Does Politics Matter in the Conduct of Fiscal Pol-icy ? Political Determinants of the Fiscal Sustain-ability : Evidence from Seven Individual Central and Eastern European Countries ( CEEC ) ∗

This paper aims at assessing the fiscal sustainability and its political determinants in seven Central and Eastern European Countries (CEEC), namely Estonia, Latvia, Lithuania, Poland, Slovenia, Slovakia and the Czech Republic. First, using the recent sustainability approach of Bohn (1998) based on fiscal reaction function, econometric findings using Ordinary Least Squares (OLS) reveal a positive response of the primary surplus to changes in debt in several countries. In other words, fiscal policy is sustainable in Baltic countries, Slovenia and Slovakia, but not in Poland and in the Czech Republic. Second, by introducing political dummy variables, we test the electoral budget cycle and the partisan cycle theories. We find the presence of electoral and partisan cycle in Poland but not in the rest of our countries.


Introduction
Since May 1, 2004, the EU New Member States (NMS) 1 are classified as "member states with derogation" which means that those countries are expected to comply with the Stability and Growth Pact (SGP) limits for government budget deficit of 3% of Gross Domestic Product (GDP) and government debt of 60% of GDP.Moreover, the fiscal situation has to be judged as sustainable in the me-dium term.This is a new situation for the NMS given that prior to accession their fiscal situation was not the subject of regular and binding surveillance.
Indeed, empirical literature on public debt sustainability in transition countries gained in importance (see Llorca and Redzepagic, 2007 2 for a survey) due notably to the EU enlargement process (negotiations for EU membership in 1999, then EU enlargement in May 2004) and to the European Monetary Union (EMU) perspective (Buiter, 2004).Moreover, EU process induces a slow but real improvement of the quality and the accessibility of the data.
However, as Balassone and Franco (2000) rightly put it, there are serious difficulties in both the analytical and operational definition of sustainability.There is no consensus in the economic theory regarding the conditions for sustainability.We must distinguish between three existing empirical approaches in the literature (Llorca, 2007) to assess the sustainability of fiscal policy: (i) the present-value method which examines the sustainability issue during broader periods, studying the integration and co-integration properties of the relevant budget variables and departing directly from the transversality condition of the present value budget constraint; (ii) the accounting method which investigates whether public finances, in a year-by-year basis, are sustainable; and (iii) the normative approach, developed by Blanchard et al. (1990), aiming at evaluating how far has fiscal policy departed from sustainability in a year-by-year basis through sustainability indicators Thus, the most common empirical approach in the literature to test fiscal policy sustainability is the present-value method with influential papers such as Hamilton and Flavin (1986), Trehan andWalsh (1988, 1991) or Quintos (1995).However, this part of the literature ignore that the discount factor may be correlated with the primary surplus and debt.That is why, we adopt the methodology developed by Bohn (1995Bohn ( , 1998Bohn ( and 2004) ) who proposes a convenient framework to assess fiscal sustainability.He focuses on the response of the primary surplus to accumulated public debt that would guarantee fiscal sustainability as a test for sustainability.
Most of the empirical studies estimating the reaction of the primary surplus to debt levels focused on the American case and other industrial countries 3 .However, very few papers (Neck and Getzner, [2001]; Haber and Neck [2006] for the case of Austria; De Haan, Sturm and de Groot [2004] for the case of the Netherlands; Berenger and Llorca [2007] for the case of six developed countries 4 ) have enriching the Bohn's fiscal sustainability model with political vari-ables, whereas it is of particular interest to explore the politico-economic determinants for the growth of the central government's debt.Indeed, theoretical arguments taken from "political business cycle" theories (Nordhaus, 1975), such as the electoral and partisan cycle can explain the development of public debt.Moreover, such political issues are important for new EU countries facing fiscal challenges to join the EMU in the future.
As a result, the main purpose of this article is to assess first the fiscal sustainability in seven individual CEEC (Estonia, Latvia, Lithuania, Poland, Slovakia, Slovenia and the Czech Republic) in line with the recent literature on fiscal reaction functions developed by Bohn.Second, we combine it with the investigation of the political potential determinants of fiscal sustainability in each country by introducing political variables such as the ideologies of political parties and political electoral cycles, taking the primary surplus as the variable to be explained.
The paper is organised as follows: section 2 examines the theoretical framework, namely the Bohn's fiscal sustainability model and the introduction of political variables.Section 3 describes the sample and data, and reports the econometric findings.Section 4 concludes the study.Bohn (1995Bohn ( , 1998Bohn ( and 2004) ) emits two important critics against the presentvalue approach 5 :

Theoretical Framework
Firstly, this approach has imposed some simplifying and arbitrary conditions by assuming the economic environment as exogenous and constant which can be appropriate only under determinist circumstances and are hardly held in a stochastic environment.Indeed, tests of the transversality condition have commonly been based on the choice of the real return of debt as discount factor.Thus, the transversality condition would require an economic environment of dynamic efficiency.
Secondly, Bohn questioned the validity of cointegration tests as they require restrictive assumptions about the stochastic process driving fiscal policy and ignore that the discount factor may be correlated with the debt and primary budget surplus.Therefore, econometric tests for stationary are not appropriate to provide convincing evidence for sustainability.
As a result, in order to circumvent these drawbacks, Bohn (1998) has proposed a more flexible test to assess sustainability than does the present-value approach as it does not demand any assumptions on the real rates of interest, growth rates and uncertainty.
5See Ehrhart and Llorca (2007) for a detailed description of the present-value approach.
The strategy consists in investigating whether a government acts to increase surpluses in response to increases in government debt in order to ensure long term government solvency.Bohn (2004) proved that such a corrective action would satisfy the transversality condition in a stochastic setting.
The test 6 suggests running the following regression: where s t denotes the primary surplus to GDP and d t is the debt to GDP ratio respectively.
Z t is a vector which consists of variables which are assumed to affect the primary surplus, namely the level of temporary government spending and a business cycle indicator, GVAR and YVAR, respectively.Thus, the possible counter cyclical effect of the budget surplus is captured by the business cycle variable YVAR.The coefficient of YVAR is expected to be negative, as an increase in the output gap is accompanied by an increase in the budget deficit.For the same reason, the coefficient on temporary government spending GVAR is also expected to be negative.
Thus, fiscal policy can be shown to be sustainable if in equation ( 1), ρ is positive and sufficiently large, meaning that fiscal policy makers react to a high stock of debt at the beginning of period t by increasing the primary surplus (or reducing the primary deficit) in period t.This means that government acts to reduce the exponential growth of its debt by a factor r which is sufficient to fulfil the intertemporal budget constraint.
We propose to extend the basic Bohn model for the determination of government fiscal policy to include political determinants of fiscal policy.Such political factors are related to the "political budget cycle" theories.Among this large literature 7 , we focus on the electoral and partisan cycles on budget deficits.
Thus, the theory of budget electoral cycle, developed first by Nordhaus (1975) and Tufte (1978) predicts that politicians, motivated by their re-election, take advantage of voters' myopia (i.e fiscal illusion), so that deficits are higher immediately before elections than at other moments of the electoral cycle.
Moreover, according to the theory of the partisan cycle (exposed by Hibbs, 1977), politicians want to be elected in order to implement their preferred economic policy.As a result, fiscal policies are more expansionist when the government is dominated by the left.Left-wing parties favour government inter-6The background of this specification is the neoclassical theory of tax-smoothing (Barro, 1979) 7See Corsetti and Roubini (1992), Alesina and Perotti (1994), and Imbeau and Chenard (2002) for a theoretical literature on the political economy of the budget deficit.vention (i.e.public expenditure) and deficits; whereas right-wing (or conservative) parties favour less expenditure and a lower deficit Overall, the literature shows that budgetary performances are affected by electoral cycles but not by partisan cycles (see Berenger and Llorca, 2007 for a survey).
Finally, among the empirical literature about political factors in CEEC, we can refer to the following studies: Csontos et al. (1998) has shown, the fiscal illusion where voters are not aware of the costs of public services and the taxes they pay, is very persistent in Hungary.
Hallerberg and Vinhas de Souza ( 2000) look for evidence of opportunistic political cycles in the CEECs.They find that governments operating under fixed exchange rate regime pursue fiscal expansions during election years.Thus, they confirm that the NEW-8's governments acted very much like their OECD counterparts, and, where possible, manipulated the economy before elections.Berger, Kopits, and Szekely ( 2004) look for any deviation in the CEE fiscal policies from what could be expected from the standard empirical factors influencing deficits and debt.They find that a political-economic bargaining position prior to EMU entry seems to tilt the larger CEE countries more to profligate policies, whereas the smaller countries are more conservative since they lack a bargaining position.
Finally, the study of Schneider and Zapal (2005) do not reveal the existence of statistically significant electoral cycles for 2001-2003.

Sample and Data
Appendix A describes the sources, the definitions of the data.The political determinants of the fiscal sustainability are assessed for seven CEEC 8 , namely Estonia, Latvia, Lithuania, Poland, Slovakia, Slovenia, and the Czech Republic.We use quarterly data collected from Eurostat.The sample covers the period 1999:1-2006:1 for all countries.We use following economic variables: the primary budget balance, the gross public debt and the temporary government spending 9 .All the previous variables are measured in terms of their ratio to nominal GDP.Moreover, they are quarterly seasonally-adjusted by using the Tramo/Seats method10 ( Gómez and Maravall, 1996).Finally, we integrate political variables such as the election year and the ideology of political parties to respectively test for political business cycle and partisan effects.Appendix B describes the political evolution in each country; the political variables are extracted from Europe-Politique11 and Perspective-Université de Sherbrooke12 .

Empirical Results
All the regressions use Ordinary Least Squares (OLS) with heteroskedasticconsistent standard errors and we include an autoregressive term for the case of Lithuania, Slovakia and the Czech Republic in order to remove serial correlation of the residuals (Cochrane-Orcutt procedure).
Besides, the regression includes the lagged primary budgetary surplus S prim (-1) in order to take into account the inertial process typically associated with fiscal policy.
Finally, we test the eventual influence of an electoral cycle by using the dummy variable ELEC (=1 for the election year).Then, we examine the effects of political ideology by introducing the dummy variable PART (=1 when leftwing parties governs) in country where a change of government political orientation occurred during the last decade (this was the case in Lithuania and in Poland, see Appendix B).
The findings are summed-up in the table 1, 2 and 3.As indicated in table 1, the coefficient on the lagged debt-income ratio, for the Baltic States, is positive (he has the expected sign) and statistically significant indicating the absence of a mean-reverting process in the series.As a result, the positive response of the primary surplus to changes in debt shows that fiscal policy in the three Baltic States is satisfying the intertemporal budget constraint.However, estimations show no significant influence of the political variables tested in Baltic States.According to the sign of the coefficient of Debt(-1), it appears that public debt is sustainable in Slovenia but not in Poland.Moreover, estimations in table 4 display partisan and electoral effects in Poland, so that such political factors are one of the determinants of the fiscal unsustainability in this country.Finally, we find that public debt is sustainable in Slovakia but not in the Czech Republic; whereas the political variable (ELEC) is not significant.

Conclusion
If we comment our different results for each country, it appears first that by using fiscal reaction function, fiscal policy is sustainable in the Baltic States, Slovakia and Slovenia, but not for Poland and the Czech Republic.It can be interested to compare such results obtained via the Bohn's fiscal reaction function with the findings that we obtained by using the present-value approach in a panel perspective (Llorca and Redzepagic, 2007).The results in these two approaches seems coherent with on the one hand the existence of fiscal sustainability in a panel perspective in the present-value framework and on the other hand a fiscal policy sustainable in Baltic states, Slovenia and Slovakia.
Second, we find budget electoral cycle and partisan effects in Poland but not in the rest of our countries.We can notice that such political influence of fiscal unsustainability appears in a big country such as Poland and not in small countries such as Baltic states or Slovenia.Moreover, the political influence in Poland seems to follow some European cases due to political structure in this country with a clear opposition between left and right wing parties since 1997 (see Appendix B).

Table 1 .
Fiscal sustainability and influence of political variables in Estonia, Latvia and Lithuania

Table 2 .
Fiscal sustainability and influence of political variables in Poland and Slovenia

Table 3 .
Fiscal sustainability and Influence of political variables in the Czech Republic and Slovakia