Searching for Political Fiscal Cycles in Hungary

Hungary has had a remarkably high public debt throughout the transition, and it has continued to increase during recent years, exceeding 80% of the GDP. Its debt and fiscal deficit were the highest among the Visegrád countries during the transition. One factor triggering the debt increase may be elections-related fiscal policies. By analyzing quarterly data for Hungary, we found clear empirical evidence of fiscal expansion before elections and contractions afterwards. These events are widely known as political fiscal cycles. We observed statistically significant incremental increases in fiscal deficits as elections approach, both in nominal and in GDP ratios, followed by contractions after elections. Thus, it can be concluded that incumbents in Hungary are engaged in opportunistic political fiscal cycles by embracing expansionary fiscal policy before parliamentary elections. Our findings also suggest that political fiscal cycles in Hungary may be an underlying factor contributing to the accumulation of public debt.


Introduction
According to the opportunistic political business cycles (PBC) theory that was originally developed by Nordhaus (1975), the incumbent engages in expansionary economic policies and behaves in an opportunistic manner before an election to increase the likelihood of winning. The Nordhaus political business cycle theory asserts that governments stimulate economic growth before elections, thus benefiting from the short-run Philips curve and price rigidities in the short term. Inflation increases after an election because of the preelection economic expansion. After elections, incum-bents revert to tight economic policies to stabilize or reduce inflation.
An alternative view is the partisan PBC theory, which substantially differs from the opportunistic PBC theory because the former is based upon an ideological approach rather than on an opportunistic approach focused only on the incumbent's re-election. Hibbs (1977)  The framework of an opportunistic PBC is considered relevant not only for developing/transition countries but also for developed countries -the opportunistic cycle theory regarding government spending inclinations has been observed in Germany (Galli & Rossi, 2002), and with respect to lower fiscal balance, it has been observed in several OECD developed countries with weaker fiscal transparency (Alt &Lassen, 2006).
Hungary has had remarkably high debt throughout the transition, and it had the largest fiscal deficit during the transition of all Visegrád group countries.
The Hungarian debt has continued to increase during the last several years, exceeding 80% of the GDP and prompting the European Union to suspend funds of 495 m euros ($655 m; £417 m) due to the country's budget deficit. This is the first case of the EU taking action over the budget deficit of any of its members (BBC, 2012). Despite the recent attempt by the government to curb public spending, debt is still high and remains a major concern for the Hungarian economy.
Furthermore, there are concerns that the government may follow expansionary economic policies prior to the 2014 elections (Than & Szakacs, 2012), which may trigger increasing of debt. The Hungarian political system provides a basis for a strong government that can more easily persuade fiscal expansionary policies. The Hungarian parliament is unicameral, meaning it does not have an upper chamber. Furthermore, the President does not have veto power over legislative proposals. Therefore, as long as the government holds the majority in the parliament, it can easily pursue its fiscal policies (Haggard, Kaufman, & Shugart, 2001). In the last two elections, the winning party received the necessary majority in the parliament to establish a government with its own votes; thus, there was no need for a coalition, a situation that may be favorable political ground for conducting PFC (see Streb, Lema, & Torrens, 2009).
The lack of strict and practically enforced fiscal rules in Hungary may have also been a supporting factor for PFC, as was the case for a panel of American states according to Rose (2005).
The objective of our work is to identify the possible existence of political fiscal (or budget) cycles (PFC) in Hungary based on an opportunistic PBC framework.
Given Hungary's political and economic history, its institutional organization, its macroeconomic developments during the last two decades of democracy and its Searching for Political Fiscal Cycles in Hungary market economy, we hypothesize that and empirically test whether incumbents' behavior in Hungary toward fiscal policy is considerably similar to the implications of an opportunistic PBC framework. The high deficit policy of Hungarian incumbents (shown above) can be explained, to some extent, by its institutional and political systems, while one factor underlying the debt increase could be elections-related fiscal policies.
Testing for PFC in Hungary is conducted by analyzing the dynamics of the public (fiscal) deficit. We statistically test the hypothesis that the governments may engage in opportunistic behaviors following an expansionary fiscal policy by increasing the deficit to reduce unemployment and increase output before/ during elections.
There is a wider consideration of contractionary post-election effects in theoretical works regarding PBC, which has not been considered as frequently by empirical research and has only been supported occasionally by empirical evidence. Few studies find evidence of expected post-election effects on different fiscal variables (e.g., Ames, 1987;Persson & Tabellini, 2003;Streb et al., 2009) In a recent study, Streb et al. (2012) found that political budget (fiscal) cycles contribute to public debt buildup in OECD countries but not in Latin America.
The remainder of this paper continues with an explanation of the methodology in Section 2, where the data, variables and empirical approach are presented. Section 3 presents the empirical (econometrical) findings of this study, and Section 4 provides the conclusions of the study. The Appendix contains detailed information on each estimated econometrical model.

Data and variables of interest
Consistent with the opportunistic PBC model, we expect that the government may follow an expansionary fiscal policy before/during elections and return to the long-term path or tighten fiscal policy after the elections as constrained by the necessity to sustain public finance and/or reduce inflation. In this regard, the variable of interest we analyze in this article is the fiscal balance of the government. We statistically test the hypothesis of fiscal deficit expansion before elections and normalization or contractions post elections.
We analyze the time series data on net lending (+)/ net borrowing (-) of the general government of Hungary, generally referred to as the overall fiscal balance, which is sourced from the EUROSTAT database. 3 We base our analysis on quarterly data, which, in addition to providing more robust statistical results due to a higher number of observations (compared to a yearly based analysis), most importantly allows inclusion of any inter-annual election effects. Empirical analysis based on annual data has been one serious drawback of many empirical studies analyzing several aspects of PBC, both in developed and developing countries. Streb et al. (2012) argue that the failure of many studies to show econometrically important opportunistic PBC is due to their reliance on annual data. Streb et al. (2012) conduct econometric analyses on both quarterly and annual panel data for a group of Latin American and OECD countries and conclude that the annual data strongly underestimates the presence of political budgetary cycles, particularly when pre-electoral expansion is followed by post-electoral contrac-tion. Based on their results, Streb et al. (2012) argue that temporal aggregation, which is inherent in annual data, is a strong underlying factor that accounts for the non-evidence of PBC in most of the existing empirical research on developed countries. Opposite-sign shifts in fiscal policy within less than a year of elections offset each other, and consequently, PBC is underestimated if annual data are used. Akhmedov and Zhuravskaya (2004) are even more critical of the inter-annual frequency of the time series. In their monthly panel data study that investigated opportunistic PBC in a set of regions in Russia, they argue that even analyses based on quarterly data tend to underestimate the PBC. They find that with respect to Russia, only with monthly frequency data is it possible to correctly estimate the magnitude and timing of generally short-lived but sizable election-related cycles.
In addition to overcoming this potential deficiency of temporal aggregation that is inherent in annual data, we could analyze election effects on the fiscal policy through two different time perspectives by utilizing quarterly time series. First, we could adopt the common approach of analyzing possible opportunistic electoral effects on fiscal deficit during different quarterly cumulative time intervals around elections, ranging from one quarter to eight cumulative quarters (two years) before and after elections. Second, we could adopt another approach, which, to the best of our knowledge, has not been used in the existing empirical research on PFC. This second approach aims to analyze possible electorally driven shifts of fiscal policy (fiscal deficit) during different yearly time windows before and after elections. We do this by creating yearly political dummy variables (PDy) (to be explained in the next sub-section). On the one hand, this new approach avoids the potential problem of temporal aggregation, given that the specification of yearly time

Econometrical approach
Following the standard approach in this area of research, we apply the intervention analysis based on Box and Tiao (1975), which is known as the Box-Tiao approach. This approach has been applied in several where E t denotes an independent error sequence. The simplest term, which corresponds to the t-test in a non-time series setting, is the intervention term/variable. In this case, the intervention variable takes the form of a pulse intervention, meaning an abrupt jump in the series followed by a gradual decline at the normal level of the series. The pulse intervention term can be formally expressed as: Note: For convenience, we denote ( ) T t P by PD representing the political dummy variable.
The first set consists of sixteen cumulative pre-election and post-election political dummy variables defined as: These PDs, which are individually incorporated into the specified models, aim to separately capture the election impact toward fiscal deficit before or after the election, throughout the different cumulative periods of time preceding or succeeding the EQ as well as during the EQ. The second set consists of four yearly political dummy variables (PDy i ) defined as: . Note: In this case, PDsym -1 indicates the election quarter (EQ).
Both PDs from each couplet are incorporated and estimated simultaneously in the statistical models. The main scope of these variables is to indicate the magnitude of the possible fiscal deficit expansion relative to the contraction in exactly the same time interval surrounding the election, which permits inferring any net effect of PFC on the accumulation of public debt.
The fourth set of political dummy variables we design contains twelve V-shaped variables. These are certain ordinal variables intended to mimic the possible shape of the political cycle, which may be a significant determinant in the evolution of the fiscal deficit during an incumbent's full electoral tenure. 4 The empirical technique with these variables was originally used by McCallun (1978) and later by Grier (1987;1989;2008), Beck (1987)  PDcycle 9 model the case in which the expansion of the fiscal deficit occurs before the election (or around the second half of incumbency) and has a greater magnitude and/or lasts longer than the contraction, which occurs after the election (or around the first half of incumbency). Therefore, if statistically significant, the expansion-biased V-shape variables could provide statistical evidence that, in addition to the existence of PBC, the fiscal consolidation that occurs after the election is not enough to fully offset the pre-electoral fiscal expansion, thus contributing to the further accumulation of public debt. While the last four variables, PDcycle 10 to PDcycle 12 , model the opposite case, an expansion of deficit, which is more than offset by a stronger contraction after the election, and implies an overall reductive effect on public debt stock. Figure  1 shows the specific designed form of each V-shaped dummy variable during an incumbent's full tenure (or during the time interval between two consecutive parliamentary elections). Note that some of the PD variables from different sets are identical, such as PDcum 4 and PDy 1 , PDcum e and PDsym -1 , as well as PDcum i and PDsym i for all (i).

Specification and estimation of the statistical models
In the first stage, we precisely followed the Box-Jenkins methodology (Box & Jenkins, 1970), specifying the most appropriate ARMA model for the fiscal deficit denominated both in nominal terms and as a ratio of the GDP. 5 We investigated both time series on the presence of any seasonal pattern as well as on the stationarity. 6 In the event of a non-stationary time series, we transformed the data into a stationary times series by applying the appropriate transformation approach.
We employed an iterative process of identification, estimation and diagnostic checking of several ARMA models until we finally settled on the most plausible model -the one considered the "best" model for each ber of observations used in the empirical analysis was 51. 7 Among all competing possible models we estimated and diagnosed, the "best" model for DLNNETDEF was determined to be the ARMA with two autoregressive terms of lag two AR(2) and four AR(4) and two moving average terms of lags one MA(1) and two MA(2). The "best" model for DNETDEF_GDP was an ARMA with a moving average term of lag one MA (1) and an autoregressive term of lag four AR(4). 8 In the second stage, based on the Box-Tiao intervention analysis, we incorporated one defined political             dummy variable at a time (or in pairs in the case of PDsym i ) into the "best" tentatively found ARMA model and re-estimated all parameters of each final model.
The statistical significance of the political dummy variables, tested using a t-test, reveals any possible impact of the elections on the fiscal deficit. 9 The obtained results are discussed in the following section.

Empirical Results
The In the case of a nominal deficit, the parameter estimates for all PDcum -i and PDcum e variables have a positive sign and are statistically significant at a less than 5% level of significance, while in the event of a deficit denominated as a ratio to GDP, all but PDcum e are statistically positive at less than 5%, strongly implying an election-related expansionary fiscal policy.
The increase in both measures of the deficit before elections intensifies as election day approaches, as indicated by the increase in the magnitude of several consecutive PDcum -i (PDcum -5 <PDcum -4 <PDcum -3 <PDcum -2 <PDcum -1 ). The nominal level of fiscal deficit spikes by almost 80% one quarter before the election quarter (EQ) as indicated by PDcum -1, and almost 70% in the EQ (PDcum e ) compared to its natural longterm pattern modeled by the ARMA components.
The average nominal expansion of the deficit during other cumulative periods before the EQ ranged from approximately 18% to 40% more than its natural pattern, as indicated by estimated coefficients of PDcum -2 to PDcum -7 and was significant at less than 5%.

Conclusions
In our search for political fiscal (budget) cycles in In addition to the presence of a PFC in Hungary and consistent with the common knowledge in the field of political business cycles, we also found some evidence that such politically motivated fiscal cycles may contribute to the accumulation of public debt stock. The fiscal deficit increases prior to elections and is followed by contractions after elections, a phenomenon that leans toward the expansionary side. Nevertheless, we could not conclusively confirm with conventional statistical certainty that existing PFCs in Hungary are also an underlying factor causing the accumulation of public debt in Hungary. Further analysis, however, could reveal a statistically conclusive answer to this hypothesis.
These empirical findings support the hypothesis regarding the existence of politically motivated fiscal policy cycles in Hungary. We found indications, although not statistically significant, that PFCs may contribute to the accumulation of public debt. Therefore, the existence of a PFC can be inefficient for the economy of