examine

institutional features of the euro area national banking systems that might affect the strength of the

bank lending channel. In particular, as emphasized by Ehrmann et al. (2003), four aspects could

matter for monetary transmission: the importance of state influences in determining credit flows,

the prevalence of relationship lending, the size of deposit insurance guarantees, and the extent of

bank networks. We would expect that each of these features would reduce the sensitivity of bank

lending to changes in monetary policy.

We begin by considering the countries for which the hypothesis of IRC dominance seemed

to receive most support: Finland, Luxembourg and Spain. The two available country case studies

on the bank lending channel (Topi and Vilmunen (2003) and Hernando and Martínez-Pagés

(2003)) do not find clear evidence of loan supply effects on the monetary transmission. In the

Spanish case this finding is reinforced by an interesting observation about the impact of the

phenomenal growth of mutual funds in Spain. The deposit outflows that accompanied the growth

were uneven across banks, but the lending changes that followed did not track the deposit shifts.

This is unlikely to be due to any loan demand differences and is instead most naturally interpreted

as showing that loan supply and deposits in Spain are not tightly linked. However, the structure of

the Spanish banking system shows none of the institutional factors that might insulate lending

decisions from monetary policy. In Finland, Topi and Vilmunen find that the main bank

characteristics that might be expected to influence loan supply (size, capitalization and liquidity)

do not lead to any significant differences. A limited role of bank supply in the transmission

appears broadly consistent with the presence, in Finland, of an important network for the many

co-operative banks.

In Italy and Austria the prior evidence identified financial factors as playing a role in

explaining firms’ responses to monetary policy. In Italy, the bank level analysis presented in

Gambacorta (2003) indicates that monetary policy does alter loan supply. Specifically, he finds

that the amount of liquidity on individual banks’ balance sheets significantly influences the degree

to which they change loans after a monetary shock: the lower the level of liquidity the stronger the

loan supply response. Overall, these findings are consistent with previous results and the VAR

evidence we examined. Indeed, Gambacorta (2001; Table 2) reports that there is near unanimity in

the past literature that a broad credit channel exits.

19

The Austrian evidence is more ambiguous. Kaufmann (2003) only finds cross bank

differences in lending responses during recessions, when lending from banks with more liquid

assets is significantly less affected than that from otherwise comparable banks. But the strength of

this finding depends on how the recession periods are selected and also sometimes is accompanied

by other anomalous findings (e.g. higher policy rates leading to higher lending). The VAR

evidence does suggest that M1 and loans both contract after a monetary tightening. One possible

explanation for the relatively weaker role of Austrian banks in the transmission process could be

the importance of networks and relationship lending between firms and their banks.

Germany is the only country where our preliminary classification more clearly suggested

that financial factors, if they were to matter, would operate through consumption, but not

investment. Logically this would suggest investigating whether loan-supply effects (or other

financial factors) are particularly important for households. We have no direct evidence on this

question, but there are some suggestive pieces of information. First, the VARs shows that

household borrowing falls much more quickly after a monetary tightening than does business

borrowing. Actually, business borrowing is estimated to rise in the first year. Second, the bank

level analysis of Worms (2003) shows significant loan supply effects that are related to the

liquidity position of the banks. A relevant aspect of this paper is that it uses data on the customer

mix of each bank to build a variable that reflects the average income of each bank’s borrowers. As

this income proxy should reliably control for loan demand, we expect lending changes to

genuinely reflect supply shifts. At any rate, similar results are obtained when the standard controls

for demand conditions employed by Ehrmann et al. (2003) are used. The general picture also

seems to fit with certain structural features of Germany’s banking system. On the one hand, the

level of concentration is relatively low and banks are not particularly well capitalised. On the

other hand, banks tend to belong to networks and the hausbank lending relations are often very

strong. The latter feature could explain why corporate borrowers are insulated from credit

restrictions, while households remain exposed.

In our preliminary classification Belgium was the country where it was unclear as to how

much weight to assign to investment movements in accounting for GDP movements, but it was

clear that cash-flow seemed important for observed investment responses. This suggests that

liquidity effects would be expected to matter for business investment and possibly household

expenditure. The VAR evidence, as in Germany, points to household borrowing responding much

more quickly to a policy shift than business borrowing. As estimates based on bank level data are

20

not available, we are not able to draw any strong conclusions for Belgium . It is worth stressing

13

however that de Bondt (2000) suggests that bank loan supply may be affected by monetary policy.

Based on the macro and the micro evidence on firms in France, financial factors appear to

play an important role in driving investment responses to monetary policy. The French evidence

on micro-bank data suggests that loan supply does shift when monetary policy changes. In

particular, Loupias et al. (2003) show that less liquid French banks are more responsive to

monetary policy. Previous studies using microeconomic data also find a role for banks – see

Rosenwald (1998). The evidence does not speak to the question of whether the relative

importance of these effects for business as opposed to consumer lending (both of which might be

expected to matter based on the evidence in sections 4 and 5).

Finally, we are left with the four countries (Ireland, Greece, the Netherlands, and Portugal)

where, absent firm-level evidence, our “testing” of the IRC dominance hypothesis was only

partially, if at all, implementable.

Among these, Greece stands out because the small share of GDP movements accounted by

investment already points at an important role for consumption adjustments in the transmission

mechanism. This suggests looking for evidence of financial factors having a role for consumption

(and possibly for investment). Brissimis et al. (2003) find that both smaller banks and banks with

lower levels of liquidity are more responsive to monetary policy. Smaller banks, with less liquid

assets are estimated to be especially sensitive to policy changes. However, we do not have enough

information to tell whether these loan supply shifts are more relevant  fo r households or

businesses. Other evidence seems supportive of a non-negligible role of banks.

14

In Portugal, the aggregate evidence was ambiguous about the role of investment. It appears

that loan supply is affected by monetary policy changes. Farinha and Robalo (2003) conclude that

bank capital plays an important role in shaping banks responses to monetary policy, with less

capitalised banks being more sensitive. This is consistent with the institutional characteristics of

13

The structural information is also ambiguous. On the one hand, the industry is concentrated (dominated by 12

banks) and the banks hold a relatively low percentage of assets in the form loans. On the other, the lack of

relationship lending, government guarantees, deposit insurance and bank networks means that there are not some of

the mechanisms that would cushion bank lending from monetary policy.

14

For instance, the Greek banking system is characterised by banks holding relatively low levels of capital and liquid

assets. Moreover neither networks nor relationship lending are believed to be significant. Finally, Brissimis and

Kastrissianakis (1997) conclude that the bank lending channel appears to exist in Greece, although it may it have

weakened in the 1990s.

21

the banking sector in Portugal, whereby networks are unimportant and relationship lending is not

typical, thus supporting the operation of a lending channel.

In the other two countries the aggregate evidence was broadly consistent with the

hypothesis of IRC dominance (in the weak sense that investment appeared to be an important

mover of a g g r egate demand responses to a monetary policy shift). In Ireland, we lack a bank level

analysis and the indicators that are available provide ambiguous readings so that we draw no firm

conclusions.

15

For the Netherlands, De Haan (2003) finds that unsecured bank lending is responsive to

monetary policy. These effects are larger for small banks, banks with low liquidity and banks with

low capital – although the interactions between these characteristics do not appear to be important.

In contrast, secured lending seems to be unaffected by policy changes. Finally, household lending

is little affected by policy changes, and this reinforces the view that monetary policy operates

primarily by affecting investment. This relatively clear-cut evidence however contrasts with priors

based on the fact that in the Dutch banking system liquidity, concentration and capital levels are

relatively high and relationship lending is prevalent.

7.

Summary of the evidence for individual countries

The logic of our “testing strategy” on how the macro-evidence, the firm level investment

demand evidence and the bank level credit supply evidence affect the likelihood of IRC

dominance leads to the two-way classification presented in Table 9.

Austria

Austria is a country where interest sensitive components of GDP seem to account for a

large part of the movements of GDP in the wake of a monetary shock. But, firm-panel results

show that there is a non-negligible role for liquidity variables in determining investment. Looking

at the bank side, the results from the panel estimates suggest that the lending channel of monetary

policy is not likely to be strong. This may be due to the strong bank networks and bank-firm

relationships. Hence, any monetary policy effects beyond those going through the IRC should

work largely through other channels (e.g., firm balance sheets).

15

Bank networks are prominent, with most banks belonging to one, and there is a lot of relationship lending. But, the

largest banks control a relatively small share of the total market, the share of loans in banks assets is very high, and

banks do not seem to carry high levels of liquidity or capital.

22

Belgium

The evidence appears to point against IRC dominance in Belgium. The VAR and the

structural models provide conflicting indications about the role of investment in accounting for

aggregate demand movements (small when considering GDP, large when considering domestic

demand). The evidence on investment is that financial factors probably matter for investment. On

the bank side, previous evidence concluded that the role of bank loan supply shifts seem to play a

role in the transmission of monetary policy in Belgium. We have no evidence that can be used to

challenge that finding.

Finland

In Finland, the IRC seems to offer a satisfactory account of monetary transmission. The

VAR and national econometric model suggest that this would be the case and we find the micro-

economic evidence on investment to be consistent with this prediction as well. The prior banking

evidence available on Finland was very limited. Our bank panel estimates (from the post-banking

crisis period) signal that loan supply does not appear to be very responsive to monetary policy. In

any case, considering that the IRC seems to be dominant, loan supply behaviour of banks should

not play an important role

16

in the overall mechanism.

France

As in Belgium, we had difficulty getting for France a consistent assessment of the role of

interest sensitive spending components. Traditionally, it has been difficult to identify cost of

capital effects on investment in France. Our evidence confirms that, in keeping with the findings

of past studies, the cost of capital does not have a strong effect on investment, while financial

factors — as captured by a cash-flow variable — appear important. On the bank side, the earlier

literature placed France, alongside with Germany and Italy, among the candidates for a strong

bank lending channel. The evidence we have does not fully confirm this. Bank liabilities do not

appear to be strongly affected by monetary policy; however, loan rates react strongly to monetary

policy. The panel evidence on banks suggests that loans supply is responsive to monetary policy.

Germany

The German evidence is also complicated. On the one hand, investment spending plays a

smaller than average role in accounting for GDP movements in the wake of a monetary policy

16

Our main caveat surrounding this conclusion is that much of the evidence comes in the post banking crisis

environment. Some prior studies had found that liquidity variables might matter for investment. We leave open the

possibility that this may be the case again now that the adjustment to the crisis is over.

23

shift. On the other hand, the IRC seems to be the dominant, indeed almost the only relevant

channel in explaining monetary policy effects on investment. At the same time, the monetary

shifts appear to alter loan supply. One possible reconciliation is that loan supply effects influence

consumption, but cannot be independently checked using the evidence in this book.

Greece

The large changes in the Greek economy over the last decade make it difficult for us to fit

a stable VAR. But, based on the structural model of the Greek economy, it appears that

consumption is an important component of the adjustment to a monetary shock. Earlier authors

have pointed at Greece as a candidate for significant loan supply effects, and the new econometric

evidence we quoted points in this direction.

Ireland

Ireland is the country where our evidence on monetary transmission is scarcest. The only

available one comes from the structural model of the economy and it suggests that the IRC could

be quite important. But we lack any findings about firms or banks that allow us to test this

conjecture.

Italy

Interest sensitive spending in Italy seems to largely account for output movements in the

wake of a monetary policy shift. The investment response, however, shows clear signs of being

affected by financial factors. This seems to reject the IRC dominance. Moreover, bank lending

responds to policy shifts. This picture is confirmed by a host of studies. Overall Italy is a country

where the case for the presence of lending channel seems strong.

Luxembourg

Our evidence for Luxembourg is somewhat incomplete. We do not have the data needed to

fit a VAR, so relying solely on the structural econometric model we conclude that interest

sensitivity spending movements dominate the monetary induced changes in GDP. The firm level

evidence moreover suggests that investment does not appreciably depend on firms’ liquidity

holdings. The (limited) previously available evidence for Luxembourg suggests that bank loan

supply is not likely to play a major role in monetary transmission. Our informal evidence supports

this, though we lack any econometric evidence on bank behaviour in Luxembourg.

24

Netherlands

We also have incomplete information on the Netherlands. The VAR and econometric

model each suggest that the investment plays an important role in accounting for output responses

to monetary policy, but we lack the firm level analysis to verify whether this corresponds to a

dominant IRC. Past evidence is ambiguous as to whether investment responses can be fully

explained by interest rate effects. There is clear evidence that bank loan supply changes following

changes in monetary policy. The extension of household credit, however, does not appear

affected. Thus, the outstanding question is whether the estimated change in unsecured business

credit is relevant for Dutch corporate investment.

Portugal

As with Greece, Ireland and the Netherlands, the assessment of Portugal is impaired by

lack of data. The structural changes in the economy limit our ability to estimate a VAR, while the

evidence based on the econometric model was ambiguous about the role of investment. We lack

the firm level evidence needed to further sharpen this assessment. However, it appears that bank

loan supply does change following a shift in monetary policy. As in the Netherlands, we cannot

determine whether this is material for the transmission.

Spain

Spain is the case where the evidence most consistently points towards a pure IRC

explanation for monetary transmission. Following a monetary policy shift, investment movements

are substantial, yet they do not appear to be dependent on financial factors. Loan supply also

appears to be disconnected from monetary policy; the evidence on how banks also shielded their

lending after regulatory induced deposit outflows reinforces this presumption. This all fits

together and suggests that financial factors do not play an important role in the Spanish monetary

transmission.

8.

Conclusions

Taken together, the findings from this project paint a rich, composite and, to some extent,

surprising picture of the monetary transmission for the euro area as a whole. This picture can

hopefully also serve as a point of departure when sufficient information to document, and

measure, any changes in the transmission process resulting from the introduction of the single

currency becomes available.

25

Starting with the unsurprising aspects, the VAR analysis suggests that an unexpected

increase in the short-term interest rate temporarily reduces output, with the peak effects occurring

after roughly one year. Prices respond more slowly, with inflation hardly moving during the first

year and then falling gradually over the next few years. Structural econometric models, though not

strictly comparable, provide a picture with similar qualitative features. Despite the somewhat

artificial nature of the synthetic data for the area as a whole, these findings are theoretically

sensible and broadly consistent with a large body of empirical literature analyzing either

individual countries of the euro area or the other large currency area in the world, namely the U.S.

Moreover, the delayed response of prices relative to that of output suggests that studying the

transmission of policy to spending and output is a logical step, even if the aim of monetary policy

is defined primarily or exclusively in terms of prices.

A further aspect of the assessment based on aggregate data at the area level is that both the

VARs and the structural models highlight the importance of investment in driving output changes

in the wake of a monetary policy tightening. This feature distinguishes the transmission

mechanism in the euro area from that in the U.S., where much of the output adjustment appears to

be due to changes in consumption – a topic we explore further in Angeloni et al. (2003).

Moving to the question posed in our title, the answer seems to be no. Neither the IRC nor a

broadly construed financial channel emerges as clearly and exclusively dominant. Although

somewhat obvious, this conclusion contradicts both the presumption of an IRC dominance based

on the just mentioned, quantitatively large role of interest sensitive spending in explaining output

movements, and the presumption of a strong and widespread lending channel based on the

overarching role of banks as providers of finance in the euro area. While making a precise

quantitative estimate of the contributions of each channel is not yet possible at this stage, we can

nonetheless further qualify the overall picture in several ways.

First, while not dominant on the whole, the IRC is still a prominent channel in the

transmission. For the euro area as a whole, the interest sensitive demand components account for

the bulk of the change in GDP after a monetary policy shift. Moreover, in a group of countries,

accounting for about 15% of the euro area GDP, the IRC is unambiguously the dominant channel.

In all other countries for which we have the evidence (covering, together with the first group,

about 90% of the euro area GDP) interest rate effects are always a sizeable, and sometimes the

virtually unique, source of investment movements. It is interesting that there generally seems to be

a significant effect of the user cost of capital on investment. This finding, based on a careful

26

analysis of micro level data, is somewhat in contrast with the often more ambiguous results

coming from aggregate data.

Second, financial factors influence the transmission of monetary policy in several ways.

Importantly, the cases where the IRC dominance do not find much support do not point to a

single, prevalent alternative. In some of the countries it looks like the role of banks in supplying

business credit to finance investment may be important. But this does not appear to hold

everywhere, as there are cases in which if loan supply matters it is not likely to be true for

investment. Thus, in terms of monitoring bank lending it is probably necessary to track both

household and business lending. Moreover, there are also cases in which financial factors are

important but banks are not likely to be an important ingredient in the picture.

Thirdly, the overall role of banks in the transmission mechanism is somewhat different,

and perhaps smaller, than what might have been expected based on prior work. There are

countries where bank lending appears irrelevant for transmission. In some, we suspect that

government guarantees to support banks, the propensity of banks to operate in networks, and

strong borrower-lender relationships may mitigate the strength of any loan supply effects. Taken

together this means that even though the banks dominate the supply of credit in all euro area

countries, they do not appear to be uniformly important.

Lastly, in assessing the role that the banks do play in the transmission, the relevant

characteristics that appear to affect the potency of the lending channel are not always those that

we (and probably others too, based on our reading of the past literature) would have guessed.

Bank size and bank capital seems not to play much of role in shaping loan supply responses to

monetary policy. We find the institutional reasons discussed by Ehrmann et al. (2003), and noted

in section 6, to be a plausible explanation for this result. But this means that the vast heterogeneity

in terms of size both across and within countries is probably not very important. In contrast, bank

liquidity positions seem to be important in virtually all the countries where loan supply effects

appear to be present. But there are other potential supply effects that remain to be isolated.

All of our analysis has been based on the analysis of data from before the launch of the

euro. Capital markets, bank market structure, and business financing already have changed

substantially over the last four years. One obvious caveat to our analysis is that we cannot say

whether this has changed the operation of the monetary policy transmission channels, as a result

e.g

. o

f increased monetary and financial integration. Angeloni and Ehrmann (2003) explore these

issues by looking at post 1999 data.

27

Whether or not our assessments are confirmed in future work, we believe that they still

provide guidance about useful next steps in studying monetary transmission in the euro area. For

instance, it would be very useful to fill in the missing pieces of evidence that would be needed to

complete and make more robust our analysis for all the countries. In particular, being able to

gauge the macroeconomic importance of the lending supply effects in those countries where our

evidence confirms their presence would be very important. Also, for several of the countries the

implication of the analysis is that consumption adjustment is surprisingly important for

transmission. Compiling direct evidence on this is another obvious next step.

28

Bibliography

Agresti A. and B. Mojon (2003), Some stylised facts on the euro area business cycle, Chapter 1, in

I. Angeloni, A. Kashyap and B. Mojon (eds.) Monetary Transmission in the Euro Area,

Cambridge University Press.

Angeloni, I. and M. Ehrmann (2003), Monetary transmission in the euro area: early evidence.

Forthcoming in Economic Policy 37 (October).

Angeloni, I., A. Kashyap, B. Mojon and D. Terlizzese (2003), The output-composition puzzle: a

difference in the monetary transmission mechanism in the euro area and the U.S.

Forthcoming: Journal of Money, Credit and Banking.

Angeloni, I., A. Kashyap and B. Mojon (eds.) (2003), Monetary Transmission in the Euro Area,

Cambridge University Press.

Brissimis S. and Kastrissianakis (1997), Is there a credit channel in the Greek economy? , Bank of

Greece Economic Bulletin, 10.

Brissimis S., N. Kamberoglou and G. Simigiannis (2003), Is there a bank-lending channel of

monetary policy in Greece? Evidence from bank-level data, Chapter 18, in I. Angeloni, A.

Kashyap and B. Mojon (eds.) Monetary Transmission in the Euro Area, Cambridge University

Press.

Butzen P., C. Fuss and P. Vermeulen (2003), Business investment and monetary transmission in

Belgium, Chapter 8, in I. Angeloni, A. Kashyap and B. Mojon (eds.) Monetary Transmission

in the Euro Area, Cambridge University Press.

Chatelain J.-B. and A. Tiomo (2003), Monetary policy and corporate investment in France,

Chapter 10, in I. Angeloni, A. Kashyap and B. Mojon (eds.) Monetary Transmission in the

Euro Area, Cambridge University Press.

Chatelain J.- B., M. Ehrmann, A. Generale, J. Martínez-Pagés, P. Vermeulen and A. Worms

(2003a), Monetary policy transmission in the euro area: new evidence from micro data on

firms and banks, Journal of the European Economic Association 1 (2-3) April-May.

Chatelain J.-B., A. Generale, I. Hernando, U. von Kalckreuth and P. Vermeulen (2003b), Firm

investment and monetary policy transmission in the euro area, Chapter 7, in I. Angeloni, A.

Kashyap and B. Mojon (eds.) Monetary Transmission in the Euro Area, Cambridge University

Press.

Christiano, L., M. Eichenbaum and C. Evans (1999), Monetary Shocks: what have we learned and

to what end?, Chapter 2, in John Taylor and Michael Woodford (eds.) The Handbook of

Macroeconomics, Elsevier Science Publication.

De Haan L. (2003), The impact of monetary policy on bank-lending in the Netherlands, Chapter

20, in I. Angeloni, A. Kashyap and B. Mojon (eds.) Monetary Transmission in the Euro Area,

Cambridge University Press.

De Bondt G. (2000), Financial Structure and the transmission mechanism in Europe: a cross-

country study, Edward Elgar, Cheltenham, UK.

29

Ehrmann M., L. Gambacorta, J. Martínez-Pagés, P. Sevestre, A. Worms (2003), Financial systems

and the role of banks in monetary policy transmission in the euro area, Chapter 14, in I.

Angeloni, A. Kashyap and B. Mojon (eds.) Monetary Transmission in the Euro Area,

Cambridge University Press.

Els P. van, A. Locarno, B. Mojon and J. Morgan (2003a), New macroeconomic evidence on

monetary policy transmission in the euro area, Journal of the European Economic

Association, April-May, Vol. 1 (2-3).

Els P. van, A. Locarno, J. Morgan and J-P. Villetelle (2003b), The effects of monetary policy in

the euro area: evidence from structural macroeconomic models, Chapter 5, in I. Angeloni, A.

Kashyap and B. Mojon (eds.) Monetary Transmission in the Euro Area, Cambridge University

Press.

Fagan G., J. Henry and R. Mestre (2001) An area-wide model (AWM) for the euro area, ECB

Working Paper 42.

Farinha L. and C. Robalo (2003), The bank-lending channel of monetary policy: identification and

estimation using Portuguese micro bank data, Chapter 22, in I. Angeloni, A. Kashyap and B.

Mojon (eds.) Monetary Transmission in the Euro Area, Cambridge University Press.

Gaiotti E. and A. Generale (2003), Monetary policy and firms' investment in Italy, Chapter 11, in

I. Angeloni, A. Kashyap and B. Mojon (eds.) Monetary Transmission in the Euro Area,

Cambridge University Press.

Gambacorta L. (2003), The Italian banking system and monetary policy transmission: evidence

from bank-level data, Chapter 19, in I. Angeloni, A. Kashyap and B. Mojon (eds.) Monetary

Transmission in the Euro Area, Cambridge University Press.

Gambacorta L. (2001), bank specific characteristics and monetary policy transmission: the case of

Italy, ECB Working Paper No. 103.

Hernando I. and J. Martínez-Pagés (2003), Is there a bank-lending channel of monetary policy in

Spain?, Chapter 16, in I. Angeloni, A. Kashyap and B. Mojon (eds.) Monetary Transmission

in the Euro Area, Cambridge University Press.

Kalckreuth U.

von (2003), Investment and monetary

transmission in Germany:

a

microeconometric investigation, Chapter 9, in I. Angeloni, A. Kashyap and B. Mojon (eds.)

Monetary Transmission in the Euro Area, Cambridge University Press.

Kaufmann S. (2003), The cross-sectional and the time dimension of the bank-lending channel: the

Austrian case, Chapter 21, in I. Angeloni, A. Kashyap and B. Mojon (eds.) Monetary

Transmission in the Euro Area, Cambridge University Press.

Loupias C., F. Savignac and P. Sevestre (2003), Is there a bank-lending channel in France?

Evidence from bank panel data, Chapter 17, in I. Angeloni, A. Kashyap and B. Mojon (eds.)

Monetary Transmission in the Euro Area, Cambridge University Press.

Lünnemann P. and T. Mathä (2001), Monetary transmission: empirical evidence from

Luxembourg firm-level data, ECB Working Paper No. 111.

McAdam P. and J. Morgan (2003), Analysing monetary policy transmission at the euro area level

using structural macroeconomic models, Chapter 4, in I. Angeloni, A. Kashyap and B. Mojon

(eds.) Monetary Transmission in the Euro Area, Cambridge University Press.

Mojon B. and G. Peersman (2003), A VAR description of the effects of monetary policy in the

individual countries of the euro area, Chapter 3, in I. Angeloni, A. Kashyap and B. Mojon

(eds.) Monetary Transmission in the Euro Area, Cambridge University Press, 2003.

30

Mojon B. and G. Peersman (2001), A VAR description of the effects of monetary policy in the

individual countries of the euro area, ECB Working Paper 92.

Peersman G. and F. Smets (2003), The monetary transmission mechanism in the euro area:

evidence from VAR analysis, Chapter 2, in I. Angeloni, A. Kashyap and B. Mojon (eds.)

Monetary Transmission in the Euro Area, Cambridge University Press.

Reifschneider, D., R. Tetlow and J. Williams (1999), Aggregate disturbances, monetary policy

and the macroeconomy: the FRB/US perspective, Federal Reserve Bulletin (January).

Topi J. and J. Vilmunen (2003), Transmission of monetary policy shocks in Finland: evidence

from bank-level data on loans, Chapter 23, in I. Angeloni, A. Kashyap and B. Mojon (eds.)

Monetary Transmission in the Euro Area, Cambridge University Press.

Valderrama M-T. (2003), The role of trade credit and bank-lending relationships in the

transmission mechanism in Austria, Chapter 13, in I. Angeloni, A. Kashyap and B. Mojon

(eds.) Monetary Transmission in the Euro Area, Cambridge University Press.

Valderrama M-T. (2001), Credit channel and investment behavior in Austria: a micro-econometric

approach, ECB Working Paper No. 108.

Worms A. (2003), The reaction of bank-lending to monetary policy measures in Germany,

Chapter 15, in I. Angeloni, A. Kashyap and B. Mojon (eds.) Monetary Transmission in the

Euro Area, Cambridge University Press.

NB: Tugas lengkap ada di softcopy