Does tourism reduce the shadow economy? An international evidence

Abstract

This paper investigates one of the positive contributions of tourism to the economy through the lens of its influences on the shadow economy. Specifically, our study analyzes the effects of five indicators of tourism consumption (including domestic tourism spending, international travel and tourism consumption, business tourism spending, leisure tourism spending, and outbound tourism spending on the percentage of shadow economy to GDP) in 129 economies between 1996 and 2015. We find interesting results that contribute to the existing literature about tourism economics. Firstly, the development of the inbound tourism industry reduces the shadow economy significantly, while outbound tourism causes higher underground economic activities. Secondly, the influence of tourism on the shadow economy is significant in both the short-run and long run with a stronger effect in the long run. Thirdly, the effect of tourism on the shadow economy is more significant in the 42 High-Income Economies and 54 Low and Lower-middle Income Economies, while it is less obvious in the 33 Upper-Middle Income Economies. These findings have been checked by a battery of robustness checks ensuring their statistical consistency.

1. Introduction

In recent decades, the fast growth of the tourism industry around the globe generates huge attention and strong enthusiastic debates among scholars in the tourism economics [[1], [2], [3]]. Many empirical studies have investigated the influences of tourism on the socio-economic, environmental and cultural realities of our contemporary societies [[4], [5], [6]]. Broadly speaking, tourism development contributes to the economy through several channels such as foreign currency earnings, international investment attraction, tax revenues and employment. In recent years, the role of tourism development has also brought interesting solution to solve other social issues such as gender inequality, for instance Ref. [7].

However, tourism can also have negative impact. Recently, Din, Habibullah [8] raised an interesting question about the relationship between the tourism industry and the shadow economy by examining both short-run and long-run relationships between these aspects. Working on a sample of 149 countries over the 1995–2008 period, they found that tourism receipts and the shadow economy have a cointegration and a long-run relationship. These authors showed that a higher shadow economy was found to have a negative impact on tourism development. This observation has been confirmed by Nguyen and Nguyen [5] who reached a similar conclusion for another sample. The link between tourism and shadow economic often results from the fact that a vast part of employment in the tourism industry (such as accommodation and restaurant services) is mainly fulfilled by self-employment or family business that contributes to the development of informal economic activities [9]. In this context, the results of Din, Habibullah [8] and Nguyen and Nguyen [5] opened several avenues for further studies on the relationship between tourism development and the shadow economy.

The recent literature emphasizes that a high level of informal economies might cause several new issues [[10], [11], [12]]. For instance, Canh, Schinckus [12] explained that a higher informal sector might cause high level and intensity of energy consumption that can have negative impacts on environment. In relation to that, Nguyen and Nguyen [10] documented that a high level of the shadow economy might cause stronger deforestation around the world in the long run. Other studies show that a high shadow economy could cause higher income inequality and several social issues [13,14]. Nowadays, addressing the shadow economy is an important policy objective [15].

The motivation of this study is to delve into the potential relationship between the two aforementioned economic areas (tourism and shadow economy). Precisely, this paper examines the influences of tourism development on the shadow economy in a global sample of 129 economies over the period covering 1996–2015. With this purpose, tourism development is added as an augmented factor to explain the percentage of shadow economy to GDP in comparison to other economic drivers including income level, unemployment, trade openness, and FDI inflows. We conducted our empirical analysis through various econometrical techniques dealing with balanced panel data (Robust Pool OLS, Fixed Effects Models (FEM), Random Effects Model (REM), Feasible Generalized Least Square (FGLS), and the Panel-Corrected Standard Error (PCSE) estimator). We also used the Granger-causality tests and cointegration tests for panel data. Finally, we estimate the short-run and long-run effects of tourism development on the shadow economy using the Autoregressive Distributed-Lagged Dynamic Fixed Effects estimator (ARDL DFE model).

Generally speaking, our empirical results confirm Din, Habibullah [8] and Nguyen and Nguyen [5] who wrote that the shadow economy affects tourism. However, our study contributes to the existing literature by showing that there is a reciprocal effect of tourism development on the shadow economy. Precisely, a higher tourism development, especially domestic tourism, business tourism, and leisure tourism can help in reducing the level of the shadow economy. In contrast, outbound tourism has an increasing effect on the shadow economy. The major contribution of this paper is provide a nuanced perspective on the potential role that each form of tourism could play in controlling informal activities. The policy implication of these observations are discussed in this article.

The rest of this study is organized as follows: the next section reviews the literature regrouping the existing works dealing with the links between tourism development and the shadow economy. Section 3 presents the methodology and data while the empirical results are reported and discussed in Section 4. The final section concludes our research with some policy implications and economic recommendations.

2. Literature review

Until 2018, WTTC [16] emphasized that the tourism industry is one of the most contributing sectors to the economy for employment, export stimulation, and prosperity. Interestingly, they reported a 10.4 % contribution to global GDP and one of five net jobs being related to the tourism industry in the last decade. Therefore, many studies have tried to examine the influences of tourism on the economy [4,17]. Of course, COVID-19 affected these trends but economies gradually came back to their pre-COVID level so one can expect the same relative importance of tourism.

Tourism is usually perceived as having a positive contribution to economic development through the tourism-led growth analysis in levels of sector, industry, country, regions, and across nations.1 For example, Balaguer and Cantavella-Jorda [18] and Inchausti-Sintes [17] documented evidence of the positive effects of tourism on economic growth in Spain. In the same vein, Dogru and Bulut [4] found that tourism can stimulate economic growth (and vice versa) in 7 European countries. Lee and Brahmasrene [19] extended the positive effect of tourism on economic growth in 27 European Union countries between 1988 and 2009. Other evidence of the positive effects of tourism on economic growth has been found in 5 Central America countries [20], China [21], Iran [22], Brazil [23]. Interestingly, Jucan and Jucan [24] argued that tourism development could be seen as a form of policy stimulation to reduce the economic effect of a crisis. Some empirical studies found contradictory results on the contribution of tourism. Sokhanvar, Çiftçioğlu [25], for instance, nuanced and minimized the effects of tourism on economic growth in Brazil, Mexico and the Philippines, while the effects of economic growth on tourism are unveiled in China, India, Indonesia, Malaysia and Peru. Inchausti-Sintes [17] emphasized that tourism could be an economic boost, but it has a negative effect through the Dutch disease effect, which could alter resource allocations towards non-tradable sectors.

In relation to economic growth, tourism is found to have a strong positive influence on employment through job creation [26]. Tourism-related sectors are argued to create jobs mostly for lower-skilled wage workers [9], from unskilled labours to semi-skilled labours [27]. Banerjee, Cicowiez [28], for instance, found that tourism investment in Haiti reduces the unemployment rate from 26 % to 23 %. Interestingly, Çiçek, Zencir [29] indicated that tourism has created many job opportunities for women in Turkey since the 1970s. However, other studies emphasized that employment in tourism, especially in accommodation and restaurant services, is usually fulfilled by self-employment or family business [9], Lundmark, Ednarsson [30].

Din, Habibullah [8] asked an interesting question about the relationship between tourism and informal activities. These authors investigated both the short-run and the long-run relationships between international tourism and the shadow economy in a sample of 149 countries over the 1995–2008 period. They found a cointegration between tourism and the shadow economy. They also showed that a higher shadow economy has a negative impact on tourism. This cointegration and long-run relationship between tourism development and the shadow economy calls for further investigation of the relationship between tourism and informal economic activities. This is the avenue explored in this article.

Shadow economy can be defined as the market-based unregistered economic activities [31] for which there is a desire to evade tax or avoid business regulations [32]. Furthermore, tourism development is related to many sectors such as accommodations, energy, transportation, and telecommunications [16] so the specialized literature pointed out that the shadow economy has various channels to be linked with the tourism industry. The current literature in tourism economics recognizes the existence of the shadow economy as an obstacle to tourism development [33]. Interestingly, Din, Habibullah [8] paved the way for a deeper investigation of the linkages between tourism and the shadow economy and the recent database on shadow economy developed [34] provide a good picture of informal economic activities – these works open new avenues for the analysis of the drivers of the shadow economy. An understanding of these mechanisms is vital for policymakers [35].

Economists agreed that governments can not control or count all economic activities [36]. The shadow economy is usually defined as all currently unregistered economic activities that would contribute to the official GDP [35]. Put in other terms, the shadow economy refers to the legal economic and productive activities that would contribute to GDP if recorded [37]. In their calculation, Mai and Friedrich [37] don't count illegal or criminal activities, do-it-yourself, charitable or household activities in measuring the shadow economy. From the perspective of taxation, Schneider and Williams [35] opined that their calculation of shadow economy only considers all market-based production of legal goods and services that are deliberately concealed from public authorities to avoid one or several of the following constraints: payment of income, value-added or other taxes, or payment of social security contributions, legal labour market standards, etc., or compliance with certain administrative obligations.

In this context, most of the existing literature has pointed out two main reasons for the existence of the shadow economy. The first reason states that the shadow economy is due to tax evasion - the economic agents would want to hide their legal incomes from paying taxes if taxes or social security burdens are high (see Tanzi [38], Tanzi [39], Frey and Pommerehne [40], Schneider [41]). Moreover, the shadow economy activities are untaxed so a higher taxation would create a higher trade-off between working and pleasuring time for workers so that they have higher incentives to hide their income from official sectors [31]. Several empirical studies have investigated this framework [42]. The second reason argues that tax evasion is related to the government control [31], especially the quality of the government [43]. In this view, a bad institutional framework (such as bureaucracy, regulatory discretion, rule of law, corruption, and a weak legal system) would lead to tax evasion [44]. The intensity of regulations through the number of laws and requirements (e.g., licenses, labour market regulations, labour restrictions for foreigners, and trade barriers) is also perceived as a weak institutional framework since it would increase labour costs in the official economy and it can increase the incentives for workers\businesses to hide their activities in the shadow economy [31].

Macroeconomic realities also affect the shadow economy. Kanniainen, Pääkkönen [45], for instance, documented that the state transfers and unemployment quota besides taxation, social security and the tax morale variables are the major determinants of the shadow economy in 21 developed OECD countries for the period 1989/90–2002/03. Salahodjaev [46] unveiled strong evidence that intelligence is negatively associated with the shadow economy in a sample of 158 countries over the period 1999–2007. Moreover, the roles of external factors including foreign direct investment (FDI) and trade openness in the evolution of the shadow economy got stronger attention in the literature [47].

The FDI inflows are documented to be significant contributors to economic development in advanced economies, while it is an indispensable external source for promoting economic activities in developing economies [47]. Many empirical studies documented the positive effects of FDI inflows on economic growth Blomstrom et al., 1992; [[48], [49], [50], [51]]. This literature showed that an increase in FDI inflows fosters official economic activities and then it could limit the size of the shadow economy. However, there is also evidence reducing these negative effects of FDI inflows on the domestic economy [52].

Trade openness appears to play a more important role in business generation in low-income countries [53] where it could also impact the shadow economy. Trade openness is shown to have benefits for both trading partners through the transmitting advanced technology [18]. Importantly, a company, that wants to join the global market, must be a legally registered firm sharing the full relevant information. In this context, a higher trade openness might create incentives for some domestic companies acting in the shadow economy to move to the official registered sectors (if the benefit of globalization is higher than the costs of aligning their activities with all forms of compliance).

Furthermore, the trade openness-related processes induce a higher specialization in domestic economies [54]. Such specialization often requires more sophisticated activities including more specific needs related to technology patents and licenses, which cannot be provided for shadow economic agents. Meanwhile, trading activities with international firms would limit the scope for domestic firms to hide their incomes due to the legal and ethical standards of their trading partners. Consequently, a higher trade openness could reduce shadow economic activities.

However, it is worth mentioning that trade openness may increase the shadow economy due to the competitive effects. Indeed, higher competition from international producers could impact domestic producers from official sectors. There exists some evidence documenting that trade openness has no benefit for economic growth (for example, see Redding [55], Yanikkaya [56], Tekin [57], Menyah, Nazlioglu [58]). Similarly, Vashisht [59] showed that, in some contexts, the overall employment gain from trade openness has been minimal while lower labour demand is related to the shadow economy.

3. Methodology and data

• a.Model

This study starts with the analysis proposed by Din, Habibullah [8] that we extended to tourism economics by incorporating the dynamics of the shadow economy. Specifically, our research examines the effects of tourism development on the level of the shadow economy in a global sample of 129 economies and three sub-samples including 54 Low and Lower-Middle Income Economies (LMEs), 33 Upper-Middle Income Economies (UMEs), and 42 High-Income Economies (HIEs) over the period 1996–2015. We recruit the common drivers of the shadow economy from many existing studies (see Dreher, Kotsogiannis [60], Friedman, Johnson [44], Singh, Jain-Chandra [61], Salahodjaev [46]) including income level, unemployment, urbanization, trade openness, and FDI inflows. We then added tourism development as an augmented driver to explain the dynamics of the shadow economy. In this context, our model can be defined as,

$${SE}_{it}={\beta }_0+{\beta }_j{X}_{it}+{\delta }_1{TR}_{it}+{\epsilon }_{it}$$ [1]

In which: i, t refer to country i at year t; SE is the proxy for the shadow economy; X is a vector of common drivers of shadow economy including income level (Income), unemployment rate (Unem), urbanization (Urban), trade openness (Trade), and FDI inflows (FDI). TR is the tourism development which is proxied by five categories of tourism spending to GDP including the percentage of domestic tourism spending to GDP (TR1), the percentage of internal travel and tourism consumption to GDP (TR2), the percentage of business tourism spending to GDP (TR3), percentage of leisure tourism spending to GDP (TR4), and percentage of outbound travel & tourism expenditure to GDP (TR5), respectively. $\beta ,\delta$ are coefficients while $\epsilon$ is the classical residual term. The following sub-section describes how we handle with our data.

• b.Data

The data related to the shadow economy were collected from Medina and Schneider [34]. Even though, there are debates over the reliability of the shadow economy's measurements since they are unobservable [62]; the method used in Medina and Schneider [34] has been recognized and applied in several previous studies [[63], [64], [65]].

Economic factors including GDP per capita, unemployment rate, urban population, trade openness, and foreign direct investment (FDI) net inflows were collected from the World Development Indicators database of the World Bank (https://datatopics.worldbank.org/world-development-indicators/). Tourism development data including domestic tourism spending, internal travel and tourism consumption, business tourism spending, leisure tourism spending, and outbound tourism spending, were collected from the database of World Travel and Tourism Council (https://wttc.org/).

After collecting all variables, some adjustments have been made. Countries with missing data in main variables (tourism and the shadow economy) were dropped and our final sample includes 2580 country-year observations for a global sample of 129 economies between 1996 and 2015. We use the GDP per capita in its log form to normalize data, urban population is calculated as a percentage of total population, while all other variables are calculated as a percentage of GDP. Data description and other detail are presented in Table 1 below.

Table 1.

Data, definitions and description.

Variable	Definitions	Source	Obs	Mean	Std. Dev.	Min	Max
SE	Shadow Economy (% GDP)	Medina and Schneider (2018)	2580	30.69	13.05	6.16	70.57
Income	Log of GDP per capita (constant 2010 US$)	WDI	2580	8.40	1.53	5.35	11.43
Unem	Unemployment, total (% of total labor force) (modeled ILO estimate)	WDI	2580	7.94	5.33	0.16	29.77
Urban	Urban population (% of total population)	WDI	2580	56.20	22.38	11.35	100.00
Trade	Trade (% of GDP)	WDI	2580	83.62	46.18	15.64	441.60
FDI	Foreign direct investment net inflows (% of GDP)	WDI	2580	4.89	15.76	−43.46	451.72
TR1	Domestic Tourism Spending (% GDP)	WTTC	2580	3.53	1.88	0.57	13.88
TR2	Internal T&T Consumption (% GDP)	WTTC	2580	5.45	3.28	0.50	29.32
TR3	Business Tourism Spending (% GDP)	WTTC	2580	0.96	0.69	0.02	5.12
TR4	Leisure Tourism Spending (% GDP)	WTTC	2580	2.86	2.56	0.00	20.82
TR5	Outbound Travel & Tourism Expenditure (% GDP)	WTTC	2580	2.38	1.75	0.01	24.32

Notes: WTTC is World Travel and Tourism Council database on Tourism Development (assess on Jan/2019); WDI is World Development Indicators of World Bank (databases on Jan/2019).

Following the income classification suggested by the World Bank, we divided our data into three sub-samples in order to study the effects of tourism development on shadow economy by income levels (see Table A1 in Appendix for country list - the data description of three sub-samples is reported in Table A2).

Table 1 above shows that the ratio of the shadow economy to official GDP is 30.69 % on average with a standard deviation of 13.05 %. The period 1996–2015 witnessed a higher trade openness on average of 83.86 % of GDP, while the FDI inflows are 4.89 % of GDP. Interestingly, the average percentage of tourism spending to GDP is 1.88 % coming from domestic spending, 3.28 from internal travel and tourism consumption, 0.69 % for business tourism spending, 2.56 % from leisure tourism spending, and 1.75 % from outbound tourism spending.

Fig. 1 illustrates the relationships between economic factors (including income level, unemployment, urbanization, trade openness and FDI inflows) with the shadow economy. It shows that most of the economic factors (excluding unemployment) seem to have a negative relationship with the shadow economy. Notably, income level and urbanization have a strong negative relationship with the shadow economy. This implies that a country with higher income levels and/or urbanization would have a lower level of informal economic activities. Meanwhile, the economic openness in terms of trade openness and FDI inflows shows some signs of negative relationships with the shadow economy. Fig. 2 below documents the relationship between tourism development and the shadow economy.

Fig. 1.

Shadow Economy and Economic factors.

Fig. 2.

Tourism development and the shadow economy.

It shows that most of the tourism spending (TR1, TR2, TR4, and TR5) have a negative relationship with the shadow economy. This confirms the conclusion provided by Smith [33] and Din, Habibullah [8] who explained that the shadow economy might be a constraint for tourism development. However, this observation might also suggest another explanation according to which tourism development could help in reducing informal economic activities. Moreover, it seems that there might be non-linear relationship between tourism and the shadow economy such as the cases of internal travel and tourism consumption, leisure tourism and outbound tourism.

Let us analyze the correlation matrix for our full sample in Table 2 below (the correlation matrix of three sub-samples is reported in Table A3 in the Appendix).

Table 2.

Correlation matrix.

Correlation	SE	Income	Unem	Urban	Trade	FDI	TR1	TR2	TR3	TR4
Income	−0.70***	1.00
p-value	0.00
Unem	0.02	0.07***	1.00
p-value	0.31	0.00
Urban	−0.51***	0.84***	0.07***	1.00
p-value	0.00	0.00	0.00
Trade	−0.17***	0.22***	0.01	0.18***	1.00
p-value	0.00	0.00	0.64	0.00
FDI	−0.04**	0.08***	0.00	0.10***	0.29***	1.00
p-value	0.03	0.00	0.97	0.00	0.00
TR1	−0.23***	0.26***	−0.04**	0.23***	−0.17***	−0.06***	1.00
p-value	0.00	0.00	0.02	0.00	0.00	0.00
TR2	−0.12***	0.21***	0.09***	0.15***	0.06***	0.04*	0.40***	1.00
p-value	0.00	0.00	0.00	0.00	0.00	0.06	0.00
TR3	0.10***	−0.18***	−0.04*	−0.17***	0.17***	0.03	0.19***	0.37***	1.00
p-value	0.00	0.00	0.07	0.00	0.00	0.19	0.00	0.00
TR4	−0.05***	0.17***	0.04*	0.14***	0.20***	0.15***	0.25***	0.89***	0.23***	1.00
p-value	0.01	0.00	0.07	0.00	0.00	0.00	0.00	0.00	0.00
TR5	−0.11***	0.16***	0.09***	0.18***	0.35***	0.11***	−0.08***	0.29***	0.15***	0.23***
p-value	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00

Notes: *, **, *** are significant levels at 10 %, 5 %, and 1 %, respectively.

Table 2 shows that the shadow economy has a positive correlation with the unemployment rate and business tourism, while it has negative correlations with income level, urbanization, trade openness, FDI inflows, and the rest of tourism spending. Importantly, the correlation between urbanization and income level is positive and large (0.84 for the full sample, 0.68 for LMEs, 0.49 for UMEs, and 0.25 for HIEs). The positive and large correlation among independent variables could cause a problem of multicollinearity in the estimation [66]. In this context, the urbanization is dropped out of empirical estimations that we detail in the following sub-section.

• c.Empirical estimation

Our econometric analysis includes a large cross-section number (129 countries) with a short time range (1996–2015, 20 years). According to Meuleman and Billiet [67] or Faber and Fonseca [68], there might be a problem if the sample size is too small. The literature suggests that the minimum sample size should be 30 and it should be maximized as much as possible [67]. In our analysis, there are 129 countries, which is relative large enough for panel data estimates. However, the analysis with a large country sample might face to issue of cross-section dependence within variables among countries [69]. This section explains the way we deal with this issue. It is worth mentioning that there exist be heteroscedasticity among groups of countries within a range of condition (e.g., economic development). Consequently, we divided our sample into subsamples for further checking and control this aspect.

Therefore, we first use Pesaran's Cross-sectional dependence test [69] to examine the existence of cross-sectional dependence. We show the result in Table 3 hereafter.

Table 3.

Cross-sectional Dependence test.

Test	Cross-sectional Dependence test
Group	Full sample	LMEs	UMEs	HIEs
SE	212.4***	85.64***	68.80***	62.47***
Income	259.9***	101.1***	78.72***	87.67***
Unem	21.13***	7.377***	9.435***	12.89***
Trade	83.68***	27.08***	12.23***	56.41***
FDI	46.83***	29.19***	11.31***	21.87***
TR1	5.352***	10.41***	5.599***	29.44***
TR2	8.566***	14.33***	4.650***	48.49***
TR3	30.21***	13.61***	9.407***	23.74***
TR4	6.473***	24.65***	8.119***	29.00***
TR5	6.913***	3.028***	3.406***	2.277**

Notes: In Cross-sectional Dependence test: the null hypothesis of cross-section independence, CD ∼ N(0,1), P-values close to zero indicate data are correlated across panel groups. *, **, *** are significant levels at 10 %, 5 %, and 1 %, respectively.

Results in Table 3 show the evidence of cross-sectional dependence in all variables for the full sample and also for our three sub-samples. Whereas we use the Granger-causality test developed by Dumitrescu and Hurlin [70] to examine the causal relationship between each independent variables with the dependent ones as detailed in the table below.

Results in Table 4 show an interesting finding since all independent variables have a mutual Granger-causality with the dependent variables, especially the mutual causality between tourism development (TR1, TR2, TR3, TR4, TR5) with shadow economy (SE). This finding confirms the current literature (Smith [33]; Din, Habibullah [8] stating that tourism development influences shadow economy similarly to the negative effect of the shadow economy on tourism. However, the mutual causal relationship between tourism and shadow economy creates a problem of endogeneity for our econometric estimations. In this case, we estimate our equation [1] by using all 1-year lag of independent variables to avoid this issue. This tactics is supported by several existing studies (see Din, Habibullah [8]) who documented long-run relationship between tourism and shadow economy so that the effect of tourism development on shadow economy could be lagged in time. If we drop the urbanization out the empirical estimation due to the strong positive correlation with income level (see the previous sub-section); our final equation for the empirical estimations is given by

$${SE}_{it}={\beta }_0+{\beta }_1{Income}_{it-1}+{\beta }_2{Unem}_{it-1}+{\beta }_3{Trade}_{it-1}+{\beta }_4{FDI}_{it-1}+{\delta }_1{TR}_{it-1}+{\epsilon }_{it}$$ [2]

Table 4.

Granger-causality test.

Dumitrescu & Hurlin (2012) Granger non-causality test
Full sample
Variable: X	X does not Granger-cause SE		SE does not Granger-cause X
	Z-bar	p-value	Z-bar	p-value
Full sample
Income	22.89***	0.000	8.300***	0.000
Unem	12.38***	0.000	15.86***	0.000
Trade	13.03***	0.000	11.85***	0.000
FDI	6.322***	0.000	7.786***	0.000
TR1	9.619***	0.000	11.29***	0.000
TR2	10.25***	0.000	11.43***	0.000
TR3	6.360***	0.000	14.83***	0.000
TR4	9.574***	0.000	14.39***	0.000
TR5	6.508***	0.000	13.56***	0.000
LMEs
Income	18.08***	0.000	4.370***	0.000
Unem	8.651***	0.000	15.17***	0.000
Trade	10.69***	0.000	7.584***	0.000
FDI	7.354***	0.000	7.306***	0.000
TR1	8.845***	0.000	4.364***	0.000
TR2	7.721***	0.000	4.879***	0.000
TR3	6.861***	0.000	6.971***	0.000
TR4	4.952***	0.000	4.420***	0.000
TR5	4.804***	0.000	9.239***	0.000
UMEs
Income	14.92***	0.000	3.403***	0.001
Unem	8.388***	0.000	6.055***	0.000
Trade	6.285***	0.000	6.413***	0.000
FDI	2.002**	0.045	2.601***	0.009
TR1	2.869***	0.004	4.830***	0.000
TR2	4.472***	0.000	7.639***	0.000
TR3	1.122	0.261	10.60***	0.000
TR4	6.796***	0.000	5.784***	0.000
TR5	5.765***	0.000	7.266***	0.000
HIEs
Income	6.386***	0.000	6.573***	0.000
Unem	4.461***	0.000	5.232***	0.000
Trade	5.148***	0.000	6.494***	0.000
FDI	0.965	0.334	3.055***	0.002
TR1	4.285***	0.000	10.56***	0.000
TR2	5.253***	0.000	7.728***	0.000
TR3	2.371**	0.017	8.703***	0.000
TR4	5.138***	0.000	15.08***	0.000
TR5	0.847	0.396	6.851***	0.000

Note: *, **, *** are significant levels at 10 %, 5 %, and 1 %, respectively.

Before estimating equation [2], we recruit six categories of stationary tests for panel data including Pesaran [71]'s CIPS (Z (t-bar)) test; Levin-Lin-Chu unit-root test [72], Harris-Tzavalis unit-root test [73], the Im-Persaran-Shin unit root test [74], Hadri LM test Hadri [75], and Fisher based on Phillips–Perron type unit root test [76]. The results in Table 5 below show that most of the variables are stationary.

Table 5.

Stationary tests.

Test	CIPS test	Levin-Lin-Chu unit-root test		Harris-Tzavalis unit-root test		Im–Pesaran–Shin test		Hadri LM test		Fisher unit root test
Var.		Z-t-tilde-bar Statistic	p-value	Inverse chi-squared	p-value	Adjusted t*	p-value	Z-statistic	p-value	Z-statistic	p-value
Full sample
SE	−2.130**	−1.128	0.129	4.102	1.000	4.918	1.000	105.1***	0.000	185.2	0.999
Income	−2.164***	−3.269***	0.001	8.666	1.000	9.805	1.000	124.0***	0.000	222.1	0.948
Unem	−1.726	−5.396***	0.000	−0.341	0.366	2.125	0.983	72.03***	0.000	263.5	0.393
Trade	−1.898	−4.790***	0.000	−3.657***	0.000	−0.937	0.174	81.84***	0.000	283.6	0.130
FDI	−3.178***	−10.02***	0.000	−19.48***	0.000	−12.63***	0.000	11.83***	0.000	912.6***	0.000
TR1	−1.482	−6.568***	0.000	−4.301***	0.000	−0.906	0.182	80.30***	0.000	358.9***	0.000
TR2	−1.748	−6.996***	0.000	−3.557***	0.000	−2.033**	0.021	82.67***	0.000	464.6***	0.000
TR3	−1.747	−6.176***	0.000	−8.634***	0.000	−2.788***	0.003	62.51***	0.000	468.5***	0.000
TR4	−2.201***	−5.084***	0.000	−3.929***	0.000	−0.638	0.261	84.48***	0.000	389.7***	0.000
TR5	−1.899	−7.977***	0.000	−8.051***	0.000	−1.996**	0.023	66.57***	0.000	483.0***	0.000
LMEs
SE	−2.456***	0.061	0.524	2.710	0.996	4.393	1.000	66.35***	0.000	64.43	0.999
Income	−1.909	2.005	0.977	0.985	1.000	12.07	1.000	79.55***	0.000	45.52	1.000
Unem	−1.787	−2.624***	0.004	−0.622	0.266	3.154	0.999	54.39***	0.000	102.4	0.631
Trade	−2.270***	−2.403***	0.008	−3.847***	0.000	−0.591	0.277	52.56***	0.000	119.4	0.212
FDI	−3.184***	−5.713***	0.000	−10.97***	0.000	−6.678***	0.000	24.54***	0.000	305.6***	0.000
TR1	−2.094**	−3.080***	0.001	−4.230***	0.000	−0.849	0.197	48.64***	0.000	143.0**	0.013
TR2	−2.148**	−4.587***	0.000	−1.285*	0.099	−1.881**	0.029	56.73***	0.000	160.1***	0.001
TR3	−2.391***	−3.875***	0.000	−5.193***	0.000	−2.115**	0.017	43.06***	0.000	207.5***	0.000
TR4	−2.337***	−4.612***	0.000	−2.353***	0.009	−2.294**	0.011	57.47***	0.000	139.9**	0.021
TR5	−1.780	−4.604***	0.000	−9.078***	0.000	−2.469***	0.007	32.28***	0.000	174.6***	0.000
UMEs
SE	−1.998	−1.747**	0.040	1.961	0.975	2.267	0.988	55.18***	0.000	38.53	0.997
Income	−1.666	−5.079***	0.000	4.481	1.000	5.917	1.000	65.11***	0.000	23.34	1.000
Unem	−2.069*	−3.810***	0.000	0.397	0.654	−0.855	0.196	47.68***	0.000	75.37	0.201
Trade	−1.637	−3.590***	0.000	−2.259**	0.011	−1.150	0.125	37.41***	0.000	86.03	0.049
FDI	−3.000***	−5.913***	0.000	−7.818***	0.000	−6.515***	0.000	13.69***	0.000	219.6	0.000
TR1	−2.101*	−5.070***	0.000	−2.675***	0.003	−2.180**	0.014	39.19***	0.000	137.9	0.000
TR2	−1.883	−2.078**	0.018	−4.027***	0.000	−2.533***	0.006	34.92***	0.000	159.6	0.000
TR3	−2.100*	−2.447***	0.007	−4.750***	0.000	−2.058**	0.019	26.77***	0.000	162.0	0.000
TR4	−2.110**	−1.848**	0.032	−2.100**	0.017	−1.054	0.145	41.65***	0.000	140.3	0.000
TR5	−1.826	−3.510***	0.000	−0.475	0.317	−0.021	0.491	44.62***	0.000	155.5	0.000
HIEs
SE	−1.689	−0.295	0.383	2.441	0.992	1.629	0.948	60.08	0.000	82.29	0.532
Income	−1.885	−5.187	0.000	3.130	0.999	−1.757	0.039	66.06	0.000	153.3***	0.000
Unem	−1.478	−6.117	0.000	−0.455	0.324	0.905	0.817	28.06	0.000	85.65	0.429
Trade	−1.725	−2.597	0.005	−0.310	0.378	0.048	0.519	49.67	0.000	78.21	0.657
FDI	−3.419***	−6.264	0.000	−11.23***	0.000	−8.796	0.000	5.829	0.000	387.3***	0.000
TR1	−2.263***	−3.867	0.000	1.436	0.924	1.307	0.904	55.52	0.000	77.88	0.667
TR2	−2.126**	−5.017	0.000	−0.557	0.288	0.815	0.792	51.16	0.000	144.8***	0.000
TR3	−2.007	−4.334	0.000	−6.150***	0.000	−0.664	0.253	33.73	0.000	98.97	0.126
TR4	−2.347***	−2.495	0.006	−2.673***	0.004	2.417	0.992	40.78	0.000	109.4**	0.032
TR5	−2.024	−5.649	0.000	−4.198***	0.000	−0.679	0.248	34.92	0.000	152.9***	0.000

Notes: In Pesaran Panel Unit Root Test (CIPS test): H0 (homogeneous non-stationary): bi = 0 for all I; In Im-Pesaran-Shin unit-root test: Ho: All panels contain unit roots, Ha: Some panels are stationary; In Fisher-type unit-root test: Ho: All panels contain unit roots, Ha: At least one panel is stationary; In Levin-Lin-Chu unit-root test: Ho: Panels contain unit roots, Ha: Panels are stationary; In Harris-Tzavalis unit-root test: Ho: Panels contain unit roots, Ha: Panels are stationary; In Hadri LM test: Ho: All panels are stationary, Ha: Some panels contain unit roots. *, **, *** are significant levels at 10 %, 5 %, and 1 %, respectively.

These results support our plan to examine the effect of tourism on the shadow economy with a low probability of spurious regression due to the non-stationary data. In our context characterized by a large N and a short T panel data with the existence of cross-sectional dependence; the Panel Corrected Standard Errors model (PCSE) estimator is recruited as the main estimation.2 This approach is suggested by many previous works in the specialized literature [77]. Our robustness check is done by using the following techinques: Robust Pool OLS, Fixed Effects Models, Random Effects Models, and Feasible Generalized Least Squares (FGLS) [78]. Due to the variety of economies and time period, the year- and country-fixed effect were also added into our estimations.

As suggested by Din, Habibullah [8], there is a long-run relationship between the shadow economy and tourism so there might be a cointegration between these two realities. Moreover, the literature also suggests that there might be a long-run relationship between tourism with specific economic factors [8]. In this context, we use three different cointegration tests including Kao cointegration test [79], Pedroni cointegration test [80], and Westerlund cointegration test [81] for each relationship between tourism development (TR1, TR2, TR3, TR4, and TR5) with shadow economy. The results in Table 6 below show evidence of the cointegration.

Table 6.

Cointegration tests.

Test	Westerlund test for cointegration		Pedroni test for cointegration		Kao test for cointegration
Model with	Variance ratio	p-value	Modified Dickey-Fuller t	p-value	Modified Phillips-Perron t	p-value
Full sample
TR1	−1.672**	0.047	8.260***	0.000	1.464*	0.071
TR2	−1.114	0.132	7.643***	0.000	1.368*	0.085
TR3	−1.785**	0.037	7.689***	0.000	1.392*	0.081
TR4	−1.484*	0.068	7.707***	0.000	1.569*	0.058
TR5	−1.583**	0.056	7.444***	0.000	1.570*	0.058
LMEs
TR1	−0.478	0.316	5.372***	0.000	1.242	0.107
TR2	−0.805	0.210	5.163***	0.000	1.454*	0.072
TR3	−0.668	0.252	5.757***	0.000	1.400*	0.080
TR4	−0.734	0.231	5.506***	0.000	1.513*	0.065
TR5	−0.101	0.459	5.254***	0.000	1.494*	0.067
UMEs
TR1	−1.561*	0.059	3.588***	0.000	−1.097	0.136
TR2	−0.133	0.446	3.575***	0.000	−1.169	0.121
TR3	−1.798**	0.036	3.186***	0.001	−1.196	0.115
TR4	−0.429	0.333	3.566***	0.000	−1.168	0.121
TR5	−0.950	0.170	3.598***	0.000	−1.125	0.130
HIEs
TR1	−1.005	0.157	4.288***	0.000	2.590***	0.005
TR2	−0.921	0.178	4.245***	0.000	2.450***	0.007
TR3	−0.776	0.218	4.970***	0.000	2.509***	0.006
TR4	−1.389*	0.082	4.561***	0.000	2.550***	0.005
TR5	−1.816**	0.034	3.914***	0.000	2.582***	0.005

Notes: In Westerlund cointegration test: Ho: No cointegration, Ha: Some panels are cointegrated. In Kao cointegration test: Ho: No cointegration; Ha: All panels are cointegrated. In Pedroni test for cointegration: Ho: No cointegration; Ha: All panels are cointegrated. *, **, *** are significant levels at 10 %, 5 %, 1 %, respectively.

In such methodological context, we use the autoregressive distributed lag models (ARDL models) for panel data as suggested by the specialized literature [82]. More precisely, due to the integration of the year effect and the country effect, we use the ARDL Dynamic Fixed-Effects models [83] is applied. Finally, many studies (e.g., see Aidt, Dutta [84], Wang, Cheng [85], Paramati, Shahbaz [86], Azam, Mahmudul Alam [87]) documented some differences in the influences of tourism across countries and regions. Therefore, we decided to replicate the procedures explained above for three sub-samples: LMEs, UMEs, and HIEs. The next section discusses our empirical data.

4. Results and discussions

The results from our PCSE estimator and ARDL DFE models for the case of 129 economies and three subsamples are reported in Table A4 to A11 in the Appendix. The major findings are summarized in Table 7, Table 8 below. Robustness checks have been conducted with Robust Pool OLS, FEM, REM, and FGLS estimator and they all showed consistent and robust findings (these tests can be provided upon requests).3

Table 7.

The effects of tourism development on the shadow economy: PCSE estimators.

Dep. Var: The shadow economy (% GDP)	Full sample	LMEs	UMEs	HIEs
Domestic Tourism Spending (% GDP)	–a	–	–	–
Internal T&T Consumption (% GDP)	–a	–c	+	–a
Business Tourism Spending (% GDP)	–a	–b	–	–a
Leisure Tourism Spending (% GDP)	–a	+	–	–b
Outbound Travel & Tourism Expenditure (% GDP)	+a	+c	+a	+

Notes: −/+ denote negative/positive coefficients (effects); a, b, c denote statistical significance at 1 %, 5 %, and 10 %, respectively. The full results are reported in Tables A4-A7, Appendix.

Table 8.

The short-run and long-run effects of tourism development on the shadow economy: ARDL DFE estimators.

Dep. Var: The shadow economy (% GDP)		Full sample	LMEs	UMEs	HIEs
The short-run effects	Domestic Tourism Spending (% GDP)	–a	–	–c	–
	Internal T&T Consumption (% GDP)	–c	–	–	–
	Business Tourism Spending (% GDP)	+	+	–	–
	Leisure Tourism Spending (% GDP)	–a	–b	–b	+
	Outbound Travel & Tourism Expenditure (% GDP)	–	–	+	+
The long-run effects	Domestic Tourism Spending (% GDP)	–	–	+	+
	Internal T&T Consumption (% GDP)	–b	–c	+	–c
	Business Tourism Spending (% GDP)	–	–	+	–a
	Leisure Tourism Spending (% GDP)	–a	–b	–	–
	Outbound Travel & Tourism Expenditure (% GDP)	+	+	+	–

Notes: −/+ denote negative/positive coefficients (effects); a, b, c denote statistical significance at 1 %, 5 %, and 10 %, respectively. The full results are reported in Tables A8-A11, Appendix.

In the full sample, all variables related to tourism spending (domestic tourism spending (TR1), internal travel and tourism consumption (TR2), business tourism spending (TR3), and leisure tourism spending (TR4)) have a significant negative impact on the shadow economy. This means that tourism development can help in reducing or limiting the size of informal economic activities on a global basis. Our analysis is one of the very first attempts (after the seminal paper from Din, Habibullah [8] to show such dynamic in the relationship between the tourism industry and the shadow economy. Our empirical results actually confirm [8] and indicate a mutual relationship, in which a higher shadow economy could be an obstacle for the tourism industry (especially for international tourism receipts). However, tourism development, in return, can reduce the size of the shadow economy. This is the major contribution of this paper. This finding also has important policy implication as discussed below. Consistently with many empirical studies documented the positive gains of tourism development for job creation, export stimulation, foreign exchange currency earnings, and economic growth [4], this study contributes to the existing literature by providing evidence that tourism development helps in reducing the informal economic sector.

The other proxy for tourism development captured by the outbound tourism spending (TR5), has a significant positive effect on the shadow economy, implying that a higher outward tourism economic activity could induce a higher level of informal economic activities. This is a surprising observation that can be understood as follows: the outward tourists reduce the domestic demand for tourism (especially the tourists’ expenditures), inducing fewer activities in the tourism industry. Consequently, a lower tourism development induces a lower level of employment, incentivizing people to work in informal activities to compensate for their losses. Such a configuration can favor the growth of the shadow economy.

The various sub-samples provide interesting findings. In case of 54 Low and Lower-Middle Income Economies (LMEs), the major results show that the tourism development (including domestic tourism spending (TR1), internal travel and tourism consumption (TR2), business tourism spending (TR3)) have a negative effect on the shadow economy while, the leisure tourism spending and outbound tourism spending have a positive effect on informal economic activities. Interestingly, we have here a result that is specific to the LMEs: leisure tourism can contribute to informal activities – such observation probably results from the fact that leisure tourism in LMEs is not an established sector; it is therefore not very well institutionally structured so that significant number of activities might not be recorded either counted. Domestic tourism or international tourist arrivals can help reduce these countries’ shadow economy. Moreover, it is important to note that LMEs include many transition countries, which are witnessed a high level of shadow economy [34]. Despite the political diversity of these countries, our study provides some policy implications for these countries in fighting with informal economic activities.

For the upper-middle income economies (UMEs), the resutls show that tourism development including domestic tourism spending, business tourism spending, and leisure tourism spending have a negative effect but statistically insignificant. In contrast, outbound tourism spending significantly positively affects the shadow economy. This result implies that tourism development in UMEs is not truly the main driver in reducing informal economic activities in UMEs. This result is interesting since it suggests that the UMEs should focus on transforming the quality of economic development, especially the institutional framework, to solve the problem of informal economic activities instead of concentrating only on policies increasing the quantity of economic activities.

For the high income economies (HIEs), the results on the effects of tourism development on the shadow economy are statistically significant in the same way as observed in LMEs - but the statistically no significance measured in UMEs may suggest that tourism development and the shadow economy may have a dynamic non-linear relationship. This would be consistent with some empirical works showing the non-linear relationships between tourism development and economic factors in tourism economics. For instance, Field Wan and Song (2018) indicated that a non-linear estimation would be more suitable for forecasting tourism development. In the same vein, Rasoolimanesh, Ali [88] unveiled a non-linear algorithm between residents’ negative perceptions and their advocacy for tourism, whereas Katircioğlu [89] documented the non-linear relationship between tourism and emissions in Singapore. Therefore, our results emphasize the importance of considering the dynamic relationship between tourism and the shadow economy or other social-environmental-economic factors in future studies.

There are some observations from our control variables to discuss. The results in Table A4, Appendix, for the full sample show that income level has a significant negative effect on the shadow economy. This is consistent with our primary data and the idea that a higher income context would create less room for informal economic activities. This statement can be explained by the good socio-economic structure provided in high-income economies in capturing, counting and supporting social welfare. Moreover, institutional quality is usually better in higher-income economies so there are fewer opportunities to hide income in the shadow economy [44]. This result is consistent with many existing works. Schneider and Enste [31] documented that, between 1990 and 1993, a high level of the shadow economy in low-income economies such as Nigeria, Egypt, Tunisia, Morocco, Guatemala, Mexico, and Peru (from 40 % to 76 % of the GDP), while this ratio was around 13 %–23 % high-income economies such as Hong Kong, Singapore, and most OECD countries. More recently, Medina and Schneider [34] confirmed this perspective by providing a clear picture of the size of the shadow economy along with income level (see Table 8 in their paper for detail). The significant positive effect of unemployment on the shadow economy implies that the higher the former, the higher the latter. This is consistent with our observation in the previous section and the theoretical background according to which unemployment would induce a higher level of informal economic activities to avoid the tax burden (see Tanzi [38], Schneider [41]).

Our findings also show that economic integration (including trade openness and FDI inflows) has a negative effect on the size of the shadow economy. This means that a higher economic openness would reduce informal economic activities. Interestingly, the negative effect of trade openness is statistically significant, while the effects of FDI inflows are not. This result confirms that trade openness would create pressure on domestic economic agents through competition, which stimulates them to operate in official sectors for a better chance to compete with external producers. Meanwhile, the insignificant effect of FDI inflows on the shadow economy might imply that the latter may not necessarily influence domestic economic activities. More specific research might be needed on this matter. In relation to this point [90], wrote about the emergence of a new kind of informal economic activity through the digital shadow economy. These authors indicated that the digital shadow is not still included in the estimations of the shadow economy despite increasing volumes of e-trade and e-transactions. This raises a question for further investigation about the role of e-trade or e-transaction and their relationship with the shadow economy (and eventually with the tourism industry).

The short-run and long-run effects of tourism development on the shadow economy were estimated by ARDL DFE estimations and are reported in Tables A8 to A11 in Appendix. The major findings are summarized in Table 8 below.

It is worthy to noticing that the coefficients of ECT (Error Corrected Term) in all cases (Tables A8 to A11) are significantly negative and range between −1 and 0 confirming our appropriate use of the ARDL model. Tourism spending from domestic tourism spending, internal travel and tourism consumption, business tourism spending, and leisure tourism spending have a significant negative effect on the shadow economy, whereas outbound tourism spending has a positive effect on the latter. These results are consistent with our findings estimated earlier: four of five proxies of tourism (TR1, TR2, TR3, TR4) have a negative effect on the shadow economy in the long run. The results exhibited in this section confirm the important role played by tourism development in limiting and reducing informal economic activities, especially in the long run.

5. Conclusion

The new comprehensive database of shadow economy from influential studies (e.g. Schneider and Enste [31], Schneider [91], Medina and Schneider [34]) in combination with the interesting findings in tourism economics (e.g. Smith [33] and Din, Habibullah [8]) have opened a room for further investigation on an interesting question: the relationship between tourism development and the shadow economy. This study contributes to tourism economics by extending the studies of Smith [33] and Din, Habibullah [8] on the influences of tourism development on the shadow economy.

Our study examines the effects of tourism spending, including domestic tourism spending, internal travel and tourism consumption, business tourism spending, leisure tourism spending, and outbound tourism spending on the level of shadow economy (expressed as a percentage of the official GDP). A balanced panel of 129 economies with three sub-samples following income levels, including 54 Low and Lower-Middle Income Economies, 33 Upper-Middle Income Economies and 42 High-Income Economies in the period 1996–2015 have been investigated in details under various econometric techniques.

Overall, our analysis shows that inbound tourism development appears to reduce the shadow economy significantly, but outbound tourism induces a higher shadow economy. The effects of tourism are statistical significance in both the short-run and long run but the long run effects are stronger. The effects of tourism on the shadow economy are statistical significant in the HIEs and the LMEs, while they are less obvious in the UMEs. These findings suggest that policymakers should set up suitable policies to support sustainable tourism development, especially in relation to domestic tourism, which would not only help in solving the negative effects of the shadow economy on tourism but would also be benefits the economy by reducing the level of the shadow economy. More importantly, our findings emphasize that government should form out long-term development strategies for tourism since the benefits of the latter are stronger in the long run.

Finally as a limitation of this study, it is important to notice that we only investigated the direct effects of tourism development on the shadow economy without considering other important catalysts such as institutional frameworks [92] or climate changes [2]. Future studies would be welcome to explore into the catalyst roles of these factors in the relationship between tourism and the shadow economy.

Data availability statement

The data can be provided upon reasonable requests.

CRediT authorship contribution statement

Canh Phuc Nguyen: Conceptualization, Data curation, Formal analysis, Methodology, Validation, Writing – original draft, Writing – review & editing. Christophe Schinckus: Conceptualization, Validation, Writing – original draft, Writing – review & editing. Binh Quang Nguyen: Data curation, Validation, Writing – original draft.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article.