Bilateral Remittances and Convergence: An Empirical Analysis

Bilateral Remittances and Convergence:

An Empirical Analysis

Ibrahim Gerguri, 2052164

MSc. Economics, Rijksuniversiteit Groningen

Supervisor: prof. dr. B.W. (Robert) Lensink

Abstract:

In this paper, I investigate the impact of international bilateral remittances on economic convergence measured as the change in GDP per capita gaps between source and destination countries over the time period 2010-2015. In order to do this, I employ a large dataset of bilateral remittance flows for more than 200 countries. Firstly, I estimate my model with ordinary least squares (OLS) using both random effects and fixed effects in the estimation. Secondly, I estimate the model using an instrumental variable regression (IV). Tests show that the IV estimation with country fixed and time fixed effects produces the most reliable results. These results indicate e positive relationship between remittances and GDP per capita gaps between countries. That is, larger amounts of remittance flows across countries lead to a larger increase in GDP per capita gaps between countries over time, meaning that remittances cause income divergence rather than convergence. Additionally, I use several control variable which are previously proven to be relevant in this setting.

2

1. INTRODUCTION

In this paper, my objective is to analyze the effect of remittances on the economic convergence between countries from both ends of the remittance flows. In general, remittances are send by immigrants to their home country which makes it interesting to include migration in explaining the why and how of the interaction between remittances and economic development. As stated by Hagen-Zager (2008), migration is as old as society itself and recently, the international migration of people has increased remarkably. Think of the crises in the middle east with the Syrian civil war and the uprising of ISIS playing a significant role in the displacement of millions of people. These relatively recent events make the role of migration and remittances an extremely interesting topic of discussion. In the case I mention here, people are forced to leave their home country due to conflict. In most other cases, the prospective of a better job and living standards are closely linked to the decision to migrate. Intentionally or not, these movements of people and all accompanying aspects (such as remittances) have a significant influence on countries at both sides of the route. As has been stated in a report put out by the European Commission back in 2007, remittances are ever-increasing in volume and because of their potential to reduce poverty, policymakers at the highest levels are increasingly turning their attention to this phenomenon.

3

There are different theories we can follow when we can look at the role of remittances in economic development. The main theoretical strands I will discuss here are: Growth Theory, International Trade Theory and The New Geographical Economics Theory. All of these theories use different models and make different assumptions when analyzing the effects of migration and remittances on economic development. Therefore, they predict rather different outcomes in terms of convergence, making the issue of the impact of remittances on economic development an interesting dilemma.

If we compare the above-mentioned theories, we see that the neoclassical growth theory and international trade theory predict a convergence between economies as a result of bilateral migration and remittance flows between them. On the contrary, endogenous growth- and new geographical economy theories predict divergence as a result of these bilateral flows of people and remittances. I will go deeper into this in the literature review part of this paper.

Having been informed about the anticipated outcomes of these theories, it becomes very interesting to see what the empirics have to say about the effects of remittances on economic convergence. For this purpose, I make use of a dataset of bilateral remittance flows for more than two hundred countries over the period 2010 up to and including 2015. In this setting all countries are paired with each other, giving us information about the destination- and origin countries of remittances. The big size of this dataset makes it possible to effectively analyze the influence of remittance flows on the economic convergence between pairs of countries.

4

2. Why migrate in the first place?

As mentioned earlier, remittances would be hardly, if at all, existent if people did not leave their home country. Their motivation to relocate to a different country can takes several forms which have something to do with their home country or with the country they would like to relocate to, or both. Table 1 summaries the different factors, divided into push- and pull factors, that can play a role in the decision to migrate. As can be seen from the table, the decision to migrate is hardly an easy one and there seem to be several distinctive reason why people do so. However, searching for a better live in all its forms, seems to be the common denominator.

Table 1: Motives for migration1

Push factors Pull factors

Economic and demographic

Political

Social and cultural

Poverty

Unemployment Low wages High fertility rates Lack of basic health and education

Conflict, insecurity, violence Poor governance

Corruption

Human rights abuses Discrimination based on ethnicity,

gender, religion and the like

Prospects of higher wages

Potential for improved standard of living

Personal or professional development

Safety and security Political freedom

Family reunification

Ethnic (diaspora migration) Freedom from discrimination

Besides economic-, social, cultural and political concerns all can be an important motivator to relocate. In this regard, there is a distinction between economic- and non-economic migrants. For instance, refugees and asylum seekers who are escaping their home country due to conflict or human rights abuse, are considered as non-economic migrants. Furthermore, family reunification can also be an important factor in pushing people to leave their home country.

5

Massey et al (1993)2 _{provide a collection and summary of the different kinds of international}

migration theories. In this paper, the analysis is centered around the theory that people migrate because of prospects of earning a higher wage.

In the 50’s and 60’s of last century, Lewis’ (1954) model was given a central place in research concerning migration. In his mode of economic development, there was put an emphasis on the labor supply which was assumed to be unlimited and even though it was not explicitly developed as a migration model, it functioned as a tool to explain internal migration from rural- to urban areas. Here, the need for laborers in the urban sector is fulfilled by workers from rural areas. This theory and its extensions, as developed by Harris and Todaro in the late 1960’s and early 1970’s ( Todaro, 1969; Harris and Todaro, 1970), suggest that the differences in the demand for- and supply of labor cause internal migration. In their model, Harris and Todaro (1970) suppose there exists a politically determined urban wage which is substantially higher than the agricultural one in which domain laborers maximize their expected income and are willing to move accordingly. This model is later on extended to try to explain international migration as well.

In light of the analysis in this paper, the focus lies on the international relocation of people due to wage differences between their home country and their destination country. According to this view, migrants will leave labor abundant, capital scarce countries and move to labor scarce, capital abundant countries where their expected wage is higher. This movement of people across borders eventually leads to an equalization of wages between regions; labor supply in in initially labor abundant countries drops and in the process wages in those countries increase while the increase in laborers in initially labor scarce countries leads to a drop in wages. The decision to migrate can also be approached from a micro-economic point of view in which case individuals weigh out the benefits of migrating against the costs of doing so. As stated by Sjaastad (1962), in this case migration is seen as a form of investment in human capital. People will migrate only when, given their skills level, the expected benefits of migration exceed the initial costs of relocating abroad. Their initial investment ranges from travel costs to the costs associated with learning a new language. As such, the decision to migrate can be very

6

heterogenic, leading to different behavior from people originating from the same country. This approach is less frequently used than the one discussed above since it focuses on individuals’ socioeconomic characteristics.

So far, the theories that have been discussed all look at the decision to migrate from an individual’s point of view. A different approach to look at migration comes from the new economics of migration theory which suggests that the decision to migrate is not solely made by an individual who’s looking for a better live abroad. Instead, Stark and Bloom (1985) point out this theory implies that the decision to migrate is made by complete households and families. These households and families decide upon migrating in order to hedge against several market failures in their home country. Since the necessary institutions to contain the risk of loss of income , such as insurance- and financial markets and government aid, are underdeveloped or completely absent, such families from less developed countries will resort to migration to diversify their income stream. According to this theory, wage differentials are not a necessary condition for international migration since families have the incentive to diversify against risk either way.

As Mansoor and Quillin (2006) argue, differences in wages or income are a necessary but not sufficient factor in explaining migration. They argue that a broader set of factors which define quality of life in general are also essential in order to properly explain the reasons migration. Furthermore, they add that ‘differences in political stability, human rights situations and the general rule of law may also affect migration since these factors can proxy for the level of individually perceived security’ (Mansoor and Quillin, 2006, p.78). This means that risk-averse individuals that enjoy a comfortable life, will not be eager to migrate as opposed to their risk-seeking counterparts who do not have the same comfort and would jump at the opportunity for a better life abroad.

7

3. Data and Methodology

3.1 Data

In order to properly answer the research question of this paper, I make use of a dataset of estimates obtained from The World Bank which are based on the methodology developed by Ratha and Shaw (2007). This dataset provides bilateral remittances data for more than two hundred countries over the period 2010 up to and including 2015 where every observation represents a pair of countries. See Appendix I for a list of these countries. In every set of countries, one is a ‘source’ country and one is a ‘destination’ country. As such it represents the flow of remittances from every source country to every destination country. In this setup, I have estimated for every pair of countries a coefficient indicating the convergence between the two countries, the flow of remittances and some control variables. As mentioned earlier, the remittances dataset are estimates for five consecutive years, 2010 up until 2015, which makes it possible to track the effect of remittances over the years. The remittances at year t are estimated using host country and origin country incomes and estimated migrant stocks from year t-1. Since remittances are inherently linked to migration, this setup makes it possible to measure the effects of migration from country A to country B on the convergence between those two countries. As of date, it is the most comprehensive data on bilateral remittances (and as such also migration) that I’m aware of in the literature. In Appendix II you can find a detailed description of the acquired data used in this analysis.

3.1.1 Dependent variable: Convergence

8

set of countries then currently is the case. The GDP per capita data (PPP converted, in current international $) are obtained from the World Development Indicators, World Bank.

To measure convergence, I follow the methodology of Ben-David and Kimhi (2000) since their analysis is focused on the impact of bilateral trade on convergence. This is very similar to the question I want to answer in this paper which makes it a very suitable method to use here as well. The measure of convergence in this paper is as follows:

|𝐺𝐴𝑃

| − |𝐺𝐴𝑃

|,

Where i refers to remittance sending country (i.e. source country of remittances), j refers to remittance receiving country (i.e. destination country of remittances), t is the time indicator referring to a year in the period of interest (i.e. if t=2010 then t+1=2011), |𝐺𝐴𝑃𝑖𝑗,𝑡| is the

absolute value of the gap in GDP per capita between source and destination country at year t and |𝐺𝐴𝑃_{𝑖𝑗,𝑡+1}| is the absolute value of the gap in GDP per capita between source and destination country at year t+1. According to this methodology, the convergence between countries is measured by the decline in the gap of GDP per capita between them over time.

3.1.2 Independent variable of focus: Bilateral remittances

9

illustrates how remittance flows to the developing world have developed over the years. As can be seen, they amount to more than three times the official development aids (ODA) and as such are expected to play a significant role in the developing world.

As mentioned earlier, remittances are inherently linked to migration and as noted by Bauer and Zimmermann (1999), countries with a large population may have a higher number of emigrants. Since these emigrants send remittances back home to their country of origin, it is important to account for the differences in population size of the remittance receiving countries i.e. country of origin. In order to control for this country size effect, the amount of remittances sent to destination country is weighted by the population of the destination country. However, due to the large amount of small values for the remittance/population ratio, it is highly skewed to the left, so in order to make it more normally distributed, I transform this variable by taking the natural logarithm.

3.1.3 Control variables

In order to account for factors other than remittances that might influence the relative performance of countries, I incorporate a set of control variables in my estimation. In order to

10

select the appropriate control variables I follow the standard growth models which have been widely used by others (Barro, 1996; Chami et al, 2003; Mundaca, 2005). These variables are:

(i) Investment: Gross fixed capital formation as a share of GDP (ii) Human Capital: gross secondary school enrollment rates (iii) Inflation: GDP deflator (annual %)

(iv) Openness: Export plus Import as a share of GDP (v) Government effectiveness

All of these variables enter the equation for both source and destination countries. Except for government effectiveness, all of them are obtained from the World Development Indicators, World Bank. The government effectiveness variable is obtained for the Worldwide Governance Indicators.

Government effectiveness captures perceptions of the quality of public and civil services and the degree of the independence of civil services from political pressure, the quality of policy formulation and implementation as well as the credibility of the government’s commitment to such policies. Its estimate can take values between approximately -2.5 (weak) and 2.5 (strong).

3.2 Methodology

To investigate the effect of remittances on the change in income per capita gabs, I estimate a number of equations. The specifications of the equations depends on the assumptions made about the error term and the endogeneity of remittances. Since we have a dataset that covers a period of five consecutive years, the correct procedure is to set the dataset as panel data and run the regressions. There are several kinds of regressions which could be suitable for the analysis in this paper. I’m basically interested to analyze the following:

11

(ii) If country A sends more remittances to country B than to country C, does the income gap (measured as the difference in GDP per capita) between A and B decline more than the gap between A and C?

The hypothesis to be tested in this regard is of two parts: a) An increase in remittances sent from country A to country B leads to a drop in the income gap between the two countries, and (b) if country A sends more remittances to country B than to Country C, there will be a stronger decline in the income gap between countries A and B than between countries A and C. A negative coefficient for the remittance variable in my regression will answer both parts of this hypothesis with a ‘yes’. A positive coefficient for the remittance variable would suggest the exact opposite. That is, an increase in remittances from country A to country B leads to a divergence of GDP per capita between the two countries.

Keeping this in mind, there are several ways to perform the necessary regressions. One way would be by estimating a panel model with random effects; the other would be by estimating the model with fixed effects. As stated by Green and Turkey (1960), the choice between a fixed effects and a random effects model is dependent on whether the sample that is used exhausts the population (i.e. is rather big) or that it is just a small part of the population. This analysis makes use of bilateral remittances data of 210 countries which means it represents a big part of the population and as such, a fixed effects model would be the appropriate one to use.

However, to illustrate the differences between the two methods, I estimate the model using both fixed- and random effects. The fixed effects model captures the panel variable-fixed effects (pair of the source and destination country) controlling for unobserved country-level heterogeneity. Since I expect that part of the time series variation in the change in GAP is to be explained by overall time trends, I add a time-series dummy to capture the time-fixed effects as well. This being said, the Hausman test results suggest I should use a fixed effects model. As such, the only results that are reliable enough to interpret accordingly, are the results from the fixed effects estimations. However, it remains interesting to compare both random and fixed effects to see what differences in estimations these two methods produce. The general model to be estimated in this analysis is as follows:

|𝐺𝐴𝑃_{𝑖𝑗,𝑡+1}| − |𝐺𝐴𝑃_{𝑖𝑗,𝑡}| = 𝛽₀+ 𝛽₁ln (𝑟𝑒𝑚𝑖𝑡𝑡𝑎𝑛𝑐𝑒𝑖𝑗

12

Where |𝐺𝐴𝑃𝑖𝑗,𝑡+1| − |𝐺𝐴𝑃𝑖𝑗,𝑡| is the change in the absolute income gap between source (i)-

and destination (j) countries of remittances; 𝑟𝑒𝑚𝑖𝑡𝑡𝑎𝑛𝑐𝑒_𝑖𝑗 refers to the amount of remittances (in thousands of dollars) sent from country i to country j, i refers to source country (i.e. remittance sending country), j refers to destination country (i.e. remittance receiving country), 𝑋_𝑖 is the matrix of control variables for the source country, 𝑋_𝑗 is matrix of control variables for the destination country and 𝑒_𝑖𝑗 is the error term.

There is however a risk of biased results due to measurement errors or omitted variables. This concern comes from the fact that the quality of remittances is based on the migration stocks in countries that are dependent from the quality of the population consensus, as also noted by Ratha and Shaw (2007). Furthermore, the omission of a variable that influences both remittances and income per capita might also cause biased results. Equation (1) above assumes that all of the independent variables are exogenous. However, since remittances might not be exogenous to the changes in income per capita gaps, estimating solely this OLS regression might lead to biased results. That is, I expect that countries with a high level of GDP per capita will attract more immigrants who in turn will send remittances to their home country.

In order to account for these concerns, I transform the estimation by using an instrumental variable (IV) approach. The two-stage least square method conducted in this matter uses several instrumental variables to be instrumented on the endogenous variable of interest, remittance.

𝑙𝑛 (𝑟𝑒𝑚𝑖𝑡𝑡𝑎𝑛𝑐𝑒𝑖𝑗

𝑝𝑜𝑝_𝑗 ) = 𝛼0+ 𝛼1𝐼𝑖𝑗 + 𝑒𝑖𝑗 (2)

13

have a positive sign since it decreases the cultural distance between countries. Geographical distance is expected to have a negative effect since, firstly , the larger the distance between two countries, the harder it is to migrate between the two, and secondly, sending remittances over a larger distance brings along higher transaction costs and thereby reduces remittance flows. Almost all studies on migration find a negative and significant effect of distance on migration and thus on remittances as well.

All of these instruments are obtained from CEPII3 and all of them, except for geographical distance, are dummies. Contiguity takes value 1 if the two countries share a border and 0 otherwise; common language takes value 1 if the two countries share a language and 0 otherwise and colonial ties takes value 1 if the two countries have ever shared a colonial link and 0 otherwise. Geographical distance between countries in kilometers is measured following the great circle formula, which uses latitudes and longitudes of the most important cities/agglomerations (in terms of population).

However, upon performing the tests for the validity of the above mentioned instrumental variables, I found that only contiguity and weighted distance are valid to be used in this setting. Running the regressions with instruments that are not valid leads to biased results and is obviously not what we want here. This means I end up using an IV model with just contiguity and distance as the two instrumental variables for remittances. These instruments’ validity and this model’s specification is tested by means of a test of overidentification, which is explained in more detail below.

Furthermore, in order to make sure that our variable of interest, remittances, is indeed endogenous, I perform an endogeneity (see Appendix III) test and conclude that there is endogeneity . This means I have made the correct decision to transform the model by using an IV regression. In order for these instruments to be appropriate, they need to be valid and fulfill two conditions: (i) they need to be relevant and (ii) they need to be exogenous. In this regard, the results of equation (2) are provided in Table 2 in the next section as well as the covariance’s of the instruments in Appendix V which tell us that the instruments contiguity and weighted distance are indeed relevant to be used in this setting. To test whether these instruments are valid, I perform a test of overidentifying restrictions (Appendix IV) and conclude that, with a

14

Sargan-Hansen statistic of 0.5340 𝑐ℎ𝑖2 (1) and a P-value of 0.4649, they indeed are valid and the model is correctly specified.

3.2.1 Limitations of the data and model

It is worth noting that, even though I have used a very large bilateral dataset with over 200 countries, the dataset contains many missing observations. This could bias the results since it doesn’t provide us with a full picture of the interrelationships of these variables. Understandably, it is rather hard to keep track of all remittance flows between all countries in the world, especially in developing and underdeveloped countries where unstable economic and political conditions could have a significant effect on its ability to accurately keep track of such flows. As is stated in a report recently published by the European Parliament’s Committee on development, bearing in mind that most datasets only keep track of remittance flows that go through official channels (Kamuleta, 2014), a vast amount of remittances goes by unnoticed by the authorities. This again reinforces the statement made above that the data is far from complete and only provides part of the picture on the complex effects of remittances.

Furthermore, the timespan of the dataset of five years is rather shot, only providing an illustration of the short run and leaving out possible significant long run effects. This, combined with many missing observations, makes it hard to accurately track the effects of remittances on income gaps in the longer run and produce results which are reliable in the real world.

4. Literature review

15

the multiplier effect. When demand for goods and services increases, supply follows and as a consequence economic development takes place. Remittances are used for varying purposes, ranging from investments to consumption and even though their main use is for consumption, it isn’t the only one. Albeit in lower frequencies, investment opportunities as much as charities or insurance are just as well financed through remittances sent from abroad by emigrants. Moreover, these migrants use their savings to invest in real estate, small business or other assets in their home country because they know the market there better than in their host countries and probably plan to return there later. Furthermore, as argued by Brown et al (2013), it is hypothesized that the deposit of remittance receipts in banks in remittance receiving countries makes them more able to increase the availability of loanable funds to the population. From economic theory we know that an increase in the availability of loanable funds gives a boost to investments, consumption and as such, helps harness economic growth and increase income levels. This being said, the main use of remittances remains consumption, leaving little use for investments and other initiatives for these funds (Chami et al., 2003; Taylor et al., 1996; Durand and Massey, 1992; and Papademetrious and Martin, 1991).

In the following part, I discuss some of the theories that predict the effects of migration and remittances on economic development. They are largely focused on migration, but since migration is inherently linked to remittances, by which migrants can influence their country of origin, it is very much of interest to be considered here as well.

4.1 Growth Theory

As stated by Hinojosa Ojeda (2003), Terry and Wilson (2005), Giuliano and Ruiz-Arranz (2009), Mundaca (2009) and Ramirez and Sharma (2009), the growth-focused analysis allows for a possible interaction between remittances and overall financial development in estimating growth equations for countries on the receiving end of remittances.

16

These models, however, do not seem to take into account the role of factor mobility in influencing economic convergence. As evidenced by Taylor and Williamson (1995), we indeed should take into account the role of factor mobility because the international movement of people does appear to have a great role in pushing for convergence. Furthermore, Abramovitz (1986) suggests that on top of factor mobility, international trade is also needed to be taken into account.

Open economy growth models are a step forward in this direction as they do account for factor mobility and its influence on convergence. Faini (2003) explains that if there would be perfect factor mobility, there would be instantaneous convergence. In other words, if laborers could freely migrate from capital scarce to capital abundant countries, real wages would eventually equalize and economies would reach a steady state in which they have the same income levels. In this model, the migration of laborers would go on until economies have fully converged with each other. However, as Baumol (1986) states, empirical research suggests absolute convergence only holds for a relatively small group of countries which share similar preferences and technology. Logically, the perfect factor mobility assumption in these models does not apply to most countries in the world. This attribute only seems to hold for highly integrated economies such as the European Union. Furthermore, as emphasized by Sala-i-Martin (1996), it is rather strange that these neoclassical models assume countries are the same in every way except for their initial capital level. If there is taken account for initial differences in preferences, technology and capital levels explicitly, we arrive at the conditional growth model. However, the empirics do not seem to support this model either.

17

higher than in less developed countries where there might not even be any growth at all (Romer, 1986). The long-run growth rate in this model is driven by the knowledge accumulation of forward-looking, profit-maximizing agents and physical capital. Both Romer (1986) and Bertola (1992) suggest that economic integration does not necessary lead to efficient allocation of labor and as such, may even deepen the gap between countries in terms of economic performance.

It seems that neither exogenous nor endogenous growth model can provide us with a clear-cut picture of the role of remittances and migration in economic convergence and divergence.

4.2 Trade Theories

According to O’Rourke (1996), Taylor and Williamson (1994) and Williamson (1996), at least in the period from 1870 until the First World War, migration flows have had a stimulating effect on convergence. The evidence they found suggests that even then, migration played an important role in bringing economies closer to getter. After WW II, it has often been argued that growth is also influenced by the openness of an economy. Poor countries saw this as a key part of their strategy to catch up with more developed, open economies (Sachs and Warner, 1995). The underlying idea of openness is that countries are forced to specialize and export goods and service they are relatively good at producing. This is achieved by exploiting their abundant factors, such as labor, capital or natural resources, and increase the efficiency of production. The Heckshler-Ohlin model predicts that this international trade in goods and services equalizes factor prizes. Capital-abundant countries produce relatively more capital-intensive goods and labor-abundant countries produce relatively more labor-capital-intensive goods and both types of countries import the other’s products.

18

These findings reassure that migration has always played an interesting role in economic convergence between countries and regions. The derivation hereof, remittances, are obviously an interesting aspect to be analyzed. As put forward by Leon-Ledesema and Piracha (2001), ‘although the poorest seldom have the means to migrate, remittances have been shown to play an important role in poverty alleviation for migrant households and in sub-national areas of out migration, especially in countries with high income and wage disparities’. Mundaca (2005) states that, if canalized efficiently, remittances can be an important source of economic development and thereby lead to convergence with economically stronger countries. Taylor (1999) adds to the discussion by suggesting that remittances lead to a rise in demand for goods and services by which it helps foster economic growth. Additionally, according to Ratha and Mohapatra (2007), remittance flows tend to increase in times of crises, making it a strong financial stabilizer in volatile times.

4.3 New Geographical Economics Theory

The central critic of this strain of theory to the international trade theory, as stated by Krugman (1991), is the fact that in the international trade theory, countries are treated as one-dimensional points on a map where there are no transportation costs to speak of. In economic geography studies, the focus lies on explaining the tendency of populations to agglomerate in a certain area within and across countries, regions and the world. Krugman developed a model which is based on free labor movement to explain this behavior. Agriculture, which is characterized by constant returns to scale, and manufacturing, which is characterized by increasing returns to scale, are the only two sectors in this model. Demand for manufacturing products comes from both sectors and firms face transportation costs for moving between regions/borders.

19

Let us start with a base scenario where the population is evenly distributed between two regions, Region 1 and Region 2, and let us assume that this is a long-run equilibrium. Slight changes in the distribution of the manufacturing workforce set in motion a process of adjustment that results in a different long-run equilibrium. For instance, for some reason, when a small portion of the mobile workforce decides to move to another region, firms follow workers because of now-larger market. As larger market attracts more firms, the varieties produced in the regions increases. This increase in the number of varieties pulls down the price of each variety because of a “love-of-variety-effect” (Brakman et al, 2001). The decrease in prices leads to an increase in real wages, creating an extra stimulus for further migration to this region. Thus, as Krugman (1991) argues, firms follow workers and workers follow firms. This is what Myrdal (1957) refers to as “circular causality”. Reallocation encourages further reallocation and consequently the two regions diverge.

If we try to apply the theory of geographical economics to the real world, we arrive at divergence as a result of migration. As the model suggest, people react to differences in real wages and tend to move to regions where real wages are high. This is exactly in line with the theories of migration, especially the one developed by Harris and Todaro (1970). What follows from the model described above is that migration will lead to the agglomeration of economic activity in migrant receiving countries which increases the gap between the migrant receiving and the migrant sending countries rather than decreasing it.

20

investment of remittances in countries that the lack necessary economic infrastructure and policy to channel these funds effectively.

All in all, the theories discussed above seem to predict contradictive results in terms of the effects of migration and the subsequent remittance flows on economic development. Next, we turn to the empirics to see what effect remittances have on income gaps between remittance sending and remittance receiving countries.

5. Estimation Results

As mentioned earlier, I estimate the model using both random and fixed effects to illustrate what differences there are in the results of both methods. The tables below show the results from the OLS and IV random effects and fixed effects estimations, with the IV fixed effects estimations being the most reliable ones. The results of the OLS estimations of the effect of remittances on the change in income per capita gaps between 2010 and 2015 are reported in Table 2. Both the random and fixed effects results are reported next to each other so we can easily compare them. The suffix i indicates the remittance-sending countries (source countries) while the suffix j indicates the remittance-receiving countries (destination countries). Even though the IV results are the more accurate ones because of our endogeneity problem when conducting regular OLS regressions, let us first examine the OLS results.

5.1 OLS estimation

21

For now, let us have a look at the other variables estimates and see whether they are what we would expect.

Table 2

:

Estimation method RE FE

Dependent variable Change in

Ln( / ) 0.027 0.295 (0.17) (0.32) -0.019 -0.041 (8.53)** (7.40)** 0.008 0.018 (2.98)** (4.12)** -0.014 -0.022 (7.69)** (5.11)** 0.003 -0.014 (1.84) (2.19)* -0.008 -0.006 (4.63)** (1.81) -0.014 -0.016 (6.58)** (3.61)** 0.002 0.006 (0.63) (1.42) -0.010 -0.012 (5.06)** (2.94)** -0.002 -0.006 (1.35) (1.09) -0.009 -0.011 (4.61)** (3.50)** Constant 30.300 64.684 (6.02)** (3.44)** N 49,694 49,694 * p<0.05; ** p<0.01

22

insignificant while government effectiveness is significant at the 5% level. However, not all of them are of the initially expected sign. Investment in the source country seems to have a negative effect on the income gap between source and destination country. This seems rather counterintuitive since we would expect the exact opposite: if the source country invests more in its economy, this will increase the income-gap with the destination country since source-country incomes will rise due to the investments that are made.

When we look at the human capital variable, which is proxied by gross secondary enrollment , it seems to have the correct (positive) sign in both fixed effects and random effects estimations. Indeed, when the source country has a higher-educated population than the destination country, incomes there will generally be higher. So, all things being equal, an increase in the overall education level in the source country, will lead to an increase in the income gap with the destination country.

Looking at the variable for Openness, which is estimated by imports plus exports as a share of GDP, it seems to have the correct (negative) sign on the income gap. This can be explained by economic theory which suggests that more open borders will make it easier for people and companies to operate abroad and trade with each other. This helps countries interchange knowledge and benefit from each other to ensure economic development, leading to a drop in income gaps between these trading countries.

Government effectiveness seems to be statistically insignificant when estimated with random effects but according to the fixed effects results, it does seem to have a significant effect with an unexpected negative sign. We would have expected a positive sign here since having a government that is more effective and trustworthy will increase the ability to ensure economic growth and stability, leading to higher wages and overall economic development. This is expected to have a diverging effect with countries that score lower on government effectiveness. But then again, we will just have to see what the IV results tell us in order to appropriately judge these estimates.

23

economy and are expected to diverge from destination countries which is the opposite of what this result suggests.

When we look at the estimates for the destination countries of remittances, we see that only investment, openness and inflation seem to be statistically significant in both random and fixed effects estimations. All the estimates seem to have the correct (negative) sign, except for the statistically insignificant human capital variable.

When we look at the estimate for the investment variable, we can see that an increase in investments in destination countries seems to lead to a drop in the income gap with source countries. Looking at this from an macro-economic perspective, we can safely state that investments lead to more economic development in a given country. In this particular case, an increase in investments in remittance-receiving countries will thus lead to convergence with more developed remittance-sending countries.

Even though the human capital variable is insignificant, both methods produce a positive sign where we would expect the opposite. Intuitively, we would expect a negative sign since higher education leads to an increase in an individual’s job opportunities and subsequent earnings. Reasonably, we could state that, given the structure of the economy and government in these less developed economies, having a higher education level does not necessarily lead to more job opportunities. This being said, we have to note that the reported effect is statistically insignificant and as such not very reliable for the analysis in this paper.

As expected, openness seems to have a statistically significant negative effect on the income gap. This coefficient is significant according to both random and fixed effects methods. It can be explained in the same line of thought as for the source country, i.e. conducting more trade with other countries leads to economic convergence ,and consequentially, of income levels. Government effectiveness seems to have the correct sign but unfortunately not significant enough. The same explanation holds here as for the source country but the conclusion is the exact opposite, however. For the destination country, having a better functioning and more effective government which stimulates economic development leads to higher income levels, and thus to convergence with the more developed source countries.

24

source countries. Still, the effects of the inflation rate is rather complex and hard to explain since it could go both ways, depending on the initial level of inflation. Apparently, in the period analyzed in this paper, the inflation rate in destination countries was such that its increase has led to economic convergence with source countries.

This initial OLS regression using both random and fixed effects has already given us an idea of what to expect in terms of the relationship between remittances and income gaps between countries. In the next section, I lay out the results of the IV estimations.

5.2 Instrumental Variable estimation

As stated earlier, in order to control for endogeneity, measurement errors or a potential omitted variables bias, I run an instrumental variables regression to try to produce more reliable estimates.. These results are reported in Table 3. Again, both random effects and fixed effects estimates are reported in order to illustrate the differences in outcomes between the two methods. Both methods seem to produce similar results in magnitude and sign, except for the one crucial variable of interest in this paper, remittances. The estimate of remittances leads to very different conclusions depending on which method (random or fixed effects) of estimation is used. According to the random effects estimation, an increase in remittances leads to a drop in the absolute income gap and as such steers countries to economic convergence. In line with the main hypothesis of this paper, this result can be interpreted as follows: (i) when country A sends more remittances to country B, the income per capita gap (estimated by GDP/Capita) between the two countries will become smaller (i.e. they will converge). The second interpretation of this result is: (ii) when country A sends more remittances to country B than to country C, the income per capita gap between country A and B will become smaller than the gap between country A and C (i.e. country A will converge more strongly with country B than with country C).

25

remittances on economic growth. Namely, higher real exchange rates leads to less exports (and more imports) which in turn hurts economic growth and causes them to get farther behind source countries in terms of economic growth rates. This diverging effect of remittances is also found by Rodrik (2007), who states that the production of tradable goods suffers from market failures and weak institutions as a consequence of an exchange rate appreciation. This in turn undermines economic growth, leading to divergence. Furthermore, as argued before, remittances are mostly spent on consumption and according to Lipton (1980) and Brown and Ahlburg (1999), this consumption likely consists of foreign goods which adds little to nothing to their home country economy. Additionally, Chami et al (2003) also find that remittances have a negative effect on income levels and as such lead to divergence instead of convergence. They state that when remittances are saved or invested, it is typically done in real estate, property and jewelry and that this does not necessarily add to the economy as a whole.

When we look at the other variables, we see that most of them are of similar magnitude and sign as the OLS regressions. The explanation about their sign given in the OLS results part above are therefore still accurate, with just a slight difference in their magnitudes. For both OLS and IV estimates in general, the fixed effects results, which are the ones most reliable in this analysis, seem to be larger in magnitude that the random effects results.

Table 3

:

Estimation method RE FE

Dependent variable Change in

26 0.006 0.008 (1.90) (1.75) -0.010 -0.013 (5.31)** (2.89)** -0.000 -0.008 (0.10) (1.33) -0.009 -0.012 (4.78)** (3.52)** Constant 4.218 76.293 (0.45) (8.65)** N 47,636 47,636 * p<0.05; ** p<0.01

Ln( / ) is the instrumented variable Instruments: contiguity and weighted distance

5.3 The instruments

In addition, Table 4 below provides the first stage results of the two-stage-least-square (IV) regression. As we can see, all instrumental variables seem to have a highly statistical significance at the 1% level and are of the correct sign. As expected, sharing a border has a positive effect on the magnitude of remittances send. Even though small in magnitude, the estimate for the other instrumental variable in this analysis, the weighted distance, seems to be of the correct sign as well. I believe this low magnitude can be partially explained by the fact that, even though people are more likely to migrate to nearby countries, with the current costs of transportation, distance has become much less of an impending factor to migration than it used to be several decades ago. Having noted this, the coefficient for distance remains to be statistically significant and a strong instrument for remittances.

Table 4: Estimates from the instrumental variables regression on the endogenous variable

27

5.4 Additional robustness checks

To test the robustness of these results, I have created a subsample with remittances flows from several highly developed remittance sending countries to the rest of the world to see whether the same results hold in a smaller sample as well. The countries in the subsample are selected on the basis of two criteria: 1) they most have a high GDP/capita and 2) they must have a significant immigrant population. Furthermore I have looked for countries from more than one single continent for diversification. The nine countries I have selected based on these criteria are: Australia, Austria, Belgium, Canada, Denmark, France , Germany, United Kingdom and the United States. Table 5 and 6 report the OLS and IV results of this subsample analysis.

Table 5

:

Estimation method RE FE

Dependent variable Change in

28

Constant 87.506 -96.785

(1.79) (0.70)

N 6,188 6,188

* p<0.05; ** p<0.01

Again, I have included both random effects and fixed effects estimates for comparison. The initial OLS results seem to be promising and suggest that what we initially would expect, namely that remittance flows cause convergence between countries. The fixed effects estimator for remittances seems to be quite a bit stronger than the random effects estimator, suggesting that when we control for country fixed effects and time fixed effects, remittances seem to be having a strong negative effect on the income gap between two countries. Most of the other variables seem not to be statistically significant, except for the random effects estimate for destination country investment; both estimates for destination country human capital formation, openness and inflation and the random effect estimate for destination country government effectiveness. Except for the destination country human capital formation, these statistically significant estimates do seem to have the correct sign. However, the problem of endogeneity still exists in this subsample so an IV regression is still the way to go here. When we look at the IV results, we see that the remittance estimate seems not to be statistically significant. This can be explained by the fact that the size of the subsample might not be large enough to accurately capture the relationship between remittances and convergence. This shows us that when examining the relationship between convergence and migration-related topics such as remittances, one must try to gather as much data as possible to be able to perform an accurate analysis. This robustness check has shown that the relationship of remittances and economic convergence between nations is very complicated and sensitive to dataset size.

Table 6

:

Estimation method RE FE

Dependent variable _{Change in}

29 0.004 0.172 (0.56) (1.35) -0.046 0.052 (1.15) (0.68) 0.002 -0.006 (0.33) (0.86) -0.014 -0.014 (3.42)** (1.61) 0.035 0.034 (5.72)** (4.28)** -0.029 -0.043 (7.99)** (4.74)** -0.018 -0.008 (5.45)** (1.15) -0.037 -0.049 (9.86)** (8.22)** Constant 63.606 -76.597 (1.30) (0.54) N 6,004 6,004 * p<0.05; ** p<0.01

Ln( / ) is the instrumented variable Instruments: contiguity and weighted distance

All in all, this analysis has provided some interesting results on the effects of international remittances on the convergence of per capita incomes across countries. The appropriate method of estimation on this panel data set seems to be using instrumental variables in which I also control for country fixed and time fixed effects. While this method produces an estimate for remittances which suggests economic divergence, the random effects estimate suggests the exact opposite. Having consulted theory and statistical testing and bearing in mind the limitations of this study, I must conclude that the only reliable estimates are those with fixed effects and that indeed remittances flows between countries result in economic divergence.

6. Conclusion and Recommendations

30

income per capita is defined by the GDP per capita since this is the most widely available income measurement for a broad set of countries, making it the most suitable one to use for the analysis in this paper. Ideally, I would have used real income, wages or labor productivities but since these data are not available (yet) for such a broad set of countries, these variables were not suitable to use here. When such data becomes available, it would be interesting to see whether the results would be different. For now, I leave this endeavor to be dealt with by future studies on this extremely interesting subject.

As noted earlier, remittances are inherently linked to migration since they are send by people who live and work outside of their home country. The recent international movement of people, a majority due to war, makes this topic of discussion particularly interesting. Aside from war and conflict, there are several other factors which push people to leave their country of origin, most of which have an economic perspective. Moreover, there are social and cultural factors such as freedom from discrimination and family reunification, which motivate people to migrate. Willingly or not, this international migration influences countries on both sides of the route and its effects need to be considered carefully. Hence, this paper has tried to lay out the impact of one of the biggest accompanying aspects of migration, namely remittances, on a worldly scale. Although having been studied for several times, the results are far from being coherent. Some studies find economic convergence due to remittances, while others find divergence instead. Therefore, I have taken up the effort and tried to add to the existing literature by making use of a large bilateral dataset consisting of 210 countries.

31

completely due to the endogeneity problem that is present in these regular OLS regressions. When we turn our head to the IV results, we see a contradiction between the random effects and fixed effects estimate. While the random effects estimate suggests a convergence of GDP per capita, the fixed effects estimate suggests divergence of GDP per capita due to remittance flows. After testing the model by means of a Hausman test, the most accurate results seem to be produced by the fixed effects method which incorporates country fixed effects as well as time fixed effects. This model provides enough evidence that remittances lead to divergence rather than convergence of GDP per capita. Other papers that have found the same result suggest that this has to do with the appreciation of local currencies due to incoming remittances as well as with the eventual usage of these remittances, which seems to be mainly consumption of imported goods. Furthermore, remittances that are not used for consumption, are mainly used for investments in real estate, property and jewelry. In this sense, remittances add little to the home country economy as a whole.

To check whether the same results are found when using a select set of rich remittance sending countries, I have estimated the same model using only Australia, Austria, Belgium, Canada, Denmark, France , Germany, United Kingdom and the United States as the remittance sending countries. The initial OLS results produce significant negative results, suggesting economic convergence due to remittance flows between two countries. However, as we know, these results cannot be completely trusted due to the endogeneity problem between remittances and GDP per capita. The IV results, just as with the regular regressions, suggest contradictive results depending on whether random effects or fixed effects are used. The random effects estimates suggest convergence, while the fixed effects estimate suggests divergence. Even though this fixed effects estimate seems not to be statistically significant, it still suggests a diverging effect of remittances on the GDP per capita gap between countries.

32

flows. Lastly, the time period covered here is rather short and only provides an illustration of the short run, leaving out potential implications of remittance flows in the long run. Consequently, there is much room for improvement for future studies on this subject. This includes using a more complete dataset over a longer time period as well as more accurate measurement of income per capita, such as real income, wages or labor productivities instead of ‘just’ GDP per capita to measure the effects of remittances on economic convergence.

Appendix

Appendix I List of countries

Afghanistan Albania Algeria American Samoa Andorra Angola

Antigua and Barbuda Argentina Armenia Aruba Australia Austria Azerbaijan Bahamas, The Bahrain Bangladesh Barbados Belarus Belgium Belize Benin Bermuda Bhutan Bolivia Bosnia and Herzegovina Botswana Brazil Brunei Darussalam Burundi Cambodia Cameroon Canada Cape Verde Cayman Islands

Central African Republic Chad Channel Islands Chile China Colombia Comoros

Congo, Dem. Rep. Congo, Rep. Costa Rica Côte d'Ivoire Croatia Cuba Cyprus Czech Republic Denmark Djibouti Dominica Dominican Republic Ecuador

Egypt, Arab Rep. El Salvador Equatorial Guinea Fiji Finland France French Polynesia Gabon Gambia, The Georgia Germany Ghana Greece Greenland Grenada Guam Guatemala Guinea Guinea-Bissau Guyana Haiti Honduras Hong Kong SAR, China

Hungary Iceland India Indonesia

Iran, Islamic Rep. Iraq

33 Bulgaria Burkina Faso Jordan Kazakhstan Kenya Kiribati

Korea, Dem. Rep. Korea, Rep. Kosovo Kuwait Kyrgyz Republic Lao PDR Latvia Lebanon Lesotho Liberia Libya Liechtenstein Lithuania Luxembourg Macao SAR, China Macedonia, FYR Madagascar Malawi Malaysia Maldives Mali Malta Marshall Islands Mauritania Mauritius Mexico Micronesia, Fed. Sts. Moldova Monaco Mongolia Montenegro Morocco Mozambique Myanmar Namibia Nepal Netherlands New Caledonia Eritrea Estonia Ethiopia Faeroe Islands Oman Pakistan Palau Panama

Papua New Guinea Paraguay Peru Philippines Poland Portugal Puerto Rico Qatar Romania Russian Federation Rwanda New Zealand Nicaragua Niger Nigeria

Northern Mariana Islands Norway

Samoa San Marino

São Tomé and Principe Saudi Arabia Senegal Serbia Seychelles Sierra Leone Singapore Slovak Republic Slovenia Solomon Islands Somalia South Africa Spain Sri Lanka

St. Kitts and Nevis St. Lucia

St. Vinc. and the Grenadines

Israel Italy Jamaica Japan Turkey Turkmenistan Turks and Caicos Islands Tuvalu Uganda Ukraine United Arab Emirates United Kingdom United States Uruguay Uzbekistan Vanuatu Venezuela, RB Sudan Suriname Swaziland Sweden Switzerland Syrian Arab Republic Tajikistan Tanzania Thailand Timor-Leste Togo Tonga Trinidad and Tobago Tunisia Vietnam Virgin Islands (U.S.)

West Bank and Gaza

34

Appendix II

:

Variable Obs. Mean STD Min Max

Remittance Remittance/𝑝𝑜𝑝𝑗 Ln(remittance) Ln(remittance/𝑝𝑜𝑝𝑗) 166,390 166,390 58,252 58,252 .3796003 19.04183 -8.397842 -.8995471 16.18848 280.133 6.190465 3.596393 0 0 -31.83998 -16.23136 4839.526 24323.19 8.484572 10.09918 GDP/𝑐𝑎𝑝𝑖𝑡𝑎𝑖 GDP/𝑐𝑎𝑝𝑖𝑡𝑎𝑗 |𝐺𝐴𝑃𝑖𝑗| |𝐺𝐴𝑃𝑖𝑗,𝑡+1|-|𝐺𝐴𝑃𝑖𝑗,𝑡| 258,300 264,600 258,300 215,250 484.5935 482.469 400.3192 2.116423 348.174 348.9223 286.3401 157.5819 1 1 0 -1095 1102 1102 1101 1090 𝐺𝑐𝑓𝑖 𝐺𝑠𝑠𝑒𝑖 𝐼𝑚𝑒𝑥𝑖 𝐺𝑑𝑝𝑑𝑓𝑖 𝐺𝑒𝑓𝑖 𝐺𝑐𝑓𝑗 𝐺𝑠𝑠𝑒𝑗 𝐼𝑚𝑒𝑥𝑗 𝐺𝑑𝑝𝑑𝑓𝑗 𝐺𝑒𝑓𝑗 258,300 258,300 258,300 258,300 258,300 264,600 264,600 264,600 264,600 264,600 391.5537 184.6057 428.8846 515.2667 711.6642 386.4881 184.2222 425.6786 511.2198 711.5341 322.1309 222.4604 332.1415 331.854 396.7582 323.4385 222.3274 335.4627 331.6726 395.194 1 1 1 1 1 1 1 1 1 1 986 680 1035 1117 1381 986 680 1035 1117 1381

Appendix III

:

Robust score 𝑐ℎ𝑖2 40.0784 (p = 0.000) Robust regression F(1, 54759) 40.2237 (p = 0.000)

Appendix IV: Test of overidentifying restrictions, result: instruments are valid

Cross-section time-series model: xtivreg fe Sargan-Hansen statistic

P-value

35

Appendix V Covariance matrix instruments, result: instruments are relevant

Ln(remittance/popj)

Contiguity

Distance (weighted)

0.157426 -5288.9

Appendix VI Hausman test, result: use fixed effects

𝐻₀: differences in coefficients not systematic 𝑐ℎ𝑖2 ₍₁₄₎

P > 𝑐ℎ𝑖2

59.0 0.000

References

Abramovitz, M. (1986), “Catching up, Forging Ahead, and Falling Behind”, Journal of

Economic History, No. 46, pp. 385-406

Ahlburg (1991), “Remittances and their Impact: A Study of Tonga and Western Samoa” Pacific Policy Paper No. 7. Canberra: National Centre for Development Studies, Australian National University

Ahlburg, Dennis A., Brown, Richard P.C., (1999) "Remittances in the South Pacific", International Journal of Social Economics, Vol. 26 Issue: 1/2/3, pp.325-344,

Amuedo-Dorantes and Pozo (2004), “Workers' Remittances and the Real Exchange Rate: A Paradox of Gifts”, World Development, vol. 32, issue 8, pp. 1407-1417

Barro, R. J. (1996), “Determinants of Economic Growth: A Cross Country Empirical Study” NBER Working paper No. 5698

36

Bauer, T. K, Zimmermann, K. F. (1999), “Assessment of Possible Migration Pressure and Its

Labor Market Impact Following EU Enlargement to Central and Eastern Europe”, IZA

Research Reports 3, Institute for the Study of Labor (IZA)

Baumol, W. J. (1986), “Productivity Growth, Convergence and Welfare: What the Long-Run Data Show”, American Economic Review, No. 76, pp. 1072-85

Ben-David, D., Kimhi, A. (2000), “Trade and the Rate of Income Convergence”, Journal of

International Trade & Economic Development, Vol. 13, No. 4, pp. 419-441

Bertola, G. (1992), "Models of Economic Integration and Localized Growth," CEPR Discussion Papers No. 651

Boyer, G., Hatton, T. J., O''Rourke, K. H. (1993), "The Impact of Emigration on Real Wages in Ireland 1850-1914" CEPR Discussion Papers, No. 854

Brakman, S., Garretsen, H, van Marrewijk, C., An Introduction to Geographical Economics, Cambridge University Press, Cambridge UK, 2001

Cass, D. (1965), "Optimum Growth in an Aggregative Model of Capital Accumulation", Review

of Economic Studeis, Vol. 32, p.233-40

Chami, R., Fullenkamp, C., Jahjah, S. (2003), “Are Immigrants Remittance Flows A Source of Capital for Development?” International Monetary Fund Working Paper, WP/03/189

European Commission (2007), ‘Remittance flows to and from the EU’, Luxembourg: Office for Official Publications of the European Communities

Faini, R. (2003), “Migration and Convergence in the regions of Europe: A bit of Theory and Some Evidence”, Flowenla Discussion Paper, No. 9

Green, B.F., and Turkey, J.W., (1960), “Complex analyses of variance: General problems”,

Psychometrika, Volume 25, Issue 2, pp 127–152

Hager-Zanken, J. (2008), “Why Do People Migrate? A Review of the Theoretical Literature”, Maastricht Graduate School of Governance, Working Paper, No. 002

37

Joyl, D. (2000), “Some Structural Effects of Migration on Receiving and Sending Countries”,

International Migration, Vol. 38, No. 5, pp. 25-40

Kamuleta, K.M. (2014), “The Impact of Remittances on Developing Countries”, European

Parliament Policy Department Directorate-General for Policy Recommendations.

Katseli, L. T., Lucas, R. E. B., Xenogiani, T (2006), “Effects of Migration on Sending Countries: What Do We Know?”, OECD Development Centre, Working Paper, No. 250 Koopmans, T. C. (1965), "On the Concept of Optimal Economic Growth", Pontificae

Academiae Scientiarum Scripta Varia, Vol. 28, p.225-300

Krugman, P. (1991), “Increasing Returns and Economic Geography”, The Journal of Political

Economy, Vol. 99, No. 3, pp. 483-499

Leon-Ledesema, M., and Piracha, M. (2001), ‘International Migration and the role of remittances in Eastern Europe’, University of Kent Discussion Paper, No. 01/13

Lewis, W. A. (1954), “Economic Development with Unlimited Supplies of Labor”, The

Manchester School of Economic and Social Studies, Vol.22, pp.139-191

Lipton, M. (1980), “Migration from rural areas of poor countries: The impact on rural productivity and income distribution”, World Development, vol. 8, issue 1, pp. 1-24

Lopez, H., Bussolo, M., & Molina, L. (2007), “Remittances and the real exchange rate”, World Bank Policy Research Working Paper 4247

Mansoor, A., Quillin, B. (2006), “Migration and Remittances. Eastern Europe and the Former Soviet Union”, The World Bank Report

Massey, D. S, Arango, J., Hugo, G., Kouaouci, A., Pellegrino, A., Taylor, J. E. (1993), “Theories of International Migration: A Review and Appraisal”, Population and Development

Review, Vol. 19, No. 3, pp.431-466

Mundaca, Gabriela, (2005), "Can Remittances Enhance Economic Growth? The Role of Financial Market Development" Available at SSRN http://ssrn.com/abstract=799484

Myrdal, G. (1957), “Rich Lands and Poor”, New York: Harper and Row.

38

Ramirez, M., Sharma, H., (2009), “Remittances and Growth in Latin America: A Panel Unit Root and Panel Cointegration Analysis”, Estudios Economicos de Desarrollo Internacional, 2009, vol. 9, issue 1

Ratha, D., & Mohapatra, S. (2007). “Increasing the macroeconomic impact of remittances on development”, World Bank.

Ratha, D., & Shaw, W. (2007). “South-South migration and remittances” World Bank Publications, No. 102.

Ritchey, P. N. (1976), “Explanations of Migration”, Annual Review of Sociology, Vol.2, pp.363-404

Rodrik, D. (2007). The real exchange rate and economic growth: theory and evidence”, CID Harvard University Working Paper No. 123

Romer, P. (1986), “Increasing Returns and Long-Run Growth”, Journal of Political Economy, No. 94, No 1002-37

Sachs, J. D., Warner, A. (1995), “Economic Reform and the Process of Global Integration”,

Brookings Papers on Economic Activity, No. 1, pp. 1-118

Sala-i Martin, X. X. (1996), “The Classical Approach to Convergence Analysis”, The Economic

Journal, Vol. 106, No. 437, pp. 1019-1036

Sjaastad, L. A. (1962), “The Costs and Returns of Human Migration”, Journal of Political

Economy, Vol. 70, No. 5, pp. 80-93

Solow, Robert M. (1956), "A Contribution to the Theory of Economic Growth." Quarterly

Journal of Economics, Vol. 70, pp. 65-94

Stark, O., Bloom, E. B. (1985), “The New Economics of Labor Migration”, The American

Economic Review, Vol. 75, No. 2, pp. 173-178

Taylor, E. J. (1999), “The New Economics of Labour Migration and the Role of Remittances in the Migration Process”, International Migration, Vol. 37, pp. 63–88

39

Todaro, M. P. (1969), “A Model Labor Migration and Urban Unemployment in Less-Developed Countries” The American Economic Review, Vol. 59, pp.138-48

Williamson, J.G. (1996), “Globalization, Convergence and History”, The Journal of Economic

History, Vol. 56, No.2, pp.277-302

Websites

Data Sources

 Bilateral remittances, source: worldbank.org

URL: http://www.worldbank.org/en/topic/migrationremittancesdiasporaissues/brief/ migration-remittances-data

Accessed on: 10-03-2017

 GDP per capita, source: data.worldbank.org

URL: http://data.worldbank.org/indicator/NY.GDP.PCAP.CD

Accessed on 17-03-2017

 Inflation (GDP deflator), source: data.worldbank.org

URL: http://data.worldbank.org/indicator/NY.GDP.DEFL.KD.ZG

Accessed on 17-03-2017

 Gross capital formation (% GDP), source: data.worldbank.org URL: http://data.worldbank.org/indicator/NE.GDI.TOTL.ZS

Accessed on: 17-03-2017

 Gross secundary enrolment ratio, source: data.worldbank.org URL: http://data.worldbank.org/indicator/SE.SEC.ENRR

Accessed on: 17-03-2017

 Exports of goods and services (% GDP), source: data.worldbank.org URL: http://data.worldbank.org/indicator/NE.EXP.GNFS.ZS

Accessed on: 17-03-2017

 Imports of goods and services (% of GDP), source: data.worldbank.org URL: http://data.worldbank.org/indicator/NE.IMP.GNFS.ZS

Accessed on: 17-03-2017

 Population, source: data.worldbank.org

40 Accessed on: 17-03-2017

 Contiguity, source: http://www.cepii.fr/francgraph/bdd/distances.htm

Accessed on: 20-04-2017

 Distance, source: http://www.cepii.fr/francgraph/bdd/distances.htm