Académique Documents
Professionnel Documents
Culture Documents
Contemporary Economics
Didier Laussel William Marois
Antoine Soubeyran (Eds.)
Spri nger-Verlag
Berlin Heidelberg New York
london Paris Tokyo
Editorial Board
D. BOs G. Bombach
B. Gahlen
K. W. Rothschild
Editors
Didier Laussel
Professor of Economics
University of Aix-Marseille II, Faculty of Economics
13621 Aix en Provence, Cedex, France
William Marois
Professor of Economics
University of Orleans, Faculty of Economics
BP 6749, 45067 Orleans, Cedex 2, France
Antoine Soubeyran
Professor of Economics
University of Aix-Marseille II, Faculty of Economics
13621 Aix en Provence, Cedex, France
ISBN-13: 978-3-540-50322-4
DOl: 10.1007/978-3-642-74104-3
e-ISBN-13: 978-3-642-74104-3
This work Is subject to copyright All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, re-lJse of Illustrations, rocitation, broadcasting, reproduction on microfilms or in other ways, and storage In data banks. Duplication of this
publication or parts thereof is only permitted under the provisions of the German Copyright Law of
September g, 1965, In its version of June 24, 1985, and a copyright fee must always be paid. Violations
fall under the prosecution act of the German Copyright Law.
2142/3140 - 54321 0
INTRODUCTION
Didier LAUSSEL, William MAROIS and Antoine SOUBEYRAN
IV
markets.
The
central to
formalization
of
expectations
hypotheses
is
two hypotheses
: the
Muth
Rational
based on
the second
describe the
micro
hypothesis and
post 1980
markets,
hypothesis is
he
deceleration of
also
shows
inflation. Turning to
empirically
that
the
ARE
His demonstration
i.e. the
on the
relies on
the existence
differences between
true model
and the
of bayesian errors
concept of
and to
provide the
in a
are also
neo-keynesian and
HICKS (1974),
he develops
two sectors
intermediate
disequilibrium tradition. As
is perfectly
indexed on
the price
flexible but
the
of
price
sluggish process
As in
the
of adjustment
good
shows
regimes (a
glut and
between inflation
employment when
a shortage
regime) and
and unemployment
steady state
with a
finds a relation
maximum
inflation is
nil.
level
of
Inflation and
AMENDOLA and
restore the
the heart
an
of HICKS
J.L. GAFFARD's
analysis of
economy
which
highlights
technical intertemporal
production.
is essential
In
the
time
structure
complementarities of
the
and
process
the
of
v
assets can
be understood
formulating short
on its
either as the
of a mistake in
~esult
"routine path"
structural change
or
in the
as the
signal of
process of
the search of a
is then
to
analyse
innovative choice
of the
early phase
the
three
phases
when the
output of
both routine
and
innovative
processes is on the market, the late phase when only the output
of the innovative process is still on the market.
At the
survey, BARRO
as one
model.
In
separable utility
optimizing
function, BROCK
(1974)
models
and
using
other
authors
divergent
price
paths
hyperinflationary).
In
the case
function, OBSTFELD
(1984) has
(hyperdeflationary
of a
non
shown that
or
separable
utility
price paths
may be
stable.
D. LAUSSEL
uniqueness of
SOUBEYRAN examine
equilibrium in
separable utility
demonstrate a
and A.
a general
function and
the problem of
framework with
a non
price paths and give a set .of necessary and a set of sufficient
conditions
to
different and
have
divergent
price
complements-substitutes relationships
consumption is
the speed
paths.
They
provide
money
and
of decrease
three papers
are more
concerned with
the
(1967)
constraints.
gave
an
answer
based
on
cash-in-advance
VI
J.M.
where the
ROUSSEAU
elaborates
choice between
theoretical
framework
two characteristics,
liquidity function
and store of
cases, money
could be
held although
it is
it
examines the
is an
characteristics of
this market
asymmetry in
functionning of
important matter
a
are an
the credit
debt
contract.
The
imperfect information
the relationships
between the
main
and an
borrower and
the
lender.
In a
debt contract
bankruptcy of
either a
the firm
there
In both
some
probability
of
partial reimbursement
the firm.
is
or the
cases FREIXAS
possibility to control
in
repeated
firms must
effect on
game
consider the
one player
imperfect
with
reputation
problem
the
one
instruments with
credit
knows
market
the
the financial
which
actual
are
of
development
deregulation or
peculiar
of
credit
the amount of
relationships
monetary aggregate
popular VAR
between
the
counterparts
of
the
model developped
initially in
the SIMS's
(1980)
seminal work. This empirical procedure does not need to have "a
priori" strong
variables and
differences between
imposes less
econometric models.
box without
endogenous
constraints
than
and
in
exogenous
traditional
theory" but
VII
in ANTAO'
study the
that. in
sources of
Portugal. during
out effect
was strong
reserves were
monetary creation.
He
shows
but does
not influenced
not last
by the
and
that
external
credit component
of the
counterparts.
II. MONETARY POLICY CAMES
In the last ten years the analysis of economic policy
has
made
an
concepts.
increasing
There
were
developments which
KYDLAND and
use
of
two
game
basic
were initiated
PRESCOTT (1977)
theoretic
tools
motivations
for
and
such
on one
hand. of
HAMADA 1974).
private sector
only to the present rate of growth of the money supply but also
to its
future value
as announced
initially by
the
monetary
This gives
not
announced since
if
PRESCOTT (1977)
monetary
: the
policy
initially
for its
announcement
implementation date.
As argued by
one may
an open-loop
private
partly designed
its planned
the
to
the
monetary authority
framework of
manipulate
rise to
allowed
SALMON (1985)
between the
to
KYDLAND and
implement
it was
effects before
MILLER and
order
analyzed by
monetary authority.
t >
in
and the
Stackelberg dynamic
WHITMAN (1986.
More recently (see TABELLINI (1986
between monetary
been modelled
Treasury"
and fiscal
as a
each
game between
endowed
the interactions
with
the "Central
a
different
loss
function.
In the
no more
place in
possible to
one country
affects the
rest of
the
world
through
various
VIII
channels which
(1983):
the other
hand the
current account)
policymakers
same variables
may enter
in
way
simultaneous attainment
modelling of
such
as
of their
international
coordination in
(the exchange
the utility
it
rate or the
functions of
makes
several
impossible
the
economic
policy
conflict
and/or
this part
distinguished with
definition of
dynamic or
analyze
respect to
which
are
their basic
the
repeated) of
~Q1jcy
players,
sa~
and may be
constituents: the
the
nature
(static,
theoretical and
empirical, related
to
both
monetary authority
and of
be increased
They conclude
through
that coordination
inconsistency-credibility
problem
PRESCOTT
closed
(1977)
in
inconsistent policies
international
may help
coordination.
to solve
raised
by
economy
the time-
KYDLAND
context.
and
If
time
the costs
rationaly expect
that
these
of reneging
conclusions
do
not
contradict
SALMON's
(1987)
only time-consistent
policies:
CHOURAQUI's argument
is precisely
cooperative policies
policies implies
solution (one
the essence
that time
simultaneously reverting
has then
of
CLINTON
and
inconsistent but
to
non-cooperative
cooperative time-inconsistent
policies to
the
interactions
conflict
between
the
monetary
IX
x
A. LAVIGNE
and P.
interactions between
attempt to
Each of
monetary and
reduce the
these
WAECHTER
has
the
strategic
public debt
authorities
analyze
stock within
different
one country.
quadratic
loss
function
stock from
zero but,
stabilize the
in addition,
monetary growth
minimize deviations
conflict between
of
the
the Central
while
budget
the
Bank wants to
Treasury
deficit
tries
from
zero.
to
The
game framework
first without
private sector
and then
analyzing the
modelling a
behavior
of
private sector
characterized by
its demand
for money
and
behavior. Closed
Loop Nash,
Open Loop
Nash
its
the
which is
anticipative
and
Cooperative
cooperation.
When
explicitly
sector's behavior
slows down
accounted
for,
the
private
stock.
AOKI, GIRARDIN
rules
in
model
of
three
large
interdependent
economies where policy actions not only affect each own economy
but also
affect
other
decompose policy
shocks when
economies.
spillover or
other countries
AOKI exhibits
functions that
The
main
problem
is
to
such
cross effect
actions
on
recursive
allows for
of policy
decomposition
decentralized
policy
of
objective
actions.
He
the weighted
variances. The
sum of
individual
author decomposes
this
country's
objective
output
into
three
the world
output variance
and the
two
others
its output
and the
world output
(the
a-
XI
monetary instruments
or exogenous
disturbances affect
of the
through
aim of
current account
for the
determinant is
by the
seen
also the
empirical
paper.
seven main
the anticipated
an
E. GIRARDIN's
OECD countries.
The
floating
The
major
inflation, unemployment.
real rate
of growth
the
current
foreign impulses
are both
account
equations,
considered and
domestic
and
GIRARDIN shows
the
the non
American budget
current accounts.
surpluses
do
not
the importance
of US
fiscal
policies
in
the
this
period.
other
countries,
the
LDC
problem
when the
paper) debt
paper is
first
about the
studied
by
debt-with-potentialEATON
and
GERSOWITZ
servicing is
not exogenously
enforceable.
PICHT
problem. He
tries to
techniques the
establish with
the 1980's
: first
the 1970's
had changed
finance
in
favor
relative share
the help
of statistical
of cross-default clauses in
the relative
debt
of direct
and,
consequently,
investment in
reduced
the
of the
XII
especially of
itself was
lenders
the return
the cause
had
strong
strategies (voluntary
on invested
of the
incentives
lending became
choose
non-cooperative
marginal
on
one
hand,
In publishing
some additional
all those
research in
texts, we
hope to raise up
monetary theory
We are
the organization
very grateful
of the
Conference and
to C.
FONTENEAU
for
REFERENCES
BARRO R.J.
and S.
FISCHER (1976)
Monetary Theory",
- "Recent
Developments
in
133-167.
BROCK, V.A.
(1974) -
Perfect-Foresight",
International
Economic Review,
IS,
pp. 750-77.
CANZONERI, H.B.
Policy
an
Interdependent
Finance Discussion
World",
International
Some Recent
Theoretical Developments",
Federal Reserve
and
under
Inflation
Discretion
and
and H.
GER50VITZ
Repudiation
"Debt
(1981)
Theoretical
and
with
Empirical
Potential
Analysis",
Interdependence
Monetary
International
National
System,
(1976)
the
in
Policies
K.
and
Policies",
Monetary
Ch i cago, pp.
HAMADA,
of
Financial
13-33.
A
Strategic
Analysis
of
Monetary
XIV
Blackwell.
KYDLAND, P.W.
and E.C.
Discretion
Journal of
PRESCOTT (1977)
the
Inconsistency
Potitical Economy,
- "Rules
of
Rather than
Optimal
85, n03,
Plans",
491.
LANCASTER, K.
(1966) -
"A New
Approach to
Consumer Theory",
Time
Inconsistency
of
Optimal
Policy
in
an
qpen
(1986) -
(1986) -
(1986) -
LIST OF AUTHORS
AHENDOLA Hario
ANTAO Hario
AOKI Hasanao
AZAH Jean-Paul
BASAR Tamer
CARRARO Carlo
CHOURAQUI Jean-Claude
CLINTON Kevin
FREIXAS Xavier
GAFFARD Jean-Luc
GIRARDIN Eric
LAUSSEL Didier
LAVIGNE Anne
HAROIS William
PICHT Hartmut
ROUSSEAU Jean-Harie
SALHON Hark
SOUBEYRAN Antoine
STEIN Jerome
WAECHTER Philippe
University of Roma
University of Lisboa
University of Cal ifornia
University of Clermont-Ferrand I
University of Illinois
University of Venice
OECD
OECD
University of Toulouse I
University of Nice
University of Bordeaux I
University of Aix-Marseille II
University of Paris X-Nanterre
University of Orleans
Indiana University
University of Brest
University of Warwick
University of Aix-Marseille II
Brown University
University of Paris I
CONTENTS
J. L. STEIN
A Neo-Structuralist Model of Inflation and Unemployment
39
J.P. AZAM
Towards a Monetary Theory of a Process of Change
55
73
the Random
Characteristics Approach
93
J.M. ROUSSEAU
Debt Contract under Imperfect Information
a Survey
X. FREIXAS
Causal
Relations Among
103
the Sources
of Money
Supply
the
Portuguese Case
M. ANTAO
117
149
151
173
XVIII
On
the Convergence
of Beliefs
and
Policy
to
Rational
Debt. Inflation
207
and the
Coordination of
Fiscal and
Honetary Policies
A. LAVIGNE and P. VAECHTER
Decentralized
Monetary Rules
in a
225
Three-Country Model
and
Expectations and
269
Current Account Surplus of the Main
OECD Countries
E. GIRARDIN
The
Political Economy
297
of Debt Repudiation and Expropriation
in LDCs
H.R. PICHT
329
characterized by
the controversy
found
past decade
markets
to
and
very sharp
is that
be
international
finance
controversy. The
the "conventional
inconsistent
with
the
have
been
wisdom"
has
empirical
been
evidence.
speculative markets
futures markets
in
general
in commodities,
and,
in
particular
financial instruments,
stock
markets can
evaluate the
controversy surrounding
the various
macroeconomic models.
First, the
Monetarists,
discussed
macroeconomic controversies
Keynesians
and
evaluated
disagreements;
and
New
Classical
determine
results
of
the
the studies
the
Economists
sources
simultaneous
are summarized.
derived from
are brought
Second:
of
tests
are
the
of
the theories
of speculative markets
evidence from
speculative markets,
part, concerning
with the
to
the
alternative hypotheses
and evidence
and
between
the anticipations
discussed
hypotheses,
in
the
is
second
consistent
first part.
I. MACROECONOMIC THEORY AND POLICY
A. The Polarization of Macroeconomics
Disenchantment
during the
post 1968
declined, the
rate of
with
Keynesian
economics
developed
rose,
and
the
rate
of
-hu(t-l) + v(t).
pi
rate of inflation
u = U - Ue = unemployment rate less
its equilibrium value ; v(t) = non-systematic variable with
zero mean.
The period of stagflation was inconsistent with the
Keynesian point
of view. A search occurred for a more
satisfactory theory of macroeconomics which could explain the
paradox of stagflation and the observed economic phenomena.
The stagflation paradox has not only been observed in
the United States during the 1970's, but also in other
countries. In Germany, the "Wirtschaftstwunder" of the 1960's
turned into the "Wirtschaftsfrage" of the 1980's where sluggish
output growth
and record
unemployment rates
have been
accompanied by moderate inflation.
In Argentina, the rate of
inflation rose drastically from the first half of the 1970's to
the second half of the decade but the growth rate of output
declined.
The New Classical Economics (NeE) developed as the
total rejection of Keynesian economics. The Keynesians claimed
that their
demand management
policy contributed to the
5
obsolescence of the business cycle and successfully eliminated
the gap between full employment (potential) output and actual
output. The NCE argued just the opposite: there is no way that
the monetary authority can follow a systematic activist policy
that would achieve a rate of output that is, on average, higher
over the business cycle than would occur if the monetary
authority did not respond to varying business conditions. Their
main tenet is that the unemployment rate or rate of growth of
real output is insensitive to demand management policy choi~es.
(Lucas; Sargent and Wallace).
Crucial
to
their
analysis
is
the
MUTH
RATIONAL
framework, the
expectations, because
publicly
the market
available
is alleged
information
to be
efficiently.
using all
The
monetary
authority can control the actual rate of inflation, but not the
so the monetary policy cannot be used for
unanticipated rate
demand management.
Milton Friedman
NCE.
has never
been a
supporter of
the
Schwartz
stated
their
theme
main
concerning
the
Great
Contraction as follows.
"The
monetary
consequence of
factor which
collapse
other forces,
exerted a
events. The
was
powerful influence
failure of
the
no~
inescapable
the Federal
on the
course
of
the
particular
authorities and,
policies
in smaller
followed
degree, the
by
the
monetary
particular
monetary
authorities could
of money-
have prevented
indeed, could
the decline in
have produced
almost
any
th'e
dec line
substitution of
in
the
stock
of
monetary expansion,
contraction's severity
and almost
money,
could
let
have
as certainly
alone
the
reduced
the
its duration.
over one-half
and prices
by over
one third
in the course of
(1963 : 300-301).
The NCE
unemployment
rate,
or
growth
of
output,
= au(t-l)
where mu(t)
is the
- b[mu(t) - Emu(t;t-l)]
rationally expected
disagree with
the stock
of money,
Friedman, and
per se,
stock of
was no
consensus
amon~
economists which
state,
but
in macroeconomics
rather
the
dynamics
steady
the unemployment
deviates from
agree that
is at
in the
rate or
between
states, where
steady state:
its natural
rate, which
policy
(2) the
rate of
monetary
expansion
is not
affected
inflation is
less
the
long
by
equal to
run
the rate of
growth
rate.
monetary
The
If there is
accelerated by
monetary policy
? What
will be
the resulting
facilitate an
upper half
of the
diagram. On
y(t) to
simple way
axis is
figure is
the horizontal
demand schedule
is
the
the
familiar
equations.
aggregate
The
demand
f[k(t)l, where
effective labor.
real aggregate
understanding of
capacity output
capital to
M(t)
is a
k(t) is
the ratio of
demand. The
height of
the
aggregate
level anticipated
are held
to prevail
in the
at time
previous period
Its
better way
height
depends
the
real
money wage
factors
(a)
the
anticipated price
the height
by more
W(t) is
of the
than W(t),
set at
unemployment
value
of
money,
level p*(t;t-1).
u(t-l)
and
Monetary policy
(b)
the
can raise
latter depends
upon p*(t;t-1),
lower
unemployment rate
part
U(t) to
of
the
the ratio
diagram
of
actual
relates
to
the
capacity
Suppose that
curve [M/p]
at Oy.
the aggregate
demand is
described
by
An increase
aggregate demand
in the
to curve
money supply
[M/p]'
if
will
the
only
increase
anticipated
price
is
the
rationally
expansion. Therefore,
shift upwards,
if the
the aggregate
as a
rise in
The MRE
rate
demand
of
monetary
curve
will
only
hypothesis states
variables with
anticipated
zero expectations
serial correlation).
and no
Consequently, on
structure
average,
(such
as
rate
of
the
monetary policy
aggregate demand
the NCE
will not
curve. That
is unanticipated
which is
is why
money growth
[mu(t) -
Emu(t;t-1)],
p(t;t-1)
is
"sticky"
variables. Therefore,
change M/p
argue that
monetary policy,
and shift
the aggregate
which changes
M,
can
argumentsare
not
based
upon
explicit
optimizing
state of
affairs induced
me to
1982). The
general macro-dynamic
schools
of
thought
as
parameter specifications.
by the
stock of
three schools
capital and
research strategy
model which
special
was to
can imply
cases,
develop
depending
upon
that each
the
school of
thought is
It
a special case of a
10
<--
E(pi)
<--
E(mu)
<--
mu(t-1)
11
its view
concerning the
volatility of
the growth
of the
supply
(the
the rate
the desired
and (iii)
the level
its estimate
of defective
money
variance of
of confidence
size corresponds
Second, there
expected trend
economists have
to the
lag between
mu
and
rate of
monetary expansion
rate of
(ii)
required sample
E(mu) .
rate of
different models
concerning how
quickly
the
The relation
between money
stressed by
Monetarists Milton
lags between
money and
simple relation
quarter and
the rate
speaks of
prices. They
between the
relation between
this well
the intellectual
Friedman who
variable
rate of
of inflation
money and
leader of the
long and
monetary expansion
next quarter.
this
Instead, the
on average,
and independent
of the
horizon, monetary
the
the lag
expected
anticipated rate
Wet).
risk
It
[Wet) -
trend
rate
of inflation
is shown
W(t-1)]/W(t-1) <--
used in
[Stein, 1982
of
inflation
E(pi)
and
setting nominal
the
wages
the price
output or
prices is
reduced;
these three
whereby the
converges asymptotically
lags are
anticipated rate
to the
of
inflation
pi-
12
pi
mu
percentage change in M
Figure 2
describes this
monetary expansion
of inflation
expansion;
affected by
of
much
as
the
rate
are obtained.
1 (parameterized
of
monetary
The aggregate
by M/p),
cannot be
monetary policy.
figure 2,
curves in
are obtained
as
NCE results
in figure
the rate
changes by
and the
demand curve
the two
rise.
equation. Let
whereby monetary
policy can
temporarily
affect
M/p.
The statistical
specification of
the
a,b, ...
hypotheses implied
complex
dynamic
by my monetarist
structural
model
are
the coefficients
1982 :
chapter 31.
The first
states that
inflation between
upon the
years t
growth of
the change
and t-l
real balances
in
[pitt) [mu(t-l)
rate
of
pi(t-l)l,
the
pi(t-l)l depends
the
rate of
inflation is
independent of
inflation and
and 2.
The anticipated
finite, as
in figure
equation can
be understood
from
described in
rises.
Keynesian excess
the shift
of this
in the
Say
demand of
that
output
was
initially
Oy.
aggregate demand
curve.
The
rise
in
the
initially, by
balances.
a multiple
"a" of
the growth
of
real
13
aggregate
demand
(M/p*)' (M/'Il) ,
/p*)
________
y
~
____
(M/W)
1
___________________
y/f(k)
Figure 1
rates of
monetary
expansion
mu/2
Figure 2
T'
time
14
been pointed
example, Benderly
and Zwick
restrictive equation
be obtained
real balances
the
several
(4a) which
strong Monetarist
the acceleration
by
authors
this
is
[for
an
unduly
from equation
Keynesian and
out
I985b) that
of inflation
is a
synthesis of
equations. Equation
the
(4a) for
Monetarist equation.
semi-reduced form
the Keynesian
if there
is
equation differs
view, because
high
rate
of
inflation,can
unemployment.
unemployment (or
statistical semi-reduced
from the
NCE view.
form equation
growth
differs
Monetarist equation
of
output)
fundamentally
(5) states
that
the
and the
as mu(t-I) - pi(t-I).
(5) u(t)
Since the
negatively related
growth rate
to the
of
the
change in
economy
the
G[y(t)]
unemployment
is
rate,
from figures
depends upon
real unit
depends upon
the growth
price change.
initial
of the
The growth
unemployment
and
2.
labor cost
The
unemployment
rate
rate
and
the
certainty
equivalent
rate
depends
negatively
upon
the
initial
15
unemployment
and
rate,
negatively
upon
the
unexpected
inflation.
Suppose that
same rate,
money and
there is
is unemployment
prices were
growing at
the
output is Oy. Then wages will grow at a slower rate than prices
and money.
The real
balances M/W
aggregate demand
curve in figure
aggregate demand
of AB
reduces the
in wage
leads to
an increase
in output which
in equation (5).
The second
balances,
is
term, which
explained as
expansion rise.
follows. Let
change in real
the rate
of monetary
monetary expansion
a smaller
contains the
amount ("c"
is finite).
Therefore,
the
KeyneSian
excess demand
excess demand
of AB
rise at
a faster
rate than
nominal wages,
when there
resulting
decline
increases employment
rate. A
in
real
is an
unit
labor
cost
W/p
Keynesian could
accept
this
argument,
because
the
because
of "unanticipated"
the
underlying
money growth
theory
of
is used
anticipations
is
and
unemployment
variables stressed
simultaneously
by each
which
important. This
school of
sets
is the
equations
of
which
thought,
variables
nested tests
contain
and
are,
or
the
to
test
are
not,
the
process
alternative hypotheses
shown that
statistical
during the
testing
period 1953-86
of
the
it has been
whereas the
with the
of
Monetarist unemployment
evidence;
(B) the
hypothesis is
weak Monetarist
consistent
hypothesis is a
16
better
explanation
hypothesis. This
of
inflation
than
is
the
Keynesian
(6)
equation
contains
the
NCE
and
= constant + aU(t-1)
- pi(t-1)1 + v(t)
L
mu(t-1)
pi(t-1)
change in
real balances
in the previous
in
(6) ,
period
The
NeE
hypothesis,
equation
that
is
unanti c i pated money growth matters, but not the previous year's
growth
of
real
hypothesis.
balances
Thus
the
as
NeE
implied
claims
by
the
Monetarist
vector
that
"b n
is
zero. The
Monetarist hypothesis
is
just
the
opposite.
The NeE
nested tests
their equation
such
as
do not
follow
of hypotheses
(2),
general
estimates of
in
this
research
but simply
strategy
consider
variants
of
of
equation
non-observable
(6).
They
construct
anticipations,
which
arbitrary
cannot
be
interesting that
they cannot
19861. They
even the
explain the
Great
NeE are
forced to
Depression
[Rush,
does not affect the validity of their theory. The NeE school of
thought
cannot
particularly the
explain
the
event that
major
macro-economic
events,
In every' country and in every period the NeE fails when tested
against
the
"unanticipated"
Monetarist
money
hypothesis.
growth,
following
The
the
measures
of
NeE
of
ways
17
measuring it
growth of
(Stein,
Wohar,
are never
real
balances
1982
1984
significant;
ch.
are
always
4). Several
significantly
authors 1983
negative
Turnovskyand
have considered
the interwar
period. Others
results are
always the
output equation
The NCE
period and
examined foreign
1952:3-1980:4 (Domenech,
several periods
the post
countries such
1987) and
in the US :
World War
II
as
Argentina
Germany (Sauer,
1987). The
always fails
are not
involved here,
because they
hypotheses;
involved. The
coefficient of
negative, and
the Monetarist
the lagged
Keynesian
view
unemployment rate
the coefficient
and the
is
NCE is
that
the
is significantly
of real
is that
(i) the
significantly positive.
hypothesis is that,
in addition to (i),
[pi(t) - pi(t-l)]
The logic
figure 1.
When
aggregate demand
of this
the
real
equation can
balances
curve shifts
be conveyed through
rise
upwards. The
in
year
t-l,
Keynesian
the
excess
18
change
in
auction
markets,
which
then
get
transmitted
excess
Viewed
demand
in
of
this
will
follow
AB
way,
increases
the
the
Monetarist
rate
of
inflation
trajectory
or
c'
and
converge
United States
41;
[1957-76. Carlson,
[1958-80, Domenech),
strong Monetarist
the Keynesian
view is
considered jointly
1929, and
is
in equation
not
19781
(7). For
to 1953,
demonstrably
(1962-79),
and
view is
accepted, when
possibly prior
hypothesis
(1958-79), Canada
Argentina
rejected and
the
in the
superior
US the
to
the
prior to
Monetarist
Keynesian
[Sauer) the
explanation.
expect that
money is
during the
indicates that
for Western
as yet
succeeded in
finding an
inflation equation
which
is
"a" on
the inflation
rate of
expansion. The
the growth
of the
monetary
"half-life" is
of the
full effect
the US
of inflation
times as
implied by
cause is
react quickly
where inflation
and people
to the
is lower,
place more
analysis than
current fiscal
policy.
In the countries
weight upon
input.
19
In terms
1980 Price
Federal
Experience
Reserve
and
the
the
Dilemma
of
the
Monetarist/Keynesian
Hypotheses
In october
operating procedure
the federal
1979, the
and switched
funds rate
the growth
of the
the Federal
summarized his
Reserve
changed
its
Reserve to
monetary growth.
Federal
de-emphasize the
policy of targetting
the use
of HI as a policy
tool.
"Experience over
the
difficulty
monetary policy
two simple,
-I
in current
HI alone
impossibility-
of
conducting
circumstances according
during this
institutional transition
future price
the broader
is not
economy and
to one or
period
today a
of
economic
reliable measure
more restrained
well as
the
and
of
performance of
performance
of
the
The
concerning the
money as
say
suggests that
(quoted in
the first
would
1987).
difficulty
use of
a policy
that
the
Federal
Reserve
faced
data.
rate of growth of
1967-82
1980-86
HI
H2
9.2
9.4
7.4
7.7
7.0
4.9
20
The growth
decelerate,
but
of the
there
was
significant
deceleration
of
of the
and inflation
inflation rate
incorrect?
The way
then become
? Is
to resolve
these
questions
is
to
re-
(7b ; c) pi(t)
RBAR-SQUARE 0.89
OBS = 24
OW 2.27
(7b) 1957-1980
SEE 0.93
variable
coefficient
t-statistic
CONSTANT
1.7
1.98
LAG INFL
0.72
5.98
LAG GHPM
0.37
3.22
-0.40
LAG UNEMP
-2.5
equation
This
is
most
consistent
with
monetarist theory.
lagged growth
unity.
The
.w.e.a.k.
significant.
( 7 c) 1967 -1986
OBS
RBAR-SQUARE 0.77
OW 2.7
20
SEE 1.08
variable
coefficient
t-statistic
CONSTANT
0.57
0.41
LAG INFL
0.75
6.2
LAG GHPM
0.72
3.2
-0.68
-4.3
LAG UNEMP
the
21
The Keynesian
more important
in explaining
the deceleration
inflation. The
lagged growth
of high
of the
powered money
recent
is still
is that
the
-which is
arguments- are
elements
of
the
weak
still valid
in
explaining
the
of
estimate
unemployment,
increased.
however,
inflation decelerated.
concerned about
has
The
been
effect
of
main
reason
the
the
high
why
the
the effect
growth
of
Monetarist hypothesis
the
economy
and
explains more
the
variati.ons
about
in
the
in particular
used the
Monetarist model
why they
seem to
to explain
inflation.
Their argument is that real stock returns in year t-l
denoted by R(t-1) reflect the growth of the economy from t-l to
year
t.
Monetarist
The
growth
monetary aggregate,
1, denot'ed
of
equation
the
(Sa)
economy
above.
Given
G[y(t)]
is
given
the
growth
of
by
the
by pi(t-1),
22
occurs at
all. There
is a
if it
determine,
in
is almost
an
objective
way,
what
appropriate equation
for anticipations.
composite hypotheses
in
equation for
of this
macro
price anticipations.
paper
concerning:
examines
evidence
is or is
There
model
to
are
not
too
disentangle
the
many
the
from
micro
markets
foreign exchange.
A. Bayesian Errors. Muth Rationnal and asymptotically
rational expectations
The forecast
p(t+l), and
denoted by
equation
the subjective
or market
E*p(t+l;t), can
anticipation at time t,
(S).
[p(t+ll - Ep(t+ll] +
[Ep(t+l) - E*p(t+l;t)]
(Sa)
u + B(t>
The first
term is
price p(t+l)
and the
abbreviated
the expectation
objective expectation
that would
be known
Ep(t+l), which is
by u,
is an
second
term
is
the
difference
between
the
I call difference Ep
23
- E*p
a BAYESIAN
ERROR, denoted
by B(t).
way to
that there
distributed numbers
the mean
from the
is
the Bayesian
large
urn
error is
with
as
normally
price, but
an individual
think of
incurs a
given cost
estimates of
the
estimate E*p
B(t)
mean
and
obtains
subjective
posterior
these Bayesian
information
errors. The
available
Bayesian error
at
time
will be related to
contained
currently drawn.
Bayesian
The MRE
hypothesis assumes
error.
The
expectation. There
public
is no
that B
always
=0
knows
attention paid
in
to
the
sample
; there is no
the
the
objective
process
of
If
E[B(t)] and
the variance
did not
converge, then
it is possible
that there are growing positive errors which are exactly offset
by negative errors.
The
ASYMPTOTICALLY
HYPOTHESIS is
concerned with
convergence of
the Bayesian
(ARE)
EXPECTATIONS
RATIONAL
MRE hypothesis
is a
Bayesian errors
are serially
(1) the
is attached
the weight
attached
is on
to
the
to the
prior
declines
is positively
the urn,
relative
to
that
related to
the variance
of the
24
items sampled,
denoted by
var p/N(t)
The variance,
errors converge
N(t) of
and the
asymptoti~ally
items sampled
that the
expectation, of
the Bayesian
grows indefinitively
large:
provided
sampling, the
; and
convergence. The
there will
be a
uncertainty, the
slower
will
be
the
speed
of
way
the
ASYMPTOTICALLY
RATIONAL
EXPECTATIONS
HYPOTHESIS.
In the
MRE hypothesis,
zero,
is
with the
that the
nominal interest
[B)
[C) pricing
nominal
decomposed as
the sum
which changes
rate
of
interest
i(t)
is
usually
r+ E*pi(t+1;t)
The difference
between the
25
u + B(t)
and the
nominal rate
of
interest
is
equation
(10c).
=-
(10c) pi(t+1)
r + i(t) + B(t) + u
If the
no Bayesian
error, then
First:
there should
rate of
interest and
the rate
regression of
of interest
should be
the rate
on an
unity;
and the
included in
Bayesian error
contemporaneous variables,
a predictor
It
in
: no
rate as
Second
rate. Third
period.
of inflation.
B(t), should
which
add to
would
be
the interest
will be
shown that
each of
these
statements
is
over the
that
long
"the data
run
1867-1975.
remarkable
inflation and
nominal assets
average much
falling prices ... " [po 10]. The correlation between the rate of
inflation and the nominal rate of interest is not significantly
positive.
The relation
"After World
War II,
began to
change in
the financial
markets began
to
behave
the United
prior pattern...
inflation, so
variable" (p.
interest rates
real
11).
States and
returns
on
the United
start to
nominal
Kingdom, from
parallel
assets
rates
become
the
of
less
26
2. Recent Evidence
Fama and
(10c) and
the
others
FISHER
evidence. Their
addressed
hypothesis,
hypotheses were
themselves
on
the
that the
to
basis
equation
of
recent
coefficient of
the
the
period
1953-71,
Fama
reported
that
the
one-to-one
correspondence
inflation of
a subset
with
the
subsequent
rate
of
examination of
several results.
there is
a lot
of available
subsequent inflation
(b)
information that
In all periods,
helps
predict
of interest.
Several equations (11.1)-(11.3), which are subsets of
equation (10c)
hypothesis
above, show
the serious
Standard errors
are shown
(Barro
1986) .
1953-71 (11.1) pitt)
.01
+ .82 i ( t-1 )
( . 15)
( .006)
This
equation
coefficient b
.82
gives
is not
of the
mixed
results.
significantly different
(i)
The
from its
real rate,
equation is
misleading.
and lagged
is obtained.
money growth
mutt-h) to (11.1),
27
.001
( . 007 )
. 73 P i ( t - 1 )
( . 35 )
-.30 pi(t-2) +
( . 17)
Now the
.14 i(t-l)
( .32)
.26 mu(t-2)
.49 mu(t-l)
( . 18)
( .23)
(a) The
and is
(b) The
hypothesis.
1972-82 (11.3) pi(t) = .061
( .025)
.19 i(t-l)
( .30)
is that,
insofar as
the Fisher
hypothesis is
is shed
upon
the
anticipations
instruments
and
traditional
commodities.
These
markets are
spot markets.
For
Treasury bond
times that
easier to
the
daily
volume
of
trade
in
markets can
no price
example,
rigidities or
test anticipations
hypotheses on
a micro
It is much
level in
markets we
focus
upon,
and
have
data
28
which calls
for delivery
consider the
futures price
Japanese yen
to be
on January
delivered in
9 of
a US
March. Denote
dollar for
this price as
the approach
error (p-q)
taken in
equation (8a).
(p-q)
The first
term (p-Ep)
u +
B + R
contract.
The evidence
from a
market. Then,
futures markets
would
not
affect
the
has
theoretically and
the convergence
been
shown
(Stein
1986
ch.
5),
both
Bayesian errors
development, the
and Idaho,
development of
year
A(t)
in some
there were
cobweb cycles
prior
to
the
depended
previous year's
positively
price at
and
significantly
upon
the
Idaho
potatoes.
The
cycle
continued
very
29
However it disappeared in the case of Maine potatoes, such that
there was
no
longer
any
significant
relation
between
the
clearly very
markets
were
developed,
the
Bayesian
errors
were
reduced significantly.
Second : consider the case of "storable" commodities.
A commodity
is purchased
the commodity
at time
t becomes
physical depreciation,
positive
marginal
including risk
the commodity
m<1
m units
and
productivity,
premium is
from t
expected price at t
if
at t+l.
the
m>l.
If there is
commodity
The
has
interest
rate
to t+l
risk premium,
should equal
the subjectively
including
expected
price.
Th i s i s (13. 1 ) .
(13.1) E*p(t+l;t) = [(I+i)/ml p(t)
Using equation
(8a), which
no Bayesian
were correct
errors, then
the price
at t+l
should
depend upon the current price, but not on earlier prices p(t-h)
where h>O.
This
hypothesis
was
examined
for
many
storable
no active
second period,
commodity.
futures market
there was
an
active
in the
futures
commodity.
market
In the
in
the
30
and a
vector of
(13.3) p(t+1)
a + b pet) + c. pet-h) + u
The MRE
hypothesis is
errors. Therefore,
that there
the contribution
are
no
of c.p(t-h),
Bayesian
h>O to
the
concerning the
calculated for
contribution of
each commodity
c.
periods: with and without futures trading. The results are the
same for
in the
each commodity.
commodity, the
F statistic
no Bayesian
errors.
commodity, then
(2)
futures
markets
equilibrium.
seems to
accelerates
the
convergence
to
MRE
be a
necessary but
another important
not correct
arise from
makes a
several sources.
smaller Bayesian
superior forecasting
not exist
particular
objective
; and
group
costs of
One is
that the
is
not
Ep
ARE claims
sampling will
that a
error than
ability. The
expectation
expectation. The
MRE assumes
subjective expectation
significantly
which
is
that the
take larger
group
different
equal
to
group with
sized samples
E-p of
any
from
the
the
market
the smaller
of current
31
information and,
on the
market, denoted
(15) EZ.
f(l-b.)
If the
Bayesian error
of a
particular
group
were
done theoretically,
unobservable. However,
since
the
the expected
Bayesian
errors
are
if its
= cov(B.,B)/varB
positive function
Bayesian error
= market
; B
of this
group; EZ.
Bayesian error ; B.
the
evidence
in
speculators and
is
that
forecasting
small
there
abilities
(amateur)
are
between
significant
professionnal
speculators
and
between
agents
profits of
the
(either
larger
superior forecasting
The larger
a whole
Bayesian errors
errors are
than does
a systematic
or
speculators).
generally
derive
from
than from
smaller Bayesian
whereas
not random
differ in
agents
abilities rather
agents make
market as
commercials
The
their
risk premia.
the smaller
agents make
larger
variables with
way among
zero
groups as
means,
but
they
predicted by ARE
but one
profit from
the risk
from commercials
this risk
group, usually
premium (R
who want
referred to
to hedge
as speculators,
Keynes called
32
"normal backwardation".
hypothesis which
the
There is
profit
speculators
from
the
"normal
backwardation".
According to
non-commercial traders
daily positions
in
the futures
December 31,
reporting commercial
the losses
of the
pure speculators.
table 1
markets.
(who
are
the
required
markets during
1981. The
profits
and non-commercial
losses
traders are
of
the
equal to
contradicts
short
positions
the
in
MRE/normal
the
($ millions)
Commodity
mean return
commercial
non-commercial
0.35
0.03
commercial
non-commercial
pork bellies
commercial
non-commercial
live cattle
commercial
non-commercial
feeder cattle
commercial
non-commercial
T.bonds
commercial
non-commercial
0.45
0.04
2.80""
0.07"
-1.02"
0.55""
0.49
0.64""
5.95"
-0. 11
futures
backwardation
TABLE 1
wheat
to
hypothesis.
oats
large
long and
This
returns to
shows that
both their
the daily
33
T-bills
1.20-0.04
commercial
non-commercial
All markets
commercial
non-commercial
Source:
Hartzmark (1987,
table 3)
; his original study is
cited in Stein, 1986 : ch. 5). The -(--) = significant
at 10~ (1~) level
foreign
exchange
markets
are
important
and
many years.
exchange market
hypothesis.
The
forward/futures
A remarkable
is that
existence
markets
phenomenon in
they are
of
seems
the foreign
to
be
and
necessary,
developed
but
not
well known
and
be shown
to see
to
connection with
ways. A
total violation
corresponds
in many
of the
the
FISHER
FISHER
CLOSED
equations (10)
concerning futures
OPEN
hypothesis,
hypothesis
(11) above,
which
discussed
or equation
in
(12)
the latter, repeated here as equation (16), when the spot price
= E-p-
not been
failed to
written explicitly
find any
evidence for
because researchers
have
exchange market.
(16)
[p(t+1) - pIt)]
The HRE
hypothesis which
claims that
there are
no
time t+1 less the (logarithm) of the spot price at time t, upon
the (logarithm)
time t+1
less the
(logarithm) of
the spot
price at
time t.
34
should yield
a regression
coefficient,
denoted
by
"b",
of
unity.
There is
covered interest
premium on
another way
rate parity
the foreign
to look
hypothesis is
exchange [q(t+l;t)
that the
forward
on securities
of comparable
risk, denoted
by
i(t)
foreign exchange will be at a premium if
and (17)
then imply
(18),
equation
i>O,
i(t) + B(t) + u
domestic interest
rate exceeds
the
foreign
relative to
the crucial
concerning exchange
the analysis
It makes
no difference to
MRE hypothesis.
have to
rate dynamics.
It is
search for
interest rates
on securities
with
equal
risk.
Every study
the hypothesis
done by
Table
that b
the Bank
2
"b"
the
that the
hypothesized value
standard error
is so
foreign exchange
The most
and
the
is
standard
one
month
significantly
value is
market rejects
(Longworth, Boothe
values
coefficient
asterisk indicate
definite.
1.
concerning
Generally, this
from the
of Canada
presents
coefficient
of the
and Clinton).
errors
of
forward
negative.
significantly
the
rate.
The
different
large that
35
TABLE 2
EVIDENCE FROH FOREIGN EXCHANGE HARKETS 1970-1981
ONE HONTH FORWARD RATE RELATIVE TO US DOLLAR
COUNTRY
REGRESSION COEFFICIENT
STAND. ERROR
.389--
Canada
.550)
France
- 1.8376-
.8756)
Germany
.5333
.4786-.4803
- 1.5330--
(1.4169)
( .4716)
.4174)
.8434)
Italy
Japan
UK
Source
Contrary
domestic interest
to
the
rate exceeds
foreign exchange
tends
contradicts the
Fisher
to
significantly diffe-
Open
hypothesis,
the foreign
decline.
The
if
the
evidence
literature.
The HRE hypothesis, or its counterpart in the finance
literature THE
the data.
and can
The ARE
hypothesis is
explain why
a negative
is contradicted by
1986:
found (Stein,
ch.5).
Hy
conclusions
are
as
follows.
First
futures
a rapid
among
commercial firms,
groups
and to
in their
a lesser
Bayesian
errors.
extent large
Large
speculators.
Third
large commercial
their costs
of sampling
smaller than
even
firms make
when
must be
there
or
smaller Bayesian
obtaining
incurred by
are
well
the
market
other
developed
36
very significant
hypotheses are
with the
: It is
incorrect to
use the
macro model,
which implies
MRE hypothesis
data. Fourth
the speed
as an integral part of a
of convergence
(eqn.9) is a superior
results obtained
in part I above.
REFERENCES
Barro,
~obert
(1986) Futures
Markets and the Fluctuations
Inflation, Journal of Business, 59, April, S21-38
in
Barro, Robert (1978) Unanticipated Money, Output and the Price Level
in the United States, Journal of Political Economy, 86, 549-80.
Benderly, Jason and Burton Zwick (1985a)
Inflation, Real Balances,
Output and Real Stock Returns, American Economic Review, 75,
December, 1115~23
Benderly, Jason and Burton Zwick (1985b) Money,
Unemployment and
Inflation, Review of Economics and Statistics, 67, February,
139-43
Blinder, Alan (1987) Keynes,
Lucas and Scientific Progress, American
Economic Review, Papers and Proceedings, 77, 130-36
Carlson, Keith (1978)
Inflation, Unemployment
Federal Reserve Bank of St. Louis, n09
and
Money,
Review
of
37
in the
Chicago
Hartzmark, Hichael
(1987) Luck
Versus Forecasting
Ability
Determinants of Trader Performance in Futures Markets, Working
Paper, Department of Economics, University of Michigan
Longworth, David, P. Boothe and Kevin Clinton (1983) A Study of the
Efficiency of Foreign Exchange Markets, Ottawa: Bank of Canada
Lucas, Robert E. (1973) Some International Evidence on the OutputInflation Tradeoff, American Economic Review, 63, 3, 326-34
Hodigliani, Franco and L. Papademos (1976) Monetary Policy for the
Coming Quarters, Federal Reserve Bank of Boston, New England
Economic Review, March/April
Neutzel, Philip (1987)
Uncertain Times.
February
the
The Economics
of Futures
and
New
Classical
Markets. Oxford:
Ja~es
(1975) Monetary Policy.
Inflation and
Papers on Monetary Policy. Bank of Australia
Unemployment.
Turnovsky, Stephen and Hark Vohar (1984) Monetarism and the Aggregate
Economy:
Some Longer-Run Evidence. Review of Economics and
Statistics. 66. November. 619-29.
University of Clermont-Ferrand
INTRODUCTION
On the
development theory,
which can
one can
be more
approach to
be found
find a
or less
bunched to
inflation. Although
in the
set of
analytical themes
are so labelled,
in a proper sense.
But some
in the
basis of
behaviour of
supply or
the value
or rates). For example, the fiscal system matters more than the
level of
taxation. Other
concerned with
following, where
Findlay,
H The
( 1988) .
propositions from
international trade,
we
only
analyse
and are
the
closed
economy
(see
1973).
French version
40
In the
present article,
model of
inflation and
flavour,
emphasizing
assumptions
and
characteristics.
we are
unemployment
inelasticity,
deriving
having
the
structuralist
rigidity,
results
we
Consequently,
microeconomic foundations
trying to produce a
and
provide
asymmetry
"structural"
from
some
rigourous
to some
this
reason,
the
label
"Neo-Structuralist"
is
more
use
two-sector
producing intermediate
in Hicks
(1974). To
been called:
as the
odel,
goods, whose
including
sector
model developed
here
uses
some
of
the
assumptions
we assume
"flexprice" markets,
allowing for
on relative
prices. Similarly,
goods plays
it is
rather reasonable
between this
the
market
for
intermediate
aim of
the present
model is
not
only
to
maximized in
the long
corollary, this
costs
of
model provides
perfectly
employment.
run when
expected
Curiously
then,
prices
are
constant.
As
in
terms
Neo-Structuralist
of
lost
model
the second
the aggregate
one describes
supply curve
which
it
implies.
No
41
effort is
done to
the demand
side,
mechanism to
and
one
only
assumes
prevent equilibrium
curve, whether
in its
dynamic long-run
from
static short-run
version, which
is the
that
lying
there
on
the
version,
topic
is
or
of
no
supply
in
the
its
third
section.
1. THE ASSUHPTIONS OF THE HODEL
1.1 The technologies of the sectors
We
analyse
here
model-economy
including
goods are
sold respectively
denote L y
and Lx
sectors,
and
at the
two
in sector I,
money prices
p and v. We
the
decreasing returns
money
prevail in
wage.
Lastly,
the two
we
assume
sectors, so
that
that
the
(2)
FL > 0, Fx > 0
( 3)
(4)
(5 )
(6)
F(L y
X)
(1)
X = H(L x )
H' > 0, H" < 0
In the
the subscripts
expressions of
denote the
assumptions (2),
we
as the
working of
select
assumptions
"flexprice" markets
market, we
exist side
assume that
the
so
various
that
markets
"fixprice"
is
and
full employment
is not
feasible, and
In
other words,
42
of
unemployment
equilibria.
Moreover.
we
assume
that
the
sp
In order
to (at
least) four
theories. First.
appealing rationale
assume that
trade unions
imposing the
real wage
following Friedman's
has decided
and
this
is
the
more
powerful, and
in
advice (Friedman,
to use
succeed
an indexation
1974), the
mechanism of
government
wages on
the
prices of
one could
as well
which could
For example,
for a
one could
labour hypothesis
constant MRS
find justifications
be applied
(Lewis, 1954),
between leisure
or to Sen's assumptions of a
and consumption
in the
peasant
perfect flexibility
akin to
Postulate of
the "First
maintained in
being very
apply this
which Keynes
common, used
well justify
the Classics",
it by
by nearly
analysis to
the very
short run
(see Azam,
1986,
chap. II).
1.4 Sluggish
adjustment
on
the
intermediate
good
market
Lastly, for
mixed assumption.
quite natural
the intermediate
As Hicks
to assume
good market,
(1974) explicitly
that the
we use a
does, it
seems
43
44
whether supply
XS or
demand Xd
intermediate-good market.
where supply
of that
is
We call
good is
the
short
shortage
side
regime
a bottleneck,
of
the
and we
the
case
call the
functions that
we use
to form
the
(8 )
X"
( 10)
max vX - wLx
H (Lx)
s.c. X
( 11 )
Under the
(11) cannot
vX
F (Ly,X)
s.c. Y
(9 )
stated assumptions,
constraints
(9)
and
borderline
between
problem similar
of quantity
has been
the
two
disequilibrium
constraints, giving
previously dealt
regimes.
with by
the present
author (Azam,
behaviour of
the case
the firms
where X
<
XS,
in
sector
I,
pFL
(13)
pFx
Taking the
d(W/ P \
d(V/PY
45
for labour
and
notional demands
words, the
In other
intermediate inputs in sector I are decreasing functions of the
s) and of the real price of good X (vIp) .
real wage rate (w/p
It follows that the notional supply of final good Y is a
decreasing function of the real wage s and of the nominal price
v, and an increasing function of the price p.
Except in the border-case where the intermediate-good
market is
sector I
cleared, the
is only
notional behaviour
effective in
of
the
firms
in
In this case,
(15)
Lx
H- 1 (X d
= Ly +Lx
is a decreasing function
shortage regime,
results. First,
sector II
one gets
is not
rather
different
vH'
One
supply of output
H'
( 17)
price p,
(H' ) ..
Verbally, the
sector is
a decreasing
on which
notional demand
function of
for
the real
X-, .by
d(W/P
d(v/p>
labour
by
this
46
pFL = w
pFx > v
R
C~:)-hLJ (FL
-FLX)
("(W/P~
FLLFx - FLXFL
dX-
This gives the structural forms of the demand for labour - and
supply of output functions of the firms in sector I in the
shortage regime.
To get the reduced form of these functions, in order
to describe the global behaviour of the economy in the shortage
regime, one must replace dX- in (18) by its expression coming
from (17). One thus gets results which are very different from
the previous case. Let's begin by describing the total demand
R
47
48
3. THE LONG RUN SUPPLY CURVE
3.1 Adjustment of the intermediate-good price
In the
same way
rigidity assumption
as it
is never
easy to
warrant a
it is never easy to
describe then
justified the
vision".
invoking
intermediate
the precarious
good
sector.
customers' markets.
violent than
where
which
price
elsewhere. But
an instrumental
position
part. to
leads
of
firms
them
to
fluctuations
in
the
establish
must
be
less
v. for
inflation and
do not
mean to
the analysis
unemployment. On
carryon
assuming
like
the wage
acts instantaneously.
act simultaneously
where only
we assume
as time
one of
here that
passes. even
if one regards it as
model
of
the
dynamics
of
inflation
and
interested in
of perfectly
here.
expected inflation.
seeking to
analyse in
trend rate
of inflation
fashion of
a "Long-run
short-run dynamics.
our model
we
a relationship
(expected) . and
Phi 11 ips Curve" .
are
between
employment.
We thus
in
only
the
the
neglect the
the final good market in this model. and we only need to assume
that steady
which is
states. defined
equivalent) exist
by the
and
are
Hence.
we
are
of money
49
that real
cash balances
demand. We
do not
problem set.
v as
do not
need to
affect at
time passes,
we denote v,
whi~h
is an increasing function
= V(p,X
function, and
of this
Vp and
that the
=a
function play
below, and
X-
X-), V(O,O)
We denote
we must
a crucial
present them
dynamic adjustment
of
carefully. One
the
price
could assume
satisfies
the
following inequalities:
(21)
This squares
the
Vp < 1 and
<
above
(21) as
Vx
well with
exercises.
expressed in
<
But
one
a bit
00
regard
too restrictive.
the
conditions
We can
get the
(22)
Vx
<
00
- Vp
The marginal
benefit in
one can assume nearly perfect indexation (Vp ----> 1), provided
Vx has
the required
hypothesis
aims
at
properties
ruling
p, on
out
for
(22)
perfect
the first
to
and
hold(2).
This
instantaneous
indexation of
v on
excess demand
50
Then, restricting
(where v
p),
one gets
as a
"long-run Phillips
Curve" the
corresponds
to
the
inflationary steady
regime. To
glut
regime,
whereas
the
set
of
(23)
Vx
dp
- Vp
(24)
- XB)
d(X d
FL.L.
d(v/p)
P(H'
)2
< 0
H"v
(23) to
regime, and
already
LE in
studied.
the level
the glut
Denoting
of employment
LP in the shortage
(25)
d(X d
dp
is positive
and finite.
negative. from
described in
of (vIp)
XB)/d(v/p)
dL1/d(v/p)
From assumption
we have
(24).
(25) is
(22), the
The numerator
Consequently,
of the
the
sign
second term is
of
the
slope
regime. This
shortage regime.
fact explains
and decreasing
the increasing
in the
function
~lut
relating
which prevails
when the
reverse case
51
CONCLUSION
In the model that we have just analysed, steady state
employment is
maximum when
constant, for
a given
concerns the
Phillips
the price
value
of
of the
the
real
"neo-structuralist" equivalent
Curve".
It
assumptions, but
and deflation,
as compared
is
not
it equally
only
good
rate.
of the
non-vertical,
implies real
measuring these
final
wage
is
This
"long-run
under
our
therefore end
up with
assimilates long-run
constancy.
Then,
even
"immutable", there
if
the
remains in
real
and
wage
steady-sta~e
is
price
regarded
as
fight unemployment.
To get
a very
labour market.
final good
its
price.
fast indexation
Downstream, we
have assumed
a market
for the
intermediate good
these
market has
two,
we
a price
have
assumed
reacting to
that
the
these
two
REFERENCES
Azam, J.P.
(1982), "L'impact
commerciale en
1089-1114.
macroeconomique de
la politique
52
Azam.
(1986).
J.P.
Theorie
macro-economique
et
monetaire,
Paris, Nathan.
Basu. K.
Benassy. J.P.
(1982), The
(1973),
lEA.
Hicks. J.R.
Blackwell.
Lewis.
W.A.
(1954),
"Economic
Development
with
Unlimited
Okun.
Sen. A.K.
(1966) ,
"Peasants
Surplus Labour",
425-450.
and
Journal of
Duallism
Blackwell.
with
or
without
53
p
;II,
r----+.---L
L(s)
The divorce
of changes
shows the
in the
failure of
and monetary
disequilibria into
contemporary
crises
technological
full
impacts
call
and
economists seem
technique of
on
transformations
that the
phenomena only
the
while
radical
under
way.
appraisal of the
requires short
term
of
Monetary
significant restriction
the short
model. Thus.
attention
on thinking
of monetary
votaries
a proper
the
productive
to go
analytical frameworks.
: "Modern
: they
Analysis
assume
the
introduce
most
organization
and
run}
( ... ).
The reader
all the
including the short run phenomena and problems - hinge upon the
incessant creation of new
that. because
of this.
framed upon
this restrictive
and above
all. economic
policy
unless
the heart
of changes
of monetary
different representation
in the productive
theory. This
will require
preference. We
shall thus
(*) This paper stems from: M. Amendola and J.L. Gaffard "The
Innovative choice.
An Economic Analysis of the Dynamics of
Technology" Blackwell. Oxford. 1988.
56
truly sequential
up for
the analysis
of a
process of
change
in
monetary
economy.
The analytical framework
Sequence analysis
requires in
the first
place that
process of
output (Hicks,
and
what happens
the
1973). The
technical
intertemporal
complementarities
of
the
process of
"making of
the machine"
of change
of productive
capacity.
they are
identified with
seen as
produced, cannot
transfered;
a stock
a result
process of
classical tradition
capital and
propose to
of physical
(although an
production.
That
kept
essentially
processes of
it and
intermediate
is
why
result)
going
alive
by
the
shared
by
Schumpeter
conceive capital
available for
exist outside
as a
financing the
Austrian
fund,
carrying them
of
the
to
the
theory
(1934)
made
labour required
production and
back
up
of
of
we
money,
Fund).
in fact,
pointed out
real sequence
is a
process conceived
periods makes
(1973),
economies, that
sequential learning
~s
money
is,
in
becomes
economies
predetermined, although
dates, succession
by Hahn
of decisions.
it possible
environment than
is
becomes relevant
less as
to
available
the character
of liquidity
"liquidity is
not a
in
more
the
information
first
on
period,
the
money
property of
it.
a single
In
this context,
choice;
it is
57
matter of
a sequence
1974. p. 38)
of choices.
a related
sequence" (Hicks
there is
new and
the feeling
or the
different. although
perception
not
yet
clearly
is
liquidity. as
easily revoked
commitment to
a given
course of
action. emerges
as the most
the agents
decrease in
expect
to
the confidence
learn.
the
greater
In fact.
is
the
reflecting
change
it
the old
in the
the appearance
will
considering the
longer formed
in
fact
be
one. although
of a
a
advisability of
same
way.
case for
signal
then
qualitative
that
substituting a
will
agents
are
which will define the new model. have still to be learned. This
will result in a search for flexibility. expressed in the first
instance. as
demand for
we have
waiting, postponement
liquid assets
(money) appears
then as
the
most
appropriate
the
sources
answer.
However. when
process
df
learning
the nature
involved
and
are
considered.
of
the
the
problem
of
implies. may
in fact
detailed
concerns existing
is expected
assets, with
a
binding
correct
decisions
response
opportunities, about
of the
that
when
this
learning
new opportunities
as the
Flexibility.
in
to diminish
the options
then a
and
be
the possibility
result of
this case,
choice that
does not
reduce the
"not
A flexible choice is
future
alternatives
58
associated
with
the
it
most
liquid
in
choice,
this
of
But when
learning can
waiting,
holding
only be
liquid
assets
is
no
longer
the
not become
available independently
of the
choice that
will itself
bring about
new opportunities,
the
of this
of the
passive
concrete expression
shift from
concept of
posture
of
a passive to an
flexibility
waiting,
is
which
the
finds
to start
becomes that
the innovative
their evolution
in time.
disequilibrium, as
expectations to
processes of
In fact,
a result
which
the
of a
production,
the appearance
modification of
existing
capacity
is
and
on
of a stock
long
term
no. longer
between current
a financial
It
in the
It
breaking of
the previous
an essential
makes it
possible
role
in
this
analytical
existence of
an exogenously
given technological
of confidence
in the
for something
59
allows the
the
case for
model,
to
disequilibrium
acquire
kind of
analytical
resulting
expectations can
points to
a qualitative
from
take a
relevance.
modification
monetary form,
the qualitative
The
and this
stock
in
the
immediately
down the
sequence in
in which
only real
difference between
process
interpreted
as
qualitative
presence of
case for
acquire analytical
a qualitative
relevance.
It
change to
also plays
appear
and
to
a marked role in
policy aimed
at modifying
it) on
the evolution
of the
where
a' k
and
c'
81, . . . 8
= (ak
= [C~+l"
1,
The a's
required for
.l, ... 8
rn + H
ak2,'" aka) ,
c]
. C~+H]
are
the
(heterogeneous)
labour
inputs
it out through the m+M periods of its life. and the c's are the
final outputs
the m
periods which
phase of
are assumed
to make
up the construction
particular technology.,
In steady-state.
the processes
will have
with an
60
of the
economy) and
the age
composition of
the
xo(t)
while the
other relevant
period t, will be :
- the employment of the different types of labour, LD(t), given
by
aoxo(t) + Ax(t)
where A = (a"a 2
- total output,
a~+M)
and x'(t)
output given by
where PR(t)
is
the
price
(routine) processes,
and
equal
[X~+l(t)
, ...
to
the processes
the
call it R(t),
one
x~+M(t)l
of
in
output
of
the
existing
steady
state),
and
xu(t)
of production
in the
Wages Fund,
that is
the amount
of financial resources
w'(t)LD(t)
where w'
(Wi'
W2, . . . ws
exogenously determined
correspond to
the different
types of
labour employed
in the
processes of production
- the
money value
61
one
is
of
the
components
of
P*(t)
and
which
is
exogenously determined
-' the exogenously determined inflow of money
~M( t)
and between
~JHt)
which shows
financing
that the
the
consumption)
processes
have
proceeds of
an
of
production
internal
the sales
source,
(and
their
represented
by
own
the
an external
source
~M(t),
where
W(t) + Q(t)
which shows
only routine
income
available,
since
there
is
no
search
for
flexibility;
P(t)
= eP*(t) =
(l+g)P*(t-l)
which shows
that the
each period
is determined
value of
aggregate final
on the
demand in
62
the latter results from a growth rate equal to the one realized
in the
previous period
state) .
Assume now
state sequence.
expressed by
that there
This will
is a
a reduction
in the
<
A flow
Rs(t)
i1M(t) - O(t)]
o(t)[W(t) + O(t)]
P(t) >
P-(t) -
which implies
R(t) -R-(t)
assumed that
that stocks
of final
output-
price changes
do
not
take
place
within
each
period.
This
considered by
formulating
short
(rightly) as
case the
disequilibrium
the producers
terms
the signal
producers give
appearance of
consists in
either
result of
expectations
of a
(i.e.
mistake
eP-(t.
structural change.
a quantitative
(wrongly)
be
in
or
In the first
interpretation to the
investment targets
according to
can
as the
revision of
the
final
production
and
interpretation to
what is
going on,
and their
aoxo(t) + Ax(t)
(2)
P(t)
(3)
W(t)
w'(tlLD(t)
(4) P-(t)
(5) ott)
(6)
W(t)
PR(t)R-(t)
P(t) - W(t)
its
63
where SR(t-l)
[R"(t-l) - R(t-l)]
unvoluntarily accumulated
by the
shifted to
period so
the following
>
consumers and
automatically
pet)
(8)
pet)
where KR
price changes
in each
w'(t)
(under
PR(t)W'
the
hypothesis
of
fixed
real
wages) .
When the economy starts moving on an innovative path,
the profile
change
of the
at
each
expression
of
(innovative) processes
successive
the
step,
ongoing
of production will
this
change
technological
and
being
the
productive
(t)
[b k
(t), b k 2 (t),
; d(t)]
... b k
...
(t) ]
0,
1, ... n+N
input and
with
taking place
of the
new
in time,
innovative processes
the moment
the
an innovative
productive
problems,
choice was
first
made
up
to
the
of research
specify on
the way
the profile
of the innovative
those processes
new skills
possible to
; thus
leading to
and qualifications
devise
and
which will
implement
themselves make
altogether
new
forms
it
of
64
production (and
consumption). A
productive options,
resources,
sets in
is
process of
associated with
creation
a modification
of
of
new
human
as the
result of
an innovative
choice
and
of
the
at time
fact the
labour availability
the different
the human
successive period.
is in
be the
skills of
vector
whose
the heterogeneous
Each element
result of
at
the
beginning
of the vector,
of
each
in each period
where gh
is the
proportion of
of skill
h due
process. We
upgrading resulting
assume, further,
existing ones
size of
to the
in each
dimension is modified.
Given the
the skills
devise, the
on the
labour availability
type and
be carried
range of
on in
each given
can be started and/or kept alive (and in the amounts) for which
the
req,uired
proportions.
labour
However,
inputs
this
are 'available
constraint
is
in
the
modified
right
by
the
moment T
= 0,
(1973), we
shall call
65
During the
that is.
when
production has
magnitudes of
the
preparatory phase
output
not yet
of
the
appeared on
the economy
(from T
innovative
=m
0 to T
processes
the market)
of
the relevant
be :
(la) LD(t)
where yo(t)
is the rate of starts of innovative processes and,
yc(t) is the vector [yC , (t),yC n (t)] whose elements represent
the
innovative
processes
of
different
age,
still
in
the
( 3a) \.[( t )
pet) - wet)
(6a) wet)
(7a) P-(t)
~M(t)
- OCt)]
(9a) PR(t)
that is
early phase
(from T
m+1
to T
= m+M
where yet)
[Yl (t),
elements represent
all the
is
innovative processes
vector
whose
of different
where y"(t)
I(t-I) - I-(t-l)
>
0, where I
66
(3b) W(t) = w'(t)LD(t)
(4b) P-(t)
PR(t)R-(t) + Px(t)I-(t)
(Sb) Q(t)
P(t) - W(t)
(6b) W(t)
where Sx(t-l>
(7b) P-(t)
(8b)
eP-(t)
P( t)
[1 + g(t-l)]P-(t-l) + PR(t-l)SR(t-l) +
Px(t-l)S:r(t-l)
PR(t)
(9b)
px(n+l)
= 6(n+l)P(n+l)/CI(n+l)
= p(t)w'
= T(t)P(t)/PR(t)
= 6(t)P(t)/P:r(t)
R-(t) = T-(t)P-(t)/PR(t)
R(t)
I(t)
(lib)
I-(t)
where Toutput
= 6-(t)P-(t)/Px(t)
and 6of
express the
the
routine
respectively, and
T(t) + 6(t) = l,T-(t) + 6-(t)
and 6(t)
= 6-(t-l)
During the
all the
output of
the
the innovative
(from T
have been
= b o ( t ) y ( t) +.B ( t ) y ( t )
= m+M+l
scr~pped
processes keeps
innovative
for
the
processes
late phase
routine processes
preferences
consumers~
and
onwards when
and
flowing on
only
the
to
the
67
(2c) P(t)
(3c) W(t)
Pz(t)[d'(t)yU(t) + I. (t-1)]
w'(t)LD(t)
= Pz(t)I(t)
(4c) P-(t)
(5c) Q(t)
P(t) - W(t)
(6c) W(t)
r(t)[p-(t-1) - pz(t-1)Sr(t-1) +
~M(t)
- Q(t)],
with r(t) >1
(7c) P-(t)
= eP-(t) = [1
(gc) Pr(t) = Pz(t-1) +
(10c) w'(t) = Pz(t)w'
(8c)P(t)
On the
+ g(t-1)]p-(t-1)
+ Pr(t-1)Sr(t-1)
Kr (t)Pr(t-1)[I-(t-1) - I(t-1)/I(t-1)]
given by
vectors x(t),
the values
of the
u(t), y(t),
whose elements
scalar yo(t),
v(t), the
represent the
and
latter being
of
the
the vector
output (both
growth
the
(first the
a reduction
processes will
nearness
expected
stringent
in
financial
utilization phase
there is
less
and its
there is
constraint. When
the
of
an index
its amount
be
older, then
of resources,
the
first
to
the younger).
When
cut.
followed
by
the
a routine choice
The analysis
will be
stochastic simulations
would happen
developed
that will
under alternative
exogenous variables
and the
by
means
of
non-
parameters of
terms
the model,
of
the
set at
the articulation
of the
sequence in
68
in order to bring to light the relevant moments and connections
of this sequence.
On a
routine path.
The labour
is already perfectly
available financial
activity of
affected
resources that
the economy,
mainly
by
determine
the
levels
of
the
values
taken
by
the
exogenously
the lower
final demand
and/or
existing state
is
- which
reflecting,
(and 0)
~M
follows a
reduction in
the value
in
of r
of affairs
to be.
The growth
growth rate
production
scrapped, and
this
in
will
however,
the
go
implies
utilization
on
as
long
that
phase
as
the
some
must
be
inherited
the corresponding
degree of
reduction in
the growth
stronger financial
investments and
rate of
AM. This
constraint, a
a fall
smaller
in current
will result in a
Wages
Fund,
production. The
lower
immediate
period
the
disequilibrium down
carrying
the sequence
of
the
will then
resulting
cause the
stock
growth
at which
the sequence
actual growth
situation. Two
the economy
which has
phases can
economy converges
brought about
rate, thus
is kept growing at
trying to
the
reduction
return
be considered.
to
the
in
the
original
oscillatory movement.
69
in the rate of starts that have occurred during the first phase
begin to
effects
produce their
becomes
explosive
enormously:
and
disarray. The
attempt to
level
the
the economy
the
of
indebtedness
gets into
an intolerable
get
to
back
movement
oscillatory
potential
increase
state
of
growth
by
because it
induces an
increasing variance in
of instability
fluctuations in
in the
the productive
sense that
it
capacity that
brings
about
innovative path,
the effects
of
learning
must be
human resources
releasing of
taken into
by bringing
account.
into higher
Learning
affects
choice finally
depends. However,
the
learning
follow the
there is
a human
resources
actual growth
constraint at
available
effectively
be
for
used.
The
rate of
work, not
employment
in
resulting
fall
L\.M (and 0)
the economy.
all the
When
financial
production
in
is
final
can
demand
way in
in the
model, and
of the
economy that,
preparatory phase
when learning
disappear. From
is stabilized,
and even
in production
beyond -
last all
through
the
causes the
labour constraint
to
rate of
70
available to replace the external ones). thus making possible a
reduction of the indebtedness.
scenario portrays
The second
AM
If
(and 0)
the initial
o.
to return
the economy
of
the attempt
the
human
fluctuations in
constraint
is
particularly
and/or
the
strong.
also a
the phase
serious increase
lenghtening of
scrapping of
of construction.
in
the
the period
degree
of
processes of
such as
production
to result
indebtedness
and
in a
in
Conclusion
While it
of money
one is
133), the
both the
the sequence
undergone by
appears as
analysis carried
for determining
time of
is certainly
which portrays
the economy.
a policy
out shows
nature and
In
aimed at
this
1973 p.
process
context
rendering an
evolution in
of
change
monetary
policy
innovative choice
of the
intervention rather
71
BIBLIOGRAPHY
Amendola, M. and J.L. Gaffard (1988), The Innovative Choice. An
Economic Analysis of the Dynamics of Technology, Oxford,
Basil Blackwell.
Hahn, P.H.
Parkin
and
A.R.
Economics,
Nobay
London,
(eds.),
Longman
Essays
Group,
in
in M.
Modern
reprinted
in
Hicks, J.R.
(1974, The
Basil Blackwell ..
Lundberg,
E.
(1937),
Studies
in
the
Theory
of
Economic
(1934),The Theory
of Economic
Development,
intertemporal
infinite
horizon
problem
and
where
the
to
provide
assumption
widely
models. Up
to now
original
confusing.
It
used
been contradictory
can't be
[1974]
the
saddle
original
ruled out
path
expectations
from Brock's
and
Rogoff
[1983],
of
Gray
is rather
Obstfeld
[1984],
ruling
[1983]
an
rational
framework the
and incomplete.
Rogoff
in
which evolved
exception
this simple
possibility of
allowing to
the
the separable
Even in
Brock's [1974]
for
monetary
(Obstfeld
has, with
Obstfeld and
in
Brock
concentrated on
growth case.
by
justification
the literature
contribution
[1984], Obstfeld
been the
introduced
out
proved
conjecture,
results have
The debated
question
divergent
that,
in
price
has
paths.
opposition
hyperinflationary
to
paths
sufficient to
rule out
hyperdeflationary
paths
steady
state
equilibrium paths
may
may not
occur
only
be unique
in
infinite
time)
(Obstfeld [1984]).
and
In a
deals with
non zero
74
monetary growth
necessity and
model. The
sufficiency, for
optimizafion problem,
Value Condition
of the
the
consumer's
intertemporal
[19821, but the conditions of their theorems are not met in the
monetary model
of equilibrium
price paths.
We first
price paths.
conditions and
a set
of
unstable (divergent)
different from
necessary
price paths.
come from
analyses
[19831, Gray
give a
of
which substantially
conditions
sufficient
to
rule
out
are markedly
hyperdeflations
[19841 and
set of
These results
corresponding ones
These differences
previous
Then we
from the
modifies the
(Obstfeld
and
Rogoff
Obstfeld and
importance for
prove entirely
the long-run
demonstrating
the
between real
money and
consumption.
optimality of
decrease of
[19831),
substitutes relationship
Then we
Rogoff
hyperdeflations of
the
speed
of
2. The model
Let us
an infinitely
The consumer
continuous
consider an
lived representative
faces a
function
constraint M(t)
price level
of
P(t)
time
(y -
and
c(t
an
+
is a
piecewise
instantaneous
budget
consumer's
the net
constant
transfers from
exogeneous
income.
The
Government
created to
constraint 9MS(t)
the consumer
according
to
its
budget
T(t).
75
mIt~
MaxI: u(c(t).
e- 6 t dt
s.t.
(1)
M(t)
(2)
( M( t). c ( t
(3)
M(O)
u :
R2+
R+ the
6 E
PIt)
(0)
(y E
cIt~
+ T(t).
R2+.
Mo.
->
c-{t)
y. all t E [0,00]
(S)
WIt)
Note
that the
budget constraints
R2+
-> R
is twice-continuously
differentiable and
I:
u~uc~
ucu~~
>
O.
(b)
ucuc~
u~ucc
>
O.
i.e.
O.
(A.S)
(a)
y > O.
(b) Mo > O.
Assumption (A.4)
which.
despite
the
necessarily satisfied
price level
enters the
states a
normality
very natural
assumption
requirement
(A.3)
is
not
via the
real
money
(A.4)
is
equivalent
to
direct
assumption
on
the
76
properties of
we make
the utility
no assumption
function:
on the
conditions" since
function case
such assumptions
properties of
u=/u c
in the
non-separable utility
would be
are the
which
Note that
called "Inada
o.
+ mu c = >
Uc
the
useless:
sufficient
the
conditions
consumer's optimal
with
step to
for
{c(t),
M(t)}
to
solve
the
transversality
condition
at
infinity.
While
the
and Kurz
[1971]), Halkin's
"terminal value
the concave
[1974]
condition" (TVC)
valuation function
example
is not
shows
that
this
always necessary.
In
[1982] showed the necessity of the above TVC. However they used
assumptions which
model:
their "tpchnology
and their
when the
are not
always satisfied
set" depends
"interiority assumption"
monetary growth
level tends
rate is
in
the
monetary
does not
hold at infinity
negative or
is available.
Hence we
prove here
new
theorem
providing
necessary and
and
{c(
be a
t) ,
fixed function
'"
M(t ) }
of
candidate
time,
solution
Obviously the
control path
ERa
to
(0).
E [O,t],
Definitions
(D.1) An
-v
ALhLl~~~h~e
is a
comparison path
{c(t)
a parameter
value a.
all arbitrage
function h(.).
77
(0.2) n(h(.
is the
corresponding to
(cit), M(t)}
set of
a given
feasible
arbitrage
function h(.),
schemes
i.e. the
set of
t E [0,00].
(0.3)
A(h(.
is the set of
I:
PIs)
h(s)ds} E n(h(. .
Only the
other
families
feasible paths
:f
that n(h(.
interest.
(or.
of
arbitrage
( a)
(b )
of
Problem
A(t)
A (t)
X (t)
if
.G:$ if
.G
if
~ ( t)
are
N(O)
R-
A(h(.C
R+
of
(N(O)
( wit h
J (t) =
at
is
continuous
:f
P ( s) h ( s ) d s )
is a neighbourhood of 0),
,...
- ah(t)}
the criterion
depends only
on the
parameter a
and we
write
v (a )
I:
value of
path (cIt)
exists
J ( t)]
A(h(.C.
Necessity.
h(.), the
there
- un.! P ( t ), all t
any arbitrary
A(h(.
iff
all t
G = 1 i m [A (t) e - <> t
t-;>oo
exists and
.G
(0)
such that
P (t) = u c
(c) for
~.
:f
A(h(.
a solution
function
including
i.e. such
'"
'"M(t)}
(A.2), a path (c(t),
equivalently,
schemes
solution,
u ( c( t) - ah (t) ,
PIt)
iff
E aArg Max
EA (h ( . ) )
78
Under assumptions
(A.1) -
I:
V' (0)
Integrating by
[-U c hIt)
parts the
P( t>
obtain :
100
0
100
h(t)q(t)dt - lim J(t)e- 6t t _____ e- 6
t->oo
PIs)
u~
V' (0)
-u c e- 6t + PIt)
where q(t)
I:
and '"
M(t)
Since cIt)
u~
e-
(s- t1
ds,
6sds ,
PIs)
are strictly
positive for
all
I:
h(.)
such
O.
for
~N(O).
q(t)h(t)dt
q(t)
= o
J(t)
such that
that
To
11 (t)}
is to so I ve (0), every
Hence there
must
res~ect
(since k
constant), we obtain
is a
The necessity
of
the
time and
Euler
must
k E R
exist
q(t)/P(t) with
to
P(t>h(t>dt
verify
such
that
uc/P(t>, derivating
result to 0
equating the
equation
(T.1
(a)
(b
is
established.
From the
V'(O) above
M(t)}
is
necessity of
we deduce
to solve
(a) and
that V'(O)
problem (0).
(b) and
must equal
The necessity
results
(-G)
if
on
{c(t),
of (c) follows
trivially.
Sufficiency. By concavity of u (A.1), u('C(t> , }l'(t)/P(t u(~(t)
u~
VIOl - VIa)
PIt)
aJ(t).
- aV' (0)
79
V is obviously a concave function and the sufficiency
of (al -
(bl -
us with
used directly.
(cl
implies the
t-> ...
Theorem 2.
LimA,(tle- 6 t H(tl
t-> ...
Necessjty Let
fr..Q.Qi.
[O.T]' g(tlP(tl
J(tl
iff T.1.
= H{t)
(cl holds
"-
P(tlg(tl
M(tl +
all t ~ T.
M(t).
i:P{s)g(SldS
MolT. all
Obviously
for all t
= [- 1 a 1.
T. 1.
t->OI
(c) - >
that when
6 t
t->OI
given that
and A('h(.C
R+ -> G
Let
simply
us
= 0,
leads.
6t
for
A(hL C R- -> G
O. Q.E.D.
note
to
conclude
here
that
=a
4. Equi libria
x(t)
We shall
A (t)M(t).
Differentiating with
equation (T.1
(al -
respect to
T.1 (a),
x(t)
uom.
A candidate
Equilibrium Path
(CEP) is
path
80
(x(t)} defined by
x (t )
(6 )
u,"
(9 + 6 -
(y,m
x(t),
Lemma
below
correspondence between
It will
enable us
will
a price
to speak
establish
one-to-one
.l...emm.a-1..
Under assumptions
(A.1)
(A.4), x(t)
From T.1
(a)
(P(t) u cc )-l(u c
0,
+ mu c ,")'
(u c + mu c ,") > O.
<
is a strictly
Now
dx(t)/dP(t) = -
it follows that
(P(t-2M(t)
(u c + mu c ,")
O. QED.
is equivalently:
all t
0,
such that (y,MS(t)} solves Problem (Q) where the consumer faces
(P-(t)} parametrically;
(b)
candidate
equilibrium
path
V-
{x-ttl}
(CEP)
(x-(t ~
all t
lim x-(t)e- 6 t = 0
t -> + ro
(M-(t)/P(t
= ~(P(t,
It would be
Let us
write x(t)
easy to
= uc(y,M-(t)/P(t
show that
the feasibility
x-(t) =
V-
condition
(x(t ~
imply that
The separable
Uy
It
write x
(Figure 1.8)
0 corresponds
the
>
below (Figure
cases where
and
1.A) together
\>(0) =
0 but V'(m)
to
O.
is pictured
other possible
:f
u'c(y,m)
V' (m)
~'(m)
is larger than
is lower than u y
).
81
v(m)
FIGURE 1.A.
______________________________
~m
v(m)
u (y,O)m
y
v(O)
v(m)
FIGURE 1.C.
82
SinceV : R+ -> R+ is a strictly increasing function,
there is a well-defined, strictly
V- 1
'\J
=-
u'"
increasing
(y,"y- 1 (x
inverse function
Uc
u",/u c
equation in
x,
the
solutions
x(t)
(1)
Lemma 2.
(6
of
-X (x(t
which
are
the
Candidate
Under Assumptions
x(t)
(A.1),
(A.3)
and (A.4)
the
only
e ;
6 +
Divergent
(Hyperdeflationary paths)
X'
or dx(t)/dt
(x(t) )x(t).
that
<
dx(t)/dt
(A.2)
from
But
(Hyperinflationary
: we
dx(t)
it
dx(t)/dt > 0
p.a.t.h.s) .
~.
x, where x is
it! t)
dx(t)
is
V-
easy
1
to
show
is a strictly
o.
Lemma 2
follows trivially.
The main
it allows
as have
key to
this result.
multiplicity of
equilibrium paths
it turns
in Obstfeld
out that
the
paths) which
would become
possible with
consumption being
Giffen good over some ranges of values of m.
We can
now proceed
to
study
the
set
of
Full
83
L~mma
3. Lim
t->oo
x(t)e- 6 t
0 ->
i: f(x(sds
r-
f(x)
Xo x(cS-f(x
- .....
- e, x = lim x(t)
t->oo
to t yields
f(x(sds.
e)t -
r"
r"
rK'
mUm
84
path.
hyperinflationary
E.r..wU..
(a)
equivalently written as :
(8 )
+ e)
V (m)
Equation
CEP.
may
be
- mUm
~ '(m)
>
0,
all
hyperinflationary
o.
CEP
to
be
is
lim(m) < O.
m->O
one concludes that
since
(7)
0 ; a similar argument
0)0
X->O)
f(x(sds
<
<
0 ->
f(x(s
from
Lemma 3,
<
0, all s >
S, and
the transversality
.(x) :s -k[f(x)]"',
X->O)
then
fIx)
X
f'(x)
:s - - - -
kf(x)'"
<
i:(Xo)
df(x)
kf(x)"'(cS-f(xo
f(xo)l-",
k(cS-f(xo(l-a)
<
+0)
T.3 (a)
is a sufficient (and necessary) condition for ruling
out hyperinflationary CEP on infeasibility grounds (obviously,
the transversality condition is always satisfied along an
85
hyperinflationary CEP).
One should note that the condition
~.
< + 00,
or
< + 00,
mUm
then lim --- > 0
m->O Ue
The
reader. If
~.
86
I:
sufficient that
f(x(tdt (to show that this is not true,
let f(x(t = lIt . . ). The economic interpretation of T.3. (c)
is the more straightforward in the zero monetary growthseparable utility function case
it then means that the
implicit (or subjective) rate of return on real money balances,
u~/uc, has
an elasticity with respect to real balances m which
is bounded above by increasing function of m
under this
assumption the rate of return on money holdings decreases
quickly enough while real balances increase so as to make its
integral converge along any hyperdeflationary CEP. Note that
T.3. (c) implies Brock's [1974] sufficient condition for ruling
out hyperdeflationary CEP: f(x) S x- k Let us take a = 0 in
T.3. (c), and, now
E. (x) S -k, all x <! xo, imply that log
(f(x S -klogx and. hence. f(x) S x- k
Theorem 4. Under Assumptions (A.1) to (A.5)
mu~
if lim
o
m->O U c
belongs to the equilibrium set 0
(a)
<
0 ->
I:
It follows that I
I: f(x(sds
<!
I:(X
o )
df
kf(x)a(o - f(x
f(x) l - a
[---------]f(xo) = + 00.
(1-a)k6 o
Now the transversality condition is obviously satisfied.
>
87
L..~mlluL5..
If (a)
u~
lim
m -> 0
<
OD
Uc
mu~/uc=O
or
<
and
00
u~
+
Uc
00
Uc
(y,O) = +
00,
or
00
below :
u(c,m)
[cxc +
~m
+ log (W m + JJ
1,
0 <
<
1, cx,
~,W,
JJ > 0,
where
u~
Uc
~ W m + ~JJ +W
aW m + CXJJ
W )/cxJJ.
generated by
88
The
second
part
of
T.4 (bl
condition:
f(xl
A similar
case is
suggested to
Fernandez.
[ko: 10g<1 +
f(xl
function
previous statements
transversality
in the
condition
(k,o:l E R+x[I,+oo)
[logx)-l,
the
Rogoff (1986)
Obstfeld and
These
xl)-1/~,
are
so
example
by G.
many
enough
to
R.
counterexamples
to
rule
hyperdeflationary CEP.
APPENDIX
I:
f(a,tldt.
V'
(a
df
(a,tldt.
was
Calvo and
literature according
is
which
to which the
out
all
89
THEOREM
Let yea}
I:f(a,t}dt, f
If:
(i) f
af
( ii )
~a
(iii)
[- , ]x
wo
-> R+
[- , ]x R+
is continuous,
-> R+ is continuous,
I:f(a,t}dt exists V a E [- ,
( i v) Sup
aE [-, ,, ]
af
T.
1.
---
oa
(a,t}dtl
-> 0 as T,T'-> +
Conditions (i),
00
(a,t)dt
u(.
exists.
1 im
t->oo
Proof
a) from concavity of the utility function u(c,m), a E [- , 1
->
f(a,t)
is concave t
since
Jt
(b)
90
is concave
aE[-C,l
1--1
a E [- /2. /2],
L(-) - L(-f,/2)
f:,/2
da
dL
Max
[I
L() - L( /2)
/2
I.
1]
We can now
under the sign
S kTT'
use
the standard
theorem of
derivation
(a,t)dt
(a)
da
Since
dL(a)
1--1
da
I:'
S kTT'
/2] it
df (a,t)dt
da
[-12. 1 2 ]
But kTT"
the existence of
I:
T,T'->oo
1=
0,
Q.E.D,
VaE[-,l
k TT
= max [1--1,
da
L( ') - L(
/2
12)
11 with
concav i ty, ,
It is not difficult to show that
dL(a)
1--1
da
S k TT
, by
91
da
zero
as
time
tends
to
infinity.
Q.E.D.
REFERENCES
D'Autume, A. and P. Michel: "Transversality Conditions. Budget
Constraints and
Equilibrium
in
of a Perfect-Foresight
Model".
European
the Determinacy
a
Monetary
Growth
Benveniste, L.M.
Dynamic
and J.A.
Sheinkman
Continuous Time
"Duality
Models
Optimization
Case", Journal
Theory
Economics
of
of Economic
for
the
Theory. 27
( 1982). 1 - 19 .
Brock, V.A.
Foresight" International
Economic
Review.
15
(1974),
Monetary
Model".
750-77.
Brock, V.A.
"A
simple
Perfect-Foresight
: "On
Models
of
Money
and
Perfect-Foresight".
: "Dynamic
Models:
Instability
in
Rational
Expectations
92
Obstfeld, M.
"Multiple
Stable
Equilibria
in
Perfect-
and Rogoff,
Maximizing Models
K.
: "Speculative Hyperinflations in
17 (1986). 349-
62.
Tirole, J.
"Asset
Bubbles
and
Overlapping
Generations",
THE RANDOH
Yhen CLOVER
double constraints
double function
(1967) establishes
in the
which is
consumer objective
fulfilled
by
he recalled the
money
as
medium
of
selling
monetary balance,
of
the
in contrast
products
it is
gives
the
agent
equal
marginal rate
to
one.
It
of substitution
follows
that
between the
optimally,
two
the
corresponding
hence possible
account, by
P.
to take
associating to
DAVIDSON
(1973),
D.
the two
them a
FISHER
and
functions of
characteristic
more
lately
by K.J.
consumer choices
not
LANCASTER who
to
the
goods
but
to
the
of the
financial assets
LANCASTER's
approach
to
the
ARTUS et
J.M. ROUSSEAU
(1987
and
secondly
to
94
As regards this last point we have already formulated
the main
results in
the case
of certainty
(J.M. ROUSSEAU
1984) .
We deal
characteristics of
the assets
of this
),
money/bond is done.
In order
of certainty
and uncertainty,
we will
the agent
expectation drawn
is supposed
from this
Under
characteristics
(J.
J.
LAFFONT
1985) .
The sharing
deduce money
other side.
two sorts
money/bond in
demands on
allows
to
The obtention
of assets
the portfolio
of the
can be
done by
characteristic approach.
The
holding
of
an
asset
allows
to
obtain
A(l +
which yields
the rate
r allows to obtain
2. A liquidity service
As the
agent decides
or
AU +
where I
1)
=r
- c(l + r) or
approximately I
Let
r - c
95
store
value
of
composition of
the
agent
is' reaching
his portfolio
so as
to
determine
to maximize
his
the
utility
(1 - 9}w
and
The problem
9w where 9 E [0,1]
consists then
to determine
the optimal
value of
(01) and
a store
of
value
service
(02)
according
to
the
following array.
+ r)
(1
TO + 1)
p
M
M
p
Portfolio
02
01
p02
w(l + 9r}
Under the
constraint rOl
l.p.02
(r -
l}w which
(r
1)
Yo
1)
.w
and
-1
96
e-.
e-
--r.l
for
e-
>
1 + 2r
If c
<
r +
Money is
dominated by
yield rate,
r, liquidation
cost c, and
of the
agent is
function expectation.
The agent objective is then
(r
l)w
97
Ee:-)
This value
interval (0,1).
can be
As usualy
shall
study
the
first
order
condition
2{r.T.w + (a 2
now study
the
influence
of
lip.
the
and
1st. case
cov(X,Y)
=0
v X,Y
r.T
The examination
the following
portfolio is
results
first order
of the
if
uniquely composed
>
condition gives
1) then the whole
Money is completely
T < o.
98
The
condition
expectation of
expectation)
<
the
(i.e
nominal
yield
is
a necessary
optimal portfolio.
~he
For insuring
come again
to conclusion
case.
If
when
the balance
then 8
< 0,
demand not
identical to
M
to be zero, we
those of
the
certain
(there will
<
be some
if c
r + -------.
>
If c
2r 8
will
1 + 2r
always be zero
In a more general way we can then state the following
conclusion :
In the case where the dispersion of the values of the
random variables
be zero
or neglected)
the optimal
sharing of
the
portfolio
of
will be
c (c
will
be
intermediate values
<
composed
of
exclusively
c (c
of
conclusions
2F) the
money.
For
>
for
are
strictly
similar
c.
to
those
cov(X,Y)
=0
+ r
= 1..----------
We have
X,Y
except for X
Y=r
99
69
sign
6r
The share
of bonds
will increase
with the
average
69
sign
61
The share
of bond
demands decreases)
liquidation cost)
The two
with those
use of
increases (and
when l{average
nominal yielding
- average
increases if r2 > 02 r .
results should
be compared
and
contrasted
a quadratic
liquidity in
utility
the optimal
function
portfolio
leads
which
to
decreases
share
of
when
Furthermore
69
sign
6c
The
share
liquidation cost if r2
of
>
bonds
order moments
first case
by
with
the
average
02r (2)
Integrating uncertainty
the first
decreases
through other
modifies the
elements than
question:
money in his portfolio 7" the answer is always the same because
money dominates
bonds as
100
by
1 =r - c
<
0).
Taking
elements
other
account
into
for
facing
3rd case
In this case, the formula of
r
becomes
possible to
prove that
contains money
despite the
positivenes of
(1
yielding of bond
In other
by the
net
expected
0).
>
words, although
portfolio.
As an example let us take the case when c
It is
that
r>
then possible
0 and
r.T
necessary
02r
to find
ar + b
cov(r,c) < O.
condition
for
having
there
two
> Or since a
It is
when the
hence possible
suppose
the bond
T =0
provides the
of this
(i.e
r = c).
same service
The difference
If the
risk of
interest rate
risk of
liquidation
cost
the
Or
whole
will
be
can be
extended to
the one of
c.
Taking
certainty
standard
allows
to
deviation
predict
into
some
is greater than
account
portfolio
under
an
sharing
not
conclusion
characteristics allows
a portfolio.
It is
and a
demand
bonds
the
approach
by
the
random
then possible
apart
from
conventional
portfolio
theory.
But, furthermore
it is
also possible
is
hypothesis we
nevertheless
have formely
necessary
assumed
to
about
release
the
some
magnitude
of
covariances.
More
influence of
explicitly,
it
a price variance.
relatively to
is
necessary
to
assess
the
If 9 is homogeneous of degree
take place
I and
through covariances
c involved
the investor
in
the
utility
with
the
maximising
is assumed to be constant.
Taking into
consideration the
random
character
of
service yielded
optimal sharing
to depend
on other
102
REFERENCES
& Viallet,
Aftalon, F.
C.
(1977)
- Theorie
du portefeuille.
& Rousseau,
P.
caracteristiques
(1987)
J.H.
"Valorisation
qualitatives
Communication au
Colloque de
des
des
obligations"
l'A.E.A. sur
les modeles
(1987)
H.
Les
vertiges
de
la
finance
of Monetary
; reimprime
in CLOWER
Ed.
P.
Money
(1973)
and
the
Real
World,
London
MacMillan.
Fisher, D.
(1978) -
(1985) -
Economie de
l'incertitude et
de l'information.
Paris,
Economica.
Lancaster, K.
(1966) -
"A New
Approach to
Consumer Theory",
J.H.
(1984)
portefeuille:
-"Services
le cas
de
monetaires
deux
actifs."
et
choix
de
Document
de
A SURVEY
The classical
appear irrelevant
when the
problem
at
hand
is
related
to
lending, and to the credit market. The limits that are imposed
to the level of a firm's debt by its creditors, or the effect
of a
firm's capital
would still
remain unexplained
credit
characterized by
market
the
( 1 i ke
labour
the
heterogeneity
of
the
market)
agents.
If
is
the
lent and
be obtained.
up taking
place in
unique
market
where
the
relationships between
two reasons
in
able to
modify the
charac~eristics
and from
the
credit
104
105
106
if
control of
the firm
and/or 2),
the firms'
creditors take
repayment to
its
types
of
models
have
established
the
use.
It
depending on
viewed as
is possible
to distinguish
the characteristic
primordial
the
of
the
revenue
two
approaches,
bankruptcy
stream
or
that
the
is
firm's
control.
2. Optimality of the standard debt contract
Two models,
have established
contract to
Still, they
by Townsend
(1979).
information concerning
will examine
auditing
for the
standard ctebt
firm's cash
quite similar,
first developed
asymmetry of
features are
sufficient conditions
be optimal.
Hellwig (1985)
that we
.whose main
first,
flows provided
cost),
while
the
the
models
firm's
assume
cash
an
flows.
he pays
this
Both
a cost
is
(for instance
impossible
in
an
Diamond's
framework.
Since
this
model
revelation principle
use of
uses
dominant
strategies,
the
direct mechanisms.
The firm
whether to
not.
would imply
(Mixed
a very
high penalty).
A debt
of the
contract will
cash flows
cash flow
in Y
Then, for
the cash
incentive for
that are
it is
announce the
flows restricted
the firm
is to
107
associated with
the minimal
flow y.
Min R(yl
YE Y
For such
a mechanism
to be incentive compatible, we
need
R(yl = R
in other words, the repayment function has to be constant.
The implication
firm does
not repay
observation cost
Therefore,
it
announce a
R. On
so
will
given the
is the
following:
R, the
creditor will
as
check
to
never
cash flow
the other
of this
be
the
optimal
choose to
realized
for
if the
the
pay the
cash
flow.
borrower
to
hand, when
existence of
a limited
liability constraint,
it is
the firm's
cash flow.
the incentive
shown that
compatible credit
contracts.
It
remains to
be
so, Gale
are risk
contracts are
those that
minimize
that the
the
Pareto efficient
expected
observation
given expected
if
the cash
repayment, the
set y,
flow is
superior or
equal
to
R,
then
the
repayment is R
iiI
if
the cash
maximum
that
clause. This
flow is
is
inferior to
permitted
defines the
given
R the
the
repayment is the
limited
firm's bankruptcy
liability
as the case in
which the cash flows are observed and become the property of
the creditors.
108
The risk
role, since
plays
neutrality assumption
otherwise
there
could
be
an
important
trade-off
between
much"
risk
consequence of
the firm
rate of
aversion
the
result
will
still
hold.
will invest
model considers
a bankruptcy
cost
which is
may
the
be
opportunity
cost
of
entrepreneur to
deal with
determines then
simultaneously the
the non
pecuniary cost.
pecuniary,
Since
spent
optimal debt
the
by
procedure.
bankruptcy
the
Diamond
contract and
cost
is
non
efficient contract
with the
will minimize
the bankruptcy
firm must
will be
time
every Pareto
cost. Yet,
the
the bankruptcy
here characterized
flow is
the cash
bankruptcy that
cash flow.
superior to
R and
to the
is equal
pecuniary cost of
firm's point
From the
a non
of view
the
cost
of
the
that R(y)
and
~(y)
Diamond's model.
One of
that is
the interesting
developed by
financial
a contract
between a
firm and
In both
information related
Financial
that it
view,
justifies
cases there
to the
a financial
number of
firms and
asymmetry
of
diminish with
the number
To understand
why there
larger number
exists an
intermediation is
fact that
to a
the fact
Diamond is
this
its depositors.
flows.
implications of
pecuniary
costs
by
firm
will
of firms
we have
109
E(w>
=E
[i!l gi ] - HN
[P(~gi ~
Hn > +
P(~gi
is assumed to be
<
Hn>]
[~
gi
<
HN ]
110
almost completely
risky projects,
financed by
since if
limited liability
the firm
view, the
choice of
the project
tend to
take
will retain
moral hazard
investment projects,
very
project succeeds
point of
debt would
if the
issue is
(1987) a
projects are
model in
available to
which two
every firm,
types of
investment
one risky,
the firm's
in addition, that
the lender
which is
to
quality of
risk averse
is unable
identify
the
two investment
projects
differ
only
by
their risk, the optimal contract should induce the firm to take
the less
risky decision,
the less
risky returns
while giving
stream. This
only with a
models focus
on
the
incentives
in
the
by taking
analyze the change in the firm's property rights that will take
place in the event of bankruptcy. We know that in this case the
creditors will
making and
choose either
to liquidate
or to
reorganize
the
firm.
Aghion and
of the
credit contract.
contracts are
too costly
They assume
to be
used, thus
restricting their
considers only
strategies:
two states
continuation of
of nature, and
the same type of
111
management,
innovation
flow that
obtains is
depend on
the state
action that
state of
or closing
a
that is
so that
nature that
it is
distribution
realized and
on
the
flow. Still,
that changes
variable
of nature
is taken,
realized cash
random
down of
the probabilities
of the
two states
of
nature
investors differ
control the
is 9 1
parties cannot
approach the
issue contingent
it
abandon the
contingent
on
~he
If the two
states
of
on the
value of
correlated with
is more
the state
interesting for
is to
of
nature.
In
other
to compensate
them. The
easy to
debt holders
inferior to
make contracts
that the
variable is
it is
nature, so
words,
in such
nature realized
way to
establish that
will only
be of
some critical
the passing
over of control to
interest if
level corresponding
to the nominal
level of debt.
5. Dynamic contracts and reputation
It has
is less
important in the long run since the fact that the game
is repeated
may induce
may clearly
be the
firm's bankruptcy
to the
absence of
case here,
has an
future lending.
has been
formalized in
the repeated
of reputation
: the
reputation is
the effect on
112
of his
actions into
like an
periods to
investment, entails
a loss
in
the
first
been applied
by Diamond
(1986) to
exists a
represents the
that the
riskless
asset,
and
the
investor,
which
is risk neutral so
investor cannot
observe the
have a
ones that
have a
large probability
larger profit
of default
(risky) and a
their investment
project, that
cash flows
for a
successful 1 firm
risky project.
such that
Also, the
cash flows
of the risky
that go
lend to
sequential
characterized by
its cash
flow is
repay their
Bayesian
is
then
shown
to
exist,
debt (ii)
the debt,
if it is inferior
the firms
(i)
if
prefer to
limits its
loans to
113
effect,
with
the
strategic
firms
building
To conclude,
it is important to understand that all
this litterature is at its start. The different characteristics
of the debt contract are now better explained. On the one hand,
auditing or observation costs for the cash flows will make
optimal for the firm to issue standard debt. Yet, if this is
the case, the moral hazard issue related to the firm's choice
of rnvestment projects is worsened. Taking into account this
problem leads to an endogeneous characterization of the capital
114
structure as
understanding of
why the
control of
the
firm
goes
to
its
creditors when the firm goes bankrupt. All these problems still
arise in
fact that
the firm
It
is clear
will
be
and
in
that in
explored again.
the years
to
come
possibly within
this
dynamic
problem
setting.
REFERENCES
Ashion, P.
et P.
approach
Bolton (1986).
to
"An
bankruptcy
"Incomplete
and
the
optimal
contracts"
financial
(1984). "Financial
Monitoring" Review
Intermediation
of Economic
and
Delegated
414
Diamond, D.
(1986). "Reputation
WP 134.
Graduate
School
Business.
University
of
Chicago.
Freixas, X.
(1987). "On
Debt and
Stock as
Optimal financial
et H.
Hellwis
Contracts
(1985).
"Incentive
Compatible
Debt
and
R.
Wilson
(1982)
"Reputation
and
Imperfect
and J.
Roberts (1982)
115
Sapington, D.
(1983),
"Limited Liability Contracts between
Principal and Agent", Journal of Economic Theory, 29,
pp. 1-21.
Townsend, R.M. (1979), "Optimal Contracts and Competitive
Markets with Costly State Verification", Journal of
Economic Theory, 21, pp. 265-293.
INTRODUCTION
I.
The global
is usually
monetary situation
of a national economy
DLX
where DLX
CLSP
is net
CLEP
HZ + DIV
foreign reserves.
public administrative
sector. CLEP
the
made
private
sector
individuals. HZ
is the
up
of
volume of
internal credit to
companies
and
private
second
half
instrument employed
(total internal
of
the
seventies
major
policy
credit). However.
monetary planning
the
has in
outline of
accordance
balance of
with
the
payments been
the expansionist
monetary
approach.
has
the
fundamentally determined by
or contractionist
nature of
credit
118
A number
type
of
question.
and
identification
of
possible
direct
been
possible
autoregressive
adjusted and
systems
using
modelling
stationary variables
of
new
method-vector
previously
- hereinafter
seasonally
referred
to
simply as VAR.
An alternative
consists
of
specifying
equations. from
which
models
in
for
general
structural
simplified
simultaneous
structures
in
to
nature.
disadvantages
be
made
Weighing
normally
could lead
correct procedure
in an
and
suggestions thereby
dynami~
econometric
necessarily
not
up
expressed
methodologies (2)
specification
is
the
of
advantages
for
each
to situations
of
a
and
these
subsequently,
obtained, to
using
the
information
and
go on to specify a classical
be seen below.
In summary
endogenous
and
exogenous
and
for
excluding
relatively precise
number of
model ;
(1) C.
Sims (1980)
produced an important pionnering paper. In
terms of VAR specifications for the Portuguese economy. only
two references are known prior to this - Barbosa, A.
Pinto
(1984) and Teixeira dos Santos, F. (1986). The first case is a
bivariate model on the relation between inflation and output
while the second is a trivariate model adding those variables
to the money supply.
(2) On this issue see Genberg. H. et al ..
(1984), for example.
119
through
which
macroeconomic
on target
action governing
administration of
be inferred.
There are
discussion
two basic
should
be
reasons why
subject
to
the problem
particularly
under
careful
expectations play
a vital role in
future behaviour
could be
not anticipated
influence the
present
and
future
course
of
target variables.
In
this
context
it
these
simplification
empirical
models
which
is
therefore
e~pectations
require
which
they
hypotheses
distance
might
that
should be adopted.
strong
considerably
realities
suggested
them
explain
from
In
for
the
predominate.
demands that
be forecast.
the future
The
quality
behaviour
of
of
estimation
also on
eliminating possible
structure of
multicolinearity
monetary
and
foreign
exchange
policies
significant forecasting
errors on
are frequent.
Moreover,
in
the Portuguese
between the
seventies and
changes and
financial
probably affected
eighties, there
innovations
the nature
predominant variables
case for
and the
of some
of causal
a long period
were institutional
significance which
stability of
the parameters of
120
we will
begin with
a closer
examination
of
monetary programming.
In a stylised way the programming which has been made
for one
or more
yardstick. To
explanation (3),
increase in
summarise
the
the increase
in
traditional
demand
central
for
H2
bank
which
the
and u
levels
of
administratively)
variation (4),
residually,
rates
(to
be
in
development of
established
accordance
the CAB
with
the
is determined,
forecast
available
on
finally, having
be maintained
period from
objectives or
pursued, and
stages of
the emphasis
with which
they
were
in
the international
this period.
Furthermore, since
the financial
markets,
investment been
alternative to
which is
by the
provided
not
with
bank credit,
1986 with
only
an
has
the stimulation of
financing
important
and
of
private
attractive
position is
(5))
which,
In this
adopted by
placed
in
BT issues
the
primary
market
with
financial
133.
121
central bank,
we will
illustrate them
by summarising
to adopt
122
explains the
for two
periods is
1972.111-1986.IVand
there was
selected
of policy
; in Section III
longer
one
from
another corresponding
most homogeneity
1986.IV
of the
methodology adopted
mixes. from
1976.1
to
(impulse
means of
response
two types
functions)
forecast error
variance). and
also presented
on possible
of analyses
and
DFEV
called IRF
(decomposition
certain final
of
observations are
with trend-free
stringent (7)
i.e. series
information contained
variances and
on and
relations
are finite
with
in data
zero
in a
covariances {n t . a }
summarised.
series in
(1969) causal
ensure that
weakly stationary.
moment (
analyse Granger
such a
mean)
is
removed
sequence of
matrices
and
of
{E(Yty'a)} is concentrated
matrices
depend
on
the
t-s
first
stage
has
just
been
referred
to
it is
sufficient to
the original
series. Seasonal
generally be
eliminated with
influence. when
the
help
of
the
suspected. can
operator
of
or more
(1986).
123
function of
autocorrelation plays
respect, as
is known,
respective graph
or are
the
and in
makes it
not autoregressive
residual
variables
critical
many cases
role
in
examination of
this
the
moving averages in
supposedly
the
stationary
deseasonalised series.
Two
are
tests
frequently
used
to
test
the
the t
analysis
stationality or
As
regards
to
corroborate
and
deepen
awareness
of
the
latter
study
of
the
Durbin
accumulated
periodgram graph and concomitant application of the KolmogorovSmirnov test are particularly important.
All
these
procedures
have
been
adopted
in
the
Y(t)
of
first stage
there is
vector
the functionning
in dynamic terms.
The second
stage consists
the first,
which we
will call
Sims first
suggested it
in 1980,
all variables
of the
second, which
using an
optimal criteria
in general
1982) and
in this
model the
number of
between
As regards
the first
variables.
imposed ad
the Hsiao
adopted.
We will
now go
on to
the question
of choosing the
(8) The results of these tests have not been included so as not
to overload the text. Figures are available on request, as are
other results quoted but not presented.
124
second mode I .
In
formal
terms
VAR
model
proposes
an
Y( t)
where D(t)
is an
i
is the
variable
error
1, ... ,H, as
vector, not
a white
noise
correlated to
type.
The
any Y(t-i),
1, ... ; H)
Ai ( i
the
zero
precisely
restrictions
from
criterion. This
many steps
of
Hsiao criterion
as there
estimation the
which
application
might
the
be
imposed
aoresaid
result
efficiency
consists of minimising,
in as
Akaike statistic
relative to
final prediction
N+P
SSR
N-p
N
It must be noted that adoption of the Hsiao criterion
to specify
allows the
concept of
operationally,
and
Granger causality
(1969) to be defined
brief
on
digression
this
matter
is
Granger
concept
a simultaneous
defines
equation
causality
model
between
according
to
moment t
stationary and
".
'" ...
on the
P,E and
deseasonalised series
transformation of the
respectively obtained by
125
Each pair
different types
identified in
of
identify
In
Granger
a trivariate
causal relations
figures.
of variables
causal
Granger
relations
causal
which
I defines
relations
to the
among
the
be
all these
have been
Portuguese
can
G statistic
relations among
these relations
variables, corresponding
model. Annex
by quantitative
this study
seventies and
will be
applied to
P,
case
and
during
R
the
outlined below.
Merely for
formal accuracy,
the
purposes
the group
of
illustration
of possibilities
and
for the
for
P and R
directly causes
P iff
(if
and
only
if)
the
Annex I,
it is
said
that
directly
i.e. either in
I spurious
II spurious
"R causes 2P, but less than E, and R does not cause 3P
(as in
b)
in addition,
opposed to
b) and
reverse of
b)".
E is
used, R
used the
E causes
not cause
2R and
causes
3E
3R
(as
(the
does not
cause E. However,
criterion suggests
situation could
R does
occur when
that R
if E is not
causes
P.
This
is a "proxy" of E.
e) There is a direct causal relation ("feedback") between
Rand
observed.
whether
relation
a)
or
the
reverse
is
126
f) Finally.
R does
not cause
f.l)
f.2)
It must
saying that
"R
does not
now clear
the causal
problem of
veri fy i ng the
furthermore under! ie
consisting of
the various
outlined. There
nature
all VAR
comparing the
of
these
models. and
is
therefore
relat ions.
there is
the
which
method
which seems
rather significant.
and we
will
the outline
of
the
second
stage
the
[i] [::1.
A( )
[ D( )
G( )
(3 )
where the
procedure
in
steps
for
minimising
the
statistic.
In the third stage the accuracy of the model selected
in
the
previous
stage
is
tested.
comparing
it
to
close
127
most important
explained and
cases in
methodological
which in
In the following
elements
will
be
the
fourth
plausible restricted
mean square
for out
stage
the
predictive
of
error) and
to analyse
capacity
the predictive
~e
also decided
capa~ity
model.
After
this
procedure
definitively
selected
approached, where
(to the
the system's
ortonormalised responses
studied. This
model to
in the
necessitates changing
error variables)
the form
is
of the selected
innovations
(moving
transformation is
to which
response to innovations
average
representation).
This
any stochastic
by the
sum of
of the
moving average
(2) can
therefore be
rewritten as
D(t) + e(t)
(4)
[I - J(L)]
where I
y(t)
functions in
the L
operator, with
orders between
Denoting the
non-deterministic component
and
M.
is
therefore
the
aforesaid
moving
average
representat i on ..
Having reformulated
question to consider
a) How
the innovations
128
explaining
the
variables
forecast
of
the
(decomposition
error
model
of
in
variance
future
of
periods
variance
error
forecasting
the
analysis).
To conclude
this methodological
explanation certain
different
units
the variables
of
measurement.
the
impulses
must
be
the
error
contemporaneously related.
It is
variables
therefore
are
not
generally
possible
to
disturb only one equation in the system each time. with each of
the
possible
diagonalising
residuals. The
n.
unitary
the
innovations.
matrix
of
the
diagonalisation procedure
without
previously
covariances
adopted by
of
the
Sims (in
article of
this
way
the
formula
(5)
can
be
rewritten.
(5')
Wt
= C(L)r t =
I
Ctrt-l
i=O
j. With
this error
variables
is the innovation occuring at t and which
at time only effects the j equation of the n equations in the
system. However.
in subsequent periods the effects of an
impulse on rJt spread through the whole system, propagated by
the C 1 matrices.
As Buiter (1984) observes.
the problem with this
procedure is that there are n!
linearly independent error
=
0,
with OrlrJ
representation. rJt
129
orthogonalisation procedures.
The C l
matrices
can
therefore
Doan and
Litterman (1981)
have stated,
prOcess will
provided
not
the
significantly
correlation
influence
between
the
Cl
matrices,
innovations
is
not
very
significant.
It must
applied rather
the Gram-Schmidt
in fact
procedure (which
it corresponds exactly to
is frequent
in
non-linear
Furthermore,
the
simplified considerably
procedure in
particular,
matrix arising
as
linear
information as
Yl
of
innovation
are
triangular
block
contains
these
the
same
type
of
vectors, established
and
finite variance
data
vectors,
vectors
the data
group of independent Yl
constant and
prediction
non-singular
invariable in
innovation
group of
the
combinations
of
of
combinations are
group
problems
generated by
in
weakly
stationary
the sub-group YJ
1, ... k
the
present
study,
and
to
certain
extent
and
models
were
estimation
procedure,
(seemingly
unrelated
subject
which
to
we
will
regressions),
estimated coefficients
with the
the
so
recognised
abbreviate
as
to
Zellner
as
SUR
correct
the
information contained
in the
all results mentioned in this and the following section are set
130
Y-(t)
131
It was
then observed
(76.I-86.IV) in
were more
previous
which monetary
consistent and
= 8.
establish H
years
was
in
observed that
the interval
was, however,
seen that
prediction error
was considered
for the
sequence
(FPE) very
to be
Table I,
of
to
period
to
to
rise
demanded and
the
it was
exceed 72.III-86.IV.
estimates
It
G statistic
values, the
of
the
final
perfectly sufficient
exaggerated to establish H
a
have
original series,
could not
> 8
sufficient
such
the lags
period
exchange policies
H would
that
for H
shorter
was
extending
view of
data available
and foreign
noted
that in
generally presented
the
coordinated, it
However,
it
considerably. so
limitation of
that for
and
perhaps
even
Annex II,
presents
the
main
statistical
the three
equations estimated
generally acceptable,
except
in each
for
the
period. The
dependent
be noted
both periods
possibility of
suggests crowding
explaining the
DLX fluctuations
by
means
of
0-
= N(N+2)
I (N-k)-l 2(k)
k=l
version of
the Box-Pierce
132
(7 )
(8 )
c(O)]
[i1.
[AlB)
D(6)
[it
[A( 16)
8(6)
E(8) F(O)
G(O) H(6) 1(8)
D(6)
G(O)
respectively for
[i] [::1.
8(3) C (0) ]
E(8) F(O)
H(3) I ( 16)
[i]
1976.1-1986.IVand
[::],
for 1972.111-1986.IV.
It
and (8)
to the
Hsiao criterion
and therefore
in
theory should not be modified, which means that the third stage
referred to
should
application of
minimums for
than the
excluded.
the Hsiao
G for
Furthermore,
criterion shows
each equation
if
that
are not
the
empirical
absolute
significantly less
fewer lags,
of the
be
the problem
additional lags
other hand,
than what
if M
has been
corresponds to
first step
of the
Hsiao criterion,
of G
the significance
of
the
therefore our
or excluding
the third
clarifies a
heuristic
criterion
methodological problem
just
suggested
which has
suitably
correctly answered.
This problem is important in so far as it might arise
whenever the case under study is characterised by the existence
133
and I2
covariances obtained
by OLS
estimation applied
to
each
VAR
model pair which only differ from each other because of the lag
interval being
the so
tested. N is the
called multiplication
number of observations, c is
correction factor
third step
and as
it was
usual
observed
(9 )
( 10)
[it
[i1.
[A(8)
0(6)
8(4)
E(O) C(4)]
F(O)
G(O) H(4) 1(8)
[A(0(3)16)
G(O)
[p] ["']
E
R
B(3) C(O) ]
E(8) F(3)
H(3) I (8)
[i]
then
e2
e",
[::1.
periods~
identified
the
models
accordance
implied in
these
defined in
Annex I.
in
Granger
causal
with
the
relations
criterion
conclusions of
FPE
For E
G(6.8.0) =
G(0.8.0) =
G(7,8.6)
G(0.B.6) =
G(6.B.7)
X 10 3
var
0.309
0.415
0.293
0.348
0.505
For R var
G(0.6.8) = 336.5
G(6.0.8) = 498.0
G(0.0.8)
546.1
G(7.6.8) = 600.1
G(6.7.B) = 600.2
134
as follows
a)
In
model (7),
there is
direct
feedback
relation
P indirectly
be fulfilled
than E",
by the condition:
this necessarily
less
cause of R.
b) The
interesting result
direct
relations,
feedback
compatible with
the next
the variables
the logic
which
have
is
paired
perfectly
Section, although
it
is
not
necessary
condition.
c) In
(8), E
directly causes
spurious causal
P while
relation in
governs R
in explaining
neither P
nor R
cause E.
R has
relation to
a type
P.
II
thus
P fluctuations. Furthermore,
Finally, E
governs P
and
directly causes R.
Comparison of
the years
(in the
72 to
long period
referred to
above, while
an exogenous
for the
which give
short period,
rise to
the transformations
(10) also
on (8)
causal relations.
For the
modifications are
first
equation
in
which
135
R now
directly cause
E. Finally,
despite the
equation are
unchanged
E governs
P and
directly causes R.
We believe
of analysis.
the relations
Identification of
implicit in
VAR models
postulating the
intrinsic dynamics
usefulness of
is very
dominant
for
terms, there
implication relations,
and
system under
information
classical econometric
important,
mechanisms
of the
such
in both
particularly
directions
for
in
the
constructing
are still
models
in
reservations
it is
estimated with
the Zellner
estimations for
The
SUR procedure
main
the (7)-(8)
Table II.
quality of
and that
in the
construction of a
previous Section.
shown in
fourth step
Thus it
the (9)-(10)
both are
statistical
and (9)-(10)
can be
referred to
indicator
in the
in
these
superior to
the corresponding
VAR
(Sims)
models in Table I.
An additional
(10) pair
should be
performance, using
terms of
comparative predictive
models to
the others
and also
to the "naive"
it can
therefore be
(9)-(10). We
will now
go on
136
ANALYSIS OF
obtained with
response functions
the reactions
of the
innovation in
used was
naturally standard
deviation for
each
innovation.
For model
response functions
(9) Figures
of P,
Figures
P responses
the expected
for ip and i e
b) The
responses
differentiated:
have
the
when the
expected
ie., for i r
signs
and
).
are
is attenuated
subsequently
(the
137
the
CIT
ceiling
established
in
For model
(10), corresponding to the 1972.111-1986.IV period,
the basic evidence is as follows:
d) P responses are not long lasting, they are more intense
when the impulse is on the variable itself and they
are irrelevant for i r .
e) E responses are temporary too and the response of ir is
also irrelevant. The crowding out hypothesis is still
suggested by the evidence and is particularly intense
during the first year, as in the previous case.
f) Although there are increased fluctuations the response
pattern for R is similar to that identified for model
(9) .
Impulses on R have
variability of P and
138
and in
all cases
The
combined
effect
of
reasonably significant
impulses
on
in explaining
and
is
the variability
of R.
In summary, this analysis confirms the previous one relative to
the impulse
answers to
response functions,
the two
and
both
suggest
that
the
follows
1) Crowding
out effects
the adoption
of more
In
this period,
DLX fluctuations,
internal credit
does not
seem
to
observation
latter
could
be
reconcilied
the
with
of a
an additional
possibility that
there were
consideration throughout
former sometimes
therefore not
and toward
evident that
the latter
other
the trade-off
times.
It
is
have a
extent they
situations or
high positive
are complementary
time intervals
and contraction
relation since
of CIT are
to a
instruments. However.
in which
certain
a priori
139
nature of
monetary and
clarified by
foreign exchange
empirical analysis.
They stimulate
and
suggest
importance of
extensions are
which could
Given the
expected
from the
reliability of
confirm them.
Naturally such
be drawn
that
external impulses.
the
most results
foresaid
extension
presented,
and
it is to be
refinement
will
140
ANNEX I
Types of Granger causal relations resulting from application of
the Hsiao criterion to a trivariate YAR model.
1. Notation
A
Let Z(t)
and R.
A '" "
(P,E,R)
be all
which the
explanatory variables
endogenous variable
are,
in
is
rising sequence,
and
their
the
lagged
stage of
application of
the Hsiao
same way
after separate
addition of
criterion obtained
in the
'"
of these
"
It is accepted that
Let G(PIZ)
then be
application of
addition of
the minimum
the Hsiao
the R(t-l),
'"
criterion obtained
1 from
to
"
corresponding G(PIP,E)
model.
as before
regressors
after
to
the
di~ectly
...
"
,..
"
"
.A
be seen
'"'
....
...
""
"""
"""
G(PIP,E)
~
""
is type
141
,.A
""'''''''
,.
142
ANNEX II
TABLE I
Statistical indicators of OLS estimation of VAR (Sims) models
H
=8
Equation
0R- 2
Value l.s.
SSR
P
E
21.7 .24
14.8 .68
9.5 .95
.249
.006
6.41
76.1-86. IV
.95
.39
.12
c. V.
P
E
R
= 16
72. I I 1-86. I V
0*
Value l.s.
R~2
SSR
CV
20.3
23.9
13.8
.84
.62
.63
3.95
.004
4.52
P,E
E,P
E
.48
.30
.88
TABLE II
Statistical indicators of SUR estimation
of restricted VAR models
Hodel
P
E
R
Hodel (8)
(7)
0"
R- 2
Value l.s.
SSR
18.6 .41
15.4 .63
18.5 .42
.336
.033
7.96
.95
.05
.29
Hodel (9)
0*
Value l.s.
20.0 .52
9.2 .99
14.0 .87
R-:Z
SSR
.68
.37
.59
5.94
.026
5.54
Hodel (10)
R- 2
0*
Value 1.s.
SSR
13.1 .79
16.9 .53
.018
8.81
0R- 2
Value 1. s.
SSR
18.7 .60
19.7 .54
.017
6.01
------------------------------------------------18.5 .43
.95 .295
22.8 .35
.75 4.43
P
E
R
.42
.44
.44
.57
Note
l.s. means level of significance and
variables;
i.e., variables whose values
significance.
CV means causal
have F < 0.1
143
ANNEX II
TABLE III
Predictive performance of VAR models selected
VAR (Sims) Hodels
76.1-86. IV
M = 8
No future
quarters
p
RMSE U
RMSE U
M=12
RMSE U
RMSE U
72.1I1-86.IV
E
RMSE
RMSE U
--------------------------------------------------------------1 .158 .241 .015 .761 .406 1. 47 .598 1. 01 .008 .647 .262 2.53
RMSE
(7)
E
RMSE
76.1-86. IV
U
R
RMSE
RMSE
E
RMSE
R
RMSE
--------------------------------------------------------------1 .167 .255 .024 1.20 .216 .788 .517 .874 .014 1.07 .115 1. 12
RMSE
E
RMSE
R
RMSE
1. 01 .218 1.55
.766 .246 1. 53
.572 .208 1. 75
RMSE
E
RMSE
R
RMSE
--------------------------------------------------------------1 .134 .205 .020 .990 .211 .780 .514 .869 .008 .589 .067 .652
.130 .929
.095 .594
.128 1.07
144
TABLE IV
~
DECOHPOSITION OF VARIANCE
of prediction error variance in future k quarters explained
by each innovation
Hodel (9) 76.I-86.IV
Forecast
variable
K=l
Future quarters
K=4
K=8
K=12
K=16
K=20
100.0
0.0
0.0
58.0
32.2
9.8
59.7
29.7
10.6
61.8
27.0
11.2
64.0
24.8
11.2
65.6
23.2
11.2
ip
i ..
ir-
5.1
94.9
0.0
8.7
86.8
4.5
9.3
82.2
8.5
9.3
81.3
9.4
9.3
81.1
9.6
9.3
80.9
9.8
ip
i ..
i r-
7.9
1.5
90.6
6.4
22.3
71.3
6.5
23.4
70.1
6.7
23.8
69.5
6.7
24.0
69.3
6.7
24.0
69.3
Impulses
--------------------------------------------------------------
TABLE V
Hodel (10) 72.III-86.IV
Forecast
variable
Impulses
p
Future quarters
K=l
K=4
K=8
K=12
K=16
K=20
K24
100.0
0.0
0.0
72.9
62.4 58.4
37.'1 40.2
0.5 1.4
61.2
37.2
1.6
57.3
41.2
1.5
51.4
47.2
1.4
27.0
0.1
ip
8.6
11.4
11.7 12.6
12.7 12.9
15.3
87.3
85.8 84.3
84.3 84.1
91.4
81.9
i ..
0.0
1.3
2.5 3.1
3.0
3.0
2.8
ir-----------------------------------------------------------R
ip
6.8
6.5
7.8 8.3
8.6
8.6
8.6
0.0
25.9
37.3 38.6
41.7 41.2
41.7
i ..
93.2
67.6
54.9 53.1
49.7 50.2
49.7
i r-
145
Si,i,i~
l p
rS
Model (9)
Model (10)
1976.1 - 1986.1V
1972.ill - 1986.IV
--- .......... , - ,
......
_---------
Fig. 1 De P
Fig. 4 De P
Fig. 2 De E
Fig. 5 De E
Fig. 3 De R
Fig. 6 De R
P, E and R responses to ip
1-++
P, E and R responses to ie
P, E and R responses to ir
....
..................
..... _ - /
"
,/
146
BIBLIOGRAPHY
AKAIKE,H.
(1969)
"Statistical predictor
identification",
Anals of the Institute of Statistical Mathematics, vol.
21, 1969.
AOKI, H.
(1986), Notes on Economic Time
System theoretical
perspectives,
Berlin, 1986.
Series Analysis
Springer
Verlag,
BARBOSA, A.
PINTO (1985)
- "Infla9ao e produ9ao en Portugal
vol. IX, n l, Jan/85.
1953-1980, Economia.
O
BRANSON, W.
(1984a) "Exchange rate policy after a decade of
floating" in Exchange Rate Theory and Policy ed. BILSON,
J.E. MARSTON, R., Univ. of Chicago Press, USA, 1984.
BRANSON, W.
(1984b) "A model of exchange rate determination
with policy reaction: evidence from monthly data", Open
Economy Macroeconomics Conference, Lisbon, June/84.
A.
(1984)
"Testing for the
BURBRIDGE, J.
and HARRISON,
International
effects of oil price rises using VAR",
Economic Review, vol. 25 N2, June/84.
(1986) "Monetary policy, fiscal policy
CHOWDHURY, A.
et al
Southern Economic Journal,
and investment spending",
Feb/86.
DOAN, R. and LITTERHAN, R. - "User's manual rats version 4.30",
VAR Econometrics, Minneapolis, USA, 1981.
FACKLER, J.
(1985) "An empirical analysis of the markets for
goods, money and credit",
Journal of Money Credit and
Banking, vol. 17, n l, Feb/85.
O
GENBERG, H.
et al (1987) - "The relative importance of foreign
and domestic disturbances for aggegate fluctuations in
the open economy", Journal of Monetary Economics, vol.
19, 1987.
- "Deficit budg'taire et
GIRARDIN, E.
and HAROIS, W. (1984)
Universit' Paris I,
d'ficit externe" , Document n091,
Sept/84.
GRANGER,
C.
(1969)
"Investigating causal
relations by
econometric models
and
cross
spectral
methods",
Econometrica, vol. 37, n03, July/69.
HSIAO, C.
(1981) "Autoregressive modelling and money income
causality detection",
Journal of Monetary Economics,
vol. 7, n l, Jan/81.
O
HSIAO,
C.
(1982)
"Autoregressive modelling and causal
ordering of economic variables",
Journal of Economics
and Dynamic Control, vol. 4, Aug/82.
147
NELSON, C.
OCDE (1982)
RIBEIRO, C.
SAMUELSON, P.
SANTOS. F.
TEIXEIRA (1986)
"Money prices and output
in
Portugal:a vector autoregression analysis". Association
Sud Europea
de Economia
Teorica.
Conf.
Papers,
Marseille. May/86.
In
assessing
the
appropriate
stance
of
monetary
models.
degree of
confidence in
In
received much
attention in
the
recent
analytical
which the
enhanced through
efficacy
of
monetary
international cooperation.
policy
More
could
be
details
on
of economists
discretionary monetary
stability. This
concept of
policy may
conclusion is
classical assumptions,
it takes
A policy
fully into
to be
future.
a monetary
this sense,
zero inflation
argued
be incompatible
derived from
discretionary decisions
In
recently
that
with price
some strong
neo-
time consistenCY.
consistent" when
have
made by
is said
to
be
"time
152
public
believe
does
that
wi 11
prices
stable.
remain
to
expansion
inflationary monetary
problem.
If
renege on
output
boost
policy may
the public
(1).
Thus.
non-
suspects
that
the
authorities
will
inflation will
cause nominal
inflation will
turn out
to be lower
reasons. ex
surprisingly high.
get output
Consequently.
closer to
inflation to
words. a
post real
commit all
only by
expectations of
time-consistent policy
unless the
their future
rates
will
appear
th~
full employment
match the
some inflation.
interest
allowing
the public.
(2) will
in general
monetary authorities
policy decisions
some
In other
inv.olve
can credibly
to the objective of
price stability.
This line
of reasoning
has led
some economists
to
non-inflationary
policies.
interest for
their audacity
difficult to
envisage how
important
are
the
Such
suggestions
than their
they
implications
discretionary monetary
policy as
These implications
are
consistency throws
them into
not
novel
it
implemented.
time
of
is
but
of
realism. since
be
might
are
consistency
actually
the
sharp relief.
notion
First.
more
it is
More
for
practised.
of
time
arguments
153
A central
creates money
rapidly,
increasing the
inflation
output-inflation
stably linked
and
Second,
tradeoff.
monetary targets,
The central
if
in circumstances
to ultimate
worsening
the
short-run
arguments
for
announced
requires
authorities.
But
policies is
that such
once
the
taken into
authorities to
has rested
incentive
for
in
the
monetary
time-inconsistent
confidence can
considerable time,
confidence
exist. In
or some
practice,
very sharp
establish credibility.
on demonstrated
it
shocks,
will
for
take
monetary
Monetary targets, where they have been met, have helped in this
respect by
the authorities.
larger
in
the
discretion is
While the
absence
likely to
for expectations
in
the
1980s
room
of
for
targets,
be more
discretion
the
may
exercise
appear
of
such
involved
reduced
output
and
increased
studies
potential gains
a number
that
have
attempted
to
Small gains
can be
rational insular
derived from
policies. In
coordinated
vis-a-vis
154
1980s, the
gains stem
monetary policies
in increased
imbalances.
in large
part from
in cooperative
output and.
Inflation
more
relaxed
in some models.
is higher.
in reduced trade
but not
by so
much to
in dynamic
models may
regime so
is less,
is the
that the
present value
although the
same as
of
the
in a non-cooperative regime to
potentially
important
developing countries.
to the
side
benefit
accrues
to
the
demand and
two interacting
First. policy
price stability
relative to
policy --
i.e. an
disinflationary (inflationary)
depreciat~on
(appreciation)
preference for
price
easing
(tightening)
effect abroad
of
stability
the
of
because
domestic
itself
has
of
the
currency.
The
would
of
course
in conjunction
with the
second
factor
it
has
important
qualifications
from coordination
attach
to
the
empirical studies :
155
i)
ii)
in this
last
respect
is
that
the
estimated gains in dynamic models often rely implicity on timeinconsistent policies. Such policies are not necessarily more
credible just because an international agreement is made, and
they may be less credible. Because fear of exchange rate
depreciation is reduced, international coordination could yield
solutions that are too inflationary and on balance worse for
welfare over time than non-cooperative solutions. Some authors
conclude that unless binding constraints or the authorities'
strong reputations can make non-inflationary policies credible,
cooperation is futile.
III. A LITERATURE REVIEW
This Section
provides a
brief explanation
of
analytical concepts used above, and highlights some of the more
important findings in the recent literature on the issues of
time inconsistency, credibility and international coordination
(4). Although the focus is on monetary policy, the issue of
a "Pareto-efficient" situation,
can be better off without some
156
cooperation
involves
necessarily
monetary-fiscal
mix.
empirical setting
The
analysis
of flexible
discussion
some
is
limited
of
to
the
today's
independently,
reactions of
non-cooperative
adapting
decisions
the others.
cooperative, or
other parties'
game,
The most
to
countries
actual
or
act
expected
common assumption
of non-
taken as
"Stackelberg assumption"
leader, setting
as best
as
multicountry models
that
countries
assumption, with
be a
country acts
they can
other OECD
that one
U.S.
but
policy
not
vice
significantly
versa,
the
affects
Stackelberg
more relevant
games yield
outcomes in
better off,
with no
other being
worse off,
by a cooperative
thus be
calculated as
non-cooperative solution
(5). For
any gain
(i)
must hold
on ultimate
domestic
countries
the difference
between a
optimal cooperative
solution
negligible spillover
countries' policy
and an
instruments must
(ii) foreign
instruments
must
not
appropriately
have
enough
and
(iii)
instruments
individual
to
achieve
(1968)
anticipated
much
of
the
recent
157
research in
reserve
a theoretical
currency
cooperation. tax
too
tight.
system.
He
policy might
since
maintaining
study of
the
external
employment [cf.
fixed
exchange
concluded
that.
rate.
without
latter
balance
is
mainly
and
the
directed
former
towards
towards
full
1976). focusing
fixed
deflation
exchange
or
and
the
individual objectives
the
exact
bias
of individual
growth
of
in his
exceeds
be overly
an example
be
the sum
(exogenously determined
policy will
might
inflation.
relation between
objectives
rates
biased
towards
depending
balance
on
of
If
supply
the
of
the
payments
international
model).
the
too
reserves
sum
of
reserves
the
then
of conflicting
country objectives.
Canzoneri
and
the structure
situations).
situations) and
policies are
asymmetric.
the
be a
In the
are deflation-biased
symmetric-negative
countries attempt
case.
to offset
round of
competitive exchange
symmetric-positive case.
since no
country gives
policies
enough weight to
the beneficial
Biases from
("locomotive"
example would
rate depreciations.
of the
In
inflation-biased as
mutually negative
A concrete
symmetric-positive
asymmetries. but
mixes across
they can
lead to
conflicting policy
United States
can be
by Asikoglu
instrument to
instrument that
another to
does not
asymmetric situation
affect the
real output
price level
(fiscal
policy)
and
Europe
158
it
tells us
where to
does not say how large the gains might be. The
landmark
coordination is
solution with
and
study
Oudiz and
of
potential
gains
from
an optimal
cooperative solution.
the characteristics
of governments'
objective functions
from
the multipliers of the models and from the assumption that each
country (the United States. Germany and Japan) does the best it
can without
cooperating. Then.
from synthetic
to 1986.
in the
United States
current account
three countries
and Germany.
surpluses in
and the
and
in
the 1984-86
period as
well as
models.
The gains
Sachs stem
from coordination
mainly from
rates. With
suggest more
a coordinated
by
reduction
Oudiz
in
and
interest
expansion everywhere.
more fiscal
derived
contraction in
all three
fiscal expansion
in the
countries. those of
United States
MeM
and more
fiscal restraint in Germany and Japan. This odd result from the
MeM --
that countries
already doing
on fiscal
is a
consequence
of
the
159
revealed
preferences
approach
and
of
the
fact
that
no
If
fiscal and/or
increased U.S.
(1985).
trade deficits.
Sachs
and
McKibbin
expansionary monetary
(1985)1.
policies does.
but
that
for
more
output and
increase
in
employment more
inflation.
given
than compensate
policymakers'
for
apparent
preferences.
A common
assessed from
finding is
the
that
objective
countries (invariably
a group
1 per
Taylor (1985)
neo-classical
to
welfare
the
gains.
cooperating
so of
are empirically
that tend
net
functions.
usually no
more than
the
negligible. Since
assumptions
their analysis
(including
rational
is based on
expectations)
eliminate it
entirely in
effects of
answer seems
in a
to be
wide range
Canzeroni and
has some
For example.
Minford (1986)
cases coordination
of
spillovers.
The
very large
(1986) allows
especially
from the
results derived
monetary spillovers.
yields only
for uncertainty
by
agree that
in many
parties. neither
real world.
of which
is an
exact representation
of the
coordination.
Two factors.
not be
overlooked.
timing of
more favourable
coordination can
Eir~.
policies [e.g.
to cooperation. should
Sachs (1983)1.
result in
better
of monetary
to appreciate
policy. which
and
dampens
causes
domestic
the
real
inflation
160
rapidly.
Competitive
selfish
international deflation.
participant attempts
advantage, each
loss is
policies
then
With cooperative
to exploit
imply
sharp
policies, since
the exchange
no
with competitive
so, with
policies, but
normal rates
outcome yields
of
developed countries
higher demand
might have
time
higher welfare.
industrialized countries
is spread
preference,
~e~Qnd,
yields
and improved
participating countries
lower
LDC terms
gain from
the
cooperative
side-benefits
(LDCs), favouring
more to
to
the
interest
of trade.
lessrates,
Indeed
LDCs
themselves [e.g.
Sachs
and
McKibbin
(1985)].
3. Other approaches
i) Exchange rate and world money growth rules
McKinnon (1984) proposes a monetary agreement between
the United
States, Germany
and Japan
supply. The
global policy
Although currency
stance cause
by currency
substitution is
not
important
is
show that
in the
his
empirically
McKinnon's proposal
context of
because
substitution.
nevertheless
has
scheme,
like
the
full
cooperative
proposals have
exchange rate
encouraging
code
substitute for
suggested the
from underlying
of
behaviour
that
the
could
to limit
the
intention
serve
as
of
a
[e.g. Williamson
exchange rates
been made
flexibility, with
target zones
(1985)]. Critics
do little
good if
macro policies
for major
countries
[e.g. Dornbusch
attention away
(1983)].
The
161
majority view
of the
system (1985)
of improvement
are only
of the
present situation.
in special
cases. Although
manipulation
of
distortions in
exchange
they might
rates,
policy mixes,
they
also
migh~
lead
to
(1986)].
The system
attempts to
encourage more or
realignments at
more frequent
intervals
than,
operation of
the system
implications
for
symmetry of
EMS
have explicitly
countries
adjustment between
concentrated on
themselves
members
--e.g.
on
the
the
broader internatIonal
these studies
the rules
on how
of the
system might
be best
number of
authors have
amount of
Buiter (1985)
the world
explicit theory
and empirical
evidence.
in response to a
United States
remains unchanged.
would then
Within the
be offset
(improved U.S.
such that
by a
the world
interest rate
real depreciation
competitiveness). Another
of
the
response would
dollar
be a
permanent increase
also offset
cost of
permanently higher
inflation.
Some
questions
about
162
confidence obviously
similar package
arise from
of measures.
United States.
these
proposals.
However
fiscal expansion
before. the
reduced current
account
imbalances
seems
worthwhile
with the
neutrality.
necessarily produces
classical property
inflationary
an
of
monetary
additional
inflation
run is
policy
inflationary policy
firmly believes
long-run
one that
that the
allows no
inflation.
monetary authority
is
not
polic~
in
If the public
is committed
to
favourable outcome
is likely.
i.e. no inflation
"natural" rate).
precommitment, the
But
given
policymakers can
public
achieve
belief
an
in
even
their
better
society and
output,
central bank
even
beyond
can then
increasing the
if the
the central
the
bank have
a preference
full-employment
improve welfare
in
the
rate.
short-run
for
The
by
and the
increase in output.
A "time-consistent"
policy can
be
defined
as
one
the future
[Kydland and
Prescott (1977)].
In
163
this sense,
monetary
it can
policy
be said
is
can in
is credibly
general achieve
commitment of
are clear
to inflate,
outcome by
For this
reneging on
the
guarantee such
this commitment,
commitment
the
will revise
inflation. When
involves
it
a better
price stability.
constraints that
divert from
since
inflationary policy
constant non-inflationary
"time-inconsistent"
committing authorities
be optimal
that a
no constraints
the situation is likely to slip over time into one in which the
inflation rate
expects
just
the
of
inflation
that
is
a time-consistent
policy in
general
allows
some inflation.
These
treatment in
ideas
recent
have
been
literature
given
on
the
rigorous
theory
of
formal
economic
164
ACTUAL POLICY
No inflation
{Precomm'i tment}
***
Some inflation
{Precommitment
ignored}
**
****
Time inconsistent
policy solution
There are
Time consistent
solution
outcomes illustrated
(i.e. one
Time inconsistent
expectations
solution
Precommitment
solution
here, yielding
outcome for
system). Policymakers
four
possible
solutions
may not
target of no inflation
the public mayor may not believe
them.
The asterisks (stars)
indicate the ranking of the
solutions in terms of social welfare ; the ranking rises as the
stars increase. The only two possibilities that are sustainable
in the long run are the precommitment solution and the timeconsistent solution,
because only
in those
solutions are
eL
policy, if
it is
the long run, gets only three stars. At the other end of the
spectrum, the
{one star}
worst solution
is when
the
policymakers are committed but lack credibility -- an output
loss is then caused by the central bank's refusal to accomodate
the higher
wages and
consistent policy,
by the
stars.
prices built
which just
public, avoids
into
contracts.
time-
this output
loss and
so receives
two
be acheived
only when
the public
believes that
the means
165
worst, and
might achieve
preferences under
each solution
that will
output of
conclusions with
establish
policy without
manoeuvre.
the true
the
credibility
some external
for
of
an
constraint
anti-inflationary
on
its
freedom
of
observing its
countries the
resources to
its intentions
actions over
private sector
"central bank
recognize that
an announced
In
their
model
announced targets
of time
inferred
and
in
by
some
policy
is
monetary
increases. Backus
and
Meltzer
in
reputation
the precision
be
watching". Cukierman
credibility as
increases as
can only
a period
(1986) define
occurred.
wide ramifications,
has
actually
parameter
control
and Driffill
in
which
hitting
(1985) pose
(inflation
Moreover the
authors define
lost, cannot
be regained.
credibility as
As long
as the
the announced
policy, once
the policymaker
reveals
this kind
have radically
the arguments
for monetary
rules.
The
recently been
argued, must
visibly and
shifted
required rule,
it has
permanently bind
the
on the
central bank,
the stability
(1983)
than a
constant-money-growth rule.
of demand
describe
as
permanently constrain
The debate
for money.
discretionary
Thus,
a
Barro
system
that
and
Gordon
does
price stability.
not
In
166
their model,
which has
discretionary system
strong classical
properties.
such
the
central
credibility
is led to focus on
and
bank
establishes
price
stability
attractive short-run
opportunities
generally recognised
that this
this case
to
is a
constraints
external
high
reputation
foregoing
inflate,
for
apparently
it
very valuable
will
be
asset.
In
endogenous incentive
unnecessary.
a
by
pre-commitment solutions
and
which is
repu~ation.
and
Barro
policymakers' concern
the
on
central
(1983)
Gordon
for reputation
bank
show
become
that
the
has a
low enough
rate of
time discount)
will be
reputation.
It
weights highly
the prospect
that the
and hence
tradeoff.
2.
Dynamic models
issue implicitly
raise no
that do
assume that
credibility problem.
not explicitly
the
address
time-inconsistent
the
policies
If precommitment by policymakers
Ploeg (1986)].
cooperation is
are admissible.
more
inflation
considers that
existence of
.if only
since cooperative
than
non-cooperative
This is because,
inflate the
time-consistent solutions
solutions of that kind have
cooperation between
worsen welfare.
incentive to
But the
not assured
ones.
Rogoff
(1985)
model this
raises the
rate of
inflation without
167
might improve
of inflation
social
if it
w&lfare,
increases
the
despite
the
stability
of
agreement between
that time
of reneging
very big
this kind.
REFERENCES
Asikoglu, Y.
Coordination Between
Theoretical Approach",
the U.S.
and Europe
Discussion paper
Game-
n0631, Queen's
University, March.
Backus, D.
(1986), "Reputation
Incomplete
Information",
Journal
Economics, vol. 17, n 0 1, pp. 3-20.
Barro, R.J.and
D.D. Gordon
Monetary Policy
in a
(1983),
"A
Political
Natural Rate
of
Monetary
Theory
of
Model", Journal
of
K. Klinton,
International
A.
Asset
Cote
and
D.
Longworth
Substitutability
(1985),
Theory
and
168
Buiter,
Y.H.
(1985),
"Macroeconomic
Interdependent World
Policy
Economy:
An
Design
Analysis
in
of
an
Three
in
an
Interdependent
Finance Discussion
World",
International
and P.
Minford
Policy Coordination
Discussion Paper
(1986),
Matters:
Series,
"When
International
An Empirical
n0119,
Centre
Analysis",
for
Economic
and J.B.
Mobility and
J.S. Bhandari
Taylor (1985),
the Coordination
(ed.),
Exchange
"International
of Monetary
Rate
Capital
rules",
Management
in
under
Credibility
and
Inflation
Under
and
pp.
1099-1128.
Dornbush,
R.
(1983),
Interdependence",
"Flexible
Exchange
rates
and
30.
Frankel,
J.A.
(1986),
"The
International Macro
Sources
Models and
of
Disagreement
among
F.
and
A.
Giovanni
(1986),
"European
Currency
Ten (1985),
"The Functioning
Monetary System",
IMF
of
the
International
169
(1974),
Hamada, K.
"Alternative
Interdependence
Monetary
System,
of
Monetary
Policies
and
Policies,
the
in
National
International
Financial
Chicago, pp.13-33.
Hamada,
K.
(1976),
"A
Strategic
Analysis
of
Monetary
W. McKibbin
Policy Mix,
Aspects
and J.
Sachs
(1985),
"The
Economic
of
Macroeconomic
United States,
Interdependence
Among
the
L.
(1982),
"A
Note
International Equilibrium
Journal
of
Kydland, F.W.
the
Possibility
of
an
International
February, pp.
on
Economics,
vol.
98,
nOl,
Rather
than
1-23.
and E.C.
Discretion
Prescott (1977),
The
Inconsistency
"Rules
of
Optimal
Plans",
(1985),
a Risk",
n014,
in International
Economics,
December.
Melitz, J.
(1985),
"The
and J.
and Fiscal
Sachs (1986),
Policies in
"Coordination of Monetary
the OECD" ,
n01800, January.
McKinnon, R.I.
(1984),
Stabilization",
Analyses
in
Internattonal
170
Institute for
Economics, n08,
International Economics,
Washington, March.
(1963), "Capital Mobility and Stabilization Policy
Mundell. R.
Under
Fixed
Journal of
and
Flexible
Economics and
Exchange
Political
Rates",
Canadian
Science
29,
n04,
in
Open
Niehans. J.
"Monetary
Economies Under
and
Fiscal
Fixed Exchange
Approach", Journal
of Political
Policies
Rates:
An
Optimizing
n04, Part
Economy 76,
( 1985) ,
G.
"Strategies
Economiques
Confrontation 7,
Coordination ou
Europ~ennes
INSEE Working
Paper.
n08506, June.
Oudiz.
G.
and
J.
Sachs
Coordination among
G.
J.
and
"Macroeconomic
(1984).
Policy
Sachs
1, pp.
( 1985) ,
1-75.
"International
Policy
International
Buiter and
R.C.
Economic
Marston
Policy
eds.),
Coordination
Cambridge
(W.H.
University
Press.
K.
Rogoff.
( 1985) .
Cooperation
be
"Can
Monetary
International
Counterproductive
International Economics.
7"
vol.18. n03/4,
Policy
Journal
May. pp.
of
199-
217.
Sachs, J.
(1983),
Dynamic
"International
Macroeconomic
Policy
Model",
Coordination
NBER
Working
in
Paper.
n01166, July.
Sachs, J.
(1985),
"The
Dollar and
the Policy
Mix
1985" ,
117-97.
171
and V.
the
OECD
HcK i bb i n (1985),
and
LDC
"Macroeconomic Policies in
External
Adjustment",
Centre
for
Ploeg. F.
Long-run
(1986),
"Capital Accumulation,
Conflict
Discussion Paper
in
International
Inflation and
Objecti ves" ,
Research, London.
Villiamson. J.
International
Economics
International Economics
1983) .
Policy
Institute for
Analyses
in
1.
Introduction
The
analysis
economic policy
of
by using
effectiveness
game theory
contemporary macroeconomics.
and the
and
is a
optimality
of
crucial issue
of
analysed.
Most papers
the following
dilemma, proposed
by Kydland-Prescott
(1977)
re-phrased in
way:
the announced
Barro-Gordon
trigger strategy
equilibrium concept
to his announcement,
same result
is
shown
for
incomplete information
by Backus-Driffill
equilibrium concept
This paper
aims at
providing more
strategy,
finite
games
with
on
the
Stackelberg
174
be dominated
by any
theorem
if
policymaker can
which Pareto-dominate
we
allow
achieve any
the time-consistent
for
trigger
of the
outcomes
Nash outcome
(more
theorem
if
a hierarchical,
we
introduce
trigger
sequential, framework,
where
of the
Nash outcome
(iv) either
a coordination
problem characterizes
a general
problem or a credibility
outcome indeterminate.
In order to simplify the analysis, we will consider a
specific
economic
example,
the
Monetary
Policy
Game
(see
Policy Game
and conflict
between the
sector. The
crucial point
monetary policymaker
analyses the
interactions
has a
zero inflation
implies that
only
by
the policymaker
inducing
private sector.
inflation
shocks,
i.e.
However, being
private
sector's
expectations
rational,
behaviour is
is sub-optimal,
the
policymaker's
by
time-in'consistent
the
monetary
inflation target
monetarist rule
to output
policymaker
by committing
that money
natural rate
can
achieve
himself to
at
the
least
his
old-fashioned
exists a
time-consistent monetary
175
the policymaker
can achieve
both
his
inflation
and
output
show that
The novel
organization literature,
is that
hierarchical, sequential,
a Stackelberg
framework
(a
of the game between the policy authority and the private sector
is
indeed
hierarchical
application of
and
the results
organization literature,
makes
it
proposed in
being
those
unsatisfactory
the
conceived
for
next section,
we
consider
the
same
model
Barro (1986),
in
section 3,
perfect equilibria
the monetary
we show
subgame
policymaker. This
infinite policy
that infinite
extends
games. Furthermore,
previous
results
on
a coordination problem of
is shown to arise.
also threaten
the private
represented
player),
as
monolithic
sector
and
we
(initially
analyse
the
assumption.
equilibria
It
will be
In
are
particular,
considered
shown that
Stackelberg
(see
Carraro,
Trigger
1985,
176
outcomes can
bliss
be achieved
point
whenever
coordination
his
problem
credibility problem.
exists a
In particular,
is greater
is
believed.
The
into
transformed
therefore
is
monetary policy
and output
announcement
than its
the game.
Section 5
Policy Game
discusses the
when the
monolithic player,
solution
private sector
but is
of
is not
the
Monetary
represented as a
of atomistic players.
Finally, section
previous results
of
respecifications of
functions,
6 analyses
the monetary
model, of
the players'
loss
set.
2. A Model of Monetary Policy
A standard,
discuss the
role of
models (see,
Gordon,
ingredients of
the following
real
negatively
targets
may
i.e.
the
economic
inflation. Fourth,
over optimal
increases in
Third.
of
the
not
Barro-
a discussion)
1986).
the inflation
there
is
policymaker
activity
by
first.
general price
an
rate
on
policymaker.
coincide
with
cash
aims
at
inducing
balances
Finally.
the
the
public's
level.
expand
"expectational
achieving
unexpected
inflation tax
by
1977)
1985a,b ; Cukierman,
affecting the
activity.
Phillips Curve".
expansions
to
in rational expectations
1986. for
works by
economic
used
(see Barro.
Secondly. unexpected
been
Kydland-Prescott,
; Backus-Driffill.
monetary policy
has often
monetary policy
for example,
1983a,b
The basic
simple, model
is
inflation
considered
policymaker's
targets.
For
the objectives
of
the
policymaker
and
the
public.
177
178
The first
inflation shocks,
inflation. The
rate),
term of
the equation
whereas the
parameters
second
(which
term
is
depends
the
on
cost
the
of
natural
eq.
(2.3)
can
be
re-written
in
the
output
and
following way:
Defining X t
potential output
e~/o
by
and
as the
rates of
difference
between
growth (i.e.
then re-normalizing
Xt
Yt - Y*t), X
we obtain:
(~/O)2,
(2.4)
The policymaker's
control variables
bo(~/o)2
are
the
money
above definitions,
we can
re-write
eqs.
( 2 . 2) as
(2.5.1)
and
(2.5.2)
mt
(2.5)
imply:
met
(2.6)
where met
stock rate
agents'
so
is the
expected deviation
of growth
from
its
natural
expectations
may
not
coincide
announcement mat.
Solving equations (2.5), we obtain
level
with
of
the
Y*t.
the
money
Private
monetary
179
(2.7)
1 +
(2.8)
pet. +
pt.
1 +
From eqs.
of output
from its
natural
rate
can
be
obtained
only
by
Therefore, the
x only by cheating,
which,
if
is the
the monetary
the announced
that,
the private
monetary policy
sector anticipates
and the
(i.e. Xt.
only if
the policy
outcome of
the
time-inconsistent
the game
is
sub-optimal
information is
time horizon
assumed, can
monetary policymaker
that Xt.
is still
to be
is
infinite
a reputation
committed to
or
incomplete
policy.
The above
equations
define
the
standard
monetary
game, but do not define yet all the elements of the constituent
game (furthermore,
eq.
(2.1),
being a
reaction function,
is
redundant).
the "rules
of the
will completely
game" must
define the
be specified.
constituent game
equilibrium outcomes.
3. Friedman's Trigger Strategies and the Coordination
Problem
This section
Gordon's (1983a,b)
depart from
analyses
the
their analysis
in that
implications
of
Barro-
expli~itly
write
the
180
that
and
Barro-Gordon's
derived by
Backus-Driffill's
assuming that
results
can
be
the
next
section.
Furthermore,
we
assume,
for
In contrast, the
We emphasize
that the
results below
can
be
The private
sector's
loss function
can be written
as
Vpt
(3. 1 )
(Yt - Y*t)2 +
so that
- pe t )2
~(Pt
X2t +
- pe t )2
~(Pt
private sector's
is given by (2.1).
Private sector's decision variables are output Xt and
price expectations pet.
Define the players' strategies as Spt
SMt
(mat,m t ), where
SMt. The
to Spt
define the
players
Spt belongs
loss vector
of
the
constituent game
Vt
game
(X t , pet) and
= SMtXSPt.
Furthermore
(two
at time
is
in
this
section).
denoted by
Then,
the
r(N,St,V t ). Notice
Monetary Game,
that mt
(2.8. Consequently,
becomes mt
players'
0,
The monetary
= mat
the
met
implies Xt
0 (see eq.
which implies
losses are:
Pt
0 (see
eq.
(3.1.
The
181
X*2
(3.2)
and,
loss function
by setting
mt
the game
is X t
m*t.
~(1+~)x-/(~2+b)
~2x-/(~2+b)
The
and Pt
~x-/(~2+b).
losses are:
(3.3)
bx-2/(~2+b)
~3(1+~)x-2/(~2+b)2
(iii).D.i.l;tc. r..e...tLo...n.
anticipates that
The
private
monetary authority's
time inconsistent
sector
correctly
announcement mat = 0 is
sector's expectations is
(3.4)
~2
+ b
(3.5)
and (3.4)
imply:
~x*/b
which substituted
the outcome
players'
(2.6)
of the
into (3.4)
gives mt
game is X t
0 and Pt
~x*/b.
~x-/b,
Then
so that the
losses are
(3.6)
The
relevant
outcomes
of
the
game
can
also
be
182
Table 1.
Private Sector
Monetary
Authority
mat=O, mt =(3.4)
announcement is
the outcomes
when the
announcement is
not
believed
by
the
private-sector.
Table
has
r
strategy m t
in Cukierman (1986).
that
YC Mt
always
an
yr Mt .
so
that
incentive
to
cheat.
<
is therefore
the
monetary
The
optimal
yrMt, so
(the
domi nant
strategy
is
sub-optimal).
Th i s i s
Kydland-Prescott's dilemma.
Consider now
the repetition
defined.
'vi i =
T
L
t=O
players'
(T times)
Each player
at i Y it.
(T
of the
now
= M, P.
discount
repeated game
the game
r(N.St.Yt.a.T) which
constituent game
minimizes
the
The vector
is just
previously
intertemporal
a = (aM, a p
loss
def i nes
the
If complete
information is
imply
the
r(N,St,Yt,a,T)
game is
that
is
equilibrium
(pdt,X d t ) at
the
monetary
game
if the
repeated an
infinite number
of times
and a
is
sufficiently
183
close to
one (Friedman,
assumed (Kreps-Wilson,
(at least
in the
Backus-Driffill,
first stages)
of
the
is based
on
three
outcome
~he
repeated
1985a,b ; Fudenberg-Maskin,
This result
First, the
game
(see
1986).
basic
assumptions.
if mT =0 at all
(3.7)
<
otherwise
This implies
sector to
an implicit
threat
from
the
private
the private
sector (who
anymore), whenever
does not
a money
believe the
stock rate
monetary
rule
zero is chosen.
Suppose the
(see Frtedman,
adopts the
game is
1971) close
thus forcing
the monetary
Then, tfie
private sector
authority to
1986).
By using
same conclusion
punishment lasts
1983a,b the
Green and
can be
obtained even
a finite
if the
private
s~ctor
punishment lasts
only one
period,
so
that
the
equilibrium
sequential
Driffill (1985a,b)
mt =0
to show
concept
used
by
Backus-
(at almost all stages of the game) even when the game is
finite and
incomplete
information
incomplete information
is necessary
is
assumed.
only
if
be used
sector's trigger
forced to
exist, Friedman's
to
show
strategy such
that
(1985) and
there
Notice
that
unique
Nash
if multiple
Abreu's
exists
(1986)
private
184
finite games.
The second
raised
only
by
crucial assumption
the
private
authority is
simply allowed
being Pareto
superior to
sustained
by
sections, we
Backus and
the
sector's
that Barro
Driffill's (1985a,b)
allowing the
threats
whereas
the
are
monetary
private
will see
sector,
is that
threats.
In
and Gordon's
the
next
(1983a,b) and
relaxing
obtained.
In
basic assumption
belong to
this
is that
all equilibrium
assumption
more
general
results
can
be
rational outcomes
are equilibrium
outcomes of
the game
(see
Fudenberg-Maskin, 1986).
For the moment, we want to emphasize that even in the
simpler
strategic
setting
Backus-Driffill (1985a,b)
used
and
by
many
Barro-Gordon
others,
(1983a,b),
the
problem
of
that
the
private
sector
forms
his
(3.8)
otherwise
= mat,
we have
does not
yrpt =
have any
Yd pt = 0, so that the
incentive to
cheat.
If the
game is
is an
easy application
of Theorem 3.3 in
Friedman (1986).
Hence the
time-consistent outcomes
of the
infinite
185
(3.10 )
The same
conclusion can
if
(i )
( incomplete
186
trigger strategies.
function of
unique
state variables
time-consistent
Monetary Policy
stages of
(subgame
Game is
the game
(the inflation
rate),
perfect)
then
the
of
the
solution
(see Stanford,
1986a ; Robson,
1986). This
E-equilibria
of
the
instead of
discounted loss
functions are
1986a)
strategy
are
with
game
1985)
also
considered
depends
(see
on
the
Carraro,
other
1986b).
player's
This
last
Trigger Strategy
Problem
Let us now consider the Monetary Policy repeated game
r(N,St,Vt,a,T) and
without imposing
let us
determine its
the restriction
equilibrium outcomes
hierarchical
structure
monetary policymaker
sector.
In
other
authority adopts
impose his
to
raise
words,
a more
the game,
we
the
game,
threats
assume
by
allowing
against
now
that
active monetary
own desired
we consider
into
the
the
the
private
monetary
policy, by trying to
desired solution
to the
Finally we
assume no
discounting.
The basic
attempt by
produce
feature
the policymaker
more
than
policymaker's effort
only by
Appendix,
Y*t.
is vain
game
if he
As
is
therefore
the private
previously
tries to
becomes effective
appropriately threatens
Trigger
the
to induce
Yt
cheating, but
Stackelberg
of
if
sector
shown,
achieve X t
the
the
to
the
policymaker
Strategy
equilibrium
described
in
the
str~tegy
187
and to show under what conditions his threats are effective and
credible. For
a monolithic
and many
The game
atomistic producers
will be
discussed in
the
next
section.
Suppose
the
monetary
following continuous
authority
announces
(discontinuous strategies
are considered
- x)
(4.1)
In other
words, the
the
if X t
S x
private sector
authority's desired
does not
comply
output growth x.
with
the
In contrast,
monetary
if X t
x,
mt
desired output
when X t
an important
is
x
In other
policymaker's desired
monetary
property of
growth. which
desired outcome.
the
authority's
=0
in this case.
the monetary
implies the
words. we
credibility of
can also
outcomes must
policymaker's
belong to
say
that
the
all
function.
Is strategy
private
sector
(4.1) effective
accepts
to
produce
x?
Xt
Will
the
coincide with
the constituent
the "rules
Then credibility
by using
the above
of t~e
questions we
game which
derived in
the
game" assumed
and effectiveness
the results
first derive
in this section.
(1985,
1986a)
and
strategy (4.1)
Let us
is effective
first assume
and credible.
that the
In this
trigger
case,
as
188
shown in
is also time-consistent,
sector finds
implies mt
players'
by (4.1)
(ii)
but not
credible and
x-.
In
the policymaker
whenever the
other words,
punishment but
pet.
=0
by (2.6.2). The
V p t = x- 2
The trigger
DiB~~tiQn.
the policymaker's
pet
V- Mt
announced threat
and Pt
losses are
(4.2)
Xt
it optimal
strategy is effective,
does not
reaction to
his
choice
will
not
be
This implies
mt
Pt
~x-/b.
the
~x-/b
thus choosing X t
= O.
The players'
losses are :
(4.3)
This equilibrium
of the
xd
standard Monetary
(i.e. pOt
( iii) .
credible,
but
punishment.
the
The
policymaker
trigger
actually
strategy
is
carries
out
not
the
private sector's
but
now the
expectation is
threat is
still met
(4.4.1)
and
(4.4.2)
p8 t =
ktx t + (1-
189
(4.5.1)
and
(4.5.2)
Therefore, the
relevant outcomes
I.a.b.Le__.2...
Private Sector
Xt=X,
m- t =(4.1),m t =(3.4)
p8 t =mol3l. t
Monetary
Authority
m- t =m~_ = ( 4. 1 )
As stated
only equilibrium
"rules of
in the
if
Appendix (see
of
this
repeated a
finite number
standard
section,
is
{pOt=~x*/b,xOt=O}
complete information
obtained the
the game"
(discretion) equilibrium
game,
YOM,Y Op
Y*M'Y*P
of
result
is
assumed
times.
We
previously
the
stable
Nash
the
have
game
therefore
described
is
re-
for
the
the private
policymaker finds
punishment), i t
is possible
it optimal
to
carry
out
the
0,
Xt
x)
190
is the
sequential
equilibrium
of
the
game
under
suitable
first check
authority's strategy.
(Pt =
0,
Xt
effective)
the
effectiveness
of
monetary
x*)
is
inducible
lie
monetary
policy
is
(4.6)
Furthermore, the
trigger strategy
(4.1)
is shown to
be credible if :
(4.7)
; that is i f :
(4.8)
where 6
reputation,
can
i.e.
>
----------------------
be
interpreted
as
the
policymaker's
actually carried
out by
the policymaker
whenever Xt
is
not
equal to x*.
Furthermore, the
the
game
is
repeated
credibility condition
an
infinite
number
is 6
of
>
times
0 if
(see
is if
trigger strategy
bx* >
O. Assuming
therefore x*
>
0,
any
191
(4.6) and
of the
3) .
In particular,
section,
using the
in this
.Er:.Q.,p..9...s...i..t..i.Q1LA
where p
results presented
: If k t <
and the
[~(1+~)11/2,
-[p-~l-l[(1+~)b+p(~-b)l
credibility
0,
<
condition
>
[b2(1+(1-kt)~)1/[~(1+~)1-kt)b-~)21
implies that
is threatened by the
a large
monetary policymaker
produce X t
= x
inflation burst
sector's
monetary expansion
in order
and output
loss,
so
that
he
prefers
to
comply
with
the
proves
definition of
the policymaker
output grows
the
the Introduction,
consistent (by
such that
first
of
the
results
achieves
his
bliss
point,
where
changes
the
nature
of
Kydland-Prescott's di lemma
and
result
can be
(the
second
derived from
was
proved
in
the
Proposition 1 and 3 of
the Appendix, which define the inducible region of the game and
the credibility
condition. The
monetary authority
can indeed
(4.9)
IRs
{(x*t,P*t)
>
>
(V*pt-VOpt)/(VPpt-VOpt)
t
where V*Pt
loss when
(x* t, p* t) .
= VPt(x*t,P*t)
the monetary
is the
private
= 1 ... n
sector's
192
Again, the
However,
in
problem
th i s case,
policymaker can
of
multiple
solutions
arises.
minimize his
loss function
where x"
and mt
is a
l ... T}
In
this
is unique.
reputation,
is
be assumed
to be
a subjective
Hence, the
policymaker may
6,
the
policymaker's
if
prior
induce a
desired solution
which does not belong to IRs, so that either he carries out the
announced punishment or he acquiesces. In both cases, multiple
(sub-optimal) equilibrium paths are possible (see CarraroSiniscalco, 1987)
be avoided.
Furthermore,
understood from
the
fourth
result
can
easily
be
which arises
when the
authority in
the announced
optimal
private sector
rule,
has
now
threatens the
become
monetary
credibility
the
previous
section
the
Stackelberg
Trigger
193
~(Pt
- p8 1t )2
194
Table 3.
Player i
-2
,x - 2
Player j
YOpt,Y+it
is
defined by
i's los&
(5.2),
x, pe. t
Xit
YO p t
when he
the announcement
player
YOpt,YOpt
i's
best
reply
believes the
function
Y+. t
is player
their rational
reaction strategy
(5.2). We have:
(5.3)
Notice
XJt
Pt
that
= x, aggregate
= O. Hence player
when
outcome is
Xt
so that mt
Xit
adopt
x and
0, so
is easy
RF.(m t
to
prove
that
Y+. t
>
YO p t ,
so that
the game.
An analogous
case, that
threat if
is when
Xt
result can
be obtained
the policymaker
in
the
actually carries
other
out the
game is summarized by :
195
Table 4.
Player i
*2
,X
*2
Player j
for all j:fi
x.Jt = Rp i
(m t
YPPt,Y-
it
(4.4.2).
player knows
choose Xit
that when
different from
x,
he
all
finds
the
it
other
196
Prisoners' Dilemma
therefore arises.
Suppose 6
is such
that
(4.8)
but
has
always
an
incentive
(whatever
the
monetary
loss.
The
above
discussion
~x/b)
that.
being
the
producers identical.
x. Pt = 0)
implies
(x t
= O.
Pt
not carried
the atomistic
players is
still described
by
Table 2.
As shown
the incentive
not to
strategy (4.8)
Dilemma can
1986;
above. there
is however
choose Xit
is credible.
be found
The solution
in several
Benoit-Krishna. 1985).
ways (see
[~:.21
(5.5)
Prisoners'
Fudenberg-Maskin.
i.e.
otherwise
shown
assumption all
by
Kreps
and
players choose
Xit
al.
(1982).
under
this
and
announcement is
some t.
As a
consequence.
credible. the
if
the
monetary
sequential equilibrium
.
x
of
the
game is
(Xt
We
uncertain not
only
need
to
assume
only whether
punishment whenever
producers actually
desired strategy.
each
the policymaker
x.
accept to
In other
that
but
comply
also
carries
whether
with
words. each
producer
the
out
the
is
the
other
policymaker's
player does
not know
197
with certainty
the loss
function of
the game.
Summing
up,
under
the
above
general
incomplete
producers
when
the
monetary
strategy
(4.1)
is
credible.
Finally, if
multiple Nash
equilibria of
atomistic producers
possible to
RFi(m t ) is
not a
contraction (so
the single-play
exist), and
game
that
among
the
is
finite
is possible
and
to
use
no
incomplete
the
bounded
(x it
finite repeated
x*) to
game among
the atomistic
producers when
the
conclusions of
by introducing
Kydland-Prescott's dilemma
policy problem
eliminated.
and the
an
the previous
atomistic
constitutes a
multiplicity of
section are
private
sector
partial view
solutions
of the
cannot
be
6 Conclusion
This paper
policy. The
has shown
Folk
Theorem
of
monetary
198
defining more
precisely beliefs
and behaviours
out of
equilibrium (see, for a first attempt, Kohlberg-Mertens, 1986).
Finally, we proved that there may exist a credible
strategy which minimizes the monetary policy's loss function
and which is not dominated by any other strategy (i.e. the
po1icymaker achieves his most desired outcome).
The economic fact which explains why the private
sector can be induced to comply with the po1icymaker's desired
outcome is that inflation is costly.
In the simple monetary
model previously considered, only unexpected inflation can be
used by the monetary authority to threaten the private sector.
In this case, the private sector evaluates the probabilistic
cost of
being punished
when he
does not
choose the
po1icymaker's desired strategy, i.e.
the cost of having an
inflation rate higher than expected, and then he decides
whether to comply with the policymaker's decision.
However, if both expected and unexpected inflation
are costly, then the monetary authority's strategy is more
effective. In particular, the announced monetary strategy, i.e.
a monetary expansion whenever output growth is not greater than
its natural rate of growth, is more likely to be credible when
199
both expected
and unexpected
inflation are
included
in
the
economic rationale
the
private
variability is
for including
sector's
costly and
loss
expected
function
that output
is
inflation
that
output
variability depends on
is fairly
indeed realistic
out a
trigger monetary
plausible and
strategy proposed
probably "deja
vu".
in
It is
not achieved.
APPENDIX
Let VO(SO,SF)
policymaker's loss
function, which
and VF(SO,SF)
function
and
are assumed
preferences over
be,
the
to
respectively,
private
represent
the respective
sector's
the
targets. The
the
two
loss
players'
subscript D
is
the game.
the subscript F,
The loss
functions Vi'
defined, continuous
assumed to
D,F,
is
i =
and bounded
be concave
the strategy
on S
with respect
of each
D,F, are
= SOXSF
assumed to be
and euch Vi is
player and
Si is
i =
the strategy
space.
The constituent game is therefore r(N,S,V) where N is
the number
consider T
repetitions of
the players
we obtain
ai =
the constituent
If we
i = D,F,
1/(1+rl)'
=D,F.
Both
finite
and
infinite
horizon
Carraro
Strategy equilibrium
dominant player's
stage of the game.
(1985),
Stackelberg
Trigger
200
Let us consider the constituent game r(N,S,V).
Suppose
that
optimum outcome,
(SO,SF)
is
the
dominant
i.e.
(A. 1 )
arg
min VO(SO,SF)
SO,SF
player's
the
dominant
player's
losses.
Stackelberg
Trigger
Strategy is defined as :
if SF = s F
(A.2)
otherwise
where sPo
player threatens
comply with
defines the
to carry
the dominant
punishment that
out any
the dominant
player's desired
strategy SF'
The
arg
min
SF
VF(stBo,SF)
and
(A.3.2)
where Ri(sj)
(4)
defines player
i.e.
(A.4)
arg
D,F j
In other
the follower
D,F
words, the
is induced
dominant player's
=1=
min
Si
Vi(Si,Sj)
to choose,
desired strategy
in his
s F'
own
Then,
interest,the
the
dominant
must be
201
Trigger Strategy
for the
policymaker to
that the
achieve (S*O,S*F) at
stso must be
~~,
must assign
a positive
dominant player
threat whenever
i.e.
actually carries
out the
announced
SF
detailed
credibility problem
found in
analysis
of
the
for Stackelberg
Carraro (1985)
effectiveness
and
where the
shown
F'Lo"p"Q..s..i..t..i. Q.ll-l
then it
cannot be
induced by
trigger strategy.
Hence,
Proposition
Stackelberg trigger
defines
strategy and,
the
most
powerful
IR
(A.6)
are
where
follower's loss
when SF
player actually
carries out
eq.
(A.S)
the
differs from
dominant
S-F
and
player's
the
and
dominant
implies:
202
VP F
min
SF
max
So
VF(SO,SF)
consider now
where we assume at = I,
D,F.
If
> VOo ,
and VP o
the only
finitely repeated
complete information
sub-game
perfect
is assumed
solution
of
the
where (SOO,SOF)
complete information,
player never
finds it
profitable
whenever the
follower does
to
carry
not believe
the dominant
out
his
threat
SOF'
dominant player's
loss
function,
so
that
he
is
uncertain
the dominant
6 the
assigns to
the first
player's desired
prior probability
the possibility
stage of
that
that the
the game
strategy.
the
Let
private
us
sector
and let
us
suppose
that
this
If
in
1985)
complete
information
is
assumed, the strategy stBo can induce the follower to adopt S*F
at any stage of the finitely repeated game if and only if :
(A.7.1)
(S*O,S*F) E IR
(A.7.2)
Then the
sequence {SOt
be sustained
as a
203
= 00,
1,2, ... }
s"o, SFt
S"F ; t
consequence, the
minimum
conditions are
of
his
loss
dominant player
function
satisfied. Hence,
at no
any
achieves the
time
the
above
strategy
st- d
which
is
therefore
time-
consistent.
The above
results define
REFERENCES
Abreu,
D.
(1986),
"Extremal
Equilibria
of
Oligopolistic
(1976), "Rational
Monetary Policy",
Expectations and
Journal of
the
Role
Monetary Economics, 2,
of
1-
32.
Barro, R.J.
versus Discretion",
Economic Journal,
Supplement,
96,
Discretion
and
23-37.
Barro, R.J.
and Cordon,
Reputation in
D.
a Model
Monetary Economics,
(1983a),
"Rules,
of Monetary
12, 101-121.
Policy", Journal of
204
Barro, R.J.
and Gordon,
Monetary Policy
D.
(1983bl,
in a
"A
Positive
Natural Rate
Theory
of
Model", Journal
of
(19851,
C.
(19851,
Policy
"Hierarchical
Analysis",
Games
chapter
Macroeconomic Policy
Analysis",
Princeton University.
Also in
D.
(19871,
eds., Developments
for
Macroeconomic
of
"New
Ph.
D.
Methods
for
Dissertation,
Control
Theory
for
(1986al,
Martos,
B.
and
Modelling and
Ziermann,
Control of
C.
(19871,
eds.,
National Economies,
in
Dynamic
Pergamon
Press, Oxford.
Carraro, C.
(1986bl,
(19871,
C.
"Credibility,
Indeterminacy of Macroeconomics",
Y. and McKenzie, G.
Reputation
and
the
and Siniscalco,
D.
(19871,
"Producers' Strategies
(19861,
Some Recent
Theoretical Developments",
Federal Reserve
J.\I.
(19711,
"A
Noncooperative
Equilibrium
1-12.
for
205
Friedman.
(1985),
J.V.
"Cooperative
Horizon Noncooperative
Equilibria
in
Finite
J.V.
(1986),
Game
Theory
with
Applications
to
and
D.
Equilibria
D.
Levine,
of
Finite
(1983),
and
"Subgame
Perfect
Horizon
Games",
Infinite
and Haskin.
Repeated
Games
E.
with
(1986),
"The
Discounting
Folk
or
Theorem
with
in
Incomplete
under Imperfect
Price Information",
Econometrica,
52,
87-100.
Kalai, E.
and Stanford,
Strategies in
V.
(1985),
"Conjectural
Variations
and
Hertens,
J.F.
(1986),
"On
the
Strategic
and Vilson,
R.
(1982),
D.H.
et
Finitely
al.
(1982),
Repeated
"Rational
Prisoners'
Cooperation
Dilemma",
in
Journal
the
of
and Prescott.
Discretion
the
E.C.
(1977),
Inconsistency
"Rules
of
rather
Optimal
than
Plans",
206
Lucas. R.E.
(1973), "Some
International
Inflation Trade-Offs",
Evidence
American Economic
of
Output
Review,
326-
334.
Robson. A.J.
(1986), "The
Existence
Reaction Functions
of
for Dynamic
Nash
Equilibria
Models
of
in
Oligopoly",
(1986),
K.
"Reputational
Constraints
on
Monetary
Price Wars
during Booms",
American
Economic
A.
Schotter.
(1980) ,
The
Theory
Economic
of
Social
R.
(1978),
"The
Chain-Store
Paradox",
Theory
and
Decision, 9, 127-159.
( 1986a) ,
"On
Continuous Reaction
Stanford. V.
Equilibria in Duopoly Supergames with Mean
Function
Payoffs".
V.
(1986b).
Equilibria
in
"Subgame
Discounted
Perfect
Duopoly
Reaction
Function
Supergames
are
1. Introduction
This
expectations
asymmetric
the
and policy
in
model
of
monetary
policy
with
private sector.
the
policy maker
the
optimal policy
Certainty
In this
is
n2n~quadra~
equivalence does
and
the
policy maker
policy
and
action on
also on
when
for the
into account
the information
his ability
the private
capacity
must take
to affect
sector forms
policy
maker
set facing
its
to
Despite
intervene
this
in
the
expectation
formation process
there
to
is a
which both
the expectations
optimal
policy converge.
problem
which is
we
the
describe a
In section
solved analytically
number of
factors affecting
private sector
2 we
introduce the
in section 3.
and
the
policy
In section 4
the convergence
to the
unique maximizing
consider the
208
Salmon
(1987). The
monetary authority
faces
finite
horizon
(1)
N
(~
- 1/2(m P )2)}
~l(eixi
i=O
instrument is
the
planned
rate
of
monetary
The
private sector's
information set
is imperfect
since it
is
mPi' Actual
which has
- E [mi I Ii 1
i
ei
mi
mi
mP + III i ,
where
(3)
that
to deviate from
(2)
The
is assumed
authorities
III i ,
by
monetary growth
llIi
is then given
mP - E [mpi L 1 + llIi
0,1, ...
N(O,02)
'"
private
sector
is
assumed
not
to
know
the
basic
inflation in
it's objective
preference parameter
allowed
to change
components
is considered
over time
leading to
with both
function in
to
be
each period.
stochastic
Xl
and
1,2, ...
209
The
then
information set
given by
{mi-l,
mi-2,""
structure
mol,
together with
and parameters
constructs
(except
a knowledge
Xi)'
The
of
the
monetary
Ii
model
authority
preferences
Xi as
particular
well
as
Ii,
nested information
hence
the
asymmetric
and
in
(5)
Under
these informational
able
to solve
1 ,
xo,
Ii)
assumptions the
the private
monetary authority is
sector's prediction
that
quadratic the
due to
the presence
will affect
when
it forms
that
it
may
mechanism
also
choice
of monetary
linear
reflect
this
message
the
monetary
It is this
the effect
of the
optimal
linear
quadratic.
While
conditional
of
of
to
the
past
be
the problem
objective
function
monetary
linear
to a
growth
function
may
rates.
of
the
take into
manner by
authority.
period
that
affect the
problem
clear
expectation
when we
shall
qu?dratic
formation
set returns
one for
also
a linear
is
expectation
non
inflation
information
below
function
a nonlinear
Forcing
is
affect
growth on
function of
surprise
While it
influence through
the problem
expectation
of
the information
directly
and hence
channel of
objective
it's expectations
second
makes
the
problem is
objective function.
authority
and
although
to find
The discussion
case for
reasons of
without introducing
the optimal
here will
policy for
the
be restricted
tractability and
to
monetary
the
two
excessive notation.
The
reader
is
210
referred
to Basar
complete
discussion of
papers,
our intention
in the
present paper
is to
consider the
convergence
equilibrium
in the
of
the general
interval
we are
economic
parameters on
able to
consider
the
the iterative
effect
of
process and
the
various
also
observe
the
two
period
problem
(with
1) ,
the
E{
maxJ
(6 )
i=O
9xo + A ( 1-9) +
x,
subject to
l3 i c(xi,Ii,m P )}
Vi,
Vi
fIl i
c(x,I,m P )
(7)
where
The
change in
discussed
'"
N (0,0 2
) ,
N (0,0 2
'"
i =0, 1
mP(x-E[xIIl) - 1/2(m P )2
the objective
fully in
function from
Basar and
is
equivalence
E{mPxi - E{mPlli}x i
1I2(m P )2}
1/2(mP)2}
E{mPxi - mPE{xill
which
211
(9)
E[E{mPII1}E{xlll.})
i
= E[mPE{x.II.})
1
The optimal cost function for the two period problem can then be
written. having substituted the optimal policy rule for the final
period. as
max J a = max
E{1/2~(x.
- 1/2(m P )2}
a
( 11)
as
(12)
~a)
v.
212
( 13)
1/2~a2
max
E{1/282~(xo
- E[xo\m P + O])2
o
which
(x o - E[xo\Io])m P - 1/2(mP)2)
o
as,
( 14)
1/2~a2
+ max F(a o )
ao
where
F(a o
difficulty
policy
is
in solving
this problem
o
growth
(or the
lies
is part
in
that
the
optimal
of the conditioning
preference parameter)
is taken
in period 1. The
optimal
1,
predictor (expectation)
say
a. Since
will
depend on
government's
sector's
of
a
information
way
the maximand
the government's
optimal policy
forecast function,
policy rule,
rule
will
we need
depend
and in
on
turn
the
the
private
the response mappings 6(a) and a(6). We now show that there is
unique fixed
point
corresponding
to
the
unique
maximising
the
unknown
predictor
function,
6,
in
the
G(6,a)
E{1/282~(6(I,)-xo)2
( 16)
mP
o
noting
that the
problem"
error
= a ( x o , 10
through its
this,
the prediction
1/2(m P )2)
0
1, = m P + o
o
information set
facing the
recognising
),
+ (x o - E[xo\Io]lm P
10 will
be empty.
The "policy
6.
will determine
The
monetary
its optimal
authority,
0:,
by
upon
maximising
213
max F(o:)
( 17)
0:
0:
6(0:)
(18)
or alternatively,
such saddle-point
pair (6 M,ex M)
we have
from
max F(ex)
(20)
0:
and furthermore ex
is the unique maximising
M
M
(6 ,ex ) is unique as a saddle-point solution.
Before
some
presenting the
notation. Let
L =
La be
main result
solution above
we first
introduce
equation
La 2
( 21>
(L a
2
'"
92~
+ 02
if
= geL)
)2
a
'"
that provides the largest value for F(ex) and let Ka be given by
X
(22)
214
(1-Ke2~)
r (K)
(23)
=
(1-K2e2~)
(24)
< 0
)02
x
..
THEOREM
r(
o
<
(i)
~
equation (21)
The polynomial
(L,
is identical with
and admits a
L
with
Lo < 1.
If Lo satisfies (24), the game G admits the unique
(ii)
saddle-point solution
Xo + KoIt
Lo(xo-i o )
(25)
;5
;5
xo + ~ (Lo)I.
r(Ko)(xo - E[xoIIo])
1 -
(i)
Existence and
(Lo)2e2~
>
<=>
1 - K2
o
e2~
>
Proof
solution,
Lo
solution
lies
of gel)
fact that
(21)
in
the
changes sign
is
open
right
Here we
interval
maximising
at most
any a
a-,
this conditional
and
(0,1),
the
(L
side inequality,
into (23).
for
hand
F(a)
the
of (21)
derivative
The
uniqueness of
G(6-,a)
is a quadratic function
215
(27)
[1 -
L2(02
0
"0
(L 2 02 + 0 2 )2
o
The
condition T
concave
)282~
<
function of
"
= T
G(6~,a)
being quadratic,
is a strictly
it admits a unique
solution which is
ex(x a
and by (i)
ex(x o )
This
verifies the
the
Lo(x - xo)
left hand
o.
condition T <
side of
be simplified to
T = 1/2(Lo (I-L o )(02
Xo
and
Note
and
hence the
point
that
this condition,
is
G(6~,a)
interchangeability property
equilibria [Basar
(25)
1)
concavity condition
that under
using the
/0 2 ) -
and Olsder
indeed the
of
(1982)]
multiple
saddle-
readily
follows
it
(24) .
(iii) This follows readily by noting that
condition (24)
solution
problem
of (21),
of the
Theorem is
and this
depends on
given
the
o
terms
of
the
parameters
of
the
in
parameters
0 2
K
,0 2 ,
...
8 and
216
(28)
< 40 2 /0 2
82~
which
xa
(29)
.0 2
x
(L 2 02 +
a x
a
82~
a
<
0 2 )2
but since
L 2 02
a x
Max
.0 2
x
(L 2 02 +
a x
a
La
82~
x
0 2 )2
(28), or
for the
the less
satisfaction
of (24)
not
saddle point
admit a
larger
simply means
(that is
value is strictly
than the lower value) ; however this does not rule out the
possibility
that
the
maximising
solution
for
F(a)
is
still
linear.
In fact
if we restrict the monetary authority to linear
(or affine) policies at the outset, say
(30)
then
we can
show through
routine manipulations
that a solution
unique
217
(31 )
r (K' n')
and r are defined by (22) and (23), respectively. Note
where
that if the sequences {K'n'} and {L'n,} generated by (31), for
any initial conditions, converge, then the limits have to be the
optimal solutions Ka and La respectively, defined earlier. The
fact that the pair
Xa + KaI,
218
would
as
simplest
approaches is
factors in (31)
through the
llsL(n> +
This
Over
relaxation
or Jacobi
(31). The
into
the
relaxation
process.
one
step
levels of
This
memory
simple Gauss-
provides
than the
to
another.
more
basic iteration,
be
Different
seen
as
confidence that
most recent
corresponding
than the
Successive
parameters could
different
of
behavioural interpretation
from
other's
the class
relaxation terms
buffering against
party
is in
algorithms rather
adjustment
reasonable
(l-~s)r{K(n
iterative scheme
Relaxation (SOR)
Seidel
by
of
leading to ;
(32)
in
introduction
response. The
for
corresponding
the parties
level of
relaxation parameter
values
moves
to
the
place in
confidence
away
the
from
each
as
the
the
value
shows
the policy
the
graphs
of
the
two
functions
The
and L
exact shape
economic
of these
e, f>,
parameters
a unique
equilibrium
Starting
from any
the
unique fixed
can
(1-Ke 2 f
(1-K 2 e 2 f
functions depends on the values of the
0
and
2
H
be seen
but the
clearly
initial position,
point proceeds
existence of
..
from the
say <Ko,Lol
by iterating
figure.
convergence to
between
the
two
functions.
Table
been
the
carried out
rate with
which the
and
219
02
noise
policy
02
02
K
number of
observations
simulations.
The first
expectations
equilibrium has
unable
to cause
through
values
of K
and L
all
of F(a)
results
when
the
the weaker
been
instance run
and always
to
the
7)
same
and 8
factors,
discussed
It is
from 110
and
for
is also obtained
to 14.
8 were
reduces
~8'
In general
relaxation factors
required to
in run
Aside
the value
of
sensitivity of
it can
and when
it is
closer
to the
(L=l).
Similarly when
or
and
~8
exist.
certainty
the product of
of the
~p
very small
myopic
small (eg.
of
large
the square
number
from the
of
the
that reduce
becomes
above
that convergence
~p
calculate theoretically
used
to a.
2 ,
for these
iterations
conditions
relaxation
iterations
4, 5
seen from
values
sufficient
with respect
for runs
importance
and
we have
(see for
been rapid
violated
to
rational
has invariably
concavity
to
the
in that
to converge
numerical
when
of
the
cases
be
from
convergence to
been global
the simulations
initial values
convergence
follow
being that
differing choices
divergent
rate of
equivalent
~
and 9
2 ,
solution
value
to
10
shows the
walk
and there
greatest
is
again
case when
is no
deviation from
a
rapid
government preferences
discounting. This
follow a
solution
random
provides
the
convergence
to
the
rational
expectations
220
equilibrium.
more
fully explored
should
also be
matrix
with the
equations
in Basar
noted that
but in
SOR method
and
Salmon
the spectral
is affected
various experiments
(1987).
Finally.
it
we found
ordering of the
no effect on the
convergence to equilibrium.
Figure 1
L
K=f(L)
.........
_-/
5. Conclusions
In
and
empirically
equilibrium
the
this paper
the
we have
convergence
to
rational
expectations
expectations generating
221
non
conditions
found
unde~
to be
behaviou~al
to the
the
~educing
va~iance
in
Salmon (1987),
a
"Stackelbe~g"
Cukie~man
and
Meltze~
mo~e
the optimal
solution
fo~
(1986) has
Table 1 -
gene~al
~elaxed
and
ite~ations
In
the
pa~amete~
~atio
of
the
unce~tainty
in
the
is
small
ve~y
difficult. As discussed in
de~ived
the same
de~ived
Nume~ical
its
it seems,
gene~al
dominant
2 ,
~atio
policy
we~e
of the
noting.
this
demonst~ated
impo~tance
of
of
theo~etical
simulations and in
the initial
When
is potentially
conve~gence
and
be
that the
the value
p~efe~ences.
gove~nments
wo~th
noise to
moneta~y
The
at least,
is
conve~gence
could
numbe~
is
inte~p~etation
The
p~oblem.
nume~ical
and global.
these calculations
f~om
affecting
cont~ol
conve~gence
~apid
in
ite~ation
which
~elevant
was
conve~gence
to
equivalent
ce~tainty
above
policy
Basa~
co~~esponds
p~oblem
that
a "Nash" solution.
simulations
Initial
0.95
0.25
0.95
0.95
0.95
0.95
0.95
0.95
0.95
1.0
1.0
0.1
Relaxation
~un
8 when
0.8 1 . 0
0.8 1.0
O. 1 1.0
0.8 0.1
0.8 0.5
0.8 10.0
0.8 1.0
0.8 O. 1
0.8 0.01
1.0 1.0
1.0 10.0
0.1
1.0
pa~amete~s
~p
6
4
3
110**
6*
7
7
14**
FAILED
9
8
3
and
~p
0.9 and
..
** Violates
..
(0,0)
0.4905 0.8220
(0,0)
0.4995 0.9593
(0,0)
0.4999 0.9976
(0,0)
0.9548 0.9411
(0,0)
0.6997 0.8181
(0,0)
0.0872 0.9514
0.4905 0.8220 ( 1000 , 1000)
(0,0)
0.9548 0.9411
TO CONVERGE***
(0,0)
(0,0)
0.4656 0.6823
(0,0)
0.0849 0.9217
(0,0)
0.5
1.0
set to unity
0.45
conve~gence
fo~
all
~uns
except
222
02xo/02~
4 and
e2~o2x%2~
xo/02~
<
REFERENCES
Basar
T.(1987a), "Solutions
control
problems
with
active
learning",
submitted
for
publication
T.
Basar
(1987b),
algorithms
"Relaxation
for
equilibria",
on-line
techniques
computation
Proceedings of
the IEEE.
and
of
asynchronous
noncooperative
Conf. on
Decision
T. and
G.J. Olsder
(1982), Dynamic
Non-Cooperative
Game
T. and
H. Salmon
information
(1987), "Credibility
and the
value
of
a time
consistent government
?", CEPREMAP
discussion
paper.
Cukierman
A. and
A. Heltzer
credibility
information",
(1986),
and inflation
"A
theory
of
ambiguity,
1009-
1128.
Kydland
inconsistency of
Economy, n03.
Hiller
of
optimal
in
an
open
economy",
223
Salmon
H.
(19861,
"Rational Consumption Behaviour", Working
Paper, Department of Economics, University of Warwick.
Whitman
C.
(19861,
"Analytical policy design under
expectations", Econometrica, vol. 54, n"6.
rational
INTRODUCTION
In
industrial
most
institutions, the
implement
countries
central bank
fiscal
and
distinct
two
monetary
policies.
The
degree
of
countries
depending
considerations. For
Federal Reserve
historical
instance the
Bank have
independence compared
France. But
on
institutional
Deutsche Bundesbank
a well
to the
and
establish~d
Bank of
and
~he
reputation
of
policymaker underlying
the optimal
coordination between
should be abandoned.
The design
question
of
instruments
coordination
than
of
alternative
of
question
l~ss
objectives
cooperation
between
a
and
two
monetary and
the budget deficit can raise two kinds of conflicts between the
central bank and the Treasury (I)
(I) Treasury,
fiscal authorities and government are used ~o
indicate the authority which implements fiscal policy. Central
bank and monetary authorities indicate the authority which
implements monetary policy.
2~
il Both
The Treasury
of monetary
minimizes the
nominal growth
thus alleviates
this view
base
against
of the
its
liabilities
it
the method
In
the
situation prevailed
end
of
in some
exhibited a
despite the
potential crowding
in
favor
strong
Therefore the
of
the
authorities led
central
to a
this
aversion
conflicting
since
to
central
inflation.
And
were reluctant
bond issuance.
1970's
Western countries
banks there
the governments
the
to reduce
significantly their
the
decreasing
preference
monetary
of
financing
both
of
the
the reduction
of the
public
debt
should
occur
through
an
ongoing monetization
of the
central bank
achieve both
the objective
reducing the
public debt
of reducing
outstanding. Since
the beginning of
banks and
United States.
analyse the
governments.
particularly
in
the
linear quadratic
dynamic
game
this
paper
first part.
does not
private sector
two authorities.
The evolution
under
institutional
settings.
different
We
show
that
the
227
cooperation between
reduction of
the two
public debt
behavior of
authorities leads
outstanding.
the private
sector
is
In
to the fastest
a second
explicitly
part, the
modelled.
We
Taking this
behavior into
account,
both
neutral vis
vis
the strategic
and fiscal authorities. The central bank and the Treasury share
a common
goal:
outstanding.
the reduction
But
the
creation whereas
through a
the
budget
central
on the
the
of
nominal
wishes
wants
Thus
way
face the
the
bank
Treasury
deficit.
Treasury disagree
outstanding and
of
to
public
to
limit
boost
central
reducing
debt
the
bank
the
following dilemma
money
economy
and
the
public
debt
either
they
cooperate to slow down the growth of the public debt but forego
their own objectives; or they forswear cooperation in order to
achieve their
own objectives,
policymaker will
In doing
bear the
in the
hopes
that
the
other
the public
debt. Thus
monetary and
government
fiscal objectives
and
on
the
of the
institutional
central bank
framework
and
the
determining
linear
quadratic
dynamic
game
la
respect
to
the
weights
assigned
to
the
budget constraint
gives the
law
of
228
All variables
denotes the
interest payment
liabilities of
r- is
scaled
of
creation of
to
is the
is the
nominal
budget
income
deficit
d
net
of
the real
of growth
are
public debt
income.
monetary base
The
central
whereas the
bank
controls
budget deficit
is
the
in
of central
bank and
of the Treasury
(2) L"
According to
(2) and
(3) both
policymakers wish to
with the
absence of
lump sum
taxes. A
larger
public debt.
introduces distorting
debt target
effects on
is normalized
evolution of
the
public
debt
without
changing
the
the monetary
consistent with
inflation or
base. The
objectives such
as
the
control
of
central bank
is definitely
independent:
its monetary
target. Conversely
central bank
merely finances
when
229
fiscal
authorities.
The
parameter
represents
time-
authorities on
the other
hand
wish
to
(2). The
assigned to
parameter r
indicates the
relative
weight
peculiar. since
expressed
economic policy.
relationship
of these
in
Thus we
between
loss functions
the preferences
terms
make
final
of
a
the
LF
of policymakers
ultimate
crucial
LM and
goals
assumption
macroeconomic
of
the
objectives
and
commit itself
t=O
bank
takes
into
account
its
influence
on
ft
+ 1
(2) With target not normalized to zero the time path of public
debt would be for the benchmark simulation
dt + 1
X d t + y(f - ffi) - z(d F - d M) where x is the solution of
the game in a closed loop Nash framework and y and z are
positive constant. Therefore x is independent of the targets f.
m. d F and d M.
230
on mt
+1'
optimal
response
strategy). Thus,
order to
fiscaf strategy)
strategy (resp
The monetary
the
to
we get
fiscal
a closed
strategy
(resp
is the
monetary
Equilibrium.
loop Nash
In
and
(2)
(3)
we
obtain
the
indirect
loss
is given
(6b) oV M (d 3
6d 3
)/
Td 3
(6c) 6V F
(d 3
6d 3
)/
rd 3
By backward
strategies for
the append i x)
recursion we
both authorities
get the
Nash
02 + ar 2 (1+aT)
~ raT (---------------) d
x
(8)
closed loop
02 +
~r2(1+~r)
~rr(
{me,f e }
2
where
the
superscript
"c"
stands
for
closed
loop
Nash
equi 1 ibrium.
with 0
1 + aT +
~r
From these
public debt:
strategies we
get the
evolution of
the
231
The analysis
of the
closed
loop
Nash
equilibrium
The
existence
associated with
means that
rate of
of
stationary
the condition
equilibrium
that r
is less
is
usually
by two
equilibrium is assured if :
(10)
1 +
<
ii) When
condition (10)
equilibrium
absence of
fiscal
+ aT
~r
holds,
strategies
are
precommitment on
authorities,
these
then
time
the
closed
consistent.
the part
of
Thus,
Nash
in
the
monetary
and
strategies
are
both
non-cooperative
loop
credible.
iii) When
burden of
the reduction
the Treasury.
The
lower
the
T,
slower
the
rate
of
debt
Treasury. Thus
(T
0), the
the debt
reduction. Conversely,
give priority
to the
when the
deficit, the
fiscal
authorities
adjustment of
the debt is
the monetary
present (a
Treasury. As
reduction (3)
bank and
-+ 0
is higher
of
today than
fact,
the
cost
tomorrow for
of
debt
the central
authorities have
high time
preferences for
the
present
the
232
reduction of
rate of
the public
interest is
case, both
lower than
authorities will
In this
condition (10)
cooperative behavior,
debt. We
will see
there is
now if
holds, and
of one
authority only.
as the
parliament or
1/2
central
the Treasury
present for
bank
objectives
relative
to
authority
which
solves
the
T
(12) Min 1/2 [ ~ 6 t (ft + pm t
"'t,r t
t=l
s.t. d t
+ 1
= rd t
+ ft
)2
+ (r +pT)d 2
t
+ 6T
+ 1
]
(r + pT)d 2
T+1
- mt
233
( 13b) P S = (r + Il T) d:3
:3
where p~ is the shadow cost of d:3.
:3
The solutions
the values
of this
of monetary
optimal control
and fiscal
problem give
authority
~
where
I[-( ~-:_~_:_T-J
r_(
In the
[ - ::-:-:-:-)- ]
,6 d ,
+ (1/1l)]6
[1
[r + IlT]
r6 d 1 ,
cooperative equilibrium
( 16) (d S
d S ,d S )
:z
:3
Let us
now compare
run by
single
controller.
According
equation
to
(9)
and
(16),
( 17)
(3r
+ eXT)1l
(1+1l) (r+IlT)
< 6
This condition
reduction is
with two
not always
authorities,
This result
implies that
higher with
i.e.
depends on
coordination does
the value
of Il
(the
weight
of
the
of
the
aRd of
single
the
time
controller.
preference
When
it
for
the
neglects
the
234
objectives of
the central
always fulfilled
point.
(17) still
nil),
(~
(17)
condition
is
Conversely
objectives of
bank
when
the
the Treasury
holds and
controller
single
the monetary
ignores
to infinity),
(~tends
the
condition
authorities are
at
their
bliss point.
Figure
and fiscal
authorities. The
is the reaction
M (resp F)
fiscal authorities).
equilibria with
The dotted
curve represents
the value
respect to
neighborhood of
zero, the
neighborhood of
F. Conversely
of
cooperative
When
~.
cooperative equilibrium
when
is in
the
infinity, the
tends to
is in the
both the
authority have
the same
time preference
for
the
present, condition
(17) is
without ambiguity,
always verified.
Consequently and
Monetary
and
fiscal
strategies
with
precommitment
Cooperative institutional
feasible. However
decentralized
even if
between
two
distinct
authorities,
each authority
central bank
precommits itself
thus determines
better
It is the
on a
sequence of
its own
sequence of
moves with the knowledge of the moves done by the Treasury. The
sequence of
moves to
response to
the sequence
same conditions
obtained is
which it
commits itself
is the optimal
thus an
235
figure.
236
H
( 18) H = a t /2
( 19) H r
H
[rd t
[m'" + Td"') + a t + 1 p
t+1
t
t
+ ft
- mt - d t + 1 )
- d +1)
t
pF
:3
rd:3
0 + r"'a
~aTr02
0 + r
r2(a2T+~2r)d1' -~rr
The superscript
"0"
j)
02 + rZ(a 2 T+j)zr)
in equations (21).
these
open
loop
equilibrium
strategies
is
1'0
-------------------d 1
(23)
0 2 + r 2 (a Z T+j)"'r)
When monetary
and fiscal
-0-z--+--r-2--(-a-:-:-+-~-Z--r-)d)
237
loop framework.
they act
closed loop
in a
framework. This
can be
when
seen from
The
ambiguous and
the superiority
of one
be shown
in the
in order
to ease
equilibria.
ambiguity
with
comparison
strategies is
cooperative
equilibrium
Table
the
the
rate
allows
of
of the
us
debt
to
different
conclude
reduction
is
kinds
that
of
without
higher
in
the
different
increased. This
institutional
is the
settings
this
both
monetary
case when
rate
can
and
be
fiscal
authorities have the same time preference for the present. When
a single
controller is
policies. the
draw
rate of
similar
when
regards its
vis
debt reduction
conclusions
precommitment as
accelerated vis
in charge
is accelerated.
each
authority
own strategy.
the closed
loop but
We can
makes
The reduction
is
not vis
vis the
cooperative equilibrium.
behavior of
both authorities
and the
soon as
model. Private
money and
we introduce
sector is
by its
the private
characterized
anticipation behavior
model of
the economy.
It has
will not
directly interact
no
in the
sector in
the
demand
for
by its
in
one
simplified
strategic behavior. so
game
between
fiscal
it
and
238
monetary
Nevertheless
authorities.
behavior. pri.vate
sector put
strategic choices
of both
by
its
an additional
anticipation
constraint on the
authorities. This
will change
the
with Xt
Pt - pt-l
(25) Yt
with x"
p" - pt-l
(1
d)(Xt-l
All variables
level. y
denotes the
output at
current
its natural
rate
inflation. I
of
t-l
inflation.
the nominal
x"
stock of
the
anticipated
rate
of
is an
aggregate demand
Equation (26)
aggregate supply
function
describes an
adaptative expectations
behavior.
y~
= h [(x t
t-l = e( at
- Yt-l
(28) Yt - y
- Xt-l) - (x" t - x .. t - xt )
1 )]
x" t
[( I-d) /h]Yt-l
Xt
and
239
(32) LM = 1/2
[~ at (m 2
(33) LF = 1/2
[~ ~t(f2
t=l
t=l
+ an 2 + Td 2 ) + a T+ 1 (an 2
~n2
t
+ rd 2 ) +
t
T+l
~T+l(~n2
T+l
+ Td 2
T+l
)]
+ rd 2 )]
T+l
where a and ~ are the weight of inflation for the central bank
and the Treasury respectively.
We make the further assumption that the target
inflation of both authorities is zero. Including two different
targets n M and n F would have simply added inertia in the
solution of each equilibrium.
The loss functions (32) and (33) with the constraints
(1) and (31) define the new strategic model (5).
As above. the various equilibria of the dynamic game
are derived from a two-period model. However the analytical
240
solutions
are
more
comparison between
cumbersome
the
impeding
different
equilibria.
Loop Nash,
any
superficial
We
have
thus
Open Loop
Nash and
Cooperative
equilibria
The central bank chooses a closed loop rule for money
creation in
which each
inflation and
move depends
for budget deficit. Therefore the central bank assumes that the
government will choose the optimal current state feedback rule.
The government
plays
equilibrium strategies
methods since
the closed
the
way.
are computed
The
closed
with dynamic
loop
Nash
programming
loop Nash
(Analytical resolution
(AlB) and (Al9)
equilibrium
is given
is
by
dynamically
equations
consistent
(AlO),
(All),
in the appendix).
Consider now
the open
decentralized authorities
moves. But
same
the central
loop Nash
equilibrium.
Both
strategies.
Hence
each
authority
precommits
to
the
announced by
strategies are
its
computed by
(see analytical
opponent.
means of
resolution in
These
open
loop
optimal control
the appendix).
Nash
methods
With no binding
a single
fiscal policies
policymaker
it weights
implements
both
6,
0,
~,
r,
T,
241
0.6. 0.1
and 0.3
respectively. Each
strategy is
given as
parameter in
within the
order
t~
out in Table I.
2.2.1. General properties
i)
The
relationship
monetary and
between
inherited
inflation
and
both
of the parameters are and for each kind of equilibrium. For the
monetary authority.
high inflation
involves a
contractionary
the public
fis~al
authorities
Whatever
the
value
equilibrium concept
of
the
is used.
parameters
the inherited
and
whichever
Ceteris paribus.
the higher
the inherited
period
1.
authorities to
closed loop
generate a
is higher
public
debt
leads
contrast.
in the
in
period
the monetary
The
authorities plan
the
authority in
reason
the
is
reduction
that
of
their strategy
in
the
closed
debt
of
debt
loop.
both
over
strategy) for
debt is
2.
monetization.
takes
fiscal
equilibrium.
budget surplus
than in
existing
Nash equilibrium
the cooperative
drawn for
the
account
the
period 1
monetary
and 2.
clearly assessed
strategy
the
two
in period
(resp
fiscal
all along
the two
periods.
In open
within the
same period.
So the
behavior
of
each
242
Table ,
Comparative strategies in closed loop Nash equilibrium, in open loop Nash equilibrium and in cooperative equilibrium
benchmark
CLH
OLH
cooperation
CLH
OLH
cooperation
CLH
OLH
cooperation
I
I
lid,
1,1"'-,
l itr,
m,ld,
mid,
-0,372
-0,381
-0,505
-0,106
-0,17 2
-0,072
-0,100
-0,129
-0,059
0,332
0,342
0,307
0,085
0,078
0,059
-0,025
-0,025
-0,507
-0,534
-0,533
-0,504
cooperation
-0,580
-0,580
-0,507
ClH
OLH
-0,277
-0,290
cooperation
-0,5"04
CLH
OLH
l,Id,
ns
-0,187
I
I
I
-0,015
-0,015
-0,111
-0,110
-0,109
-0,072
-0,189
-0,188
-0,111
I
I
I
-0,071
-0,067
-0,072
-0,104
I
I
-0,010
I
I
I
I
I
-0,108
0,361
;,"',"t'
m ,,.,,.,
mi",
-0,248
-0,158
ns
-0,04'
-0,060
-0,142
-0,173
-0,012
-0,040
-0,151
I
I
-0,270
I
I
-0,053
-0,068
0,040
-0,121
0,015
-0,170
ns
-0,344
-0,200
0,015
-0,137
-0,075
-0,083
-0,151
-0,190
-0,031
-0,136
-0,051
ns
-0,137
Table'
ComparatiYe strategies in closed loop Nash equilibrium, In open loop Nash equilibrium and In cooperative equilibrium
, ,Jd,
lid,
1,1"-,
CLH
OLH
-0,372
cooperation
-0,106
-0,100
-O,1J87
-0,172
-0,505
CLH
OLH
-0,361
-0,369
-0,458
-0,100
-0,094
-0,080
'I',
m,Jd,
mid,
m,J....,
ml1fl
0,085
0,078
0,059
-0,248
-0,158
ns
-0,041
-0,060
-0,142
0,094
0,088
0,061
-0,033
-0,032
ns
-0,017
-0,016
-0,076
benchmark
cooperation
CLH
OLH
cooperation
CLH
OLH
cooperation
ClH
OLH
cooperation
-0,381
I
I
-0,382
-0,390
-0,554
-0,367
-0,376
-0,458
-0,377
-0,386
-0,554
I
I
I
-0,112
-0,106
-0,090
-0,104
-0,098
-0,080
-0,109
-0,103
-0,090
-0,072
0,332
-0,059
0,342
-0,104
0,307
ns
changing the weights 01 inflation
<if = 0
-0,050
-0,017
0,350
-0,047
-0,017
0,359
-0,056
0,361
ns
IS = 2
-0,274
-0,115
0,31 B
-0,092
-0,197
0,329
-0,145
ns
0,248
<p = 0
-0,132
-0,060
0,336
-0,089
-0,045
0,346
-0,056
ns
0,361
9= 2
-0,210
-0,084
0,329
-0,168
-0,071
0,339
ns
-0,145
0,248
-0,129
I
I
I
I
I
I
I
I
I
I
I
0,079
0,073
0,064
0,088
0,081
0,061
0,083
0,077
0,061
I
I
I
-0,427
-0,263
0,017
-0,223
-0,132
ns
-0,271
-0,182
ns
I
I
I
-0,052
-0,092
-0,200
-0,029
-0,048
-0,076
-0,052
-0,072
-0,200
243
Table ,
Comparative strategies in closed loop Nash equilibrium, in open loop Nash equilibrium and in cooperative equilibrium
l,td,
lid,
-0,372
-0,381
-0,505
-0,106
-0,100
'i"11',
1,1,1",
m,td,
mid,
m,;....,
ml....,
0,332
-0, 7 2
-0,059
0,342
-0,104
0,307
time preferences
fJ = 0,9
-0,069
0,349
-0,055
0,359
0,085
0,078
0,059
-0,248
-0,158
ns
-0,041
benchmark
CLN
OLN
cooperation
CLN
OLN
CLN
OLN
-0,087
-0,104
-0,098
-0,387
-0,396
-0,112
-0,106
cooperation
-0,492
-0,091
CLN
OLN
0 '590
1--0,590
-0,198
-0,198
-0,120
I
I
-0,158
-0,125
I
I
""=
I
I
U,9
-0,163
-0,123
-0,069
-0,056
0,308
0,317
I
I
0,092
0,085
0,084
0,078
I
I
-0,060
-0,142
-0,239
-0,155
-0,231
-0,147
-0,037
-0,058
-0,041
-0, 5 8
.)= 0,9
----0,470
cooperation
CLN
OLN
-0,344
-0,352
-0,172
-0,129
ns
changing
0 '687
1--0,688
-0,160
-0,159
ns
-0,025 0
-0,025
ns
~=
-0,094
T= 0
ns
ns
-0,050
0,303
0,313
-0, 16 71
-0,138
0,066
0,102
0,047
0,047
0,024
0,024
1
ns
ns
-0,595
-0,358
-0,130
-O,OBl
-0,106
-0,134
Table'
Comparative strategies in closed loop Nash equilibrium, in open loop Nash equilibrium and in cooperative equilibrium
I,td,
lid,
I,"'",
-0,372
-0,381
-0,505
-0,106
-0,100
-0,087
-0,331
-0,33 8
-0,454
-0,089
-0,084
-0,073
-0,172
-0,072
0,332
-0,129
-0,059
0,342
ns
-0 104
307
changing the other parameters
r = 0,9
-0,154
-0,068
0,296
-0,128
-0,059
0,304
n.
-0,105
0,271
m,td,
Ii",
mid,
m,t"lf,
ml",
-0,248
-0,041
-0,060
-0 142
benchmark
CLN
OLN
cooperation
CLN
OLN
cooperation
l'
CLN
OLN
cooperation
CLN
OLN
cooperation
-0,414
-0,425
-0,555
-0,373
-0,382
-0,482
-0,125
-0,117
-0,102
-0,1 ()8
-0,101
-0,095
0,085
0,078
-0,185
-0,130
-0,010
-0,135
-0,090
-0,0'4
059
-0,158
ns
0,070
0,060
0,049
-0,227
-0,150
ns
-0,046
-0,060
-0,141
1,1
-O,a77
-0,059
-0,'02
b =
-0,061
-0,040
-0,067
0,370
0,382
0,344
0,326
0,337
0,324
O,! 01
0,093
0,070
0,083
0,076
0,051
-0,270
-0,157
ns
-0,227
-0,134
0,011
-0,035
-0,060
-0,143
-0,031
-0,049
-0,127
244
The cooperative
equilibrium is
reduction of
the stock
of public
reduction of
inflation in
characterized
debt in
period 2.
period 1
by
and by a
public debt
and of
inflation in
the objectives of
both authorities.
2.2.2.
Changing
the
weights
of
public
debt
and
inflation
- Changing the weights of public debt (r,T)
A variation in the value of the debt parameters (r,T)
has a
in its
own parameter
weight
of
the
public
fUnction, their
own strategies
strategies. For
fiscal
the debt,
lower money
creation. Conversely
the weight
debt
a variation in the
in
their
objective
authorities,
assigned to
the higher
than to
parameter of
modify the
the higher
the
higher
the budget
the
weight
assigned to
the debt,
or T
are nil,
(0,
~)
245
A variation
no significant
closed loop
is.
effects on
of debt
reduction in
Nonetheless
strategies is
matter of
of the
the
variability
a varies
greater when
fact. a
of
monetary
than when
or
fiscal
varies. As a
variation of
the debt
In
weight of
has no
a cooperative
sensitive to
above, this
when
changing the
the debt
weights of
was modified.
inflation than
As was
explained
equilibrium, characterized
by a
tight monetary
policy in the
of the
two authorities
has a
higher time
by the other one. When the fiscal authorities has a higher time
preference for
the present
lower in
periods.
both
creation rises,
adjustment.
return
the central
Conversely
(~is
In
the
bank bearing
when
amount
of
the burden
decreases.
money
monetary
of
the
creation
smoother evolution.
monetary and
Compared to
the money
When a
fiscal policies
the benchmark
creation are
simulation. the
2.
2.3. Debt
evolution in
Open Loop
Equilibrium
The equilibrium
time of
the public
made simulations
strategies shape
debt. As
in order
to rank
246
is always
open loop
fastest in
Nash equilibrium
and, finally,
in the
closed loop
Nash equilibrium.
ii} For
Changing
the
weights
of
public
debt
and
inflation
- Changing the weights of public debt {r,T}
The more
reduction the
both authorities
faster
the
stock
care
of
about
public
public
debt
debt
decreases.
debt
is
low
authorities to
the Treasury
and
this
in
turn
contrains
itself primarily
with its
substantial reduction
public
debt
debt
fiscal
occurs
the
fiscal
target
objectives of
of the
only
and
concerns
stimulus.
through
A
the
stock of
the public
247
Table 2
The comparative evolution of public debt in the open loop Nash
equilibrium, in the closed loop Nash equilibrium and in the
cooperative equilibrium
(
d 2 /d1
d 3!d1
d 2!W1
d 3!W1
(-----------:--------:---------:----------:----------)
(
benchmark
)
(
(
(
CLN
OLN
(cooperation
0,294
0,276
0,187
0,101
0,045
0,040
0,288
0,270
0,179
0,094
0,081
0,037
0,075
0,028
-0,016
0,044)
0,029)
0,021
)
(---------------:----------:----------:----------:----------)
(
changing the weights of inflation
)
(
cr=O)
(
(
CLN
OLN
(cooperation
(
(
-0,016
-0,015
If" ..
(
(
CLN
OLN
(cooperation
0,299
0,280
0,197
0,107.
0,101
0,042
(
(
4I
(
(
CLN
OLN
(cooperation
(
(
(
(
0,296
0,277
0,179
(cooperation
0,292
0,274
0,197
0,153
0,065
-0,024
)
0,091)
0,066)
0,029
)
0,090
0,043
ns
0,099
0,093
0,042
I )
s O )
0,106
0,098
0,037
41
CLN
OLN
ns
-0,017)
-0,016)
0,011)
=2
0,060)
0,045)
0,011)
)
0,060
0,013
-0,024
)
0,029)
0,014)
0,029
)
)
(---------------:---------:----------:----------:----------)
(
changing the weights of public debt
)
(
(
(
CLN
OLN
(cooperation
(
(
(
(
CLN
OLN
(cooperation
0,415
0,414
0,316
0,101
0,047
-0,023
r
0,223
0,211
0,134
0,056
0,053
0,021
(
(
CLN
OLN
(cooperation
(
(
(
(
CLN
OLN
(cooperation
(
0,396
0,395
0,316
0,228
0,214
0,134
0,193
0,193
0,108
0,061
0,058
0,021
T ..
0,103
0,043
-0,023
T -
)
0,080)
0,050)
0,033
)
)
2
0,057
0,020
-0,010
(
(
0,222
0,222
0,108
0,056
0,021
-0,010
0,027)
0,019)
0,015
)
)
0,076)
0,049)
0,033
)
)
)
0,029)
0,020)
0,015
)
248
'ftble 2
d 2 /d l
d 3/d l
d 2 /lTl
d 3/lTl
(-----------------:----------:----------:----------:----------)
(
benchmark
(
CLN
(
OLN
(cooperation
0,294
0,276
0,187
0,101
0,045
0,040
0,075
0,028
-0,016
0,044)
0,029)
0,021)
(-----------------:----------:----------:----------:----------)
(
(
(
CLN
OLN
(
(
)
cc=O,9)
(
(
CLN
OLN
0,303
0,286
0,107
0,101
0,067
0,024
0,040)
0,026)
)
(
(
0,9
-0,014
(cooperation
0,204
0,047
0,402
0,402
0,308
0,204
0,204
0,116
0,021
)
)
(-----------------:----------:----------:----------:----------)
(
)
(
IY=
(
CLN
(
OLN
(cooperation
0,
c O )
-0,025
-0,025
-0,012
-0,030)
-0,030)
0,170)
(
(
(
CLN
OLN
0,265
0,264
0,080
0,079
T ..
0,
0,144
0,073
)
)
0,083)
0,069)
(
(
CLN
(
OLN
(cooperation
(
(
(
CLN
(
OLN
(cooperation
(
0,270
0,257
0,174
0,315
0,290
0,199
0,084
0,080
0,033
O,9)
0,039)
0,029)
0,020
)
)
r=1,1)
0,119
0,082
0,049)
0,111
0,026
0,029)
0,047
-0,017
0,020)
0,067
0,030
-0,016
(
CLN
(
OLN
(cooperation
(
0,299
0,280
0,193
0,108
0,101
0,017
0,092
0,044
-0,026
0,061)
0,046)
0,033
)
)
249
,.,
.'.,,
\,
0,35
.,
0,25
.'.,.,
.'., .,
.,
.'.,
0,2
.'..... ...
....
-'., .....
...... .....
......
.... .........
CL
OL
Figure 2a
0,35
0,25
CL
OL
0,2
Figure 2b
250
(o,~)
251
~
c:
"
l-'
~ t.
I
I
.f II
t
.,.
I
I
.,.
<t
I
t. I
.f
.f
I
I
t, I
'"
i,
+ I
I
.,.
t I
. I
fI
f I
,I
!,
!I
\
~
.+
u:
'"
.f
.f
4-
bO
.j I
::s
"
"
""v....
:l-
,"
~
""- lC.,
.,.""",-
"'""""'-.
.,.
--
.,. -..,. -
~,
\.
,
" "
'''foo..
252
'0
\
I
I
,
I
I
I
I
-\-
..
-\ 0 ,
I
I
'I
..
\
-t
I'
\
.
\..
,oT
1 -\
I
I
\
-\
'
\.
-\-
~~~
~c
"0
r:
"
'V
__~~__--__---- ~----~----~O
\
00
N
'"
N
0,28
0,29
0,30
0,31
.-
- .-- . 5
0,1
0,2
0,]
0,'1
0.5
0.6
0,7
0,8
b=c=O .. S
0.9
b=c=O .. 5
_____,.. a
(eLl
\OL
(OL)
(eL)
(OL)
(eLl
Ben.chmark
c=0 .. 9
_ c = 0 .. 9
. - ---_.-.-..---
-.
....-.-.~.
figure
. + . +. -+-. -+-. -+-.-+-. +.-+-. +. -+- .-+-. +.+ . -+-.+ .+.+ .-f-
.... +
--
0,]2
0.33
d 2 /d 1
254
inflation (31)
will change
(31) the
will not
the level
stock of
of the
public debt
(31) -
higher c i s , the
higher the
be higher
inertia
of
see Figure
increase with
will always
evolution of
will always
whatever the
in equation
it
public debt
benchmark simulation,
(parameter a
in the
inflation
is
the value
of
c.
The
inflation, and
the lower
is the monetization of
the debt.
- The
Specialization of
= 2,
The specialization
fiscal authorities
the public
authorities (0
2,
0,
when
the
T = 0)
of authorities
are mainly
occurs
(T
nil and
in the
benchmark simulation
but this
rate is
both
and
nil,
the
central
bank
reaches its bliss point. The fiscal authorities are thus forced
to achieve
a high
constraint. These
those described
since the
budget surplus
and Wallace
the Treasury
was
possible for
it
necessary for
exogenous
to
achieve
and
in their
positive
such
1981 paper,
it
surplus.
would
It
not
be
would
be
and fiscal
reduction is
(~
255
2.4. Comparing
the private
3 and
constraint that
sector is
Figure 6).
conclusions
parameters but
are wrong
private sector
(8) the
policy
is
which
authorities. This
hold
when
for
value T
correctly
equals
0
large
zero.
range
Without
of
the
anticipated
by
the
fiscal
of debt
one hand.
because of
same size
with or
the high
budget surplus
without private
sector) and.
on the other
cooperative equilibrium.
clear. When
authorities is
the balance
strongly marked
between fiscal
and
monetary
When
policymaker's behavior
and to
or
is to
reduce inflation
sector the
reduction of
are
superior
reduce debt
in the
to
the
second. Without
the debt
unity,
the
private
periods.
Thus the behavior of the private sector significantly
modifies both
monetary
prevailing institutionnal
and
fiscal
strategies
settings. In
whatever
general this
the
behavior
authorities exhibit
256
'!'able 3
The comparative evolution of public debt with and without
private sector
with)
private sector
)
without
private sector
--------------------: -----------------)
d 2 /d 1
d 3 /d 1
d 2 /d 1
d 3 /d 1
(-----------:----------:---------:----------:---------)
(
)
~
( benchmark
0,290
0,096
0,294
0,101)
0,400
0,222
0,400
0,222
0,301
0,301
0,267
0,200
0,055
0,200
0,055
0,103
0,103
0,080
0,415
0,223
0,396
0,228
0,306
0,303
0,271
0,222)
0,056)
0,193)
0,061)
0,109)
0,107)
0,084)
(r
0
( r .., 2
(T
0
(T
2
0,9
( ~ c
0,9
( r '" 0,9
(3
(------------:----------:---------:-------:-------)
(
)
~
0,272
0,091
0,276
)
0,095)
0,400
0,210
0,400
0,210
0,283
0,283
0,254
0,200
0,052
0,200
0,052
0,097
0,097
0,070
0,414
0,211
0,395
0,214
0,288
0,286
0,257
0,222)
0,053)
0,193)
0,058)
0,103)
0,101)
0,080)
0,172
0,034
0,187
0,040)
0,070
0,110
0,041
0,272
0,200
0,272
0,200
0,188
0,159
ns
0,013
ns
0,090
0,018
0,090
0,018
0,041
0,028
0,072
0,106
0,034
0,316
0,134
0,316
0,134
0,204
0,174
0,036)
0,013)
ns
)
0,108)
0,021)
0,108)
0,021)
0,047)
0,033)
( benchmark
( reO
(r
2
( TeO
(T
2
(3
(
0,9
= 0,9
( r '" 0,9
(------------:---------:----------:----------:----------)
(
S
)
)
( benchmark
( I" :.
0,1
( ~ '" 5
("
20
( r '" 0
(r
2
T = 0
T = 2
o
0,9
r '" 0,9
---------------------------------------------)
257
...
-'"
o '"
II ~
II
I
I
I:Q
-t
,:
t
:
..... ,,
+
'
.... :''
+
./.
,il
N
il
./
...
!'
I
I
'!'
t :
.I
..
...
...
..
. I
~ I
.. I
. I
'!I
.1
I!.
It
,
l(
...
\\
:.-
,,
/"
VI
.If\ 1/
/
"'!.
0
VI
"!.
0
Y.\
-\'
I
I
....
"'!.
,
,,
,
/'
,
,
t ,
....... ,,,
+ ,,
,
.... ,,
,
,,,
,
t
I-'
II
II
t-~J.o;.
I-'
.-l
'".
'.
"
'+
'"
.............
...
'+
'
' ............ + ..
VI
VI
.....
.....
c:::
258
present.
Conversely. the
policies by
the public
coordination of
a single
debt. The
authorities will
the central
be much
bank weigh
more effective
if the objectives of
CONCLUSION
During the
1970s. most
combined expansionary
policy. While
limited the
fiscal policy
central banks
monetization of
adopted increasingly
the economy.
of the
developed countries
stock exceeds
reduction of
orthodoxy.
the public
However,
implemented by
common goal,
These
the
two
fiscal
the
and fiscal
policies
are
the
authorities
public
debt.
share
yet
they
goal through
reach this
GNP. Hence
a monetary
and
decentralized
reduction
the way
of the
debt requires
monetary
two distinct
the Treasury.
disagree on
the growth
easily
theoretic framework.
conflict beween
be
analyzed
monetary and
within
fiscal
dynamic
game
adds
macroeconomic
objective function
strategies
private sector
are
of the
constraint
and
target
policymakers. Monetary
significantly
modified
and
the
and
in
the
fiscal
adaptive
259
anticipatory behavior
of the
debt reduction.
APPENDIX
We present
closed loop
game in
Nash and
appendix the
The
analogous to
derivation of
the
complete information
Government.
with a
in this
derivation
the derivation
single controller.
cooperative
equilibrium
is
authorities neglect
the
influence of
private sector
the relevant
equilibria are
computed with
o.~.a,b
and c equal
to zero.
1. The closed loop Nash equilibrium
The indirect
loss functions
of monetary
and fiscal
1/2/ Min(f2 + rd 2 +
f
~U2
) +
the constraints
(1) and
1.2
~yF(dt+l.Ut+l).
t
subject to
1.2
terminal conditions:
(A3a) d 1 U1 are given
(A3b) 6yM(d 3 .u 3 )/6d 3
Td 3
rd 3
ou 3
~U3
The closed
recursion.
loop
Nash
equilibrium
is
computed
by
backward
260
minimum are
0
mt +a[(6V M (d t + i , rrt+i)/6dt+i,)(6dt+i/6mt)]
+ a[ (6V M (d t + i , rr t + i ) /6rr t + i ) (6rr t +i /6m t )]
(AS) 6V F (d t ,rr t )/6f t
ft +
+
~[(6VF(dt+i,rrt+i)/6dt+i)(6dt+i/6ft)]
a(Td 3
(A7) f",
acrr 3 )
+ brr3)
~(rd3
Given
the
inflation, we have
of
motion
of
public
f '"
= -
and
- aaac, M3
Fo
debt
(AS) m",
(A9)
laws
~rr,
F,
~(r-cb),
F",
= -
~ab,
F3
l+a(T+ac"'),
+
~(r+b"')
mC
. '"
(All) fC
'"
with
Ho
(FoM3+F,Mo)/(F3M3-F,M,)
H",
(F,M",+F",M3)/(F3M3-F,M,)
Go
(MoF3+M,Fo)/(F3M3-F,M,)
G",
(M,F",+M",F3)/(F3M3-F,M,)
261
Using these
strategies,
we
derive
the
motion
of
the FOGs
for a
Xod z + Xzrrz
(AIS) f1
Yod z + Yzrr z
with
Xo
Yo
-~(r+~r(r-Go)(r+Ho-Go)
Xz
a(aT(r+Ho)(Hz-G z ) - co - aoc(a+bHz)(a+bHz+cG z
Yz
-~(~r(r-Go)(Hz-Gz)
c~
~~b(a+cGz)(bHo+cGo
~~b(a+cGz)(a+bHz+cGz
the strategies :
(AI7)f 1
= (P o d 1 + P 1 f 1 + P Z rr1)/P 3
= (Ood 1 + 01f1 + OZrr 1 )/03
with Po
(AI6) m1
01
rX O ,P 1
Yo-cYz,Oz
Xo+bXz,P z
aY Z ,03
We express
aX Z ,P3
I-Yo-bY z
the closed
262
fe
(A19)
with
Ro
R2
So
S2
(OoP3+O.PO)/(03P3-0.P.)
(O.P2+02P3)/(03P3-0.P.)
(P003+p.OO)/(03P3-0.P.)
(P.02+P203)/(03P3-0.P.)
Using theses
(A21)
Ue
2
open
loop
equilibrium
Nash
of two
simultaneous solution
strategies
optimal
control
derive
problems.
from
The
t,
t,
t,
+ at,+'[pH
t,
+ qH
(rdt,+ft,~mt,-dt,+l)
t+l
(A23) HF
t
~t,/2(f2
~u2+rd2}
t
(rdt,+ft,-mt,-dt,+,) + qF
~t,+l[pF
t+l
where pHt,
t+l
pFt,)
(resp
variables which
(aut,+bft,+cmt,-u t +.)]
t+l
et
qHt,
the central
(resp
qFt,)
are
the
costate
FOGs for
a minimum
in period
written
mt
a(pH
t .... t
- cqH
t .... 1
may
be
263
(A2S) 6H F /6ft. = ft. +
t.
~(pF
t. ... 1
+ bqF
t. ... 1
(A26) 6H M/ 6d t.
t.
Tdt. + ocrpM
t. ... 1
pM
t.
(A27) 6H F /6dt.
t.
rdt. +
pF
t.
(A28) 6H M/ 6lT t.
t.
OlTt. + ocaqM
(A29) 6H F / 6lT t.
t.
lTt. +
~rpF
t. ... 1
=qM
t"'1
~aqF
t. ... 1
t.
qF
t.
(A30c) pF
3
(A30d) qM
3
(A30e) qF
3
rd oo
OlToo
lT oo
(A32) f O
,..
Hence
(A33) dO = (r+Ho_Go)d,.. + (H,..-G,..)lT,..
3
264
Knowing
mz
and
fz,
we
determine
the
costate
variables :
(A3S) pM
z
Uod z + Uz 1l z
(A36) pF
z
Wod z + Wz 1l z
(A37) qM
z
Vod z + Vz 1T z
(A38) qF
z
zod z + Zz1lz
with
Uo
T(l+ar(r+H o -G o
Wo
r(l+~r(r+Ho-Go
Vo
o(aa(bHo+cG o
Zo
(~a(bHo+cGo
Uz
T(ar(Hz-G z ) )
Wz
r(~r(H2-GZ
Vz
o(l+aa(a+bH z +cG z
Zz
(l+~a(a+bHz+cGz
mt
(A40) f
FOGs (A24)
11.
(Kodt+Ktft+Kz1T.)/K3
(Lodt+Ltmt+Lz1Tt)/L3
with Ko
ar(Uo-cV o ) ,K t
a(Uo+bUz-c(Vo+bV z ,
Kz
aa (Uz-cV z ) . K3
Lo
-~r(Wo+bZo)
Lz
-~a
,L t
~(Wo-cWz+b(Zo-cZz'
(Wz+bZ z ) ,L3
l+~(Wo+bZo+b(Wz+bZz
We finally get
in
265
(A42) f O
and
(A43) dO
2
REFERENCES
Alesina, A. and G. Tabellini (1986), "Rules and Discretion with
non Coordinated Monetary and Fiscal Policies" , Mimeo,
University of California
Anderson, T.M. and F. Schneider (1986), "Coordination of Fiscal
and
Monetary
Policy
under
Different
Institutional
A.
(1986),
"Les
anticipations
rationnelles
dans
Working Paper n.
1986-12,
and P.
Hichel(1984),
Precommitment. I
: An
"Toward a
Analysis of
Theory
of
Optimal
the Time-Consistent
266
Fisher,
S.
(1986),
"Time
Consistent
Monetary
and
Fiscal
Kydland, F.
Optimization and
Economic and
the
Stabilization
Assignment
Policies
Problem",
Social Measurement,
Annals
of
261
Kydland, F and E. Prescott (1977), "Rules rather than Di scret ion
the
Inconsistency
of
Optimal
,vol.
in
an
Economy
Oudiz.
G.
and
J.
Buiter et
of
Capital",
Journal
of
(1984) ,
Sachs
Coordination in
Journal
without
Plans",
"International
R.C. Marston,
Policy
Models",
in W.H.
International Economic
Policy
Dynamic Macroeconomic
(1976),
Policy
"The
Cost of
Formulation",
Conflicting
Annals
of
Objectives
Economic
and
in
Social
Federal
Reserve
G.
( 1986a) ,
Monetization of
"Central
Deficits;
Bank
of
Minneapolis
1-17
Bank
Reputation
the 1981
Italian
and
the
Monetary
(1986b),
"Money
Debt and
Deficits in a Dynamic
(1977).
267
Turnovsky, S. and Y.A.
Brock
Optimal
Foresight
(1980),
Government
"Time
Consistency
Policies
Equilibrium",
Journal
in
and
Perfect
of
Public
1. Introduction
Coordination of actions of several policy makers is
fraught with difficulties of theoretical and practical nature.
To understand the difficulties,
it helps to note that policy
coordination problems are in essence control problems with
multi decision-makers
under imperfect
and
non-identical
information patterns. Since policy makers in several countries
must act
on information
sets which are not identical,
incomplete and imperfect, pooling or exchange of information
may be desirable, were it not for the cost and delays in
gathering, exchanging and processing information, or incentives
for transmitting false information. Decentralized controls can
sidestep some of the costs and delays involved in information
exchange since they are control actions based mostly on locally
available information, possibly supplemented by information
supplied from outside on a few items. (1)
Decentralized coordination
schemes for
a small
country are easier to implement since policy decisions of small
open economies can be based on information set containing
mostly locally available variables and some "key" international
variables which define the international economic environment
which the small open economy faces, and thus they avoid many of
* I thank the participants in the International Macroeconomics
Workshop at the Center for Economic Policy Research, London,
and Marcus Miller in particular, for their discussions and
comments on an earlier version.
(1) Seminal
attempts at
the coordination
question
in
international economiCS via decentralized control are Aoki
(1976) based on Aoki (1972). Alternative approaches to policy
coordination via non-interacting control are discussed in Aoki
(1974). A connection with Mundell's assigment problem and its
dynamic generalization is discussed in Aoki (1976) as well.
270
the difficulties
associated
with
implementation
aspects
of
policy coordination.
with its
it can
own policy
into account.
small open
economies policy
economies but
effects of
In such
also affect
however, only
if decomposition
interactions, and
of policy
external shocks
objectives, dynamic
are realized
in the
sense
To
demonstrate
decentralized policy
simply
schemes, the
the
possibility.
paper adopts
the
of
average-
effects in
a benchmark
model composed
of
three
followed by
section lays
out the
policy objectives
that of
is
dynamics in
framework for
examined
Section 3.
analyzing
deviations
in
This
of
is a
three real
brief account
Theoretically, when
three
elasticities,
when three
for the
of time
size of
economy with
the real
countries
~NP
have
identical
set
of
different scales),
then the
average
deviations
economic
decomposition of
period of
characteristics
are
dynamic interactions
1965 through
not
the
same,
roughly holds
for
this
the
5.
2. Decomposition of the Performance Index
Let Yi
be the
271
the presentation.
we carry
r13Y3 where
in a
constant
expressed at
common
currency.
(1982) constant
Deviation of
denoted by
rlJ is
output from
Y1
is
dollar. called
some
reference
the
US
GNP
or
trend
path
is
trend (reference)
path is
given by
6Y 1 +
r 12 6Y 2 +
r 13 6Y 3 +
6r 12 Y2 + 6r 13 Y3 .
Define a lower case letter Y. as the relative change
of real GNP. 6Y , /Y , i = w. 1.2.3. Then
(1)
where
Yw = alYl+a2Y2+a3Y3+(6rI2/rI2)a2+(6r13/rI3)a3
are positive
shares of
the real
GNP. This
relation
can
be
rewritten as
'-1
are
smaller
a.Yl'
In
than
the
other
terms.
We
assume
that
the
sum of the first three terms in (1). we must carry these in our
analysis. For
Later in
GNP of
simplicity. we
Section 4.
deviations of
InYue.lnY g and InY J . from a common trend time path are regarded
as Yue.Ys and YJ. Note that 61nY.
3
Ya=~
i=l
( 3)
= 6Y,/Y,.
g.1.2.3. Define
yt =Yi - Yw '
=1,2.3.
272
Other d-vGriables
are analogously
y?=(yj+Yw)2=(Yw)2+2yw yj+(yj)2,
we see that
Thus. the
var Yi
expected value
country's output
of a
= 1,2,3.
y~
S ,
(4 )
L?=l ai var Yi =var Yw + L:'l ai var yt
because the cross product term vanishes because of (2).
individual
objectives of
country's
the three
variances
output
is
the
policy
minimizing the
left side
objectives. minimization
of minimizing
(2). Eq.
as "one
of var yd i
the average
(4) may
optimize the
natural to
second term
assign the
in (4)
role of
subject to (2)(2).
controlling y.
It may be
to the country
with the largest weight, country 1, say. The reason for this is
explained below.
nature of
assignment
dynamic interactions.
country problem,
rewrite
Such an
Eq.
~31_1alydi=O,
well-known in
(4)
by
using
is
independent
of
the
rate literature,
definitional
relation
as
+ a 3 [ 1 + a 3 / a 1J var Y ~ + 2 a 2a 3 / a 2 E (y ~ ~) .
The cross-product
reasons
a2a3/a1
term may
be
negligible
for
two
273
or nearly
so,
countries 2
unless
there
(As we
and 3.
is
global
will show
shock
common
to
then a 2 a 3 /al
above suggests
the largest
and
.046.
a
decompositi~n
country in
countries
var yd 3
a3
,2 and a
that a2
.15.
illustration, suppose
.65.
and 3
individually
dynamics are
a-variables affect
shall show
recursive,
above assignment,
is stated as
Min E (~ajY?)
dj
where
'
i=1,2,3
=Min
dw
The conditional
if the
[varyw
+ Min
J]
dj
i=1,2
d-disturbances are
uncorrelated.
coordination assumes
countries 2
and 3
the form
take Yw
In
form,
of Stackelberg
as given
the
game,
and jointly
in
policy
which
minimizes J
f5,
II
L aj
j=1
One of
control yd n
conflicts in
yl= o.
the n-countries,
but take
anyd n
coordination of
11-1
country
-.L ajyf
,=1
say,
must
not
as given,
if
274
3. Decomposition of dynamics
The decomposition
benchmark cases
and d-variables
in the
sense
In
conversely.
in which
a-variables
these
benchmark
affect
d-variables
cases,
but
decentralization
not
of
to be
several models
have the
decoupled from
by Aoki
same set
size,
recursive:
the
the
paper
shows
a-dynamics
that
affect
dynamics
the
the
d-dynamics
become
but
not
conversely (3).
Here to
of three
coun~ries
in real
interest rates
and"due
to
sluggish
adjustments
of
is a
complicate the
as wealth
changes and
decoupled a-
and d-dynamics.
regarded merely
the
as a
(recursive)
decentralization of
The models
used here
should be
structure
in
supporting
(3) Models
in which countries are alike in the sense of
possessing the same values for the structural parameters
(elasticities) but of unequal size are analyzed in Laffargue
(1985).
275
Model Dynamics
The model
of this
paper is
specified in log-linear
(7)
sj/c (t
+ lIt)-sjk
and
( 10)
Pj (t+l)
Time index
which is
omitted if
standard LM
Its price
in (9)
is PJ.
may be
from
although the
this
case
under the
the j
of each
Eq.
country price
the expected
Sii
is done in Aoki
country consume
procedure for
since it
the a-d
(7)
(6)
is the
goods. We
complicates
algebras,
interest parity
unit of
0 for
1986a. b)
all three
capital mobility
of the
(1985.
is the
expected value
is a
argument
Eq.
assumed perfect
(8)
or
is likely.
this refinement
logarithm. Thus
as subscript
no ambiguity
preferable as
as well.
denotes the
value.
is denoted
curve. Country
since residents
refrain
j, k == 1,2,3.
country k's
1.2.3. The
formed at
of
relation
currency in
argument
the
next
t+ilt
period
output good
price adjusts
11j .
slowly.
Finally.
i. e ..
( 10)
indicates that
6 is sma 11.
Here the
276
3.1. a-Hodel
imply that the a-variables are governed
by a
set of
appears.
In
equations in
taking the
variables average
details.
Note
The a-model
out
no
exchange
weighted average
to
also that
become
out in
model
Aoki
rate
variable
zero.
the a-model
is therefore
originally pointed
which
Appendix
gives
the
contains no d-variables.
of
closed
economy
as
is specified
by
or
where
weighted average of
and where ~ is a
~ij
and
Reduced Form
Substitute (12)
where
277
where
are not
V~t=O
is assumed,
observed at
We drop
a,
and
to
lighten
notation
in
this
1t =
y (m - e) - 'YP + lCT\
where
and
( 20 )
=[ - (1 -
05)(m - e - P ) + 11211 ] I D
a-dynamics
Eq.
(14),
(17) and
(19)
yield
the
price
dynamic
equation as
(21)
and the
278
Monetary Rules
This section
examines how
by feedback rules on m.
m =Ki
i
where DK
'"
K(l-crcS)/D~.
for the
=[ (1 -
as) (p + e) + J..Li'lll DK
=[ - ex (p + E) + ~ T\ ]B
where
( 23 )
=- ex (p + e) + ~T\ .
Pt+l
where
p=l-cxa
Eq.
(24)
= PPt + nt
and (23)
=( - a e + ~T\) B+ v.
imply that
under this
feedback rule.
the
i.e .. y,a
Note that
is
ARMA (1.1).
Note that
<
<
1. unless crcS = O.
279
and with opposite signs. The condition -l<p<l
is equivalent
to
0<
Os
DK
After we
a- and
=(cr~-2D)/2(1-a~).
<2, or K >Kc. Kc
develop the
d-variables, we
the "effective"
show that
exchange rate
of country
j and
pf
through a
vector proportional to
The sign
of the
complicated function
in the
derivative is positive.
the
is thus
cr~=O
avary/aK
derivative
derivative
is
negative.
Thus,
by
but
continuity
rate with
disturbances dominate
feedback
are in
O.
the LM
not be
~hould
interest rate,
K >
the case
disturbances
used to
i.e., K
In
the
stabilize the
where
the
IS
interest rate
world
average
It is
interesting to
random walk
for the
completely from
unequal in
observe that
{P:J
the rule
produces
the d-dynamics
even when
the
countries
are
size. Under this rule we can see from (22) that the
the effect
of
Comparing this
vIa
now
~~+1
with the
corresponding
expression
under
.. -
the
280
cr;
and
This inequality
than cr~
confirms our
cri
is larger
p rule.
3.2. d-Hodel
Define the
d-variables as
the deviations
Yj - ya and pd J
from
the
pj _ pB
~jajydj
obtain
The difference
The result is
of (8)
and (12)
281
d +s ) - -I;yd. + -9 .y a +".d
,. -1t.d) + d (
_po
Y)d =-(}" (.d
))
)).))
'I)
or
(28)
where the
9j = ~ ( 1-3aj) / aj ,
division by
and -
denotes
(l+~).
forwardly obtained:
(29)
Pj(t+l)=Pj(t)+8yj(t)+vl-
(30)
Note that
are decoupled
of country
k,k
the d-model
for the
individual countries
disturbances. When
the a-dynamics
3 countries
do not
are about
affect the
9j
d-dynamics, because
is
nearly zero for each j in (28). To give some idea, suppose that
~.
O. 1, a z
a3
.3
and a
.4,
then
9z
93
.03 and
9 1 = - .05.
93
regarded then
as a
small open
interest
rate,
economy can
etc.,
as
be analyzed
exogenous.
The
just
of
monetary
Reduced Form
Substitute (28)
282
1 =(1 - 08)! D,
where
g =1
=J.l2! D,
-nd,
d-dynamics
Let
From (27),
(30),
(31)
and (32)
where
=g ,
a22 = 1- <X;.
cp = (8Ill! J.l2)n
a 12
=BIll' D
Two facts are noteworthy about (33). The dynamics for country j
do not
depend on
the a-dynamics.
Zkt,k
283
Monetary Rule
Eq.
(34)
makes
clear
that
so
long
of
md J
country
is
determined as
a function
of the
policy makers
of country
of the
d-variables
as
j,
country j
are jointly
affected by
mdJ and
m j d =K.dR
Ij
ule
From (31), the md
Kid determines i d to be
because
and
c + KYn t (1 + KI) =[
~:1'
We
284
(35)
where
where
DK =D (1 +Kl)
and where
D 1 =<5(J..ll+K)dID K
D2 = 1- <5
and
now give
interactions between
P~t
Zjt.
Eq.
(21).
be written jointly as
(34)
where
[1-1
=
-X~2 A
-x83 0
va
I
nIa --
0
0
HG=H
0
Y
x83 0
+[ ;'c]
83 c
J..llTi,a / D
~]
285
Eg.
(34)
shows
variables.
In
that
can
affect
only
the
j-country
The monetary variable rna can, however, affect all the variables
if 6:j= 0.
(If
The monetary
sense that
rule for
it affects
the model
rna is
more important
in the
md J , however,
the
where
and the
vector
since x
disturbance
~c/D,
~~
impinge
Effects of disturbance
Under the
p-t,F
rule mat
complete dynamic
on pa,Z't
is modified
to
diag(l,A,A}
and
286
respectively
Dynamic Structure
Dynamic multiplier matrices summarize all information
regarding how
exogenous
disturbances
informative. Eq.
(22),
transfer function
outputs to
~Q
y,;i
and monetary
can be
1,2,3
and
are
more
(dynamic multiplier)
the real
response to
affect
(32)
matrix connecting
shocks.
obtained from
For
example,
the
the
"
yQ =
rl-
(0"1 D)(1
where
The impact
shift in
multiplier is
T1Q
(~l/D).
Effects
of
-y(1-y)'
once-for-all
, t
0,1 .... ;
~ll
D + OLld I
Dl9] ~.
where
This becomes
where
The transfer
287
obtained. Putting these expressions together. we obtain
y"a
={[
00]
[o][lOOl}
g4 + g:zg3 e
g3
g4
where
and
In both
cases we
variables while
shows that
note that
11- and
m-
e"
~d and
md,
affects yB
affects only
and yd' S
Eq.
yd
(37)
component responding
transfer function
to
are relevant
either to pB or
transfer function
transfer function
or to pB
components responding
of the
Z3,
to
have
Z2
non-zero
element
of
the
to our
next section.
These transfer
functions can
be equivalently stated
288
dynamic structure
lost. This
With e
so clearly
is one
0
of the
exhibited in
Unequal country
shocks and
(37) are
(three countries
summarized by
(36) and
sizes add
glgze or
monetary shocks,
respectively. When
6 is
zero in
and
have unstable
function
g4 thus
gz and
characteristic of rational expectation models.
The
roots,
- 1986.2
dollars), West
Germany and
constant deutsche
models for
of the
mark and
random trends
(the
latter
respectively),
two
to
at
1980
construct
currency with
289
where Yw
6Y w/Y"",yu
analogously defined.
Ys and Yj are
regarded as
deviations
of
the
fixed. One
logarithms
can
of
interpret
the
Yu
etc.
corresponding
as
level
+ as
since au
dropped as
+ aj
being small
construct models
in two
compared
with
steps. First,
the
first
we build
three.
a model
We
for
"trend" component
y,,]
In
[ In Yg
In y.J
where
=C't,j + Y,
,
represent
trend dynamics
a "trend"
reference time
is of the form
't'+1 =A't,
+ short-run
Yt+l
=HZt+l + et+l
=H(Fzt +Get)+et+i
= HFH+(yt - e,) + HGet + et+i
290
where
In the
then, SHFH+S- l
a-d framework
and aJ
( 38 )
.64,a s
.14
.23 we obtained
SHFH+S-l =
With the
- .796
.63,
shares au
.12 and
as
aJ
becomes
(39) [
- .063l
.792 -.071
-.800
.189 -.265
-.294 -.217
.702
The (1,2)
and (1,3)
effects of
elements
yd's ori
compared with
y- and
the other
in
both
not equal
cases
to
represent
zero.
but
the
"small"
in connection
with (37).
the non-zero
entries in
variables but
to the
countries. For
modeling
further details
method
used
in
on
this
state
paper
space
are
time
found
series
in
Aoki
(1987a,b,c).
There are
weighted average
not be
exactly decoupled
real GNP
291
the relative
sizes of
Second, elasticities
Germany and
Japan are
the three
of the
countries are
the GNP
rates. However.
the idealized
it is
of these
countries
grew
at
different
theoretical model
is rather
close to empirical
paper
provides
informationally decentralized
model by
economy
applying the
macroeconomic
demonstrates a
well as
framework
policy rules
for
examining
in a multi-country
average-difference formulation
models.
In
particular.
the
to open
paper
set of elasticities.
The quarterly
three
country
real GNP
empirical
decomposition (decoupling)
dynamics is
data are
models.
Although
no
perfect
countries.
REFERENCES
Aoki. H.
(1972) -
163-173.
292
Aoki. H.
Aoki. H.
Aoki. H.
Aoki. H.
Aoki. H.
293
Aoki, H.
(1986b) -
and Coordinated
Monetary Accomodation",
Econ.
Study
Aoki, H.
(1987a)
State
Space
Modeling
of
Time
Series,
Springer Verlag.
Aoki, H.
(1987b) -
"How
to
Build
State
Space
Models
for
Aoki,
(1987c)
H.
"Time
Interdependence of
presented at
Series
Evidence
the USA,
the 1987
Far
of
Real
GNP
Meeting
of
the
Buiter,
V.
(1986)
"Macroeconomic
Interdependent World
Policy
Economy:
Design
An Analysis
in
an
of Three
Buiter, V.
and
H.H.
Hiller
Overshooting and
(1982)
the Output
"Real
Lost
of
Exchange
Rate
Bringing
Down
Fukuda, S.
of
Rules
of
Monetary
Intervention", Mimeographed.
Coordination
and
294
Modeling
of
Time
Series
with
Trends
and
APPENDIX
>
countries. Define
where the
superscript a
denote the
weighted average
of the
is zero because
sJ
0 and IJaJs Jk ~ IJtkajSjk = 0 because
Similarly the right hand side is iA - in = O.
In (8),
we assume
that d Jk
r.j
aj
r.bj
=d
r.j
=d
r.j
=d
r.j
=dpa -a
=0.
r..J
-a
=d
ak
SJk
for all
=-
SkJ.
j.
The
295
Since the
elasticity
S'J represents
the
effects
of
output
changes in country j
on that of country i.
(3) we
assume that
size aJ/a"
i. e.,
Then
I A. 1 )
(3yD _y D)
=2SyD
Noting IA.l)
take the
(8) and
(12).
j. the
difference of
Then
=-
2S yD
- TlD
2~ yD.
Using the
posited relation
IA.2)
=d(-Pj+s),
The
fourth
posited relation
term
becomes,
upon
substituting
the
296
E"j ~j" y" - 2~ ya
~j" -~) ya
=-~Yf+ej ya
where
where Iakyd k
1/3.
For the
payments
of
OECD
disequilibria
and
countries. the
to 2
GDP in
of
last fifteen
years. the
countries
large
has
scale
current balance of
been
marked
variations.
current account.
Thus.
by
deep
for
most
1985. Furthermore.
the divergence
of
in the direction of
striking contrast
States and
on the
between on
other hand,
the one
Indeed, there
hand, the
Germany and
United
of their GNP.
These developments
have been
different ways.
Price-effects have
the
terms
guise
of
specifically to
the two
the dollar-shock(s).
approach to
as the
account developments.
between fiscal
trade
fluctuations.
referred to in
This
refers
A second
the balance
monetary impulses
the other
of
often been
of payments.
lays
main explanatory
However, in
the
emphasis
factor of
recent years,
on
current
the contrast
U.S.
(BOURGUINAT,
1987
MARRIS,
1987)
are
good
paper we
shall estimate
the floating-dollar
a current
period (1973-1986)
account
for the
298
Graph n' I : Structural budget surplus as a deviation from average
o United States
.,
.J
-Japan
2.5
2
15
1
Po
..
....
.5
0
- .5
JI
-1
-1.5
-2
-2.5
-3
-3_~1
70
74
72
.,
-'
76
\~
78
80
84
82
86
88
OFrance
Germany
Q.
..
8....
0
0
-I
-2
-3
-4l70
6
7'2
76
74
78
80
D \taty
() United Kingdom
2
/
0
Q.
.
Q
....0
60
/I..
'~
-2
'0
-4J
-6
-10
~, "
-12
70
7'2
-8
86
o Canada
84
82
74
i-----'
'\.--
76
78
80
82
'a........ ~
84
Source; computed from PRICE and MULLER (1984) and OI.C.D. (1987a)
86
86
299
isolate
contribution
the
determination
of
contribution of
these
fiscal
of
external
the other
impulses
accounts
the
in
compared
to
the
forward.
On a
shown that
theoretical level.
the relevant
budget deficit
VINALS,
impulse
instance:
of
is
empirical point
measured by
with conjonctural
using the
studies
the
have
anticipated
can be
fiscal
(see for
a number
1985 ;
1987).
impulse
an anticipated
(see FELSTEIN,
the best
is
However, such
to accept
the existence
acceptable predictor
that agents'
foresight.
To
get a more
the determinants
a reaction
of perfect
function of
fiscal policy
the influence
of future
anticipated fiscal
afterwards be
budget surplus
used as
account equation
an explanatory
alongside
the
thus
forecasted
will
price
and
monetary
impulse
variables.
Contrary to
confine our
many works
study to
variables. Thus,
we
deviation between
these impulses
the only
will
in this
influence of
take
into
these domestic
for the
main
OECD
account
impulses and
countries,
domestic
not
fiscal
only
the
the average of
but
also
the
300
1. FUTURE
STRUCTURAL
BUDGET
SURPLUSES
AND
CURRENT
BALANCE
First. we
surplus (SBn t
for the
shall assume
as a
+ 1
year t+l
this same
structural
budget
year. Along
current account
that the
(CA. as a
(CUT). The
the increase
in the
For all
monetary impulse
monetary base.
these variables.
compared to
the average
compared to
one of
we will
of the
will be measured by
use
seven main
these countries.
their
deviation
OECO countries or
for the seven main OECO countries. using ordinary least squares
are given
in Table
those estimates
1. The
will be
main conclusions
presented first.
to be drawn from
We will
then
give
Italy on
direction of
country
the other.
this influence
considered.
deficit induces
Thus
a current
varies a
an
the
lot
anticipated
account deficit
size
according
structural
in a
one
and
the
to
the
budget
to half
a deviation
spillover
France. Tnis
negative. This
that the
However.
positively on
spill-over amounts
sign difference
the current
to 44
can be
of such
account in
in Italy. where it is
explained by
the fact
future fiscal
301
United States
CA = 0,:>3:>. 0,727 SBuS + 1 0,162 TDEus - 0,081 TDEj - 0,03 ClITusj - 0,138 TEMPS
(2,67)
<3,63)
(-,B)
H,50)
(-2,38)
R2 .. 0,928
DW ,,2.15
F = 20,S
SCR = 1.89
Japan
CA =0,006 - 0,077 TDEus 0,087 TDEj - 0,326 SBusall.l 0,04 MGROusj 0.163 MGROus
(-1.8'0
(<(,32)
(<(,5:
(2,31)
(2.10)
R2 =0,956
DW .. 2,23
F .. 3,9
SCR .. 1.62
Germany
CA =2,989 - 0,609 SBus71 - 0,080{ ClITital- 0,363 SBukl 0,09 TDEa11- 0,06 MGROjall
(-2.18)
(-2,20{)
H,63)
(2,58)
0,83)
France
- O,075TDEuk
0,27)
R2 =0,93
DW =2,07
F =15,6
SCR .. 2,02
CA .. 1.39 - 0,235 ClITfall- 0.17 MGROfa11 + 0,60{3 SBus+ 1 +0.185 SBf71 - 0.133 MGROusf
(-6,37)
(-.12)
(5,25)
(<(.15)
(-3.00()
R2 .. 0.92
DW -1.76
F -18.5
SCR - 1.16
United l:insdom
CA .. - 0,555 1,009 SBus71 O.99SBukl 0.073 ClIT usuk 0,256 SBa.1171
(5,0)
(5,75)
0,86)
0.87)
R2 .. 0.910{
DW =2.68
F .. 23,8
SCR "3.63
Italy
CA .. - 7.20{ - 1.33 SBpl 0,961 SBa1!71 1.52 SBf7 1 - 0.57 SBuk71 - 0.5 SBit7-l
(-H2)
(2.63)
(3.58)
(-2.52)
H.7)
0,076 TDEit 0.152 TDEal!
(2.04)
(2,81)
R2 =0,927 DW" 2.56 F" 10,87
SCR" 3.39
Canada
CA =- 1.778 - 1.38 SBus2 - 0,13 TDus O,O9TDEusca-l - 0.371 SBus7.1
(-5.77)
(-3,9)
(2,68)
0,81)
R2 .. 0,865 DW .. 1,87
F IH
SCR .. 2.33
302
year t+2.
(1 % of GOP)
influences positively the Canadian
current account (for 1.4 % of GOP). Thus a permanent American
budget deficit influences positively the Canadian current
account surplus,
An American budget deficit amounting to 1 % of GOP is
associated with a French current account deficit representing
0.6 %
of GOP.
the deviation
303
influence
differs
rather
widely
between
of the
American and
one
country
and
another.
The growth
Japanese
terms
of
a 10
increase
in American
terms
of
trade
improves this
worsened by
contrast, the
~.
This surplus is
~.
By
10
rise
associated with
a 1
country, but
with a
Moreover,
Italy, a
in
in
the
German
current
1.5
In Germany,
this influence
an impact
trade
is
account
deficit.
of
account surplus
Italian current
10
terms
of the
of GOP.
of the
terms
of
trade
labor costs.
The latter
matters only
Indeed, when
the growth
in unit
in France
and Britain.
labor costs is 1
higher in
of
GOP.
In
differential with
only by 0.07
the U.S.,
impulaeB.
similar
deteriorates the
unit
labor
cost
current account
current account
Japanese current
rate of
of GOP.
~~ry
determine the
Britain,
balance is
growth of
the U.S.
of three
countries. Thus,
the
~he
monetary
growth differential
growth in
the U.S.
monetary
of GOP.
By contrast a 1
and the
of
GOP.
For France
differential is
higher in
very contrasted
France than
in Germany.
when
monetary
growth
is
304
1.2. The U.S. case
To explain completely the U.S. current account we had
to introduce
a time
variable. The
latter's
contribution
is
account balance
% of GNP.
this phenomenon
as the
quality of
1976 and
1981 and
improves the
precision of
the
satisfactory.
Omissions" (N.E.O.)
explain this.
importance
in the
of
"Net
Errors
and
Indeed. N.E.O.
U.S. GNP
the large
in 1981
The
(when a
deficit
is
forecasted
while
the
fact that
estimate a
current account
thus
(CA-N.E.O.). the
quality of
the regression
improves.
shrink considerably
from 1981
to 1983
corrected
Errors
factors
coefficient of
value)
main
differences
(Table
the time
2.
trend is
are
noticeable
equation
much
n03).
higher
in
First.
(in
the
the
absolute
in place
305
CA
(1)
SCR
3,22
0.535 + 0.727 SBus. 1 + 0.162 TDEus - 0.081 TDEj - 0,034 CUTusj - 0.138 TEMPS (2)
(2.67)
<3.63)
(-.(.13)
(-1.50)
(-2.38)
R2 = 0.928
DW = 2.15
F = 20.5
SCR = 1.89
306
~: ---~~\---------
Po.
-1
(!)
......
0
...c::
'-'
a)
;::)
b\~
-1.5
u
u
...c::
..s
-2
4)
-2.5
;::)
-3
''""''
-3.5+-~-""----"--r---'--r--~-r-~-""--..---.-~--..-~-..
72
74
1.5
;;::;
z
(!)
b)
76
78
80
82
84
86
as
o U.S. observed
U.S.adjusted
.5
......
0
-.5
...c::
;::)
-)
u
u
-1.5
4)
-2
-2.5
......sc::
'"'
8'"'
-3
.~
-3.5
72
;;::;
z
(!)
74
76
78
80
82
84
86
88
o U.S. observed
U.S.adjusted
c.:..
0
~
....CI
'-'
;::)
c)
u
u
..s
....CI
4)
''""''
u
;::)
-1
-2
-3
"0
Su
-4
''""''
-s
4)
72
74
76
78
80
82
84
86
88
307
of the
level of
the American
these
two
amounting
to
coefficient of
impulses
induces
0.25
of
GNP.
current
account
Finally,
now
the
two countries
is negative.
differential
induces
an
non-American countries,
the quality
of the
quite good.
This is
Germany (except
in 1982),
and the
U.K.
The regressions
(except in 1985)
(3).
in 1976 and 1978, and for Italy, except in 1976, 1981 and 1984.
The regression is bad for Canada from 1976 to 1979, and in 1982
and 1986 (Chart 3).
The first
method used
future structural
nature
of
the
to
budget surplus
assumption
of
measure
the
is limited
perfect
anticipated
by
the
ad
foresight.
hoc
It
is
future
fiscal
deficits
informations available
deficits. One
today
must then
economic
on
the
estimate
agents
use
determinants
fiscal
policy
all
of
the
these
reaction
function.
2. FISCAL POLICY REACTION FUNCTIONS
We shall
structural budget
estimate a
deficit for
objective-variables of
the seven
main OECD
thus regress
reaction function
year
t+1
in which the
is a function of the
the 1973-1986
period, we
(from 1974
=-
+ 0,533 SB UK +
(7,31)
OW
2,4
31,9
SCR
2,77
308
Graph n 3 : The quality of the regressions for the non-American
current account balances.
a::
Japan
'"~5
a::
o Jap observed
Jap adjusted
'Q
~4
..
~
~
..
c:
;;;
"'c:"
~
;;;
"l
"'c:"
u
u
c:f -t
...
a -2n
74
76
78
eo
c:
'"'Q<.:>
F adjusted
1.5
..
i......
ee
116
lI'I
82
Geeaany
Ger adjusted
4
3.5
3
2.5
2
1.5
1
.5
0
-.5
-1
-1.5
-2
Feance
eo
o Ger observed
82
OF observed
.5
'"~
;;;
-.5
-3.5
"'c:" -1
::>
e -1.5-2
li
c: -2.5
~ -3
...
a::
8
74
76
o Ita observed
'"uli
;;;
&>
;;;
c:
C
~
8u
~ -3
i...
..
"'" -1
S-2
...::>
...::>~ -~
'-'
86
Ita adjusted
'Q
.~
lI'I
Italy
OUK observed
Ci
..
82
a::
United K.inldoa
UK adjusted
eo
78
-1
-2
..
-3
ee
74
a::
Can adjusted
c;
Canada
o Can observed
.5
1;,"::
tc:
~
"'c:"
-.5
-1
-1.5
e -2
-2.5
c: -3
~
li
::>
u
88
309
- the
rate of
growth of
the consumer
price level
( INF) ,
- the unemployment rate (TCHO),
- the share of public debt in GDP,
- the rate of growth of real GDP,
(DET),
(TCR),
dollar,
(the
U. S. )
we also
consider the
these variables
influence
compared to
of
the average
the
of the
estimation covers
floating,
respect to
Italy, Canada
budget surplus
and Japan.
the period
3), the
future structural
regime of
(Table
controlled floating,
fiscal policy
countries such
and France
structural budget
and the
In
against the
German and
the
Franc
U.K. on
the exchange
rate
other.
Indeed,
the
currency's exchange
latter.
reacts to
as Germany
rate in
other words
Dollar is
an appreciation
associated with
U.S. structural
(Pound)
of the
Deutsche Mark
an improvement
of the
against
the
Dollar
is
linked
to
deviation
of
the foreign
the
domestic
one influences
and Canadian
structural budget
British one.
Moreover,
in
unemployment
surpluses, but
Canada, the
rate
level of
positively the
the
dom~stic
310
United States
United l:in8dom
SB =18.97 -17.9 (S/)-1 - 0.205 DETuk-l + 1.217 TCHOuk7-1 + 2.08 DUM77 -1.7 DUM79
(-5.87)
(-5.67)
(6 ..{9)
(2.85)
(-2.50
R2 =0.961 DW =2.19
F =39.6
SCR = 3.19
Italy
SB =-0.988+ 0.335 DETus7-1-0.626 SBf-l + 1.69 DUMn. 1.92 DUM80 + 1.68 DUM83
(8.62)
(-2.46)
(2.80)
(3.13)
(2.73)
R2 =0.929 DW=2.50
f=21.0
SCR.2.51
Canada
SCR .2.20
311
312
year t-l would thus call for an Italian deficit of about 0.63 %
of GOP in year t.
2.2. The contribution of different variables
The graphical
the
different
representation of
variables
surpluses allows
to
the
the contribution of
explanation
of
structural
to explain
is true
three
can be
"justified" in
one
countries.
the
inflatiQn
which
structural budget
can
measure.
seems to
this price
the German
in the inflation
have played
especially during
fluctuations in
the increase
the U.S.
we
differential also
influence on
for 1976.
explained by
1976 a
fiscal policy.
U.S. except
differential between
the
OECO
in the
that year
main
the
inflation
the years
the inflation
during some
years (1973
to 1976,
then
1978
and
unemployment rate
compared to
Germany and
by the
German real rate of growth. The first factor explains the large
French deficit
in 1978,
while the
of the
growth in Germany.
In the United Kingdom, the evolution of the deviation
of the unemployment rate "justified" a deficit greater than the
one measured
in 1977,
while in
1979. the
evolution
of
the
313
Unlled Slales
AU_S_adjusled
0 U_S_ obsened.
;;:
""
<.!>
~
Japan
A jap
'0
'"::>
::>
'"
-I
...
Q.
...
Q.
.5
f;;:
il
co
::>
-2
'"
il
co
-3
'"::> ....
"'-;;;..."
-5
'"::>
o jap observed
adjusted
!!
1.5
!!
;;:
"'-;;;..." -I
e...::> -1.5
n
V;
E
<>
-5
-{,
::>
_0
A(lIDMX-1l
O(lIUS7X-1l
.lDTUS7H)
7.
76
7e
80
82
12
OINfJ7(-1l
10
...
-I
<.!>
-2
""
'0
...
'0
...
... -.
-3
-2
-5
-{,
-7
-4
-{,
7.
76
76
80
82
6.
86
80
62
Ger.any
""<.!>
'0
o Ger observed
Ger adjusted
!!
::>
'"
.5
'"
~
-.5
...::>
;;:
United IUnado.
A
'0
o UK observed
UK adjusted
!!
I~
'"
::>
.
...::>
Q.
Q.
..,il::>co
-I
'"::>
-1.5
-2
... -2.5
E
-3
...::><>
V;
-;;;
-;;;
...
-I
-2
-3
E ....
74
7G
7f:.
60
82
64
66
~
"V;
-5
7.
76
18
80
"(SID.M X-I)
7~,
25
.,.
..
...
'0
...
_I
-2
-2.5
-5
-7.5
-10
125
-IS
-17.5
86
8L
8-4
6'
76
80
'<t
<:>
Q-
os
!!
.
.....''""
Q.
.Q
....
g...
Vi
"'-
...
8"'-
'IS
-.5
-I
-1.5
-I
-2
-3
-~
-5
35
30
25
20
15
10
5
-5
-10
-15
74
78
F adjusted
76
78
ft"l
80
DrCA~K-1I
80
.INFn-1I
80
82
82
82
/
86
86
Ii:)
s...
!!
~
e
II>
'"
'a
...'"
"2
g
...
-7
-8
-9
-10
-II
-12
72
-I
-.5
.5'
1.5J
Vi -13
"'-
S
'S
..
-2
-1.5
-2.5
74
78
IJO
76
78
80
62
&2
o Ita observed
86
'0
86
~
84
&4
OS8f71-11
76
~,
74
"S
.e
!!
-~
-I
1.5
-.5
2.5
-3
'S
.8
-4
72
25
1.5
-6
-5
-4
-3
-2
-I
~sl
1.5
-I
"'- s
'S
Vi
~ -35
e'"
c;
...1
...
.~
15
Graph
quality of the regressions for the structural budget surplus
. n .fb : the
(cont.)
86
of observed
84
.(l/ffX-1I
UCHOI,U(-II
eo.triti. . .r til.....1.....
76
76'
DINFrolll-l1
7~
DOEFn-1I
7~
74
&
76
78
CDd.
80
Can adjUSted
78
Can observed
/104
B6
82
OlhfcM(-1)
80
80
84
OTCAc..71-11
/104
B6
"P
86
/"
arCHOt-.,,(-1)
76
76
78
erCHCo:InI-1J
.01(17(-1)
74
315
for year
use the
(that
by the reaction functions introduced in the preceding section together with relative price and monetary impulse variables. to
explain the
U.S .. France
first two
countries.
than the
surplus compared
against 0.78)
higher than
on the
expected
current account
in the
in
one for
to the
the deviation
OECD average
because the
in
French-Italian
of
the
domestic
(0.17
table
deviation
coefficient of
the
Thus,
Italian surplus
is smaller
is now
compared to
explanatory variable.
to the
influence of
budget surplus
i.e. the
fiscal impulse
show an
is
as
much
nearly 0.27
In Italy.
surplus both
we now
have an
in
influence
the
~f
regression
the
budget
(4) For the non corrected current account balance (equation 4),
the coefficient of the time trend increases at the exp.ense of
the coefficient of the American budget surplus. In the absence
of the trend. the regression becomes:
CA
= - 0,686
0.81
DW
= 1.65
9.61
SCR
4.93
316
United States
CA =0,073 + 0,528 SBPus+ 1 + 0.165 TDEus - 0,09.( TDEj - -0,018CUTusj - 0,173 TEMPS
(1,99)
(3,21)
(-3,9.()
(-0,85)
(Z,9Z)
RZ .0,908
DW 1.53
F -15,8
SCR 2,<{
(Z')
(CA-E.O.N'> =- 0,258+ O,Z72 SBusj+ J +0.1<{ TDEus - 0,059TDEj - 0,038CUTusj - 0,191 TEMPS
(Z,65)
(3,25)
(-3,09)
(-Z,O<{)
(-Z,58)
- 0,05MGROusj
(-1,50)
R2 =0,953
Japan
(3' )
DW = l.n
F =23,6
SCR = U6
CA =0,.457 + 0,062 MGOusj+ 0,157 MGROus - 0,092 TDEus +0.105 TDEj - 0,303 SBPusall+ 1
(3,07)
(2,47)
(-2,05)
(5,21)
(-<{,24)
R2 =0,952
DW = 1.9Z
F =31,4
SCR = 1.79
Germany.
CA = 3,186 - 0.102 DETita11 - 0.089 MGROjaU- 0.104 TDEall - 0,135 TDEuk -0.357 SBPuk-l
(Z,89)
(2.76)
(-2.61)
(-4,62)
(3.16)
- 0,166 SBPusj.l
( -1.66)
F = 15.1
SCR =2,07
France
CA = - 1.1 3 + 0.662 SBPus1 - 0.07 MGROfall- 0,118 CUffall- 0,078 TDEall-l 0.17 SBPfit.l
(Z.33)
(UO)
U.56)
(-4,28)
(-3.10)
R2 =0,837
DW =Z,OO
F ~ 8,Z4
SCR = 2.36
United Kingdom
Italy
DW=2,01
F5.13
SCR c 8,58
SCR = 3.62
Canada
317
French
surplus.
The
explanatory variable
Its coefficient
the
only
level of
is the
presented above.
more than
country
where
the
is slightly
structural surplus
spillover of
is
U.K.
half of
However, this
a British
confirms a
structural
budget
fiscal
to explain
between the
American
and
German
structural
surpluses
thus
of GOP.
budget surplus,
paribus, a
the
German
deficit of
deviation is
current
account
shows,
ceteris
nearly 0.17
current account
has a
surplus in
excess of
1 %
of GOP.
In
a remarkable
In fact an American
and year
t+2 now
balanced budget
for
the
account surplus
of nearly
average)
1.25 %
of GOP
Canadian
current
(as against
1.03 %
above) .
The influence
similar to
find the
the one
level of
account equation
as against
of non
American
explained in
the British
with an
-0.363).
The
fiscal
the first
surplus in
impulses
is
section. We first
the German current
current
account
is
still
Japanese surplus
British one,
while the
compensates for
influence of
the
surplus
is
the OECO
variables still
318
nearly all
the current
channel of
the terms
of unit
account equations.
of trade in Japan,
relative costs
in the
United Kingdom.
Both channels
terms
vanishes. By
of
trade
on
the
contrast, German
Italian
terms of
current
account
role
negligible way
of
in France.
growth differential
differential with
before. By
monetary
impulses
The influence
with the
in
non
U.S. disappears
Germany still
contrast,
varies
completely. The
contribution of
the different
variables and
three
countries,
"rational predictor"
the
regressions
using
the
to
of the
understand
the
origin
of
the
current
account
develooments.
Thus,
in
explained by
the positive
the American
monetary
American and
German budget
deterioration of
is of
base
and
trade, but
the
deviation
surpluses expected
the Japanese
course caused
terms of
contribution of
by the
also by
between
the
for 1983.
The
of the Japanese
in
the
American
the years
between the
recent years,
we must
1981 to
American and
German fiscal
the explanatory
emphasize the
1983 is
mainly due
policies.
to the
gap
In the most
0)
(,!)
""
0:
~
'-
'";
-;
c:
,s:,
"ouv -,
0$
-3
-2
c:
72
1.2
IA
.6
.6
.4
~l
-:;
-.4
-.61
n
2.5
1.5
.5
74
76
76
eo
62
eo
+ TOE ...
eo
oJ
.MGR<Mj.
66
.TOEj.
y-
L.~
ocurusj.
~-4,
,
M
Q:
'0
"c
v
~
c
'-
"
a.
'0
...
Q.,
-1
-2
72
1.5
-5
-,
-,
-2
72
-1.5
@
c
~.
-2
72
-3
Japan
Jap ad jus ted
76
78
DMGROJs
82
;.,
66
o Jap observed
.at
pi
84
66
O'SBPusall + I
,._.. ~_~--<r-1,
64
,-~
62
6 TD[usa
~--~~-
ao
/\
/
\;1
~v
74
eo
76
aTOI, ..
76
60
~~~-~-.,----~--
eMGPrAlSJ
74
(,!)
""
'o
.,
c:
,s:,
"ov
4 .
-I
-,
.1.5
-1
-.5
.5
-1.51
72
-5
~ -21
t 72
U
"
'0
....
a.
o
;:
$;-10
-2
-251
72
Ger adjusted
Gerllany
v-'.
eo
~
OTDEry
eo
eo
aS6Pru + I
o Ger observed
o SBPusj+ I
.MGAO,.,
erOhtl
64
united States
U.S.adjusted
01 U.S. observed
OS8Pusj.
-51-I-----~-~--~~~-_:r:_'
-4
t...
"
'"'
""3u
''c""'
c..
13
(;
...
p..
'0
13
.~
-.5
-II
C"')
'.5
1
15
2
25
3 S.
72
lj
15
51
7<
&
fruce
60
62
66
o F observed
64
GMGROtIIi
66
&SBflIPhl)
\ /'J'Q
60
OTO,1111)
60
66
Ii:
'0
It>
fa
!:!
.,
..
c:
.6
:a
c:
'0
80-
..
0.
-I
-2
-3
-4
-I
1.5
.~
0
.5
5
....
74
74
74
CUTus"
76
OSBPoI17f+1)
76
78
78
78
80
eo
60
82
82
82
64
OS6Pus7('1)
84
.5BPru(+1)
84
66
86
66
76
UDite. ~iDl.o.
o OK observed
.. UK adjusted
76
F adjusted
76
OS8USPlt 1)
'/', / \
'd
I \
fJ-.-.. ....-D
.curr,11
j ~
-1.5,
72
25.
ISl~~
7<
;il/~~ =:;~
72
Ii:
'0
1:5
!:!
c:
8
ii
~
-
-I
-2
-I
7<
&
Ita adjusted
Italy
o If a observed
oS6pr,UflJ
66
./
io
-\
OS6PJ7C+1I
.S8P11I71+ II
'S6P,l7{tl)
eTDflll
if
O~6Put.7(.I)
"
~.,~----../...
~~'b
<I __~~__~~__~____~__~~__~~~e6
~ -4L
U
0Q
2
3
30
25
IS
20
10
-3
-I
-s
-10
'0
0-
'0
N
0.
'0
...
41
321
1984,
the
monetary
growth
in
differential
1985,
and
the
in 1979.
negative contribution
United Kingdom.
account deficit
in 1985
the forecast
of
current
the deviation
of the
American fiscal
OECD average.
The inverted
V shaped
the current
similar
evolution
expected structural
astonishing that
of
the
contribution
of
is
the
domestic
budget
surplus.
It
of
course
most
the terms
of trade
should not
appear
here
States. which
for 1981).
In
two
countries
the
quality
of
the
regression
improves. This is true for the United States with the corrected
current account balance (which includes N.E.O.)
(5).
factors play
trade from
1973 to
can better
understand the
evolution of the
the quality
322
improved for
1977.
the picture
to 1977
1979-80
can be
explained by
for 1976
trade in
1975. The
evolution of
is due
1980 error
the expected
between the
impulse.
can be
explained both
American fiscal
by the
In 1985-
American fiscal
expectations, the
and Canada,
much worse
than before.
the Appendix).
overall movement
followed.
Thus,
explained by
in
(for Italy,
the first
current account
curve effect
graphs are
country,
is still
the
quite well
Japanese, German
as Canada
However,
of the
the J
is
and British
concerned,
it
expected fiscal
seems
that
impulses. As far
one
should
use
Indeed, the
explained both
by the
negative
for 1981
and
contribution
1982
of
the
can
be
fiscal
CONCLUSION
The estimation
seven main
OECD countries
(1973-1986)
led
us to
over
the
isolate the
floating
dollar
period
323
contribution of
fiscal expectations
the
expected
assumption
of
structural
perfect
budget
foresight
surplus
enables
us
to
American fiscal
on the
OECD countries.
American budget
deficit is
for
p~Lt~
the German
and
Japanese
current
accounts
but
of
the
domestic
fiscal
impulse
is
Italy and
effects are
quite diverse.
relatively minor.
This measure
of the
States in
account the
importance of
American balance
Net Errors
of payments,
and
we were
Omissions
able to
in
the
that the
specific.
In
both cases
trend among
towards a
0.2 %
the
exogenous
deterioration of
of GOP
might reflect
every year
the erosion
variables.
the external
has not
of the
This
tendency
surplus at a rate of
It
of the
has
been
emphasized
by
KRUGMAN
and
324
APPENDIX 1
Definition and sources of variables
CA
CUT
DET
INF
MGRO
SB
TCHO
$/N
TCR
TOE
XX. - yy,
325
the quality of the international specialisation of this country
could thus
the robustness
by
using
all
the
available
information
on
the
of
the
of the
reaction functions,
United States,
of
the
structural
led us
Japan and
main explanatory
determinants
Germany where
variable,
to
medium
economies
like
France, the United Kingdom and Canada where this role is played
by
the
past
unemployment
rate.
Moreover,
plays
major
either
role
in
is
(Germany, United
debt in
have
it
exchange rate
of public
can
when
significant, the
positive
almost
all
that the
Germany, Japan
main results.
impulse on
impulse influences
United States,
the fact
as in
using a
the
contribution of
is fairly stable.
first
that the
domestic
current account
France, the
for
the French
half
of
relative price
fiscal
in
United Kingdom
rational predictor
perfect foresight
evolution of
balance in
robust
significantly the
as well
Italy. Moreover,
of the
are
countries, and
assumption of
equations
American fiscal
main OECD
current account
rather than
the
or
the
eighties.
current account
Finally,
the
P.:
w
-;
~
...
I.S
I
,5
IS
2-
-2.5
-I
~-:I
ii
~ --,
~ -,-~
~
'1
-2
"
74
78
Append ix 2
0us observed
.S8Pus(+ fl.
84
eo"82"
86
~ ~
OCUTusj.
78
78
eo
,,/\
eo
~
84
Ol1lEj,
84
86
86
72
IS,
78
76
.TDEtlS
60
62
0 ",
64
64
.b"~"
&6
66
eS6Pu57(+ll
80
e....
., "'-"'," '"
16
,--.----
7.4
76
OS8Pusl+21
74
82
merican
the contrib ution of variabl es to the explana tion of North-A
- current acount balance s
p;
'-
.S
-.S
-I
;; 'J
~
-;
J:l
~
-IS
~
.
:::>
-2
~ ~S
L:
~
::
~ -_~
-272
~ ~I.S
A l1lEus.
.......
76
A US adjusted
J\
I~~ f
~
-4
-2
-6
25
-I
-15
~-IO
~ -~
...
~L
327
A SELECTED BIBLIOGRAPHY
Bourguinat,
(1987)
H.
Les
de
vertiges
finance
la
(1986) -
"The Budget
Deficit and
et Razin,
A.
and Internationl
(1985)
the Dollar".
1986.
Economic Interdependence"?
Economic
l'equilibre
externe
dans
economies
des
International
(1987)
Financial
Statistics.
Annual
Edition.
Harris. R.
(1987) -
Les deficits
et Ie
dollar
l'economie
O.E.C.D.
O.E.C.D.
Price, R.W.R.
et Hijller
structurels et
P.
(1985)
- "Indicateurs budgetaires
interpretation de
politique budgetaire
des pays
l'orientation de la
de l'O.C.D.E.".
Revue
(1986) -
fu~
Weltwi~schaft
The LDC
only a
~isk
debt crisis
of
te~ms
~elatively
add~ess
of the
c~isis
d~amatical
early 1980s
~isk
In
~ole.
debt
1980s has
been
inc~ease
the
f~om
sove~eign
solvency
p~oblems
them
concept of
in LDC
banking which
manageable.
mo~e
sove~eign
~isk
is
majo~
played
conside~ations
p~ima~ily
is
onwa~ds
mainly
capacity of the
debt-se~vicing
sove~eign
mino~
the
pe~spective.
from the
the
while
count~ies.
made to
R.F.A.)
Introduction
analyzed in
debtor
Weg 120
Dusternb~ooke~
I.
Economics
Wo~ld
In
~educed
what
desc~ibed
sove~eign
follows
befo~e
c~oss
~isks
the
basic
the main
th~ust
as
p~ope~ty~ights.
I'\ational
domain.
c~edit
but
That
contractual
obligations
contracts depend
hono~ing
of
intertempo~al
The
~esea~ch
cont~a~y
debt
enfo~ceable.
claims
c~edit
is
(1).
As
of
obligations
cost-benefit
histo~ically
enfo~cement
cont~acts
to
becomes a
the
c~edit
whe~e
matte~
pa~t
the
enfo~ce
the
self-enfo~cement
calculus on
in the
exogeneously
exist
cor.~equence
autho~
g~atefully
acknowledges
assistance by Angela Husfeld.
acknowledged
not
inst~umentalities
on mechanisms
cont~actual
to
se~vicing
is. no
usually
a~e
of
the
the
of the debtor
ve~y
efficient
(Bo~n
330
only.
between
The relationship
basically a
it
relationship between
involves
an
concluded and
asymmetry
the capital
substantial range
308-9). The
in
of discretion
debtors is
after
the
is transferred,
relationship may
matrix for
creditors and
the debtor
(Jensen and
be described
i.e.,
contract
is
has
Meckling 1976
by a simple payoff
assumed to have the choice between the cooperative (A1) and the
noncooperative strategy
contract. Similarly,
choose to
(A2)
with
respect
the principal
provide capital
to
honoring
(creditor)
(cooperative strategy
may
B1),
i.e., before
the
either
or
to
in the
combinations of
strategy of
agent A
choices.
It
is
the
noncooperative
Agent's
A1
Principal's
Strategy
Strategy
A2
B1
B2
However, the
the rational strategy on the part of the agent is only one form
of sovereign
risk that
for discussion.
LOC debt
basic
In order
crisis,
it
substitute
applies to
(2) For
( 1988) .
foreign
One has
and noncooperative
a related,
credit,
i.e.,
to distinguish
behavior on
the principal
this set
of
investment (FOI).
government) and
is of
of players
formulation
the part
foreign
between cooperative
of the
agent
(external investor).
as well.
see
direct
The
(host
Figure
noncooperative
Chakravarty
and
Holler
strategy of
the agent
expropriations. The
expropriates is,
now stands
danger that
for large
a
host
scale or
government
macrofor
FOI
up in this analysis.
What matters
either case
is to
from the
principal's point
provide capital
of view in
(d l
an LOC
some reasonable
>
d2
Further,
).
agent must
> C3)'
In the
present picture
course,
game-theoretic
framework
just
presented
is
crisis has
occurred.
It
first.
argued
that
with
respect
to
debt
sovereign risks
principals. while
with respect
to
noncooperative
behavior
concerning
foreign
external LOC
the
relationship exists
and the
It is
equity
hypothesis
between the
is
advanced
that
performance.
in
the shift
had
negative
influence
on
the
economic
it
behavior with
(3) For
325).
as
is argued
result
that the
respect to
a compatible
new
of
the
induced
move towards
capital
statement see
(Figure
solvency
noncooperative
1)
Folkerts-Landau
seriously
(1985
332
threatens
the
international
financial
Substantial
system.
is
expressed
principal's perspective
behavior are
formal
the
choices were
sovereign
predictability
In
debt
of
From
the
noncooperative
the different
Section III
risk
of
first.
determinants of
considered.
terms
probabilities
estimated both
potentially relevant
reduced
in
exposure
and
repudiation
strategic
the
changed
increased
the
relative
is applied.
on the
structure of
adverse implications
It is
made
then shown
why the
noncooperative
increased solvency
strategies
more
risks,
attractive
in turn,
again.
The
A(t)
denotes
the
stock
of
external
of return
preference rate.
to
be
paid
Let dS(t)/dt
on
A(t),
be the
and
the
time
pecuniary equivalent of
potential future sanctions per unit of time other than the loss
of future
suppose
that
captures
perceived
disutilities per
unit of
country. Thereby
invested in the
and avoided
of noncooperative
noncooperative strategy
the opposite
positive (4).
it is
holds.
if
exp - r(t-F
dt < 0
the present
net value
is
present value
general form
applies to
direct investments.
foreign credit
In what
as well as to foreign
follows 1(.)
to whom
to this
decision-making
horizon?
And what
best case
relevant determinants
behind the
decision-making
constraints.
In
formulation is
drawn from
the
present
advanced. Ten
repUdiation and
the
paper
options
much
less
partial hypotheses
issue.
They are
and
its
ambitious
on the
debt
respectively.
put together to
are
explain
of the
net present
its
unit.
expropriation
the literature.
the strategic
value in
the negative.
positive
relevant
the part
unit?
are the
fundamental variables
the
sign.
which is
equation (1)
expressed in
respectively.
noncooperative behavior
since
decreases if
a negative
the
the
sign
and
probability
net
present
of
value
163)
for a
related.
but
different
334
In what
follows,
noncooperation are
for the
in Section
hypotheses on
first presented
FOI case.
procedure (logit
the
In Section
cooperation and
C further
0 the
logit regression
results for
such as
Bates(1984), to
however,
have presented
Feder and
note only
focussed
countries, and
more
If
les~
repudiation issue
Just (1977)
a few
on
models on
of them.
the
the
debt
and Saini
The studies
debt-servicing
and
have,
capacity
of
part competing
listed below
as H1
hypotheses
to H1D.
were
When
presented.
rival
They
are
interpretations
are
first hypothesis
expectations on
strategies. Eaton
and Gersovitz
income prospects
allow that
relates to
the choice
(1981b;
the impact of
between
the
two
current consumption
is
divorced
the present
at the
may be induced to borrow from abroad, and pay back later on the
assumption that
its
the marginal
constituency
is
high
income
is
low.
If
debt
is high
the
banking
intertemporarily, then
community
to
smooth
high-income prospects
risk of
noncooperative behavior
is that
"rapidly growing
consumption
will
lower
the
repay loans, since they do not expect to enter the market again
after the
(Eaton and
future external
net present
: 16).
credit I(dA/dt)
value of
future capital
flows may
move
to
the
335
The
second
hypothesis
refers
to
another
the international
community in
the case
credit
markets
by
the
banking
counterthreat. A
single borrowing
his odds
to get
agent may
fairly large
problem for
the banks.
although
each
single one of them would not have seriously affected the banks.
Under these
circumstances banks
future
credit
forgone
I(dA/dt)
may
In formal terms.
then
actually
be
is the
high economic
simultaneous defaults.
here in
that the
for parallel
interdependence
Their hypothesis
which
explains
differs from
the one
debt repudiation.
formulation.
(H3) Another hypothesis pertains to the international
importance of
a country.
The standing
of some
countries
in
ways of
are conceivable.
position in
of course.
one
of
which
is
the
relative
leading indicator
with respect
to future
world
trade.
Hard
hence. are
important countries.
be relatively
positively with
In formal terms.
small for
probability that
less likely
to be imposed upon
the noncooperative
the political
weight of
in
investment.
in
It
real
has.
fourth.
absorption.
response to
been
argued
i. e .
external shocks
that
consumption
large
and
336
(Eaton
and
Gersovitz
adjustment costs,
8-9) .
I98Ib
governments may
In
have a
order
to
reduce
strong incentive to
borrow from
are subject
to severe
external shocks.
It follows,
in turn,
that the
since in
the light
foregone future
of effective
credit I(dA/dt)
large. Notice
vulnerability may
also mean
that
the
country
faces
liquidity problems.
(HS) A further explanation relates to fluctuations in
the domestic
one
economy. The
forward
put
as
consumption patterns
argument parallels
HI.
Governments
strongly to the
may
wish
to
smooth
versa.
If
markets I(dA/dt)
future access
would be
severely hampered
in the
case
of
higher is
the probability
chosen (H6a).
I(iA) and
Th i s app 1 i es,
of course,
to interest
payments
(Eaton and
Gersovitz I98Ib
amounts of
agent in
credit may
also reflect
the international
conclusion,
i.e.,
: 302).
On the
other hand,
large
governments could
speculate to
get hold of
and Gersovitz
I98Ib:
9).
In
the case
exporters may
wish to
continue trade
of
willful
only on
a cash
default,
basis,
i.e., external
default
decision.
probability
of
noncooperative
337
behavior is,
country on
thus, negatively
imports. An
related to
the dependence of a
Sachs (1983
: 20)
embargo.
In
is that
of
the
of
p~ssibility
trade
(1)
further argument
(1985). He
argued that
higher when
income is
rationale is
assumed to
this:
- would
The hypothesis,
low. This
and the
Lachler
expected.
higher when
The
it is
that
the
probability
the actual
of
debt
income is unexpectedly
and contrasts
Lachler's
includes
changes. H5
that it
prev i ous 1 y
related to
in situations
H5. The
reasoning
variations, whereas
(H5)
by
thus, states
both parallels
that
than
be lower
repudiation is
presented
be positively
agent's country.
was
refers
the argument
anticipated
as
contrasts with
would state
critical difference
to
unexpected
income
of intertemporal consumption
well
as
unanticipated
the hypothesis
a negative
is
income
presented here
relationship
also
when
in
the
hypothesis addresses
on noncooperative
it may
rand/or private
charity
the magnitude
behavior.
In
of
the case of
of
affected
lending
organizations.
In
other words,
I(dS/dt) may
probability
of
noncooperative
behavior
is
is that
negatively
final hypothesis
relates to
the level of
If the difference is
invested
on
the
part
of
the
borrowing
country.
The
338
difference between
of the
choice between
I(dA/dt)
is
marginal rate
costs
Gersovitz 1981b
negative
noncooperation,
since
between the
borrowing
cooperation and
of
return
(Folkerts-Landau
: 8).
on
1985
The hypothesis,
relationship
exists
noncooperative behavior
330
thus,
between
and the
capital
the
Eaton
and
states
the
difference
and
that
probability
between
the
a
of
two
rates (6).
In Table
of cooperative
and noncooperative
-/+
+
+/+
+
+
-/+
+
+/-
+
+
+
+
+
and
components
the
negatively with
of
condition
the determinants
(1)
vary
denoted in
positively
or
parentheses. The
(6) One may note. however, that from the portfolio theory point
of view a higher rate of return may simply be a compensation
for the
higher systematic
risk associated with an LDC
investment (Sharpe 1970: 97). The implicit assumption here is
that borrowers from LDCs accept a certain trade-off between
risk and return.
339
to understand
noncooperative behavior
provide the
is comparable
Similar attempts
have been
Schneider
hypotheses are
theoretical basis
macro-level that
(1986),
and
for expropriations on a
to the
undertaken
Frey
labelled as
(1985),
Gil to
by
Burton
and
G20.
Frey
In
and
Inoue
(1985).
The
The
stability (7).
to the
the basic
: 9).
refers
subject
to
political
presence
of
to
large
forces
foreign
political
internal
may,
capital
among
in
the
visibility of
over the
be
conflict between
relate
country.
hypothesis
Countries may
tensions. The
others,
first
foreign investments
popular concern
in the
In a
stable political
acknowledge the
presence of
balance of
environment a
the benefits
government
would
This balance
may translate
and national
Stoever talks
about the
emergence
could be
stable capital
adjustment of
love-hate
attitude
: 8).
the actual
clear
then one
about
its
noncooperative behavior
cases where
ratio and
total investments,
is quite
of
the
upon the
adjustment to
new ratios
dealing
with
FOI.
The
risk
of
internal
tensions may,
tensions
are
large,
because
translate into
stop-and-go policies.
I(eA) captures
the effect
discussed.
: 6-7),
the
require
and they
may
hypothesis,
thus,
340
states
that
stability and
negative
relationship
consistency of
policies
exists
towards
between
FDI
(G2)
withdrawal of
next hypothesis
aid to
: 102-7
payments
; i(eA
(included
follows
the
70-71).
the
potential
1985:
may have
may be
191-2). The
hypothesis
level is,
the support
foreign
offset by
The
gains an
confiscating
from
l(dS/dt.
in
The higher
expropriating government
assets (IliA)
and
relates to
the
reads
as
(G3)
hypothesis is
relates government
economic policy
behavior to
G1
in
that
it
performance of
of doing
similar to
to assess.
One way
so, however,
is to
look
at
country as
their consistuencies
performance within
the LDC
noncooperative behavior
performance is
country group.
The probability of
lagging behind,
since then
the
policy-makers
Multinational cooperations
the scapegoat
talked about
drive governments
called it
(1968:
terms of
equation (1),
r may
In the formal
The
question
was
may
as
trigger
to
whether
unanticipated
income
behavior. The
that there
increases
raised
noncooperative
is
is a
of expropriation
in these
situations the
value
of
the
assets
expropriating
~1
governments may
get hold
definition residual
of is
larger. Equity
claims are by
may
then
be
translates into
value. Since
assumed
a lower
be
relatively
and potentially
the focus
argument matters
to
is
more if
on
which
unexpected
the time
in equation
large,
developments,
the
Income
influence on
(1979:
profit
the probability
10) argued
the prospects
it is
and
expectations
may
have
an
for future
In this case
more likely that the value of the assets seized and thus
the net
present
value.
is larger,
Alternatively,
it
was
since in
these situations
profits I(iA),
foreigners in
There are,
a government
which might
order to
thus, two
may wish
to take the
stimulate economic
rival hypotheses
performance (G5b).
that
are
related
to
that
may also
create an
strategy.
large
In
the present
variations
income
respective economy
is mono-structured.
income variability
is to
attract new
be larger
the hypothesis
behavior
is
way
may be
that
of
the
reducing
In formal terms,
I(dA/dt) may be
states that
smaller
in
the probability
the
presence
of noncooperative
of
large
income
variability.
(G7)
The
noncooperative
next
behavior
hypothesis
with
on
respect
cooperative
to
foreign
and
direct
the debt
counteract potential
much better
case
as
well.
sanctions on
larger
the part
country
of the
may
investors
342
role in
that the
expropriation risk
the weight
of a
assumed to
vary negatively
country.
In
It may, thus,
formal terms,
with the
if
I(dS/dt)
importance of
may
be
a country
and, thus, make the net present value smaller, everything being
equal.
(G8) The
also be
assumed to
stock invested
was
probability of
argued
in
reference is
country and
vary positively
in an
economy,
G1,
explanation is
an expropriation
event may
reasonable
benchmark
for
FOI.
This
year to
define
expropriation
as
macro-event
Another
cooperative and
payments pressures.
first one
determinant
of
the
noncooperative behavior
is that
between
payments problems
choice
relates to balance of
of extraordinary balance-of-
pressures could
be reduced
by transferring
the ownership
drains caused
by the
The hypothesis,
expropriation
(G9a)
6)
thus, states
increases
(Stoever
(9).
repratiation of
1982:
at
thus, states
lower at
it
times
of
(G9b).
of foreign
that the
times of
In
horizon chosen
macro-
pressures
may
large
well
be
external
argued
imbalances
i.e., expropriations
exchange. The
probability of
that
may cause
set
a
counterhypothesis,
macro-expropriations is
formal
interpretations,
probability of
balance-of-payments
counterintentional signals,
further drainage
that the
with
Alternatively,
expropriations
it
terms,
appears,
and the
the
is
difference
a matter
between
of the
the
two
actual time
decision-making unit.
343
(Gl0) The
refers to
the developmental
this hypothesis
creates a
may be
last hypothesis
there is
income economy
risks
crucial reference.
proxied by
on expropriation
The magnitude
the mean
per capita
of this reference
with an
income close
said to
choose the
countries which
capita incomes
have either
A
much higher
more
or
readly
much
curvilinear relationship
is
than
lower
per
is said to exist
instrumentalizing FDI
development process
rising, whereas
firms is
and the
the instrumental
run those
usefulness
of
firms
is
nationalized
economic interactions
get more
terms,
to
I(
eA)
seems
be
and more
a
complex.
function
of
In
the
formal
stage
of
development.
The summary table for the hypotheses on expropriation
risks is
presented as
whether the
to
vary
Table 2.
The signs
displayed indicate
discussed in
or
hypotheses Gl
negatively
to Gl0.
with
The
the
determinants
definitions
of
the
344
Expropriation
Risk
(DUM)
(AIDCRQ) (WRYM)
+
(DWRY)
(WRGDY)
+/(GDPVAR) (POP)
(IDBSR)
(BOP)
+/Gl0 (DGDPKA)-
Gl
G2
G3
G4
G5
G6
G7
G8
G9
Note:
+
_
+
+
-/+
The variables
in the parentheses are used as proxies.
MNote that a large r puts a larger discount on flows in
the more distant future.
It can also be read as
reflecting a
relative short time horizon on the part of
the agent.
has
already
theoretical model
been
exists that
said
that
no
would capture
comprehensive
all the
effects
discussed above
choices between
cooperative and
all related
to the
crucial variables
noncooperative behavior
relationships were
form. For
debt
the
the
specified in
repudiation
probability
dependent variable
in
equation
were
condition
for
(1).
the
but
functional
(Gl-Gl0). respectively.
where
displayed in
not explicitly
the present
probability of
noncooperative behavior
(Hl-Hl0)
and
expropriation
noncooperative
and proxies
were chosen
behavior
is
the
to operationalize
dependent variable
measured in
"1" if
is a
noncooperative behavior
cooperative behavior
either-or-categories only.
was observed.
~5
problems. While
probit and
logit
analysis
comparable (Altman,
logit
analysis
iteractions are
the
for
equal
Avery, Eisenbeis
offers
results
data
sets
for
are
computational
necessary. An
comparable studies
estimation
quite
31-3),
advantages
alternative technique
of
since
used
in
Frank and
models in
it avoids
noncooperative agents,
hypotheses
While
this
relatively small,
example, were
(HI-HIO,
explains
countries like
not taken
tensions during
the time
GI-GIO)
in
simultaneously
part
why
Nicaragua
into account,
the
and
because
period covered
sample
Turkey,
the
country
reference to
thus, part
capture potential
of the
set of
is
for
internal
were
of
parallel behavior.
determinants of
They
are,
346
cooperative and
noncooperative behavior.
mentioning that
one half
seventeen most
of the
heavily indebted
Finally,
country set
it is worth
belongs to
countries (World
the
Bank 1986 :
the theoretical
dependent
variable.
rescheduling events
been triggered
First,
data
are
available
for
by solvency
case.
However,
possible explanations
contrary,
the
it
on
the
of default
is argued
hypotheses
in
present
analysis
were taken
not
all
willful
default
do
in
fact
provide
reasonable explanation.
Second,
in
order to
dependent variable,
dimensions, one
case.
In
had to
absence of
expropriations of
considered.
the reschedulings
decide on
data on
FOI, the
all
had
macroeconomic
the equivalent
the actual
in the
volumes involved
FOI
in
legal transfers
of foreign
finally lead
1980:
used which
were made
all
forms
four
whenever three
year, the
to an
68-9).
effective transfer
of
of
or more
condition for
involuntary
divestment.
expropriation events
macro-expropriation
Specifically,
were listed
was
assumed
per
to
prevail (12). Finally, the time periods covered are 1976-85 and
1961-77 in the debt and the FOI case, respectively.
With respect
to the
(12)
This assumption seems plausible,
since expropriation
actions may upon occasion be motivated by other than macropolitical configurations.
It may be noted that the empirical
results did not change much when the critical value was raised
to 4 or more expropriation actions per country and year. For a
thorough discussion on measuring involuntary divestiture see
Williams (1975), Kobrin (1980), Burton and Inoue (1984).
347
Appendix. Several
cases the
G8)
in
reciprocal of
order to
model more
linear
meaningful. Since
transformation
in
approach zero.
variable name
the logit
the
sure that
noncooperative behavior
appropriate.
in some
in H6 and
transformations make
and G8
first
model implies
place.
the partial
approach zero
Second. an
these
variable
probabilities
if GDL
"L"
a non-
was
of
and IDBSR in H6
attached
to
the
choices
should
be
taken
as
first
approximation.
Third. with
of political
respect to
instability on
the probability
(G1). additional
econometric calculations
well-known stock
adjustment
stocks of
FDI. Let
model
SFDT(t) be
of expropriation
(Koyck)
was
applied
for
time t.
time t.
According to
change in
the stocks.
the stock
adjustment model
the
actual
is proportional to
the gap between the current desired level SFDI*(t) and the past
actual level. which is SFDI(t-1)
(2 )
SFDI(t) - SFDI(t-1)
(equation 2).
T(SFDI*(t) - SFDI(t-1
along the
6K(t)
(1-T)SFDI(t-1) + T6K(t)
348
(5) FDI (t)
(l-T)FDI(t-l) + T6I(t)
where K(t)
sample.
In
was tested
for
the
FDI(
selected
t)
country
was sufficiently
country to
high. the
reflect stable
dummy value
investment conditions.
in all other
D. Empirical Evidences
The
findings
for
the
debt
repudiation
case
are
presented in
indicated by
the average
reasonable (0.86).
corresponding to
The t-ratios
hypotheses HI
for the
to HI0
single
coefficients
are displayed
in
the
parentheses.
(13)
The criteria actually applied were
(a)
correct and
plausible signs for the parameters. and (b) an adjusted RSquare
of 0.5 or more as an
indication of
the overall
explanatory
power of
the stock adjustment model. Morocco was excluded due
to an
insufficient data base. for Sri Lanka and Thailand the
expected mean adjustment
lags E(I)
were considered to be
unrealistically high and low.
respectively. The procedure as
well as the classification derived should be taken as a first
approximation only.
The regression results are provided by the
author on request.
349
Table
Logit Estimates
Repudiation
(Constant)
the
Probability
of
Debt
2.57
(0.71 )
-0.83
( -1 .89) 2.88
( 1 .88)8.36
(0.78)
19.22
(2.19)--23.02
(-2.06)--12.99
(-1.31)
-13.22
(-1.39 )
-0.07
(-0.44)
0.03
(0.10 )
0.05
(0.42)
0.86
Hl
(WRGDYL)
H2
(AND)
H3
(POP)
H4
(EXPVARL)
H5
(GDPVARL)
H6
(GDL)
H7
( I MPQL)
H8
( DWRYL)
H9
( AIDCRQL)
HI0
( DZKPL)
(AVG. LH. )
Note
for
Source
significant at
hypotheses
the 10
have
to
insignificant. The
data confirm
prospects
two-tai led
the debt
be
which
negatively
with
(willful
as
the
are
other
statistically
stated
the
debt
that
high-income
probability
defaults)
are
that
chosen.
repudiation strategy
Mexico, Brazil
balance of
hypotheses
or better
considered
Hla
noncooperative actions
Second, LDC
percent level
four
hypothesis
vary
test,
and Argentina
if
other
- take
the
governments
lead
(H2).
here
Third,
the probability
of noncooperative
hypothesis that
interest in
smoothing consumption
negative impact
confirmed.
behavior (H4).
Fourth, the
350
Further,
positively with
indicated by
competing
size
the
of
the probability
a negative
hypotheses
the
external
debt
varies
advanced
as
H6a
and
H6b,
the
evidence for
probability of
the agent
on imports
measure what
is at
export credits,
sanction in
a negative
correlation
noncooperative behavior
and the
between
the
dependence of
or else,
the case
if a
trade embargo
of noncooperative
is imposed as a
behavior.
All
other
to the
expropriation case
it becomes immediately
is
much lower
only three
in a
than in
hypothesis were
two-tailed significance
test at
to conjecture
of the
that rational
decision-makers as
economic calculus on
significant,
hypotheses
then
obviously
the
advanced
as
(14)
The
population size variable
introduced to capture
hypotheses H3 was not adjusted for the fact that India had a
population more as ten times as high as that of Thailand, which
followed next.
(15) Note again that the condition of whether the cooperative
or the noncooperative strategy
is chosen in equation (1) does
not assume that anyone maximizes social welfare.
351
G2
(AIDCRQ)
G3
(WRYM)
G4
(DWRY)
G5
(WRGDY)
G6
(GDPVAR)
G7
(POP)
G8
(IDBSR)
G9
(BOP)
GlO
(DGDPKA)
AVG.LH.
Notes
-3.51
(-1.75)-1.84
(-1.19 )
-0.25
(-0.81)
-0.21
(-2.02)-0.28
( 1 .79)0.17
( 1 .07)
-2.35
(-0.75)
-0.31
(-0.06)
-0.43
(-0.60)
0.06
(0.36)
-4.39
(-0.33)
0.76
Sources
was G3.
This hypothesis
performance in
makers in
become a
stated that
relatively strong
in comparison
to
what
was
the debt
case, all
that the
knowledge about
the
left with
agent's
choice
352
III. The
Effect
of
Cross-Default
Clauses
on
the
FDI
in
the
repudiation can
previous
be better
indicators for
which are
section
have
predicted than
shown
that
debt
expropriation.
The
0.86 in
second, the
the debt
number of
understanding
of
and 0.76
in the
hypotheses which
the
choice
FDI
case
and,
contribued towards an
between
cooperative
and
striking that
the 1970s,
obviously the
i.e.,
the
major
introduction
financial
of
cross-
default clauses,
hypotheses which
were confirmed
previous section,
threat of
not less
denying future
markets (Table
than three
access to
3). Cross-default
were
related
to
the international
the
credit
since
enforcement of
party enforcement,
debt case
it
means
which both
(see condition
(1
that
mechanism
reduced sovereign
and
made them
of
self-
risks in the
more manageable
was
observed
in
the
case
of
foreign
direct
follows it
cross-default clauses
is shown
translated into
that the
a change
adoption
of the
of
price
the two
determinants
(Section B).
of
types of
sovereign
LDC finance.
risk
costs
In a first step
are
presented
353
investors) and
the agents
decision problem
whether or
for a
(debtors or
principal is
(creditors or
direct
to
decide
on
the principals
two players,
has also
to assess
noncooperation occurs,
what the
and whether
costs are
and at
if
actually
i.e., whether he
sovereign
agent,
required which
compensation
covers the
charge
(loss
expected costs
premium)
is
now,
what
matters
configuration of
in
the cooperative
are chosen
by the
prediction of
to
first,
the
choices are,
with which
agent
addition
and the
and,
the agent's
the
costs
of
any
prior probabilities
noncooperative strategies
second,
the
quality
of
the
and 81
monitoring
exist for
alternative without
situation
where
the
the
agent
the principal
contract.
alone
and
the
i.e.,
costs
may
be
to
move
to
the
had erroneously
adhered to the
e3
principal if
the agent
stayed with
(Al), and
represent
the loss,
e4
principal's
information
the principal
cooperative strategy.
exit
and
sovereign risk
noncooperative strategy.
loss for
the
described by cooperation,
sides, additional
assumed to
terms, suppose
positions are
denotes
the
loss
the cooperative
if
he
had
moved
to
the
strategy
to
the
354
(17).
Principal's
Strategy
81
82
From the
Agent's Strategy
Al
A2
assumption
is
conjunction with
logit analysis.
in
Figure
that
recognizing that
(d j
the
Consequent ly,
rules
(d j
the
decided to
probability
move
from
moves from
cooperative
(1-[3)
R (A2, d j
(6)
(7)
to
A2
Al
(beta-error).
Agent's Strategy
Al
A2
cd d j )
(1-[3(d j
[3(d j )
(I-cdd j
to assess
the costs
of
engaging
in
the
decision rule
Al
not
and (I-cd
81
82
In order
sovereign risk
i.e.,
in
Principal's
Strategy
for any
of
Al
principal erroneously
with
some
denotes
the agent
(alpha-error). Beta,
stays
follows
The probabilities
Alpha
3.
he
and AZ,
dj
for
which are
any given
the risk
functions R(AI,d j
and
Al
R(Al,d j
el(I-[3(d j
+ e 3 [3(d j
A2
R(AZ,d j
ezo:(d j
Suppose now
+ e4
that data
l-o:(d j
are available
on
the
prior
probabilities of
cooperative (AI)
and noncooperative behavior
(AZ) on the part of the agent. and denote these probabilities
with pI
and pZ
(I-pI). Then the expected loss for any given
(17) Note that the
losses refer to absolute reductions of the
present value of an existing contract,
i.e.,
a
loss could
exceed the initial face value of a given capital transfer.
355
decision rule
may
be
defined
as
the
weighted
average
of
R(Al,d j ) and
of Al and A2,
loss B(d j )
The expected
for decision
and
rule d j
R(A2,d j )
are
absolute
probabilities, B(d j )
of B(d j )
to the
costs,
is measured
and
pI
and
(l-p1)
are
initial capital
is an
it is obvious
chosen decision
forecasting or
logit functions
rule d j
estimated
in
Section
probabilities of
simply
to
explanatory
up an
fill
variables
probabilities of
such a
is
related
in
in
used
for
One
the
II
can
be
appropriate
order
to
get
noncooperative behavior
principal should
probability
of
some
noncooperative behavior.
data
the
of the
stay
with
critical cutoff-rate
q*j, and
the
agent. Given
one can set
qj.
The
strategy
noncooperative
for
predicted
critical
to
predicting the
needs
which
rules
Bl.
action
is
if
say
the
smaller
that
predicted
than
the
B2 otherwise.
Errors are
rules.
If
one moves
one choses
unavoidable by
a different
critical cutoff-rate
decision
upwards,
the
rule
dj .
By
alpha-error
shifting
the
probabilities
a trade-off
beta-errors. The
a(d j )) and
avoiding
alpha-
and
(1-~(dj))
It is
exists between
change accordingly.
obvious that
for any
(e. to e4) and prior probabilities of cooperative and noncooperative behavior by the agent (pI and (I-pI)). the Bayesian risk
356
in equation
(8) changes.
This
becom~s
immediately obvious.
if
into (8).
rate q" j. and hence the decision rule d"j is selected such that
the Bayesian
risk B(d"j)
which
the
calculated both
required
for the
sovereign
loss
premiums
can
be
B(d"j). and
accordingly.
course.
The
corresponding
sovereign
loss
premium.
of
transfer C.
C. The
Decline
in
the
Relative
Costs
of
Debt
Repudiation'
Suppose now
form of
either debt
the optimal
e4. 1 .
and
debt
risks as
Bl(d"j.l) and
to
e 4 .2.
The
prior
probabilities
of
case.
and
pl.2
and
(l-pl.2)
in
Similarly. the
alpha- and
beta-errors depend
decision
d"j.l
d"j.2
rule
B2(d"j.2).
cooperative and
the
Bayesian loss
respectively. The
to
and
chosen.
the
FDI
on the
in
case.
optimal
respectively.
Then
I B2 ( d" j
B 1 ( d" j
[pl.l
(el.l(l-~(d"j.l)
1)
2 )
+ e3.1~(d"j.l
+ e3.2~(d"j.2 +
(R) .
It is
induced by
obvious that
the change
of the
price
ratio
357
to
d(a(d- j . 1 ))
innovation. while
and
p1 and
induced
(~(d-j.l))
by
this
both
~(d-j.l)
of increased
constant for
of the
relative sovereign
condition means that the loss when exit is chosen. although. the
country did
not move
"excess" monitoring
an unanticipated
Since adjustment
to willful
default.
is
have changed
behavior
equation
in
expression for
would be
part of
(11
has
forgone future
higher in
the case
the creditors.
shown
that
capital inflows
of an
Assuming that
I(dA/dt)
(if
as
an
positive)
decision-
indebted agents.
The
partial
derivative
of
with
[e , . , (l-~(d*j.l)) + e 3 . ~(d*j.l)
- e Z . 1 a(d*j.l) - e 4 . 1 (1-a(d*j.ll]/B2(d*J.z)
358
that the
term is
alpha- and
beta-errors as
smaller than
again
negative for
e2.1 and
quite
would be
e4.1
strategy (A2).
(sufficient
and
e3.1
are
of
both
otherwise
the
combinations
an asymmetry
Now,
e1.1
assumptions,
which come
larger than
there is
long as
reasonable
expected losses
all possible
if
(R(A1,d j
in (6)
in
generates
in (7).
the present
leads
result of
this analysis,
thus,
is that
under very
magnitudes
of
Figure 4
the
loss
matrix
(Figure 4),
the
Principal's
Strategy
A1
1
2
B1
B2
of
emergence
relevant
cross-default
Agent's Strategy
clauses
A2
4
3
reduced
the
relative
argument referred
forecast techniques, while leaving the prior risks (p1 and (1 p1) unchanged.
The second
cross-default clauses
of cooperative
debtor, while
reasoning took
into
account
that
and noncooperative
leaving the
actions on
forecast
technique
unchanged.
Of
1980s, although
(World Bank
the
indeed falling
so-called
from the
spreads
on
early 1970s
syndicated
to the early
1985:
359
and consistent
magnitudes of
with the
prediction derived
charge (ignoring
assumptions about
risk-aversion) for
the loss
quite plausible
estimated at 3,85 % on
matrix was
Impact
External
of
Sovereign
Capital
Loss
Structure
Costs
on
the
and
Economic
Performance.
A. The Hypothesis
The hypothesis
basically of
three
suggests that
relation to
if the
in
First,
this
section
simple
consists
economic
calculus
the other,
conditions change
to the
advanced
parts.
Applied
for incurring
finance of
debt has
the country.
section that
relative to
the sovereign
the
result
loss premium
of
the
for debt
previous
went
down
the
question
external finance
the agent's
Modigliani-Miller
arises
as
is irrelevant
to
whether
the
in
the
business
finance
If
it would
matter,
in
which
direction
would
the
360
impact go
that the
relative to
equity capital
of the
debt finance
displays the
percent
it
reasoning
behind
this
finance,
and
that
magnitude of
public involvement
in LDC
all
guaranteed. Hence,
the external
the
is public
investments are
borrowing from
80
hypothesis,
in practice
foreign direct
Table 5
to understand
debt
incurred
any reasonable
is
puhlic
or
publicly
to public
decision makers
foreign direct
and
investments relate
the
related
to priyate
361
Table 5
Public and
Publicly
Guaranteed
Private
NonGuaranteed
Total
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
49.49
60.13
68.16
83.72
103.87
125.72
154.82
193. 14
246.58
292.33
351.76
393.32
443.81
513.07
552.96
16.96
14.87
22.52
25.52
31.53
36.46
40.04
46.37
55.54
62.75
75.30
96.67
103.09
110.05
112.53
66.45
75.00
90.67
109.24
135.40
162. 18
194.86
239.52
302. 11
355.08
427.06
489.99
496.41
622.12
665.49
Sources
In what
follows the
government agents
quite popular
0.74
0.80
0.75
0.77
0.77
0.78
0.79
0.81
0.82
0.82
0.82
0.80
0.89
0.82
0.83
view that
governments and
number of arguments.
First, De
Alchian and
Alessi
Demsetz that
has
argued
in a
along
the
classical firm
lines
of
centralized
and rewards
i.e., the
shirking information
solved by
assigning residual
entrepreneur
agents,
in
(1980
turn, there
directly from
gain from
since
perquisites, i.e.,
civil-service
suggests that
rights to
For
is hardly
such profits.
residual incomes
expense of
15).
in a
problem of
team production is
income to
politicians
a
and
possibility
the ownergovernment
to
benefit
regulations
preclude
forms
the
of
income,
payment
of
increased reliance
(private) foreign
upon (public)
direct investments
debt
at
the
reduces
the
362
aggregate rate of return on capital (20).
Second, Fama
time horizon
quality
of those
of
the
of ordinary
hold for
Hence,
increased
It
horizon of
business
men.
to western-type
regimes who
broader sense
decisions.
the time
certainly applies
who manage
investment
acknowledged that
than that
and Jensen
has
widely
been
politicians is shorter
While
this
assessment
of democracies,
it may also
public involvement
would indicate
in
the
economy
in
the
favor
of
cross-default
individual borrowers
is supported
may
covering
publicly
within
country,
since
hypothesis
clauses
have significantly
deliquent borrower
is not
the
from Folkerts-Landau's
of
guaranteed debt
risk among
argument in
will have
particularly strong
in a
syndicate to
to bail
borrowing
agent
effectively.
a given
the additional
leading
discretionary profit
banks,
monitoring would
while
the
accrue to
margin that
overall
all
benefits
banks
goes to
of
the
effective
participating
in
the
syndicate.
Fourth,
in
most developing
centralization within
government is
countries the
degree of
i.e.,
federal elements are the exceptions rather than the rule (Hicks
1978). Since
concern of
external relations
the central
intermediate and
government.
lower levels
(20) Wallich
(1986:
part of the borrowed
investment.
role in comparison to
is increased
by
the
extended
70). for
instance. argued that a large
money went
into consumption rather than
command over
is true
that a
money more
decentralized system
efficiently (Nelson
borrowing would
also mean
1987),
that the
If it
then
extended
quality of
the
public
spending
note
that
substantial
part
of
the
questionable
productive use
whether
abroad due
operations. But
the
capital
to the
is
illegal
put
into
character
most
of
the
suggests
that
the
capital
received
is
spent
less
efficiently (21).
The two
be combined.
premium does
not only
structure away
part of
solvency that
(1958:
265), a
run into a severe solvency problem which may have been minor in
the first
place (22).
default clauses
make it
was to
While the
primary
purpose
of
cross-
long-run
may
be
that
solvency
risks
are
substituted
for
sovereign risks
may be
lead to
to B2)
the agents.
In
turn,
advantageous in
for
the
agents
it
may
then
become
the light
from the
cooperative (AI)
as well.
In the
collapses for
selected above.
external finance
investment to
as measured
In Table
by the
6 the
structure
ratio of
of
foreign direct
sovereign loss
Debt ratio.
The use
premium should
The hypothesis
of cross-default
relative loss
is, thus,
clauses has
premiums, but
not
has also
only
lead to
reduced
the
a shift in the
second and
displays
the country
there is
also
data
on
group (unweighted
moving averages
(WRGDY). Again,
that the
of
GDP
growth
a substantial
the
economic
average). As
were
calculated
rates of
returns on
(see the
data
generated
in
unweighted averages
capital were
the
estimated for
International
are presented
as RR.
Monetary
each country
Fund).
These figures
The
also
365
FDI/Debt
0.41
0.48
0.35
0.10
0.38
0.46
0.22
0.16
0.01
0.17
0.07
0.08
0.04
0.19
NA
NA
NA
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
Note
RR
20.08
15.0
16.0
15.5
15.0
15.6
13.8
13.3
13.3
12.5
11.9
12.0
12.3
16. 1
NA
NA
NA
NA
0.71
0.73
0.72
0.81
0.81
0.72
0.84
0.86
0.83
0.89
0.86
0.86
0.85
0.88
0.88
NA
Growth
Prospects
WRGDY
Libor
(average)
(real)
6.8
6.3
6.3
6.7
6.7
6.0
5.4
5.8
5.6
5.8
4.7
3.6
2.7
2.1
2.4
2.3
NA
NA
NA
3.28
0.97
-3.68
-7.90
-2.25
0.97
-0.05
1.05
-0.41
0.09
7.77
11.29
8.42
8.54
13.40
Following now
would expect that the decline in the FDI-Debt ratio had lead to
a less
Table 6
favorable economic
support this
coefficients
are
performance. The
hypothesis.
displayed.
In
They
data presented in
Table 7
are
the correlation
large
and
highly
share of
only been high at the outset. 0.71 percent in 1970. but that it
increased continuously
366
Table 1
FDI/Debt
(AVR)
FDI/Debt
(AVR)
RR
(AVR)
WRGDY
(AVR)
0.692
( 13)
P=0.004
0.637
( 13)
P=0.010
1.00
0.520
( 13)
P=0.034
1.00
1.00
WRDGY
(AVR)
RR
(AVR)
Note:
P-Values denote
covers 13 years
Source
the analysis
Table 6.
One might now argue that time-series data are not the
appropriate basis
forces might
on the
the
for
testing
hypothesis,
since
other
external capital
estimated
the
structure, the
aggregate
rates
of
return
on
capital
were
therefore averaged over the sample period (1969-82), and crosscountry correlation
are displayed
coefficients were
in Table
8. Although
FDI/Debt
( M)
WRDGY
1. 00
WRGDY
(M)
0.294
( 10)
P=0.205
1.00
(M)
RR
RR
(M)
0.840
( 10)
P=0.001
0.431
( 10)
P=0.107
1.00
(M)
Note
P-Values denote the significance levels; (M) denotes
the country means (1969-82). The parentheses indicate that 10
countries were looked at.
Sources: See variable list in the Appendix.
367
The correlation
average real
rate of
significant at
prefer.
between the
return on
FDI-Debt ratio
and the
the
time-series
analysis
or
the
cross-sectional
data
presented
in
Table
do
even
provide
sovereign risks.
return on
capital RR
1980s, it
also reached
whether
the
to
come up
returns
foreign capital.
rate
of
up to the early
where doubts
generate
attract new
marginal
only decreased
a level
potential
sufficient to
The average
had not
on
as
to
capital
is
on capital,
one can
see that
this
rate
has
(Picht 1988),
the stage
are
may have
not
able
already
reached where
many countries
capital. That
enough return
on the
capital invested
to
service
been
new
the principal's point of view. Recall from Section III that the
the minimal
estimated at
be high
Bayesian (loss)
risk in
the debt
enough to
this
distinction between
solvency
hypothesis
holds.
sovereign
risks
however.
on
the
then
one
hand.
the
and
risks
other hand
becomes
empirically
on the
inoperational for new capital. One may recall that the initial
discussion
had
focussed
on
cooperative
and
noncooperative
behavior on both sides of the game, the agent and the principal
(Figure 1). Even if one were to assume
that the agent is
willing to play the cooperative strategy, principals will still
expect excess premiums over the risk-free lending rate to cover
the expected
sovereign risk
loss
and
the
non-diversifiable
368
risk.
If
the economies
enough return
return on
on the
do not
seem to
capital invested
investments including
be able
to meet
adequate risk
to
generate
the
required
compensations,
quite
consequence,
marginal
in
(World
turn, the
Bank
1986)
(23).
But
as
cut-off
equation (1)
from
future
capital
anyway
(I(dA/dt)
in
drops).
Both
the
principals
and
the
agents
have
acted
: 4).
noncooperation of
both sides
much a
reduced trust,
matter of
lack of
sufficiently attractive
LDCs covering
but -
investment
opportunities
in
V. Conclusions
While the international banking community was able to
organize collective
action quite
investors in
LDCs were
debt gained
ground at
the cost
of private foreign
direct investments.
The unfortunate
public debt
generate
sufficient
severely affected.
returns
In the
on
the
early 1980s
capital
invested
the solvency
was
problems
(23) Note that the FDI-Debt ratio for the country sample chosen
has increased substantially in 1982 (Table 6) and onwards due
to a drastical reduction of new bank loans (see also World Bank
1985 : 112).
369
induced began
mechanism.
to spill
The
suggested not
even
more
back to
reduced
only a
profitability
of
LDC
investment
important
from
future
cooperation
(voluntary
lending).
In turn. LDC governments were lead to minimize losses
by moving
to the
point of
noncooperative strategy
a sanction
related to
a steady
to generate
cover the
enough to
does not
required risk
In either case
was
thereby
recurrent reschedulings
outward form
sanctions
certainly
of noncooperative
other
than
applicable.
It
the market
sanction of
advantageous
rather than
behavior. since
foreclosing
was shown
in the
to
choose
outright default
fresh
as
the
in this
money
are
case
hardly
foregone future
choice between
strategy. Within
rational for
the
the cooperative
noncooperative
the agents
to keep
it
the sanction
is.
thus.
costs down
by
has also
been advantageous
for the
lending
respect to
as the
stability of
was crucial to get time for adjustment (Sachs 1982 ; Grauwe and
Fratianni 1984
In
the
creditors move
strategy?
the
Obviously it
situation.
how
noncooperative
can
to
debtors
the
and
cooperative
governments need
to find
ways
which
make
370
investments more
may infer,
profitable. From
first,
investments has
that
an opening-up
of course,
was much
uses than
guaranteed debt
debt. Second,
may increasingly
be
The
shirking-problem
coverage
case
of
failure
incentive is
set towards
first place.
Third, and
the efficiency
to be
put into
converted
into
associated
would
be
publicly
private
with
reduced.
public
Thus,
an
public management
these suggestions
chance that
public and
of investment
be enhanced.
much as
direct
obligations.
need to
towards foreign
more productive
in
analysis one
Debt-for-equity swaps,
since there
the preceding
are taken
In as
banks an
will
increase in
lower
the
the
required
tradeability
(future)
return
of
on
The
Bayesian
presented in
lower in
this case.
easier to
approach
Second, by
diversify at
lowers the
to
sovereign
loss
premiums
least a
required rate
trade it
is
of return
and,
hence,
lowers
the
be
may increasingly
lending. Both
since the
lending may
lower the
substituted
required premium
principals are
in a
for
unconditional
the
related
if the
devices
outlined
have
merits,
the
(25)
For
a detailed discussion on the
sovereign risk management see Picht (1988).
possibilities
of
371
a third
capital
investment at
(27).
structure
high rather
inflows. The
advanced by
effectively protect
support the change
in
favor
the World
may
Bank is
bonding
serve
of
than shrinking
foreign
levels
direct
of
activities
the
foreign
on the
capital
Guarantee Agency (MIGA)
Multilateral Investment
Voluntary
governments
which
would
same
on
the
purpose,
part
since
of
LOG
brand-name
be crucial
behavior of
LOG governments.
get integrated
imports, the
welfare
the
back
towards
cooperative
countries and
world
economy,
the
lower
is
the
potential for
credit is
gains
move
from industrialized
The analysis
into
sovereign risk.
to future
for the
both investors
from
increased. Another
liberalized
trade
reason
may
is
be
that
assumed
the
to
372
financial system
by opening
their economies
LDes.
APPENDIX
1. Variable List
AIDeRQ
AND
BOP
Balance of payments deficit divided by the threeyear moving average of balance of payments deficit.
Source:
IMF, Balance of Payments Yearbook, various
issues.
DGDPKA
DUM
D\/RY
DZKP
EXPR
EXPVAR
GDL
Reciprocal
of either public or publicly guaranteed
debt, outstanding and undisbursed, in nominal US $.
Source:
The World Bank, World Debt Tables, various
issues.
GDPVAR
Variation
average
coefficient
(in prices
of GDP-five-year
of
1980).
Source
moving
IMF,
373
International Financial Statistics Yearbook. various
issues.
IDBSR
Reciprocal
of gross
investments
(in prices of
1980). in millions.
Source
IMF.
International
Financial Statistics Yearbook. various issues.
IMPQ
Imports
per unit
of GOP.
Source
IMF.
International Financial Statistics Yearbook. various
issues and World Bank 1985.
LIBOR
London
Interbank Offer Rate on US deposits
(3months. deflated by GOP price index (US. Source:
IMF. International Financial Statistics Yearbook.
various issues.
POP
Reciprocal
of population.
in mil~ions.
Source
IMF. International Financial Statistics Yearbook.
various issues.
RSCDG
Reschedulings either in
Club. or both. Source
1986) .
RR
WRGDY
WRYM
2. Note
on the
Estimation Procedure
for the
Aggregate
Real
marginal rates
country. First.
estimated and.
to the
had to
the real
of return
net capital
stock K(t)
had
to
be
capital stock
calculated as
be solved
were required.
RR(t)
could
then
been
374
stock K(t).
In order to get an estimate of the real capital stock
available for
productive use
in an
the insight
theory that
macroeconomic variables
product,
gross domestic
effective
labor
progress) all
stocks grow
the same
real investment,
money
balances,
Harrod-neutral
rate in
and
technical
steady states.
It is
same rate.
steady-state rate,
period t,
real
(assuming
supply
grow at
important in
savings,
such as
what the
size of
the underlying
the variation
coefficient (standard
growth rates
of the
variables was
by the
deviation/mean)
three-year-moving averages
of
for
the
the
four
data availability
International
Financial Statistics
the
estimate
capital
stock
available, all
other capital
adding (sub-tracting)
for
the
stock
the net
reference
data
(total)
were
year
was
generated
investments
for
by
each
procedure
comparing the
coefficient
suggested
estimates on
with
the
was
the basis
estimates
first,
checked,
of the
based
on
best
the
by
variation
second
best
did not
external check
differ
much
from
one
another.
An
partially possible,
second,
countries)
some of
included
them net
capital stock
comparison showed
that the
results.
given
Hence,
estimates are
because
the
estimates
procedure
that
not available
sample
for
at all,
are
suggested
most
LDCs
(fifty-four
available.
yields
capital
the procedure
used
good
stock
was
considered as appropriate.
In the second step, the estimate capital incomes P(t)
were estimated.
Data for
the operating
surplus OS,
which is
375
defined as the excess of value added over the sum of (domestic)
compensation of
employees,
fixed capital,
consumption of
(UN, Yearbook
of National
issues). The
were taken
operating surplus
income which,
of course,
employed, and
function of
as a starting point
corresponds to entrepreneurial
relates to
persons
incomes.
Basically,
to property
the Cobb-Douglas
1, various
type was
who
are
self-
production
were run where the natural logarithms of OS(t), K(t) and of the
proxies
for
available)
the
number
were
of
entered
Industrial Statistics,
self-employed
as
Vol.
variables
1,
persons
(UN,
(where
Yearbook
various issues).
of
In the actual
assumption (Cobb-Douglas)
cases it
was necessary
to run
was not
imposed.
regressions
using
In
some
the
first
In
can directly
the present
(based on
OS) were
be
interpreted
income
of interest.
Generally,
as
the
regressions
actual OS
share estimated,
one directly
level of
interest and
the real
marginal rates
were derived
figures, and
by dividing
on the
in
the
the
magnitudes derived
discard the
inclusion
sample
of
made
it
major
plausible
industrialized
to
check
the
method applied,
i.e., the
derived
results
were
plausible.
Finally,
are largely
it
External Relations
Fund (Khatkhate
in the
was striking
Department of
1985). For
present study
correlation coefficient
the
International
Monetary
a direct
comparison was
between the
possible.
The
It
376
is significant
upward or
at the
downward bias
procedure in
comparison to
the
estimates
generated
by
the
REFERENCES
Altman, E.J.,
R.R.
(1981),
Avery,
R.A.
Application
Business, Banking
Economic
of
Eisenbeis
and
Classification
and Finance.
J.F.
Sinkey
Techniques
in
Contemporary Studies in
: JAl Press.
Basche, J.R.Jr.
(1986) ,
K.E.
"Erfahrungen
aus
Erfahrungen.
Socialpolitik, vol.
internationalen
Ursachen - Konsequenzen -
Schriften
des
Vereins
& Humblot,
fur
pp.
9-29.
Burton, F.N.
and H.
Owned Firms
Inoue (1984),
in Developing
and H.
of
"Expropriations of Foreign
Countries", Journal of World
18(5).
Inoue (1986),
Foreign
Asset
"A Country
Expropriation
Risk Appraisal
in
Developing
"International Debt:
377
(1986),
Claassen, E.H.
"Die
Funktion
des
Kreditgebers
del'
Konsequenzen
des Vereins
fur
Historische
Socialpolitik,
World Economy.
Institute for
International
Economics.
Economics n 4, London
(1981),
"Beyond
Political Risk
the Earthquake
Allegory: Managing
L.
(1980),
"The Economics
of Property
Rights.
Eaton, J.
1-47.
and
H.
Gersovitz
Repudiation
(1981a),
Theoretical
"Debt
and
with
Potential
Empirical
Analysis",
Private Financial
"Poor-Country Borrowing in
Markets and
the Repudiation
Issue",
E.F.
and
H.C.
Jensen
(1983),
"Agency
Problems
and
and V.
Regev (1975),
Foreign Investment,
"International Loans,
and Optimal
Direct
Capital Accumulation",
I.K.
and
R.L.
Feierabend
(1966),
"Aggressive
378
Behaviors within
Study", The
10(3),
pp. 249-71.
Folkerts-Landau, D.
Bank
Lending
in
Development
Finance"
International
and
V.R.
Cline
Servicing Capacity
Analysis",
Frey.
An
Journal
1(3),
~p.
B.S.
(1985),
(1971),
of
"Measurement
Application
of
International
of
Debt
Discriminant
Economics,
vol.
327-44.
Internationale
Politische
Okonomie.
P. and
H. Fratianni (1984),
International
Research Paper
Lending",
n012,
International
Katholieke
Economics
Universiteit
Leuven.
(1968),
T.
"A
Causal
Comparative Analysis
Model
Using New
of
Civil
Indices", The
Strife
American
1104-24.
International
Lending.
Studies
in
International
J.A.
(1974),
Lending", The
"Optimal
International
Borrowing
and
616-30.
Hauser, H.
(1986),
"Promotion
Developing Countries
Aussenwirtschaft,
Hicks.
U.
(1978),
Exercise
Federalism
Comparative Study.
in
Cooperation",
London and
Failure
and
Basingstoke
Success.
MacMillan
379
Press.
IHF,
and W.H.
Managerial
Heckling (1976),
Behavior,
Structure", Journal
Agency
of Financial
and
Ownership
pp. 305-60.
Jodice, D.A.
(1985), Political
Bibliography.
Westport
Risk Assessment.
(ET)
and
An Annotated
London
Greenwood
Press.
Jones, J.Jr.
(1980),
"A
Latin America
(1984),
"Empirical
of International
Business
Studies,
vol.
15
(Spring/
Khatkhate, D.R.
Interest Rates
Relations
"Assessing
in Less
Department.
the Level
and
Impact
Developed Countries",
International
of
External
Monetary
Fund.
(Document DM/85/8).
Kindleberger,
C.P.
(1958)
Economic
Development.
New
York,
London: McGraw-Hill.
Knudsen,
H.
( 1972) ,
Expropriation
Investments
in Latin America.
~~~~~~~~~~~~~.
of
Foreign
University of
Private
Oregon
Ph.D. Dissertation.
Kobrin, S.J.
(1978),
Increased Investment
World Business, vol.
Risk 7"
13(3), pp.
The Columbia
113-22.
Journal
of
380
Kobrin. S.J.
(1980),
in LDCs" ,
International Organization,
vol. 31(1),
pp.
65-88.
U.
Lachler.
(1985),
"Debt
Versus
Equity
in
Development
B.
Lloyd.
(1974),
"The
Political Risk
Identification
in the
and
Assessment
of
(1974) "Conflicts
Between Host
Hodigliani. F.
and H.H.
Hiller (1958),
Corporation Finance,
(1987), "Searching
Extension", The
pp.
American Economic
Comment
Review, vol.
and
77(1),
198-204.
J.
Niehans,
for Leviathan
(1986) ,
"Internationale
undurchsetzbaren Forderungen" ,
Internationale Schuldenkrise.
Historische
Erfahrungen.
Socialpolitik, vol.
In A.
Kredite
mi t
Ursachen - Konsequenzen -
Schriften
des
Vereins
fur
151-79.
OECD
Geographical
Distribution
of
Financial
Flows
to
(1988)
"Sovereign
Diversification", Kiel
Loan
Exposure
Institute
of
and
World
Portfolio
Economics,
and D.
Rubinfeld (1981),
Econometric Models
and
381
Forecasts, Second
Edition, New
McGraw-Hill Book
York
Company.
(1971>,
Robock, S.H.
"Pol itical
Assessment", The
Identification
Risk
Columbia Journal
of
World
and
Business,
(1968),
"The
Companies. What
Horizons, vol.
Sachs. J.
(19821,
In P.
Expropriation Experience
happened
to
38
of American
companies",
Business
"LDC
Structure. Lexington,
Toronto:
Lexington
Books,
pp.
197-243.
Sachs.
(1983),
J.
"Theoretical
Borrowing", Princeton
Issues
Studies in
in
International
International Finance
n 54.
Sachs. J.
and D.
Cohen (1984),
"LDC Borrowing
with
Default
K. G.
and
Bates
Ph.S.
Quantitative
(1984),
to
Approaches
"A
Country
Survey
Risk
of
the
Analysis",
Schlaifer. R.
Decisions. New
Introduction
Company.
Schneider. F.
and
Determinants
B.
of
Development, vol.
Sharpe. V.F.
Frey
(1985)
Foreign
13(2), pp.
(1970) Portfolio
"Economic
Direct
and
Political
Investment",
World
161-75.
(1974),
"Foreign
Investment in
382
U.S.", The
Columbia Journal
of
World
Business,
vol.
Shihata, J.
Multilateral
"The
Role
of
Investment
JSID
(1972),
"Some
Corporate Finance
the
Projected
Agency
(MIGA)",
and
Guarantee
105-22.
Aspects of
the Pure
: Bankruptcies
Theory
of
(1982),
An Analytical
"Endowments,
Scheme for
Country Policy
J.F.
(1974),
Investment,
Expropriation
of
Private
Foreign
UN Yearbook
of National
Accounts Statistics,
vol.
I, Various
Issues.
Wallich, H.C.
(1986),
"The
American View
In A.
Gutowski, ed.
Ursachen
Schriften
M.,
R.
Die Internationale
Konsequenzen
des
Vereins
Berlin: Duncker
Watson,
Historische
fur
Erfahrungen.
Socialpolitik,
vol.
155,
Kincaid.,
C.
Folkerts-Landau (1986)
Developments
Schuldenkrise.
and
Atkinson,
E.
Kalter
"International Capital
Prospects",
International
and
D.
Markets.
Monetary
M.L.
(1975)
Nationalization of
"The
Extent
and
Foreign-Owned Assets
Significance
in
of
Developing
Countries
1956-1972",
Oxford
Economic
Papers,
vol.
Bank (1986),
World Debt
Developing Countries.
Tables. External
Debt
of