Political drivers of accounting discretion in local governments: evidence from Spain

Over the last years, there has been a growing global trend among many countries to adopt accrual accounting systems in public sector organizations as a means to improve financial management and implement standardized international public sector financial systems (Brusca et al., 2015; Hyndman and Connolly, 2011). Unlike cash accounting, accrual accounting involves a greater subjectivity since it heavily relies on professional judgment, which provides substantial discretion in recognizing revenues and expenses, and reporting assets and liabilities. Consequently, in a highly political environment, this discretion inevitably increases the risk of opportunistic behavior by politicians and public administrators, who face significant pressure from stakeholders to be accountable for public resources (Cohen et al., 2019; Pilcher and Van der Zahn, 2010).

The manipulation of financial information, known as earnings management (EM) in academic literature, refers to the deliberate adjustment of financial variables to mislead stakeholders or influence transactions dependent on reported accounting and financial information. This practice involves the use of accounting techniques to present a desired financial position, which may not accurately reflect the underlying economic reality (Schipper, 1989; Healy and Wahlen, 1999). While EM has been a common practice for private sector firms, its application has been increasingly observed in the public sector, raising significant concerns about the reliability and integrity of financial reporting in public entities (Cohen et al., 2019). Indeed, the study of EM has recently become a subject of interest for public sector scholars (Bisogno and Donatella, 2022). Focusing on the particular context of local governments, empirical evidence has shown that public managers may use accrual-based accounting variables to manipulate reported financial outcomes, often driven by political motivations, such as electoral cycles and political competition (Cohen et al., 2019; Ferreira et al., 2020; Ferreira, 2024).

Unlike the private sector, where financial decisions are primarily driven by profit maximization, public sector financial reporting pursues different objectives. The discretion permitted by accounting standards allows politicians to strategically adjust financial figures to align with their performance objectives (Arcas and Martı́, 2016). Accruals close to zero suggest limited use of accounting discretion, as this reflects a closer alignment between accounting results and actual cash flows. In this context, the magnitude of accruals, particularly deviations from expected or “normal” levels, serves as a useful proxy for discretionary reporting choices. In local governments, reporting high surpluses may suggest excessive taxation relative to service quality or an inappropriate level of transfers from higher authorities, while deficits could indicate poor financial management and failure to achieve targets, attracting criticism from political opponents and scrutiny from oversight authorities (Cohen et al., 2019; Bisogno and Donatella, 2022). These dynamics create incentives for politicians to manipulate financial reporting in ways that serve their personal and political interests. Public choice theory (PCT) (Buchanan and Tullock, 1962) offers a relevant framework for understanding such behaviors, as it suggests that politicians act in self-interest, prioritizing re-election over objective financial reporting.

Based on PCT, this study examines the influence of political factors in shaping financial reporting quality at the municipal level, with a focus on EM practices in Spanish local governments. Specifically, we study how political factors, referring to the political party of the governing body and the mayor, triggered opportunistic behaviors. To do so, we employ a novel one-stage modified Jones (Dechow et al., 1995) model that integrates both accounting and political variables into a single regression. This empirical approach addresses the limitations of traditional two-stage approaches, which often produce biased coefficient estimates and nonrobust standard errors (Chen et al., 2018). Moreover, as some of the variables we are going to use are time-invariant, following Mundlak (1978), we define a specific treatment that allows the decomposition of between-units and within-units effects in the analysis. This ensures a more nuanced understanding of how stable municipality-specific characteristics (e.g. political power or alignment) and dynamic intra-municipal political factors (e.g. mayoral wages) influence financial reporting practices, an issue that is highly relevant in the analysis of political cycles.

We use a balanced panel dataset of 3,190 Spanish municipalities for the 2016 to 2018 period. The results reveal significant EM in net profit distribution, influenced by political and individual-level factors. Higher political power and political alignment between municipal and regional governments foster financial manipulation, indicating that reduced scrutiny and low external oversight in politically concentrated environments, along with shared political agendas, may incentivize opportunistic financial practices. In terms of mayoral characteristics, higher salaries appear to discourage manipulation, whereas full-time dedication is linked to increased discretionary accruals. However, the mediation between these two variables shows a positive between (stable) effect and a negative within (transient) effect, meaning that well-paid mayors who do not have full-time responsibilities tend to manipulate less, but this effect decreases with full-time contracts. Interestingly, gender and political signs do not significantly influence financial manipulation.

The present study further deepens the understanding of EM in the public sector and provides three key contributions. First, while previous research on EM and political factors has mainly focused on political competition and election years (see, for instance, Cohen et al., 2019; Donatella, 2020; Ferreira et al., 2013, 2020), this study broadens the scope by examining both the characteristics of the governing political party and those of individual politicians in power. Second, from a methodological perspective, it enhances the measurement of EM in the public sector by integrating a one-stage model that accounts for the inclusion of time-invariant variables, improving the reliability of results. Finally, to our knowledge, this is the first study to analyze EM in the Spanish public sector, a particularly relevant context due to its decentralized structure (Balaguer-Coll et al., 2010; Carrillo, 1997), politically fragmented governance, and some cases of corruption reported in the last two decades (Quesada et al., 2013). Additionally, the 2010 adoption of accrual accounting in the Spanish public sector under the International Public Sector Accounting Standards (IPSAS) underscores the relevance of this study.

The article is organized as follows. After this introduction, section 2 gives an overview of the previous literature on EM at the local level, while section 3 provides the theoretical framework on which we focus and develop the hypotheses. Section 4.2 establishes the institutional framework of Spanish municipalities. Section 4 provides a detailed description of the sample, data and the empirical strategy proposed to measure EM in our particular context; and section 5 presents and comments on the main results. Finally, section 6 outlines the most important conclusions and implications as well as avenues for future research.

As mentioned in the introduction, the topic of EM has traditionally been a common practice among private sector firms. However, in recent years, it has attracted considerable attention from scholars in the public sector (see Bisogno and Donatella, 2022, for a comprehensive literature review on EM in public sector organizations). In particular, within the context of local governments, although the literature remains relatively limited, empirical studies have increased notably over the past two decades, providing insights from various perspectives and multiple countries. For instance, EM behavior at the local level has been examined in Australia (Pinnuck and Potter, 2009; Pilcher and Van der Zahn, 2010; Pilcher, 2011; Drew, 2018), the United States (Beck, 2018), the United Kingdom (Arcas and Martı́, 2016), Greece (Cohen et al., 2019; Cohen and Malkogianni, 2021; Malkogianni, 2024), Italy (Anessi-Pessina and Sicilia, 2020), Portugal (Ferreira et al., 2013, 2020; Ferreira, 2023, 2024), Sweden (Stalebrink, 2007; Donatella et al., 2019, 2024; Donatella, 2020; Donatella and Tagesson, 2021) and Norway (Haugdal et al., 2024). Despite geographical diversity, these studies generally converge on the idea that local governments engage in EM practices.

Regarding the research methodologies used for detecting EM at the local government level, the most common approach is the use of discretionary accruals as a proxy, frequently employing the Jones model (Jones, 1991) or its modified versions (e.g. Dechow et al., 1995). This methodology has been widely applied in studies such as those by Beck (2018), Cohen et al. (2019), Cohen and Malkogianni (2021), Ferreira et al. (2013, 2020), Ferreira (2023, 2024), Donatella et al. (2024), Haugdal et al. (2024) and Malkogianni (2024). Beyond total accruals, other studies have proposed alternative models or focused on specific accruals, such as depreciation or impairment (Stalebrink, 2007; Pilcher and Van der Zahn, 2010; Pilcher, 2011; Arcas and Martı́, 2016; Drew, 2018; Donatella et al., 2019; Donatella, 2020). Furthermore, some studies have examined revenue manipulation in local governments, highlighting how financial outcomes can be influenced by revenue adjustments (Anessi-Pessina and Sicilia, 2020; Donatella and Tagesson, 2021).

Apart from measuring EM, scholars have investigated its key drivers, often employing two-stage models in which discretionary accruals are first estimated and then regressed against potential explanatory variables. In the context of local governments, some studies highlight political factors as key drivers of EM practices, particularly in relation to competition and election cycles, where financial figures are adjusted to present a more favorable fiscal position before elections. Ferreira et al. (2013, 2020) and Ferreira (2024) found that mayors use discretionary accruals to report earnings close to zero, especially under strong political competition or when seeking re-election. Similarly, Cohen et al. (2019) and Cohen and Malkogianni (2021) showed that earnings manipulation intensifies during electoral cycles, particularly when incumbents seek re-election, reinforcing the idea that politicians adjust financial reports to enhance their electoral prospects. Moreover, Donatella (2020) highlighted that political competition increases the likelihood of accrual-based reporting discretion, but only when there is a majority government, where weaker institutional oversight facilitates such practices. Additionally, Malkogianni (2024) found that both pre-election periods and weak political opposition (i.e. concentrated political power) significantly influence EM in local governments, particularly among municipalities that are more dependent on subsidies.

Finally, recent studies have also explored how the individual characteristics of bureaucrats influence financial reporting choices, adding a micro-level perspective to the broader political dynamics at play. Donatella and Tagesson (2021) found that bureaucrats with longer tenure and prior experience are less likely to engage in opportunistic accounting choices, suggesting experience reduces such behavior. Moreover, Anessi-Pessina and Sicilia (2020) further highlighted that gender and educational background influence financial misrepresentation, and external oversight mechanisms, such as auditors and political opposition, help limit opportunistic financial reporting.

In this article, we use the PCT framework to understand how political dynamics can incentivize the manipulation of financial reporting in local governments. PCT (Buchanan and Tullock, 1962) applies economic principles to political decision-making, assuming that individuals in the public sector behave rationally and are primarily guided by their own self-interest when participating in collective decisions. This self-interested behavior manifests in various forms, including vote-maximization strategies for re-election, rent-seeking (Krueger, 1974), and the manipulation of public resources for electoral or career-related gains (Mueller, 1976).

Building on this theoretical foundation, Boyne (1997) applies this perspective to the context of local governments, using a market-based analogy in which politicians and public officials are viewed as “sellers” of public services, while citizens and interest groups act as “buyers.” He also points out two key assumptions underlying PCT in local governments. The first concerns the “self-interest axiom,” under which politicians will engage in strategic behavior to maximize self-interest over the broader public interest. The second concerns the “pressure of the competition,” such as the fragmentation of political power, which incentives politicians to act in the public interest instead of their own. Therefore, conflicting interests are key conditions that can either drive or constrain the manipulation of financial data in local governments.

PCT offers valuable insights into financial discretion in local governments, where officials may manipulate financial reports to secure political advantages or obtain personal benefits such as higher salaries or career advancement. Scholars have empirically applied this framework to investigate how political incentives influence financial reporting choices (Stalebrink, 2007; Cohen et al., 2019; Ferreira et al., 2020; Cohen and Malkogianni, 2021; Ferreira, 2024; Malkogianni, 2024). Therefore, based on this theoretical framework, we formulate our hypotheses in the following paragraphs.

PCT suggests that when political power is highly concentrated, oversight mechanisms weaken (Buchanan and Tullock, 1962), allowing greater discretion in financial reporting and fostering opportunistic behavior. Boyne (1997) argues that in environments with highly concentrated power, politicians behave like monopolistic actors, reducing transparency to maintain control over public resources.

While prior research has predominantly examined political competition (typically understood as the intensity of electoral rivalry and the chances of re-election), this study focuses instead on political power concentration, defined as the extent to which political authority is centralized in a single party or cohesive governing coalition in a legislative term. The mechanisms underlying financial manipulation differ substantially across these two dimensions. In competitive contexts with intense electoral pressure, manipulation is often motivated by electoral incentives, where politicians seek to improve reported performance in order to enhance their probabilities of being re-elected (Ferreira et al., 2013, 2020; Donatella, 2020; Cohen et al., 2019; Ferreira, 2024). In contrast, when political power is highly concentrated, external oversight tends to weaken, increasing the scope for opportunistic behavior. Indeed, Donatella (2020) finds that the positive effect of political competition on the use of accrual-based discretion is more pronounced under majority governments, where institutional oversight tends to be weaker. Moreover, Malkogianni (2024) confirms that weak opposition (i.e. strong political power) in the municipal council is aligned with more extensive EM. In such contexts, political strength facilitates financial misreporting, reflecting PCT's core axiom that weak oversight shapes politicians' incentives to manipulate financial reporting.

Considering these arguments, the hypothesis is stated as follows:

PCT suggests that politicians act in their own interests (Buchanan and Tullock, 1962). In low-oversight environments, this self-interest may lead to greater fiscal discretion, potentially facilitating financial misreporting in vertically concentrated structures (Boyne, 1997), such as when local governments share the same political affiliation as the regional or central government. Empirical evidence supports this claim. In the private sector, partisan affinity has been shown to influence financial disclosures by increasing optimism in earnings forecasts and reducing accounting conservatism (Arikan et al., 2023). Similarly, Gross (2022) find that firms headquartered in politically aligned states engage in greater EM. Moreover, political alignment is associated with higher corporate fraud rates, primarily due to weaker regulatory oversight (Cordis, 2024).

In the public sector, although there are no previous studies that directly explored the link between political alignment and EM behavior, it is reasonable to hypothesize that political alignment, by potentially weakening or reducing oversight, may facilitate EM in local governments. This proposition is consistent with evidence from multi-level governance literature. For instance, Solé-Ollé and Sorribas-Navarro (2008) and Bracco et al. (2015) show that political alignment across government tiers systematically influences the allocation of intergovernmental transfers. In such contexts, aligned municipalities may benefit from preferential treatment and less stringent external scrutiny. These dynamics create conditions under which EM is more likely to occur.

Drawing on these arguments, the hypothesis is formulated as follows:

PCT posits that self-interested politicians respond to personal incentives, and the nature of their compensation may influence their behavior (Boyne, 1997; Mueller, 1976). Mayors who receive higher wages or serve in full-time roles may have stronger incentives to sustain their political careers, potentially increasing their engagement in earnings manipulation to signal financial stability or political success. Conversely, where political positions offer limited economic benefits, the motivation for financial distortion may be lower, as the personal gains are reduced.

Although no previous study has directly analyzed the relationship between mayoral employment conditions and earnings manipulation in the public sector, related research supports the hypothesis that such conditions influence financial reporting practices. Benito et al. (2021) find that full-time dedication among mayors is associated with increased service efficiency, indicating that employment conditions affect managerial behavior. However, the study also highlights that these same factors, in the absence of strong institutional oversight, can foster opportunistic practices. Moreover, Pique (2019) finds that higher mayoral wages and full-time dedication do not consistently lead to better financial outcomes, suggesting that the effect of employment conditions on earnings manipulation may depend on governance context.

From the private sector, we find clearer evidence supporting the relationship between employment conditions and earnings manipulation. For example, Bergstresser and Philippon (2006), Burns and Kedia (2006), Duong and Evans (2016) and Laux and Laux (2009) show that managers' employment conditions, such as performance-based compensation, are linked to a higher likelihood of earnings manipulation, underscoring the role of financial incentives in shaping reporting behavior. These studies collectively suggest that when professional success is tied to financial figures and when short-term performance metrics are emphasized, individuals are more likely to manipulate financial reports. Li et al. (2016) further contribute to this understanding by showing that higher compensation thresholds tied to performance also increase the likelihood of opportunistic reporting behavior. Drawing on these parallels that can be applied to the public sector context, we hypothesize that better-paid and fully dedicated mayors have stronger incentives to manipulate financial reporting in local governments, as these factors align with personal incentives.

In light of these considerations, the following hypothesis is proposed:

This study examines a sample of Spanish municipalities with populations exceeding 5,000 inhabitants during the period 2016–2018. All years analyzed are within the same electoral cycle and exclude local election years, which took place in 2015 and 2019, respectively. The analysis period concludes before the onset of the COVID-19 pandemic to avoid potential biases related to the economic and financial disruptions caused by the health crisis. The final sample is a balanced panel data set comprising 3,190 municipalities, representing approximately 39.2% of the total number of municipalities in Spain, after excluding observations with missing data for any year within the 3-year period.

The data used in this study to measure EM behavior were sourced from the Financial Statements of Spanish municipalities, including the Balance Sheet, Profit and Loss Statement and Cash Flow Statement, accessed through the Public Accountability website, overseen by the Spanish Court of Audit (Tribunal de Cuentas) and the Regional Audit Institutions. In our search for political factors, we use data provided by various institutions. Information on mayors' salaries and their dedication regime is obtained from the ISPA statistics published by the Spanish Ministry for Digital Transformation and Public Function (Ministerio para la Transformación Digital y de la Función Pública). Additionally, data to compute the variables on political power, political alignment, ideological position and mayor gender are sourced from the Spanish Ministry of the Interior (Ministerio del Interior), specifically from the 2015 local elections. Meanwhile, information on mayors' gender is also obtained from the Local Information System (SIL) published by the Spanish Ministry of Territorial Policy and Democratic Memory (Ministerio de Política Territorial y Memoria Democrática).

The Spanish territorial organization consists of three levels established by the 1978 Constitution: the State, Autonomous Communities (or Regions), and Local Government, with municipalities representing the primary tier of local governance ^[1]. Considering 2024 data, there were 8,132 municipalities, marked by significant heterogeneity in population, economic activity and geographic location. Over 90% of these municipalities had fewer than 20,000 inhabitants, and 4,994 had populations below 1,000, accounting for only 12% of the national population. This demographic structure presents challenges for service provision, financial sustainability and governance capacity.

The core competencies of municipalities are outlined in the Spanish law regulating the local system (Ley 7/1985, Reguladora de las Bases de Régimen Local). Depending on their size, municipalities must provide certain basic minimum services, such as public lighting, waste collection and street maintenance, and, for those with populations exceeding 5,000 or 20,000, additional services like public libraries, parks and/or public transport. Municipal revenue is derived from a combination of own-source revenues, including local taxes such as the Property Tax, business activity taxes and fees for services, as well as intergovernmental transfers from both regional and national governments.

In terms of accounting, accrual accounting was adopted in the Spanish public sector in 2010 under the International Public Sector Accounting Standards (IPSAS) as part of the European Union's efforts toward accounting harmonization, aimed at enhancing transparency and comparability in public sector financial information (Brusca et al., 2021). The Spanish General Public Accounting Plan (Plan General de Contabilidad Pública), adapted for local administration, requires municipalities to follow an accrual-based system to issue annual financial statements, such as the balance sheet and income statement, in accordance with IPSAS. However, budget accounting at the local level follows a modified cash basis. It is regulated by the Local Finance Law (RDL 2/2004, de Haciendas Locales) and the Budgetary Stability and Financial Sustainability Law (Ley 2/2012, de Estabilidad Presupuestaria y Sostenibilidad Financiera), which establish rules for revenue and expenditure management to ensure fiscal discipline, budget stability and financial sustainability.

Finally, municipal elections in Spain are held every four years to determine the composition of municipal councils, which are responsible for electing the mayor – the head of local government. The mayor is elected by an absolute majority vote within the council and, if no candidate achieves this majority, the position is automatically assigned to the candidate from the party list with the most votes. As the highest authority in local governance, the mayor plays a crucial role in managing municipal services, implementing council policies and representing the municipality.

Accruals models are used to measure managers' discretion in accounting information as they are prone to opportunistic behavior (Cohen et al., 2019). According to Dechow et al. (2010), the most common models for the estimation of discretionary accruals in previous literature are the aggregate Jones (1991) model and the modified Jones (Dechow et al., 1995) model, being this last one an extension that adjusts for the growth in credit sales. These models define the normal accrual process (related to working capital and depreciation) as a function of the growth of sales and the investment in Property, Plant and Equipment (PP&E). In doing so, they assume that the residual represents the nondiscretionary component of total accruals, which can be used in a second step as a proxy of the manipulation of accounting information through the reporting of earnings with poor quality.

To estimate earnings manipulation using accruals models, standard models (see Exhibit 1 from Dechow et al., 2010) define the measurement of Total Accruals (⁠$TACCi,t$⁠) calculated as Net earnings minus Operational Cash Flows:

According to the modified Jones model, the estimation of discretionary accruals requires, as a first step, the estimation of the following regression model:

where $TACCi,t$ are the total accruals in year $t$ for municipality $i$⁠; $ΔREVi,t$ are the growth in revenues in year $t$⁠, $ΔRECi,t$ are the growth in receivables in year $t$ for municipality $i$ and the difference between them represents the growth in sales collected; $PPEi,t$ is the gross property, plant and equipment in year $t$ for municipality $i$⁠; and $TAi,t−1$ are the lagged total assets. After this, the nondiscretionary accruals (⁠$NDAi,t$⁠) are the predicted level of model 3 for each unit, while the discretionary accruals (⁠$DAi,t$⁠) are computed by deducting the predicted nondiscretionary accruals from the total accruals:

In the second stage, discretionary accruals (used as a proxy for the manipulation of accounting information through the EM) become the new dependent variable; hence, a model including potential variables that are expected to have an impact on the level of discretionary accruals $DAi,t$ are considered.

However, despite there exist an extensive number of empirical articles – with a massive use in accounting papers after Jones (1991) seminal paper – that define a two-step process in the estimations to decompose (1) a dependent variable into the predicted and residual components and (2) to use the residual of previous step as dependent variable in a second regression, Chen et al. (2018) pointed out that this two-step procedure can provide biased coefficients and nonrobust standard errors, which can produce incorrect inference with Type I and Type II errors.

In order to overcome these estimation issues, in the present article, we employ the one-stage model proposed by Chen et al. (2018). As our final objective is to determine the impact of political factors on EM behavior at the local government level, we include in only one regression the two samples of independent variables (i.e. the ones corresponding to the modified Jones model as well as the others corresponding to the political factors that can increase the manipulation of accounting variables) and obtain the estimations in only one step. From the results, the signs and standard errors related to regressors of the political variables can be considered to test the hypotheses around the real impact of these variables on the discretionary accruals. Specifically, our model includes several political variables related to (1) the political party of the local governing body and (2) the mayor:

where $HIi,t$ is the Herfindahl Index that represents the level of political power, which takes a value between 0 and 1 with higher values indicating a higher political concentration or strength in terms of town councilors per political party; $ALIGNi,t$ is a dummy variable, which takes the value of 1 if the local governing body of the municipality is aligned with the one from the regional government, and 0 otherwise; $RIGHTWINGi,t$ is a dummy variable that measures the municipal governments' ideological position, which takes the value of 1 for right-wing and 0 for left-wing political orientation, respectively; $WAGEi,t$ is the mayors' salary in year t in municipality i; $FULLTIMEi,t$ is a dummy variable for the mayor's dedication, which takes the value of 1 if the mayor has a full-time dedication, and 0 otherwise; and $FEMALEi,t$ is a dummy variable for the mayors' gender, which takes the value of 1 for female mayors and 0 otherwise. Additionally, we also include the interaction $WAGE*FULLTIMEi,t$ to analyze whether the effect of wage is different for full-time mayors.

In the estimation, we also faced an additional issue related to the characteristics of the variables. Indeed, in the specific context we are dealing with we have several variables that are time-invariant (i.e. they relate to the full political cycle under analysis) and biased results can appear on the coefficients corresponding to time-invariant variables. One way to correct this problem is the proposal formalized by Mundlak (1978). The correction proposed consists of redefining the panel data estimation model by adding as new variables the average of the values of the original time-variant variables. This allows to break down the effects of the variables into two components: a specific one that is stable and constant between units (between effect or cross-municipal) and the other that represents the existing changes among the units (within effect or intra-municipal). Adding the two effects, we have the total effect of the variables considered. The adaptation of random effects models to Mundlak (1978) proposal is as follows:

where $1̅TAi$⁠, $ΔREVi−ΔRECi̅TAi$⁠, $PPEi̅TAi$⁠, $WAGEi̅$ and $WAGEi*FULLTIMEi̅$ represent the time-invariant levels (arithmetic means) of the respective variables that allow us to estimate the between effects of these variables.

Table 1 presents the descriptive statistics of the variables used in the EM model (Equation 4). All the accounting variables used are deflated and weighted by the assets of the previous year.

It is observable that, considering the low average value of the index $HI$⁠, there is a strong fragmentation with a concentrated distribution; the inter-quartile range is only 0.1254. The political alignment affects almost half of the municipalities (44.44%), local governments oriented toward right-wing positions are around one third of the sample (32.57%). Almost two-thirds of the mayors declare a full-time dedication (62.33%) and, finally, female mayors are around 20% of the total. These percentages indicate that the phenomenon we want to analyze is relevant and has a substantial impact on the management of Spanish local governments.

As a first sight of the potential manipulation of financial magnitudes, following Burgstahler and Dichev (1997), we first analyze the distribution of net profits following the assumption that, in the absence of $EM$⁠, this distribution would be smooth. On the contrary, if a large number of municipalities report earnings in an interval close to zero, this can be interpreted as a signal of $EM$ practices (Ferreira et al., 2013; Cohen et al., 2019; Cohen and Malkogianni, 2021). Figure 1 shows the frequency distribution of Net Earnings of Spanish municipalities for the period 2016–2018.

As we can observe, the distribution of Net Earnings shows a positive skewness with most of the Spanish municipalities reporting slightly positive earnings, which provides evidence of the manipulation of accounting variables and justifies a more rigorous analysis regarding their discretionary accruals.

To present empirical results, we will follow a double perspective. The first approach is to estimate model [4] by using an OLS pooled regression. The second approach consists of the estimation of the model [5] using a panel data random effect, including the treatment proposed by Mundlak (1978) adjusted for the presence of time-invariant variables.

Table 2 shows the results for the estimations of the total accruals in Spanish municipalities for the period 2016–2018. Specifically, this table represents five pooled regressions that start from including only accounting variables to increasingly adding political variables: Model 0 considers the accounting variables originally included in the adjusted Jones model; Models 1 and 2 extend model 0 by including political variables related to the government or the mayor, respectively; Model 3 extends model 0 by including all political variables; while Model 4 extends model 3 by including the interaction between the mayor's dedication and wages. It should be noted that, despite the incremental inclusion of political variables, the results remain robust and homogeneous across the different model specifications.

In general, as expected, the results show that the accounting variables included in the classical modified Jones model – specifically, those that define the generation of nondiscretionary accruals – have a significant impact on the total accruals. In the base regression model (Model 0), $R2$ is very high, reflecting the assumption that most accruals are explained by the normal activities of municipalities. Specifically, the level of total accruals increases with the growth in sales collected (⁠$Δ$ REV$(t)$− $Δ$ REC$(t)$/TA $(t−1)$⁠), while it decreases with both the size of the municipality (⁠$1/TA(t−1)$⁠) and the relative level of PP&E (⁠$PPE(t)/TA(t−1)$⁠) ^[2].

As for the inclusion of political variables, in Models 1 to 4, we identify certain factors that would explain the generation of the discretionary component of total accruals beyond the accounting variables. In these models, the adjusted $R2$ increases, though not by a very large amount. In any case, these additional variables have a marginal impact, while remaining statistically significant. Regarding the government variables included in models 1, 3 and 4, it is found that both the political power or concentration (measured by the Herfindahl index, $HI$⁠) and political alignment between the municipality and the regional government (⁠$ALIGN$⁠) exhibit a positive and significant relationship with the generation of accruals, while the variable for the political sign (⁠$RIGHTWING$⁠) remains nonsignificant.

The positive sign of political power suggests that local politicians facing greater political concentration and reduced scrutiny from the opposition are more likely to engage in EM. This result aligns with the findings of Donatella (2020), which indicate that a majority government weakens oversight, allowing greater financial discretion in municipal reporting. Thus, these results validate Hypothesis 1 (⁠$H1$⁠), reinforcing the argument that municipalities with higher political concentration have increased opportunities for accrual-based financial manipulation. Similarly, political alignment between local and regional governments also shows a positive significant association with accrual manipulation, suggesting that financial discretion is more prevalent when both governing levels are aligned. When local and regional governments share the same political affiliation, oversight mechanisms may become less rigorous due to fewer institutional constraints and reduced accountability pressures. This supports Hypothesis 2 (⁠$H2$⁠), as political alignment can facilitate opportunistic behaviour by weakening monitoring incentives and increasing the discretionary power of aligned governments.

Regarding the variables related to the mayors included in models 2, 3 and 4, the results show a negative and significant relationship between mayors' salaries (⁠$WAGE$⁠) and accrual generation, suggesting that better-paid mayors are less prone to financial manipulation. Moreover, we find a positive and significant relationship between mayors' full-time dedication (⁠$FULLTIME$⁠) and accruals, which suggests that full-time mayors have stronger incentives to manipulate financial reports compared to those who split their time between public office and other economic activities. However, when including the interaction between the variable $WAGE$ and $FULLTIME$ in model 4, the results reveal a negative and significant effect. This indicates that poorly paid full-time mayors are more likely to engage in EM, suggesting that while full-time dedication may encourage manipulation, higher salaries can mitigate this effect, demonstrating a nonlinear relationship between employment conditions and financial discretion.

Overall, the findings offer a nuanced perspective on the influence of mayoral employment conditions on EM in local governments. The negative association between mayors' salaries and accrual generation implies that higher compensation may mitigate opportunistic behavior, contrary to the initial expectation of Hypothesis 3 (⁠$H3$⁠), which predicted that better-paid mayors would be more inclined to manipulate financial reports. At the same time, the positive relationship between full-time dedication and accrual generation supports the idea that greater municipal engagement increases manipulation incentives. Overall, although these results partially contradict $H3$ by showing that higher salaries can function as a deterrent, they simultaneously confirm that intensive engagement in municipal affairs is associated with increased financial discretion. Finally, no significant relationship was found between mayors' gender (FEMALE) and the generation of discretionary accruals.

Otherwise, Table 3 presents the results for the estimations of total accruals in Spanish municipalities for the period 2016–2018, considering panel data estimations with random effects, following the variation proposed by Mundlak (1978). This approach addresses the presence of time-invariant variables and allows us to disentangle the influence of stable cross-municipality characteristics from that of dynamic intra-municipal political changes, which is especially relevant for understanding how political factors affect financial reporting practices. The table presents four regressions corresponding to Models 1 to 4.

Results show that, in the between effect (i.e. means that do not vary across units), total accruals increase significantly with municipality size, sales growth and the relative level of PP&E. Higher average wages tend to reduce manipulation, although the interaction between $WAGE$ and $FULLTIME$ is positive, indicating that higher wages reduce manipulation less for full-time mayors. Within the effect, results align with the pooled estimation, except for the municipal size, sales growth and $WAGE$⁠, which become nonsignificant. The interaction between $WAGE$ and $FULLTIME$ is negative, suggesting that increasing wages for full-time mayors reduces their tendency to manipulate EM.

In summary, the results for the political variables indicate: (1) municipalities with higher political concentration and (2) aligned with upper-level government have a higher predisposition to manipulate accruals, (3) full-time mayors tend to manipulate more, (4) higher wages reduce manipulation incentives, especially for part-time mayors and (5) increased wages for full-time mayors reduce their manipulation. Moreover, (6) political ideology and (7) mayor’s gender show no explanatory power for EM, a result consistent with previous findings in related literatures such as transparency, efficiency or financial performance (Balaguer-Coll and Brun-Martos, 2021; Balaguer-Coll and Ivanova-Toneva, 2021; Ríos et al., 2022).

This article investigates accounting manipulation in Spanish municipalities within the framework of PCT. It examines how political factors, such as the governing party and mayors, trigger opportunistic behaviors in the management of local government. The study analyzes a sample of 3,190 municipalities from the 2016–2018 political cycle. Our empirical strategy employs a one-stage modified Jones model (Dechow et al., 1995) to address biased coefficients and nonrobust standard errors found in traditional two-stage procedures (Chen et al., 2018). Additionally, we tackle the inclusion of time-invariant variables in panel-data estimations (Mundlak, 1978), which allows us to separate the influence of long-term structural differences across municipalities from short-term political dynamics, an essential distinction when studying behavior across a political cycle.

The results indicate significant manipulation of accounting information in the distribution of net profits, evidenced by positive skewness and a marked discontinuity around the transition from losses to profits. The application of the modified Jones (1991) model clearly explains the generation of discretionary accruals, beyond standard accounting variables. Government-related factors are robust and consistent across all estimations, showing that political power (as opposed to diverse representation) encourages manipulation of financial information. Additionally, political alignment with higher levels of government introduces further incentives for misreporting. Findings also show that full-time dedication by mayors increases manipulation, while higher wages reduce such incentives, particularly in part-time positions. Other factors, such as the political orientation of the local government and the mayor's gender, are nonsignificant in all estimations.

The implications of our results differ across stakeholders. Researchers and practitioners should refine empirical models by using one-stage regressions and adapting the treatments of time-invariant variables. From a citizen's perspective, voting orientation significantly influences both political concentration and the extent of financial manipulation. Large governing majorities with weak opposition fail to ensure adequate financial reporting. Moreover, political alignment with higher levels of government weakens vertical oversight, suggesting that nonaligned administrations deliver higher-quality financial information, regardless of their political orientation.

From a regulatory perspective, part-time politicians, who balance public duties with private professional activities, tend to exhibit greater financial reporting quality than full-time politicians who rely solely on public income. At the same time, adequate wages for the public service positively influence the accuracy of financial reporting. In this context, full-time dedication might be less advisable if the aim is to enhance the quality of financial disclosures. It might be desirable to promote frameworks that enable mayors to combine their professional careers with political responsibilities. Furthermore, given that wages moderate the likelihood of financial manipulation, imposing strict salary caps on public officials could hinder efforts to improve financial reporting quality and transparency. Additionally, to further enhance the quality of financial information, regulators should strengthen the role of Audit Commissions and enforce stricter sanctions for noncompliance with accounting standards. Finally, financial reports should play a more active role in guiding policy and target setting, moving beyond the mere presentation of budget figures.

Future research could offer additional perspectives for a more nuanced understanding of this phenomenon. Extending the time span might allow for the inclusion of factors such as local government re-election probabilities or mayoral turnover. While focusing on a single electoral cycle ensures internal consistency and avoids distortions from election years or extraordinary events like the COVID-19 pandemic, it may limit the generalization of the findings. Broader external validity could be achieved by examining other institutional contexts or periods marked by different political and economic conditions. Moreover, future studies could also explore broader political dynamics within local governments, such as the gender composition of the municipal council, which may be more influential than the mayor's gender alone when explaining financial reporting behaviors.

Another promising direction for future research involves linking the quality of local governance or sustainable development goals with the reliability of financial information. Methodologically, developing a model based on nonlinear patterns would enable a deeper exploration of the variability in the analyzed effects. Additionally, financial manipulation could be studied through alternative mechanisms, such as distortions in budget magnitudes or real operations.