Determining and examining the multiple causes of remittances with use of the Gravity equation
Name: Emanuel Kolta
Student Number: S2492946
E-mail address: e.kolta@student.rug.nl
University of Groningen
Abstract
The aim of this paper is to present the characteristics and drivers of remittances from a macro prospective and to introduce an extended Gravity model. The top ten senders and receivers of remittances are providing the sample for the Gravity model, which covers 54% of the whole inflow worldwide. According to the model, it has been scientifically proven that the economic mass of a country, differences in the sender and the receiver countries GDP per capita, common border between countries, distance and the GDP per capita of the remittance sender country, are all significant explanatory variables of the remittances inflows.
Table of content
I. INTRODUCTION ... 4
II. THEORY ... 5
1.DESCRIPTION OF THE DEFINITIONS ... 5
2.DESCRIPTION OF WORLDWIDE REMITTANCES FLOWS ... 6
3.MIGRATION AND REMITTANCES ... 10
4.HYPOTHESES ... 11
III. DATA AND METHODS ... 12
1.GRAVITY MODEL ... 12 2.EXAMINED COUNTRIES ... 13 3.VARIABLES ... 14 4.MODEL ... 19 5.ESTIMATION TECHNIQUES ... 20 IV. RESULTS ... 22
1.RESULT OF THE MODEL ... 22
2.LIMITATIONS ... 23
3.CONCLUSION ... 24
REFERENCES: ... 25
I. Introduction
There are more than 216 million people in the world who live outside of their home country. The topic of remittances incorporates a lot of aspects (sociological, political, economic) and the importance of this theme is continuously growing due to falling tangible and intangible barriers, a fast level of flowing information and globalization. In the following pages, the driving forces behind the flow of remittances will be examined. The above topic has not been thoroughly researched; the first paper that directly estimates flow of remittances is an IMF working paper from 2006. (E. Lueth, M Ruiz-Arranz, 2006). The rapid growth of the amount of money being sent home is quite remarkable: the World Bank recorded 10.9 per cent for the year of 2011. (Migration and Development Brief 20, 2013). Furthermore, while the scientific background of the migration has been widely examined, the topic of remittances has been much less investigated. Despite its importance, migration remains a topic on which available data is relatively weak or merely not available (Thorogood, 2005). According to previous research, useful results can be drawn from those studies which investigate variables connected to migrants home and host countries. Few papers have focused on one or a small set of countries. (Bouhga-Hagbe, 2004). While these articles are related to specific regions or countries, some of them are globally examining the topic of the remittances. (Gupta, 2005). The aim of this paper is to present the most up to date literature about remittances and to apply the Gravity model, by investigating the background of remittances using the most current and high coverage version of the databases.
comprehensive global database, which covers 275,661 million dollars of remittances. It means that 54% of the global remittances inflow that is contained in the following analysis, comes from the year of 2011. The aim of this paper is to present the latest theoretical findings, to summarize the recent global trends and determinatives, to build a cross-country model and to interpret the most significant macro-based explanatory variables. Based on previous research, determinants can be classified by three distinct groups. Namely, variables associated with the migrant’s home country, variables related to the migrant’s host country and variables that describe the relationship between the migrant’s home and host countries. (Lueth, Ruiz-Arranz, 2006)
The paper is organized as follows; section II discusses the theory, definitions and describes the global remittances flows, section III introduces the database, and presents the variables and the model, and finally section IV concludes the results.
II. Theory
1. Description of the definitions
First of all, there are several different definitions for remittances and differing measuring techniques. The second edition of the Migration and Remittance Factbook from 2011 uses the following simple definition: “migrant remittances are defined as the sum of workers’ remittances, compensation of employees, and migrants’ transfers.”
According to the International Monetary Foundation, the definition of remittances is the following: “Current private transfers from migrant workers who are considered residents of the host country to recipients in the workers’ country of origin.” (Balance of Payments Manual, 6th edition)
Nations Technical Subgroup on the Movement of Persons accepted the conceptual definition of remittances based on the issue paper Definitions of Remittances (Alfieri and Havinga, 2006) and discussed different measuring techniques. Personal transfers were described as “all current transfers in cash or in kind made or received by resident households to or from other non-resident households.” (Alfieri and Havinga, 2006). These personal transfers include all current transfers from the resident to the home country households.
The Bilateral Remittance Matrix (The World Bank, 2011) provides the basis for the later presented model. The compilers of this matrix encountered with several difficulties and lacked missing values. The irregular status of migrants and restrictions on outward remittance flows are causing an underestimation of remittances flows. (Ratha, 2007) According to the World Bank working paper South-South Migration and Remittances, there are three main methods as to how the issuer of the Bilateral Remittance Matrix is addressing the problem. Firstly, with weights based on migrant stocks abroad. Secondly, with weights based on migrant incomes, proxied by migrant stocks multiplied by per capita income in the destination countries. Thirdly, weights that take into account migrants’ incomes abroad as well as source-country incomes.
2. Description of worldwide remittances flows
According to the World Bank’ s Bilateral Remittance Matrix from 2011, many connections can be detected between the remittances flows and the geological location.
Asia
countries, the GDP per capita is the biggest one in China with 5447 (World Bank, 2011), so they can be categorized as emerging economies. (Hoskisson et al., 2000). Thirdly, there is a group of countries in Asia, which has a significant proportion of remittances, in comparison to their GDP. The head of this list is Tajikistan, where the money that the migrants sent back equates to 47% of the GDP. Furthermore Kirgizstan (29%) and Armenia (13%) are in the top ten of the list of recipients of migrant remittances as a share of the GDP in 2011. Common characteristics of this group are that they are post-soviet countries, with a relatively small population (less then 8 million) and have a common border with Russia. It is important to be within proximity to Russia, because of the rapidly growing wages and job possibilities in some highly frequented Russian cities. After this example, it can be assumed that geographical closeness and a common border could have significant effect on remittances flows.
Europe
Romania see over 50 per cent of flows going to rural areas. (IFAD, 2013) The globally significant remittances flows in western Europe are made by a relatively small number of migrants. (World Bank, International migrant stock) This could be caused by the number of educated, better-paid migrants. It is important to highlight two forces, which affect the remit level of the educated migrants. If a larger part of the migrants are skilled and highly educated, they are more likely to earn a higher income and remit more. However, educated migrants tend to have more favorable backgrounds and so their families are less dependent on remittances. Therefore, it is still not that obvious and quite a context-sensitive question, as to which force is dominant. (Ozden, 2006)
Africa
The two biggest remittance receivers according to the Bilateral remittance matrix are Nigeria (5.) and Egypt (7.) Furthermore, northern African countries are the major receivers of remittances and the most important senders are western European countries. East African countries have the biggest dependence on these money transfers, especially Somalia. (IFAD, 2010) It is important to highlight the underdeveloped financial environment and that they have the highest cost of transfer remittances, 8-12%. (Sending money home to Africa, 2010) In the fifth biggest remittances receiver, Nigeria, almost 80 per cent of the money transferred is handled by one MTO1. Regarding to Sending money home to Africa (2010), further conclusions can be drawn about remittances inflows. Overall, the level of competition in the finance sector is significantly lower than in other parts of the world and there is a limited payout presence in rural areas. The lack of access for the payout locations and the fact that most regulations in Africa permit only banks to pay remittances may impact the remittance flows.
1Money transfer operator (MTO): A payment service provider that receives payment, in cash or by
North America
Canada and the United States are main targets for migrants: both of the two countries have significant net migration2. According to the World Bank, the United States has 4,954,924 migrant surplus, while Canada also has a positive balance with 1,098,444 migrant, which is an unique value compared to the 33,476,688 inhabitants. (Statistics Canada, 2012) Furthermore, USA is the biggest sender of remittances, Canada is the third largest sender and the biggest remittances corridor was also between US and Mexico, with more than 23 billion dollars. According to the above-mentioned fact, it can be assumed that economic mass, and big differences in GDP per capita could be strong explanatory variables of the remittances flow.
Latin America
According to the bilateral remittance matrix in 2011, Mexico is third in the top ten receiving remittances of 2011. Furthermore, almost all the remits which Mexico receives (23,588 million dollars) come from United States (23,175 million dollars). This means that 98.25% of the remittances are coming from one country, the United States. This dependence ratio is decreasing in other parts of the Latin American region. It can be assumed, that the common border between the sender and the receiver could have an affect for the remittances flows. Other important corridors can be find between Spain and Dominican Republic, Ecuador, Peru, Colombia and furthermore between Portugal and Brazil. It can be assumed that common language could increase the size of the flow, even in inter-regional situations.
Middle East
Saudi Arabia (2th) and United Arab Emirates (7th) are both leading the list of the top remittances senders. After the oil crisis in 1973, oil-exporting countries in the Middle East had a shortage of competitive, skilled labour to carry out their ambitious development plans. (Arnod, 1984) To implement these (mainly construction) projects, foreign workers were brought to the Middle East. The main sources of the invited
3Mexico alone with 117 million inhabitants has almost as many payout locations as the entire African
workers was firstly South Asia and after Southeast Asia. According to the World Bank’s Bilateral Migration matrix, there are 1.5 million Indians and 1 million Pakistanis living in Saudi Arabia and a further 2.2 million Indians and 453 thousand Pakistanis in the United Arab Emirates. Due to the above-mentioned fact, the migration corridor between India and United Arab Emirates is the 9th biggest (excluding the former Soviet Union as 5th) with 14.255 millions sent remittances.
3. Migration and remittances
Relevant conclusions can be drawn about the different the explanatory variables of migration and remittances after investigating the Europe. A noticeable difference can be found between the migrant and remittances flows. European countries are representing themselves in the top recipients of migrant remittances with France (6.), Germany (8.) and UK (11.). Here a noticeable difference can be found between the main sources of migrants and the main flow of remittances. The comparison between the above used remittances database, with the World Bank’s Bilateral Migration Matrix 2011 can highlight important differences.
and Remittances Factbook, 2011), French remittances inflow are mainly coming from developed, mainly European countries. It may be inferred, that the remittances/migrant ratio is connected with the economic differences between the sender and the receiver country.
4. Hypotheses
Most of the variation in bilateral remittance flows can be explained by a few gravity variables. (Lueth, 2006) Overall, in the following pages a method will be presented to investigate the background of the remittances. The research question can be defined as follows: Which of the home and host country factors can be seen to have a direct effect on the remittances inflow, and how can the relationship of the two countries influence the remittance inflow. The hypotheses below will be tested:
H1a: The logarithmic form of the remittance sender country’s GDP is significantly
increasing the level of the remittances.
H1b: The logarithmic form of the remittance sending country’s GDP per capita is
significantly increasing the level of the remittances.
H2a: The logarithmic form of the remittance receiving country’s GDP per capita is
significantly increasing the level of the remittances.
H2b: The logarithmic form of the remittance sending country’s GDP per capita is
significantly increasing the level of the remittances.
H3a: The logarithmic form of the remittance receiving country’s unemployment level
is significantly increasing the level of the remittances
H3b: The logarithmic form of the remittance sending country’s unemployment level is
significantly decreasing the level of the remittances.
H4: The logarithmic form of the distance between the remittance sender and the
receiver country is significantly decreasing the level of the remittances.
H5: Land link connection between the remittance sender and the receiver country is
H6: Ex-colonial status between the remittance sender and receiver country is
significantly increasing the level of the remittances.
H7: A common official language between the remittance sender and receiver country
is significantly increasing the level of the remittances.
H8: The logarithmic form of the GDP per capita difference between the remittance
sender and receiver country is significantly increasing the level of the remittances. H9a: The logarithmic form of the nominal cost of the passport in the remittance
receiver country is significantly decreasing the level of the remittances.
H9b: The logarithmic form of the relative cost of the passport (proportion of Gross
National Income) in the remittance receiver country is significantly decreasing the level of the remittances.
III. Data and Methods
1.Gravity ModelA widespread way to explain international trade flow or foreign direct investment is the Gravity model. The main idea can be derived from Newton and his theory of the “Law of Universal Gravitation” in 1687. It assumed that forces between objects i and j derive from two masses divided by the square of the distance between them. In 1962 the Dutch economist, Jan Tinbergen suggested using this model to describe international trade flows.3 Since then it has been applied in several distinct topics for instance tourism, migration or other spatial areas. The general gravity law for social interaction can be expressed by the following:
3The main difference from Newton’s law is that trade is inversely proportionate to distance whereas
(1.)
The Fij is the flow between country i and j. Mi and Mj are the economic mass of i and j
locations. Dij is the distance between i and j location. The gravity model presented
above will provide as a core to estimating remittances.
In its simplest form the gravity equation for trade states that trade which flows between two countries is proportional to the two countries’ economic sizes (GDPs) and inversely proportional to the distance between them. The model often includes variables to account for income level (GDP per capita) and physical and cultural proximity (shared border, language relationship, and colonial history). Other factors that either enhance or impede trade, such as price levels and tariffs, have also been included in extended specifications. Therefore, the framework of the above-presented model is based on a gravity model.
However, while the top 10 receivers and the top 10 senders are included in the model, overall only 18 different countries were observed. This is due to the fact that France and Germany were included in the top ten senders and top ten receivers as well. The top 10 remittance sender countries (Table 1) had been paired with the top 10 remittances receiver countries. Reverse flows are also included in the database. Without the German and the French duplication there was would be 200 observations, but for the above-mentioned reasons, the France-France and the Germany-Germany line were deducted to avoid duplication.
2. Examined Countries
According to this database (Bilateral Remittance Matrix, 2011), 507,600 million dollars flow between 213 countries. It is important to not only mention countries included in the list. For instance Faeroe Islands or West Bank and Gaza are not traditionally independent, individual countries in the eyes of the UN, however, examining these territories can provide other useful scientific results.
The ten biggest remittance sender and receiver countries are covered in this paper. The amount of remittances that countries sent or received in 2011
Top 10 Remittance-sending country (million dollars)
Top 10 Remittance-receiving country (in million dollars)
United States 120,155 India 63,011
Saudi Arabia 24,187 China 61,365
Canada 23,399 Mexico 23,588
United Kingdom 23,164 Philippines 23,065
Germany 21,740 Nigeria 21,619
France 19,994 France 19,483
Spain 18,988 Egypt 14,324
United Arab Emirates 18,219 Germany 13,393
Hong Kong (1) 17,439 Pakistan 12,263
Australia 14,684 Bangladesh 12,068
Table 1, Source: World Bank
Overall, 196 flows are being examined between top remittances senders and top remittances receivers. In total, 507,600 million dollars had been sent home by migrants in 2011, the later applied model covers 275,661 million dollars, which means 54% is presented.
Bilateral inflow of remittances
The dependent variable, which is being examined, is the inflow of remittances in 2011, which is the money that migrants sent back to their home country for family or friends. However, the correlation between the number of migrants and amount of remittances could be weak as in the case of a small amount of highly educated people could send back a relative large amount of money (i.e.: the case of Germany and France). Likewise a large number of migrants could also remit a negligible amount of money for many reasons. The source of the exact amount of money sent back is the World Bank’s bilateral remittance matrix 2011. In this 213 x 213 Matrix, we can observe each channel and how many millions of dollars have flowed between i and j countries in 2011. The dependent variable will be named as flow.
GDP
The GDP is an essential characteristic of a country from the viewpoint of the Gravity model. It represents the economic mass of the countries which are included in the model. These dates have been gathered from the World Bank (International Comparison Program database 2011) and they are corrected with purchasing power parity. Due to the features of the Gravity model, two separate variables are included in the model, one of them represents the sender country’s economic mass, while the other represent the receiver’s economic mass. These independent variables will be named as gdprec and in the model.
GDP per capita:
indicator of economic performance is included in the model two times, firstly describing the remittances receiver country. It can be assumed that the home countries’ economic environment has an impact on the amount of money that migrants sending back. This independent variable will be named as gdpcaprec in the model.
Secondly, the impact of GDP per capita in the remittances sender country will be examined as well. This indicator helps to account for income levels in the host countries. The first usage of estimating remittances with GDP per capita is related to E. Lueth and M. Ruiz-Arranz from 2006. This independent variable will be named as gdpcapsend in the model.
Level of unemployment:
It is assumed in the in the hypothesis 3a and 3b that remittance receiver country’s
unemployment (β2) may increase the level of the remittances because of two main reasons: firstly, unpleasant labor market conditions may raise the migration to other countries. Secondly, these unpleasant labor market conditions may increase the motivation to send back money from other countries, to support friends and family at home. This independent variable will be named as unemprec in the model. It is also assumed, that remittance sender country’s unemployment (β4) has a vital impact for
our dependent variable. The possibility of getting a job which is engaging and potential labor force shortage could also raise the amount of the money which migrants send back to the receiver country. This independent variable will be named as unempsend in the model.
Distance
Common Border
This dummy variable indicates whether the two countries are contiguous. It can be assumed that countries with a common terrestrial border have a significantly higher level of remittances, than countries without this characteristic. This independent variable will be named as commonborder in the model.
Ex Colony
Ex colonial connections between country i and j could also have significant effect on the amount of remittances as well. If ever in history, the i or j country have been the colony of the other, the excolony dummy will turn to one. This means that there was a colonial relationship between the examined two countries and colonizers of the country for a relatively long period of time and with a substantial participation in the governance of the colonized country. Hypothesis 6 was created with reference to the section regarding the characteristics of South America. This is based on the fact that sharing similar institutions, culture and numerous connections is helpful in increasing the flow of remittances. The source of the excolony variable is CEPII.
Common language
If i and j country have the same language officially, this dummy variable will turn to 1. As it is specified in the hypothesis 7, common language could increase the amount of the money that migrants are sending back. It would create a better labor market position for migrants and less cultural borders. The source of the database is the French research center in international economics (CEPII), namely the source is GeoDist.
Difference of the GDP
This variable demonstrates the difference between the above-presented GDP per person (gdppercapsend – gdppercaprec) between i and j country. The differences between i and j countries GDP assumedly count in terms of remittances and the bigger the economic differences, the more motivation there is to send home money.
Passport cost
According to McKenzie (2005) , the cost of obtaining a passport can affect remittances flows. This variable represents the price (in dollars) of one passport. Passport costs were collected in local currency, and converted into United States dollars at the prevailing interbank exchange rate. This independent variable will be named as passportcost.
Passport cost to GNI per capita
The relative price of a passport may have a more significant effect on remittances flow. According to McKenzie (2005), this variable presents the price of one passport relatively to the issuing country’s GNI. This independent variable will be named as passportcostperGNI.
Country effects
each of which refers to a country that is presented in the model. The coefficients of the variables are irrelevant for this particular paper, however they treat the above-mentioned bias.
4. Model
In the following section, an extended gravity model will be presented, which has been created to estimate remittances flows. The ordinary gravity model can be described as:
The Fij is the flow between country i and j. Mi and Mj are the economic mass of i and j
locations. Dij is the distance between i and j location. The logarithmic form of the
above-mentioned gravity equation will prove the basis of model. It can be described as:
lnF ln m1 m2 r
In the previous academic literature, many examples of expanded gravity models can be seen in the topics of FDI, commerce, etc.
remittances. The final model used is shown here:
e
The structure of this model is consistent with trade studies and has some similarities to a previous study (E. Lueth, M Ruiz-Arranz 2006). However, the variables and the database here is more comprehensive.
5. Estimation techniques
The Ordinary Least Squares (OLS) method will be used during the model. With this method the sum of squared vertical distances is minimizing. During the estimation part, multiple factors can arise that could distort the results of the model, for example multicollinearity and heteroskedasticity.
independent variables, they have been omitted.
Secondly, heteroskedasticity is likely in cross-sectional databases. The Breusch– Pagan (appendix 3.) test will be applied to detect the presence of heteroskedasticity. The null hypothesis of this test is based on the variance function and expresses that if the variances of all observations are the same, the data is homoscedastic. If the null hypothesis is approved, then variance is constant and there is no heteroskedasticity present. According to the result of the Breusch–Pagan test, the database is homoscedastic. (appendix 4.) The H0 cannot be omitted because the variance of the
explanatory variables is constant.
Thirdly, the result of the F-test is less then 0.05, meaning that all coefficients in this
model are different from zero.
Fourthly, according to the graphical and numerical tests of the variables normality, normality is not characterized for every variable. Variables on the countries’ GDP and GDP per capita are not normally distributed and this may be due to the small number of examined countries. Overall, the lack of the normal distribution in some of the explanatory variables doesn’t distort the result greatly.
The first model is the basic Gravity model, with the lnGDPi and lnGDPj which
represent the economic masses of the i and j countries, and the logarithmic form of the distance between i and j country.
The second model is an extended version of the first one, with each countries logarithmic form of the GDP, logarithmic form of the GDP per capita, logarithmic form of the unemployment level.
The third presented model includes three other dummy variables, namely common
border, common language and ex-colony.
The fourth model, has additional variables: the logarithm of the difference between the sender and the receiver countries GDP per person, country effects for sender and receiver countries, logarithmic form of the nominal passport cost and the logarithmic form the passport cost to GNI per capita.
IV. Results
1.Result of the model
After the model estimation and model selection process, the final version of the model explains 62% of the reality, with eight significant (p‹0.05) independent variables.
1 2 3 4 5
VARIABLES ln (flow) ln (flow) ln (flow) ln (flow) ln (flow)
ln(GDPi) 1.123*** 0.615*** 0.556*** 0.553*** 0.516*** ln(GDPcapitai) 0.663*** 0.677*** 0.799*** 1.251*** ln(unempi) 1.018*** 0.994*** 0.899*** 0.987*** ln(GDPj) -0.0169 0.649*** 0.590*** 0.797*** 0.611*** ln(GDPcapitaj) -1.246*** -1.232*** 0.0614 ln(unempj) 0.237 0.223 0.18 0.31 ln(distance) -0.659*** -0.679*** -0.460** -0.526** -0.531** commonlanguage 0.279 0.269 0.349 excolony 0.889 0.74 0.396 commonborder 2.779*** 2.766*** 2.568*** c. effect receiver -0.0192 -0.0234 c. effect sender 0.0445 0.0584* ln(diffcdppercap) 0.234 0.605*** ln(PassportCost) -1.553*** -1.339*** ln(PasportcostperGNI) 0.658 Constant -6.504* -5.458 -6.215 -18.12*** -24.02*** Observations 196 196 196 196 196 R-squared 0.243 0.574 0.613 0.638 0.615
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Returning to H1a and H1b, economic masses proved as significant explanatory
level of the remittances. The assumed GDP per capita refers to H2a and H2b. The
remittances receiver country’s GDP per capita didn’t prove to be significant, while the remittances sender’s GDP per capita did. According to the model, if the remittances sender country’s GDP per capita increases, the remittances flow will increase. H3a and
H3b refer to the unemployment of the sender country (a) and receiving country (b).
H3b was omitted, because of lack of significance. Unemployment in the sender
country is a significant variable with a positive sign. H4 refers to the distance between
the sender and the receiver country. It has been proven that it became a significant independent variable with a negative sign. This implies that if the distance is increasing between the two countries, the remittances inflow will decrease, like in regular gravity models. H5 is also proven to be right. Land connection between the
sender and the receiver is a highly important explanatory variable with a positive sign. A common border between the remittances sender and the receiver country will raise the amount of the remittances. H6 was referring the ex-colonial status between the i
and j country. After the estimation process it became insignificant. This means that ex-colonial status is not affecting the level of remittances, H6 is s rejected. H7 presents
the common used language and it proved to be insignificant. Ergo, common language between countries has no effect for remittances. H8 refers to economic differences
between the remittances sender and receiver country. According to the result of the model, this hypothesis is accepted. GDP per capita differences are increasing the level of remittances inflow. So according to the model, the bigger the economic differences between the sender and the receiver country, the larger the amount of the remitted money. H9a and H9b have regard to nominal passport cost and to the passport cost
compared to GNI per capita. H9a is proven to be significant with negative passport
cost. Ergo if the cost of obtaining a passport is increasing in dollars, the level of the remittances inflow will decrease. H9b has to be omitted because of a high level of
correlation with other independent variables. (VIF test, appendix 2.)
2. Limitations
such data are reported, they may not be accurate, because funds channeled through international banks may be attributed to a country other than the actual source country.” In addition, irregular status of migrants and restrictions on outward remittance flows lead to further difficulties.
It is important to highlight, that other factors may influence the level of remittances as well. Characteristics of the migrants like age, education, industry experience may affect the remit levels. These factors could be integrated in the following research.
3. Conclusion
References:
Aggarwal, R., Horowitz, A., (2002) Are International Remittances Altruism or Institute? Evidence from Guyana Using Multiple-Migrant Households. World Development, Vol30, pp. 2033-2044.
Alfieri, A., Havinga, I. (2006), Definition Of Remittances, United Nations, Department Of Economic And Social Affairs Statistics Division, Meeting of the Technical Subgroup of the Task Force on International Trade in Services
Barajas,A., Chami, R., Fullenkamp, C., Gapen, M. and Montiel,P. (2009), Do Workers’ Remittances Promote Economic Growth?, IMF Working Paper, Middle Eastern and Central Asia Department
Bouhga-Hagbe, J. (2004), A Theory of Workers’ Remittances With an Application to Morocco, IMF Working Paper, Middle East and Central Asia Department
Duanny, J. (2010), To Send or Not to Send: Migrant Remittances in Puerto Rico, the Dominican Republic, and Mexico
E. Lueth, M. Ruiz-Arranz (2006), A Gravity Model of Workers’ Remittance, IMF Working Paper
El-Sakka, M.I.T. and R. McNabb (1999) The Macroeconomic Determinants of Emigrant Remittances, World Development, Vol27, No.8, pp. 1493-1501. Fred, A., Nasra, S. M.,(1984) Asian Labour Migrationto the Middle East, International Migration Review
Gupta, P. (2005), Macroeconomic Determinants of Remittances: Evidence from India, IMF Working Paper, Vol. 18, No. 2 (Summer, 1984), pp. 294-318
Head, K (2007), Element of Multinational Stategy, Springer, Sauder School of Business University of British Columbia, ISBN: 978-3-540-44765-8
Hill, R. C., William, E. Griffiths, Lim, C., Principles of Econometrics Fourth Edition (2010), John Wiley & Sons, Inc.
Hoskisson, R. E., Eden L., Ming, L., Wright, M., Strategy in Emerging Economies, American Management Journal, 2000
Ilahi, N., Jafarey, S., (1998) Guestworker Migration, remittances and the extanded Family: Evidance from Pakistan, Journal of development economics, Vol.58, pp. 485-512.
Mayer, T., Zignago S. (2005), “Market Access in Global and Regional Trade”, CEPII Working Paper 2.
Mayer, T., Zignago, S. (2006), “Notes on CEPII’s distances measures”, MPRA Paper 31243.
Mayer,T., Zignago, S. (2011), Notes on CEPII’s distances measures (GeoDist), CEPII Working Paper 2011-25.
McKenzie, David J (2005), Paper Walls Are Easier To Tear Down : Passport Costs And Legal Barriers To Emigration, World Development Volume 35, Issue 11, November 2007, Pages 2026–2039
Mélitz, J. (2002), “Language and Foreign Trade”, CEPR Discussion Paper 3590 Niimi, Y., Özden, C. (2006), Migartion and Remittances Causes and Linkages, The World Bank Development Research group
Orozco, M. (2003), Worker Remittances in an International Scope, Inter Amarican Dialogue, Research series: remittances project
Ozden, C. (2006) Remittance flows in Latin America and the Caribbean: Patterns and Determinants, World Bank
Ratha, D., Shaw, W., (2007), South-South Migration and Remittances, The World Bank, World Bank Working Paper 102.
Concepts and Definitions.Paper presented at CEIES Seminar: Migration Statistics— Social and Economic Impacts with Respect to Labour Market, Riga, 14–24.
Bilateral Migration Matrix (2010), World Bank, Migration and Remittances Unit, Bilateral Remittance Matrix (2011), World Bank, Migration and Remittances Unit
World Bank, Migration and development brief 19 (2012), Migration and Development Unit, Development Prospect Group
Databases:
2011 Census: Population and dwelling counts, Statistics Canada. February 8, 2012. Retrieved February 8, 2012.
Migration and Remittances Factbook 2011, Second Edition, 2011, The International Bank for Reconstruction and Development / The World Bank
Sending money home to Africa, Remittance markets, enabling environment and prospects, IFAD, 2009
Sending money home to Asia, Trends and opportunities in the world’s largest remittance marketplace, IFAD, 2013
Sixth Edition of the IMF's Balance of Payments and International Investment Position Manual, IMF, 2010
World Bank, Data bank, GDP per capita (current US$), 2001 World Bank, Data bank, International migrant stock, total
World Bank, Migration and development brief 20 (2013), Migration and Development Unit, Development Prospect Group
Appendix:
1.Summarizing variables
Variable | Obs Mean Std. Dev. Min Max ---+---
Flow of remittances | 196 1406.436 8732.778 1 117779 GDP, sender country | 196 2349629 3403861 111879.1 1.50e+07 GDP per cap, sender c. | 196 25233.59 18792.24 1489 60979 ---+---
Unemployment sender c. | 196 8.602551 5.232379 3.3 24
GDP, receiver country | 196 2349629 3403861 111879.1 1.50e+07 GDP per cap, receiver c | 196 25233.59 18792.24 1489 60979 Unemployment receiver c. | 196 8.689286 5.306072 3.3 24 Distance | 196 7297.021 3996.341 342.9475 16975.46 ---+--- Commonlanguage | 196 .255102 .4370349 0 1 ex-colony | 196 .0918367 .2895349 0 1 Commor border | 196 .0510204 .2206029 0 1 GDP difference | 196 0 33861.97 -59490 59490 Passport cost | 196 57.65589 30.69224 8.945 115.2212 ---+---
Passport cost to GNI | 196 1.712364 2.425154 .09 9.945
2.
VIF test for full model:
commonlanguage1.16 0.860455 commonborder 1.05 0.950355 ---+--- Mean VIF | 4.49
3. VIF test for the corrected model Variable VIF 1/VIF ---+--- Ln GDPcapitai 1.93 0.517306 LnPassportcost 1.85 0.540415 Ln diffGDP 1.51 0.662611 Ln GDPj 1.48 0.676317 Ln distance 1.42 0.704340 Ln GDPi 1.36 0.735221 excolony 1.27 0.785897 sender 1.23 0.812269 receiver 1.19 0.842408 Ln unempj 1.18 0.847473 Ln unempi 1.16 0.859975 commonlanguage 1.16 0.860534 commonborder 1.16 0.865354 ---+--- Mean VIF | 1.38 4. Breusch-Pagan test:
Breusch-Pagan / Cook-Weisberg test for heteroskedasticity Ho: Constant variance
Variables: lnGDPi lnGDPcapitai lnunempi lnGDPj lnGDPcapitaj lnunempj lndistance commonlanguage excolony commonborder fr fs lndiffcdppercap lnPassportCost lnPasportcostperGNI
