The impact of budget transparency on tax revenue forecasts: Evidence from Croatian local government units

doi:10.21203/rs.3.rs-3440688/v1

Download PDF

Research Article

The impact of budget transparency on tax revenue forecasts: Evidence from Croatian local government units

https://doi.org/10.21203/rs.3.rs-3440688/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

The tax revenues that are actually collected by local government units for a fiscal year frequently differ from what is planned in their enacted budgets. Despite the literature on budget credibility, little is known about the effect of budget transparency on budget forecasting deviations and errors. This paper analyses whether the level of online local budget transparency (OLBT) affects tax revenue forecasting and the impact of other control variables (e.g., budget balance, income, population, political ideology and strength) on budget deviations for all 556 Croatian local government units (LGUs; cities and municipalities) during the period 2016–2022. This paper also analyses public budget cycles in tax revenue planning at the Croatian local government level. The results show that in LGUs where OLBT increases in pre-election years, underestimation of tax revenues is higher, and overestimation is lower. In election and post-election years, with increased OLBT, underestimation of tax revenues is lower and overestimation is higher.

public budget cycles

tax revenues

budget transparency

budget credibility

local government units

Croatia

Local government units’ (LGUs’) executed (actual) tax revenues for a fiscal year frequently differ from what they planned (forecasted) in their enacted budgets. This paper examines tax revenue forecasting using two measures to analyse deviations in budget planning. The first measure, tax revenue deviation, is the share of planned minus actual tax revenue in planned tax revenue. This measure is used, for example, by Benito, Guillamón & Bastida (2015), Ríos et al. (2018), Cuadrado-Ballesteros, Guillamón & Ríos (2021) and Prijaković, Mačkić & Bronić (2023). The second is tax revenue error, which is the share of planned tax revenue (in the enacted budget) in the actual tax revenue (in the budget year-end report). This measure is used, e.g., by Bischoff & Gohout (2006, 2010) and Goeminne, Geys & Smolders (2008). Tax revenue overestimation arises from planned amounts being greater than executed amounts, and tax revenue underestimation results from planned amounts being lower than the executed amounts. The constitutions of many countries and LGUs place limitations on budget deficits and debt. Budgets that do not adhere to these limitations should not be approved by the legislative authority or carried out by the executive authority. Public and legislative budget oversight are key components of fiscal discipline in these governments (Bischoff & Gohout, 2010). The budgetary process must be open for control to succeed (Alt & Lassen, 2006a). This specifically calls for accurate and objective future tax revenue collection.

Enacted budgets are the plans that all governments use to outline the goals set by decision-makers to state how resources will be acquired and how they will be used to achieve the goals. Enacted budgets often include some years beyond the current fiscal year, allowing governments to make progress towards their objectives systematically over time. Forecasting the resources of revenues or expenditures is a key component of budgeting since financial planning requires perspective (Williams & Calabrese, 2016). Conservative tax revenue forecasting means underestimating tax revenues; such biases making it simpler to fulfil the statutory and constitutional financial obligations arising from the prevalent balanced budget requirements. These pessimistic biases also ensure that there will be money for budget revisions, which might be useful for incumbent politicians (to win over either other elected officials or the general public). In contrast, optimistic tax revenue forecasting means overestimating tax revenue, which might make tax or policy choices that will develop and/or hide structural deficits feasible.

Researchers and practitioners across a range of policy sectors recognise the importance of budget transparency. The roots of public sector budget transparency are the policies, institutions, and practices that disseminate budget information in ways that increase political effectiveness, improve knowledge of public policies, and lessen policy uncertainty. Taxpayers, who provide the majority of funding for the public sector, need open budget information to be able to keep track of the operations of LGUs that use budgets to carry out specified policy goals each fiscal year. Incumbent politicians might not be motivated to use a transparent budget process because they might try to hide the true budget balance to mislead voters (Alesina & Perotti, 1996). In reality, consistently implemented budgets are essential to maintain the legitimacy of choices in policy. Forecasting mistakes can occasionally reflect political decisions as well as technical shortcomings, leading to systemic errors. These systematic budget deviations could indicate that independent auditors or other third parties should keep an eye on the budget process, especially if the budget deviations confuse budget users. In view of the methods politicians employ to influence the budget, budget deviations may also indicate that legislation requiring a balanced budget needs to be reevaluated (Mayper, Granof & Giroux, 1991).

Budget transparency may be a strategy for fiscal responsibility regarding budget credibility planning during the electoral cycle, given that budget openness is necessary for effective political control and public sector monitoring. The definition of budget transparency according to Bronić et al. (2023: 3) is as follows: it “… implies providing an insight into complete, accurate, timely and comprehensible information regarding the budget. This information enables citizens to participate and potentially impact the efficiency of collection and spending of public funds, to demand more accountability from national and LGUs and, consequently, to reduce potential corruptive acts.” Previous research has examined how budget/fiscal transparency affects many fiscal performance indicators, including budget balance, debt and expenditures (e.g. Aguiar-Conraria, Magalhães & Veiga, 2019; Alt & Lassen, 2006a, 2006b; Arbatli & Escolano, 2015; Bronić, Stanić & Prijaković, 2022; Cuadrado-Ballesteros & Bisogno, 2022; Glennerster & Shin, 2008; Mačkić, 2021; Montes, Bastos & de Oliveira, 2019; Prijaković, 2023; and Vicente, Benito & Bastida, 2013). However, the impact of the level of online local budget transparency (OLBT) on tax revenue forecasting has been little tested empirically (Cuadrado-Ballesteros, Guillamón & Ríos, 2021; Ríos et al., 2018). To fill this gap, this paper aims to answer the following research question: Does the level of OLBT affect tax revenue forecasting?

Adequate budget planning is an essential signal for voters; politicians are incentivised to influence tax revenue forecasts, which is a key element of budget planning. According to the theory of political economy, politicians who are members of a left-wing government give more optimistic or less pessimistic tax revenue forecasts than those of a right-wing party. According to common pool theory, a fragmented council tends to make more optimistic or less pessimistic tax revenue forecasts (Jochimsen & Lehmann, 2017). Political budget cycles (PBCs) play a role in tax revenue planning. Here, we analyse all Croatian LGUs – 128 cities and 428 municipalities – during the period 2016–2022. Our results show that in Croatian cities and municipalities where OLBT increases in pre-election years, the underestimation of tax revenues is higher and the overestimation is lower. In pre-election years, LGUs with higher OLBT plan conservative tax revenues rather than those that are excessively optimistic. In election and post-election years, in LGUs where OLBT increases, underestimation of tax revenue is lower and overestimation is higher. Unlike their practice in pre-election years, in election and postelection years, LGUs plan tax revenues too optimistically rather than conservatively, allowing them to plan higher expenditures.

This introduction is followed by the second part, which offers a literature review; the third explains the data and methodology; the fourth gives the results; and the fifth contains the conclusions, limitations and suggestions for further research.

Underestimation or overestimation of tax revenues may result from poor forecasting or/and ineffective tax revenue collection. Therefore, to understand forecasting deviations, there is a need to identify the factors that hinder government efforts to accurately predict future revenue sources and/or those that result in ineffective tax collection. This paper argues that a government’s political and institutional characteristics may influence (or encourage) deviations in tax revenue forecasting. To our knowledge, only one article analyses the influence of an index of budget transparency on budget forecast tax revenue deviations – Ríos et al. (2018). They used a sample of the 100 largest Spanish municipalities during the period 2008–2014, showing that budget transparency influences budget forecast tax revenue deviations. Less budget-transparent municipalities tend to overestimate their tax revenues, which enables them to expand public goods and services without raising taxes immediately. These municipalities may spend less than they planned since they are aware of the overestimation of their tax revenues. While less budget-transparent municipalities overestimate their tax revenues, more budget-transparent municipalities appear to be more conservative in their estimates and underestimate their tax revenues, allowing them to spend more than projected.

The public policy cycle in LGUs is a process of six stages: determination of the agenda, policy formulation, budget forecasting, budget implementation, evaluation and policy change (Ríos et al., 2018). This paper focuses on budget forecasting and forecasting revenues, which is one of the most significant issues in public budgeting. Budget analysts might forecast revenues and expenditure line items with few inaccuracies, but forecasting also serves additional purposes that benefit politicians. In this sense, budget transparency may be a tactic to improve the decision-making process and lessen the motivation for politicians to manipulate the predictions for their objectives. In light of this, this article empirically analyses the impact of OLBT on budget forecasting in tax revenues.

Lago-Peñas & Lago-Peñas (2004), using Spanish municipalities during the period 1985–1995, demonstrated the importance of the political power of the incumbent and the electoral cycle. Deficit forecasting deviations are lower for incumbents with a single-party majority and more significant during election years and for smaller municipalities. Additionally, population size is inversely connected with the variability of deviations; this may be explained if officials’ technical capacity increases with LGU size. In contrast, their estimations indicate that the impact of ideology and of the disintegration of ruling coalitions on deficit deviations is significant. As a result, political variables play a role in the positive association between budgetary flexibility and actual budgets observed in empirical research on the causes of public deficits. They tested how the search for political support ultimately leads to overestimating public deficits when (i) budget procedures are soft, (ii) breaking promises made on higher expenditures and the lowering of taxes is costly in political terms, and (iii) ex-post control by voters and political opposition is imperfect. According to econometric forecasts, incumbents with a single-party majority are less likely to change forecasted budgets. They tend to have more significant anticipated deficits but smaller actual deficits, which might be attributed to greater regularity in the financial process. Second, deficit upward deviations typically increase during election years. Forecasted deficits do not change during election years, but actual deficits do. Additionally, elections result in consistent declines in revenues. The incumbent’s ideology, on the other hand, is irrelevant when attempting to explain deficit deviations.

Bischoff & Gohout (2006; 2010) used ten West German states during the period 1992–2002. Bischoff & Gohout (2006) showed that governments often overestimate their forecasts of tax revenues, while Bischoff & Gohout (2010) found that governments commonly overestimate tax revenue during election years. Bischoff & Gohout (2006) contend that incumbent state lawmakers may do this to raise expenditures, increasing the debt and, therefore, their chances of winning re-election. The degree of overestimation is not affected by elections or partisanship, but it is discovered that the politician’s popularity has a significant impact. The degree of overestimation in tax revenues increases with the popularity of the incumbent’s party and, thus, with the likelihood of its being re-elected. This finding lends credence to the idea that due to the uncertainty of re-election, governments accept deficits to accelerate expenditures that were initially budgeted for future periods. Since tax estimates are invisible ex ante, unlike open deficits, they have a special allure when used for this purpose. As a result, they are exempt from constitutional limitations on public deficits. Furthermore, they are probably less damaging to the incumbent’s perceived valence and consequent popularity. To improve external control, the budget process must be transparent. Bischoff & Gohout (2010) found that governments that are unsure whether they will be re-elected should provide overly optimistic tax revenue forecasts every year. The greater the degree of overestimation, the less popular politicians are. These findings support the idea that the uncertainty of re-election induces governments to tolerate deficits to accelerate expenditures that were initially budgeted for later periods and/or impose more debt on future governments. According to the same article, overestimated forecasts of tax revenues are a substitute for explicit deficits. Governments that overestimate tax revenues can incorporate increased expenditures without obviously increasing debt levels if they disclose exaggerated tax revenues. In contrast, they can lower explicit deficits without reducing expenditures. According to the literature, an incumbent party may benefit from this potential in two ways. It can first exploit overestimated forecast tax revenues to boost its chances of winning re-election. Second, it may advance future expenditures that benefit its constituency or put financial constraints on future governments if re-election appears questionable or improbable.

Goeminne, Geys & Smolders (2008), using a sample of 242 Flemish municipalities during the period 1992–2002, conclude that two-party governments are more optimistic than single-party governments. Governments with at least three parties are much more cautious (or less optimistic) in their forecast revenues than single-party or two-party governments. Extending this study, Chatagny & Soguel (2012) used 26 Swiss cantons between 1980 and 2002 to show that underestimating tax revenue considerably lowers fiscal deficits. Furthermore, they demonstrated that a decline in expenditures mediates this benefit. Overestimation and underestimation have symmetrical consequences on budgetary deficits. Thus, this study proves that conservative tax revenue forecasts lower fiscal deficits and supports the inverse claim that conservative tax revenue budgeting additionally increases budgetary deficits.

Benito et al. (2015), using 2,644 Spanish municipalities with populations of over 1,000 during the period 2002–2010, examine electoral cycle-influenced opportunism in the drivers of municipalities’ budget forecast quality. In their efforts to gain popularity and electoral support, incumbents overestimate revenue in order to spend more, and they underestimate expenditures. These authors prove that the budget forecasting process has to be open and subject to external auditing or oversight. Their findings demonstrate that opportunistic behaviour is influenced by the electoral cycle. They identify two manipulation techniques: (1) since incumbents match anticipated revenues to budgeted expenditures, they overestimate revenue, which enables them to spend more; (2) actual spending in the pursuit of greater approval and electoral support surpasses projected spending. As a result, if there is no legal consequence or public disapproval, governments can manipulate budget estimates in accordance with the timing of elections.

Sustainable budgets serve as crucial quality indicators for the electorate. Politicians may therefore be motivated to make an impact on tax revenue forecasts, which are frequently viewed as a crucial component of long-term national budget planning. Jochimsen & Lehmann (2017), using 18 OECD countries during the period 1996–2012, assessed whether national tax revenue forecasts are skewed as a result of political manipulation. Using political economy theories, they found support for partisan politics. Left-wing governments tend to produce more optimistic or less pessimistic tax revenue forecasts than right-wing governments. Theoretically, the common pool problem predicts that more fragmented governments will create more pessimistic or less optimistic tax revenue forecasts. Politicians may be likely to have an incentive to manipulate tax revenue projection because it serves as the foundation for the government’s expected total revenues, and there are various justifications for this behaviour in the theoretical political economics literature.

Cuadrado-Ballesteros et al. (2021), using a sample of 140 Spanish municipalities for the period 2008–2018, examine the impact of the gender of the mayor and the proportion of female councillors in the local council on tax revenue deviations. Municipalities with female mayors and a higher proportion of female councillors tend to overestimate tax revenues, leading to poor financial conditions. These effects are reversed when the proportion of female councillors is relatively high: municipalities then tend to underestimate tax revenues, which leads to better financial conditions. These findings highlight the relevance of considering the nonlinear influence of women councillors in local councils because women’s expected effect might contribute to LGUs’ financial health (Cuadrado-Ballesteros et al., 2021). Female involvement in LGUs enhances the likelihood that fiscal and financial stability legislation will be followed. The findings suggest that because women tend to adopt a more pessimistic budgeting perspective, causing LGUs to underestimate budgeted tax revenues, women’s presence in LGU councils can improve their financial situation; this is in addition to the moral obligation to promote gender diversity and equality of opportunities between men and women. However, to achieve this favourable outcome, their numbers must be relatively substantial; otherwise, the interests and viewpoints of the few women may be diluted among their male counterparts, who hold various viewpoints as per the stereotypes put out by social role theory.

Bohn & Veiga (2021) proved three propositions of political forecast cycle (PFC) models. First, incumbent politicians may lower their revenue forecasts if the ex ante forecast winning margin is higher. Second, when the incumbent politician is not running, his or her incentive to win the elections is diminished, and opportunistic manipulation is reduced. Third, any form of budget manipulation is more successful if there are more uninformed voters. These three propositions were tested and proven using 308 Portuguese municipalities during the 1998–2017 period. The findings unmistakably show that there is revenue forecast manipulation in election years and that the manipulation is smaller when higher-win margins were evidenced at the previous elections, when the incumbent politicians do not offer themselves for re-election, and when the municipality’s population has a lower level of education.

Couture & Imbeau (2009), Anessi-Pessina & Sicilia (2015), and Boukari & Veiga (2018) have used political and electoral variables, such as political ideology, the closeness of local elections, and the fragmentation of government, to explain the presence of variations in local planned tax revenue. Additionally, they have accounted for other socioeconomic aspects, including unemployment, economic level, fiscal/budget transparency, financial situation, and fiscal regulations. Some researchers claim that politicians may be motivated to interfere with public expenditures to accomplish personal goals. Alesina and Perotti (1999) point out that politicians may purposely overestimate tax revenue, leading to more significant deficits. Politicians might use this tactic to blame deficits on the technological challenges involved in forecasting future tax revenue. Van der Ploeg (2010) has theoretically demonstrated that it may be preferable to purposely underestimate the tax base to restrain expenditure and politicians’ financial appetites and minimise debt. The government reduces debt by raising taxes and lowering public expenditures to protect against future shocks. Chatagny & Soguel (2012) contended that such a buffer may be produced by underestimating tax revenues to restrain expenditures, generate extra revenue as a result, and reduce the deficit. However, a precautionary buffer may be created to boost expenditure in an election year (Bischoff & Gohout, 2010; van der Ploeg, 2010). Therefore, careful planning of tax revenues does not ensure that deficits and debt will be minimised.

According to two theories, incumbent politicians typically choose to be ambiguous. First, fiscal illusion theory suggests that uninformed and naive voters frequently underestimate the costs of ongoing and upcoming public projects, especially when budgets are not fiscally transparent. However, governments can distort budget estimates to suit their preferences due to a lack of transparency. Politicians may, therefore, be influenced by incentives to be either optimistic or pessimistic (Mayper et al., 1991; Chatagny & Soguel, 2012; Goeminne et al., 2008). Governments might enhance services without raising taxes right away or meet the need for a balanced budget by being too optimistic about revenues. Additionally, optimistic revenue projections have a smaller political cost than higher taxes in terms of votes lost. The second, principal-agent theory, claims that voters and incumbent politicians may not have the same goals in mind since politicians occasionally prioritise their interests, which do not always maximise voters’ welfare. Voters can likewise not assess incumbents’ competency, which is their capacity to offer particular service levels at the lowest taxes. To enhance their chances of being re-elected, incumbents make an effort to demonstrate their competence to the public, especially when elections are close. As a result, a lack of budget transparency may provide politicians with a competitive advantage in attaining their objectives and allow LGUs to customise budget predictions according to their preferences to appear competent to voters and, as a result, increase their chances of re-election. The models of PBCs suggest that LGUs would try to implement more popular policies shortly before an election. Similarly, more unpopular policies are implemented shortly after elections. To increase their prospects of winning elections or gaining re-election, politicians may make overly optimistic projections of budget variables, which causes revenues to be overestimated (Drazen, 2008). On the one hand, they can predict optimistic revenues and receive more support from the public and politicians before elections (but they also run the danger of damaging their brand if they do not fulfil their forecasts). On the other hand, they might offer gloomy budget revenue forecasts, which have a negative impact on both ex ante and ex post reputations. Pessimistic revenue projections (underestimation) may provide a buffer for unforeseen costs or revenue shortfalls and demonstrate that wise management produces year-end operational savings.

Third-generation models of PBCs include models based on the limited rationality of voters and have been proposed by scholars such as Bohn (2018), Bohn & Veiga (2021) and Crombach & Bohn (2022). The previous two generations include models based on information asymmetry (Rogoff, 1990; Rogoff & Sibert, 1988) and a model based on moral hazard (Shi & Svensson, 2006). They cannot explain why political budget manipulations occur and systematically increase the chances of re-election or that the PBC is conditioned by a specific institutional context, i.e., in budget non-transparent governments. PFCs introduce unsophisticated (poorly informed) voters and explain that the vote share of the current politician in power increases due to his or her electoral manipulations. The increased suspicions of unsophisticated voters are reinforced by the PBC, while the manipulations of politicians must be stronger than the suspicions of unsophisticated voters to be effective. The effect produced by unsophisticated voters on political cycles is studied in the context of forecasting manipulation. Some of the motives for forecast manipulation are as follows: (a) the politician is too optimistic (overestimates revenues) to have more room for manipulation before the election, or (b) the politician is too pessimistic (underestimates the budget balance) to demonstrate his own level of competence by being able to implement unexpected expansionary fiscal policies. Manipulations based on overly optimistic forecasts before upcoming elections create PBCs and lead to budget deficits before elections despite the existence of a balanced budget obligation.

As previously stated, although certain theories contend that some politicians lack the motivation to provide financial information, budget transparency is increasing due to, among other things, outside demands and limits. However, there are meaningful variations in the level of OLBT and the precision of budget forecasts between LGUs, which raises the possibility that these discrepancies are not accidental (Williams & Calabrese, 2016). The lack of budget transparency may give politicians more chances to use optimistic or pessimistic budget forecasts to persuade voters. In this regard, we state the hypothesis that the level of OLBT has a significant impact on tax revenue forecasting.

The Institute of Public Finance has annually calculated the online local budget index (OLBI) for all Croatian counties, cities and municipalities since 2014. The OLBI can serve as an indicator of the political responsibility of local politicians because it measures whether the management of public funds is in accordance with the principles of good governance. The OLBI shows the availability of five key budget documents on the official websites of LGUs at the time of the research (budget proposal, enacted budget, citizen budget, year-end and mid-year reports) and assigns a score ranging from 0 to 5, depending on the number of budget documents available.

Graph 1. OLBI, all 556 LGUs, 2016–2022, number of online published budget documents

Budget transparency for all Croatian cities and municipalities from 2016–2022 has been continuously improving except in 2021, when there is a slight decrease compared to the previous cycle (Graph 1). The average number of budget documents published during the observed period is 4.0.

Figure 1 gives the best illustration of the increasing OLBT among Croatian cities and municipalities during the observed period. In 2016, there were 43 LGUs with no budget document available on their official websites, while in 2022, there was one such LGU. In 2022, there were 417 LGUs with all five budget documents available, in comparison with only 106 in 2016. OLBI values from 0 to 3 refer to LGUs with low OLBT. In 2016, there were 322 LGUs with low OLBT, and there were only 44 in 2022. Furthermore, OLBI values of 4 and 5 indicate high OLBT. In 2022, there were 512 LGUs with high OLBT, in comparison to 234 in 2016.

Graph 2. Tax revenue deviation, all 556 LGUs, 2016–2022, in %

In Graph 2, presenting a measure of tax revenue deviation, the overestimation of tax revenues is the highest in the election years (2017 and 2021), while underestimation is the highest in the post-election year (2018). In the pre-election year (2016), there is an underestimation of tax revenues, while in (2020), due to the pandemic, tax revenues are overestimated.

Graph 3. Tax revenue error, all 556 LGUs, 2016–2022, in %

Graph 3, presenting a measure of tax revenue error, shows the highest overestimation of tax revenues due to the coronavirus pandemic in the election year (2021). This measure moves around one during the other years (collected tax revenues tend to plan tax revenues).

The paper’s primary goal is to conduct a regression model with panel data to test the impact of OLBT on tax revenue forecasting in Croatian LGUs. Fixed-effects (FE) regression (within the regression estimator) and random-effects (RE) regression were conducted using the generalised least squares (GLS) estimator (producing a matrix-weighted average of the between and within results) (Baltagi, 2013). After applying a Hausman test on estimates from the FE and RE models, this paper reports the results of the FE regression. Table 1 gives all variables used according to previous research and theory. The control variables used to explain tax revenue forecasting are budget balance pc, income pc, population, political ideology and strength.

Table 1

Definitions of variables
Variable	Definition	Source
Dependent variables
Tax revenue deviation (devtax)	Tax revenue deviation is defined as the share of the difference in plan and actual tax revenues in the plan tax revenues. A share higher than zero means the plan was higher than actual tax revenue (overestimation or optimistic planning of tax revenues). A share lower than zero indicates the plan was lower than actual tax revenue (underestimation or conservative planning of tax revenues). (in %)	Data on actual tax revenues from the Ministry of Finance – MoF (2022) and planned tax revenues from Enacted budgets on the official websites of LGUs.
Tax revenue error (errortax)	Tax revenue error is defined as the share of the plan in actual tax revenues. A share higher than one means that the plan was higher than actual tax revenue (overestimation or optimistic planning of tax revenues). A share lower than one means that the plan was lower than actual tax revenue (underestimation or conservative planning of tax revenues). (in %)
Independent key variable
OLBI	Budget transparency is measured annually as the online availability of five key budget documents on the official websites of LGUs (budget proposal, enacted budget, citizen budget, year-end and mid-year reports), ranging from 0 to 5.	Bronić et al. (2023)
PBC variables[2]
Pre-election	Dummy variable: 1 in a pre-election year (2016 and 2020); 0 otherwise.	Author
Election	Dummy variable: 1 in an election year (2017 and 2021); 0 otherwise.
Post-election	Dummy variable: 1 in a post-election year (2018 and 2022); 0 otherwise.
Independent control variables
Balance pc	The budget balance is total revenues minus total expenditures, calculated per capita (pc). It could be positive (surplus) or negative (deficit).	MoF (2022)
Income pc	Residents’ average annual income pc in thousands HRK.	Ministry of Regional Development and EU Funds – MRRFEU (2022)
Population	The population estimate for 2016–2020, Census from 2021 for 2021–2022.	Croatian Bureau of Statistics – CBS (2022)
Ideology	Political ideology, dummy variable: 1 left wing; 0 otherwise.	State Electoral Commission – SEC (2022)
Strength	Political strength is the measure of fragmentation of local councils, calculated using the following equation: ${\sum }_{i=1}^{n}\frac{{S}_{i}^{2}}{{S}^{2}}$, where $S$ is the total number of seats in the local council and ${S}_{i}$ is the number of mayor’s party councillors in the council. Values range from 0 (no councillors belong to the mayor’s party) to 100 (all councillors belong to the mayor’s party).	SEC (2022)
Note: All variables refer to the period 2016–2022. Source: Author.

[2] Regression analysis uses some interaction variables from variables mentioned in Table 1.

To analyse the impact of the level of OLBT on tax revenue forecasting, the estimated model is as follows:
$${dependent}_{it}={\beta }_{0}+{\beta }_{1}OLB{I}_{it}+{\beta }_{2}balance{pc}_{it}+{\beta }_{3}income{pc}_{it}+{\beta }_{4}populatio{n}_{it}+{\beta }_{5}ideolog{y}_{it}+{\beta }_{6}strengt{h}_{it}+{\beta }_{7}{PBC}_{it}+{\beta }_{8}{\delta }_{i}+{\epsilon }_{it},$$

where $i$ indicates LGU, $t$ indicates time period, ${\beta }_{0}$ is constant, ${\beta }_{1}, \dots , {\beta }_{8}$ are parameters to be estimated, ${\delta }_{i}$ is time dummies and ${\epsilon }_{it}$ is the error term. In the model time dummies are included. Our dependent variables are calculated as follows:

$$dev{tax}_{it}=\frac{plan tax revenu{e}_{it}-actual tax revenu{e}_{it}}{plan tax revenu{e}_{it}}$$

and

$$error{tax}_{it}=\frac{plan tax revenu{e}_{it}}{actual tax revenu{e}_{it}}$$

where planned tax revenue refers to the value of the planned collection of tax revenue in the enacted budget, and actual tax revenue refers to the value of collected tax revenues in the year-end report. If the value of tax revenue deviation is higher than 0 (tax revenue error higher than 1), then the planned collection is higher than the actual collection of tax revenue; i.e., the LGU is overestimating or optimistically planning tax revenues. On the other hand, if the value of tax revenue deviation is lower than 0 (tax revenue error lower than 1), the planned collection is lower than the actual tax revenue collected; i.e., the LGU underestimates or conservatively plans tax revenues.

LGUs in Croatia generate tax revenue from various sources. Tax revenues are recorded in the report on revenues and expenditures, receipts and outlays. The Law on Local Taxes (2023) defines LGUs’ taxes and other sorts of own revenues. The personal and income tax (PIT) and surtax, as well as taxes on local consumption, vacation homes, the use of public lands, and real estate transactions, are examples of municipality or city taxes. A municipal piggyback tax that is paid on top of the PIT (imposed by the central government) is known as a surtax. The central government’s tax PIT base serves as its tax base, while LGUs choose the surtax rate. The LGU where taxpayers reside (and not where they work) collects the surtax. The surtax may be implemented by all municipalities and cities. Municipalities may impose the surtax at a rate of up to 10%, cities with fewer than 30,000 residents may do so at a rate of up to 12%, and cities with 30,000 or more residents may do so at a rate of up to 15%. Zagreb may impose up to 18%. In the continuation of the paper, there is an analysis of the tax revenues.

Table A1 presents descriptive statistics for all used variables and interaction variables. The mean values for the dependent variables are different due to different measurements; tax revenue deviations are negative (underestimation), while tax revenue errors are positive (overestimation). This is because of the enormous amount of tax revenues; deviations are centred around zero, while errors are centred around one. Furthermore, numerous LGUs exhibit high OLBT. Table A1 shows larger forecasting deviations and errors in election years than in nonelection years, which Bohn & Veiga (2021) proved. According to rational expectations theory, LGUs want to increase their forecasts of additional revenues until the marginal deficit cost and the punishment effect (which are very small for low levels of forecast manipulation) exceed the marginal benefit from expenditures on public goods and services, which translates into a marginal benefit from increasing the chances of being re-elected. This implies higher deficits, which have to be paid for in nonelection years.

Table 2 presents the results for tax revenue deviations during 2016–2022. This paper examines OLBT as well as PBC in tax revenue forecasting. The mean value for tax revenue deviations is negative (-0.09), which indicates underestimation or conservative planning. Model (1) in Table 2 suggests that in election years, LGUs’ underestimation of tax revenues is lower; i.e., collected tax revenues are closer to planned values. Model (2) in Table 2 suggests that LGUs exhibit greater underestimation of tax revenues in the pre-election years.

In all tested models, balance pc has a negative and significant relationship with tax revenue forecasting (tax revenue deviations and errors). This implies that in LGUs in which surplus pc is growing, underestimation of tax revenue is greater (too conservative planning) and overestimation is lower (collected tax revenues are closer to planned values). This is in line with the previous research of Chatagny & Soguel (2012), who showed that underestimating tax revenue considerably lowers fiscal deficits, while conservative planning of tax revenue exhibits higher fiscal surpluses.

Furthermore, income pc has a negative and significant relationship with tax revenue forecasting (tax revenue deviations and errors) in all tested models. This implies that in LGUs where income pc is growing, underestimation of tax revenues is higher (too conservative planning) and overestimation is lower (collected tax revenues are closer to planned values). This is in line with it being better to underestimate revenues, for in this way planned expenditures are lower than in the case in which revenues are overestimated, when planned expenditures are higher but are not matched by the necessary revenues.

In contrast, in most of the tested models, the population has a positive and significant relationship with the forecasting of tax revenues (tax revenue deviations and errors). This implies that in LGUs where the population is growing, the underestimation of tax revenues is lower (collected tax revenues are closer to planned values) and overestimation is higher (too optimistic planning). This is in line with research from Lago-Peñas & Lago-Peñas (2004), who proved that population is inversely connected with the variability of positive and negative deviations.

Political ideology is a dummy variable equal to 1 if the mayor’s party is left-wing. In the tested models, the variable ideology has a positive and significant relationship with tax revenue deviations only when the models include election and post-election years. This implies that in election and post-election years, incumbent politicians supported by a left-wing government underestimate tax revenues less; i.e., the collection of tax revenues is closer to planned values. Jochimsen & Lehmann (2017) proved that left-wing governments tend to produce more optimistic or less conservative tax revenue forecasts than right-wing governments. Politicians may be likely to have an incentive to manipulate tax revenue forecasts because they serve as the foundation for the government’s expected total revenues, and there are various justifications for this behaviour in the theoretical political economics literature. Moreover, ideology has a negative and significant relationship with tax revenue deviations in models for pre-election years. This indicates that incumbent politicians with support from the right make higher underestimations of tax revenues in pre-election years.

The variable of political strength is not significant in any of our models, although Goeminne et al. (2008) conclude that two-party governments are more optimistic than single-party governments and that governments with at least three parties are much more cautious (or less optimistic) in their revenue forecasts than single-party or two-party governments.

Table 2

Results for tax revenue deviation, 2016–2022
devtax	(1)	(2)	(3)	(4)
OLBI	0.0057
OLBI	(0.471)
Pre-election	-0.0335
Pre-election	(-0.533)
Election	0.4802***
Election	(10.18)
Post-election	0.0731
Post-election	(1.167)
Preelection$\bullet$OLBI		-0.0340**
Preelection$\bullet$OLBI		(-2.479)
Election$\bullet$OLBI			0.0108
Election$\bullet$OLBI			(0.698)
Postelection$\bullet$OLBI				0.096
Postelection$\bullet$OLBI				(1.624)
Balance pc	-0.00003***	-0.00003***	-0.00001	-0.00008**
Balance pc	(-3.334)	(-3.181)	(-1.048)	(-2.387)
Income pc	-0.0206**	-0.0155**	-0.0210***	-0.0354
Income pc	(-2.205)	(-2.260)	(-2.950)	(-1.525)
Population	0.00001	0.00002	0.00002**	0.00002
Population	(1.576)	(1.202)	(1.977)	(1.251)
Ideology	0.0636*	-0.0593**	0.0933**	0.2535**
Ideology	(1.705)	(-2.279)	(2.007)	(2.378)
Strength	-0.0002	-0.002	0.0014	-0.0008
Strength	(-0.166)	(-1.437)	(1.086)	(-0.207)
Constant	0.3787	0.411	0.3647	-0.125
Constant	(1.305)	(1.570)	(1.623)	(-0.157)
2017.year	-0.3884***
2017.year	(-4.977)
2018.year	-0.5732***
2018.year	(-7.048)
2019.year	-
2020.year	0.24***	0.2672***
2020.year	(3.757)	(5.044)
2021.year	-		0.3878***
2021.year			(5.607)
2022.year	-			0.6275***
2022.year				(3.547)
Observations	3,867	1,100	1,106	1,107
R-squared	0.162	0.117	0.141	0.135
Number of id	555	555	555	555
Note: Robust t-statistics in parentheses. p < 0.01, p < 0.05, * p < 0.1. Source: Author.*

Table 3 presents the results for tax revenue errors during 2016–2022. The mean value is higher than one, which indicates the overestimation of tax revenues. Model (1) in Table 3 suggests that in the pre-election years, LGUs’ overestimation of tax revenues is lower; i.e., tax collection is closer to planned values. Furthermore, in the election and post-election years, LGUs’ overestimation of tax revenues is growing, and it is higher in the election years. Models (2) and (4) prove that in the pre-election years, LGUs’ overestimation of tax revenues is lower, and in post-election years, it is higher.

Table 3

*Results for* tax revenue errors, *2016–2022*
Errortax	(1)	(2)	(3)	(4)
OLBI	0.00825
OLBI	(0.701)
Pre-election	-0.256***
Pre-election	(-3.534)
Election	1.190***
Election	(14.80)
Post-election	0.309***
Post-election	(6.279)
Preelection$\bullet$OLBI		-0.0267**
Preelection$\bullet$OLBI		(-2.519)
Election$\bullet$OLBI			0.0289
Election$\bullet$OLBI			(0.639)
Postelection$\bullet$OLBI				0.0813***
Postelection$\bullet$OLBI				(2.660)
Balance pc	-0.00003***	-0.00003***	-0.00008**	-0.00004***
Balance pc	(-3.333)	(-3.411)	(-2.219)	(-2.979)
Income pc	-0.0547***	-0.0206***	-0.0922***	-0.0167*
Income pc	(-5.134)	(-2.668)	(-3.553)	(-1.885)
Population	0.00002*	0.00002	0.00004*	0.000008
Population	(1.741)	(1.395)	(1.853)	(1.159)
Ideology	0.0633	-0.0401	0.210	0.0482
Ideology	(1.073)	(-1.251)	(1.368)	(0.472)
Strength	0.000619	-0.00112	-0.00149	0.00205
Strength	(0.373)	(-0.907)	(-0.373)	(1.195)
Constant	2.403***	1.490***	3.270***	0.810***
Constant	(6.414)	(5.576)	(4.103)	(2.752)
2017.year	-1.212***
2017.year	(-10.43)
2018.year	-0.549***
2018.year	(-8.117)
2019.year	-
2020.year	0.463***	0.287***
2020.year	(5.842)	(4.878)
2021.year	-		1.500***
2021.year			(6.005)
2022.year	-			0.290***
2022.year				(4.726)
Observations	3,867	1,100	1,106	1,107
R-squared	0.284	0.179	0.282	0.151
Number of id	555	555	555	555
Note: Robust t-statistics in parentheses. p < 0.01, p < 0.05, * p < 0.1. Source: Author.*

Tables 4 and 5 display the results of forecasting tax revenues in pre-election years. Models (1)-(4) in Table 4 show a negative and significant relation between OLBT and tax revenue deviations. In pre-election years, LGUs with higher OLBT underestimate tax revenues more. 2016 and 2020 were pre-election years, and the tested models confirm that 2020 had higher tax revenue deviations.

Table 4

Results for deviations in pre-election years, 2016–2022
devtax	(1)	(2)	(3)	(4)	(5)	(6)
OLBI	-0.0317**	-0.0316**	-0.0341**	-0.034**
OLBI	(-2.291)	(-2.295)	(-2.472)	(-2.479)
Low OLBI					-0.0375
Low OLBI					(-1.263)
High OLBI						-0.0123
High OLBI						(-0.535)
Pre-election	-	-	-	-	-	-
Preelection$\bullet$OLBI	-	-	-	-
Preelection$\bullet$Low OLBI					-
Preelection$\bullet$High OLBI						-
Balance pc			-0.00003***	-0.00003***	-0.00007**	-0.00003***
Balance pc			(-3.118)	(-3.181)	(-2.427)	(-3.196)
Income pc			-0.0163**	-0.0155**	-0.0067	-0.0156**
Income pc			(-2.432)	(-2.260)	(-0.436)	(-2.505)
Population			0.00001	0.00002	0.0001	0.00002
Population			(1.068)	(1.202)	(0.634)	(1.182)
Ideology		-0.0571**		-0.0593**	-0.0804	-0.046
Ideology		(-2.205)		(-2.279)	(-0.850)	(-1.369)
Strength		-0.002		-0.002	-0.0006	0.0008
Strength		(-1.377)		(-1.437)	(-0.159)	(0.684)
Constant	0.03644	0.1296	0.3535	0.411	-0.1826	0.1707
Constant	(0.755)	(1.415)	(1.469)	(1.570)	(-0.221)	(0.617)
2020.year	0.1663***	0.1713***	0.2674***	0.2672***	0.1482	0.2564***
	(10.02)	(9.851)	(5.195)	(5.044)	(1.343)	(5.628)
Observations	1,100	1,100	1,100	1,100	384	716
R-squared	0.091	0.097	0.111	0.117	0.216	0.437
Number of id	555	555	555	555	325	492
Note: Robust t-statistics in parentheses. p < 0.01, p < 0.05, * p < 0.1. Source: Author.*

Models (1)-(4) in Table 4 show a negative and significant relation between OLBT and tax revenue errors. In pre-election years, LGUs with higher OLBT overestimate tax revenues less.

Table 5

Results for errors in pre-election years, 2016–2022
errortax	(1)	(2)	(3)	(4)	(5)	(6)
OLBI	-0.0235**	-0.0235**	-0.0268**	-0.0267**
OLBI	(-2.163)	(-2.161)	(-2.521)	(-2.519)
Low OLBI					-0.0349
Low OLBI					(-1.015)
High OLBI						-0.00554
High OLBI						(-0.191)
Pre-election	-	-	-	-	-	-
Preelection$\bullet$OLBI	-	-	-	-
Preelection$\bullet$Low OLBI					-
Preelection$\bullet$High OLBI						-
Balance pc			-0.00004***	-0.00004***	-0.00008**	-0.00004***
Balance pc			(-3.352)	(-3.411)	(-2.664)	(-3.186)
Income pc			-0.0211***	-0.0206***	-0.00973	-0.0283***
Income pc			(-2.765)	(-2.668)	(-0.549)	(-3.582)
Population			0.00002	0.00002	0.000163	0.00003
Population			(1.305)	(1.395)	(0.778)	(1.201)
Ideology		-0.0370		-0.0401	-0.0809	-0.0334
Ideology		(-1.146)		(-1.251)	(-0.739)	(-0.785)
Strength		-0.00108		-0.00112	-0.00167	0.00124
Strength		(-0.819)		(-0.907)	(-0.391)	(0.869)
Constant	1.076***	1.128***	1.460***	1.490***	0.840	1.450***
Constant	(29.66)	(15.37)	(5.730)	(5.576)	(0.880)	(4.139)
2020.year	0.156***	0.159***	0.287***	0.287***	0.160	0.340***
	(8.889)	(8.626)	(4.958)	(4.878)	(1.251)	(5.891)
Observations	1,100	1,100	1,100	1,100	384	716
R-squared	0.123	0.126	0.175	0.179	0.191	0.371
Number of id	555	555	555	555	325	492
Note: Robust t-statistics in parentheses. p < 0.01, p < 0.05, * p < 0.1. Source: Author.*

Tables A2 and A3 show the results of forecasting tax revenues in election years. The models do not show a significant relation between OLBT and the forecasting of tax revenues (tax revenue deviations and errors). 2017 and 2021 were election years, and the tested models confirm that 2021 had higher tax revenue deviations. The analysis includes dummy variables for the years; the base year is 2016, in which LGUs underestimated tax revenue in their forecasts. They overestimated tax revenues in 2017 and 2021, which were election years, but underestimated tax revenue in 2018, 2019, 2020 and 2022 (non-election years).

Tables 6 and 7 give the results of forecasting tax revenues in post-election years. Models (1) and (2) in Table 6 show a positive and significant relation between OLBT and tax revenue deviations. In post-election years, LGUs with higher OLBT are less inclined to underestimate tax revenues; i.e., collection of tax revenues is closer to planned values. The post-election years were 2018 and 2022, and tax revenue deviations were higher.

Table 6

Results for deviations in post-election years, 2016–2022
devtax	(1)	(2)	(3)	(4)	(5)	(6)
OLBI	0.1019*	0.1012*	0.0963	0.096
OLBI	(1.710)	(1.699)	(1.629)	(1.624)
Low OLBI					-0.2687
Low OLBI					(-0.760)
High OLBI						-0.0057
High OLBI						(-0.0506)
Post-election	-	-	-	-	-	-
Postelection$\bullet$OLBI	-	-	-	-
Postelection$\bullet$Low OLBI					-
Postelection$\bullet$High OLBI						-
Balance pc			-0.00008**	-0.00008**	-0.0007**	-0.00006*
Balance pc			(-2.506)	(-2.387)	(-2.415)	(-1.908)
Income pc			-0.0346	-0.0354	0.3075*	-0.0464*
Income pc			(-1.495)	(-1.525)	(1.895)	(-1.838)
Population			0.00002	0.00002	0.0043	0.00002
Population			(1.241)	(1.251)	(1.568)	(0.588)
Ideology		0.2893***		0.2535**	1.956	0.2487
Ideology		(2.678)		(2.378)	(0.680)	(1.334)
Strength		-0.0008		-0.0008	-0.0302	-0.0005
Strength		(-0.203)		(-0.207)	(-0.985)	(-0.123)
Constant	-1.014***	-1.048***	-0.1197	-0.125	-19.709*	0.7828
Constant	(-4.341)	(-3.664)	(-0.155)	(-0.157)	(-1.884)	(0.790)
2022.year	0.3726***	0.3860***	0.6133***	0.6275***	0.6327	0.6084***
2022.year	(5.236)	(5.277)	(3.491)	(3.547)	(0.826)	(3.989)
Observations	1,107	1,107	1,107	1,107	208	899
R-squared	0.113	0.117	0.132	0.135	0.393	525
Number of id	555	555	555	555	179	0.093
Note: Robust t-statistics in parentheses. p < 0.01, p < 0.05, * p < 0.1. Source: Author.*

Models (1)-(4) in Table 7 show a positive and significant relation between OLBT and tax revenue error. In post-election years, LGUs with higher OLBT overestimated tax revenues more.

Table 7

Results for errors in post-election years, 2016–2022
errortax	(1)	(2)	(3)	(4)	(5)	(6)
OLBI	0.0837***	0.0840***	0.0808***	0.0813***
OLBI	(2.726)	(2.734)	(2.647)	(2.660)
Low OLBI					-0.0168
Low OLBI					(-0.137)
High OLBI						0.0357
High OLBI						(0.615)
Post-election	-	-	-	-	-	-
Postelection$\bullet$OLBI	-	-	-	-
Postelection$\bullet$Low OLBI					-
Postelection$\bullet$High OLBI						-
Balance pc			-0.00004***	-0.00004***	-0.000214*	-0.00004**
Balance pc			(-3.013)	(-2.979)	(-2.054)	(-2.467)
Income pc			-0.0166*	-0.0167*	0.0820	-0.0256*
Income pc			(-1.903)	(-1.885)	(1.456)	(-1.964)
Population			0.000008	0.000008	0.00107	0.000006
Population			(1.112)	(1.159)	(1.114)	(0.445)
Ideology		0.0692		0.0482	0.813	0.0667
Ideology		(0.663)		(0.472)	(0.815)	(0.694)
Strength		0.00205		0.00205	-0.00515	0.00260
Strength		(1.192)		(1.195)	(-0.484)	(1.181)
Constant	0.462***	0.356**	0.908***	0.810***	-4.344	1.274**
Constant	(3.689)	(2.361)	(3.227)	(2.752)	(-1.196)	(2.494)
2022.year	0.174***	0.173***	0.292***	0.290***	0.219	0.318***
	(5.764)	(5.553)	(4.889)	(4.726)	(0.826)	(4.052)
Observations	1,107	1,107	1,107	1,107	208	899
R-squared	0.130	0.132	0.149	0.151	0.332	0.108
Number of id	555	555	555	555	179	525
Note: Robust t-statistics in parentheses. p < 0.01, p < 0.05, * p < 0.1. Source: Author.*

Tables A4 and A5 present the results for LGUs with low and high OLBT. The results in Table A4 support what was previously stated. In pre-election years, LGUs with higher OLBT underestimated tax revenues more. In contrast, in election and post-election years, LGUs with higher OLBT underestimated tax revenues less; i.e., collection of tax revenues was closer to planned values. The results in Table A5 support what was previously mentioned for LGUs with high OLBT. In pre-election years, LGUs with higher OLBT underestimated tax revenues more. In contrast, in election and post-election years, LGUs with higher OLBT underestimated tax revenues less; i.e., collection of tax revenues was closer to planned values. Additionally, LGUs with lower OLBT underestimated tax revenues less, and the collection of tax revenues was closer to the planned values.

To conclude, it is shown that in LGUs with growing surplus pc, the underestimation of tax revenue is higher (excessively conservative planning) and overestimation is lower (collected tax revenues are closer to planned values), as in (Chatagny & Soguel, 2012). Additionally, it is shown that in LGUs with growing income pc, underestimation of tax revenues is higher and overestimation is lower, as in Bohn & Veiga (2021). Furthermore, in LGUs with a growing population, the underestimation of tax revenues is lower (collected tax revenues are closer to planned values) and overestimation is lower (excessively optimistic planning), in line with Lago-Peñas & Lago-Peñas (2004).

In conclusion, the impact of the OLBT on tax revenue forecasting is revealed in pre-election and post-election years. In pre-election years, LGUs with higher OLBT underestimate tax revenues more and overestimate tax revenues less. In post-election years, LGUs with higher OLBT overestimate tax revenues more. Moreover, the results prove that in election and post-election years, LGUs’ underestimation of tax revenues is lower; i.e., collected tax revenues are closer to planned values, and overestimation of tax revenues is higher. In the pre-election years, LGUs’ overestimation of tax revenues is lower; i.e., collection of tax revenues is closer to planned values, and underestimation of tax revenues is higher.

The paper uses data from all 556 Croatian cities and municipalities during 2016–2022 to analyse the impact of OLBT on budget forecasting deviations and errors. The results indicate that higher levels of OLBT entail higher underestimation and lower overestimation of tax revenues in pre-election years. It is better for LGUs to underestimate revenue because planned expenditures are higher in the case of overestimation, which leads to higher deficits and debt. Moreover, the results indicate that higher levels of OLBI imply a lower underestimation and a higher overestimation of tax revenues in post-election years. This is in line with the finding of Bischoff & Gohout (2010) that governments often overestimate tax revenue forecasts, especially in election years. They argue that incumbent politicians may overestimate revenues to raise expenditures, which leads to increasing the debt and their chances of being re-elected. Our results are in line with those of Lago-Peñas & Lago-Peñas (2004), who demonstrated the overestimation of tax revenues in election years, while we showed that Croatian LGUs had the same relation in election and post-election years.

Regarding the results of the PBC, dummy variables of pre-election, election and post-election years were used, as were some interaction variables described above. PBCs in Croatian LGUs regarding tax revenue forecasting are revealed. In pre-election years, they exhibit a higher underestimation and a lower overestimation of tax revenues. PBCs also indicate a lower underestimation and a higher overestimation of tax revenues in election and post-election years. The aforementioned findings suggest that there is no power in local councils, citizens (voters), or national revisor agencies to stop Croatian LGUs from providing inaccuracies in tax revenue forecasting. Budget transparency is good, but citizen participation is poor because the public and legislative debate on the budget process is predicated on faulty inputs in the crucial process of tax revenue plans.

What are the policy implications of these results? In Croatian LGUs, budget transparency is good, considering the availability of key budget documents on official websites. In contrast, there is not enough citizen participation in the budget process. This could be improved with simple institutional reforms. Participating citizens and media can compare tax revenue forecasts with those of other LGUs. LGUs must explain any deviation publicly, mainly if it is higher than tax revenue forecasting. This also improves the efficiency of the state auditing ex-post control. According to Alt and Lassen (2006a), these actions are crucial for maintaining discipline and promoting budget transparency. From a broader standpoint, the results emphasise the significance of having independent tax revenue forecasts. Assigning the duty of tax revenue forecasting to an impartial organisation and making their projections a mandatory component of the budget process is a practical means of guaranteeing this. Inaccurate revenue tax forecasts are still far too commonly generated at present, as tax projections are a crucial factor in deficit estimates.

One limitation of this research might arise from the OLBT measurement, the OLBI, which depends on the number of key budget documents published on the official websites of LGUs and ranges from 0 to 5. Future studies could examine the impact of qualitative OLBT measures on tax revenue deviations. A second limitation concerns the methodology’s ability to demonstrate the correlation or relationship between the variables without showing a causal relationship. Future research could use different methods to examine the impact of OLBT on tax revenue forecasting. It would also be interesting to see comparable research on a different country with distinct fiscal regulations and variables.

Aguiar-Conraria, L., Magalhães, P. C., & Veiga, F. J. (2019). Transparency, Policy Outcomes, and Incumbent Support. Kyklos, 72(3), 357–380. https://doi.org/https://doi.org/10.1111/kykl.12203
Alt, J. E., & Lassen, D. D. (2006a). Fiscal transparency, political parties, and debt in OECD countries. European Economic Review, 50(6), 1403–1439. https://doi.org/10.1016/j.euroecorev.2005.04.001
Alt, J. E., & Lassen, D. D. (2006b). Transparency, Political Polarisation, and Political Budget Cycles in OECD Countries. American Journal of Political Science, 50(3), 530–550. https://doi.org/10.1111/j.1540-5907.2006.00200.x
Arbatli, E., & Escolano, J. (2015). Fiscal Transparency, Fiscal Performance and Credit Ratings. Fiscal Studies, 36(2), 237–270. https://doi.org/10.1111/1475-5890.12051
Baltagi, B. H. (2013). Econometric Analysis of Panel Data. 5th ed. Chichester, UK: Wiley.
Benito, B., Guillamón, M. D., & Bastida, F. (2015). Budget Forecast Deviations in Municipal Governments: Determinants and Implications. Australian Accounting Review, 25(1), 45–70. https://doi.org/10.1111/auar.12071
Bischoff, I., & Gohout, W. (2006). Tax projections in German states – manipulated by opportunistic incumbent parties? 74.
Bischoff, I., & Gohout, W. (2010). The political economy of tax projections. International Tax and Public Finance, 17(2), 133–150. https://doi.org/10.1007/s10797-008-9103-y
Bohn, F. (2018). Political cycles: Beyond rational expectations. PLoS ONE, 13(10), 1–23. https://doi.org/10.1371/journal.pone.0203390
Bohn, F., & Veiga, F. J. (2021a). Political forecast cycles. European Journal of Political Economy, 66(April 2020), 101934. https://doi.org/10.1016/j.ejpoleco.2020.101934
Bohn, F., & Veiga, F. J. (2021b). Political forecast cycles. European Journal of Political Economy, 66(August 2020), 101934. https://doi.org/10.1016/j.ejpoleco.2020.101934
Bronić, Mihaela;, Ott, K., Petrušić, M., Prijaković, S., & Stanić, B. (2023). Budget transparency in Croatian Counties, Cities and Municipalities: November 2022 - April 2023. Institute of Public Finance Notes, 1–16. https://doi.org/10.1787/9789264227118-8-en
Bronić, Mihaela, Stanić, B., & Prijaković, S. (2022). The Effects of Budget Transparency on the Budget Balances and Expenditures of Croatian Local Governments. South East European Journal of Economics and Business, 17(1), 111–124. https://doi.org/10.2478/jeb-2022-0008
Budget Act (2022): Zakon o proračunu. Retrived from: https://www.zakon.hr/z/283/Zakon-o-prora%C4%8Dunu
Chatagny, F., & Soguel, N. C. (2012). The effect of tax revenue budgeting errors on fiscal balance: Evidence from the Swiss cantons. International Tax and Public Finance, 19(3), 319–337. https://doi.org/10.1007/s10797-011-9189-5
Croatian Bureau of Statistics – CBS (2022): Procjena stanovništva za 2017.-2020. godinu, Popis stanovništva za 2021. godinu, Retrieved from: https://podaci.dzs.hr/hr/podaci/stanovnistvo
Crombach, L., & Bohn, F. (2022). Uninformed voters with (im)precise expectations: Explaining political budget cycle puzzles. Economics and Politics, May. https://doi.org/10.1111/ecpo.12237
Cuadrado-Ballesteros, B., & Bisogno, M. (2022). Budget transparency and financial sustainability. Journal of Public Budgeting, Accounting and Financial Management, 34(6), 210–234. https://doi.org/10.1108/JPBAFM-02-2022-0025
Cuadrado-Ballesteros, B., Guillamón, M. D., & Ríos, A. M. (2021). Does gender matter in budget deviations? An empirical assessment of Spanish local governments. Public Management Review, 00(00), 1–24. https://doi.org/10.1080/14719037.2021.1912816
Drazen, A. (2008). Political Budget Cycles. U The New Palgrave Dictionary of Economics (str. 1–8). https://doi.org/10.1057/978-1-349-95121-5
Glennerster, R., & Shin, Y. (2008). Does Transparency Pay ? IMF Staff Papers, 55(1), 183–209. https://doi.org/10.1057/palgrave.imfsp.9450028
Goeminne, S., Geys, B., & Smolders, C. (2008). Political fragmentation and projected tax revenues: Evidence from Flemish municipalities. International Tax and Public Finance, 15(3), 297–315. https://doi.org/10.1007/s10797-007-9021-4
Institute of Public Finance. (2017). Local budget transparency (November 2016 - March 2017), Retrieved from: https://www.ijf.hr/en/transparency-2017
Institute of Public Finance. (2023). Local budget transparency (November 2022 – April 2023), Retrieved from: https://www.ijf.hr/en/transparency-2023
Jochimsen, B., & Lehmann, R. (2017). On the political economy of national tax revenue forecasts: evidence from OECD countries. Public Choice, 170(3–4), 211–230. https://doi.org/10.1007/s11127-016-0391-y
Lago-peñas, I., & Lago-peñas, S. (2004). Explaining Budgetary Indiscipline: Evidence From Spanish Municipalities. 21(04), 1–43.
Mačkić, V. (2021). Fiscal Conservatism and Re-Election Prospects: It Is the Same Principle, the Rest Is Just Details. 10th International scientific symposium Region, entrepreneurship, development, 189–203. https://www.webofscience.com/wos/woscc/full-record/WOS:000728594100013
Mayper, A. G., Granof, M., & Giroux, G. (1991). An Analysis of Municipal Budget Variances. Accounting, Auditing & Accountability Journal, 4(1), 29–50.
Ministry of Finance – MoF (2022): Financijski izvještaji JLP(R)S: PR-RAS i RAS-funkc za razdoblje 2014 – 2020, Retrieved from: https://mfin.gov.hr/istaknute-teme/lokalna-samouprava/financijski-izvjestaji-jlp-r-s/203
Ministry of Regional Development and EU Funds – MRRFEU (2022): Data for income pc, Retrieved from: https://razvoj.gov.hr/o-ministarstvu/regionalni- razvoj/indeks-razvijenosti/112
Montes, G. C., Bastos, J. C. A., & de Oliveira, A. J. (2019). Fiscal transparency, government effectiveness and government spending efficiency: Some international evidence based on panel data approach. Economic Modelling, 79, 211–225. https://doi.org/10.1016/j.econmod.2018.10.013
Prijaković, S. (2023). Proračunska transparentnost i dug gradova u Republici Hrvatskoj. Ekonomska misao i praksa, 1–29.
Prijaković, S., Mačkić, V., & Bronić, M. (2023). Budget transparency and budget credibility: The case of local government units in Pannonian Croatia. 12th international scientific symposium region, entrepreneurship, development, 165–180.
Ríos, A. M., Guillamón, M. D., Benito, B., & Bastida, F. (2018). The influence of transparency on budget forecast deviations in municipal governments. Journal of Forecasting, 37(4), 457–474. https://doi.org/10.1002/for.2513
Rogoff, K. (1990). Equilibrium Political Budget Cycles. The American Economic Review, 80(1), 21–36.
Rogoff, K., & Sibert, A. (1988). Elections and Macroeconomic Policy Cycles. The Review of Economic Studies, 55(1), 1–16.
Shi, M., & Svensson, J. (2006). Political budget cycles: Do they differ across countries and why? Journal of Public Economics, 90(8–9), 1367–1389. https://doi.org/10.1016/j.jpubeco.2005.09.009
State Electoral Commission – SEC (2022): Rezultati dosadašnjih lokalnih izbora 2013., 2017. i 2021. godine, Retrieved from: https://www.izbori.hr/site/UserDocsImages/2266
The Law on Local Taxes (2023). Zakon o lokalnim porezima. Retrieved from: https://www.zakon.hr/z/875/Zakon-o-lokalnim-porezima
van der Ploeg, F. (2010). Political economy of prudent budgetary policy. International Tax and Public Finance, 17(3), 295–314. https://doi.org/10.1007/s10797-009-9117-0
Vicente, C., Benito, B., & Bastida, F. (2013). Transparency and political budget cycles at municipal level. Swiss Political Science Review, 19(2), 139–156. https://doi.org/10.1111/spsr.12036
WB. (2022). Fiscal Decentralisation in Croatia. U Fiscal Decentralization in Croatia. https://doi.org/10.1596/38261
Williams, D. W., & Calabrese, T. D. (2016). The status of budget forecasting. Journal of Public and Nonprofit Affairs, 2(2), 127–160. https://doi.org/10.20899/jpna.2.2.127-160

Graphs are available in the Supplementary Files section.

No competing interests reported.

Appendix.docx
Graph1.png
Graph 1. OLBI, all 556 LGUs, 2016-2022, number of online published budget documents Source: Author’s calculations based on data from Bronić et al. (2022).
Graph2.png
Graph 2. Tax revenue deviation, all 556 LGUs, 2016-2022, in % Source: Author
Graph3.png
Graph 3. Tax revenue error, all 556 LGUs, 2016-2022, in % Source: Author

Download PDF

Version 1

posted

You are reading this latest preprint version

The impact of budget transparency on tax revenue forecasts: Evidence from Croatian local government units

Status:

Version 1

Abstract

Figures

1. Introduction

2. Literature review and theory

3. Data and methodology

4. Results

5. Conclusion

References

Graphs

Additional Declarations

Supplementary Files

Status:

Version 1