Sage Journals: Discover world-class research

Abstract

International organisations assert that political corruption is a major global problem. However, it is still unclear how to combat corruption. Previous research has studied corruption in central or local governments, rarely in regional ones. In this paper, we aim to identify factors influencing political corruption in regional governments. Using panel data, we analyse the statistical association between 29 economic, financial, political and socioeconomic variables and the corruption perceived by citizens, from January 2006 to September 2019. The results reveal that perceived political corruption is affected by government spending and ideological distance, among other factors. We propose several measures to reduce political corruption.

Points for practitioners

The study has found that political corruption needs to be combatted through different measures for each administrative level. Our results support the idea that the following variables influence the perceived level of corruption within the regional governments: the volume of government borrowing, the ideological distance between the voter and the government, the percentage electoral participation and the volume of dependent and immigrant population of the region. Likewise, we also confirm that regional governments with higher values on the Quality of Government Index are perceived as less corrupt.

Keywords

political corruption regional governments financial factors political factors socioeconomic factors Spain

Introduction

The impact of political corruption on national development is a major problem worldwide, often producing severe economic damage. Accordingly, it has been the subject of numerous studies in many countries (Bernardes do Nascimento et al., 2019; Chabova, 2017; Guillamon et al., 2021; Jetter and Parmeter, 2018).

Specifically, research findings show that high levels of corruption in the public sector can hamper economic development (Billger and Goel, 2009; Goel and Nelson, 2010), reduce foreign investment and aggravate the fiscal deficit (Bernardes do Nascimento et al., 2019; Guillamon et al., 2021). Thus, the widespread existence of fraud and bribery justifies the timeliness and interest of investigating how political corruption in the public sector may best be combated.

The question of eradicating political corruption has also been addressed in international organisations, such as the United Nations Convention against Corruption. More recently, the World Bank, the OECD and the UN Office on Drugs and Crime joined forces to develop the Good Practices Guide to Preventing and Managing Conflicts of Interest in the Public Sector (UN, 2019). Also, the OECD Anti-Bribery Convention (OECD, 2020) established legally binding rules to criminalise the bribery of public officials. Moreover, the World Bank has denounced the risks of corruption in the context of the COVID-19 pandemic (World Bank, 2020) and the IMF (2019) has highlighted the enormous damage inflicted by corruption on public finances. It has been estimated that the cost of this corruption is equivalent to 1.25% of the world's GDP.

In this international context, researchers have sought to identify the factors which facilitate or promote this behaviour, to design and implement dissuasive measures. Levels of political corruption have been measured according to various criteria, including the number of cases brought to prosecution (Lopez-Iturriaga and Sanz, 2018; Vela-Bargues et al., 2022), indices of corruption prepared by trustworthy institutions (Billger and Goel, 2009) and public opinion surveys (Chabova, 2017).

Other studies have researched the relationship between the level of political corruption and various explanatory factors: (a) economic and financial (Benito et al., 2018; Lopez-Iturriaga and Sanz, 2018); (b) political (Kotera et al., 2012); and (c) socioeconomic (Bernardes do Nascimento et al., 2019; Jetter and Parmeter, 2018).

However, almost all these papers have focused exclusively on the national administrative situation within a given country (Bernardes do Nascimento et al., 2019; De Mingo and Cerrillo-i-Martínez, 2018).

Although some analyses have been conducted of the effects of administrative decentralisation on levels of corruption (Fisman and Gatti, 2002; Serra, 2006), very few studies have been made of local government corruption (De la Higuera et al., 2020; Guillamon et al., 2021) and even fewer have considered the situation of regional governments (Vela-Bargues et al., 2022), despite previous calls for studies in this area (Alcaraz-Quiles et al., 2020; Campos et al., 2021; Chen and Neshkova, 2020; De la Higuera et al., 2020; Guillamon et al., 2021). Indeed, Gottschalk and Stanislas (2017) and Bastida et al. (2018) concluded that the risks of corruption are greater at the levels of regional and local government since internal controls are often weaker.

Following those considerations, the study of the regional governments of Spain can contribute to the analysis of corruption by offering specific and novel findings on a different administrative level. Regional governments have their own identity that justifies their separate analysis with respect to other levels of government owing to their peculiarities (Schuster and Vallier, 2010), such as specific competencies, financial autonomy or individual government structure, which can condition the honest behaviour of the rulers (Alcaraz-Quiles et al., 2014a; Navarro-Galera et al., 2021). In fact, Eurostat (2022) and the European Commission (2022) consider regional governments as independent economic agents. Thus, certain peculiarities of the regional governments make them suitable for analysing how to fight corruption, such as: (a) managing a large budget (Ministry of Finance, 2020); (b) filling the administrative gap between the national government and local governments (Navarro-Galera et al., 2016); (c) rising cases of corruption (Gomez Fortes and Urquizu, 2015); (d) own legal control instruments (Alcaraz-Quiles et al., 2014a); and (e) own transparency laws to fight corruption (Rodriguez, 2017).

In view of these considerations, our aim in the present study is to identify factors that may influence the level of political corruption in Spain through the perceptions of citizens residing in different regions. Success in this task would enable us to define measures to effectively combat this undesirable behaviour. Accordingly, we conducted an empirical study of the situation in Spain, a European country with high levels of political corruption. Moreover, owing to its decentralised system of government, Spain has 17 regional administrations, each of which manages abundant resources, provides a wide spectrum of public services and has suffered very numerous cases of political corruption in recent years. Our analysis is based on data from Spain's 17 regions, or Autonomous Communities, covering the period from January 2006 to September 2019. With the information included in this data panel, we generated fractional response models (homoscedastic and heteroscedastic), incorporating the economic, financial, political and socioeconomic variables considered.

Theoretical framework

In this study, our empirical analysis is based on the following well-established theories, with which we examine possible relationships between political corruption and various explanatory factors.

Agency Theory;

Theory of Political Opportunism;

Public Choice Theory;

Institutional Theory; and

Stakeholder Theory.

According to Agency Theory, politicians (agents) have more and better information for decision-making, which allows them to take decisions that benefit themselves, but which may be contrary to the interests of society (principals) (Jiménez et al., 2017; Riera et al., 2013). When politicians are motivated by self-interest, they may take advantage of information asymmetry to adopt policies that are prejudicial to the objectives of the population (Verbeeten, 2008).

Building on this theory, Campos et al. (2021) and De la Higuera et al. (2020) have also suggested that corrupt activity to satisfy private interests may be governed by the postulates of the Theory of Political Opportunism (Veiga and Veiga, 2007), according to which the ideology of the governing party may influence political behaviour. Specifically, the theory suggests that political parties with a conservative ideology usually favour market mechanisms, resulting in closer relationships with private companies, which may foster corruption (Starke et al., 2016). In contrast, representatives of progressive political parties may be less prone to corrupt behaviour, owing to their greater distance from the business world (Riera et al., 2013).

The theory of Public Choice has a long history, from early work by Niskanen (1971), later developed by authors such as Rose-Ackerman (1978), to recent work by Chen and Neshkova (2020), among others. These authors suggest that public officials take decisions aimed at maximising their personal interest and may engage in corrupt behaviour when the potential benefits outweigh the possible costs. Under this theory, therefore, corruption is less attractive when the probability of discovery is high, the penalties are severe, public sector salaries are acceptable and the moral cost of public exposure is prohibitive (Williams, 2011).

According to the postulates of Institutional Theory, organisations respond to environmental pressures, adopting structures and practices that are considered legitimate and socially acceptable by comparable bodies (Ribeiro and Scapens, 2006). Organisations tend towards conformity with prevailing norms, traditions and social influences, thus making organisational structures and practices largely homogeneous (DiMaggio and Powell, 1983). Institutions tend to adopt the practices that are generally accepted in their cultural and socioeconomic contexts, in a process termed mimetic isomorphism. In this context, the absence of legal norms against corruption and the generalised sensation of impunity regarding corrupt behaviour detected elsewhere can encourage politicians to behave corruptly.

Finally, Stakeholder Theory, according to Freeman (1984), defines a stakeholder as any group or individual that can affect or be affected by the efforts of an organisation to achieve its objectives. This theory asserts that the long-term goal of any organisation (including government) is to maximise the well-being of its stakeholders (Rusconi et al., 2007). Accordingly, public entities should seek to understand, respect and respond to the needs of their stakeholders (Freeman, 1984). One consequence of this is that the scant presence of persons with special needs (such as the disabled, the unemployed or immigrants) or of those with a low level of formal education could foster the appearance of corrupt behaviour.

Sample, variables and statistical methodology

Sample selection

Reflecting the motivation for this investigation and the background described in the Introduction, our study sample is composed of the 17 regional governments of Spain. This country is an appropriate choice for analysis for the following reasons. In recent years, opinion surveys have consistently identified the question of public sector corruption as one of the most important problems facing society (see, for example, the surveys published by the Spanish CIS – Centre for Sociological Research). A parallel, international, perspective is provided by the Global Corruption Barometer – European Union 2021 (Transparency International, 2021), according to which 86% of the Spanish population believe political corruption is a major problem, making this country a very suitable context for our analysis of the factors influencing corruption.

Furthermore, many instances of political corruption have been detected within regional governments. Spain's regional governments provide essential services such as health, education and social services. They enjoy considerable financial autonomy and manage a total budget that in 2019 amounted to €204 billion (Ministry of Finance, 2020). Yet, despite its evident importance in Spanish society, this area of public administration has attracted very little research attention.

Likewise, the study of corruption in regional governments is particularly interesting for several reasons. Following Schuster and Vallier (2010), as well as other up-to-date pieces of work (Navarro-Galera et al., 2021), each regional government, individually considered, has different peculiarities that condition its internal management such as specific competencies and political and financial autonomy.

Also, previous research has shown that these distinctive characteristics of regional governments can affect transparency, financial sustainability and service quality (Alcaraz-Quiles et al., 2014a and 2014b; Navarro-Galera et al., 2016), showing the interest and opportunity to specifically analyze corruption at this level of government. However, the problem of corruption at this level of government has not been specifically analyzed, although its identity allows justification of its in-depth analysis.

Outcome variable

In our empirical research, the outcome variable is the perceived level of political corruption. Previous studies of this phenomenon have used various measurement criteria, which can be grouped into three types. The first group is composed of studies that have measured corruption in terms of the number of cases prosecuted (Lopez-Iturriaga and Sanz, 2018). In a very numerous second category, studies have used indices defined by independent institutions (Transparency International's Corruption Perceptions Index and the World Bank's Control of Corruption Index; Billger and Goel, 2009; Serra, 2006). Finally, some studies have used public opinion surveys to measure the level of corruption perceived by the population (Chabova, 2017).

In the present study, considering the socioeconomic context of the Spanish regional governments and the availability of appropriate data for our analysis, the measurement instrument used was that of public opinion surveys. Specifically, the outcome variable in our analysis is quantified from the results of the surveys conducted by the CIS, an autonomous office within the Ministry of the Presidency aimed at facilitating the scientific study of Spanish society via systematic analyses.

In this paper, we examine the results of 629 CIS surveys, the earliest of which was No. 2633, carried out in January 2006, up to No. 3261, from September 2019. This focus enabled us to generate a data panel with 17 cross-sectional units (corresponding to the 17 regional governments) and 629 time periods. The level of perceived political corruption was measured according to the percentage of people who, month-by-month, considered the fraudulent behaviour of political officials to be among the three main problems facing the country.

Several reasons have encouraged us to make use of this measurement. First, the cases of corruption in regional governments have been numerous and considerable in terms of their economic impact in the country as a whole (Gomez Fortes and Urquizu, 2015). Therefore, the country's public opinion on this topic is heavily impacted by corrupt behaviour within these regional governments. In fact, previous research concerning Spain shows that corruption at this administrative level has a relevant effect on the public opinion concerning the whole country (Campos et al., 2021; Lopez-Iturriaga and Sanz, 2018; Vela-Bargues et al., 2022). Second, we have focused our analysis on the regions of Spain, including as the outcome variable the percentage of the population within the region that consider corruption to be one of the three main problems the country faces, month by month. The CIS surveys in our database are representative for the entirety of the country and they include every regional government. Third, our main goal is to analyse, in each region, the relevant factors determining the level of perceived corruption as a problem for the country. Fourth, authors such as Chabova (2017) state that corruption perception needs to be studied from a general perspective that includes all available levels of government: national, regional and local.

Explanatory variables

Following previous research into the causes of political corruption and the postulates presented in Section 2 (Theoretical framework), our empirical analysis considers three types of potential influences (explanatory, or independent variables) on levels of perceived corruption within regional governments: (a) economic and financial factors; (b) political factors; (c) socioeconomic factors. Table 1¹ summarises the variables considered, listing their definitions and calculation methods and the expected sign of each relationship with the political corruption perceived among the population.

Economic and financial factors

Previous studies have reported evidence of a non-conclusive relationship between national economic growth and perceived levels of corruption (Bernardes do Nascimento et al., 2019; Chen and Neshkova, 2020; Corrado and Rossetti, 2018; Rodriguez Bolivar et al., 2018). A positive association would be congruent with the postulates of Public Choice Theory, since a greater availability of resources might increase the possibility of obtaining personal benefits through fraudulent behaviour. Therefore, we selected the variable GDP per capita within the region.

Previous research findings also suggest that we should examine the influence on corruption of the regional government's financial autonomy (Tavares and Da Cruz, 2014). Under Agency Theory, this variable could be a relevant factor, because if the administration's own resources constitute a large proportion of the total, citizens may conclude that the politicians’ interests are in line with their own, limiting information asymmetry and reducing the scope for corrupt behaviour.

In terms of public spending, we followed previous research in selecting the following variables: Total public spending, Current spending and Investment spending. Evidence has been reported of a positive relationship with the level of corruption, which is in line with Public Choice Theory, according to which the greater the volume of managed resources, the greater the possibilities for and benefits from corrupt behaviour (Bernardes do Nascimento et al., 2019; Corrado and Rossetti, 2018; Fisman and Gatti, 2002). Similarly, the Budget balance and the Cash surplus might also bear a positive relationship with political corruption (Alcaraz-Quiles et al., 2020; Rodriguez Bolivar et al., 2018) and were selected for analysis.

Finally, the variable Government borrowing was included, from the view that increased borrowing tends to reduce available resources, limiting incentives to adopt corrupt behaviour. In addition, higher levels of public debt can spur citizens to demand greater information from politicians. If this were the case, it would reduce information asymmetry (Benito et al., 2018).

Political factors

Piotrowski and Van Ryzin (2007) emphasised the importance of the ideological position of citizens regarding their perception of political corruption. For this reason, our analysis incorporates the variable Ideological distance. According to the Theory of Political Opportunism, the greater a person's ideological distance from the ruling party, the stronger their perception of political corruption.

In addition, considering the Theory of Political Opportunism, our analysis includes the variable Political sign of the governing party. In this respect, it has been suggested that progressive parties have a greater propensity to spend/invest than conservative parties (Alcaraz-Quiles et al., 2020; Bastida et al., 2018, Rodriguez Bolivar et al., 2018).

Another interesting variable is that of the Ideological alignment between the parties in power in central government and in each regional government. The justification for this variable is based both on the evidence obtained by previous research (Guillamon et al., 2021; Rodriguez Bolivar et al., 2018) and on Institutional Theory, since corrupt behaviour at other levels of government may promote isomorphism in regional governments.

Another variable incorporated into our analysis is the existence or otherwise of an Absolute majority for the governing party. The possible influence of this factor on political corruption has been suggested by Tavares and Da Cruz (2014) and Serra (2006). According to Agency Theory, a political party that governs alone is more readily able to escape political control.

Furthermore, following Alcaraz-Quiles et al. (2020) and Rodriguez Bolivar et al. (2018), we have included a variable based on the degree of Political fragmentation in the regional parliament. According to Agency Theory, the greater the fragmentation, the greater the possibility of political control by other parties, which would reduce information asymmetry, thus helping combat political corruption.

The Number of consecutive legislatures enjoyed by the party currently in power can also increase the level of perceived corruption (Bastida et al., 2018; Tavares and Da Cruz, 2014). Thus, according to Public Choice Theory, the greater the number of consecutive legislatures by the same party, the greater its possibilities of concealing evidence of corrupt practices, and this facility might influence its attitude towards committing fraud.

The current timepoint within the Legislative cycle can also affect the level of political corruption (Benito et al., 2018). Under Agency Theory, the proximity of new elections can motivate the governing party to publicise its actions. This would reduce information asymmetry. Furthermore, according to Public Choice Theory, greater proximity to elections reduces the time available in which to participate in corrupt practices.

The level of Electoral participation in the last elections is another variable of interest (Tavares and Da Cruz, 2014). Following Stakeholder Theory, a high rate of participation could be interpreted as great interest among voters in governmental effectiveness, leading to a greater demand for information. In consequence, asymmetries would be reduced according to Agency Theory.

Serra (2006) and Fisman and Gatti (2002) concluded that the level of administrative decentralisation might be associated with the level of political corruption. Hence, we include the Number of provinces. According to Agency Theory, the governments of regions with a greater number of provinces and, therefore, a higher degree of decentralisation through provincial delegations might receive more pressure to reduce information asymmetry, limiting political corruption.

Finally, we have taken into consideration the work provided by Charron et al. (2021) concerning the Quality of Government Index (EQI) for the regional governments present in the European Union. By including the EQI in our analysis we measure homogeneously how citizens perceive the quality of their own regional government. The study analyses the ability of regional governments to manage essential public services, such as health, education and social services, as well as the prevalence of corruption in the region. Regional governments with higher levels in the EQI are expected to be linked with lower levels of perceived corruption.

Socioeconomic variables

Recent studies have concluded that a higher level of education among the population is associated with reduced political corruption, although on a different level of analysis rather than the regional. Also, the postulates of Agency Theory and Stakeholder Theory suggest that citizens with more advanced educational qualifications will demand more information from governments, reducing information asymmetry and making corrupt behaviour less likely. Thus, we include the following variables reflecting levels of education among the population: Primary studies, Secondary studies and Higher education.

Furthermore, according to Bernardes do Nascimento et al. (2019), Alcaraz-Quiles et al. (2020) and Goel and Nelson (2010), increasing public access to the internet would limit the scope for political corruption. This hypothesis is in line with Agency Theory, as greater access to the internet reduces information asymmetry. Similarly, Public Choice Theory holds that wider internet access would increase the possibility of corrupt behaviour being discovered. Accordingly, the variable Internet access is included in our analysis.

Another variable that might affect the level of perceived corruption is Fiscal pressure (Campos et al., 2021; Rodriguez Bolivar et al., 2018). Under Stakeholder Theory, a heavier tax burden would make voters more demanding regarding the use and control of public resources, reducing the scope for corrupt behaviour.

Our analysis also includes the Unemployment rate in the region, a variable previously identified as potentially useful (Alcaraz-Quiles et al., 2020; Tavares and Da Cruz, 2014). Stakeholder Theory states that rising unemployment could be associated with a parallel increase in levels of public corruption, owing to the conflict of interests between the unemployed population and government agencies. On the other hand, according to Agency Theory, the special needs of certain population groups, such as the unemployed, might enhance awareness among politicians and reduce their propensity for corruption.

Population volume and density has been also included, since previous studies suggest it may influence perceived levels of corruption (Chen and Neshkova, 2020). Following Agency Theory, in highly and/or densely populated regions, governments are under greater pressure to reduce information asymmetry lowering the level of corruption.

We include two more population variables, the Dependent population and the Immigrant population. In this respect, Campos et al. (2021) and De la Higuera et al. (2020) concluded that each of these variables could present an inverse association with the level of corruption, following Stakeholder Theory.

Finally, following the work by BBVA Foundation (2018), we have segmented all 17 Regional Governments of Spain according to their general level of welfare. By differentiating regions with higher levels of welfare – considering a range of attributes that includes general satisfaction with life in the region, economic security and healthcare – we can test which factors impact the most in the level of corruption perceived for each kind of region: (a) a high level of welfare; and (b) a low level of welfare.

Thus, we distributed all 17 regions into two different groups according to the welfare ranking provided by BBVA Foundation (2018). High dispersion in the results encouraged us to segment the regions using the median value. The resulting groups include eight regions with high levels of welfare and nine regions with low levels of welfare.

Model and econometric method

This study is based on a randomly selected sample with i observations $(i = 1, \dots, N)$ . For each value of i, there are T timepoint observations $= 1, \dots, T$ . The response variable, $y_{i t}$ , is bounded as follows: $0 \leq y_{i t} \leq 1$ . There exists a 1 × K vector of $x_{i t}$ explanatory covariates.

In a cross-sectional context, with fixed effects, and a response variable, such as $y_{i t}$ , a fractional response model can be specified for additive unobserved heterogeneity (Papke and Wooldridge, 2008):

E (y_{i t} | x_{i t}, μ_{i}, h_{i t}) = G (x_{i t} β, μ_{i}, h_{i t})

(1)

where

β

are the parameters to be estimated and μ_i and

h_{i t}

are the unobserved heterogeneity. We assume that

μ_{i}

is invariant in t, and that

h_{i t}

varies both in i and in t. G is the standardised normal or probit distribution function.

Since model (1) contains exogenous, continuous variables, we define the partial effect of the jth covariate $x_{t j}$ . For the sake of simplicity, we then eliminate the subscript i:

P E_{j} = \frac{\partial E (y_{t} | x_{t}, μ, h)}{\partial x_{t j}} = \hat{β_{j}} \times G (x_{t} \hat{β}, μ, h)

(2)

Papke and Wooldridge (2008) propose that whenever the partial effects can be averaged across the sample, in

x_{t}

, we must analyse the influence of the covariates, estimating the mean value of

P E_{j}

, APEs, as follows:

A P E s_{(μ, h)} = E_{(μ, h)} [{\hat{β}}_{j} \times G (x_{t} \hat{β}, μ, h)] = \hat{β_{j}} \times E_{(μ, h)} [G (x_{t} \hat{β}, μ, h)]

(3)

Let us now assume there is a vector

1 \times M

of the variables

z_{i t}

which depend on the covariates

x_{i t}

. In this case, a fractional response model must be specified, with multiplicative heteroscedasticity (Wooldridge, 2010):

E (y_{i t} | x_{i t}, μ_{i,} h_{i t}) = Φ [(x_{i t} β, μ_{i,} h_{i t}) \times \exp (\frac{- z_{i t} γ}{2})]

(4)

where

Φ (\cdot)

is a standardised normal or probit distribution function and

γ

are the parameters of the heteroscedasticity equation. To obtain a consistent estimate of β, μ and h, we must specify the Bernoulli quasi-likelihood function (Bluhm, 2013). Let

ω_{j}

be a continuous exogenous variable, which can belong to x_it, z_it or both simultaneously. We define the partial effect of the jth variable

ω_{t j}

(Bluhm, 2013) as follows:

P E (ω_{t j}) = \frac{\partial E (y_{t} | x_{t}, μ, h)}{\partial ω_{t j}} = \frac{(ω_{j} \in x_{t}) \hat{β_{j}} - (ω_{j} \in z_{t}) \hat{γ_{j}} (x_{t} \hat{β,} μ, h)}{\exp (z_{t} \hat{γ})} \times ϕ (\frac{x_{t} \hat{β}, μ, h}{\exp (z_{t} \hat{γ})})

(5)

As in the homoscedastic fractional response model, we interpret the effect of the covariates

x_{t}

on the endogenous covariate

y_{i t}

, analysing the mean value of the partial effects (APEs), through the distribution of µ and h:

A P E s_{(μ, h)} (ω_{j}) = \frac{(ω_{j} \in x_{t}) \hat{β_{j}} - (ω_{j} \in z_{t}) \hat{γ_{j}} (x_{t} \hat{β,} μ, h)}{\exp (z_{t} \hat{γ})} \times E_{(μ, h)} [ϕ (\frac{x_{t} \hat{β}, μ, h}{\exp (z_{t} \hat{γ})})]

(6)

To determine which exogenous variables generate heteroscedasticity, we performed the LnSigma test (Stata Corporation, 2019), which is an adaptation of the test proposed by Harvey (1976).

In both the homoscedastic and heteroscedastic models, only one unobserved effect $μ_{i}$ is specified, in order to capture heterogeneity. The unobserved effect h was excluded from the estimate in order to avoid possible problems of multicollinearity.

Empirical results

Descriptive analysis

Figure 1a and b shows the perception of corruption but differentiating between regions with higher and lower levels of welfare. Navarre is the region with the highest percentage of people perceiving corruption as one of the three main problems, while in the group of regions with the lowest levels of prosperity, the Balearic Islands occupies the first place in the ranking. If we compare the perception of corruption in the two categories of regions, we observe that the percentage of the population perceiving corruption as a problem is somewhat higher in the case of the regions with a higher level of welfare.

On average, the percentage of the population perceiving corruption as a major problem is higher in the regions with the highest levels of welfare than in the regions with the lowest levels of welfare.

The sample is structured as a Data Panel, and the mean values for the set of 17 regional governments analysed are shown in Figure 2. The highest level of perceived corruption was reached in November 2014, when 64.67% of citizens believed that this was one of the three main problems in the country. The lowest level was recorded in March 2008, when only 0.14% of those surveyed shared this belief. Figure 2 also shows that in January 2014 the falling trend in perceptions of corruption changed, and growing numbers of citizens stated that this was one of the country's main problems. However, the trend turned downwards again in November 2014.

Table 2 shows the descriptive statistics for the dependent variable (level of perceived corruption) and the covariates. From an analysis of the dependent variable, the sample may be considered homogeneous, as is corroborated by the dispersion coefficient. Among the covariates considered, the explanatory variable that presents the greatest dispersion is that of Cash surplus, with a coefficient of −5.9. The remaining independent or explanatory variables present greater homogeneity.

Analysis of the results of the statistical model

Table 3 shows the results of the homoscedastic and heteroscedastic estimates of the fractional response model. Part A shows the estimated coefficients and robust standard errors, and part B the statistical tests which allow us to diagnose the estimates made, and the Lnsigma test.

The values obtained for the pseudo R² coefficient show that the homoscedastic model presents a better fit than the heteroscedastic estimate. However, the log-likelihood test shows that both models properly explain the differences in perceived corruption within regional government.

Comparison of the estimates made, based on AIC and BIC selection criteria, reveals that both the homoscedastic and the heteroscedastic models are appropriate instruments for describing the perceived political corruption considered, although the heteroscedastic estimate does present slightly better ratios.

In addition, the Lnsigma test shows that we must accept the hypothesis that at least some of the covariates specified contribute to generating heteroscedasticity, namely Unemployment, Internet access, Number of provinces, Electoral participation and Government borrowing (see Table A.1, Appendix).

Table 4 shows the average values of the partial effects (APEs) of the proposed model. Our results for the economic and financial variables show that an increase in GDP per capita might aggravate the levels of perceived corruption, which is contrary to previous findings (Bernardes do Nascimento et al., 2019; Chen and Neshkova, 2020; Corrado and Rossetti, 2018; Rodriguez Bolivar et al., 2018) for both national and local governments, although these earlier conclusions were not entirely consistent. Thus, our findings suggest that the economic growth of the region might cause the citizens to distrust the governmental actions. Two variables – Total spending and Investment spending – present an inverse association with the level of corruption, which is contrary to earlier findings (Bernardes do Nascimento et al., 2019; Corrado and Rossetti, 2018) for national governments. However, the direct association for Current spending is in line with previous literature. These findings imply that public expenditure policies meant to fulfil the needs of the citizens (Stakeholder Theory) might contribute to lowering of the level of concern held by the population towards fraudulent behaviour of the government, which can be an obstacle to fighting corruption. Our results also indicate that higher levels of government borrowing are associated with higher levels of perceived political corruption, which is contrary to the conclusions of Benito et al. (2018) for local governments. Finally, according to our results, higher levels of financial autonomy are correlated with a lower perception of corruption. However, we did not find any conclusive statistical evidence for the variable Cash surplus.

With respect to the political variables considered, our results show that eight variables significantly influence the perceived level of corruption, in line with that reported in previous studies. A novel contribution to the literature is our finding that greater ideological distance between the respondent and the governing party is associated with higher levels of perceived corruption. So, perceived corruption is unavoidably affected by ideological attributes of the citizens, as stated by the Theories of Political Opportunism. This finding suggest that corruption control should be stricter in those governments with lower ideological distance from the governing party. In this respect, our analysis extends the conclusions reported by Campos et al. (2021) and Piotrowski and Van Ryzin (2007) for local governments. We also confirm that regional governments with higher values in the EQI are perceived as less corrupt, in line with Charron et al. (2021). However, we observed no such relation for two other variables (Political sign and Ideological alignment), in contrast to earlier studies (Alcaraz-Quiles et al., 2020; Rodriguez Bolivar et al., 2018; Tavares and Da Cruz, 2014).

In addition, the data analysed suggest that the variables Absolute majority and the absence of Political fragmentation are associated with stronger perceptions of corruption, which are congruent with the Agency Theory. However, greater numbers of provincial delegations are associated with higher levels of perceived corruption, which is contrary to previous findings in the literature. Moreover, the later the stage of the legislative cycle, the more likely it is that citizens will perceive corruption through those actions performed by the government to improve their electoral results as election day approaches. Finally, we found that increments in the variable Electoral participation are correlated with lower levels of perceived corruption, as supported by Stakeholder and Agency Theories.

Our results for the socioeconomic variables show that internet access is positively associated with perceptions of political corruption, contrary to that observed in previous research (Bernardes do Nascimento et al., 2019; Goel and Nelson, 2010) for national and local governments. This relation might be linked to the fact that citizens become aware of corrupt actions more easily. We also confirm previous findings for the variable Fiscal pressure in national governments (Campos et al., 2021). A higher fiscal pressure might cause the population to become more demanding, reducing the scope for corrupt behaviour, following the statements of the Stakeholder Theory.

We also present evidence that the size of the dependent population is positively associated with perceived corruption. This, too, is contrary to previous research outcomes (Chen and Neshkova, 2020; De la Higuera et al., 2020) for national and local governments, but can be justified from the perspective of the Stakeholder Theory. The special needs of that population group might cause an increase in the level of perceived corruption when they are more numerous. On the other hand, the negative association obtained for the variable Immigrant population is in line with previous literature for national and local governments. Our results suggest as well that rising unemployment is associated with higher perceptions of political corruption, which is contrary to the findings of Tavares and Da Cruz (2014) for local governments. We found no evidence of any significant influence from the variables Higher studies and Population volume. We found significance negative influence for the variables Primary studies and Secondary studies; however, this is only true for the heteroscedastic analysis.

Finally, an extended model has been estimated, in which the interaction effects of the exogenous variables with the level of welfare of the regions have been specified. We chose to differentiate between two types of regions: high-income regions and middle- and low-income regions (see the Appendix). We have found that the effect of variables such as dependent and immigrant population, as well as the unemployment rate, is more intense in those regions with a higher general level of welfare, which might be due to the fact that these factors can threaten the optimal state of welfare achieved. On the other hand, we have found that the effect of the variables primary studies, secondary studies and internet access is more prominent in those regions with a lower general level of welfare, as the access to basic education and information makes a bigger difference in regions with less economic and social development.

Conclusions

Our findings represent a significant advance on the results of previous research. They suggest that the factors influencing the perceived level of corruption may differ between regional governments and other levels of analysis. Therefore, to effectively combat political corruption it could be necessary to employ different measures and instruments for each of these administrative levels. These conclusions must be interpreted considering that our outcome variable is the percentage of people who, at the level of each region, perceive corruption as one of the three main problems in the country as a whole. However, the level of corruption in each region has an important effect on the perception at the country level. However, our results have a limitation. The perceived corruption is only a proxy indicator of the actual level of corruption, as the level of corruption perceived by citizens also may be influenced by the press.

We have detected the influence of the following variables on the perceived level of corruption at the country level of analysis (a novel finding): (a) the volume of government borrowing (positive relation); (b) ideological distance (positive relation); (c) electoral participation (inverse relation); and (d) dependent population and immigrant population (positive and inverse relation, respectively). Likewise, we also confirm that regional governments with higher values in the EQI are perceived as less corrupt.

Regarding the economic and financial variables, our results show that perceptions of political corruption are reduced when the volume of managed resources increases, although they are not when economic growth is enhanced. This discrepancy with previous research might be caused by the origin of these funds, mostly transfers from the central government. These findings suggest that political corruption might be effectively combated via the implementation of strict controls on the destination and execution of transferred funds. Moreover, our findings imply that public expenditure policies meant to fulfil the needs of the citizens might contribute to lower the level of concern held by the population towards fraudulent behaviour of the government.

On the other hand, our findings for the political variables show that ideological distance and the timepoint of the electoral cycle may both contribute to heightening public perceptions of political corruption. This is also true for the absence of an absolute majority in the regional parliament and that of a low degree of political fragmentation. However, such perceptions are reduced when there is greater electoral participation, especially when the region has a low number of provincial delegations.

Some of our findings with respect to the socioeconomic variables are contrary to those obtained by previous studies on corruption in central governments. In our case, the results obtained show that greater internet access may aggravate the level of corruption perceived by citizens. Our results also suggest that a situation of decreasing unemployment, as well as a higher fiscal pressure, may favour the fight against corruption.

We show that an increase in another two socioeconomic variables – population density and immigrant population – and a reduction in the dependent population may contribute to combating political corruption. However, no conclusive evidence was obtained of any relation between this corruption and the level of education.

Our results show that the level of welfare of the citizens of each region can influence the level of perceived corruption in the country as a whole. These findings indicate that the effect of certain variables is more intense in regions with higher levels of welfare, which encourages stricter controls in those regions.

Finally, our findings provide evidence on the relevance of specifically studying corruption in regional governments, as suggested by previous research. According to our results, three factors (internet access, immigrant population and unemployment) are related to corruption inversely to previous research on central and local governments. This finding implies that the peculiarities of the regional governments that apply can exert a different influence on corruption. This may be due to their greater proximity to the citizens (Stakeholder Theory) and its specific ability to adapt to the environment (Institutional Theory). In addition, our results also support the opposite sign to that obtained in central governments for two variables (total spending and investment spending), which could be explained, following the Agency Theory, considering the special commitment to honesty from these governments.

Supplemental Material

sj-docx-1-ras-10.1177_00208523231159244 - Supplemental material for Factors influencing political corruption. An empirical research study of regional governments

Supplemental material, sj-docx-1-ras-10.1177_00208523231159244 for Factors influencing political corruption. An empirical research study of regional governments by Daniel Raya-Quero, Andrés Navarro-Galera and José Luis Sáez-Lozano in International Review of Administrative Sciences

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This work was supported by Instituto de Estudios Fiscales (https://campus.ief.es) under Grant ‘Proyecto de Investigación del Instituto de Estudios Fiscales’ and Junta de Andalucía () under Grants ‘P18-TP-4475’, ‘A-SEJ-154-UGR18’ and ‘CV20-45590’.

ORCID iD

Daniel Raya-Quero

Andrés Navarro-Galera

Supplemental material

Please see tables, figures and appendix at

Notes

Daniel Raya-Quero is a Professor of Corporate Law at the University of Granada, Spain. As a young researcher in Economic Law, he is currently working on his PhD dissertation at the University of Bologna, Italy, as a member of the Reale Collegio di Spagna. His research interests include Banking and Technology Law, Artificial Intelligence in Law, and Institutional Transparency.

Andrés Navarro-Galera is a Full Professor at the Granada University, Spain. He is a member of the European Accounting Association. Andrés has published books and books chapters edited by the Spanish Ministry of Treasury as well as articles in international journals. Likewise, he currently leads several research projects on the financial management of public entities.

José Luis Sáez-Lozano is a Full Professor of Economics at the University of Granada, Spain. He has published several pieces of work in international journals on different topics concerning economic and statistical analysis. His main research areas are Decentralization, Voting, Subjective Well-being, Economic Policy, Public Policy and Artificial Intelligence Applications.

References

Alcaraz-Quiles

Navarro-Galera

Ortiz-Rodríguez

(2014a) Factors influencing the transparency of sustainability information in regional governments: an empirical study. Journal of Cleaner Production 82(4): 179–191.

Alcaraz-Quiles

Navarro-Galera

Ortiz-Rodríguez

(2014b) A comparative analysis of transparency in sustainability reporting by local and regional governments. Lex Localis-Journal of Local Self-Government 12(1): 55–78.

Alcaraz-Quiles

Navarro-Galera

Ortiz-Rodriguez

(2020) The contribution of the right to information laws in Europe to local government transparency on sustainability. International Environmental Agreements: Politics, Law and Economics 20(1): 161–178.

Bastida

Guillamon

Ríos

(2018) Corruption and electoral outcomes in Spanish local governments. In: Retos de la contabilidad y la auditoría en la economía actual: Homenaje al profesor Vicente Montesinos Julve. Valencia, Spain: Universidad de Valencia, pp.519–532.

BBVA Foundation (2018) Las facetas del bienestar: una aproximación multidimensional a la calidad de vida en España y sus comunidades autónomas (2006–2015). Bilbao, Spain: Fundación BBVA.

Benito

Guillamon

Rios

, et al. (2018) Can salaries and re-election prevent political corruption? An empirical evidence. Revista De Contabilidad-Spanish Accounting Review 21(1): 19–27.

Bernardes do Nascimento

JCH

Macedo

MADS

Siqueira

JRMD

, et al. (2019) Government corruption and Internet access diffusion: global evidence. Revista de Administração Pública 53(6): 1011–1039.

Billger

Goel

(2009) Do existing corruption levels matter in controlling corruption? Cross-country quantile regression estimates. Journal of Development Economics 90(2): 299–305.

Bluhm

(2013) fhetprob: A fast QMLE Stata routine for fractional probit models with multiplicative heteroskedasticity. Unpublished manuscript.

10.

Campos

De la Higuera

Pérez

, et al. (2021) Drivers of contracting back in local governments: Analysing efficiency, opportunistic political cycles, political corruption and financial factors. Journal of Economic Policy Reform 24(3): 347–360.

11.

Chabova

(2017) Measuring corruption in Europe: Public opinion surveys and composite indices. Quality & Quantity 51(4): 1877–1900.

12.

Charron

Lapuente

Bauhr

(2021) Sub-national Quality of Government in EU Member States: Presenting the 2021 European Quality of Government Index and its Relationship with COVID-19 Indicators. University of Gothenburg Press, Gothenburg.

13.

Chen

Neshkova

(2020) The effect of fiscal transparency on corruption: A panel cross-country analysis. Public Administration 98(1): 226–243.

14.

Corrado

Rossetti

(2018) Public corruption: A study across regions in Italy. Journal of Policy Modeling 40(6): 1126–1139.

15.

De la Higuera

Plata

Zafra

, et al. (2020) A dynamic model for analysing partisan political corruption in local government: the Spanish experience. Lex Localis – Journal of Local Self Government 18(1): 1–33.

16.

De Mingo

Cerrillo-i-Martinez

(2018) Improving records management to promote transparency and prevent corruption. International Journal of Information Management 38(1): 256–261.

17.

DiMaggio

Powell

(1983) The iron cage revisited – Institutional isomorphism and collective rationality in organizational fields. American Sociological Review 48(2): 147–160.

18.

European Commission (2022) NextGenerationEU funding, information. Available at: https://ec.europa.eu/info/strategy/eu-budget/eu-borrower-investor-relations/nextgenerationeu_en.

19.

Eurostat (2022) Government finance statistics, information. Available at: https://ec.europa.eu/eurostat/web/government-finance-statistics.

20.

Fisman

Gatti

(2002) Decentralization and corruption: Evidence across countries. Journal of Public Economics 83(3): 325–345.

21.

Freeman

(1984) Strategic Management: A Stakeholder Approach. New York: Cambridge University Press.

22.

Goel

Nelson

(2010) Causes of corruption: History, geography and government. Journal of Policy Modeling 32(4): 433–447.

23.

Gomez Fortes

Urquizu

(2015) Political corruption and the end of two-party system after the May 2015 Spanish regional elections. Regional & Federal Studies 25(4): 379–389.

24.

Gottschalk

Stanislas

(2017) Public Corruption: Regional and National Perspectives on Procurement Fraud. New York: CRC Press.

25.

Guillamon

Cifuentes

Faura

, et al. (2021) Effect of political corruption on municipal tax revenues: efecto de la corrupción política en los ingresos impositivos municipales. Revista de Contabilidad-Spanish Accounting Review 24(2): 231–240.

26.

Harvey AC (1976) Estimating regression models with multiplicative heteroscedasticity. Econometrica: Journal of the Econometric Society: 461–465.

27.

IMF (2019) The cost of corruption. Finance and Development, 26–29 September. International Monetary Fund, Washington, DC.

28.

Jetter

Parmeter

(2018) Sorting through global corruption determinants: Institutions and education matter – not culture. World Development 109(3): 279–294.

29.

Jiménez

Nombela

Suárez-Alemán

(2017) Tourist municipalities and local political corruption. International Journal of Tourism Research 19(5): 515–523.

30.

Kotera

Okada

Samreth

(2012) Government size, democracy, and corruption: An empirical investigation. Economic Modelling 29(6): 2340–2348.

31.

López-Iturriaga

Sanz

(2018) Predicting public corruption with neural networks: An analysis of Spanish provinces. Social Indicators Research 140(3): 975–998.

32.

Ministry of Finance (2020) Presupuestos Generales de las Comunidades Autónomas, November. Madrid: Ministerio de Economía y Hacienda.

33.

Navarro-Galera

Alcaraz-Quiles

Ortiz-Rodríguez

(2016) Online dissemination of information on sustainability in regional governments. Effects of technological factors. Government Information Quarterly 33(1): 53–66.

34.

Navarro-Galera

Alcaide-Muñoz

López-Subires

, et al. (2021) Identifying risk determinants of the financial sustainability of regional governments. Public Money & Management 41(3): 255–263.

35.

Niskanen

(1971) Bureaucracy and Representative Government. New York: Transaction.

36.

OECD (2020) OECD convention on combating bribery of foreign public officials in international business transactions. Available at: https://www.oecd.org.

37.

Papke

Wooldridge

(2008) Panel data methods for fractional response variables with an application to test pass rates. Journal of Econometrics 145(1–2): 121–133.

38.

Piotrowski

Van Ryzin

(2007) Citizen attitudes toward transparency in local government. The American Review of Public Administration 37(3): 306–323.

39.

Ribeiro

Scapens

(2006) Institutional theories in management accounting change. Qualitative Research in Accounting & Management 3(2): 94–111.

40.

Riera

Barberá

Gómez

, et al. (2013) The electoral consequences of corruption scandals in Spain. Crime, Law and Social Change 60(5): 515–534.

41.

Rodríguez

(2017) El impacto de la Ley 19/2013 de Transparencia, Acceso a la Información Pública y Buen Gobierno en la publicación de información relacionada con la rendición de cuentas en los portales de datos abiertos autonómicos españoles. Auditoría Pública 70(1): 143–152.

42.

Rodríguez Bolivar

Galera

Subires

, et al. (2018) Analysing the accounting measurement of financial sustainability in local governments through political factors. Accounting, Auditing & Accountability Journal 31(8): 2135–2164.

43.

Rose-Ackerman

(1978) Corruption: A Study in Political Economy. New York: Academic Press.

44.

Rusconi

Freeman

Dorigatti

(2007) Teoria Degli Stakeholder, Vol. 7. Milano: Franco Angeli.

45.

Schuster

Vallier

(2010) Local and regional government in Europe. Structures and Competences. Brussels, CEMR.

46.

Serra

(2006) Empirical determinants of corruption: A sensitivity analysis. Public Choice 126(1-2): 225–256.

47.

Starke

Naab

Scherer

(2016) Free to expose corruption: The impact of media freedom, internet access, and governmental online service delivery on corruption. International Journal of Communication 10(1): 4702–4722.

48.

Stata Corporation (2019) Stata Base Reference Manual Release 16, 16th ed. Lakeway, College Station, TX: StataCorp.

49.

Tavares

Da Cruz

(2014) The determinants of local government transparency: A preliminary analysis. In: Proceedings of the 8th International Conference on Theory and Practice of Electronic Governance. New York, United States of America: Association for Computing Machinery, pp.117–123.

50.

Transparency International (2021) GCB – Global Corruption Barometer, June . Berlin: Transparency International.

51.

United Nations (2004) United Nations convention against corruption. Available at: https://www.unodc.org.

52.

Veiga

(2007) Political business cycles at municipal level. Public Choice 131(12): 45–64.

53.

Vela-Bargues

Polo-Garrido

Emilio

, et al. (2022) The relationship between auditing, political corruption and information transparency: A comparative analysis of the Spanish Autonomous Communities. Revista de Contabilidad – Spanish Accounting Review 25(1): 31–44.

54.

Verbeeten

(2008) Performance management practices in public sector organizations. Accounting, Auditing & Accountability Journal 21(3): 427–454.

55.

Williams

(2011) Shining a light on the resource curse: An empirical analysis of the relationship between natural resources, transparency, and economic growth. World Development 39(4): 490–505.

56.

Wooldridge JM (2010) Econometric Analysis of Cross Section and Panel Data. 2nd ed. Cambridge, Mass; London: MIT.

57.

World Bank (2020) Addressing corruption risks in the age of COVID-19: Preventing and managing conflicts of interest in the public sector. Available at: https://blogs.worldbank.org.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.81 MB