Anti-corruption barriers, financial transparency and accounting quality in political parties: Evidence from Spain

ABSTRACT

The aim of this paper is to study the transparency and quality of the Spanish political parties' accounting information, as well as the factors that determine it, in a temporal context marked by the introduction of the institutional anti-corruption barriers, which harden the obligations in terms of financial transparency. The sample is composed by all the political organizations audited in the Auditors Court (AC) during the period analysed. The econometric model is estimated through the statistical methodologies of the linear Panel Data fixed effects and the ordered logistic regression, both with different types of accounting information. One formed with original accounting data, and another corrected and improved using the AC reports. The results show that the quality of the accounting information of the political parties is deficient and so is the level of the financial transparency. Likewise, the explanatory power of the models estimated with corrected accounting information is more significant, being the ratio of the private and public income, age, debt and the saving rate, variables that strongly affect the quality of the accounting information. Finally, the institutional anti-corruption barriers imposed in terms of financial transparency have not been enough. In fact, last year was the worst in the time series, since there is no party that discloses the quality financial information.

Barreras anticorrupción, transparencia financiera y calidad de la información contable en los partidos políticos: evidencia en España

RESUMEN

Este trabajo estudia la transparencia y la calidad de la información contable de los partidos políticos españoles, así como los factores que condicionan la misma, en un contexto temporal marcado por la implementación de una serie de barreras institucionales anticorrupción, que endurecen las obligaciones en materia de transparencia financiera. Se utiliza una muestra exhaustiva compuesta por todas las organizaciones políticas fiscalizadas por el Tribunal de Cuentas (TC) en el período analizado. Se implementan modelos lineales de Datos de Panel con efectos fijos y una regresión logística ordenada, ambos con diferentes tipos de información contable, una información original y otra corregida y mejorada utilizando los Informes del TC. Los resultados muestran que, la calidad de la información contable de los partidos es deficiente y, por lo tanto, también el nivel de transparencia financiera. Asimismo, el poder explicativo de los modelos que se estiman con información contable corregida es mayor, siendo la relación ingresos privados y públicos, la antigüedad, el endeudamiento y la tasa de ahorro, variables que afectan de forma significativa a la calidad de la información contable. Se concluye que las barreras institucionales anticorrupción impuestas en materia de transparencia financiera no han sido suficientes, siendo el último ejercicio evaluado, el peor de la serie estudiada, ya que no existe ningún partido que revele información financiera de calidad.

1. Introduction

Corruption is a global problem that affects both developing and high-income countries (Chen et al., 2020; Xu et al., 2019; Benito, 2016). In Spain, the concern about this controversial issue has increased according to the data revealed by the Center for Sociological Research (CIS, 2019). Likewise, the evidence suggests that cases of political corruption are more relevant than those related to civil servant corruption (Villoria & Jiménez, 2012). The magnitude of the cases that have spilled over into the parties during the last decade, as well as the deterioration they represent for the democracy, explains this concern, as well as the search for strategies to combat them.

Political corruption can be analyzed from the point of view of the agency theory (Jensen & Meckling, 1976). In the relationship established between the political parties and citizens, the political party assumes the role of “agent” and acts on behalf of the “principal”, the citizens, representing their interests and working for the collective goodness. Nevertheless, there are individuals, organizations and interest groups, more or less visible but institutionalized, that try to alter this "contract" through corrupt acts (Argandoña, 2001). Thus, the phenomenon of political corruption could be defined as the secret transgression by the political party (agent) in the contract established with the citizenry (principal), with the aim of favoring a third party (Della Porta & Vannucci, 1997).

From this perspective, in order to limit the political corruption, it is crucial for the “principal” to control the “agent”. However, the divergences between the information available to both actors, take on a great barrier in this control process (Martínez-Cousinou & Andersson, 2009). Accordingly, the democratic political systems such as the ones in Spain, have created institutional barriers to fight against the political corruption. In this regard, the accounting standardization (Rodríguez López & Fidalgo Cerviño, 2009; Rodríguez López, 2011), the obligations in terms of transparency (Rodríguez López, 2010), the tightening of oversight by the Auditors Court (Rodríguez López & Fidalgo Cerviño, 2011b; Ríos et al., 2018), as well as the legal provisions regulating the economic and financial activities of the parties, are basic institutional strategies to combat the corruption involving these organizations. Thus, this research seeks to analyze the quality and transparency of the accounting information of the Spanish political parties as institutional barriers against corruption and the factors that condition it.

For Everett et al. (2007), such strategies are part of an orthodox approach or conception of corruption proposed by international bodies such as the World Bank, the International Monetary Fund or the NGO Transparency International. Likewise, Neu et al. (2013), argue that while such barriers may hinder corrupt practices, they do not eradicate the problem entirely. From this perspective, the institutional barriers to the corruption configure new environments that condition and define new ways of buying influence and fraud (Bourdieu and Wacquant, 1992).

From a Latin American perspective, Santano (2019) have developed an approach to the legalistic corruption based on a proposal for anti-corruption legal measures. According to this research, the establishment of control and accountability mechanisms for the parties is essential to strengthen the fight against corruption. More specifically, Santano (2019) proposes as anti-corruption measures, a set of sanctions that may affect the personality and legal status of the parties, stating that such measures may act as a deterrent to the corrupt agents that use party structures for criminal practice.

Finally, in the Spanish context, the approval of the Organic Law on Party Financing (OLFPP, Organic Law 8/2007), as well as its subsequent modifications (2012 and 2015) attempted to increase the legal barriers to corruption. For some authors, these regulations respond to the matter surrounding the opacity and the financial irregularities of the parties and their connection with corruption (Villoria & Jiménez, 2012). Nevertheless, other authors believe that the regulations (OLFPP 2012, OLCAEFPP 2015) left several cracks and unresolved issues open, which could be exploited by corrupt actors (García-Viñuela & García Aguilar, 2011; Castro, 2014; Rodríguez Teruel, 2016; Sanjurjo Rivo, 2018).

Otherwise, the barriers related to the party accounting seek to ensure that the information provided by them serves as a tool to control the origin, magnitude and destination of the party's financial resources in order to reduce the asymmetry of information between the political parties and the citizens (Van Helden, 2016; Ouda & Klischewski, 2019). Nevertheless, Rodríguez López (2011) detects numerous inaccuracies, errors and an incorrect application of the accounting principles and rules, raising the need to develop a specific accounting plan for the political parties in Spain, which would homogenize the accounting information prepared by them.

After this analysis, the institutional accounting barriers, which range from the norms of accounting harmonization to the systems of control and auditing of the party accounts, are not efficient enough to achieve the condition of "effective" strategy to prevent corruption. Thus, the analysis of the factors that affect the disclosure of low-quality accounting information by the parties, and consequently the susceptibility to illicit corrupt behavior, is an issue that has not been analyzed to date by the academic literature. In fact, except for the references cited, which focused on pointing out descriptively the accounting errors of these organizations, there is no scientific research that has explored, with analytical models, the factors that affect the majority of the Spanish political parties to reveal financial information of low quality and, therefore, not very transparent. Likewise, the relevance of the party, as an essential pillar which guarantees the proper functioning of the democratic system, along with the controversy raised by the corruption and the lack of accounting diligence, justify the importance and originality of this research in order to address and fill this major gap in the academic literature.

The aim of this work is to analyze the factors affecting the quality and transparency of the accounting information disclosed by the Spanish political parties in the period 2013-15, considering that the accounting information quality contributes to reducing the asymmetry of information between political formations and citizens and therefore, the agency conflict. To this aim, we propose a system to measure the quality of the accounting information provided by the parties to different stakeholders. In order to achieve this objective, the statistical methodologies of the linear Panel Data fixed effects and the ordered logistic regression, are used to estimate the econometric models. Likewise, each model is estimated with different types of accounting information, one formed with original accounting data and another corrected and improved, using the AC reports. Lastly, this paper contributes to broadening the debate about the efficiency of the accountability process of the political parties. In addition, the findings provide valuable insights to the regulator to revise the current regulation.

2. Transparency and accounting quality in political parties

2.1. Transparency of accounting information

From a functional conception, financial transparency allows organizations to be more consistent, playing a relevant and fundamental role in the accountability process (Benito et al., 2006; Iyoha & Oyerinde, 2010; Bauhr & Grimes, 2014; Murphy & Albu, 2018). Thus, parties must adopt transparency as one of their essential operating principles because of the important role they play in democratic societies (Lederman et al., 2005), also, because of the large volume of public funds they receive as a source of financing (Rodríguez López et al., 2011b) and finally, because of the implications of this principle as a tool to control corruption (Lindgreen, 2004; Lehman & Thorne, 2015; Sargiacomo et al., 2015; Grossi & Pianezzi, 2018). This highlights the functional approach to accounting, which explains how accounting helps fight against corruption by building institutional barriers and elevating accountants to virtuous actors in this fight (Gray, 1992; Everett et al., 2007; Johnston, 2015).

Nevertheless, with the mission of ensuring that parties comply with the principle of transparency, the legislature has approved a set of rules¹ with a clear vocation for control. These rules include control measures in the economic and financial activity of the parties and require the disclosure of certain financial and accounting information, such as the annual financial report, private donations, public subsidies, and banking debt on their websites. In addition, this should be reported to AC.

Among the reasons that motivated this intense legislative activity, we can highlight first, the high degree of disaffection of the citizens towards the political activity derived from the financial crisis and the notorious corruption and illegal financing scandals in which some parties were involved. At the same time, we must consider the rise of new emerging political formations, which exerted pressure for an improvement in transparency.

Secondly, it is necessary to highlight the pressure received from supranational organizations in which Spain is a member. In this regard, GRECO (Group of States against Corruption), a body of the Council of Europe, plays a key role in exerting pressure on the legislator. Thus, GRECO, with the aim of placing limits on the party corruption, proposes a series of recommendations with the objective of increasing financial transparency. Spain, with the approval of the regulations, has managed to comply with all GRECO's recommendations, managing to improve the legal framework regarding the financial transparency of the parties (Smulders & Maddens, 2016).

On the other hand, despite the fact that the regulations described above are not free from criticism, the increase in demands for financial transparency has had a positive effect on the volume of information disclosed by the parties. This has evolved from a situation in which there were generalized problems of opacity -especially in 2011 and 2012- (Cuadrado-Ballesteros & Vaquero-Cacho, 2015), to a slight increase in the information provided, in quantitative terms, through political parties’ Internet portals. The results of the transparency index elaborated by the Foundation “Compromiso y Transparencia” during the period 2011-2017, corroborate this, and can be seen in Table 1.

Table 1. Financial transparency of Spanish political parties: % economic information published on Internet portals

	2011	2012	2013	2014	2015	2016	2017
Economic information	3%	14%	21%	40%	29%	43%	56%
Balance Sheet and Income Statement	7%	22%	33%	56%	45%	56%	74%
Explanatory memory	-	-	17%	28%	23%	35%	35%
Breakdown of income sources	7%	28%	33%	56%	45%	56%	74%
Breakdown of expenditure	7%	28%	33%	56%	45%	56%	74%
Audit report	0%	0%	0%	0%	0%	4%	0%
Loans and credits	-	60%	17%	39%	14%	43%	52%
AC Report	0%	0%	17%	44%	33%	46%	81%

Source: own elaboration from Reports "Transparency, the best slogan", years 2011-2017, Foundation Compromiso y Transparencia.

Even though the quantitative levels of financial transparency have increased slightly since the approval of the set of norms, the index of financial disclosure has certain limitations, which prevent us from knowing the real situation of the parties’ transparency. The most important is that this index only measures the number of pieces of evidence, in the quantitative terms that parties provide on their websites, leaving aside the qualitative evaluation of the information disclosed and not assessing whether the documentation is correctly prepared or complete, that is, if it is of good quality.

In this vein, the possibility of accessing data does not guarantee a proper understanding of it. Financial transparency should not be defined exclusively by the amount of information available, but rather by its explanatory capacity, which allows the understanding of the economic and financial reality of the organization analyzed. Likewise, the accounting disclosures should contain a sufficient level of detail to facilitate their understanding by external users, but they should avoid overload information that could generate confusion (Barth & Schipper, 2008).

2.2. Quality of accounting information

Consequently, it is considered that, for the financial information to be truly transparent and to contribute as a barrier to corruption, it must meet both, quantitative requirements and qualitative attributes to satisfy the needs of internal and external users, provided that such attributes are consistent with the objectives defined by the organization (Benito & Martínez, 2002; Hope et al., 2017; Caruana & Farrugia, 2018).

However, until the approval of the Accounting Plan Adapted to Political Formations (APAPF) in 2013, the parties did not have a standardized accounting framework adapted to the intrinsic and specific particularities of their economic-financial activity (Rodríguez López & Fidalgo Cerviño, 2012a; 2012b). Thus, the basic quality attributes of financial information described in the conceptual framework of the APAPF (Resolution 8 of the Presidency AC, 2013, p. 85.361), are two; to be more precise, reliability and relevance. Therefore, these are the attributes that the accounting information must contain in order to be considered of quality (Ruíz Barbadillo et al., 2002; Van Beest et al., 2009; Batta et al., 2014; Herath & Albarqi, 2017).

In response to GRECO's demands, the Organic Law 3/2015 introduces, for the first time, as an explicit function of the AC, the oversight of the economic-financial activity of the parties, as well as the supervision of foundations or other entities dependent on them. Thus, the objective of this audit "is to issue an opinion on the reliability of the information provided by the financial statements of an entity based on its compliance with the principles, criteria and accounting standards applicable to it" (Auditors Court, 2013, p. 16).

In this regard, the AC issues an opinion on the reliability of the annual accounts, which may be expressed as unqualified or clean opinion, qualified or with qualifications, and adverse or rejected. Such opinion allows to assess whether the information collected in the accounting statements of the parties contains the attributes of reliability and relevance that must have the quality financial information. From these possible opinions that the AC can grant, only an unqualified opinion on the accounting statements assures the quality of the financial information, turning such information into a necessary condition for the accountability process (Ruíz Barbadillo, 2002). Therefore, it is assumed that the opinion issued by the AC, measures the quality of the accounting of the parties under audit. Thus, this opinion is used as an element of reference to define the different levels of quality, as will be developed in section 4.

3. Hypothesis

First, the parties, as well as the rest of the non-profit organizations (NPOs), capture the financial resources both public and private, to finance their activity. The subsistence of these organizations, on many occasions, depends on the contributions of public administrations to be able to develop their projects and provide the services that are proper to them. Thus, those NPOs that receive a greater proportion of public funds reveal quality financial information, in order to forge trust with the administrations and not lose the subsidies granted (Ruíz Lozano et al., 2008). In this regard, when a public administration receives funds from another government entity, the effort to reveal a greater volume of accounting information is accentuated (Rodríguez Bolívar et al., 2013).

Spanish parties have historically been funded with a much higher proportion of public resources than private ones (López, 2010; 2011; Rodríguez López, 2012; 2014), so it is presumed that they face the same pressure to disclose quality information as NPOs or the public administrations. For this reason, it could be deduced that the higher proportion of public funds acts as an incentive for parties to ensure that they produce accounting with high standards of quality and transparency. All parties are subject to the control of the AC to receive public subsidies. However, the receipt of public funds is associated with the election results, and not with the justification for the transparency of the projects in which they are involved. So, the volume of public income they receive as subsidies does not ensure greater diligence in the process of preparing accounting information.

On the other hand, funding from private sources has proven to be more problematic in terms of transparency and corruption during the democratic stage (García-Viñuela, 2019). In this sense, anonymous donations from individuals or legal entities have been a focus of controversy, derived from the permissiveness of the party funding regulations for many years (Rodríguez López, 2010, 2018). Until their prohibition in 2007 (OLFPP, 2007), anonymous donations from individuals and legal entities were common in center-right parties (Rodríguez Teruel, 2016). These types of contributions of unknown origin continued after their limitation, indirectly to party-controlled foundations (García-Viñuela, 2019).

It is not until the approval of the OLCAEFPP, in 2015, that the contributions made by the enterprises are definitively prohibited. These contributions could be motivated by an economic counterpart, rather than the promotion of political debate (Rodríguez Teruel, 2016). Therefore, those formations financed with a high volume of private donations could have more incentives to formulate non-transparent and low-quality accounting. Thus, it is considered that those parties that receive a greater proportion of funds from private sources may have less incentive to disclose quality accounting information. Hypothesis 1 is:

H1: There is a negative relationship between the weight of private funding received in relation to public funding and the quality of accounting information disclosed by the political party.

Secondly, the commitment of the parties to the financial transparency has not been constant and firm since the beginning of the democratic stage (Rodríguez López, 2011a). There have been important elements that have influenced this lack of commitment to transparency: the absence of a specific regulation on party financing, which was not approved until 1987 (Organic Law 3/1987) and the shortcomings of that regulation (Martínez-Cousinou, 2013), as well as the immature democratic and accountability culture of the time.

Rodríguez López & Fidalgo Cerviño (2010, 2011b) and Rodríguez López (2011), highlight the continued lack of rigor in accounting practice, materialized in numerous errors and omissions, calling into question the transparency of the parties audited by the AC until 2006. This dynamic is perpetuated to this day, as a result of the immobility of some formations and the lack of a real and effective internal and external control system (Rodríguez López et al., 2011b).

Otherwise, the temporal context analysed is characterized by important changes in the Spanish political field, such as the end of the two-party system (Lobera & García, 2017) and the rise of emerging parties (Garrido & Dominguez, 2019). This, along with the cases of corruption and illegal financing that have been uncovered and their judicialization (Rodríguez López, 2018), have given young parties the opportunity to use this as a differential factor to attract votes. Besides, they differ from the traditional organizations, achieving greater transparency, fighting corruption and being zealous in complying with the applicable legislation in this area. In this regard, the emerging parties could have increased transparency to reduce the information asymmetry and the agency costs. Therefore, the second hypothesis is:

H2: There is a negative relationship between the age of the party and the quality of the political party’s accounting information.

Third, research in the business sector shows that companies with high financial leverage are more likely to obtain unfavorable audit reports due to the risk assumed by such organizations (DeFond et al., 2002; Mareque et al., 2019).

In this way, several authors have questioned the viability and survival of the Spanish political parties, derived from the excessive accumulated indebtedness (Rodríguez López, 2011; Rodríguez Teruel, 2016). Other line of research shows that after the modifications introduced in the reform of the OLFPP in 2012 the number of parties with high levels of indebtedness has decreased – this law limits the cancellation of debt and demands greater disclosure of information- (Rodríguez Gutiérrez & Rodríguez López, 2018; Cavero-Rubio & Gónzalez, 2021).

Despite all this, there is still a group of political formations that maintain important financial imbalances, an issue that could affect their survival (Rodríguez Gutiérrez & Rodríguez López, 2018). Thus, the economic-financial management of those formations could be questioned, and they could be less compliant with the accounting regulations, in order to hide or disguise a deficient administration of the party's resources.

In addition, the financial relationship between the political parties and the banking sector has been marked by the opacity and the lack of transparency (Sanjurjo Rivo, 2018), at least until 2012. The origin of this problem lies in the bank loans, which meant that on many occasions, debts were postponed indefinitely, or simply condoned (Rodríguez Teruel, 2016) becoming de facto in hidden donations (Rodríguez López, 2011).

For all these reasons, it is expected that political formations with high financial leverage do not provide quality accounting information. Thus, according with the above, the third hypothesis is:

H3: There is a negative relationship between debt and the quality of the political party’s accounting information.

Fourth, the elaboration of accounting plans by the legislator has been developed to systematize accounting operations in favor of transparency (Chivite & Nogales, 2016). Until the approval of the APAPF in 2013, the political party did not have a standardized accounting framework of reference adapted to the intrinsic and specific particularities of its financial and economic activity (Rodríguez López & Fidalgo Cerviño, 2009).

Thus, Rodríguez López & Fidalgo Cerviño (2009, p. 5), defend the need to develop this standard, "with the aim of harmonizing the mode, procedures, criteria for the preparation, synthesis and supply of the accounting information", which will allow, among other things, to overcome the limitations regarding the lack of representativeness of the accounting information revealed by the AC. Along with the same lines, Rodríguez López et al. (2011b) and Rodríguez López & Fidalgo Cerviño (2012) argue the need to implement an accounting plan that homogenizes the economic and financial information of the political parties, with the aim of increasing the quality and transparency in this area.

Despite the fact that the norm was approved in 2013 -although the APAPF was not obligatory until 2019-, throughout the years investigated, we have found political parties that formulate their annual accounts following the criteria and models of other accounting plans, such as the General Accounting Plan of 2007 or the Accounting Plan of NPO. In this regard and considering the existence of this reference standard, we understand that the accounting information formulated following the APAPF of 2013, must be better than the one formulated by the rest of the parties that do not follow this sectorial plan. Therefore, it is formulated as hypothesis 4:

H4: The use of the APAPF, as a reference framework for the elaboration and formulation of accounting information, positively affects the quality of accounting information produced by the political parties.

Finally, like NPOs, political parties are not profit-oriented and therefore the result or surplus on the income statement should be positive, but moderate. In this sense, according to the APAPF, "the result of the year reflects the level of the savings produced during the year as a function of the party's own funds, which the party has decided to maintain in order to ensure the permanence of economic activity as an instrument of its political activity" (APAPF, 2013, p. 8). In other words, the APAPF shows that the result of the exercise is articulated as a source of self-financing that reinforces the solvency of the party, but it cannot be considered as an indicator of the efficiency of the performance of the political party, since its objective is not the "maximization of profit". Thus, the registration of positive surpluses or high "savings rates" would not be congruent in this type of organization, since it would seem that they are placing the achievement of lucrative objectives before the politicians and consequently, they would have more incentives to be less transparent and secretive in the compliance with the accounting regulations.

Consequently, it is argued that there is a negative relationship between the "savings rate" and the quality of the political party's accounting information because many qualifications could be due to the non-recognition of certain positive results or surpluses linked to profit-making purposes. Therefore:

H5: There is a negative relationship between the savings rate and the quality of the political party’s accounting information.

4. Methodology

4.1. Sample and time series

The political parties under study are those that have been audited by the AC in any of the years analyzed (2013-15) (88 observations²). The total number of formations belonging to the sample is 32, including the parties’ coalitions, federations and the political parties themselves. This is an exhaustive sample made up of all the parties audited in the years that make up the time series. The organizations that make up the sample, in alphabetical order, are specified in Table 2.

Table 2. Sample of political parties

Alternatiba Eraikitizen	AE
Amaiur	AM
Aralar	AR
Bloc Nacionalista Valenciá	BNV
Bloque Nacionalista Galego	BNG
Chunta Aragonesista	CHA
Ciudadanos	CDN
Coalición Canaria	CC
Compromís	COM
Compromís Q	COMQ
Convergencia Democrática de Catalunya	CDC
Convegencia i Unió	CIU
Esquerra Republicana de Catalunya	ERC
Esquerra Unida i Alternativa	EUiA
Euskal Herria Bildu	EH Bildu
Eusko Alkartasuna	EA
Foro de Ciudadanos	FAC
Geroa Bai	GB
Iniciativa per-Catalunya Verds	ICV
Izquierda Unida	IU
Nueva Canarias	NC
Partido Aragonés	PAR
Partido Comunista Español	PC
Partido Nacionalista Vasco	PNV
Partido Popular	PP
Partit dels Socialistes de Catalunya	PSC
Partido Socialista Obrero Español	PSOE
Podemos	POD
Sortu	SOR
Unió Democrática de Catalunya	UDC
Unión del Pueblo Navarro	UPN
Unión Progreso y Democracia	UPD

Source: own elaboration

The accounting years analyzed goes from 2013 to 2015. The limitation of the available information explains the time series. Since the quality of the accounting information is measured through the opinion issued by the AC on the representativeness of the annual accounts of the parties, and the opinion of the audit body does not begin to be published until the 2013 Audit Report, it is not possible to extend the time series to the previous years. In turn, the 2015 Audit Report is the last one published at the end of this work.

The sample has been built from data collected from the Audit Reports of the Accounting Statements of the parties issued by the AC and from the contributions received by the foundations and other entities linked or dependent on them, as well as other secondary sources of information, such as the web page of the parties.

Part of the data that makes up the sample is of accounting origin. This type of information has several limitations, such as the lack of reliability derived from accounting make-up (Blake et al., 2000; Gowthorpe & Amat, 2005). Likewise, authors such as Hopkin (2004) or Scarrow (2007) have shown that the information collected in the accounting statements of the parties can be questionable and may contain inaccuracies and only a part of their patrimonial reality. As a consequence, some accounting items are sometimes overvalued or undervalued. In order to filter out such errors and to use more reliable and accurate information in the statistical analysis phase, the variables of an accounting nature, such as the relationship between the private income and the public one, debt or party size, have been elaborated twice, using original and modified data, as explained in the following paragraph.

On the one hand the variables described have been constructed with the information from the financial statements provided by the parties to the AC, without any manipulation or correction, that is, original data. On the other hand, the same variables have been prepared by correcting the errors that the AC collects as qualifications, named, modified data. We expect that the variable constructed with corrected information contains more reliable information than the first one, given that the information provided by the parties has been improved, with the contributions of the AC's auditors.

4.2. Variables

4.2.1. Dependent Variable (QUAL)

Audit reports are considered a valid tool to know the veracity and reliability of the information provided by the Financial Statements (Sunder, 2016; Slack & Tsalavoutas, 2018; Jeppesen, 2019; Mareque et al., 2019). Thus, the evaluation of the relevance and reliability of the accounting information can be assimilated to the opinion of the audit reports, which is the judgment issued by the auditor on the quality of the accounting disclosures (Ruíz Barbadillo et al., 2002; Duréndez & Sánchez Vidal, 2008; Xu et al., 2022). Consequently, we use the opinions issued by AC on the representativeness of the parties' financial statements to measure the quality of the accounting information.

In this way, the AC refers to Auditing Standards that define its scope of action (Auditors Court, 2013, Rule 79) as the duty by the AC to issue in the Audit Reports the opinion about the accounting statements, classifying it as: unqualified, qualified, adverse and disclaimer, the last when AC does not have the necessary information to carry out the relevant audit tests or when such tests cannot be carried out. In this regard, Ruíz Barbadillo et al. (2002) proposes a dichotomous variable to codify the quality of the accounting information. The reason for this is that, in auditing accounts in the business sector, auditors are not usually inclined to issue unfavorable or negative opinions (Barbadillo et al., 2006; Duréndez & Sánchez Vidal, 2008; Xu et al., 2011).

However, in the case of parties, the Audit Reports show very different results than in the corporate sector (Table 3). The number of adverse opinions issued by the AC during the period studied is 20.7%, while the number of opinions denied is 4.3%. On the other hand, 67.4% of opinions issued are qualified with exceptions or qualifications and only 7.6% of observations correspond to unqualified opinions. Thus, considering the significant weight of opinions with qualifications and unfavorable opinions, and in order to analyze the different accounting quality realities that exist, we do not define the quality of accounting information with a dichotomous variable.

Table 3. Distribution of opinions of parties' accounting statements

Opinion	Cases	%
Unqualified	7	7,60%
Qualified	62	67,40%
Adverse	19	20,70%
Disclaimer	4	4,30%
Total	92	100,00%

Source: own elaboration based on AC data, 2013-2015.

On the other hand, analyzing the observations qualified with exceptions, there are parties that present only one exception, while others have up to 8. In this set of observations, the average number is 2.42 and the media is 2. Thus, we take this reference to group two levels of accounting quality in a qualified audit report. Therefore, 4 levels of accounting quality are defined as follow:

• Level 3: when the accounting statements have obtained an unqualified opinion.

• Level 2: we include accounting information qualified with one or two qualifications or reservations.

• Level 1: is defined by the accounting statements when the number of qualifications is three or more.

• Level 0: the one with an adverse opinion. This is the lowest level and implies that the accounting information does not represent a true picture of the equity.

Only Level 3 corresponds to the financial information that, the AC experts qualify as representative of the faithful image of its equity situation and therefore, meets the conditions to be qualified as quality (Ruíz Barbadillo, 2002). It is assumed that the quality of the other levels is deficient, not reaching the standards to which quality accounting information should aspire and, consequently, it cannot be said to be fully transparent. Furthermore, the observations in which the parties have a disclaimer of opinion (4 in total) are not considered, since the limitation of the information available to the AC makes it impossible to assess the accounting documentation and to determine whether it is reliable and relevant. This is the case of the accounting statements of the parties UDC and UPD for the years 2014 and 2015. Therefore, they have not been included in the study.

4.2.2. Independent variables

Ratio between private and public revenue (PRIPUB). To measure the relationship between the two basic sources of parties funding, a continuous variable is defined by the following ratio: total volume of private income to total public income. This quotient allows us to capture the weight of the funds of private origin versus those of public origin. Considering the absence of previous statistic models and the specific characteristics of the Spanish political parties, it is thought that this quotient allows to capture, adequately, those parties that are proportionally financed with a greater weight of private income (Rodríguez López, 2014). The reason is that such a source of financing generates a lot of controversy, in terms of its origin and transparency (García-Viñuela, 2019), issues that are not applicable to the rest of the NPOs or commercial firms.

Age (ANTIQ). The age variable should be transformed by applying logarithms, when a very wide range is observed, (Duréndez & Sánchez Vidal, 2008). However, this transformation is not considered necessary since the maximum value of the observed data is 41 years. The age of the party is measured by the number of years that this has been active, taking as a reference the year of its foundation. In addition, to calculate the longevity of the formations founded before the Franco dictatorship, we take as a reference the year 1976, when political parties became legal again in Spain.

Debt (LEV). To measure the level of indebtedness, we have chosen to use the ratio of debt to total assets, which is usually used in research on the business environment (Barbadillo et al., 2006; Mareque et al., 2019), and also in the field of NPOs (Jegers, 2011; García-Rodríguez & Jegers, 2017).

Accounting standard (STAND). The accounting standard used by the parties to formulate their accounts is measured using a dichotomous variable, where the value 1 is assigned to the observations that have prepared the annual accounts following the APAPF. The value 0, is assigned to the observations that indicate that the accounting statements have been formulated following another accounting plan. This type of variable has been used by authors such as Chivite and Nogales (2016) to measure the type of accounting standard implemented - IAS/IFRS or GAP 90³ - and the effects on the financial transparency for profit organizations.

Savings Rate (PSAV). The savings rate variable reflects the relative capacity of the political party to "save" during the year. Thus, to determine this rate, we have chosen to use the quotient of the positive or negative result of the year to the equity (Rodríguez López, 2011).

SIZE. The size or dimension variable is introduced as a control variable. There are investigations that relate size and result of audit reports in the sector of NPOs and public foundations (Keating et al., 2005; Ruíz Lozano et al., 2008; Petrovits et al., 2011; González-Díaz et al., 2013). The relationship in the literature review is positive, the larger the size, the better are the control mechanisms and therefore, the better the quality of the accounting information. In NPOs and in the parties, the absence of profit explains why the volume of assets is not the most appropriate variable to measure their size. Therefore, the total income for the year is used to measure this variable (Tate, 2007; Ruíz Lozano et al., 2008; González-Díaz et al., 2013).

Type of political party (TPP). The AC audits different sorts of political entities, such as coalitions, federations, as well as political parties. Thereby, these political entities are among the sort of formations analysed by Rodríguez Gutiérrez & Rodríguez López (2018). In this respect, the specific characteristics of each of these political entities could also affect the quality of the accounting information. Thus, the type of political entity is introduced as a categorical control variable. The value 1 is assigned to the political parties, value 2 to the coalitions and 3 to the federations.

Law on political party financing (OLFPP)

During the period analysed, the OLCAEFPP 2015 was passed, reforming the OLFPP. Thus, a variable dummy is defined to identify the law in force during the research period (Cavero-Rubio & Gónzalez, 2021).

Territorial scope of parties (LOCAPP). Besides the different types of political parties, the sample is composed of political formations with a different territorial scope of action. On the one hand, value 0 is assigned for parties with regional scope; on the other hand, value 1 identifies the national parties. This control variable has been used by authors such as Cavero-Rubio & Gónzalez (2021).

In Table 4, the definition of the variables is shown, as well as the expected sign:

Table 4. definition of variables

Variable	Definition	Expected Sign
QUAL	0, adverse opinion; 1, opinion with more than two qualifications; 2, opinion with one or two qualifications; 3, unqualified opinion.
SIZE	Annual revenue.
PRIPUB	Quotient between private income figure and public revenue.	-
ANTIQ	Number of years the party has been in existence.	-
LEV	Ratio of debt to total financial structure (sum of liabilities and net).	-
STAND	0, political parties that do not use the new Accounting Plan; 1, political parties that do use the especial Accounting Plan (APAPF)	+
PSAV	Quotient of the result or accounting surplus between equity.	-
TPP	Type of political party: 1- political parties; 2- coalitions; 3- federations.
OLFPP	Dummy of law on political party financing: 0- 2013 and 2014 years and 1- 2015 year
LOCAPP	0- Local/regional political parties; 1- National political parties.

Source: own elaboration

4.3. Method

The econometric model defined to contrast the H1, H2, H3, H4 and H5 hypotheses is the following:

\[\begin{equation} \label{eq1} \small \begin{split} QUAL(I)_{it} =&\ \beta_{0} + \beta_{1}SIZE_{it} + \beta_{2}PRIPUB_{it} + \beta_{3}ANTIQ_{it} \\ & + \beta_{4}LEV_{it} + \beta_{5} STAND_{it} + \beta_{6}PSAV_{it} + \beta_{7}TPP_{i} \\ & + \beta_{8}OLFPP_{i} + \beta_{9}LOCAPP_{i} + Y_{t} + Z_{i} + u_{it} \ \ \ \ (1) \end{split} \end{equation}\]

In the proposed model, QUAL (accounting quality) is the dependent variable, while PRIPUB (private/public income ratio), ANTIQ (age), LEV (debt), STAND (accounting plan) and PSAV (surplus between equity) are the independent explanatory variables, alongside the following control variables: SIZE (size), TPP (type of political party), OLFPP (organic law on political parties financing) and LOCAPP (local or national parties). Finally, \(\gamma_{t}\) is a time dummy for every season and \(z_{i}\) includes a set of political parties dummies to take out the individual fixed effects.

To analyze the quality of the parties' financial statements, descriptive statistical methodology is used. The statistical methodology used to contrast the hypotheses is the Panel Data fixed effects, using OLS regression controlled by each political party. Following the results of Hausman test (Hausman, 1978), we have rejected the use of random effects. Therefore, the fixed estimator is more appropriate to determine individual effects, in our case, the political parties. On the other hand, dependent variable takes 4 values, thus the ordered model should be used. However, there would be bias associated with using fixed effects in the context of nonlinear procedures (as logit or probit models) stemming from the incidental parameter problem (Greene, 2002). Moreover, not to control by fixed effects, capturing the individual effects over time, would surely introduce much more bias due to unobserved heterogeneity (Wooldridge, 2002; Kennedy, 2008; Bai, 2009) as well as this source of endogeneity (Bliese, 2019; De Jager, 2008). In addition, this technique has a greater capacity to model the complexity of the human behavior than cross-sectional or time-series techniques (Hsiao, 2007). At last, for a comparative purpose, equation (1) has also been estimated with ordered logistic regression, using ologit command in Stata 14, controlling by political parties.

Finally, to avoid possible problems from different sources of heteroscedasticity of groups and correlation of observations in the same group and over the time (Angrist and Pischke, 2008), the variance of errors has been estimated robustly with the VCE option (INDIV cluster) in Stata 14.

5. Results

5.1. Descriptive results

Table 5. Distribution of observations by quality levels

QUAL	observations	%
Level 3	7	7,95%
Level 2	37	42,05%
Level 1	25	28,41%
Level 0	19	21,59%
TOTAL	88	100,00%

Source: own elaboration.

Table 5 shows the classification of the quality levels proposed in the definition of the dependent variable accounting quality. It should be remembered that, in the description of the quality levels, the observations with a disclaimer opinion -4 in total - have not been considered. With regard to this, it should be highlighted that the observations of political parties CDN and POD for the financial year 2013, and of PCE for the financial years 2014 and 2015, are not available, due to the fact that they were not audited by the AC⁴ for different reasons. Finally, Level 2 is the one with the highest proportion of observations, corresponding to the qualified opinions with one or two qualifications, while Level 3, which is the one with the highest quality, has the lowest number of observations.

Table 6 discusses the evolution of the accounting quality throughout the time series. Thus, it can be observed that out of the 7 observations with Level 3 quality, six correspond to the 2013 financial year, while in 2014 there is only one, and none in the 2015 financial year. As for Quality Level 2, there is a significant decrease in observations over the period 2013-15, but not as pronounced as with Level 3. At level 1, observations have no significant variations. On the contrary, level 0 observations increase considerably. In addition, 70% of the observations in the 2013 financial year correspond to Levels 3 and 2, while in 2015 the proportion drops to 34.5%. In addition, Levels 1 and 0 accumulate 31% observations in the 2013 financial year and in 2015 65.5%. Consequently, it is observed that the evolution of the accounting quality throughout the series is decreasing, implying a worsening over the period analyzed.

Table 6. Distribution of quality levels by accounting year

	Accounting exercise
QUAL	2013	2014	2015
Level 3	6	1	0
Level 2	15	12	10
Level 1	9	7	9
Level 0	0	9	10
Total	30	29	29

Table 7 analyzes the distribution of Quality Levels based on the accounting plan used. Given that the APAPF is approved at the end of 2013, but that this plan is not mandatory until 2019, the data reveal that 58.6% of the accounting statements have used the APAPF, while the rest of the formations employ another plan. In relation to Level 3 – unqualified observations – less than half have been prepared according to the APAPF, while 57.1% elaborated the accounting information with another plan. On the contrary, the data reveal that 78.9% of Level 0 observations have been developed with the specific sector-based plan. Therefore, it is highlighted that the parties that used the APAPF do not provide with higher quality information than the rest of the formations.

Table 7. Distribution of quality level by accounting plan

QUAL	APAPF	OTHER STANDARD
Level 3	3	4
Level 2	21	16
Level 1	12	13
Level 0	15	4
Total	51	37
%	58,60%	42,50%

Source: own elaboration.

Table 8 shows that the average of the dependent level variable accounting quality is 1.36, which implies, on average, low quality in the reported accounting information. The size of the parties, as a function of mean income, has been approximately nine million euros. However, there is a great divergence between parties since the standard deviation is much higher than the average. In the sample, the average of the results of the ratio from private income to public one is 1.29, increasing moderately after the recognition of adjustments from qualifications.

Table 8. Statistics of variable model

		MODIFIED DATA		ORIGINAL DATA
Variable	Obs.	Mean	Sth. Dev.	Mean	Sth. Dev.
QUAL	88	1.36	0.91	1.36	0.91
SIZE	88	9,190,359	20,300,000	9,325,017	20,400,000
PRIPUB	88	1.29	5.19	1.28	5.28
ANTIQ	88	20.28	13.79	20.28	13.79
LEV	88	0.91	1.73	0.96	1.82
PSAV	88	0.61	2.53	0.25	3.76
STAND	88	0.58	0.50	0.58	0.50
OLFPP	88	0.33	0.47	0.33	0.47
LOCAPP	88	0.17	0.38	0.17	0.38

Source: own elaboration

The average age of the analyzed parties, ANTIQ, is 20 years, the average debt ratio, LEV, is 0.96, and it is reduced to 0.91 in the modified model. However, the high standard deviation in both cases implies a high range of values in the sample. PSAV, the rate of patrimonial savings is 25%, which is very high for non-profit entities. The average rate increases even more when adjustments from qualifications are made reaching 61%. The STAND variable indicates that 58% of the accounts presented have been prepared under the APAPF or sector-based Accounting Plan. Moreover, OLFPP implies that the 38% of financial reports have been elaborated under the political parties’ financial law reformed in 2015, while the rest, the 62% under the OLFPP 2012. Finally, LOCAPP variable shows that national parties are only the 17%, while regional parties imply the rest (83%).

Table 9 presents the distribution of each type of political party, from an organizational point of view, denoting that political parties are the 58%, while the 17,05% are coalitions or federation respectively:

Table 9. Description of variable TPP

TPP	Freq.	Percent.	Cum.
1	58	65.91	65.91
2	15	17.05	82.95
3	15	17.05	100
Total	88	100

Source: own elaboration.

Otherwise, in order to have a better idea of the structure and subsequent impact of the indebtedness, a variable of three homogeneous levels has been created in Table 10, based on LEV, called LEV3, both in the original data and in the modified data.

Table 10. Description of variable LEV3

	MODIFIED DATA			ORIGINAL DATA
LEV3	Mean	Sth. Dev.	Freq.	Mean	Sth. Dev.	Freq.
1	0.09	0.08	30	0.10	0.09	30
2	0.46	0.13	29	0.51	0.12	29
3	2.21	2.58	29	2.30	2.73	29
Total	0.91	1.73	88	0.96	1.82	88

Source: own elaboration.

5.2. Results of the Panel data

5.2.1. Results of model with modified data

Based on the data collected in Table 11, in the model, no multicollinearity problems are observed due to the correlations between the variables. However, there are significant positive relationships between size (SIZE) and age (ANTIQ), and between age and debt (LEV). It is also noteworthy that in the correlation matrix all the relationships between the dependent variable, quality (QUAL) and the independent variables (excluding TPP) are negative. Furthermore, the relationships between QUAL and private and public income (PRIPUB), as well as QUAL and the financial law (OLFPP) are significant.

Table 11. Correlation matrix of modified variables

	QUAL	SIZE	PRIPUB	ANTIQ	LEV	PSAV	TPP	STAND	OLFPP	LOCAPP
QUAL	1.00
SIZE	-0.11	1.00
PRIPUB	-0.18	-0.08	1.00
ANTIQ	-0.15	0.28	-0.06	1.00
LEV	-0.12	0.08	0.14	0.22	1.00
PSAV	-0.09	-0.09	0.03	-0.17	0.01	1.00
TPP	0.06	-0.15	-0.13	-0.10	0.21	-0.07	1.00
STAND	-0.17	0.09	0.12	0.04	0.02	-0.11	0.15	1.00
OLFPP	-0.28	0.09	-0.03	0.02	0.16	-0.04	0.01	0.40	1.00
LOCAPP	-0.08	0.62	-0.04	0.05	0.15	-0.07	-0.07	-0.04	0.00	1.00

Source: own elaboration

Table 12 summarizes the coefficients of the linear panel data fixed effects, as well as the ordered log-odds estimations on the accounting quality levels. Albeit the different interpretation of coefficients, both models are basically coincident in significance and signs, in QUAL (modified data) and in QUALO (original data). For example, if a political party increases his age by one year, his ordered log-odds of being in a higher quality category would decrease by -0.50 while other variables are held constant, this negative relationship is also similar in the linear model of fixed effects.

Nevertheless, as we have mentioned previously in the proposed model section, according to the literature, the incidental parameter problem could be introduced as a consequence of controlling by fixed effects in a logit context, being this a problem not statistically solved yet. Thus, this model represents a complementary contrast of the main linear model of the panel data with fixed effects, which is analyzed in the following paragraphs.

Table 12. Results of the panel data fixed effects and ordered logistic model

	Panel data fixed effects (vce clustered)				Ordered logistic model (vce clustered)
	MODIFIED DATA		ORIGINAL DATA		MODIFIED DATA		ORIGINAL DATA
Variable	QUAL (I)		QUALO(I)		QUAL (I)		QUALO (I)
SIZE	5.23E-09		2.44E-08		5.39E-08		1.85E-07
	8.58E-09		1.76E-08		4.25E-08		1.33E-07
PRIPUB	-0.05	***	-0.05	***	-0.50	***	-1.34	**
	0.01		0.01		0.22		0.63
ANTIQ	-0.03	***	-0.03	***	-0.19	***	-0.24	***
	0.00		0.00		0.05		0.08
LEV	-0.15	**	-0.11		-1.44	**	-1.47
	0.07		0.07		0.78		1.40
PSAV	-0.09	***	-0.05		-0.65	***	-0.38	*
	0.02		0.04		0.26		0.22
TPP 2	-0.19		-0.23		0.16		-1.70
	0.15		0.15		1.68		2.63
3	-1.20	***	-1.50	***	-8.27	***	-11.16	***
	0.26		0.28		2.30		3.46
STAND	-0.08		-0.08		0.47		0.14
	0.32		0.34		1.53		1.73
OLFPP	-0.94	***	-0.92	***	-8.30	***	-8.18	*
	0.30		0.31		3.59		4.37
LOCAPP	0.17		0.19		1.66		2.81	*
	0.18		0.18		1.18		1.58
INTERCEPT/CUT1	2.87	***	2.85	***	-16.87		-18.45
	0.17		0.16		5.83		7.30
CUT2					-10.97		-13.31
					3.72		5.46
CUT3					-1.74		-3.54
					0.90		1.88
	R2=0.80		R2=0.76		PSEUDO R2=0.65		PSEUDO R2=0.62
Hausman test. Prob>Chi2= 0,000; Parties dummies=Yes; Year dummies=Yes
Std. Err. adjusted for 32 clusters INDIV;    * p < 0,1; ** p < 0,05; *** p < 0,01

Source: own elaboration

First, the QUAL (I) model shows that the variables PRIPUB, LEV and ANTIQ are significant. The relationship between the private and public income (PRIPUB) presents a negative sign, so parties that are financed with a greater volume of private income are more likely to have a lower quality of accounting information (H1). The negative sign of the age variable (ANTIQ) reveals that parties with longer trajectories in time are more likely to produce accounting information of poorer quality (H2).

Moreover, the financial dependency variable (LEV) also shows a negative sign. Therefore, parties with high financial dependency are more likely to disclose lower quality accounting information than the parties that do not.

To all of the above, it should be added that in QUAL (I) the STAND variable is not significant, which implies that even though the APAFP was approved in 2013, and 58% of the accounts have been prepared under its principles, this has not implied any improvement in quality, so we must reject the H4 hypothesis.

Finally, this model also includes the variable PSAV savings rate, which is significantly negative; that is, the greater the savings surplus, the lower quality of the accounting information. This effect is largely due to the excessive rates of equity savings, 61% on average, when the data have been modified following the qualifications of AC's auditors. This is far from the logic of a non-profit entity and very far of the previous 25% in the original data showed in Table 8, according to H5. Lastly, regarding the control variables, TPP is negative in level 3, corresponding to federations of parties, so is OLFPP, denoting that the last financial law could not improve “quality” in respect to the previous law.

5.2.2. Results of the original data model

Table 12 also shows the results of the model with variables elaborated from original accounting information. In this model, only the variables PRIPUB and ANTIQ result significant, as well as the control variables TTP and OLFPP. However, the explanatory variables LEV and PSAV were not.

The difference between the results obtained from the estimation of the models constructed with the modified and unmodified accounting data, derives from the multitude of exceptions filtered in the reconstruction of the modified variables. These differences highlight the need to correct and improve the accounting data disclosed by political parties.

Finally, albeit the sign of the regressors is the same in both models, obviously the proposed hypotheses should be considered about the first model with modified data since they collect all the caveats and therefore present a much more realistic picture of the economic and financial parties’ image.

5.2.3. Additional analysis

We find academic literature explaining that the favorable judgment of the audit report implies that when the political parties improve the quality on the financial information, so does the access to credit (Blackwell et al., 1998; Vico & Pucheta, 2005; Duréndez & Sánchez, 2008).

In this regard, the OLFPP reforms of 2012 and later in the one of 2015, placed restrictions and controls on the bank financing, including measures to improve transparency and limit the influence and pressure of the banking sector. Thus, the granting of credit by this type of organization should attend the habitual criteria of credit risk and the uses and customs of the commercial bank practices.

Therefore, it could be that political formations, in order to access financing, seek to improve quality accounting information reducing information asymmetry between the lender and the borrower. Specifically, this would be the case when the party has moderate levels of debt but not when it has an excessive level of indebtedness.

To have a better understanding of the structure and subsequent impact of the indebtedness, in the Table 13, using the linear panel data fixed effects model, a variable of three homogeneous levels, based on LEV has been created, called LEV3. The results show a negative impact on the quality of the parties, specially corresponding to the level 3, which maintains the highest debt rates, the debt is 2 times over the total financial structure (see the explanation in Table 10).

Level 2 represents a lower and less significant adjustment to the origin (Table 10 shows a composition of 0.46 times debt over the financial structure). In turn, the origin, where is Level 1, is positive respect to the rest of levels (this level has a very reduced level of debts of only 0.09 over the total of financial structure in Table 10).

These results are also in accordance with the hypothesis H3. However, before incorporating the recommendations of AC’s auditors, in the original data, LEV3 had not resulted significant, highlighting that a part of observations has been due to an incorrect recognition of debts.

Table 13. Results of additional analysis

	MODIFIED DATA		ORIGINAL DATA		MODIFIED DATA		ORIGINAL DATA
Variable	QUAL (II)		QUALO (II)		QUAL (III)		QUALO (III)
SIZE	1.93E-09		2.02E-08		1.02E-09		1.74E-08
	7.83E-09		1.94E-08		7.74E-09		1.88E-08
PRIPUB	-0.05	***	-0.05	***	-0.05	***	0.01	*
	0.01		0.01		0.01		-4.87
ANTIQ	-0.02	***	-0.02	***	-0.02	***	-0.02	***
	0.00		0.00		0.00		0.00
LEV3 2	-0.40	*	-0.47		-0.40	*	-0.48
	0.22		0.32		0.23		0.32
3	-0.80	***	-0.45		-0.84	***	-0.50
	0.28		0.57		0.28		0.56
PSAV	-0.08	***	-0.05		-0.07	***	-0.05
	0.02		0.04		0.02		0.03
PSAV*PRIPUB	-----		-----		-0.01	***	-0.01	***
	-----		-----		0.00		0.00
TPP 2	0.30		0.03		0.33		0.06
	0.28		0.59		0.28		0.58
3	-1.02	***	-1.60	***	-1.00	***	-1.56	***
	0.34		0.55		0.34		0.54
STAND	0.04		0.00		0.07		0.04
	0.36		0.36		0.37		0.36
OLFPP	-0.83	***	-0.89	***	-0.81	***	-0.85	***
	0.32		0.32		0.32		0.32
LOCAPP	0.51	**	0.54	**	0.55	**	0.58	*
	0.26		0.32		0.27		0.32
INTERCEPT	2.74	***	2.77	***	2.72	***	2.75	***
	0.17		0.17		0.17		0.17
	R2=0.81		R2=0.77		R2=0.81		R2=0.78
Hausman test. Prob>Chi2= 0,000 in each model; Parties dummies=Yes; Year dummies=Yes
Std. Err. adjusted for 32 clusters in INDIV;    * p < 0,1; ** p < 0,05; *** p < 0,01

Source: own elaboration.

Otherwise, in the case of the Spanish political parties, throughout the democratic period, the progressive increase in the electoral and operating expenses resulted in a situation of insufficient income. This lending to continuous deficits of many organizations (Martínez-Cousinou, 2009), aggravated after the cuts in the public aids in 2012 (Castro, 2014). Thus, parties tried to compensate that with private sources.

Moreover, considering H1, about the doubts in the transparency of the private contributions received by the parties, there could be a relationship between high positive savings in the income statement and the private income of the parties. In this way, the QUAL (III) model has introduced the iteration of PSAV (savings) with the PRIPUB variable and shows that the effect on quality is worse for those parties where excess savings come in greater proportion of private donations.

6. Discussion

First, the descriptive results are aligned with the contributions of Rodríguez López (2011), which reveal a multitude of errors in the accounting of the parties, so that the financial information provided remains deficient with respect to the previous periods. Likewise, despite the need detected by Rodríguez López (2011) to approve a sector-base Plan for the parties, we can state that the formations that use the APAPF do not reveal better quality accounting information, since the dichotomous variable created to assess the impact of the APAFP is not significant. Consequently, the implementation of an APAFP has not had the effects desired by the authors. In our opinion, the content of this standard is adequate and does not present any complications of interpretation. Consequently, it does not seem reasonable that the APAPF 2013 is the cause by which the parties have provided accounting information with deficient levels of quality. Thus, it is understood that such deficiencies may be due to a lack of qualified personnel, inadequate organizational structures or to conscious errors aimed at "masking" their financial reality.

Second, the research of Ruíz Lozano et al. (2008) show a positive relationship between the volume of public subsidies received by public foundations and the favorable opinion of the audited reports. This result is in line with the work of Rodríguez Bolívar et al. (2013), which justify that the behaviour in terms of transparency of the public administrations is conditioned by the volume of funds they receive from other public organizations. Meanwhile, in our research we found a negative relationship between a higher proportion of private income than public income and accounting quality. The results of these works seem to be aligned, but the explanation for these differs. Thus, in the case of political parties, the receipt of public funds is not considered a factor that positively affects the quality of the accounting information, as in the case of public foundations or public administrations. Furthermore, it is the higher proportion of private income which causes non-transparent behavior.

Thirdly, with regard to the age variable, it is noted that those parties that in the past were making accounting errors, as stated by Rodríguez López (2011) and Santano (2019) and therefore have a longer political career, are more likely to continue to formulate poor quality accounting information. In contrast, younger parties produce higher quality accounting information.

Fourth, there is a negative relationship between the level of leverage and the quality of the accounting information since parties with high financial leverage perform worse in the AC audits. This work is agreed with the contributions of DeFond et al. (2002) and Mareque et al. (2019), in which both parties and companies obtain worse ratings in the audit reports when they are observed to be highly indebted. On the other hand, the results show that low or moderate indebtedness improves the effect on the accounting quality of the political parties respect to the higher levels. Thus, Blackwell et al. (1998), Vico & Pucheta (2005), Duréndez & Sánchez Vidal (2008), highlight that the opinion of the audit report facilitates the process of granting credit to companies. Consequently, political formations with moderate financing rates try to reduce the "agency conflict" between borrower and lender, providing higher quality and transparent information.

Finally, concerning the "savings rate" variable, Ruíz Lozano et al. (2008) state that in the NPOs, given the absence of economic incentives as criteria for evaluating the efficiency of their management, there is no relationship between the accounting result for the year and the type of financial information provided. That is, they do not conceive that a positive or negative result encourages the greater or lesser commitment of the organization to the accounting quality and transparency. In this way, González-Díaz et al. (2013) neither find a significant statistical relationship between the adverse audit opinion on state public foundations and the existence of a negative or positive economic result. However, this research shows that in the case of political parties, there is a relation between the accounting result and the quality, because parties with an extraordinary capacity for "savings", once the data has been modified from the AC’s audit recommendations, reveal accounting information of lesser quality.

7. Conclusions

Even though the transparency indexes (see Table 1) indicate an improvement in the amount of financial information provided by the political parties, the quality of this information is deficient, with a minority of cases of political parties presenting quality financial information. Likewise, the evolution of the quality of the financial statements has been decreasing, since throughout the years investigated, there is a notable decrease in the quality of the financial statements. The last year evaluated, 2015, is the worst in the series studied, with no party having disclosed the quality financial information. As a result, we conclude that the institutional barrier imposed by the public authorities in terms of financial transparency is not effective. Moreover, although leading authors have suggested that the approval of the APAPF in 2013 should help overcome the problems identified in terms of accounting quality, this research concludes that this has not been the case.

Furthermore, it is necessary to review, filter and correct the information of an accounting nature issued by the parties, in order to analyze more reliable and closer to reality financial information. Therefore, in the econometrics model, it has been very useful to correct the deficiencies manifested as qualifications in the AC's Audit Reports, in order to introduce information closer to the reality of the party's assets. In this regard, in the model with modified data all the hypotheses are accepted, except the one related to the APAPF. However, the models estimated with original accounting data lose explanatory power. Only the hypotheses related to the weight of private funding received and the age of the political party are accepted.

Otherwise, the relevance of this research lies in the fact that it analyses the factors that affect the quality of the accounting information and, therefore, the financial transparency of the political parties. In this regard, this paper contributes to broadening the debate about the efficiency of the accountability process of the political parties. In addition, the findings provide valuable insights to the regulator to revise the current regulation. In this way, enforcement and sanction measures could be applied to improve the compliance of the political formations and it is recommended that the competent authorities take the necessary measures to resolve this problem. For example, article 17 of the Organic Law (8/2007) typifies the presentation of the Annual Accounts in a deficient or incomplete manner as a serious infraction subject to sanction. However, the regulations do not specify what constitutes deficient or incomplete accounting information. Therefore, in order to improve the accounting quality and financial transparency of political parties, we propose that annual accounts, whose opinion is different from unqualified, should be considered as deficient and subject to sanctions.

Likewise, in order to consolidate the commitment of the parties and improve the financial transparency, this research proposes that public subsidies received are partially associated with the quality of the accounting information and the degree of transparency revealed. In this regard, considering the delay with which the AC publishes the Audit Reports, we propose that the political organizations that do not obtain a clean qualification should reimburse part, or all of the amounts received as subsidies, depending on the qualification obtained, as well as on the qualifications and errors exposed in it.

One limitation of this research is based in the fact that the quality of the accounting information has been measured according to the opinions issued by the AC. In this regard, some cases in the corporate sector (such as Enron and Parmalat) have drawn the attention on the opinion of the audit reports. However, the audit body of the political parties -AC- is constituted by an external control body of a technical and independent nature, which does not have commercial or contractual ties with the political organizations audited. Therefore, it is understood that the limitation stated above is reduced. Nevertheless, despite being an independent body, the bias of the auditors themselves may constitute a limitation in the drafting of the reports. Another limitation is that during the years under investigation, political parties are not legally obliged to include their local activity and although some parties have already included their local structure activity during 2012-15, most parties did not.

Finally, considering that the APAPF is not mandatory until 2019, it would be interesting to carry out a study with data collected from this year when possible, in order to analyze the changes in the accounting quality of the political formations. Likewise, this research could be applied to analyze the financial transparency in European countries with a similar legal framework.