Akram Thesis

download Akram Thesis

of 50

Transcript of Akram Thesis

  • 8/3/2019 Akram Thesis

    1/50

    THE EFFECT OF MIGRANT REMITTANCES ON ECONOMIC GROWTH

    A Thesissubmitted to the Graduate School of Arts & Sciences

    at Georgetown Universityin partial fulfillment of the requirementsfor the degree of Master of Public Policyin the Georgetown Public Policy Institute

    By

    Theresa A. McCaffrey

    Washington, D.C.April 10, 2007

  • 8/3/2019 Akram Thesis

    2/50

    ii

    THE EFFECT OF MIGRANT REMITTANCES ON ECONOMIC GROWTH

    Theresa A. McCaffrey

    Thesis Advisor: Susan Fleck, Ph.D.

    ABSTRACT

    This study investigates the impact of migrant remittances on economic growth

    in the developing countries that receive the remittances. After reviewing the relevant

    literature on remittances at both the micro and macro levels, it outlines the research

    design for a macro level study on the impact of remittances on growth. The hypothesis

    is that remittances can have a positive impact on growth and the model includes

    various interaction terms in order to determine the macroeconomic situation under

    which this can happen. Ordinary least squares (OLS) and fixed effects regressions are

    performed with cross country macroeconomic data from 152 low and middle income

    countries from 1990 to 2005. The results suggest that remittances have a positive

    impact on growth and that this impact increases at higher levels of remittances relative

    to GDP. Additionally, remittances are found to have a more positive impact in

    countries with certain characteristics: low domestic credit available, low capital

    formation, and low inflation. These results provide further justification for the policy

    recommendations currently advocated by many in the development community and

    suggest some new policies to increase the developmental impact of remittances that

    fall into two broad categories: steering remittances toward the most productive areas

    and creating a macroeconomic climate where remittances can have the greatest impact.

  • 8/3/2019 Akram Thesis

    3/50

    iii

    TABLE OF CONTENTS

    Chapter 1. Introduction .................................................................................... 1

    Chapter 2. Literature Review ........................................................................... 2

    Surveys........................................................................................... 2

    Data on remittances ........................................................................3

    Micro level studies ......................................................................... 5

    Macro level studies......................................................................... 7Policy Implications ....................................................................... 13

    Chapter 3. Hypothesis and Statistical Model .................................................. 14

    Hypothesis.................................................................................... 14

    Model........................................................................................... 14

    Chapter 4. Research Design ........................................................................... 18

    Data Sources................................................................................. 18

    Analytic Methodology .................................................................. 18

    Chapter 5. Findings........................................................................................ 20

    Descriptive Statistics..................................................................... 20Regression Results........................................................................ 24

    Chapter 6. Conclusions and Policy Recommendations ................................... 38

    Appendix: Correlations between variables ......................................................... 41

    References......................................................................................................... 43

  • 8/3/2019 Akram Thesis

    4/50

    1

    Chapter 1. Introduction

    Remittances, the money sent by migrants working abroad to their home

    countries, are increasingly being proposed as a development tool. This is not surprising

    considering that remittances constitute a huge international capital flow. Remittances

    accounted for an estimated US$232 billion across all countries in 2005, $167 billion of

    which was sent to developing countries (World Bank, 2006a). Remittances are second

    only to foreign direct investment as a flow of capital to developing countries, and they

    are approximately double the amount of official foreign aid given. Although they are

    often cited as a good source for development capital, little is known about the actual

    impacts of this currency flow. A recent study by the World Bank found remittances to

    be effective in reducing poverty (Adams & Page, 2005). However, an IMF study found

    them to reduce GDP growth (Chami, Fullenkamp, & Jahjah, 2005). This paper

    investigates the impact of remittances on economic growth in developing countries.

    Chapter 2 reviews the literature, on both the micro and macro levels, of remittances

    and growth. Chapter 3 sets out a hypothesis that remittances can have a positive impact

    on growth and proposes a theoretical model for testing this relationship. Chapter 4

    contains a research plan and analytic methodology for the research. Chapter 5 presents

    the results of the investigation. Chapter 6 concludes and outlines the policy

    implications resulting from the study.

  • 8/3/2019 Akram Thesis

    5/50

    2

    Chapter 2. Literature Review

    International migration is increasingly seen as a development opportunity. The

    worker who migrates generally improves his or her wages and often increases the

    household income through remittances sent to the home country. Little is known about

    why the money is sent and its impact on the household even less is known about the

    impact on the overall economy of the receiving country. However, international

    organizations are beginning to eye this massive capital flow, estimated at around $232

    billion in 2005, as a source of funding for development (IMF, 2005 World Bank,

    2006a). Though the literature on remittances is growing, what is emerging shows great

    variation across countries and studies on the impacts of remittances, both at the micro

    and macro levels.

    Surveys

    For general background on the topic, several authors provide surveys of the

    economic implications of migration and remittance sending. Rapoport and Docquier

    (2005) review the economics of remittances with a focus on the theoretical aspects.

    They include both micro level determinants of remittances and macro level

    implications of remittances. Page and Plaza (2005) provide a comprehensive survey of

    the issues of remittances and development and consider the policy implications for

    both migration and remittances. Brown (2006) looks at recent patterns in remittance

    sending and examines the literature, both theoretical and empirical, on the relationship

  • 8/3/2019 Akram Thesis

    6/50

    3

    between remittances and development. Ellerman (2003) examines the relationship

    between migration and development for both skilled and unskilled workers between a

    variety of contexts, north north, south south, and north south. Ratha (2003)

    summarizes some key issues dealing with remittances, with an eye toward using them

    as a possible development tool. Acosta, Calderon, Fajnzylber, and Lopez (2006)

    review the literature associated with Latin American migration and consider micro and

    macroeconomic studies of the effect of remittances on poverty.

    Data on remittances

    Although accurate figures on international migration are scant, remittance

    flows are one of the better measured indicators. Even so, available data on remittances

    generally capture only money sent through official channels, such as the banking

    system or a wire transfer service, and so undercount the amount of money sent.

    There are three categories in the balance of payment statistics into which the

    money associated with international workers falls (IMF, 1993). Workers

    Remittances appears in the current transfers section of the current account and

    consists of money transferred by migrants who move to another country for a period of

    one year or longer. Compensation of Employees is a component of income in the

    current account and represents the wages, salaries, and benefits earned by nonresident

    workers who migrate for a period of less than a year. And finally, Migrants

    Transfers appears in the capital account section and represents the financial

  • 8/3/2019 Akram Thesis

    7/50

    4

    transactions that accompany the change of residence of the worker from one country to

    another. Although only the first of these three is officially called remittances, studies of

    the economic effects of remittances generally include at least the first two and most

    often all three.

    There are concerns about the reliability of these data on several counts.

    Although the IMF provides a framework, Reinke and Patterson (2005) point out

    differences in accounting methods across countries that introduce variation. Some

    countries include transfers by migrants under Other Current Transfers instead of the

    named category. Other countries count migrants abroad for over a year as residents, so

    all of their income appears as compensation of employees, even though much of it is

    likely spent in the host country. These problems can be seen in the global imbalances

    in credits and debits that exist for workers remittances and compensation of

    employees, amounting to about $12 billion in 2003 (Reinke & Patterson, 2005).

    Although this difference in definitions is a serious problem, difficulties in obtaining the

    necessary data proves to be a still larger one.

    A related concern about data accuracy is the extent to which the official flows

    capture the total flows of remittances sent by international workers. Acosta et al.

    (2006) address this question by comparing balance of payment (BOP) statistics to

    household survey data for 10 Latin American countries representing over two thirds of

    remittances to Latin America for the period of 2000 2004. They find that remittances

  • 8/3/2019 Akram Thesis

    8/50

    5

    in household data are generally less than in BOP data, but by a predictable amount.

    Regressing remittance levels from household surveys on those from BOP data,

    household survey data can explain about 80 percent of the variance in BOP data, and

    BOP remittances statistics tend to be about 70 percent larger than household survey

    statistics. They also point out, however, that household surveys may not be taken from

    a representative sample of the economy and that these surveys often have error

    associated with imperfect recall. However, their analysis does support the acceptability

    of using official remittance data to proxy for total remittances. Ratha (2003) cites

    anecdotal evidence that the proportion of official remittances is increasing as a result of

    international efforts to combat money laundering and the financing of terrorism (p. 43).

    Micro level studies

    Motivation

    There is a wide literature on the distribution and effects of remittances on the

    micro level. Economists disagree on whether the sending of remittances is motivated

    by altruism or whether they fulfill some economic goal, such as insurance or

    investment. Sana (2005), using data from male Mexican migrants in the U.S., finds that

    social status and identity are key motivations for remittances. Roberts and Morris

    (2004) review the reasons for remitting that are generally given, altruism, insurance,

    and investment, and add a fourth reason, access to village based employment

    networks.

  • 8/3/2019 Akram Thesis

    9/50

    6

    Distribution

    A second question on the micro level is who the migrants sending these

    remittances are and what type of households are receiving these remittances. DeSipio

    (2000) looks at how individual migration decisions are made and finds that remittances

    decrease as migrants spend more time abroad. Sana and Massey (2005) examine how

    migrants remit money using data from migrant communities in four Latin American

    countries. They examine the household and family structure in the country of origin

    and find that remittances differ across countries in this respect. Acosta et al. (2006)

    look at Latin America and find different patterns of remittances by income and

    education levels. In some countries, such as Mexico, Guatemala, and El Salvador,

    migrants come from and send remittances back to households in the lower levels of the

    income distribution. In others, such as Haiti, Peru, and Nicaragua, migrants are

    positively selected and come from richer households. There are likewise different

    patterns seen in educational attainment levels of migrant and non migrant households

    across the sample of countries.

    How remittances are used

    A key question for determining the economic impacts of remittances is how the

    money is used, i.e., whether it is spent or invested. Basok (2003) examines the

    investment decisions made by workers in the Canadian Mexican Seasonal Workers

    Programme and finds that migrants from the least endowed communities (with respect

    to land, infrastructure, and market access) are most likely to invest their earnings in

  • 8/3/2019 Akram Thesis

    10/50

    7

    agricultural land. Likewise, from surveys on a household level in Guatemala, Adams

    (2005) finds that increased remittances are associated with lower consumption and

    higher investment. Orozco (2004) describes the emergence of Hometown Associations

    in Mexico, which provide a formal mechanism for migrants to invest in their

    communities. In this way, remittances could be an option for financing local

    development projects on the grassroots level. Lowell and de la Garza (2004) discuss

    this potential throughout Latin America.

    Macro level studies

    Macroeconomic determinants of remittances

    On the macro level, many authors point out that the amount of remittances sent

    is influenced by various macroeconomic factors. El Sakka and McNabb (1999)

    examine the effect that macroeconomic factors have on remittance sending decisions

    using the case of Egypt. The authors find that both exchange rate and interest rate

    differentials are significant determinants of official remittance flows. They also find a

    high income elasticity for imports paid for by remittances. Studies by Gammeltoft

    (2002) and Buch and Kuckulenz (2004) compare remittances to other capital flows and

    find them to be more stable than private and official capital flows. Chami et al. (2005)

    point out that remittances operate in a counter cyclical manner, with migrants remitting

    more money when their relatives at home are worse off. They point out the moral

    hazard inherent in this situation since the sender is not able to verify whether or not

  • 8/3/2019 Akram Thesis

    11/50

    8

    those at home are working diligently. Bugamelli and Patern (2005) also mention the

    countercyclical nature of remittances and point out that this may reduce the probability

    of current account reversals. However, Giuliano and Ruiz Arranz (2005) dispute this

    conventional wisdom and show that the counter or pro cyclical nature of remittances

    varies by country and highlight the role of financial development (discussed below).

    Lueth and Ruiz Arranz (2006) develop a bilateral database of remittance flows based

    on eleven countries in Europe and Asia that track this data. They develop a gravity

    model for determinants of remittances and find that more than 50 percent of the

    variation in remittances can be explained by the GDP differences, distance, and

    common language between the two countries. They find trade linkages and colonial

    ties to be significant as well.

    Poverty and inequality

    Although remittances transfer money to households in poor countries, the

    question of whether or not they reduce poverty on a macro level depends on other

    factors. Adams and Page (2005) examine the relationship among migration,

    remittances, and poverty, using census data from the U.S. and OECD countries in

    Europe to find migration numbers by country of origin. They acknowledge various

    incomplete aspects of their data, including their lack of migration statistics from the

    Arab Gulf countries and their reliance on official remittance data. To account for the

    endogenous nature of migration and remittances, they use an instrumental variables

    approach with three instruments: distance from remittance sending area (U.S., OECD

  • 8/3/2019 Akram Thesis

    12/50

    9

    Europe, or Arab Gulf) to receiving country, education (percent of population over age

    25 that has completed secondary school), and government stability. In their analysis,

    the authors control for income level, since the migrant abroad no longer contributes to

    the family income. They find that both migration and remittances have a statistically

    significant negative impact on poverty, and a larger impact when using the

    instrumental variables approach to control for endogeneity. Using OLS regression,

    they find that, for a 10 percent increase in share of international migrants, poverty

    headcount (measured at $1/person/day) will decrease by 2.1 percent on average. After

    instrumenting for endogeneity, this decrease in poverty rises to 3.5 percent. Gupta,

    Pattillo, and Wagh (2007) similarly find that remittances reduce poverty in their cross

    country analysis, with an even larger impact seen in Sub Saharan Africa,. They deal

    with the endogeneity of remittances through a three stage least squares system of

    regressions.

    However, other studies provide contrasting results. Bracking (2003) looks at

    remittances in Zimbabwe and finds that, although they benefit households who receive

    them, they have an adverse effect on households without migrants, due to both asset

    price inflation and the inflationary effects of parallel currency markets. Acosta et al.

    (2006) examine the literature of micro level studies of the impact of remittances on

    inequality and poverty throughout Latin America and find that a one percentage point

    increase in remittances as a share of GDP lowers the poverty headcount by about 0.4

  • 8/3/2019 Akram Thesis

    13/50

    10

    percent, which they judge to be an insubstantial impact. They also find no substantial

    impact of remittances on inequality in Latin America. In addition to their effect on

    poverty, remittances can also improve other important indicators, especially health and

    education indicators, for receiving households (Kanaiaupuni & Donato 1999).

    Growth

    Chami et al. (2005) published a seminal account of remittances and growth, in

    which they develop a model of remittances, based on the unit of the family and

    motivated by altruism. Their model shows that remittances vary counter cyclically and

    contain an element of moral hazard. They regress growth against initial income,

    investment, and remittances and find that remittances are negatively correlated with

    growth. Although they acknowledge the endogeneity of remittances in their

    countercyclical model, they do not correct for it. Nonetheless, most of the subsequent

    literature on remittances and growth uses their article as a starting point, whether or not

    it supports their conclusions.

    Another study (IMF, 2005) uses an instrumental variables approach to examine

    the relationship of remittances to growth and to control for endogeneity. The

    instruments employed are the distance between the remittance receiving country and

    the remittance sending country with the largest number of migrants and whether these

    two countries share a common language. It finds no significant link between growth

  • 8/3/2019 Akram Thesis

    14/50

    11

    and remittances in the general sample of countries nor in a subset of remittance

    dependent countries in which remittances account for more than one percent of GDP.

    However, there are studies with a more positive view of the contribution of

    remittances on growth. Ziesemer (2006) examines the role of remittances, through

    contributions to physical and human capital in economic growth, by using two

    different open economy models. He uses a general method of moments with

    heteroskedasticity and autocorrelation with pooled data for four groupings of countries

    receiving remittances in 2003. He finds that the countries that benefit the most from

    remittances are those with per capita income below $1,200. For these countries,

    remittances contribute about 2 percent to steady state level of GDP per capita versus

    having no remittances. The effect of remittances on growth in richer countries is found

    to be much smaller. Glytsos (2005) uses a Keynesian econometric model to estimate

    short and long run multipliers of remittances, and then determines the impact of

    remittances on growth in five Mediterranean countries. He maintains that remittances

    can have a positive impact on growth, not only if they are directed towards investment,

    but also through the increased consumption and imports. He considers only the demand

    impact of remittances, but finds great fluctuations across time and countries for the

    effect of remittances on growth.

  • 8/3/2019 Akram Thesis

    15/50

    12

    Growth and financial development

    Another thread of the literature on remittances and growth looks at their impact

    particularly in countries with low levels of financial development, on the theory that

    remittances may provide capital for investment in countries without widespread access

    to credit. Giuliano and Ruiz Arranz (2005) examine the impact of remittances and

    financial development on growth and find that, although in general remittances do not

    have a significant effect on growth, the impact is positive and significant in countries

    with low financial development (measured in four ways loans, deposits, credit, or M2

    as a percentage of GDP). Mundaca (2005) uses data from Central America, Mexico,

    and the Dominican Republic to look at the impact of growth volatility, remittances, and

    financial development on growth, on the theory that economic volatility will influence

    how remittances are spent or invested. She finds that remittances have a positive and

    significant effect on growth in all specifications of her model, although none of the

    specifications include interaction terms between remittances and volatility or

    remittances and financial development, both of which would seem to be of primary

    importance. Looking at the issue from a different angle, Aggarwal, Demirguc Kunt,

    and Martinez Peria (2006) and Gupta et al. (2007) find a positive impact of remittances

    on financial development. So financial development could prove to be a pathway by

    which remittances bring about economic growth.

  • 8/3/2019 Akram Thesis

    16/50

    13

    Policy Implications

    The development literature and discourse is now filled with recommendations

    on what should be done to enhance the development impact of remittances (Kapur,

    2004 Sriskandarajah, 2005 Terry & Wilson, 2005 Brown, 2006). Suggestions cover a

    huge range of policy topics and include relaxing foreign exchange controls, lowering

    the costs of sending remittances, making international labor markets more flexible,

    giving international institutions a role in a remittance sending framework, combating

    brain drain, promoting circular migration, and encouraging collective remittances.

    Orozco and Fedewa (2006) mention a range of the policy options that can be

    pursued in order to enhance the developmental impact of remittances, whether or not

    remittances lead to growth in an absolute sense. The policies advocated include:

    reaching out to diasporas working abroad, reducing the costs of sending remittances,

    providing banking services for unbanked remittance senders and receivers, providing

    investment and microenterprise opportunities for the remittance monies, encouraging

    the use of hometown associations to channel money towards local development, and

    marketing tourism and nostalgic trade to migrants. However, the range of policy

    options will be greater with more information about the behavior of remittances in

    various macroeconomic contexts.

  • 8/3/2019 Akram Thesis

    17/50

    14

    Chapter 3. Hypothesis and Statistical Model

    This study extends the existing literature on remittances and growth, namely

    the work of Chami et al. (2005), by considering the impact of remittances on economic

    growth, while allowing for the impact of remittances to vary along several

    macroeconomic lines. Like the work of Giuliano and Ruiz Arranz (2005), it includes

    financial development, and its interaction with remittances, in the model, along with

    several other macroeconomic country characteristics.

    Hypothesis

    The hypothesis of this study is that there are conditions under which

    remittances will lead to economic growth. Giuliano and Ruiz Arranz (2005) have

    demonstrated low financial development, measured by the depth of the banking

    system, to be one of these cases. This research project investigates the impact of

    remittances on growth at varying levels of gross capital formation and inflation as well.

    Model

    In order to probe the impact of remittances on growth, the following model is

    estimated:

    growthit = b0 +b1Remit +b2Zit +b3 (ZitRemit) +b4Xit + mt + hi + eit

    where growthit is annual growth in per capita GDP in constant 2000 dollars, Remit is

    remittances (workers remittances and compensation of employees received) as a

    percentage of GDP, Zit is a vector of variables whose interactions with remittances are

  • 8/3/2019 Akram Thesis

    18/50

    15

    being considered, and Xit is a vector of control variables whose interactions with

    remittances are not being considered. Of interest are the coefficients b1 and b3, which

    show respectively the impact of remittances on growth and how this effect varies with

    other characteristics.

    The Zit vector includes the following variables:

    financial development (domestic credit provided by banking

    sector/GDP)

    investment/GDP (gross private capital formation/GDP)

    inflation (annual percentage change in consumer prices)

    Allowing all three of these variables to interact with remittances will paint a more

    detailed picture of how remittances interact with other facets of the economy and show

    the conditions under which they are the most economically productive. The Xit vector

    consists only of population growth, expressed as annual percentage change.

    Other variables tested for inclusion represented factors of production (arable

    land and population density), capital sources (foreign direct investment), and degree of

    openness (imports plus exports as a percentage of GDP). These variables were all

    found to be insignificant and some, including arable land and population density,

    varied by country but not over time, thus causing multi collinearity issues and proving

    ineffective for fixed effects regressions. Additional variables of interest included fiscal

  • 8/3/2019 Akram Thesis

    19/50

    16

    balance and measures of human capital, but these were not available annually for a

    sufficient number of countries.

    This study also estimates a model with a quadratic remittance term, as used in

    one of the models of Chami et al. (2005). This model specification is:

    growthit = b0 +b1Remit +b2Remit2 +b3Zit +b4 (ZitRemit) + b5Xit+ mt+ hi + eit

    This model tests whether the impact of remittances on growth varies at different levels

    of remittances.

    Remittances are expressed as percentages of GDP throughout, as in Giuliano

    and Ruiz Arranz (2005), rather than being differenced as in many of the models of

    Chami et al. (2005), because the research question is concerned with the impact of the

    flow of remittances on growth, rather than in the impact of marginal changes to this

    flow on growth.

    As stated in the hypothesis, the expectation is that remittances can positively

    contribute to growth, so one would expect the coefficient on remittances to be positive.

    The quadratic model specification is harder to predict. If remittances provide a

    multiplier effect on growth, then the quadratic term should be significant. Glytsos

    (2005) hints at this possibility in his research on how remittances impact growth in

    Mediterranean countries through the pathways of consumption, investment, imports,

    and income. If the impact of remittances on GDP growth is positive but decreasing

    over time, then one would expect the quadratic term of remittances to have a negative

  • 8/3/2019 Akram Thesis

    20/50

    17

    coefficient, while the linear term will have a positive coefficient. If the impact of

    remittances is positive and cumulative over time, then one would expect the quadratic

    term as well as the linear term to have positive coefficients. If the impact of

    remittances is only felt in the present period, then the quadratic term should be

    insignificant.

    Based on literature in the field of economic growth, strong financial markets

    and investment contribute to positive growth, so one would expect that the coefficient

    of the measures of these two variables financial development and investment will

    be positive. Given that inflation weakens the value of investment, and population

    growth mitigates gains in GDP, the coefficients for these variables are expected to be

    negative. Based on the results of Giuliano and Ruiz Arranz (2005), one would expect

    the coefficient on remittances times credit to be negative, due to a crowding out effect.

    With the exception of credit, the coefficients on the interaction terms of these variables

    with remittances are difficult to predict, not being covered extensively in the literature.

    The coefficient on the interaction of remittances and capital formation is complex

    because growth depends on whether remittances are consumed or invested, whether

    remittances crowd out other capital formation, and what type of returns to scale might

    be present. Assuming, as this model does, that some portion of remittances would be

    invested, the coefficient on remittances and inflation would likely be negative because

    invested remittances are more likely to lead to growth at low levels of inflation.

  • 8/3/2019 Akram Thesis

    21/50

    18

    Chapter 4. Research Design

    Data Sources

    The dataset used in this analysis contains cross country, time series,

    macroeconomic data from low and middle income countries, as classified by the World

    Bank, for the period from 1990 to 2005. These data come from a variety of sources, but

    are compiled in the World Banks World Development Indicators database. This

    period of time is significantly shorter than those used by others previously. Chami et

    al. (2005) look at the period from 1970 1998. Giuliano and Ruiz Arranz (2005)

    examine 1975 2002, using averages over five year periods. This study focuses on more

    recent data because the reliability of the data on remittances has increased in recent

    years as more money is sent through formal systems and as countries have become

    more adept at tracking the flow of remittances.

    Analytic Methodology

    This research uses two main statistical methods. Five models of ordinary least

    squares (OLS) regression are run. The first uses only remittances as an independent

    variable to understand the degree of correlation present. Then linear and quadratic

    models are run, both with and without regional dummy variables.

    Next, fixed effects regressions are used to deal with the issue of autocorrelation

    of the macroeconomic data. Three models are employed, the first using only

    remittances as an independent variable, then the linear and quadratic models.

  • 8/3/2019 Akram Thesis

    22/50

    19

    Finally, the OLS and fixed effects regressions are repeated with the sample

    broken apart by region to see if any region specific patterns emerge.

    These models are compared to selected models employed by Chami et al.

    (2005) and Giuliano and Ruiz Arranz (2005). Chami et al. performed OLS regressions

    of remittances (as a percentage of GDP) against growth. They included lagged GDP

    per capita, investment to GDP ratio (logged), and net private capital formation (as a

    percentage of GDP). In one specification, they included remittances squared as well. In

    subsequent OLS and fixed effects regressions in their paper, they considered the

    impact of the change in remittances on growth, which is beyond the scope of this

    thesis, which considers the impact of the overall level of remittances on growth.

    Giuliano and Ruiz Arranz (2005) regress a similar model with the dependent

    variable being growth in real GDP per capita and the independent variables being

    lagged real GDP per capita (logged), population growth (logged), fiscal balance,

    investment to GDP ratio (logged), years of education (logged), imports plus exports (as

    a percentage of GDP), inflation, and remittances (as a percentage of GDP). They use

    OLS and fixed effects, along with generalized method of moments (GMM), to estimate

    these models. They then estimate another model using OLS (and then GMM), adding

    credit and an interaction term for remittances and credit (and then three other measures

    of financial development, which this thesis does not address).

  • 8/3/2019 Akram Thesis

    23/50

    20

    Chapter 5. Findings

    Descriptive Statistics

    The dataset used contains information for 152 low and middle income countries

    (as classified by the World Bank) from 1990 to 2005, with 1,409 complete

    observations. Tables 1 through 4 present an overview of the variables that are included

    in the models. Tables 1 and 2 present the means of country quintiles ranked by real

    GDP per capita in 1990 and 2000 tables 3 and 4 present the means of country quintiles

    ranked by remittances as a percent of GDP in 1990 and 2000. More observations are

    available for 2000 than for 1990 and for GDP per capita than for remittances, as seen

    in the number of countries in each table. From Tables 3 and 4 one sees that the

    difference between the first and fifth quintiles in remittances as a percentage of GDP is

    much larger in 1990 than in 2000, so variation is decreasing over time. No

    straightforward relationship between remittances and real GDP per capita emerges.

    Table 1: Means of variables by quintiles of real GDP per capita (low to high)1990, 132 countries

    Variable

    1st

    quintile

    2nd

    quintile

    3rd

    quintile

    4th

    quintile

    5th

    quintileReal GDP per capita

    (US $, 2000) 216.51 486.99 1066.57 2038.00 4309.71Remittances

    (as a percentage of GDP) 0.02 0.08 0.09 0.03 0.07Credit

    (as a percentage of GDP) 25.02 60.40 38.62 46.48 46.60Gross capital formation

    (as a percentage of GDP) 16.84 25.49 24.39 26.47 25.48Inflation (annual growth in

    consumer prices) 23.54 430.32 15.28 412.34 326.00Population growth

    (annual percentage) 2.48 2.39 2.05 1.73 1.00

  • 8/3/2019 Akram Thesis

    24/50

    21

    Table 2: Means of variables by quintiles of real GDP per capita (low to high)2000, 141 countries

    Variable1st

    quintile2nd

    quintile3rd

    quintile4th

    quintile5th

    quintile

    Real GDP per capita(US $, 2000) 216.27 491.66 1091.75 2245.10 5227.11

    Remittances(as a percentage of GDP) 0.02 0.06 0.07 0.04 0.01

    Credit(as a percentage of GDP) 32.67 20.46 45.37 34.76 66.39

    Gross capital formation(as a percentage of GDP) 17.88 24.65 21.52 21.22 24.93

    Inflation (annual growth inconsumer prices) 30.98 8.44 26.82 11.36 4.51

    Population growth(annual percentage) 2.72 1.57 0.60 1.36 1.03

    Table 3: Means of variables by quintiles of remittances as a percentage of GDP(low to high), 1990, 82 countries

    Variable

    1stquintile

    2ndquintile

    3rdquintile

    4thquintile

    5thquintile

    Real GDP per capita(US $, 2000) 1682.98 1830.81 1768.69 1072.10 1771.25

    Remittances

    (as a percentage of GDP) 0.00 0.01 0.02 0.04 0.22Credit(as a percentage of GDP) 46.09 44.48 32.40 31.03 44.59

    Gross capital formation(as a percentage of GDP) 19.22 22.84 20.80 22.90 33.50

    Inflation (annual growth inconsumer prices) 197.89 516.26 18.16 14.61 14.60

    Population growth(annual percentage) 2.11 2.55 2.12 2.35 1.62

  • 8/3/2019 Akram Thesis

    25/50

    22

    Table 4: Means of variables by quintiles of remittances as a percentage of GDP(low to high), 2000, 120 countries

    Variable

    1st

    quintile

    2nd

    quintile

    3rd

    quintile

    4th

    quintile

    5th

    quintileReal GDP per capita

    (US $, 2000) 3021.13 2086.25 1946.64 1264.13 1335.45Remittances

    (as a percentage of GDP) 0.00 0.01 0.01 0.04 0.14Credit

    (as a percentage of GDP) 36.00 39.33 44.03 37.54 50.98Gross capital formation

    (as a percentage of GDP) 21.32 22.51 22.72 21.80 23.59Inflation (annual growth in

    consumer prices) 10.21 6.69 12.53 3.83 11.07Population growth

    (annual percentage) 1.61 1.46 1.71 1.67 0.40

    Charts 1 and 2 present scatter plots of remittances as a percentage of GDP

    versus real GDP per capita on log scales for the years 1990 and 2000. This form of

    presentation demonstrates where countries and regions stand relative to one another.

    Some countries change position significantly during this ten year period. China

    experiences a growth in both GDP per capita and remittances (as a percentage of

    GDP), while Egypt and Lebanon have an increase in GDP per capita but a decrease in

    remittances (as a percentage of GDP). Other countries, such as Mexico and El

    Salvador, remain in essentially the same position. However, from this representation

    no straightforward relationship between the two main variables of interest is observed.

  • 8/3/2019 Akram Thesis

    26/50

    23

    Chart 1: Remittances/GDP vs. real GDP per capita, 1990

    Oman

    Seychelles

    St. Kitts and Nevis

    Mexico

    Trinidad and Tobago

    St. Lucia

    South Africa

    Dominica

    Costa Rica

    Brazil

    Lebanon

    Grenada

    Jamaica

    Panama

    Malaysia

    St. Vincent and the Grenadines

    Turkey

    Belize

    Botswana

    Suriname

    Colombia

    Algeria

    Peru

    El Salvador

    Namibia

    Jordan

    Dominican Republic

    Paraguay

    Tunisia

    Thailand

    Guatemala

    Swaziland

    Ecuador

    Tonga

    Egypt, Arab Rep.

    Samoa

    Morocco

    Vanuatu

    Philippines

    Honduras

    Syrian Arab Republic

    Bolivia

    Cape Verde

    Cameroon

    Cote d'Ivoire

    Zimbabwe

    Haiti

    Indonesia

    Sri Lanka

    Papua New Guinea

    Pakistan

    Yemen, Rep.

    Kenya

    Kiribati

    Comoros

    Senegal

    China

    Lesotho

    Nigeria

    Mauritania

    Gambia, The

    India

    Bangladesh

    Benin

    Sudan

    Madagascar

    Togo

    Rwanda

    Lao PDR

    Ghana

    Burkina Faso

    Guinea Bissau

    Mali

    Niger

    Mozambique

    Ethiopia

    0.0001

    0.001

    0.01

    0.1

    1

    $100 $1,000 $10,000

    Real GDP per capita (Log scale)

    Remittances/GDP(Logscale)

    Chart 2: Remittances/GDP vs. real GDP per capita, 2000

    Ethiopia

    Sierra Leone

    Malawi

    Niger

    Guinea Bissau

    Eritrea

    Mali

    Mozambique

    Nepal

    Rwanda

    Burkina Faso

    Madagascar

    Uganda

    Togo

    Ghana

    Tanzania

    Cambodia

    Moldova

    Benin

    Gambia, The

    Lao PDR

    Bangladesh

    Nigeria

    Guinea

    Sudan Comoros

    Mongolia

    Mauritania

    KenyaSenegal

    India

    Haiti

    Lesotho

    Yemen, Rep.

    Pakistan

    Kiribati

    Armenia

    Cote d'Ivoire

    Ukraine

    Papua New Guinea

    Georgia

    Azerbaijan

    Cameroon

    Solomon Islands

    Nicaragua

    Indonesia

    Sri LankaHonduras

    Congo, Rep.

    China

    Guyana

    Philippines

    Bolivia

    Serbia and Montenegro

    Syrian Arab Republic

    Bosnia and Herzegovina

    Cape Verde

    Morocco

    Albania

    Kazakhstan

    Belarus

    Vanuatu

    Ecuador

    Samoa

    Swaziland

    Paraguay

    Egypt, Arab Rep.

    Bulgaria

    West Bank and Gaza

    Iran, Islamic Rep.

    Romania

    Guatemala

    Algeria

    Russian Federation

    Jordan

    Macedonia, FYR

    Namibia

    Colombia

    Thailand

    Tunisia

    FijiPeru

    El Salvador

    Maldives

    Dominican Republic

    St. Vincent and the Grenadines

    Turkey

    Botswana

    South Africa

    Jamaica

    Lithuania

    Latvia

    Belize

    Brazil

    Slovak Republic

    Mauritius

    Dominica

    Gabon

    Malaysia

    Panama

    Estonia

    Grenada

    Costa Rica

    Croatia

    St. Lucia

    Poland

    Hungary

    Venezuela, RB

    Lebanon

    Chile

    Czech Republic

    Mexico

    Trinidad and Tobago

    Libya

    St. Kitts and Nevis

    Argentina

    Oman

    0.0001

    0.001

    0.01

    0.1

    1

    $100 $1,000 $10,000

    Real GDP per capita (Log scale)

    Remittances/GDP(Logscale)

  • 8/3/2019 Akram Thesis

    27/50

    24

    Regression Results

    Ordinary Least Squares

    Table 5 presents the results of OLS regression of growth in real GDP per capita

    against various independent variables using five models that replicate, to some degree,

    those used by Chami et al. (2005) and Giuliano and Ruiz Arranz (2005). The inclusion

    of the quadratic term for remittances in Models III and V mirrors a similar term in one

    of the models of Chami et al. (2005). The first model uses only remittances (as a

    percentage of GDP) as an independent variable and finds them positively and

    significantly correlated with growth. Model II includes other independent variables

    domestic credit provided by the banking sector and gross capital formation (both

    expressed as percentages of GDP) as well as inflation and population growth. In this

    model, remittances are no longer significant whereas the other variables are. Model III

    uses all of the variables in Model II and adds a quadratic term for remittances and

    interaction terms for remittances with credit, gross capital formation, and inflation. In

    this specification, both remittances and remittances squared are significant. Models IV

    and V repeat Models II and III respectively with regional dummies, using the regional

    categories employed by the World Bank: East Asia and Pacific, Europe and Central

    Asia, Latin America, Middle East and North Africa, South Asia, and Sub Saharan

    Africa.

  • 8/3/2019 Akram Thesis

    28/50

  • 8/3/2019 Akram Thesis

    29/50

    26

    Comparing Models II and III (or IV and V), one sees that the addition of

    remittances squared and the various interaction terms with remittances does not have a

    great impact on the sign, magnitude, or significance of most of the coefficients. The

    big exception to this is the coefficient on remittances, which changes both in

    significance and magnitude.

    Comparing Models II and III to Models IV and V with regional dummy

    variables, one also sees that the regional dummies reduce the magnitude of the

    coefficients, but only slightly. The regional dummy coefficients themselves are

    interesting to examine. With Latin America as the base category, all of the other

    regions experience significantly higher growth, ranging between 1 and 2 percentage

    points, with the exception of Sub Saharan Africa. The fact that Latin America joins

    Africa in lower than average growth is probably due to the years selected for the

    sample. The debt crisis and slow growth were characteristic of Latin America in 1990s.

    Models III and V find both remittances and the square of remittances to be

    significant and to positively contribute to growth. This implies that as remittances

    increase, they have an increasingly large positive impact on growth, i.e., that there is

    some sort of multiplier effect at work. This result reinforces the work of Glytsos (2005)

    who hypothesizes and calculates a Keynesian multiplier for the effect of remittances on

    growth, through various pathways consumption, investment, imports, and income.

    The coefficient of credit (that is, domestic credit supplied by the banking

    sector) is negative and significant, meaning that growth is higher in countries that have

  • 8/3/2019 Akram Thesis

    30/50

    27

    less access to credit. This appears to be counterintuitive and is inconsistent with the

    results of Giuliano and Ruiz Arranz (2005) who find that domestic credit available is a

    good proxy for financial development and contributes to growth. If the coefficient for

    credit had switched signs from Model II to Model III, then it could possibly reflect a

    crowding out effect of remittances on credit. However, because the coefficient of credit

    remains near constant in both magnitude and significance, this does not appear to be

    the case. The negative correlation between credit and growth is explored in greater

    detail in the Appendix, where it is broken apart by year and region. It appears to be

    another artifact of the 1990s debt crisis.

    Gross capital formation has the expected positive sign, underlining that

    countries with more capital will experience greater growth. Inflation and population

    growth are both also found to contribute negatively to growth, which is also as

    expected.

    Looking at the interaction terms of remittances and the other variables in

    Models III and V suggests interesting features of how remittances work. The

    coefficient on credit times remittances is negative and significant. This would suggest

    that remittances have a more positive impact on growth in countries with less access to

    credit. Remittances may serve as a substitute for credit when it is not available. This is

    consistent with the findings of Giuliano and Ruiz Arranz (2005) although, as

    mentioned above, the coefficient on credit alone is not. The coefficient on gross capital

    formation times remittances is also negative and significant, suggesting that

  • 8/3/2019 Akram Thesis

    31/50

    28

    remittances have a more positive impact on growth in countries with lower levels of

    capital. This result shows that remittances play a different role in countries with greater

    capital. In countries with higher capital levels, remittances are not used for investment

    in the way that they appear to be in countries with lower capital levels. Finally, the

    interaction term of inflation and remittances is also negative and significant, suggesting

    that remittances have a larger impact on growth in countries with low inflation. This

    could be due to the fact that countries with lower inflation rates enjoy a better

    investment climate and remittances are more likely to be invested in that setting. The

    dollarization or dual currency economy in high inflation economies may also impact

    the effect of remittances on growth, but little research has been done in this area.

    These results differ quite significantly from other studies of remittances and

    development, namely Chami et al. (2005) and Giuliano and Ruiz Arranz (2005). In the

    OLS regressions of Chami et al. (2005), remittances were found to be insignificant,

    both with and without regional dummies. Including the quadratic term of remittances

    as well, they found the squared term to be significant at the 10% level with a negative

    coefficient. The linear term, while insignificant, had a positive coefficient. In

    subsequent OLS regressions, Chami et al. include a differenced term for remittances

    (dlog) and find it to be significant and negative.

    In the OLS model of Giuliano and Ruiz Arranz (2005), they find remittances to

    be insignificant in contributing to growth. In their next set of model, which add various

    measures of financial development (credit being one) and an interaction term for

  • 8/3/2019 Akram Thesis

    32/50

    29

    remittances and financial development, they find remittances to be significant and to

    positively contribute to growth. They also find that the coefficient on credit (along with

    the other three measures of financial development) is significant and positive, which is

    a departure from the negative coefficient on credit here. (For further explanation on

    this issue, see the Appendix.) Finally, they find that the interaction term of remittances

    and credit is significant and negative, indicating that remittances have a greater

    contribution to growth in countries with low levels of financial development. Their

    results are consistent with the findings of this paper.

    The models presented here have relatively low R2values, reaching only about

    0.20. These low levels ofR2 are consistent with other literature in the field, for Chami

    et al. (2005), R2reached 0.27 and for Giuliano and Ruiz Arranz (2005), it reached

    0.36. Higher values would be obtained with the inclusion of a lagged GDP variable,

    but adding lagged GDP creates problems with autocorrelation.

    The problem of autocorrelation bias arises here with macroeconomic data used

    in cross country analysis. Because of the high degree of serial correlation of

    macroeconomic data, the error terms for each country are likely correlated over time.

    To deal with this problem, fixed effects regressions are run below. Other possible

    sources of bias are mentioned below after the fixed effects results are presented.

  • 8/3/2019 Akram Thesis

    33/50

    30

    Fixed effects estimation

    Table 6 presents the results of fixed effects estimation of the same dataset. The

    three models considered correspond to the first three models in Table 5 and the results

    are consistent in sign and significance, though generally the coefficients are smaller.

    Table 6: Remittances and growth: Fixed effects cross sectional estimation (1990 2005, 152 countries)

    Dependent variable:

    Annual growth in realGDP per capita

    I II IIIwith interaction

    terms

    Constant0.714

    (1.26)0.286

    (0.33)1.127

    ( 1.22)

    Remittances(as a percentage of GDP)

    19.550***(5.43)

    5.583(1.33)

    21.867**(2.42)

    Remittances2 49.027***

    (3.30)

    Credit(as a percentage of GDP)

    0.052***( 6.29)

    0.044***( 5.13)

    Credit * remittances 0.173*( 1.91)

    Gross capital formation(as a percentage of GDP)

    0.202***(7.72)

    0.250***(8.80)

    Gross capital formation *remittances

    0.853***( 4.04)

    Inflation (annual growthin consumer prices)

    0.0015***( 3.45)

    0.00088*( 1.65)

    Inflation * remittances 0.191**( 2.27)

    Population growth(annual percentage)

    0.462***( 3.42)

    0.393***( 2.92)

    number of observations 1,621 1,409 1,409

    R2 0.2466 0.3441 0.3604

    Adjusted R2 0.1720 0.2729 0.2886

    Durbin Watson statistic 1.933 1.899 2.099

    Note: Value oftstatistics in parentheses * significant at 10 percent ** significant at 5 percent*** significant at 1 percent.

  • 8/3/2019 Akram Thesis

    34/50

    31

    The one exception to the coefficients being smaller is in Model I, which regresses only

    remittances against growth. The coefficient is significant, positive, and about four

    times larger in the fixed effects regression, compared to the OLS regression.

    As in the OLS regression, remittances are not found to be significant in the

    linear model. The signs, significance, and magnitude of the coefficients suggest the

    same interpretations presented with the OLS regressions above. The fixed effects

    regression, by considering each country separately, does address possible

    autocorrelation problems. The Durbin Watson statistics presented in Table 6 indicate

    that there is no autocorrelation present in these regressions.

    Comparing these models to the fixed effects models of Chami et al. (2005) and

    Giuliano and Ruiz Arranz (2005), several notable differences again arise. Chami et al.

    include only the differenced remittances term in their fixed effects regressions and find

    it to be negative and significant, though smaller in magnitude than in the OLS

    regressions. As in their OLS regression, Giuliano and Ruiz Arranz find the remittance

    term in their fixed effects regression to be positive but insignificant. They do not

    perform the subsequent regression including the interaction of remittances and credit

    with the fixed effect model.

    As in the OLS models above, the fixed effects regressions also exhibit low R2

    values. For these fixed effects regressions, they range up to 0.36. For comparison, the

    highest R2values obtained in the fixed effects regressions by Chami et al. (2005) and

  • 8/3/2019 Akram Thesis

    35/50

    32

    Giuliano and Ruiz Arranz (2005) were 0.39 and 0.68 respectively. Higher R2could be

    obtained with the inclusion of other variables discussed below.

    Several variables were tested in these regressions, both OLS and fixed effects,

    but found to be insignificant and excluded. These include the percentage of arable land

    and population density, as indicators of factors of production imports plus exports

    (both as a percentages of GDP), as used in the models of Giuliano and Ruiz Arranz

    (2005) and foreign direct investment (as a percentage of GDP), another international

    capital flow and possible driver of growth. Other variables of interest, but with limited

    availability for a sufficiently large portion of the sample on an annual basis, included

    fiscal balance (i.e., government surplus) and human capital. Studies in the literature

    (Giuliano and Ruiz Arranz, 2005 Adams and Page, 2005) use these variables as

    indictors of the basic macroeconomic relationship and the effect of education on

    investment. The exclusion of these variables raises the possibility of omitted variable

    bias. In the case of human capital, this likely leads to a positive bias in the coefficient

    on remittances because high levels of human capital are associated both with higher

    levels of growth and with higher levels of migration and remittance sending. The bias

    caused by the omission of fiscal balance is likely to be downward, understating the

    impact of remittances on growth. There is generally a positive correlation between

    fiscal balance and growth, but assuming, as Chami et al. (2005) do, that remittances are

    countercyclical, they will be lower when fiscal balance is higher.

  • 8/3/2019 Akram Thesis

    36/50

    33

    Another potential source of bias in the results is the endogeneity of remittances.

    Remittances are very stable over time, and, as mentioned above, often assumed to be

    counter cyclical. As seen in Table 9 in the Appendix, there is a negative correlation

    between remittances and real GDP per capita. Because of this endogeneity, the positive

    impact of remittances estimated here is likely understated.

    Regional breakdown (using OLS and fixed effects)

    Although the results in Tables 5 and 6 present a positive and convincing view

    of remittances, the case is weaker when the regions are considered separately. Tables 7

    and 8 present the results of OLS and fixed effect regressions respectively where each

    region is considered in its own model. The model estimated corresponds to Model III

    in Tables 5 and 6, containing remittances squared and all interaction terms. In the OLS

    estimations in Table 7, remittances and remittances squared, along with most of the

    interaction terms, lose all of their significance. The variation in growth is explained

    mostly by gross capital formation, along with credit, inflation, and population growth

    to a lesser degree.

    Although both the linear and quadratic terms of remittances are largely

    insignificant in these models, it is interesting to note the variations in sign across the

    regions. In East Asia and the Pacific and the Middle East and North Africa, both terms

    have positive signs, suggesting that the positive impact of remittances on growth

    increases as remittances rise. In Europe and Central Asia, Latin America, and South

  • 8/3/2019 Akram Thesis

    37/50

  • 8/3/2019 Akram Thesis

    38/50

    35

    A note of caution is necessary for these regressions. Breaking the data apart by

    region in this way leads to very small sample sizes, ranging from 79 observations in

    South Asia to 417 in Sub Saharan Africa. Although they form an interesting point of

    comparison, they do not offer as much statistical significance as the worldwide cross

    country regressions presented above.

    In the fixed effects estimations in Table 8, remittances are significant in Europe

    and Central Asia, Latin America, and the Middle East and North Africa. All of these

    regions find a positive impact of remittances on growth. Remittances squared are

    significant in East Asia and the Pacific and Europe and Central Asia, although they are

    negative in the former and positive in the latter.

    The impact of remittances on growth by region is markedly different for these

    fixed effects regressions than for the OLS regressions above. Again, putting aside the

    issue of statistical significance, the region of Europe and Central Asia is the only one

    that has both coefficients positive, leading to the multiplier effect of remittances on

    growth seen in the world wide analysis. Latin America and the Middle East and North

    Africa have positive coefficients on the linear term, but negative coefficients on the

    quadratic term, which shows that remittances initially contribute to growth, but at

    higher levels inhibit growth. East Asia and the Pacific and South Asia, have negative

    coefficients on both terms, meaning that remittances inhibit growth in all cases.

    Finally, Sub Saharan Africa has a negative coefficient on the linear term and a positive

  • 8/3/2019 Akram Thesis

    39/50

  • 8/3/2019 Akram Thesis

    40/50

    37

    They add to growth in an absolute sense with ever increasing returns. Their impact is

    strongest in countries with low levels of domestic credit available, low levels of capital

    formation, and low levels of inflation.

    This study draws from and builds on previous studies, but provides a more

    robust analysis for several reasons. By employing recent data, going back only as far as

    1990, the remittance figures are more accurate. As remittance sending fees have been

    falling over time, more money is being sent through official channels and global

    financial tracking programs are getting better at determining the capital flows around

    the world. Other major advantages of the models presented here are the inclusion of

    numerous interaction terms, which allow a greater understanding of how remittances

    and growth interact in differing economic situations, and the inclusion of the quadratic

    term for remittances, which allows the impact of remittances to vary with their relative

    size in the economy.

  • 8/3/2019 Akram Thesis

    41/50

    38

    Chapter 6. Conclusions and Policy Recommendations

    This study has contributed to the literature on remittances and development

    with a model that accounts for the increasing reliability of remittance data in recent

    years and that allows for variation in the impact of remittances by several

    macroeconomic factors. The results suggest that remittances have a positive impact on

    growth and that this impact becomes even stronger at higher levels of remittances

    relative to GDP. Additionally, remittances are found to have a more positive impact in

    countries with certain characteristics: low domestic credit available, low capital

    formation, and low inflation. Further study both on the micro and macro levels should

    investigate the pathways through which remittances promote economic growth, e.g., by

    funding imports that can promote efficiency and make investment more profitable

    (Glytsos, 2005).

    The conclusions of this research project can help to inform several levels of

    policy that could encourage economic growth in developing countries. Since higher

    levels of remittances are found to lead to accelerated economic growth, the study

    provides justification for the policy recommendations mentioned in the literature

    (Kapur, 2004 Sriskandarajah, 2005 Terry & Wilson, 2005 Brown, 2006 Orozco &

    Fedewa, 2006), which include policies to encourage migration and remittances

    (lowering the costs of sending remittances, making international labor markets more

    flexible, promoting circular migration, and encouraging collective remittances) and

  • 8/3/2019 Akram Thesis

    42/50

    39

    policies to help those remittances become more economically productive (providing

    banking services for unbanked remittance senders and receivers and providing

    investment and microenterprise opportunities for the remittance monies).

    But the policy implications of this study go even further, since information is

    obtained on the behavior of remittances in various macroeconomic contexts.

    Remittances appear to foster the most growth in countries that have lower levels of

    credit available, lower levels of capital formation, and lower levels of inflation. This

    information could be used by international organizations or donor agencies to

    concentrate their remittance and development projects on countries fitting this profile,

    in order to produce the largest positive effect on growth, or to couple their work on

    remittances with efforts to control inflation. On a country level, these results can be

    used to help form a macroeconomic context that will be the most receptive to

    remittances fostering economic growth. Although countries would understandably be

    hesitant to limiting credit or restrict capital formation, the central banks role in

    controlling inflation becomes obviously important, knowing that doing so can help

    remittances to become more economically productive in that environment. In time the

    economic growth stimulated by these remittances may help create a situation where

    families no longer need to be divided by continents or national borders in order to find

    productive work and support one another. This growth stimulating aspect of

    remittances and migration should not be ignored in the current political debates about

  • 8/3/2019 Akram Thesis

    43/50

    40

    immigration in developed countries. The potential for economic growth and poverty

    reduction in developing countries must be considered among the complex set of issues

    involved in the formulation of immigration policy.

  • 8/3/2019 Akram Thesis

    44/50

    41

    Appendix: Correlations between variables

    This appendix examines correlations between the dependent and independent

    variables of interest, concentrating on remittances (the main variable of interest) and

    credit. The focus on credit arises because there is a negative coefficient on credit in

    both the OLS and fixed effects models, contrary to the theory that credit constraints

    can restrict growth and contrary to the findings of Giuliano and Ruiz Arranz (2005).

    Table 9: Correlations between remittances, credit, and other variables

    remittances

    (as a percentage of GDP)

    credit

    (as a percentage of GDP)

    Variable correlationcoefficient

    probability of

    significance

    correlation

    coefficient

    probability of

    significance

    Annual growth in realGDP per capita

    0.06661 0.0073 0.04761 0.0331

    Credit(as a percentage of GDP)

    0.02671 0.2748 1.0000

    Remittances(as a percentage of GDP)

    1.0000 0.02671 0.2748

    Gross capital formation(as a percentage of GDP)

    0.26013

  • 8/3/2019 Akram Thesis

    45/50

  • 8/3/2019 Akram Thesis

    46/50

    43

    References

    Acosta, Pablo, Cesar Calderon, Pablo Fajnzylber, and Humberto Lopez (2006).

    Remittances and Development in Latin America. The World Economy, 29(7):

    957 987.

    Adams, Richard H., Jr. (2005). Remittances, Household Expenditure and Investment in

    Guatemala. World Bank Policy Research Working Paper, No. 3532.

    Adams, R., and J. Page (2005). Do International Migration and Remittances Reduce

    Poverty in Developing Countries? World Development, 33(10): 1645 69.

    Aggarwal, Reena, Asli Demirguc Kunt, and Maria Soledad Martinez Peria (2006). Do

    workers' remittances promote financial development? World Bank Policy

    Research Working Paper, No. 3957.

    Basok, Tanya (2003). Mexican Seasonal Migration to Canada and Development: A

    Community based Comparison. International Migration, 41(2): 3 26.

    Bracking, Sarah (2003). Sending Money Home: Are Remittances Always Beneficial

    To Those Who Stay Behind? Journal of International Development, 15(5):

    633644.

    Brown, Stuart S. (2006). Can Remittances Spur Development? A Critical Survey.

    International Studies Review 8(1): 55 76.

  • 8/3/2019 Akram Thesis

    47/50

    44

    Buch, Claudia M., and Anja Kuckulenz (2004). Worker Remittances and Capital Flows

    to Developing Countries. Centre for European Economic Research Discussion

    Paper No. 04 31.

    Bugamelli, Matteo, and Francesco Patern (2005). Do workers remittances reduce the

    probability of current account reversals? World Bank Policy Research Working

    Paper, No. 3766.

    Chami, Ralph, Connel Fullenkamp, Samir Jahjah (2005). Are Immigrant Remittance

    Flows a Source of Capital for Development? IMF Staff Papers, Vol. 52.

    DeSipio, Louis (2000). Sending Money Home...for Now: Remittances and Immigrant

    Adaptation in the United States. Available online at

    http://www.thedialogue.org/publications/Desipio.pdf.

    Ellerman, David (2003). Policy Research on Migration and Development. World Bank

    Policy Research Working Paper, No. 3117.

    El Sakka, M. I. T., and Robert McNabb (1999). The Macroeconomic Determinants of

    Emigrant Remittances. World Development, 27(8): 1493 1502.

    Gammeltoft, Peter (2002). Remittances and Other Financial Flows to Developing

    Countries. International Migration, 40(5): 181 209.

    Giuliano, Paola, and Marta Ruiz Arranz (2005). Remittances, Financial Development

    and Growth. International Monetary Fund Working Paper 05/234.

    Glytsos, N.P. (2005). The contribution of remittances to growth. Journal of Economic

    Studies, 32(6): 468 496.

  • 8/3/2019 Akram Thesis

    48/50

    45

    Gupta ,Sanjeev, Catherine Pattillo, and Smita Wagh (2007). Impact of Remittances on

    Poverty and Financial Development in Sub Saharan Africa. IMF Working

    Paper 07/38.

    International Monetary Fund (1993). Balance of Payments Manual, Fifth Edition.

    Washington, D.C.

    International Monetary Fund (2005). Two Current Issues Facing Developing

    Countries. In World Economic Outlook 2005: Globalization and External

    Imbalances, pp. 69 107. Washington, D.C.

    Kanaiaupuni, Shawn Malia, and Katharine M. Donato (1999). Migradollars and

    Mortality: The Effects of Migration on Infant Survival in Mexico. Domgraphy,

    36(3): 339 353.

    Kapur, Devesh (2004). Remittances: The New Development Mantra? United Nations

    G 24 Discussion Paper Series, No. 29.

    Lowell, Lindsay, and Rodolfo O. de la Garza (2004). The Developmental Role of

    Remittances in US Latino Communities and in Latin American Countries.

    Institute for the Study of International Migration, University of Texas Austin.

    Lueth, Erik, and Marta Ruiz Arranz (2006). A Gravity Model of Workers

    Remittances. IMF Working Paper 06/290.

    Mundaca, Gabriela (2005). Can Remittances Enhance Economic Growth? The Role of

    Financial Market Development. Downloaded from SSRN:

    http://ssrn.com/abstract=799484.

  • 8/3/2019 Akram Thesis

    49/50

    46

    Orozco, Manuel (2004). Mexican Hometown Associations and Development Finance

    Opportunities. Journal of International Affairs, 57(2): 31 52.

    Orozco, Manuel, and Rachel Fedewa (2006). Leveraging Efforts on Remittances and

    Financial Intermediation. INTAL ITD Working Paper 24. Downloaded from

    www.thedialogue.org/publications/2007/winter/orozco_intermediation.pdf.

    Page, John, and Sonia Plaza (2005). Migration Remittances and Development: A

    Review of Global Evidence. Paper presented at the Plenary Session of the

    African Economic Research Consortium, May 29, 2005.

    Rapoport, Hillel, and Frdric Docquier (2005). The Economics of Migrants

    Remittances. IZA Discussion Paper, No. 1531. Downloaded from

    http://ssrn.com/abstract=690144.

    Ratha, Dilip (2003). Workers Remittances: An Important and Stable Source of

    External Development Finance. In Global Development Finance: Striving for

    Stability in Development Finance, pp. 20 51. Washington, D.C.: World Bank.

    Reinke, Jens, and Neil Patterson (2005). Remittances in the Balance of Payments

    Framework. Presented at International Technical Meeting on Measuring

    Remittances, World Bank, Washington, D.C., January 24 25, 2005. Available

    online at http://www.imf.org/external/np/sta/bop/pdf/rem.pdf.

    Roberts, Kenneth D., and Michael Morris (2004). Fortune, Risk and Remittances: an

    Application of Option Theory to Participation in Village based Migration

    Networks. International Migration Review, 37(4): 1252 1281.

  • 8/3/2019 Akram Thesis

    50/50

    Sana, Mariano (2005). Buying membership in the transnational community: migrant

    remittances, social status, and assimilation. Population Research and Policy

    Review. 24(3): 231 261.

    Sana, Mariano, and Douglas S. Massey (2005). Household Composition, Family

    Migration and Community Context. Migrant Remittances in Four Countries.

    Social Science Quarterly 86(2): 509 528.

    Sriskandarajah, Dhananjayan (2005). Towards fairer flows: policies to optimise the

    impact of migration on economic development. A paper prepared for the Policy

    Analysis and Research Programme of the Global Commission on International

    Migration.

    Terry, Donald F., and Steven R. Wilson, eds. (2005). Beyond Small Change: Making

    Migrant Remittances Count. Washington, D. C.: Inter American Development

    Bank. http://0 site.ebrary.com.library.lausys.georgetown.edu/lib/georgetown/

    Doc?id=10089796

    World Bank (2006a). Global Economic Prospects 2006: Economic Implications of

    Remittances and Migration. Washington, D.C.

    World Bank (2006b). World Development Indicators. Accessed online at http://0

    devdata.worldbank.org.library.lausys.georgetown.edu/dataonline/.

    Ziesemer, Thomas (2006). Worker Remittances and Growth: The Physical and Human

    Capital Channels. UNU MERIT Working paper #2006 020.