THE DEMAND FOR MONEY IN GREECE: AN EXERCISE IN ECONOMETRIC MODELLING WITH COINTEGRATED VARIABLES
| DOI | http://doi.org/10.1111/j.1468-0084.1993.mp55002005.x |
| Date | 01 May 1993 |
| Published date | 01 May 1993 |
OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 55,2(1993)
0305-9049 $3.00
THE DEMAND FOR MONEY IN GREECE: AN
EXERCISE IN ECONOMETRIC MODELLING
WITH COINTEGRATED VARIABLES
Zacha rias Psaradakis
I. INTRODUCTION
During the last 15 years the methodology of econometric analysis has
attracted a great deal of attention, in an attempt by the profession to identify
good' ways of doing empirical econometric research and to codify them in
the form of methodological prescriptions. A number of diverse modelling
strategies has thus been proposed, each one claiming to provide the most
effective way of taking the con out of econometrics (see the contributions to
Granger, 1990). In the light of such substantial developments, it is
undoubtedly of considerable interest that important economic issues be
readdressed and data sets by reanalysed. The work in this paper is intended
as a contribution towards this end, the principal objective being the empirical
re-examination of an important aspect of macroeconomic behaviour in the
Greek economy within an econometric framework which encompasses some
prima facie. conflicting methodological suggestions. The modelling strategy in
question is the one recently proposed in Hendry and Mizon (1990) and
Monfort and Rabemananjara (1990), where linear dynamic structural
econometric models (SEMs) are derived by sequential reduction of an under-
lying statistically well-specified vector autoregressive (VAR) representation
of the data. Hence, the latter, rather than being a competitor to structural
modelling, provides a general unrestricted model against which reductions
associated with specific structural-form parameterizations can be tested
validly.
The econometric strengths of such a modelling strategy are demonstrated
in the context of an empirical analysis of the relation between the Greek
demand for narrow money balances and its determinants over the period
1960-89. Such a relation has undoubtedly been the object of a phenomenal
tThe author is greatly indebted to Grayham Mizon for much invaluable advice and numer-
ous inspiring discussions on topics related to the material herein. The comments and sugges-
tions of Isabel Andrade and Anindya Banerjee have also helped considerably to improve the
paper. Financial support from the State Scholarships Foundation of Greece is gratefully
acknowledged.
215
216 BULLETIN
amount of empirical econometric analysis for more than two decades.
However, despite the impressive research effort there still seems to be little
consensus over the appropriate specification of the demand for money func-
tion. To make matters worse, the unprecedented economic developments of
the post-i 973 period have posed considerable demands on estimated money
demand models, many of which appear to be badly misspecified and
temporally unstable.
The aim of this paper is the development of an empirical econometric
model of the Greek money demand which is constant, data-coherent and
capable of encompassing competing data characterizations. Given the signifi-
cant implications that time-series features such as stochastic trends and
cointegration have for econometric analysis (see, e.g., Phillips, 1991), our
modelling exercise pays particular attention to the problem of obtaining
adequate representations of the non-stationary features of the data. In addi-
tion, much emphasis is also placed on the importance of establishing the
validity and credibility of inferences used for the interpretation of the empiri-
cal evidence provided by econometric models, something which can most
effectively be done by testing the model's congruence with all information
that is available.
The outline of the paper is as follows. Section II highlights some distinctive
characteristics of the Greek financial system and their implications for the
modelling of money demand, whilst Section IIi sketches out the methodo-
logical framework within which the modelling exercise is conducted. The
issue of non-stationarity is addressed in Section IV, where the integration and
cointegration properties of the data are analysed. In Section V a conditional
SEM is developed and its ability to encompass the underlying VAR is
assessed. Finally, Section VI summarizes and concludes.
II. FINANCIAL STRUCTURE AND THE DEMAND FOR MONEY
In spite of various reforms and institutional changes introduced by the
monetary authorities in an attempt to foster the development and promote
the efficiency of the Greek financial system, there is little doubt that the latter
still remains rather underdeveloped and inadequate An organized money
market does not exist in Greece, whilst the capital market is very thin and
relatively inactive. This leaves the (oligopolistically structured) banking
system as the dominant financial intermediary in the economy, and bank
deposits (especially savings deposits) as the main financial asset in the port-
folio of the domestic non-bank sector.'
As far as monetary policy is concerned, although important deregulatory
measures have been adopted during the post-1987 period, the authorities
For recent comprehensive analyses of the Greek financia! structure and policy see Garganas
(1988) and Alexander and Demopoutos (1989).
To continue reading
Request your trial