Tax me if you can!
The Role of Taxation in the
Local and Global Shadow Economy
by
Christina Lueck
S2439972
Thesis Submitted for the Degree of
Master of Science in International Economics and Business and
Master of Science in International Financial Management
in the
Faculty of Economics and Business
UNIVERSITY OF GRONINGEN
Semester 2
June 2015
Supervision by Dr. Hein Vrolijk
Co-assessment by Prof. Dr. Niels Hermes
Abstract
Table of Contents
Abstract ... ii
Table of Contents ... iii
List of Tables ... iv
List of Figures... iv
List of Acronyms ... v
1. Introduction ... 1
2. Theoretical and Empirical Considerations ... 3
2.1. Defining the Shadow Economy ... 3
2.2. Previous Studies ... 5
3. Data and Methodology ... 14
3.1. Dependent Variable ... 14
3.2. Independent and Control Variables ... 14
3.3. Sample ... 17
3.4. Descriptive Statistics ... 18
3.5. Estimation Equation ... 20
4. Results and Discussion ... 21
4.1. Results ... 21
4.2. Discussion ... 32
5. The Role of MNEs in the “Global Shadow Economy” ... 35
5.1. The Multinational Enterprise and the Global Shadow Economy ... 35
5.2. MNEs Tax Avoidance Activities ... 36
5.2.1. Thin Capitalization ... 36
5.2.2. Trade Mispricing ... 37
5.2.3. Tax Haven Usage ... 38
6. Linking the Global and the Local Shadow Economy: The Effect of Multinational Activity on the Size of the Shadow Economy... 39
7. Conclusion ... 41
References ... 43
Appendix A. Estimation Methods for the Size of the Shadow Economy ... 46
Appendix B. Variables Used in Previous Studies... 47
Appendix C. List of Countries Included In The Analysis Sorted By Income Class ... 48
Appendix D. 2SLS Estimation Results ... 49
Appendix E. Comparison of the Tax Burden Coefficients Across Sub-samples ... 51
Appendix F. Empirical Evidence of Illicit Financial Flows ... 52
Appendix G. List of Tax Havens and Secrecy Jurisdictions ... 56
Appendix H. Empirical Evidence of Tax Haven Usage ... 58
List of Tables
Table 2.1. Summary of Previous Studies ... 9 Table 2.2. Expectations for the Differential Effect of Tax Burden on the
Shadow Economy ... 12 Table 3.1. Descriptive Statistics ... 18 Table 3.2. Correlations of Institutional Variables ... 19 Table 4.1. OLS Estimation Results for the Size of the Shadow Economy as
Measured by Schneider et al. (2010) ... 22 Table 4.2. OLS Estimation Results for the Size of the Shadow Economy as
Measured by Schneider et al. (2010) ... 24 Table 4.3. OLS Estimation Results with Interaction Term ... 29 Table 4.4. Summary of the Impact of Tax Burden on the Shadow Economy
Size Across Sub-Samples and Compared to the Main Sample ... 31 Table 4.5. Expectations and Findings for the Effect of Tax Burden on the
Size of the Shadow Economy ... 32
List of Figures
Figure 2.1. The Shadow Economy, Illicit and Licit Real and Financial Markets ... 4 Figure 2.2. Composition of the Overall Effect of Tax Burden on the Size of the
Shadow Economy ... 11 Figure 4.1. Relationship of Tax Burden and the Shadow Economy Size per
List of Acronyms
GFI Global Financial Integrity
HMO Hot Money Outflow
IFF Illicit Financial Flow
IMF International Monetary Fund MNE Multinational Enterprise
1. Introduction
Taxation is one of the main contributions to a country’s national budget. Tax avoidance and evasion may thus seriously threaten income redistribution and the provision of public goods, especially in developing countries. The phenomenon that comprises such hidden economic activities on a national level is: the shadow economy. To effectively tackle the shadow economy, governments and policy makers strive to answer: What drives entrepreneurs underground?
Due to the fact that economic activities in the shadow economy are hidden, estimating the size of the shadow economy and investigating its determinants can be described as a “scientific passion for knowing the unknown” (Schneider, 2005). Previous studies, which dealt with potential determinants of the shadow economy found that the quality of institutions is significantly and negatively associated with the size of the shadow economy. However, when it comes to the effect of tax burden the results are mixed: some scholars find a negative relation while others find a positive relation to shadow economy size.
The aim of this study is (1) to investigate the effect of a country’s tax burden, quality of public institutions and level of economic development on its size of the shadow economy, (2) to prove a differential effect of tax burden on shadow economy size depending on the country’s level of economic development and (3) to shed light on the international dimension of the shadow economy and its main actor – the multinational enterprise (MNE).
of the shadow economy, taxation is considered on a national scale. I contribute to the literature by not only investigating the national dimension of the shadow economy but also international dimension and thus tax avoidance activities on a global scale.
My analysis reveals evidence that higher tax burden is associated with a larger shadow economy. I find this effect to be stronger for developing countries than for developed ones. Furthermore, higher quality public institutions are associated with a smaller shadow economy. A stronger legal environment, more effective bureaucracy and lower levels of corruption can be related to a smaller shadow economy. Taking a look at the “global shadow economy” and thereby at the tax avoidance activities of MNEs, I can raise suspicion of a potential relationship between multinational activity and the size of the national shadow economy.
2. Theoretical and Empirical Considerations
This chapter gives an introduction to the relationship between the shadow economy and taxation. At first, the overall framework of the shadow economy is described. I will explain the construction of the shadow economy, its actors and their activities as well as the composition of licit and illicit flows of goods and money as well as the role of taxation. In the second step, previous studies on the relationship between the shadow economy and taxation are reviewed, with particular focus on their theoretical view and empirical findings. At the end of this chapter I will develop expectations for my empirical analysis.
2.1. Defining the Shadow Economy
The “shadow” or “unofficial” economy is difficult to characterize, as the occurrence of numerous different definitions have shown. The broadest definition of the shadow economy goes back to Smith (1984) who states that the shadow economy includes all market-based legal and illegal goods and services that escape inclusion in official accounts. Various researchers, including Alm and Embaye (2013), Feige (1989) and Dell’Anno and Schneider (2003) have followed Smith’s definition.1
Fig. 2.1. describes the broad definition of the shadow economy based on Smiths’ view. The shadow economy can be divided into the shadow real and the shadow financial economy. Both, the shadow real and the shadow financial economy are home for licit and illicit activities.
1
Others have focused their definitions of the shadow economy on the proceeds of the production and trade of only legal goods and services. For example Alm et al. (2004) find “the shadow economy could be defined as all market based but unreported income from the production of legal goods and services, either from monetary or barter transactions that would normally be taxable if they were reported to the tax authorities”. A somewhat narrower definition of the shadow economy is based on Schneider (Schneider 2005, Schneider et al. 2010 and
The real economy part represents the productive economy, which captures the goods and services sector. The licit real economy includes the production and trade of goods and services in formal commercial and non-commercial sectors that give rise to licit proceeds. The illicit economy covers the production and trade of goods and services from criminal activities (such as human trafficking, drug and illegal arms trade, counterfeits and smuggling), corruption and bribery and commercial activities (mainly tax evasion activities such as trade mispricing) that give rise to illicit proceeds. Both licit and illicit proceeds from activities in the real market flow into the respective licit and illicit markets of the financial economy.
Figure 2.1. The Shadow Economy, Illicit and Licit Real and Financial Markets
The shadow financial economy represents the financial sector of an economy that comprises the proceeds of financial flows from the production and trade of goods and services. The licit proceeds do not necessarily stay in the licit financial market nor do the illicit proceeds stay in the illicit financial markets: illicit financial flows can be observed between both licit and illicit financial markets. Illicit proceeds can flow to the licit market due to activities such as money laundering, which is often practiced through commercial and real estate purchases. Licit proceeds may flow to the illicit financial market mainly to avoid and evade taxation.
Thinking outside the box and across the borders of the national economy, the illicit financial market gives rise to illicit cross-border flows. These illicit flows may stem from (1) illicit proceeds of the illegal production and trade of goods and services through criminal, corrupt or certain commercial activities and (2) licit proceeds of the legal production of goods and services in the formal commercial and noncommercial sectors. Illicit cross-border flows and the issue of tax avoidance by multinational enterprises is further discussed in chapter 5.
Summing up I can conclude that the shadow economy is comprised of all market based activities and proceeds arising from the production and trade of illegal and legal goods and services that escape inclusion from official accounts, mainly as to avoid taxation.
2.2. Previous Studies
In this section I will review previous studies on the relation of taxation and the shadow economy and draw my own expectations for the empirical analysis.
discretion in the administration of rules leads to a higher effective burden on firms, more corruption and a greater incentive to move to the shadow economy.
Based on their previous work, Johnson et al. (1998) propose that the shadow economy is larger when the tax burden on firms is higher, with tax burden capturing the outcome of how the tax system is administered as well as the tax rates. Furthermore, a larger shadow economy should be associated with weaker public institutions. The quality of public institutions is proxied by two indices measuring corruption and the efficiency of the legal system. Johnson et al. (1998) find support for their propositions by using pooled 1990s cross-country data on the shadow economy in Latin America, various OECD countries and the former Soviet Bloc.
Nagac (2015) wrote the most recently published study investigating determinants of the shadow economy. Nagac analyzes the impact of a set of non-rate qualitative aspects of the tax system as well as the impact of tax burden, the quality of the legal system and labor market regulations on the size of the shadow economy. The panel dataset comprises data from 44 countries over a period from 2004 to 2005. Nagac uses data on the size of the shadow economy from two different sources, namely Schneider et al. (2010) who estimates the size of the shadow economy by using the MIMIC method and Alm and Embaye (2013) who use the currency demand method2. After estimating pooled OLS regressions and taking into account the potential endogeneity of tax burden, Nagac finds that tax burden positively affects the size of the shadow economy. The quality of the legal system and the complexity of the tax system negatively affect the size of the shadow economy, while the effect of labor market regulations is positive.
A further study that shows a positive relation of tax burden and the size of the shadow economy was conducted by Schneider (2005). By investigating a panel dataset on 110 countries and spanning the 1990s, Schneider finds that an increasing tax burden is a driving force for the size and growth of the shadow economy.
As shown above, empirical studies have produced mixed results about the relationship of tax and the size of the shadow economy - even the sign of the coefficients differ across the studies.
Lee (2005) claims that these mixed results of previous studies can be reconciled if one introduces the possibility of differential effects of taxation on the size of the shadow economy. Lee bases this claim on Friedman et al. (2000) which found countervailing effects between tax and the size of the shadow economy. Theoretically Lee separates the impact of taxation on the shadow economy into two effects, namely the direct “incentive effect” and the indirect “institution effect”. The incentive effect describes the direct effect of taxation on the size of the shadow economy. Higher taxation increases the incentive of taxpayers to hide their activities in order to reduce their tax liabilities. Subsequently this will lead to more engagement in the shadow economy. The institution
effect on the other hand hypothesizes an indirect effect of taxation on the size of the shadow economy due to the role of public institutions. Public institutions are important determinants of business performance. A well-functioning legal system that enables companies to enforce contracts, improved education which leads to a higher skilled workforce or a better infrastructure that enables the transport of goods on rails and roads are only some examples how public institutions improve successful business performance. Higher tax rates lead to higher tax revenues that enable governments to enhance the productivity of public institutions and infrastructure. This increases the opportunity costs for taxpayers to engage in unofficial activities and subsequently leads to a smaller shadow economy.
With the development of public institutions and infrastructure, Lee expects that the effect of institutions decrease as the marginal cost of developing institutions increases. This means under further development of a given country, the indirect institution effect is gradually outweighed by the direct incentive effect. In other words, Lee expects a negative relation of taxation on the shadow economy for developing countries and a neutral or positive relation for developed countries.
Lee empirically investigates these theoretical predictions by using cross-sectional and time-variant data. His dataset covers 62 countries, covering the years between 1989 and 1998. Data on the shadow economy is taken from Friedman et al. (2000) and Schneider and Enste (2000) and relies upon three different estimation methods: the currency demand approach, the physical input approach and the multiple indicators multiple causes (MIMIC) approach.3 Due to the inclusion of estimates from different estimation techniques, Lee finds systematic differences in the estimates – averages of the size of the shadow economy vary greatly. As an independent tax variable, Lee uses a country’s tax burden, which is a ratio of tax revenues to GDP. To analyze the differential effect of taxation on the size of the economy, Lee later constructs an interaction term between tax burden and the logarithm of GDP per capita.
Lee’s findings support his expectations of differential effects of taxation on the size of the shadow economy. The relationship between tax and the size of the shadow economy is negative in developing countries and becomes less negative the more developed a country is. In other words, higher taxes lead to a smaller shadow economy in developing countries, but this effect is getting weaker and less significant the more developed a country is. The results are evidence for Friedman et al.’s and Lee’s expectations that the institution effect of taxation dominates the incentive effect in developing countries and that the institution effect is offset by the incentive effect in developed countries. Table 2.1. summarizes the aforementioned studies.
Table 2.1. Summary of Previous Studies Study Countries Period Estimation
Method Dependent Variable: Shadow Economy Estimation Method Independent Variable: Main Tax Variable Finding Johnson et al. (1998) 49 Cross-country data: 1990s
OLS Mixed methods Tax Burden +
Friedman et al. (2000)
69 Cross-country data: 1990s
OLS Mixed: Currency demand, physical input and MIMIC method Tax Rates Tax Burden - - Nagac (2015) 44 Cross-country data: 2004 and 2005
OLS MIMIC method Currency demand method
Tax Burden +
Schneider (2005) 104 Panel data: 1990s
OLS and Random Effects
MIMIC method Share of direct taxation (in % of GDP) Share of indirect taxation and customs duties (in % of GDP) + + Lee (2005) 78 Unbalanced panel data: 1960 and 1990
OLS Mixed: Currency demand, physical input and MIMIC method Tax Burden Tax Burden *Log (GDP per capita) - + My study 80 Cross-country data: 2002-2006
OLS MIMIC method Currency demand method
Tax Burden ?
Similar to previous studies, I will investigate determinates of the shadow economy by using ordinary least squares (OLS) estimations. The cross-sectional dataset under investigation consists of 80 countries and comprises average values of data from the years 2002 to 2006. The dependent variable is the size of the shadow economy. By following Nagac (2015), I use data on the size of the shadow economy from two different sources; namely Schneider et al. (2010), who estimate the shadow economy with the MIMIC method and Alm and Embaye (2013), who are using the currency demand method (an explanation of the estimation techniques can be found in Appendix A). These two different estimates are used in order to control for the influence of dissimilar estimation techniques. As a tax determinant of the shadow economy I have chosen tax burden, whereas institutional quality will be proxied by three indices, namely rule of law, governmental effectiveness and corruption perception. The use of dependent and independent variables is further explained in chapter 3.
All aforementioned previous studies on the shadow economy suggest a direct negative effect of the quality of public institutions on the size of the shadow economy. Citizens’ identification with the state, and thus their willingness to pay taxes increase, if they perceive that their interests are adequately represented by political institutions and public services. Weak legal systems, ineffective public services and high levels of corruption have been found to undermine the willingness to stay in the formal economy and pay tax (Torgler & Schneider 2009). Thus, I expect that the institutional quality of a country is negatively associated with the size of the shadow economy. In other words, a lower level of institutional quality, ceteris paribus, raises the size of the shadow economy.
Figure 2.2. Composition of the Overall Effect of Tax Burden on the Size of the Shadow Economy
As displayed in Fig. 2.2., the overall effect of taxation on the shadow economy depends on two “sub-effects” with opposing forces: the incentive and the institution effect. Positive incentive effects are expected to arise because a higher tax burden will increase the incentive of citizens to hide their taxable activities and rather engage in the unofficial shadow economy than in the official economy. This naturally leads to an increase in the size of the shadow economy. Then again the overall effect may also depend on a indirect component: the institution effect. Higher taxation goes along with higher tax revenues that allow governments to further develop their public institutions. If citizens believe they are better protected by the rules of law, their public institutions are more effective and less corrupt, the opportunity costs for the tax payer to engage in the shadow economy subsequently increase, naturally leading to a decrease of the shadow economy.
The overall effect of tax burden on the size of the shadow economy thus hinges on the interplay of the incentive and the institution effect. In case the incentive effect dominates the institution effect, we can expect the overall impact of tax burden on the size of the shadow economy to be positive. If the opposite is the case, meaning that the institution
Incentive Effect
Higher tax burden directly leads to a larger shadow economy: higher tax burden increases the incentive for tax payers to hide taxable income.
Institution Effect
Higher tax burden indirectly leads to a smaller shadow economy: higher tax burden leads to more tax revenues that enable governments to enhance public institutions, thereby increasing the opportunity costs for citizens to engage in the shadow economy.
Larger shadow economy
effect dominates the incentive effect, we expect the overall impact of tax burden on the shadow economy to be negative. If the institution effect and the incentive effect balance out, we can assume tax burden to have a neutral effect on the shadow economy.
Like Lee (2005) I believe that tax burden has a differential effect on size of the shadow economy depending on the level of a country’s development (see Table 2.2.).
Table 2.2. Expectations for the Differential Effect of Tax Burden on the Shadow Economy
Relationship Explanation Developing countries Negative
Neutral
The institution effect dominates the incentive effect.
The incentive effect and the institution effect outweigh each other.
Developed countries Neutral Positive
The incentive effect and the institution effect outweigh each other. The incentive effect dominates the institution effect.
For developing countries I expect the negative institution effect to dominate (or at least outweigh) the positive incentive effects of taxation, leading to a negative (neutral) relation of tax burden and the size of the shadow economy. This means a higher level of tax burden in developing countries, ceteris paribus, decreases (does not affect) the size of the shadow economy.
With further development of public institutions and infrastructure in a given country, it can be expected that the effect of institutions decrease as the marginal cost of developing institutions increases. This gives room for the incentive effect to outweigh the institution effect.
Thus, for developed countries the relationship is expected to be the opposite: the direct incentive effect dominates (or at least outweighs) the indirect institution effect, leading to an overall positive (neutral) relation of tax burden and the size of the shadow economy. In other words higher levels of tax burden, ceteris paribus, increases (does not affect) the size of the shadow economy.
income levels, I can more adequately generalize my results across countries all over the world. Secondly, the data covers the period 2002 to 2006. This is the most recent data available on the size of the shadow economy - most of the previous studies have dealt with data from the 1990s. Third, I take two different estimates of the shadow economy for all countries into account. Most previous studies constructed their datasets by mixing estimates from different estimation techniques. By using two different estimates of the shadow economy for each country, I can control for particular influences of the estimation technique on the results. Fourth, I thoroughly analyze the interaction between tax burden and GDP per capita. With this analysis I test, whether the effect of tax burden on the size of the shadow economy depends on the level of economic development of a country. To the best of my knowledge this is the second study, besides Lee (2005), that examines the differential effect of taxation on the size of the shadow economy; but then again the first study that uses consistent data on the size of the shadow economy from two different estimation techniques.
3. Data and Methodology
This chapter firstly describes the dependent variable – the size of the shadow economy. The second section introduces the independent tax and institutional variable, followed by section 4 that gives information about the main and sub-samples, namely the countries and time period under investigation. I round this chapter off by presenting the descriptive statistics and the estimation equation.
3.1. Dependent Variable
Shadow EconomyThe shadow economy is measured in percent of GDP. Data is taken from Schneider et al. (2010) and Alm and Embaye (2013). With the help of indirect estimation techniques and following the MIMIC and the currency demand approach, Schneider et al. (2010) cover estimates of the shadow economy of 145 countries between 1999 and 2007. Up to date, this is the most complete dataset available on the size of the shadow economy. Alm and Embaye (2013) use a different estimation technique, namely the currency demand method. Their data set comprises estimates of the size of the shadow economy of 111 countries for the years 1984 to 2006. Further information about the most used techniques to estimate shadow economy size can be found in Appendix A.
3.2. Independent and Control Variables
Tax BurdenLevel of Economic Development
The level of economic development of the country is captured by real GDP per capita, measured in constant 2005 USD. Values are extracted from the World Bank’s World Development Indicators Database. As economic development improves, the quality of institutions should improve as well; therefore GDP per capita can also be seen as representing the quality of institutions in a country. Countries with a higher level of development have a greater capacity to collect taxes, and have a higher demand for income elastic public goods and services (Chelliah 1971 and Bahl 1971 as cited by Torgler & Schneider, 2009). We therefore expect a negative relation between the level of per capita income and the size of the shadow economy.
Similar to Lee (2005), I furthermore include an interaction term of tax burden and GDP per capita to the model to look for a potential impact of the level of economic development on the relationship of tax and the size of the shadow economy. The introduction of the interaction term allows me to test if better institutions and infrastructure change the effect of tax burden on the shadow economy.
Institutional Quality
The direct effect of institutional quality on the size of the shadow economy is estimated by using three different variables – the “Rule of Law Index”, the “Governmental Effectiveness Index” and the “Corruption Perception Index”.
The level of corruption in a country is proxied by Transparency International’s Corruption Perception Index (CPI). Due to the fact that corruption is comprised of illegal activities, it is not possible to adequately measure absolute levels of corruption. Transparency International therefore measures perceived relative levels of corruption in the public sector. CPI is an aggregate index, drawing on results of business surveys and expert assessments of corruption by various independent institutions that specialize in governance analysis. Based on this data, countries are scored from 0 (highly corrupt) to 10 (very clean).
World Bank’s Worldwide Governance Indicators Database. To investigate a potential impact of the countries legal environment on the size of the shadow economy the index “Rule of Law” is used. This index measures the extent to which citizens have confidence in and compliance with the country’s law. Included are perceptions of the fairness, effectiveness and enforceability of legal procedures. The country’s score may range from -2.5 (low confidence in the legal system) to 2.5 (high confidence in the legal system). A further key explanatory variable for institutional quality is the index “Government Effectiveness”. Government Effectiveness captures the perceived quality of bureaucracy and public institutions, such as public and civil services, the quality of national policies and the government’s commitment to such policies. A country’s score may range from -2.5 (low government effectiveness) to 2.5 (high government effectiveness).
All three indicators are widely used in other empirical studies and are proven to have significant statistical influence on illicit economic activities (Torgler & Schneider, 2009; Kaufman et al., 2004). Lee (2005) is using indicator variables for the country’s level of corruption and quality of institutions from the International Country Risk Guide, since only these indicators provided sufficient cross-section and time-series coverage in the 1980s and 1990s. Due to the fact that data on these two indicators is not freely available, I have chosen the aforementioned indicators – CPI, Government Effectiveness and Rule of Law - instead.
Unemployment Rates
3.3. Sample
Full data on the size of the shadow economy from both sources (Schneider et al., 2010; Alm & Embaye, 2013) is available for the period of 1999 to 2006 across 107 countries. The independent variables GDP per capita and tax burden do cover the years 1999 to 2006, however with some missing values across the years. In order to achieve a sample with no missing values for the three main variables - namely the size of the shadow economy, the tax ratio and GDP per capita - I first constructed a panel dataset for 80 countries covering the period 2002 to 2006. Because yearly changes are almost nonexistent I then took the average values, ending up with a cross-sectional dataset.
3.4. Descriptive Statistics
Table 3.1. presents the descriptive statistics of all variables.
Table 3.1. Descriptive Statistics
Variables N Mean SD Min Max
Dependent Variables
Shadow Economy (Schneider et al., 2010) 80 31.21 13.18 8.56 65.88 Shadow Economy (Alm & Embaye, 2013) 80 29.08 9.64 11.70 53.10
Independent Variables GDP per capita 80 13,147.00 17,476.00 153.30 77,892.00 Tax Burden 80 17.04 8.17 1.86 60.05 Rule of Law 80 0.23 1.03 -1.66 1.94 Government Effectiveness 80 0.33 1.03 -1.61 2.20 Corruption 80 4.66 2.40 1.14 9.56 Unemployment 80 8.02 4.22 1.46 25.08
well as at countries in which the citizens’ confidence in the legal system is very high. Similar minimum and maximum values are found for the Government Effectiveness Index. The country with the weakest legal system and the lowest government effectiveness is the Democratic Republic of the Congo, scoring -1.66 and -1.61 respectively. The country with the strongest legal system (1.94) and highest governmental effectiveness (2.20) is Denmark. Moreover, the dataset comprises countries with corruption levels between 1.14 (high level of perceived corruption) and 9.56 (low level of perceived corruption). Bangladesh is the country with the highest perceived level of corruption, scoring 1.14. The cleanest country in terms of corruption is Iceland with a score of 9.56, showing almost no evidence of perceived corruption at all. The mean average unemployment rate across the main sample is 8% with large variations between 1.46% and 25.08%.
The results from the descriptive statistics let me suspect a high correlation between the institutional variables rule of law and government effectiveness. For both variables we found the almost similar minimum and maximum values registered for the same two countries, the Democratic Republic of the Congo and Denmark respectively. By cross checking the pairwise correlations of all variables, we find high correlations between the three institutional variables rule of law, government effectiveness and corruption (see Table 3.2.). Their correlation is very high with values larger than 0.9. I therefore decided to include the institutional variables separately from each other in different models.
Table 3.2. Correlations of Institutional Variables
Rule of Law Government Effectiveness Corruption Rule of Law 1.00
Government Effectiveness 0.97 1.00
3.5. Estimation Equation
The following equation will be estimated:�= � + � � ��+ � �+ � �+ � �
+ � � + � �+ ��
where SHADOWi is the size of the shadow economy, �� �� is the logarithm of GDP per capita,
� � is the tax ratio, � is the Rule of Law Index,
� � is the Government Effectiveness Index, �� � � is the Corruption Perception Index,
� is the unemployment rate and �� is the error term.
4. Results and Discussion
In the first section of this chapter I will present the estimation results. In the second section I will then discuss findings about the relationship of tax burden and the size of the shadow economy, taking into consideration the theoretical and empirical considerations of chapter 2.
4.1. Results
Main SampleTable 4.1. presents the first results of OLS regressions. In this first step, the estimations are based on the size of the shadow economy as measured by Schneider et al. (2010). The main variables across the 5 different models are the logarithm of GDP per capita and tax burden. I gradually include the institutional variables and unemployment to the model.
Table 4.1. OLS Estimation Results for the Size of the Shadow Economy as Measured by Schneider et al. (2010)
Dependent Variable: Shadow Economy by Schneider et al. (2010)
(1) (2) (3) (4) (5) Log of GDPpc -6.012*** (0.553) -2.092** (0.868) -1.704** (0.842) -0.983 (1.121) -2.112** (0.840) Tax Burden 0.0878 (0.0920) 0.177** (0.0731) 0.221*** (0.0745) 0.151* (0.0882) 0.171** (0.0827) Rule of Law -7.850*** (1.441) -8.912*** (1.399) Government Effectiveness -9.400*** (1.946) Corruption -3.364*** (0.625) Unemployment -0.512*** (0.177) -0.453** (0.191) -0.545*** (0.206) Constant 80.21*** (4.905) 47.54*** (6.857) 47.88*** (6.859) 43.65*** (8.683) 66.08*** (5.006) Observations 80 80 80 80 80 R-squared 0.536 0.637 0.662 0.640 0.627 Notes. T-statistics in parentheses. Significance levels: *** p < 0.01, ** 0.01 < p < 0.05, * 0.05 < p < 0.10. Regressions with robust standard errors. Beta coefficients reported.
The empirical results in Table 4.1. suggest strongly that tax burden plays a significant role in determining the size of the shadow economy. The coefficient of tax burden is positive and statistically significant across all estimations. That said higher taxation is associated with a larger shadow economy.
GDP per capita significantly affects the size of the shadow economy across almost all estimations. The coefficient is negative throughout all models. GDP per capita is a proxy for the country’s level of development. This result suggests that a higher GDP per capita is associated with a smaller shadow economy, i.e. weakly developed countries have a larger unofficial economy. This goes along with the expectations.
statistically significant. The large coefficient shows the strong effect of law and order on the size of the shadow economy. The result suggests that a more efficient legal system goes along with a smaller shadow economy. Adding unemployment as a control variable in model 3 does not change the signs and significance levels of the remaining independent variables. The R² is higher and the coefficient of unemployment highly significant, I therefore decide to keep it as an adequate control variable throughout our upcoming models. The positive coefficient of unemployment tells us that countries with a larger share of unemployed citizens, ceteris paribus, have a larger shadow economy. This goes along with my expectations. In models 4 and 5 government effectiveness and corruption are separately added to the model. I remove the variable rule of law as it is highly correlated with the other independent variables measuring the institutional quality - government effectiveness and the corruption level. Government effectiveness and corruption both show highly negative and significant coefficients. A negative coefficient of governmental effectiveness means that countries with better bureaucracy and public institutions, ceteris paribus, have a smaller shadow economy. We need to be cautious in interpreting the coefficient for corruption: A negative coefficient of corruption does not mean that countries with higher levels of corruption have a small shadow economy - the opposite is the case. The corruption index ranges from 0 to 10 with 0 standing for a high level of corruption and 10 standing for a clean corruption level. Therefore the correct interpretation is that countries with a high level of corruption, ceteris paribus, have a large shadow economy. The results of all three institutional variables support my expectations. I can conclude that a higher level of institutional quality is correlated with a smaller shadow economy, holding all other variables constant.
Table 4.2. OLS Estimation Results for the Size of the Shadow Economy as Measured by Schneider et al. (2010)
Dependent Variable: Shadow Economy by Alm and Embaye (2013)
(1) (2) (3) (4) (5) Log of GDPpc -5.032*** (0.459) -1.344* (0.715) -1.575** (0.687) -1.271 (0.809) -2.718*** (0.790) Tax Burden 0.309*** (0.0799) 0.393*** (0.0670) 0.367*** (0.0575) 0.312*** (0.0751) 0.315*** (0.0711) Rule of Law -7.383*** (1.010) -6.753*** (1.005) Government Effectiveness -6.675*** (1.290) Corruption -1.841*** (0.486) Unemployment 0.304** (0.128) 0.360** (0.160) 0.349** (0.158) Constant 66.08*** (4.094) 35.34*** (6.021) 35.14*** (5.798) 33.77*** (6.461) 52.32*** (5.311) Observations 80 80 80 80 80 R-squared 0.605 0.776 0.792 0.755 0.709 Notes. T-statistics in parentheses. Significance levels: *** p < 0.01, ** 0.01 < p < 0.05, * 0.05 < p < 0.10. Regressions with robust standard errors. Beta coefficients reported.
These results almost entirely support the findings achieved earlier. Tax burden is positive and highly statistically significant across all models. If the institutional variables and the unemployment rate are taken into account the finding stays robust and R² rises. The coefficients of rule of law, government effectiveness and corruption are negative and highly statistically significant when separately included in the regression. The R² of both approaches is the highest if I control for the efficiency of the legal system as institutional variable. This suggest that rule of law is the most adequate institutional variable to choose.
unemployment have a larger shadow economy. For Alm and Embaye’s estimates we find the opposite: countries with a high rate of unemployment have a small shadow economy. The variable unemployment may be problematic in itself. The unemployment rate of a country may depend on the existence and structure of the social service system of a country. In case citizens do not receive financial support when they are unemployed, they do not have an incentive to register as being unemployed. The minimum unemployment rate across countries in my dataset is 1.46% and can be found for Thailand. I do not expect that Thailand actually has the lowest share of unemployment across all countries under investigation, but rather that there is no incentive for citizens to register as being unemployed. Therefore unemployment should be interpreted cautiously. Nevertheless, by comparing model 2 where unemployment is excluded to model 3, where unemployment is included, I find that including or excluding unemployment does neither change the sign nor the significance of the coefficient of tax burden.
To sum up, the results of OLS estimations portrayed in Tables 4.1. and 4.2. show that tax burden and institutional quality are highly relevant in explaining the size of the shadow economy. This finding holds for two different estimation techniques of the shadow economy size.
Endogeneity
(2005) and Nagac (2015) it can be argued that the independent variable of tax burden is potentially endogenous to the size of the shadow economy. Tax burden may be influenced by the size of the shadow economy i.e. countries with a large shadow economy may generally choose high taxation.
A commonly used way to encounter endogeneity issues is to perform an instrumental variable regression analysis (IV). IV estimates the causal effect of a potentially endogenous independent variable on the dependent variable. The underlying idea is to use exogenous instrumental variables that are not correlated with the error term to separate some part of the variation in the independent variable that is exogenous and to estimate the causal effect of this part on the independent variable. The instrumental variables estimation is also called two-stage least squares (2SLS) because one needs to perform two steps to obtain the estimates. In the first-stage regression I estimate the reduced form of tax burden, including on the right hand side of the equation the exogenous independent variables and the instrumental variables that are not included in our original estimation equation. The instrumental variables need to be found to be statistically significant independent variables. If this is the case the second-stage regression can be conducted: the fitted values from the first-stage regression will be included and used as an independent variable in place of tax burden in the main equation. The resulting coefficient estimates are adequate IV or 2SLS estimates. I consider a set of instrumental variables used by previous empirical studies on the relationship of taxation and the size of the shadow economy (Friedman et al., 2000; Torgler & Schneider, 2009; Nagac 2015). Instruments for tax burden are taken from La Porta et al. (1999) who use a set of variables to estimate institutional development. I use four of these variables as instruments: (1) ethno-linguistic fractionalization, (2) the share of each country’s population that is Protestant or Catholic, (3) a dummy variable indicating whether the country’s commercial law has a British, French or German origin and (4) the geographical location of a country, as measured by the absolute value of countries’ latitudes.4
4 The use of these instrumental variables is extensively discussed in Friedman et al. (2000),
The two-stage least squares regression results can be found in Appendix D. Taking into account endogeneity, the coefficient of tax burden stays positive and statistically significant throughout all regressions. If tax burden is high, ceteris paribus, the size of the shadow economy will rise. The coefficient of GDP per capita is negative across all models and significant except for model 3 in Panel A and model 2 in Panel B. A 1% increase of GDP per capita is associated with a 1-5% decrease of the shadow economy, holding all other variables constant.
Turning to the institutional variables, it can be seen that the coefficient of rule of law is negative and highly significant in all models. If rule of law increases by 1 unit, ceteris paribus, the size of the shadow economy decreases by 8-10% based on the estimates by Schneider et al. (2010) and by around 7% based on the estimates by Alm and Embaye (2013). Similarly governmental effectiveness, significantly and negatively affect the shadow economy after using 2SLS. Holding other variables constant, a one index point increase of government effectiveness causes the shadow economy to decrease by 10.17% or 6.59% respectively. The coefficient of corruption is negative as well and furthermore highly significant for both estimates of the shadow economy. For Schneider et al. (2010)’s estimate a one point increase of the corruption cleanness index can, ceteris paribus, be associated with a decrease of the shadow economy of 3.74%.
Differential Effects
As discussed in chapter 2, I expect the effect of taxation on the size of the shadow economy to depend on the country’s level of economic development. To investigate differential effects I firstly construct an interaction variable and include it in a simple OLS regression. In a second step I carry out a margins plot in order to graphically analyze the interaction term. For a last step, the main sample will be split up to estimate the effect of taxation on the shadow economy in separate regression for different levels of economic development.
I start off by constructing an interaction variable of the logarithm of GDP per capita and tax burden. Generally models with interaction terms should include the main effects of the variables used in the interaction term; otherwise the effects of the main variables and interaction effects may get confounded. Therefore I include all three terms – the logarithm of GDP per capita, tax burden and the interaction of both, resulting in the following simplified estimation equation:
�= � + � �+ � � ��+ � �
∗ � �� + ��
where SHADOW� is the size of the shadow economy, � � is the tax ratio,
� �∗ �� �� isthe interaction term of tax burden and the log of GDP per capita and
�� is the error term.
Results are shown in Table 4.3. Model 1 is based on the shadow economy estimate by Schneider et al. (2010), model 2 on the estimate by Alm and Embaye (2013).
Table 4.3. OLS Estimation Results with Interaction Term (1) (2) Log of GDPpc -5.084*** (1.363) -2.919** (1.441) Tax Burden 0.717 (0.929) 1.740** (0.861) Tax Burden*Log(GDPpc) -0.0674 (0.0944) -0.153* (0.0877) Constant 71.79*** (12.89) 46.92*** (13.68) Observations 80 80 R-squared 0.539 0.633 Notes. T-statistics in parentheses. Significance levels: *** p < 0.01, ** 0.01 < p < 0.05, * 0.05 < p < 0.10. Regressions with robust standard errors. Beta coefficients reported.
The coefficient of tax burden is positive in both models. However the interaction of tax burden and the logarithm of GDP per capita is negative for both models and significantly negative for model 2. Thus, the results of our regressions with the interaction term between taxes and GDP per capita show that the relationship between the tax ratio and the size of the shadow economy becomes less positive as GDP per capita increases. This supports the idea of a differential effect of taxation on the size of the shadow economy: the level of development decreases the effect of tax burden on the size of the shadow economy. In other words, the effect of tax burden on the size of the shadow economy is stronger the less developed a country is.
class is Bangladesh with a GDP per capita of 405 USD. The lower middle income class is represented by the Republic of the Congo with a GDP per capita of 1680 USD. Latvia with a GDP of 6,582 USD per capita is representing the upper middle income class, and Australia with a GDP per capita of 33,225 USD per capita is the representative of the high income class. I have chosen the representative countries as those countries that have a GDP per capita closest to the average GDP per capita in their respective income class.
All four graphs in Fig. 4.1. show a positive relation of tax burden on the size of the shadow economy. Moreover, we can see that this effect varies for different levels of GDP per capita. The graph shows that, as the level of development of a country increases, the effect of tax burden on the size of the shadow economy decreases.
Figure 4.1. Relationship of Tax Burden and the Shadow Economy Size per Income Class
Note. Based on OLS estimation of the shadow economy as measured by Alm and Embaye (2013) with interaction term.
20 40 60 80 0 10 20 30 40 50 60 Tax Burden
Low Income Country Lower Middle Income Country
In the third step, the main sample will be split up in two sub-samples: One for low and lower middle income countries and a second one for upper middle and high income countries. A list of country categorization can be found in Appendix C. Table 4.4. summarizes regression results for the relationship of tax on the size of the shadow economy per sub-sample and in comparison to the main sample.
Table 4.4. Summary of the Impact of Tax Burden on the Shadow Economy Size Across Sub-Samples and Compared to the Main Sample
Low and Lower-Middle Income Countries
Upper-Middle and High Income Countries
All Countries Shadow Economy 0.812*** 0.294*** 0.393*** Note. Based on the shadow economy as measured by Alm and Embaye (2013). Detailed estimation results can be found in Appendix E. T-statistics in parentheses. Significance levels: *** p < 0.01, ** 0.01 < p < 0.05, * 0.05 < p < 0.10. Regressions with robust standard errors. Beta coefficients reported.
The results support the existence of a declining effect of tax burden on the shadow economy with increasing levels of development. In low and lower middle income countries a rise of the tax burden by 1%, ceteris paribus, causes a rise of the shadow economy by 0.81%. Estimating the effect of taxation of the shadow economy for upper middle income and high income countries shows that this effect is somewhat weaker: a rise of the tax burden by 1%, ceteris paribus, increases the shadow economy by 0.29%. A comparison to the coefficient of tax burden in our main sample supports the validity of our results.
4.2. Discussion
To close the empirical analysis, I will now discuss the findings for the effect of taxation on the size of the shadow economy with regard to the original expectations from chapter 2. Table 4.5. displays the expectations and findings for the effect of tax burden on the size of the shadow economy, as well as the interpretation of the findings.
Table 4.5. Expectations and Findings for the Effect of Tax Burden on the Size of the Shadow Economy
Expectation Finding Interpretation Developing countries Negative
Neutral
Strongly positive Incentive effect dominates strongly.
Developed countries Neutral Positive
Weakly positive Incentive effect dominates weakly.
For developing countries I expected to find a negative (neutral) relationship between tax burden and the size of the shadow economy, due to a dominating (outweighing) institutional effect. In other words, higher level of tax burden in developing countries, ceteris paribus, decreases (does not affect) the size of the shadow economy.
With further development of public institutions in a given country, I expected the institutional effect to be outweighed or dominated by the incentive effect. For developed countries I thus expected the overall effect of tax burden on the size of the shadow economy to be positive (neutral). Or in other words, higher levels of tax burden in a developed country, ceteris paribus, increases (does not affect) the size of the shadow economy.
tax burden on the size of the shadow economy is stronger for Iow and lower middle income countries than for upper middle and high income countries.
Based on the theory by Friedman et al. (2000) and Lee (2005), I find evidence for a dominating incentive effect: taxpayers encountering larger taxation have the incentive to hide their activities in such a way that their overall tax liabilities are reduced.
Taking into account a differential effect of tax burden on the shadow economy size, Lee (2005) finds that with the development of public institutions the effect of institutions decreases. Contradictory to Lee (2005) I find that in higher income countries the institutional effect does counteract the incentive effect more than in developing countries. In other words, in higher income countries the incentive to hide taxable activities is counteracted more by the institutional effect (representing benefits of engaging in the official economy) than in lower income countries.
How can we interpret our from Lee (2005) divergent findings that (1) the institutional effect seems to be weaker in low income than in high income countries and (2) that the overall effect of tax burden on the size of the shadow economy is positive?
General reasons for the diverging results may be the time period under investigation. Lee (2005) uses data from the 1990s, whereas I use data from the 2000s. Furthermore, Lee mixes data on the size of the shadow economy from various sources all using different estimation techniques. I use data on the size of the shadow economy separately for each country from two sources as to control for influences of different estimation techniques.
spending needs to rise above a higher threshold level before those public institutions start being effective and enhance the tax payers productivity. I believe this threshold level to be higher because governments in low income countries encounter costs of setting up public institutions. In this situation tax revenue is spent, but the improvement of the institutions does not yet have a productivity enhancing effect on the taxpayer and subsequently does not reduce his incentive to engage in the shadow economy. This would explain why the effect of tax burden on the size of the shadow economy is stronger for low income than for high income countries.
5. The Role
of MNEs in the “Global Shadow
Economy
”
The shadow economy has so far mainly been described as being a national phenomenon. However, as shown by Fig. 2.1. and described in chapter 2, the domestic shadow economy is directly linked to illicit financial cross-border flows. In this chapter I will thus introduce the international dimension of the shadow economy – the so called “global shadow economy”. Particularly, I focus on the main agent in this picture: the multinational enterprise (MNE). Appendix F shows that illicit financial flows have increased during the last years. In this chapter I will describe the kind of activities by MNEs that might have contributed to this growth of illicit cross-border flows. In the first part I will investigate the role of MNEs in the global shadow economy, as well as the characteristics allowing MNEs to act as facilitator for tax avoidance. I will then shed light on typical tax avoidance activities by MNEs, namely thin capitalization, trade mispricing and the usage of tax havens.
5.1. The Multinational Enterprise and the Global Shadow
Economy
An institution that plays a vital role in the global shadow economy is the multinational enterprise. Tax avoidance activities by MNEs can be described as a shadow economy active in most globalized sectors, particularly in the extractive industry, banking and finance, aviation and shipping, communications and media, traded commodities and the weapons industry (Christensen & Kapoor, 2005).
Most tax avoidance and tax evasion activities imply but are not limited to the use of offshore trusts, special purpose vehicles, foundations, charities, holding companies and artificial transactions (Christensen & Kapoor, 2005).
5.2. MNEs Tax Avoidance Activities
It is difficult to draw a line between tax avoidance and tax evasion, yet I will attempt to define both concepts. Tax evasion, or illegal tax fraud, comprises illegal business-activities to evade tax. It often entails the misreporting of corporate income on which the company would encounter a tax liability. Tax avoidance schemes on the other hand, are legal modes of conduct by MNEs to reduce or avoid their tax liabilities. The US Senate defines these methods as “legalized tax fraud” (United States of America. Budget Committee, 2015). Although tax avoidance is generally considered to be lawful the distinction to tax evasion is blurred (Otusanya, 2011). Tax avoidance is possible due to certain loopholes MNEs use in different tax systems across various countries. Transactions may for example be specifically designed to avoid taxation in a certain jurisdiction. Major activities MNEs choose in order to reduce their tax bill are thin capitalization, trade mispricing and the use of tax havens.
5.2.1. Thin Capitalization
5.2.2. Trade Mispricing
MNEs may furthermore engage in trade mispricing in order to avoid taxation and manipulate customs duties as well as domestic levies. Mispriced transactions between related parties located in differently taxed jurisdictions are found to offer huge opportunities for MNEs to engage in tax avoidance and evasion (Desai et al., 2006). Technically speaking trade mispricing is the deliberate forgery of traded goods’ prices, quantities and qualities in commercial transactions. It mainly takes on the form of either import over-invoicing or export under-invoicing. These methods allow MNEs to shift profits across borders to low-tax jurisdictions. An MNE’s subsidiary in a low-tax jurisdiction may for example transfer an over-priced invoice to another party in a high-tax jurisdiction thereby transferring profits from a high-tax to a low-tax jurisdiction. Subsequently this method reduces their overall corporate tax bill.
5.2.3. Tax Haven Usage
Subsidiaries located in tax havens may play an important role in the global structure of the MNE as they may operate as a major facilitator for tax avoidance and evasion. Despite substantial research and increasing debates about tax havens there is no straightforward definition of the term itself. Prior studies (Dharmapala & Hines, 2009; Tobin & Walsh, 2013) find three specific characteristics of tax havens and secrecy jurisdictions. First of all, tax havens are characterized by low tax rates. Tax havens offer beneficial tax regulations, such as no or only nominal corporate income taxes. Secondly, tax havens have beneficial financial and administrative regimes that prevent exchange of information between tax authorities. Thirdly, tax havens may permit the reallocation of income to their low-tax jurisdictions by rising low taxes on foreign income (Desai et al., 2006). An overview of tax havens and secrecy jurisdictions worldwide can be found in Appendix G.
6. Linking the Global and the Local Shadow
Economy: The Effect of Multinational Activity on
the Size of the Shadow Economy
This chapter links the global and the local shadow economies by shedding light on potential effects multinational activity may have on the shadow economy.
On the one hand MNEs may positively affect the size of the local shadow economy, due to their tax avoidance activities in the "global shadow economy". By engaging in legalized tax fraud activities MNEs can indirectly place a higher burden on the local factors of production in the host economy, where their subsidiary is located. Governments across the world have been competing against each other by offering tax incentives to attract MNEs to their country (Davies, 2005). Especially low income and lower middle income countries are found to react to tax policies in other countries, leading to tax incentives that significantly undermine their tax revenue (International Monetary Fund, 2015). It was found that tax competition attracting foreign investment, has led governments to cut down tax on income by nonresidents, thereby shifting the tax burden from capital to labor and consumption (Smith & Thomas, 2015). As a result, the combination of less tax revenues from MNEs and high taxes for labor and consumption will stimulate the shadow economy.
On the other hand MNEs may negatively affect the size of the shadow economy in two ways: Firstly, MNEs may include activities by domestic firms in their production line. This may increase the official productivity of domestic firms, subsequently leading to a smaller shadow economy. Secondly, the MNE may act as an institutional entrepreneur in the host economy. MNEs may initiate and actively adjust existing institutions that facilitate tax fraud due to their home country regulations (Smith & Thomas, 2015). In this way MNEs may indirectly decrease the size of the shadow economy in the host country.
They argue that MNEs act as institutional entrepreneurs and agents of change that positively shape the formal sector within the country, thereby reducing engagement in the shadow economy.
The two different effects MNEs may potentially have on the shadow economy can further be related to the theoretical concepts based on Lee (2005), which are described in chapter 2. On the one hand multinational activity may strengthen the incentive effect, because the tax system stimulates tax evasion and tax avoidance in the local economy. On the other hand it may strengthen the institution effect. MNEs need effective public institutions in order to maintain their reputation, as their behavior in the host country is constrained by pressure from their home country regulations and the international business environment. Thus MNEs may actively participate in the development of institutions, giving support for the institution effect.
7. Conclusion
All the way through my empirical analysis across 80 countries, the estimation results suggest that tax burden significantly and positively affects the size of the shadow economy. As tax burden increases, the size of the shadow economy increases too. This empirical finding supports the results from Johnson et al. (1998), Schneider (2005) and Nagac (2015) who also find that tax burden positively affects the size of the shadow economy. Furthermore low-quality public institutions appear to be an important reason to engage in the shadow economy. Under a weak legal system, rather ineffective government institutions and high levels of corruption, there is a higher incentive to engage in the shadow economy. Both findings hold true for two different estimates of the shadow economy - estimated by the currency demand method on the one hand and the MIMIC method on the other. When taking into account a country’s level of development I found the positive effect of tax burden on the size of the shadow economy to be stronger in lower income than in higher income countries.
The shadow economy is not only a national phenomenon; it also exist on a global scale. The “global shadow economy” is mainly comprised of tax avoidance activities by MNEs in globalized sectors. The international group design of MNEs allows for the transfer of goods and money across borders, thereby raising opportunities to reduce their overall corporate tax bill. MNEs can perform legalized tax fraud in various ways that include, but are not limited to, thin capitalization practices, trade over- and under-invoicing as well as the usage of tax havens and secrecy jurisdictions. By analyzing the U.S. FORTUNE 500 companies I found evidence for an extensive use of secrecy jurisdictions and tax havens as host locations for MNEs’ affiliates: 362 companies hold almost 8,000 subsidiaries (3 of them more than 200 each) in such secretive and tax-friendly locations.
I suspect a link between the local and the global shadow economy due to the tax avoidance activities of MNEs. Investigating this suspicion lies beyond to scope of this study and is a relevant topic for future research.
economic development of countries, in some more than in others. My findings suggest that policy makers should focus on improving the quality of public institutions such as the legal system and bureaucracy, as well as reducing corruption in order to reduce the size of the local shadow economy. Furthermore, tax burden plays a significant role in determining shadow economy size that needs cautious handling. Due to the complexity of the shadow economy on a local and global scale, governments that aim to decrease the shadow economy size do not only have to focus on their national institutional and taxation environment, they also need to take tax avoidance happening on the global scale into consideration, as I suspect multinational activity to have an effect on the local shadow economy.
References
Alm, J., Martinez-Vazquez, J. & Schneider, F. (2004). Sizing the problem of the tax. In: Alm, J., Martinez-Vasquez, J. & Wallace, S. (Eds.), Taxing the hard-to-tax: lessons from theory and practice (pp. 11-75). Amsterdam, The Netherlands: Elsevier B. V. – North Holland Publishers.
Alm, J. and Embaye, A. (2013). Using dynamic panel methods to estimate shadow economies around the world, 1984-2006. Public Finance Review, 41, 510-543. Baker, R. W. (2005). Capitalism’s Achilles heel: dirty money and how to renew the
free-market system. Hoboken, United States of America: John Wiley & Sons. Christensen, J. and Kapoor, S. (2005). Tax avoidance, tax competition and
globalisation: making tax justice a focus for global activism. Accountancy Business and the Public Interest, 3(2).
Davies, R. B. (2005). State tax competition for foreign direct investment: a winnable war? Journal of International Economics, 67(2), 498-512.
Dell’Anno, R. & Schneider, F. (2003). The shadow economy of Italy and other OECD countries: what do we know? Journal of Public Finance and Public Choice, 21(2), 97-120.
Desai, M. A., Foley, C.F. & Hines Jr., J. R. (2004). A multinational perspective on capital structure choice and internal capital markets. The Journal of Finance, 59(6), 2451-2487.
Desai, M., Foley, C. F. & Hines Jr., J. R. (2006). The demand for tax haven operations. Journal of Public Economics, 90, 513 – 531.
Dharmapala,D. & Hines Jr., J. R. (2009). Which countries become tax havens? Journal of Public Economics, 93, 1058–1068.
Dyreng, S., Hanlon, M. & Maydew, E. (2008). Long-run corporate tax avoidance. The Accounting Review, 83(1), 61–82.
Feige, E. L. (1989). The underground economies: tax evasion and information distortion. Cambridge, United Kingdom: Cambridge University Press.
Friedman, E., Johnson, S., Kaufman, D. & Zoido-Lobaton, P. (2000). Dodging the grabbing hand: the determinants of unofficial activity in 69 countries. Journal of Public Economics, 76, 459-493.
