THE POLITICS OF STRATEGIC BUDGETEERING
Christina J. Schneider
University of California, San Diego
cjschneider@ucsd.edu
Vera E. Troeger
University of Warwick
v.e.troeger@warwick.ac.uk
Very preliminary draft, please do not circulate. Comments welcome.
Abstract
Governments want to be reelected but increasing fiscal transparency has imposed many challenges for them to pursue opportunistic fiscal policies. This paper demonstrates that governments
can electioneer even when fiscal transparency is high. We argue that governments can use several strategies to increase public good provision before elections, and they choose the strategies
that are most effective given the level of transparency. Whereas deficit spending is an effective
strategy to artificially increase voters’ perceived welfare before elections when voters cannot observe these distortionary practices, increasing fiscal transparency makes these strategies more
costly (conservative voters would punish the government). Consequently, when fiscal transparency is high governments resort to less visible strategies, such as the redistribution of budgetary
resources from long-term efficient investment spending to short-term consumption spending. We
test the predictions with data on the composition of government spending for 32 countries over
up to 38 years and data on individual budget items for 17 OECD countries over 35 years. The
preliminary findings suggest that governments indeed redistribute resources from long-term efficient investment to short-term public goods provision before elections especially if elections are
contested.
_______________________________
Notes: Equal authorship. For great comments and insights, we would like to thank Mark Hallerberg, Mark Kayser,
Thomas Pluemper, Branislav Slantchev, XXX. The paper was presented at IPES (2014), NYU Abu Dhabi (2014),
Princeton-Warwick Political Economy workshop in Venice (2014). We thank all the participants for their helpful
comments and suggestions. We gratefully acknowledge financial support of the German Research Foundation.
1
INTRODUCTION
Politicians prefer winning elections to losing them. For this very reason, governments in democratic countries have strong incentives to implement policies that maximize their political support in the short run even at the expense of long-term costs. One strategy to improve the likelihood to get reelected is by distorting fiscal policy. A political budget cycle occurs when incumbents increase the provision of (usually highly visible) public goods before the election, financed
either through higher deficits or increased taxes in the long run.1 At the same time, voters are fiscal conservatives. They prefer governments who improve their economic welfare without imposing extensive debts or increase the tax burden in the long run (Peltzman 1992; Alesina et al.
1998; Arvate et al. 2009; Brender and Drazen 2008; Drazen and Eslava 2010). The ability of
voters to punish politicians for opportunistic fiscal policies crucially depends on the availability
of information about politicians’ fiscal policies. Institutions, which improve fiscal transparency,
provide this information to voters and serve to impede governments from creating a political
budget cycle (Alt and Lassen 2006; Shi and Svensson 2006; Klomp and de Haan 2013; Vicente
et al. 2013). This has led many scholars to argue and show that political budget cycles are unlikely to occur in developed democracies, and restricted to new democracies or developing countries
in general (Brender and Drazen 2005; Drazen and Eslava 2009).
Have these reforms eliminated the ability of governments in developed democratic countries
to pursue opportunistic fiscal policies before elections? We argue that this conclusion is too optimistic. When governments manipulate fiscal policies before elections, they have several instruments at their disposal. They can raise deficits or taxes to provide more public goods before
elections, but they can also change the composition of the budget strategically to achieve the
same goal. Compositional spending occurs governments shift resources from investment spending, which is efficient to generate economic welfare in the long run, to consumption spending,
which improves political support in the short run. This allows governments to boost spending on
electorally relevant items, without having to increase the domestic deficit. Whereas deficit
spending is more easily detected in fiscally transparent systems, compositional spending is more
difficult to observe and governments can rely on this strategy when deficit spending is not feasible anymore. Our theory of strategic budgeteering analyses incentives and opportunities to make
use of these different strategies, and thereby produces a number of empirical implications about
the conditions under which governments rely on one or the other fiscal strategy opportunistically
before elections.
We test our theory using two different data sets. First we look at compositional government
spending where spending is either consumption or investment to distinguish between long term
efficient spending and short term public good provisions. As alternatives to compositional spending we also consider deficits, tax revenues and monetary instruments in 32 countries over a period of up to 38 years. The second set of analyses uses data on individual budget items that can be
either predominantly consumptive or long-term efficient investment. This data is available for 17
OECD countries and 35 years. Evidence for redistribution of resources from investment to consumption before elections if fiscal transparency is high should be detectable in both empirical
investigations.
Our findings provide interesting insights into the existing literature. First, whereas previous
approaches have analyzed individual fiscal strategies in preelection periods, such as deficit
1
Alternatively, governments could use monetary policy to create political business cycles. See Franzese (2002) for
an excellent overview of these approaches.
2
spending or compositional spending, separately (Schuhknecht 2000; Block 2002; Alt and Lassen
2006a,b; Veiga and Veiga 2007; Schneider 2010; Katsimi and Sarantides 2012), we use the insights of these analyses to develop an integrated theory of strategic budgeteering. Importantly,
the theory sheds light on how governments choose between alternative fiscal instruments in the
pre-election period. To the best of our knowledge, we provide the first empirical analysis that
analyses these different fiscal strategies in a common theoretical framework, thereby showing
how fiscal transparency affects the governments choice of different fiscal instruments. Second,
recent work on political budget cycles has focused on analyzing developing countries and new
democracies because it was assumed that budget cycles can only exist in these countries (Block
2002; Shi and Svensson 2006; Drazen and Eslava 2009; Saporiti and Streb 2008; Vergne 2009).
Our results support the underlying notion that voters are fiscally conservatives who would punish
incumbents for pursuing costly fiscal strategies before elections. At the same time, we show that
budget cycles in developed democracies are likely, and exist. Incumbents in industrialized countries just use different fiscal strategies to achieve electoral gains. The result is important because
it demonstrates that governments can substitute fiscal strategies when particular strategies are not
available to them. Finally, most of the recent efforts to analyze compositional spending focus on
local elections (Khemani 2004; Veiga and Veiga 2007; Drazen and Eslava 2009; Schneider
2010; Vicente et al. 2013). In contrast, our paper analyses compositional spending on the national level in a sample of 19 developed democracies.2
A VERY SHORT HISTORY OF STRATEGIC ELECTIONEERING
It is common wisdom that governing parties have a significantly higher chance to win elections
during periods of economic booms. This simple logic provides incumbents with ample incentives
to stimulate the economy shortly before elections even at the expense of the long-term detrimental effect. If governments act purely opportunistic and if voters consider current conditions
more than previous conditions, then the incumbent is tempted to use all available policy instruments to create a business cycle hike before elections at the cost of worsening economic conditions shortly after the elections (Wright 1974, Tufte 1978, Frey and Schneider 1978a,b, Golden
and Poterba 1980, Schultz 1995, Price 1998).3
Initially, scholars applied the logic of the Philips curve (which constitutes an inverse relationship between inflation and unemployment) to provide an opportunistic explanation of political business cycles (e.g., Nordhaus 1975; Tufte 1978).4 In these models, the electorate votes retrospectively and governments expand monetary policies to lower unemployment before elections. Later approaches abolish the assumption that expectations about the inflation are adaptive
and treat voter expectations as rational and anticipatory (Alesina 1987, 1988; Rogoff and Sibert
1988; Persson and Tabellini 1990). Governments cannot simply generate political business cycles because voters (who anticipate that lower unemployment before the election will lead to
higher inflation after the election) adjust their behavior. Consequently, incumbents are attracted
2
3
4
The article by Katsimi and Sarantides (2012) is one exception where the authors analyze budget composition in
high-income OECD countries.
This idea can be traced back at least to Schumpeter (1939), Kalecki (1943), Nordhaus (1975), Hibbs (1977, 1978)
and Tufte (1978).
Governments can still generate business cycles (in the short term) if one assumes a NAIRU-augmented (nonaccelerating inflationary rate of unemployment) version of the ‘Philipps curve’.
3
to expansionary monetary policies only if they can signal economic competence to the electorate
(Persson and Tabellini 1990).
The relevance of monetary business cycles has declined because – in the course of international economic and financial integration and the emergence of independent central banks – few
governments still autonomously command monetary policy instruments (Obstfeld and Rogoff
1996; Lohmann 1992, 1997; Jonsson 1995; Simmons 1996; Boix 1998, 2000; Clark et al. 1998;
Garrett 1998; Franzese 1999, 2003; Oately 1999; Clark and Hallerberg 2000; Way 2000; Bernhard et al. 2002; Broz 2002; Clark 2002a; Broz and Frieden 2001). With the declining relevance
of monetary political business cycles, scholars started to focus on fiscal political business cycles
(also called political budget cycle). The theory of the political budget cycle is based on the premise of rational and forward-looking voter. Voters prefer candidates who are able to provide more
public goods for given levels of taxation and private consumption (e.g. Rogoff and Sibert 1988;
Rogoff 1990; Shi and Svensson 2002; Alt and Lassen 2006a, b). In these models, the electorate
is not fully informed about the incumbent’s competence. Additionally, they do not observe the
current levels of debt. Governments try to appear competent by temporarily raising economic
growth or improving the welfare of large numbers of citizens before an election by providing
more public goods.5 Whereas truly competent governments could pursue policies that increase
public good provision without distorting the economy in the long run, less competent governments cannot do so. Strong electoral incentives, however, will make them want to appear competent by financing increased public goods provision with higher deficits. Voters observe the increased provision of public goods and attribute this to the economic competence to the government because they cannot observe the distortionary budgetary politics behind this (i.e. the fact
that the government appears economically competent by raising deficits). Since voters would
increase their support for incumbents whom they deem economically competent, incumbents
have incentives to use deficit spending to appear competent even if it distorts the economy in the
long run. Consequently, both competent and incompetent governments use increased public
goods provision as a strategy to get reelected.
The straightforward theoretical predictions find surprisingly little empirical support. Whereas there is evidence for an increase in targeted expenditures before elections (Tufte 1978, Alesina
and Roubini 1992; Franzese 2002b), only few studies report opportunistically motivated deficit
spending (Alesina et al. 1992, 1997; Shi and Svensson 2006; Persson and Tabellini 2003a). Other scholars find no or a negative relationship between the pre-election period and higher spending or deficits (Brender and Drazen 2004; Jochimsen and Nuscheler 2005; DeHaan and Sturm
1994; Drazen and Eslava 2005; Seitz 2000). The inconclusive findings raised doubts about the
very existence of political budget cycles. To explain why many empirical cross-country and single-country analyses fail to find consistent evidence for deficit-induced budget cycles, recent
work has been concerned with the conditions under which incumbents cannot imbalance their
finances to manipulate the economy before elections. These models are based on the empirical
observation that voters tend to punish governments if they increase deficits before elections expecting that “a more ‘competent’ policymaker can expand government spending or reduce taxes
and still not induce the distortion that a less ‘competent’ policymaker would induce” (Drazen
5
Partisan theories to political business cycles argue that parties have different affinities for example to increase
spending on different policy fields (Hibbs 1977; Alesina 1989; Cusack 1997; Boix 2000).
4
2000b: 101).6 In other words, the argument that governments signal competence remains unconvincing in this case, since no competence is needed to run deficits.
The ability to observe deficit spending thus became the central focus of a new generation of
research on political budget cycles. The claim is that incompetent governments may only feign
competence if the share of informed voters is relatively low and the rents of remaining in power
are high (Shi and Svensson 2002). The amount of information the voter receives about the incumbent’s actions (and thus the extent to which electioneering is possible) may depend, for example, on the transparency of fiscal institutions within a country (Alt and Lassen 2006; Shi and
Svensson 2006; Klomp and de Haan 2013; Vicente et al. 2013). This argumentation explains
some of the mixed findings very well. Research has consistently shown that political budget cycles are restricted to weak and new democracies, that is, to those countries where voters have restricted capacity to monitor and evaluate the fiscal policy process (e.g., Schuhknecht 1994;
Akhmedov and Zhuravskaya 2004; Hallerberg, de Souza and Clark 2002; Persson and Tabellini
2002b, 2003; Shi and Svensson 2000, 2002, 2006; Brender and Drazen 2005; Drazen and Eslava
2009). State-enforced media reinforces the effect of low transparency (Ferejohn 1999). In sum,
all these findings point to the importance of information and fiscal transparency as an important
factor restraining the incumbents’ ability to pursue opportunistic deficit spending in the preelection period.7 As Brender and Drazen (2005) famously concluded: “In new democracies it is possible to carry out such manipulation, whereas in more established democracies, voters have the
ability to identify fiscal manipulation and punish such behaviour, so that politicians avoid it.”
The new generation of political budget cycle research therefore increasingly focused on analysing political budget cycles in developing countries (Schuknecht 2000; Block 2002; Sapority and
Streb 2008; Drazen and Eslava 2009; Vergne 2009; Vicente et al. 2013).
Is the political budget cycle dead in the developed world as much of the literature suggests?
In this paper we argue that it is not. To support this claim we make two interrelated arguments.
First, the theoretical literature points to the importance of fiscal transparency. Whereas fiscal
transparency tends to be higher in high-income maturely democratic countries than in lowincome newly democratic countries, there is still significant amount of variation in the level of
fiscal transparency across high-income democracies. Thus, we would expect deficit spending to
take place in countries where voters cannot observe distortionary opportunistic spending. In fact,
Alt and Lassen (2006) show that low fiscal transparency increases the likelihood that OECD
countries use deficit spending before elections. Second, and more importantly, we build on the
literature of compositional spending to argue that even if fiscal transparency is high in developed
democracies, they can still expand public goods provision before elections if they redistribute
spending from long-term investment spending to short-term consumption spending.8 The redistribution of the budget enables governments to increase public goods provision without having to
raise deficits. We will argue below that compositional spending is less easily detectable than deficit spending. Based on this assumption we will show that governments use deficit spending
when fiscal transparency is low, but revert to compositional spending when fiscal transparency is
6
For empirical support see Alesina, Perotti and Tavares 1998; Brender 2003; Brender and Drazen 2005; Drazen and
Eslava 2005; Peltzman 1992; Schneider 2007.
7
Milesi-Feretti (2004) and Rose (2006) further examine the impact of fiscal rules on the scope for political budget
cycles.
8
The literature analyzes compositional spending usually independent of deficit spending, and mainly in the context
of developing countries (Schuknecht 2000; Block 2002; Sapority and Streb 2008; Drazen and Eslava 2009;
Vergne 2009; Vicente et al. 2013). Schneider 2010 and Katsimi and Sarantides are exceptions as they find some
evidence for compositional spending in developed countries.
5
high. In fact, our argument is consistent with some more recent findings about compositional
spending in developed economies (Schneider 2010; Katsimi and Sarantides 2012). The strategic
use of different fiscal strategies for reelection purposes is what we call strategic budgeteering.
A THEORY OF STRATEGIC BUDGETEERING
We develop a theoretical argument about how opportunistic governments choose between different fiscal strategies in order to maximize their chances to get re-elected by rational and prospective voters. In a nutshell, we argue that incumbents want to spend more on public goods if elections are close even though these short-term investments are not welfare enhancing in the long
term. Whereas they prefer to use deficit spending to generate public goods, higher fiscal transparency decreases their opportunities to pursue this strategy. Rather than deciding not to electioneer, increasing fiscal transparency will induce governments to substitute deficit spending
with compositional spending. Consequently, deficit spending should be more prevalent in countries with low fiscal transparency, whereas compositional spending should be more prevalent in
countries with high fiscal transparency.
Our theory is based on a number of standard political economy assumptions. Voters care
about their expected wellbeing and choose whom to vote for based on their expected income in
the time after the elections. Whereas voters value increased consumption possibilities, they are
fiscal conservatives. They want the government to increase public good provision because they
are competent, not because they increase deficits or use other distortionary policies to achieve
that goal (Peltzman 1992; Alesina et al. 1998; Arvate et al. 2009; Brender and Drazen 2008;
Drazen and Eslava 2010). To assess whether the incumbent is likely to perform in the future, the
voter can use past performance as an indicator. This incentivizes incumbents to signal economic
competence before elections. To keep the theory parsimonious we assume a set of domestic institutions where opportunistic monetary policies are not feasible in the preelection period. This
simply implies that countries either experience high central bank independence, or have pegged
exchange rates, so their ability to use expansive monetary policies is limited. According to Clark
and Hallerberg (2000), governments who have no monetary policies at their disposal are more
likely to rely on fiscal policies to maximize voter welfare. Whereas we make this assumption to
render the theoretical argument more parsimonious we will show in the empirical section that
this assumption in fact holds for our sample of countries (all of which experience high central
bank independence or pegged exchange rates).9
Consequently, politicians can only rely on fiscal policies when trying to appear competent to
their electorate. To improve the welfare of their voters before elections, incumbents may choose
between alternative fiscal strategies. They can (a) raise taxes, (b) increase deficit spending, (c)
increase compositional spending, and (d) they can reduce taxes.10 Strategies (a), (b), and (c) all
9
An interesting extension to our theory strategic budgeteering would be to integrate monetary policies. Governments that possess these policies would then be able to choose between monetary policies, deficit spending, and
compositional spending. Importantly, we would assume that the choice between monetary and fiscal strategies
more broadly depend on the international economic pressures (Clark and Hallerberg 2000). The choice between
compositional and deficit spending should be more relevant in situations where exchange rates are fixed or central bank independence is high.
10
We distinguish between long-term and short-term spending unlike Drazen and Eslava (2005) who separate targeted from non-targeted spending. We concur that this is a useful distinction but argue that both targeted and nontargeted spending are part of short-term public good provision before elections, and targeted spending is electorally more relevant in the developing context.
6
serve to pay for the increase in public goods provision, whereas strategy (d) increases voters’
perceived welfare directly. In the following, we discuss these strategies and how attractive they
are to incumbents under different conditions (see also Table 1 for a summary). Importantly, we
assume that incumbents prefer to use some fiscal strategy rather than abstaining from pursuing
opportunistic strategies in the electoral period altogether.
When choosing between different fiscal instruments incumbents have to assess the benefits
and costs of each individual strategy relative to all other strategies. Let’s first consider the feasibility of raising taxes before elections in order to finance greater public goods provision (option
a). Whereas this strategy gives the incumbent a guaranteed and long-term access to additional
resources for public goods provision, it comes with a relatively high cost: higher taxes are immediately visible to voters in the, and reduce their welfare in the preelection period, therefore likely
reduces the voters’ support for the government. Raising taxes is therefore a strategy that may actually undermine the incumbent’s goal to get reelected. Supporting this interpretation, we are not
aware of any empirical results that support such a strategy.
Table 1: Fiscal Strategies in the Preelection Period
Advantages
(a) Raise Taxes
Prevents raising deficits
and does not distort the
economy in the long run
(b) Deficit Spending
Less visible to raising
taxes
(c) Compositional
Least visible
Spending
(d) Reduce Taxes
Disadvantages
Very visible to voters who
dislike higher taxes
Distorts economy in the
long run (greater debt)
Distorts economy in the
long run (less investment
spending)
Visible to voters who Inflexible instrument for
like low taxes
short-term electoral politics
Governments could also decrease taxes in order to provide an immediate welfare benefit to
voters (option d). Since the change in taxation is immediately visible to voters it could provide
the government with the desired boost in public support. The empirical evidence for this strategy
is mixed, and evidence in favour of the tax-reduction hypothesis is not very robust (Alesina et al.
1993; Persson and Tabellini 2003; Brender and Drazen 2003; Veiga and Veiga 2007; Katsimi
and Sarantides 2012). For examples, the findings do not tend to be robust across different operationalizations of tax revenues and oftentimes are inconsistent across studies (e.g. Katsimi and
Sarantides (2012) find a significant effect for direct taxation (a null effect for indirect taxation),
whereas Erhart (2010) finds a significant effect for indirect taxation (a null effect for direct taxation). In addition, as Eslava (2011) points out, it is not clear whether the reduction in revenues
owes to a decrease in taxes or some other strategy the government uses to reduce revenue collection before elections. One of the main disadvantages of using taxation as an instrument to boost
public goods provision are the difficulties in achieving tax cuts on short notice. Tax cuts have to
go through the whole legislative process and are therefore much less flexible to control. Most
importantly, tax cuts are very difficult to reverse undetected after the elections, and may therefore be counterproductive in the medium term.
Consequently, whereas some governmnents may rely on lowering taxes as a electoral strategy, we assume that governments typically choose between deficit spending and compositional
7
spending as most attractive and flexible policies. The main advantage of using deficit spending
and compositional spending over tax policies is the flexibility and the lower visibility of distorting actions. It is important to note that although we focus on these two strategies, we will
analyse in the empirical section whether our assumption holds, that is, governments do not systematically exploit taxation policies before elections to gain political support.
To understand the choice between compositional spending and deficit spending we have to
analyse the relative costs and benefits of using each of these fiscal instruments in the electoral
period. Deficit spending refers to the strategy where the government increases the domestic deficit in order to finance an increased level of public goods provision before the election. This implies that incumbents have to somehow reduce the deficit after the elections in order to balance
the budget. Deficit spending is therefore costly to voters in the long run.
Compositional spending refers to a strategy where the governments shift resources from
budget items that are efficient to generate economic welfare in the long run to budget items that
improve political support in the short run.11 Public budgets are divided into a number of different
budget items (such as defense, social security, education, etc.) and governments may choose to
increase or decrease spending on either of them. We assume that governments may principally
raise expenditures on two different types of public goods – long-term efficient investment and/or
short-term efficient consumption. Public goods that are efficient in the long term have no immediate welfare implications for the voters. Rather, the welfare implications arise in the future (and
therefore after the election). Examples of such long-term public goods are investments in public
order and safety, education or health. In other words, the provision of these public goods is important for voter welfare in the long-term, but they do not increase voter welfare right before the
election. Short-term public goods are efficient only in the short term. That is, they increase voter
welfare in the short-term, but since they are purely consumptive they do not have a long-lasting
positive effect on voter welfare. Examples of such highly visible short-term policies are labor
market programs, social security spending, or other social transfer payments. Whereas voters
generate utility from private consumption and short-term public goods right before the election
they only gain from long-term public goods after the election. Opportunistic governments can
therefore increase the voters’ welfare directly before elections by providing more short-term efficient public goods in the preelection period. Investing in short-term public goods for electoral
reasons implies a reduction in resources available for investment in long-term public goods,
which bears distortionary effects in the long run.
To sum up to this point, both deficit spending and compositional spending are less visible to
voters than strategic taxation, so they should be the preferred instrument to generate political
budget cycles. In addition, both strategies are distortionary in the long run – deficit spending
leads to lower spending or higher taxes after the elections and strategic budgeteering leads to
long-term costs through low investments in the short term.
The big advantage of deficit spending is its flexibility vis-à-vis compositional spending.
Governments are restricted to remain within the budgetary limits when they change the composition of the budget from investment to consumption spending. In addition, some investment
11
In the literature, there are two types of compositional spending. One strategy is to shift resources from less visible
to highly visible expenditure items (i.e. more consumption spending than investment spending). The other strategy is to shift resources from non-targeted items to targeted items. The importance of targeted spending appears to
be more pronounced in developing countries that are oftentimes characterized by highly clientelistic domestic
politics (Drazen and Eslava 2005). Since we are interested in political budget cycles in high-income countries,
where voters care about public goods provision in general, we focus on the strategy that shifts spending towards
consumption spending, thereby increasing short-term public goods provision.
8
spending cannot be revised easily in the short term, which provides additional limitations for
governments before elections. These limitations limit the government’s ability to expand public
good provision. In contrast, governments can raise deficits much more easily and are less restricted in the amount of deficits that they incur as they are not forced to keep a balanced budget
in the short term. Consequently, deficit spending should provide the government with more leverage to provide more public goods before elections than compositional spending.
The big disadvantage of deficit spending is its feasibility if fiscal policies are transparent visà-vis compositional spending. Voters derive their utility from voting for the incumbent depends
on their welfare in the short and the long term. Whereas they can directly observe their welfare in
the preelection (current) period, any welfare benefits in the long-term are not fully anticipated by
voters and therefore discounted. At the same time, they expect lower future well-being the higher the public deficit incurred by the government in the preelection period. Consequently, and in
line with the literature, voters are fiscally conservative and prefer governments not to incur high
deficits. How well voters can observe debt-creation by the government depends on how transparent the fiscal system is (e.g. Alt and Lassen 2006).12 We define fiscal transparency as
“public openness about the structure and functions of government, fiscal
policy intentions, public sector accounts, and projections. It involves ready access to reliable, comprehensive, timely, understandable, and internationally
comparable information on government activities (…) so that the electorate
and financial markets can accurately assess the government’s financial position
and the true costs and benefits of government activities, including their present
and future economic and social implications” (Kopits and Craig 1998, 1).
Fiscal transparency determines the visibility of deficit-creation by the government and indicates
the need to re-balance the budget in the period after elections. We argue that the deficit spending
is more visible and problematic for two reasons. First, higher deficits are more visible than lower
investment spending even in high transparency countries. Rogoff (1990), for example, argued
that voters can observe investment spending only after the elections whereas deficit spending is
immediately visible as long as fiscal transparency is high. The problem for voters to correctly
assess investment spending is that it usually involves long-term projects where any assessment
about the inputs is difficult as long as they are on-going (Block 2002). Voters cannot observe the
output of investment expenditures immediately when they are incurred during the election year,
but rather observe the results of investment spending from previous years (i.e. those investment
projects that bore fruit by the time the election comes). Consequently, voters would have to track
the government’s investment expenditures in the current budget itself because those are usually
not discussed by the media or even by opposition parties. The media and opposition parties
mainly use deficit-creation of the incumbent government to point out her incompetence in the
political competition. It is therefore more difficult for voters to assess the government’s investment spending patterns than to assess the government’s deficit spending patterns. Second, voters
are more likely to interpret higher deficits as distortionary and opportunistic policy than lower
12
Note, since we assume that our incumbents are elected, we do not need to take into account the quality of democratic institutions to measure the visibility of governmental fiscal policies (e.g. Shi and Svensson 2000, 2002,
2006). However, both – fiscal transparency and democratic quality – have the same notion as they measure the
extent to which governmental debt-creation is visible. Consequently, we would expect the substitution effects between deficit spending and compositional spending to hold for different levels of democratic quality.
9
investment spending. Empirical research shows that voters perceive large deficits as signal that
the economy is not doing well and tend to punish the incumbent for bad economic policy outcomes (Peltzman 1992; Kraemer 1997; Brender 2003; Brender and Drazen 2005; Drazen and
Eslava 2005). The same pattern has not been observed for incumbents who increase short-term
transfers in the year before elections. To the contrary, Veiga and Veiga (2006) show at least for
Portugal that preelectoral increases in highly visible spending tends to increase public support;
whereas changes in long-term efficient spending has no effect on public support.
Consequently, the main trade off for governments when choosing between deficit spending
and compositional spending is the question of flexibility versus feasibility (with respect to
achieving political support) under different levels of fiscal transparency. When fiscal transparency is low, governments can use both fiscal strategies to increase the welfare of their voters, but
they prefer to use deficit spending because this is a more flexible instrument that allows them to
provide more public goods before elections. When fiscal transparency is high, deficit spending
becomes an infeasible strategy to governments who want to maximize political support. Even
though compositional spending may not be as flexible of an instrument it will be more feasible
and therefore more likely to be employed by opportunistic incumbents. This leads us to the following hypotheses:
Hypothesis 1: The lower fiscal transparency, the greater the likelihood that governments
use deficit spending to increase public goods provision in preelection periods, ceteris paribus.
Hypothesis 2: The lower fiscal transparency, the smaller the likelihood that governments
use compositional spending to increase public goods provision in the preelection period,
ceteris paribus.
Hypothesis 3: The greater fiscal transparency, the less likely governments use deficit
spending to increase public good provision in the preelection period, ceteris paribus.
Hypothesis 4: The greater fiscal transparency, the more likely governments use compositional spending to increase public goods provision in the preelection period, ceteris paribus.
Whereas hypotheses 1-4 reflect the most straightforward predictions that can be deducted from
our theory, there are other empirical implications that allow us to test the theoretical mechanism
more closely.
First, the existing literature suggests that governments’ incentives to use distortionary fiscal
policies before elections should increase the greater the electoral uncertainty. If governments are
not certain to be reelected they will be more willing to incur the long term costs of deficit spending and compositional spending. There are various sources of electoral uncertainty, but most important to the government may be the closeness of elections. The more competitive the electoral
race, that is, the closer the opposition and the incumbent are in the public opinion polls, the
greater the incentive for the government to get an electoral advantage using fiscal instruments.
Consequently:
Hypothesis 5: The closer the election, the more likely governments use deficits spending
and compositional spending to increase public good provision in the preelection period,
ceteris paribus.
10
Second, above we argue that voters perceive the costs of deficit spending as more distortionary
than the costs of reducing investment spending. Another way of analyzing the effect of long-term
costs on the likelihood that distortionary fiscal policies are implement is to analyze the reaction
of governments to variations in the long-term costs of deficit spending when fiscal transparency
is high. There are two potential sources of costs that seem important. First, there may be variations in the long-term fiscal costs of opportunistic deficit spending. The negative effect of fiscal
transparency on the likelihood that incumbents can use deficit spending is particularly pronounced if maintaining deficits is more costly. In particular, the higher the interest rates the more
costly it will be for the government (and the voters) to balance the budget after the elections.
Consequently, higher interest rates should provide an additional impetus to refrain from deficit
spending when fiscal transparency is high. Second, there may be variations in the institutional
setting with effects for opportunistic government spending. If governments have to abide to domestic or international institutions that have budget balance requirements, then the likelihood
that governments use deficit spending should decline. One example of such institutions are is the
Stability and Growth Pact in the European Union, which restricts EMU countries to limit their
budget deficits to 3% of domestic GDP. Such requirements will limit the feasibility of deficit
spending (although there clearly are differences in the credibility of threats across member
states) and increase incentives to use compositional spending. This provides us with additional
interesting empirical implications that shed more light on the theory of strategic budgeteering:
Hypothesis 6: The higher interest rates, the less likely governments use deficits spending to
increase public good provision in the preelection period, particularly when fiscal transparency is high, ceteris paribus.
Hypothesis 7: EMU governments are less likely to use deficits spending to increase public
good provision in the preelection period than non-EMU countries, particularly when fiscal
transparency is high, ceteris paribus.
Hypothesis 8: EMU governments are more likely to use compositional spending to increase public good provision in the preelection period than non-EMU countries, particularly when fiscal transparency is high, ceteris paribus.
In a way of summarizing, governments in mature and developed democracies still have both motive and opportunities to use fiscal policies to increase their chances of staying in power after
elections. Our theory of strategic budgeteering provides insights into how governments choose
alternative strategies, in particular deficit spending and compositional spending, given their flexibility and feasibility. Whereas deficit spending is preferred due to its relative flexibility, which
allows the government to boost consumption spending more than would be possible under compositional spending, its feasibility declines when fiscal transparency is great. In this situation,
compositional spending substitutes for deficit spending as an instrument to increase public goods
provision opportunistically. We derived several hypotheses from our theory that we will test in
the next section.
EMPIRICAL ANALYSIS
In the following we attempt to put the predictions of the theoretical model to the test. We use
two different data sets to investigate our theoretical propositions. First we simply look at the
composition of government spending into consumption (current spending) or investment (capital
spending) to investigate whether governments indeed move resources from long-term to short11
term expenditure if fiscal transparency is high. In addition we check for electoral cycles in deficit
spending and under what conditions (level of transparency) deficit spending is used to finance
short term public good provision. Alternatives to strategic budgeteering (tax rates and revenues,
monetary policy measures) are investigated for electoral cycles as well.
In a second step we investigate data on the composition of public expenditure for 17 OECD
countries over 35 years.13 The data provides information on 23 spending categories ranging from
expenditure for defense, public order and safety to education, social security and welfare, agriculture, mining and manufacturing, as well as infrastructure such as roads and railways. Table 2
describes all items in more detail. We employ both total expenditure and relative expenditure per
item as the dependent variable.
Compositional Spending
We start our empirical investigation by looking at the two interesting components of government spending: consumption (current spending) and investment (capital spending). We use
OECD national accounts data, specifically general government final consumption expenditure
(consumption) and general government gross fixed capital formation (investment). Our first dependent variable is the ratio of the two (consumption/investment, both measured in millions of
US dollars). This allows testing for a direct shift from investment to consumption (the ratio increases) in the pre-electoral period. We also use both variables individually as left-hand-side
(LHS) variables as percentage of GDP to see direct pre-electoral effects. In order to analyze both
aspects of strategic budgeteering we investigate electoral cycles in deficit spending (measured as
annual changes in debt per GDP (OECD and WDI if missing) as well as the effect of preelectoral deficit creation to increase public good provision (consumption)14. In addition to these
main features of strategic budgeteering we test for the existence of electoral cycles in fiscal and
monetary alternatives, specifically we look at inflation and real interest rates (WDI), tax revenue
per GDP (WDI), statutory corporate tax rates and effective average corporate tax rates (Devereux
et al. 2002) as well as average income tax rates (OECD).
We are mostly interested in pre-electoral effects. To determine the pre-election period we
simply generate a dummy that takes the value one for the 12 month prior to a legislative election
and zero otherwise. The data comes from the Database of Political Institutions (DPI 2012) of the
World Bank (Keefer 2004).15 In addition we look at the effect of pre-electoral deficit spending
on the composition of the budget.
The theoretical model predicts that strategic reshuffling of resources depends on the fiscal
transparency of the country and (to a lesser extent) the ex ante probability of winning an election
(the closeness of the electoral competition). We operationalize the first condition by using data
from the Open Budget Survey that is conducted on the bi-yearly bases by the International Business Partnership. The survey consists of a battery of 125 question that range from questions
about the budgetary process, e.g. at what point in the process the executive has to provide de13
The 17 countries are: Australia, Austria, Belgium, Canada, Denmark, Finland, france, Germany, Iceland (not for
all models), Italy, Japan, The Netherlands, New Zealand, Norway, Sweden, the United Kingdom, and the US.
The data collection was funded by the German Science Foundation (DFG).
14
We use changes in debt instead of deficit since the data is more comprehensive and covers more cases. Also we
believe that changes in debt are the more accurate measure because of the complicated computation of deficits.
15
We experimented with two pre-election years with no substantive changes to the main estimation results. For a
better operationalization of the pre-electoral period and a more detailed discussion see Franzese (2002, 2003)
12
tailed reports, what institutional actors are involved in the budgetary decisions, who has veto
power, who needs to be consulted (e.g. whether the executive is required to seek input from the
legislative when shifting resources between budget items - question 103), whether there is an audit process, who carries out the audit process, legal and regulatory restrictions on spending decisions etc. In addition, questions about involvement of the public are asked, e.g. how often does
the government have to publish reports on the budget, how detailed does the reports have to be
and what information has to be disseminated in what form to the public, does the government
only have to publish information about the agreed budget or also about projected spending and
changes, does the government have to lay out how the proposed spending is related to stated policy goals, is the government formally required to engage the public, what channels and formal
mechanisms (e.g. public hearings) exist for the public to feed input into the budgetary process,
does the executive provide formal feedback on how public input has affected budget decisions,
are audit reports publicly available etc. These survey questions are combined into sub-indexes
e.g. for the executives budget proposal, the audit process and the public engagement and an
overall index, the Open Budget Index (OBI), which ranges between 0 and 100, whereas larger
values indicate more open budgets and thus more fiscal transparency. We use the overall index
since it correlates highly with the sub-indexes. Unfortunately, OBI scores are only available for
2006, 2008, 2010, 2012. We follow a two-fold strategy to fill in the missings. First, we use the
score for the previous year to fill in scores for following years, second we follow a regression
approach to inpute missing values. The OBI score is highly correlated to other political measures
of governance and freedom of press, thus we use the transparency index provided by Alt and
Lassen (see below), the government effectiveness estimates from the World Bank governance
indicators, and the Freedom House freedom of press and political pressures and controls of media scores to generate a regression score to fill in missings of the OBI score. 16 Figure 1 shows
average OBI scores for the countries in our sample, ranked from the lowest to highest scores.
Figure 1: Average OBI score across countries
16
Other governance indicators such as corruption are possible candidates too, but we used the variables that gave
the best and most parsimonious fit.
13
Mexico
Indonesia
Turkey
Slovak Republic
Portugal
Japan
Italy
Russia
Czech Republic
Spain
Belgium
India
Ireland
Germany
Iceland
Switzerland
Canada
Korea
Finland
Denmark
Poland
Brazil
Netherlands
Australia
Austria
Slovenia
United States
Sweden
Norway
United Kingdom
France
New Zealand
40
50
60
70
80
fiscal transparency (open budget index)
90
The ranking seems to make intuitive sense and is very similar to the Alt and Lassen score
(correlation 0.58). Basically, we argue that in more fiscally transparent countries, incumbents
cannot easily use deficits to provide public goods before elections but have to reshuffle the
budget from long-term efficient to short-term beneficial spending. Thus, high transparency
should make strategic budgeteering more likely.
As a robustness check we use the fiscal transparency index provided by Alt and Lassen
(2005) which is based on a 1999 OECD questionnaire sent to all Budget Directors of OECD
member countries. The main problem of the variable consists in its time invariance.17
Figure 2: fiscal transparency score by Alt and Lassen
17
Since we use country dummies in all our models, we use interaction effects with long- and short-term items as
well as election periods to identify the effect of transparency.
14
Japan
Germany
Norway
Switzerland
Belgium
Denmark
Ireland
Italy
Austria
Canada
Finland
France
Iceland
Sweden
Netherlands
Australia
United Kingdom
United States
New Zealand
0
1
2
3
4
5
6
7
8
fiscal transparency (Alt/Lassen)
9
10
11
The second condition – closeness of elections – states that governments only resort to heavy
budgeteering if the elections are highly contested. To capture this factor we use a variable provided by the DPI (2004) which measures the margin of the majority by dividing the number of
seats won by the largest government party by the total number of seats (government plus opposition plus non-aligned). Thus the larger value of the variable the less contested the elections are.
Of course, this operationalization cannot capture situations of unexpected landslide victories but
comes very close to our theoretical definition.18
Clearly and as stated elsewhere (Franzese 2002), electoral budget cycles are expected to be
flatter when governments have some discretion over setting the actual election date. Much research has been done on so called endogenous election timing (e.g. Smith 1996, 2003, 2004),
that we do not need to re-iterate here. The main argument we take from the discussion in the literature is that if incumbent governments have discretion over setting the election date they can
chose a period when the economy is doing well and thus have less incentive to use fiscal
measures to manipulate the budget cycle. If this argument held, we should not find a significant
impact of our main interesting variables because the effect of endogenous election timing would
erase our hypothesized effect. However, we still are interested in testing this counterbalancing
effect directly and thus use data from the Institutions and Elections Project (Regan and Clark).
We combine the information on two variables, the election timing variable considers whether
there is a no formal schedule (1 – full discretion), exact intervals for elections (2 – no discretion),
the timing is determined by the political process within a fixed interval (3 – some discretion).
Most countries in our sample fall in category three. However, we only code the countries falling
in this category as “endogenous election timing” if the executive has the power to set the election
date within the specified interval.
18
It is always problematic to use ex post outcome data to model ex ante expectations. In future iterations of the paper we will use survey data on pre-electoral preferences to operationalize the incumbents ex ante expectation of
winning the election.
15
I addition, to these main interesting variables we use a number of political and economic
control variables in order to avoid omitted variable bias19.
We estimated fixed effects models20 (thus only use within country variation to identify effects), include a trend, and compute country-wise clustered standard errors.
Table 2 presents the regression results for the ratio of government consumption/investment.
Table 2: Ratio of government consumption/investment
DV: consumption/investment
∆ debt
Pre-election period
Fiscal Transparency
(OBI/Alt-Lassen)
Preelection*transparency
Preelection*debt
Debt*transparency
Peelection*debt*trans
trend
EMU membership
Preelection*EMU
Preelection*debt*EMU
Endogenous election timing
(1=yes)
Preelection*endog elect
Margin of Majority
Preelection*Margin of Majority
Does Party of Executive
Control All Houses?
Vote Share of Largest Government Party
Legislative Electoral Competitiveness
Government Fractionalization Index
Checks and Balances
Polarization
Baseline
OBI
0.006***
(0.002)
-0.116**
(0.052)
-0.004***
(0.001)
0.002**
(0.001)
0.002**
(0.001)
0.000***
(0.000)
-0.000**
(0.000)
0.006***
(0.002)
Baseline
Alt/Lassen
0.007***
(0.002)
-0.108***
(0.034)
Dropped due to
collinearity
0.034**
(0.012)
0.002***
(0.000)
0.001**
(0.000)
-0.001**
(0.000)
0.005***
(0.002)
extended
OBI
0.007***
(0.001)
-0.096
(0.072)
-0.003***
(0.000)
0.002**
(0.001)
0.001**
(0.001)
0.000***
(0.000)
-0.000**
(0.000)
0.007***
(0.001)
-0.032*
(0.018)
-0.017
(0.066)
0.000
(0.001)
0.030
(0.028)
-0.011
(0.022)
-0.099*
(0.056)
-0.001
(0.089)
Pol. controls
OBI
0.007***
(0.001)
-0.131
(0.081)
-0.004***
(0.000)
0.002**
(0.001)
0.001**
(0.001)
0.000***
(0.000)
-0.000**
(0.000)
0.007***
(0.001)
-0.026
(0.019)
-0.035
(0.069)
0.001
(0.001)
0.026
(0.029)
-0.001
(0.025)
-0.346***
(0.098)
0.035
(0.105)
-0.028
(0.019)
0.003**
(0.001)
0.027
(0.077)
0.196***
(0.060)
-0.001
(0.005)
-0.017*
full
OBI
0.002
(0.001)
-0.238*
(0.127)
-0.003***
(0.001)
0.003*
(0.002)
0.002
(0.002)
0.000***
(0.000)
-0.000
(0.000)
-0.002
(0.002)
-0.075***
(0.026)
-0.008
(0.063)
0.000
(0.001)
Dropped due to
collinearity
0.007
(0.028)
-0.047
(0.123)
0.139
(0.114)
-0.047*
(0.028)
0.001
(0.002)
-0.120*
(0.062)
0.125*
(0.075)
-0.023**
(0.009)
0.015
19
For parsimony, the names of the variables can be found in regression tables, the sources are the Worldbank WDI,
OECD and Database for Political Institutions (DPI, Worldbank, Keefer et al. 2004).
20
Random effects results are similar, but Hausman tests suggest the use of country fixed effects.
16
(0.009)
0.603
(0.545)
(0.014)
-0.005***
(0.001)
-0.005*
(0.003)
-0.000*
(0.000)
-0.026***
(0.002)
0.004***
(0.001)
-0.008***
(0.002)
0.018***
(0.007)
-0.006
(0.004)
2.056***
(0.472)
0.409
696
30
21.276
0.645
342
30
18.452
Foreign direct investment
(% of GDP)
GDP per capita growth
GDP per capita
Gross savings (% of GDP)
Trade (% of GDP) WDI
Inflation CPI
Age dependency ratio
Tax revenue (% of GDP)
Intercept
0.733***
(0.061)
0.589***
(0.069)
0.753***
(0.044)
R2 (within)
0.399
0.413
0.405
N
822
652
764
Number of countries
32
19
31
F statistic
10.343
23.921
32.616
Standard errors clustered on country level in parentheses, *** p≤0.01, ** p≤0.05, * p≤0.1
In order to be able to interpret the main effects directly we have graphed the interaction effects
from the baseline model.
100
.02
.004
.015
.01
0
.005
0
-.002
-.004
80
.005
40
60
fiscal transparency
0
20
Kernel Density Estimate of OBI
.015
.01
Kernel Density Estimate of OBI
marg effect of preelectoral deficit
0
.002
.2
.1
0
-.1
-.2
-.3
pre-election effect
Figure 1b: pre-electoral effect of deficit spending
on consumption/investment ratio
.02
Figure 1a: pre-election effect on consumption/investment ratio
0
20
40
60
fiscal transparency
80
100
Our theory predicts that when fiscal transparency is high, resources should be redirected from
investment to consumption in the pre-electoral period (hypothesis 4) and that is exactly what figure 1a shows: when transparency is high, the pre-electoral effect is positive implying that the ratio changes in favor of consumption. When fiscal transparency is low this is not the case and
governments do not shift resources from investment to consumption (hypothesis 2). Moreover,
figure 1b depicts the pre-electoral effect of deficit spending. When fiscal transparency is low,
higher deficit spending is used to increase government consumption (hypothesis 1). This effect
vanishes with above average fiscal transparency (hypothesis 3). Thus fiscal transparency pre17
vents governments from using the “easier” measure deficit spending to increase public good provision before elections and forces them the reshuffle resources within the budget.
With respect to compositional spending, hypothesis 5 is only partially confirmed; the actual
closeness of election has an effect throughout the electoral period but is not reinforced in the preelectoral period. As said this measure has to be taken with some caution since it is an ex-post
measure. Better survey data should enable us to get a more precise operationalization of this notion. The EMU membership (hypothesis 8) hypothesis cannot not be confirmed for compositional spending, EMU members generally are less likely to use fiscal measures to manipulate the
economy but we cannot detect a cycling effect since there is no additional effect in the preelectoral period. Finally, endogenous election timing does not seem to reduce incentives for governments to use compositional spending to manipulate the electoral cycle. Even though it seems
that governments should have less inclination to use fiscal measures before elections when they
can time the election date, they can only do so within a limited interval and prediction economic
conditions is not an easy thing to do. Moreover, these timing decisions are complex and do not
just depend on economic considerations but also political ones (e.g. domestic and international
conflicts).
The control variables generally influence the outcome in the expected way, better economic conditions reduce incentives to budgeteer, veto players reduce the ability to electioneer and fractionalization of the government increases budgeteering which confirms the coalition government and
size of government hypotheses (Persson/Tabellini).
To confirm this initial analysis we test the pre-electoral effects of transparency on the change in
governmental debt. The previous analysis suggested that indeed governments use deficits in the
pre-electoral period to finance higher consumption if fiscal transparency is low.
Table 3 presents similar results for the change in debt on the left hand side.
Table 3: Pre-electoral Effect on Change in Debt
DV: ∆ debt
Pre-election period
Fiscal Transparency
(OBI/Alt-Lassen)
Preelection*transparency
EMU membership
Preelection*EMU
Endogenous election
timing (1=yes)
Preelection*endog elect
Margin of Majority
Preelection*Margin of
Majority
GDP per capita growth
trend
Vote Share of Largest
Government Party
Government Fractionali-
OBI
0.818
(0.936)
-0.020*
(0.012)
-0.008
(0.016)
OBI
2.089
(2.647)
-0.014
(0.013)
-0.009
(0.015)
-2.252***
(0.825)
0.568
(1.105)
2.294*
(1.297)
1.078
(1.016)
-1.236
(4.831)
-5.164
(4.141)
-0.936***
(0.083)
-0.034
(0.030)
-0.096
(0.071)
-2.501
Alt/Lassen
2.691**
(1.065)
Dropped due to
collinearity
-0.431*
(0.230)
Alt/Lassen
2.233
(2.729)
Dropped due to
collinearity
-0.373*
(0.194)
-1.932**
(0.824)
-0.603
(1.124)
2.577**
(1.175)
-0.193
(1.090)
-6.280
(4.940)
-0.391
(4.256)
-1.252***
(0.094)
-0.074***
(0.027)
-0.132*
(0.071)
-2.012
18
zation Index
Checks and Balances
(3.108)
-0.301
(0.207)
Intercept
1.870***
6.896***
0.820***
(0.655)
(2.556)
(0.278)
R2 (within)
0.006
0.184
0.011
N
839
750
652
Number of countries
33
32
19
F statistic
1.642
10.585
3.433
Standard errors clustered on country level in parentheses, *** p≤0.01, ** p≤0.05, * p≤0.1
(3.096)
-0.495**
(0.211)
11.964***
(2.755)
0.290
585
18
16.098
Generally the effects for deficit spending are not as strong as the previous results. However, the
main interesting theoretical effect exists and is captured in figure 221:
-5
-3
-1
1
pre-election effect
3
5
Figure 2: pre-electoral effect on change in debt
0
2
4
6
fiscal transparency Alt/Lassen
8
10
The result in figure 2 represents the necessary condition for hypotheses 1 and 3 to hold: if fiscal
transparency is low governments increase deficit spending before elections, which in turn is used
to increase public consumption (figure 1b). If transparency is high, deficit spending is not significantly increased over non-election years. From table 3 we can also see that in general governments increase deficits in the pre-election period.
EMU membership decreases the use of deficit spending in general but there is no additional reducing effect in pre-electoral periods (hypothesis 7). Interestingly, endogenous election timing
increases deficit spending contrary to expectations that governments should have fewer incentives to engage in fiscal electioneering if they can set the election date. If we speculated to a certain extend as to the underlying mechanism, one could argue that since voters know that governments with discretion would prefer to schedule the election during a period of economic prosperity, they might use fiscal measure to appear more competent to the public who expects elections to take place during an economic high.
21
Note however that the effect is only significant for the less informative fiscal transparency measure provided by
Alt and Lassen.
19
To support the previous empirical findings we run the same models for consumption (as % of
GDP) separately. Empirical results are shown in table 4:
Table 4: Pre-electoral effects of changes in debt and transparency on consumption per GDP
DV:
∆ debt
Pre-election period
Fiscal Transparency
(OBI)
Preelection*transparency
Preelection*debt
Debt*transparency
Peelection*debt*trans
consumption
0.058***
(0.018)
-1.297
(0.853)
-0.023
(0.014)
0.020**
(0.008)
0.020**
(0.008)
0.001***
(0.000)
-0.000*
(0.000)
EMU membership
Preelection*EMU
Preelection*debt*EMU
Endogenous election timing (1=yes)
Preelection*endog elect
Margin of Majority
Preelection*Margin of Majority
GDP per capita growth (annual %) WDI
trend
Vote Share of Largest Government Party
Government Fractionalization Index
Checks and Balances
Intercept
R2 (within)
N
Number of countries
F statistic
18.979***
(0.445)
0.209
839
33
10.745
consumption
0.050***
(0.010)
-1.246
(0.848)
-0.025***
(0.005)
0.012
(0.009)
0.012*
(0.007)
0.000***
(0.000)
-0.000
(0.000)
0.245
(0.210)
-0.094
(0.764)
0.004
(0.010)
0.120
(0.329)
0.045
(0.261)
-4.009***
(1.001)
0.431
(1.051)
-0.138***
(0.023)
0.018**
(0.008)
0.047***
(0.014)
2.245***
(0.630)
-0.071
(0.052)
19.552***
(0.646)
0.274
750
32
13.889
The empirical findings in table 4 support our previous conclusions. Pre-electoral deficit spending
is used to increase consumption if transparency is low (figure 3a) but has not effect if transparency is high (hypotheses 1 and 3). Re-shuffling of the budget of course cannot be tested with the
set up in table 4.
Figure 3a: pre-electoral effect of deficit spending on consumption
20
.02
0
.005
.01
Kernel Density Estimate of OBI
.015
.04
.02
0
-.02
0
20
40
60
fiscal transparency
80
100
Competitiveness of elections impacts consumption as expected: the more competitive an election
is the higher turns out to be public consumption (hypothesis 5). EMU membership has no effect
(or only mildly negative effect) on short-term public good provision.
In these models above the pure pre-electoral effect on public consumption turns out to be insignificant. We also ran the same models for established democracies only because there the electoral incentives to provide wide spread public goods to the median voter before elections seems
to be more prevalent. In this case, as expected, the pre-electoral effect of government consumption turns out to be significantly positive but declining in fiscal transparency. This is consistent
with the argument that while governments always have an electoral incentive to increase public
good provision, fiscal transparency reduces all opportunities to electioneer but makes reshuffling
of the budget more likely. Figure 4 displays this result graphically:
Figure 4: Pre-electoral effect on government consumption depending on fiscal transparency
21
4
3
2
1
0
-1
0
20
40
fiscal transparency
60
80
This finding is also consistent with figures 5a-5c below which show effects for single budget
items.
Finally we test electoral cycles for monetary and fiscal alternative to strategic budgeteering (deficit and compositional spending). We analyze monetary measures, e.g. inflation and interest
rates, other fiscal measure, e.g. tax revenue and tax rates. Table 5 shows the main results22.
Table 5: Pre-electoral effects on alternative measures to deficit spending and budgeteering
DV:
inflation
Real interest rate
Tax revenue
Tax revenue
Pre-election period
-2.620
(21.866)
-0.074
(0.302)
-0.044
(0.371)
-1.968
(2.040)
-0.030
(0.029)
0.046
(0.034)
0.132
(0.470)
0.000
(0.004)
-0.003
(0.007)
32.179**
(15.928)
0.000
1235.000
7.283***
(1.587)
0.003
949.000
18.254***
(0.265)
0.001
547.000
-0.040
(0.482)
-0.003
(0.005)
-0.000
(0.007)
-0.781
(0.496)
0.014*
(0.008)
18.450***
(0.291)
0.008
546.000
Fiscal Transparency
(OBI)
Preelection*transparency
Post-election period
Postelection*transparency
Intercept
R2 (within)
N
Statutory
corporate
tax
0.007
(0.014)
-0.001***
(0.000)
-0.000
(0.000)
Effective
average corporate tax
0.001
(0.010)
-0.001***
(0.000)
0.000
(0.000)
Average
income tax
0.447***
(0.011)
0.072
468.000
0.323***
(0.007)
0.050
468.000
18.068***
(0.414)
0.008
821.000
0.156
(0.585)
-0.017**
(0.007)
-0.001
(0.010)
22
Results in table 5 are reduced to the main interesting variables, models including controls can be found in the appendix.
22
Number of countries
F statistic
37.000
0.111
35.000
0.806
36.000
0.182
36.000
0.840
18.000
11.583
18.000
7.763
29.000
2.185
Generally, we cannot find electoral cycles in any of the measures presented in table 5. As argued
above we didn’t expect to find electoral effects in these variables but we wanted to confirm our
suspicion empirically, otherwise findings for budgeteering could be blurred by the use of other
fiscal and monetary measures. Especially in our sample of countries, governments don’t have
monetary policy autonomy since they face independent central banks, are member of a currency
union or have fixed exchange rate regimes. Tax measures on the other hand are not flexible
enough to be exploited for electoral purposes, changing tax laws and regulations is a difficult
legislative undertaking and usually faces considerable veto-power. Interestingly though, fiscal
transparency seems to reduce taxes overall which seems to make sense given that observability
should increase public pressure on tax authorities.
Re-shuffling Resources across Single Budget Items
In a second step we attempt to tackle the issue of strategic budgeteering by looking at redistribution of resources between 23 single budget items. In order to test the hypothesis on strategic
reshuffling from long-term efficient to short-term beneficial public spending in pre-election periods, we have to decide which items rather constitute long-term efficient and which short-term
strategic public goods. The last column of table 1 shows our allocation which relies heavily on
the categorization provided by Drazen and Eslava (2005) and we also consider work on German
partisan spending preferences (Bawn 1999, Koenig / Troeger 2005). We only allocated the most
obvious items and exclude the residual category from the empirical analysis. Drazen and Eslava
(2005) mostly distinguish spending items with respect to targeted and non-targeted spending. We
argue that both targeted (e.g. sectoral, regional) spending and non-targeted spending, especially
social security transfers can be used in the short term to manipulate electoral behavior. We therefore find it more plausible to distinguish long-term efficient investment from short-term inefficient public good provision either to the median voter or targeted groups. Admittedly the assignment of items to a long-term and a short-term category is still pretty ad hoc and needs more
theoretical derivation. Yet, for a first empirical investigation this allocation seems to provide us
with relatively robust and consistent results23. The assignment of items to long-term efficient and
short-term beneficial public spending allows testing the first prediction on strategic budgeteering
that incumbents tend to reshuffle resources from efficient investment to inefficient public good
provision before elections in order to increase their chances staying in office.
Table 6: Composition of Government Spending
23
Also as a robustness check we randomly allocated items to long-term and short-term spending and re-run the
analysis 1000 times. None of these random allocation produced significant results. See appendix for the simulation
analysis.
23
Expenditure
Item
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Description
General Public Services
Defense
Public Order & Safety
Education
Health
Social Security & Welfare
Housing & Community Amenities
Recreational, Cultural & Religious Affairs
Fuel & Energy
Agriculture, Forestry, Fishing & Hunting
Mining, Manufacturing & Construction
Transportation & Communication
Other Economic Affairs & Services
Other Expenditures
of which: Interest Payments
Adjustment to Total Expenditure
Economic Affairs & Social Services (9-13) (only if 9-13 not
available)
Environmental Protection
Electricity, Steam, Water
Roads
Inland & Coastal Waterways
Other Transport & Communication
Other Economic Services
Longterm/ shortterm
Longterm
Longterm
Longterm
Longterm
Shortterm / non-targeted
Shortterm
Shortterm/ Targeted
Shortterm/ Targeted
Shortterm/ Targeted
Shortterm/ Targeted
Shortterm
Longterm
Shortterm
Shortterm
We conduct the second empirical analysis in two steps. First we run models for alternative
strategies to reshuffling the budget such as relaxing monetary policy, debt creation, and tax reduction. In a second step we estimate the extent of strategic budgeteering depending on the two
conditions (transparency (Alt-Lassen) and closeness of elections) as well as controlling for alternative strategies. We also test for partisan effects but do not exhaustively cover this question and
leave it for future research.
In a first stage we re-examine whether governments in OECD countries use strategic
measures before elections in order to increase their odds of being re-elected. We look at debt
creation (debt per GDP, OECD), monetary policy and monetary outcomes (interest rate and inflation – both WDI), and taxation (consumption tax revenue – OECD, average effective tax rates
on capital and labor – own calculations, see Troeger 2008). We regress these instruments on the
pre-election period, transparency and their interaction. We control for the unemployment rate
and GDP per capita (both WDI), left and right cabinet portfolios (both DPI), the electoral system
(1 – majoritarian, 0 – proportional, DPI) and Checks and Balances to the executive 24 (DPI).
Since the theoretical model predicts an impact of interest rates on the ability of policy makers to
create higher deficits we include the main lending rate (OECD) to the right-hand-side of the debt
ratio equation. We also add a dummy for EMU members that adopted the Maastricht criteria in
order to test for the effect of stricter constraints to deficit spending and debt creation for those
governments.
24
We use different measures for veto players/ points, and institutional constraints to the executive with essentially
the same statistical results.
24
We employ the fixed effects vector decomposition method suggested by Plümper and
Troeger (2007) since the transparency measure is time invariant and the electoral system is slow
moving at best, but certainly cross-sectionally dominant. This allows us to keep the beneficial
characteristics of the fixed effects estimator for all time varying right-hand-side variables but still
efficiently estimate coefficients for transparency and electoral system. We also employ a PraisWinston transformation to eliminate existing serial correlation.25 Table 6 depicts the estimation
results.
The most interesting (for our purpose) conclusion is that we cannot find any business cycle
or at least pre-electoral activity for taxation or monetary policy which corresponds to our theoretical discussion. Only for debt creation we can observe that governments on average increase the
debt prior to elections in order to provide public goods that serve their electoral chances. Yet, in
more fiscally transparent countries, incumbents are less likely to use debt creation for electoral
purposes – the electoral business cycle becomes flatter. This finding supports the first hypothesis
of our theoretical model: incumbents in fiscally transparent systems are less likely to use deficit
spending prior to elections in order to increase their probability of being re-elected. In addition
we find that higher interest rates indeed reduce deficit spending in general since debt services
become more expensive. Also, as expected EMU members have lower debt ratios on average.
Table 6: Electoral Business Cycles for Taxation, Monetary Policy and Debt Creation
25
We are painfully aware of all the endogeneity and simultaneity issues of this approach. It would certainly be better to estimate a simultaneous equation model. Yet, our most important results do not change.
25
Model 1a
Debt
Pre-election
Model 1b
Inflation
Model 1c
Interest
rate
-0.112
(0.470)
0.038
(0.121)
-0.206***
(0.073)
-0.211***
(0.066)
-0.000***
(0.000)
-0.006
(0.007)
-0.009
(0.007)
-0.089*
(0.059)
0.478**
(0.190)
Model 1d
Consumption
tax
28.059
(34.833)
-4.210
(6.872)
18.858***
(2.529)
0.204
(2.626)
0.026***
(0.002)
-0.511
(0.400)
-0.231
(0.488)
-11.256**
(5.612)
347.782***
(25.752)
Model 1e
Labor tax
4.901**
0.207
-0.578
(2.394)
(0.430)
(0.531)
Pre*Transp.
-0.695**
0.004
0.112
(0.363)
(0.108)
(0.103)
Transparency
-0.197*
-0.365***
0.546***
(0.121)
(0.090)
(0.032)
Unemployment
3.757***
-0.768***
0.647***
(0.192)
(0.061)
(0.052)
GDP per cap.
0.001***
-0.001***
0.000***
(0.000)
(0.000)
(0.000)
Left cabinet
0.038**
-0.012***
0.012*
(0.015)
(0.004)
(0.006)
Right cabinet
0.014
-0.012***
0.003
(0.018)
(0.004)
(0.007)
Checks & Bal.
-1.176***
-0.042
-0.036
(0.327)
(0.051)
(0.112)
Majoritarian
-19.502***
0.766***
-12.162***
(0.964)
(0.122)
(0.249)
Interest Rate
-0.302**
(0.128)
EMU
-2.207**
(0.945)
Intercept
12.810***
26.998*** 18.968*** -324.723*** 28.365***
(0.449)
(0.294)
(0.269)
(18.732)
(0.289)
Fevd
fevd
fevd
fevd
fevd
R² adj.
0.878
0.644
0.561
0.897
0.909
N
439
477
426
480
492
F
14595.300
707.237
42.018
184.135
1429.866
Prob. > F
0.000
0.000
0.000
0.000
0.000
Robust White Standard Errors in Parentheses; *** p<0.01, ** p<0.05, * p<0.1
Model 1f
Capital
tax
-0.419
(0.959)
0.104
(0.209)
0.108
(0.106)
-0.085
(0.115)
0.000
(0.000)
0.008
(0.014)
0.010
(0.010)
-0.126
(0.164)
8.913***
(0.409)
22.831***
(0.395)
fevd
0.786
485
395.294
0.000
If governments cannot create debt, do they strategically reshuffle the budget from long-term
efficient investment to short-term beneficial public goods in order to improve their electoral prospects? This section tries to empirically answer the question. Since our dependent variable is a
composite of different budget items, efficiency issues due to substitution effects arise. We have
yet to find a suitable way to address these issues in a satisfactory manner. However, this model
exactly estimates possible substitution effects between long-term efficient investments and shortterm public goods because it only includes these budget items into the model. We use dummy
variables for these two categories and interact these with the most important theoretical variables. We are aware that this approach does not account for any other substitution effects between
single spending items but we are confident that the empirical results give us some insights into
the mechanisms of strategic budgeteering. To further address substitution issues we use both relative expenditure per item as well as total expenditure per item (with overall spending on the
RHS of the model capturing the budget constraint) and compare the results.26 The data structure
26
We also use yearly changes in spending per item with mostly the same findings, yet significance levels drop.
26
is three-dimensional: spending item per country per year (15*17*35). Spending data are consistently available for about 13 spending categories which leaves us with about 7735 observations
from which we can use 5000-6000 observations for 16-17 countries due to missing data for some
of the institutional variables especially before 1980. We use other budgeteering measures (debt,
monetary policy and taxation) and the same political and institutional variables as in the first
stage models as control variables.
We include country dummies into all models so that the level effect of transparency is
soaked up by the country effects. We still can identify the pre-electoral effect of fiscal transparency which is sufficient for the test of our hypothesis. We again employ a Prais-Winston transformation to control for existent serial correlation. This specification is also useful since we are
interested in short-term adjustment effects in spending.
The results presented in Table 7 test the argument that governments reshuffle resources from
long-term investment to short-term public goods before elections and whether this is conditional
on fiscal transparency. The model fit is reasonably good for total expenditure but poor for relative expenditure which is to be expected due to the high share of pre-determined spending. Still
all models in table 3 support our prediction that governments indeed engage in strategic budgeteering before elections. Long-term efficient investment is reduced in the pre-electoral period
and more money is spent on short-term public goods, especially social security, manufacturing
and other sectoral subsidies. In addition, fiscal transparency dampens the electoral cycle and incumbents in more transparent countries are less likely to reshuffle the budget before elections but
invest in long-term efficient projects such as infrastructure and education. These findings are robust throughout all models in table 7. Yet, the effect is not as strong as for deficit spending suggesting that even though fiscal transparency constrains electioneering ceteris paribus it has a
smaller impact on strategic budgeteering. The more transparent a country the higher the probability that voters can see thru all kind of measures whose sole purpose is to manipulate the electoral
business cycle.
Table 7: Strategic Budgeteering and Transparency
27
Model
2a
Relative
exp. per
item
Total gov. exp.
Preelection longterm
spending
Pre*long*Transp.
Preelection shortterm
spending
Pre*short*Transp.
Model 2b
Total exp.
per item
-0.04***
0.088***
(0.006)
-3369.75***
(0.010)
0.004**
(0.002)
0.043**
(0.017)
-0.003
(0.003)
Model
2c
Relative
exp. per
item
Model 2d
Total exp.
per item
Model
2e
Relative
exp. per
item
Model 2f
Total exp.
per item
-0.02***
0.091***
(0.008)
-3808.30***
-0.022**
0.091***
(0.008)
-2394.334*
(933.888)
290.745**
(123.602)
5654.359**
(0.006)
0.002*
(0.001)
0.057***
(1061.337)
343.850**
(139.392)
6567.498**
(0.009)
0.002*
(0.001)
0.048*
(1263.005)
267.070**
(134.216)
5304.239*
(2454.492)
-608.965*
(321.497)
(0.016)
-0.005*
(0.003)
(2856.804)
-734.906**
(373.468)
(0.024)
-0.004*
(0.003)
-0.000
(0.000)
0.000
(0.000)
0.000
(0.000)
0.000
(0.000)
0.001
(0.000)
-0.000
(0.000)
0.000
(0.000)
0.061***
(0.016)
yes
0.009
5801
2.949
(3552.546)
-665.477*
(382.674)
-24.356
(18.933)
22.458
(50.112)
15.916
(22.294)
75.641**
(31.492)
17.573
(61.696)
74.523*
(44.688)
-0.071
(0.828)
-5429.18**
(2666.167)
yes
0.278
5801
57.754
Pre*long*debt
Pre*short*debt
Debt per GDP
0.000
14.940
(0.000)
(21.620)
Inflation
0.000
75.165**
(0.000)
(31.427)
Labor tax
0.001
18.970
(0.000)
(61.560)
Capital tax
-0.000
74.479*
(0.000)
(44.694)
Consumption tax rev.
0.000
-0.091
(0.000)
(0.826)
Intercept
0.072*** -1378.037* 0.061*** -5440.959**
(0.004)
(739.683)
(0.016)
(2666.214)
Country dummies
yes
yes
yes
yes
R² adj.
0.004
0.283
0.009
0.278
N
6268
6268
5801
5801
F
4.106
79.183
3.073
62.387
Robust White Standard Errors in Parentheses; *** p<0.01, ** p<0.05, * p<0.1
In addition we check whether incumbents who are unable to use deficit spending for electoral purposed compensate by reshuffling more money across budgetary items. Yet, the results
remain inconclusive since the estimates turn out statistically insignificant. If anything it seems
that governments who use deficit spending before elections also have a higher propensity to engage in strategic budgeteering which would support the results found with respect to transparency.
The main findings from tables 6 and 7 are graphically depicted in figure 5:
Firgure 5a: pre-election effect on
deficit
Figure 5b: pre-election effect on
long-term items
Figure 5c: pre-election effect on
short-term items
28
0
2
4
6
Fiscal Transparency
8
10
0
.08
0
.02
.04
.06
pre-election effect on relative shortterm spending
Mean of Transparency
-.02
-.06
-.04
-.02
0
pre-election effect on relative longterm spending
.02
10
8
6
4
2
0
pre-election effect on debt
-2
-4
-6
Mean of Transparency
2
4
6
Fiscal Transparency
8
10
Mean of Transparency
0
2
4
6
Fiscal Transparency
8
10
The most striking effect is that when fiscal transparency increases, pre-electoral deficit spending
turns from positive to insignificant, however there is still reshuffling going on from long-term
items (negative effect) to short-term items (positive effect).
Most of the control variables do not exert a significant effect on overall spending, just inflation and capital taxation lead to higher expenditure on all spending items.
The models in table 8 test for the second condition – the closeness of election as well as
some partisan budgeteering. We interact pre-electoral long- and short-term spending with the
margin of the majority and expect that the smaller the margin the higher the propensity for redirecting resources from long-term investment to short-term electoral gifts. And we find exactly
this: the more contested the elections the more money is reshuffled. The larger the margin the
smaller the need for the incumbent to engage in strategic budgeteering and thus the positive sign
for long-term spending and negative coefficient estimates for short-term expenditure. The signs
are consistent throughout all models but the coefficients only turn out to be significant for the
effect on long-term spending. The findings for pre-electoral reshuffling and transparency remain
stable throughout all models.
Finally, we test some partisan hypotheses. Based on predictions by Bawn (1999) and Koenig
and Troeger (2005) we chose the most obvious partisan items for right-wing and left-wing parties – military (defense) spending and social security & welfare spending. We also examine
whether governments increase social security spending before elections in case the economy
does badly – the unemployment rate is high. The results are shown in the bottom right corner of
table 4 (Model 3e and 3f, grey shaded area). The findings are strong and as expected. Left-wing
incumbents spend more on social welfare before elections and right-wing incumbents increase
expenditure on defense issues before elections in order to cater to their specific constituency. Independent of partisanship governments increase pre-electoral welfare spending in case the unemployment rate is high.
Table 8: Strategic Budgeteering and Closeness of Elections, Partisan Budgeteering
29
Model 3a
Relative
exp.
Model 3c
Relative
exp.
-0.031**
Model 3b
Total exp.
per item
0.090***
(0.007)
-6532.98**
Model 3e
Relative
exp.
-0.020*
Model 3d
Total exp.
per item
0.093***
(0.009)
-6033.321**
-0.019*
Model 3f
Total exp.
per item
0.102***
(0.010)
-6738.809**
(0.015)
0.003*
(0.002)
0.006
(2850.240)
280.040**
(125.950)
5810.986*
(0.013)
0.002*
(0.001)
0.010
(2952.013)
346.340**
(140.816)
3907.761*
(0.014)
0.001
(0.001)
0.015
(2674.757)
372.816***
(129.299)
4695.976
(0.023)
0.039*
(4086.931)
10098.556*
(0.023)
0.043
(2221.841)
11205.359*
(0.025)
0.027*
(3048.595)
3630.990*
(0.026)
-0.004*
(0.003)
-0.017
(0.061)
(7194.097)
-636.766*
(330.225)
-7487.568
(10161.018)
(0.036)
-0.005*
(0.003)
-0.028
(0.063)
0.000
(0.000)
0.001
(0.000)
-0.000
(0.000)
0.000
(0.000)
(7522.284)
-729.984*
(374.857)
-8321.073
(10429.683)
84.923**
(35.008)
8.118
(72.049)
85.560*
(52.075)
-0.388
(1.008)
(0.018)
-0.003*
(0.001)
-0.004
(0.031)
0.000
(0.001)
0.000
(0.001)
-0.000
(0.000)
-0.000
(0.000)
0.000
(0.001)
0.000
(0.000)
-0.002*
(0.001)
0.034
(0.031)
0.014
(0.017)
-0.000
(0.000)
0.000
(0.000)
0.037***
(0.002)
0.002***
(0.000)
0.000***
(0.000)
0.030
(0.040)
yes
0.172
4916
28.652
(2412.679)
-502.394*
(313.175)
-1.08e+04
(9827.533)
214.210*
(109.373)
-30.326
(102.137)
87.858*
(54.160)
2.523*
(1.785)
188.375
(187.133)
-0.302*
(0.182)
-83.586
(152.240)
23124.490**
(10130.891)
1383.456
(2413.600)
-3.783
(11.113)
-1.047
(9.808)
3554.101***
(516.566)
199.355**
(91.215)
12.898*
(9.051)
-2.60e+04**
(10842.177)
yes
0.325
4916
35.008
Total gov. exp.
Preelection
longterm spending
Pre*long*Transp.
Pre*long*Margin
maj.
Preelection
shortterm spending
Pre*short*Transp.
Pre*short*Margin
Inflation
Labor tax
Capital tax
Consump. tax rev.
Checks & Balances
GDP per capita
Unemployment
Majoritarian
Margin of Majority
Left cabinet portf.
Right cabinet portf.
Pre*unemp*SoSec
Pre*Left*SoSec
Pre*Right*Def.
Intercept
0.073*** -1623.061* 0.062*** -5915.837*
(0.005)
(861.843)
(0.016)
(3041.554)
Country dummies
yes
yes
yes
yes
R² adj.
0.004
0.277
0.010
0.273
N
5742
5742
5391
5391
F
4.003
68.301
3.008
55.134
Robust White Standard Errors in Parentheses; *** p<0.01, ** p<0.05, * p<0.1
30
CONCLUSION
The literature on political business cycles in fiscal policies has come a long way. Most importantly, scholars have highlighted different fiscal strategies which are used by incumbents to
increase the voters’ well-being before elections. Additionally, they have investigated into the
constraints governments face particularly when employing deficit spending in the pre-election
period. They thereby elucidate the conditions under which deficit spending as a strategy to win
re-election is effective.
Based on the insights of these models which analyze different strategies in isolation, this paper developed an integrated formal model of fiscal strategies in the pre-election period. Most importantly, we analyzed how opportunistic incumbents choose between different fiscal strategies,
such as deficit spending and strategic budgeteering, to increase their electoral prospect in the period before the election takes place. One of our main departures from the literature was that we
assumed that governments may either spend on long-term efficient investment or short-term efficient public good provision. They increase public good provision in the pre-election period by
either raising deficits or by redirecting resources from long-term efficient investments to shortterm efficient public goods.
We find that governments principally have an incentive to increase short-term public good
provisions if they fear fierce electoral competition and small chances of getting re-elected. To
finance these opportunistic policies, governments increase the deficit in the pre-election period.
Deficit spending becomes less attractive if fiscal transparency is high (and consequently, fiscally
conservative voters would be able to observe the distortive policies of the government) and interest rates rise. The incumbent then faces a trade off between short-term public good provision and
long-term efficient investment. Because they cannot use deficit spending under fiscal transparency, governments tend to change the composition of the budget if elections are close. Strategic
budgeteering goes at the expense of long-term efficient investments. In other words, long-term
efficient fiscal policies are less likely the higher fiscal transparency and the smaller the probability that the incumbent gets re-elected.
In a way of summarizing, the model presented here elucidates under which governments either rely on deficit spending or the redistribution of budget resources – what we call strategic
budgeteering – in order to increase the voters’ welfare in the period before elections. The model
offers a parsimonious account of opportunistic governmental strategies. It thereby provides the
basis for several important extensions. First of all, governments do not only face a trade off when
choosing alternative fiscal policies. Under certain conditions – e.g. if exchange rates are flexible
or central bank independence low – monetary policy instruments become effective leaving fiscal
strategies ineffective. An important extension of our model would thus include the possibility to
use monetary policies to generate political business cycles.
Additionally, we have neither considered partisan preferences nor different forms of strategic
budgeteering in our baseline model. However, we expect that different political parties would
serve different voters, and thus increase spending on different budget items or have different
preferences of raising the deficit. This is directly linked to different forms of re-distributing the
budget. While our baseline model simply assumes that governments choose between long-term
efficient investment and short-term efficient public goods, redistribution could take several
forms. In an extension to the baseline model, it would be thus important to distinguish between,
for example, (a) functional, (b) sectoral, and (c) regional budgeteering.
31
The empirical analysis supports most of our predictions, governments engage in strategic
budgeteering and they do so the more contested the elections are. However, fiscal transparency
seems to reduce all activities that aim at manipulating the electotral business cycle. Moreover,
we found some support for partisan budgeteering.
The empirical models however need to better take substitution effects as well as simultaneity
and endogeneity issues into account.
Include: shoud be generalizable. As developing countries become more democratic and
open, there should be more evidence of compositional spending over deficit spending. Using
OECD countries is a hard test for the argument.
(INCOMPLETE) BIBLIOGRAPHY
Adolph, C, 2001: Parties, Unions, and Central Banks: an Interactive Model of Unemployment in
OECD Countries. Unpublished Working Paper.
Akhmedov, Akhmed and Ekaterina Zhuravskaya, 2004: Opportunistic Political Cycles: Test in a
Young Democracy Setting, in: Quarterly Journal of Economics 119(4): 1301-1338.
Alesina, Alberto (1987): Macroeconomic Policy in a Two-Party System as a Repeated Game, in:
Quarterly Journal of Economics 102:651-678.
Alesina, Alberto (1988): Macroeconomics and Politics, in: O. Blanchard/S. Fischer (Hrsg.):
NBER Macroeconomics and Annual. Cambridge. MA: MIT Press.
Alesina, Alberto, 1989: Politics and Business Cycles in Industrial Economies, in: Economic Policy 5, 55-98.
Alesina, Alberto, Gerald D. Cohen and Nouriel Roubini, 1992: Macroeconomic Policy and Elections in OECD Democracies, in: Economics and Politics 4: 1-30.
Alesina, Alaberto, Nouriel Roubini, and Gerald D. Cohen, 1997: Political Cycles and the Macroeconomy. Cambridge: MIT Press.
Alesina, Alberto and Nouriel Roubini, 1992: Political Cycles in OECD Economies, in: Review
of Economic Studies 59: 663-88.
Alesina, Alberto, Roberto Perotti and José Tavares, 1998: The Political Economy of Fiscal Adjustments. Brookings Papers on Economic Activity 1998(1): 197-266.
Alt, James, 1985: Political Parties, World Demand, and Unemployment: Domestic and International Sources of Economic Activity, in: American Political Science Review 79(4): 10161040.
Alt, James E. and David Dreyer Lassen, 2006a: Transparency, Political Polarization, and Political Budget Cycles in OECD Countries, in: American Journal of Political Science 50(3):
530-550.
Alt, James E. and David Dreyer Lassen, 2006b: Fiscal Transparency, Political Parties, and Debt
in OECD Countries, in: European Economic Review 50: 1403-1439.
Andrikopoulos, Andreas, Ioannis Loizides, and Kyprianos Prodromidis, 2004: Fiscal Policy and
Political Business Cycles in the EU, in: European Journal of Political Economy 20: 125152.
Bawn, Kathleen, 1999: Money and Majorities in the Federal Republic of Germany: Evidence for
a Veto Players Model of Government Spending, in: American Journal of Political Science, 707-736.
Bernard, William and David Leblang, 1999: Democratic Institutions and Exchange-Rate Commitments, in: International Organization 53: 71-97.
32
Bernard, William and David Leblang, 2002: Democratic Processes and Political Risk: Evidence
from Foreign Exchange Markets, in: American Journal of Political Science 46(2): 316333.
Bernhard, William T., Lawrence J. Broz and William R. Clark, 2002: The Political Economy of
Monetary Institutions. Cambridge: MIT Press.
Boix, Charles, 2000: Partisan Governments, the International Economy, and Macroeconomic
Policies in Advanced Nations, 1960-93, in: World Politics 53: 38-73.
Bräuninger, Thomas, 2005: Do Preferences Make a Difference? Parties and the Composition of
Budgets in Nineteen OECD Countries, in: Public Choice 125: 409-429.
Brender, Adi, 2003: The Effect of Fiscal Performance on Local Government Election Results in
Israel: 1989-1998, in: Journal of Public Economics 87: 2187-2205.
Brender, Adi and Allan Drazen, 2005: Political Budget Cycles in New versus Established Democracies, in: Journal of Monetary Economics 52: 1271-1295.
Buti, Mario and Paul van der Noord, 2003: Discretionary Fiscal Policy and Elections: The Experience of the Early Years of EMU. OECD Economics Department Working Paper No.
351.
Clark, William R., 2002a: Capitalism Not Globalism: Capital Mobility, and Political Control of
the Economy. Ann Arbor: University of Michigan Press.
Clark, William R., 2002b: Partisan and Electoral Motivations and the Choice of Monetary Institutions under Fully Mobile Capital, in: International Organization 56: 725-749.
Clark, William R., Usha N. Reichert, Sandra L. Lomas and Kevin L. Parker, 1998: International
and Domestic Constraints on Political Business Cycles in OECD Economies, in: International Organization 52(1): 87-120.
Clark, William R. and Mark Hallerberg, 2000: Mobile Capital, Domestic Institutions, and Electorally-Induced Monetary and Fiscal Policy, in: American Political Science Review 94:
323-346.
Cusack, Thomas R., 1997: Partisan Politics and Public Finance: Changes in Public Spending in
Industrialized Democracies, 1955-1989, in: Public Choice 91: 375-395.
De Haan, Jacob and Jan-Egbert Sturm, 1994: Political and Institutional Determinants of Fiscal
Policy in the European Community, in: Public Choice 80: 157-172.
Drazen, Allan, 2000a: Political Economy in Macroeconomics. New Jersey: Princeton University
Press.
Drazen, Allan, 2000b: The Political Business Cycle after 25 Years, in: NBER Macroeconomics
Annual 15: 75-117.
Drazen, Allan and Marcela Eslava, 2005: Electoral Manipulation via Expenditure Composition:
Theory and Evidence, in: NBER Working Paper No. W11085.
Drazen, Allan and Marcela Eslava, 2005: Pork Barrel Cycles, in: NBER Working Paper No.
W12190.
Easterly, W, 1999: When is Fiscal Adjustment an Illusion?, in: Economic Policy 14: 57-76.
Ferejohn, John, 1999: Acountability and Authority: Towards a Model of Political Accountability,
in A. Przeworski, B. Manin, and SC Stokes (eds.): Democracy, Accountability, and Representation. New York: Cambridge University Press.
Franzese, Robert J., 1999: Partially Independent Central Banks, Political Responsive Governments, and Inflation, in: American Journal of Political Science 43: 681-706.
Franzese, Robert J., 2002a: Electoral and Partisan Cycles in Economic Policies and Outcomes,
in: Annual Review of Political Science 5: 369-421.
33
Franzese, Robert J., 2002b: Macroeconomic Policies of Developed Countries. Cambridge:
Cambridge University Press.
Franzese, Robert J., 2002c: The Positive Political Economy of Public Debt: An Empirical Examination of the OECD Postwar Experience. Social Science Research Network Working
Paper Series.
Franzese, Robert J., 2003: Multiple Hands on the Wheel: Empirically Modeling Partial Delegation and Shared Policy control in the Open and Institutionalized Economy, in: Political
Analysis 11(4): 445-474.
Frey, BS and F Schneider, 1978a: An Empirical Study of Politico-economic Interaction in the
United States, in: Review of Economic Statistics 60: 174-183.
Frey, BS and F Schneider, 1978b: A Politico-Economic Model of the United Kingdom, in: Economic Journal 88: 243-253.
Golden, DG and JM Poterba, 1980: The Price of Popularity: the Political Business Cycle Reexamined, in: American Journal of Political Science 24 (4): 696-714.
Goodhart, LM, 2000: Political Institutions, Elections, and Policy Choices. PhD thesis, Harvard
University, Cambridge, MA:
Hallerberg, Mark, Lucio Vinhas de Souza and William Clark, 2002: Political Business Cycles in
EU Accession Countries, in: European Union Politics 3(2): 231-250.
Hibbs, Douglas A., 1977. Political Parties and Macroeconomic Policy, in: American Political
Science Review 23: 1467-1488.
Hibbs, Douglas A., 1987. The American Political Economy: Macroeconomics and Electoral
Politics in the United States. Cambridge: Harvard University Press.
International Monetary Fund, 2003: Germany. Report on Observance of Standards and Codes—
Fiscal Transparency. IMF Country Report No. 03/286.
Jochimsen, Beate and Robert Nuscheler, 2005: The Political Economy of German Länder Deficits. Unpublished working paper.
Kalecki M, 1943: Political Aspects of Full Employment, in: Political Quarterly 7: 322-331.
Koening, Thomas and Vera E. Troeger 2005:Budgetary Politics and Veto Players, in Swiss Political Science Review 11(4): 47-75.
Leblang, David and William Bernard, 2000a: Political Parties and Monetary Commitments, in
William Bernard, J. Lawrence Broz and William R. Clark (eds.): The Political Economy
of Monetary Institutions. Cambridge: MIT Press.
Leblang, David and William Bernard, 2000b: The Politics of Speculative Attacks in Industrial
Democracies, in: International Organization 54: 291-324.
Lohmann, Susanne, 1998: Rationalizing the Political Business Cycle: a Workhorse Model. Economic Politics 10: 1-17.
Midtbo, Tor, 1998: The Open Politiconomy: A Dynamic Analysis of Social Democratic Popularity and Economic Policies in Scandinavia, in: British Journal of Political Science 28(1):
93-112.
Milesi-Feretti, GM, 2004: Good, Bad or Ugly? On the Effects of Fiscal Rules with Creative Accounting, in: Journal of Public Economics 88: 377-394.
Mink, Mark and Jakob de Haan, 2005: Has the Stability and Growth Pact Impeded Political
Budget Cycles in the European Union? CESIFO Working Paper No. 1532.
Nordhaus, William D., 1975: The Political Business Cycle, in: Review of Economic Studies 42:
169-190.
34
Oatley, Thomas H., 1999: Monetary Politics. Exchange-Rate Cooperation in the European Union. Ann Arbor: University of Michigan Press.
Peltzman, Sam, 1992: Voters as Fiscal Conservatives, in: Quarterly Journal of Economics 107:
327-261.
Persson, Torsten and Guido Tabellini, 1990: Macroeconomic Policy, Credibility and Politics.
London: Harwood.
Persson, Torsten and Guido Tabellini, 2002: Political Economics. Explaining Economic Policy.
Cambridge: MIT Press.
Persson, Torsten and Guido Tabellini, 2003a: The Economic Effects of Constitutions. Cambridge, MA: MIT Press.
Persson, Torsten and Guido Tabellini, 2003b: Do Electoral Cycles Differ Across Political Systems? IGIER Working Paper No. 232.
Phillips, Alban W. 1958. The relation between unemployment and the rate of change of money:
wage rates in the United Kingdom, 1861-1957. Economica 25 (100):283-299.
Price, Simon, 1998: Comment on “The Politics of the Political Business Cycle.”, in: British
Journal of Political Science 28: 201-210.
Rogoff, Kenneth, 1990: Equilibrium Political Budget Cycles, in: American Economic Review
80: 21-36.
Rogoff, Kenneth and Anne Sibert, 1988: Election and Macroeconomic Cycles, in: Review of
Economic Studies 55: 1-16.
Romer, David. 2001. Advanced Macroeconomics (2nd ed.). New York: McGraw-Hill.
Rose, Shanna, 2006: Do Fiscal Rules Dampen the Political Business Cycle?, in: Public Choice
128: 407–431.
Schneider, Christina J., 2007: Politischer Opportunismus und Haushaltsdefizite in den westdeutschen Bundesländern, in: Politische Vierteljahresschrift 48(2): 221-242.
Schneider, Christina, J., 2008: Fighting with One Hand Tied Behind the Back, or: Can Incumbents Use Fiscal Policies for Electoral Gain? Manuscript. Department of Politics and International Relations, University of Oxford.
Schultz, Kenneth A., 1995: The Politics of the Political Business Cycle, in: British Journal of
Political Science 25(1): 79-99.
Schumpeter, J, 1939: Business Cycles: A Theoretical, Historical, and Statistical Analysis of the
Capitalist Process. New York: McGraw-Hill.
Shi, Min and Jakob Svensson, 2000: Political Business Cycles in Developed and Developing
Countries. World Bank Working Paper.
Shi, Min and Jakob Svensson, 2002: Conditional Political Budget Cycles. Manuscript, IIES,
Stockholm.
Shi, Min and Jakob Svensson, 2006: Political Business Cycles: Do they Differ Across Countries
and Why?, in: Journal of Public Economics 90(8-9): 1367-1389.
Tufte, Edward R., 1978: Political Control of the Economy. Princeton: Princeton University
Press.
Von Hagen, Jürgen, 2003: Fiscal Discipline and Growth in Euroland. Experiences with the Stability and Growth Pact. ZEI Working Paper No. B062003.
Wright, Gavin, 1974: The Political Economy of New Deal Spending: an Econometric Analysis,
in: Review of Economic Statistics 56: 30-38.
35
APPENDIX
Table A1: Observations for Baseline model:
country
No. of years
Australia
22
Austria
38
Belgium
38
Brazil
5
Canada
38
Czech Republic
17
Denmark
31
Finland
38
France
38
Germany
38
Iceland
31
India
15
Indonesia
17
Ireland
37
Italy
38
Japan
37
Korea
7
Mexico
10
Netherlands
38
New Zealand
18
Norway
38
Poland
1
Portugal
38
Russia
7
Slovak Republic
8
Slovenia
5
Spain
35
Sweden
38
Switzerland
20
Turkey
5
United Kingdom
38
United States
38
36