Transparency in Portuguese Local Government: A Study of its Determinants

: Transparency of public management is fundamental to bringing citizens closer to public managers and as such universally considered an element of good governance. The main aim of this study is to identify the factors determining transparency in public management of Portuguese local authorities. Using the Municipal Transparency Index of 2013 and 2014, socio-demographic, political and economic-financial variables are studied for the population of 308 Portuguese local authorities by the means of testing the previously formulated research hypotheses in the light of the agency theory. The results show that transparency in public management of Portuguese local authorities is determined by their size, the number of individuals aged between 15 and 74 who use the internet, unemployment rate, financial autonomy and the net debt.


Introduction
Transparency in public management, besides being an obligation for managers and politicians, contributes to broader citizen/voter participation in exercising public functions (Tavares and Rodrigues, 2012). It is fundamental for their access to transparent information allowing on their part an assessment of the managerial efficiency as well as identification with the system. The new form of action in local government, namely accountability to voters/taxpayers, means of communication, controlling bodies and other stakeholders in accompanying governments' activities, increases the attention paid to transparency and its determinants. The principal objective of this study is to contribute to the scientific literature, seeking to answer the following research question: "what are the determining factors of transparency in Portuguese local authority management?" The main purpose is to identify, for the Portuguese local authority situation, the factors determining the level of transparency. To achieve this aim, the results of the recently developed Municipal Transparency Index (MTI) for the years 2013 and 2014 are used. Following this introduction, the article is structured in 3 sections. Section two relates the agency theory with accountability and transparency. The third section presents the empirical analysis, stating the hypotheses and variables, how the data will be treated, and the results and discussion of the hypotheses. The last section presents the main conclusions of the study. Jensen and Meckling (1976) define the agency relationship as a contract through which one or more people (the principalcapital holder) hire another person (the agentmanager) to perform a service or work in their name. In this contractual relationship the principal delegates some decision-making authority to the agent, a fact that gives rise to conflicts of interest due to the separation of ownership and control. The agent does not always act in the principal's interest, which means agency costs are incurred, resulting from adopting monitoring mechanisms and residual losses. The principal-agent dilemma involves a relational conception of accountability, bearing in mind the matter of information (Ferejohn, 1999). The problem of accountability is related to the problem of delegation, since representative democracies mirror the relation of authority between the person who delegates and the one who is delegated. The principal delegates to the agent the power to carry out something in the former's interest, and it is the agent's duty to act in the principal's interest. The agency theory has been widely applied to the business situation, but in recent decades it has been used increasingly in other contexts, particularly in representative democracy (Ferejohn, 1986). Considering the agency theory approach to representative democracy, voters (principal) elect politicians/leaders (agents) to represent their interests through the policies proposed by governments and the guidance given to Public Administration on how to carry them out. In addition, by financing public activities, tax-payers also delegate to their representatives the decision to allocate public resources.

Agency Theory, Accountability and Transparency
From the agency theory's perspective, the lack of transparency can create an advantage for political decision-makers in achieving their objectives: agents (politicians/leaders) can have their own interests that do not always maximize their voters' well-being. Therefore, and according to the above mentioned theory, accountability takes place in the scope of an electoral democracy, according to a minimalist conception of the democratic process fundamental for the institutionalization of accountability which consists of reducing the information deficit between principals and agents by enhancing transparency. Alt, Lassen and Skilling (2001) indicate that transparency reduces the information asymmetries between political agents and citizens/voters. The authors suggest two ways to diminish this problem: i) inform voters about the actions taken by elected politicians (Ferejohn, 1999), or ii) facilitate coordination in budget results between the bodies alternating in power. The concept of transparency emerges as linked to the information economy, where the citizen is seen as an investor and consumer of public goods. In the view of Ferejohn (1999), the principal invests its resources in the action carried out by the agent (politicians/leaders), hoping for a positive return on the undertaken investment. The balance in the relationship between the principal (voters/taxpayers) and the agent (politicians/leaders) depends on the existence of transparent entities that are able to reduce the uncertainty of the investment (Ferejohn, 1999). As mentioned above, transparency maximizes accountability by allowing a reduction of information asymmetry between the principal and the agent, thereby ensuring a system of possible accountability through reducing public bodies' secrets. According to this approach, for the agency theory, transparency strengthens accountability in the scope of the information economy, with the citizen being seen as a consumer of public goods, i.e., financing public sector activities.

Research hypotheses
theory, a set of research hypotheses was formulated, relating the level of transparency with certain variables considered as the potentially explanatory factors. For better understanding, the analysed hypotheses were grouped based on socio-demographic, political and economic-financial factors. DIM -Logarithm of the inhabitants residing in the local authority.
INTER1574 -Logarithm of the number of residents aged between 15 and 74 who use the internet. ESCOL -Percentage share of the population with higher education on the total population of the local authority. IENV -Number of people aged 65 and over for every 100 people under the age of 15. TXDESEMP -Number of unemployed per 100 active people. FGOV -Dummy variable taking on the value '1' if the political party that manages the municipal executive obtained majority (more than 50.1% of the votes), otherwise '0' (zero). IDEOL -Dummy variable with a value '1' when the municipal executive is of the right party ideology and '0' (zero) otherwise. PARTELEI -Number of voters in municipal elections divided by the number of voters registered in the electoral rolls. AUTFIN -Own revenues divided by total revenues. ENDLIQ -Differences between liabilities, in whatever form, including borrowings, leasing contracts and debts to suppliers, and assets, namely the cash balance, bank deposits, treasury applications and credit claims.

Data treatment
To determine the relationship between each of the independent variables and the MTI, so as to study the various hypotheses defined, a multivariate analysis is performed using the STATA program.
As the linear regression model has been widely adopted (e.g., Guillamón, Bastida and Benito, 2011;Vicente, Benito andBastida, 2013 andCruz, 2014) for specifying econometric models that explain the determining factors of local authority transparency, the same method is chosen in this study. More specifically, the multiple linear regression model is applied by the Ordinary Least Squares method since we expect that the various (independent) variables may explain the local authority transparency index. According to Marôco (2010), estimation of the linear regression model can only be performed if certain assumptions are fulfilled. Therefore, various tests will be carried out to complete that analysis. Considering MTI as the dependent variable, and the different variables already mentioned as independent, the following model will be studied: MTI = α + 1 DIM + 2 INTER1574 + 3 ESCOL + 4 IENV + 5 TXDESEMP + 6 FGOV + 7 IDEOL + 8 PARTELEIT + 9 AUTFIN + 10 ENDLIQ + Ɛ, where DIM is dimension; INTER1574 stands for number of residents aged between 15 and 74 who use the internet; ESCOL is used for educational level; IENV expresses ageing index; TXDESEMP represents unemployment rate; FGOV expresses party fragmentation; IDEOL is used for political ideology; PARTELEIT stands for turnout; AUTFIN is financial autonomy and ENDLIQ means debt.

Results and Discussion
The resulting estimated model is presented in the form of the global model (SOCDEM_PLO_ECOFIN) which includes socio-demographic, political and economic-financial variables. Within partial models, each group of factors was estimated individually. Therefore, Model 1 is related to socio-demographic factors (SOCDEM), Model 2 to political factors (POL) and Model 3 to economicfinancial factors (ECOFIN). Analysis of Table 2 reveals that for most variables, partial models present results that are quite consistent with estimations of the global model. In this context, we will now move to an analysis of results of that model (SOCDEM_POL_ECOFIN), finding that it presents an R2 of 0.214. The explanatory variables of the model express 21.4% of the variations occurring in the MTI. It is noted that the explanatory power is not very high, but it agrees with the results obtained by Guillamón, Bastida and Benito (2011) and Tavares and Cruz (2014), respectively to 29% and 25%. The results of the F test (10,605) = 34.23 allow the conclusion that, for a 1% level of significance, there is a sufficient statistical evidence to state that the regressors present values different from zero and as a whole explain satisfactorily the variations found in the MTI. Considering the previous results, tests were made to analyse a possible violation of the model's assumptions. Concerning multicollinearity, no variable was found to present a VIF value above 10, which indicates a non-violation of the assumption. Nor the 1/VIF values of tolerance are close to zero, showing that the regression coefficients are stable, both in magnitude and sign (Marôco, 2010 To test for heteroscedasticity, the White test was used. As the test statistic is 77.52 above the p-value = 0.1032, it is concluded that the assumption of heteroscedasticity was not violated. Autocorrelation of errors was carried out using the Durbin-Watson test, reaching the conclusion that the model does not suffer from error autocorrelation.
The results obtained reveal that for the maximum level of significance of 10%, the variables of DIM, INTER1574, TXDESEMP, AUTFIN and ENDLIQ are statistically significant.
The obtained results will now be discussed. Based on the agency theory, voters (principal) elect with their votes politicians/leaders (agents) to represent their interests through policies, proposed by governments and instructions given to Public Administration on how to carry them out. As there is information asymmetry between agents and the principal, based on the results obtained it is important to discuss which socio-demographic, political and economic-financial factors determine the transparency in public management of local authorities. Out of the ten research hypotheses formulated, the results obtained allow us to validate five. Validation of H 1 corroborates the idea that municipal authorities are more transparent in larger councils. This result agrees with those obtained by Gallego-Álvarez, Rodríguez-Domínguez and García-Sánchez (2010), Guillamón, Bastida andBenito (2011), Esteller-Moré andOtero (2012), Sol (2013), Mendes (2012), Pires (2012), Batalha (2014) and Sedmihradská (2015).
The results obtained also confirm that the wider the population with access to the internet, the higher the local authority transparency, and so H 2 is also validated. Easy access to the internet increases the possibility of access to information about the local authority. In this context, citizens seem to be closer to local authorities. The hypothesis related to the unemployment rate is also corroborated, and so H 5 is also validated. This result is consistent with that obtained by Alegre et al. (2013), Tavares and Cruz (2014) and Batalha (2014). There seems to be the underlying idea that, on one hand, the more precarious the council's economic situation, the greater the temptation for local authorities to conceal this fact. On the other hand, the unemployed population may have many other personal worries, and is therefore less concerned about seeking information on the local authority, outside the area of employment. In relation to economic-financial factors, the results allow validation of the two hypotheses defined. Therefore, H 9 is validated, corroborating the opinions of Tavares and Cruz (2014), Alegre et al. (2013) and Vicente, Benito and Bastida (2013). Here, we find that local authorities with higher financial autonomy, possibly even able to charge more taxes, feel the need to inform citizens, and so they are more transparent. This fact may contribute to citizens not feeling they are paying high taxes as they may have a precise idea of municipal management. The results obtained also allow the validation of H 10 , corroborating those obtained by Rueda (2007), García and García (2008), Pires (2012) and Batalha (2014).
Here, there seems to be statistical evidence confirming that greater net debt leads local authorities to be more transparent. There is a general opinion that such a result can be explained by local authorities wanting to pacify citizens. Finally, it is underlined that no results were found to allow validation of the hypotheses related to political factors (H 6 , H 7 & H 8 ).

Conclusion
Recent studies in the field of Public Administration have highlighted the role of transparency in making information available to citizens, allowing them to participate in the political process, as well as promoting accountability, improving the quality of government decision-making and assisting to anticipate and lessen corruption. The information provided can reveal private interests that may go against the collective interest and hold actors responsible for all decisions and measures taken or omitted. This study contributes to scientific knowledge by investigating the determinants of local government transparency, seeking to answer on the specific case of the Portuguese situation the following research question: "what are the factors determining transparency in public management of Portuguese local authorities". In particular, the intention was to identify which factors determine the level of transparency in public management of Portuguese local authorities. To this end, a set of hypotheses were tested, considering the 308 Portuguese local authorities. Use was made of the results of the recently developed MTI for 2013 and 2014.
The study was based on the agency theory, considering the agency relationships applied to voters/tax-payers (principal) and politicians/leaders of local administration (agent). In general terms, the results obtained identify as the explanatory determinants of transparency in public management of Portuguese local authorities: the population size, individuals aged between 15 and 74 who use the internet, the unemployment rate, financial autonomy and the net debt. No results were obtained allowing validation of the hypotheses related to the level of education, aging index or political factors (form of governance, political ideology and political/voter participation). Finally, it is highlighted that this research aimed to make a significant contribution to the literature studying the determinants of transparency in public management of Portuguese local authorities, largely due to the limited number of studies carried out in this area, and particularly considering the MTI.