Shadow Economies and Corruption All Over the World: Revised Estimates for 120 Countries

Estimations of the shadow economies for 120 countries, including developing, Eastern Europe and Central Asian and high income OECD countries over 1999 to 2006 are presented. The average size of the shadow economy (as a percent of "official" GDP) in 2004/05 in 76 developing countries is 35.5%, in 19 Eastern and Central Asian countries 36.7% and in 25 high income OECD countries 15.5%. An increased burden of taxation and social security contributions, combined with labour market regulations are the driving forces of the shadow economy. Furthermore, the results show that the shadow economy reduces corruption in high income countries, but increases corruption in low income countries. Finally, the various estimation methods are discussed and critically evaluated.


Introduction
As corruption and shadow economic activities are facts of life around the world, most societies attempt to control these activities through various measures like punishment, prosecution, economic growth or education. To gather information about the extent of corruption and the shadow economy and its relationship or who is engaged in corrupt and/or underground activities, the frequency with which these activities are occurring and magnitude of them, is crucial for making effective and efficient decisions regarding the allocations of a country's resources in this area. Unfortunately, it is very difficult to get accurate information about the relationship between corruption and shadow economy activities on the goods and labour market, because all individuals engaged in these activities wish not to be identified. Hence, doing research in these two areas can be considered as a scientific passion for knowing the unknown.
Although substantial literature 1 exists on single aspects of the hidden or shadow economy and a comprehensive survey has been written by Schneider (one author of this paper) and Enste (2000), the subject is still quite controversial 2 as there are disagreements about the definition of shadow economy activities, the estimation procedures and the use of their estimates in economic analysis and policy aspects. 3 Nevertheless around the world, there are some indications for an increase of the shadow economy but little is known about the development and the size of the shadow economies in developing, Eastern European and Central Asian (mostly the former transition countries) and high income OECD countries over the period 1999 to 2005/06 using the same estimation technique and almost the same data sample.
Hence, the goal of this paper is threefold: (i) to undertake the challenging task of estimating the shadow economy for 120 countries all over the world, 4 (ii) to provide some insights into the main causes of the shadow economy, and (iii) to explore the relationship between the shadow economy and corruption. In Section 2 an attempt is made to define the shadow economy and some theoretical considerations about the reasons why it is increasing. Section 3 presents the econometric estimation results and the calculation of the size of the shadow economy in 120 countries over the period _________________________ 1 The literature about the "shadow", "underground", "informal", "second", "cash-" or "parallel", economy is increasing. Various topics, on how to measure it, its causes, its effect on the official economy are analyzed. See for example, survey type publications by Frey and Pommerehne (1984); Thomas (1992); Loayza (1996); Pozo (1996); Lippert and Walker (1997); Schneider (1994aSchneider ( , 1994bSchneider ( , 1997Schneider ( , 1998aSchneider ( , 2003Schneider ( , 2005Schneider ( , 2007; Johnson, Kaufmann, and Shleifer (1997), Zoido-Lobatón (1998a, 1998b); Belev (2003); Gerxhani (2003) and Pedersen (2003). For an overall survey of the global evidence of the size of the shadow economy see , Schneider and Enste (2000, 2006 and Alm, Martinez and Schneider (2004), and Kazemier (2005a) 2 Compare e.g. in the Economic Journal, vol. 109, no. 456, June 1999 the feature "controversy: on the hidden economy". 3 Compare the different opinions of Tanzi (1999), Thomas (1999), Giles (1999aGiles ( , 1999b and Pedersen (2003), and Janisch and Brümmerhoff (2005). 4 This paper focuses on the size and development of the shadow economy for countries and does not show any disaggregated values for specific regions. Lately some first studies were undertaken to measure the size of the shadow economy as well as the "grey" or "shadow" labour force for urban regions or states (e.g. California). Compare e.g. Marcelli, Pastor and Joassart (1999), Marcelli (2004), Chen (2004), Williams (2004aWilliams ( , 2004bWilliams ( , 2005aWilliams ( , 2005bWilliams ( , 2006, Williams and Windebank (1999, 2001a, 2001b, Flaming, Haydamack, and Jossart (2005) and Alderslade, Talmage and Freeman (2006), and Brueck, Haisten-DeNew and Zimmermann (2006).
www.economics-ejournal.org 1999/2000 to 2005/06. In Section 4 two hypotheses about the relationship between the shadow economy and corruption are derived and some empirical results are shown. In Section 5 a summary is given and some policy conclusions are drawn. Finally in the three appendices (chapters 6, 7, and 8) the various methods to estimate the shadow economy are presented and critically evaluated, a definition of the variables and data sources are given, and the descriptive statistics of the variables are shown.

2
Some Theoretical Considerations about the Shadow Economy

Defining the Shadow Economy
Most authors trying to measure the shadow economy face the difficulty of how to define it. One commonly used working definition is all currently unregistered economic activities that contribute to the officially calculated (or observed) Gross National Product 5 . Smith (1994, p. 18) defines it as "market-based production of goods and services, whether legal or illegal, that escapes detection in the official estimates of GDP." Or to put it in another way, one of the broadest definitions of it includes…"those economic activities and the income derived from them that circumvent or otherwise avoid government regulation, taxation or observation". 6 In this paper the following more narrow definition of the shadow economy is used 7 : The shadow economy includes all market-based legal production of goods and services that are deliberately concealed from public authorities for the following reasons: (1) to avoid payment of income, value added or other taxes, (2) to avoid payment of social security contributions, (3) to avoid having to meet certain legal labour market standards, such as minimum wages, maximum working hours, safety standards, etc., and (4) to avoid complying with certain administrative procedures, such as completing statistical questionnaires or other administrative forms.
Hence, in this paper, we will not deal with typical underground, economic (classical crime) activities, which are all illegal actions that fit the characteristics of classical crimes like burglary, robbery, drug dealing, etc. We also exclude the informal household economy which consists of all household services and production. This paper also does not focus on tax evasion or tax compliance, because it would get too long, and

Public Sector Services
An increase of the shadow economy can lead to reduced state revenues which in turn reduce the quality and quantity of publicly provided goods and services. Ultimately, this can lead to an increase in the tax rates for firms and individuals in the official sector, quite often combined with a deterioration in the quality of the public goods (such as the public infrastructure) and of the administration, with the consequence of even stronger incentives to participate in the shadow economy. Zoido-Lobatón (1998a, 1998b) present a simple model of this relationship. Their findings show that smaller shadow economies appear in countries with higher tax revenues if achieved by lower tax rates, fewer laws and regulations and less bribery facing enterprises. Countries with a better rule of law, which is financed by tax revenues, also have smaller shadow economies. Transition countries have higher levels of regulation leading to a significantly higher incidence of bribery, higher effective taxes on official activities and a large discretionary regulatory framework and consequently a higher shadow economy. Their overall conclusion is that "wealthier countries of the OECD, as well as some in Eastern Europe, find themselves in the 'good equilibrium' of relatively low tax and regulatory burden, sizeable revenue mobilization, good rule of law and corruption control, and a [relatively] small unofficial economy. By contrast, a number of countries in Latin American and the former Soviet Union exhibit characteristics consistent with a 'bad equilibrium': tax and regulatory discretion and burden on the firm is high, the rule of law is weak, and there is a high incidence of bribery and a relatively high share of activities in the unofficial economy." (Johnson, Kaufmann, and Zoido-Lobatón, 1998a, p. 1). First results of the influence of corruption on the shadow economy and vice versa are reported in chapter 4 of this paper.

3
The Size of the Shadow Economy for 120 Countries

Econometric Results
In Tables 3.1 1995/96, 1997/98, 1999/2000, 2001/02, 2002/03, 2003/04, 2004/05, and 2005/06. For the developing countries we use as cause variables the following seven: share of direct taxation (direct taxes in percent of overall taxation), size of government (general government final consumption expenditure, in percent of GDP) as proxy for indirect taxation and fiscal freedom (an index consisting of top individual income tax rate, top individual corporal tax rate, and total tax revenues as percent of GDP) as three tax burden variables in a wide sense; the business freedom index (which has the elements: time to open a business, financial costs to start a business, minimum capital stock to start a business, and costs for obtaining a licence) for state regulation, the state of economy with the two variables: the unemployment rate and GDP per capita and finally an index of economic freedom. As indicator variables we use growth rate of GDP per capita, the employment quota (people over 15 economically active in % of total population), and the annual rate of local currency per capita. 11 For the Eastern European and Central Asian (mostly former transition) countries, we use as cause variables the share of indirect taxes and an index of fiscal freedom as the two tax burden variables, the state regulation, the business freedom index, and for the state of the economy the unemployment rate, inflation rate and openness (sum of export and imports of goods and services in percent of GDP). As indicator variables, we use GDP per capita, the growth rate of total labour force, and the growth rate of local money per capita. For the 25 high income OECD countries, we use for the two tax burden variables the total tax burden (total tax revenues in percent of GDP), the fiscal freedom index, for _________________________ 10 The classification which country is a developing, or an Eastern European and Central Asian or a High Income OEC country follows the one done by the World Bank (2002) e.g. using a benchmark per capita income of USD 9.265 or less for developing countries. 11 Here we have the problem that in some developing and Eastern European and Central Asian countries, the US Dollar (or the Euro) is also a widely used currency, which is not considered here, because we could not obtain any reliable figures of the amount of US Dollar (Euro) in these countries.
www.economics-ejournal.org the state regulatory burden the two variables business freedom (index) and regulatory burden (index, Heritage Foundation), and for the state of the economy the unemployment rate. As indicator variables, we use GDP per capita, the labour force participation rate and currency as ratio of M2.
The estimations results for 57 developing countries, including the unemployment rate over the period 1999 up to 2006 are shown in Table 3.1.1 and the estimation results for 76 developing countries (excluding the unemployment rate) over the same period are shown in Table 3.1.2. In both estimations, most estimated coefficients of the cause variables have the theoretically expected signs. All cause variables are statistically significant, at least at the 90 percent confidence level. In both estimations, the share of direct taxation and the size of government are highly statistically significant, as well as the business freedom variable. The unemployment variable has the expected positive sign, and GDP per capita is in both equations highly statistically significant with the expected negative sign. If we turn to the indicator variables, the employment quota and the growth rate of local money per capita are in both equations highly statistically significant. The test statistics are quite satisfactory too.
In Table 3.2, the MIMIC estimations results for the 19 Eastern European and Central Asian (mostly former transition) countries over the period 1999 to 2006 are presented. If we begin with the cause variables, the share of indirect taxes and the fiscal freedom variable, both capturing the overall state burden, are highly statistically significant and have the expected sign. Turning to regulation, the business freedom variable has the expected negative sign and is highly statistically significant. As these countries experienced periods of high inflation, the inflation rate has the expected positive sign and is highly statistically significant. The variable openness, modelling in a certain way the transition process, is not statistically significant. Considering the indicator variables, the growth rate of the total labour force is statistically significant, as well as the growth rate of local money per capita. Also, here the test statistics are quite satisfactory.
Finally, in Tables 3.3.1 and 3.3.2, the estimation results for the 25 high income OECD countries are shown. Table 3.3.1 shows the estimation without the tax morale  variable for 25 countries over a data set from 1996 up to 2006, and Table 3.3.2 the  results including the tax morale variable for only 15 high income OECD countries from  1996 up to 2005. If we consider first Table 3.3.1, the results without the tax morale variable, the two variables capturing government burden (total tax burden and fiscal freedom) are highly statistically significant and have the expected sign. The unemployment rate has the expected sign and is at 90 percent confidence level statistically significant. Turning to the indicator variables, the labour force participation rate and currency as ratio of M2 are both highly statistically significant. Also, the test statistics for this equation is quite satisfactory. Turning to Table 3.3.2, where we present the results including tax morale as an additional cause variable, we have fewer countries and fewer observations but see that the tax morale variable is highly statistically significant and has the expected sign, as well as the other cause variables. 12 Summarizing the econometric (MIMIC) results, we can demonstrate that for all three groups of countries, the theoretical consideration of the causes of the shadow economy in Section 2 can be confirmed. Tax burden variables (direct and/or indirect _________________________ 12 The importance of this variable with respect to theory and empirical investigations is also shown in Frey (1997), Feld and Frey (2002, and Torgler and Schneider (2005).
www.economics-ejournal.org and/or overall tax burden or other indices) as well as indices measuring the fiscal freedom in a country are driving forces for the growth of the shadow economy in all three types of countries. Followed by the measures of regulation (measured in the business freedom variable and regulatory intensity) and by measures of the official economy, the unemployment rate, and for the developing countries, GDP per capita have a highly statistically significant influence. 549 Notes: t-statistics are given in parentheses *; **; *** means the t-statistics are statistically significant at the 90%, 95%, or 99% confidence level. All variables are used as their standardized deviations from mean. 1) Steigers Root Mean Square Error of Approximation (RMSEA) for test of close fit; RMSEA < 0.05; the RMSEA-value varies between 0.0 and 1.0. 2) If the structural equation model is asymptotically correct, then the matrix S (sample covariance matrix) will be equal to Σ (θ) (model implied covariance matrix). 3) Test of Adjusted Goodness of Fit Index (AGFI), varying between 0 and 1; 1 = perfect fit. 4) The degrees of freedom are determined by 0.5 (p + q) (p + q + 1) -t; with p = number of indicators; q = number of causes; t = the number for free parameters.
www.economics-ejournal.org 27 Observations 720 Notes: t-statistics are given in parentheses *; **; *** means the t-statistics are statistically significant at the 90%, 95%, or 99% confidence level. All variables are used as their standardized deviations from mean. 1) Steigers Root Mean Square Error of Approximation (RMSEA) for test of close fit; RMSEA < 0.05; the RMSEA-value varies between 0.0 and 1.0. 2) If the structural equation model is asymptotically correct, then the matrix S (sample covariance matrix) will be equal to Σ (θ) (model implied covariance matrix). 3) Test of Adjusted Goodness of Fit Index (AGFI), varying between 0 and 1; 1 = perfect fit. 4) The degrees of freedom are determined by 0.5 (p + q) (p + q + 1) -t; with p = number of indicators; q = number of causes; t = the number for free parameters. 27 Observations 132 Notes: t-statistics are given in parentheses *; **; *** means the t-statistics are statistically significant at the 90%, 95%, or 99% confidence level. All variables are used as their standardized deviations from mean. Estimated the model using the government share of the real GDP per capita as proxy for indirect taxation gives similar results. 1) Steigers Root Mean Square Error of Approximation (RMSEA) for test of close fit; RMSEA < 0.05; the RMSEA-value varies between 0.0 and 1.0. 2) If the structural equation model is asymptotically correct, then the matrix S (sample covariance matrix) will be equal to Σ (θ) (model implied covariance matrix). 3) Test of Adjusted Goodness of Fit Index (AGFI), varying between 0 and 1; 1 = perfect fit. 4) The degrees of freedom are determined by 0.5 (p + q) (p + q + 1) -t; with p = number of indicators; q = number of causes; t = the number for free parameters.
1) Steigers Root Mean Square Error of Approximation (RMSEA) for test of close fit; RMSEA < 0.05; the RMSEA-value varies between 0.0 and 1.0. 2) If the structural equation model is asymptotically correct, then the matrix S (sample covariance matrix) will be equal to Σ (θ) (model implied covariance matrix). 3) Test of Adjusted Goodness of Fit Index (AGFI), varying between 0 and 1; 1 = perfect fit. 4) The degrees of freedom are determined by 0.5 (p + q) (p + q + 1) -t; with p = number of indicators; q = number of causes; t = the number for free parameters. 27 Observations 74 Notes: t-statistics are given in parentheses *; **; *** means the t-statistics are statistically significant at the 90%, 95%, or 99% confidence level. All variables are used as their standardized deviations from mean. Estimated the model using alternative measures for the tax burden gives similar results. Estimated the model using alternative measures for the tax burden (i.e. direct and indirect taxation separately) gives similar results. We have also used the share of people who find it justifiable claiming government benefits to which they are not entitled to proxy tax morality but find no significant impact of this variable. 1) Steigers Root Mean Square Error of Approximation (RMSEA) for test of close fit; RMSEA < 0.05; the RMSEA-value varies between 0.0 and 1.0. 2) If the structural equation model is asymptotically correct, then the matrix S (sample covariance matrix) will be equal to Σ (θ) (model implied covariance matrix). 3) Test of Adjusted Goodness of Fit Index (AGFI), varying between 0 and 1; 1 = perfect fit. 4) The degrees of freedom are determined by 0.5 (p + q) (p + q + 1) -t; with p = number of indicators; q = number of causes; t = the number for free parameters. www.economics-ejournal.org

The Size of the Shadow Economies for 120 Countries for 1999/2000 to 2005/2006
In order to calculate the size and development of the shadow economies of 120 countries, we have to overcome the disadvantage of the MIMIC approach, which is, that one gets only relatively estimated sizes of the shadow economy and one has to use another approach to get absolute figures. In order to calculate absolute figures of the size of the shadow economies from these MIMIC estimation results, we use the already available estimations from the currency demand approach for Australia, Austria, Germany, Hungary, Italy, India, Peru, Russia and the United States (from studies of Schneider (2007), Chatterjee, Chaudhury and Schneider (2006), Dell'Anno and Schneider (2004), Schneider (2003, 2005), Alexeev and Pyle (2003), Schneider and Enste (2002) and Lackó (2000)). As we have absolute values of the shadow economy (in % of GDP) for various years for the above mentioned countries, we can use a benchmark procedure to transform the index of the shadow economy from the MIMIC estimations into absolute values. 13 When showing the size of the shadow economies over the five periods of time (1999/2000, 2001/2002, 2002/2003, 2003/2004 and 2004/2005) for the 120 countries which are quite different in location and developing stage, one should be aware that such country comparisons give only a rough picture of the ranking of the size of the shadow economy in these countries and over time, because the MIMIC and the currency demand methods have shortcomings; these are discussed in the appendix (chapter 6) 14 . Due to these shortcomings a detailed discussion of the (relative) ranking of the size of the shadow economies is not conducted.

76 Developing Countries 15
As we presented two different MIMIC estimates with respect to the developing countries due to the fact that the unemployment variable was only available for a much smaller country sample (57 developing countries instead of 76), the calibration of the size and development of the shadow economy of the developing countries is done for both sets of estimations. 16 In Tables 3.4.1 and 3.4.2, the size of the shadow economy in _________________________ 13 This procedure is described in great detail in the paper Dell'Anno and Schneider (2005. 14 ) See also Thomas (1992Thomas ( , 1999, Tanzi (1999), Pedersen (2003) and Ahumada, et al. (2004), Janisch and Brümmerhoff (2005), Schneider (2005) and Breusch (2005aBreusch ( , 2005b. 15 For an extensive and excellent literature survey of the research about the shadow economy in developing countries see Gerxhani (2003),who stresses thoroughout her paper that the distinction between developed and developing countries with respect to the shadow economy is of great importance. Due to space reasons this point is not further elaborated here; nor are the former results and literature discussed. Compare Schneider and Enste (2000). 16 Calibration is performed separately for each country. Having calculated the ordinal MIMIC index by applying the estimated coefficients to the standardized time series, we add a constant to this MIMIC index in order to satisfy the usual condition that the shadow economy as percentage of official GDP is in the base period equal to the chosen base value. Changes of the shadow economy are then determined by the dynamics of this index. The base values for the high income OECD countries and the eastern European and central Asian countries originate from the year 2005. Regarding the developing countries we opted for base values originating from the year 2000 because of better data availability in that year compared to 2005. 57 developing countries (including the unemployment variable in the MIMIC estimation) is presented in Table 3.4.1 in alphabetical order, and in Table 3 Large shadow economies in some developing countries is only to some extent an issue of tax burden and regulation, given the simple fact that the limited local economy means that citizens are often unable to earn a living wage in a legitimate manner. Working in the shadow economy is often the only way of achieving a minimal standard of living. It should also be noted that the average size of the Asian shadow economies are smaller than the shadow economies of African and Latin American countries.

_________________________
17 It should be mentioned that Mainland China and Vietnam are still communist countries with partly market economies, so that the figures of these two countries may be biased.

19 Eastern European and Central Asian (mostly former transition) Countries
The measurement of the size and development of the shadow economies in the transition countries has been undertaken since the late 1980s starting with the work of Kaufmann and Kaliberda (1996), Johnson et al. (1997) and Lackó (2000). They all use the physical input (electricity) method and come up with quite large figures. In the work of Alexeev and Pyle (2003) and Belev (2003) the above mentioned studies are critically evaluated arguing that the estimated sizes of the unofficial economies are to a large extent a historical phenomenon and partly determined by institutional factors. In Tables 3.5.1 and 3.5.2 the size and development of 19 Eastern European and Central Asian (mostly former transition) countries in percent of GDP are presented. 18 Table 3.5.1 presents again the countries in alphabetical order and 3.5.2 with respect to size. If we first consider the average of the shadow economy of these 19 Eastern European and Central Asian countries, it was 35.8% in 1999 and increased to 36.

25 High-Income OECD Countries
The size and development of the shadow economies of 25 High Income OECD countries is shown in Tables 3.6.1 and 3.6.2. Table 3.6.1 presents again the countries in alphabetical order and 3.6.2 with respect to size. If we first consider the average development of the shadow economies of the 25 High Income OECD countries with respect to the size, the size was in the year 1996 14.2% and increased to 15.8% in the year 2006. Some high income OECD countries, like Greece has up's and down's, others (like Belgium, Australia) show a steady increase. The lowest shadow economies have Switzerland, the United States and Austria with an average size of the shadow economy over the period 1996 to 2006 from 7.0, 7.9 and 8.1 percent. The highest shadow economies among these 25 high income OECD countries have Mexico with 31.5, Korea with 26.6 and Greece with 25.3 percent.
In Tables 3.6.3 and 3.6.4, the size and development of the shadow economies of 15 high income countries are presented; for these 15 countries we could include the tax morale variable. Due to the fewer data points for the tax morale variable a comparison is difficult between the two estimations; however, one result is that those countries, which have high tax morale, have a somewhat lower shadow economy, ceteris paribus. 18 Only 19 countries could be included because for the Republic of Kyrgyzstan we have only one observation point.      , Open-Assessment E-Journal 29 1) Bold values calibrated ones (method Dell'Anno and Schneider (2009: 122)), other interpolated ones.

Corruption and the Shadow Economy: Substitutes or Complements? 19
Quite often shadow economy and corruption 20 are seen as "twins", who need each other or fight against each other. This means for a social scientist that, theoretically, corruption and the shadow economy can be either complements or substitutes. Choi and Thum (2005) present a model where the option of entrepreneurs to go underground constrains a corrupt official's ability to ask for bribes. McCorriston (2005a and2005b) extend the model to the explicit specification of institutional quality. The model shows that corruption and shadow economy are substitutes in the sense that the existence of the shadow economy reduces the propensity of officials to demand graft. Johnson et al. (1997), on the contrary, model corruption and the shadow economy as complements. In their full-employment model, labour can be either employed in the official sector or in the underground economy. Consequently, an increase in the shadow economy always decreases the size of the official market. In their model, corruption increases the shadow economy, as corruption can be viewed as one particular form of taxation and regulation (driving entrepreneurs underground). Hindriks et al. (1999) also show that the shadow economy is a complement to corruption. This is because, in this case, the tax payer colludes with the inspector so the inspector under-reports the tax liability of the tax payer in exchange for a bribe. 21 More recently, Echazu and Bose (2008) also demonstrate-considering different types of corrupt bureaucrats in the official and the shadow economies-that corruption and the shadow economy can be complements.
Theoretically, the relationship between corruption and the shadow economy is thus unsettled. There is, however, reason to believe that the relationship might differ among high and low income countries. In high income countries, the official sector provides public goods like the rule of law, enforcement of contracts, and protection by an efficient police. Usually, only craftsmen or very small firms have (or take) the option of going underground. In this case, the shadow economy is hidden from tax inspectors and other officials. In other words, there are no bribes necessary or possible to buy the way out of the official sector. In high income countries-typically showing comparably small levels of corruption-individuals confronted with a corrupt official always have the choice to bring the official to court. Moreover, in high income countries corruption quite often takes place, for example, to bribe officials to get a (huge) contract from the public sector (e.g. in the construction sector). This contract is then handled in the official economy and not in the shadow economy. Hence, corruption in high income countries can be a means to achieve certain benefits which make work in the official economy easier, e.g., winning a contract from a public authority, getting a licence (e.g. for operating taxes or providing other services or getting the permission to convert land into "construction ready" land, etc.). In high income countries people thus bribe in order to be able to engage in more official economic activities. As Schneider and Enste (2000) point out, at least two thirds of the income earned in the shadow economy is _________________________ immediately spent in the official sector. The shadow economy and the official sector might thus be complements. The corresponding increase in government revenue and strengthened institutional quality is likely to decrease corruption. The prediction of a negative (substitutive) relation between corruption and the shadow economy in high income countries is in line with the models of Choi and Thum (2005) and Dreher, Kotsogiannis, and McCorriston (2005a). 22 In low income countries, on the contrary, we expect different mechanisms to prevail. Instead of working partly in the official sector and offering additional services underground as in high-income countries, enterprises completely engage in underground activity. Examples for enterprises operating completely underground are restaurants, bars, or haircutters-and even big production companies. One reason for this is that public goods provided by the official sector are, in many developing countries, less efficient compared to high income countries or do not exist at all. Big companies, however, are comparably easy to detect and-in order to escape taxation and punishment-they have to bribe officials, thereby increasing corruption. Corruption often takes place in order to pay for activities in the shadow economy, so that the shadow economy entrepreneur can be sure not to be detected by public authorities. Here, the shadow economy and corruption are likely to reinforce each other, as corruption is needed to expand shadow economy activities and-at the same timeunderground activities require bribes and corruption. To get some additional income from the shadow economy entrepreneur, it is natural for public officials to ask for bribes and thus benefit from the shadow market. In low income countries, we therefore expect a positive (complementary) relationship between corruption and the shadow economy. This corresponds to the predictions of the models of Johnson et al. (1997), Hindriks et al. (1999), and Echazu and Bose (2008). In summary, we thus formulate the following two hypotheses: Hypothesis 1: In low income countries, shadow economy activities and corruption are complements. Hypothesis 2: In high income countries, shadow economy activities and corruption are substitutes.
To begin with, the two hypotheses are tested for a cross-section of 120 countries and a panel of 70 countries for the period 1994 to 2002. 23 Table 4.1 summarizes the empirical results of Dreher and Schneider (2006). Overall, they show that an increase in perceived corruption over time also increases the shadow economy. This confirms the models of Johnson et al. (1997) and Hindriks et al. (1999). Across countries, however, greater perceived corruption does not lead to a greater shadow economy. To some extent this also supports the results of Méon and Sekkat (2004) showing the withincountry variation to be important in their analysis of corruption on foreign direct investment and exports.
Regarding the impact of the shadow economy on perceived corruption, these results for the overall sample are similar to those for the other way round. In the cross-country regressions, all coefficients are completely insignificant. An increase in the shadow _________________________ 22 Consequently, Dreher, Kotsogiannis, and McCorriston (2005a) test their model employing data for OECD countries only. 23 For the description of the data, the estimation techniques used, and the various specification see Dreher and Schneider (2006, chapters 3 and 4). economy over time increases corruption according to the fixed and random effects estimator, but not when the endogeneity of the shadow is controlled for. Turning to the sub-samples, the results show that higher perceived corruption significantly reduces the shadow economy in high income countries, confirming the models of Choi and Thum (2005) and Dreher, Kotsogiannis, and McCorriston (2005a). In low income countries, on the contrary, corruption tends to increase with a higher shadow economy, again confirming the models of Johnson et al. (1997) and Hindriks et al. (1999). This is true for the impact of perceived corruption in the within-groups specification and actual corruption in all specifications. -modelling corruption and the shadow economy as unobservable variables using a structural equation model with two latent variablesprovide evidence for a complementary relationship between corruption and the shadow economy. Their analysis considers 51 countries around the world over the period 2000 to 2005, the majority of them being developing countries. Using the typical determinants for corruption and the shadow economy, they can confirm most of the findings of previous theoretical and empirical research for both latent variables. Figure  1 shows specification (1) of their estimations.
The estimated coefficients for the paths between corruption and the shadow economy and vice versa measure the influence of the latent variables (i.e. corruption and the shadow economy) on each other. Although the mutual relationship between corruption and the shadow economy is positive across all estimated specification (not shown here), the coefficients for the two paths differ substantially in magnitude. That is, the causal effect of the shadow economy on corruption is stronger than the effect of corruption on the shadow economy. One possible explanation for this is that corruption functions as an additional tax in the official economy-which, in turn, increases the size of the shadow economy. Likewise, the shadow economy induces higher corruption as bureaucrats exploit their positions of power and as firms or individuals willingly pay bribes and hide their underground activities. In addition, the shadow economy can also be seen as an indication of overall deterioration of social and cultural norms, which results in even more widespread corruption.  (2) Ethnic Fractionalization, Religious Fractionalization, Latitude, French Legacy, Socialist Legacy, German Legacy, Scandinavian Legacy. * denotes significant at 10% level; ** significant at 5% level; *** significant at 1% level Source: Dreher and Schneider (2006 , Table 12).  Note: *; **; *** indicate significance of the coefficients at the 90%, 95%, and 99% confidence level.
Clearly, the structural equation model presented in  is only an additional step in furthering our understanding of corruption and the shadow economy. The findings however reveal that a large shadow economy is linked to high levels of corruption. In countries with large shadow economies, firms and individuals often rely to a large extent on shadow economic activities. In order to avoid detection, taxation, and punishment, they bribe bureaucrats. Moreover, low tax revenues reduce the quality of public services and infrastructure. This in turn reduces the incentives to remain in the official economy. Weaker legal systems and unstable conditions for economic activity increase corruption. Acting like an extra tax corruption drives individuals underground. Thus, the empirical relationship between corruption and the shadow economy analyzed using a structural equation model confirms the findings of Johnson, Kaufmann, and Shleifer (1997), Johnson, Kaufmann, and Zoido-Lobatón (1998b), Hindriks, Muthoo, and Keen (1999), Friedman et al. (2000), and Echazu and Bose (2008).

Summary and Conclusions
There have been many obstacles to overcome to measure the size of the shadow economy, to analyze its consequences on the official economy and the interaction between corruption and the shadow economy, but as this paper shows some progress has been made. We provided estimates of the size of the shadow economies for 120 countries for five periods of time (1999/2000, 2001/2002, 2002/2003, 2003/04, 2004/05 and 2005/06) using the MIMIC procedure for the econometric estimation, and the currency demand approach for calibrating the estimated values of the size of the shadow economy into absolute ones. Coming back to the headline of this paper, some new knowledge/insights are gained with respect to the size and development of the shadow economy of 120 countries, 24 and to the relationship between the shadow economy and corruption leading to four conclusions: The first conclusion from these results is that for all countries investigated the shadow economy has reached a remarkably large size of an average value of 32.3% of official GDP over 120 countries over 1999/00 to 2005/06. However, the average size of the shadow economies of all three groups of countries (76 developing countries, 19 Eastern European and Central Asian (mostly transition) countries, and 25 high income OECD countries) increased only modestly from 31.8% of official GDP in 1999/00 to 32.7% of official GDP in 2005/06.
The second conclusion is that shadow economies are a complex phenomenon present to an important extent in all type of economies (developing, transition and highly developed). People engage in shadow economic activity for a variety of reasons, among the most important of which we can count are government actions, most notably, taxation and regulation.
Considering a public choice perspective a third conclusion for highly developed countries is that a government may not have a great interest to reduce the shadow economy due to the fact that: (i) tax losses my be moderate, as at least 2/3 of the income earned in the shadow economy is immediately spent in the official economy, (ii) income earned in the shadow economy increases the standard of living of at least 1/3 of the working population, (iii) between 40 and 50% of the shadow economy activities have a complementary character, which means that additional value added his created, which increases the official (overall) GDP, and (iv) people who work in the shadow economy have less time for other things like going to demonstrations, etc.
Considering these three conclusions, it is obvious that one of the big challenges for every government is to undertake efficient incentive orientated policy measures in order to make work less attractive in the shadow economy and hence to make the work in the official economy more attractive. In a number of OECD countries this policy direction has been successfully implemented and this has led to a stabilisation or even reduction of the size of the shadow economy.

_________________________
The fourth conclusion is that the shadow economy reduces corruption in high income countries (substitution effect) and increases corruption in low income countries (complementary effect).

6
Appendix 1: Methods to Estimate the Size of the Shadow Economy: The DYMIMIC and Currency Demand Approach It has already been mentioned in chapter 3, estimating the size and development of a shadow economy is a difficult and challenging task. In this appendix, we give a short but comprehensive overview of the currency demand and the MIMIC approach; each is briefly discussed as well as critically evaluated. 25

The Currency Demand Approach
The currency demand approach, which is also called an "indicator" approach, is a macroeconomic one and uses various economic and other indicators that contain information about the development of the shadow economy (over time), and leaves some "traces" of the shadow economy. This approach was first used by Cagan (1958), who calculated a correlation of the currency demand and the tax pressure (as one cause of the shadow economy) for the United States over the period 1919 to 1955. 20 years later, Gutmann (1977) used the same approach but without any statistical procedures. Cagan's approach was further developed by Tanzi (1980Tanzi ( , 1983, who econometrically estimated a currency demand function for the United States for the period 1929 to 1980 in order to calculate the shadow economy. His approach assumes that shadow (or hidden) transactions are undertaken in the form of cash payments, so as to leave no observable traces for the authorities. An increase in the size of the shadow economy will therefore increase the demand for currency. To isolate the resulting "excess" demand for currency, an equation for currency demand is econometrically estimated over time. All conventional possible factors, such as the development of income, payment habits, interest rates, and so on, are controlled for. Additionally, such variables as the direct and indirect tax burden, government regulation and the complexity of the tax system, which are assumed to be the major factors causing people to work in the shadow economy, are included in the estimation equation. The basic regression equation for the currency demand, proposed by Tanzi (1983), is the following: ln (C / M 2 ) t = β O + β 1 ln (1 + TW) t + β 2 ln (WS / Y) t + β 3 ln R t + β 4 ln (Y / N) t + u t with β 1 > 0, β 2 > 0, β 3 < 0, β 4 > 0 where ln denotes natural logarithms, C / M 2 is the ratio of cash holdings to current and deposit accounts, TW is a weighted average tax rate (to proxy changes in the size of the shadow economy), _________________________ 25 A discussion and critical evaluation of all used approaches is given in Schneider (2005Schneider ( , 2007. www.economics-ejournal.org WS / Y is a proportion of wages and salaries in national income (to capture changing payment and money holding patterns), R is the interest paid on savings deposits (to capture the opportunity cost of holding cash) and Y / N is the per capita income. 26 Any "excess" increase in currency, or the amount unexplained by the conventional or normal factors (mentioned above) is then attributed to the rising tax burden and the other reasons leading people to work in the shadow economy. Figures for the size and development of the shadow economy can be calculated in a first step by comparing the difference between the development of currency when the direct and indirect tax burden (and government regulations) are held at their lowest value, and the development of currency with the current (much higher) burden of taxation and government regulations. Assuming in a second step the same velocity for currency used in the shadow economy as for legal M1 in the official economy, the size of the shadow can be computed and compared to the official GDP.
The currency demand approach is one of the most commonly used approaches. It has been applied to many OECD countries, 27 but has nevertheless been criticized on various grounds. 28 The most commonly raised objections to this method are: Not all transactions in the shadow economy are paid in cash. Isachsen and Strom (1985) used the survey method to find out that in Norway, in 1980, roughly 80% of all transactions in the hidden sector were paid in cash. The size of the total shadow economy (including barter) may thus be even larger than previously estimated.
Most studies consider only one particular factor, the tax burden, as a cause of the shadow economy. But others (such as the impact of regulation, taxpayers' attitudes toward the state, "tax morality" and so on) are not considered, because reliable data for most countries are not available. If, as seems likely, these other factors also have an impact on the extent of the hidden economy, it might again be higher than reported in most studies. 29 As discussed by Garcia (1978), Park (1979), and Feige (1996), increases in currency demand deposits are due largely to a slowdown in demand deposits rather than to an increase in currency caused by activities in the shadow economy, at least in the case of the United States. Blades (1982) and Feige (1986Feige ( , 1996, criticize Tanzi's studies on the grounds that the US dollar is used as an international currency. Instead, Tanzi should have _________________________ 26 The estimation of such a currency demand equation has been criticized by Thomas (1999) but part of this criticism has been considered by the work of Giles (1999aGiles ( , 1999b and Bhattacharyya (1999), who both use the latest econometric techniques. 27 See Karmann (1986and 1990), Schneider (1997, 1998a, 2005, Johnson, Kaufmann, and Zoido-Lobatón (1998a), and Williams and Windebank (1995). 28 See Thomas (1992Thomas ( , 1999; Feige (1986); Pozo (1996);Pedersen (2003) and Ahumada, Alvareda, Canavese A., and P. Canavese (2004); Janisch and Brümmerhof (2005); and Breusch (2005aBreusch ( , 2005b. 29 One (weak) justification for the use of only the tax variable is that this variable has by far the strongest impact on the size of the shadow economy in the studies known to the authors. The only exception is the study by Frey and Weck-Hannemann (1984) where the variable "tax immorality" has a quantitatively larger and statistically stronger influence than the direct tax share in the model approach. In the study of Pommerehne and Schneider (1985), for the U.S., besides various tax measures, data for regulation, tax immorality, minimum wage rates are available, the tax variable has a dominating influence and contributes roughly 60-70% of the size of the shadow economy. See also Zilberfarb (1986). considered (and controlled) the presence of US dollars, which are used as an international currency and are held in cash abroad. 30 Moreover, Frey and Pommerehne (1984) and Thomas (1986Thomas ( , 1992Thomas ( , 1999 claim that Tanzi's parameter estimates are not very stable. 31 Most studies assume the same velocity of money in both types of economies. As argued by Hill and Kabir (1996) for Canada and by Klovland (1984) for the Scandinavian countries, there is already considerable uncertainty about the velocity of money in the official economy, and the velocity of money in the hidden sector is even more difficult to estimate. Without knowledge about the velocity of currency in the shadow economy, one has to accept the assumption of "equal" money velocity in both sectors. Ahumada, et al. (2004) show that the currency approach, together with the assumption of equal income velocity of money in both the reported and the hidden transaction is only correct if the income elasticity is 1. As this is not the case for most countries, the calculation has to be corrected.
Finally, the assumption of no shadow economy in a base year is open to criticism. Relaxing this assumption would again imply an upward adjustment of the size of the shadow economy.

6.2
The Model Approach 32 All methods described so far that are designed to estimate the size and development of the shadow economy consider just one indicator that "must" capture all effects of the shadow economy. However, it is obvious that shadow economy effects show up simultaneously in the production, labour, and money markets. An even more important critique is that the causes that determine the size of the shadow economy are taken into account only in some of the monetary approach studies that usually consider one cause, the burden of taxation. The model approach explicitly considers multiple causes leading to the existence and growth of the shadow economy, as well as the multiple effects of the shadow economy over time. _________________________ 30 In another study by Tanzi (1982 a or b?, esp. pp. 110-113) he explicitly deals with this criticism. A very careful investigation of the amount of US-$ used abroad and the US currency used in the shadow economy and to "classical" crime activities has been undertaken by Rogoff (1998), who concludes that large denomination bills are the major driving force for the growth of the shadow economy and classical crime activities are due largely to reduced transactions costs. 31 However in studies for European countries Kirchgaessner (1983Kirchgaessner ( , 1984 and Schneider (1986) reach the conclusion that the estimation results for Germany, Denmark, Norway and Sweden are quite robust when using the currency demand method. Hill and Kabir (1996) find for Canada that the rise of the shadow economy varies with respect to the tax variable used; they conclude "when the theoretically best tax rates are selected and a range of plausible velocity values is used, this method estimates underground economic growth between 1964 and 1995 at between 3 and 11 percent of GDP." (Hill andKabir [1996, p. 1553]). 32 This part is derived from a longer study by Aigner, Schneider, and Ghosh (1988, p. 303), applying this approach for the United States over time for the first time; for Germany this approach has been applied by Karmann (1986 and1990). The pioneers of this approach are Weck (1983), Frey and Weck-Hannemann (1984), who applied this approach to cross-section data from the 24 OECD countries for various years. Before turning to this approach they developed the concept of "soft modeling" (Frey, Weck, and Pommerehne (1982), Frey andWeck (1983a and1983b)), an approach which has been used to provide a ranking of the relative size of the shadow economy in different countries.

www.economics-ejournal.org
The empirical method used is quite different from those used so far. It is based on the statistical theory of unobserved variables, which considers multiple causes and multiple indicators of the phenomenon to be measured. For the estimation, a factoranalytic approach is used to measure the hidden economy as an unobserved variable over time. The unknown coefficients are estimated in a set of structural equations within which the "unobserved" variable cannot be measured directly. The MIMIC (multiple indicators multiple causes) model consists in general of two parts, with the measurement model linking the unobserved variables to observed indicators. 33 The structural equations model specifies causal relationships among the unobserved variables. In this case, there is one unobserved variable, or the size of the shadow economy; this is assumed to be influenced by a set of indicators for the shadow economy's size, thus capturing the structural dependence of the shadow economy on variables that may be useful in predicting its movement and size in the future. The interaction over time between the causes Z it (i = 1, 2, …, k), the size of the shadow economy X t , in time t, and the indicators Y jt (j = 1, 2, …, p) is shown in Figure 6.1.  ...

Causes (i)
The burden of direct and indirect taxation, both actual and perceived. A rising burden of taxation provides a strong incentive to work in the shadow economy. (ii) The burden of regulation as proxy for all other state activities. It is assumed that increases in the burden of regulation give a strong incentive to enter the shadow economy. (iii) The "tax morality" (citizens' attitudes toward the state), which describes the readiness of individuals (at least partly) to leave their official occupations and enter the shadow economy: it is assumed that a declining tax morality tends to increase the size of the shadow economy. 35

Indicators
A change in the size of the shadow economy may be reflected in the following indicators: Similarly, increased activities in the hidden sector may be expected to be reflected in shorter working hours in the official economy. (iii) Development of the production market. An increase in the shadow economy means that inputs (especially labour) move out of the official economy (at least partly), and this displacement might have a depressing effect on the official growth rate of the economy.
The latest use of the model approach has been undertaken by Giles (1999aGiles ( , 1999bGiles ( , 1999c and by Giles, Tedds and Werkneh (2002), Giles and Tedds (2002), Chatterjee, Chaudhury and Schneider (2006), , Pickhardt and Sarda-Pous (2006), Schneider (2007), and Buehn, Karmann, and Schneider (2009). They basically estimate a comprehensive (sometime dynamic) MIMIC model to get a time series index of the hidden/measured output of New Zealand, Canada, Germany, India or Australia, and then estimate a separate "cash-demand model" to obtain a benchmark for converting this index into percentage units. Unlike earlier empirical studies of the hidden economy, they paid proper attention to the non-stationary, and possible cointegration of time serious data in both models. Again this MIMIC model treats hidden output as a latent variable, and uses several (measurable) causal and indicator variables. The former include measures of the average and marginal tax rates, inflation, real income and the degree of regulation in the economy. The latter include changes in the (male) labour force participation rate and in the cash/money supply ratio. In their cashdemand equation they allow for different velocities of currency circulation in the hidden and recorded economies. Their cash-demand equation is not used as an input to determine the variation in the hidden economy over time-it is used only to obtain the long-run average value of hidden/measured output, so that the index for this ratio predicted by the MIMIC model can be used to calculate a level and the percentage units _________________________ 35 When applying this approach for European countries, Frey and Weck-Hannemann (1984) had difficulty in obtaining reliable data for the cause series, besides the ones for the direct and indirect tax burden. Hence, their study was criticized by Helberger and Knepel (1988), who argue that the results were unstable with respect to changing variables in the model and over the years.
www.economics-ejournal.org of the shadow economy. Overall, this latest combination of the currency demand and MIMIC approach clearly shows that some progress in the estimation technique of the shadow economy has been achieved and a number of critical points have been overcome. However, objections can also be raised against the (DY)MIMIC method 36 , i.e.: (1) instability in the estimated coefficients with respect to sample size changes, instability in the estimated coefficients with respect to alternative specifications, difficulty of obtaining reliable data on cause variables other than tax variables, and (4) the reliability of the variables grouping into "causes" and "indicators" in explaining the variability of the shadow economy.