Does globalisation affect the shadow economy?

DOI: 10.1111/twec.12549 ORIGINAL ARTICLE Does globalisation affect the shadow economy? Aziz N. Berdiev 1 James W. Saunoris 2 1 Department of Economics, Bryant University, Smithfield, RI, USA 2 Department of Economics, Eastern Michigan University, Ypsilanti, MI, USA 1 INTRODUCTION The presence of the shadow economy exists in all countries to varying degrees (Loayza, 2016; Schneider, 2005; Schneider & Enste, 2000). 1 For instance, among developed countries, it is estimated that around 16% of total production takes place in the shadow whereas for developing countries approximately 38% of their production is underground, while in some cases (Bolivia; 66%) a majority of their economy is underground (Schneider, 2005). Interest in the shadow economy is partly in response to the potentially large costs they impose on the official sector. For example, the shadow economy weakens the government s ability to collect taxes, undermines established institutions, distorts relative prices and allocations of resources to the extent the less productive underground economy competes with the official sector, and misrepresents official statistics in which many policies are based (see G erxhani, 2004; Schneider & Enste, 2000). Alternatively, it is possible that the shadow economy provides benefits to the formal sector. For example, Dell Anno and Solomon (2008) find that the shadow economy employs unemployed individuals from the formal sector, thereby lessening the negative consequences of unemployment on the formal sector. Moreover, the shadow economy contributes to economic growth and development by the creation of markets, increase financial resources, enhance entrepreneurship, and transform the legal, social, and economic institutions necessary for accumulation (Asea, 1996, p. 166, cited in Schneider & Enste, 2000). The widespread prevalence of the shadow economy and the costs and benefits it imposes has thus prompted researchers and policymakers to better under the factors driving the size of the shadow economy in an attempt to curtail its spread or encourage shadow activities to transition to the formal sector (G erxhani, 2004). In addition to the growing concern over the consequences of the shadow economy worldwide, there has been an increasing acceptance towards opening borders and becoming more globally integrated (economically, politically and socially) to exploit the benefits of globalisation. Globalisation as defined by Keohane and Nye (2000, p. 2) is a state of the world involving networks of interdependence at multi-continental distances. Globalisation contributes to, among other things, the sharing of ideas, reducing trade barriers, increasing capital flows and lowering costs of 1 Although we use the terms informal and shadow economy interchangeably, we are referring to firms and workers that operate outside the legal and regulatory framework (see Loayza, 2016); that is, shadow economy production includes activity that, had it been recorded, would be included in official GNP estimates; therefore, it excludes the criminal sector of the economy (e.g. drug production). 222 2017 John Wiley & Sons Ltd wileyonlinelibrary.com/journal/twec World Econ. 2018;41:222 241.

BERDIEV AND SAUNORIS 223 transportation (World Bank, 2002). Likewise, the benefits of globalisation show up in increasing, for example, economic growth (see Dreher, 2006). Whereas the effects of globalisation on economic growth have been well documented in the literature, the effects on the shadow economy are less forthcoming. 2 To fill this void, in this paper we study the impact of globalisation on the shadow economy. Since the process of globalisation is comprised of numerous dimensions (Dreher, 2006; Potrafke, 2010), we also investigate the effect of economic, political and social globalisation on the shadow economy; however, it is important to recognise that the disaggregated measures of globalisation are not mutually exclusive and instead are intricately related. We use globalisation variables developed by Dreher (2006) and further summarised in Dreher, Gaston, and Martens (2008). Moreover, we employ the measures of the shadow economy calculated by Elgin and Oztunali (2012) and Schneider, Buehn, and Montenegro (2010). Using panel data for 119 countries, we find that globalisation matters in mitigating shadow development. In particular, after controlling for important factors that affect the shadow economy, our evidence suggests that it is the political (rather than economic or social) element of globalisation that is primarily responsible for impeding shadow development. These findings are robust after accounting for an alternative measure of the shadow economy, outliers, endogeneity, and alternative model specifications. The rest of the paper is organised as follows: Section 2 describes the relationship between globalisation and the shadow economy; Section 3 describes the data and the model; Section 4 presents the empirical results, while Section 5 offers a number of robustness tests. We summarise our major findings in the final section. 2 GLOBALISATION AND THE SHADOW ECONOMY Understanding the effects of globalisation and the shadow economy can be best understood using a standard two-sector growth model that is, formal and informal sectors where labour, entrepreneurs and managers have the ability to escape the formal sector (including a government) and migrate to the informal sector (see, e.g. Loayza, 1996, 2016; Rauch, 1991). 3 As defined above, the informal sector comprises firms and workers that operate outside the legal and regulatory framework. The attractiveness of the informal sector is determined by the cost-benefit differential between the two sectors. For example, the development of the informal economy is determined by the trade-off between informal firms benefiting from avoiding high labour costs (e.g. minimum wage laws, hiring and firing regulations) at the cost of higher capital costs and lower productivity (see, e.g. Loayza, 1996, 2016; Rauch, 1991). 4 Therefore, the taxes and regulatory burdens imposed in the formal sector encourage economic actors to evade these by employing their resources in the informal economy (see, e.g. G erxhani, 2004; Johnson et al., 1997; Loayza, 1996; Loayza et al., 2009; Schneider, 2011; Schneider & Enste, 2000; Tanzi, 1982). 5 Additionally, the quality of formal institutions distorts the 2 For a relevant exception, see, for example, Fugazza and Fiess (2010) and more recently Pham (2017). 3 Related to this literature is the economics of tax evasion by Allingham and Sandmo (1972), according to which, rational individuals weigh the benefits from evading taxes against the probability of detection. 4 Loayza (1996) and Loayza, Serven, and Sugawara (2009) provide an excellent discussion on the costs of formality and informality. Kaufmann (1997) also notes that companies examine the costs and benefits of participating in the official versus the shadow economy. 5 Related to these causal factors includes the citizen s perception of the state referred to as tax morality, which has been shown to significantly influence shadow development (see Torgler & Schneider, 2007).

224 BERDIEV AND SAUNORIS cost-benefit trade-off. Consequently, shadow participants weigh the benefits from evading taxes, bypassing government regulations and circumventing the high transactions costs generated by poor institutions against the direct costs (e.g. fines) and indirect, or opportunity, costs (e.g. forgone use of official sector institutions such as the court system, police and financial system). Globalisation distorts the relative costs and benefits of producing in the shadow economy relative to the formal economy. In particular, globalisation may mitigate the spread of the shadow economy by improving institutional quality (Bonaglia, De Macedo, & Bussolo, 2001; Dreher, Kotsogiannis, & McCorriston, 2009; Friedman, Johnson, Kaufmann, & Zoido-Lobaton, 2000; Simmons & Elkins, 2004). For instance, the process of global integration facilitates the diffusion of institutions as it erodes national boundaries, integrating national economies, cultures, technologies, and governance (Norris, 2000, p. 155). For example, if, according to De Soto (1989), the shadow economy is a response to an over-regulated economy, then the competitive pressures resulting from globalisation will likely put pressure on governments to deregulate. Moreover, for those formal firms that engage in international trade, globalisation likely leads to improved worker productivity and capital investment, especially in those countries with low domestic finance capacity. Consequently, firms are incentivised to shift production from the informal sector to the formal sector. 6 In what follows, we define the three main facets of globalisation (economic, social and political) following Dreher (2006), Dreher et al. (2008) and Keohane and Nye (2000), and identify several scenarios in which globalisation impacts the relative costs and benefits of underground activities that ultimately influence its spread. 7 Economic globalisation refers to long-distance flows of goods, services, and capital, and the information and perceptions that accompany market exchange (Keohane & Nye, 2000, p. 4). Our measure of economic globalisation consists of actual flows (trade, foreign direct investment, portfolio investment and income payments to foreign nationals) and restrictions (hidden import barriers, mean tariff rate, taxes on international trade and capital account restrictions) (see Dreher, 2006; Dreher et al., 2008). Research suggests that burdensome taxes and regulations provide incentives for firms to operate in the informal sector by raising the costs of producing in the official sector (see, e.g. G erxhani, 2004; Schneider & Enste, 2000). Economic globalisation through, for example, reducing trade barriers may mitigate the spread of the shadow economy. Greater restrictions on international trade may encourage firms to move to the shadow sector to evade costs associated with, for example, higher tariffs. Specifically, the presence of trade barriers increases the costs of transporting goods through the formal sector, thereby motivating firms to transport their goods through the informal sector (e.g. smuggling) in order to circumvent tariffs (Buehn & Farzanegan, 2012). Mishkin (2009, p. 166) also suggests that importers have incentives to pay customs officials to look the other way when the importers avoid tariffs by smuggling in goods. 8 In fact, the empirical literature finds that greater trade restrictions promote smuggling (see, e.g. Buehn & Farzanegan, 2012). Schneider and Enste (2000) thus argue that policies to promote greater economic integration through, for example, eliminating trade restrictions may generate incentives for participants to migrate from the informal sector to the formal sector. 6 We thank an anonymous referee for this suggestion. 7 It is important to recognise that there is a significant overlap among the three types of globalisation. Also, the three elements of globalisation are somewhat narrowly defined for example, economic globalisation is focused on trade. Therefore, caution should be used when interpreting the empirical results. 8 See also Bhagwati and Hansen (1973) for a theoretical model of smuggling.

BERDIEV AND SAUNORIS 225 Social globalisation represents movements of ideas, information, and images, and of people who of course carry ideas and information with them (Keohane & Nye, 2000, p. 5). Our measure of social globalisation contains data on personal contact (telephone traffic, transfers, internal tourism, foreign population and international letters), data on information flows (Internet users, television and trade in newspapers), and data on cultural proximity (number of McDonald s restaurants, number of Ikeas and trade in books; see Dreher, 2006; Dreher et al., 2008). Social globalisation may reduce the spread of the shadow economy by improving institutional quality (e.g. improvements in the rule of law, macroeconomic policies, property rights, constitution, democracy and reduction in corruption). In particular, social globalisation through for example diffusion of scientific knowledge (i.e., transfer of ideas and information) (Keohane & Nye, 2000) can facilitate the spread of sound institutions. 9 This transfer of ideas and information can take place through the Internet, television, and the presence of foreign population, to name a few. For example, the diffusion of knowledge through the Internet about the presence of local corruption may cause the general population to demand that the government fights corruption (Goel, Nelson, & Naretta, 2012). In fact, Goel et al. (2012) find that the frequency of corruption-posting activity on the Internet reduces corruption. Similarly, it is possible that the spread of knowledge about the pervasiveness of the domestic shadow economy promotes citizens to require that their government mitigates the spread of the shadow economy. Elgin (2013) finds that countries that experience high Internet usage enjoy lower shadow economies. The transmission of ideas, information and knowledge also enables countries to discover, for example, the amendments in government policies in neighbouring countries (Starr, 1991). It is possible that then decision makers and peoples might use that information as cues and sources of emulation as argued by Starr (1991, p. 360). Indeed, Starr (1991) shows the spread of democracy across national borders. Likewise, Keohane and Nye (2000, p. 4) emphasise that a critically imperative feature of social globalisation includes imitation of one society s practices and institutions by others. To the extent that voters rely on policies of neighbouring countries to determine the competence of their policymakers, social globalisation through movement of information, ideas and knowledge could enhance this so-called yardstick competition and lead to more effective government policies that then reduce the size of the shadow economy (see, e.g. Rogoff & Sibert, 1988). Political globalisation epitomises the diffusion of government policies, or of international regimes (Keohane & Nye, 2000, p. 5). Our measure of political globalisation contains data on embassies in countries, memberships in international organisations, participation in UN Security Council missions and international treaties (see Dreher, 2006; Dreher et al., 2008). Political globalisation, through diffusion of sound government policies, can inhibit shadow economic development. Government policy diffusion can take place by, for example, holding memberships in international organisations. Sandholtz and Gray (2003) argue that international organisations are a round-table for the diffusion of government policies (e.g. policies to fight corruption). Because most international organisations are filled with prosperous countries, the norms and ideals of these prosperous countries will prevail in these establishments (Sandholtz & Gray, 2003). In fact, Sandholtz and Gray (2003) find that countries that participate in international organisations enjoy lower corruption. Research documents that corruption increases the costs of producing in the official sector, thereby providing an incentive for individuals to migrate underground in order to avoid dealing with corrupt public agents (see, e.g. Hindriks, Keen, & Muthoo, 1999; Hibbs & Piculescu, 2005). This complementarity between corruption and the shadow economy is supported in the empirical literature (see, e.g. Buehn & Schneider, 2012; Johnson, Kaufmann, & 9 See Knack and Keefer (1995) for a broader discussion of the role of institutions.

226 BERDIEV AND SAUNORIS Shleifer, 1997). 10 Thus, participation in international organisation by foreign governments may provide an environment where governments, through policy diffusion, learn from each other on ways to mitigate the spread of the shadow economy. Moreover, it is through political globalisation that governments can learn from other governments on ways to improve institutions that ultimately reduce shadow production. For instance, according to Simmons and Elkins (2004), these information externalities prompt governments to liberalise following their competitors, and implement policies of successful countries. Institutional and policy improvements brought about by political globalisation effectively raise the opportunity cost of producing in the shadow economy. Furthermore, participation in international organisations by foreign governments would likely keep these governments in check and thus raise their credibility (Dreher & Voigt, 2011). If individuals have more trust in their government, they are less willing to migrate to the shadow to evade taxes (see Wintrobe, 2001). Furthermore, an increase in trust in government likely raises tax morale, which has been shown to be an important deterrent of shadow economic activity (see, e.g. Torgler & Schneider, 2007). In summary, the above arguments lead us to our main hypothesis that globalisation economic, social and political mitigates the spread of the shadow economy. 3 DATA AND EMPIRICAL METHODOLOGY 3.1 Data We use two measures of the shadow economy. The first measure of the shadow economy (shadow1) comes from Schneider et al. (2010), who estimate the shadow economy (% of GDP) by employing the multiple indicators multiple causes (MIMIC) model. The MIMIC model involves estimating the size of the shadow economy as a latent variable using a factor analytic and structural model of observable indicators and causes. The second measure of the shadow economy (shadow2) is obtained from Elgin and Oztunali (2012), who construct a measure of the shadow economy (% of GDP) using a two-sector dynamic general equilibrium model. The globalisation data come from Dreher (2006) and Dreher et al. (2008). 11 The authors build an index of globalisation that includes three main dimensions: economic, political and social. These three indexes are combined to create the overall globalisation index (see, for details, Dreher, 2006; Dreher et al., 2008). All globalisation indexes range between 1 and 100, where higher values denote a greater degree of globalisation. The previous section defines these three facets of globalisation and describes the variables that are utilised in their construction. We provide a purely descriptive look at the relationship between the globalisation variables and the size of the shadow economy (shadow1) in Figure 1. As shown, these figures illustrate a negative correlation between the size of the shadow economy and the various measures of globalisation. However, these figures present only a cursory look at the data; the next section proceeds with a formal analysis on the relationship between globalisation and the size of the shadow economy. 10 Goel and Saunoris (2014) also show a complementary relationship between corruption and the shadow economy when accounting for neighbouring interactions. Whereas Goel and Saunoris (2014) focus their analysis on local spillovers, globalisation likely influences the shadow economy on a more global scale. However, to the extent that globalisation reduces trade barriers, this would help prevent these local spillovers for example, by reducing the need for smuggling. 11 The globalisation data have been used in more than 100 studies (see, for a detailed literature review, Potrafke, 2015). For example, the literature has utilised these globalisation indexes to study the effect of globalisation, among many other areas, on electoral turnout (Steiner, 2010), life expectancy (Bergh & Nilsson, 2010), corruption (Potrafke, 2012), social expenditures (Gaston & Rajaguru, 2013), poverty (Bergh & Nilsson, 2014) and democracy (Gassebner, Lamla, & Vreeland, 2013).

BERDIEV AND SAUNORIS 227 0 0 20 20 40 40 60 60 80 80 20 40 60 80 Overall Globalisation 100 0 20 40 60 80 Economic Globalisation 100 Fitted Values Size of Shadow Economy (% of GDP) (Shadow1) Fitted Values Size of Shadow Economy (% of GDP) (Shadow1) 0 0 20 20 40 40 60 60 80 80 0 20 40 60 80 100 0 20 40 60 80 Political Globalisation Social Globalisation 100 Fitted Values Size of Shadow Economy (% of GDP) (Shadow1) Fitted Values Size of Shadow Economy (% of GDP) (Shadow1) FIGURE 1 Globalisation and the shadow economy Specifically, it is imperative to control for the determinants of the shadow economy before drawing any strong conclusions about these correlations. Following the extant literature, we thus control for the main determinants of the shadow economy (see, for reviews, G erxhani, 2004; Schneider, 2005; Schneider & Enste, 2000) see also Table 1 for variables details. We employ (log) real GDP per capita to control for the level of economic development. More prosperous countries are likely to have better institutions that discourage shadow operations. We also account for the level of education because more investment in human capital raises the opportunity costs of producing in the shadow economy (Berdiev, Pasquesi-Hill, & Saunoris, 2015; Buehn & Farzanegan, 2013; G erxhani & van de Werfhorst, 2013; Loayza et al., 2009). Next, we control for the stringencies of regulations related to credit markets and labour markets. More burdensome regulations induce firms and workers to operate underground or outsource to the underground (Schneider & Enste, 2000). Alternatively, regulations that apply to both formal and informal sectors may reduce the size of the shadow economy. Finally, we account for government size. On the one hand, larger governments possess greater resources to combat shadow activities, but on the other hand, larger governments proxy for an increase in government overreach that could encourage migration to the shadow economy. The definitions, data sources and descriptive statistics for all the variables are displayed in Table 1.

228 BERDIEV AND SAUNORIS TABLE 1 Variable definitions and sources Variable Description [mean; SD] Source Shadow1 Size of the shadow economy (% of GDP) calculated by employing the multiple indicators multiple causes method [33.06; 12.81] Schneider et al. (2010) Shadow2 Overall globalisation Size of the shadow economy (% of GDP) calculated by employing a two-sector dynamic general equilibrium model [36.55; 14.80] Overall globalisation index on a scale from 1 to 100; higher values denote greater overall globalisation [45.14; 17.74] Economic globalisation Economic globalisation index on a scale from 1 to 100; higher values denote greater economic globalisation [49.87; 19.30] Political globalisation Political globalisation index on a scale from 1 to 100; higher values denote greater political globalisation [46.90; 26.83] Social globalisation Log GDP per capita Education Credit market regulation Labour market regulation Government size Social globalisation index on a scale from 1 to 100; higher values denote greater social globalisation [41.26; 21.16] The natural log of gross domestic product per capita in constant PPP adjusted [8.55; 1.27] Secondary school enrolment as a percentage of gross enrolment [61.27; 34.05] A component of regulation of credit, labour and business on a scale from 0 to 10; higher scores indicate lower regulation [7.67; 1.43] A component of regulation of credit, labour and business on a scale from 0 to 10; higher scores indicate lower regulation [5.49; 1.40] Size of government by taxes, expenditures and enterprises on a scale from 0 to 10; higher scores indicate smaller government [6.29; 1.45] Elgin and Oztunali (2012) Dreher (2006), Dreher et al. (2008) Dreher (2006), Dreher et al. (2008) Dreher (2006), Dreher et al. (2008) Dreher (2006), Dreher et al. (2008) World Bank (2012) World Bank (2012) Gwartney et al. (2009) Gwartney et al. (2009) Gwartney et al. (2009) 3.2 Empirical model The model to be estimated is as follows: shadow it ¼ a i þ s t þ bglobalization it þ c 0 X it þ e it ; (1) where i and t index country and year, respectively. The dependent variable shadow it denotes either shadow1 or shadow2; the variable globalization it corresponds to either overall, economic, political or social globalisation; X it represents the vector of explanatory variables as discussed above; e it is the error term; country-specific fixed effects are denoted by a i ; and time effects are denoted by s t. We estimate Equation (1) using the ordinary least squares (OLS) method while assuming different structures of the composite error term (labelled at the bottom of the table). Specifically, we estimate Equation (1) using (i) OLS by assuming a homogeneous intercept; (ii) OLS with regional dummies; (iii) panel fixed effects (the within estimator); and (iv) panel fixed effects (the within

BERDIEV AND SAUNORIS 229 estimator) with time dummies. Due to limited data availability for all the variables, the empirical analysis is based upon panel data for 119 countries over the period 2000 2007. The list of countries is reported in Table A1 in Appendix. 4 RESULTS The regression estimates for the dependent variable shadow1 are displayed in Table 2. We find that overall globalisation reduces the size of the shadow economy, at the 1% level of significance (column (1)). To account for possible heterogeneity in the sample, we include regional effects in column (2), and control for country fixed effects in column (3). In column (4), we control for time and country fixed effects. The evidence continues to suggest that overall globalisation significantly shrinks the shadow economy. 12 As we emphasised earlier, globalisation is comprised of several dimensions (Dreher, 2006; Potrafke, 2010). The advantage of the globalisation data is that it enables us to distinguish between economic, political and social integration. We therefore analyse the impact of economic, political and social globalisations on the shadow economy. As suggested in Dreher (2006), these globalisation variables are estimated separately (columns (5) (7)) and jointly (column (8)) in Table 2. We estimate each equation using time and country fixed effects. The results here suggest that economic and political globalisation, and not social globalisation, significantly reduce the size of the shadow economy. In Table 3, we report the results after re-estimating the models in Table 2 while controlling for other factors that contribute to the size of the shadow economy. Again we find that the coefficient on overall globalisation is negative and statistically significant, at least at the 5% level in all equations, except when time and country fixed effects are controlled for, the effect of globalisation is still negative, but insignificant (column (4)). Regarding our control variables, higher GDP per capita reduces the shadow economy, at the 1% level of significance. Moreover, smaller governments significantly increase the shadow economy. Overall, we find that education, credit market regulation and labour market regulation have no robust impact on the shadow economy. We find that political globalisation is negative and significant at conventional levels in both equations (columns (6) and (8)), whereas social and economic globalisations have no significant influence on shadow development. Interestingly, once we control for factors that can be considered economic factors, the coefficient on economic globalisation is insignificant. 13,14 Our finding on the economic globalisation is broadly consistent with those of Torgler and Schneider (2007) and Teobaldelli (2011), who find that trade openness has no robust influence on shadow operations. 15 In addition, Buehn and Farzanegan (2012) show that trade openness has no significant impact on smuggling. The remaining explanatory variables are in line with our earlier findings. The exception is the variable labour market regulation, which is now positive and significant. 12 As a robustness check, we also used the Global Connectedness Index from Ghemawat and Altman (2014) as an alternate measure of globalisation. These results confirmed our main results and are available upon request. 13 We thank an anonymous referee for pointing this out. 14 After some additional analysis, GDP proves to be the main variable muting the effects of economic globalisation. This makes sense given the high correlation (0.79) between economic globalisation and GDP. 15 Interestingly, when economic globalisation is decomposed into actual flows and restrictions, only restrictions (i.e., relaxing restrictions) negatively influence the size of the shadow economy. Likewise, when social globalisation is decomposed, the variables personal contact, information flows and cultural proximity are statistically insignificant at conventional levels. These results are available upon request.

230 BERDIEV AND SAUNORIS TABLE 2 Regression estimates (dependent variable: shadow1) (1) (2) (3) (4) (5) (6) (7) (8) Overall globalisation 0.504*** 0.621*** 0.252*** 0.078*** (0.021) (0.040) (0.026) (0.024) Economic globalisation 0.032** 0.028** (0.014) (0.014) Political globalisation 0.037*** 0.034*** (0.012) (0.012) Social globalisation 0.011 0.00005 (0.024) (0.023) Constant 62.421*** 63.951*** 46.013*** 33.616*** 32.457*** 31.272*** 28.988*** 33.000*** (1.570) (3.286) (1.682) (1.625) (0.890) (0.981) (1.457) (1.658) Country effects No No Yes Yes Yes Yes Yes Yes Regional effects No Yes No No No No No No Time effects No No No Yes Yes Yes Yes Yes Number of observations 685 685 685 685 685 685 685 685 R 2 0.367 0.516 0.335 0.602 0.592 0.599 0.583 0.607 Notes: Robust standard errors in parenthesis. ***p <.01, **p <.05, *p <.10.

BERDIEV AND SAUNORIS 231 TABLE 3 Regression estimates (dependent variable: shadow1) (1) (2) (3) (4) (5) (6) (7) (8) Overall globalisation 0.207*** 0.280*** 0.051** 0.028 (0.051) (0.056) (0.023) (0.019) Economic globalisation 0.012 0.012 (0.012) (0.012) Political globalisation 0.015* 0.015* (0.009) (0.009) Social globalisation 0.005 0.009 (0.014) (0.014) Log GDP per capita 6.442*** 7.230*** 7.839*** 5.547*** 5.632*** 5.632*** 5.813*** 5.509*** (0.742) (0.712) (1.009) (1.161) (1.135) (1.143) (1.121) (1.157) Education 0.103*** 0.051* 0.005 0.006 0.007 0.007 0.008 0.006 (0.030) (0.026) (0.007) (0.006) (0.007) (0.006) (0.006) (0.007) Credit market regulations 1.020*** 1.111*** 0.144 0.060 0.062 0.063 0.067 0.061 (0.372) (0.348) (0.090) (0.097) (0.098) (0.096) (0.098) (0.097) Labour market regulations 1.124*** 0.092 0.093 0.207*** 0.210*** 0.197*** 0.200*** 0.207*** (0.252) (0.253) (0.089) (0.075) (0.077) (0.075) (0.075) (0.076) Government size 1.670*** 0.869*** 0.161** 0.217*** 0.224*** 0.206*** 0.222*** 0.207*** (0.276) (0.246) (0.070) (0.069) (0.070) (0.067) (0.070) (0.067) Constant 80.849*** 88.775*** 104.266*** 79.823*** 80.483*** 80.999*** 80.247*** 79.992*** (5.060) (5.700) (8.279) (10.424) (10.181) (10.068) (10.277) (10.186) Country effects No No Yes Yes Yes Yes Yes Yes Regional effects No Yes No No No No No No Time effects No No No Yes Yes Yes Yes Yes Number of observations 685 685 685 685 685 685 685 685 R 2 0.462 0.600 0.667 0.713 0.712 0.714 0.711 0.715 Notes: Robust standard errors in parenthesis. ***p <.01, **p <.05, *p <.10.

232 BERDIEV AND SAUNORIS In summary, our evidence indicates that globalisation significantly reduces the size of the shadow economy. This evidence suggests that the benefits from global integration and interdependence spillover to the shadow economy; that is, globalisation appears to significantly raise the perceived costs (or lower the benefits) of informal activities, thereby mitigating the spread of the shadow economy. Because globalisation is multidimensional, we use a disaggregated measure of globalisation to determine the channel through which globalisation influences the shadow economy. We find that political globalisation significantly reduces the size of the shadow economy, whereas economic and social globalisations have no significant effects after controlling for important factors that affect the size of the shadow economy. These results highlight the importance of countries opening their political borders to prevent the development of the shadow economy. 5 ROBUSTNESS TESTS To verify the validity of our previous findings, we conduct a number of robustness tests, which are displayed in Tables 4 6. In particular, we check the sensitivity of our results to (i) an alternate measure of the shadow economy (Table 4); (ii) the presence of outliers (Table 5, columns (1) (4)); (iii) the possible endogeneity of globalisation (Table 4, columns (5) (8)); and (iv) the inclusion of alternate control variables based on initial conditions (Table 6). 16 5.1 Alternate measure of the shadow economy For obvious reasons, activities in the shadow economy are concealed, thereby making it difficult to accurately measure its size. Consequently, we utilise an alternative measure for the size of the shadow economy (shadow2) estimated by Elgin and Oztunali (2012) to check the robustness of our earlier findings. Table 3 shows the results from re-estimating all equations using the dependent variable shadow2. The results show that overall globalisation is negative and significant at the 5% level in all specifications, in line with our previous finding that more globalised countries experience smaller shadow economies. Moreover, political globalisation reduces the size of the shadow economy, at the 5% level of significance. As before, we find no evidence that economic and social globalisation impact shadow development. The results for the control variables are similar with the exception that government size is not robustly significant across specifications and labour market regulation is now negative and significant coefficient (in four of eight equations). 5.2 Outlier test Next, we check the sensitivity of our findings to the presence of outliers using robust regressions. To do this, we use Cook s distance (<1) to eliminate outliers and subsequently follow Li (1985) and perform Huber iterations followed by biweight iterations. As can be seen, the results in columns (1) (4) of Table 5 continue to indicate that overall and political globalisation decrease the 16 We also test the sensitivity of our results to using alternate control variables (i.e., minimum wage regulations, corruption and price controls) that have been considered in the literature (see, e.g. Dreher & Schneider, 2010). The results are qualitatively similar across various specifications, thereby confirming our previous findings. Moreover, we created a cross-section of our data based on the unweighted average of each variable from 2000 to 2007 to mitigate problems associated with such things as measurement error. Our evidence suggests that it is political globalisation that continues to have a significant impact in mitigating the spread of the shadow economy. These results are available upon request.

BERDIEV AND SAUNORIS 233 TABLE 4 Robustness checks (dependent variable: shadow2) (1) (2) (3) (4) (5) (6) (7) (8) Overall globalisation 0.197*** 0.287*** 0.069** 0.066** (0.050) (0.056) (0.029) (0.028) Economic globalisation 0.015 0.013 (0.020) (0.020) Political globalisation 0.035*** 0.034** (0.013) (0.013) Social globalisation 0.022 0.015 (0.014) (0.016) Log GDP per capita 6.220*** 6.860*** 4.624*** 4.082*** 4.456*** 4.294*** 4.570*** 4.089*** (0.695) (0.673) (1.181) (1.465) (1.584) (1.331) (1.430) (1.453) Education 0.087*** 0.036 0.007 0.006 0.007 0.006 0.008 0.005 (0.028) (0.025) (0.005) (0.005) (0.006) (0.005) (0.005) (0.005) Credit market regulations 1.287*** 1.371*** 0.064 0.094 0.083 0.085 0.081 0.092 (0.353) (0.334) (0.079) (0.091) (0.092) (0.088) (0.093) (0.088) Labour market regulations 1.035*** 0.114 0.117* 0.115 0.119 0.137* 0.130 0.127* (0.239) (0.241) (0.069) (0.077) (0.074) (0.080) (0.082) (0.071) Government size 1.583*** 0.921*** 0.064 0.072 0.087 0.050 0.088 0.053 (0.263) (0.231) (0.091) (0.104) (0.106) (0.095) (0.104) (0.096) Constant 77.014*** 84.410*** 75.972*** 70.231*** 70.429*** 71.272*** 71.890*** 70.487*** (4.721) (5.233) (9.022) (12.448) (13.432) (11.832) (12.826) (12.698) Country effects No No Yes Yes Yes Yes Yes Yes Regional effects No Yes No No No No No No Time effects No No No Yes Yes Yes Yes Yes Number of observations 696 696 696 696 696 696 696 696 R 2 0.470 0.596 0.496 0.508 0.490 0.511 0.489 0.516 Notes: Robust standard errors in parenthesis. ***p <.01, **p <.05, *p <.10.

234 BERDIEV AND SAUNORIS TABLE 5 Robustness checks (dependent variable: shadow1) Robust regressions Two-step efficient GMM (1) (2) (3) (4) (5) (6) (7) (8) Overall globalisation 0.017** 0.054* (0.008) (0.030) Economic globalisation 0.0005 0.023 (0.005) (0.019) Political globalisation 0.009** 0.061*** (0.004) (0.023) Social globalisation 0.012 0.053 (0.007) (0.040) Log GDP per capita 6.422*** 6.655*** 6.422*** 6.654*** 5.357*** 5.481*** 5.196*** 6.231*** (0.282) (0.281) (0.274) (0.275) (0.731) (0.714) (0.750) (0.668) Education 0.005* 0.006** 0.006** 0.006** 0.006 0.006 0.004 0.009** (0.003) (0.003) (0.003) (0.003) (0.004) (0.005) (0.005) (0.004) Credit market regulations 0.075*** 0.093*** 0.096*** 0.088*** 0.052 0.059 0.057 0.080 (0.029) (0.029) (0.029) (0.029) (0.060) (0.060) (0.062) (0.060) Labour market regulations 0.137*** 0.134*** 0.141*** 0.133*** 0.214*** 0.218*** 0.192*** 0.203*** (0.028) (0.028) (0.028) (0.028) (0.051) (0.056) (0.051) (0.051) Government size 0.061* 0.070** 0.066** 0.071** 0.208*** 0.226*** 0.162*** 0.202*** (0.031) (0.032) (0.031) (0.031) (0.056) (0.055) (0.061) (0.053) Constant 83.831*** 99.850*** 92.945*** 100.713*** (1.543) (3.017) (2.738) (2.995) Number of observations 681 683 684 683 679 679 679 679 (Continues)

BERDIEV AND SAUNORIS 235 TABLE 5 (Continued) Robust regressions Two-step efficient GMM (1) (2) (3) (4) (5) (6) (7) (8) Kleibergen-Paap rk Wald F-statistic 51.30 29.58 15.69 9.251 Kleibergen-Paap rk LM statistic 51.60 39.93 17.64 11.41 [0.000] [0.000] [0.000] [0.003] First-stage F-statistic 51.30 29.58 15.69 9.25 [0.000] [0.000] [0.000] [0.000] Hansen s J statistic 0.283 0.00661 0.866 1.630 [0.595] [0.935] [0.352] [0.202] Notes: Columns (1) (4) report the estimates from the outlier test (robust regression), while columns (5) (8) display the two-step GMM estimates with robust standard errors in parentheses and probability values in brackets. Each globalisation measure is instrumented with its second and third temporal lag. The critical values for the Kleibergen-Paap rk Wald F-statistic are in Stock and Yogo (2005). ***p <.01, **p <.05, *p <.10.

236 BERDIEV AND SAUNORIS shadow economy, at the 5% level of significance. As expected, we also find that the coefficients on credit market regulations and education are now negative and significant at conventional levels across all specifications, suggesting that lower credit market regulations and a more educated populace reduce the size of the shadow economy. As before, the log of GDP per capita continues to exhibit a negative and significant coefficient, indicating that economic development lowers shadow activities. The results for the remaining independent variables are similar to our previous findings. 5.3 Endogeneity of globalisation Finally, it is important to address the possible endogeneity of globalisation. For example, endogeneity could arise from reverse causality such that countries with smaller shadow economies are conceivably more likely to implement policies favourable for globalisation. To account for this possible endogeneity of globalisation, we employ the general method of moments (GMM) approach. In the presence of unknown forms of heteroscedasticity, the GMM estimator is more efficient than the traditional two-stage least squares (2SLS) method; thus, we re-estimate Equation (1) using the two-step efficient feasible GMM. 17 Given the lack of readily available external instruments, we rely on internal instruments using the second and third lags of globalisation. 18 We provide the GMM regression estimates in columns (5) (8) of Table 5. The results continue to reinforce our main findings that overall and political globalisation reduce the shadow economy. Also, the coefficients on both overall and political globalisation are larger in absolute value compared to the OLS results, thus suggesting an upward bias in the OLS results. To verify the instruments are both valid and relevant (i.e., correlated with the endogenous variables and orthogonal to the errors), we utilise four diagnostic tests. First, we employ the weak identification test using the Kleibergen-Paap (2006) rk Wald F-statistic to test whether the instruments are only weakly correlated with the endogenous variables. Second, we use the underidentification test using the Kleibergen-Paap (2006) rk LM statistic to test whether the instruments are adequate to identify the equation. Third, we utilise the first-stage F-statistic to test whether the instruments are correlated with the endogenous variables. Lastly, we employ the overidentification test using the Hansen s J statistic to test whether the orthogonality conditions are valid. Rejection of the first three tests and failure to reject the Hansen s J tests is indicative of valid and relevant instruments (see, for details, Baum, Schaffer, & Stillman, 2003; Baum et al., 2007). The diagnostic tests, which are reported at the bottom of Table 5, verify that the instruments are relevant and valid in all equations. 5.4 Alternate control variables based on initial conditions To account for the possibility that the contemporaneous control variables are muting the effects of globalisation, we transform the control variables to their initial values. 6 Specifically, we estimate the impact of globalisation variables on the shadow economy by employing the control variables that are based on initial conditions. These results are reported in Table 6. Each model includes regional and year effects. As expected, these findings confirm our main results that overall globalisation reduces the size of the shadow economy (column (1)). Furthermore, political globalisation also continues to have a negative and statistically significant effect on the shadow economy 17 The two-step procedure is employed to make efficient GMM feasible. The first step involves estimating the variance covariance matrix assuming i.i.d errors, and the second step uses this estimate to calculate the optimal weighting matrix (Baum, Schaffer, & Stillman, 2007). 18 The results are robust to using alternate lag structures.

BERDIEV AND SAUNORIS 237 TABLE 6 Robustness checks (dependent variable: shadow1) (1) (2) (3) (4) (5) Overall globalisation 0.192*** (0.058) Economic globalisation 0.005 0.051 (0.036) (0.040) Political globalisation 0.072*** 0.063** (0.026) (0.026) Social globalisation 0.183*** 0.196*** (0.049) (0.053) Log GDP per capita (initial) 7.435*** 8.817*** 8.662*** 6.706*** 6.765*** (0.691) (0.679) (0.633) (0.779) (0.758) Education (initial) 0.084*** 0.073*** 0.082*** 0.076*** 0.082*** (0.021) (0.021) (0.021) (0.020) (0.021) Credit market regulations (initial) 0.167 0.338 0.447 0.179 0.411 (0.272) (0.276) (0.278) (0.282) (0.285) Labour market regulations (initial) 0.116 0.117 0.059 0.235 0.031 (0.236) (0.234) (0.254) (0.226) (0.241) Government size (initial) 1.245*** 1.414*** 1.305*** 1.238*** 1.150*** (0.266) (0.267) (0.278) (0.256) (0.266) Constant 88.868*** 89.880*** 96.299*** 80.079*** 86.179*** (5.781) (6.055) (6.627) (6.064) (6.902) Country effects No No No No No Regional effects Yes Yes Yes Yes Yes Time effects Yes Yes Yes Yes Yes Number of observations 685 685 685 685 685 R 2 0.610 0.602 0.608 0.612 0.618 Notes: Robust standard errors in parenthesis. ***p <.01, **p <.05, *p <.10. (columns (3) and (5)); however, social globalisation now has a negative and statistically significant effect on the shadow economy (columns (4) and (5)). In summary, the robustness tests confirm our main results that globalisation successfully mitigates the development of the shadow economy and this is achieved primarily through political integration and interdependence. 6 CONCLUSION Many studies have been devoted to better understanding the effects and significance of globalisation. We contribute to this important literature by examining the impact of globalisation on the shadow economy. The prevalence of the shadow economy along with the general tendency towards globalisation among many countries makes the interaction between these two phenomena

238 BERDIEV AND SAUNORIS an interesting topic. Using panel data for 119 countries, the results suggest that globalisation is a useful tool in combating shadow activities. Given the multidimensionality of globalisation, we use a disaggregate measure of globalisation (i.e., economic, political and social) to find that political globalisation proves to be the main driver in mitigating shadow development after controlling for important factors that affect the size of the shadow economy. However, we stress that although political globalisation is robustly statistically significant there remains considerable overlap among the three disaggregated measures of globalisation. Moreover, economic globalisation significantly reduces the size of the shadow economy; however, once we account for other important factors these effects are muted. Therefore, we caution the reader to not view the disaggregated measures of globalisation as mutually exclusive. Countries considered to be more politically globalised are more likely to be involved in international treaties, participate in UN Security Council meetings, be a member in International Organization (e.g. IMF, NATO), and house embassies (see, for details, Dreher, 2006). Conceivably, these endeavours provide a number of benefits that help prevent the spread of the shadow economy. First, they enhance the credibility of institutions by promoting policy diffusion and by enabling countries to learn effective policies adopted by other countries. Second, these help constrain corrupt leaders that would otherwise encourage individuals to migrate underground to avoid, for example, having to pay bribes (see Buehn and Schneider (2012) for the relation between corruption and shadow economy). Finally, they promote official sector production that raises the opportunity cost of producing in the shadow sector. Overall, our findings demonstrate the beneficial effects of globalisation for reducing the spread of shadow operations. Thus, policies aimed at opening borders, especially through political integration, would reduce incentives to participate in the shadow economy. Although this study is by no means the last word on this topic, it opens up new avenues of research. For instance, a deeper understanding of the channels through which globalisation, especially economic and social, impact the shadow economy is warranted. We will continue to investigate the linkages between globalisation and the shadow economy in our future work. ACKNOWLEDGEMENTS We thank the editor, an anonymous referee and participants at the 2015 American Public Choice conference in San Antonio and the 2015 European Public Choice conference in Groningen for their valuable suggestions. All remaining errors are our own. ORCID James W. Saunoris http://orcid.org/0000-0002-6304-3070 REFERENCES Allingham, M. G., & Sandmo, A. (1972). Income tax evasion: A theoretical analysis. Journal of Public Economics, 1, 323 338. Asea, P. K. (1996). The informal sector: Baby or bath water? A comment. Carnegie-Rochester Conference Series on Public Policy, 45, 163 171. Baum, C. F., Schaffer, M. E., & Stillman, S. (2003). Instrumental variables and generalized method of moments: Estimation and testing. The Stata Journal, 3, 1 31. Baum, C. F., Schaffer, M. E., & Stillman, S. (2007). Enhanced routines for instrumental variables/gmm estimation and testing. The Stata Journal, 7, 465 506.

BERDIEV AND SAUNORIS 239 Berdiev, A. N., Pasquesi-Hill, C., & Saunoris, J. W. (2015). Exploring the dynamics of the shadow economy across US states. Applied Economics, 47, 6136 6147. Bergh, A., & Nilsson, T. (2010). Good for living? On the relationship between globalization and life expectancy. World Development, 38, 1191 1203. Bergh, A., & Nilsson, T. (2014). Is globalization reducing absolute poverty? World Development, 62, 42 61. Bhagwati, J., & Hansen, B. (1973). A theoretical analysis of smuggling. The Quarterly Journal of Economics, 87, 172 187. Bonaglia, F., De Macedo, J. B., & Bussolo, M. (2001). How globalisation improves governance (Center for Economic Policy Research (CERP) Discussion Paper No. 2992). Retrieved from https://papers.ssrn.com/sol3/pape rs.cfm?abstract_id=288354 Buehn, A., & Farzanegan, M. R. (2012). Smuggling around the world: Evidence from a structural equation model. Applied Economics, 44, 3047 3064. Buehn, A., & Farzanegan, M. R. (2013). Impact of education on the shadow economy: Institutions matter. Economics Bulletin, 33, 2052 2063. Buehn, A., & Schneider, F. (2012). Shadow economies around the world: Novel insights, accepted knowledge, and new estimates. International Tax and Public Finance, 19, 139 171. De Soto, H. (1989). The other path. New York: Harper & Row. Dell Anno, R., & Solomon, O. H. (2008). Shadow economy and unemployment rate in USA: Is there a structural relationship? An empirical analysis. Applied Economics, 40, 2537 2555. Dreher, A. (2006). Does globalization affect growth? Evidence from a new index of globalization. Applied Economics, 38, 1091 1110. Dreher, A., Gaston, N., & Martens, P. (2008). Measuring globalization: Gauging its consequences. New York: Springer Science & Business Media. Dreher, A., Kotsogiannis, C., & McCorriston, S. (2009). How do institutions affect corruption and the shadow economy? International Tax and Public Finance, 16, 773 796. Dreher, A., & Schneider, F. (2010). Corruption and the shadow economy: An empirical analysis. Public Choice, 144, 215 238. Dreher, A., & Voigt, S. (2011). Does membership in international organizations increase governments credibility? Testing the effects of delegating powers. Journal of Comparative Economics, 39, 326 348. Elgin, C. (2013). Internet usage and the shadow economy: Evidence from panel data. Economic Systems, 37, 111 121. Elgin, C., & Oztunali, O. (2012). Shadow economies around the world: Model based estimates (Bogazici University). Retrieved from http://www.econ.boun.edu.tr/public_html/repec/pdf/201205.pdf Friedman, E., Johnson, S., Kaufmann, D., & Zoido-Lobaton, P. (2000). Dodging the grabbing hand: The determinants of unofficial activity in 69 countries. Journal of Public Economics, 76, 459 493. Fugazza, M., & Fiess, N. (2010). Trade liberalization and informality: New stylized facts. United Nations Conference on Trade and Development (Policy Issues In International Trade And Commodities Study Series No. 43). Retrieved from https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1596030 Gassebner, M., Lamla, M. J., & Vreeland, J. R. (2013). Extreme bounds of democracy. Journal of Conflict Resolution, 57, 171 197. Gaston, N., & Rajaguru, G. (2013). International migration and the welfare state revisited. European Journal of Political Economy, 29, 90 101. G erxhani, K. (2004). The informal sector in developed and less developed countries: A literature survey. Public Choice, 120, 267 300. G erxhani, K., & van de Werfhorst, H. G. (2013). The effect of education on informal sector participation in a postcommunist country. European Sociological Review, 29, 464 476. Ghemawat, P., & Altman, S. A. (2014). DHL Global Connectedness Index 2014. Bonn: Deutsche Post DHL. Goel, R. K., Nelson, M. A., & Naretta, M. A. (2012). The internet as an indicator of corruption awareness. European Journal of Political Economy, 28, 64 75. Goel, R. K., & Saunoris, J. W. (2014). Global corruption and the shadow economy: Spatial aspects. Public Choice, 161, 119 139. Gwartney, J., & Lawson, R. with Grubel, H., de Haan, J., Sturm, J.-E., & Zandberg, E. (2009). Economic freedom of the world: 2009 annual report. Vancouver, BC: The Fraser Institute.