THE ROLE OF MEDIA PRESSURE IN PROMOTING TRANSPARENCY OF LOCAL GOVERNMENTS

Abstract This paper analyses the role of media pressure on the level of public transparency of local governments. Making use of a sample composed by 110 Spanish cities for the 2008-2010 period, the results show that local governments tend to be more transparent under media pressure, especially in relation to information about economic, fi nancial, urbanism and public works issues. Additionally, we take into account the electoral moment, since the analysis of the ways in which public affairs are conducted is still superior when elections are approaching. Empirical results indicate that media plays a fundamental role in that moment because governments are willing to decrease the level of information transparency.


Introduction
As a consequence of several fi nancial scandals, citizens' interest is focused on the behavior of the public sector, especially on topics such as accountability and transparency.Public demand for transparency in public institutions is increasing constantly (Piotrowski and Van Ryzin, 2007).Citizens, as providers of resources, require information in relation to the activities that are defrayed by their resources (Guillamón, Ríos-Martínez and Vicente-Oliva, 2011), knowing where they are allocated and how they are being used (Jorge et al., 2011).Indeed, good governance is fi rstly assessed by citizens in terms of transparency in decision-making and policy implementation (Sharman and Chaikin, 2009).
Information opacity enables corruption in public bodies, but this could be avoided with higher levels of information transparency (Hood, 2001).In Spain, information opacity has led to several corruption cases (e.g.irregular adjudications, capturing illegal subsidies, kickbacks within political parties, etc.), especially at the local level.Here, one of the main problems for the identifi cation of these cases is the lack of protective laws for informers, as Spain was one of the few European countries that do not had a public transparency law, until December 2013.At present, Law 19/2013, of December 2013, on transparency and access to public information and good governance, requires all public administrations and public entities to actively publicize their activities.
In absence of specifi c legislation to ensure access to government activities in a clear, permanent and timely fashion, media could be seen as a tool to keep citizens informed about the daily administration of the local government (Gandía and Archidona, 2008).Media provide information to citizens and incite the government to promote accountability (Macdonell and Pesic, 2006).The coverage given to organizations by the media is crucial to the image formed of them within society (Carroll and McCombs, 2003).Thus, it could be expected that higher media pressure push governments to act more responsibly (Besley and Burgess, 2002).This is our research question, which is empirically tested by using a sample composed by 110 Spanish municipalities, whose information on levels of transparency has been published by Transparency International Organization Spain 1 .
Microscopic analysis of the public sector by the media is still superior in election periods with the aim of informing citizens about the management capacity of the ruling party, and in the face of possible re-election.Media plays a fundamental role in transmitt ing information to mass audiences, so they could be interpreted as a way to make available information that could be used by citizens in the voting process (Strömberg, 2004).Accordingly, this study also takes into account the electoral moment in testing the role of media on information transparency of local governments.
The remainder of this paper is divided into the following sections.In the second section, we review previous literature and present the working hypothesis.The third section contains the methodology.The fourth section presents and discusses the research results and fi nally, we set out the main conclusions, implications and limitations.

The concept of transparency and its determinants
Although there is not universal defi nition of transparency, all of them hold the availability of information.The OECD (2002) defi nes transparency as openness on policy intentions, formulation and implementation.On the basis of OECD's defi nition, Armstrong (2005) defi nes this concept as the access by the public to timely and reliable information on decisions and performance in the public sector.Similarly, Alt, Lassen and Rose (2006) noted the overall degree to which citizens, the media and fi nancial markets can observe the government's strategies, its activities and the resulting outcomes.In short, transparency can be viewed as the degree of access to information about the government, being accessible on time to all relevant stakeholders (Kaufmann and Kraay, 2002), and allowing the knowledge of the ways in which public aff airs are conducted (Heald, 2006).
Previous studies are focused on the relationship between transparency and several socio-economic and political factors.For instance, Piotrowski and Van Ryzin (2007) found that factors such as age, political ideology, confi dence in government leaders, the income level, etc., aff ect the public's demand for transparency.Alt, Lassen and Rose (2006) showed that more equal political competition and power-sharing are related to greater levels of fi scal transparency, but political polarization is related to lower transparency.Meanwhile, Laswad, Fisher and Oyelere (2005) found that leverage, municipal wealth, press visibility and types of council are related to fi nancial disclosures of local governments in New Zealand.Jorge et al. (2011) showed a positive link between transparency and the size of Italian local governments, and a negative impact of voting participation on the level of transparency in the case of Portuguese and Italian local governments.
Transparency has been highly studied in Spanish local governments.For instance, Gandía and Archidona (2008) found that disclosure levels in Spanish local governments depend on political competition, public media visibility, access to technology and educational levels.Cárcaba-García and García-García (2010) show that size, capital investment and political competition are positively related to information disclosure by Spanish local governments, but they found a negative link between voluntary fi nancial disclosure and press visibility of Spanish local governments.Guillamón, Ríos-Martínez and Vicente-Oliva (2011) found a positive link between the level of transparency and taxes, and transfers per capita, population, and left-wing ideology.Guillamón, Ríos-Martínez and Vicente-Oliva (2011) also showed that government fragmentation is related to higher levels of transparency, but that the unemployment rate impacts negatively on levels of fi nancial transparency.The positive link between population and left-wing governments has been also found by Albalate del Sol (2013); additionally, he found a negative link between transparency and political strength, unemployment rate, and tourism index, and no infl uence of voters' participation and the level of economic activity.

Research hypotheses
Regarding previous literature, one of the determinants of public transparency is press visibility.Laswad, Fisher and Oyelere (2005) indicated that press visibility is positively associated with the internet fi nancial reporting practices of local governments in New Zealand, and similarly, Gandía and Archidona (2008) found that web disclosure levels in Spanish local governments are positively related to public media visibility.Although Cárcaba-García and García-García (2010) showed that press visibility is negatively linked to voluntary reporting of fi nancial information on the internet by Spanish local governments, Grimmelikhuijsen and Welch (2012) refi ned previous results, by using a sample of Dutch municipalities; they found that media att ention aff ected positively on a specifi c type of public transparency, the policy information.
In general, media pressure may make a political system more transparent in three ways (Balking, 1999): (i) it enables people to understand the operations of the government, (ii) it involves the public in policy decisions, and (iii) it makes the responsible offi cials accountable.These eff ects are shown thanks to an increase in information availability, through which citizens know operations and activities of governments and the use of resources that they provide to governments.In addition, the media provides information to citizens, inciting the authority to promote accountability (Macdonell and Pesic, 2006), so it would be expected that media might be a determinant of the level of municipal transparency.
The agenda-sett ing theory (McCombs and Shaw, 1972) defends that highlighted items of media agenda infl uence the agenda of society, thus society uses the most relevant news to decide what topics and people are the most relevant (Carroll and McCombs, 2003).The media are highly persuasive in focusing the public att ention on certain events and topics, determining the importance that people att ach to them (Shaw 1979).
Thereon, the media make more information available, and it is a powerful tool to push governments to act more responsibly (Besley and Burgess, 2002).This is highly important when information access right is under regulated.Citizens require information on the public activities fi nanced with resources they provide; and the access to such information should be considered a right.However, there is no universal law regarding transparency.Regulation 1049/2001 which forms part of European Union law could be an att empt to regulate this topic, but, in general, countries have developed their own national laws.In general, regulations on transparency are very recent, due to demands of information have specially grown from the global economic crisis of 2008.In the case of Spain, until 2013 there was not a specifi c regulation on access to public information, namely Law 19/2013 from 9 December 2013 on transparency and access to public information and good governance 2 .
In such situations, in which regulation is weak and government has discretionary power to disclose some information and do not disclose other information, media plays a fundamental role in transmitt ing information to mass audiences.As a consequence, we proposed the following hypothesis with the aim of testing the relationship between media pressure and public transparency on local government: H1 -Local governments tend to be more transparent when the media exert pressure on them.
Greater transparency creates clarity of responsibility by enabling citizens to access the decisions taken by the government, reliably and in good time (Armstrong, 2005).Transparency increases the visibility of government actions (Alt and Lassen, 2006), which is always important for citizens, but it is even more relevant when elections are close.Nordhaus (1975) proposed the pioneer model of political business cycles, in which ruling governments have incentives to manipulate economic variables in order to be re-elected.From this opportunistic point of view, the maximization of re-election probability is usually considered to be a political parties' main aim, which determines the ruling government's policies (Downs, 1957).In this context, governments could create favorable conditions in pre-election periods to infl uence the opinions of voters about their mandate.Favorable conditions refer to higher public spending or public debt -which results in more/bett er public services -and lower taxation; these policies improve the image of the ruling party and may increase the probability of re-election.
A way for citizens to know the ruling government activities is through information disclosure -i.e.public transparency.This idea is deduced from the Ferejohn's (1999) model, in which voters tend to allocate their resources to candidates that provide a more precise signal of the political agent, through information disclosures.Thus, it could be expected that ruling government increases transparency in the pre-election period with the aim of showing the 'favorable conditions' that they have created (such as increasing public spending, reducing taxes, etc.) and improve their image in the 'political market'.Following these arguments, we formulate the second hypothesis: H2 -The closeness of the election means that local governments tend to be more transparent.
However, the role of the media logic 3 is important, since media does not necessarily act on behalf of the public, but also media has fi nancial aims (Cárcaba-García and García-García, 2010).Following Mazzoleni and Schulz (1999), it has come to refl ect a commercial logic, including the typical aims of commercial communication activities.This has generated situations in which journalists' interpretations become negative (Hallin, 1992), and media coverage has risen from healthy skepticism to automatic negativity about governments.Probably, news about negative action (e.g., corruption cases) are more commercial than others.The consequence is the 'mediatisation'4 of governments, who are dependent on and shaped by the media (Mazzoleni and Schulz, 1999).
With fi nancial objectives, media usually prefers to publish news on cases of corruption and political scandals, on the basis on commercial logic.Such news has a negative eff ect on electoral support and damage ruling government re-election (Costas-Pérez, Solé-Ollé and Sorribas-Navarro, 2011); even the magnitude of electoral punishment depends on the media coverage.According to these arguments, the negativism of media and their att ention on political scandals create a negative image of the ruling government.Thus, we expect that they tend to entrench themselves and reduce their level of public transparency, such as we proposed in the following hypothesis: H3 -Media pressure reduces the eff ect of the electoral cycle on the level of transparency of local governments.

Sample
To test our research hypotheses we use a sample composed by 110 Spanish local governments in the period 2008-2010, because of the availability of information about the Spanish local governments' transparency on Transparency International Spain website.These local governments have a population over 50,000 inhabitants, and also include all province capitals 5 .This sample is adequate to test the hypotheses since these local governments are legally urged to promote citizens' participation in local governments and enhance information transparency, according to the Act 57/2003 of measures to modernize local government.

Model for analysis
In order to test the hypotheses, we use the following dependence models for panel data: The dependent variable (IT_INDEX) represents the level of local governments' transparency through the index published by Transparency International Spain, and it takes values between 0 and 100 (i.e. the higher the level of IT_INDEX, the higher the level of transparency).Additionally, this index in disaggregated into fi ve groups depending on the type of information to which they relate (IT_SUB-INDEX): (a) Information about the municipal corporation (e.g., biographical data on the mayor and councilors, contact addresses, salaries and remuneration, etc.); (b) Relations with citizens and society (e.g., city council webpage appearance, possibility of publishing administrative processes online, publishing public jobs, traffi c, environmental situation, discussion forums, etc.); (c) Economic and fi nancial information (e.g., the publication of the annual accounts of the city council, municipal budgets, audit reports and external control, data defi cit/surplus and income/expenditure per capita, amount of municipal debt, etc.); (d) Information about municipal service contracts (e.g., calls for tenders for bids, if the public is the supplier associated with each service, etc.); (e) Information about urban development and public works (e.g.information on urban planning and public works such as zoning plans and planning agreements, decisions on re-qualifi cations and urban adjudications, ads and bids for public works, information on deals and decisions, etc.) To test hypothesis H1, we use the independent variable called MEDIA, which represents the number of news items related to each local government over each year of the sample.This measurement has been previously used, for instance by Laswad, Fisher and Oyelere (2005), Gandía and Archidona (2008), Cárcaba-García and García-García (2010), and Grimmelikhuijsen and Welch (2012) to represent press visibility.The former made a count search on Newzindex; the last made a count search on local Dutch newspapers; the rest of scholars used the number of Google quotes for each local government by a search in 'google.com'.Due to the relevance of internet in the last decade, we consider that this technique is the most adequate, since most of the newspapers have an online version.Accordingly, to create the variable MEDIA, we conducted a 'Google News' search with the name of each local government in quotation marks as the search term.Using advanced search tools, we fi ltered the results by year, from 2008 to 2010.To obtain validity of our measure, the search was performed on both Google.comand Google.es,obtaining identical results.The Google News search allows us to determine the relevance of each local government to any newspaper, not just the specialized economic press, as local government behavior may be subjected to pressure from any medium.This broad selection allows us to identify all the media pressure on the local governments, taking into consideration pressures generated from diff erent interest groups, not only those specialized in economic issues.Google's search engine has been used by other authors in very diff erent areas of research (e.g.Reverte, 2009;Aguinis et al., 2012; Cuadrado-Ballesteros, Frías-Aceituno and Martínez-Ferrero, 2014; García-Sánchez, Cuadrado-Ballesteros and Sepúlveda, 2014).
To test the hypothesis H2, we considered the election proximity, using a dummy variable, called PREVIOUS YEAR that takes the value 1 in the year before elections and 0, otherwise.In this case, as the time period analyzed is 2008-2010, this variable takes the value 1 in 2010 and 0 in all other years, since local elections were held in 2011 in Spain.Finally, with the aim of testing the third hypothesis, we create the variable PREVIOUS YEAR*MEDIA, through the interaction between the two previous variables.This shows the eff ect that media pressure may exert as a moderator of the impact of the electoral cycle on municipal transparency level.
Finally, results are controlled by some political and socioeconomic factors: fi rstly, we consider the political ideology, through the variable LEFT, which is a dummy that takes the value 1 if the ruling party is of left ideology and 0, otherwise.Progressive ideology parties, who prefer a larger public sector (Borge, 2005), tend to be more transparent (Guillamón, Ríos-Martínez and Vicente-Oliva, 2011; Guillamón, Bastida and Benito, 2011) in gaining the trust of voters which are the providers of public resources (Ferejohn, 1999).
Furthermore, we use a Herfi ndal index to represent the political strength, calculated according to Ratt sø and Tovmo (2002), and Borge (2005) as, where n represents the number of political parties in local government, and p i represents the number of councilors of the party i in local government.This index takes values between 0 (maximum fragmentation) and 1 (maximum strength).Maximum fragmentation (0) implies that each councilor is of a diff erent party and maximum strength (1) indicates that all councilors are of the same political party.Roubini andSachs (1989a and1989b) showed that the problems of coordination in fragmented and coalition governments lead to higher defi cits than single party governments.This leads to less information transparency in order to avoid showing such internal problems to citizens.
Finally, to control the size of local government, we take into account population density (DENSITY), represented as the number of inhabitants per km 2 .In general, large cities present agglomerations, which can be considered to att ract large companies and businesses and be a symbol of modernity and economic power, but can also be considered to present large social and economic inequalities (Camagni, 2002).
Considering that our dependent variables take values between 0 and 100, we use the Tobit methodology for panel data.The random error term is decomposed into two parts: μ it , which varies among municipalities and over time, and the individual eff ect ε i , which characterizes the municipality and is invariant over time.These can be interpreted as the factors that are not included in the regression, and which are specifi c to each municipality.

Descriptive analysis
The descriptive statistics are shown in Table 1.We can see that the mean value of the global transparency index, IT_INDEX, is 62.45 (in the range from 0 to 100); this means that, in general, Spanish local governments show a medium level of transparency in the analyzed period.In particular, local governments tend to be more transparent about relationships with citizens and society, about municipal corporations, and about urban development and public works.The lowest level of transparency relates to economic and fi nancial information.subscript which reports on the municipal corporation; B refl ects information about relationships with citizens and society; C refl ects economic and fi nancial information; D refl ects information about the hiring of public services; E refl ects information about planning and works public; MEDIA represents the number of news related to each municipality; PREVIOUS YEAR is a dummy variable that takes the value 1 in the year prior to the election and 0 otherwise; PREVIOUS YEAR*MEDIA refl ects the interaction between the PREVIOUS YEAR and MEDIA.LEFT is a dummy variable that takes the value 1 when the ruling party is leftist ideology and 0, otherwise; STRENGTH refl ects the strength of local government policy through the Herfi ndal index calculated according to Rattsø and Tovmo (2002); DENSITY represents the inhabitants per km 2 .

Source: Authors' computations
Furthermore, the average amount of news for local governments in this period of time is 19,660, of which 9,902 news items are provided in the year before elections (mean value of variable PREVIOUS YEAR*MEDIA).Finally, about 42% of the sampled municipalities are governed by left-wing parties.The Herfi ndal index, which measures political strength, has a mean value of 0.48, in the range between 0 and 1.This mean value refl ects that, in general, councilors represent diff erent parties in local governments.In addition, population density is about 2,711 inhabitants per km 2 , although the standard deviation is high.The bivariate correlations are shown in Table 2.Although several coeffi cients are statistically relevant, any correlation between the dependent and independent variables or between the independent variables is large.

Explanatory analysis
Table 3 shows the empirical results.Concretely, we found a positive link between media pressure and the level of public transparency, which is relevant at a confi dence level of 95%.This means that when media pressure is strong, the level of municipal transparency is high, according to the hypothesis H1.Media is associated with a higher level of public transparency, favoring citizens' knowledge of how resources are managed (Gandia and Archidona, 2008); thus, the local political system is more transparent (Balkin, 1999).
In addition, the estimated coeffi cients of MEDIA and PREVIOUS YEAR are positive and statistically relevant at the 99% confi dence level.However, when we considered the interaction between these two variables, it is negatively related to IT_IN-DEX and it is relevant at 95% confi dence level.The coeffi cient of PREVIOUS YEAR indicates that in the year before elections, the level of transparency at the local level is higher, according to the hypothesis H2.This means that governments tend to be more transparent in pre-electoral periods with the aim of improving the image that voters have on their mandate and provide a more precise signal of their future policies (Ferejohn, 1999).
However, it is necessary to take into account that this relationship is changed by media pressure because the variable PREVIOUS YEAR*MEDIA has a negative coeffi cient.This shows that local governments tend to be less transparent when media pressure is strong in the year before elections, in line with the hypothesis H3.Media is a great mechanism for citizens to know the behaviors of the ruling government.In general, the negativism and commercialization of the media industry (Mazzoleni and Schulz, 1999) generates media that is focused on fi nancial scandals about local governments and cases of corruption, especially when elections are approaching.Thus, ruling governments tend to reduce the level of transparency, as a mechanism of entrenching themselves.
In general, the three control variables are positively related to IT_INDEX at diff erent confi dence levels.This means that municipalities governed by left-wing parties, with high political strength and more densely populated areas tend to be more transparent with the aim of being supported by citizens who provide the public resources (Ferejohn, 1999).In addition, Roubini andSachs (1989a and1989b) showed that the problems of coordination in fragmented and coalition governments lead to higher defi cits than in single party governments.This generates problems that are not well considered by citizens and so, local governments tend to avoid this becoming known.Finally, larger cities, with higher populations, may pressure government favoring information transparency.MEDIA represents the number of news related to each municipality; PREVIOUS YEAR is a dummy variable that takes the value 1 in the year prior to the election and 0 otherwise; PREVIOUS YEAR*MEDIA refl ects the interaction between PREVIOUS YEAR and MEDIA; LEFT is a dummy variable that takes the value 1 when the ruling party is leftist ideology and 0, otherwise; STRENGTH refl ects the strength of local government policy through the Herfi ndal index calculated according to Rattsø and Tovmo (2002); DENSITY represents the inhabitants per km 2 .

Robust empirical analysis
Table 4 shows the same models but here, the dependent variable represents the diff erent sub-indexes that compose the global transparency index (IT_INDEX).Regarding sub-index A as the dependent variable, which is related to transparency according to information about municipal corporations, the results are less conclusive.The only statistically relevant variables are those control variables that represent political strength and density, whose coeffi cients are positive.This means that when more councilors represent the same party and when the municipality is more densely populated, the level of information transparency is greater.
When the dependent variable represents the level of transparency according to relationships with citizens and society (sub-index B), and the level of economic and fi nancial transparency (sub-index C), respectively, results are similar to those obtained for the global transparency index.MEDIA and PREVIOUS YEAR show positive coeffi cients, and PREVIOUS YEAR*MEDIA shows negative coeffi cients.This means that when elections are approaching the level of municipal transparency increases.However, if media pressure is strong in that moment, the level of transparency may be reduced.This means that local governments tend to be more transparent in terms of information related to relationships with citizenship and information related to economic and fi nancial information when elections are approaching.However, if media pressure is strong in that moment, the level of transparency may be reduced.
As for sub-index A, for IT_SUB-INDEX D -which represents the level of transparency according to public services contracts -, results are less conclusive, since the only statistically relevant variable is PREVIOUS YEAR at the 99% confi dence level.This means that local governments tend to disclose information about public services contracts in the year before the election, independent of media pressure.
Finally, for IT_SUB-INDEX E, the level of transparency is related to information about urban development and public works.The variable MEDIA is positively and signifi cantly (at 95%) related to the dependent variable.The variable PREVIOUS YEAR has a positive coeffi cient, and it is signifi cant at the 99% confi dence level, meaning that the level of public transparency increases when elections are approaching.However, media pressure changes this relationship, reducing the disclosure of public information about urban development and public works in the year before elections.
The individual analysis of each sub-index provides an interesting conclusion.On the one hand, sub-index C represents transparency about economic and fi nancial information (municipal annual accounts, public budgets, public incomes and expenses, and level of public debt).In Spain, although municipal public debt is not a problem within all public administrations, it is a problem for councils.Access to this information should not be limited for citizens (Transparency International Spain, 2010) with the aim of avoiding the loss of their confi dence.
On the other hand, sub-index E represents transparency about urban development and public works (information about urban development plans, urban agreements, urban reclassifi cations and adjudications, bidding processes).This sub-index represents one of the most important problems of corruption in Spain.Information opacity is more important in municipalities that have suff ered cases of urban corruption (Transparency International Spain, 2010).
Owing to the relevance of these incidents in Spain, the eff ect of media pressure as a mechanism of monitoring and controlling public managers is more statistically relevant for sub-indexes C and E. In conclusion, Spanish local governments tend to be more transparent when media pressure is strong, especially for economic and fi nancial information and information about urban development and public works, topics in which Spanish local governments have traditionally suff ered more cases of corruption.In addition, the level of transparency, especially in relation to this information, is higher when elections are approaching.However, this relationship is moderated by media pressure, reducing the level of transparency when media pressure is strong, to avoid a loss of confi dence of citizens just when elections are close.

Concluding remarks
Using a sample composed by 110 Spanish cities for 2008-2010, this paper analyzed the behavior of local governments about public transparency, considering additionally the electoral moment.The empirical results show that the level of municipal transparency is higher when media pressure is strong, suggesting that media could be seen as a powerful tool to push governments to act responsibly (Besley and Burgess, ).This tool is even more necessary when regulation on access to information is low; media enables citizens to access the decisions taken by the government and to observe the strategies, activities and results of public policies (Balking, 1999).Thus, media may incite the government to promote accountability (Macdonell and Pesic, 2006), replacing regulation on access to public information.However, our fi ndings show that the proximity of elections moderate such relationship.In general, the negativism and commercialization of the media industry (Mazzoleni and Schulz, 1999) generates media that is focused on fi nancial scandals about local governments and cases of corruption involving politicians, especially when elections are approaching.Thus, as a mechanism of entrenchment, politicians are willing to decrease the level of transparency.
In addition, empirical results suggest that Spanish local governments tend to be more transparent about economic and fi nancial information and information about urban development and public works, when media pressure is strong.These topics are relevant in the case of Spain because local governments have traditionally suffered more cases of corruption related to them.The ruling governments increase the level of information disclosures about these topics when elections are approaching, but this level is reduced when media pressure is strong in that moment; this behavior could be understood as an entrenchment strategy -due to the fear to lose the confidence of voters.
To avoid such situation, a reliable transparency law is necessary.Until December 2013, Spain was one of the few European countries that do not have a public transparency law.However, we think that this law is the fi rst step, but it is not enough.It posits that public administrations must disclose periodical institutional, organizational and planning information, relevant legal information, and economic, budgetary and statistical information.However, it does not determine specifi cally what information must be disclosed, and the frequency of data provision.
This study has some limitations.The use of the Google tool may be doubted, since not all newspaper reports or regional issues can be found on Google.A more specific database for press could provide more valid data about media pressure.Furthermore, the possibility of distinguishing between good and bad news about local governments could enrich this study.Furthermore, we were focused on Spain due to the availability of data, and our empirical evidence is not generalized to other countries or periods.We believe that these limitations could be overcome in future studies.
Our future lines of research are in accordance to our present limitations.It would be useful to conduct the same analysis using a larger sample.In order to ensure more robust results, future research would corroborate our evidence through alternative measures of transparency and media pressure.In addition, it would be interesting to carry out a comparative analysis between local governments from diff erent countries, controlling for institutional contexts.Thus, whether the characteristics of each country related to legal aspects about anti-corruption, and how free media may aff ect the relations proposed in the present study could be analyzed.More concretely, it would

2 .
* and *** indicate signifi cance at the 1%, 5% and 10% level respectively.IT_INDEX represents the municipal transparency index; A, B, C, D and E represent the different sub-indexes that compose the previous global index; MEDIA represents the number of news related to each municipality; MEDIA represents the number of news related to each municipality; PREVIOUS YEAR is a dummy variable that takes the value 1 in the year prior to the election and 0 otherwise; LEFT is a dummy variable that takes the value 1 when the ruling party is leftist ideology and 0, otherwise; STRENGTH refl ects the strength of local government policy through the Herfi ndal index calculated according to Rattsø and Tovmo (2002); DENSITY represents the inhabitants per km Source: Authors' computations

Table 3 :
Empirical results

Table 4 :
Empirical results for sub-indexes MEDIA represents the number of news related to each municipality; PREVIOUS YEAR is a dummy variable that takes the value 1 in the year prior to the election and 0 otherwise; PREVIOUS YEAR*MEDIA refl ects the interaction between PREVIOUS YEAR and MEDIA; LEFT is a dummy variable that takes the value 1 when the ruling party is leftist ideology and 0, otherwise; STRENGTH refl ects the strength of local government policy through the Herfi ndal index calculated according to Rattsø and Tovmo (2002); DENSITY represents the inhabitants per km 2 *, ** and *** indicate signifi cance at the 1%, 5% and 10% level respectively. .Source: Authors' computations