Trust in Spanish Governments: Antecedents and ... - CiteSeerX

Economic Analysis & Policy, Vol. 43 No. 2, september 2013

Trust in Spanish Governments: Antecedents and Consequences Gloria Alarcón García Dpto. Hacienda y Economía del Sector Público, Facultad de Economía y Empresa, Universidad de Murcia, España (Email: [email protected]) and Edgardo Ayala Gaytán Tecnológico de Monterrey, Eugenio Garza Sada 2501, 64849 Monterrey, N. L., México (Email: [email protected]) Abstract:

We used the Fiscal Observatory Survey of the University of Murcia (2007) to examine the antecedents of trust in the Spanish government (at city, regional and state levels). We found that the main drivers of trust are individuals’ perception of the government’s performance and the socio-economic environment in which they live. Regarding the consequences, we tested a variant of the rule compliance hypothesis: the higher the trust in the government, the higher the citizens’ fiscal awareness, meaning individuals are more conscious of their fiscal obligations and more intransigent on fiscal fraud. Our results suggest that trust levels shape tax awareness, producing virtuous or vicious loops.

I. Introduction There is a large body of literature showing that trust and social capital, understood as trust, social norms and codes, are critical ingredients in the economic welfare recipe (Beugelsdijk and Van Schaik 2001, Fukuyama 1995, La Porta et al. 1997, Dasgupta 2000, Gambetta 2000, Durlauf and Fafchamps 2004, Keefer and Knack 1997). More specifically, recent research suggests that citizens’ trust in their governments improves government’s performance. This is presumably due to the fact that trust is associated with a higher participation of citizens and thereby with a higher accountability of government officers, with a decrease in polarization, 177

Trust in Spanish Governments: Antecedents and Consequences

more likelihood of achieving agreements, and more social-oriented preferences of individuals (Knack and Zack 2003, King et al. 2002, Boix and Posner 1998). In this paper, we examine the antecedents or determinants of trust in all three Spanish government systems (city, autonomous community and state) by conducting the Fiscal Observatory Survey of the University of Murcia (2007), FOSUM. Regarding the antecedents, we test the hypothesis that assumes that the main drivers of trust are both citizens’ perception of government’s performance, and the objective performance of the economic and social context in which they live. With regard to the consequences, we test the Boix and Posner (1998) rule compliance hypothesis which proposes that the higher the trust, the likelier are the agents to obey the rules. In our case, we test this hypothesis for the case of fiscal rules or what we define as fiscal awareness (Hasseldine and Kaplan 1992), which encompasses fiscal ethics and fiscal morale. In this study, we deal with the compliance of fiscal rules by using the construct of fiscal awareness instead of fiscal morale. We agree with Torgler (2005) and Cummings et al. (2009) on the fact that it consists in the “intrinsic motivation to pay taxes” and that it arises from the moral obligation to pay taxes as a contribution to society” (Cummings et al. 2009). But this is why we distinguish between individual awareness “fiscal awareness” and the collective morale “tax morale”. It also includes the reflective knowledge of individuals and their assessment of public revenues and expenditure considering their correct compliance with the tax system (Alarcón, Martínez and Quiñones 2007). Following this line, our work extends previous studies concerning the Spanish case, mainly by Torgler and Schneider (2007), Alm and Gómez (2008) and Martinez-Vazquez and Torgler (2009). In this paper their hypotheses are tested by means of a new survey exclusively designed to tackle fiscal awareness among Spaniards using the same data as Mª Dolores et al. (2010). Besides, a psychosocial concept of tax compliance as part of fiscal awareness is applied (James, Murphy and Reinhart 2005), and the idiosyncratic and context drivers of trust in governments are carefully tested in the Spanish context. In accordance with the view initiated by Torgler and Murphy (2004) and Torgler and Schneider (2005), our study aims to examine the elements that form the fiscal awareness. In this case, we focus on proving that trust in government is one of these constitutive elements and may have a multiplying effect on tax awareness.

II. Trust in governments: A brief survey Trust is especially important in government’s performance. La Porta (1997) proved that a high level of trust is associated with high indexes of government efficiency and public servants’ honesty. In a more detailed work, Knack (2000) found a significantly positive relation between citizens’ trust in governments and government’s performance using objective measures of performance based on administrative audits. Why should we expect trust to improve public performance? Putnam (1993) and Knack (2000) proposed three mechanisms that explained this relation. They argued that trust encourages civic participation and this makes public officers more accountable, which Boix and Posner (1998) call the rational voters and the competitive elites effect. This is because, in countries 178

Gloria Alarcón García and Edgardo Ayala Gaytán

with high levels of trust, citizens participate more often and spend more time monitoring their authorities. In addition, trust increases the likelihood to reach agreements by reducing polarization as well as encouraging government innovation. Boix and Posner (1998) add two other micro-foundations to the link between trust and public performance: rule compliance and civic virtue. Trust might induce rule compliance, because when individuals rely on their fellow citizens, they comply with their public obligations (that is, they pay their taxes and obey the rules). In other words, they are more encouraged to comply with their own obligations. Additionally, as they expect governments to use public funds in a suitable way, they would be pleased to contribute to finance the public funds that governments produce and provide. The civic virtue argument proposes that high levels of trust might contribute to a shift in the citizens’ preferences from a personal interest (what is good for me?) to a more community-oriented one (what can we do for our neighborhood?). This shift would increase cooperation between citizens and governments. Now in order to propose a model of the determinants of trust in governments, we make a brief review of the research about the determinants of interpersonal trust and trust in the institutions that are clearly antecedents of trust in governments. Regarding interpersonal trust, the large body of evidence on trust determinants suggests that most deciding factors are divided into three sets of variables: the individuals’ socio-economic profiles, the context in which they live, and their perceptions on the performance of the main institutions of society. Among the first set of variables, education seems to be the most important driver of trust. In this regard, the studies by Helliwell and Putnam (1999) and Knack and Zak (2003) conclude that the higher the individuals’ level of education, the higher is the trust. Thus, education and human capital promote trust, which directly enhances individuals’ trust in their fellow citizens, and indirectly improves the quality of institutions in charge of enforcing the fulfillment of contracts. Among the context-related determinants, Putnam (1993) found a positive relation between trust and civic norms and the density of horizontal networks in society. Nonetheless, Olson (1982) argued the opposite, suggesting that the denser the networks, the more vulnerable is the society to special interest groups that might lobby for privileges imposing disproportionate burdens to the rest of society (Knack, 2003). Knack and Zak (2003) found that the factors related to a low social polarization and the formal institutions that bound the domain of action of governments have a positive effect on trust. On the other hand, using data from the GSS, Alesina and La Ferrara (2002) found lower levels of trust in individuals characterized by any of the following factors: traumatic experiences, belonging to a discriminated group (e.g. Afro-Americans), low income, low levels of education, ethnic heterogeneity, and unequal income distribution. The same evidence that was presented by Rahn and Rudolph (2005) applied to certain cities of the United States. In addition, Martinez, Ayala and Aguayo (2012) found evidence applied in the Mexican case that proved that trust diminishes with the presence of segregation, the absence of participation in networks other than family, and the size of communities. Finally, the perception citizens have on the different institutions’ performance affect trust. Chanley, Rudolph and Rahn (2000) showed that individuals’ satisfaction with public services and their perception on safety shape trust in governments in the United States. In this regard, Tyler (1994, 1995) found that citizens’ perception on the existing fairness in the most important procedures in society, like the fiscal system, increases interpersonal trust. 179


III. A model of trust and fiscal awareness in Spain Following the literature we presented, we have built an empirical model of trust for Spanish governments based on the premise that trust is a function of the socio-demographic profiles of individuals; the individuals’ perception on the quality of the public services that the different governments provide; their perception on fairness in the procedural systems, mainly in the fiscal one; and the context variables related to the environment in which they live, such as polarization and economic growth, among others. The proposed model is based on the premise of the rational citizens, that is, trust is partially the result of citizens’ perceptions on the government’s performance. This means that they are more trustful when they identify the outcomes of good government actions. These outcomes can be measured at two levels: at a micro level, where they are synthesized according to the subjects’ reported perception of the public services quality (Chanley et al., 2000); and at a macro level, where they are characterized by some relevant context-related variables, such as economic growth and income inequality, among others (Alessina and Ferrara, 1999 and 2000). The equation (1) shows a generic model for Trust: Tig = α g + Di' β g + Si'θ g + C 'jδg + ηgj + uig

(1)

where “I” denotes individuals, “g” level of government and “j” autonomous communities. Di is a kx1 vector containing all socio-demographic characteristics of the individual I, Si is a Hx1 vector containing individual “i” perceptions of government “g” performance, Cj is a Lx1 vector with the main context variables of the community j in which individual “i” lives in. The parameters for each level of government are represented by vectors α, β, δ, and θ, and finally ηgj represents a fix effect of community j in government g. On the one hand, the set of equations (1) is designed to explore the antecedents or determinants of trust in the Spanish governments. On the other hand, we want to model some hypotheses on the consequences of trust. Although there are many factors suggested by theorists and previous studies, such as participation rate in elections and politics in general (Beugelsdijk and Van Schaik 2001), government’s performance (Knack and Zack, 2003; King et al., 2002), the size of companies (Fukuyama 1995, La Porta et al. 1997), and the general economic performance (Dasgupta 2000, Gambetta 2000, Durlauf and Fafchamps 2004, Keefer and Knack 1997), we focus on just one possible consequence of trust: citizens’ fiscal awareness. Previous studies have focused on tax morale, usually associated with the degree of acceptance or tolerance to tax fraud. For example, Alm and Torgler (2006) analyse the influence of fiscal conditions on the tax morale in Europe and the United States, and highlight some significant differences. Torgler (2003) studies the tax morale in post-communist countries and Germany after the reunification, while Alm et al. (2006) analyse the tax morale in Russia before and after the political transition, detecting an increase in the degree of trust in government, and finding that national pride improves fiscal morale. Martínez Vázquez and Torgler (2005, 2009) analyse the correlation between the fiscal reform implemented in Spain in 1977 and the tax morale. Later, Alm and Gómez (2008) and Torgler and Schneider (2009) found a strong negative correlation between unofficial economic activity and tax morale: the lower the tax morale, the higher the unofficial activity. 180


By the ‘‘slippery slope’’ framework, Kirchler et al. (2008) suggested a framework for tax compliance as a conceptual tool, in which both the power of tax authorities and trust in the tax authorities are relevant dimensions for understanding enforced and voluntary compliance. In the same sense, Barone and Mocetti (2011) studied the link between government efficiency and tax morale. To sum up, there are many factors that may have an impact on tax morale: cultural factors, institutional factors, trust in the system, sociological factors, and even globalization. (Alm et al. 2003, Torgler 2003, Torgler and Schneider 2005, Murphy 2005, Wahl et al. 2010, Hammar et al. 2008, Richardson 2008 and Hofmann 2008). In the present paper, we prefer to use the concept of fiscal or tax awareness. Fiscal awareness, therefore, includes both unconscious components derived from some principles and values that belong to the moral and ideological domain, individual’s psychosocial factors and technical components related to the objects and subjects involved in the fiscal process. In this paper, we understand fiscal awareness as the individual perception of each citizen regarding fiscal ethics and fiscal morale of the society in which they live (Alarcón and Tipke 2007). From a wider perspective, fiscal awareness is defined as the individual behaviour when tackling tax obligations. It is made up with (a) ethical factors, every individual’s values and principles regarding taxes1; (b) psychosocial factors, satisfaction with the efficiency of Administrations when applying tax revenues, and trust in fiscal administrations, among others; and (c) technical factors of knowledge on the tax system and demographic determinants. Someone with a high tax awareness will pay their taxes voluntarily regardless of what their neighbours do, what the collective moral of the group considers that should be done, and regardless of the individual’s possibilities to commit tax fraud. Awareness is a dynamic phenomenon that is in a constant feedback process where psychosocial factors play a vital role (Kornhauser 2007). The validation contexts of human behaviors, in this case related to the voluntary tax compliance, are the personal and social contexts in keeping with individual awareness and social morale. This means that tax compliance has to do with three different levels: the individual, the society, and the State, which regulates these relations and is constituted from them. It is important to highlight that it is not a question of dichotomous perspectives or positions, but rather all of them integrated in the individual by mechanisms of adjustment, adaptation and assimilation. Acceptation of fiscal fraud is the negative definition of tax awareness as well as tax morale. That is, tax fraud tolerance is the clearest expression of low or inadequate fiscal awareness that fosters the development of low fiscal morale in the community where individuals with low tax awareness live. This is why the most immediate way to measure fiscal awareness or morale is getting to know the individuals’ attitude towards fiscal fraud. This is the parameter we used in the study, the most-used one in the literature on fiscal morale, even though the concept cannot be identified (Torgler and Schneider 2009), and its validity discussed due to the multidimensional nature of fiscal awareness and morale (Alm and Torgler 2006, Torgler 2012). There are also recent works on tax evasion by McGee and Gupta (2008). 1

Being a taxpayer or a member of the legislature, the judiaciary or the executive.

181


The fiscal awareness criteria was applied by Alarcón, De Pablos and García (2009), who used the information produced by the initiative of The Fiscal Observatory Survey of the University of Murcia (FOSUM 2006) to model fiscal awareness according to individuals’ age, education, labor status, income and reports on their satisfaction with the Spanish fiscal system. In a follow-up study in this line of results, Prieto et al. (2006) found that the political view is the main driver of the degree of tax evasion and fiscal awareness by means of an exercise based on an ordered probit model. Alarcón, Beyaert and de Pablos (2012) incorporate gender interactions and knowledge of the fiscal systems to explain the variations in fiscal awareness. In line with the above-mentioned attempts, we suggest that trust in Spanish governments influences fiscal awareness, mainly because of the increasing intolerance to fiscal fraud, following the literature that takes it as the only variable of measurement. The hypothesis is that the higher the level of trust in governments, the higher the degree of compliance with fiscal obligations measured negatively by aversion to fraud. The rationale of this proposal is disclosed by two explanations. On the one hand, from a rational perspective, reliable governments might promote fiscal compliance because the funding of public services has a high return to tax payers, and vice versa. On the other hand, from a psychological perspective, the individual cognitive dissonance produced by evading taxes might be more powerful if the government is reliable than when it is not. In both cases, the hypothesis establishes a direct linkage between trust and fiscal awareness. Of course, tax evasion also depends on variables, such as the labor status of the individuals or the efficiency of revenue agencies, which can promote this illegal activity, but we model just the individuals’ commitment to their fiscal obligations, not the degree of tax law observance. That is, an employee may have no chance to commit a fiscal fraud because he is a captive taxpayer according to the prevalent tax system. Nevertheless, a distrusted government might make him not willing to comply with his fiscal obligations. Equation (2) describes this relation: _

Fi = α F + ϕ T i + γ Z i + u Fi

(2)

where “T” represents the average trust at the different levels of government of individual “i”, and “Z” contains socio-demographic variables of the individuals. Equations (1) and (2) can be combined to introduce vicious or virtue circles. When people trust their governments, the levels of rule compliance may increase (by equation 2), which leads to a decrease in transaction costs for individuals and the government, and to an increase in public funds, enhancing the quality of public services and increasing the level of trust in governments (by equation 1). On the other hand, distrusted governments will find it more difficult to collect taxes and achieve citizens’ cooperation, and that can surely affect the quality of public services. Thus, a variation in the fundamentals of trust, such as an increase in crime rates or a scandal involving corrupt public officers, can cause an enduring low level of trust, through the trust-fiscal awareness multiplier mechanism described above, making the total effect higher than the initial impact on trust.

182


IV. The data In this paper, we use the FOSUM conducted on November 2007 especially designed to gather information on the values that shape Spaniards’ fiscal awareness. The studied population are all the residents in Spain over 24 years of age. A multilevel probabilistic approach was applied, with a first and second stage stratification of drawing on an independent sample from each regional autonomous community. The survey main sections are: (1) the perception of fiscal fairness in the Spanish tax system, (2) Spanish taxpayers’ fiscal awareness, and (3) sociodemographic information. Table 1 and Table 2 present the description of the socio-demographic profile of the sample that was almost evenly divided by gender with 49 being the average age. 60% of the respondents declared a monthly income between €1,240 and €3,340; 40% of them are employees in the private sector or public officers, and 60% have secondary or higher education. Table 1: Descriptives of Socioeconomnic Characteristics Demographic Profile Male Age Bachelor Married Divorce Widow Children

Mean

Std. Deviation

0.487 49.441 0.214 0.670 0.032 0.078 0.741

0.500 15.900 0.410 0.470 0.176 0.268 0.438

0 to 1,240 € 1,240 to 2,300 € 2,300 to 3,340 € 3,340 € or more

0.329 0.398 0.208 0.065

0.470 0.490 0.406 0.246

Employee Employer Self-employed Bureaucrat Student Unemployed

0.284 0.027 0.069 0.105 0.025 0.070

0.451 0.161 0.254 0.307 0.155 0.255

Primary education Secondary education Tertiary education

0.293 0.288 0.309

0.455 0.453 0.462

Income (monthly)

Labor Status

Education achievement

Source: Authors‘ estimations using the FOSUM, 2007.

183


Table 2: Trust and Fiscal Awareness variables Mean

Std. Deviation

2.219

0.992

2.021

0.885

1.922

0.925

Fiscal system is fair

2.036

0.910

Fiscal fraud is justified because of the excess fiscal burden

0.073

0.260

0.239

0.427

0.107

0.309

0.536

0.499

Trust

City Government

Autonomic Community Government

State Government

Perception about public performance Public services reduce poverty

Public services enhance welfare

Fiscal Awareness

Fiscal fraud is justified because it does not affect anyone

Fiscal fraud is justified because the government is inefficient

Fiscal fraud is never justified

Knowlede of fiscal system

2.201

2.144

2.567

0.918

0.961

1.057

Source: Authors‘ estimations using the FOSUM, 2007.

Trust was measured by the item: “Please, rank the degree of trust you have in your city (autonomous community or state) government in a scale from one to four, where one means no trust and four is full trust”.

Mean values of trust are around 2.2, being marginally larger for city and autonomous community scenarios than for the state government. The proportion of the respondents who scored the lowest levels of trust (1 or 2) rose from 62% for city and autonomous governments up to 72% for the state government. The way individuals perceive government’s performance was measured through three items related with the benefits of public services and the fairness of the fiscal system. Concretely, the questions were the following: “Please, rank the degree in which public services contribute to reduce poverty in a scale from one to four, where one means ‘no contribution at all’, and four means ‘great contribution’”; “Please, rank the degree in which public services enhance welfare in a scale from one to four, where one means ‘no contribution at all’ and four means ‘great contribution’”; and “Please, rank the degree of fairness of our actual fiscal system in a scale from one to four, where one means ‘totally unfair’ and four means ‘very fair’”.

The first two items reflect the satisfaction felt by individuals with two current critical issues in Spain: social policies and the quality of the public goods provided by the government. The other item concerns fairness in the fiscal rules for which governments and parliaments are 184


responsible. In view of the sample results, Spaniards are really sceptical about the contribution of governments to reduce poverty, to enhance welfare, and about the fairness of the fiscal system (75%, 66% and 73% of the respondents scored low levels on the above-mentioned items, respectively). The corresponding mean values are 1.9, 2.1 and 2, respectively, for these variables. On the other hand, we measure fiscal awareness as Alarcón, De Pablos and García (2009) did, that is, by using a dummy variable that takes value 1 if the respondent chooses the option “Fiscal fraud is never justified” to the question “In your opinion, fiscal fraud is justified because...”. The rest of the alternatives were the following: “Because of the excessive fiscal burden”; “Because it does not affect anyone”, and “Because the government is inefficient”. We believe that asking the question in that manner helps to eliminate the social desirability bias, i.e. to prevent individuals from answering what is politically right instead of what they really think. Clearly, if the question was directly asked as, for instance, “Do you commit fiscal fraud?” or “Do you think that committing fiscal fraud is wrong?”, most respondents would answer “no” to the first question, and “yes” to the second one. Results show that 47% of the respondents chose one of the three justifications for committing fiscal fraud. Most of them justify fiscal fraud because the evasion does not affect anyone. It seems that providing individuals with more options helps reveal whether they have strong principles about this issue or they are looking for a good excuse to commit it. In contrast, Alm and Gomez (2008) used the item: “Do you mostly agree or mostly disagree with the following statements? [...] Actually, it is not that bad to hide part of your income since nobody is really affected by it”.

However, 81% of their sample disagreed on the statement, either because the question framing induces social desirability bias, or because some individuals think fiscal evasion is not necessarily bad, although the reason is not what the question proposes (“since nobody is really affected by it”). Therefore, using the FOSUM item on tolerance to fiscal fraud allowed us to count on a dependent variable with a higher variance. The variance of the proportion of respondents who are intolerant to tax fraud is almost three times higher than the variance of the proportion in the question found in the study by Alm and Gomez (2008). Finally, the context-related variables included in the model for the autonomous communities were economic growth, homicide rate, income inequality, and education level. Economic growth was measured as the GDP growth rate per capita. The homicide rate corresponded to the number of homicides per one thousand residents. The income inequality was calculated through the Gini coefficient, which took the income distribution and education as the average number of years of education. The source of all variables was the Spanish Statistical Office (Instituto Nacional de Estadística de España).

V. Estimation and Results We started obtaining estimations from both pooled ordinary least squares (OLS) and the random-effect method that included the context variables considered in this paper (homicide rate, economic growth, income inequality and average educational achievement). In addition, 185


dummies for the autonomous communities were used to identify any remaining unobserved fixed factors associated with the political and geographical division of Spain. We chose random effect estimators because the context variables are repeated measures in the pooling equation, that is, the context variables (e.g. economic growth) refer to the autonomous communities so that all individuals of a particular community share the same context variables. Therefore, applying only fixed effects would omit the context variables in the estimation, and it would not be possible to test the effect of the context on trust. Another relevant issue to deal with is the metric of the dependent variable. Adopting OLS or Panel data implicitly assumes that trust is a continuous variable, although it is not actually so, because it is a discrete variable that takes values in a scale from 1 to 4, according to FOSUM. However, we decided to deal with the estimations under such a restrictive assumption only after comparing them with the ordinal probit estimations of equation (1) (without any random effects). This equation recognizes the discrete nature of the variable and verifies that the main conclusions are the same for both approaches. The ordinal probit results are available under request to the authors. The results for the trust equations are presented in Table 3. Few socio-demographic variables proved to be significantly different from zero. The exceptions are the educational achievement dummies and, in some cases, the dummy for individuals that are either an employer and/or self-employed, as well as the male condition. It is interesting to note the presence of a reversal effect of education on trust depending on the level of government. Trust in city authorities decreases as the level of educational achievement increases. Nevertheless, trust in the state government increases as the educational achievement level rises. Education has no effect on trust felt in autonomous community governments2. Only in the trust equation in autonomous community governments, both the employer/self-employed condition and the male one are significantly different from zero, being even negative. The important driver of trust in the different levels of Spanish governments is the quality of government’s performance, which is represented by the scores of three questions related to the agreement on the following sentences: “Public services reduce poverty”; “Public services enhance welfare”; and “The fiscal system is fair”.

The coefficients of these items are positive, as expected, and significant (p