Voice, responsiveness, and alternative policy venues: An analysis of citizen complaints against the local government to the national Ombudsman

The Ombudsman is one of the most relevant and institutionalized tools to give voice to citizens, so they can express their dissatisfaction with local public services. The literature has yet to produce a systematic explanation for the variation in complaints to the national Ombudsman. This article uses classic literature from public administration and public policy to consider arguments of voice, responsiveness and alternative policy venues concerning the role of the Ombudsman in citizen complaints against local governments. Data on all 2139 citizen complaints submitted to the Ombudsman in Portugal (2012 – 2015) are analyzed using negative binomial regression models. Policy implications drawn from the results show not only that the presence of alternative voice venues reduces the number of complaints to the national Ombudsman but also that context matters. Local politics stimulates civic engagement and difficult local conditions, such as unemployment, crime, and aging populations, activate citizens to resort to the Ombudsman.


| INTRODUCTION
The establishment of formal complaints mechanisms and a rapid increase in Ombudsman institutions are characteristics of a trend in administrative reforms in the 1980s and 1990s that sought to improve the delivery of public services (OECD 2001;Overman 2017;Schläpfer 2017;van de Walle 2018).The Ombudsman is one of the most relevant and institutionalized mechanisms to give voice to citizens.Voice is the process through which citizens (individually or collectively) express their dissatisfaction with public services through speaking out.According to van de Walle (2018), they can do so in different ways: by complaining directly to a service provider during the service interaction or afterwards, complaining through media and, nowadays, social media or by addressing the complaints to those delivering the service, those managing it, those politically responsible for the public service or higher authorities.
A burgeoning literature in public administration investigates the factors that affect citizen satisfaction at many levels (e.g., DeHoog et al. 1990; van de Walle 2018).The majority of the studies to date look at local complaints made directly to local public services or tend to study the use of surveys of individuals to assess satisfaction or dissatisfaction with service delivery (DeHoog et al. 1990).In many of these studies, satisfaction is assessed with either perception-based data or self-reported measures rather than objective measures such as complaints to an external body.Others extensively analyze the role of Ombudsman institutions in general (Hossu & Dragos 2013;Ladi 2011) or in managing complaints against all types of public services (Magruder et al. 2018;van Roosbroek & van de Walle 2008) and do not particularly address complaints against local governments.Interestingly, this literature has yet to produce a systematic explanation for the variation in complaints to the national Ombudsman.This article relies on classic arguments from public administration and public policy, including the exit-voice-loyalty (EVL) framework, to develop such an explanation and analyze the role of the national Ombudsman as a voice solution of last resort for citizen dissatisfaction with their local governments.
Complaints are the result of citizen experiences with "inappropriate or discourteous treatment, omissions, mistakes, faults, inconsistencies, misleading guidance, unclear procedures, or displayed bias or injustices" in public services (Brewer 2007, p. 550).Hence, the inability or unwillingness to respond to negative citizen perceptions are likely to result in complaints, starting at the local level of government and moving to other venues 1 if this level proves to be unresponsive.Following Schattschneider's claim, losers in one particular policy venue may seek to expand the scope of conflict in the hopes of winning at a different venue (Schattschneider 1960).The presence of additional venues of political representation provides alternative channels to accommodate citizen dissatisfaction with municipal governments.The national Ombudsman is one such alternative channel and the degree of variation in citizen complaints against their respective local governments to the Ombudsman is one avenue of research worth pursuing.
Ombudsman complaints can contribute to increase accountability in local government.Brewer (2007) argues that the opportunities to use complaints as administrative appeals outside the local sphere can serve to bolster the legitimacy of the political system (Aldons 2001), and may even provide some compensation for the absence of democratic legitimacy (Scott 2005) by bringing about an alternative policy venue.Complaints to the national Ombudsman become a mechanism, albeit limited, for balancing the power between the local state and the ordinary citizen and a means of accountability (Devereux & Weisbrod 2006).
In order to expand research and knowledge about the role of Ombudsman institutions, we collected data on all citizen complaints submitted to the national Ombudsman of Portugal between 2012 and 2015 filed against the municipal governments (2139 complaints in total).These data are part of a larger project designed to assess the quality of local government in Portugal.Data are analyzed using negative binomial regression models to test our hypotheses.
After this introduction, Section 2 explores the literature and prior research examining the determinants of citizen satisfaction with local public services.The subsequent section discusses how the concepts of EVL can be combined with ideas about responsiveness and alternative venues of political representation to provide an explanation of the role of the Ombudsman in accommodating citizen dissatisfaction with local governments.Section 4 presents the hypotheses to be tested.Section 5 describes the data and methods employed in the empirical analysis.Section 6 discusses the findings and Section 7 concludes.

| PRIOR RESEARCH ON THE DETERMINANTS OF CITIZEN SATISFACTION
Satisfaction with public services has become a highly relevant field of study for public administration scholars (van de Walle 2018).When stated preferences are translated into revealed preferences, through formal mechanisms, they become critical behaviors to explain the performance of public services (van de Walle 2016).Many studies have tried to understand citizen complaints against local governments and explain the dissatisfaction with the provision of local public services (e.g., Kosecik & Sagbas 2004), particularly related to the environmental sector (e.g., Carvalho & Fidélis 2009;Zeng et al. 2019) or health care (e.g., Bloemen et al. 2015;Schlesinger et al. 2002).
The improvement of complaints procedures has also been analyzed, given its importance in building trust in and loyalty towards governments while reducing costs (e.g., Brewer 2007).In fact, in recent years, several governments have attempted to use technology-based complaints systems (Brewer 2007).These voice mechanisms are considered to be vital in improving accountability in the public services (Brewer 2007;Grossman et al. 2017).
The literature also points to critical factors that determine the territorial pattern of public complaints, particularly against municipalities and local services.Most factors are linked with the socio-economic characteristics of residents, including: 1. income (e.g., Bonaiuto et al. 1999Bonaiuto et al. , 2003Bonaiuto et al. , 2006;;Izazola et al. 1998;Kosecik & Sagbas 2004), with citizen complaints generally higher in more economically developed areas (Zeng et al. 2019); 2. literacy rate (e.g., Carvalho & Fidélis 2009;Kosecik & Sagbas 2004), with lower educational levels associated with fewer complaints (Jilke & Van de Walle 2013); 3. individual features, such as age, sex, race, and occupation (e.g., Brown & Coulter 1983;Kosecik & Sagbas 2004) also related to different levels of public participation (e.g., an elderly population and higher unemployment rates usually correlate with higher numbers of complaints); 4. length of residence in the locality and housing tenure, with homeowners and people who have lived in the locality for a long time more inclined to complain about local government's services (e.g., Kosecik & Sagbas 2004).
Apart from these socioeconomic factors, there are also individual-level explanations for citizen satisfaction with public services (DeHoog et al. 1990).DeHoog et al. (1990) argued that these are related to an attitudinal perspective, linked to an array of political behaviors (e.g., protesting and rioting as proposed by Sears and McConahay (1973), exiting the jurisdiction as defended by Tiebout (1956) or Ostrom et al. (1961), or contacting behaviors as studied by Sharp (1984)).
Other studies focus on the jurisdiction level highlighting the effects of different configurations of local governments in a region (DeHoog et al. 1990).Consolidation advocates argue that larger local governments are better equipped to respond to the complexities of urban life and present citizens with clearer lines of accountability (Lyons & Lowery 1989).In contrast, the enthusiasts of polycentrism stress that regions with multiple, smaller local governments provide plural exit options to citizens and tend to experience higher levels of service performance and accessibility as a result (Bish & Ostrom 1973).This debate is primarily interested in determining which configuration is capable of producing more "competent, responsive, and cost-effective local governance" (Carr & Tavares 2014, p. 270).This multiplicity of arguments and perspectives on citizen satisfaction and dissatisfaction with local governments' actions and the diversity of explanations on how to understand complaints, still presents a challenge to researchers.In this article, we focus on how the mechanisms of voice, responsiveness and alternative policy venues interact with contextual variables to account for the variation in the number of complaints against local governments brought by citizens to the national Ombudsman.The following sections detail these arguments and derive the hypotheses to be tested in the empirical section.

| EXIT, VOICE, AND ALTERNATIVE POLICY VENUES
Hirschman's (1970) exit-voice-loyalty (EVL) framework has proved to be a very useful way of categorizing behavioral reactions, and it is increasingly being tested in a public service context (Dowding & John 2012).Here, we extend the EVL framework by also including arguments related to local government responsiveness and alternative policy venues to explain the variation in complaints to the national Ombudsman.This extension must be able to accommodate two different and possibly competing reasons for the variation in complaints.On the one hand, it should explain the conditions likely to trigger citizen dissatisfaction.Adverse social contexts related to crime and unemployment, for example, may generate an increase in complaints to the Ombudsman if citizens perceive inaction or inability of local government to respond to these problems.On the other hand, the explanation should also account for the different types of policy venues at the disposal of citizens that may work as alternatives to the national Ombudsman.
The reasoning is that citizens will complain to the Ombudsman only if there are not alternative venues at the local level at their disposal or if these venues, while present, are unresponsive.Following Hirschman's classic work, voice has been primarily associated with politics, whereas exit is thought of as a market mechanism.Traditionally, if voice is seen as an active and constructive way to communicate an opinion, exit is regarded as a silent and destructive mechanism that can only be assessed after the fact (Peeters et al. 2020;Pierre & Røiseland 2016).Recent work, however, suggests that exit has a place in democratic theory and can be equally constructive for empowering citizens and institutionalizing choice in democratic systems (Warren 2011).
Complaints-the focus of our work-are considered as a form of voice as citizens express dissatisfaction with public service provision by using existing complaints systems to communicate their opinions (Pierre & Røiseland 2016).Voice is an active and constructive response, as it has the advantage of providing feedback to those responsible for service delivery, thus contributing to organizational responsiveness and possible improvements (Hirschman 1970;Peeters et al. 2020).Furthermore, in systems where residents face limited exit options, either due to the absence of institutionalized forms of exit or because individuals are not empowered to exercise exit options (Warren 2011), voice becomes an important mechanism to express discontent and citizens are more likely to get involved in local politics (DiPasquale & Glaeser 1999).
In contrast with voice, exit was traditionally regarded as a destructive mechanism, because it entails severing ties with a service provider (Dowding & John 2012;Peeters et al. 2020): internal exit (switching to another public provider), private exit (switching to a private provider) or geographical exit (moving to another local jurisdiction).Democracies institutionalize exit in many forms, including through interparty competition in electoral systems, horizontal competition between local governments, competing associational memberships, and (quasi-) market competition for local public goods (Warren 2011).
According to Hirschman's definition, loyalty drives citizens to remain in a situation, with "the expectation that someone will act or something will happen to improve matters" (p.78).In this sense, loyalty bears some traits in common with neglect, which entails "putting in less effort, not working at a relationship, and letting it fall apart" (Withey & Cooper 1989, p. 522;Golden 1992).However, in contrast with neglect, loyalty can be cultivated by organizations in order to prevent the exit of their most committed members, who are better able to promote reforms (Warren 2011).
Thus, citizen complaints against local governments to the national Ombudsman can be interpreted in light of the EVL framework.First, complaints are a form of voice, as they serve to communicate citizen dissatisfaction with the local government and "provide corrective feedback that can help to get faltering institutions back on track" (Sharp 1984a, p. 68).In the presence of an unresponsive local government, citizens are incentivized to search for alternative venues to express voice.The Ombudsman offers one such venue, particularly if those available at the local level fail (e.g., voting to replace an incumbent with a poor performance).
Second, beyond a voice role, the national Ombudsman can be seen as an institutionalized form of (quasi-)exit (Warren 2011).Individual decisions to complain to an external body (the Ombudsman) serve to elicit responsiveness from local governments, while avoiding the destructive effects of actual relocation to another municipality.In other words, citizens may complain to the Ombudsman not only to voice their dissatisfaction with an unresponsive local government, but also to signal an intention to exit if conditions fail to improve.
Third, Stoker (2011, p. 21) argues that "one of the most established functions of local governments is the expression of identity", which is stronger in Southern European countries such as Italy and France.Citizens are connected to their local governments by a strong sense of identity and attachment to place.Identity claims provide a source of legitimacy (Stoker 2011) and of loyalty that serve as a mechanism to suppress exit and encourage voice (Dowding & John 2008).Furthermore, citizens expect local governments to be more effective than upper level governments so they begin by placing demands on lower levels of government before moving up.The corollary of this argument is that the Ombudsman is a venue of last resort, that is, citizens will complain to the Ombudsman when other local alternatives are unavailable or unresponsive and to avoid hard exit (relocating to another municipality).
Lastly, local citizens may also rely on other forms of institutionalized exit.Competitive party systems are one such form and "votes function as empowerments just to the extent that voters can exit one candidate or party in favor of another" (Warren 2011, p. 692).When geographical exit is difficult or unlikely, as in the case of those who are "socially excluded" (Dibben 2006), this form of institutionalized exit can be an effective way of communicating dissatisfaction with local services, particularly if combined with the alert mechanism described by Laver (1976).
Increasing numbers of complaints may operate as an alert mechanism to the fact that the quality of local government services is declining, thus forcing local officials to act in order to prevent exit, in whichever form (Hirschman 1970;Laver 1976).
The loyalty mechanism combined with competitive party systems reduce the likelihood of geographical exit (Deng et al. 2007) and are likely to increase the use of voice and other institutionalized forms of exit, such as those described by Warren (2011).Our explanation extends the EVL logic by arguing that these mechanisms interact with the level of responsiveness and the number of alternative policy venues available at the local level to produce the variation in the number of complaints against local governments to the national Ombudsman.With this extension in mind, we now turn to the hypotheses of this study.

| HYPOTHESES
Citizen complaints to the national Ombudsman are the product of a local government's failure to address local problems experienced by citizens.Citizen dissatisfaction with local government service delivery may emerge from adverse local conditions such as high crime rates (Daniele & Marani 2011;Engel et al. 2012), low economic performance and persistent unemployment (Hickman & Piquero 2009;Worrall 2002), environmental degradation (Dong et al. 2011;Zeng et al. 2019), police abuse of power (Cao & Huang 2000;Hickman and Piquero, 2009) or inadequate government response in general (DeHoog et al. 1990;White & Trump 2018).Widespread problems such as these are likely to be associated with poor local government responsiveness and one reason for complaining to the national Ombudsman.
In the absence of these negative circumstances, citizens will be less likely to complain and, even if they do, local governments will be in better shape to respond to their demands.In the EVL framework, complaints are a way of exercising voice when citizens experience poor responsiveness from local governments.Hence, we expect that: H1.Adverse local conditions increase the number of citizen complaints against local governments directed to the national Ombudsman.
Pluralism, competitive elections, and high turnout are indicators of a vibrant civil society and active citizenship (Lowndes et al. 2002).They are the foundation of exit-based empowerment that manifests itself in the ability to vote for a non-incumbent to "exit one representative relationship in favor of another" (Warren 2011, p. 692).When combined with adverse local conditions, these institutionalized forms of exit are likely to generate higher levels of complaints.On the one hand, a more competitive and adversarial style of local politics means that problems are more likely to remain on the political agenda (Kelleher and Lowery, 2004) and discontent more likely to be confronted in multiple ways, including using alternative policy venues (Schattschneider 1960).Heightened competition should increase the number of complaints directed to the national level, with a more adversarial style of politics operating as a stimulus for local citizens to voice discontent.Political parties are especially willing to grant access to those citizens more capable to provide knowledge and expertise about specific issues (Chaqués-Boanfont & Muñoz 2016).
This willingness can improve party mobilization efforts, enhance access to the policy making process, or even increase contributions to political campaigns (Baumgartner et al. 2009).The increase in interparty competition and political action "effectively multiplies the opportunities for exit, and thus the incentives for parties to respond to constituents" (Warren 2011, p. 692).As such, H2.Higher levels of electoral competition at the local level are associated with a higher number of citizen complaints against local governments addressed to the national Ombudsman.
On the other hand, complaints can be understood as complementary ways to other, more common public participation mechanisms, such as voting in local elections (Vedlitz 1980).Hirschmann (1970) stated that those more likely to exit public services (from middle and upper classes) are also those most able to effectively vocalize their complaints.This means that their voice and votes are likely to be important within the democratic system to make visible their satisfaction/dissatisfaction (Dowding & John 2012).It is also expected that the refusal to participate in local elections might be associated with lower levels of complaints to express dissatisfaction.Contrariwise, H3.Higher voter turnout at the local level is associated with a higher number of citizen complaints against local governments addressed to the national Ombudsman.
In contrast, when the local political scene is dominated by a political monopoly favoring the maintenance of a status quo, local governments are less likely to face dissent (Trounstine 2009).The reelection of incumbents is often associated with local political stability and lower levels of information and participation (Trounstine 2006(Trounstine , 2011) ) and fewer opportunities to express dissatisfaction locally.When dissatisfied citizens are unable to change the political monopoly, they may pursue a strategy of searching for alternative venues to express discontent, by complaining to the national Ombudsman.Conversely, incumbent reelection can also be as a sign of approval from the electorate (Krebs 1998), indicating better performance by the local government and lesser need for alternative venues outside the local community to express dissatisfaction.These arguments lead to alternative hypotheses: H4a.(political monopoly hypothesis): Local governments where the incumbent is reelected will register a higher number of citizen complaints addressed to the national Ombudsman.
H4b. (responsiveness hypothesis): Local governments where the incumbent is reelected will register a lower number of citizen complaints addressed to the national Ombudsman.
Voting for a nonincumbent candidate in a multiparty competitive election is one form of institutionalized exit.
However, multilevel governance offers additional opportunities for citizens to express discontent with local governments.Other subnational tiers of government above or below municipalities constitute alternative venues to voice discontent without resorting to Tiebout-type exit.In other words, multilevel governance institutionalizes different combinations of voice and exit mechanisms as part of a democratic system to secure communication of dissatisfaction with service delivery and avoid silent exit (Warren 2011).On the one hand, in decentralized systems, supramunicipal levels of government such as regional or metropolitan governments may function as alternative policy venues capable of absorbing citizen discontent and respond to citizen demands (Demazière 2020).Therefore: H5.The presence of an intermediate level of regional government decreases the number of citizen complaints against local governments directed to the national Ombudsman.
On the other hand, citizens may voice dissatisfaction to neighborhood or sub-municipal units of government (SMUs) that are geographically and psychologically closer to them.In municipalities with formal SMUs, citizens may pressure those to start a dialogue with the municipal administration (Tavares & Teles 2018).Citizens expect SMUs to lobby the municipal executive for a response to localized problems (Hlepas et al. 2018).Furthermore, SMUs can offer choices for citizens to relocate to a preferred neighborhood inside a municipality without incurring the costs of moving to a different jurisdiction.In other words, SMUs function as internal exit, a democratic form of empowerment that signals preferences to local officials while avoiding the destructive response associated with geographical exit to another municipality.Hence, we expect that: H6.The higher the number of sub-municipal units of government in a municipality, the lower the number of citizen complaints against local governments directed at the national Ombudsman.
The following section describes the data and methods employed in the empirical analyses to test the hypotheses developed above.

| DATA AND METHODS
The national Ombudsman in Portugal is an independent body, supported by article 23 of the Portuguese Constitution and Law 9/91, April 9.The Ombudsman is appointed by a qualified majority of the members of the National Parliament.She/he is mandated to "receive complaints of all (natural or legal) persons who feel harmed by unfair or illegal act's by public administration or when their fundamental rights are violated" (http://ennhri.org/Portuguese-Ombudsman).One of the primary targets of the complaints submitted to the Ombudsman in Portugal are local authorities, which also tend to be less forthcoming in responding to the Ombudsman's requests in a timely manner (Ombudsman Annual Report to the Parliament, 2016).
As part of a larger project intended to assess the quality of local government in Portugal, we collected data on all citizen complaints filed against municipal governments to the national Ombudsman between 2012 and 2015.A total of 2139 complaints were filed during this period.Currently, the number of municipal Ombudsmen is negligible; only 11 out of 308 municipalities have a local Ombudsman.Of these, five have been in place since 2016, but the complaints figures in the analysis predate this recent increase.
Table 1 includes a list of all complaints by theme and sub-theme.One third of the complaints concern land use planning (33%), including primarily complaints about public infrastructure (21% of the total).Urbanism and housing (33%) represent another third, particularly building works (13%) and social housing and housing support (5%).The environment and natural resources represent the remaining part (31%) with a great percentage associated with noise problems (16%).
The majority of complaints concern local public goods or services that cannot be "bought" elsewhere in the market.
The dependent variable is the total number of complaints to the national Ombudsman per municipality.The regression models control for the size of the municipality by including population as an independent variable.The dependent variable has the advantage of being an outcome variable rather than a perception-based variable, typically employed by earlier studies to assess satisfaction with service delivery (Overman 2017; Schläpfer 2017).
T A B L E 1 List of complaints to the ombudsman (by theme) (Continues) In order to test the first hypothesis concerning the effect of local adverse conditions on the number of complaints, we employ a series of "usual suspects" (e.g., Bonaiuto et al. 2006;Worrall 2002).Higher levels of crime, unemployment, and elderly citizens in the community are likely to place pressure on the responsiveness levels of local government and, all else equal, contribute to a higher number of complaints.Crime is measured as the rate of reported crimes per 1000 inhabitants, unemployment is the percentage of unemployed individuals in the active population in the municipality, and elderly population is assessed as the proportion of citizens 65 years-old or above.Communities that are less educated are also less likely to express dissatisfaction in multiple ways (Zeng et al. 2019), including complaints to the national Ombudsman.We employ the proportion of resident population with a bachelor degree as a measure of education levels.
The level of financial autonomy is also included in the analyses to control for the capacity of local governments to address some of these pressures.It is measured as the proportion of own revenues collected by the municipality.
In order to test Hypotheses 2-6, we employ several variables.First, competitive party systems offer excellent opportunities for combining voice in the political process with alternatives for exiting one representation relationship for another.Three variables account for these mechanisms.The level of electoral competition is measured using the margin of victory as the average difference between the first place and runner up in the last three mayoral elections.
Voter turnout is measured as the average proportion of citizens who voted in the previous three mayoral elections.
Hypothesis 4 is tested using a dummy variable taking the value of "1" if the incumbent party remained in office.
Second, the combination of voice with institutionalized forms of exit can also occur in the context of multi-level governance.We employ two variables to account for the interactions between these mechanisms.On the one hand, the existence of regional governments in both the Azores and Madeira archipelagos is likely to operate as a buffer, reducing the number of complaints to the national Ombudsman.In contrast, since continental Portugal does not have regional governments, citizens do not have an intermediate level of government to whom to take their grievances against local governments.Given that the archipelagos are more distant (geographically and culturally) from the head of the national government, it is more likely that they will have fewer complaints to the National Ombudsman.Nevertheless, there are deconcentrated offices of the National Ombudsman in both the Madeira and Azores archipelagos, so distance is not a factor to explain potential differences in complaints in these regions.The reason for fewer complaints might be therefore the presence of additional venues of political representation, which provide alternative channels to accommodate citizen dissatisfaction with municipal governments.This argument receives further relevance in a highly centralized country as Portugal.A dummy variable is included for all island municipalities to test Hypothesis 5.
On the other hand, SMUs of government foster closer proximity between citizens and elected officials and can operate as lobbying institutions at the local government level (Tavares & Camões 2007).A higher number of SMUs per municipality offers additional opportunities for citizens to "vote with their feet" inside the municipality, leading to a lower number of citizen complaints to the Ombudsman.The number of Parishes (the SMUs in Portugal 2 ) in a municipality is the variable chosen to capture this effect.Lastly, a dummy variable is included for the presence of mayors elected in nonpartisan (independent) lists.Given the current trend in Portuguese local elections towards the election of nonpartisan mayors, this can capture some of the dissatisfaction with local partisan politics (Tavares et al. 2020).
Table 2 contains the variables, indicators, and sources of all the data employed in the analyses.It also includes the expected signs of the coefficients.Table 3 presents the descriptive statistics for all the variables included in the analyses.
Data on the complaints to the Ombudsman are a count variable and follow a Poisson distribution.The Poisson distribution assumes that the mean and variance are the same (equidispersion).Unfortunately, our raw complaints data suffer from significant over-dispersion, as evidenced by a variance much larger than the mean (see Figure 1).
Given the over-dispersion in our data, we employ negative binomial regression to estimate the models, as it is a more flexible solution than Poisson regression.Given the data structure, using Poisson regression in this case would result in biased standard errors.The negative binomial distribution has one parameter more than the Poisson regression that adjusts the variance independently from the mean (Long 1997).

| FINDINGS AND DISCUSSION
Table 4 depicts the results of the negative binomial regression analyses of citizen complaints against the local government to the national Ombudsman.The first specification (1) includes the independent variables related to Hypotheses 2-6.The second specification (2) includes all the variables describing the local conditions that may lead citizens to place complaints to the national Ombudsman (Hypothesis 1).Column (3) contains the full model.In order to facilitate interpretation, column (4) displays incidence rate ratios (IRR) for the variables included in the full model.Results are largely in line with the expectations.The variables selected to test the effects of local conditions perform according to Hypothesis 1. Crime, unemployment, and the proportion of elderly population are all positively associated with higher levels of complaints to the national Ombudsman and their coefficients are statistically significant at conventional levels.Additionally, education is also positive and significant, confirming prior empirical findings (Jilke & Van de Walle 2013;Zeng et al. 2019).Figure 2a-c, display 95% confidence intervals for the relationship between elderly population, education, and crime rate and the predicted number of complaints to the national Ombudsman.Figure 2d shows the combined effect of crime and unemployment on the predicted number of complaints.Clearly, both variables are associated with increases in the number of complaints, but their combined effect is linked to the highest number of complaints: the difference between the highest and lowest unemployment lines in the graph is smaller at the lower levels of crime rate and much larger at the highest levels.In practice, this means that municipalities experiencing a combined effect of high crime and unemployment rates are likely to face high Margin of victory is negatively associated with the number of complaints to the national Ombudsman, thus providing support for the second hypothesis.It indicates that increased political competition at the local level translates into higher numbers of complaints to the national level.Similarly, turnout rate displays a positive and statistically significant coefficient.In municipalities where turnout rates are higher, complaints to the Ombudsman tend to be higher as well.This result is quite robust and backs Hypothesis 3.Both findings suggest that citizens may regard voting and political competition as complementary ways to complaints to express dissatisfaction with their local government through different forms of institutionalized exit.Figure 3a provides visual support for this idea, showing higher numbers of complaints in municipalities with more competitive elections and higher turnout rates.
The presence of an incumbent mayor is positively associated with the number of complaints to the national Ombudsman.This result fails to support our expectation in H4b that incumbency is a form of reward for good responsiveness levels.Instead, it supports H4a (political monopoly hypothesis) suggesting that when citizens are unable to oust the incumbent party, they search for other venues to voice discontent.Losers at the local venue seem to resort to the national Ombudsman as an alternative venue.The positive effect of the incumbent variable is supported regardless of turnout levels and the margin of victory of the mayor (Figure 3b,c).
One of the most robust results of the analysis is the coefficient for island municipalities (H5), which is negative and statistically significant across the board.Clearly, municipalities in the Azores and Madeira archipelagos display a much lower number of complaints than the municipalities in continental Portugal.The presence of an intermediate level of government is likely to be acting as a cushion for the complaints presented to the national level, providing an alternative institutional venue for citizens to express their views on policy and public services.The difference between island municipalities where citizens have access to a regional level of government (as well as to regional demands.This evidence is still tentative and it will require additional fieldwork to reach a more definitive conclusion.
The argument of proximity in political representation fails to receive support from the SMU variable.Having more SMUs is not associated with lower numbers of complaints, suggesting that SMUs are not being used to connect citizens to their local government.The presence of an independent mayor displays a negative and statistically significant coefficient.As the number of nonpartisan mayors is growing across the country, Portuguese citizens are complaining more in municipalities run by partisan mayors.
The coefficients in column (3) are not very informative, as they report the difference between the log of expected counts.Column 4 contains the incidence rate ratios (IRR) for our full model, obtained by exponentiating the negative binomial regression coefficients.The IRR allow us to report the percent change in the number of complaints per a 1-unit change in a given independent variable.
In substantive terms, local conditions play an important role in increasing the number of complaints to the national Ombudsman.The percent change in the IRR of the number of complaints is a 32.9% increase for every unit increase in the unemployment rate, a 1.5% increase for every unit increase in education and a 3.9% increase for every unit increase in elderly population.Similarly, if the crime rate in a municipality increases by 1-unit, its IRR for the number of complaints to the Ombudsman is expected to increase by a factor 1.011, while holding all other variables in the model constant.Taking into account the differences in measurement units, this result confirms the The stronger finding regarding the political variables is the overwhelming impact of the "islands" dichotomous variable.As an illustration of this effect, the IRR for island municipalities is 0.063 times the IRR for continental municipalities, while holding the other variables constant in the model.All the remaining effects are smaller than this, but some are quite substantial and worth reporting.For example, the percent change in the incident rate of the number of complaints is a 0.57% decrease for a 1-percentage point increase in the margin of victory and a 2.9% increase for every unit increase in the turnout rate.Both effects are statistically significant and display a substantial impact on the number of complaints to the Ombudsman, but their effects are smaller when compared to being a municipality under a regional tier of government.Three policy implications can be drawn from the results.First, we find that the presence of alternative voice venues reduces the number of complaints to the national Ombudsman.This underlines the importance of providing additional venues at sub-national levels which are able to absorb and solve most of the grievances experienced by local citizens.On the one hand, initiatives to implement these additional venues are likely to be more effective in addressing the complaints, given the proximity to the citizens experiencing dissatisfaction.On the other hand, these sub-national venues will also reduce the pressure placed on the national Ombudsman, freeing institutional capacity to tackle national level complaints.

| CONCLUSIONS
Second, it is commonplace to argue that context matters.In this particular case, context is relevant in two ways.
Local politics affect the likelihood that complaints will have a national impact.A more active involvement in politics, either through stronger interparty competition or higher electoral participation, is associated with a higher number of complaints to the national Ombudsman.Additionally, if the incumbent wins the local election, the likelihood that the conflict expands to the national level also increases, as shown by the higher number of complaints in municipalities where the incumbent has won.Moreover, if local politics stimulates civic engagement in a variety of ways, an intermediate level of government may function as another level to exhort this stimulus, as suggested by the findings, reaffirming other calls in the country for a more decentralized model of government towards a stronger multi-level governance system.
Third, the importance of context is also visible in the effect of the "usual suspects" upon the number of complaints to the national Ombudsman.Difficult conditions experienced at the local level-unemployment, crime, and aging populations-generate more complaints.While this is not a surprising result, it suggests that citizens resort to an upper level government when their municipal executive is unresponsive to their concerns regarding these adverse contexts.Local autonomy and institutional capacity at the local level should be addressed as tools to empower local authorities to address these wicked problems.
Despite these implications, the research is not without limitations.The major limitation stems from the absence of individual level data, which is protected due to privacy laws.Ideally, we would be able to estimate a multi-level model with both individual level and institutional level predictors to assess their relative effects in explaining the variation in the number of complaints to the national Ombudsman.A second limitation originates from the limited

F
I G U R E 2 Predicted number of complaints for different levels of socioeconomic variables.(a) Elderly population; (b) education; (c) crime rate; and (d) crime and unemployment rates delegations of the National Ombudsman) and continental municipalities where they do not have this option is staggering, as witnessed in Figure 3d.This can be seen as evidence of the benefits of having this intermediate level of government in a centralized country like Portugal, enabling alternative policy venues capable of responding to citizen

F
I G U R E 3 Predicted number of complaints for different levels of political variables.(a) Voter turnout and margins of victory; (b) margins of victory and incumbency; (c) voter turnout and incumbency; and (d) voter turnout and location of municipalities substantive impact of crime and unemployment as driving factors of citizen dissatisfaction and therefore the higher number of complaints.
This article investigated the determinants of citizen complaints against local governments placed to the national Ombudsman in Portugal.Prior works have employed perception-based surveys to highlight individual factors affecting citizen satisfaction levels with local governments.While recognizing the importance of these earlier efforts, here we focused primarily on political factors influencing varying levels of citizen complaints.These factors interact with contextual variables to account for the complaints against local governments brought by citizens to the national Ombudsman and feed the debate regarding the institutional configurations capable of fostering more responsive and effective local governance.
Note: Bold values indicate higher values and proportions of complaints.
Municipal complaints to the National Ombudsman, 2012-2015 1064 TAVARES ET AL. 14679299, 2022, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/padm.12787 by Cochrane Portugal, Wiley Online Library on [28/03/2023].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)onWiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License discontent.Unsurprisingly, the findings provide strong support to H1: local conditions are positively associated with the number of complaints to the national Ombudsman.The level of financial autonomy of Portuguese municipalities is positively associated with the number of complaints to the national Ombudsman.This is a puzzling result, as we expected that increased autonomy would result in more capacity to address citizen concerns.Instead, our models may be capturing citizen dissatisfaction with the level of fiscal effort required by the local governments to maintain higher levels of financial autonomy.T A B L E 4 Negative binomial regressions (Dep.Var: # complaints)