8/9/2019 gemoba-3b

1/28

Money and Banks: Some Theory and Empirical

Evidence for Germany

Oliver Holtemoller

November 2002

Abstract

This paper contributes to the analysis of the money supply process in Germany during

the period of monetary targeting by the Bundesbank from 1975-1998. While the stan-

dard money multiplier approach assumes that the money stock is determined by the money

multiplier and the monetary base it is argued here that both the money stock and the mon-

etary base are determined endogenously by the optimizing behavior of commercial banks

and private agents like households and firms. An industrial organization style model for

the money creating sector that describes the money creation process is developed assuming

that the main policy variable of the central bank is the money market interest rate. A vector

error correction model for the nominal money stock, the monetary base, nominal income,

short-term and long-term interest rates, and the required reserve rate is specified, and the

interaction between these variables is analyzed empirically. The evidence contradicts the

money multiplier approach and supports the presented model of the money creating sector.

JEL Classification: C32, E51, E52

Keywords: Industrial organization approach to banking theory, money multiplier, endoge-

nous money, vector error correction model

This paper is partially based upon the second chapter of the authors doctoral dissertation (Vector au-

toregressive analysis and monetary policy, Aachen: Shaker, 2002). This version will be published with the

title Money Stock, Monetary Base and Bank Behavior in Germany in Jahrbucher fur Nationalokonomie und

Statistik (2003), forthcoming. I thank Helmut Lutkepohl and Jurgen Wolters for helpful comments. Financial

support from the Deutsche Forschungsgemeinschaft (SFB 373) is gratefully acknowledged.

8/9/2019 gemoba-3b

2/28

1 Introduction

The role of nominal money in the monetary policy transmission process is still an open

question. In contemporaneous macroeconomic models that follow a general-equilibrium

approach, money is often determined endogenously by a money demand relation without

having a direct impact on real variables. The important equations for the development of the

real sector in this type of models are an aggregate demand equation, the term structure of

interest rates, an inflation-adjustment equation and a monetary policy reaction function, see

for example Walsh (1998). On the other hand, money is highly relevant in practiced mon-

etary policy strategies and in empirical macroeconomic models. The growth rate of money

plays a prominent role in the first pillar of the European Central Banks monetary policy

strategy (European Central Bank, 1999a,b), and money growth has been the key element of

the Bundesbanks monetary targeting strategy from 1975 to 1998 (Deutsche Bundesbank,

1995). Econometric models of monetary policy transmission, in which money is explicitlyconsidered, are Bruggemann (2001) and Lutkepohl and Wolters (2001) for Germany; an

overview of monetary policy transmission models for the Euro area is given in Angeloni

et al. (2002). Before analyzing the role of money in the monetary policy transmission

mechanism, it is necessary to investigate the money supply process. Two conflicting views

of money supply can be found in the literature. The older one is the money multiplier ap-

proach saying that the money stock is determined by the money multiplier and the monetary

base. In general it is assumed in this framework that the monetary base is controlled by the

monetary authority. Under certain conditions this implies that the monetary authority can

also control the money stock such that money is exogenous in the sense that it is policy-

determined. The other view is the so-called new view which stresses the importance of

commercial banks in the money supply process. According to this view, money is endoge-

nous in the sense that the money stock is not determined by a monetary policy authority but

is the result of the optimizing behavior of commercial banks and private agents given the

money market conditions set by the monetary policy authority. The purpose of this paper is

to analyze the theoretical implications of the exogeneity view and the endogeneity view of

money and to compare these implications to empirical evidence for Germany in the period

of monetary targeting from 1975 to 1998.

The paper is structured as follows. In section 2, the money multiplier approach is dis-

cussed, and the development of the money stock in Germany from 1975 to 1998 is analyzed

under consideration of the monetary policy strategy of the Deutsche Bundesbank. After-

wards, the new view of money supply is briefly reviewed. In section 3, an industrial

organization model of the money-creating sector with endogenous money is developed. In

section 4, a vector error correction model (VECM) for the nominal money stock, the mon-

1

8/9/2019 gemoba-3b

3/28

etary base and related variables is estimated and analyzed. The empirical results are com-

pared to the implications of both the money multiplier approach and the model presented in

section 3. Finally, section 5 concludes.

2 The Money Supply Process in Germany from 1975 to 1998

2.1 The Money Multiplier Approach and the New View of Money Supply

The standard textbook approach explaining the money stock outstanding and its growth

rate, is the money multiplier model. Many versions of this model are in use. They have in

common that the money stock (M) is determined by the monetary base (or high-powered

money,H) and the money multipliermm:1

M=mm H. (1)

The monetary base is controlled by the central bank, and the money multiplier depends on

the behavior of the public (constant currency-deposit ratio, d = CU/D), the commercial

banks (reserve ratio as a function of interest rates and uncertainty), and the central bank

(minimum reserve requirements). These behavioral determinants enter the money multiplier

in a nonlinear way. The simplest version of the money multiplier is the following one: The

money stock consists of currency in use (CU) and deposits (D):M=C U+ D= mm H;

and the monetary base consists of currency in use and reserves of banks (R):H=C U+ R.

The (required) reserve rate isr = R/D such that

mm= CU/D+ 1CU/D+ R/D

= d + 1d + r

. (2)

If the central bank is able to forecast the money multiplier correctly and is also able to

control the monetary base it can control the money stock. Under these circumstances, the

money stock is an exogenous variable assuming that the supply of deposits by the public is

not restricted such thatD = R/r. Exogeneity of the money stock in this context means the

ability of the central bank to control the money stock.

The money multiplier approach has some important drawbacks: First, the operating tar-

get of central banks in the USA and in Europe is not the monetary base but a money market

interest rate (federal funds rate, euro overnight index average EONIA). A theory of money

supply has to consider this and other institutional details. Second, according to studies of the

relationship between the money stock and the monetary base in Germany by Willms (1993)

and by Nautz (1998), a stable relationship between the money stock and the monetary base

seems not to exist, see also figure 1. The increase of the money multiplier from about 4.5 Insert fig-

ure 1 about

here.

1The money multiplier approach is explained in many macroeconomic textbooks. This description benefits

fromDornbuschand Fischer(1994).

2

8/9/2019 gemoba-3b

4/28

to 7.1 shows that the share of the monetary base in M3 has decreased, and that book money

created by commercial banks has become more and more important. Therefore, it may be

appropriate to model the behavior of banks explicitly instead of reducing it to variations of

the money multiplier.

In the terminology of Tobin (1967), the money multiplier approach is the old viewof money supply, while the new view interprets financial intermediaries as firms which

optimize their portfolios given the optimizing behavior of non-banks. That is, financial in-

termediaries do not possess the ability to expand deposits without limit like it is assumed

in the money multiplier approach above. Thus, the amount of deposits and the money

stock are endogenous variables determined by the portfolio selection process of commer-

cial banks and the public. Corresponding to the use of the notion of exogeneity in this paper,

endogeneity of money in this context means that the central bank is not able to control the

money stock.2 While optimizing their portfolio, banks and non-banks have to consider the

conditions set by the central bank. Advocates of the money multiplier approach refuse the

new view, and find no reason to look beyond the balance sheets of commercial banks.

(Meltzer, 1969, p. 39). Albeit weaker and not as explicit as in this quotation, this view can

also be found in a more recent work (Meltzer, 1995). The money multiplier approach is also

supported by Rasche (1993) who admits that the algebraic components of the money mul-

tiplier, however formulated, vary in response to the economic decisions of both depository

institutions and the public (p. 32) but claims that the variations of the money multiplier

are unsystematic, and of short-run nature (p. 47): Over the longer run, such random move-

ments tend to average out, so that changes in base money are the most important source of

changes in transactions money.

2.2 Monetary Targeting in Germany from 1975 to 1998

The exogeneity or controllability assumption of the money multiplier approach forms the

basis of the monetary policy strategy of monetary targeting.3 This strategy has been adopted

by the Deutsche Bundesbank from 1975 to 1998. In Deutsche Bundesbank (1995, p. 91 ff.)

it is described how the Bundesbank has tried to control the money stock. The Bundesbank

refers implicitly to the money multiplier approach and states that its monopoly for bank

notes and the minimum reserves requirement imply long-run controllability of the money

stock by means of controlling the monetary base.

A precondition for an exogenous money stock is a flexible exchange rate. This precon-

2See alsoM uller(1993); for an analysis of endogeneity, causation, and their relation see M uller(1998).3The money multiplier approach does not necessarily imply exogeneity of the money stock. If the money

multiplier exhibits unpredictable and endogenous variations, the money stock is endogenous. For reasons of

simplicity, it is supposed here that the money stock is exogenous in the money multiplier approach. A money

multiplier model with endogenous money can be found in Jarchow(1998), for example.

3

8/9/2019 gemoba-3b

5/28

dition has not been given due to the more or less fixed exchange rates within the European

Monetary System (EMS, founded 1979), but the Bundesbank has been able to sterilize in-

terventions on the foreign exchange market like it has been the case in the EMS crisis of

September 1992. Nevertheless, as can be seen from figure 2, the Deutsche Bundesbank was

not always able to achieve its announced monetary target. In the 24 years of monetary tar-geting from 1975 to 1998, the observed monetary growth rate deviated from the announced

growth rate eleven times. Among others, there are two possible reasons for a deviation of Insert fig-

ure 2 about

here.

the money growth rate from the announced target: first, the Bundesbank has also had other

objectives. The money growth rate has not been an ultimate goal but only an intermediate

target. The ultimate goal has been price stability measured in terms of the inflation rate. In

some periods there may have been trade-offs between the announced monetary target, the

price stability objective, and other objectives, like the exchange rate. Second, money could

be endogenous. That is, the central bank is not able to set the money growth rate as the

money multiplier approach or exogeneity view suggests. In a modern open economy with a

sophisticated profit-maximizing banking system, a non-banking financial sector, and rapid

international capital flows, it is at least questionable whether money is exogenous, see also

Desai (1992).

The endogeneity view is supported by the explanations of the Bundesbank for the dif-

ferences between announced target and observed money growth rate since 1992/93. Before

1992/93, the explanations of the Bundesbank for deviations from the money growth rate tar-

get were reasons for a more expansive or more restrictive monetary policy than announced.

That is, the actions of the Bundesbank have been responsible for the deviations of the ex-

ogenous money growth rate from the announced target. Since 1992/93, the explanations

refer to unforeseen changes in the demand for money implying that the endogenous money

stock has been determined by the demand for money. The following explanations have

been given for deviations:4 From 1975 to 1978, the money growth rate was higher than

the announced target. The reason was a policy of low interest rates in order to increase the

low level of real economic activity. The reasons for the excess money growth from 1986

to 1988 have been the stabilization of exchange rates (DM/US-Dollar, EMS) and provision

of liquidity to avoid a recession after the stock market crash in October 1987. In 1993, a

flight into currency and into short-term deposits has been the result of the introduction ofa withholding tax on interest yields. Therefore, the demand for money increased. In 1995,

the only year with a lower money growth rate than the announced target, the demand for

money decreased as a consequence of the permission of money market fund shares which

have not been part of M3. And in 1996, the money growth rate was too high because of

interest rate driven portfolio variations from money capital to time deposits. According to

4The explanations of the monetary development are taken from von Hagen(1998) andBaltensperger(1998).

4

8/9/2019 gemoba-3b

6/28

these explanations, at the end of the period of monetary targeting in Germany, demand side

forces have been the reasons for deviations of the money growth rate from the announced

target. This supports the endogeneity view.

3 An Industrial Organization Model of the Money-Creating Sector

In the following, the so-called (Freixas and Rochet, 1997) industrial organization approach

to banking theory is applied to model the behavior of commercial banks. Central elements

are submodels of the credit market and the deposits market. Reduced form equations for the

quantity of loans and the quantity of deposits are developed and inserted into the aggregated

balance sheet of commercial banks. As a consequence, the monetary base and the money

stock are endogenous. Before the details of this model are explained, section 3.1 gives a

brief overview of the industrial organization approach to banking theory.

3.1 The Industrial Organization Approach to Banking Theory

The definition of a bank in banking theory is mainly the legal definition of a commercial

bank in the United States of America (U.S. Banking Act of 1971): banks are financial

intermediaries that receive (demand) deposits and originate loans. The models that analyze

assets and liabilities as well as its possible dependencies can be divided into subsets.5 One

subset is the industrial organization approach. In this subset, a special focus is laid on the

structure of the banking market and the competition between banks. The banks are modeled

as optimizing agents on the market for loans and the market for deposits. The optimizing

behavior is modeled as expected profit maximization, that is banks are risk neutral. This

approach can be extended with assumptions about the cost function. Baltensperger (1980)

reviews models that consider the costs of real resources, especially labor, and Bofinger

(2001) uses a quadratic cost function to model credit default risk. Another subset of models

uses the theory of portfolio selection, where banks are assumed to be risk averse, see for

example Freixas and Rochet (1997, chapter 8).

Two articles, Klein (1971) and Monti (1972), build the basic setup of the industrial

organization approach to banking. Three markets (bonds, loans, deposits) and three types of

agents (central bank, commercial banks, households/firms) form the main structure. Whilethe bond market is assumed to be perfectly competitive implying that there is only one

interest rate for bonds, the loan and the deposits markets are not perfectly competitive. The

profit function of an individual banking firmn is:

n= iSSn+ i

LnLn i

Dn Dn C(S,L,D) , (3)

5See for example Santomero (1984), Bhattacharya and Thakor (1993), and Freixas and Rochet(1997) for

overviews of banking theory.

5

8/9/2019 gemoba-3b

7/28

where Sn, Ln, and Dnare the holdings of bonds (securities), loans, and deposits and iS, iLn ,

iDn are the respective interest rates. The bank decision variables are the interest rate on de-

posits, the quantity of loans and the quantity of bonds in Klein (1971) and the interest rate on

deposits, the interest rate on loans, and the quantity of bonds in Monti (1972). The cost func-

tion is not modeled explicitly. If the demand for loans by firms and the supply of deposits byhouseholds are specified, the first order conditions of the profit maximization problem can

be used to determine the stocks of bank assets and liabilities and the corresponding interest

rates. If the demand for loans and the supply of deposits are independent of each other, the

cross derivatives of the cost function are zero (2C/LD = 2C/DL = 0), and if no

further assumptions about the three markets (loans, deposits, bonds) are made, the decision

problem of the bank can be divided into two independent problems: the optimal choice of

interest rate/quantity on the loan market and the optimal choice of interest rate/quantity on

the deposits market.

Freixas and Rochet (1997, p. 60) formulate this model with the quantities of loans and

deposits as decision variables and interpret it as a model with imperfect competition with

two limiting cases:N = 1(monopoly) andN+ (perfect competition), whereN is

the total number of commercial banks.

The Monti-Klein model has been expanded in several ways. Dermine (1986) adds bank-

ruptcy risk and deposit insurance and Prisman et al. (1986) analyze uncertainty and liquidity

requirements. The result of both papers is that the separability result for loans and deposits

breaks down.

In the following, a version of the Monti-Klein model with interest rates on loans and

deposits as decision parameters of banks is used.

3.2 The Model of the Money-Creating Sector: Market Participants

The market participants in this model of the money-creating sector are the central bank,

commercial banks and private agents (households/firms).

The behavior of the central bank is exogenous. It fixes the interest rate on the money

marketi. The balance sheet of the central bank consists of central bank credit (CB C) on

the assets side and currency in use (CU) and reserves (R) on the liabilities side:

CBC=C U+ R. (4)

The commercial banks are profit-maximizing firms on an oligopolistic banking market.

Banks buy loans (L) and sell deposits (D). Their profit function is

n = iLnLn i

Dn Dn iCBCn + i I BPn, (5)

6

8/9/2019 gemoba-3b

8/28

whereiLn denotes the interest rate of bankn on loans, iDn the interest rate on deposits, and

IBP the net position on the interbank money market. The balance sheet equation of a

commercial bank is

Rn+ Ln+ IB Pn = Dn+ CB Cn (6)

such that the net position can be defined as

IBPn = Dn (1 r) + CB Cn Ln (7)

when it is assumed that banks only hold required reserves ( Rn= r Dn). The profit function

can also be written as

n = iLnLn + i (1 r) Dn i Ln i

Dn Dn. (8)

Commercial banks are price setters and quantity takers on the credit and the deposits mar-

ket. This type of simultaneous Bertrand competition between banks is discussed in Yanelle

(1988, 1989). One problem of simultaneous Bertrand competition is the existence of a

competitive equilibrium because a monopolist in one market automatically becomes a mo-

nopolist in the other market (Yanelle, 1988, p. XV). If a single commercial bank offers

a higher interest rate on deposits than all other commercial banks, it gets all deposits and

becomes also a monopolist on the credit market. But in a model with a central bank that

offers high powered money at a fixed interest rate there is a second refinancing possibility

for commercial banks besides deposits. On this market (the money market), the commercial

bank is a price taker.

The third group of agents are the households and firms. They have linear demand func-

tions for loans of every bankn:

Ln = 0+ 1 iLn + 2 (i

Ln i

Ln) + 3 Y (9)

with

0, 2, 3> 0, 1< 0

and for deposits of every bankn:6

Dn = 0+ 1 iDn + 2 (iDn iDn) + 3 Y (10)

with

0, 1, 3> 0, 2< 0.

6This relationship could also be called supply of deposits. The notion demand for deposits has been chosen

in analogy to the credit market.

7

8/9/2019 gemoba-3b

9/28

8/9/2019 gemoba-3b

10/28

Inserting this interest rate into the demand for loans yields

Ln = 0+ 1iL + 3Y

= 0(1 2)

21 2+

21 12

21 2i +

(1 2)321 2

Y. (17)

The aggregated quantity of loans is

L= c0+ c1 i + c2 Y, (18)

where

c0 > 0, c1< 0, c2> 0

are the coefficients of the expression for the quantity of loans of a single bank (17) multi-

plied byN, the number of commercial banks. The quantity of loans depends on the money

market interest rate and on the income of the households.

3.4 Equilibrium on the Deposits Market

The solution concept on the deposits market is the same as on the credit market. Commercial

banks are price setters and quantity takers. The first derivative of the profit function (8) of

bankn with respect to the interest rate on the deposits market is:

niDn

= i (1 r) Dn

iDnDn i

Dn

DniDn

= i (1 r) (1 2) (0+ 1iDn + 2(i

Dn i

Dn) + 3Y)

iDn (1 2). (19)

We get the following first order condition:

(0+ 2iDn) + (1 2)i(1 r) 2(1 2)i

Dn 3Y = 0 (20)

and the second order condition is satisfied:

2n

iDn2

= 2(1 2)< 0. (21)

In analogy to the credit market, every bank sets the same interest rate on deposits:

iD

= 0 (1 2)i(1 r) + 3Y

21 2(22)

and the demanded quantity of deposits for every bank is:

Dn = 0+ 1iD + 3Y

= 0(1 2)

21 2+

21 12

21 2i(1 r) +

(1 2)321 2

Y.

9

8/9/2019 gemoba-3b

11/28

The aggregated quantity of deposits

D= d0+ d1 i(1 r) + d2 Y (23)

with

d0> 0, d1> 0, d2> 0depends on the money market interest rate iand on incomeY. The coefficients diare again

the coefficients of the expression for the individual quantities multiplied byN.

3.5 Monetary Base and Money Stock

The aggregated balance sheet of the commercial banks is:

Nn=1

(Rn+ Ln+ IB Pn) =Nn=1

(Dn+ CBCn) .

WithRn = r DnandNn=1

IBPn= 0follows

Nn=1

Ln =Nn=1

Dn r Nn=1

Dn+Nn=1

CB Cn

or

H CB C=L (1 r) D= c0+ c1i + c2Y (1 r)(d0+ d1i(1 r) + d2Y). (24)

The monetary base H equals central bank credit and it depends on the quantity of loans

L =Nn=1

Ln, the quantity of deposits D =Nn=1

Dn and the required reserve rate r .

The monetary base is endogenous. The quantity of central bank lending is a result of the

profit maximization of the commercial banks.

Aggregating the balance sheet of the banking sector yields the money stockM:

L= C U+ D M. (25)

The money stock equals the aggregated quantity of loans, that is, it is determined by its

counterparts. In a more detailed model of money supply, the other counterparts of the money

stock which are neglected here, could be modeled, too. Neglected major counterparts are

net foreign assets of the banking system (F A) and non-monetary liabilities of the bankingsystem (N ML):

M = L + F A NM L O

(2262.1) (5247.7) (262.7) (3005.2) (243.0)

Numbers in parentheses are 1998 averages of German M3 and its counterparts in billions of

DM, andO denotes other counterparts.7

7The data is taken from the monthly bulletin of the Deutsche Bundesbank, February 1999, table II.2.

10

8/9/2019 gemoba-3b

12/28

3.6 Comparison of Money Multiplier Approach and Money Supply Endogeneity

In the money multiplier approach, money is determined by the monetary base and the money

multiplier. The monetary base is set exogenously by the central bank, and under the assump-

tions made here, money is exogenous. In the theory of endogenous money on the other hand,

money equals credit demand and the monetary base is a result of the optimal behavior of

commercial banks and households.

According to the money multiplier approach, changes in income and changes in the

money market interest rate should not cause changes in the money stock. However, this

statement only holds in the very simple money multiplier model of section 2.1. In more

sophisticated models, the money multiplier depends also on interest rates and on income.

Therefore, the impact of interest rates and income on the money stock cannot be used to

distinguish between the two approaches. However, the effects of changes in the required

reserve rate on the monetary base and on the money stock are different in both concepts.

Whereas an exogenous monetary base does not depend on changes in the required reserve

rate, there is a positive effect on the monetary base in the model of the money-creating

sector presented in this section:

H

r =d0+ 2d1i(1 r) + d2Y >0. (26)

A higher required reserve rate causes a decrease in the quantity of deposits but does not

affect the quantity of loans. The banks ask for more central bank credit to finance the

loans.8

The reaction of the money stock on changes in the required reserve rate is different, too.

The money multipliermmdepends negatively on the required reserve rate, and so does the

money stock, see equation (2). In the model of the money-creating sector with endogenous

money, the money stock does not depend on the required reserve rate.

A money multiplier equation can also be written in the endogenous money framework:

M

H =

c0+ c1i + c2Y

c0+ c1i + c2Y (1 r)(d0+ d1i(1 r) + d2Y)=mm(Y,i ,r). (27)

The interpretation of (27), however, is different from the interpretation of (2). A discussion

of the money multiplier approach, the counterparts approach and their relation can also befound in Artis and Lewis (1990).

8Regardless of the considered theoretical model, the monetary base will always increase if the required

reserve rate is increased, at least in the very short run. This is due to the definition of the monetary base which

is the sum of currency in use and reserves. In the money multiplier approach, however, the required reserve rate

and the monetary base are assumed to be more or less independent policy variables.

11

8/9/2019 gemoba-3b

13/28

4 The Econometric Model

4.1 Data and Unit Root Tests

Quarterly data for Germany from 1975-1998 is used in the econometric analysis. The vari-

ables are denoted as follows: m is the logarithmic money stock M3, h is the logarithmicmonetary base,y is logarithmic gross domestic product in current prices,s is a short-term

interest rate, is a long-term interest rate, and r is the average required reserve rate. The

data is not adjusted for the German unification in 1990 and not seasonally adjusted. Fur-

ther details can be found in the data appendix. The calculations have been performed with

EViews 4.1 and with Mathematica 4.0.

Unit root tests show that the endogenous variables can be assumed to be integrated of

order one, see table 1. The mean shift inm, h, and y is considered in the unit root tests

using a modified Dickey-Fuller type unit root test proposed in Lanne et al. (2002). The

results are quite robust to variations of the number of included lagged differences in the test

regression.

Table 1: Unit Root Tests, Sample Period: 1975:1-1998:12

Variable Lags Statistic Det. Terms Test

m 1 1.38 c, sd, t LLSm 1 10.37 c, sd, di ADFh 1 1.75 c, sd, t LLSh 1 6.96 c, sd, di ADFy 5 2.12 c, sd, t LLSy 4 4.32 c, sd, di ADF 1 1.44 c ADF 1 6.50 c ADFs 2 2.40 c ADFs 1 4.33 c ADFr 1 2.14 c, t ADFr 1 7.82 c ADF

Notes: Variable names are described in the text. ADF is the Augmented Dickey-Fuller Test and LLS is the

test (

+

int) proposed by Lanne et al. (2002). The LLS test is used for time series that exhibit a structural break(mean shift) due to the German unification. Deterministic terms are included as indicated in the fourth column:

constant (c), seasonal dummies (sd), linear trend (t), impulse dummy (di). Three asterisks denote significance

at the 1% level.

12

8/9/2019 gemoba-3b

14/28

4.2 Cointegration Analysis

An econometric framework that is suited to analyze the relationships between integrated

variables is the vector error correction model (VECM):9

xt = t+ xt1+k1i=1

ixti+ ut, (28)

wherextis an(p 1)-vector of endogenous variables, here xt = (mt, ht, yt, t, st, rt) and

thus p = 6; ut N(0, u) is a p-dimensional error process, andt contains determin-

istic terms. Here a constant, centered seasonal dummies (sdit), and impulse dummies are

included:

t = 0+3

i=1

sdi sdit+k1i=0

didti, (29)

wheredt is an impulse dummy variable that is one in the second quarter of 1990 and zero

otherwise. The matrix can be decomposed into a (p r)-adjustment matrix and a

(p r)-matrix containing the cointegration relations: =. Here,r denotes the cointe-

gration rank.10

The econometric model (28) differs from the theoretical specification in the following

way: first, in addition to the money market interest rate st, the long-term interest ratet is

included such that the empirical analysis can be compared to existing money demand stud-

ies. Second, it is assumed that the relations, which determine money stock and monetary

base, can be approximated by log-linear simplifications.

The lag length k is determined applying information criteria and set to two, k = 2,according to the Hannan-Quinn criterion (HQ). A cointegration rank of two is imposed in

the following implying that two stationary linear combinations of the variables exist. This

is supported by a cointegration rank test (Johansen, 1995), summarized in table 2. At a sig-

nificance level of 5%, the hypotheses of at most zero and at most one cointegration relation

are rejected while the hypothesis of at most two cointegration relations is not rejected. A

cointegration rank of two is also compatible with the theoretical considerations of section

3.

Identification of the cointegration vectors can be achieved by imposing linear restric-

tions:

= (H11, H22), (30)

9The econometric analysis of vector error correction models is for example described in Johansen(1995),

Lutkepohl(1993), andL utkepohl(2001).10r denotes the cointegration rank and rt the required reserve rate. Though this may be confusing, this

notation is quite usual in the literature.

13

8/9/2019 gemoba-3b

15/28

Table 2: Cointegration Tests, Sample Period: 1975:1-1998:12

H0 H1 LRtrace

r= 0 r >0 116.12

r 1 r >1 77.25

r 2 r >2 42.87r 3 r >3 18.56r 4 r >4 10.12r 5 r >5 4.11

Notes: The table shows the Johansen trace statistic used to test the hypothesis of at most r cointegration

relations. The lag length of the VAR in levels is k = 2, an unrestricted constant, seasonal dummies and impulse

dummies as in equation (29) are included. symbolizes rejection of the null hypothesis at a significance level

of 5%.

where the matricesi contain the unrestricted estimates andH1and H2are defined as

H1

=

1 0 0 0 0 0

0 0 1 0 0 0

0 0 0 1 0 0

0 0 0 0 1 0

0 0 0 0 0 1

and H2

=

1 1 0 0 0 0

0 0 1 0 0 0

0 0 0 1 0 0

0 0 0 0 1 0

0 0 0 0 0 1

. (31)

The restricted cointegrating relations can now be described by the two equations:

mt = 31yt 41t 51st 61rt + ec1t

mt ht = 32yt 42t 52st 62rt + ec2t,(32)

where ec1t andec2t are the error correction terms which can be interpreted as deviations

from the long-run equilibria. These restrictions satisfy the rank criterion given in Johansen

and Juselius (1994, Theorem 1) such that they are identifying:

rk(H1

H2) =rk(H

2H1) = 1 1. (33)

The just identified cointegration vectors are given in the upper part of table 3. This identi-

fication scheme implies that the variablesyt,t, st andrt are not cointegrated, which can

be confirmed by testing for the cointegration rank of(yt, t, st, rt). The hypothesis that the

cointegration rank is at most zero is not rejected (LRtrace = 40.86, 5% critical value: 47.21).

The required reserve rate measures a policy variable that is assumed to be exogenous,

that is, it is assumed that the central bank does not follow a reaction function for setting

the required reserve rate. Therefore, the model is now conditioned on the required reserve

rate in order to get a slightly more parsimonious model. This transformation into a partial

model is supported by testing for weak exogeneity of the required reserve rate: imposing

61 = 62 = 0 in the unconditional model gives a LR test statistic of 0.79 (p-value:

14

8/9/2019 gemoba-3b

16/28

Table 3: Long-Run Relations

m h y s r

just identified cointegrating vectors

1 0 0.95 5.14 2.76 3.971 1 0.59 10.42 7.98 16.34

overidentified cointegrating vectors, partial system

1 0 1.13 3.63 1.00 0

(0.02) (0.78) (0.43)

1 1 0.40 8.99 6.44 12.25

(0.08) (1.78) (1.08) (1.47)

Notes: Reduced rank estimation of the VECM (28) as described in Johansen (1995) and Lutkepohl (2001),

the cointegration rank is two, the number of lags is two. Centered seasonal dummies and impulse dummies

according to (29) are included. The estimates reported are the entries of. Asymptotic standard errors in

parentheses. The sample period is 1975:1-1998:4.

0.67).11 Additionally, it cannot be rejected that the coefficient of the required reserve rate

in the first cointegration relation is zero (LR test statistic of 2.07, p-value: 0.15). The

overidentified cointegration relations in the partial system are given in the lower part of

table 3. Abstracting from deterministic terms they can be interpreted as a nominal money

demand function

mt= 1.13 yt 3.63 t+ 1.00 st+ ec1t (34)

and a money multiplier relation (recall thatm,h andy are measured in logarithms)

mt ht = 0.40 yt 8.99 t+ 6.44 st 12.25 rt+ ec2t. (35)

These two long-run relations imply that

ht = 0.73 yt+ 5.36 t 5.44 st+ 12.25 rt+ (ec1t ec2t). (36)

This is a long-run relation between the monetary base and the variables that determine it

in the theoretical model of section 3 (yt, st, and rt). It can be seen that all coefficients

including the required reserve rate have the expected signs.

The money demand function (34) cannot be compared directly to other money demand

studies for Germany. While (34) is a demand function for nominal money, most other stud-

ies focus on real money balances. The general result that a stable money demand function

can be specified for the unified Germany is among others also supported by Wolters et al.

(1998), Bruggemann (2001), and Lutkepohl and Wolters (2001). However, a stable rela-

tionship between the monetary base and the broad money stock M3 has not been found in

11A variable is said to be weakly exogenous if its adjustment coefficients in front of all cointegration relations

(ec1,t1 and ec2,t1) are zero: i1 = i2 = 0. For a discussion of (weak) exogeneity and cointegration see

Ericssonet al. (1998), partial VECMs are discussed inHarboet al. (1998).

15

8/9/2019 gemoba-3b

17/28

other studies. The reason is presumably that the impact of the required reserve rate is mod-

eled explicitly here while Willms (1993) and Nautz (1998), for example, do not include the

required reserve rate in the model but use a monetary base that is adjusted for changes in

the required reserve rate.

4.3 Adjustment and Short-Run Dynamics

The cointegrating vectors do not show the complete information about the relationships

between the variables in the system. It is still an open question which variables react on

deviations from the long-run equilibria and how innovations or shocks affect the variables.

When the cointegrating vectors are known, the other parameters of the VECM can

be estimated by OLS. The adjustment to the long-run equilibria can be characterized by the

adjustment parameters that are stored in the matrix . These parameters are the coefficients

ofec1,t1 and ec2,t1 in the equations for the first differences. The adjustment parameters

are summarized in table 4, where the results of tests on weak exogeneity for each variable

can be found, too. These tests suggest that both the long-term and the short-term interest

rates are weakly exogenous. Nominal money, the monetary base, and nominal income

adjust in direction of the long-run equilibria if equilibrium-deviations occur and are rejected

to be weakly exogenous. The table shows also the results of some diagnostic tests for

the equations of the VECM. The remaining serial correlation in the monetary base and

income equations disappears if the lag length is increased. However, this does not change

the qualitative results such that the parsimonious specification with two lags can still be

accepted. The non-normality indicated by the Jarque-Bera test for the short-term interestrate is due to two large outliers (1979:4 and 1981:1); after deleting these outliers, the test

statistic has a value of 2.41 withp-value 0.30. Overall, it can be supposed that the VECM

is well specified.

The required reserve rate that does not occur in the money demand function is also

not important for the short-run development of the money stock. The coefficients tort

andrt1 in the equation for the first differences of the money stock are 0.08 (0.35) and

0.33 (1.29) which are not significant (t-values in parentheses). On the other hand, the

coefficient ofr is strongly significant in the monetary base equation of the VECM. Its

value is 4.34 with at-value of 6.55 such that the positive impact of the required reserve rate

on the monetary base that is predicted from the theoretical model can be confirmed. The

coefficient ofrt1is not significant: 0.24 (0.32).

The reaction of the variables to innovations in other variables can be analyzed with

impulse response functions and forecast error variance decompositions. They are calculated

16

8/9/2019 gemoba-3b

18/28

Table 4: Adjustment Coefficients (Loading Matrix ) and Diagnostic Tests

m h y s

ec1,t1 0.15 0.10 0.07 0.00 0.04

(5.01) (1.11) (1.36) (0.05) (1.67)ec2,t1 0.01 0.14 0.05 0.01 0.02

(0.97) (4.04) (2.69) (1.96) (2.70)

Weak Exogeneity 18.21 9.78 8.36 4.48 7.52[0.00] [0.02] [0.04] [0.21] [0.06]

R2

0.94 0.77 0.94 0.05 0.21

JB 3.93 1.67 4.19 0.25 131.20

[0.14] [0.43] [0.12] [0.88] [0.00]

LM(1) 0.86 2.34 6.94 0.79 1.63

[0.36] [0.13] [0.01] [0.38] [0.21]

LM(4) 1.13 3.30 9.82 0.78 1.23[0.35] [0.02] [0.00] [0.54] [0.30]

ARCH(1) 0.47 8.03 1.20 0.00 0.02

[0.49] [0.01] [0.28] [0.98] [0.88]

ARCH(4) 0.86 5.38 2.31 2.97 0.35

[0.49] [0.00] [0.06] [0.02] [0.85]

RESET(1) 0.35 1.22 0.43 0.20 0.07[0.56] [0.27] [0.51] [0.65] [0.79]

Notes: Upper part: coefficients of the error-correction terms in the equation for the variable in the first row.

Ratio of coefficient and respective asymptotic standard error in parentheses. Middle part: The test on weak exo-

geneity is a likelihood ratio test of zero restrictions on . Weak exogeneity is rejected if thep-value (in brackets)

is smaller than 0.05. Lower part: diagnostic tests. JB denotes the Jarque-Bera test for normality, LM(k) the

Lagrange multiplier test for serial correlation of the residuals (k lagged residuals included), ARCH(k) theLagrange multiplier test for autoregressive conditional heteroskedasticity, and RESET(1) the Regression Spec-

ification Error Test considering the second powers of the fitted values from the original regression, R2

is the

adjusted sample multiple correlation coefficient.

from the level representation:

xt = t+ A1xt1+ A2xt2+ ut, (37)

with A1 = 1 + +I5 and A2 = 1. BecauseA1 and A2 are calculated from the

VECM representation, the cointegration restrictions and the overidentifying restrictions are

imposed on the level coefficients. Now, the impulse responses and their asymptotic standard

errors as well as the forecast error variance decompositions can be calculated, see Lutkepohl

(1993, Chapter 11). Not all 25 impulse responses are discussed here, only the reaction of the

money stock and the monetary base to innovations are of interest here. Figure 3 shows the

generalized impulse responses, which do not depend on the ordering of the variables, see

Pesaran and Shin (1998). The forecast error variance decompositions in figure 4 confirm

that innovations in the monetary base have only a weak impact on the money stock but

17

8/9/2019 gemoba-3b

19/28

that innovations in the money stock have a considerable impact on the monetary base. The

largest impact on both variables have innovations in the short-term interest rate. Insert fig-

ures 3 and

4 about

here.

The economic scenario can now be described as follows: the interest rates seem to

be exogenous with respect to money stock and monetary base. This is supported by tests

for weak exogeneity and by the impulse responses ofst and t to innovations inmt andht, which are not significant (not depicted here). Stable demand functions for the money

stock and the monetary base exist, in which interest rates play a significant role. Therefore,

it can be supposed that the Bundesbank has influenced money stock and monetary base

by controlling the short-term interest rate. A stable relationship between money stock and

monetary base does also exist. However, the money multiplier approach is strongly rejected.

The monetary base adjusts in direction of the equilibrium between these two variables but

the money stock does not. Similar results are reported by Brand (2001) who estimates a

state space model for interest rates, bank reserves, and the money stock. He concludes that

(p. 114 f.) it would be misleading to view the money supply process in terms of a money-

multiplier model, since interest rates and money are exogenous to bank reserves and not

vice versa.

5 Conclusions

A model of the money-creating sector with endogenous money has been developed. In this

model, money equals credit, and the monetary base is determined by the profit-maximizing

behavior of commercial banks.

The implications of the theoretical model have been tested in the framework of a cointe-

grated vector autoregressive model. Strong empirical evidence against the money multiplier

approach has been found: A stable relationship between the monetary base and the money

stock can be specified, but the nominal money stock does not adjust if deviations from this

long-run relation occur; the adjustment is done by the monetary base instead. Furthermore,

the required reserve rate has not the negative impact on the money stock that is predicted

from the money multiplier approach. On the other hand, the empirical evidence is very

much in line with an industrial organization style model of the money creating sector. The

money stock and the monetary base are determined endogenously after the central bank has

set the money market interest rate. While the model of the money creating sector seems

to be a better description of the money supply process in Germany than the money multi-

plier approach, it has still some shortcomings. The money stock is simply set equal to the

quantity of loans, which are only one item of the counterparts of M3. There are also other

important counterparts like net capital formation, for example, which are neglected here.

The main conclusion of this analysis is therefore that the Bundesbank has been able to

18

8/9/2019 gemoba-3b

20/28

influence the money stock M3 via interest rate changes to a considerable extent but that

the standard textbook money multiplier approach is not appropriate to describe how the

Bundesbank has affected the development of M3. Because a stable money demand relation

can be specified for the period of monetary targeting in Germany it is reasonable to suppose

like for example Brand (2001) that the Bundesbank followed a policy of indirect monetarytargeting by changing money market conditions.

19

8/9/2019 gemoba-3b

21/28

Data Appendix

M3: End of month money stock M3 (currency in use plus sight deposits of domestic non-

banks at domestic banks in Germany plus time deposits for less than four years of

domestic non-banks at domestic banks plus savings deposits at three months notice

of domestic non-banks at domestic banks in Germany) in billions of DM, seasonally

unadjusted. Monthly data (TU0800) from the Compact Disc Deutsche Bundesbank

(1998a), continued with data from the monthly bulletin of the Deutsche Bundesbank,

table II.2. 1975:01-1990:5 West Germany, and 1990:06-1998:12 Germany, not ad-

justed for German unification. Quarterly data are end of quarter stocks.

Monetary base: sum of currency in use (TU0048), required and excess reserves (TU0062),

liabilities of the Deutsche Bundesbank against domestic banks (TU0084) and cash of

banks (OU0312), in billions of DM, seasonally unadjusted. Monthly data from the

Compact Disc Deutsche Bundesbank (1998a), continued with data from the monthly

bulletin of the Deutsche Bundesbank, tables II.2., III.2 and IV.1. Quarterly data are

end of quarter stocks.

Nominal GDP: Gross domestic product in current prices, in billions of DM, seasonally

unadjusted. Quarterly data (WH12011N) for 1975:01-1990:02, West Germany, from

Deutsches Institut fur Wirtschaftsforschung (DIW) Berlin (DIW-statfinder: http://

www.diw-berlin.de) continued with GDP for Germany (GH12011N).

Short-term interest rate: Daily money market interest rate, Frankfurt/Main, monthly av-erages, fractions, monthly data (SU0101) from the Compact Disc Deutsche Bundes-

bank (1998a), continued with data from the monthly bulletin, table VI.4. Quarterly

data are the respective values of the last month in a quarter.

Long-term interest rate: Yields on bonds outstanding issued by residents, monthly aver-

ages, fractions, monthly data (WU0017) from the Compact Disc Deutsche Bundes-

bank (1998a), continued with data from the monthly bulletin, table VII.5. Quarterly

data are the respective values of the last month in a quarter.

Average required reserve rate: Ratio of required reserves (IU3006) and reserve base of

banks subject to reserve requirements (IU3156), monthly data from the Compact Disc

Deutsche Bundesbank (1998a), continued with data from the monthly bulletin of the

Deutsche Bundesbank, table V.2. Quarterly data are the respective values of the last

month in a quarter.

20

8/9/2019 gemoba-3b

22/28

References

Angeloni, I., A. Kashyap, B. Mojon, D. Terlizzese (2002), Monetary transmission in the Euro

area: Where do we stand?, Working Paper 114, European Central Bank, Frankfurt/Main.

Artis, M. J., M. K. Lewis (1990), Money supply and demand, in: Bandyopadhyay, T.,

S. Ghatak (Eds.), Current issues in monetary economics, Hemel Hempstead: Harvester

Wheatsheaf, pp. 162.

Baltensperger, E.(1980), Alternative approaches to the theory of the banking firm. Journal

of Monetary Economics, Vol. 6, pp. 137.

Baltensperger, E.(1998), Geldpolitik bei wachsender Integration (1976-1996), in: Deutsche

Bundesbank (1998b), S. 475559.

Bhattacharya, S., A. V. Thakor (1993), Contemporary banking theory. Journal of Financial

Intermediation, Vol. 3, pp. 250.

Bofinger, P. (2001), Monetary policy: goals, institutions, strategies, and instruments, Ox-ford: Oxford University Press.

Brand, C. (2001), Money stock control and inflation targeting in Germany: a state space

modelling approach to the Bundesbanks operating procedures and intermediate strategy,

Heidelberg: Physica.

Br uggemann, I. (2001), Zum geldpolitischen Ubertragungsmechanismus in Deutschland.

Eine Analyse im Vektorfehlerkorrekturmodell, Aachen: Shaker.

Dermine, J. (1986), Deposit rates, credit rates and bank capital. The Monti-Klein model

revisited. Journal of Banking and Finance, Vol. 10, pp. 99114.

Desai, M. (1992), Endogenous and exogenous money, in: Newman, P. K. (Ed.), The new

Palgrave dictionary of money and finance, London: Macmillan, pp. 762764.

Deutsche Bundesbank (1995), Die Geldpolitik der Bundesbank, Frankfurt/Main: Deutsche

Bundesbank.

Deutsche Bundesbank (1998a), Funfzig Jahre Deutsche Mark. Monetare Statistiken 1947-

1998 auf CD-ROM, Munchen: Beck.

Deutsche Bundesbank (Hrsg.) (1998b), Funfzig Jahre Deutsche Mark: Notenbank und

Wahrung in Deutschland seit 1948, Munchen: Beck.

Dornbusch, R., S. Fischer(1994), Macroeconomics, sixth edition , New York: McGraw-

Hill.

Ericsson, N. R., D. F. Hendry, G. E. Mizon (1998), Exogeneity, cointegration, and economic

policy analysis. Journal of Economic Dynamics & Control, Vol. 16, pp. 370387.

European Central Bank (1999a), The stability-oriented monetary policy strategy of the Eu-

rosystem. Monthly Bulletin, January, pp. 3950.

European Central Bank (1999b), Euro area monetary aggregates and their role in the Eu-

21

8/9/2019 gemoba-3b

23/28

rosystems monetary policy strategy. Monthly Bulletin, February, pp. 2946.

Freixas, X., J.-C. Rochet(1997), Microeconomics of banking, Cambridge: MIT Press.

Gebauer, W.(Ed.) (1993), Foundations of European Central Bank policy, Heidelberg: Phys-

ica.

Guth, W.(1994), Markt- und Preistheorie, Berlin: Springer.Harbo, I., S. Johansen, B. Nielsen, A. Rahbek(1998), Asymptotic inference on cointegrating

rank in partial systems. Journal of Economic Dynamics & Control, Vol. 16(4), pp. 388

399.

Jarchow, H.-J.(1998), Theorie und Politik des Geldes I, Zehnte Auflage , Gottingen: Van-

denhoek & Ruprecht.

Johansen, S. (1995), Likelihood-based inference in cointegrated vector autoregressive mod-

els, New York: Oxford University Press.

Johansen, S., K. Juselius(1994), Identification of the long-run and the short-run structure.

An application to the ISLM model. Journal of Econometrics, Vol. 63, pp. 736.

Klein, M. A.(1971), A theory of the banking firm. Journal of Money, Credit, and Banking,

Vol. 3, pp. 205218.

Lanne, M., H. Lutkepohl, P. Saikkonen (2002), Comparison of Unit Root Tests for Time

Series with Level Shifts. Journal of Time Series Analysis, forthcoming.

Leschke, M., T. Polleit(1997), Zur Validitat der M3-Konzeption der Deutschen Bundesbank.

Konjunkturpolitik, Vol. 43, pp. 1642.

Lutkepohl, H.(1993), Introduction to multiple times series analysis, second edition , Berlin:

Springer.

Lutkepohl, H.(2001), Vector autoregressions, in: Baltagi, B. (Ed.), Companion to theoreti-

cal econometrics, chap. 32, Oxford: Blackwell, pp. 678699.

Lutkepohl, H., J. Wolters (2001), The transmission of German monetary policy in the pre-

Euro period, Working Paper 2001-87, SFB 373, Humboldt-Universitat zu Berlin.

Meltzer, A. H.(1969), Money, intermediation, and growth. Journal of Economic Literature,

Vol. 7, pp. 2756.

Meltzer, A. H. (1995), Monetary, credit and (other) transmission processes: a monetarist

perspective. Journal of Economic Perspectives, Vol. 9(4), pp. 4972.

Monti, M. (1972), Deposit, credit and interest rate determination under alternative bankobjective functions, in: Shell, K., G. P. Szego(Eds.), Mathematical methods in investment

and finance, Amsterdam: North-Holland, pp. 430454.

Muller, M.(1993), Endogenous money and interest rates in Germany, in: Gebauer (1993),

pp. 3548.

M uller, M. (1998), Endogenitat des Geldes: Eine Untersuchung zum Beitrag des Kredi-

tes zur Endogenitat des Geldes am Beispiel der Bundesrepublik Deutschland, Frankfurt:

22

8/9/2019 gemoba-3b

24/28

Haag und Herchen.

Nautz, D. (1998), Wie brauchbar sind Multiplikatorprognosen fur die Geldmengensteue-

rung der Bundesbank? Kredit und Kapital, pp. 171189.

Pesaran, H. H., Y. Shin(1998), Generalized impulse response analysis in linear multivariate

models. Economics Letters, Vol. 58, pp. 1729.Prisman, E. Z., M. B. Slovin, M. E. Sushka(1986), A general model of the banking firm un-

der conditions of monopoly, uncertainty, and recourse. Journal of Monetary Economics,

Vol. 17, pp. 293304.

Rasche, R. H. (1993), Monetary policy and the money supply process, in: Fratianni, M.,

D. Salvatore (Eds.), Monetary policy in developed economies, Amsterdam: Elsevier, pp.

2554.

Santomero, A. M.(1984), Modeling the banking firm. Journal of Money, Credit, and Bank-

ing, Vol. 16, pp. 576602.

Tobin, J.(1967), Commercial banks as creators of money, in: Hester, D. D., J. Tobin (Eds.),

Financial markets and economic activity, New York: Wiley, pp. 111, reprinted from

Carson, D. (Ed.) Banking and monetary studies. Homewood: Richard D. Irwin, 1963.

von Hagen, J.(1998), Geldpolitik auf neuen Wegen, in: Deutsche Bundesbank (1998b), S.

439473.

Walsh, C. E.(1998), Monetary theory and policy, Cambridge: MIT Press.

Willms, M. (1993), The money supply approach: empirical evidence for Germany, in:

Gebauer (1993), pp. 1134.

Wolters, J., T. Ter asvirta, H. Lutkepohl(1998), Modeling the demand for M3 in the unified

Germany. The Review of Economics and Statistics, Vol. 80, pp. 399409.

Yanelle, M.-O.(1988), On the theory of intermediation, Dissertation, Rheinische Friedrich-

Wilhelms-Universitat, Bonn.

Yanelle, M.-O. (1989), The strategic analysis of intermediation. European Economic Re-

view, Vol. 33, pp. 294301.

Dr. Oliver Holtemoller, Humboldt-Universitat zu Berlin, Wirtschaftswissenschaftliche Fakultat,

Sonderforschungsbereich 373, c/o Freie Universitat Berlin, Institut fur Statistik und Okonome-

trie, Boltzmannstr. 20, D-14195 Berlin, phone: +49(0)30-838-56358, fax: +49(0)30-838-54142,

EMail: oh@holtem.de.

23

8/9/2019 gemoba-3b

25/28

Figure 1: Money Multiplier, Germany, 1975-1998

75 80 85 90 95

4.5

5.0

5.5

6.0

6.5

7.0

7.5

Notes: Money Multipliermm = MH

, whereMis the money stock M3 andH is the monetary base.

8/9/2019 gemoba-3b

26/28

Figure 2: Money Growth Rate and Monetary Target in Germany

75 77 79 81 83 85 87 89 91 93 95 97

2

4

6

8

10

75 77 79 81 83 85 87 89 91 93 95 97

2

4

6

8

10

Notes: Thick line: Money growth rate in %, shaded area: announced target (point target from 1975 to 1978,

and in 1989, other years: upper and lower bound). From 1975 to 1987, the money stock under consideration has

been central bank money, from 1988 to 1998, M3. Up to the complete year 1990, the targets and the realized

growth rates are for West Germany. From 1991 on, the targets and the realized growth rates are for united

Germany. The data are taken from Leschke and Polleit (1997), and from the monthly bulletin of the Deutsche

Bundesbank.

8/9/2019 gemoba-3b

27/28

8/9/2019 gemoba-3b

28/28

Figure 4: Forecast Error Variance Decompositions for Money Stock and Monetary Base

4 8 12 16

0.20

0.40

0.60

0.80

1.00Money Stock

4 8 12 16

0.20

0.40

0.60

0.80

1.00Money Stock

4 8 12 16

0.20

0.40

0.60

0.80

1.00Monetary Base

4 8 12 16

0.20

0.40

0.60

0.80

1.00Monetary Base

y s { h m

Notes: Generalized forecast error variance decompositions of money stock (m) and monetary base (h), calcu-

lated from the level representation that is recovered from the VECM. For m, the upper dark layer depicts the

impact ofh, and forh, the upper dark layer depicts the impact ofm.