The Influence of Diaspora Engagement and Government Policies on Remittance Inflows: A Panel Data Analysis of 182
Countries between 1990 and 2014
Msc International Business and Management University of Groningen, Faculty of Economics & Business
Robbert Jenne Cazemier Master Thesis 19-02-2016 Supervisor: Dr. D.H.M. Akkermans
Co-assessor: Dr. R. de Vries
Robbert Jenne Cazemier Feanbei 24 9231NB Surhuisterveen Student number: 2328054 E-mail: r.j.cazemier.2@student.rug.nl
2 TABLE OF CONTENTS
LIST OF FIGURES AND TABLES ... 4
LIST OF ABBREVIATIONS ... 5
ACKNOWLEDGEMENT ... 6
ABSTRACT ... 7
1. INTRODUCTION ... 8
2. LITERATURE REVIEW ... 10
2.1MIGRATION ... 10
2.2REMITTANCE INFLOWS ... 12
2.3DETERMINANTS FOR REMITTANCE INFLOWS IN THE RECEIVING COUNTRIES ... 13
2.3.1PERSONAL DETERMINANTS (MICRO-ECONOMIC LEVEL) ... 14
2.3.2COUNTRY LEVEL DETERMINANTS IN RECEIVING COUNTRIES ... 14
2.3.3GOVERNMENT POLICIES ON REMITTANCES IN COUNTRY OF ORIGIN ... 19
3. METHODOLOGY ... 24
3.1SAMPLE ... 24
3.2DEPENDENT VARIABLE ... 24
3.3INDEPENDENT VARIABLES ... 25
3.3.1GDP PER CAPITA ... 25
3.3.2PUBLIC HEALTH EXPENDITURE PER CAPITA ... 25
3.3.3POLITY2 INDEX ... 25
3.3.4GOVERNANCE QUALITY INDEX ... 25
3.3.5REFUGEE SHARE AS A PERCENTAGE OF THE EMIGRANT STOCK ... 26
3.3.6LEVEL OF VIOLENCE IN A COUNTRY ... 26
3.3.7DIASPORA ENGAGEMENT INDICATOR ... 26
3.4DUMMY VARIABLES ... 27
3.4.1DEVELOPING COUNTRIES ... 27
3.4.2MINI STATES ... 27
3.4.3TAX HAVENS ... 27
3.5EMPIRICAL MODEL ... 28
3.6MODELS ... 29
4. EMPIRICAL DATA & ANALYSIS ... 30
4.1DESCRIPTIVE STATISTICS ... 30
4.2QUALITY OF DATA ... 31
4.3MULTI-COLLINEARITY ... 32
4.4MODEL TESTS ... 33
5. EMPIRICAL RESULTS ... 35
5.1REGRESSION RESULTS ... 35
5.2MODEL FIT ... 37
5.3POST ESTIMATION RESULTS ... 38
5.4ANALYSING THE RESULTS OF THE REGRESSION ... 39
5.4.1REGRESSION RESULTS AND HYPOTHESES TESTING FOR THE REGRESSORS ... 39
3
5.4.2REGRESSION RESULTS FOR THE DUMMY VARIABLES ... 41
5.5RESULTS TRANSLATED BACK INTO THE EMPIRICAL MODEL ... 42
6. CONCLUSION & DISCUSSION ... 44
6.1CONCLUSION ... 44
6.2DISCUSSION ... 44
6.2.1LIMITATIONS ... 45
6.2.2SUGGESTIONS FOR FUTURE RESEARCH ... 47
REFERENCES ... 48 APPENDIX I – WORLDWIDE REFUGEE AND EMIGRANT STOCK ...
APPENDIX II – TOP 20 COUNTRIES DIASPORA SIZE (ABSOLUTE) ...
APPENDIX III – RELATIVE CHANGE IN REMITTANCE INFLOW AND EMIGRANT STOCKS ...
APPENDIX IV – REMITTANCES RECEIVED IN 2013 COUNTRY TOP 37 ...
APPENDIX V – COUNTRY LISTS ...
APPENDIX VI – SECRECY JURIDICTIONS ...
APPENDIX VII – DIASPORA INSTITUTIONS PER COUNTRY ...
APPENDIX VIII – LIST OF VARIABLES, SOURCES, CODES ...
4 LIST OF FIGURES AND TABLES
Figure 1 - Emigrant stock and refugees worldwide
Figure 2 - Twenty countries with the largest diaspora populations and flows Figure 3 - Development of remittance flows worldwide
Figure 4 - Relative remittance inflows Figure 5 - The Rise of Diaspora Institutions
Figure 6 - Value distribution of GDP_PCAP versus log(GDP_PCAP) Figure 7 - Different transformations compared for RI_EMI
Figure 8 - Graphical correlation between independent variables Figure 9 - Distribution of residuals versus normal distribution Figure 10 - Qnorm (QQ-plot) and Pnorm (PP-plot) of the residuals
Table 1 - Absolute Remittance inflow, remittance per emigrant and remittance inflow ratio Table 2 - Secrecy Jurisdictions
Table 3 - Top 20 countries FSI Value, Secrecy, & Global Weight Score Table 4 - Ministry levels of Diaspora Units
Table 5 - Sub ministry levels of Diaspora Units Table 6 - Diaspora Institutions at the Local Level Table 7 - Diaspora Institutions at the National Level Table 8 - Quasi-governmental diaspora institutions Table 9 - List of all variables used in this research Table 10 - Overview of models
Table 11 - Descriptive statistics
Table 12 - List of extreme refugee percentages from the dataset Table 13 - Correlation matrix for all independent variables Table 14 - Output of VIF test
Table 15 - Hausman Test results in Stata
Table 16 - Regression Results Fixed Effect Models Table 17 - Regression Results Random Effect Models Table 18 – Test outcomes for all Fixed Effects Models Table 19 – Test outcomes for all Random Effects Models
5 LIST OF ABBREVIATIONS
FDI Foreign Direct Investments
GDP Gross Domestic Product
ILO International Labour Organization IOM International Organization for Migration ODA Official Development Assistance
OECD Organisation for Economic Co-operation and Development UNHCR United Nations High Commissioner for Refugees
6 ACKNOWLEDGEMENT
This master marks the end of my master program at the University of Groningen. Writing this thesis was a time consuming and challenging part of my study. Nevertheless, I learned some valuable insights and skills, which I hope to employ in my future career.
This work, however, would not been possible without the support of many people. First, I would like to thank my supervisor Dr. D.H.M. Akkermans from the University of Groningen for his valuable guidance, patience, support, conversations and feedback. Secondly, I would like to thank my family, friends, and girlfriend for their time and support during my university career. Although studying and writing a master thesis is an individual task, this task would not have been possible without the support of these people.
February 2016,
Rob Cazemier
7 ABSTRACT
The total amount of remittances sent back to their home countries by emigrants has grown rapidly during the last decade, much faster than the steady growth of the total number of emigrants worldwide. In 2014, a total amount of 583 billion US$ was remitted by a total number of 238 million emigrants. For some countries, these remittance inflows are more substantial than the development aid they receive. Or even come close to the GDP for some countries. This explains the growing interest of governments in their emigrant stocks (also referred to as diasporas), and why governments actively develop policies and institutions to regulate the remittance inflows into their economies.
This research aims to assess which country specific determinants influence these remittance inflows, and especially test whether the level of diaspora engagement in a country is related to the amount of remittance inflow. In doing so, this research answers the following question: To what extent does diaspora engagement influence the differences in remittance inflows between countries, or can these differences be explained by other country specific factors? By answering this research question, this research addresses a research field (the tapping perspective) introduced by Gamlen (2013), and explores whether this tapping perspective is can be confirmed by analysing data on remittances and emigrant stocks. For this purpose, data was obtained from the databases provided by the World Bank, the UN Department of Economic and Social Affairs, and others. The analysis was conducted using panel data for 182 countries over the period from 1990 until 2014. For this empirical analysis, a model was built and tested using both random and fixed effects regression.
The findings support the conclusion that countries with higher levels of diaspora engagement indeed receive higher remittance inflows, although causality was not tested. Other findings show that other country specific determinants, such as GDP, the level of violence, the refugee share of the emigrant stock and a country’s political system are also significantly related to differences in the remittance inflows between countries, and between years per country.
Keywords: Remittances, remittance inflow, emigrants, diaspora engagement, countries
8 1. INTRODUCTION
International migration and its economic, social, and political implications currently receive a lot of attention from researchers, scholars, and policy makers worldwide. One recent cause of this growing attention is the rapidly growing influx of Syrian refugees into Europe and other countries in the Middle East. Already four million refugees left Syria since the conflict started five years ago, and their numbers are still growing (Edwards, 2015).
Although this number of emigrants is substantial, the total worldwide migrant stock was estimated at 232 million in 2013 (UN Department of Economic and Social Affairs, 2013).
Despite the fact that this comprises only 3.2 per cent of the total world population, the average annual growth has been 2.5 per cent from 2000 until 2013. So migration has been a continuously growing phenomenon for decades.
Among the most important economic effects of migration are remittances1. Remittances are money transferred back to their families in the country or origin by emigrants, also referred to as the diaspora of a country. Although remittances are small scale and individual by nature, the totals are of economic importance. According to the bilateral remittance data provided by the World Bank (2015), the global remittance inflow amounts to 583 billion US$ in 2014, of which 414 billion US$ flows into developing countries. Compared with the total amount of FDI net inflow of 681 billion US$ (UNCTAD, 2015) and a total amount of 161 billion US$
Official Development Assistance in 2014 (OECD, 2015), remittance flows are a significant factor on a macroeconomic scale for developing countries. Due to these impressive and growing financial flows, and their potential for economic development, governments and policy makers in both sending and receiving countries contribute substantial amount of attention towards regulation and policies.
This research focuses on differences in remittances between countries corrected for the number of migrants (emigrants), and investigates the influence of several determining factors for these differences. In particular the refugee share in the total migrant stock, and the influence of diaspora institutions and government policies on remittances inflows are
1 “household income received from abroad, resulting mainly from the international migration of workers (Yang, 2011)”.
9 investigated. In doing so, this research addresses a gap in the literature introduced by Gamlen, Vaaler & Rossouw (2013), Gamlen (2014), and Hatton (2014) all state that more research is necessary on the impact of diaspora engagement on remittances. And especially, how diaspora engagement moderates international financial flows such as remittances. Gamlen et al. (2013) theorize that governments actively engage with their diasporas in order to ‘tap’
financial resources from more developed countries. This research will further investigate this
‘tapping perspective’ by exploring the relation between government diaspora engagement and remittance inflows.
Therefore, the following main research question is formulated:
To what extent does diaspora engagement influence the differences in remittance inflows between countries, or can these differences be explained by other country specific factors?
This paper is structured as follows. First, relevant theories will be reviewed and discussed.
Based on this, hypotheses are formulated that will be tested in the results section. Section three will discuss some of the findings and provide some suggestions for future research.
Conclusions are drawn on the hypotheses based on these findings in the final section.
10 2. LITERATURE REVIEW
2.1 Migration
Migration is a natural process and has been going on since the history of mankind. The size, speed, distance and composition of these migration flows, however, changed substantial due to technological developments, increased mobility, and globalization. Although migration can also occur within country boundaries, this research will focus on international migration.
International migration is conceptualized as a move from an origin to a destination, or from a place of birth to another destination across borders (country of destination).
According to the UN Department of Economic and Social Affairs Statistics Division (1988), an international migrant is defined as “any person who changes his or her country of usual residence”. Although this definition incorporates all movements across international boundaries, most research and papers make a clear distinction between several types of migrants. Examples are: transit migrants (Suter, 2012), regular vs. irregular migrants (King, 2012), temporary migrants (Dustman & Weiss, 2007), seasonal migrants (Khandker, Khalily
& Samad, 2012), labour migrants (ILO, 2014), and refugees (UNHCR, 2015).
For this research, migrants are divided into two main categories: immigrants and emigrants.
Since this research focuses on emigrants and the remittance inflows they generate in their home countries, immigrants and remittance outflows are not included. A common term used to describe all types of emigrants stemming from a country is a ‘diaspora’. Although different definitions of diaspora are used in literature (Cohen, 2008; Collier, 2013; Dilip & Rahta, 2011; Ratha & Plaza, 2011; Vertovec, 2009), the main idea is that communities living in another country retain certain connections to their country of origin. Brubaker (2005) distinguishes three core characteristics of diasporas: dispersion in space; orientation to homeland; and boundary maintenance (by preservation of distinctive identity). This research uses the term ‘emigrant stock’ instead of ‘diaspora’ for all people originating from one country and residing in the rest of the world.
A migrant ability to relocate is affected by the presence of political boundaries and regulation.
In particular, regulation about entry and exit. Breunig, Cao & Luedtke (2012), for example, explains that a country’s regime affects both the volume of immigration and the volume of emigration. As they explain, a regime’s willingness to allow migrants depends on: its ability
11 to limit rights granted to workers, and the necessity to respond to anti-immigrant public demands. Their findings also show that non-democracies (e.g. autocracies) grant fewer rights (e.g. limited social and political rights) to workers, and often have favourable guest worker programmes, so their labour market are more flexible, allowing them to make use of labour immigrants at will. Some countries even develop so-called ‘dual labour markets’, which means that the least attractive jobs are primarily reserved for immigrants (Tsuda et al., 2013).
Political regimes also differ in their willingness to grant exit to their citizens. While some authoritarian regimes (e.g. the Democratic Republic of Korea) restrict emigration, all democracies allow both citizens and residents the freedom to leave. In fact, many emerging democracies have policies in place that actively encourage citizens to work abroad. Countries also cooperate with other countries to create a borderless area in which people can move freely from one country to another country within that area. One example of this type of collaboration is the Schengen Agreement (signed in 1985, effective in 1995).
Researchers often use push or pull models to explain migration. Push factors are associated with low wages, poverty, civil conflict, political instability, social unrest, the paucity of jobs, and unfavourable working environment. Pull factors, on the other hand, are associated with factors such as the labour demand, health facilities, better wages, improved standard of living, and political and religious freedom (Azam, 2015). Although safety and living conditions are important factors, in 2013 only 11.7 million migrants can be characterized as refugees (UNHCR, 2013). So the major part of all migrants is led by economic reasons.
Empirical research also shows that the majority of migrants is not motivated by political factors (e.g. wars, political stability and democracy) in their choice of departure and destination, but instead are motivated by economic prosperity at the country of destination (Collier 2013; Leblang, Fitzgerald, & Teets, 2007; Masey et al, 1993).
The development of the worldwide emigrant stock and refugees is shown in in Figure 1 Appendix I over the period 1990-2013. The data shows that the number of refugees in the world is relatively constant, and only comprises a small but significant portion of the migrant stocks. This indicates that only a minor part of the international migrant stock can be characterized as refugees according to the Refugee Convention. So the majority are labour emigrants, who follow the normal procedures for emigration to the country of destination.
12 A more detailed graph (Figure 2) on diaspora size and emigrants flows for the top twenty countries in the world in absolute numbers can be found in Appendix II. As can be seen in Figure 2, India is currently the country with the largest diaspora in absolute numbers of people, followed by Mexico and the Russian Federation. These three countries are relatively stable. However, this list also includes countries, which are relatively unstable (e.g.
Afghanistan, the State of Palestine, and the Syrian Arab Republic).
2.2 Remittance inflows
As can be seen in Figure 3 below, the total worldwide remittance inflow in absolute numbers has grown at a remarkable fast pace from 2001 onwards.
Figure 3 - Development of remittance flows worldwide (source: World Bank, 2015)
These worldwide remittance inflows grow much faster than the worldwide emigrant stock (see Figure 4, Appendix III), so the average remittance per emigrant is growing.
The differences between countries are also interesting. The top five receiving countries in absolute numbers are India, China, the Philippines, France and Mexico. But these counties also have a large population and a large diaspora (see Table 1, Appendix IV). Dividing the remittance inflow by the number of emigrants (diaspora size) produces a totally different ranking with French Polynesia, New Caledonia, Bermuda, and Luxembourg as countries with the highest remittance inflow per emigrant (as shown in the same table). Since it is well
0 100.000 200.000 300.000 400.000 500.000 600.000
1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 Remittance inflow
13 known that these countries are characterized as ‘Tax Havens’, a relation between the remittance inflow per emigrant and tax havens in general can be expected. This can also be explained, because ‘workers compensation’ is a component in the remittance inflows, and some tax havens are well known for tax evasion on income earned abroad. Some other tax havens (e.g. Cook Islands and the Bahamas) do not provide any remittance data, so this list could have been longer. But due to the lack of data, these countries are excluded from this research.
When the remittance inflow per emigrant is corrected for the standard of living in the receiving country, the country ranking shifts again. By dividing the remittance inflow per emigrant by GDP per capita of each country, the ‘remittance inflow ratio’ is obtained. Using this ratio shows the relative weight of the remittance inflows per emigrant for a specific country. A ratio of 1.00 means, that the remittance per emigrant equals the GDP per capita.
Only 36 of the 182 countries included in this research, have a remittance inflow ratio higher than 1. As shown in Table 1, a totally different set of countries appears on top, such as Tajikistan, Nigeria, Nepal, and Madagascar. These countries receive relatively high remittances per emigrant.
This research tries to explain the differences between countries over a number of years, whereas many other researches focus on the economic effects of remittances. Examining the differences in ranking as discussed, the first question is which parameter should be used to rank the countries when analysing the differences. Obviously, the absolute remittance inflow per country does not reflect the differences between countries very well. For this research, the remittance inflow per emigrant per year is used to compare countries. Analysing these large differences between countries raises the question: which causes can be found to explain these differences between countries? Since the remittance inflow ratio is not used to compare countries, the GDP per capita is used as an important independent variable to explain the differences between countries.
2.3 Determinants for remittance inflows in the receiving countries
There are numerous possible causes, which might explain the large differences in remittance inflows between countries. Although remittances are personal by definition, the reasons for emigrants to remit money to their home country differ, and can be divided in three categories:
personal determinants, country level determinants, and government policy determinants.
14 2.3.1 Personal determinants (micro-economic level)
The reasons, determinants and motivations for migrant workers to send money back to their country of origin have been examined extensively (Bollard, Mckenzie, Morten, Rapoport, 2009; Stark, 1995; Yang, 2011), An overview of some theoretical and empirical concepts on migrant’s remittances are provided in a paper by Rapoport & Docquier (2005). In their article, theories on motives such as ‘altruism’, ‘insurance’, ‘exchange’, ‘attachment to the homeland’
and the ‘investment motive’ are explained and summarized.
The motives and determinants mentioned here might also be influenced by the personal characteristics of the remittance sender. Faini (2003), for example, found that more educated migrants are likely to remit less because their migration is often more permanent. The income level in the country of residence is another factor, which might determine the size of the remittance send back. Although it is reasonable to believe that emigrants in developed countries earn more money and therefore send back higher remittances, no research clearly describes this relationship. The duration of the emigration might also influences the size of the remittance flows. This is based on the remittance decay hypothesis, which states that the longer migrants live away from home, the less likely it is for them to remit money to their country of origin (Grigorian & Melkonyan, 2008).
To investigate the personal determinants, more detailed (bilateral) data is required on each emigrant stock in each country of residence, and this research only focuses on comparing remittance inflow into countries. Therefore, these personal determinants are omitted in this research, and this research focuses on country level determinants.
2.3.2 Country level determinants in receiving countries
The circumstances in the country of origin are also likely to have an effect on the remittances sent by their emigrant stock. For example, access to financial services, the provision of public goods such as electricity and water supply, and income level in the country of origin could be important determinants for the decision of emigrants to remit money to their families in the country of origin.
One of the most distinguishing factors between countries is GDP, and GDP per country is a very common variable within all research done on remittances. For this research, GDP per capita is used to compensate for the relative size each country and its economy. The
15 difference between remittance inflow per emigrant and GDP per capita show the relative importance and size of the remittance inflow for each country. On one hand, worker migrants stemming from countries with a high GDP per capita are expected to remit higher incomes because its not likely that they emigrated to earn less than in their home country. On the other hand, the families residing in countries with high GDP are less dependent on the remittance sent by their emigrant relatives. Moreover, emigrants from countries with high GDP are more likely to take their family with them when they emigrate. Therefore, the relation between GDP per capita and remittance inflows per emigrant is less predictable than might be expected, which can also be seen in the country rankings in Appendix IV Table 1. For this research, the following hypothesis is used:
H1: Countries with higher GDP per capita attract higher remittance inflow per emigrant
In developed countries, a large portion of public spending goes to healthcare. Developing countries spend considerably less. But health problems and care for their elders are expected to be important incentives for emigrants to remit, especially when the government in their country of origin does not provide these public services. Therefore, a negative relation can be expected between public healthcare expenditure and remittance inflow per emigrant. In order to compare between countries, this research uses public healthcare expenditure per capita as a determinants for remittance inflows. Based on this, the following hypothesis is formulated:
H2: Countries with higher public healthcare expenditure per capita attract lower remittance inflow per emigrant
Remittances can also compensate the families and communities of emigrants for a lack of stability and weak government. Remittance inflows enable local families and communities to organize public goods and stability that otherwise would be absent. In most developing countries, governments rely heavily on foreign aid and loans to provide their citizens with a minimum of public goods and services. In some cases, the government even fails to provide a minimum level of public goods (e.g. education, healthcare, and roads). To counter this, emigrants provide or investment some of their financial resources into collective efforts (public goods). This is in line with the altruistic and investment motive mentioned earlier.
16 Ebeke (2012), for example, found a negative relationship between remittances and the provision of public goods (measured in public spending per social sector) by governments. As he explains this is caused by two reinforcing effects. The public moral hazard (governments divert resources because remittances can replace these resources), and the second effect is the household moral hazard (households do not have an incentive to monitor their government since they are insured by the remittances send from abroad).
However, researching the effect of public expenditures in combination with GDP and healthcare expenditure per capita is problematic, since there is a very strong relation between GDP and public expenditure. Furthermore, the variable GDP per capita is already very unevenly distributed over countries in the world. In other words, there are few countries with high GDP per capita and those countries also have high public expenditure. The consequence is that the effects of public expenditure and GDP on remittances cannot be distinguished.
Therefore only GDP per capita is used.
In fact, the same goes for public healthcare expenditure per capita, but the relation between GDP and healthcare expenditure is expected to be weaker. This research also distinguishes between developed and developing countries, to check for differences in the results between these two groups of countries.
Comparison across countries depends on the correctness of the statistics per country. Because a considerable number of countries can be characterized as “mini states”, the statistics for these countries are susceptible to measurement errors caused by the small number of people involved. Therefore, a check will be done if the results differ for the group of mini states versus other, bigger countries. For this research, a country is characterised as a mini state if it has a population smaller than 500.000 people.
Besides the economic differences between countries and economic development within countries, there also political and governmental differences and developments. Although some argue that politics and economics cannot be completely separated, this research incorporates also some specific political determinants. Since countries around the world obviously differ on political and governmental aspects, their effect on remittance inflows has to be considered.
17 One common and often used measure determining political regime is the Polity index, which is determined by the Polity IV project of the Center for Systemic Peace. The polity2 index indicates how the political regime in a country can be characterized (ranging from +10 for a stable democratic regime to -10 for a totally authoritarian regime). Although this index is a valuable starting point, a paper by Norris, Frank, Ferran, & Martinex I Coma (2013) indicates, that this index has some limitations when measuring political regimes. As they explain,
“evaluating the quality of elections is equally important” to determine whether regimes are democracies or autocracies. In 2013, they introduced the Electoral Integrity Project, which also provides data on a huge set of political variables for comparing countries in more detail across times. For further research it might be valuable to compare the scores with the Polity IV project and to specify the differences in more detail. This research includes the Polity2 index, also because this index is available for many countries and the entire research period, and this index seems to be the best available determinant to characterise the differences in political regimes between and within countries. The relation between the different political regimes of countries and their remittance inflow is, however, ambiguous. Although literature reveals a relation between political regimes and emigration, the remittance inflow per emigrant is probably not influenced by the country’s political regime in itself. Therefore, the following hypothesis is formulated for this variable:
H3: A country’s polity2 index does not have a significant influence on the remittance inflow per emigrant
Nevertheless, regimes do implement legislation and governance in countries, and thereby determine the financial liberties for their citizens. For instance, the freedom for citizens to own houses or to run a business is determined by the government of a country. Gwartney, Lawson & Hall (2011) define ‘economic freedom’ as “the degree to which a jurisdiction’s policies and institutions protect the rights of corporation and individuals to pursue their economic objectives without interference”.
Civil liberties are likely to have an effect on remittance inflows, since the emigrant’s remittances will go to their families. According to Bang, Mitra, & Wunnava (2015), for example, financial liberalization in the country of origin has a positive and immediate impact on remittances sent by emigrants. In their paper they included 84 countries over the period 1986 till 2005. There are several governance indicators, but this research uses the WGI
18 indicators provided by the World Bank and produced by Kaufmann & Kraay (2015). From their six main indicators, three are used for this research to indicate the quality of governance per country. More specific, the ‘Government Effectiveness’, the ‘Regulatory Quality’ and the
‘Rule of Law ‘ indicators are combined into one quality of governance indicator. Regulatory quality is the ability of the government to formulate and implement sound policies and regulations that permit and promote private sector development. Increasing government quality, on the other hand, also implies an increasing level of public goods and services, reducing the need for remittances. The assumption in this research is that a higher government quality leads to higher remittance inflows.
H4: A higher governance quality leads to lower remittance inflow per emigrant
One special aspect of governance is corruption. The WGI also provides a special indicator
‘control of corruption’. The influence of corruption on remittances has also been investigated, but led to ambiguous results. Weng, Woo, Chen, Ho, & Horowitz (2015), found that public trust in government and nongovernment organizations is essential to citizens’ willingness to donate money to those organizations. Others describe a political “tug-of-war” over remittance benefits (Tyburski, 2014). Two perspectives lead this debate. The first perspective, the substitution perspective, argues that remittances divert government spending away from public goods provision and to spending money on private goods and ‘patronage’. Citizens enable this, according to this view, by sending remittances and making themselves less reliant on government to provide public goods (Abdih, Chami & Dagher, 2011). This causes the citizens to hold the government less accountable, and consequentially government can freeride and use more resources for their own interest. The second perspective is referred to as the accountability perspective (Tyburski, 2012). This view identifies incentives for migrants and their families in the country of origin to use remittances as policy resources to reduce corruption from the bottom up. Because of these two contradictive views, and the fact that both describe an indirect relation between remittances and corruption, no determinant for corruption is included in this research.
Another important difference between countries is the presence or absence of violence, affecting both emigration and remittances. This is certainly a dominant factor for the number of refugees from a country, but also an important reason for emigration. As mentioned earlier, the total number of refugees worldwide is fairly constant, but the numbers per country vary
19 over time. And since refugees are not expected to be economically active, it seems reasonable to assume that the share of refugees in the diaspora of a country has a negative effect on the remittance inflow per country. Therefore the following hypothesis will be tested:
H5: Countries with a higher proportion of refugees in their emigrant stock have a significantly lower remittance inflow per emigrant
The only problem is that countries with a lot of violence have other things on their mind than statistics; so lacking data could be a problem. The presence of violence is also measured by the Center of Systemic Peace, which has recorded all major episodes of political, civil, and international violence (from 1946 onwards and most of the countries in the world). For this research, the total scores of civil violence (within a country) and international violence (between countries, either with neighbouring countries or elsewhere, e.g. the USA in the Gulf War) are combined to one total score for the presence of violence in a country over time. The assumption is that the uncertainty resulting from violence will lead to lower remittance inflows. Therefore the following hypothesis is formulated:
H6: Emigrants send lower amount of remittances to their home countries during violent periods
2.3.3 Government policies on remittances in country of origin
Because remittance inflows surpass official development assistance in size for some countries, governments have come to realize that they can benefit from higher remittances.
Literature reveals two perspectives on these policies. The first, positive, perspective emphasizes the benefits that remittance inflows have for the country, the communities and the families. The second, pessimistic, perspective highlights the risk of large foreign capital inflow on the economy and governance. This pessimistic view is not unrealistic, since economies can really suffer from large remittance inflows (e.g. Dutch Disease). Vacaflores, Kishan, & Trinidad (2011) also argue that remittances give rise to inflation. However, this research is not aimed at analysing the economic impact of remittances in the recipient countries, so this category of policies is not included in this research.
Both the positive and the negative effects of remittances are motives for governments to actively develop policies on remittances. Policy development on remittances is an example of
20 active governance, directly aimed at the remittance inflows. This differs from all the determinants described in the previous paragraph, which merely characterize the situation in a country at a specific point in time. Because these government policies are directly aimed at remittances, this research distinguishes them from the more passive determinants described in the previous paragraph.
Numerous examples of all kinds of policies on emigration and remittance policies can be found in literature. Ragazzi (2014), for example, distinguishes five categories of diaspora policies: symbolic policies, religious and cultural policies, social and economic policies, citizenship policies, and state and bureaucratic control.
As described by Gamlen et al. (2013), a lot of research in political science and international relations has focused on immigration policies made by migrants’ country of destination, while overlooking emigration policies made by the emigrant’s country of origin. One particular policy is the establishment of ‘Diaspora Institutions’, varying from ministry level to local government levels. Gamlen (2014) defines diaspora institutions as “formal state offices dedicated to emigrants and their descendants”, and explains that these engagement initiatives take several forms (ranging from Ministry to local decentralized institutions). Figure 5 below graphically shows the rise of these institutions worldwide. As can be seen in the graph, these institutions started to grow significantly from 2000 onward.
Figure 5 – The Rise of Diaspora Institutions, adapted from Gamlen (2014)
21 Gamlen et al. (2013) provides three explanations for the emergence of these diaspora institutions, namely: the tapping perspective, the embracing perspective, and the governing perspective. The tapping perspective focuses on the country of origins’ material interest in investments, connections, remittances, skills, and strategic capabilities as drivers for engagement. The second perspective, the embracing perspectives, emphasizes the role of nationalists and transnationalists to re-incorporate lost members of the nation. The last perspective, the governing perspective, describes how diaspora institutions are driven by
“efforts to form a coherent but decentralized system of global migration governance (Gamlen, 2014)”. From this perspective, governments try to make their diasporas
“governable” through diaspora policies, despite the fact that these emigrants are beyond their legal jurisdiction. This research will mainly focuses on the tapping perspective, so the other two perspectives will not be included in this research.
According to Foucault (1982), The capacity to exercise power on citizens abroad consists in three types of relationships: relations of power, relationship of communication, and finalized activities. First, states produce a relationship of communication (e.g. a systems of symbols or signs). Secondly, states aim to develop ‘objective capacities’ for the realization of power by establishing specific diaspora institutions. Finally, governments also exercise power via transnational citizenship (finalized activities). In this case governments extend rights and rights to extract obligations to non-residents.
One example of diaspora engagement via policies is dual citizenship. Citizenship can be based on place of birth (referred to as jus soli), by the parental origins (jus sanguinis) or both.
Most emigrants, however, cannot access citizenship through birth, so citizenship must be acquired through a process called ‘naturalization’. Globally, there exist differences in countries’ naturalization requirements and citizenship laws, but minimum requirements usually involve a period of legal residency, and a demonstration of some knowledge about the country and dominant language (Bloemraad, 2006). Some countries even exclude citizenship rights for immigrants (e.g. countries in the Golf region). According to Basch, Schiller, &
Blance (1994), migrants “through their daily life and social, economic and political relations create social fields that cross national boundaries”. Via hometown associations (HTAs), political ties, and religious affiliation these migrants retain ties to their country of origin.
22 Diasporas themselves can also push particular agendas on policy makers in both the host and or sending countries. Dual citizenship (or multiple citizenship) is one manifestation of this type of deterritorialized membership. The number of states using these forms of citizenship has been growing over the years. Sometimes caused by international conventions, and other times because of policies in the country of origin. The way countries apply these rules can differ per country. Gorny, Krzymala-Kazlowska, Korys, & Weiner (2007), for example found that Poland allowed dual citizenship for emigrants living abroad, while restricting dual citizenship rights for immigrants living in their country. Although some research suggest that this type of policies has an influence on the remittance inflows, no specific data could be found on the years for which countries allowed dual citizenship and which year not. Further research should therefore investigate whether dual citizenship indeed has an influence on remittance inflows.
This research will not investigate the reasons for governments to develop these policies, nor the exact definition or aim of these policies. Merely their presence in time and their influence on remittance inflows are investigated. Therefore, a “diaspora engagement indicator” is introduced; reflecting the number of government levels involved in all the remittance policies of a particular country. A one-point score is given for each government level, while five government levels are distinguished (ministry level, sub ministry level, national level, local level, and quasi-government level), thus resulting in a 0 to 5 score. This is done for each country in each year researched. Based on this diaspora engagement score, and following Gamlen’s tapping perspective mentioned earlier, the following hypothesis is formulated:
H7: Countries with a higher level of diaspora engagement generate higher remittance inflows per emigrant
This research also distinguishes between countries having some form of diaspora engagement and countries without. This is done because the diaspora engagement policies are only known for a limited number of countries and therefore the differences in results between these two groups of countries have to be checked.
Special kinds of policies are tax policies. Because they strongly affect capital flows, they deserve special attention. Remittance inflows (including workers compensation) are therefore also susceptible to tax policies. It is therefore reasonable to expect that countries characterized
23 as tax havens have an influence on the remittance flows. The country ranking in Table 2 also clearly shows some well-known tax havens on top with strikingly high remittance inflows per emigrant. For this research, only a handful of secrecy jurisdiction could be included since data on remittances were not known for these countries. This is no surprise, since they are called secrecy jurisdiction for the same reason.
Cobham, Jansky & Meinzer (2015) indicate that a lot of academic research and public policy debate around tax havens and offshore finance suffer from definitional consistency. Some papers refer to Offshore Financial Center, while other use secrecy jurisdiction to indicate whether countries are tax havens. An often-used term to indicate tax havens is secrecy jurisdiction. Murphy (2008) characterizes a secrecy jurisdiction by the following two characteristics: (1) “The secrecy jurisdiction creates regulation they know is primarily of benefit and use to those not resident in their geographical domain”; and (2) “the creation of a deliberate, legally backed, veil of secrecy that ensures that those from outside the jurisdiction making use of its regulation cannot be identified to be doing so”. Cobham et al. (2015) also use the term ‘secrecy jurisdiction’. They define a secrecy jurisdiction as “a jurisdiction that provides facilities that enable people or entities to escape or undermine the laws, rules and regulations of other jurisdictions elsewhere, using secrecy as a prime tool”. Both definitions indicate that government actively structure their tax and legal codes to attract commerce or capital flows from abroad. Palan (2002) even speaks of the ‘commercialization of sovereignty’. Briefly summarized as: “the decision by certain jurisdictions to obtain economic advantage by allowing selected political decisions (e.g. the taxation of non- residents) to be dictated by those likely to benefit from the decision”.
Being a tax haven or not has a binary character, applies to only a few countries, and has minimal variation in time per country. Therefore, the characteristic “tax haven” cannot be considered a variable for this research. However, since being a tax haven is expected to affect remittance inflows, this research distinguishes between tax haven countries and non-tax haven countries in order to check for differences in results between these two groups of countries.
24 3. METHODOLOGY
3.1 Sample
In total, 215 countries were originally selected for this research, based on remittance data from the World Bank. Of these countries, 33 countries were excluded because of insufficient data (24 without any remittance data and 9 with less than 4 observations). Consequentially, 182 countries are included in this research. A complete list of countries included and excluded is provided in Appendix V.
The World Bank remittance database provides annual data for all countries over a period from 1960 till 2014. The research period for this research is limited to a time period of 25 years from 1990 to 2014, for three main reasons. The most important reason is the moderate growth of remittances over the period up to 1990, and the rapid and progressive growth since then.
Furthermore, just before this period, several new countries have been formed through the dissolution of the Soviet Union. The third reasons for selecting this time period is because before 1990 only a handful of diaspora institutions were found, while after 1990 these numbers rapidly increased.
Thus, the resulting sample for this research includes a maximum of 25 years for 182 countries, resulting in a panel with a maximum of 4.550 observations.
3.2 Dependent variable
As explained in the previous section, the remittance inflow per emigrant is the dependent variable in this research. This variable is calculated by dividing the total annual remittance inflow per country per year by the total emigrant stock of that country in that year. The reason for this is to make this variable suitable for comparing countries, since it eliminates the differences in size.
Data on emigrant stock is obtained via UN Department of Economic and Social Affairs. They provide estimated data on migrant stocks for the years 1990, 1995, 2000, 2005, 2010, and 2015 (International migrant stock: the 2015 revision). These estimates were interpolated for each of the four intermediate years, as proposed by the statisticians of the UN. Since annual data was available for most of the other variable, interpolating the migrant stocks significantly increased the total number of observations.
25 In the data from the World Bank and the UN a number of countries have different names, or are not recognized by all sources (e.g. “West Bank and Gaza” versus “The State of Palestine”). Therefore the names and countries have been harmonised for this research, with the World Bank names as reference.
3.3 Independent Variables
For this research several independent variables are included. A complete list with all the variables, descriptions, codes, and sources can be found in Appendix VIII Table 9. Although this list already provides the sources, codes and basic descriptions, the following paragraphs will explain the data, the measures and the variables in more detail.
3.3.1 GDP per capita
For this variable the GDP per capita (in current US$) from the World Bank is used. From their WDI database, the available annual data for all countries from 1990 until 2014 was used.
3.3.2 Public Health Expenditure per capita
For this variable, the total health expenditure per capita and the public health expenditure as a percentage of total public health expenditure per capita were used to calculate the public health expenditure per capita. Both parameters were obtained from the World Bank database WDI database. Again the available data for all countries from 1990 until 2014 was used.
3.3.3 Polity2 index
For this variable the Polity2 index was used as provided by the Center for Systemic Peace (2015). This index captures a countries’ regime on a 21-point scale ranging from -10 (autocracy) to +10 (democracy), and is calculated on the basis of 6 components. All available annual data on 167 countries covering the period 1946 until 2014 was downloaded from this database. After removing all data until 1989 and harmonising countries these data were added to the panel dataset.
3.3.4 Governance Quality index
For this variable, the annual data from the World Bank WGI database was used. As described in the previous section a total score was calculated using three indicators from this database:
the ‘Government Effectiveness’, the ‘Regulatory Quality’ and the ‘Rule of Law’ indicators.
These three were combined into one indicator called ‘Governance Quality’, by just adding up
26 the three underlying indicators. This data is only available for a limited number of countries and years (at most 1996-2014), but was added to the dataset.
3.3.5 Refugee share as a percentage of the emigrant stock
For this variable, two parameters were combined. One is the interpolated annual emigrant stock per country, also used for calculating of the dependent variable as explained earlier (see paragraph 3.2). The second is the refugee stock obtained from the World Bank WDI database (2015) but originating from the UNHCR data (annual data per country from 1990 until 2014).
The refugee share was calculated as a percentage, dividing refugee stock by the total emigrant stock, and added to the panel dataset.
3.3.6 Level of violence in a country
For this variable, the necessary data was obtained from the database on Major Episodes of Political Violence kept by the Center of Systemic Peace (2015). This database contains several variables, but only two variables are used for this research. The first variable is the variable INTTOT, which combines two separate variables named INTVIOL (international violence) and INTWAR (international warfare). The second variable is INTTOT, which is a combined variable including the variables CIVVIOL (civil violence), CIVWAR (civil warfare), ETHVIOL (ethnic violence), and ETHWAR (ethnic warfare). All of these underlying variables are scaled on a 0 to 10 scale. The total score of both variables INTVIOL and INTTOT are combined into one variable named VIOL_TOT. Although the theoretical value of this variable ranges from 0 to 50, the actual values of the dataset range from 0 to 13.
3.3.7 Diaspora Engagement indicator
For this variable, the necessary data was obtained from a paper by Gamlen (2014), a research done by the International Organization for Migration (IOM) and Migration Policy Instititute (MIP), and own research on the year of establishment for the known institutions. A complete list of all Diaspora institutions can be found in the Tables 4-8 provided in Appendix VII.
To calculate the Diaspora Engagement score, the presence of a diaspora institution on five different governance levels (represented by 0= absent or unknown, and 1= present) were added up to one score, ranging from 0 to 5. This led to a categorical score per country per year, so an increasing number indicates a higher government engagement from the receiving country.
27 3.4 Dummy variables
As explained in paragraph 2.3, three subgroups of countries will be checked for deviations in the results against all other countries. These three subgroups are explained in the following paragraphs.
3.4.1 Developing Countries
This dummy variable is added to the regression analysis to check for structural differences in remittance inflows between developing and developed countries. The dummy variable indicating whether a country is a developing country or not, was obtained from the UNCTAD classification and database. The resulting 0/1 values (0= developed country, 1= developing country) per country were added to the panel dataset. A complete list of developing countries can be found in Appendix V.
3.4.2 Mini States
This dummy variable is added to the regression analysis to check whether the statistics for extremely small countries structurally differ from normal countries. This is done because a considerable number of mini states is included in the dataset (33 mini states on a total of 182 countries), and the quality and accuracy of their statistics are possibly influenced by their small population size.
The dummy variable indicating whether a country is a mini state was determined by the population size obtained from the World Bank WDI database. For all countries and all years with a population of less than 500.000 the value 1 was calculated, while the value for all other years and all other countries remained 0. The resulting 0/1 values (0= normal country, 1=
mini state) per country were added to the panel dataset. A complete list of mini states can be found in Appendix V. Over the years seven mini states became normal countries because their population rose above 500.000.
3.4.3 Tax Havens
Because tax policies strongly influence the worldwide flows of money, and a number of well- known Tax Havens show strikingly high remittance inflows per emigrant (as explained in paragraph 2.3.3), the dummy variable Tax Haven is added to the regression analysis in order to check for structural differences between Tax Havens and normal countries.
28 The dummy variable indicating whether a country is a Tax Haven or not, was obtained from the Tax Justice Network (2015). Their classification of secrecy jurisdiction is used, incorporating only the countries for which the year it became a tax haven was known and being a tax haven was clear. The list of countries, which are considered to be a Tax Haven, is provided in Appendix VI Table 2 and Table 3. The resulting 0/1 values per country per year were added the panel dataset.
3.5 Empirical Model
Using the variables described above, the following model was constructed to analyse the relationship between the dependent variable (remittance inflows per emigrant) and the independent variables:
𝑅𝐼!"#, 𝑖𝑡 = 𝛼 + 𝛽1 ∗ 𝐺𝐷𝑃!"#!, 𝑖𝑡 + 𝛽2 ∗ 𝐻𝐸𝐴𝐿𝑇𝐻!"#!, 𝑖𝑡 + 𝛽3 ∗ 𝑃𝑂𝐿𝐼𝑇𝑌2, 𝑖𝑡 + 𝛽4
∗ 𝐺𝑂𝑉!"#$, 𝑖𝑡 + 𝛽5 ∗ 𝑅𝐸𝐹%, 𝑖𝑡 + 𝛽6 ∗ 𝑉𝐼𝑂𝐿!"!, 𝑖𝑡 + 𝛽7 ∗ 𝐷𝐼𝐴𝑆𝑃!"#, 𝑖𝑡 + 𝛽8 ∗ 𝐷𝑢𝑚𝑚𝑦, 𝑖𝑡 + 𝜀"
With the following dependent and independent variables:
− RI_EMI, it = remittance inflow per emigrant for country i in year t
− GDP_PCAP, it = GDP per capita for country i in year t
− HEALTH_PCAP, it = Public health expenditure per capita for country i in year t
− POLITY2, it = The polity2 index for country i in year t
− GOV_QUAL, it = The governance quality index for country i in year t
− REF%, it = The refugee percentage of the emigrant stock for country i in year t
− VIOL_TOT, it = The total violence score for country i in year t
− DIASP_ENG,it = The diaspora engagement score for country i in year t And the following dummy variables, which are introduced one by one in the model:
− MINI_STATE, it = A dummy which indicates whether a country is a mini state
− DEV_CNTRY, i = A dummy variable whether a country is a developing country
− TAX_HAV, it = A dummy which indicates whether a country is a tax haven in year t
− DIASP_UNIT, it = A dummy variable indicating whether a country has some kind of Diaspora Institution
29 3.6 Models
In order to test the empirical model described above, six regression models are used in the to analyse the relationship between the dependent and the independent variables. A description and summary of these models can be found in Table 10.
Overview of models
Model 1 Main model without distinction between countries and the first six independent variables.
Model 2 The same as model 1, but with the dummy variable MINI_STATE included to check for differences in results between all mini states and all normal countries.
Model 3 The same as model 1, but with the dummy variable DEV_CNTRY included to check for differences in results between all developing and all developed countries.
Model 4 The same as model 1, but with the dummy variable TAX_HAV included to check for differences in results between Tax Havens and normal countries.
Model 5 The same as model 1, but with the dummy variable DIASP_UNIT included to check for differences in results between all countries with some form of Diaspora Institution and all countries without.
Model 6 The same as model 5, but with the factor variable DIASP_ENG included to investigate the effect of the level of diaspora engagement on remittance inflows for all countries with diaspora institutions.
Table 10 - Overview of models
All these models are based on the empirical model and formula states in the previous paragraph. The only difference is the use of different dummy variables per model and the use of the additional categorical variable DIASP_ENG in model 6.
By using six different regression models with one dummy variable per model, the differences in the results between the subgroups of countries are much clearer and easier to understand.
30 4. EMPIRICAL DATA & ANALYSIS
4.1 Descriptive statistics
The content of the dataset for all the variables are summarized in Table 11, presenting the descriptive statistics per variable. As described in the Table, most variables have a high number of observations.
Variable Obs. Mean Std. Dev. Min. Max. Remark:
CNTRY_NR 4550 1 215 182 countries
YEAR 4550 1990 2014 25 years
logRI_EMI 3750 6.5098 1.8311 -3.8694 13.235 Dependent variable
logGDPpcap 4319 7.9607 1.5953 4.1715 11.6416 Independent variable (regressor)
logHEALTHpcap 3270 2.8369 1.9783 -4.9767 7.6247 dropped: multicollinearity
POLITY2 3605 3.7157 6,4056 -10 10 Independent variable (regressor)
GOV_QUAL 2834 144.364 77.046 1.4612 299.5098 dropped: multicollinearity
REFp 4017 0.0603 0.1833 0.00 2.76 Independent variable (regressor)
VIOL_TOT 3469 0.6705 1.6438 0 13 Independent variable (regressor)
DIASP_ENG 4550 0.1954 0.5009 0 4 categorical variable
DIASP_UNIT 4550 0.1504 0.3574 0 1 dummy variable DEV_CNTRY 4550 0.6978 0.4593 0 1 dummy variable MINI_STATE 4550 0.1670 0.3730 0 1 dummy variable
TAX_HAV 4550 0.1756 0.3801 0 1 dummy variable
Table 11 – Descriptive statistics
A special comment on the variable REFp must be made, since some countries have abnormal high refugee numbers compared to their emigrant stock. Obviously a refugee percentage of over 100% is impossible. Table 11 shows the countries and the years involved.
Countries Years Refugee Percentage
Afghanistan 1990-1991 94% - 103%
Bhutan 1992-1994, 1997, 2001, 2002, 2006 -2011 92% - 118%
Burundi 1993 189%
Djibouti 1994-1997 105% - 276%
Iraq 1992, 2007 94% - 121%
Liberia 1990-2000, 2003-2004 92% - 160%
Rwanda 1994 128%
Sierra Leone 1991-1993, 1999 98% - 116%
Timor-Leste 1999 122%
Togo 1993 111%
Table 12 – List of extreme refugee percentages from the dataset
