Are Politically Connected Firms More Likely to Evade Taxes? Evidence from Tunisia

Are politically connected firms more likely to evade taxes? Using tax, social security, and customs records from Tunisia in which firms owned by former president Ben Ali and his family are identified, this paper demonstrates that connected firms were more likely to evade taxes. Ceteris paribus, connected firms are 4.6% more likely not to submit a tax declaration in spite of registering workers and/or customs transactions and 8.4% more likely to report anomalously low sales when submitting a tax declaration.

Are politically connected firms more likely to evade taxation? The jury is out, in spite of the prevalence of political connections (Faccio 2006) and their association with anticompetitive practices (Claessens et al. 2008; Faccio et al. 2006; Mitten and Johnson 2003; Rijkers et al. 2014b). The question is especially relevant for developing countries, since lackluster revenue mobilization (Slemrod and Yithzaki 2002) and problematic state-business relationships are both symptoms and causes of underdevelopment.

Taking advantage of unique administrative and tax data from Tunisia in which firms with ownership connections to former president Ben Ali confiscated in the aftermath of the Jasmin Revolution are identified, this paper answers this question by assessing whether connected firms were more likely to either (i) not declare or (ii) underreport their earnings to the tax authorities.

Tunisia, a small Northern African country at the forefront of the Arab Spring, provides an interesting context to examine this issue. The Ben Ali family had extensive business interests, owning, fully or in part, at least 662 firms, which together accounted for approximately 5% of private sector output. With a revenue to GDP ratio of 21%, its revenue collection performance is not atypical of a country at its stage of development. Findings for Tunisia are thus likely to be relevant for other middle-income and developing countries.

Moreover, Tunisia has a wealth of administrative data, but information sharing between (and within) different government agencies was extremely limited during the Ben Ali era. This in turn facilitates testing for tax evasion through an examination of discrepancies between different administrative databases. More specifically, triangulating social security and customs data with tax data enables us to assess whether firms that reported hiring workers and/or customs transactions submitted a tax declaration and, if so, whether the earnings reported in these declarations are likely to be credible. This simple but powerful strategy of testing for tax evasion can easily be replicated in other countries.

Previewing our main findings, tax evasion is rife in Tunisia; 9% of firms do not submit a tax declaration in spite of reporting hiring workers to the social security administration and/or participating in international trade. Amongst those that submit declarations, 15% report anomalously low sales. Connected firms are among the worst evaders, being ceteris paribus 5% more likely not to declare taxes and 8% more likely to report anomalously low levels of output when submitting tax declarations.

These results contribute to the literature on political connections by highlighting tax evasion as a mechanism by which connected firms reap rents. Estimates of the premium on being connected that do not consider such evasion, which is the norm in the literature, are likely to be downward biased. The findings furthermore add to the literature on the determinants of tax compliance. Our findings also resonate with a companion study in which we demonstrate that connected firms were more likely to evade tariffs by underreporting prices (Rijkers et al. 2015).

I. Hypotheses

Why might politically connected firms be more likely to evade taxes? Models of tax evasion predict it to be a function of rewards, the risk of being caught, and penalties conditional on being caught (see Alm 1999, Andreoni et al. 1998, Slemrod 2007, and Slemrod and Yithzaki 2002, for reviews of the literature). These factors may vary between connected and nonconnected firms. The Ben Ali family was notorious both for being above the law and for abusing it for personal purposes (Government of Tunisia 2011). Officials might have been more willing to tolerate evasion by connected firms because of career concerns, to ingratiate themselves with the ruling elite, or because of fear of reprisals.¹ Alternatively, politically connected firms may have been less risk averse because their owners were wealthier or better capable of coping with the consequences of being caught. However, being associated with corrupt behavior could constitute a more severe reputational risk for connected entrepreneurs with political exposure than for nonconnected entrepreneurs. Connected firms seem to have had better access to information and thus might have been better able to anticipate controls and to judge where and when evasion was more likely to remain undetected.

While distinguishing between these different channels is beyond the scope of this note, they form the basis for the null hypothesis that connected firms are more likely to evade taxes. Moreover, one may expect potential differences between connected and nonconnected firms to vary with the strength of connections.

II. Data

Identifying political connections: In the aftermath of the Jasmin revolution, the Tunisian government ordained by presidential decree (Decret-Loi 2011–13) the confiscation of the assets of 114 individuals belonging to the Ben Ali clan, including Ben Ali himself, his relatives, and his in-laws. The list almost exclusively comprises family members, which helps assuage endogeneity concerns, since only a handful of entrepreneurs managed to select into the family through marriage. The confiscation process is still ongoing. Amongst the assets confiscated thus far are boats, yachts, houses, bank accounts, and 662 firms, which we use to identify firms as politically connected. Of these, there were 261 firms for which listed individuals had 100% ownership. These are referred to as “Fully Ben Ali owned.” Another 226 firms were co-owned by the Ben Ali clan and nonlisted individuals. Such firms are referred to as “Partially Ben Ali owned.” For another 175 firms, the exact ownership share of the Ben Ali family could not be determined on the basis of the information provided to us.² Such firms are referred to as having “Unknown Ben Ali Ownership shares.” This variation in the extent of Ben Ali ownership is used as a proxy for the strength of political connections.³

Administrative data: To assess whether connected firms are more likely to evade taxes, we examine the prevalence of discrepancies between different administrative databases, notably tax records, social security records, and firm-level customs transactions data.⁴ This comparison is useful, since interaction and data sharing both between and within different government agencies was minimal—and entrepreneurs were aware of this. Only after the Revolution were we able to assemble this dataset with the help of the Tunisian authorities, the Tunisian Institute National de la Statistique, and the World Bank. In addition, social security and customs data provide useful objective indicators of economic activity and the quality of social security data is high.⁵

The database we assemble covers all registered private sector firms in Tunisia, including the self-employed. We exclude firms in the regime forfaitaire, for whom controls are limited (typically these are very small firms with sales up to 100.000 TND), and focus exclusively on private firms. Firms in the agricultural and hydrocarbon sectors are also excluded since they face different reporting requirements. We focus on the period 2006–2009 since for this period the richest set of firm characteristics is available.⁶ Because of these sample restrictions, we have a total of 432 connected firms, of which 192 are fully Ben Ali owned, 147 are partially Ben Ali owned, and 93 have unknown degrees of Ben Ali ownership. Descriptive statistics for the estimation sample are presented in appendix table A1.

III. Nonreporting: Missing Tax Declarations

To start with, we test whether connected firms are less likely to submit tax declarations conditional on being active, that is, reporting (i) hiring workers to the social security administration and/or (ii) having customs declarations on importing and/or exporting. Second, we examine the prevalence of reporting anomalously low levels of sales, defined as declarations in which sales reported to the tax authorities are lower than at least one of the following (i) the wage bill reported to the social security administration (ii) total exports or (iii) total imports recorded in customs transactions data. Entering and exiting firms are excluded from the analysis, since discrepancies in the declarations of such firms may be due to starting up and filing of bankruptcy.

Results of OLS regressions⁷ are presented in table 1 in which the dependent variable is an indicator of whether a firm has a missing tax declaration despite being active, that is, recording having hired and/or paid workers to the social security administration and/or recorded customs transactions. The first thing to note is that tax evasion is widespread; 9.0% of nonconnected firms did not submit a tax declaration despite being economically active (see the bottom row).⁸ One might wonder why firms would report hiring workers and/or customs transactions but not submit a tax declaration. A possible explanation is that being registered with the social security administration entitles employees to various benefits, including health insurance, while engaging in international trade without going through customs is arguably difficult (though smuggling does happen).

Second, such evasion is even more widespread among active connected firms. The estimates presented in column 1, which control for being connected and year dummies only, suggest that connected firms are 1.9% more likely than nonconnected firms not to submit a declaration, but the differential with nonconnected firms is not statistically significant. Once their industry, location, and superior size and trade performance are controlled for, as is done in column 2, the evasion premium on being connected rises to 4.6% percentage points and becomes significant at the 1% level.

To assess whether the evasion propensity varies with the strength of political connections, columns 3 replicates the second specification 2, but now separates Ben Ali firms into those that are (i) known to be owned only in part by the Ben Ali family (ii) firms which are fully Ben Ali owned and (iii) a residual group for which the exact Ben Ali ownership share is not known. While coefficient estimates are positive for all three groups, they are only significant for the group of fully Ben Ali owned firms, for whom the connectedness premium is highest; ceteris paribus they are 6.4% more likely not to submit any tax declarations than nonconnected firms.

IV. Underreporting

We now turn to examining the prevalence of underreporting amongst those firms that submitted tax declarations and reported hiring workers and/or customs transactions, by examining the incidence of anomalously low levels of sales defined here as sales that are lower than either the wage bill, total imports, or total exports.⁹

Table 2 presents results using specifications that mirror those presented in table 1 but using as dependent variable having submitted an anomalous declaration. Such anomalous declarations are prevalent; no less than 15.3% of nonconnected firms submit a tax declaration in which they claim their sales were lower than their wagebill, total exports, or total imports. Amongst Ben Ali firms submitting tax declarations, this percentage is an additional 6.6% higher (column 1), and the differential with nonconnected firms is statistically significant. Controlling for other firm characteristics, such as industry, location, age, size, exports, and imports, connected firms are 8.4% more likely to submit an anomalous declaration (column 2). Differentiating by type of connection, as is done in column 3, demonstrates that fully Ben Ali owned firms are most likely to submit anomalous declarations – in spite of being least likely to submit tax declarations in the first place.

Note that an anomalous declaration does not constitute proof of tax evasion by itself; it may reflect poor management, measurement error, and/or time lags in the production or sales process. Table A2 in the appendix presents robustness checks that demonstrate that the results also obtain when focusing on profit making firms only (columns 1 and 2). Moreover, they are robust to using more stringent definitions of anomalous declarations, which allow for lags between the production and sales processes, notably (ii) defining a declaration to be anomalous if cumulative output over a two year period is lower than either the wagebill, exports, or imports in the base year (columns 3 and 4) and (iii) average output between 2006 and 2009 is lower than either the average wagebill, average imports, or average exports over the same period (columns 5 and 6). Using either of these alternative measures, connected firms are consistently significantly more likely to report anomalously low levels of sales.

V. Conclusion

By examining discrepancies between administrative data from Tunisia that were curated without coordination between different government agencies, this paper unveils widespread tax evasion; 9% of firms do not submit a tax declaration in spite of registering workers, and 15% report earnings which are anomalously low compared to declarations made to other government agencies. Firms owned by the Ben Ali family were among the worst perpetrators, being ceteris paribus both 5% less likely to submit tax declarations and 8% more likely to underreport earnings to the tax authorities. Fully Ben Ali owned firms have the highest evasion propensity.

Overall, our findings not only attest to tax evasion being a mechanism by which connected entrepreneurs reap rents (see also the companion paper, Rijkers et al. 2015, which demonstrates that connected firms were more likely to evade tariffs) but also suggest that estimates of the impact of political connections on firm performance that do not consider such evasion are likely biased downwards. While the methods used in this paper are simple, they are powerful and can easily be replicated in other countries where information sharing across institutions is imperfect.

Appendix Tables

References

Author notes