1 Introduction

Residents in District III of Budapest have faced challenging times since 2020. The well-known impact of COVID-19 on daily life, spanning healthcare, employment, education, and social care, has been compounded by other adverse developments unfolding amidst the pandemic. In 2020, per capita central grants for municipal projects plummeted by a staggering 86%, dropping from a 5-year average of HUF 20,900 (EUR 58) to HUF 2900 (EUR 8). Similarly, discretionary grants witnessed a sharp decline, decreasing by 73%. Until 2020, the municipality encountered no issues securing central approval for borrowing initiatives, but subsequently, even attempts to amend existing loans were consistently denied by the Cabinet. The roots of these central decisions trace back to the 2019 local elections, where opposition parties surprisingly secured victories in flagship settlements, including the metropolitan municipality of Budapest, most of its districts, and major towns in the countryside. Despite the governing coalition maintaining a similar number of settlements, their budgetary influence at the local level dropped from 58.5% to 42.2%. Citizens may have been aware that voting against the government candidate could bring hard times for the settlement, as several studies, journal notes (e.g., Vasvári, 2019), and political statementsFootnote 1 suggested that aligned municipalities might continue to receive favorable treatment post-election. However, opposition candidates triumphed in numerous settlements, indicating voter acceptance of potential disfavor in central decisions. District III's experience underscores the accuracy of these expectations.

While political favoritism in central decisions is a well-explored topic in Hungary (e.g., Gregor, 2020; Kornai, 2014; Muraközy & Telegdy, 2016; Vasvári, 2022), the recent local elections' unfolding impacts on local affairs provide a compelling basis for further research. Two key factors drive this need: the government's post-pandemic steps toward centralization, including the legal trimming of local revenues, and the increasing role of discretionary grants in central financing, nearly doubling from 2019 to 2020. In response, our research investigates the discretionary grant allocation mechanism in Hungary from 2015 to 2020, with a primary focus on understanding how changes in political affiliation influence transfers received. Using a dataset covering HUF 1193 billion (approx. EUR 3.3 billion) in grants, we employ two identification strategies: fixed-effect (FE) regression and regression discontinuity (RD) design.

Our contribution lies in demonstrating how an unwelcome local election outcome from the government's perspective affects fund allocation mechanisms in Hungary. By delving into the political conversion of municipalities and employing the RD approach, we offer detailed insights into the impact on local governments. To our knowledge, this is the first attempt to investigate the political repercussions of the latest local election, providing potential lessons for the upcoming 2024 election. The topic aligns with recent rule-of-law concerns raised by the European Commission (Politico, 2022) and democratic backsliding noted by the European Parliament (European Parliament 2022a). The EP has also adopted a report suggesting that Hungarian local governments may submit funding applications directly to the European Union (EU), bypassing government institutions (European Parliament, 2022b).

The rest of the paper is organized as follows. The next chapter briefly introduces the theoretical background and reviews relevant literature. The third section outlines the Hungarian institutional setting. Subsequently, we introduce the data and methodology, which is followed by the presentation of the results. Finally, conclusions are drawn.

2 Theoretical background

Fiscal federalism delineates the optimal allocation of public tasks between central and subnational governments to foster social welfare. The decentralization theorem, as articulated by Oates (1977), underscores that social welfare flourishes when local authorities provide public goods rather than relying on a uniform, centrally dictated supply. A crucial prerequisite for decentralization is vertical fiscal balance,Footnote 2 which asserts that those who benefit from expenditures should bear the associated costs in both time and space (Vasvári, 2022). Long-term loans to finance current expenses, a breach of the golden rule, have long been recognized as leading to an inequitable distribution of burdens across time (Musgrave, 1959). Similarly, flawed grant design and biased transfers can distort the outcomes of decentralization, favoring citizens in certain settlementsFootnote 3 while causing welfare losses for those in less-favored areas, assuming uniform taxation at the central level. Initiating local projects from central funds, an increasing share of discretionary grants, and unconditional bailouts can create winners benefiting from central decisions and losers left out. These dynamics may foster patron–client relationships, where local leaders are more inclined to align with the government's preferences to ensure a steady flow of resources (Coman, 2018). Financial difficulties may emerge in disfavored municipalities, leading to deficiencies in public services, project postponements, or infrastructure deterioration. Ultimately, “collective irresponsibility” may manifest, with local and central governments blaming each other for failures in delivering local public services (Vigvári, 2010).

The allocation of funds and central decisions often exhibits a political pattern, investigated thoroughly in the literature. Early studies on political favoritism focused on how governments utilize financial resources to win swing districts (Lindbeck & Weibull, 1987) or reward core supporters (Cox & McCubbins, 1986). In this regard, Ward and John (1999) explored how the UK government directed public resources to marginal constituencies and flagship local governments in the 1994/1995 financial year. Johanson (2003) examined the Swedish case, demonstrating that political competition influences the allocation of intergovernmental transfers, with municipalities having more swing voters receiving more grants. Kauder et al. (2016) found evidence on the core supporter hypothesis in Germany, revealing that settlements with more voters aligned with the incumbent state government receive more discretionary grants. Veiga and Veiga (2013) showed that in election years, the central government of Portugal provided extra grants to municipalities where they expected greater loss of votes. A broad stream of research studied partisanship between central and local leaders. Migueis (2013) investigated municipal elections in Portugal between 1989 and 2001 and found that aligned municipalities received approximately 19% more targetable transfers than unaligned municipalities. Solé-Ollé and Sorribas-Navarro (2008) analyzed grants received by the Spanish municipalities and argued that political alignment with the upper-tier governments is rewarded by increased transfers. Curto-Grau et al. (2018) further investigated the Spanish case and provided evidence of political bias in capital transfers from regional to local governments. Bracco et al. (2015) explored the issue from theoretical and empirical aspects and found that politically aligned Italian municipalities receive more grants by 36–47%.

Pork-barreling in Hungary was also investigated in detail by several papers. Gregor (2020) found evidence that during the national elections of 1998 and 2002, swing and poorer local governments were politically targeted and received larger intergovernmental transfers. Muraközy and Telegdy (2016) investigated the allocation schemes of European Structural Funds between 2004 and 2012, revealing that politically aligned local governments had higher project acceptance rates, and EU subsidies benefited incumbent mayors in local elections. Vasvári (2022) explored the effect of alignment from both local and central perspectives, concluding that political clientelism has been on the rise since 2010, particularly regarding EU funds. Kornai (2014) found evidence of political patterns in bailouts of local governments between 2011 and 2014. Vasvári (2020) argued that unaligned local governments are less likely to receive central approval for their credit applications, limiting their access to financial markets.

3 Institutional background

A two-tier governance system was reinstated in Hungary in 1990 immediately after the transition period. The local tier, marked by significant fragmentation, assumed responsibility for a diverse range of public services, setting Hungary apart within the European landscape (Vasvári, 2022; Vigvári, 2010). The economic foundation of municipal autonomy rested on inherited assets, the authority to levy local taxes, block grants from central governments, and unregulated financial market activities (Kopányi, et al., 2004). This resulted in a highly decentralized and market-driven model of local governance. However, post-2010, the newly elected government, securing a two-thirds majority, prioritized recentralization in public reforms (Dobos, 2021; Karas, 2021). From 2011 onward, deconcentrated forms of public services (e.g., government offices) were favored over municipal (decentralized) service delivery (Jankovics, 2016; Kákai, 2021), with counties now responsible only for regional development conceptualization. Following the reallocation of public services in 2013, municipal expenditure as a percentage of GDP declined from 12.4% to 6.6% between 2010 and 2020. Central grants became predominantly earmarked, and shared taxes and block grants diminished, while discretionary grants assumed an increasingly pivotal role in financing public services and municipal projects. The granting scheme was revised, and now settlements with high local tax income get fewer central grants or may even be obliged to make payments to the central budget. Moreover, the spending of local tax income became further limited by law. The central government also prepaid or took over the outstanding liabilities (loans, bonds) of the entire municipal sector (Kornai, 2014). Since then, municipal borrowing has become conditional on the approval of the central government.

Up until 2020, significant changes in local finances were not apparent following the 2013 reforms and the assumption of municipal debt in 2014. However, in response to the COVID-19 pandemic, additional centralization measures were implemented, further constraining the financial autonomy of local governments (Kákai, 2021). The government is now authorized to establish priority economic zones that channel local tax revenues from the settlements to the counties. Meanwhile, municipalities were also prohibited from increasing local tax rates. The previously shared motor tax income is now withdrawn entirely by the government, while the rate of local business tax paid by small and medium-sized enterprises (SMEs) also decreased by 50% by law.Footnote 4 While initially conceived as temporary measures, it became evident that the central government considered some of them permanent (e.g., motor tax). Beyond the pandemic challenges, the reduction of local autonomy may be rooted in political motives, traceable to the local election in October 2019, where opposition parties experienced unexpected success. Despite a decrease in the number of opposition municipalities from 50 to 41, their budgetary share increased from 8.4% to 19.5%. Notably, half of the settlements won by the opposition were previously aligned, and important flagship municipalities were captured from the government coalition, constituting 14% of the entire sector budget (Table 1).

Table 1 Characteristics of local governments per political affiliation, 2015–2020
Full size table

Despite thorough research on pork-barreling in Hungary, the local election held in 2019 adds a new layer of significance to this topic. Even though it might have been expected that going unaligned would be disfavored and settlements may receive fewer transfers, in many municipal strongholds the majority of the voters went against the tide, which altered the balance of power at the local level. In response, the government took decisive action: they curtailed local financial autonomy while concurrently augmenting the scope of discretionary funds. This shift implies an intensification of political patterns influencing central decisions.

4 Data and methodology

4.1 Descriptive statistics

In our empirical research, we delve into a specific aspect of discretionary decisions, the distribution of supplementary grants allocated to municipalities between 2015 and 2020. The dataset is constructed using the municipal final statements provided by the Hungarian State Treasury for each municipality-year. This comprehensive dataset enables us to categorize transfers based on their distribution mechanisms:

  1. 1.

    Discretionary grants based on government decree. These transfers have various purposes, ranging from additional support for operation (e.g., wages) and public services delivery to funding various municipal projects.

  2. 2.

    Other supplementary transfers based on the budget bill. This category includes various supplementary transfers. Some follow a formulaic structure, like compensation for wages or additional grants for the operation of cultural institutions such as libraries and museums. The calculation of certain transfers is delegated to corresponding ministries, e.g., support for chimney sweep services or public transport. The remaining grants are subject to municipal application, covering areas like additional support for residential water and sewage services or development funds for renovating public buildings and infrastructure (e.g., kindergartens, schools, roads, and pavements).

The total grants under consideration amount to HUF 1193 billion (EUR 3.3 billion), constituting 13.5% of the total earmarked and capital grants (including EU Funds) disbursed by the state during this period (Fig. 1). Our primary focus is on the allocation of discretionary grants, the share of which in total grants exhibits a noteworthy 6.1-percentage-point increase between 2015 and 2020, almost doubling from 2019 to 2020.

Fig. 1
figure 1

Supplementary grants, HUF billion. Own compilation based on data from the Hungarian State Treasury

Full size image

Our key variable for political affiliation is the directly elected mayor for each settlement, as provided by the National Election Office.Footnote 5 Mayors are considered aligned if nominated by the governing coalition (Fidesz-KDNP [Christian Democratic People's Party] during this period). Conversely, they are deemed unaligned if nominated by an opposition party or a party that did not secure parliamentary entry, or ran as an independent candidate.Footnote 6 Alignment status is based on the national elections held in 2014 and 2018, the local elections held in 2014 and 2019, and the midterm local elections between 2015 and 2020. Given Fidesz-KDNP's continuous governance since 2010, changes in alignment status are primarily attributed to local dynamics, particularly the 2019 elections.Footnote 7 Control variables are sourced from data provided by the Hungarian Central Statistics Office and municipal final statements from the Hungarian State Treasury. The characteristics of local governments are detailed in the Appendix (Table 5). Unaligned municipalities generally represent smaller settlements with lower tax capacity and higher unemployment. This trend arises from the fragmented municipal system, where many smaller local governments are outside the sphere of national politics, leading to predominantly independent mayoral candidates in local elections (approximately 80% of settlements). When considering municipalities with a population exceeding 3000, alignment becomes more balanced.

We further categorized local governments based on changes in political alignment to assess their impact on grant distribution. Two groups saw no conversion, remaining aligned or unaligned throughout the period. The third group comprises those shifting to unaligned status, while the last group includes new government allies converting to aligned status. Table 1 presents the size and budgetary power of these groups, while Fig. 2 illustrates the per capita distribution of supplementary grants in each group. Noteworthy shifts in discretionary grants emerged after the 2019 election: loyal municipalities that remained aligned received significantly more, while those who went against the tide and became unaligned earned less compared to previous years, creating a gap exceeding HUF 15,000 (EUR 41.6) per capita between these groups. Initially, unaligned settlements exhibited similar patterns: new government allies were rewarded, with their per capita grant amount more than tripling from 2019 to 2020, while municipalities that remained unaligned experienced minimal change.

Fig. 2
figure 2

Supplementary grants per capita in the conversion groups, HUF thousand (2020). Own compilation based on data from the Hungarian Election Office and the Hungarian State Treasury

Full size image

4.2 Methodology

We incorporate all local governments into our analysis, with certain limitations. Counties and the metropolitan municipality of Budapest were excluded due to their special status, and certain settlements were omitted due to insufficient data. As a result, the final dataset comprises 3155 local governments, encompassing 18,930 municipality-years (with a coverage of 98.7%). Within this dataset, we distinguish local governments with a population exceeding 3000 from smaller towns and villages, totaling 500 local governments and spanning 3000 municipality-years. This differentiation allows us to filter out smaller municipalities where political competition tends to be less intense.Footnote 8 The analysis encompasses a 6-year period from 2015 to 2020.

Our initial approach involves conducting a fixed-effect regression, wherein we account for time-invariant factors that could contribute to the correlation between alignment status and allocated grant values. Additionally, we seek to measure the variance in grant distribution among conversion groups, as illustrated in Fig. 2. The first group comprises initially aligned municipalities, with the coefficient of the Aligned dummy indicating the marginal effect of maintaining aligned status on received grants relative to municipalities that convert to unaligned (blue lines in Fig. 2). The second group includes initially unaligned municipalities, with \(\delta_{1}\) representing the marginal effect of transitioning to aligned status on received grant values relative to unaligned municipalities (gray lines in Fig. 3).Footnote 9 Our fundamental model for FE estimation is outlined as follows:

Fig. 3
figure 3

Margin of victory/loss and grants received (2020). Margin is measured on the X axis, while log grant value per capita received in 2020 is measured on the Y axis. Black lines represent the predicted values of local polynomial estimations. Gray lines are 95% confidence intervals. The dots are bin averages of 2.5% bin size

Full size image
$$\log \left( {\frac{{{\text{Grant}}\;{\text{value}}}}{{{\text{Population}}}}} \right)_{it} = \delta_{1} {\text{Aligned}}_{{{\text{it}}}} + \delta_{2} X_{{{\text{it}}}} + \delta_{3} {\text{Region}}_{i} {\text{Year}}_{t} + \alpha_{i} + \varepsilon_{it},$$

i denotes municipalities and t denotes years. Our key independent variable is the Aligned dummy (1 if the mayor is aligned to the governing coalition in the given year), indicating the marginal effect of the mayor’s political alignment on the received grant value. A positive estimation of \(\delta_{1}\) would suggest a bias towards politically aligned municipalities. Xit includes control budgetary variables of total expenditure per capita, operating balance per capita, and proportion of public service delivery in the total budget. The per capita tax income controls for the economic environment and serves as a proxy for external shocks, such as the pandemic, given its high correlation with locally produced GDP. Socioeconomic factors are accounted for by including unemployment, population density, and the proportion of elderly population.Footnote 10 Region (NUTS2) dummies interacted with year dummies enable the incorporation of otherwise non-varying regional effects. Nominal values are measured in HUF 2020 and deflated by the Consumer Price Index (CPI). The dependent variable is computed as the natural logarithm of grant value per capita, facilitating the management of outlier values and allowing interpretation in percentage terms. The same transformation is applied to control variables where necessary and applicable.

Our second approach employs a sharp regression discontinuity design to capture the grant effect of political alignment at the discontinuity point for the supplementary transfers received in 2020. To achieve this, we include the “Margin of victory or loss” variable in the model as our running variable, while “Aligned” counts as the treatment. Given that mayors are directly elected with a simple majority in each settlement and the districts of Budapest, we follow the calculation method of the running variable applied by Migueis (2013): if the aligned mayor won the election, the vote margin equals the difference between the aligned vote share and the runner-up candidate (right side); if a government coalition candidate lost the election, the vote margin is his or her vote share minus the winning candidate. Consequently, we could include only those settlements where the government coalition had a candidate in the 2019 local election. If the alignment of the mayor influences the distribution of transfers, we expect a “jump” at the cutoff point. A significant coefficient of the margin alone would suggest that the government may preferentially reward core supporters rather than the aligned mayor. Our RDD model is formulated as follows:

$$\log \left( {\frac{{{\text{Grant}}\;{\text{value}}}}{{{\text{Population}}}}} \right)_{i} = \delta_{0} + \delta_{1} {\text{Aligned}}_{i} + f\left( {{\text{Margin}}_{i} } \right) + {\text{Aligned}}_{i} \cdot f\left( {{\text{Margin}}_{i} } \right) + \varepsilon_{i},$$

where Aligned = 1 if the municipality is aligned with the government in 2020, and 0 otherwise, \(\delta_{1}\) represents the marginal impact of alignment on transfers at the cutoff, function f(Margini) represents the relationship between the running variable and the awarded per capita grant value, while the interaction term Alignedi f(Margini) allows us to estimate different slopes on both sides of the cutoff point, as the alignment status may impact not only the intercept but also the slope of the regression curve. We estimate the model using nonparametric (local) and parametric (global) approaches with different functional forms (linear, quadratic, and cubic) along with other robustness checks on the margin and the control variables. Specifically, we also estimate the impact on the subset of data within bandwidth \(h\) to the left and right of the cutoff.

5 Results

5.1 Fixed-effect regression

The results are presented in Table 2, with two columns corresponding to estimations for municipalities with populations under and over 3000. The models were estimated for per capita discretionary and other supplementary grants. The \(\delta_{1}\) coefficient of the Aligned variable is estimated to be 0.619 and significant for major municipalities. Alignment plays a role in the allocation of other supplementary resources in the case of smaller settlements (significant at 5% level), albeit to a much lower extent (0.072). Control variables mostly behave as expected (refer to Table 6 in the Appendix). Higher unemployment and elderly people ratios result in more grants, while population density and per capita tax income correlate negatively with the received grants; this may be due to the compensation mechanism of grant allocation, i.e., urbanization and higher taxing power result in fewer central resources.

Table 2 FE estimates on alignment effect
Full size table

Table 3 presents the results for the conversion groups based on their changes in political alignment. The upper part of the table includes coefficients for initially aligned municipalities, investigating the difference between loyal settlements and those that shifted to unaligned status. The lower section describes initially unaligned local governments (LGs), illustrating how the government treats new allies compared to those that remained unaligned. The estimated coefficient regarding discretionary transfers is significant for large municipalities. Specifically, when an unaligned candidate is elected after aligned leadership, the settlement receives lower discretionary grants, averaging a decrease of 86.4% relative to the loyal settlements. Furthermore, we observe that other supplementary transfers received by smaller settlements are also influenced, albeit to a much lower extent (11.9%). On the other hand, if a municipality converts to aligned status, it will receive 65.2% more grants than those that persistently remain unaligned.

Table 3 FE estimates on the alignment effect in the conversion groups
Full size table

5.1.1 Robustness checks

We performed several checks to test the robustness of the results. First, we tested possible bias stemming from the definition of political alignment and the high number of neutral settlements. As shown previously, independent mayors lead approximately 80% of the settlements, i.e., they may be out of the interest of national politics. Thus, in the first check, we simply dropped independent municipalities throughout the examined period. We found similar results, although the coefficient of the political alignment is somewhat higher (Table 7), suggesting that aligned settlements are even more favored than the ones in opposition. Second, instead of excluding independent municipalities, we introduced a new dummy into the model, indicating municipalities with a mayor in opposition, with independent settlements serving as the reference point. The Aligned dummy retained its significance, and is also confirmed that settlements in opposition may be disadvantaged relative to independent municipalities (Table 8). To avoid the loss of any municipality-years, we replaced zero grant values with one in our specification, resulting in a log value of zero for the dependent variables, a methodology applied by, for instance, Muraközy and Telegdy (2016).Footnote 11 To assess whether this data manipulation impacted our results, we re-estimated the model without the municipality-years with zero grants. Table 9 demonstrates that the strength of the political variable remained consistent and still statistically significant.

5.2 Regression discontinuity

5.2.1 Graphical representation of the discontinuity

Figure 3 illustrates the effect of the election outcome on grant distribution: the upper graphs depict the allocation of per capita discretionary grants in 2020 based on the vote margin, while the graphs below represent the distribution of per capita other supplementary grants Consistent with our prior findings, a noticeable jump occurs at the cutoff point only in the case of discretionary grants awarded to major settlements. Municipalities that marginally fall to the right of the cutoff (the aligned ones) received more discretionary transfers than those marginally to the left (the unaligned ones). Conversely, the other three scenarios (discretionary grants for settlements below 3000; other supplementary grants) appear to be free of political bias.Footnote 12 Therefore, we continue our investigation on the discretionary grants received by major settlements.

5.2.2 Conditions for internal validity

To ensure that the estimates are unbiased, we show that the conditions for internal validity are met; thus, the RD approach can be implemented. First, we show that our forcing variable (i.e., vote margin) is continuous around the cutoff by inspecting the histogram (which shows the frequency of observations in each bin) and applying a manipulation test introduced by Cattaneo et al. (2018). The graphical illustration and the statistics of the discontinuity estimate show no statistical evidence of manipulation of the forcing variable (Fig. 4). We also tested for discontinuity in the case of the controls, and we could not find any evidence of systematic discontinuity (Table 11 and Figure 5 in the Appendix).

Fig. 4
figure 4

Continuity of the forcing variable (major settlements). Robust manipulation test statistics is −1.467 (p-value = .142), nonsignificant

Full size image

5.2.3 RD estimates on the alignment effect

In Table 4 we introduce the results of the parametric (global) and nonparametric (local) RD estimates. The forcing variable is the vote margin, while the dependent variable is the received discretionary grant value per capita (in logarithm), estimated for major settlements. We mainly follow the recommendations in the literature by applying a local linear regression with optimal bandwidth and triangular kernel (e.g., Curto-Grau et al., 2018; Meyersson, 2014; Migueis, 2013). The optimal bandwidth \(h^{*}\) equals 21.8%, calculated according to Calonico et al. (2014). Aligned municipalities that are marginally to the right of the cutoff received approximately twice as many transfers as unaligned municipalities that are just to the left of the cutoff (the estimated coefficient is 1.090). The coefficients for different bandwidths are between 0.869 and 1.398, which still show significant differences in the allocation of transfers between aligned and unaligned municipalities. As the margin variable is nonsignificant, we may suggest that mayors attract the additional transfers, not the core supporters.

Table 4 RD estimate on discretionary transfers for major settlements
Full size table

Table 4 also shows that the estimated alignment effect is significantly higher than indicated by the FE results, which may suggest that the effect of political alignment increased noticeably after the election in 2019. Therefore, we investigated the difference between grants received in 2020 versus 2019. According to the graphical presentation of the results (Figure 6), a significant effect is estimated again for the discretionary grants regarding the major settlements. Therefore, not only do aligned major settlements receive more grants than unaligned ones, but the central government also rewards alignment with increased grant amounts.

5.2.4 Robustness checks

Local RD estimates are robust to the choice of the bandwidth and the kernel functions (Figure 7 in the Appendix) and the inclusion of the control variables (Table 12 in the Appendix). The global RD estimates seem to be robust regarding the functional form (refer to Table 6). We performed placebo tests based on Artés and Jurado (2018) to check whether there is evidence of jumps at nonzero cutoff points by setting it 2% and 4% in both directions (Table 13 and Figure 8 in the Appendix); the graphs show no statistically significant effect on grant values, indicating that the discontinuity can be attributed solely to the alignment effect. We also applied alternate settings (Table 14 and Figure 9). First, to test for reverse causality, we used average discretionary transfer values before the election (granted between 2015 and 2019) as an alternate dependent variable, and we found no significant relationship with alignment; this means that grants prior to the election did not influence changes in the alignment status. In the second setting, we omitted independent municipalities, which further strengthened the role of alignment, similar to FE estimates.

6 Conclusion

In fiscal federalism, centralization often accompanies welfare loss, and it is a well-established fact in a political economy that certain grants can be designed to serve political interests. Hungary has experienced a significant recentralization of local affairs since 2010, with government decisions reflecting political interests in various aspects, including the approval of local borrowing, initiation of projects, and provision of extra funds. Notably, despite this trend, in nearly 200 municipalities that were previously aligned, representing more than two million people, the majority voted against the candidate of the government coalition, leading to fundamental changes in the municipal political landscape. Subsequently, political tension escalated, and the growing opposition in local politics prompted the government to further curtail local financial autonomy. This was achieved by channeling local taxes to county municipalities and by withholding the motor tax, while enacted laws temporarily froze local tax rates and halved local business tax paid by SMEs. Major settlements received compensation primarily through discretionary grants, the extent of which nearly doubled in 2020. Shifting towards sector-wide restrictions and one-by-one compensations has made local governments even more vulnerable to central decisions.

This study delves into the mechanisms of discretionary grant allocation in Hungary from 2015 to 2020. The dataset spans 6 years, allowing us to estimate the impact of political alignment using fixed-effect regression and a regression discontinuity approach. Our findings indicate that political grant-targeting predominantly favors major settlements, while smaller municipalities appear more shielded against changes in political alignment. Considering the initial alignment status, we found evidence that loyalty or becoming aligned with the government pays off in discretionary transfers versus those converting to or remaining unaligned (86.4% and 65.2%, respectively). Regression discontinuity estimates further support these findings, emphasizing strong political patterns in changes in per capita discretionary grants from 2019 to 2020. Considering the average value of grants, rewarding the loyal or the new ally settlements is estimated to be HUF 5760 (EUR 16.0) and HUF 4340 (EUR 12.1) per capita, respectively. These results are in line with Solé-Ollé and Sorribas-Navarro (2008), who estimated the effect of political alignment to EUR 7–11 per capita in the case of Spanish intergovernmental transfers. Muraközy and Telegdy (2016) found a similar extent of political bias in the case of EU funds, estimating that aligned local governments received an additional transfer of HUF 4700–5500 (EUR 13–15) in Hungary. Our regression discontinuity estimates are quite similar to the findings of Curto-Grau et al. (2018), who also estimated that aligned settlements may receive twice the amount of transfers as the unaligned ones.

Our research underscores the fact that political favoritism in Hungary goes beyond providing additional funds to aligned mayors or swing districts. Measures such as trimming revenues, restricting the spending of local funds, and rejecting borrowing initiatives make local governments highly reliant on discretionary decisions, significantly impacting the day-to-day lives of citizens. Our study reveals that political influence in intergovernmental transfers has intensified since 2019, resulting in aligned local governments regularly receiving extra funds, while unaligned settlements receive little or nothing, burdening their residents with the fiscal consequences of political favoritism. Local governments where voters went against the tide now face postponed projects, tightened local budgets, and diminished financial resources, sparking debates over responsibilities. Consequently, a fair and more balanced distribution of resources is advised, encompassing a reduction in vertical imbalance, the restoration of fiscal autonomy at the local level, and a shift towards formulaic allocation of intergovernmental transfers that does not penalize those who opt out.