Weakening political connections by means of regulatory reform: Evidence from contracting out water services in Spain

One area of public policy where rent-seeking and favoritism is relatively common is the contracting out of public services. Private firms can improve their chances of obtaining contracts by bribing politicians or public servants and funding political parties. In the same vein, firms can gain access to policymakers by hiring influential former politicians—a practice commonly referred to as revolving-doors. In this paper, we use information from 922 privatizations of water services in Spanish municipalities between 1984 and 2016 and multinomial logistic regression techniques to study the association between specific firms securing contracts and the political parties ruling the municipalities. We find robust statistical evidence of an association between the Popular Party (Partido Popular or PP) and the firm Aqualia, part of the large Spanish holding company Fomento de Construcciones y Contratas (FCC), which is known to have funded the Popular Party. Furthermore, former PP politicians have been appointed to top positions in the FCC Board of Directors. However, this relationship weakened after the institutional reform of 2007 on public procurement and financing of political parties, which is empirically evaluated in this paper.


Introduction and background
The study of rent-seeking has long been a key element in the analysis of the motivations of governments' decisions. Rent-seeking has been defined as '… The quest for privileged benefits from government' (Aidt 2016: 144). According to Lambsdorff (2002: 101), the focus of rent-seeking is on the interaction between the state and private parties, where the government has the monopoly on allocating property rights. The study of how individuals or firms can benefit from political influence and connections has received attention in the field of economics, ever since the seminal papers of Stigler (1971) and Peltzman (1976). In particular, Stigler (1971) discussed financial support for electoral campaigns, and funding for the bureaucratic tasks of the political party machinery as potential demands that governing parties can make in exchange for regulation in favor of specific industries or firms. In recent years, there has been an accumulation of empirical evidence on the existence of strong, mutually beneficial linkages between firms and politicians (Acemoglu et al. 2016;Akin et al. 2016;Faccio et al. 2006).
One of the fields most closely linked to political rent-seeking and favoritism is that of privatization of public services (Hart et al. 1997). In this context, rent-seeking may occur when a company secures public service contracts in exchange for financial donations or board positions, among others. Empirical evidence on the relationship between political connections and financial donations, and the allocation of procurement contracts has been provided for different levels of government: federal (Boas et al. 2014;Goldman et al. 2013;Witko 2011) as well as local (Amore and Bennedsen 2013). This practice is contrary to the public interest, because it does not guarantee that the best-performing company is awarded the contract. This is therefore likely to lead to inefficiency (Dal Bó and Rossi 2007), lower service quality and extra service delivery costs (Dastidar and Mukherjee 2014), which will ultimately be borne by users or by taxpayers (Rose-Ackerman 1999;Tollison 2012). In short, it adversely affects the expected social welfare of a contracting auction (Boehm and Olaya 2006). Furthermore, Hart et al. (1997) contend that the policymakers engaging in rentseeking in public services delivery are incentivized to create over-employment when public services are delivered in-house, thus engaging in political patronage and obtaining political support. Conversely, if service delivery is contracted out, rent-seeking politicians will tend to obtain financial resources, rather than over-employment, which can be used for personal enrichment, to fund political activities (elections and party machinery), or both. Personal enrichment is closely associated with illegal behavior and corruption; however, the funding of political parties or electoral campaigns by firms is legal in most democracies. Hence, rent-seeking does not necessarily involve illegal practices, and can be seen as different from corruption. In fact, our study does not deal with corruption, but rather with potential favoritism between political parties and firms, based on funding to the party, and revolving doors between governments and firms.
According to Ariño (2009), decisions related to the procurement of public services are also one of the main sources of political rent-seeking and favoritism in Spain. On the one hand, scholars and politicians recognize that the rules on the financing of Spanish political parties establish an overly restrictive framework for securing resourcesdonations are limited to e60,000 per donor, and the contributions by party members are largely insignificant-and create a system that is overly dependent on government subsidies (Pujas and Rhodes 1999;Casal et al. 2014). In fact, this has been used as an argument to explain why Spanish political parties have irregularly obtained resources at some point (Bel et al. 2014); since the 1990s, judiciary courts and the Court of Auditors (Tribunal de Cuentas) have investigated several political parties over events such as the cancellation of parties' bank debts, corporate donations and bribery.
On the other hand, the processes for awarding contracts in Spain are not characterized by transparency and independence. The lack of transparency was partially addressed some years ago by Law 30/2007 on Public Sector Contracts. However, external control is insufficient, and the hiring process is largely under the control of local governments. Besides, an additional problem is the difficulty faced by the judiciary in obtaining evidence of malpractice, and the lengthy judicial processes. An index recently developed by the RAND Corporation (Stanley et al. 2014) assesses aspects such as the expectation of bribes and anti-bribery laws, and ranks Spain as one of the developed countries with the highest risk of bribery.
Against this background, our paper focuses on two main research questions, namely, the possible existence of political connections between the main Spanish political parties and major private firms in the urban water delivery industry; and the impact of two laws (Organic Law 8/2007 andLaw 30/2007) passed in Spain in 2007, which were aimed at increasing transparency in the funding of political parties, and ensuring objectivity in the procurement of public services. Our first hypothesis is the existence of a systematic relationship between companies that win privatization contests and the political party that holds the municipal government. This is a reasonable suspicion in Spain, judging by the many cases that are currently under judicial investigation for alleged corporate payments in exchange for preferential treatment in the awarding of public contracts for water service management (GWI 2013). Due to the slowness of the judicial system in Spain, most of these cases date back to the past decade. If our study reveals significant relationships, these could be interpreted as indirect evidence of favoritism. Our second and more interesting hypothesis is that the legal reforms implemented in 2007 made it more difficult to engage in favoritism in the awarding to private firms of contracts for urban water delivery.
The study centers on 922 agreements to privatize water services management made by local governments in Spain between January 1984 and April 2016. The methodology used is multinomial logistic regression. Our results show evidence of a statistically significant association between the likelihood of specific firms securing (or not) the contracts, and the presence of a specific political party in government in the period before reforms. These significant relationships disappeared in the period after 2007, which may be an indication of how institutional reforms can weaken political connections. In this way, we contribute to the literature in this field of research by providing evidence to suggest bias in the awarding of contracts, depending on the governing political party in the municipality, and pointing to the effectiveness of institutional reforms in ending these practices.
The remainder of the paper is organized as follows. Section 2 describes several cases of political connections in contracting out local public services in Spain. Section 3 outlines the legal framework for the privatization of the urban water service in Spain, and the reforms passed in 2007 regarding procurement of public services and funding of political parties. Section 4 describes the data and the methodology. The results are presented in Sect. 5, and the final section summarizes and sets some policy conclusions.

Political connections related to contracting of local public services in Spain
When Spaniards these days are surveyed about Spain's problems, corruption is revealed to be a major concern (Schwab 2016;EC 2014 Within the water distribution sector itself, numerous cases of alleged bribery have been subject to judicial investigation. Those cases have been frequent enough as to be the object of a report issued by Global Water Intelligence (GWI 2013), which provides extensive information on judicial inquiries affecting the two major water distribution firms in Spain, namely, Aqualia 4 and Agbar. 5 Taken together, these two firms cover about 67% of the municipalities with private delivery of urban water (around 75% of the population served by private firms). Other minor firms have also been affected, such as Emarsa and Aguas de Valencia.
Finally, mention needs to be made of the common practice of revolving-doors between politics and big business in Spain, as documented by Castell and Trillas (2013: 109-110). This practice refers to situations where politicians holding high positions in government are subsequently appointed as members of the Board of Directors of large firms that have regulatory or contractual relations with the government, after they have withdrawn from institutional political activities (Heyes 2003). Within the context of this study, several former politicians of the Popular Party have been appointed as members of the Board of Directors of FCC. In this vein, it is worth recalling that two prominent former members of the Popular Party were members of the FCC's Board at the end of 2014 (FCC 2015: 3). One was Marcelino Oreja Aguirre, minister in the Spanish government (1976)(1977)(1978)(1979)(1980) and first Popular Party candidate elected to the European Parliament in the 1989 election. He was later appointed Commissioner of Transportation and Energy for the European Commission (1994)(1995)(1996)(1997)(1998)(1999), a job which he was put forward for by the Popular Party.
The second was Gustavo Villapalos Salas, who had formerly been Minister of Education, Culture, andSports (1995-2001) in the regional government of Madrid, under the ruling of the Popular Party. It is also interesting to note that in 2001, Abel Matutes was appointed to the FCC Board of Directors just after being Minister of Foreign Affairs in the Popular Party's first government (1996)(1997)(1998)(1999)(2000). He was a member of that Board for most of the last decade.  (FCC 2015), thus becoming the major shareowner. In addition to providing water services to more than 850 Spanish municipalities, this holding company is also present in more than 50 countries, including China, Mexico, Portugal, the Czech Republic, Poland and Algeria. 5 Agbar (an abbreviation of Aguas de Barcelona) is a subsidiary of the French holding company, Suez Environment. It comprises over 150 companies and has around 13,000 employees, and provides water services not only in Spain but also in countries such as the United Kingdom, Mexico, China, Chile, Cuba, Colombia and Algeria. Moreover connections with former members of governments on the Boards of Directors of other firms in the urban water delivery sector; neither in Agbar 6 -the other dominant firmnor in other firms with lesser market share.

Background and legal framework
In Spain, municipal governments are responsible for the urban water supply, and they can choose among different forms of service delivery, which are set out in the current legislation. 7 First, the municipal government may choose to manage the service inhouse (the government itself manages the service) or to externalize it. Externalization can be effected by means of transferring delivery to a government-owned company (public firm), or privatizing it. In case of the latter option, privatization can be either partial, by means of a mixed firm (institutional public-private partnership, PPP), or full, whereby the service is contracted out to a private firm (contractual public-private partnership).
After the entry into force of Law 7/1985 on Local Government Regulations, many Spanish municipalities decided to privatize the management of water services. Three decades after that Law was passed, Spain has one of the highest levels of private sector participation in the management of urban water of any OECD country, as a percentage of the population served. According to González-Gómez et al. (2014), in 23% of Spanish municipalities, urban water supply is delivered via one of the forms of private management referred to in the legal system. However, this percentage rises to 55% when expressed in terms of population. This is indicative of the concentration of the privatization process in the more populated municipalities. 8 The main motivations for water privatization in Spain have been pragmatic and, to a lesser extent, ideological and political (González-Gómez et al. 2011;Picazo-Tadeo et al. 2012). These results align closely with those obtained in Bel and Fageda (2009) meta-regression analysis to explain local privatization decisions.
According to the legislation in force, the bidding process for water services in Spain develops as follows. First, an announcement of the public tender of the contract is made, which includes the technical and economic conditions of the contract and the procedure and assessment criteria. Details are also given on the commitments that must be fulfilled by the private firm awarded the contract. In light of all that information, private firms place a first-price sealed-bid, in which bidders must submit their technical 6 Actually, five out of the seven members of Agbar's Board of Directors are French. Its Executive President has never had any involvement in institutional activities. 7 The regulation of the management of municipal services in Spain is set out in Law 7/1985 on Local Government Regulations, and Law 57/2003 on the Modernization of Local Government. 8 Partial privatization, by means of mixed firms, is also becoming increasingly common. Ownership is shared between the government, who usually retains ownership of a large fraction of the firm's capital and is expected to ensure the public interest, and the private partner, who has industry know-how and is often in charge of the day-to-day management (Warner and Bel 2008;González-Gómez et al. 2009). and financial proposal. Then, the municipal government awards the contract to the firm that made the best proposal with respect to a number of criteria.
The jury in charge of adjudicating the bids is headed by the mayor or any other city council member or public servant. In the first place, the jury determines whether the proposals fulfill the commitments required in the public tender, and offers noncomplying firms a deadline of 3 days to amend rectifiable mistakes in their proposals. Second, it evaluates and scores the technical conditions offered by bidders admitted to the contest, which are included in the first-price sealed-bid and might refer to issues such as service quality, management or emergency plans in case of supply disruptions. Given the qualitative nature of some of these technical conditions, their assessment unavoidably requires some value judgments. As a third step, the financial conditions of each proposal-which include elements such as water tariffs and, occasionally, other proposals for infrastructure investment-are also scored. Finally, following a joint assessment of both technical and financial conditions, each bidder is awarded a final overall score, on which the decision is based.

Legal reforms in 2007 regarding procurement of public services, and funding of political parties
In 2007, as mentioned in the Introduction, two laws were passed that changed the framework of relations between political parties, local governments and private companies. This new scenario affected local procurement, and financing of political parties.  (Razquín 2008). Three changes brought about by the new law are particularly relevant. The first one is the obligation to publish online all the information related to the bidding processes promoted by local governments, which makes the process more public and transparent. A second change affects the composition of the jury in charge of evaluating the bids. While the mayor was always the president of the jury before the new law, the new regulations stipulate that the president can be a member of the city council, or a public servant. Still more important, a third change limits the discretionary power of the jury: whenever the decision gave more importance to qualitative factors than to quantitative valuations automatically generated from the bidding clauses, Article 134.2 of the new Law requires that the awarding reports must be reassessed by a committee formed by three external referees. Hence, the New regulation envisages the possibility of transferring to external referees the assessment of qualitative valuations (Moffa 2009).
Organic Law 8/2007 of 4 July, on the Financing of Political Parties, replaced the old Spanish legal framework established in 1987. A notable aspect of the new Law was the ban on political parties receiving donations from companies working for the public administration. A subsequent reform introduced in 2012 (Law 5/2012 of 22 October), extended the ban to companies belonging to the same group as those who work for the administration, or who are controlled by them. This seeks to prevent the party in power from granting contracts or services in exchange for financial contributions.

Variables, data and sources
In order to test the hypotheses stated in the Introduction, we use information from 922 privatizations of the urban water service that occurred in Spain between January 1984 and April 2016. Given that our aim is to shed some light on why local governments choose one particular firm over other alternatives, the first variable of interest in our research is the name of the firm to which the contract was awarded.
As already noted, private firms, either contractual or institutional PPPs, managed the urban water service in approximately 1800 Spanish municipalities in 2016. It is worth noting that until 2008 there was no public register with information about contracting procedures of local services in Spain. As described in Sect. 3.2, from 30 April 2008, Law 30/2007 on Public Sector Contracts required local administrations to create an institutional webpage named Contractor Profile, which is aimed at ensuring transparency and affording public access to the information relating to the contractual activity of public administrations. Accordingly, from this date onwards, when a local government privatizes the management of a given service it must grant public access through the abovementioned webpage to all information related to the privatization. We have thus made use of this resource to collect information about privatizations taking place after Law 30/2007 was passed, 9 and particularly the name of the firm awarded the contract for the urban water service provision.
In addition, we submitted postal or electronic requests to local councils that had privatized the service before that time, asking them for the date of the town council plenary session in which the decision to privatize was taken. We then referred to municipalities' Official Gazettes to obtain the name of the provider of the urban water service. Furthermore, in a few cases, we obtained this information directly from the web pages of the town councils. Taking these sources together, we collected information about 922 privatizations of the urban water service in Spanish municipalities, which represent nearly 50% of total privatizations in the period of reference. 10 For the purpose of our research, we have firstly established three possible options for the firm chosen to privatize the urban water service, namely, Aqualia, Agbar, and Others. The group Others includes all firms other than those included in the first two groups, e.g., firms that operate at national level such as Acciona, Agua & Gestión, Gestagua, Hidrogestión, Urbaser and Valoriza, in addition to some companies that, as mentioned, operate mainly at regional level, including Aguas de Valencia, Facsa, Prodaisa and Espina & Delfín.
In our sample, Aqualia was chosen to privatize the urban water service 314 times (34.1%), Agbar 308 times (33.4%), while firms included in the group Others were chosen 300 times (the remaining 32.5%) (Table 1). Furthermore, we can distinguish between the periods before and after the 2007 legal changes, in order to provide an initial description of the effect of reforms. The percentage of contracts awarded to big companies (Aqualia and Agbar) was larger in the period prior to the reform  and decreased in the period after the reform (2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016). The group of other companies, which was the one with the lowest percentage of contracts (23.2%) between 1984 and 2006, becomes the group with the highest share of contracts (45.8%) in the period 2007-2016. Aqualia secured 39% of contracts between 1984 and 2006, but just 27.1% in the following period. Similarly, Agbar was awarded 37.8% of contracts in the period prior to the reform, but just 27.1% in the period after the reform. Finally, Table 1 also shows that the two biggest Spanish political parties, the Spanish Socialist Workers' Party (Partido Socialista Obrero Español or PSOE) and the Popular Party have both awarded more contracts to Aqualia than the average that this firm obtained all over Spain, while the opposite is true in the case of Agbar, to which PSOE, and particularly PP, have awarded a lower percentage of contracts than the share of contracts that Agbar has obtained all over Spain.
These changes occurred during a period in which the water delivery service industry was highly concentrated, as described in Bel et al. (2013: 379), where concentration ratios were compared between 2003 and 2009. The two biggest companies of the sector covered 69% of population in Andalusia and 91% in Catalonia, two of the most populated regions in Spain with private contracting practices. In 2009, these shares increased to 79% in the case of Andalusia and to 92% in Catalonia. 11 Herfindahl-Hirschman indices were also very similar in 2003 and 2009. Therefore, the increase in contracts awarded to other companies after the reforms cannot apparently simply be explained by a change in the industry market structure unrelated to the policy evaluated.
Furthermore, note that changes between the periods before and after the reform are particularly large in the case of contracts awarded to Aqualia by municipal governments ruled by the big political parties. There is a difference of 21 percentage points between the two periods for PP-from 47.4 to 26.4%-and 13.5 percentage points for PSOE-from 43.9 to 30.4%. Smaller changes are found in the case of contracts awarded to Agbar, though percentages of contracts awarded by both PP and PSOE also diminished with respect to the period prior to the reform. All political parties awarded much larger percentages of contracts to other companies in the period 2007-2016.
In order to explain the choice of the firm selected to privatize the urban water service and test the hypotheses established in the Introduction, we have selected several variables relating to the policy framework, some features of the municipality, and also to the party in power of the local government at the time the decision was made. Table 2 presents some basic statistics for these variables, while the "Appendix" summarizes  A second set of control variables includes two market variables related to the municipality that privatizes the urban water service: population, measured as the number of inhabitants and taken from the Spanish Statistical Office (INE); and economic activity, measured by the indicator of economic activity in Spanish municipalities provided by La Caixa (2014). The third set of variables includes 16 geographical dummy variables that take the value 1 if the municipality belongs to the regions of Andalusia, Aragon, Asturias, Balearic Islands, Basque Country, Canary Islands, Cantabria, Castile-La Mancha, Castile and León, Catalonia, Extremadura, Galicia, Madrid, Murcia, La Rioja and the Valencian Community, respectively, and 0 otherwise. 12 These variables are used in order to capture the possible existence of regional unobservable patterns in the decision to privatize the urban water service or other unobservable features of the region that could affect the selection of the service provider.
Regarding the regional patterns observed in Table 2, Aqualia seems to do well in Andalusia while Agbar is overrepresented in Catalonia. There is a sharp difference between the market structures of water services in the two regions; while 74% of Catalonian municipalities had contracted out their urban water services in 2009, only 13% of Andalusian municipalities had done so (Bel et al. 2013). Furthermore, the distribution of local political power is also fairly different. In 2011, 34% of mayors in Andalusia belonged to the Popular Party, while PSOE accounted for 48%; conversely, in Catalonia, the Popular Party accounted for less than 1% of all mayors, and PSOE around 20%. Following the 2015 local elections, the number of municipalities governed by the Popular Party further decreased in Andalusia. Finally, whereas political ideology has been found to be a strongly influential factor in the contracting out of water services in Andalusia (Picazo-Tadeo et al. 2012), with conservative local governments privatizing more frequently, available evidence for Catalonia indicates that ideology has not played a key role in this decision (Miralles 2009).
Taking all the abovementioned facts into account, it is our belief that the strength of Aqualia in Andalusia is not related to the Popular Party being particularly strong in the region, but rather to the fact that municipalities governed by this conservative party tend to contract out water more frequently. On the other hand, the predominance of Agbar in Catalonia could be explained by historical reasons; Agbar, which is the oldest player in the Spanish water sector, was created in 1867 to deliver water services in the city of Barcelona, the capital of Catalonia, and it has its headquarters in this city. Although empirical studies on the urban water service in other Spanish regions are scarce, it is worth noting the case of the Valencian Community, where the overrepresentation of the group Others can be explained by the dominant presence of a regional company, Aguas de Valencia, which, though long-established in the area, has not expanded outside the region.
Finally, the political variables intended to capture the effect of the party in power in the local government are the following: PSOE, which is a dummy that takes the value 1 if the left-wing Spanish Socialist Workers' Party was in power when the decision to privatize was taken and 0 otherwise; and PP, which takes the value 1 if the rightwing Popular Party was ruling the local government at the time of privatization and 0 otherwise. In both cases the data come from the Spanish Ministry of Finance and Public Administration. In addition, we have included two further variables, namely PSOE*Reform 2007 and PP*Reform 2007, computed as the interaction between the variables PSOE and PP, on the one hand, and the policy variable Reform 2007, on the other. These two variables are intended to test whether the reforms weakened (or not) political connections. Furthermore, majority is a dummy variable that takes the value 1 if the party in power held a majority when the decision to privatize was taken, and 0 otherwise; the number of city councilors defines this variable, with data sourced from the Spanish Ministry of Home Affairs. And finally, continuity is a dummy that takes of the value 1 if the decision to privatize was taken in the second or a subsequent term of office of the party ruling the town council, and 0 otherwise.

A brief methodological note
In order to formalize the choice of the firm selected to privatize the urban water service, represented by the variable Y, and its determinants we use multinomial logit regression techniques (see Greene 2012: 763-766 for details). In particular, so as to identify the model, we compare the probability that municipality i selects firm m to privatize the urban water service, where m = Aqualia and Agbar, against the base category which is set to Others. In formal terms, our model is:

Ln
Probability where policy variable is a dummy for the 2007 reform, while market variables, regional variables and political variables are different vectors of variables included in our models aimed at capturing features of the municipality and the party in power of the local government, as explained in Sect. 4.1.
This expression has been fitted by maximum likelihood using Stata software, with robust standard errors to account for the presence of heterogeneity in the sample. In addition, and provided that the coefficients from the multinomial logit do not have a direct interpretation, i.e., they are relative to the base outcome, we have computed the marginal effects that measure the effect of changing the value of each explanatory variable on the probability of observing a given outcome: Aqualia, Agbar or Others.

Results
The results for the estimated marginal effects of the explanatory variables included in our analysis are in Table 3. The model is jointly significant and the goodnessof-fit is high compared to other studies conducting logistic regressions. Coefficients for variables are presented grouped according to the classification mentioned above: policy variable (Reform 2007), market variables (population and economic activity), regional variables (dummy regional variables) and political variables (dummies for political party, either PSOE or PP, interactions between political parties and the policy variable, majority and continuity).  Standard errors robust to heteroskedasticity in brackets *, ** and *** mean statistical significance at 10, 5 and 1%, respectively Our main variable used to assess the possible existence of a structural change produced by the 2007 reforms is the policy variable Reform 2007. Its associated coefficient is negative and statistically significant at the 5% confidence level for Agbar, but positive and statistically significant, also at 5%, for the group Others. This result indicates that the reform was correlated with an increase in contracts awarded to the smaller companies at the expense of one of the larger groups, Agbar. The magnitude of the marginal effect is slightly lower in the former case than in the latter. This result constitutes our first evidence of how the reforms seem to have affected contracting choices.
Market variables population and economic activity are not significant at standard confidence levels in this first model. 13 The estimated parameters for the set of regional variables reflects some regional unobservable patterns characterized by the greater relative presence of some companies in certain areas of Spain. This may be due to a business strategy phenomenon: companies that are the first to enter a particular area may have an advantage when it comes to business expansion in that area, whether due to their better understanding of the area, and/or due to an emulation effect, whereby municipalities tend to follow the choices of their neighbors.
The political variables majority and continuity are significant, in both cases at 5%, only for the contracts awarded to the group Others. Governments of political parties that enjoy a majority have a higher probability of choosing firms other than Agbar and Aqualia. Furthermore, governments that have been re-elected are less inclined to contract firms other than the two major ones. Perhaps the extended period spent in local government facilitates the creation of networks that ultimately promote the awarding of contracts to the major firms. 14 Beyond the reform of 2007, one of our aims is to analyze the role of political parties and their links to the big companies at the time when privatization took place, which allows us to test our hypothesis regarding how legal reforms weakened political connections. Table 3 above shows some evidence related to the relationship between political parties and firms. In the case of Agbar, on the one hand, the marginal effects of variables PSOE and PP are negative and statistically significant at the 10 and 5% confidence levels, respectively. This result suggests that when one of these parties governs the municipality, the probability of Agbar being awarded the contract is lower; furthermore, the magnitude of the effect is larger in the case of the Popular Party. Thus, Agbar seems to perform poorly compared to other companies in municipalities ruled by large political parties, especially in those ruled by the Popular Party. This is consistent with the actual evidence on political connections, given that Agbar is the main competitor of Aqualia and has not been officially involved with the Popular Party in rent-seeking activities or revolving doors, unlike the latter. On the other hand, we also find a positive and statistically significant (at 5%) relationship between the variable PP and Aqualia, indicating that when the Popular Party rules the municipality, the probability of Aqualia being awarded the contract increases by 13.1%. In addition, the estimated marginal effect for the variable PP*Reform 2007 in the equation for Aqualia is negative and also statistically significant at the 10% confidence level, indicating that the passing of Law 30/2007 and Organic Law 8/2007 weakened this relationship. 15 Furthermore, we have separately estimated two additional multinomial regressions for the periods 1984-2006, before the implementation of the 2007 reform, and 2007-2016, after the passing of Law 30/2007 and Organic Law 8/2007. The results obtained show, once again, that the Popular Party is positively associated with Aqualia (at 5% significance level) in the period prior to the reform and negatively associated with Agbar (in this case, at a confidence level of 1%), its main competitor (Table 4). However, this relationship disappears if the sample considers only the contracts awarded in the period after the reform, i.e., between 2007 and 2016 (Table 5). Accordingly, the probability of Agbar being awarded the contract for the management of urban water services between 1984 and 2006 was 19.8% lower when the Popular Party governed the municipality. In contrast, the probability of Aqualia being awarded the contract was 15.2% higher with a local government ruled by the Popular Party. After the reforms of 2007, as mentioned, no specific relationship is identified in terms of statistical significance between political parties ruling the local government and firms awarded contracts.
Moreover, a change in the patterns of the regional marginal effects is also observed between the periods 1984-2006 and 2007-2016. This mostly consists of an increaseafter the 2007 reform-of the probability of contracts awarded to firms in the group Others in all regions, to the detriment of the two larger firms, Aqualia and Agbar. Although these changes have no simple interpretation, we believe that they might reflect the fact that the reforms carried out 2007, which affected all Spanish regions equally by introducing transparency and reducing discretionary power in contracting out public services, fostered a convergence in the regional patterns of contracting out. They could also be interpreted in the context of a recent global ideological movement, widely-reported in the media, against privatization and the control of urban water services by large multinationals. In this context, once the decision has been made to privatize, local governments could prefer to contract regional companies in order to reduce public opposition.  Standard errors robust to heteroskedasticity in brackets * , * * and * * * mean statistical significance at 10, 5 and 1%, respectively Finally, we have implemented a couple of additional tests aimed at further checking the robustness of the abovementioned results. On the one hand, we carried out several falsification tests substituting the real reform implemented in 2007 with some placebo reforms, as if the regulatory changes had occurred at other previous points in time. Unlike the variables representing the interaction between political parties and the real reform, we expect interaction with these placebo variables not to be statistically significant. Results from this robustness test, which replicates the methods and specifications considered so far, are presented in Table 6; for the sake of simplicity, it only displays the estimated coefficients for our variables of interest in the case of the multinomial regression. Results are satisfactory, in that they show that only the real reform produces a statistically significant decrease in the contracting relationship  Standard errors robust to heteroskedasticity in brackets * , * * and * * * mean statistical significance at 10, 5 and 1%, respectively between the Popular Party and Aqualia, underlining the reforms' impact on political connections. 16 On the other hand, we also run a regression with year fixed effects considering the whole period of analysis  replacing the policy dummy variable. The year not included in the model that acts as benchmark by which to interpret the rest of the year specific dummies is 2007, the year of the reforms. Our results for this analysis are presented in Table 7, which only provides information regarding coefficients of these year-specific fixed effects and coefficients associated with political variables. Note, however, that the rest of the regressors-as well as all the other year dummies-were also included in the regression. Results show a sharp change in the marginal effects for Aqualia after 2007. The coefficients of years 2008 and 2009 are large, negative and statistically significant, indicating that Aqualia suffered a reduction in the probability of being awarded a water distribution contract in the first two years after the reform. In contrast, Agbar saw an increase in this probability in the same period. Moreover, the coefficient of the binary variable indicating whether the Popular Party rules the government remains positive and statistically significant for Aqualia, and negative and statistically significant for Agbar. Results from all these robustness tests provide further evidence to support the conclusions of our research. The reform made the market for contracting out more transparent and competitive, and both big companies-Aqualia and Agbar-lost market share with respect to the others. However, apparently only Aqualia experienced a systematic loss of the contracts that it had been receiving from the Popular Party before 2007; in the case of Agbar, the decrease in the number of contracts was not linked to any connection to a particular political party.
Are the relationships found in our empirical analysis between political parties and water distribution firms mere coincidence? Or, on the contrary, are they the result of favoritism or discrimination stemming from firms' different behavior with respect to funding political parties and/or revolving-doors practices? While there is no publiclyavailable evidence that irrefutably points to a causal link, it is public knowledge that FCC, the holding company of the water distribution firm showing a significant association with water service concessions made by PP-governed municipalities, has been the second biggest donor to that party in the past decade. 17 Moreover, several major Standard errors robust to heteroskedasticity in brackets * , * * and * * * mean statistical significance at 10, 5 and 1%, respectively. This model also includes regional fixed effects and control variables population and economic activity, as well as the rest of the year fixed effects, which are not displayed in the Table   Popular Party politicians have come to occupy seats on the FCC Board of Directors after leaving their institutional responsibilities, as described in Section 2. Conversely, the company Agbar, which showed a significant negative association with the concessions for water services granted by PP-governed municipalities, is not among the list of donors to the Popular Party, nor were prominent members of the PP appointed to its Board of Directors. Of course, showing favoritism to particular firms in local concessions can be the result of atomized decisions by local governments seeking rents or other compensation. In fact, cases under judicial review in Spanish provinces such as A Coruña, Asturias or Huelva involve municipalities governed by different political parties, within each Footnote 17 continued for instance, to US politics; by way of example, the Popular Party spent a total of e22.7 million in the 2008 national election campaign.
province. But beyond these isolated, atomized decisions, our results provide evidence that may well suggest that higher-level decisions based on the company's political connections and political party funding at the national level might have played a significant role in the awarding of water concessions in Spain over the last three decades. Furthermore, our results also suggest that legal reforms enacted in 2007, aimed at improving public procurement and the financing of political parties, weakened favoritism. Further research on these issues is definitely needed.

Summary, conclusions and policy implications
Contracting out of public services is an area of public policy in which political rentseeking and favoritism are believed to be relatively common. If that is in fact the case, eradicating such practices through legal reforms to weaken political connections would undoubtedly be in the public interest. However, a problem associated with the adoption of remedial measures is the lack of precise knowledge of the extent of these practices; since these are illegal or irregular practices, they are systematically hidden.
In this research, we evaluate legal reforms aimed at weakening the existence of potential relationships between political parties in governments that award contracts, and private firms that secure these contracts. The context of our study is the widespread concern over corruption and favoritism in Spain, where many cases of public services procurement are subject to judicial investigations, and the 2007 reforms of public sector contracting and the financing of political parties. We sought to understand whether the apparent biases in favor of a specific firm-taken as a proxy for favoritism (or the opposite, in cases of negative bias)-are only isolated events or, conversely, whether they might be part of a more complex network of decision-making aimed either at financing political parties, and/or rewarding firms' political connections. In order to shed more light on this possibility, we look for statistically significant relationships between political parties in local government and firms awarded contracts for provision of public services, which could suggest systematic favorable (or contrarian) treatment. Moreover, we analyze how legal reforms might affect and potentially weaken these systematic relationships.
The results we have obtained, which are robust to different model specifications and statistical treatments, could be taken as evidence of political favoritism shown by a political party to one of the leading business groups with interests in the Spanish water industry. City councils governed by the Popular Party were more likely to award the service contract to the company Aqualia, and were also less likely to award it to Agbar-Aqualia's main competitor. Our results are consistent with the facts that, as proven in judicial investigations, the owner of Aqualia-Fomento de Construcciones y Contratas (FCC)-was the second largest donor of funds to the Popular Party between 2002 and 2009. On the contrary, neither Agbar nor its owner company-Suez Environment-have appeared in the lists of donors to the Popular Party. Furthermore, it is publicly known that several politicians that enjoyed high positions of office, including national ministries and regional ministries, in Popular Party governments, have regularly occupied positions on the Board of Directors of Fomento de Construcciones y Contratas (after leaving institutional office), whereas no former national minister or regional minister-whether from the Popular Party or the Spanish Socialist Workers' Party-has been a member of the Board of Directors of Agbar or Suez Environment in the period we study. More interestingly, our results also suggest that this systematic relationship weakened after the legal reforms carried out in 2007, which aimed to ensure a more transparent and open contracting process.
The abovementioned results can be interpreted, in our opinion, as evidence supporting the need for changes to the regulatory and institutional framework to weaken political connections. In the case of Spain, these reforms were in the public interest, and intended to erode connections between political parties and firms. Our results were indeed consistent with perceptions of corruption shown by indicators developed by Transparency International (Transparency International 2014) and the RAND Corporation (Stanley et al. 2014). Improving the regulation of political parties' funding, as well as delivering faster and tougher sanctions in proven cases of corruption, could further reduce the temptation to engage in illegal practices. Additionally, ensuring greater transparency in public procurement processes, and building more effective mechanisms to monitor the contracting process, as in the case studied, would make it more difficult to engage in favoritism (or discrimination) when awarding public contracts. Of course, governments and legislators are responsible for introducing measures that aim to achieve such improvements.
Furthermore, it is also our belief that these policy implications are of particular relevance for those policymakers that tend to positively value cooperation between the public and private sector in the delivery of public services. In fact, suspicion of political favoritism and corruption has been one of the key drivers of remunicipalization of local services in the past; and is also a factor in the current wave of remunicipalization spreading across several countries, particularly in Europe. Making favoritism in public procurement more difficult to engage in, and thus less common, might well be a good way to reduce pressure in favor of remunicipalization, thus helping to preserve cooperation between public and private actors as a tool for public sector reform.