- Email: [email protected]
Tax morale and tax evasion reports
Economics Letters 121 (2013) 110–114
Contents lists available at ScienceDirect
Economics Letters journal homepage: www.elsevier.com/locate/ecolet
Tax morale and tax evasion reports Francesco Flaviano Russo ∗ Department of Economics, University of Naples Federico II, Via Cinthia, Monte Sant’Angelo, 80126 Napoli, Italy CSEF, Via Cinthia, Monte Sant’Angelo, 80126 Napoli, Italy
highlights • • • •
New measure of tax morale based on voluntary reports of tax evasion. Positive relationship between tax morale and social capital. Positive relationship between tax morale and immigration. Negative relationship between tax morale and dissatisfaction with public services.
article
info
Article history: Received 30 November 2012 Received in revised form 4 July 2013 Accepted 5 July 2013 Available online 11 July 2013
abstract The study proposes a new measure of tax morale for Italy based on voluntary tax evasion reports. Using this measure, I detected a positive relationship between tax morale and social capital, political participation, and immigration, but a negative relationship between tax morale and the dissatisfaction with public services and unemployment. © 2013 Elsevier B.V. All rights reserved.
JEL classification: H26 K34 Keywords: Tax evasion Tax morale Social capital
1. Introduction The concept of tax morale was introduced in the economic literature to provide a rational explanation of why people pay taxes (Slemrod, 2007). This was because traditional models of tax evasion Allingham and Sandmo (1972), which are based only on auditing probabilities and expected penalties, underestimate the empirically observed rate of tax compliance significantly (Andreoni et al., 1998; Feld and Frey, 2002; Sandmo, 2005). Thus, in an attempt to reconcile the model with the data, several authors have hypothesized the existence of a moral (Erard and Feinstein, 1994) or intrinsic motivation (Frey, 1997) for paying taxes, which was designated as ‘‘tax morale’’. Previous empirical studies have constructed several measures of tax morale based on a variety of surveys: the International Social Survey Program (Torgler, 2005b), the World Values Surveys (Alm and Torgler, 2006; Torgler, 2006), the European Values Surveys (Lago Peñas and Lago Peñas, 2010; Hug and Spörri, 2011), the British Social Attitudes Survey (Orviska and Hudson, 2002), the
∗
Tel.: +39 081675276. E-mail address: [email protected].
0165-1765/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.econlet.2013.07.004
Latinobarometro (Torgler, 2005a), and the Afrobarometer (Cummings et al., 2009). Typically, tax morale has been measured as the percentage of survey respondents who agree with statements such as: ‘‘Citizens should not cheat on their taxes’’, ‘‘Cheating on taxes is justifiable if there is the opportunity to do so’’, or ‘‘Managing to avoid tax payments is legitimate’’. In this study, I propose a new measure of tax morale, which is based on quasi-experimental data: the voluntary reports of tax evasion collected in Italy via a website evasori.info. This site allowed users to report transactions that lacked a receipt, so were hidden from the tax authority and allowed the supplier to evade taxes. It was expected that buyers who are averse to tax evasion would report such episodes, possibly as a form of social sanction, which may or may not be related to other possible social sanctions, such as exclusion from communal activities, thereby acting as a negative advertisement and causing sellers to lose customers. In the present study, my measure of tax morale was the number of tax evasion reports per unit of irregular activity given the actual level of irregular activity (which by definition entails tax evasion), so a higher number of tax evasion reports is an indicator of a less lenient attitude toward tax evasion and, therefore, of a stronger tax morale.
F.F. Russo / Economics Letters 121 (2013) 110–114
Thus, I conducted an analysis of the determinants of tax morale in Italy. Most of the variables considered have already been discussed previously, i.e., social capital, political participation, working status, education, satisfaction with public services, religion, and demographics. In contrast to previous studies, however, I also considered the relationship between tax morale and immigration in a systematic manner. I found that tax morale was positively associated with social capital, political participation, and immigration, but negatively associated with dissatisfaction with essential public services and unemployment. The most interesting result was the positive association between tax morale and social capital, because involvement in the local community can also foster a locally-oriented subculture, which may oppose national Government contributions and redistribution. The study belongs to the broad category of studies on tax morale and its determinants. The most closely related work, which was conducted by Cannari and D’Alessio (2007), analyzed opinions on tax evasion and the propensity to evade by addressing specific questions in the Survey on Household Income and Wealth (SHIW), which was administered by the Bank of Italy. They find that selfemployed, younger and less educated individuals were more likely to evade taxes, and that this propensity was positively associated with the crime rate and the unemployment rate, but negatively related to social capital. They also found that there was a generally low aversion to tax evasion, which is part of the reason why evasion is so common in Italy. Among other studies, Lago Peñas and Lago Peñas (2010) discussed the relationship between tax morale and the socio-demographic, cultural, and political characteristics of European countries. Alm and Torgler (2006) compared tax morale in the USA and in Europe. Torgler (2005b) and Hug and Spörri (2011) studied the relationship between democratic participation and tax morale in the Swiss Cantons and among countries. The new perspective introduced in the present study is a new measure of tax morale. The remainder of this paper is organized as follows: Section 2 describes the data, Section 3 summarizes the empirical results, and Section 4 provides the conclusions. 2. Measuring tax morale I constructed a measure of tax morale, which was based on voluntary reports of tax evasion submitted to an Italian website evasori.info. The website allowed anyone to submit anonymous reports of their transactions with sellers who failed to issue receipts, thereby evading taxes. These details included the amount of the transaction, the geographic location, and the type of economic activity conducted by the seller (a choice was made from a long list). The concept behind my measure of tax morale was that with the same level of tax evasion, a higher number of tax evasion reports would suggest a higher frequency of social sanctions against this behavior, so it could be used as an indicator of a more negative attitude toward tax evasion. To control for the actual level of tax evasion, I used the rate of irregular activity reported in the Italian Statistical Institute (ISTAT) database. One problem with this measure was that it was only available at the region level (21 observations), whereas the tax evasion reports were available at the province level (102 observations). ISTAT also reports a rate of irregular activity, which is disaggregated into economic activity categories, so I constructed a measure of irregular activity at the provincial level by computing a weighted average of the irregular rates for different activities, where the weights equated to the share of activities in the province. I obtained my measure of tax morale by dividing the total number of reports in a given province by this irregular activity rate. The empirical distribution of my tax morale measure was positively skewed, which reflected the presence of a few large
111
outliers. The mean number of reports per unit of irregular activity was 4.4 and the median was 1.7. The highest value in the sample was 70 (Roma) while the lowest was 0.1 (Vibo Valentia). The standard deviation was 9.5. Excluding the top two observations in the sample (greater than 20 times the median value), I arrived at a mean of 3.2, a standard deviation of 3.8, and a range of 23. My tax morale measure also had high geographic variability, i.e., among the lowest 10 (20) observations, only one (five) belonged to a province in Northern Italy, whereas only one (two) of the top 10 (20) observations belonged to a province in Southern Italy. In my opinion, there are three major problems with my tax morale measure. The first is the interpretation of zeros, i.e., the absence of reports. They could indicate the absence of tax evasion, which may or may not be a consequence of an extremely high tax morale. Alternatively, they could be a sign of a particularly low level of tax morale where evasion is much easier due to a lack of social pressure. Two elements significantly reduced the importance of this ambiguity. First, there were no zeros in the reports dataset and there was a high correlation between population size and the number of reports at the level of the province (0.61). Second, zero tax evasion was never observed in the irregular activity dataset. The second major problem was that the anonymity of reports meant that it was not possible to control for the frequency of reports by the same users. Thus, my measure could have been based on a small number of reports made by a very large sample of users or a large number of reports made by a small sample of users. The problem is that tax morale will potentially have stronger effects on all outcomes if it is more widespread, because this means there is a higher probability of being matched with an individual with a high tax morale during an economic transaction. However, this is only true if we assume an equal number of transactions per individual. Instead, if the number of reports is proportional to the number of transactions, the magnitude and diffusion of tax morale are highly correlated. Thus, I assumed that the individuals who made fewer reports simply had fewer opportunities to make reports. The third problem was related to the interpretations of the reports. Based on the information captured via the website, it is not clear whether they were reports of actual or possible tax evasion. For example, we might consider a case where a seller tries to complete a transaction without issuing a receipt but the buyer notices this and asks for a receipt, thereby preventing tax evasion, yet the buyer still reports the transaction via the webpage. For the purposes of my analysis, there was not a big difference between reports of actual or potential evasion because they are indicators of a negative social attitude, where an individual sanctions the behavior of an (actual of potential) evader. The only difference is that, if the reports only described potential tax evasion, I must refrain from any interpretation of the total level of evasion because the amounts would have been disclosed to the tax authority eventually. Another, perhaps less important, problem is that my measure might be more volatile than tax morale itself. Indeed, a higher tax morale is associated with a lower level of tax evasion and a higher number of reports, which indicates a higher numerator and a small denominator in my measure. 3. Determinants of tax morale What are the possible explanations of tax morale? I performed regressions of my tax morale measure on most of its determinants that have been highlighted in previous studies. The effects were validated based on the variability at the level of the Italian provinces.1 An important additional source of validation was
1 I used the classification of provinces at the time the website evasori.info was created, which was before the two institutional reforms that first increased (2004) and then decreased (2012) the number of provinces.
112
F.F. Russo / Economics Letters 121 (2013) 110–114
Table 1 Explaining tax morale.
Signals
Volunt
Civilright
Culture
Politics
R2
0.463** (0.213) [0.319]
0.930 (0.731) [0.136]
−0.094
0.211** (0.102) [0.474]
0.318
(0.328) [−0.051] Water
Crime
R2
Clean Signals
Signals
Pubtransp **
***
−0.118 (0.055) [−0.224]
−0.127 (0.068) [−0.177]
0.144 (0.052) [0.292]
0.261
(0.047) [−0.027]
Primary
Middle
High
College
R2
***
0.611 (0.184) [0.481]
pop2029 Signals
*
−2.361
***
(0.801) [−0.965]
−0.012
***
*
0.249 (0.194) [0.183]
0.765 (0.238) [0.533]
0.942 (0.494) [0.377]
pop3039
pop4049
pop5059
***
1.872 (0.711) [0.411]
−0.571
0.280
−0.736
0.185 (0.93) [0.029]
(0.902) [−0.091]
R2
pop60 **
0.405
(0.359) [−0.564]
Notes: Regression of the number of tax evasion reports per unit of irregular activity on the variable indicated in the column. Volunt is the percentage of volunteers in the population. Civilright is the percentage of civil right activists in the population. Culture is the percentage of participants in cultural associations. Politics is the proportion of the population who read and talk about politics at least once each week. Cleaning is the percentage of the population who are not satisfied with street cleaning. Pubtransp is the percentage of the population who are not satisfied with public transport. Water is the percentage of the population who are not satisfied with the water supply. Crime is the percentage of the population who are concerned about crime. Primary, middle, high, and college are the proportions of the population in primary school, middle school, high school, and college education, respectively. pop2029, pop3039, pop4049, and pop5059 are the proportions of the population in the corresponding age brackets. pop65 is the proportion of the population aged over 65. All of the regressions included a constant, the population in the province, the percentage of internet users, and the regional GDP. The numbers of reports were obtained from the website evasori.info. All of the other variables were obtained from the ISTAT database. The robust standard errors are shown in brackets. The standardized coefficients of the regressions are given in square brackets. * Significant at the 10% level. ** Significant at the 5% level. *** Significant at the 1% level.
the absence of significant differences in the tax pressure among different regions/provinces, because the tax pressure may be negatively correlated to the tax morale (Lago Peñas and Lago Peñas, 2010). The explanatory variables used in the regression are freely available online on the webpage of ISTAT. In all of the regression specifications, I controlled for the size of the population in the province and for the proportion of the population that used the internet frequently, both of which were positively correlated with a number of indicators of tax evasion on the evasori.info webpage. In all of the regressions, I also controlled for the GDP at the regional level. Tables 1 and 2 summarize the results. There were 100 observations in each regression because I purged the data from the two provinces that had a signal number greater than three times the third highest value in the dataset. The first set of variables included proxies of social capital and active political participation. The results are summarized in the first row of Table 1. A stronger sense of community and a more active participation in social life should foster a higher tax morale. Tax payments are only one possible way of contributing to the life of the community and should be correlated with other measures of participation. In agreement with this intuitive assumption, Cannari and D’Alessio (2007) identified a positive relationship between tax morale and social capital in Italy while Alm and Gomez (2008) found a similar relationship in Spain. However, participation in social life at the community level might also promote a locally oriented subculture, which might oppose or be hostile to making contributions to a national Government. The present study obtained mixed results. There was a positive relationship between tax morale and an important measure of social capital: the proportion of the population engaged in volunteer work. A one standard deviation (std henceforth) increase in the percentage of volunteers predicted 1.4 more reports per unit of irregular activity. However, there were no relationships between tax morale and the proportion of the population who participated in associations that promote civil and human rights, or with the proportion of the population who participate in cultural organizations. With respect to
Table 2 Explaining tax morale, continued. Women (1) Signals
Signals
Signals
Signals
Imm (2)
Self (3)
Unemp (4)
Relwed (5)
Church (6)
0.127* (0.051) [0.141]
R2 0.214
0.701** (0.156) [0.649]
0.402
0.319** (0.112) [0.174]
0.224
−0.368**
0.239
(0.105) [−0.372] Signals
−0.089**
0.239
(0.034) [−0.349] Signals
−1.608*
0.226
(0.829) [−0.274] Notes: Regression of the number of tax evasion reports per unit of irregular activity on the variable indicated in the column. Women is the proportion of women in the population. Imm is the proportion of immigrants in the population. Self is the number of self-employed people divided by all employed people. Unemp is the unemployment rate. Relwed is the number of religious weddings relative to the total number of weddings. Church is the percentage of the population who attend church every day. All of the regressions include a constant, the population in the province, the percentage of internet users, and the regional GDP. The numbers of reports were obtained from the website evasori.info. All of the other variables were obtained from the ISTAT database. The robust standard errors are shown in brackets. The standardized coefficients of the regression are given in square brackets. * Significant at the 5% level. ** Significant at the 1% level.
the first measure, it is highly likely that individuals who participate in volunteer work will be law abiding and they have a strong sense of morality, both of which should be reflected in their attitudes towards tax payments. The result related it civil rights variables was driven by the high correlation with volunteer variables, because
F.F. Russo / Economics Letters 121 (2013) 110–114
excluding this last variable from the regression produced a significant coefficient. Participation in cultural organizations was a less obvious indicator of moral beliefs because we might consider that any type of ‘‘antisocial’’ or locally-oriented subculture might belong to this generic category. For example, we might consider an organization that promotes the secession of a geographic area or another that favors fiscal federalism. If the citizens feel they are represented and they participate actively in the political life of their community, they will be more likely to pay taxes and more opposed to tax evasion. In agree with this assumption, Torgler (2005a,b) showed that institutions that fostered direct democracy were predicted to have a higher tax morale while Hug and Spörri (2011) found that these institutions strengthened the link between trust and tax morale. As an indicator of political participation, I used the ‘‘Number of people that read and talked about politics at least once a week’’ (although the results were similar if I used the ‘‘Once a month’’ analog of this indicator). I found that this indicator predicted a higher tax morale. In particular, an increase of one std predicted 1.8 more reports per unit of irregular activity. A closely related indicator considered the satisfaction with essential public services, where more satisfied citizens were more likely to contribute because they felt that their tax money was well spent. Thus, higher satisfaction with public services should predict a higher tax morale (Torgler, 2003; Cannari and D’Alessio, 2007). I considered four survey-based indicators of dissatisfaction with public services: street cleaning, public transportation, water supply, and crime. The results are summarized in the second row of Table 1. Overall, I found mixed support for the hypothesis. In particular, there was a negative and significant coefficient for the first two indicators, dissatisfaction with street cleaning and public transportation, and a negative but insignificant coefficient for the third indicator, dissatisfaction with water supply. A 1 std increase in the satisfaction with street cleaning and public transportation predicted 0.8 and 0.6 more reports per unit of irregular activity, respectively. However, I also found that dissatisfaction with the crime rate was significantly associated with a higher tax morale, which contrasted with the prediction of the hypothesis. A second set of explanatory variables were related to education. The results are summarized in the third row of Table 1. More educated individuals, who are generally better informed about the functioning of the government, may be more likely to pay taxes because they understand their importance, or less likely because they are critical about the way public funds are spent. Thus, there is no obvious theoretical prediction, which has already been noted by Lago Peñas and Lago Peñas (2010). There was a positive and significant relationship between tax morale and the proportion of the population in primary school, high school, and college, and a nonsignificant relationship between tax morale and the proportion of the population in middle school education. A one std higher proportion in college education predicted 1.4 more reports. A one std higher proportion in high school education predicted 1.7 more reports. A one std higher proportion of the population in primary school predicted 1.8 more reports. A third set of explanatory variables considered demographic factors. The result are summarized in the last row of Table 1 and in columns 1 and 2 of Table 2. In agreement with previous studies, tax morale increased with age (Lago Peñas and Lago Peñas, 2010; Torgler, 2005a,b, 2006; Alm and Torgler, 2006; Cummings et al., 2009; Cannari and D’Alessio, 2007). I found that a higher percentage of younger individuals, aged 20–29, was a significant predictor of a lower tax morale. A one std higher percentage predicted 3.5 less reports. By contrast, a higher percentage of individuals aged 30–39 was associated significantly with a higher tax morale, where a one std increase in the proportion was associated with 1.6 more reports. However, a higher proportion
113
of individuals aged over 60 years was actually associated with a lower tax morale, even controlling for the use of the internet, which was much lower among this group. A one std higher proportion predicted 2.1 less reports. The latter result is surprising because older people, who typically live on social security and take advantage of public health care, are highly likely to appreciate the importance of public spending and of tax contributions. Perhaps they are fundamentally dissatisfied with the services they enjoy, although this is not a reason for a lenient attitude toward tax evasion, which could threaten further cuts to public services if it increases. The size of the female population was positively associated with tax morale, which confirms previous demonstrations that women assign higher weights to ethical issues in their decisions (Torgler, 2006; Torgler and Schneider, 2007. A one std increase in the female percentage in the population predicted 0.54 more reports per unit of irregular activity. I also found that a higher percentage of regular immigrants predicted a higher tax morale, where a one std higher proportion of immigrants in the general population was associated with 2.5 more reports per unit of irregular activity. This is not surprising because all immigrants must renew their work permit periodically, so they have to provide proof that they have sufficient income to support themselves and their families. A high proportion of immigrants earn low wages, so evading taxes reduced their likelihood of permit renewal. The fear of non-renewal because of tax offences must also be considered. The picture is different for irregular immigrants, who must be paid illegally by their employers, thereby promoting a tax avoidance culture. However, there is no information about irregular immigrants, so I could not explore this hypothesis further. A fourth set of explanatory variables considered occupation. The results are summarized in columns 3 and 4 of Table 2. Selfemployed individuals should be more likely to favor tax evasion because they are not subject to automatic tax withdrawals (unlike employees). Thus, the tax morale should be lower when the proportion of self-employed is higher (Cannari and D’Alessio, 2007; Lago Peñas and Lago Peñas, 2010). However, I found no evidence of this negative relationship, and instead I actually found that there was a positive relationship. A one std increase in the proportion of self-employed people predicted 0.6 more reports per unit of irregular activity. However, the choice of occupation may have been related to tax morale because an individual might have been self-employed if they were willing to evade taxes, thereby indicating a low tax morale. An alternative explanation2 is that a higher proportion of self-employed people might increase the probability that a given citizen engaged in a transaction with one of them. If self-employed people evaded taxes during transactions, there would be a higher awareness of their tax evasion and a stronger social sanction on tax evasion. I also found that the proportion of unemployed individuals was negatively associated with the tax morale measure. It is possible that some of these individuals earned illegal incomes, which meant they favored tax evasion. A one std increase in the unemployment rate predicted 1.4 more reports per unit of irregular activity. The final set of explanatory variables was related to religion. The results are summarized in columns 5 and 6 of Table 2. The attitude toward tax evasion is part of the system of moral values held by a specific individual. Thus, tax morale is only a single facet of a broader concept of morality. Religion is a factor that affects tax morale, perhaps more than any other. In addition to any metaphysical beliefs, a major part of any religion is concerned with precepts and duties, which shape the behavior of an individual with respect to others and to the civil authorities.
2 I thank a referee for suggesting this possibility.
114
F.F. Russo / Economics Letters 121 (2013) 110–114
Moreover, religion is a way of strengthening the social ties between individuals and fostering a sense of community. Indeed, religiosity is consistently identified as one of the most robust predictors of tax morale within and among countries (Torgler, 2005a,b, 2006; Alm and Torgler, 2006; Lago Peñas and Lago Peñas, 2010). I did not detect this relationship in the present study. Indeed, two indicators of religious beliefs predicted less reports per capita, i.e., the relative frequency of religious weddings relative to the total number of celebrated weddings3 and the proportion of individuals who reported that they attended church once a day. A one std higher percentage of religious weddings was associated with 1.3 less reports per unit of irregular activity, while a one std higher proportion of the population attending church once a day predicted 0.96 less reports per unit of irregular activity. I cannot provide any insights into this result, other than observing that religion and tax morality are not strongly related in Italy. 4. Conclusions This study developed a new measure of tax morale based on voluntary reports of tax evasion and I attempted to identify some of its determinants based on the within-country variability in Italy. I found that tax morale was positively associated with social capital, political participation, and immigration. The positive relationship with social capital is important because involvement in the community can also promote a community-oriented subculture, which may oppose tax contributions to a national government. The positive correlation between tax morale and immigration is also important because it highlights another positive role of immigrants, i.e., their monetary contributions to tax revenue and the fact that they have a positive attitude toward tax payments, thereby setting an important example. Acknowledgments I would like to thank the founder and administrator of the website evasori.info, who asked to remain anonymous, for his
3 To make the measure more precise, I purged weddings where at least one of the spouses was divorced, because a civil wedding was the only option in this case.
cooperation and support. I also thank an anonymous referee. The views expressed in the paper and all remaining errors are only mine. References Allingham, M., Sandmo, A., 1972. Income tax evasion: a theoretical analysis. Journal of Public Economics 1 (3), 323–338. Alm, J., Gomez, J.L., 2008. Social capital and tax morale in Spain. Economic Analysis and Policy 38 (1), 73–87. Alm, J., Torgler, B., 2006. Cultural differences and tax morale in the US and in Europe. Journal of Economic Psychology 27 (2), 224–246. Andreoni, J., Erard, B., Feinstein, J., 1998. Tax compliance. Journal of Economic Literature 36 (2), 818–860. Cannari, L., D’Alessio, G., 2007. Le Opiniomi degli Italiani sull’Evasione Fiscale. In: Temi di Discussione, Bank of Italy. Cummings, R.G., Martinez-Vazquez, J., McKee, M., Torgler, B., 2009. Tax morale affects tax compliance: evidence from surveys and an artefactual field experiment. Journal of Economic Behavior and Organization 70 (3), 447–457. Erard, B., Feinstein, J., 1994. Honesty and evasion in the tax compliance game. Rand Journal of Economics 25 (1), 1–19. Feld, L.P., Frey, B., 2002. Trust breeds trust: how taxpayers are treated. Economics of Governance 3 (2), 87–99. Frey, B., 1997. A constitution for knaves crowds out civic virtues. The Economic Journal 107 (4), 1043–1053. Hug, S., Spörri, F., 2011. Referendums, trust and tax evasion. European Journal of Political Economy 27 (1), 120–131. Lago Peñas, I., Lago Peñas, S., 2010. The determinants of tax morale in comparative perspective: evidence from European countries. European Journal of Political Economy 26 (4), 441–453. Orviska, M., Hudson, J., 2002. Tax evasion, civic duty and the law abiding citizen. European Journal of Political Economy 19 (1), 83–102. Sandmo, A., 2005. The theory of tax evasion: a retrospective view. National Tax Journal 58 (4), 643–663. Slemrod, J., 2007. Cheating ourselves: the economics of tax evasion. The Journal of Economic Perspectives 21 (1), 25–48. Torgler, B., 2003. To evade taxes or not: that is the question. Journal of SocioEconomics 32 (2), 283–302. Torgler, B., 2005a. Tax morale in Latin America. Public Choice 122 (1), 133–157. Torgler, B., 2005b. Tax morale and direct democracy. European Journal of Political Economy 21 (2), 525–531. Torgler, B., 2006. The importance of faith: tax morale and religiosity. Journal of Economic Behavior and Organization 61 (1), 81–109. Torgler, B., Schneider, F., 2007. What shapes attitudes toward paying taxes? Evidence from multicultural European countries. Social Science Quarterly 88 (2), 443–470.
Contents lists available at ScienceDirect
Economics Letters journal homepage: www.elsevier.com/locate/ecolet
Tax morale and tax evasion reports Francesco Flaviano Russo ∗ Department of Economics, University of Naples Federico II, Via Cinthia, Monte Sant’Angelo, 80126 Napoli, Italy CSEF, Via Cinthia, Monte Sant’Angelo, 80126 Napoli, Italy
highlights • • • •
New measure of tax morale based on voluntary reports of tax evasion. Positive relationship between tax morale and social capital. Positive relationship between tax morale and immigration. Negative relationship between tax morale and dissatisfaction with public services.
article
info
Article history: Received 30 November 2012 Received in revised form 4 July 2013 Accepted 5 July 2013 Available online 11 July 2013
abstract The study proposes a new measure of tax morale for Italy based on voluntary tax evasion reports. Using this measure, I detected a positive relationship between tax morale and social capital, political participation, and immigration, but a negative relationship between tax morale and the dissatisfaction with public services and unemployment. © 2013 Elsevier B.V. All rights reserved.
JEL classification: H26 K34 Keywords: Tax evasion Tax morale Social capital
1. Introduction The concept of tax morale was introduced in the economic literature to provide a rational explanation of why people pay taxes (Slemrod, 2007). This was because traditional models of tax evasion Allingham and Sandmo (1972), which are based only on auditing probabilities and expected penalties, underestimate the empirically observed rate of tax compliance significantly (Andreoni et al., 1998; Feld and Frey, 2002; Sandmo, 2005). Thus, in an attempt to reconcile the model with the data, several authors have hypothesized the existence of a moral (Erard and Feinstein, 1994) or intrinsic motivation (Frey, 1997) for paying taxes, which was designated as ‘‘tax morale’’. Previous empirical studies have constructed several measures of tax morale based on a variety of surveys: the International Social Survey Program (Torgler, 2005b), the World Values Surveys (Alm and Torgler, 2006; Torgler, 2006), the European Values Surveys (Lago Peñas and Lago Peñas, 2010; Hug and Spörri, 2011), the British Social Attitudes Survey (Orviska and Hudson, 2002), the
∗
Tel.: +39 081675276. E-mail address: [email protected].
0165-1765/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.econlet.2013.07.004
Latinobarometro (Torgler, 2005a), and the Afrobarometer (Cummings et al., 2009). Typically, tax morale has been measured as the percentage of survey respondents who agree with statements such as: ‘‘Citizens should not cheat on their taxes’’, ‘‘Cheating on taxes is justifiable if there is the opportunity to do so’’, or ‘‘Managing to avoid tax payments is legitimate’’. In this study, I propose a new measure of tax morale, which is based on quasi-experimental data: the voluntary reports of tax evasion collected in Italy via a website evasori.info. This site allowed users to report transactions that lacked a receipt, so were hidden from the tax authority and allowed the supplier to evade taxes. It was expected that buyers who are averse to tax evasion would report such episodes, possibly as a form of social sanction, which may or may not be related to other possible social sanctions, such as exclusion from communal activities, thereby acting as a negative advertisement and causing sellers to lose customers. In the present study, my measure of tax morale was the number of tax evasion reports per unit of irregular activity given the actual level of irregular activity (which by definition entails tax evasion), so a higher number of tax evasion reports is an indicator of a less lenient attitude toward tax evasion and, therefore, of a stronger tax morale.
F.F. Russo / Economics Letters 121 (2013) 110–114
Thus, I conducted an analysis of the determinants of tax morale in Italy. Most of the variables considered have already been discussed previously, i.e., social capital, political participation, working status, education, satisfaction with public services, religion, and demographics. In contrast to previous studies, however, I also considered the relationship between tax morale and immigration in a systematic manner. I found that tax morale was positively associated with social capital, political participation, and immigration, but negatively associated with dissatisfaction with essential public services and unemployment. The most interesting result was the positive association between tax morale and social capital, because involvement in the local community can also foster a locally-oriented subculture, which may oppose national Government contributions and redistribution. The study belongs to the broad category of studies on tax morale and its determinants. The most closely related work, which was conducted by Cannari and D’Alessio (2007), analyzed opinions on tax evasion and the propensity to evade by addressing specific questions in the Survey on Household Income and Wealth (SHIW), which was administered by the Bank of Italy. They find that selfemployed, younger and less educated individuals were more likely to evade taxes, and that this propensity was positively associated with the crime rate and the unemployment rate, but negatively related to social capital. They also found that there was a generally low aversion to tax evasion, which is part of the reason why evasion is so common in Italy. Among other studies, Lago Peñas and Lago Peñas (2010) discussed the relationship between tax morale and the socio-demographic, cultural, and political characteristics of European countries. Alm and Torgler (2006) compared tax morale in the USA and in Europe. Torgler (2005b) and Hug and Spörri (2011) studied the relationship between democratic participation and tax morale in the Swiss Cantons and among countries. The new perspective introduced in the present study is a new measure of tax morale. The remainder of this paper is organized as follows: Section 2 describes the data, Section 3 summarizes the empirical results, and Section 4 provides the conclusions. 2. Measuring tax morale I constructed a measure of tax morale, which was based on voluntary reports of tax evasion submitted to an Italian website evasori.info. The website allowed anyone to submit anonymous reports of their transactions with sellers who failed to issue receipts, thereby evading taxes. These details included the amount of the transaction, the geographic location, and the type of economic activity conducted by the seller (a choice was made from a long list). The concept behind my measure of tax morale was that with the same level of tax evasion, a higher number of tax evasion reports would suggest a higher frequency of social sanctions against this behavior, so it could be used as an indicator of a more negative attitude toward tax evasion. To control for the actual level of tax evasion, I used the rate of irregular activity reported in the Italian Statistical Institute (ISTAT) database. One problem with this measure was that it was only available at the region level (21 observations), whereas the tax evasion reports were available at the province level (102 observations). ISTAT also reports a rate of irregular activity, which is disaggregated into economic activity categories, so I constructed a measure of irregular activity at the provincial level by computing a weighted average of the irregular rates for different activities, where the weights equated to the share of activities in the province. I obtained my measure of tax morale by dividing the total number of reports in a given province by this irregular activity rate. The empirical distribution of my tax morale measure was positively skewed, which reflected the presence of a few large
111
outliers. The mean number of reports per unit of irregular activity was 4.4 and the median was 1.7. The highest value in the sample was 70 (Roma) while the lowest was 0.1 (Vibo Valentia). The standard deviation was 9.5. Excluding the top two observations in the sample (greater than 20 times the median value), I arrived at a mean of 3.2, a standard deviation of 3.8, and a range of 23. My tax morale measure also had high geographic variability, i.e., among the lowest 10 (20) observations, only one (five) belonged to a province in Northern Italy, whereas only one (two) of the top 10 (20) observations belonged to a province in Southern Italy. In my opinion, there are three major problems with my tax morale measure. The first is the interpretation of zeros, i.e., the absence of reports. They could indicate the absence of tax evasion, which may or may not be a consequence of an extremely high tax morale. Alternatively, they could be a sign of a particularly low level of tax morale where evasion is much easier due to a lack of social pressure. Two elements significantly reduced the importance of this ambiguity. First, there were no zeros in the reports dataset and there was a high correlation between population size and the number of reports at the level of the province (0.61). Second, zero tax evasion was never observed in the irregular activity dataset. The second major problem was that the anonymity of reports meant that it was not possible to control for the frequency of reports by the same users. Thus, my measure could have been based on a small number of reports made by a very large sample of users or a large number of reports made by a small sample of users. The problem is that tax morale will potentially have stronger effects on all outcomes if it is more widespread, because this means there is a higher probability of being matched with an individual with a high tax morale during an economic transaction. However, this is only true if we assume an equal number of transactions per individual. Instead, if the number of reports is proportional to the number of transactions, the magnitude and diffusion of tax morale are highly correlated. Thus, I assumed that the individuals who made fewer reports simply had fewer opportunities to make reports. The third problem was related to the interpretations of the reports. Based on the information captured via the website, it is not clear whether they were reports of actual or possible tax evasion. For example, we might consider a case where a seller tries to complete a transaction without issuing a receipt but the buyer notices this and asks for a receipt, thereby preventing tax evasion, yet the buyer still reports the transaction via the webpage. For the purposes of my analysis, there was not a big difference between reports of actual or potential evasion because they are indicators of a negative social attitude, where an individual sanctions the behavior of an (actual of potential) evader. The only difference is that, if the reports only described potential tax evasion, I must refrain from any interpretation of the total level of evasion because the amounts would have been disclosed to the tax authority eventually. Another, perhaps less important, problem is that my measure might be more volatile than tax morale itself. Indeed, a higher tax morale is associated with a lower level of tax evasion and a higher number of reports, which indicates a higher numerator and a small denominator in my measure. 3. Determinants of tax morale What are the possible explanations of tax morale? I performed regressions of my tax morale measure on most of its determinants that have been highlighted in previous studies. The effects were validated based on the variability at the level of the Italian provinces.1 An important additional source of validation was
1 I used the classification of provinces at the time the website evasori.info was created, which was before the two institutional reforms that first increased (2004) and then decreased (2012) the number of provinces.
112
F.F. Russo / Economics Letters 121 (2013) 110–114
Table 1 Explaining tax morale.
Signals
Volunt
Civilright
Culture
Politics
R2
0.463** (0.213) [0.319]
0.930 (0.731) [0.136]
−0.094
0.211** (0.102) [0.474]
0.318
(0.328) [−0.051] Water
Crime
R2
Clean Signals
Signals
Pubtransp **
***
−0.118 (0.055) [−0.224]
−0.127 (0.068) [−0.177]
0.144 (0.052) [0.292]
0.261
(0.047) [−0.027]
Primary
Middle
High
College
R2
***
0.611 (0.184) [0.481]
pop2029 Signals
*
−2.361
***
(0.801) [−0.965]
−0.012
***
*
0.249 (0.194) [0.183]
0.765 (0.238) [0.533]
0.942 (0.494) [0.377]
pop3039
pop4049
pop5059
***
1.872 (0.711) [0.411]
−0.571
0.280
−0.736
0.185 (0.93) [0.029]
(0.902) [−0.091]
R2
pop60 **
0.405
(0.359) [−0.564]
Notes: Regression of the number of tax evasion reports per unit of irregular activity on the variable indicated in the column. Volunt is the percentage of volunteers in the population. Civilright is the percentage of civil right activists in the population. Culture is the percentage of participants in cultural associations. Politics is the proportion of the population who read and talk about politics at least once each week. Cleaning is the percentage of the population who are not satisfied with street cleaning. Pubtransp is the percentage of the population who are not satisfied with public transport. Water is the percentage of the population who are not satisfied with the water supply. Crime is the percentage of the population who are concerned about crime. Primary, middle, high, and college are the proportions of the population in primary school, middle school, high school, and college education, respectively. pop2029, pop3039, pop4049, and pop5059 are the proportions of the population in the corresponding age brackets. pop65 is the proportion of the population aged over 65. All of the regressions included a constant, the population in the province, the percentage of internet users, and the regional GDP. The numbers of reports were obtained from the website evasori.info. All of the other variables were obtained from the ISTAT database. The robust standard errors are shown in brackets. The standardized coefficients of the regressions are given in square brackets. * Significant at the 10% level. ** Significant at the 5% level. *** Significant at the 1% level.
the absence of significant differences in the tax pressure among different regions/provinces, because the tax pressure may be negatively correlated to the tax morale (Lago Peñas and Lago Peñas, 2010). The explanatory variables used in the regression are freely available online on the webpage of ISTAT. In all of the regression specifications, I controlled for the size of the population in the province and for the proportion of the population that used the internet frequently, both of which were positively correlated with a number of indicators of tax evasion on the evasori.info webpage. In all of the regressions, I also controlled for the GDP at the regional level. Tables 1 and 2 summarize the results. There were 100 observations in each regression because I purged the data from the two provinces that had a signal number greater than three times the third highest value in the dataset. The first set of variables included proxies of social capital and active political participation. The results are summarized in the first row of Table 1. A stronger sense of community and a more active participation in social life should foster a higher tax morale. Tax payments are only one possible way of contributing to the life of the community and should be correlated with other measures of participation. In agreement with this intuitive assumption, Cannari and D’Alessio (2007) identified a positive relationship between tax morale and social capital in Italy while Alm and Gomez (2008) found a similar relationship in Spain. However, participation in social life at the community level might also promote a locally oriented subculture, which might oppose or be hostile to making contributions to a national Government. The present study obtained mixed results. There was a positive relationship between tax morale and an important measure of social capital: the proportion of the population engaged in volunteer work. A one standard deviation (std henceforth) increase in the percentage of volunteers predicted 1.4 more reports per unit of irregular activity. However, there were no relationships between tax morale and the proportion of the population who participated in associations that promote civil and human rights, or with the proportion of the population who participate in cultural organizations. With respect to
Table 2 Explaining tax morale, continued. Women (1) Signals
Signals
Signals
Signals
Imm (2)
Self (3)
Unemp (4)
Relwed (5)
Church (6)
0.127* (0.051) [0.141]
R2 0.214
0.701** (0.156) [0.649]
0.402
0.319** (0.112) [0.174]
0.224
−0.368**
0.239
(0.105) [−0.372] Signals
−0.089**
0.239
(0.034) [−0.349] Signals
−1.608*
0.226
(0.829) [−0.274] Notes: Regression of the number of tax evasion reports per unit of irregular activity on the variable indicated in the column. Women is the proportion of women in the population. Imm is the proportion of immigrants in the population. Self is the number of self-employed people divided by all employed people. Unemp is the unemployment rate. Relwed is the number of religious weddings relative to the total number of weddings. Church is the percentage of the population who attend church every day. All of the regressions include a constant, the population in the province, the percentage of internet users, and the regional GDP. The numbers of reports were obtained from the website evasori.info. All of the other variables were obtained from the ISTAT database. The robust standard errors are shown in brackets. The standardized coefficients of the regression are given in square brackets. * Significant at the 5% level. ** Significant at the 1% level.
the first measure, it is highly likely that individuals who participate in volunteer work will be law abiding and they have a strong sense of morality, both of which should be reflected in their attitudes towards tax payments. The result related it civil rights variables was driven by the high correlation with volunteer variables, because
F.F. Russo / Economics Letters 121 (2013) 110–114
excluding this last variable from the regression produced a significant coefficient. Participation in cultural organizations was a less obvious indicator of moral beliefs because we might consider that any type of ‘‘antisocial’’ or locally-oriented subculture might belong to this generic category. For example, we might consider an organization that promotes the secession of a geographic area or another that favors fiscal federalism. If the citizens feel they are represented and they participate actively in the political life of their community, they will be more likely to pay taxes and more opposed to tax evasion. In agree with this assumption, Torgler (2005a,b) showed that institutions that fostered direct democracy were predicted to have a higher tax morale while Hug and Spörri (2011) found that these institutions strengthened the link between trust and tax morale. As an indicator of political participation, I used the ‘‘Number of people that read and talked about politics at least once a week’’ (although the results were similar if I used the ‘‘Once a month’’ analog of this indicator). I found that this indicator predicted a higher tax morale. In particular, an increase of one std predicted 1.8 more reports per unit of irregular activity. A closely related indicator considered the satisfaction with essential public services, where more satisfied citizens were more likely to contribute because they felt that their tax money was well spent. Thus, higher satisfaction with public services should predict a higher tax morale (Torgler, 2003; Cannari and D’Alessio, 2007). I considered four survey-based indicators of dissatisfaction with public services: street cleaning, public transportation, water supply, and crime. The results are summarized in the second row of Table 1. Overall, I found mixed support for the hypothesis. In particular, there was a negative and significant coefficient for the first two indicators, dissatisfaction with street cleaning and public transportation, and a negative but insignificant coefficient for the third indicator, dissatisfaction with water supply. A 1 std increase in the satisfaction with street cleaning and public transportation predicted 0.8 and 0.6 more reports per unit of irregular activity, respectively. However, I also found that dissatisfaction with the crime rate was significantly associated with a higher tax morale, which contrasted with the prediction of the hypothesis. A second set of explanatory variables were related to education. The results are summarized in the third row of Table 1. More educated individuals, who are generally better informed about the functioning of the government, may be more likely to pay taxes because they understand their importance, or less likely because they are critical about the way public funds are spent. Thus, there is no obvious theoretical prediction, which has already been noted by Lago Peñas and Lago Peñas (2010). There was a positive and significant relationship between tax morale and the proportion of the population in primary school, high school, and college, and a nonsignificant relationship between tax morale and the proportion of the population in middle school education. A one std higher proportion in college education predicted 1.4 more reports. A one std higher proportion in high school education predicted 1.7 more reports. A one std higher proportion of the population in primary school predicted 1.8 more reports. A third set of explanatory variables considered demographic factors. The result are summarized in the last row of Table 1 and in columns 1 and 2 of Table 2. In agreement with previous studies, tax morale increased with age (Lago Peñas and Lago Peñas, 2010; Torgler, 2005a,b, 2006; Alm and Torgler, 2006; Cummings et al., 2009; Cannari and D’Alessio, 2007). I found that a higher percentage of younger individuals, aged 20–29, was a significant predictor of a lower tax morale. A one std higher percentage predicted 3.5 less reports. By contrast, a higher percentage of individuals aged 30–39 was associated significantly with a higher tax morale, where a one std increase in the proportion was associated with 1.6 more reports. However, a higher proportion
113
of individuals aged over 60 years was actually associated with a lower tax morale, even controlling for the use of the internet, which was much lower among this group. A one std higher proportion predicted 2.1 less reports. The latter result is surprising because older people, who typically live on social security and take advantage of public health care, are highly likely to appreciate the importance of public spending and of tax contributions. Perhaps they are fundamentally dissatisfied with the services they enjoy, although this is not a reason for a lenient attitude toward tax evasion, which could threaten further cuts to public services if it increases. The size of the female population was positively associated with tax morale, which confirms previous demonstrations that women assign higher weights to ethical issues in their decisions (Torgler, 2006; Torgler and Schneider, 2007. A one std increase in the female percentage in the population predicted 0.54 more reports per unit of irregular activity. I also found that a higher percentage of regular immigrants predicted a higher tax morale, where a one std higher proportion of immigrants in the general population was associated with 2.5 more reports per unit of irregular activity. This is not surprising because all immigrants must renew their work permit periodically, so they have to provide proof that they have sufficient income to support themselves and their families. A high proportion of immigrants earn low wages, so evading taxes reduced their likelihood of permit renewal. The fear of non-renewal because of tax offences must also be considered. The picture is different for irregular immigrants, who must be paid illegally by their employers, thereby promoting a tax avoidance culture. However, there is no information about irregular immigrants, so I could not explore this hypothesis further. A fourth set of explanatory variables considered occupation. The results are summarized in columns 3 and 4 of Table 2. Selfemployed individuals should be more likely to favor tax evasion because they are not subject to automatic tax withdrawals (unlike employees). Thus, the tax morale should be lower when the proportion of self-employed is higher (Cannari and D’Alessio, 2007; Lago Peñas and Lago Peñas, 2010). However, I found no evidence of this negative relationship, and instead I actually found that there was a positive relationship. A one std increase in the proportion of self-employed people predicted 0.6 more reports per unit of irregular activity. However, the choice of occupation may have been related to tax morale because an individual might have been self-employed if they were willing to evade taxes, thereby indicating a low tax morale. An alternative explanation2 is that a higher proportion of self-employed people might increase the probability that a given citizen engaged in a transaction with one of them. If self-employed people evaded taxes during transactions, there would be a higher awareness of their tax evasion and a stronger social sanction on tax evasion. I also found that the proportion of unemployed individuals was negatively associated with the tax morale measure. It is possible that some of these individuals earned illegal incomes, which meant they favored tax evasion. A one std increase in the unemployment rate predicted 1.4 more reports per unit of irregular activity. The final set of explanatory variables was related to religion. The results are summarized in columns 5 and 6 of Table 2. The attitude toward tax evasion is part of the system of moral values held by a specific individual. Thus, tax morale is only a single facet of a broader concept of morality. Religion is a factor that affects tax morale, perhaps more than any other. In addition to any metaphysical beliefs, a major part of any religion is concerned with precepts and duties, which shape the behavior of an individual with respect to others and to the civil authorities.
2 I thank a referee for suggesting this possibility.
114
F.F. Russo / Economics Letters 121 (2013) 110–114
Moreover, religion is a way of strengthening the social ties between individuals and fostering a sense of community. Indeed, religiosity is consistently identified as one of the most robust predictors of tax morale within and among countries (Torgler, 2005a,b, 2006; Alm and Torgler, 2006; Lago Peñas and Lago Peñas, 2010). I did not detect this relationship in the present study. Indeed, two indicators of religious beliefs predicted less reports per capita, i.e., the relative frequency of religious weddings relative to the total number of celebrated weddings3 and the proportion of individuals who reported that they attended church once a day. A one std higher percentage of religious weddings was associated with 1.3 less reports per unit of irregular activity, while a one std higher proportion of the population attending church once a day predicted 0.96 less reports per unit of irregular activity. I cannot provide any insights into this result, other than observing that religion and tax morality are not strongly related in Italy. 4. Conclusions This study developed a new measure of tax morale based on voluntary reports of tax evasion and I attempted to identify some of its determinants based on the within-country variability in Italy. I found that tax morale was positively associated with social capital, political participation, and immigration. The positive relationship with social capital is important because involvement in the community can also promote a community-oriented subculture, which may oppose tax contributions to a national government. The positive correlation between tax morale and immigration is also important because it highlights another positive role of immigrants, i.e., their monetary contributions to tax revenue and the fact that they have a positive attitude toward tax payments, thereby setting an important example. Acknowledgments I would like to thank the founder and administrator of the website evasori.info, who asked to remain anonymous, for his
3 To make the measure more precise, I purged weddings where at least one of the spouses was divorced, because a civil wedding was the only option in this case.
cooperation and support. I also thank an anonymous referee. The views expressed in the paper and all remaining errors are only mine. References Allingham, M., Sandmo, A., 1972. Income tax evasion: a theoretical analysis. Journal of Public Economics 1 (3), 323–338. Alm, J., Gomez, J.L., 2008. Social capital and tax morale in Spain. Economic Analysis and Policy 38 (1), 73–87. Alm, J., Torgler, B., 2006. Cultural differences and tax morale in the US and in Europe. Journal of Economic Psychology 27 (2), 224–246. Andreoni, J., Erard, B., Feinstein, J., 1998. Tax compliance. Journal of Economic Literature 36 (2), 818–860. Cannari, L., D’Alessio, G., 2007. Le Opiniomi degli Italiani sull’Evasione Fiscale. In: Temi di Discussione, Bank of Italy. Cummings, R.G., Martinez-Vazquez, J., McKee, M., Torgler, B., 2009. Tax morale affects tax compliance: evidence from surveys and an artefactual field experiment. Journal of Economic Behavior and Organization 70 (3), 447–457. Erard, B., Feinstein, J., 1994. Honesty and evasion in the tax compliance game. Rand Journal of Economics 25 (1), 1–19. Feld, L.P., Frey, B., 2002. Trust breeds trust: how taxpayers are treated. Economics of Governance 3 (2), 87–99. Frey, B., 1997. A constitution for knaves crowds out civic virtues. The Economic Journal 107 (4), 1043–1053. Hug, S., Spörri, F., 2011. Referendums, trust and tax evasion. European Journal of Political Economy 27 (1), 120–131. Lago Peñas, I., Lago Peñas, S., 2010. The determinants of tax morale in comparative perspective: evidence from European countries. European Journal of Political Economy 26 (4), 441–453. Orviska, M., Hudson, J., 2002. Tax evasion, civic duty and the law abiding citizen. European Journal of Political Economy 19 (1), 83–102. Sandmo, A., 2005. The theory of tax evasion: a retrospective view. National Tax Journal 58 (4), 643–663. Slemrod, J., 2007. Cheating ourselves: the economics of tax evasion. The Journal of Economic Perspectives 21 (1), 25–48. Torgler, B., 2003. To evade taxes or not: that is the question. Journal of SocioEconomics 32 (2), 283–302. Torgler, B., 2005a. Tax morale in Latin America. Public Choice 122 (1), 133–157. Torgler, B., 2005b. Tax morale and direct democracy. European Journal of Political Economy 21 (2), 525–531. Torgler, B., 2006. The importance of faith: tax morale and religiosity. Journal of Economic Behavior and Organization 61 (1), 81–109. Torgler, B., Schneider, F., 2007. What shapes attitudes toward paying taxes? Evidence from multicultural European countries. Social Science Quarterly 88 (2), 443–470.