Download del02-gros 221137 en

A joint Initiative of Ludwig-Maximilians-Universität and Ifo Institute for Economic Research
CESifo-Delphi Conferences:
Managing EU - Enlargement
Delphi Conference Centre, Delphi
13-14 September 2002
Who Needs an External Anchor?
Daniel Gros
CESifo
Poschingerstr. 5, 81679 Munich, Germany
Phone: +49 (89) 9224-1410 - Fax: +49 (89) 9224-1409
E-mail: [email protected]
Internet: http://www.cesifo.de
Preliminary and incomplete
g:\daniel\conf/Munich anchor ifo.doc\2 September, 2002
Who needs an External Anchor?
by
Daniel Gros*
Abstract
It is well known that countries with high public debt, or very weak fiscal institutions, might
benefit from an external anchor to save them from the high inflation rates they would otherwise
have to endure because the market knows that the temptation for them to use surprise inflation
to reduce public debt service is so strong. This paper argues that the relationship between the
strength of the domestic framework for fiscal policy and the interest of a country to use a peg to
achieve price stability might be non-monotonic. Both countries with a very strong domestic
framework, e.g. low public debt, and those with high public debt might benefit from an exchange
rate peg. By contrast, countries with moderately weaknesses might be in a situation where they
would prefer to keep a national currency because they need some inflation to supplement
government revenues with seigniorage, but the inflation resulting from the interaction with the
market, which knows about this, is still moderate.
Applied to the enlargement process this implies that both countries with very strong and those
with very weak institutions might have an interest in joining the euro area quickly. The choices
of Estonia (very strong) and Bulgaria (very weak), to adopt the euro/DM via currency boards,
seem to reflect these considerations.
Prepared for
CESifo-AUEB conference on
"Managing EU Enlargement »
Munich December 4/5, 2001
*
Centre for European Policy Studies (CEPS) Brussels. Many thanks to Anna Maria Pinna and Peer Ritter for
important research assistance.
1
I.
Introduction
It is well known that highly indebted countries, or countries with weak fiscal institutions, can
fall into a low credibility trap. This occurs when a government loses credibility in the eyes of the
financial markets and is forced to pay a risk premium in the form of higher interest rates. The
higher debt-service burden that results, if inflation is kept low, makes it even more likely that
the authorities will abandon efforts to stabilise the situation and attempt to reduce the real value
of the debt through a surprise inflation. This further increases the risk premium demanded by
financial markets and can lead to a spiral of increasing interest rates until the government caves
in and produces the inflation the market expects.
A country in such a situation has an interest in using an external anchor. For the Central and
Eastern European countries the obvious candidate to be such an anchor would be the euro given
that most of their external trade is with the euro area. Some countries have already de facto
joined the euro area by linking their money via a currency board to the DM, or, now, the euro
(e.g. Estonia, Bulgaria and Lithuania). For all the others, which are also candidates for
membership in the EU, a key question that remains on the table is when to enter the euro area as
well. The orthodox approach to the issue of EMU membership is that countries should converge
gradually first and can join the euro area only once they satisfy the Maastricht criteria on
inflation and public finance. However, this approach misses the point that the weaker countries
would actually gain more from joining EMU than countries that have already achieved price
stability on their own. Fulfilment of the Maastricht criteria can be reasonably required of
countries that want to have a seat on the Governing Council of the ECB, but this should not
prevent weaker countries from considering to adopt the euro unilaterally, either via a currency
board (as in Estonia and Bulgaria) of via full euroisation under which the domestic currency is
fully substituted by euro notes and coins (which becomes feasible in 2002) as discussed in
Emerson and Gros (1999).
The purpose of this note is to show that the standard model, that is used to explain why a
country with a credibility problem might benefit from an external anchor, can actually lead to
the result that the relationship between the strength of fiscal institutions (in the sense of their
ability to sustain price stability) and the interest in joining the euro area is not monotonic. The
very strong might benefit (as assumed under the orthodox convergence approach), but the very
weak would also benefit and should thus contemplate a different, unilateral approach.
2
The standard optimum currency area approach (OCA) stresses different considerations, for
example the importance of asymmetric shocks. This paper shows that the standard OCA
approach can be thought of as constituting a base line: countries with fewer asymmetric shocks
1
have a stronger interest in pegging the exchange rate. The public finance arguments stressed
here constitutes an additional consideration that should also be taken into account. It remains
true that, ceteris paribus, i.e. for the same level of asymmetric shocks, both countries with very
high or very low levels of public debt might have an interest in pegging to an external anchor.
The remainder of this paper is structured as follows: the next section presents the model. This is
followed in section III by an examination of the equilibrium. The relationship between the
interest in an external anchor and the level of public debt is then discussed in section IV. Section
V describes briefly how an inefficient fiscal policy, which allows special interest groups to gain
at the expense of society can lead to excessive inflation and shows how the degree of
inefficiency influences the interest of the country in adopting the euro. Section VI concludes.
II.
The Model
The model used here is entirely conventional. The starting point is a standard social loss
function, Lt, given by:
(1)
[
Lt = αqt2 + pt2
]
α ≥0
where pt stands for the inflation rate and qt stands for tax revenues as a percentage of GDP,
which is equivalent to the average tax rate. High taxes and high inflation create distortions and
are thus socially costly. The parameter α indicates the relative weight of taxes in the social loss
function. A high α could be interpreted to mean that the tax collection system is not efficient,
i.e. that it causes high distortion costs for a given revenue. The experience in Central and
Eastern Europe has shown that there are indeed great differences in the ability of different
countries to raise taxes. In Russia, to take an extreme example, until recently, the government
was not even able to raise 15 % of GDP, whereas in Estonia government revenues amount to
1
This is the standard hypothesis, Berger, Jensen and Schjelderup (2001) make the point that
the covariance of shocks (domestic and foreign) might also be a key element under the
standard OCA approach.
3
over 30 % of GDP2. At this point it is assumed that the government reflects accurately the
preferences of society in setting taxes and inflation.
The authorities can determine inflation via their control over the money supply (or to be more
precise its growth rate), mt, but the control is not perfect as money demand is subject to shocks,
µt:
(2)
p t = mt + µ t
With E(µt) = 0 and variance σ2m.
The aim of the authorities (as usual, no distinction is made between the central bank and the
Ministry of Finance) is to minimise the social loss, subject to the expected budget constraint:
(3)
d (bt ) = g t + bt (it − mt ) − qt − mt ξ
where bt is the public debt/GDP ratio and gt represents (non-interest) expenditure relative to
GDP. The last term in this budget constraint represents seigniorage revenues under the
assumption of a constant velocity money demand function with the cash (or rather monetary
base) to GDP ratio constant and denoted by ξ. The constant velocity assumption implies that
seigniorage increases linearly with the money supply (and hence expected inflation). This is not
realistic, but this assumption was chosen in order to show that the results do not depend on a
‘Laffer curve’ for seigniorage revenues under which the revenues from the inflation tax fall with
very high inflation rates as money demand goes towards zero.
In countries with moderate inflation seigniorage revenues can reach up to 2-3 % of GDP, or,
between 5 and 10 % of overall public sector revenues. Seigniorage can thus make a significant,
contribution to public sector revenues.3 Table 1 below shows some crude estimates of
seigniorage revenues, defined as in the model, as the ratio of the increase in the monetary base
to GDP, for the average of the period 1995-2000. There are evidently wide differences between
2
An alternative interpretation would be that society and/or the politicians in power dislike high taxes (for example,
because their marginal voter is a household with a high marginal tax rate).
3
For an analysis of the importance of seigniorage revenues in transition countries and the relative stability of
money demand see Gros and Steinherr (1995) or Gros and Vandille (1995).
4
the candidate countries, with values ranging from 5.1 % of GDP for Bulgaria (with peaks during
1996/7 of over 10 % of GDP) to only 0.7 % for Slovenia. The value for the eurozone, reported
for comparison is, at 0.3 % of GDP, even lower. Seigniorage is thus definitely more important
for the candidate countries than in the eurozone.
Table 1, Seigniorage and domestic debt
Bulgaria*
CR
Estonia
Hungary
Latvia
Lithuania
Poland
Romania
Slovakia
Slovenia
Cyprus
Malta
Turkey
Eurozone**
Seignorage (change in monetary base in
period 1995 - 2000) as % of GDP
5,1
1,7
2,3
1,6
1,7
0,8
0,8
3,4
1,3
0,7
1,0
0,8
3,1
0,3
Domestic debt (line 88a) as % of GDP
60,2
13,2
0,5
47,7
3,0
6,3
19,7
12,0
17,6
n.a.
43,2
52,5
27,6
Source: Own calculations based on IMF, IFS data. * Average for individual years ** Only for
1999 and 2000
Table 1 also shows the importance of domestic debt. The debt to GDP ratio which plays a
central role in the model refers only to domestic debt because foreign (or rather foreign
currency) debt cannot be devalued by domestic inflation. The last column in table 1 shows that
there is clearly still a lot of domestic public debt in candidate countries to make the model
potentially relevant.
Another real world qualification is that it is implicitly assumed that the monetary base consists
only of cash. Introducing required reserves on commercial banks (which could be remunerated)
would not change the thrust of the analysis. If required reserves are not remunerated (which is
usually the case) the value of ξ would just be somewhat higher.
For the sake of simplicity, real growth is assumed to be zero. Government expenditure could be
made endogenous, as in a number of other contributions on the optimal choice of taxes and
inflation, but this has not been done here as it would not affect the main results of the paper,
5
which concentrate on the incentive to use surprise inflation to reduce the real value of the public
debt.4
The crucial point about the budget constraint (3) is that real interest payments, given by bt(it mt), are a function of the difference between the nominal interest rate and money supply because
the latter determines the expected future price level. The simple form of the budget constraint
used here assumes implicitly that all government debt has the same maturity, equal to the length
of the period of this model. Another interpretation would be that b represents only the
government debt that matures in this period. Interest payments on other government debt would
then be subsumed under general government expenditure. This is not a serious limitation of the
model since most emerging market countries have a relative short average duration of debt (and
the little long-term debt that exists is indexed on short-term interest rates).
The nominal interest rate, it, can be written as the sum of the real interest rate, rt, required by
investors and their expected inflation. The latter will be equal to Et(mt) by equation (2)5.
(4)
it = rt + Et (mt )
The real interest rate demanded by financial markets is here assumed (as usual) to be exogenous
to the country concerned, but not necessarily constant. It is affected by shocks to the world
financial markets, denoted by ετ, with, E(rt) = ρ and variance σε2. This shock is observed by the
financial market and by the authorities at the same time, when they converge on an interest rate
on public debt.
(4)’
it = ρ + ε t + E t ( mt )
The authorities can thus use monetary policy and also the tax rate to accommodate interest rate
shocks. Taking into account (4)’ the expected budget constraint that the authorities have to
observe can be rewritten as:
(3)'
4
d (bt ) = g t + bt ( ρ + ε t + Et (mt ) − mt ) − q t − mt ξ
See, for example, Mankiw (1987).
6
III.
Equilibrium under discretion
The authorities have to take their decisions before µ occurs. They try to minimise each period
the expected loss while observing their budget constraint. The F.O.C. for a minimum of (1),
subject to (3)' are:
(5)
∂Lt
= 0 = 2αq t − λ t
∂q t
(6)
∂Lt
= 0 = 2mt − λ t (bt + ξ t )
∂mt
Where λt is the shadow price associated with the budget constraint (3)'.
In order to simplify the notation only the steady state will be considered with a constant
debt/GDP ratio, denoted b.6 Conditions (5) and (6) then yield a simple relationship between
money supply and tax revenues (as a percentage of GDP):
(7)
mt = (b + ξ )αqt
This can be substituted into the budget constraint (3)' to obtain an expression for the steady state
"tax rate". If one assumes that the public anticipates monetary policy and hence inflation
correctly (remember that the interest rate shock is observed by the authorities and the financial
markets at the same time), Et(mt| ετ) - mt = 0. The debt-to-GDP ratio remains constant on
average only if:
(8)
qt = g t + b(ρ + ε t ) − ξ (b + ξ )αq t
If expenditure is constant at g, the optimal tax ratio changes only if the realisation of the interest
rate shock is different from zero7 :
4
5
With pedantic notation, it would be Et(mt| ε)
In a similar model, Gros (1990) shows that this should not affect the conclusions.
7
(9)
qd =
[g + b(ρ + ξ t )]
1 + ξ (b + ξ )α
The expected loss under discretion E(Ldisc ) would then be given by:
{[g + bρ ]
}
2
(10)
E ( Ldisc ) =
+ b 2σ ε2
2
{α + [(b + ξ )α ] } + σ µ2
[1 + ξ (b + ξ )α ]
As usual the discretionary equilibrium is not the first best for the country. The social optimum,
if there were no constraints on credibility, can be calculated by using the first order conditions
(5) and (6), but without the effect of surprise inflation on debt service in equation (6). This
means that in the social optimum the relationship between taxes and money supply would be
given by:
(11)
m SO = ξαq
Which differs from the corresponding relationship (7) in that only seigniorage is a valid
argument for having inflation. The optimum tax rate is then given by substituting this
expression into the budget constraint (and setting the debt-to-GDP ratio constant), which yields:
(12)
q SO =
[g + b( ρ + ε t )]
[1 + ξ α ]
2
The expected loss under the social optimum, E(LSO), would then be equal to:
 [g + bρ ] 
b 2σ ε2
2
α
αξ
E ( L SO ) = 
+
+
+ σ µ2
(
)

2
2
2
2
1 + ξ α [α + (αξ )]
 1+ξ α 
2
(13)
[
] [
]
[
]
which is lower than the loss under discretion. However, in this set-up there is no way the
country could reach this bliss point. Without an external anchor the country would be stuck at
the discretionary equilibrium.
7
Since this paper's focus is only on steady states, the time subscript will henceforth be suppressed.
8
The implicit assumption behind the limitation to the alternative: discretionary equilibrium or
pegging to an external anchor is that the institutional set-up of the country is such that it would
be impossible to establish an independent central bank that would both be credible in financial
markets and be able to react flexibly to shocks by expanding the money supply whenever
required by an adverse external shocks (and to do so in conjunction with the finance ministry,
which would have to increases taxes (see equation (12). This paper does not make any
distinction between various forms of using an external anchor (pegging the exchange rate,
instituting a currency board or full adoption of the foreign currency). For a discussion of the
choice between pegging via a currency board and an independent central bank applied to the
Baltic experience see de Haan et. al. (2000).
IV.
Welfare effects of the euro
The alternative to the discretionary equilibrium might be to join the euro area. This would
eliminate the credibility problem and would not be subject to the problems of speculative
attacks that arise in fixed exchange rate systems.8 In the EMU area prices would be determined
via purchasing power parity through the monetary policy of the ECB. It will be assumed that the
ECB maintains perfect price stability. Its independence is not in doubt and it has a clear mandate
to maintain price stability. Moreover, seigniorage is not an important source of revenue for most
EU member countries, which implies that the social optimum for the EU is anyway to set
inflation to zero.
Domestic prices will then be determined by a stochastic purchasing power condition:
(14)
p t = et
where et is the shock to purchasing power parity, subject to E(et) = 0 and variance σppp 2. (In this
case the domestic money market adjusts passively and the shocks to the domestic money market
do not matter any more.)
8
Creating an independent central bank and giving it the task to maintain price stability would in principle be
equivalent. Almost all CEECs now have independent central banks. However, experience has shown that their
9
The loss under EMU membership would then be:
(15)
Leuro = α (g + bρ ) + αb 2σ ε2euro + σ 2ppp
2
Joining the euro area would thus be advantageous if this loss is lower than the one under
discretion, i.e. when the country does not have an external anchor for expectations. Adopting the
euro has a number of consequences that are apparent once one compares equations (15) and
(10). The money demand shock is no longer relevant under the euro regime, because shocks to
domestic money demand in a small participant country do not affect the euro area price level.
But pegging to (or fully adopting) the euro instead of having a national currency introduces
another source of shocks, namely the shock to PPP.
Finally, as the notation σε2
euro
suggests, it is likely that a tight pegging to a stable currency
affects the risk premium on foreign debt. In particular, the large size of the euro area suggests a
lower variance of shocks than for each of its constituents. Moreover, for countries with large
external debt euroisation may lower the risk premium paid on foreign debt. Euroisation
eliminates the difference between external and internal debt, removing the government's
incentive to discriminate against foreign creditors. In addition, euroisation mitigates uncertainty
about a country's external debt service capacity. When a country devalues in reaction to an
external shock, the capacity of the government to service foreign debt is usually much reduced
because a large part of government revenues comes from taxes on non-tradable activities. A real
depreciation thus reduces also the value of tax revenues in foreign currency. For example, when
the price of dollars in terms of roubles increased fourfold during the Russian crisis of 1998, the
debt service capacity of the Russian government was initially also cut to one fourth because
domestic prices did not jump along with the exchange rate.
All these arguments suggest that σ ε2 euro= φσε2 disc. with φ likely to be smaller than one. The
difference between the losses under EMU and the discretionary equilibrium can then be
written as:
independence is not strongly enough anchored in either the constitution or the political reality to allow them to really
achieve price stability.
10
(16)
[
]
{
Leuro − Ldisc = α 2 (g + bρ ) + b 2σ ε2disc. [1 + ξ (b + ξ )α ] (b + ξ ) (ξ − b ) + ξ 2 (ξ + b )α
2
−2
}
− (1 − φ )αb 2σ ε2disc. + σ e2 − σ µ2
To interpret this result, we will first discuss the role of the debt-to-GDP ratio (b) and then the
role of the shocks.
Regarding the effect of the debt-to-GDP ratio, let us take a look at the term in curly brackets.
This shows immediately that the difference in welfare loss can be negative or positive
depending on the sign of (ξ − b), which is likely to be negative. The tax basis for seigniorage is
the monetary base, ξ, which in most transition countries is below 10% of GDP. There is no
country in the world where the monetary base, or cash-to-GDP ratio exceeds 1; values much
below the average of 0.1 for the transition countries can be found in LDCs with chronic high
inflation problems, but even the most stability oriented countries never have a cash-to-GDP
ratio much in excess of this value. By contrast, the debt-GDP-ratio, b, shows much more
variability (in absolute terms) and is sometimes close to 100%.9 Estonia might be one of the few
countries without any significant public debt so that in this case (ξ – b) might actually be
positive.
To discuss the role of the debt ratio it is convenient to re-write equation (17) slightly as:
(17)
[
]
{(
) (
Leuro − Ldisc = α 2 (g + bρ ) + b 2σ ε2disc . [1 + ξ (b + ξ )α ] (b + ξ ) ξ 1 + αξ 2 − b 1 − αξ 2
2
−2
)}
− (1 − φ )αb 2σ ε2disc. + σ e2 − σ µ2
Inspection of equation (17) shows that the term in curly brackets must change sign as b
increases from zero to infinity as long as the term (1 - αξ2) is negative (which is likely as argued
above). This suggests that the relationship between the interest in joining the euro area and the
debt to GDP ratio should be non monotonic. However, as the debt ratio appears also in other
terms of equation (18) it is not straightforward to sign the derivative.
However, one point is clear from general considerations: When there is no public debt, staying
outside the euro area is clearly better because the discretionary equilibrium would then be equal
9
In order to highlight the role played by the interest burden on public debt, Gros (1996) assumes that ξ = 0. This
elimination of seigniorage from that model was justified by the fact that seigniorage has not played a significant role
in EU public finances in recent years, as shown by Gros and Vandille (1995).
11
to the social optimum and by staying outside the country can use inflation to earn some
seigniorage. I.e. for b=0 the value of the RHS is clearly positive, as can be established by
inspection. Moreover, there is clearly also one value of b for which the term in the curly
brackets in equation (17) is equal to zero and hence the country is indifferent between joining
the euro and staying outside. This is at: b = ξ(1 + αξ2) / (1 - αξ2 ), which is larger than the cashto-GDP ratio, and thus a realistic value. Furthermore, it is also clear that for b larger than this
value the difference in loss becomes negative. At this point al that remains to to show the
‘hump’ shaped relationship between the difference in loss and b is that the derivative of this
expression with respect to b is positive around zero. This involves some tedious algebra. Annex
shows in detail that this is indeed the case.
Having established the general presumption for a hump shaped it might be useful to present the
relationship between the difference in welfare loss and b. This is done in figure 1 for a particlar
set of parameter values g (the ratio of non-interest expenditure to GDP) = 0.3, ξ (monetary base
to GDP ratio) = 0.1, φ=1, and two values for α). It is apparent that there is an inverted U-form
relationship between the interest in joining the euro area and the debt to GDP ratio. For
countries with low debt the loss in the euro area is larger than that from going it alone. They
have an interest in staying outside, whereas very weak countries have a larger loss outside and
should thus gain from joining the euro area.
The intuition behind this result is clear: for low levels of debt the incentive to use surprise
inflation is low. The market knows this and therefore expects only low inflation. It is thus
possible to have an equilibrium with inflation only little above what would be needed to get
some seigniorage. However, in countries with high levels of debt the inflation rate would be
much higher than the one that would be justified for optimum seigniorage because the market
knows that the government has an overwhelming incentive to use surprise inflation.
Looking at the role of shocks, a first point to note is that neither the variance of the money
demand shock nor that of the PPP shock interact with the debt-to-GDP ratio. Their relative
magnitude just shifts the curves in figure 1 vertically. The standard optimum currency area
approach stresses only the size of the variance of the PPP-shocks. However, in order to make
any statement about the loss from adopting an external anchor on has to compare two sources of
shocks: the PPP shocks and the money demand shocks that would complicate monetary
12
management if the national currency is kept. What matters is thus difference between these two
variances.
The variance of the interest rate shock interacts, however, with b. As the net expected loss from
giving up the national currency turns negative for high values of b, a higher interest rate
variance would reinforce the argument for adoption of the euro. On the other hand, for very low
levels of b, the size of the variance does not really affect the loss differential. Hence, for low
debt-to-GDP countries, interest rate variance does not constitute an additional argument to
retain their national currency.
Any reduction of the interest rate variance from adoption of the euro (i.e. assuming that φ is
smaller than one), would of course reinforce the argument in favour of its adoption more, the
higher b is. In figure 1 it was assumed that φ is equal to 1. Assuming that φ is smaller than one
would lead to a new curve that would lie lower. The distance with the curve which assumes φ
=1 would increase with the square of b.
13
V.
Inefficient fiscal policy institutions
This section introduces the concept of an inefficient fiscal policy process into the model. It has
so far been assumed that the government takes its decision in terms of the overall level of
taxation needed to finance expenditure. However, there are always special interest groups that
plead for exemptions. Any tax exemption that is granted must be paid for somehow. In this setup the only alternative source of revenues is inflationary finance. Using inflation also causes a
welfare loss to the special interest group that obtains a tax exemption. However, as the part of
any group in the overall budget will be small, this cost cannot fully offset the direct gain from
the tax cut. Each special interest group thus behaves as if the shadow price of a tax benefit were
only ηλ , where η is a fraction, between zero and one, which indicates the overall inflationary
impact any tax benefit has on the special interest group concerned. This fraction should be a
function of various elements, for example the share of the interest group in the overall budget,
the extent to which benefits have to be shared with other groups (a tax exemption for one
specific enterprise might not be possible, an exemption for all enterprises of a certain category
might be acceptable) and the extent to which fiscal policy decisions are centralised so that
demands by competing interest groups neutralise each other (see von Hagen and Harden (1994)
for an analysis of EU member countries in this respect). Velasco, (1998) uses a similar
approach with two symmetric interest groups whereas Drazen, (2000) presents (in chapter 10)
a model with a large number of competing groups, which try to extract transfers from the
government.
For simplicity all shocks are set to zero. The results of the previous section show anyway that
the introduction of shocks does not affect the central message of this paper. If the influence of
special interest groups is the same throughout all areas of fiscal policy and if all interest groups
are identical in terms of their size and influence this leads to a modified first order condition, see
equation (5), for the setting of the overall average tax rate:
(18)
∂Lt
= 0 = 2αq t − ηλ
∂q t
Setting the inflation rate hits all interest groups in the same way; the first order condition (6) is
thus not affected. However, the resulting trade-off between taxes and inflation is different:
14
(19)
pt =
(b + σ )αq t
η
This implies that, ceteris paribus, inflation will be higher as all interest groups push the
government to finance their benefits through inflation. As nothing changes in the remainder of
the model (i.e. essentially the budget constraint) the resulting tax rate under the discretionary
equilibrium is given by an equation that is identical to equation (9) above, except that α is
substituted by α/η. This implies that the welfare loss under discretion is given by:
(20)
Ldisc ,if


 [g + bρ ] 
=

1 + σ (b + σ )α 


η
2
2

 (b + σ )α  
α + 
 

 η
 
where the subscript stands for inefficient fiscal policy.
The loss under the euro is not affected by the inefficient policy as special interest groups are no
longer able to have their tax benefits financed by inflationary finance. The difference in the loss
between the euro regime and the discretionary equilibrium thus becomes:
2
(21)
Leuro − Ldisc


 [g + bρ ]  α 2 (b + σ ) 
σ 2α − 1 (b + σ )
σ
2
=
+



η
η


1 + σ (b + σ )α 
η


(
)
This solution collapses, of course, to the corresponding expression for the standard case, see
equations (16) or (17) if η equals one.
How does the difference in the loss expressed in equation (21) vary with the key parameter that
describes the strength of the fiscal framework of a country, i.e. η? As the inverse of η can vary
between 1 and infinity, it will be convenient to analyse equation (21) in this way. A high value
of η-1 indicates a weak fiscal framework.
Inspection of equation (21) shows immediately that there is one value of η-1 for which the
country will just be indifferent between joining the euro or staying out in the cold:
15
η-1ind.= -2σ / (b + σ) (σ2α - 1) (> 0 under the assumptions used so far). For values of η-1 above
this threshold the loss of staying outside would exceed that of joining the euro area. At this
threshold the country is just indifferent between joining and going it alone because two effects
just offset each other: if the country joins the euro area its inflation rate would be too low (zero
inflation would not be optimal since the country needs some seigniorage revenues), outside the
euro area its inflation rate would be too high (because of the inefficiency in the fiscal process,
which magnifies the usual time inconsistency problems).
Inspection of equation (21) also shows that as η-1 tends towards infinity the difference between
the welfare losses tends towards minus infinity (i. e. [α2 (ασ2 − 1) / σ] (g + bρ)2 . Basket cases
would thus have a very strong interest in using an external anchor.
The general picture is determined by the fact that equation (21) contains a quadratic expression
in the inverse of η and as long as the expression (σ2α - 1) is negative, which is likely given that
σ is much smaller than one, this will result in a inverted parabola.
Figure 2 shows the difference in welfare loss as a function of η-1 for two values of g (the ratio of
non-interest expenditure to GDP) assuming the monetary base to GDP ratio is equal to 0.1 and
the debt to GDP ratio is a Maastricht conform 0.6. The function seems convex and depicts an
inverted U-form relationship between the interest in joining the euro area and the strength of
domestic fiscal institutions. For countries with strong, but not perfect, fiscal systems the loss in
the euro area is larger that from going it alone. They have an interest in staying outside, whereas
very weak countries have a larger loss outside and should thus gain from joining the euro area.
VI.
The model and reality: the case of the candidate countries
The model developed here represents considerations that should inform the exchange rate
regime choice of any country. However, the theory seems particularly relevant for the
countries of Central and Eastern Europe which are candidates for membership in the
European Union (EU). Once they become members of the EU they will also become eligible
for membership in the euro area. While membership in the euro area is subject to the
Maastricht criteria, this implies that all the candidate countries have at least a medium run
perspective of being able to participate in the decision making for the monetary policy for the
euro. For these countries the political cost of abandoning national monetary sovereignty is
thus much smaller than for other emerging markets or weak economies around the world.
16
Do the choices of the candidate countries reflect the fundamental factors embodied int he
model presented here? This seems to be the case. Given the small number of observations one
has it is not possible to engage in an econometric exercise, but a quick glance at the data (see
table 2) shows that the countries with the highest and the lowest debt ratios opted for the
tightest exchange rate regimes (namely currency boards).
Table 2, Exchange rate regimes and public debt
Domestic as % of GDP
Bulgaria
CR
Estonia
Hungary
Latvia
Lithuania
Poland
Romania
Slovakia
Slovenia
60,2
13,2
0,5
47,7
3,0
6,3
19,7
12,0
17,6
n.a.
Exchange rate regime
(as classified by IMF 2001)
Currency Board
Managed Float (infl. targeting)
Currency Board
Crawling bands
Fixed peg
Currency Board
Floating (infl. targeting)
Managed float
Managed float
Managed Float (monetary target)
Source: IMF.
Another way to look at the relationship between public debt and the exchange rate regime is
provided in figures 3 and 4, which depict in each case the ratio of (domestic) debt to GDP on
the horizontal axis and on the vertical axis two different measures of the extent to which the
countries have used an external anchor. Figure 3 uses the variability of their currencies vis-àvis the euro10 and figure 4 shows the relationship between government debt and inflation. A
polynomial trend line inserted with a standard package gave in both cases immediately the
hump shaped relationship posited by the model.
10
Exchange-rate variability was measured by the standard deviation of month-to-month changes, in a given
year, in the logarithm of the nominal exchange rates of the respective currencies against the euro. In order to
eliminate possible excessive short-term disruptions the average value for the years 1999 to 2001 was used.
Inflation was measured by the CPI over the same period.
17
Domestic debt and ER variability
3
Nominal exchange rate variability (%)
2.5
2
1.5
1
0.5
0
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
50.0
60.0
70.0
-0.5
Domestic debt as % of GDP
Domestic debt and inflation
45.0
40.0
Inflation (average 99-01)
35.0
30.0
25.0
20.0
15.0
10.0
5.0
0.0
0.0
10.0
20.0
30.0
40.0
Domestic debt as a % of GDP
18
VII Concluding remarks
This paper has used on purpose a standard model to derive a result that is intuitively plausible,
but has not been recognised so far, namely that both very strong and very weak countries might
have an interest in adopting an external anchor. For the Central and Eastern European countries
this would imply joining the euro area.
The intuition is straightforward: countries with little pressure for excessive expenditure and an
efficient tax system, would anyway enjoy low inflation rates and therefore have no problems
with a peg to a hard currency. Countries with high debt, or very weak fiscal institutions, would
greatly benefit from an external anchor to save them from the high inflation rates they would
otherwise have to endure because the market knows that the temptation for them to inflate public
debt away is so strong. By contrast, countries with moderately weaknesses might be in a
situation where they need some inflation to supplement government revenues with seigniorage,
but the inflation resulting from the interaction with the market, which knows about this, is still
moderate.
This result is independent of the issues stressed in the usual optimum currency area approach
(OCA). It is clear that the likelihood of experiencing an asymmetric shock, and the gains from
transactions costs will be further elements in the choice of the exchange rate regime. However,
the OCA considerations are independent of the aspects discussed here. Moreover, the main
difference between Estonia and Bulgaria is not that the latter is much more exposed to
asymmetric shocks à la Mundell (1961), i.e. shocks to export earnings or terms of trade. Both
countries are vulnerable in this respect since they have a narrow industrial base. The really
important difference lies in the strength of the domestic fiscal framework: in Estonia the
government has so far been usually able to balance its budget. The tax service works reasonably
well, tax rates are low, but the tax base is broad and the (uncomplicated) tax code is actually
enforced. By contrast in Bulgaria the government has had always difficulties in finding enough
revenues to finance its expenditure and large state owned enterprises were a constant drain on
the budget.
The purpose of this contribution was just to point out that in an entirely conventional model the
relationship between debt levels and the need for an external anchor is non-monotonic. A high
variability of the risk premium in the international capital market reinforces this argument.
Starting from very low levels of debt, a high variability of the international interest rate might
19
initially make it more attractive to have flexibility in the exchange (and hence monetary policy)
in order to be able to finance increased debt service partially via the inflation tax. However, at
high levels of debt a shock to the international interest rate can cause such severe credibility
problems that pegging the exchange rate becomes again more attractive. The paper also shows
that a higher degree of inefficiency of the domestic fiscal policy process can be equivalent to a
higher public debt level. Pegging to a stable currency might thus be useful not only for countries
with very high or very low public debt, but also with very strong or very weak fiscal policy
institutions.
20
References
Berger, Helge, Henrik Jensen and Guttor, Schjelderup (2001) ‘To Peg or not to Peg? A
Simple Model of Exchange Rate Regime Choice in Small Economies’, Economic Letters, 73
(2001).
De Grauwe, Paul (1996) "The Economics of Convergence towards Monetary Union",
Weltwirtschaftliches Archiv, No. 132,1, pp 1-27.
De Haan, Jakob, Helge Berger and Erik van Fraansen (2000) ‘How to Reduce Inflation: An
Independent Central Bank or a Currency Board? The Experience of the Baltic Countries,
Emerging Market Review, 2(3), 2001.
De Kock, Gabriel and Vittorio Grilli (1993), ‘Fiscal Policies and the Choice of Exchange
Rate Regime’, The Economic Journal, 103, March, pp. 347-358.
Drazen, Allan (2000) Political Economy in Macroeconomics, Princeton University Press,
Princeton, New Jersey.
Eichengreen, Barry, Masson, Paul, Savastano, Miguel and Sharma, Sunil (1999) “Transition
strategies and nominal anchors on the road to greater exchange rate flexibility”, Essayes in
International Finance, N. 213, April, Dept. of Economics, Princeton University.
Emerson, Michael and Daniel Gros (1999), “The CEPS Plan for the Balkans”, CEPS
Paperback, Brussels.
Giavazzi, Francesco and Pagano, Marco (April 1989), Confidence Crises and Public Debt
Management, NBER Working Paper Series, Working Paper No. 2926.
Gros, Daniel (1990), "Seigniorage and EMS Discipline", Chapter 7 in Paul de Grauwe and
Lucas Papademos (eds.), The European Monetary System in the 1990s, Longman, London,
pp. 162-177.
Gros, Daniel and Guy Vandille (1995), "Seigniorage and EMU: The Fiscal Implications of
Price Stability and Financial Market Integration", Journal of Common Market Studies, Vol.
33, No. 2, June, pp. 175-196.
Gros, Daniel and Alfred Steinherr (1995), “Winds of Change: Economic Transition in Central
and Eastern Europe, Addison/Wesley Longman, London.
Gros, Daniel (1996), “Self-fulfilling Public Debt Crises”, CEPS Working Document No. 102,
also in “European Monetary Union, Transition, International Impact and Policy Options”, P.
Welfens (ed.), 1997, Springer, pp. 271-297.
Hochreiter, Eduard, and Tadeusz Kowalski (1999), ‘Central Banking Institutions in selected
European Emerging Market Economies’, Austrian Central Bank, Vienna, July.
21
Hochreiter, Eduard, and Ricardo Rovelli (1999), ‘The Generation and Distribution of Central
Bank Seigniorage in the Czech Republic Hungary and Poland’, Austrian Central Bank,
Vienna, July.
Mankiw, G.N. (1987), "The Optimal Collection of Seigniorage, Theory and Evidence",
Journal of Monetary Economics, Vol. 20, pp. 327-341.
Mundell, Robert (1961), “A theory of optimum currency area”, American Economic Review, 51,
pp. 657-75.
Velasco, Andres (1998) "The Common Property Approach to the Politcal Economy of Fiscal
Policy", chapter 8 in The Political Economy of Reform, edited by Federico Sturwenegger and
Mariano Tommasi, MIT, Cambridge, Massachusets.
Von Hagen, Jürgen and Ian Harden (1994) ‘National budget processes and fiscal performance’
European Economy, Reports and Studies, 3, Commission of the European Communities,
Brussels, pp. 310-418.
22
Difference in loss
Figure 1: Gain or Loss from Entering the Euro Are
Difference in loss as a function of the debt/GDP ra
40
30
20
10
0
-10
-20
-30
0
0.4
Debt/GDP ratio
23
0.8
Figure 2: Gain or Loss from Entering the Euro Area
Difference in social welfare loss as a function of alpha
Difference in social loss
250
200
Diffe
discre
150
100
Diffe
discre
s=0.1
50
0
-50
1 1.3 1.7 2.0 2.4 2.7 3.1 3.4 3.8 4.1 4.5 4.8
Inverse of fi (inefficiency of fiscal
institutions)
24
Annex: Derivative of equation (18) with respect to b around the value b=0.
To simplify the notation it is convenient to take the derivative of the logarithm of the difference
in the loss:
[
∂ ln (Leuro − Ldisc ) / ∂b = α 2 (g + bρ ) + b 2σ ε2disc.
{(
2
) (
+ (b + ξ ) − ξ 1 + αξ 2 − b 1 − αξ 2
−1
] [2(g + bρ ) + 2bσ ]− 2[1 + ξ (b + ξ )α ]
−1
−1
2
ε disc .
αξ
)} (1 − αξ )
−1
2
For b = 0 this reduces to:
[ ]
∂ ln (Leuro − Ldisc ) / ∂b = α 2 (g )
2 −1
(
)[ (
+ 1 + αξ 2 (ξ ) 1 + αξ 2
[2(g )] − 2[1 + ξ 2α ]−1 αξ
)] − {ξ (1 + αξ )} (1 − αξ )
−1
2
−1
2
Collecting terms yields:
[(
) ] [− 2αξ
∂ ln (Leuro − Ldisc ) / ∂b = 2α 2 g −1 − 1 + ξ 2α ξ
−1
2
(
) (
+ 1 + αξ 2 − 1 − αξ 2
)]
The terms in the last square brackets just cancel out. This implies that that around the value of
zero the derivative of the difference in the loss with respect to b is always positive.
25