EUROPEAN UNIVERSITY INSTITUTE
DEPARTMENT OF ECONOMICS
EUI Working Paper ECO No. 98/33
An EMU with Di®erent
Transmission Mechanisms?
Giorgia Giovannetti and Ramon Marimon
BADIA FIESOLANA, SAN DOMENICO (FI)
All rights reserved.
No part of this paper may be reproduced in any form
without permission of the author.
c
°1998
G. Giovannetti and R. Marimon
Printed in Italy in December 1998
European University Institute
Badia Fiesolana
I-50016 San Domenico (FI)
Italy
An EMU with di®erent transmission
mechanisms?¤
Giorgia Giovannetti
Universitµa di Firenze and European University Institute
Ramon Marimon
European University Institute,
Universitat Pompeu Fabra, CEPR and NBER
July 1998
Abstract
We develop and compute a dynamic equilibrium model
where economies di®er on the relative e±ciency of ¯nancial
intermediaries and, therefore on households portfolios and
currency holdings. Our model economies have some of the
features of the di®erent ¯nancial structures in countries of
the European Union and respond to monetary shocks in a
way similar to the observed responses, which we also estimate. It follows that, if di®erences on the relative e±ciency
of ¯nancial intermediaries persist in a monetary union, con°icts of interests in the pursuit of a common monetary policy
can arise.
¤ We would like to thank Elena Gennari and Christian Upper for research assistance
and Pedro Teles, as well as participants in seminars, at UPF, EUI, UCLisboa, OECD,
SED and SET, for comments. This research project has taken place within the 1996-
7 European Forum of the EUI on \The Political Economy of an Integrated Europe."
Giorgia Giovannetti acknowledges ¯nancial support by MURST.
9
Non Technical Abstract
In spite of the high level of economic and ¯nancial integration in
Europe in the last ten years, there are still marked di®erences across
countries regarding: (1) e±ciency of the banking sector; (2) portfolios
holdings and ¯rms' ¯nancing, and (3) output and price responses to
monetary shocks. This paper develops an equilibrium model with limited
participation, which is consistent with these three facts.
In the paper, we study two economies (which can be identi¯ed with
France and Germany) di®ering for the degree of e±ciency of the banking
sector (fact 1). As a result, agents in the two countries hold di®erent
portfolios and ¯rms are ¯nanced with a di®erent mix of debt and equity
(fact 2). When we calibrate the model for the two economies, impulse
responses to a monetary shock di®er in a way similar to the observed
impulse responses for France and Germany (fact 3).
We then study the e®ect of integrating these two economies in a
Monetary Union. We ¯nd that, when countries are in a monetary union,
and to the extent that the di®erences in the ¯nancial systems persist
(likely to happen in the ¯rst stage of EMU), endogenous preferences for
monetary policy may be even more diverse than when countries are separated. In particular, the same monetary policy gives rise to redistribution
e®ects not present if countries were more isolated.
Finally, we estimate VAR for France and Germany over the period
1973-1997 and we show that the output responses to monetary shocks
are very similar to the theoretical reactions derived from our model and
similar to other existing result.
1
1
Introduction
The transmission mechanism of monetary policy may be de¯ned as the
ways in which monetary impulses from the central bank a®ect output
and prices. Changes in monetary policy are transmitted to the real economy through various channels, each of them can consist of several stages.
Hence, national transmission mechanisms are likely to be di®erent: different channels may be at work in di®erent countries and the intensity
by which a monetary impulse is transmitted can vary, even substantially.
But di®erences in national monetary transmission processes between European countries, in turn, are likely to a®ect the magnitude and timing
of the price and output e®ects of alternative monetary policies of the
European Central Bank (ECB). They may also have implications for
the scope and nature of policy coordination in the current state and for
evaluating possible bene¯ts of joining a monetary union for individual
countries. Hence, even in the absence of cyclical divergences and differences in policy preferences across countries, the stance of monetary
policy to be followed by the ECB can be a source of con°icts between
member states of EU. This, in turn, may imply a decrease of support for
the monetary policy of the ECB1 .
In what follows, we do not enter the debate on the relative importance of the di®erent channels of transmission of monetary policy. We
believe that monetary policy a®ects output and prices (at least in the
short to medium run) and may do so through di®erent channels, not
mutually exclusive, simultaneously at work and likely to reinforce each
other. Our aim is to focus on the fact that a certain (common) monetary
stance may have di®erent macroeconomic consequences from one country to another. While the existing literature has mainly analyzed partial
equilibrium models emphasizing speci¯c channels or, when approaching
the issue in a general equilibrium framework, has concentrated on liquidity e®ects, we show that the di®erences in the speed and magnitude of
1 For
instance, the ERM crises of 1992-93 highlighted cross-country di®erences in
the monetary transmission mechanism which seemed to substantially a®ect the cost
of maintaining the parities.
2
a monetary impulse into economic activity depend on di®erences in the
¯nancial structure, on di®erent role of ¯nancial institutions and on di®erent portfolios composition of households and ¯rms in di®erent countries
as well as on di®erent liquidity constraints. In this framework, when (or
if) a common monetary policy is implemented (and to the extent that
the e®ects of it are di®erent for di®erent countries) unwanted distortions
and/or con°icts may arise.
An empirical (statistical) examination of the role of banks, stock
markets and portfolios compositions of the European economies suggests
that they di®er signi¯cantly in many ways. First of all, in some European
countries (e.g. UK) markets for privately issued debt and stock markets
are highly developed, so that bank credits are (almost) perfect substitutes
for bonds, while in others (e.g. Italy and Germany) these markets are
less developed and bank credit and loans cannot be seen as substitute
sources of ¯nancing. This can substantially a®ect the liquidity of markets
and the ways in which a money injection (or reduction) is translated
into households and ¯rms. There are also di®erences in regulations, in
procedures, in the relative use of short term versus long term ¯nancing,
in the relative share of ¯xed versus °oating rates, in the degree and
composition of indebtedness of ¯rms, households and governments. Also,
the ¯nancial structure of the di®erent countries has evolved di®erently in
the last two decades, with changes in the competitiveness of the banks
and growth of non bank ¯nancial intermediaries in some countries but
not in others, with di®erent evolution of stock markets and changes in
the composition of assets and liabilities of households and ¯rms. At the
same time, mainly in the last decade, capital markets have become more
integrated and some EU countries (e.g. France and Italy) have been
compelled to lift previously operating administrative controls.
There are, of course, some areas where convergence is likely at the
outset of a Monetary Union. First of all, the convergence of in°ation
rates should lead to a more uniform pattern of short-term versus long
term ¯nancing across countries, at least to the extent that these differences have emerged as a result of di®erent in°ation records; also the
increased competition across ¯nancial intermediaries should imply that
3
the pass-through of changes in market rates to lending rates should become more similar. We maintain that some structural di®erences will
stay; the transmission of monetary impulses to the real activity and the
distribution of gains and losses amongst EMU members is likely to depend on these di®erences.2
Our aim is to capture some of the existing di®erences in the ¯nancial structure and portfolios compositions of European countries in
an equilibrium model where economies di®er for the relative e±ciency
of the ¯nancial intermediaries (i.e. households portfolios and currency
holdings) and to study the implications of a common monetary policy
for di®erent economies under two di®erent regimes: with and without a
Monetary Union. We ¯nd that the same monetary policy gives rise to
redistribution e®ects, not present when countries are isolated.
The paper is organized as follows: Section 2 presents some stylized
"facts" emphasizing the heterogeneity of ¯nancial markets in 4 big European countries: Germany, France, Italy and the UK. Section 3 outlines a
simple dynamic equilibrium model which allows to account for (at least
some of) the detected di®erences and analyze the consequences for the
transmission mechanism of monetary policy. It is an overlapping generation model with cash in advance constraints. A monetary expansion
induces a decrease of the interest rate and an expansion of output through
a substitution of consumption of cash and credit goods and revisions of
plans about deposits and assets holding. The di®erent e±ciency of the
banking system implies di®erent liquidity constraints across countries, so
that the degree of substitution and therefore the real e®ects are di®erent.
Contrary to most G.E. models, using this framework, we get persistence
of an independent shock. Section 4 calculates the theoretical impulse
responses for an interest rate shock, under di®erent scenarios (autharkic
countries, Monetary Union, the same and di®erent cash/deposit ratios).
Our model economies, which we call for convenience France and Germany, have some features of the true ¯nancial structure of France and
2 Of
course di®erences exist also within the national borders. They tend, however,
to be smaller, since between countries there are more di®erences in regulations and
institutions.
4
Germany and respond to shocks in a way similar to the observed response.
Section 5 presents some VAR estimates of the e®ect of an interest rate
shock on output of France and Germany over the 1973-97 period and
Section 6 concludes. The Appendix contains the description of the data
set and unit root tests on the variables used in the VAR estimation.
2
The heterogeneity of European ¯nancial
markets: Some stylized "facts"
Despite the implementation of the single market from 1992, despite all
the changes brought about by deregulation, capital liberalization and
technological innovation in the last two decades, the ¯nancial systems of
European countries are still characterized by a high degree of heterogeneity. Furthermore, their convergence over time has been quite limited and
some of the fundamental di®erences existing in the 1980s have survived
all the changes. In the following we point out two related di®erences in
the ¯nancial markets of France, Germany, Italy and the UK3 , which we
believe can a®ect the transmission mechanism of monetary policy and
therefore induce con°icts in the monetary policy decisions of the European Central Bank: the degree of development of ¯nancial markets and
portfolio decisions of households, ¯rms and institutional investors.
2.1
The degree of development of ¯nancial markets:
There are two main channels through which funds °ow from savers to ultimate borrowers within each economy. Savers can invest directly, through
the purchases of securities such as stocks or bonds issued by a non ¯nancial corporation (direct ¯nance) or their °ow of savings can be interme3 While
here we concentrate on these four countries, Gennari and Giovannetti, 1998
provides data on the ¯nancial structures, liquidity constraints and portfolio choices
for 15 EU countries.
It must be noticed that, in Europe, there is also a signi¯cant
heterogeneity regards the links between the central banks and the ¯nancial sector, cf.
Giovannetti and Marimon, 1995.
5
diated by ¯nancial ¯rms (indirect ¯nance). Direct ¯nance takes place in
capital markets. The prevalence or absence of ¯nancial intermediation in
a national economy structures the relationships within the private sector.
European countries are very di®erent with respect to the mix of
direct and indirect ¯nance that characterizes their ¯nancial systems. According to European Economy (1997) "This is mostly explained by the
relative role of domestic banking: countries with high ¯nancial intermediation equally show a high degree of banking intermediation" (p.10). As
shown in Table 1, Germany is characterized by a much higher degree of ¯nancial intermediation (more than 50%) and bank intermediation (above
80%) than any other European country. Because of the dominant role
of bank intermediation, many ¯nancing demands which could be met by
bonds or equities are provided by bank loans. Accordingly (or because
of) the most e±cient banking sector amongst European countries -no
matter what criteria is used to assess e±ciency4 is in Germany. The existing data, not fully harmonized and therefore to be used with caution,
show that classical banking intermediation (i.e. taking deposits from consumers and making loans to people and ¯rms) is still the main channel of
saving and investment in all EU countries. However, there are relevant
di®erences in the use of loans versus shares, which re°ects di®erences in
market capitalization. In Germany, security markets are underdeveloped
with respect to other major EU countries (namely France and the UK,
see Table 1). In1995, stock capitalization represented only 29% of GDP
in Germany versus almost 150% of GDP in the UK (it was 39% in France,
and 22% in Italy, see again table 1). The number of ¯rms quoted in the
stock market is much larger in the UK and in France (both in terms of
consistency and new quotations) than in Germany and Italy and new
issues are particularly low in Germany5 . As a result, equities issues by
4 Di®erent crietria can be applied to assess the e±ciency of the banking sector.
Gual and Neven (1993) suggest to evaluate the sta® costs per deposit, which give
information on the cost side of intermediation, or the net interest income per deposti,
which also allows to account for possible lack of competition.
5 In Germany the capital market was fragmented into eight independent regional
stock exchanges till fairly recently and this can at least partially explain the di®erences
in the degree of stock market capitalization. Also, German banks conduct both direct
6
¯rms are a signi¯cant share of GDP in the UK and France (respectively
65 and 70% in 1994) but almost irrelevant in Germany (25% in 1994)6 .
It must be also noted that most European stock markets mainly trade
domestic equity and that ¯nancial integration has not changed this type
of segmentation. Only in London foreign shares are usually traded (2/3
of total trading in London is foreign shares, which amounts to around
95% of total EU trading in foreign shares).
Table 1 here
2.2
Portfolio decisions
The di®erent mix of direct and indirect ¯nance re°ects in portfolio decisions of the private sector. Even though, as far as households are concerned, the share of deposits over gross assets has fallen everywhere in
the last twenty years (table 2), the extent of the fall is very di®erent: in
Germany deposits were 59% of gross assets in 1980 and still constitute
45% of households ¯nancial assets in 1994, while, for instance in France
they dropped from 59% to 32% and in Italy from 58% to 29% (households
savings has been fairly stable in this period, despite cyclical °uctuations).
While bonds have remained fairly constant between 1980 and 1994 (see
Table 2), direct securities holding have been in general declining (France
represents an exception). Transactions costs in securities markets (including the bid-ask spread) makes it di±cult for households of average
means to diversify via direct securities holdings especially because liquidity is low in the case of direct holdings. Hence a feature of UK, Germany
and Italy has been that the share of households portfolios held in the
form of securities has tended to decline (table 2) while the proportion of
equities and bonds held via institutions has tended to increase (see again
and indirect ¯nance and have therefore made the capital markets largely endogenous
to the banking system.
To the extent that industries have access to the securities
market, their access has been governed by banks.
6 There
also seems to be a correlation in the 4 European countries between equity
market capitalization and the size of ¯nancial institutions, but a discussion of this
issue is outside the scope of this paper. Cf. Davis, 1996.
7
table 2). Only in France direct holding of securities passed from 14%
in 1980 to 32% in 1994 (possibly because of a successful privatization
process), with institutional investors also increasing their weight (from
7% to 29%). This seems to indicate that in France households tend to directly supply funds to the ultimate borrower even if this means to bypass
the ¯nancial sector.
Table 2 here
As far as non-¯nancial ¯rms are concerned, there has been an overall increase in ¯nancial liabilities in the last two decades which has been
covered with di®erent mix of debt and equities. The existing data (OECD
¯nancial accounts statistics) show particularly large di®erences in the use
of loans versus shares (see Table 3). Loan ¯nancing is particularly high
in Germany and substantially lower in France and the UK (when considering loans as proportion of gross ¯nancial assets, respectively (50%, 28%
and 12%). Hence, the role played by German banks in lending to non¯nancial corporations is substantially bigger7 . Furthermore, over time,
the loan ratio declined substantially in the UK and remained fairly constant in other European countries. The equity ratio, on the other hand,
has risen everywhere except in Germany, reaching the remarkable value
of 70% in France and 65% in the UK, while staying at a mere 25% in
Germany.
Table 3 here
Structure of equity holdings, however, has tended to move away
from the household sector and towards institutional investors everywhere
apart from France, where, as we said, households hold directly substantial shares of equities (see Table 4). In Germany, for instance, ¯nancial
institutions own 30% of the total amount outstanding (14% are directly
owned by banks and the remaining 16% by other ¯nancial institutions)8 .
Table 4 here
In 1991 more than 60% of bank loans were provided in a long term form in
Germany, while the same ¯gure was around 50% for the UK, cf. O
7
ECD Non-Financial
Entrerprises Financial Statements, 1991.
8 Many
bank customers, furthermore, keep their shares deposited with banks and
allow banks to exercise voting proxies on their behalf.
8
2.3
Monetary aggregates
Per capita currency holding (expressed in common currency, i.e. dollars)
di®ers substantially amongst European countries despite similar levels of
development9 . Germany has a much higher ¯gure than other European
countries (with the exception of Switzerland), possibly because of the
large holdings of DM abroad. Theoretically, the currency holding should
decrease over time as a result of ¯nancial innovation and use of electronic
money, but in Germany if anything, currency holding has increased.
Composition of monetary aggregates also varies substantially in
Europe. For instance, the ratio of cash to the di®erent measures of
money supply (respectively, M1, M2 and M3) is higher in Germany10
(see Table 6) than in the other countries.
Also, o±cial reserves are lower in France and the technical features
of the reserve requirements di®er signi¯cantly across countries, re°ecting
functional and structural di®erences between national ¯nancial systems.
The main di®erences are re°ected in the de¯nition of bank liabilities (type
and currency), the rates applied, and the existence and level of remuneration. In Germany for instance, a 5% reserve requirement is levied on
sight deposits and 2% on other types of deposits, without remuneration.
In France, the ratio is 1% on sight deposits, and 0,5% on other types
of deposits, also not remunerated. In Germany, the required reserves in
1994 were 1.3% of GDP and in France only 0.1%. This is likely to a®ect
the costs of the intermediation.
Table 5 here
These di®erent characteristics of the ¯nancial systems obviously
9 The presumption is that per capita currency holdings di®ers with di®erent level
of developments. In particular, less developed countries have a lower average level of
currency holding, also because of unstable environment. However, the big di®erences
existing between countries with similar levels cannot be explained merely by reference
to di®ering payment habits and rates of in°ation.
10 It must, however, be noted that Eastern European countries use DM (and no
other European currency) and this can impart a bias on the total amount of cash, cf.
Overall, Seitz concludes that roughly 40% of the German money supply
is held abroad.
Seitz, 1995.
9
re°ect in empirical analysis of the transmission mechanism, but do not
translate in clear-cut conclusions about the likely impact of a monetary
shock. At the empirical level, in fact, the di®erent characteristics of
a country in terms of ¯nancial structure can have o®-setting e®ects11 .
At the theoretical level, the work on the transmission mechanism has
mainly focussed on limited participation models (see for all, Christiano
et Al, 1997), without emphasizing the possibility of di®erent ¯nancial
structures.
Having in mind that the mix of direct and indirect ¯nance is very
di®erent in the four major European countries, that the ratio cash to
monetary aggregates also varies substantially, we propose a limited participation model where we allow for di®erent ways of saving and ¯rms'
¯nancing (which is the endogenous result of di®erent e±ciency of the
¯nancial sector across countries). This is the object of the next section.
3
A model with ¯nancial diversity
We develop a model that tries to incorporate some of the features that, as
we have done in the previous section, can be identi¯ed as potential sources
of diversity {and con°ict{ in the way that the Transmission Mechanism
may work in the early stages of the EMU. The model is an Overlapping
Generations Model with Cash-in-advance features. The OLG structure
allows for alternative savings decisions. In particular, agents live for three
periods, receive an endowment in their two initial periods and consume
in their last two periods. They can diversify their portfolios between
outside money (cash), bank deposits and equity, in the form of asset
holdings of an underlying technology, that {after two periods{ realizes a
positive real return. There is no uncertainty and Cash-in-advance constraints guarantee that the {return dominated{ outside money is being
held by households. Nevertheless, economies may di®er in the extent that
11 For instance, sluggish adjustment of bank lending rates can protect ¯rms from
shocks but banks can ration credit (non price rationing) and amplify the e®ects of a
shock.
10
goods must be purchased with cash. Agents can get a positive return on
their savings by either making deposits in ¯nancial intermediaries or directly holding the two-period asset. Whether they directly hold assets
depends on the relative e±ciency of the banking system, another feature
that will di®erentiate our economies. However, even with relatively inef¯cient banking systems, agents will typically use ¯nancial intermediaries
to obtain one-period returns.
The ¯nancial intermediaries, resembling the behavior of banks, accept funds from households and return them to the household in the form
of interest and principal payment. Financial intermediaries use their deposits (and, possibly, monetary injections) to purchase two-period assets
(as if they were lending to ¯rms). Given that they are in¯nitely-lived institutions they can provide households with one-period returns at a cost.
Given that there is perfect competition in the sector, ¯nancial intermediaries' returns correspond to the outside asset return net of operating
costs. Banks account for indirect channels of supply of funds; the stock
market on the other hand, is an example of a direct channel, since it lets
households to directly purchase assets. As in most developed economies,
central monetary authorities deal primarily (uniquely, in our model) with
¯nancial intermediaries and, therefore, new money enters the economy
by an injection from the monetary authority into the ¯nancial intermediaries. Government bonds and open market operations can easily be
incorporated in our model but, for simplicity, we do not include bonds
and we limit our analysis to the case an exogenous injection (subtraction)
of cash to (from) ¯nancial intermediaries.
With respect to the stylized facts previously discussed, our model
economies could represent, broadly speaking, France (and the UK) and
Germany (and Italy). As we have seen in Section 2, in the former household directly hold assets (shares) and, in general, they do not channel
a large part of their saving into deposits. In the latter, on the other
hand, indirect channels are the norm. Households loan to the ¯nancial
intermediaries their money and get in exchange a return.
11
3.1
Goods, assets, households and ¯nancial intermediaries
There is a continuum of consumption goods, exogenous endowments, and
a real asset giving a return of R2 after two periods12 . Consumption goods
only di®er in the form on how they can be purchased. In fact, real assets
and endowments can be {without costs and linearly- transformed into
consumption goods, independently of their type. One can think of our
economies as having goods in di®erent locations where some locations
(e.g. street vendors) only accept cash while others are willing to sell for
what, e®ectively is, credit (e.g. stores that accept debit cards, checks or
other forms of credit). In terms of consumption, however, the agent is
indi®erent on where the good is being purchased. An agent of generation
t (born in period t ¡ 1) only values consumption in the last two periods
of his life. That is, household's preferences are represented by
U(c1t ) + ¯U(c2t )
(1)
where c1t is {the average{ consumption in the intermediate period of his
life and c2t of his last period. The utility from cash and credit goods is
given by:
U(c) =
Z
0
°
u(ci )di +
Z
1
u(ci )di
(2)
°
where ° is the parameter indicating how goods can be purchased13 .
Goods in the range (location) [0; °] can only be purchased with cash
while goods in the range [°; 1] can also be purchased with credit14 .
12
R2
denotes the real return net of transactions costs.
These can be di®erent for
¯nancial intermediaries and for individual agents, as well as they can di®er across
di®erent economies.
13 We
0; u
00
0
make the standard concavity and di®erentiability assumptions. That is, u >
< 0:
14 As
we have said, ° is one of the parameters that will di®erentiate our economies.
It should be noticed that this formulation allows for a simple characterization of a
richer transactions technology, by making ° endogenous (e.g.
a function of e®ort
and society's technology). Here, however, we consider ° an exogenous technological
12
In period t ¡ 1 an agent of generation t has the following budget
constraint:
M1t + D1t + pt¡1 at · pt¡1 !0
(3)
R1
where !0 is the ¯rst period's (average, i.e.,!0 = 0 !0i di)15 endowment.
Its value is allocated in a portfolio of cash, M1t ¸ 0, nominal deposits in
¯nancial intermediaries, D1t , and holdings of the real asset, at ¸ 0. In
the intermediate period of his life, the agent faces the following budget
and cash-in-advance constraints:
M2t + D2t + pt
Z 1
0
c1t · M1t + D1t It + pt !1
M1t ¸ pt
Z
0
(4)
°
c1t
(5)
where It is the nominal return on (positive) deposits. Notice that we
already imbedded in these de¯nitions the fact that all goods face the same
prices, as well as the fact that the agent has no interest in purchasing
two-period assets in the intermediate period of his life. Finally, in the
last period of his life, the agent faces constraints
pt+1
Z 1
0
c2t · M2t + D2t It+1 + pt+1 at R2
(6)
Agents can also borrow from ¯nancial intermediaries. However, an agent
borrowing from a ¯nancial intermediary faces a higher interest rate. Such
spread corresponds to ¯nancial intermediaries costs, which are discussed
below. In the class of equilibria that we study agents do not borrow.
Notice that generations overlap for two periods. When generation t-1
decides how much to consume of respectively cash and credit goods (i.e.
how to allocate the endowment !1 ), generation t gets an endowment !0
and decides how much to deposit and how much to invest in real assets.
parameter.
15 For
simplicty of exposition, we will denote integrals simply as
13
R1
0
! from now on.
3.1.1
Financial intermediaries
In the model ¯nancial intermediaries accept loans from households, which
are repaid at the end of each period at a market interest rate, and purchase assets. Alternatively, the purchase of assets can be viewed as loans
to private ¯rms that pay back {after two periods{ a real return. Financial intermediaries also receive new cash injections from the monetary
authority16 . The balance of ¯nancial intermediaries, in absence of money
injections, can be written as:
Dt+1 = pt abt+1
(7)
denotes
where abt+1 denotes assets in the hands of banks, dt+1 = Dpt+1
t
deposits in real terms and Dt = D1;t + D2;t¡1 ; i.e. total deposits in
period t ¡ 1 are given by the sum of generation t ¯rst period deposits
and generation t ¡ 1deposits in their intermediate period.
We consider the following ¯nancial intermediation technology. First,
¯nancial intermediaries can obtain a two-period return (R + µ1 )2 from
(borrowing to) private ¯rms. µ1 ¸ 0 denotes the technological advantage
of ¯nancial intermediaries with respect to households. Second, ¯nancial
intermediaries can transform a two-period asset into a one-period asset,
with return (R+µ1 ) at a real cost µ2 +µ3 , where µ2 corresponds to the cost
of making the asset more liquid and µ3 to the cost of handling the one
period asset. In other words, the ¯nancial intermediation technology generates one-period assets with a real return (R ¡µ), where µ = µ2 +µ3 ¡ µ1 ,
from the existing two-period assets. The relative e±ciency of di®erent
¯nancial communities will be represented by di®erences in µ. The cash
°ow of ¯nancial intermediaries (CF) can, therefore, be written as:
CFt = pt abt R ¡ pt abt+1 + Dt+1 ¡ Dt It ¡ pt abt µ
16
(8)
For simplicity we do not include government bonds into ¯nancial intermediaries'
balance sheets. This can be done without any di±culty (the standard non-arbitrage
conditions will equate the returns of di®erent assets in circualtion) and will allow for
government's open market operations.
14
Since there is free entry in the ¯nancial intermediation sector, the zero
pro¯t condition implies that:
abt (R ¡ µ) =
Dt pt¡1
It ´ dt Rtd
pt¡1 pt
i.e., Rtd = R¡µ, where Rtd is the one-period real return on (real) deposits.
As we said, households can borrow from ¯nancial intermediaries,
signing one-period debt contracts. In such a case, they will face the
nominal rate It + ¼t µ3 . We assume that µ1 ¸ µ2 which guarantees that,
even when µ < 0, households will not borrow to ¯nance the purchase of
assets, since Rdt + µ3 = R + µ1 ¡ µ2 ¸ R; the last inequality following from
our assumption.
3.1.2
Monetary policy
We consider a very simple class of monetary policies. At the beginning of the initial period 0 agents of generation 0 are endowed with
per-capita money holdings of M1 and agents of generation ¡1 with percapita money holdings of M2 . Money supply is constant thereafter, although we will consider the experiment of unexpectedly increasing (decreasing) the money supply by Xt+1 in period t. This is done through
¯nancial intermediaries. In such a case, their consolidated balance sheet
is Dt+1 + Xt+1 = pt abt+1 . That is, ¯nancial intermediaries can purchase
(or sell) assets with the proceeds (the claims) of the Central Bank and
return, the following period, (Dt+1 + Xt+1 )It+1 to depositors. To maintain the deterministic nature of our model we will only consider \once
and for all surprises."
3.1.3
The initial period
Notice that in our economies there is not enough to characterize the initial distribution on money holdings, we must also characterize the initial
distribution of assets and deposits. We assume that at the beginning of
the initial period agents of generation 0 have real claims in ¯nancial intermediaries of d1;0 giving them a real return d1;0 R0d . Similarly, agents of
15
generation ¡1 start period 0 endowed with assets a¡1 and deposits d2;¡1 ,
giving them returns a¡1 R2 and d2;¡1 R0d , respectively. Finally, ¯nancial
intermediaries start period 0 endowed with ab¡1 assets and satisfy their
commitments on initial deposits d0 = d1;0 + d2;¡1 . As we will see, stationary equilibria can be easily characterized; however, their existence
requires an appropriate initial distribution of assets and deposits. For
example, we will consider economies where a¡1 = 0 and economies where
d2;¡1 = 0. Alternatively, it can be shown that, given an initial distribution of assets and deposits, the economy converges to a stationary
equilibrium from period one on.
3.2
Monetary equilibria in a closed economy
A monetary equilibrium is achieved, for a given initial distribution
(M1 ; M2 ; d1;0 ; d2;¡1 ; a¡1 ; ab¡1 ), when there are prices (p0; f¼t ; It g1
), such
t=1
that (i) ¯nancial intermediaries choose asset holdings and supply deposits, fabt ; Dt g that maximize pro¯ts, under a free-entry condition;(ii)
households choose consumptions and portfolios fc1;t ; c~1;t ; c2;t ; c~2;t ;
M1;t ; M2;t ; D1;t ; D2;t ; at g17 that maximize their utility subject to their
budget, and cash-in-advance, constraints and, ¯nally, (iii) all markets
clear. In particular, feasibility in the goods market requires that:
°c1;t +(1¡°)~
c1;t +°c2;t¡1 +(1¡°)~
c2;t¡1 +at+1 +abt+1 = !0 +!1 +at¡1 R2 +abt Rdt
(9)
In order to characterize equilibria, notice that from the ¯rst order
condition of the households maximization problem we obtain di®erent
solutions depending on whether assets returns dominate deposits or viceversa because depending on the sign of µ agents will decide to directly
purchase assets (if µ > 0), in which case they will not hold second period
deposits (i.e., D2;t = 0), or they will put all their savings into ¯nancial intermediaries (i.e., at = 0 if µ > 0). To distinguish among these
17 Since consumers decide to consume the same quantities of
period, and similarly for credit goods, we denote by
c1;t
all the cash goods of one
the generation
t
consumtion
of cash-goods in their interemediate period and c
~1;t the consumption of credit goods
for the same period, etc.
16
economies, we will denote by economy A, an economy where Assets returns dominate deposits and by an economy B one where consumers
prefer Banks to the stock market as a way to channel their savings.
In an economy of type A (µA > 0), we get the standard condition
equating the marginal rate of substitution between cash and credit goods
(of period one) to the nominal interest rate (the cost of assigning part of
the portfolio to an intermediary is the lost liquidity of not holding cash
and the gain is the interest that can be used for future purchases):
u (c1;t )
= Itd
u (~
c1;t )
0
(10)
0
where, from now on, c~ denotes a credit good and c a cash good. However,
for period 2 we get:
·
¸2
R
u (c2;t )
= Itd+1
u (~
c2;t )
R¡µ
0
0
(11)
On the other hand, in an economy of type B (µB · 0), we get the standard
condition for both periods:
u (ck;t )
= Itd 1+k ; k=1,2
u (~
ck;t )
0
¡
0
(12)
We obtain the following intertemporal Euler equations, respectively for
economy B and economy A:
u (~
c1;t ) > ¯u (~
c2;t )Rtd+1
0
0
(13)
u (~
c1;t ) > ¯u (~
c2;t )R2 (Rtd ) 1
0
0
¡
(14)
Furthermore, as we have seen, competition in the ¯nancial intermediation
sector implies that Rtd = R ¡ µ:
To simplify the analysis, we consider the case of a log utility: u(c) =
log(c):With a logarithmic utility function demands take a simple form.
17
¡1
1 (R¡µ)
Let W = !0 +!(1+
, then in economy A generation t has the following
¯)
demands
c1;t = WA ¼t¡1 ;
c~1;t = (R ¡ µA )WA ;
c~2;t = ¯R2 WA ;
d1;t = !0 ¡ [(1 ¡ °A )¯ + °A ]WA ;
at = (1 ¡ °A )¯WA ;
m1t = °A WA
1
c2;t = ¯(R ¡ µA )WA ¼t¡+1
d2;t = 0;
and m2t = °A ¯(R ¡ µA )WA
Substituting for consumptions, assets and deposits expressed in the
feasibility constraint (9) we obtain an equation in one variable, namely
the in°ation rate:
M (¼t¡1 ¡ 1) = 0
(15)
where MA = °A WA [1 + ¯(R ¡ µA )]: Notice that for µA 2 (0; R); (15) has
a solution ¼t = 1, for t ¸ 1, showing that there is a unique monetary
equilibrium which is stationary from period one on.
Similarly, for economy B we obtain the following demands
c1;t = WB ¼t¡1 ;
c~1;t = (R ¡ µB )WB ;
c~2;t = ¯(R ¡ µB )2 WB ;
at = 0;
m1t = °B WB
1
;
c2;t = ¯(R ¡ µB )WB ¼t¡+1
d1;t = !0 ¡ °B WB ;
d2;t = (1 ¡ °B )¯(R ¡ µB )WB
and m2t = °B ¯(R ¡ µB )WB
and (15) also characterizes the equilibrium in°ation rate, ¼t¤ = 1, t ¸ 1:
3.3
Open economies with segmented ¯nancial sectors
If an economy A and an economy B have a common market, but ¯nancial
disparities are maintained and -consistently with the well known \home
bias puzzle"- consumers tend to use their home ¯nancial institutions,
then the situation is similar of that of two independent closed economies.
To see this, consider a °exible exchange regime within the countries and
that the cash-in-advance constraints must be satis¯ed with the domestic
18
currency. Furthermore, assume that, in spite of the single market, there
is a cost µ4 from operating across borders, such that µB + µ4 > µA (and
¯nancial intermediaries maintain the same domestic cost structure with
¡1 < (R ¡ µ ) and ¼ ¡1 <
no arbitrage opportunities). Then, as long as ¼Bn
A
An
(R ¡ µB ) , n = t; t + 1, generation t demands are as in the close economy
case and monetary equilibrium in°ation rates are solutions to
¡1 ¡ 1) + MB (¼ ¡1 ¡ 1) = 0
MA (¼At
Bt
In particular, the stationary solution is ¼At = ¼Bt = 1 de¯nes a monetary
equilibrium for the °exible exchange regime. Notice, however, that there
is a continuum of equilibria given by
¡1 ¡ 1
¼At
MB
=
¡
¡1 ¡ 1
MA
¼Bt
satisfying the above restrictions on asset return dominance. These solutions, however, involve a trade imbalance, and a corresponding permanent devaluation of one of the currencies. We will focus in the stationary
equilibrium that parallels the closed economies case.
3.4
Monetary Union equilibria (with segmented ¯nancial sectors)
We ¯nally consider the case in which countries A and B form a monetary
union, but \national," or \regional," disparities persist. That is, \transnational" (or \trans-regional") ¯nancial transactions are subject to the
cost µ4 . We can also consider that, even if all consumers in the monetary union can satisfy their cash-in-advance constraints with the common
currency, there may still persist di®erences regarding the range of goods
that can be purchased with credit; that is, °A and °B may di®er. As in
the °exible exchange regime with segmented ¯nancial markets, demands
are as in the closed economies case. In particular, monetary equilibrium
in°ation rates for the MU are solutions to
[MA + MB ](¼t¡1 ¡ 1) = 0
19
As in the case were both countries are separate, stationary output
will di®er across countries even if they have the same underlying (asset) technologies, endowments and preferences, but they maintain their
di®erences regarding the e±ciency of the ¯nancial intermediation sector.
4
Unexpected monetary shocks
We now consider a monetary expansion [contraction] taking the form of
a once-and-for-all monetary injection [absorption] Xt+1 in period t. We
consider ¯rst the case of independent countries (which also characterizes
the stationary equilibrium of the common market with °exible exchange
rates) and then the case of a monetary union. As it is well known, the
e®ects of monetary policies (in models of limited participation and in
real economies) depend on the \when and how" monetary interventions
take place. By a monetary injection in period t we mean a unexpected
monetary intervention that takes place after period t decisions have been
made. In our model, the monetary injection is done through the ¯nancial
intermediaries, which have the following consolidated balance sheet
Dt+1 + Xt+1 = pt abt+1
That is, ¯nancial intermediaries purchase [sell] assets (make loans) and
the proceeds are paid back to the depositors who {on aggregate- receive a
return (Dt+1 + Xt+1 )I: The e®ects of an identical monetary shock will be
di®erent depending on the type of ¯nancial structure; i.e., whether the
economy is a type A economy, a type B economy or a monetary union
(of a country of type A and a country of type B). In particular, we are
interested in how prices (i.e., nominal interest) and portfolio allocations
change and the e®ect of these changes on consumption and output. We
can distinguish three types of e®ects: (i) an income e®ect (due to the
fact that only people holding deposits get a share of the shock); (ii) an
e®ect through di®erent portfolio choices, and (iii) a pure liquidity e®ect
(due to di®erent ° s).
0
20
In a closed economy the equilibrium condition in period t is
M (¼t¡1 ¡ 1) +
which implies that
¼t = 1 +
Xt+1
=0
pt
zt
M
where zt = Xptt¡+11 . If Xt+1 > 0, i.e. if we consider a monetary expansion, then there will be a contraction of cash goods in period t a®ecting
generations t¡1 and t. In economy A; generation t+1, the only one holding deposits, revises its decisions knowing that will be getting a higher
(lower) return from their deposits than the one originally foreseen. Let
c~1;t+1 be the variation on the consumption of credit goods (i.e., denote
variations). we have that generation t + 1 shares the extra returns from
his deposits, d1;t+1 , xt+1 (R ¡ µA ); as follows:
0
0
0
R ¡ µA
R ¡ µA
; m2;t+1 = ¯xt+1
(1 ¡ °A + ¯)
(1 ¡ °A + ¯)
R ¡ µA
= ¯(1 ¡ °A )xt+1
(1 ¡ °A + ¯)
0
c~1;t+1 = xt+1
0
and d2;t
Notice that after a monetary shock, an agent of generation t + 1 would
like to change his portfolio, but, in economy A, the assets, at , are not
liquid and, therefore, the agent must deposit or borrow from the bank (a
less attractive intermediation technology). When he saves (i.e. xt+1 > 0)
this results in c~2;t+1 = ¯(R ¡ µA )d2;t , while when he borrows in c~2;t+1 =
¯(R ¡ µA + µ3 )d2;t . These costs of readjusting the portfolio are a crucial
distinct feature of economy A:
0
0
0
0
If instead the economy that experiences the Xt+1 shock is of type B
, both generations, t and t + 1, holding deposits, will change their creditd
+1 be the share of deposits corresponding
goods consumption. Let ® = d +1
to generation t + 1. Then, c~2;t = (1 ¡ ®)xt+1 (R ¡ µB ) and generation t + 1
will revise their consumption plans as in economy A (except that they
only receive xt+1 (R ¡ µB )). The di®erence, however, is that since in economy B; d2;t > 0, and agents always want to have positive consumption of
credit goods, d2;t + d2;t > 0: That is, there is no borrowing from ¯nancial
i;t
0
t
0
0
21
intermediaries following a monetary contraction in economy B. In other
words, in economy B the adjustments, following a monetary shock, are
less costly than in economy A (agents use the same intermediation technology with the same returns as when they where making consumption
plans in their initial period).
The adjustments of generation t + 1, however, result in an excess
supply (demand) in the goods market in period t + 1 (due to m2;t+1 6= 0),
resulting in a variation of prices given by
·
¸¡1
¯°A
xt+1
¼t+1 = 1 +
(R ¡ µA )
1 ¡ °A + ¯ MA
0
in economy A and, similarly, in economy B
¸¡1
·
¯°B
®xt+1
¼t+1 = 1 +
(R ¡ µB )
1 ¡ °B + ¯ MB
If both countries form a monetary union but ¯nancial structures
remain the same and households use their countries' ¯nancial intermediaries, households will adjust their portfolios in the same manner as they
do when countries are separate. There is, of course, an important di®erence in that there is a unique price reaction for both countries. That is,
with a shock Xt+1 , in°ations in period t and t + 1 are, respectively,
zt
¼t = 1 +
MA + MB
and
¶
¸¡1
·
µ
®¯°B
xt+1
¯°A
(R ¡ µA ) +
(R ¡ µB )
¼t+1 = 1 +
1 ¡ °A + ¯
1 ¡ °B + ¯
MA + MB
Notice that if countries are of the same (endowment) size, the country
with a larger ° will absorb most of the shock in consumption, which will
result in a redistributive e®ect in period t (and t + 1).
4.1
The quantitative e®ect of a money shock in our
economies
The real e®ects of monetary shock can be quite di®erent depending on
the type of ¯nancial structure that a country has or wether countries are
22
integrated in an heterogeneous monetary union or not. Even if our aim
is not to exactly mimic observed economies, the parameters underlying
¯gures have been chosen as to approximate Economy A with France and
Economy B with Germany (see Table 6)18 .
Table 6 here
Figure 1a illustrates the e®ects of a monetary contraction on output
(and Figure 1b for consumption) when economies A and B are independent (the shock is in period 6). As it can be seen, the e®ect on output is
higher and slightly more persistent for economy A than for economy B.
In other words, the economy (Germany) with a higher cash/deposit ratio
and indirect ¯nance prevailing over direct access to the market (µA >µB );
because of lower transaction costs and higher e±ciency, is partially protected from the e®ects of a monetary restriction. Figure 2a, reproduces
the same experiment for a Monetary Union (and a shock twice the size).
As in the closed economies case, at the time of the shock, aggregate
output does not change, but there are important redistribution e®ects
between countries due to di®erent cash/deposit ratios, output of economy B is increased at the time of the shock, indicating that in relative
terms economy B is better o®. In the next period, however, aggregate
output decreases and also output of economy B drops, even though still
less than output of country A. This pattern can explain, at least to a
certain extent, di®erences in preferences for a tighter monetary policy
for countries of a B type (in relative terms economy B is better o® with
monetary tightening and worst o® with monetary expansion).
In order to isolate the e®ect of large di®erences in the cash/deposit
ratio we replicate the same exercise with the same ° for both countries. In other words, we concentrate on income and assets e®ects. If we
take again autarchic countries we see that, as in the case with di®erent
cash/deposit ratios, in economy A the output e®ect is larger (Figure 3).
18 The two main di®erences are in the degree of e±ciency of the banking sector approximated in our calibration by the parameter µ; substantially higher in Germany
than in France (the banking sector has a higher technological advantage with respect
to households and the transaction costs are lower, Cf also Rodriguez, 1998)- and in
the cash holding of domestic currency, i.e the parameter
23
°;again
higher for Germany.
When we consider the monetary union case, there is no redistribution
of consumption, but there are still output di®erences (Figure 4). Here
total output decreases but less than it would had been for an economy A
in isolation. In other words, endogenous preferences for monetary policy
may be diverse.
Figure 1b and 2b show the behavior of total consumptions, it increases at the time of a monetary contraction (as expected) because of
cash in advance constraints, to decrease in the subsequent periods. Total
consumption of economy A is more volatile, than that of economy B,
when economies are independent, but less volatile when countries joint
a monetary union. This is due to di®erences in ° (with equal ° the period t reaction is the same in both countries and, as in ¯gures 1b-2b, the
-negative- e®ect is more persistent in economy A).
Finally, ¯gure 5 shows how prices (i.e., gross in°ation ¼) react to a
monetary contraction, when countries are independent or in a monetary
union. As it can be seen, our model does not predict a \price puzzle",
as it has been observed in some European economies (see, for example,
Sims, 1992). Following a monetary contraction prices fall (the \puzzle"
being that it seems to ¯rst increase), to then experience a small increase,
before returning to their stationary level.
5
The calculated VAR for France and Germany
The issue of empirical testing the existence of possible di®erences in the
impact of monetary policy on output and prices in European countries
is not easy and far from having a de¯nite answer (see Dornbusch et Al,
1998). The case for asymmetric impact of a monetary shock is easy
to make, since, as we have documented in Section 2, there are marked
cross-country di®erences in the ¯nancial structure (e.g. mix of direct
and indirect ¯nance, share of ¯xed and variable rate contracts, degree
of indebtedness etc.). But di®erences in the ¯nancial structure do not
translate easily into clear-cut results and, in any case, they prompt forces
24
which are likely to o®set each other (see Gennari and Giovannetti, 1998
for a discussion). Furthermore, EMU represents a change in regime, di±cult to account for properly (we are back to \Lucas's critique"). Against
this background, many studies have tried to identify cross-country differences in monetary policy transmission19 , at least in the current set-up
(i.e. before the start of EMU), but no consensus seem to exist on the
extent or nature of possible di®erences. More precisely, it seems that
very di®erent results can be obtained for the same country using di®erent models and that the ranking of the strength of a common monetary
shock on output is not consistent across di®erent studies.
In what follows, we do not enter into the debate of what is the
best way to estimate the e®ect of a common monetary shock in di®erent
countries (see BIS, 1995) nor what is the best identifying scheme, even
though we are aware of the possibility of getting di®erent results when
using di®erent methods or identi¯cation schemes. Our aim is simply to
see whether the implications of our theoretical model are consistent with
the actual response of output to monetary shocks - i.e.e that di®erences
in the e±ciency of the ¯nancial structure and banking system a®ecting consumers' portfolio choices and ¯rms' ¯nancing re°ect in di®erent
output response to interest rates shocks, higher in the country with the
least e±cient banking system. To this aim, we estimated VAR, which
have the advantage of avoiding the need for a complete speci¯cation of a
structural model20 . In principle, to evaluate correctly the e®ects of monetary policy, we should solve an identi¯cation problem: policy actions
which are endogenous responses to current developments in the economy
must in fact be separated from exogenous policy actions. Only when the
19 Di®erent
methods have been used to this purpose:
national and multi-country
econometric models, structural VAR with their impulse response function, single
equation models among others.
Cf.
Britton and Whitley, 1997 for a comprehen-
sive survey; Dornbusch et al, 1998, and Ramaswamy and Sloeck,
1997 for estimation
on groups of countries.
20 There is an extens literature on pro and cons of VAR to study the transmission
mechanism of monetary policy well summarized in Christiano, Eichenbaum, Evans,
1998. Also there are problems related to the so-called price puzzle, pointed out by
Sims, 1992 and suggestions to include import price to avoid it and so on (see also
Bagliano e Favero, 1997).
25
latter are identi¯ed, the dynamic analysis of the VAR system can give
reliable information on the monetary transmission mechanism. In the
following, we use the Choleski decomposition21 for a parsimonious VAR
speci¯cation which includes 3 endogenous variables: output ( industrial
production), prices, and interest rates. We estimate the model over the
period 1973-9722 for France and Germany and here we only report the
impulse response functions23 , i.e. the responses of output, prices and
interest rates to unexpected shocks to interest rates.
All our variables are non-stationary (see Table 7): output (industrial production24 ) seems to be integrated of order 1 both in levels and
logarithms, while CPI is I(2) (i.e. in°ation is I(1)). Hence, we used
in°ation in our VAR estimates.
Table 7 here
For both countries, the VARs are speci¯ed with 2 lags; in our preferred speci¯cation, we add a trend, a set of orthogonal seasonal dummies
21 In
a 3 variable system this means that the last variable in°uences the ¯rst two,
without feedbacks from them and the second variable in°uences the ¯rst without
feedbacks from it.
22 Data
are described in the Appendix.
Unit root tests are done using PCGive.
Estimations are done using the package E-views, version 2.0 and PCFIM L.
corresponds to the longest available with fairly homogeneous data.
The period
We have also
reduced the period of estimation to consider only the ERM period (1979-97) and
results do not change. The same applies when exogenous variables are added to the
estimates, such as exchange rate developments, raw material prices etc. or dummies
to account for the 1992 ERM crisis and the 1993 enlargement of °uctuation bands.
23 Even though impulse responses are not a valid model selection criteria, because
they are determined by the chosen methodological framework in which a model is
built (i.e.
the imposed identifying restrictions, its speci¯cation and its estimation
method), they are widely used in the empirical literature because they easily convey the message and provide a simple graphical assessment of the di®erences in the
trasmission mechanism.
24 Industrial production is preferred to output in the empirical literature, and the
e®ects of a monetary shocks are more visible; however here we report the impulse
responseof output for consistency with our theoretical model where we have consumption and assets rather than production. Cf. Simms, 1992; Mojon, 1997 amongst
others for discussion on the use of industrial production and Gennari and Giovannetti,
1998, for VAR using the same data set and methodology but industrial production
instead of output.
26
(CPI are not adjusted) and some country dummy (for further details, see
Appendix and Gennari and Giovannetti, 1998).
Graph 6 and 7 show the impulse responses to a standardized monetary shock together with 95% con¯dence intervals. In Germany (output)
bottoms out about ten quarters after the contractionary shock, and a
similar pattern is observed in France. The numerical e®ect, however, is
higher in France (around -0.004 against -0.002) than in Germany25 . This
implies that Germany is partially protected from a monetary tightening (at the same time it bene¯t less from a monetary expansion), with
consequences on the preferences for monetary policy.
These results are very stable in the case of France. Di®erent measures of interest rates (Pibor, Tbill rate), di®erent sample sizes, inclusion
of a dummy variable to account for the EMS crises, inclusion of import
prices to deal with the price puzzle did not change the response of output in any dimension (shape, numerical size, lags) while impacting on
the price response to an interest rate shock26 .
As for Germany, however, the results seem to be more sensitive
to the sample size, most likely because German Uni¯cation represents a
change in regime which is di±cult to account for when the more recent
period has a higher weight27 . Interesting enough, when a shorter sample
25 If industrial production is used instead of output, the numerical values are respectively -0.008 for France and -0.005 for Germany, so the results are con¯rmed. When
using the output data a step dummy and an impulse dummy have to be included for
Germany in order to account for the break in the series due to German Uni¯cation.
26 As most of the empirical studies we have reviewed, we found weak evidence of
a price puzzle, i.e. a perverse response of prices. The inclusion of import prices in
the VAR reduces the positive response of prices to a monetary contraction, without
eliminating it completely. Clarida and Gertler (1996) provide two explanations for
the price puzzle: either the magnitude of an interest rate rise which represents a
policy shock is not strong enough to have a decreasing impact on in°ation, or there
is an identi¯cation problem in the sense that the Central Banks have additional news
about in°ation which are not captured by the model.
27 We replicated our exercise with the data set kindly provided by Ramaswamy
and Sloeck, 1997.
Again, the output response of Germany changes with di®erent
sample sizes while that of France is very stable. In particular the standard errors for
Germany become very large on a shorter sample. The price response, not reported in
27
size is selected (e.g. 1983-97), the e®ect of the shock on German output
is substantially weaker (not di®erent from zero) and the standard errors
much larger.
The e®ects of the shock are transitory in both countries but seem
to be slightly more persistent in France (as in the computed responses
from our model). Overall, the estimated impulse response for France and
Germany have a shape very similar to the theoretical response calculated
from our model: the e®ect of a monetary shock in the current situation
is di®erent in the two countries and has a bigger impact in France where
agents are more illiquid. The asymmetry can have important consequences for the behavior of the European Central Bank, at least up to
when(if) the ¯nancial structures in Europe will converge.
6
Conclusions
The focus of the debate on Monetary union has been so far mainly on
"real convergence". Real convergence is very important to achieve consensus on harmonization of policies in a MU. However, di®erences in
the ¯nancial structure can be crucial to achieve consensus on the -by
de¯nition- harmonized monetary policy in EMU. Cross-country e®ects
of a common monetary policy can be di®erent as a result of di®erences
in ¯nancial structures and in the transmission channels of monetary policy. While these issues are often discussed, they had not been appropriately modeled and quanti¯ed. This paper is novel in these regards. First
describes the underlying di®erences across the main EMU countries, second provides a theoretical model accounting for these di®erences, where
is possible to study the e®ects of an unexpected monetary shock, and,
third, provides new estimates of the e®ects of a monetary shock in France
and Germany. Such estimates are consistent with the predictions of the
theoretical model.
Our work suggests that there are possibilities of con°icts over monetary policy, at least in the early stages of EMU, even if countries' reptheir study, seems to be much worst.
28
resentatives in the ECB share the same principles over monetary policy and there are no cyclical di®erences across countries. It is to be
expected, however, that ¯nancial sectors will be progressively less segmented, which, according to our theory, will result in more homogeneous
e®ects. The European experience, since 1992, and the US experience,
shows, however, that such convergence may be slow. This can be important for the EMU since policy consensus will be crucial in its ¯rst
stage.
29
References
[1] Alvarez, F., A. Atkenson and P. Kehoe, 1996, Money and interest
rates in segmented assets markets, mimeo, July;
[2] Bagliano, F. and C. Favero, 1997, Measuring monetary policy with
VAR models: an evaluation, mimeo, May;
[3] Bank for International Settlements, 1995, Financial structure
and the monetary policy transmission mechanism, Basle,
March;
[4] Barran Fernando, Coudert Virginie and Benoit Mojon, 1996, the
transmission of monetary policy in the European countries, CEPII,
Paris, document de travail, 96-03;
[5] Bernanke Ben and Mark Gertler, 1995, Inside the Black box: the
credit channel of monetary transmission mechanism, Journal of Economic Perspectives, 9, pp.
[6] Bernanke B. and I. Mihov, 1995, Measuring monetary policy, NBER
wp 5145;
[7] Borio, Claudio, 1995, The structure of credit to the non-government
sector and the transmission mechanism of monetary policy: a crosscountry comparison, Bank for International settlements working paper, Basle, April.
[8] Borio, Claudio and Wilheim Fritz, 1995, The response of short term
bank lending rates to policy rates: a cross-country perspective, Bank
for International Settlements, Wp n. 27, Basle, May;
[9] Britton E. and J.Whitley, 1997, Comparing the Monetary Transmission Mechanism in France, Germany and the UK: some issues and
Results, Bank of England Quarterly Bulletin, May;
[10] CEPR, 1997, Options for the Future Exchange Rate Policy of the
EMU, mimeo, July 8, London;
30
[11] C. Carlstrom and T. Fuerst, 1995, Interest rate rules vs money
growth rules, Journal of Monetary Economics, 36, pp. 247-267;
[12] Chakravarty, S.P. and P. M. Molyneux, 1996, The Single European
Market in ¯nancial services, European Business Review, 96, 5;
[13] Christiano, L., 1991, Modeling the liquidity e®ect of a money shock,
Federal reserve Bank of Minneapolis Quarterly Review, Winter;
[14] Christiano, L. and M. Eichenbaum, 1992, Liquidity e®ects and the
monetary transmission mechanism, American Economic Review, vol
82, p346- 353;
[15] Christiano, Lawrence, Eichenbaum Martin and Charles Evans, 1996,
Sticky Price and Limited Participation Models of Money: A comparison, mimeo, October;
[16] Christiano, Lawrence, Eichenbaum Martin and Charles Evans, 1997,
Modelling Money, mimeo, paper presented in Perugia, January 1997;
[17] Christiano, Lawrence, Eichenbaum Martin and Charles Evans,1998,
Monetary Policy Shocks: What have we learned and to what end?
NBER WP 6400, February;
[18] Clarida R and M. Gertler, 1996, How the Bundesbank conducts
Monetary Policy, NBER WP 5581;
[19] Cottarelli, Carlo and A. Kourelis, 1994, Financial structure, bank
lending rates and the transmission of monetary policy, IMF working
papers, Washington, March;
[20] Davis, Paul, 1996, The role of institutional investors in the evolution
of ¯nancial structure and behavior, LSE Financial Market Group,
Special paper 89, November;
[21] Dornbusch R, C, Favero and F. Giavazzi, 1998, Immediate Challenges for the European Central Bank, Economic Policy, 26, April,
15-64 (including discussion);
31
[22] European Monetary Institute, 1994, Di®erences in monetary transmission processes in the European Union and policy implications,
interim report, December;
[23] European Economy, Supplement A, 1997, Advancing Financial Integration, n. 12, December;
[24] Fuerst T., 1992, Liquidity, loanable funds, and real activity, Journal
of Monetary Economics, 29, 3-24;
[25] Giovannetti G., 1997, Some evidence on the heterogeneity of the
¯nancial structure and the monetary transmission channels in Europe, mimeo, EUI, Florence, June;
[26] Gual J and D. Neven, 1993, Banking in European Economy, reports
and studies 1993 (3);
[27] Goodfriend M. and R. King, 1997, The neoclassical synthesis and the
role of monetary policy, mimeo, paper presented at Perugia, June;
[28] Kneeshaw J.T., 1995, A survey of non-¯nancial sector balance sheets
in industrialized countries: implications for the monetary policy
transmission mechanism, Bank for International settlements, working papers n. 25, Basle, April.
[29] Lucas, R. E.Jr., 1972, Expectations and the neutrality of money,
Journal of Economic Theory, 4, 103-124;
[30] Lucas, R. E. Jr., 1990, Liquidity and Interest Rates, Journal of
Economic Theory, 50, 237-264;
[31] Miron, J., C. Romer and D. Weil, 199., Historical Perspectives on
the Monetary Transmission mechanism, 263-306;
[32] Mishkin, Frederic, 1996, The channels of monetary transmission:
lessons for monetary policy, NBER working paper 5464, February;
[33] Ramaswamy R. and T. Sloeck, 1997, The Real E®ect of Monetary
Policy in the EU: What are the di®erences? IMF WP 97/160, forthcoming IMF Sta® Papers, 1998;
32
[34] Rodriguez Mendizabal H., 1998, Monetary Unions and the Transaction Cost Saving of a single Currency, UPF, Barcelona, May, mimeo;
[35] Seitz F., 1995, The circulation of Deutsche Mark abroad, DP 1/95
Deutsche Bundesbank, May;
[36] Sims C.A., 1992, Interpreting the macroeconomic time series facts:
the e®ects of monetary policy, European Economic Review, 36, 9751011;
[37] Taylor, John, the monetary transmission mechanism: an empirical
framework, Journal of Economic Perspectives, 9, pp.
33
7
Tables
Table 1: Structure of Financial markets, 1996
Market capitalization Trading volume
FIR BIR (% GDP)
% GDP
France
42.4 73.8 38.9
63.5
Germany 52.8 80.2 29.6
33.3
Italy
39.9 70.3 21.7
8.6
UK
na
na
149.9
66.1
EU
na
na
53.0
38.5
Source: Bundesbank, Monthly Report, January 1997 and European Economy,
Supplement A, Economic Trends, n.12, December 1997. FIR stands for Financial Intermediation Rate and is Financial assets of FI% total assets; BIR stands for Banking
Intermediation Rate and is Fin. assets of Banks%of Fin. Assets of Fin. sector.
Table 2: Financial assets of Households
(as proportion of gross ¯nancial assets)
1980
1994
Banks bonds equities inst.inv. Banks bonds equities inst. inv.
France .59 .09
Germany .59 .12
Italy
.58
.08
UK
.43
.07
Source: Davis, 1996.
.14
.04
.1
.12
.07
.17
.06
.3
.32
.45
.29
.26
.04
.14
.2
.01
.32
.06
.24
.12
Table 3: Corporate sector balance sheets, 1980 and 1994
(as proportion of gross ¯nancial assets)
1980
1994
bonds equit. loans. bonds equit. loans.
France 0.4 .34 .28 .03 .70 .28
Germany .02
.2
.52
.08
.25
.50
Italy
.04
.52
.43
.03
.46
.44
UK
.02
.37
.22
.001
.65
.12
Source: Davis, 1996 and OECD, Financial Statistics, various issues;
34
.29
.28
.09
.54
Table 4: Ownership of listed shares by sector,1995
France
Germany
Italy
UK
Households
Non ¯n. corp.
Public
tot non-¯n sector
Fin. Inst.
Foreign
19.4
14.6
17
29.6
58.0
42.1
32
4.1
3.4
4.3
28
0.2
80.8
61.0
77
33.9
8.0
30.3
19
52.4
11.2
8.7
5
13.7
Source: Deutsche Bundesbank Monthly Report, January 1997 and OECD Financial Markets Trends, November 1995.
Table 5: Monetary Aggregates, 1995
France
cash/M1
9.1
cash/M2
7.8
cash/M3
4.7
Per capita currency holding (US $)
850
Germany Italy UK
29.1
15.9 4.7
18.9
8.1
na
11.8
na
na
1983
1066 575
Sources: Banque de France, Banca d'Italia and Deutsche BundesBank, annual
reports and OECD Financial Statistics.
Table 6: Parameters of the simulations
°
R
µ1
µ2
µ3
R¡µ
!0 !1 ¯
Country A 8
6
.996 0.78 1.05 .018 .01 .012 1.046
Country B 8
6
.996 .189 1.05 .02. .005 .005 1.06
35
Table 7:Unit root tests
a) France
in° rate
gdp
import
prices
ind.
prod
call m rate
int.rate
3mth T.Bills
PIBOR
3 months
test
ADF(1)
ADF(1)
ADF(4)
ADF(2)
ADF(1)
ADF(1)
ADF(2)
ADF(2)
ADF(5)
statistic
-1.521
-2.788
-2.572
-2.18
-2.705
-2.628
-2.392
-2.385
-4.512**
levels
ADF(1)
-3.081
¢levels
DF
-6.651**
levels
ADF(1)
-3.073
¢levels
DF
-6.981**
lev
logs
lev
logs
lev
logs
lev
logs
lev
¢lev
¢logs
¢lev
¢logs
¢lev
¢logs
¢lev
¢logs
¢lev
test
ADF(1)
DF
ADF(3)
ADF(1)
DF
DF
ADF(1)
ADF(1)
ADF(5)
statistic
-4.806**
-3.968*
-3.45*
-4.493**
-5.249**
-5.646**
-4.178**
-4.285**
-3.651**
b) Germany
gdp
in° rate (cpi)
import
prices
w. mkt p.
(raw mat).
call m rate
LT int.rate
(7-15 y)
lev
logs
lev
logs
lev
logs
lev
logs
lev
test
ADF(1)
ADF(4)
ADF(1)
ADF(4).
ADF(1)
ADF(1)
ADF(3)
ADF(3)
ADF(5)
statistic
-2.116
-2.278
-1.521
-3.54*
-3.25
-3.55*
-3.49*
-3.66*
-2.077
¢lev
¢logs
¢lev
¢logs
¢lev
¢logs
¢lev
¢logs
¢lev
test
DF
ADF(3)
ADF(1)
DF
DF
DF
ADF(2)
ADF(2)
ADF(4)
lev
ADF(3)
-2.32
¢lev
ADF(2)
* signi¯cance at 5%, ** signi¯cance at 1%
36
statistic
-8.248**
-3.01*
{9.171**
-4.46**
-4.21**
-4.18**
-3.17*
-3.25*
-5.97**
-3.51*
Appendix I:Data Sources
Data are obtained from IFS and Analytical Database of the OECD.
The period used is 1973 ¯rst quarter, 1997 fourth quarter. Output is in
logs and is seasonally adjusted. The series on real GDP is de¯ned in
national currency and is obtained from the OECD database (GDPVol).
The series on consumer price index is obtained by IFS (n. 64 for each
nation). The nominal interest rate is the call money rate and is also
from IFS. We also used industrial production from OECD database, UN
commodity price index (IFS), DM- dollar exchange rate series (IFS),
French franc- DM exchange rate series (IFS). For Germany a step dummy
for GEMU was used (0-1) and an impulse dummy for changes in the mean
in 1991. For France a dummy accounting for the oil crisis and the ERM
crises.
37
Output response to a monetary contraction (Indep. co.)
1
0.999
0.998
0.997
0.996
0.995
0.994
0.993
0
2
4
Economy A: *
6
8
Economy B: + (log(y)/log(y*)
10
12
Consumption response to a monetary contraction (Indep. co.)
1.01
1.005
1
0.995
0
2
4
Economy A: *
6
8
Economy B: + (log(y)/log(y*)
10
12
Output resp. to a mon. contr., equal gammas (Indep. co.)
1
0.999
0.998
0.997
0.996
0.995
0.994
0.993
0
2
4
Economy A: *
6
8
Economy B: + (log(y)/log(y*)
10
12
Output response to a monetary contraction (Monet. Union)
1.002
1.001
1
0.999
0.998
0.997
0.996
0.995
0.994
0
2
4
6
8
10
Economy A: * Economy B: + MU: o (log(y)/log(y*)
12
Consumption response to a monetary contraction (MU)
1.015
1.01
1.005
1
0.995
0.99
0
2
4
6
8
10
Economy A: * Economy B: + MU: o (log(c)/log(c*)
12
Output resp. to a mon. contr., equal gammas (MU)
1.001
1
0.999
0.998
0.997
0.996
0.995
0.994
0.993
0
2
4
6
8
10
Economy A: * Economy B: + MU: o (log(y)/log(y*)
12
Price response to a monetary contraction (Monet. Union)
1.05
1
0.95
0.9
0.85
0.8
0.75
0.7
0
2
4
6
8
Economy A: * Economy B: + MU: o
10
12
Response to One S.D. Innovations ± 2 S.E.
Response of LGDPVOL to CALLMONEY
0.005
0.000
-0.005
-0.010
5
10 15 20 25 30 35 40 45 50 55 60
Response of INFL to CALLMONEY
0.0020
0.0015
0.0010
0.0005
0.0000
-0.0005
-0.0010
-0.0015
5
10 15 20 25 30 35 40 45 50 55 60
Response of CALLMONEY to CALLMONEY
1.5
1.0
0.5
0.0
-0.5
-1.0
5
10 15 20 25 30 35 40 45 50 55 60
Response to One S.D. Innovations ± 2 S.E.
Response of LGDPVOL to CALLMONEY
0.005
0.000
-0.005
-0.010
5
10 15 20 25 30 35 40 45 50 55 60
Response of INFL to CALLMONEY
0.0020
0.0015
0.0010
0.0005
0.0000
-0.0005
-0.0010
-0.0015
5
10 15 20 25 30 35 40 45 50 55 60
Response of CALLMONEY to CALLMONEY
1.5
1.0
0.5
0.0
-0.5
-1.0
5
10 15 20 25 30 35 40 45 50 55 60