Content uploaded by Signe Krogstrup
Author content
All content in this area was uploaded by Signe Krogstrup on Jul 03, 2017
Content may be subject to copyright.
Dealing with the De…cit Bias: Principles and
Policies
Signe Krogstrup Charles Wyplosz
Graduate Institute of International Studies, Geneva
April 13, 2007
Abstract
This Chapter formulates a common po ol model of …scal policymak-
ing including both countercyclical and productive spending motives to
run de…cits, and derives key principles for the optimal design of …scal
restraints. The model is then used to evaluate the welfare and design
aspects of …scal rules and delegation, with a view to the special cases
of the recently amended Stability and Growth Pact and National Fiscal
Councils. We …nd that simple numerical …scal rules are prone to cause
pro-cyclicality and to suppress productive public spending projects in the
presence of a de…cit bias. The amendments to the Stability and Growth
Pact introduced in 2005 are helpful in this respect. National Fiscal Coun-
cils generally satisfy the principles for optimal de…cit restraint, but neither
Councils nor Rules are enough to ensure an optimal …scal outcome, even
if optimally conditioned on economic circumstances. Both arrangement
must be combined with a budgetary institution allowing for productive
public investments and spending projects to take precedence in the budget
process.
1 Introduction
Budget de…cits and public debts increased with unprecedented pace on average
in OECD countries since the 1970s, pointing to the emergence of a de…cit bias in
Signe Krogstrup: krogstru@hei.unige.ch. Charles Wyplosz: wyplosz@hei.unige.ch. Ad-
dress for corresp ondence: The Graduate Institute of International Studies, 11A Avenue de la
Paix, C H-1202 G eneva. We are grateful to Guido Tabellini, Xavier Debrun, Heikki O ksanen,
Marco Ho eberichts, Jürgen von Hagen, Carlos Martinez-M ongay and participants at sem inars
at Bo cconi, the University of Cop enh agen, The Graduate Institute of International Studies,
University of Basel, the 2006 MAPMU conference, the 2006 EEA congress, the European
Com mission Workshop on the Role of Fiscal Rules and Institu tions in Shaping Budgetary
Outcomes, Brussels, 24 N ovemb er 2006, the European C entral Bank and the Swiss National
Bank, for helpful com ments and discussions on a previous version of this paper, and …nally to
the Europ ean Com mission (under the MAPMU network) for …nancial support.
1
…scal policy making. The policy debate regarding how to best deal with the ten-
dency of national governments to run excessive de…cits is currently centered on
the advantages and drawbacks of mainly two types of anti-bias …scal arrange-
ments, namely …scal rules and delegation. The Stability and Growth Pact is
the prime example of a …scal rule in place, and has been discussed heavily in
both academic and policy circles ever since it was proposed by Theo Weigel in
1996. The main critiques of the Pact, at least in its initial form, relate to its
seeming inability to appropriately allow for countercyclical …scal policy due to
inappropriate design and incentive incompatibility (it has in fact been accused
of producing pro-cyclical …scal policy outcomes in signatory countries), and its
tendency to suppress productive public investments. National …scal councils,
as an example of the delegation approach, have been proposed as a solution
to some of these problems, but involve more far-reaching changes to domestic
budgetary institutions than simple …scal rules.
The aim of this Chapter is to provide a model based policy evaluation of …scal
rules and delegation, with a view to the special cases of the recently amended
Stability and Growth Pact and National Fiscal Councils. A number of papers
have already taken a …rst step in this direction. Von Hagen and Harden (1995)
and Hallerberg and Von Hagen (1999) use a de…cit bias model to show that …scal
restraints are desirable and that delegation of the budget decision to a Finance
Minister reduces the bias. Beetsma and Uhlig (1999) show that a Stability and
Growth Pact can be welfare improving in the presence of a de…cit bias, and
Beetsma and Debrun (2004, 2005) and Blanchard and Giavazzi (2004) focus
on how the Stability and Growth Pact may suppress incentives to carry out
productive and therefore desirable public investments. The analyses in these
papers are based on models which are adapted to the particular issue in focus.
We instead formulate a more general common pool model of …scal policymaking
including both countercyclical and productive spending motives to run de…cits.
The resulting model allows us to derive some very general insights and principles
regarding the welfare aspects and the desirable institutional design of …scal
restraints. The principles can be implemented both through the rules based
approach and through the institutional or delegation approach. We …rst discuss
the characteristics of the necessary …scal institutions in the case of the Stability
and Growth Pact. The Pact was recently amended, notably to allow for more
‡exibility in the interpretation of the budget ceiling and to put more emphasis
on the medium term objectives, and we consider some of these amendments
in the light of the theoretical model. We also consider how National Fiscal
Councils compare to the principles derived from the theoretical model, and go
into more detail regarding the institutional mandate and desirable design and
accountability of such councils.
Throughout our discussion, we take as given that once …scal rules are in
place, they are complied with by …scal policymakers. We hence do not enter
the debate on enforcement or the possible endogeneity of …scal rules here - see
instead chapter XX (Ter-Minassian, Debrun and Kumar) of this book.
The Chapter is structured as follows. The next Section is theoretical. Based
on the previous theoretical literature on the de…cit bias, we set up a common
2
pool model of the de…cit bias with a business cycle component and two types
of public expenditures, productive spending and non-productive "pork-barrel"-
style transfers. We then derive the …scal policy outcomes in the unconstrained
case and under simple balanced budget rules, and derive the properties of opti-
mally designed …scal restraints conditional on the underlying process for dividing
the budget between unproductive and productive expenditure types. Section (3)
contains policy implications, as it applies the …ndings of the theoretical analysis
to a discussion of the Stability and Growth Pact in its recently amended form,
and to the design and mandate of National Fiscal Councils. The …nal section
concludes.
2 The De…cit Bias in Theory
In order to characterize and compare the properties of numerical …scal rules and
national …scal councils as …scal restraints, a general model of the de…cit bias is
needed. In this section, we take a brief overview of the theoretical literature.
We then proceed to set up a common pool model of the de…cit bias, which is
solved for the social optimum case, the unconstrained Nash outcome and the
outcome under various …scal restraints.
2.1 The literature
The most in‡uential theories of the de…cit bias are in the political economy
tradition1and comprise the common pool theory and the time inconsistency of
preferences theory. The common pool problem, formalized by Von Hagen and
Harden (1995), Hallerberg and Von Hagen (1999) and Velasco (1999, 2000),
arrises when there are several policymakers (ministers, parties, lobby groups)
involved in setting the budget2. Tax revenues represent a common resource for
these policymakers and since the budget de…cit represents claims on future tax
revenues, the budget de…cit can be considered a common as well. Thus, when a
policymaker wishes to increase public spending that bene…ts her constituency,
or enact a tax reform which reduces the tax incidence on her constituency, the
cost of these measures will be shared by all constituencies through a higher
general future tax burden. If unconstrained, policymakers will hence have a
tendency to push for policies which increase the budget de…cit while there will
not a be a similar push for …scally conservative budgetary measures, resulting
in a de…cit bias. The de…cit bias due to time inconsistency of government
preferences (Alesina and Tabellini, 1990) is similar in nature to the common pool
problem. It arrises when policymakers are subject to elections, and therefore do
not internalize the cost of running de…cits because they do not su¢ ciently care
about the public …nances of its potential successors.
1More complete survey s can b e found in Persson and Tabellini (2000), Chapter 9 and
Alesina et al. (1999), Chapter 9. The seminal contribution is Weingast et al. (1981).
2The delayed stabilizations case, develop ed by Alesina and D razen (1991), ca n be seen as
a case of com mon po ol.
3
A de…cit bias can also arise for purely economic reasons in an integrated
world, if national …scal policies spill over into other countries through economic
linkages. The debate surrounding the adoption of the Stability and Growth Pact
have identi…ed potential negative externalities of government debt levels on the
common interest rate, exchange rate and in‡ation rate of the EMU countries3.
The existence or strength of an international externality of budget de…cits
and public debt levels remains a matter of debate, as the empirical evidence is
very scarce. However, a number of empirical studies document that common
pool problems play a role in explaining the de…cit bias among OECD countries
(Persson et al. 2003; Fabrizio and Mody, 2006; Roubini and Sachs, 1989; von
Hagen 1992; von Hagen and Harden, 1994). There is less empirical support for
the view that uncertainty of reelection causes de…cits. Roubini and Sachs (1989)
…nd some support, but a more comprehensive and rigorous test of the reelection
uncertainty hypothesis is still missing4. For simplicity, we therefore focus on
domestic political distortions as the source of the de…cit bias5. Moreover, we
take a common pool interpretation of the domestic political distortion. The
common pool model currently has the strongest empirical support, and as shown
in Krogstrup (2006), nesting common pool problems and reelection uncertainty
in the same model complicates the modeling without gaining much insight6.
2.2 A Common Pool Model of the De…cit Bias
The model is based on Krogstrup and Wyplosz (2006) without an international
externality of debt, in which the de…cit bias is characterized as a domestic
common problem a la Velasco (2000) in a two period framework. Zoom in on
a small country in a large world economy. The country’s government, which
decides on …scal policy in the two periods, is made up of ndecision makers that
we broadly refer to as interest groups in the following. The de…cit bias arises
when interest groups seek to redistribute resources to their advantage through
the public tax and transfer system. Assume that the ninterest groups are of
the same size. Each interest group i,i= 1; :::n, decides on the amount of net
transfers, gi
t, that it obtains in period t= 1;2. Net transfers are de…ned as gross
transfers received less taxes paid by the members of that interest group.
In addition to net transfers, which only bene…t the receiving interest group,
the government can also undertake productive spending in the …rst period.
These spending items are productive in the sense that they raise public revenues
3See the pap ers collected in B runila, B uti and Franco (2001). See also Giuliodori and
Beetsma (2004, 2007), and K rogstrup and Wyplosz (2006) for a discu ssion of the these chan-
nels.
4Alesina et al. (1999) make this same observation.
5See Krogstrup and W yplosz (2006) for a similar model which also allows for an interna-
tional externality of debt.
6In fact, with m inor modi…cation s in the m odeling setup, the common pool externality can
be interpreted as an externality due to reelection uncertainty. The com mon po ol mo del hence
also allows for drawing p olicy implications based on a time inconsisten cy interpretation of th e
de…cit bias. We leave this for future work and focus on a com mon pool interpretation of the
model’s conclusions in Section 3.
4
in the second period. One interpretation is public infrastructure or human cap-
ital investments and another is costly reforms which raise future taxable income
(the interpretation given in Beetsma and Debrun, 2004, 2005). We assume
that productive spending does not directly bene…t interest groups and hence
does not feature directly in the utility functions of interest groups. Productive
public spending only serve interest groups in the sense that they increase the
intertemporal common pool from which transfers to interest groups are made.
This assumption is meant to capture the idea that common pool pressures are
stronger on targeted transfers which do not bene…t the population as a whole
than on certain types of productive expenditures/investments. Formally, de-
note productive spending in period 1as X. Spending Xin period 1raises
tax revenues by (X)in period 2. We assume 0>0and 00 <0, i.e. these
expenditures are subject to decreasing returns, which is needed for the second
order condition to be satis…ed. Since the interest groups are identical, they will
want the same level of productive spending, so we let the interest groups select
productive spending by consensus7.
To generate a motive for the government to wish to run countercyclical …scal
policy, we further extend the model with a simple business cycle. Assume that
in period one, a shock hits the budget. If < 0 (>0), the economy is in a
downturn (upturn). If the shock is positive, the country starts out in a good
year, meaning that additional tax revenues enter into the budget in the …rst
period. In a bad year (negative initial shock), tax revenues are lower than
the baseline case. We assume that if the economy starts in a good year, it always
ends with a bad year and vise versa, such that the business cycle always cancels
out across the two periods.
The country is fully …nancially open. The government can therefore bor-
row or lend internationally any amount that it wishes at the constant world
real interest rate, but is bound by the intertemporal budget constraint. The
international real interest rate is normalized to zero. The budget constraint is
understood and accepted by all interest groups. This assumption rules out de-
faults, an extremely rare occurrence in developed economies. Further, to break
Ricardian equivalence, assume that citizens and interest groups are personally
credit constrained. The …rst period government budget constraint is then given
by:
n
X
i=1
gi
1+X=B(1)
and the second and …nal period budget constraint is:
n
X
i=1
gi
2++B=(X)(2)
7We could also let the interest groups elect one amongst them selves who then selects the
level of productive spending - this assumption would yield the same outcome.
5
The intertemporal budget constraint is therefore given by:
n
X
i=1
gi
2+
n
X
i=1
gi
1+X=(X)(3)
We assume that the interest groups making up government have the same
preferences over the net transfers, represented by the following utility function
for interest group i:
Ui= log gi
1+g+ log gi
2+g:(4)
where the rate of time preference is set equal to the world interest rate8. The
term grepresents the maximum amount of net taxes that each interest group
is willing and able to pay. More precisely, we assume that there is a lower
limit g < 0for the net total transfers received by each interest group. The
maximum net revenues that can be collected is therefore G=ng.
2.3 The Socially Optimal De…cit
Suppose that a social planner decides on …scal policy and observes the …scal
shock before making the de…cit decision. The social planner’s objective is to
maximize social welfare, de…ned as a monotonic transformation of the sum of
the interest groups’utilities:
W=1
n
n
X
i=1 log gi
1+g+ log gi
2+g+ 2 log n
This expression is convenient as it reduces to
W0= log G1+G+ log G2+G(5)
under symmetry. The socially optimal …scal policy is a set of policy choices
fX; G
1gsuch that fX; G
1g= arg max Wsubject to the intertemporal budget
constraint (3), and conditional on the realization of the …scal shock, . Sine the
social planner cares equally about each interest group, she will always distribute
transfers equally in social optimum, so g
1=G
1
n. The social planner’s problem
therefore reduces to:
max
G1
W0= log G1+G+ log G2+G
8This assumption eliminates de…cits or surpluses due to smoothing, which we are not
interested in here. See Krogstrup an d Wyplosz (2006) for the general case in which 6=Rin
a similar model.
6
subject to the budget constraint (3), G2=(X)G1X. The …rst order
condition for optimal productive spending is:
0(X) = 1 (6)
where a star denotes socially optimal values. Given optimal productive spend-
ing, the …rst order condition for transfers then becomes:
G
1=(X)X
2(7)
When the business cycle is characterized as symmetric across the two periods,
it does not a¤ect the amount of the optimal …scal transfer. This means that the
business cycle shock will translate one to one into the optimal actual de…cit of:
B=G
1+X=(X)X
2+X(8)
The socially optimal de…cit consists of three components: smoothing to the
present of net future returns to current productive spending (the …rst term of
the right hand side of (8)), borrowing to …nance current productive investments
(the second term of the right hand side of (8)), and …nally, borrowing (or sav-
ing) to smooth the business cycle (the last term of the right hand side of (8)).
Considering the business cycle smoothing part of the budget in isolation, opti-
mal …scal policy dictates that in downturns, it is optimal to borrow an amount
equal to in the world markets, and pay back this amount in the subsequent
upturn. Conversely, starting in an upturn, it is optimal to save in period one,
and dis-save in period two, not borrow. Finally, note that the optimal cyclically
adjusted budget balance,
B
struc =(X)X
2+X;(9)
is only zero when there are no net returns to productive spending, i.e. when
(X)X=X= 0. The socially optimal structural budget balance hence
‡uctuates with the returns to productive spending (however de…ned).
Plugging (7) and (3) into the social welfare function (5) yields welfare in
social optimum:
W= 2 log (X)X
2+G(10)
2.4 The Actual De…cit Outcome
In the absence of a social planner, the interest groups choose their respective
net transfers, and select the aggregate level of pruductive spending by consen-
sus. The model is solved backwards. The debt level in the second period is
predetermined by the choices of the interest groups regarding net transfers and
productive spending in the …rst period. The second period budget constraint
7
hence gives us the aggregate net transfer possible to interest groups in the second
period, which we assume is shared equally among the ninterest groups:
g2=(X)B
n=(X)Pn
i=1 gi
1X
n:(11)
(11) makes the common pool problem clear. All else equal, if an interest group
increases its …rst period transfer gi
1by one unit, it will see a reduction of its
second period transfer of only 1
n, while the remaining n1
nunits of reduction on
the second period transfer will be held by the other interest groups. The cost
of increasing ones’ transfer in the …rst period is hence not fully internalized,
leading to a de…cit bias which increases in the number of interest groups n.
In the …rst period, the interest groups select by consensus the amount they
want to spend on X. They will choose productive spending so as to satisfy (6).
This is the level of Xthat maximizes the net discounted return to productive
spending. The interest groups subsequently select their net transfers simulta-
neously given the budget constraints and the observed realization of the …scal
shock. We could alternatively let the interest groups select net transfers prior
to or at the same time as selecting the level of productive spending. Since the
optimal level of productive spending is independent of the chosen net trans-
fer pro…le, the outcome would be the same. The sequencing of the game only
matters when the overall de…cit is constrained, as shown in Section (2.5) below.
Each interest group hence maximizes (4) with respect to (11) and (6), taking
all other interest groups’choices as given. The …rst order condition yields the
reaction function of interest group i:
gi
1=(X)X
21
2
n
X
f=1
f6=i
gf
1+n1
2nG
Imposing symmetry across interest groups yields the Nash equilibrium uncon-
strained transfer (denoted by N):
GN
1=n
n+ 1 ((X)X) + n1
n+ 1G(12)
(12) shows that the actual outcome for the transfer is above the socially opti-
mal one due to common pool problems. As ngoes to in…nity, the entire net
discounted return to productive spending in addition to the entire future taxing
capacity of the economy is spend on period one transfers. This excess is not
dependent on the business cycle, which cancels out across the two periods.
The unconstrained Nash de…cit outcome for the …rst period hence becomes:
BN=GN
1+X=n
n+ 1 ((X)X) + n1
n+ 1G+X(13)
Just as the socially optimal de…cit, the outcome for the unconstrained Nash
de…cit consists of borrowing to …nance productive spending and smoothing of
8
the business cycle (the last two terms of the right hand side of (13)). But now,
the smoothing to the present of future net returns to productive spending is
biased toward period one due to the common pool problem described above.
Note that it is the structural de…cit, BN
struc =n
n+1 ((X)X) + n1
n+1 G+X,
which is biased.
Welfare in the unconstrained Nash outcome is
WN= 2 log (X)X
2+G+ log 4n
(n+ 1)2!< W for n > 1(14)
2.5 Policy Responses
We use this model to analyze the properties and welfare implications of nu-
merical …scal rules and optimally designed …scal restraints below. Attention is
limited to issues of design and welfare.
2.5.1 Numerical …scal rules
Numerical …scal rules cap the maximum budget de…cit at a numerically …xed
level; usually somewhere in the range of between 0 and 3 percent of GDP in in-
dustrial countries. As such, these rules are not contingent on the business cycle,
and (unless they only apply to the current and not the capital budget) do not
take into account …nancing needs for productive public spending or productive
spending projects. To analyze the welfare implications of numerical …scal rules
using the model presented above, we focus our attention on the simplest case
of a national balanced budget rule of b
B= 0 (where hats denote de…cit ceilings
in the following). Considering balanced budget rules simpli…es the analysis rel-
ative to that of numerically …xed positive de…cit ceiling, without losing insights
regarding welfare aspects of numerical …scal rules in general.
Assume that a balanced budget rule is in place and that the rule is credible
and always complied with. When there is no borrowing, the second period net
transfer is given by the level of productive spending undertaken in the …rst
period, and the …scal shock according to
G2=(X)(15)
Before solving for …rst period transfers and productive spending, note that
reaching social optimum is only feasible under a balanced budget rule in the
special case of an initial upturn for which the above-normal budgetary revenues
() are equal to the funds needed to undertake the socially optimal level of
productive spending plus the socially optimal level of transfers, i.e. if
=(X)X
2+X0(16)
If is below (as de…ned by (16)) in upturns, then …rst period borrowing is
necessary in order to be able to carry out the socially optimal level of productive
9
spending and transfers simultaneously. Conversely, if the business cycle revenues
are so abundant that good public spending opportunities are exhausted before
reaching a balanced budget (i.e. of > ) then there is room for common pool
ine¢ ciencies to lead to excess spending in the …rst period.
To solve for the …rst period outcome, assume that the balanced budget rule
always binds, i.e. the revenue gains from the business cycle are not important
enough to allow for the unconstrained Nash outcome to occur under the balanced
budget rule, or GN+X> . There are two possible outcomes depending
on the sequencing of decisions regarding net transfers and productive spending,
henceforth referred to as the precommitment and the non-precommitment cases.
A simple balanced budget rule (non-precommitment) Assume …rst
that the interest groups cannot precommit to productive spending ahead of
deciding on net transfers. The timing of the game is as follows. In an initial
stage of the game, the balanced budget rule of b
B= 0 is observed and accepted
by all actors and the …scal shock is realized and observed by all actors. In a
second step, the interest groups select their net transfers. Finally, the level of
productive spending is determined as a residual under the balanced budget rule:
X=G1(17)
Each interest groups will choose a level of transfers that maximizes their indi-
vidual utility (4) subject to (15) and taking into account the balanced budget
rule, (17). The …rst order condition for the interest group’s problem is
X
i
gi
1!+G=0 X
i
gi
1!gi
1+g
We cannot solve explicitly for the de…cit outcome, but applying symmetry across
interest groups yields the …rst order condition for the constrained Nash outcome
under the balanced budget rule without precommitment:
0XBBR;N P =nGBBR;N P
2+G
GBBR;N P
1+G(18)
where BBR denotes outcomes under a balanced budget rule, and N P denotes
outcomes in the non-precommitment case. (18) shows that even in the special
case of =;where reaching social optimum without borrowing is feasible,
imposing a balanced budget rule in the absence of precommitment to productive
spending does not yield the socially optimal welfare level. The reason is that
the interest groups will have individual incentives to increase their net transfers
above the socially optimal levels, even if this means that public spending will
su¤er. In the general case in which 6=welfare is also below social optimum
and the deviation of welfare from its optimal value depends on the deviation of
the business cycle from the optimal level of transfers and productive spending,
in addition to the strength of the common pool distortion.
10
Finally note that under a balanced budget rule, where the overall de…cit is
acyclical per de…nition, we will tend to see a procyclical structural de…cit when
common pool distortions push the budget to the limit.
Whether or not the …scal outcome under a simple balanced budget rule is
better or worse than in the unconstrained Nash outcome depends on the degree
of the common pool problem in the unconstrained case (i.e. the size of n)
relative to the deviation of the business cycle from the optimal level of transfers
and productive spending9.
A balanced budget rule with precommitment Assume now that the
interest groups decide on the level of productive spending by consensus, and
precommit to this level before selecting transfers. The timing of the game taking
place in the …rst period is now as follows. As before, the balanced budget rule
of b
B= 0 is observed and accepted by all actors, and the …scal shock is realized
and observed in a …rst step. In a second step, the interest groups decide on the
level of productive spending to carry out and precommit to this level. Finally,
the interest groups set the transfers to comply with the balanced budget rule.
Transfers in period one are hence determined residually by the balanced budget
rule, productive spending and the …scal shock:
G1=X(19)
The optimal choice for the level of productive spending maximizes the utility of
the representative interest group (4) subject to the balanced budget rule (19)
and second period transfers (15). The …rst order condition becomes
0XBBR;P =GB BR;P
2+G
GBBR;P
1+Gfor > 0(20)
where a Pdenotes outcomes in the precommitment case. The marginal gross
return to productive spending is set equal to the ratio of the marginal utilities of
transfers in the two periods. The intuition is straightforward. Under a balanced
budget rule, there is a one-to-one trade-o¤ between the level of current trans-
fers and productive spending which the interest groups take fully into account
when selecting the public spending level in the …rst step. Productive spending
can hence be used to smooth the positive (negative) current returns from the
business cycle to the future downturn (upturn). In the second stage of the …rst
period game, the interest groups will then select transfers which take the total
de…cit to the ceiling. These transfer can either be positive or negative, depend-
ing on whether the gross returns to productive spending in the second period
su¢ ciently compensate for a possible future business cycle downturn or not.
9Plugging (15), (17) and (18) into the welfare function (5) yields welfare under the
balanced budget rule without precommitment: WBB R;NP = 2 log XB BR;N P +G+
log 0(XBBR;N P )
n:Since we cannot solve exp licitly for G1and X, we cannot compare this
expression with that of the unconstrained Nash case.
11
As before, the de…cit will be acyclical per de…nition of a balanced bud-
get rule. But the precommitment technology will reduce, if not eliminate, the
tendency for net transfers to be procyclical as the cycle is smoothed through
productive spending instead of saving. The overall structural balanced, which
includes productive spending, remains equally procyclical however. Thus, the
precommitment mechanism for productive spending combined with a balanced
budget rule partly internalizes the common pool problem. But another distor-
tion is created in that the government cannot borrow in order to undertake
additional productive spending in cases where the returns to such spending ex-
ceed the interest rate, and cannot save at the world interest rate in cases where
returns to additional public spending are low.
(20) shows that in the special case in which =;precommitment com-
bined with a balanced budget rule yields the socially optimal welfare level10 .
But in the general case in which 6=welfare is below social optimum and
the deviation of welfare from its optimal value depends on the deviation of the
business cycle from the optimal level of transfers and productive spending. If
< , we have that 0XBBR;P >1and both transfers and productive
spending are below socially optimal levels. On the other hand, when > ,
we have that 0XBBR;P <1and productive spending will be above its so-
cially optimal level while transfers will be below their optimal level. In this
case, it would have been better to invest less and run a surplus which would
be spend on transfers only in the second period. But the interest groups know
in advance that they will have to exhaust their budget "allowance" under the
balanced budget rule if they want to transfer the funds to the future, because
it will otherwise be used for current transfers due to common pool problems.
Per construction, the balanced budget rule combined with precommitment
to productive spending will always yield higher welfare than the case of the a
balanced budget rule without precommitment11 . But as before, whether or not
the …scal outcome under a balanced budget rule combined with precommitment
is better or worse than the unconstrained Nash outcome depends on the degree
of the common pool problem in the unconstrained case (i.e. the size of n)
relative to the deviation of the business cycle from the optimal level of transfers
and spending.
To sum up, the analysis of numerical …scal rules suggest that combined with
business cycles and common pool problems, such rules will have a tendency to
engender procyclical structural …scal policies, and too little productive spending.
These results raise the question of what characterizes an optimally designed
…scal restraint. We turn to this issue next.
10 (16) an d XBB R;P =Xsatis…es (20).
11 Welfare under a balanced budget rule with precom mitment is given by WB BR;P =
2 log XBBR;P +G+ log 0XBBR;P .
12
2.5.2 Optimal Policy Responses
From the above analysis, we know that an optimally designed …scal restriction
will only achieve social optimum if it can be conditioned on the cycle and on
the returns to productive spending. How would such a restriction look? Sup-
pose that a social planner can choose a de…cit ceiling after having observed
the cycle and returns to productive spending, leaving the division of the de…cit
between productive spending and net transfers for the government to decide.
Such a setup can only achieve social optimum if the government has a means
to precommit to productive spending before choosing net transfers. To show
this, we …rst solve for the outcome under an optimal de…cit ceiling without
precommitment.
No precommitment The game is structured as follows. All players …rst
observe the realization of the …scal shock . The social planner then selects a
de…cit ceiling after which the interest groups select their net transfers. Finally,
productive spending are determined as a residual under the de…cit ceiling:
X=b
B+G1(21)
Solving backwards, the interest groups choose their transfers to maximize (4)
subject to (11) and (21). The …rst order condition for net transfers becomes:
0(X) = nG2+G
G1+G(22)
(22) shows that due to common pool problems under the de…cit ceiling in the
absence of precommitment, social optimum (characterized by G2+G
G1+G=0(X) =
1) is not possible to reach. The interest groups will set their transfer too high,
leaving too little for productive spending given the ceiling over the …rst period
de…cit.
Taking (11), (21) and (22) into account, the social planner sets the de…cit
ceiling which maximizes social welfare. The …rst order condition for the problem
is
0XO;N P =n
1+(n1) @ XO ;NP
@b
B
;(23)
where
@X O;N P
@b
B=0XO;N P +n
0(XO;N P ) (n+ 1) GO;NP
1+G00 (XO;N P )
:(24)
where an O refers to outcomes under an optimal …scal rule. We cannot solve ex-
plicitly for the optimal de…cit ceiling in the absence of precommitment, but the
…rst order condition reveals a few things about the outcome. Since "(XO;N P )<
0, (23) and (24) imply (XO;N P )0>1i.e. XO;NP < X, i.e. the de…cit ceiling
is set such that productive spending is below its socially optimal level. This
13
means that the net returns to productive spending are also below their poten-
tial level - the pie is smaller than it could be. Moreover, in spite of the ceiling,
the division of the pie is suboptimally tilted toward the present in the form of
too high net transfers due to common pool problems. The larger is n, the lower
is the level of productive spending relative to net transfers under the ceiling,
and for nlarge enough, investments are driven to zero. In this case, the social
planner does not face any more a trade-o¤ between squeezing the net transfers
and reducing the net return to productive spending, and will choose a balanced
budget rule. In short, a de…cit ceiling, even if optimally conditioned on the cy-
cle and returns to productive spending, cannot deliver the social optimum when
national governments are subject to common pool problems and cannot pre-
commit to productive investments. But the de…cit ceiling nonetheless improves
upon the Nash equilibrium and on the simple balanced budget rule; otherwise
the social planner would set the ceiling at or above the unconstrained Nash
de…cit outcome1 2 .
Precommitment Suppose now that the social planner can observe the out-
come of the …scal shock and then select a de…cit ceiling, b
B. The sequencing of
the game in the …rst period is now as follows. First, the social planner observes
and selects b
B. In the second stage, the interest groups select the level of pro-
ductive spending. Finally, transfers are set residually, as the di¤erence between
the maximum allowed budget de…cit and that which is spent on investments.
Solving backwards, transfers in the second period are equal to
G2=(X)b
B(25)
and transfers in the …rst period become
G1=b
BX+(26)
Taking (25) and (26) into account, the interest groups select the optimal level of
productive spending by maximizing (4) subject to b
B. The …rst order condition
for productive spending becomes
0XO;N P =GO;NP
2+G
GO;N P
1+G(27)
The social planner then selects the de…cit ceiling which maximizes social welfare,
subject to (25), (26) and (27). The …rst order condition for this problem yields
the socially optimal de…cit ceiling:
b
BO;P =B=G
1+X=(X)X
2+X(28)
12 Welfare under an optimally determined de…cit ceiling but in the absense of precommitment
is given by WO;N P = 2 log XO;NP + +G+ log 0(XO;N P )
n.
14
The optimal de…cit ceiling is hence not a balanced budget rule, but ‡uctuates
according to the need for countercyclical …scal policy and the need for produc-
tive spending. (28) shows that the outcome for productive spending is also
socially optimal in this case, i.e. 0(X)=1. Thus, when a de…cit ceiling
is conditioned on the cycle and returns to productive spending and combined
with precommitment to productive spending before the rest of the budget is
determined, a socially optimal welfare level can be achieved.
2.6 A Welfare Ranking of Policy Responses
When domestic political distortions cause a de…cit bias, the welfare ranking of
the de…cit restraints analyzed above is as follows:
W=WO;P > W O;N P > U N
=WO;P WBB R;P > W BBR;NP 7UN
First, note that we cannot generally rank all policy options. The relative
welfare ranking of WO;NP ,WB BR;P ,WBBR;N P and UNdepends on the rela-
tive strength of the common pool problem (i.e. the size of n), the business cycle
outcome and the returns to productive spending. But a few general conclusions
emerge. First, social optimum can be reached when budgetary institutions al-
low the government to precommit to productive spending items before setting
transfers, and when the overall de…cit ceiling is optimally set according to the
business cycle and returns to productive spending. It is, in other words, not
enough to cap the de…cit at its optimal level, the budget process has to make
it possible to precommit the desirable or productive part of spending (produc-
tive in the sense that it raises output) in advance of the non-productive part
of the budget. De…cit ceilings without precommitment can never reach social
optimum. Second, in the special case in which the size and sign of the busi-
ness cycle satis…es (16), a balanced budget rule combined with precommitment
can also reach social optimum, but this is not the general case. Third, an opti-
mally designed restriction applying to the overall budget de…cit improves on the
unconstrained Nash outcome even in the absence of appropriate budgetary in-
stitutions to allow for precommitment. Fourth, depending on the business cycle
and on potential returns to productive spending, numerical …scal rules can lead
to procyclical …scal outcome which are worse than the unconstrained Nash out-
come, even if the government can precommit to productive spending under the
ceiling. We turn next to how these analytical …ndings can guide the discussion
of options for inducing …scal restraints in the European Union context.
3 Policy Implications
The model suggests a number of policy implications. First, inasmuch as a
common pool problem lies at the roots of the tendency to run budget de…cits,
any solution to the bias problem will have to constrain the interest groups’
15
ability to preempt budget decisions. Second, cyclical ‡uctuations complicate
matters. Here we make the simplifying assumption that the cyclical conditions
are known when the budget process gets under way, but this is not always the
case. Even so, we …nd that a recession intensi…es the common pool problem as
it reduces the overall surplus and raises the marginal utility of transfers. On
the other hand, an expansion creates some slack that interest groups see as even
more of a free good. Third, we …nd that quantitative budget rules –here in the
simple version of a balanced-budget rule, but any rule has the same qualitative
e¤ects –do not achieve the social optimum unless the interest groups commit to
the level of productive spending before the size of transfers is negotiated. One of
the negative e¤ects of rules is that they tend to deliver procyclical …scal policies.
Fourth, pre-commitment is a necessary ingredient of any attempt to bring the
budget to its socially desired level. Finally, the socially optimal budget must
take into account time-varying conditions, here the productivity of spending
and the stage of the business cycle.
In this section we attempt to move from theory to practical solutions. We
…rst interpret the nature of interest groups. Then we discuss the Stability and
Growth Pact in light of the model. We then develop the idea of …scal councils.
3.1 Who are the Interest Groups?
In our model, the interest groups have the power to decide on the size of the
transfers but they cannot really negotiate; they only stake out their con‡icting
claims on the common pool knowing full well that the outcome will be socially
undesirable. We also need to provide an interpretation of how the level of
productive spending is decided and who could be the social planner. As with
all model interpretations, the link with the model is neither immediate nor
compelling.
The simplest interpretation of interest groups is that they are ministries,
as suggested by von Hagen and Harden (1994, 1995). In that view ministries
are fully captured by interest groups whose utilities they simply attempt to
maximize. The Minister of Finance sets the level of productive spending and,
possibly along with the Prime Minister or the Head of State, is the only gov-
ernment agent who attempts to raise social welfare, the utilities of the interest
groups. Furthermore, the model implicitly assumes that all citizens are member
of one interest group, that all interest groups are of the same size and that each
of them is represented by one ministry.
An alternative interpretation pits the government as a whole against the
interest groups. In this view, it is the government that sets public spending.
The government also decides on the transfers to the interest groups but under
their direct control. This last description implies that the interest groups are
perfectly organized to defend their narrow interests, up to the point where they
e¤ectively exercise total control over the government in their respective areas of
interest. As before, the interpretation assumes that all citizens are members of
one interest group and that all interest groups are of the same size.
Each of these interpretations is extreme. In the …rst one, it is not literally
16
true that each and every interest group is represented by a ministry, so some
element of the second interpretation would need to be included to o¤er a better
description of reality. Alternatively, in both interpretations, it would be desir-
able to allow for some segments of the population not to be part of any interest
group. They could be represented by the Finance Minister in the …rst case and
the government in the second case.
While both interpretations can be taken as an acceptable stylized fact in a
theoretical model, it matters a lot which one is more relevant when we attempt
to draw practical implications on how to deal with the de…cit bias. For instance,
under the …rst interpretation, a reasonably simple step would be to reduce the
number of ministers since the size of the common pool problem is directly related
to the number of interest groups. Under the second interpretation, political
parties can be seen as the channels through which interest groups operate. The
de…cit bias would then be alleviated by adopting electoral laws that favor the
existence of a small number of political parties. This would, of course, raise
complex issues on how political parties relay the wishes of the interest group
that they seek to agglomerate under their banners.
In what follows, we largely disregard this important issue, although we will
at time discuss how the interpretations a¤ect the arrangements discussed.
3.2 Where Does the Stability and Growth Pact Fit?
The Stability and Growth Pact is not strictly a quantitative budget rule. At
least in its revised version, it allows for a ‡exible interpretation of the de…cit
ceiling. In addition, the preemptive arm of the Pact requests that some restraint
be exercised even when the ceiling is not binding. This allows for intertemporal
smoothing of the rule itself and stands to alleviate its potential procyclical
e¤ects.
Similarly, the recognition that the budget balance should be cyclically ad-
justed helps to deal with the risk of pro-cyclicality. The model illustrates another
risk, namely that de…cit reductions …rst and foremost a¤ect productive public
spending. The risk arises when the interest groups succeed in sheltering the
transfers of interest to them.
Indeed, one of our results is that quantitative rules are welfare-improving
only when accompanied by a pre-commitment to the level of productive spend-
ing. The Stability and Growth Pact does not include any such provision. Yet,
the numerous amendments adopted in 2005 allow for taking into account a num-
ber of spending items that can be broadly related to what the model describes as
productive spending: major reforms, R&D, investment in the quality of public
…nances, international solidarity and a few less clearly de…ned items. However,
such a pre-committed spending is not to be counted within the overall ceiling,
rather it is allowed to be treated outside the ceiling. Consequently, the ceiling
remains for what we call transfers.
From the perspective of our model, therefore, the Stability and Growth Pact
allows for pre-commitment to some productive spending alongside a ceiling.
Yet, it does not stand a chance of achieving the social optimum insofar as the
17
precommitment concerns a limited range of spending items and the amounts
committed are not allowed to be adjusted to changing economic conditions,
including cyclical conditions. In order to be optimal, the Stability and Growth
Pact could be modi…ed as follows.
Each government would identify a list of spending items that it considers as
productive. The list would provide the background for setting a limit on the
remaining de…cit. For instance, should productive spending represent 15% of
GDP, the ceiling on the remaining part of the budget would be a surplus of
15%, thus amounting to a total budget in balance, with the usual 3% tolerance
margin and with the preemptive arm arguing for larger surpluses in good years.
The productive spending list could be changed annually, to re‡ect changing
circumstances, with adequate adjustment on the ceiling.
This modi…cation would retain the objectives of the Stability and Growth
Pact while allowing for a pre-commitment regarding the productive spending.
It would e¤ectively constrain non-productive spending as identi…ed by each
government. The obvious risk is that all spending items would be identi…ed as
“productive”, which would not allow for a compensatory surplus. This would
not be a problem, though, since it would imply a ceiling on the empty remaining
set equal to the size of productive spending, in e¤ect reproducing the current
pact.
3.3 National Fiscal Councils
The Stability and Growth Pact rests on the principle that the agent of enforce-
ment of any precommitment is external, the Council on the basis of a recommen-
dation from the Commission. This feature raises a delicate issue of sovereignty.
Through the Maastricht Treaty’s Excessive De…cit Procedure (EDP), member
states have accepted the principle of some collectively enforced limit to their
budget de…cits. At the same time, in each country, a fundamental principle
attributes to the national parliament the exclusive authority to approve the
budget. Without going into legal issues, politically at least therein lies a con-
‡ict between sovereignty over budgetary matters and the acceptance of “peer
pressure”to restrain de…cits.
It has been suggested that a solution to this tension – which has led to the
2003 abeyance of the Stability and Growth Pact –would involve decentralizing
the task of de…cit restraint at the national level while allowing for a collective
oversight of the solution adopted in each country. Such an approach satis…es
both the need for restraint and the principle that monetary union members
abide by commonly accepted procedures.
Von Hagen and Harden (1994) propose to strengthen the power of the …nance
minister in the budgetary process. This is in line with the …rst interpretation
of our model. Empirical work has con…rmed that such a solution succeeds in
restraining budget de…cits (see e.g., Hallerberg and Wol¤, 2006). A di¤erent
proposal is to delegate part or all of the budget balance decision to a …scal
council (see Wyplosz, 2005), which we now examine in some detail.
18
3.3.1 Principle
A …scal council includes a group of non-elected independent persons who are
given strong independence, along the lines of central bank monetary policy
committees. Non-election and independence are the necessary conditions to
sever the link between interest groups and decision-making. Because they will
always retain the right to set the budget, delegation to non-elected o¢ cials must
be severely restricted.
The fundamental reason why budgetary decisions are owned by elected of-
…cials is that they involve income transfers. Accordingly, the …scal council’s
remit must carefully exclude any explicit income transfer. The budget balance
approximately satis…es requirement, since it only transfers income over time,
mainly across generations, all of which save one are not represented by the
current population. Thus the only decision to be delegated to a …scal council
concerns the budget balance.
Like any non-elected body, the …scal council must be given an explicit ob-
jective by elected o¢ cials. The natural objective is the evolution of the public
debt. The objective must be given by the government with parliament approval,
with changes possible but presumed to be infrequent.
Accountability is another required feature of delegation to non-elected of-
…cials. Accountability must not be the back door for reducing the council’s
independence. Various arrangements are possible, ranging from annual reports
to formal periodical auditions. Failure to achieve the objective must give rise
to sanctions. In order to preserve independence, sanctions must be carefully
designed. This issue is not pursued here.
3.3.2 Link with the theory and implications
We have de…ned social welfare as the intertemporal utility of interest groups
under the assumption that all citizens are included in one or another group. As
a consequence, we cannot examine some important issues like income redistrib-
ution and the protection of underprivileged groups. As always, the assumptions
are restrictive, yet we believe that they are adequate to study the design of
de…cit restraints in the presence of a common pool problem. We also assume
away debt sustainability, which we view as a reasonable assumption for Euro-
pean countries.
The model indicates the importance of pre-commitment to socially produc-
tive spending and the need to set a limit to the de…cit – possibly a negative
one, which implies a surplus. The limit is included in the debt target: combined
with a set horizon, it speci…es an intertemporal ceiling.
The model also attracts attention to the sensitivity of the optimal ceiling
to cyclical conditions. This is why the debt objective must be set over the
medium-term; the desirable term must be consistent with the length of business
cycles. This is also why the optimal annual budget ceiling depends on current
conditions.
The implication of our model that raises the biggest implementation di¢ -
19
culty is that it is not enough to limit the de…cit, there must be a procedure in
place to ensure that socially productive public spending not be curtailed while
transfers to interest groups are little a¤ected. The solution, we suggested, is
that the government be able to precommit to productive spending before the
transfers are decided. A …scal council that is only given authority to set the
budget balance cannot interfere with the composition of public spending and
therefore does not solve this problem. In order to achieve the social optimum
as de…ned in our model, the council would have to be combined with govern-
ment institutions that allow for productive spending to be determined before
transfers. In that sense, …scal councils do not improve upon budget rules.
3.3.3 Mandate of a national …scal council
In line with the preceding principles, a …scal council should be given a mandate
and an instrument over which it has exclusive authority. The mandate is a debt
target over the medium run. This target, which will be changing over time as,
hopefully, the debt to GDP ratio declines, is a political decision to be made by
the government and, in some countries, the parliament.
The instrument is the budget balance. The council would operate as a ‡ex-
ible budget rule. Since the evolution of the debt over the medium run is driven
by the cumulated budget balances over the intervening years, the …scal council
can map a variety of possible paths. Its margin of action is not negligible and
ought to be used to take into account the general economic situation. This
is where a …scal council dominates a budget rule: instead of context-free ceil-
ings, a council is able to optimize the ceilings given the current and foreseeable
prevailing economic conditions.
This is also where a …scal council dominates the Stability and Growth Pact:
its incentive structure is strongly conducive to operate both the preventive and
the corrective arms and, in addition, to optimize on both margins. The stronger
is action in good years, the more room will the council have to maneuver in bad
years. The less slippage it can tolerate in bad years, the easier will it be to
achieve the debt target.
The debt target has to be realistic. It might be tempting for an incoming
government to set an overly ambitious debt reduction target and then blame
the …scal council for failing to deliver. This risk does not exist if the council
is given the exclusive control of the budget balance. In that case, given an
excessively ambitious debt target, the council will enforce a restrictive strings of
budget balances for which the government will have to assume full responsibility.
Conversely, a modest debt target, which the …scal council will easily deliver, will
have to be defended by the government. We return to this question further below
when we consider the role of the European Union.
3.3.4 Design of a national …scal council
The raison d’être of …scal councils is to provide a commitment technology in a
world where signi…cant portions of national budgets are controlled by interest
20
groups through government capture. The implication is that councils be given
an iron-clad protection against interest groups, which means that they should
be composed of non-elected o¢ cials with strong personal quali…cations.
The model to follow here is the monetary policy committees that now oper-
ate independent central banks around the world. Several formulae exist but they
all share a few common characteristics. Crucially, council members ought to be
given long mandates, which cannot be discontinued except for grave breach of
duty. A good solution is that the length of the mandate exceeds the length of
legislature in parliamentary regimes or the duration of the president’s mandate
in presidential regimes. Along with staggered appointments, this solution en-
sures that the council members are not all related to the ruling majority of the
time.
A second feature of the model is the two-period setup. It is meant to capture
the medium term: the current period describes policy options over a period
for which economic forecasts are reasonably accurate, while the second period
represents the horizon when the cycle unwinds so that the overall length captures
a completed business cycle. This description is relevant for the de…nition of the
medium run over which the debt target is set. Ideally, it should encompass the
length of a business cycle. Unfortunately, business cycles do not exhibit stable
duration. What then should be the medium run chosen for the mandate?
Given the lack of regularity of business cycle duration, another criterion
should come to bear on this issue. Since, in our view, it is the government that
sets the target debt, it would be logical that the medium run corresponds to the
government’s term in o¢ ce. Indeed, a government cannot e¤ectively bind its
successor(s) in such matters. In addition, in most countries, the term of o¢ ce
is of 4-5 years, which closely matches the average duration of business cycles.
The procedure would be for a newly elected government –possibly subjected
to parliament approval – to determine the debt target that it wishes to reach
at the end of its mandate. If, as is possible in many countries, the life of the
government can be shortened, the debt target may not be reached and a new
one will be established by the successor government. This raises a question of
accountability, but otherwise adequately corresponds to the actual working of
the democracy.
3.3.5 Accountability of a …scal council
As non-elected o¢ cials, a …scal council ought to be strictly accountable for
its actions to an elected body. It would seem natural that this body be the
parliament. In order to carry out this mission, the parliament should be assisted
by a highly competent advisory body, possibly patterned at the Congressional
Budget O¢ ce of the US Congress.
Since its mandate is a medium-run target, in principle the council’s actions
can only be evaluated at the end of the period corresponding to the target. This
suggests a two-stage procedure. The major stage is the end-of-period evaluation.
There are many reasons why the best council could miss its target. Obviously,
an unfortunate period of below-average growth may result in undershooting
21
the target. This can be explained by the council and the explanation can be
evaluated by a well-assisted parliament. What happens if the explanation is
found unconvincing? There should be a sanction, but designing it is made
di¢ cult by the need for the council to be independent and therefore for its
members to hold secure positions. In serious cases of mis-performance, the
parliament could determine a breach of duty and replace the whole council,
but this would complicate the staggering process, which is crucial to achieving
non-partisanship. Another solution would be to only replace the council chair.
Because a vote of no-con…dence would greatly weaken the council, it could be
adequate as a sanction.
The other stage would be annual parliamentary reviews of the evolution
of the debt. Each year, including at the start of the medium-run cycle that
follows the appointment of a new government, the council would produce a
report explaining its strategy to achieve the mandated target. The report would
include a year by year projection of the debt, backed by the publication of each
year’s budget balance and key assumptions, in particular, growth, in‡ation and
the interest rate. Each year, the report would have to explain any discrepancy
between the projected debt and its realized level. Here again, a vote expressing
concern when the parliament considers that the council’s actions are not in line
with its mandated debt target should provide for adequate accountability.
3.3.6 The role of the European Union
The EDP enshrines the principle that, within the monetary union, one country’s
lack of …scal discipline is a matter of common concern. The precise economic
basis for this concern remains a matter of controversy but this is besides the
point given the existence of the EDP. In Krogstrup and Wyplosz (2006) we study
the role of this concern and …nd that it reinforces the need for supranational
involvement to achieve optimal …scal outcomes.
The general implication of the EDP is that each monetary union member
country must o¤er its partners a reassurance that its …scal policy is and will
remain disciplined. The weakness of the Stability and Growth Pact in its original
form was to focus on annual budget outcomes; the 2005 revision has taken a
step forward by emphasizing the evolution of the debt. This …ts well with the
proposed …scal council medium run mandate and makes it easy to reshape the
EDP. The new EDP would rest on two formal measures.
The …rst measure would be that the national legislations that create the
councils and de…ne their tasks and independence would have to be EDP-compatible,
as is the case with legislation that de…ne national central bank independence.
The measure would thus provide a solid reinsurance that each country has put
in place adequate institutions for debt stability.
The second measure would organize mutual surveillance, as mandated by
the EDP. A possible procedure would be that the medium run debt target, set
by each government, would have to gain approval of the other monetary union
members. This would take the form of an amended Stability and Convergence
Program. A newly elected government would submit its planned debt target
22
to the Council, which would approve it following a report from the Commis-
sion. Then, each year, at the same time as the national parliament conducts
its own accountability exercise, the Council would examine the council’s re-
port and would issue its own conclusions, following a recommendation from the
Commission.
This procedure has several advantages. First, it would uphold the EDP’s
principle that a country’s …scal discipline is a matter of common concern. Sec-
ond, it would allow for peer pressure, a key objective of the Stability and Growth
Pact. Third, it would focus on the evolution of the debt and not on annual
de…cits, holding each member country council to its mandate. Finally, it would
reduce the odds of unrealistic targets.
3.3.7 The asymmetry problem
A structural weakness of the EDP is that not all countries are equal when
considering the common concern. Obviously, a small member country could
display …scal indiscipline without any serious consequence for the monetary
union as a whole. On the other hand, even moderate slippages in the large
countries are bound to attract attention and to potentially weaken the euro
and to raise the euro area interest rate risk premium. This is an unavoidable
economic asymmetry.
The implementation of the Stability and Growth Pact has led to another,
political asymmetry, which aggravates the economic asymmetry. As the events
that led to the 2003 decision to put the pact in abeyance have shown, the large
countries, those whose …scal discipline matters, can use their natural in‡uence to
avoid sanctions. The result has been a serious loss of credibility for the pact. The
2005 reform has sought to restore credibility but, by emphasizing ‡exibility, the
solution –adequate as far as the economic logic of …scal restraints is concerned
–has not obviously brought renewed credibility.
Relying on national …scal councils stands to alleviate the asymmetry prob-
lem. There is no reason that the rigor of council actions will be related to
country size. It follows that we should not expect large countries to be less dis-
ciplined than smaller ones. This would not eliminate the economic asymmetry
but it would eliminate its impact. The political asymmetry is less likely to be
eliminated but, inasmuch as national councils deliver on their mandates, the
issue is unlikely to emerge.
4 Conclusions
This chapter has two main objectives. The …rst one is to clarify the sources of
the de…cit bias that has been ubiquitous in most European countries. To that
e¤ect, using the common pool framework, we describe the budgetary process
as a grab race among sel…sh interest groups. Along with time-inconsistency,
this is a standard interpretation of the de…cit. Our contribution is to enrich
the setting by considering socially productive public spending and to allow for
23
cyclical ‡uctuations. The …rst addition allows us to capture the concern that
…scal restraints may act primarily on productive spending. The second addition
illustrates the danger that de…cit-reducing policies may make …scal policy pro-
cyclical.
Our second objective is to use our theoretical results to explore what kind
of policies can be used to correct the de…cit bias. Indeed, with few exceptions,
much of the relevant literature does not directly relate policy options to an ex-
plicitly articulated interpretation of the de…cit bias. Our results provide support
to those who argue that numerical …scal rules, i.e. rules which unconditionally
restrict the unadjusted budget balance, are not generally optimal and may ac-
tually worsen the …scal outcome in terms of welfare. We also con…rm that the
common pool problem makes such rules a further source of procyclical …scal
policies. A frequently suggested response to the pro-cyclicality implication of
numerical …scal rules is to apply them to cyclically adjusted budgets. We …nd
that, when a common pool problem exists, this is likely to remove pro-cyclicality.
We also …nd that numerical rules do not prevent borrowing to …nance produc-
tive public spending when they are optimally conditioned on the associated
potential returns. Crucially, however, in our model, for such restraints to be
socially optimal, they must be combined with budgetary institutions that allow
the government to commit to productive spending before the rest of the budget
is determined. In the absence of such a precommitment technology, the domes-
tic political distortions which lie at the root of the de…cit bias will continue to
negatively in‡uence the …scal outcome under numerical …scal rules.
Viewed in the prism of our model, the amendments to the Stability and
Growth Pact introduced in 2005 are therefore helpful. They explicitly allow
for the consideration of cyclically adjusted budgets and they recognize that
spending on socially productive projects ought to be taken into account. Yet,
our model suggests that this is not enough, since it shows the importance of a
precommitment to productive spending. Our result can be seen as a particular
case of the more general observation that no rule, however detailed, will ever be
able to take into account all possible current and future events which may a¤ect
the optimal intertemporal allocation of public spending. Continuous judgment
of the economic circumstances is needed.
The more general observation is simply another manifestation of the stan-
dard rules vs. discretion debate. The relevant literature has clearly shown that
discretion is welfare-reducing when it leads to a loss of credibility of the author-
ities in charge. As is now well-understood, the resolution of the debate is to
allow for constrained discretion by delegating decision to an agent that is not
prone to political pressure – in this case, the common pool problem. In other
words, institution building is the response to the standard rules vs. discretion
debate.
This leads us to examine how such a delegation could be organized in the …s-
cal policy area. This is what lies behind the proposal of …scal councils composed
of independent wisepersons. We use the model to determine the instrument and
the mandate that should be delegated to …scal councils. The instrument is the
annual budget balance –which stands in sharp contrast to numerical rules, such
24
as the Stability and Growth Pact, which identify the budget balance as the main
objective. The mandate is the public debt level over the medium run, a hori-
zon long enough to extend beyond the business cycle and thus avoid the risk of
pro-cyclicality.
As we examine the various facets of optimal …scal councils, we consider
the de…nition of independence and the associated democratic accountability
requirement. We pay particular attention to the requirement, embodied in the
Excessive De…cit Procedure, that each country’s …scal discipline is a common
concern for all other EU members. This leads us to examine the possibility
of requiring that national legislation creating the …scal councils be subject to
collective approval. We also suggest that the annual reports submitted by the
…scal councils as part of the accountability procedure, become the main tool for
mutual surveillance.
Yet, neither the revised Stability and Growth Pact nor the …scal councils
fully meet all the requirements suggested by our model. Indeed, the theoretical
analysis reveals an important point regarding the optimality of …scal restraints.
Neither the rules based nor the institutions based approach to de…cit bias cor-
rection automatically ensures that the mix of unproductive transfers and pro-
ductive spending will be optimal under the de…cit restraint. Both arrangements
must be augmented with budgetary institutions which allow the government to
identify productive spending items and to precommit to these ahead of deciding
on the rest of the budget. Creating such a budgetary institution is complicated
by the fact that it is not always possible to draw a line between productive and
non-productive public goods. The implication is that some value judgment will
be required here as well.
25
References
Alesina, Alberto and Allan Drazen (1991) "Why Are Stabilizations Delayed?",
American Economic Review 81(5): 1170-88
Alesina, Alberto, Nouriel Roubini and Gerald Cohen (1999) "Political Cycles
and the Macroeconomy", MIT Press, Cambridge
Alesina, Alberto and Guido Tabellini (1990) "A Positive Theory of Fiscal
De…cits and Government Debt", Review of Economic Studies 57: 403-14.
Beetsma, Roel and Xavier Debrun (2004) "Reconciling Stability and Growth:
Smart Pacts and Structural Reforms", IMF Sta¤ Papers 51(3): 431-456.
Beetsma, Roel and Xavier Debrun (2005) "Implementing the Stability and
Growth Pact: Enforcement and Procedural Flexibility", IMF Working Paper
WP/05/59.
Beetsma, Roel and Harald Uhlig (1999) "An Analysis of the Stability and
Growth Pact", Economic Journal 109: 546-571.
Blanchard, Olivier Jean and Francesco Giavazzi (2004) "Improving the SGP
Through a Proper Accounting of Public Investment", CEPR Discussion Paper
No. 4220.
Brunila, Anne, Marco Buti and Daniele Franco, eds. (2001) The Stability
and Growth Pact, Basingstoke: Palgrave
Fabrizio, Stefania and Mody Ashoka (2006), "Can budget institutions coun-
teract political indiscipline ?", IMF Working Paper 06/123
Giuliodori, Massimo and Roel Beetsma (2004) "What are the Spill-overs
from Fiscal Shocks in Europe? An Empirical Analysis", ECB Working Paper
No. 325.
Giuliodori, Massimo and Roel Beetsma (2007) "On the relationship between
…scal plans in the European Union: An empirical analysis based on real-time
data. CEPR Discussion Papers 6088
Hallerberg, Mark and Jurgen von Hagen, (1999), "Electoral Institutions,
Cabinet Negotiations, and Budget De…cits in the European Union", in: James
Poterba and Jürgen von Hagen (eds.), Fiscal Institutions and Fiscal Perfor-
mance, University of Chicago Press.
Hallerberg, Mark and Guntram B. Wol¤ (2006) “Fiscal Institutions, Fiscal
Policy and Sovereign Risk Premia”, Bundesbank DP 35-2006.
Krogstrup, Signe (2006) "A note on the formal equivalence of the time incon-
sistency model and the common pool model for budget de…cits", The Graduate
Institute of International Studies, Geneva.
Krogstrup, Signe and Charles Wyplosz (2006) "A common pool theory of
de…cit bias correction", CEPR Discussion Paper 5866
Persson, Torsten and Guido Tabellini (2000) Political Economics, Explaining
Economic Policy, MIT Press.
Persson, Torsten, Gerard Roland and Guido Tabellini (2003) "How Do Elec-
toral Rules Shape Party Structures, Government Coalitions, and Economic Poli-
cies?", NBER Working Papers 10176
Roubini, Nouriel and Je¤rey Sachs (1989) "Government Spending and Bud-
get De…cits in the Industrial Countries", Economic Policy 8: 100-132.
26
Velasco, Andres (1999) "A Model of Endogenous Fiscal De…cits and De-
layed Fiscal Reforms", in: James Poterba and Jürgen von Hagen (eds.), Fiscal
Institutions and Fiscal Performance, University of Chicago Press.
Velasco, Andres (2000) "Debts And De…cits With Fragmented Fiscal Poli-
cymaking" Journal of Public Economics 76(1): 105-125.
Von Hagen, Jürgen (1992), "Budgeting Procedures and Fiscal Performance
in the European Communities", Economic Papers 96, European Commission,
Brussels.
Von Hagen, Jürgen and Ian Harden (1994) “National Budget Processes and
Fiscal Performance", European Economy Reports and Studies 3, 311-418.
Von Hagen, Jürgen, and Ian Harden (1995) "Budget Processes and Com-
mitment to Fiscal Discipline", European Economic Review 39(3): 771-779.
Weingast, Barry, Kenneth Shepsle and Christopher Johnsen (1981) "The Po-
litical Economy of Bene…ts and Costs: A Neoclassical Approach to Distributive
Politics", Journal of Political Economy 89(4): 642-664.
Wyplosz, Charles (2005) "Fiscal Policy: Institutions Versus Rules", National
Institute Economic Review 191: 70-84.
27
