This paper explores the impact of refugees on the size of the underground economy in 28 European Union countries over the period from 1998 to 2017. It applies a nonlinear methodology by employing dynamic panel threshold estimations. The main findings uncover a nonlinear connection between refugees and the informal economy with an inverted V-shape and a different magnitude of effects depending on the share of the refugee population. The underground economy is stimulated at a low level of refugee inflows (where immigrants make up <0.572% of the total population). Large inflows compress the underground economy, which increases competition in the labor market based on lower labor costs. Economic growth and international trade play a crucial role in reducing the size of the informal economy. Equally importantly, coherent unemployment policy and adequate regulation of illegal immigrants support this process.
- underground economy
- European Union
In recent years, many European Union (EU) member countries have been confronted with inflows of migrants, most of whom invoke refugee status. This demographic movement has generated heated debates, not only in the sociopolitical environment but also among researchers. A new set of challenges now face the EU space, with economic challenges assuming key importance. Generally, the inflow of both legal and illegal immigrants modifies the aggregate size of an economy, influences the level of wages and the labor force participation rate, generates revenue inequalities in respect to native population, and affects the net fiscal burden (Becker and Ferrara, 2019; Verme and Schuettler, 2021). Given many current geopolitical issues (such as violent conflicts in Syria, Iraq, Afghanistan, Pakistan, the Horn of Africa, and the Sahel), a parallel inflow of refugees accentuates the phenomenon of migration in the EU.
Unlike typical migration flows, refugees’ movement represents a special case. The typical migrant moves from his or her origin country not because of a direct threat or persecution, but to improve their standard of living, while a refugee is a person seeking international protection according to the “1951 Convention relating to the Status of Refugees” and its 1967 Protocol. At the level of EU, the “Mutual Learning Programme Peer Review” (European Commission, 2016, p. 2) stipulates that “refugees have the legal right to participate in the labour market on equal terms as native-born residents, and the barriers they face may be of a practical nature.” Noteworthy is that in his or her first step to being officially recognized as a refugee, an asylum-seeker has the right to work no later than 9 months after filing the application.
Unfortunately, despite this right, it is very difficult, if not impossible, for a refugee to find a place of work in legal economic activity. Low revenues, language barriers, status as an unskilled worker, culture, religion, or employer reticence are some of the limits in entering the formal labor market.
Therefore, the refugee often prefers to
The underground economy (shadow economy) is defined by Medina and Schneider (2018) as activities out of official rules being hidden from monetary, regulatory, and institutional perspectives. Thomas (1992) classifies those activities in four sectors: the household sector, the informal sector, the irregular sector, and the criminal sector.
The EU member states give rise to special interest regarding the implications of refugees on the underground economy. Figure 1 reveals the number of refugees in the EU countries between 1990 and 2020. There are two peaks. The first peak is shown in 1993, with >2.4 million refugees. The Iraqi invasion of Kuwait, the Yugoslav Wars from 1991 to 2001, genocide in Rwanda, and the conflict in Afghanistan are the main explanations of these inflows. After 1993, the number of refugees has continuously decreased, being <1 million people in 2013.
The second peak occurs in 2020, with >2.5 million refugees. The wave started in 2013 and is related to conflicts in several countries, such as Syria, Iraq, Afghanistan, Pakistan, and those in the Horn of Africa and the Sahel. According to the European Commission (2018), in 2015, around 72% of the budget for humanitarian aid (over €1 billion) has been used for helping refugees and internally displaced persons. This exodus generated numerous disputes between the EU member states, especially regarding the country contingents, border protection, and refugees’ real status. In parallel, the inflow has been accompanied by illegal immigration, the official control of this phenomenon being a critical problem.
Compared with other migrants, refugees give rise to particular interest related to the underground economy at the level of the EU. Their productivity can be higher if they act in the official economy than in the unofficial one, given the comfort offered by the right to work. The preference for the official economy can also reduce the budget pressure through social allowances, improving at the same time the tax base. In parallel, refusing to act in the underground economy, they can confer more stability and security for host countries, being a disincentive for illegal child labor. Not least, statistical evidence reveals that the share of the refugee population is more strongly correlated with the underground economy than the share of the immigrant population (Table A3 in the Appendix). Despite these particularities, the EU registered different experiences with the officially recognized refugees. Their economic integration depends on how resourceful the person is and/or how open the host society is (Strang and Ager, 2010).
In this context, this paper analyzes the influence of refugees on the size of the underground economy in all the 28 EU countries, over the period from 1998 to 2017. The theoretical grounding of the study is the “structural model” of the underground economy proposed by Medina and Schneider (2018), adjusted for the influence of refugee inflows. The model is inspired by the seminal works of Frey and Weck-Hanneman (1984), followed by Schneider et al. (2010), Hassan and Schneider (2016), and Buehn et al. (2009). The “structural model” of Medina and Schneider (2018) is a part of an extended approach, along with a “measurement model”. The “structural model” investigates the main determinants of the underground economy, while the measurement model estimates it. The underground economy is explained by the tax burden, the regulatory burden, economic and business freedoms, unemployment, and the level of economic development. The authors claim that economic and business freedom as well as economic development are negatively correlated with the underground economy, while the rest of determinants have positive signs.
Two main theoretical resorts connect refugee inflows with the underground economy, as Williams and Round (2010) claim. The first resort is explained by the “structuralist theory,” while the second is explained by the “neoliberal thesis.” The “structuralist theory” claims the idea of the expulsion of workers from the official labor market (Sassen, 1998; Slavnic, 2010), while the “neoliberal thesis” considers the individual choice as the key of the process, with an emphasis on skilled workers, artisans, and entrepreneurs (Maloney, 2004).
To test the impact of refugee inflows on the size of the underground economy, as well as their implications on Medina and Schneider’s (2018) model, a nonlinear empirical methodology is followed using a dynamic panel threshold strategy similar to that of Kremer et al. (2013), with origins in Hansen (1999). The estimations reveal a nonlinear connection between the share of the refugee population and the informal economy.
The contribution of this paper is threefold. First, to the best of our knowledge, the paper is one of the first studies in the field investigating the “refugees-underground economy” nexus in the case of the EU. Second, as an additional novel contribution, this paper is one of the first studies that follow a nonlinear strategy to investigate the relationship between refugees and the size of the underground economy. Finally, this work offers a new extension of Medina and Schneider’s (2018) model by entering a new determinant for the size of the underground economy: refugee inflows.
There is a vast literature regarding the main
Schneider and Buehn (2018) offer a quasi-complete framework of these determinants. Their tabular approach starts with factors such as the tax and social security burden, continuing with institutional quality, regulation, public sector services, tax morale, deterrence, the size of the economy, self-employment and unemployment, and the size of the agricultural sector.
Dell’Anno et al. (2007) claim that a higher tax burden modifies labor-leisure choices, extending the underground economy. Institutional quality seems to be negatively correlated with the underground economy, as an efficient government reduces the size of the informal sector (Williams and Schneider, 2016). Regulations and trade barriers have a positive influence, with more regulations (i.e., low international trade flows) stimulating the underground economy, as Hassan and Schneider (2016) reveal. Furthermore, according to Feld and Schneider (2010), low quality and quantity of public sector services increase the general tax rate, which extends to the informal sector. Feld and Schneider (2010) also underline that the perceptions toward taxation and strict social norms reduce the shadow economy, while the deterrence in policies combats it. The size of economy is also linked with the informal economy, but opposite in sign, as Williams and Schneider (2016) note. More precisely, an extended official economy is often associated with a small underground one. Self-employment and unemployment are two other essential determinants. Williams and Schneider (2016) argue that rates of self-employment and unemployment are positively correlated with the underground economy. Finally, Hassan and Schneider (2016) find that the agricultural sector also plays an important role. Herein, the possibility of evading the underground economy increases as the agricultural sector expands.
Over the past decades, the literature has started to consider
The seminal works in the field can be considered to be the classical models proposed by Todaro, (1969), Fields (1975), Mazumdar (1976), and Lal (1973), treating the underground economy as a temporary outlet for migrants. Further, extensive literature is developed by focusing exclusively on the link between migration flows and the underground economy, with both theoretical and empirical approaches.
For example, Borjas (1994) studied the United States, finding that immigrants are partially responsible for the fall of native earnings during the 1980s. Card (2005) reports a low correlation of local revenues with immigrant density, while D’Amuri and Peri (2014) argue that the inflow of migrants drives natives toward more skill-intensive jobs. Other papers (Borjas, 1998; Bianchi et al., 2012) show that the presence of immigrants is significantly correlated with a higher propensity of committing crimes.
A positive influence of migrant inflows on the underground economy is claimed by Crush et al. (2005) and Landau (2006) in the case of South Africa. They argue that immigration stimulates the underground economy as a result of informational barriers and issues regarding the access to the formal economy. Okkerse (2007) focuses on the case of Organization for Economic Co-operation and Development (OECD) countries, evidencing a bidirectional causality between illegal migration and the shadow economy. She also states that in Ireland, the small inflow of migrants is correlated with a small informal economy. Venturini and Villosio (2008) focus on Italy, but demonstrate that legal immigrants prefer to alternate between regular and irregular jobs. The authors explain that the majority of immigrants access informal jobs because they are better remunerated compared with formal, low-skilled jobs, which are often less stable. Not least, the literature gives special attention to the particular alternatives at the classical underground economy, namely, criminal activities and robberies. For example, Borjas (1998) links U.S. immigrants with criminal activities, while Bianchi et al. (2012) claim that the immigration phenomenon is strongly related to the incidence of robberies. By contrast, Schneider and Badekow (2006) demonstrate that the immigrants escape to underground economy only under severe constraints imposed by public institutions and bureaucratic apparatus.
While it is vast, at the same time it confuses the differences between forced and voluntary migration. Northcote (2015) highlights the fact that there are serious difficulties distinguishing between a refugee (i.e., a forced migrant) and a voluntary migrant, despite the fact that the refugee claims an independent legal personality (Feller, 2005). Therefore, a frontier approach is invoked by many researchers, such as Friebel et al. (2013), Hunter and Skinner (2003), Landau and Wa Kabwe Segatt (2009), or Vearey (2008).
Finally, a new strand of literature focuses on the impact of
There is an extended literature underlining the
Altındağ et al. (2020) consider yearly censuses of companies by targeting the case of Syrian refugees in Turkey. They underline that refugee inflows significantly impact the host economy. Many shocks were registered at the level of production and prices, but some engagements of companies in the informal economy were also observed. Moreover, the most affected companies seem to be small ones, as well as those from the construction and hospitality sectors. In the same vein, Berdiev et al. (2020) show that the refugee inflows boost the underground economy preferring cheaper goods and services provided by informal agents. In a very recent paper, Berdiev (2021) claims the same result based on an extended panel with 120 countries, over the years 1991–2017. He finds that the refugee inflows stimulate the size of the underground economy, particularly in low- and middle-income countries. The leading causes are the barriers to entry into the formal labor market, which stimulates the refugees to supply their labor and entrepreneurial skills in the underground economy.
In addition to quantitative analyses, qualitative analyses also validate the preference of refugees for the informal sector. By using such strategies, Erdoğan and Unver (2015) or Erdoğan (2017) show that Syrian refugees accept the low wages offered in the underground economy. Several sectors stand out in this sense: agriculture, construction, manufacturing, and industry services. Other notable works claiming a positive impact of refugees on the underground economy belong to Likic-Brboric et al. (2013), referring to East-Central Europe (including Latvia), Schneider (2016), in the case of Austria, or Nasser and Symansky (2014), for Jordan.
Few papers evidence the
Although this literature offers an extended picture regarding the impact of refugees on the underground economy, several research gaps have not been yet or partially been covered appropriately. The first gap is the lack of studies devoted to the case of the EU countries, many of them targeting individual EU countries. The second gap is the lack of empirical papers assuming a nonlinear approach. This strategy is invoked by the existence of a “threshold acceptance reaction” in the destination countries. Therefore, different impact magnitudes of refugees on the underground economy can be expected over time. This elasticity strongly depends on the refugees’ behavior, modeled via internal and common EU regulations, climate conditions, the rigidity of markets, or geopolitical contexts.
Given this general literature framework, the paper addresses both gaps mentioned above by offering a complex analysis of the EU’s informal economy under the last large exodus in the history of human migration.
The analysis is performed for 28 EU countries (i.e., Austria, Belgium, Bulgaria, Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Poland, Portugal, Romania, Slovak Republic, Slovenia, Spain, Sweden, and United Kingdom – UK), from 1998 to 2017 (data availability). Therefore, the dataset represents a panel with 28 cross-sections and 20 years (N > T).
The We refer to the “refugee population” offered by World Bank (2022a), including “people who are recognized as refugees under the 1951 Convention Relating to the Status of Refugees or its 1967 Protocol, the 1969 Organization of African Unity Convention Governing the Specific Aspects of Refugee Problems in Africa, people recognized as refugees in accordance with the UNHCR statute, people granted refugee-like humanitarian status, and people provided temporary protection. Asylum seekers – people who have applied for asylum or refugee status and who have not yet received a decision or who are registered as asylum seekers – are excluded.”
We refer to the “refugee population” offered by World Bank (2022a), including “people who are recognized as refugees under the 1951 Convention Relating to the Status of Refugees or its 1967 Protocol, the 1969 Organization of African Unity Convention Governing the Specific Aspects of Refugee Problems in Africa, people recognized as refugees in accordance with the UNHCR statute, people granted refugee-like humanitarian status, and people provided temporary protection. Asylum seekers – people who have applied for asylum or refugee status and who have not yet received a decision or who are registered as asylum seekers – are excluded.”
Three categories of
Migration and refugee flows can be correlated altering the quality of estimation. On the one hand, several pull factors are almost the same, and on the other hand, the common migration routes are intensively used by both illegal migrants and asylum seekers. To tackle this issue, in addition to Medina and Schneider’s (2018) model, the share of immigrant population variable (
Detailed information about the variables and their expected signs is presented in Table A1 in the Appendix, according to Medina and Schneider’s (2018) model. All variables are treated in percentages, so the marginal effect in a regression estimates their elasticities. The only exceptions are
This study of the impact of refugees on the underground economy of the EU follows a nonlinear analysis by using a threshold panel approach, which covers the period from 1998 to 2017. The reason for this assumption is given by the fact that the biological, natural, and social processes generally follow a nonlinear dynamic, with the birth rates and coefficients of interaction between the populations being the main explanations, as Jordanov and Nikolova (2013, p. 69) note.
A large number of refugees represent a serious challenge for receiving countries, which must involve additional financial and human resources to face. In this context, we assume the existence of a refugee’s thresholds generated by disturbances in the management of the reception and integration processes. Therefore, such “critical” levels have deep implications in the informal economy, as the related literature shows. It “describes the jumping character or structural break” (Wang, 2015) in connection to the refugees with the underground economy.
As consequence, the “structural model” of the underground economy of Medina and Schneider (2018), with an extension to incorporate refugees, is empirically tested using Hansen’s (1999) tool in its improved version developed by Kremer et al. (2013). This methodology assumes threshold effects in non-dynamic panels. Such models have several advantages, as Pan et al. (2016) highlight: there is no need for nonlinear equations, the identified thresholds are endogenous, the confidence interval of parameters follow asymptotic distribution, and the statistical significance is based on bootstrap routines. The country-specific effects are also taken into account.
Assuming thresholds in the share of the refugee population (
The double-threshold model is as follows:
The estimation of thresholds is performed by assuming that the sequential estimator is consistent, as Bai (1997) and Bai and Perron (1998) note. Therefore, according to Wang (2015), three main steps depict this procedure:
Step 1: The estimation of the single-threshold model, in order to fit the parameter
Step 2: The estimation of the second threshold and its related confidence interval, based on
Step 3: Re-estimation of the first threshold, given that
Testing for threshold effects means checking whether the coefficients are the same in each regime or not. Herein, two sequences are followed: the first discriminates between the linear and the single-threshold model, while the second compares the single- with the double-threshold model.
In the first sequence, the null hypothesis of no threshold effect (linear model) or the alternative one of threshold effect (nonlinear model) are:
If the null hypothesis is rejected in the simple-threshold model, the double-threshold estimator is further checked, according to a new
According to Nguyen and To (2016), in the double-threshold model, the null hypothesis tests the existence of one or two thresholds. If the null hypothesis is rejected, then the double-threshold model is more appropriate. Further, the process is similar for the models with more than two thresholds.
The refugees represent a special case of migrants who leave their countries seeking international protection. Often this desiderata is overlapped with many other classical pull factors, such as the underground economy (Talani, 2019). Hence, a potential endogeneity issue by reverse causality type can arise between the share of the refugee population and the underground economy. More precisely, the share of the refugee population can influence the size of the underground economy but also the underground economy can act as a pull factor for the refugee flow. Moreover, several other variables can be suspected by endogeneity through the reverse impact of the underground economy on them, generating bias in estimations:
To deal with any endogeneity concerns, the estimations are conducted by using an improved version of Hansen’s (1999) tool proposed by Kremer et al. (2013) The estimations are performed based on STATA code freely offered by Diallo (2020).
The estimations are performed based on STATA code freely offered by Diallo (2020).
Working with many explanatory variables, the variance inflation factors (VIF) test and matrix of correlation offer support to detect possible multicollinearity. The robustness check is done by sequentially entering the control determinants in the constructed models.
By using the threshold dynamic panel method proposed by Kremer et al. (2013), a set of models offers the results supporting the main interpretations and conclusions.
The output clearly reveals that the null hypothesis of linear estimation (i.e., no threshold effect) is rejected at the 10% level of significance; the
The outcomes reveal that both threshold
The control variables are significant in all estimations, with the exception of the tax burden, institutional quality, and the size of the government. Per-capita GDP, unemployment, trade openness, and the share of the immigrant population are negatively correlated with the size of the underground economy. The findings partially confirm the results of Medina and Schneider (2018), as the unemployment rate has the opposite sign in respect to the expected one. The negative link between unemployment and the underground economy aligns with Gora et al. (2009). A generous unemployment allowance can offer a comfortable standard of living, stimulating the employment in the official labor market given the attraction of social security benefits. The negative sign of the share of the immigrant population in respect to the underground economy does not confirm the mainstream literature (e.g., Crush et al., 2005; Landau, 2006; Bracco and Onnis, 2016; Venturini and Villosio, 2008). This is in accord with Schneider and Badekow (2006), who claim that permissive constraints created by public institutions and bureaucratic apparatus can be a significant disincentive for immigrants to access the underground economy.
Not least, the significance and positive sign of a lagged underground economy reinforce the idea that the underground economy exhibits a short-memory process in the EU countries. More precisely, the current underground economy depends on its past dynamic (e.g., a positive past change of underground economy stimulates its current level probably because of “permissive” signals come from the market).
For the full Model (7), the threshold regime coefficients have values of 239.3 and −0.826, respectively. The estimated threshold is 0.00572, meaning that refugees make up 0.572% of the total population in the receiving country. Herein, the increase in the share of the refugee population has a nonlinear effect on the underground economy, with different signs and slopes impacts, as Figure 2 shows.
Figure 2 illustrates that the increase in the share of the refugee population strongly boosts the underground economy until a peak after that slightly falls. More precisely, the inflow of refugees stimulates the underground economy at a low level of contingents (i.e., LR statistic in Figure A1 in Appendix reveals a low threshold level), contracting it as soon as a critical point is overcome.
The results confirm Maloney’s (2004) vision, where an asymmetrical effect of the share of the refugee population on the underground economy is an adaptation issue rather than a barrier one. The positive slope is in line with Altındağ et al. (2020), Berdiev et al. (2020), or Berdiev (2021), while the negative effect reinforces the results of Akgündüz et al. (2018), or Cengiz and Tekguc (2018).
The study is limited by the data availability; further research targeting the extension of the panel in terms of the period can be performed as soon as the new dataset is officially available.
This study explores the impact of refugees on the size of the underground economy in all 28 EU countries, for the period 1998–2017, using a dynamic panel threshold approach. The main results uncover a nonlinear link between refugee inflows and the size of the underground economy following an inverted V-shape, with its magnitude depending on the level of the share of the refugee population.
The positive impact of refugees on the underground economy under the threshold suggests a sort of labor market rigidity in the absorption of refugee inflows even under good macroeconomic context. Herein, as the labor supply does not significantly increase, competition in the labor market is still low, which does not reduce labor costs. This encourages the refugees to prefer the underground economy to the detriment of the official one in terms of both labor and consumption (i.e., lower prices).
With large refugee contingents, the labor market slowly adapts in demand, with supply becoming very generous at lower labor costs. In parallel, growth and international trade can support this process, expanding the economy. New labor and entrepreneurial opportunities arise in this case for newcomers, stimulating consumption in the official market. As a consequence, the budgetary constraint relaxes, counteracting any unemployment allowance pressure. This can open the door for better regulation of other immigrants, finally helping regulated persons to act in the official economy.
Several policy implications can be identified. Accepting a high number of refugees, EU policymakers must control the underground economy by maintaining economic growth and international trade flows. Special attention must be paid when refugee contingents are small. Budgetary policy requires decent unemployment allowances to counteract the propensity of jobless persons to escape into the underground economy. The past dynamic of underground economic activity seems helpful to anticipate future activity, while illegal immigration should be strictly controlled.
Tests for threshold effects
|(10%, 5%, 1% critical value)||(13.0142; 14.8043; 19.9052)|
|Test for two thresholds|
|(10%, 5%, 1% critical value)||(10.8853; 14.6780; 20.4151)|
|Test for triple thresholds|
|(10%, 5%, 1% critical value)||(12.9204; 16.9980; 27.0286)|
Matrix of correlation
Results of single threshold models
||−0.291 (0.718)||−0.866 (0.822)||−0.369 (0.663)|
|0.015 (0.011)||0.038* (0.021)||0.004 (0.011)||0.027 (0.021)|
|Wald χ2||3,399.1 Pr.=0.000||1,565.5 Pr.=0.000||1,062.8 Pr.=0.000||7,305.4 Pr.=0.000||2,287.3 Pr.=0.000||5,389.2 Pr.=0.000||8,689.12 Pr.=0.000|
|Number of observations||532||532||532||532||532||532||532|
|Number of groups||28||28||28||28||28||28||28|
Description of variables and their expected signs
||Level of underground economy as percentage of GDP.||%|
|Share of refugee population
||Refugee population as percentage of population in destination country.||%||±|
|Tax burden (
||Tax revenues as percentage of GDP.||%||+|
|GDPc (constant 2015 US$).||US$||−|
||Total of unemployed persons as percentage of total labor force.||%||+|
|Institutional quality (
||Captured as Rule of Law indicator, with range goes from −2.5 (weak) to 2.5 (strong) governance performance.||Index||−|
|Size of government (
||Volume of government expenses as percentage of GDP.||%||−|
|Trade openness (
||Volume of imports plus exports as percentage of GDP.||%||−|
|Share of immigrant population (
||Total immigrants as percentage of population in destination country.||%||±|
VIF test results for multi-collinearity detection