The impact of remittances on financial development in selected asian countries

Although the impact of remittances on economic growth and poverty has alwaysbeen a controversial problem for researchers and policy makers as remittanceinflow have been becoming one of t

UNIVERSITY OF ECONOMICS ERASMUS UNVERSITY ROTTERDAM

VIETNAM – THE NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

THE IMPACT OF REMITTANCES ON

FINANCIAL DEVELOPMENT IN

SELECTED ASIAN COUNTRIES

BY

HUYNH THI MY CHI

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, DECEMBER 2016

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES

VIETNAM THE NETHERLANDS

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

THE IMPACT OF REMITTANCES ON

FINANCIAL DEVELOPMENT IN

SELECTED ASIAN COUNTRIES

A thesis submitted in partial fulfilment of the requirements for the degree of

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

“This declaration is to certify that this thesis entitled “The Impact of Remittances

on Financial Development in Selected Asian Countries” which is conducted andsubmitted by me in partial fulfilment of the requirements for the degree of theVietnam – The Netherlands Programme

The thesis constitutes only my original works and due supervision andacknowledgement have been made in the text to all materials used.”

Huynh Thi My Chi

I am also very obliged to Prof Nguyen Trong Hoai, Dr Pham Khanh Nam fortheir valuable comments and suggestions for my Concept Note and ThesisResearch Design My special thanks to Dr Truong Dang Thuy for hisencouragements, advices and enthusiasm to help me finish the thesis I would alsolike to thank all VNP staff for their diligent assistance.

I am thankful to my friends from VNP who supported, encouraged and sharedexperiences for my thesis completion Besides, my sincere thankfulness also goes

to my company’s managers and colleagues who kindly and understandinglyfacilitated my master studying

Finally, I am most grateful to my family for their endless support andencouragements to me all the way through my journeys

FDI : Foreign Direct Investment

FEM : Fixed Effect Model

GDP : Gross Domestic Product

GDPPC : Gross Domestic Product Per Capita

GMM : Generalized Method of Moments

ODA : Official Development Assistant

REM : Random Effect Model

Although the impact of remittances on economic growth and poverty has alwaysbeen a controversial problem for researchers and policy makers as remittanceinflow have been becoming one of the largest external capital sources for manycountries, its direct effect on financial development merely attract more attentionafter the financial crisis of 2007-2008 In an effort to contribute to empiricalstudies on this issue, this study utilizes fixed effect, random effect and systemGeneralized Method of Moments (GMM) to investigate the direct impact ofremittances on two dimensions of financial development comprising thepercentage of domestic credit to private sector by banks and broad money to GDP

in thirty-seven Asian countries during the period 1990-2014 Furthermore, thisstudy also examines whether there are different effects of remittance inflows inhigh, middle and low-income countries in this area The results show that anincrease in remittances seems to have no impact on financial development ingeneral while there are mix results regarding the different income groups ofcountries in Asia In particular, while no evidence on the impact of remittances onboth measurements of financial development as the whole region are obtained inmiddle and low-income countries, there is a significant and positive effect of theseflows on the ratio of domestic credit to private sector by banks to GDP despite aninsignificant influence on broad money to GDP in high income countries

Keywords: Remittances, Financial development, Asia, Income Group

TABLE OF CONTENTS

LIST OF TABLES

LIST OF FIGURES

CHAPTER 1: INTRODUCTION 1

1.1 Problem statement 1

1.2 Research objectives 3

1.3 Scope and data of the study 3

1.4 Structure of the study 5

CHAPTER 2: LITERATURE REVIEWS 6

2.1 Theory of remittances and financial development 6

2.1.1 The concepts and channels of remittances 6

2.1.2 Definitions of financial development 7

2.1.3 The role of remittances in financial development 7

2.2 Empirical studies 10

2.3 Other determinants of financial development 15

CHAPTER 3: MODEL SPECIFICATION AND DATA 18

3.1 Model specification 18

3.2 Data sources 21

3.3 Estimation methods 23

3.3.1 Pooled OLS model 24

3.3.2 Fixed effect model 25

3.3.3 Random effect model 26

3.3.4 Tests for choosing sufficient model 26

3.3.5 The system generalized method of moment estimation 28

CHAPTER 4: THE IMPACT OF REMITTANCES ON FINANCIAL DEVELOPMENT IN ASIA 31

4.1 Overview of remittance inflows and financial development in Asia 31

4.1.1 Overview of remittance inflows to Asia from 1990 to 2014 31

4.1.2 Overview of financial development in Asia from 1990 to 2014 36

4.2 Empirical results 40

4.2.1 Descriptive statistic 40

4.2.2 Empirical results 45

CHAPTER 5: CONCLUSIONS AND POLICY IMLICATIONS 54

5.1 Conclusions 54

5.2 Policy implications 55

5.3 Limitations and further researches 56

REFERENCES

APPENDIX I

APPENDIX II: THE REGRESSION RESULTS

LIST OF TABLES

Table 3.1: The definition and expected sign of variables 22

Table 4.1: The summary statistics of variables 41

Table 4.2: The correlation between variables 44

Table 4.3: The results of tests for choosing models 45

Table 4.4: The results of FEM with robust 47

Table 4.5: The results of system GMM 49

Table 4.6: Summary of the impact of remittances on financial development in Asia and different income groups by FEM and system GMM 50

LIST OF FIGURES

Figure 4.1: Remittances received by areas in the world from 1990 to 2014 (US$

billion) 31

Figure 4.2: Top 10 remittance recipient countries in 2014 (US$ billion) 32

Figure 4.3: Top 10 remittance recipient countries in 2014 (% GDP) 33

Figure 4.4: Remittances to areas in Asia from 1990 to 2014 (US$ billion) 34

Figure 4.5: Remittances received by income groups in the world from 1990 to 2014 (US$ billion) 35

Figure 4.6: Remittances received by income groups in Asia from 1990 to 2014 ($US billion) 36

Figure 4.7: Domestic credit to private sector by banks (% of GDP) in Asia from 1990 to 2014 37

Figure 4.8: Broad money as % of GDP in Asia from 1990 to 2014 38

Figure 4.9: The average ratio of domestic credit to private sector by banks to GDP across income groups in Asia (%) 39

Figure 4.10: The average ratio of broad money as % of GDP across income groups in Asia 39

Figure 4.11: Correlation between domestic credit to private sector by banks (%GDP) and remittance inflows (%GDP) and other controlling variables 42

Figure 4.12: Correlation between broad money (%GDP) and remittance inflows (%GDP) and other controlling variables 43

CHAPTER 1: INTRODUCTION

1.1 Problem statement

Together with the augmentations of migration flows in recent decades,remittances - money sent by migrants to home countries - have grown rapidlyand appeared to surpass main conventional channels like official developmentassistant (ODA), private capital (Dilshad, 2013) According to the data fromWorld Bank, in 2014, personal remittances recorded worldwide are up to one-third of foreign direct investment (FDI) and more than three times as large asthe sum of ODA and official aids, from US$ 67 billion in 1990 to over US$

533 billion Moreover, while foreign direct investment (FDI) fluctuates wildlyduring the period from 2000 to 2014, ODA and other external sources remaineither unchanged or even decrease substantially; remittances keep its routine

to accelerate stably over time In particular, officially recorded remittancesworldwide declined the first time in 2009 since 2000 Nevertheless, thedecrease in remittance inflows by nearly 5 percent in 2009 compared to theremittance amount in 2008 due to global financial recession was trivial, incomparison with a 45 percent decrease in FDI in the same year After that,while remittances have been totally recovered and continued to rise, FDIalready fell twice in 2012 by nearly 10 percent and in 2014 by 20 percent

As the stability and increment of remittance inflows has drawn enormousattention from researchers and policy makers in recent decades, a strong wave

of papers and studies have dug into links between these flows and variousaspects of economic development However, while most investigations haveattempted to figure out effects of remittances on growth, poverty, education,less effort has been spent for studying the linkage between remittance inwardflows and financial development, which has been documented to fostereconomic growth and reduce poverty by literature extensively (Goldsmith,1969; Bencivenga and Smith, 1991; King and Levine, 1993; Beck, Demirguc-

Kunt and Levine, 2004), across recipient countries despite their importantcontribution in total external financial sources.

Moreover, various perspectives relating to this relationship argue differentchannels via which remittance flows may affect financial sectors, andempirical studies examining the impact of remittances on financialdevelopment also demonstrate inconsistent outcomes across countries andareas On the one view, these flows may enhance financial development ifformal financial channels are utilized in order to conduct transactions relating

to remittances, recipients are acknowledged financial services; therebypermitting banks to get acquainted or approach unbanked recipients At thesame time, that other financial products might be need if recipients prefer tostore surplus income fosters the development of banking products andservices (Orozco and Fedewa, 2006; Gupta, Pattillo and Wagh, 2009;Aggarwal et al., 2011; Demigüç-Kunt et al., 2011) Furthermore, playing as acollateral role, remittance flows may enlarge bank willingness to provide loansfor families with stable remittances Even in the case remittances cannotbecome sufficient collateral, overall credit might still surge through recipientcommunities due to the incremental loanable funds as a result of remittancesdeposited to banks (Chami et al., 2009)

On the other view, remittances might impede the development of financialsectors if remittance inflows loosen recipients’ budget constraints and allowthem to lower their demand for outside credit (Martínez, Mascaró andMoizeszowicz, 2008; Brown et al., 2011) Recipients might also considerremittances as additional income and utilize most of them for consumption, ortheir distrust of financial sectors might discourage them to deposit money tobanks, therefore no need for financial products and no increase in bankdeposits (Chami et al., 2009) As a result, whether, how and to what extentremittances impact financial developments in remittance recipient countries

are still complicated issues, hence the need for more studies to investigatethese effects.

In an effort to address the relationship between remittances and thedevelopment of financial sector in Asia, this study utilized a panel data from

1990 to 2014 for thirty-seven Asian countries and different econometricmethods In particular, the linkage of personal remittances and financialdevelopment’s indicators comprising domestic credit to private sectorprovided by banking system and broad money are analyzed with fixed effect,random effect and system Generalized Method of Moments method.Furthermore, based on the level of country’s income which is classified byWorld Bank, this study also examines whether there are different effects ofremittance inflows in high, middle and low-income countries in this area

1.2 Research objectives

This study is conducted in an attempt to:

 Analyze the trend of remittance inflows and the situations of domesticcredit to private sector provided by banks and broad money in Asia region,

 Access the impact of remittances on financial development in Asia ingeneral,

 Evaluate different impacts of remittance flows on financial development indifferent income-groups of countries in Asia

From the findings, this study will propose recommendations in order to fosterthe effects of remittance inflows on financial development in Asian remittancerecipient countries

1.3 Scope and data of the study

This research investigates whether remittances have significant influence onfinancial development for Asian countries due to many reasons First,remittances flowing to Asia have accelerated substantially in the recent decade

and captured nearly 50% total remittance inflows in the world Especiallymany middle and low income countries have taken large parts in bothnumerical value and percentage of Gross domestic product (GDP) in theamount of remittances flowing to this area; nonetheless, the developmentimpact of remittances on financial sectors has not been commensuratelyanalyzed in this area.

Second, researches on studying the impact of remittances on financialdevelopment are solely officially conducted for separate country or a part ofAsia For example, Chowdhury (2011) finds that there is a direct positiverelationship between remittance flows and financial breadth and depth inBangladesh while Noman and Uddin (2012) provides evidences of for anindirect impact of remittances on banking sector and economic growth inselected South Asia countries

Third, as previous studies have chosen sample basing on the division ofdeveloping and developed countries which may affect results when hightolerance among the income of countries may be occurred among those types

of countries Hence, it is necessary for studying the linkage of remittances andfinancial development in this area Specially, investigation on the impacts ofthese flows on different income groups will provide a better view for policymakers

Finally, from the impact of remittances on financial development in Asia,several recommendations may be suggested in the case of Vietnam, one ofAsian middle-income countries whose remittance inflows have been acceleratefrom US$1.34 million in 2000 to over US$13 million and account for nearly7% of GDP in 2015

Hence, there is a need for investigating the impact of remittances both ingeneral and in separate income groups in Asia Data of selected Asiancountries collected for the period from 1990 to 2014 are employed in this

study Except for financial openness indicator which is represented by TheChinn-Ito Index (Chinn and Ito, 2006), the remained data are extracted fromthe database of World Bank.

1.4 Structure of the study

This thesis consists of five chapters Chapter 1 presents the motivation forscrutinizing the nexus between remittances and financial development inAsian countries Chapter 2 provides concepts of remittances and reviews vitalliterature relating to perspectives on the role of remittances to financial sectorsand empirical studies considering the effects of remittances and other variables

on financial development Chapter 3 demonstrates the research methodology

in tandem with data utilized in this study Chapter 4 presents an overview ofremittances and financial development in Asia during the period 1990-2014and analyzes empirical results derived from regression The last chaptersummarizes main findings, proposes policy recommendations; moreover,limitations and future researches are also mentioned in this chapter

CHAPTER 2: LITERATURE REVIEWS

This chapter comprises three parts First part presents concept and channels ofremittances, definitions of financial development and main perspectives on theeffect of remittances on financial development Previous empirical studies onthe relationship between remittance inflows and financial development will bepresent in the second part The last part delivers former papers investigatingthe impact of main macroeconomic and openness factors to financialdevelopment

2.1 Theory of remittances and financial development

2.1.1 The concepts and channels of remittances

The term “Remittances” has been used commonly recently but it is rarelydefined In many previous papers, remittances are defined generally as thecross-border money that migrants send back to their home countriesRemittances may be transferred via either official or unofficial channels.Official transfers refer to transactions utilizing the banking system or moneytransfer organizations Unofficial transfers are sent mainly in cash throughfriends, family members, migrants themselves or via traditional channels,called hawala in some countries, by which money may be deposited with anunlicensed organizations in one country and may be drew at their partners inthe recipient country (Freund and Spatafora, 2008; Nyamongo, 2012; Giulianoand Ruiz-Arranz, 2005; Aggarwal et al., 2011)

As mentioned in literature, informal channels are utilized widely fortransferring remittances in the world due to the accessibility, anonymity, lowcost, reliability First, both the remittance senders and receivers do not need tohave bank account or prepare complex procedures which are required at banks

or official organizations Second, the information relating to the transactionscan be remained unknown with other people Third, the cost of these transfers

remittances are delivered via the networks of family members or friendscreates the reliability especially for people who seldom used official financialorganizations.

Nevertheless, utilizing informal channels induce some consequents First,official organizations may encounter difficulties in collecting data and make amiscalculation the volume of remittances Second, criminal organizations orindividuals can use remittances for money laundering Finally, diminishes thefinancial development impact of remittances as envisaged by the proponents

of financial development (Nyamongo, 2012)

2.1.2 Definitions of financial development

According to the definitions of World Bank, financial sector is thecombination of institutions, instruments, markets, and the legal and regulatoryframework that permit transactions to be made by extending credit Primarily,

to develop the financial sector is to overwhelm “costs” arise in the financialsystem The process of cost reduction in obtaining information, enforcingcontracts, and conducting transactions leads to the appearance of financialcontracts, markets, and intermediaries Diverse kinds and conjunctions ofinformation, enforcement, and transaction costs with various legal, regulatory,and tax systems have inspired distinctive financial contracts, markets, andintermediaries across nations and throughout history

Financial development appears when there are improvements in financialinstruments, markets, and intermediaries although these improvements may ormay not remove the burdens of information, enforcement, and transactionscosts (Levine, 2005)

2.1.3 The role of remittances in financial development

Different perspectives on the impact of remittances on financial development

in recipient countries at the theoretical level have been mentioned in manyprevious studies The first argument is that remittance inflows foster financial

sector on both demand and supply aspects With regard to the demand side,remittances is assumed to enhance financial literacy within remittancerecipient communities when migrants and remittance recipients adopt formaltransfer channels, thereby accelerating requirements for and utilization ofbanking products and services (Orozco and Fedewa, 2006; Gupta, Pattillo andWagh, 2009) For instance, the need for channels through with remittances can

be transferred or received can increase the opportunities of searchinginformation related to products offered by banks; hence, stimulating the needand usage of various financial products and services and increasing thechances for banks to approach un-banked recipients, which permit for theiroutreach expansion It is also argued that even if these flows of remittances arenot transacted through banks, money surpluses created by additional incomefrom remittances might potentially stimulate the acquisition of other bankingservices as people feel necessity for a means to store or invest the excessbalances (Aggarwal et al., 2011; Demigüç-Kunt et al., 2011)

Regardless of supply side, remittances are also presumed to promote thedevelopment of breadth and depth of financial sectors For example, banks areencouraged to provide more products and services or even more branches tosatisfy the growing requirement of remittance recipients Furthermore, banksmight be obligingly permit or increase credit for remittance recipients withstable and considerable remittance receipts as they become more attractive andpotential customers Additionally, augmentation in loanable fund deposited tobanks by remittance recipients along with the costs of transactions related tosending, receiving or storing remittances may result in rise of bank credit tocommunities (Giuliano and Ruiz-Arranz, 2005; Toxopeus and Lensink, 2008;Aggarwal et al., 2011) Consequently, financial systems in recipienteconomies may be further widening and deepening, and it is essential toinvestigate the impact of remittances on financial development at the countrylevel

In the second view, remittances may create deleterious effects on financialdevelopment since individuals with remittance receipts may relax their budgetconstraints and lower credit demand which may be needed from financialorganizations otherwise (Caceres and Saca, 2006; Martínez, Mascaró andMoizeszowicz, 2008; Aggarwal et al., 2011) Moreover, it is worse in case apattern of conspicuous consumption is developed along with the inability topromote a saving habit that can assist investments and economic growth.

In the third view, remittance inflows may have no impact on stimulatingfinancial development in many countries, especially in countries with lessdeveloped financial systems due to many reasons (Brown et al., 2011) Aboveall, migrants prefer to transfer money to home countries through informaltransfer ways rather than formal channels then there might be no need forrecipients to be aware of or utilize bank transfer services As unbanked –receivers do not have acknowledge of banking products and services, namelyfinancial literacy is not increased in the communities, the need for financialsystems does not exist or rise Moreover, remittance recipients might storeexcess money at home or utilize them for other purposes as they might distrustbanks and other financial institutions, or spend most of them for consumption;hence, there is no demand for saving products provided by banks or otherformal financial products These behaviors of remittance recipients mightresult in neither growth nor harmfulness on financial sector performance.Besides debates on the impact of remittances on financial development, theadverse causality is also argued from different level of financial sectorperformance (Brown et al., 2011) Lacking of development financial systemand accessibility of commensurate quantity of banks and other formalfinancial institutions within the remittance recipient countries may lead todistrust of this sector among unbanked individuals or insufficient adaptationfor the demand of financial products and services, hence driving individuals torely more on informal channels for transaction related to remittances In

contrast, higher level of financial development might enhance the trust ofindividuals in financial sectors and stimulate the demand for financial productsand services Furthermore, lower cost of transferring remittances as a result offinancial development might foster utilization of these services of bothremittance senders and receivers.

2.2 Empirical studies

Various methods and datasets have been utilized to access the effects ofsteadily accelerating remittance inflows on financial development in the world

in recent decades; the approaches of the above relationship in those papers can

be categorized in two main strands comprising indirect and direct link For theindirect approach, researchers explore the nexus of remittances and financialdevelopment indirectly by analyzing whether remittances foster economicgrowth at given level of financial development With regard to the directapproach, the impact of remittances on financial development is investigateddirectly to evaluate the widening and deepening effect of remittances onfinancial sector performance

The first indirect group of studies focusing on the relationship of remittancesand growth while considering interactions between remittances and financialdevelopment derives two main contrary aspects relating to whether country’sfinancial development is less developed or relatively developed On the onehand, in a country lacking in financial development, remittances are believed

to play an important role in providing additional sources for capital market,thereby promoting investment and economic growth For instance, by adding arepresentative variable for financial development in tandem with remittances

in growth equations while analyzing for sample of over 100 countries from

1975 – 2002 with GMM approach, Giuliano and Ruiz-Arranz (2009) findstrong evidence that remittance inflows enhance economic growth byproviding an alternative financial source for investment and mitigating credit

constraints for countries with less developed financial systems The findingsremains unchanged for different measures of financial system developmentutilized in the paper after controlling for the problem of endogeneity emerged

by the causal relationship between remittance inflows and financialdevelopment This paper also provide proof that there may exist an investmentchannel via which inflows of remittances can boost economic growth,especially in case the credit offered by financial sector does not fulfill to thepopulation demand The role of remittances in providing external sources forinvestment is weaker in countries with quite developed financial sectors as theavailability of credit lowers the needs of remittances for investment intentions.Ramirez and Sharma (2008) also acquires analogous outcome as Giuliano andRuiz-Arranz (2009), indicating that remittances may boost economic growthmore in country with weak financial development when utilizing unit root test,cointegration test and fully modified ordinary least square method for a dataset

of twenty-three upper and lower income countries in Latin American andCaribbean area for the period of 1990 – 2005 Nevertheless, this study showssupplementary outcome that impact of remittances on economic growth in theupper income group is stronger than in the lower income group Besides, thestudy also presents other channels through which remittances can positivelyinfluence economic growth such as the degree of education and economicliberalization

On the other hand, the impact of remittances on growth may be stronger incountries with more developed financial systems since the immenseaccessibility of financial products and services give assistance in orientatingremittances to more effective usage or investment Mundaca (2009) shows this

by utilizing a cross-countries panel of Latin America and the Caribbeancovering the period of 1970 – 2002 to analyze his theoretical model UtilizingFirst Difference GMM to solve the possible endogeneity, this paper concludesthat remittances are appeared to boost further growth in countries with proper

developed financial system which can direct remittance inflows to invest intechnology or other capital; besides, another important finding of this study isthat when remittance proxy is added in growth equation, the impact of percapita investment on economic growth increases substantially The studies ofOjeda (2003), Terry and Wilson (2005), and World Bank (2006) also supportfor the above findings by presenting the similar results that channelingremittances for better profitable investments through the formal financialsector accelerates the development impacts of remittances.

Bettin and Zazzaro (2011) also obtain similar results when investigating therelationship of remittance inflows and the level of financial development.Utilizing a panel dataset of sixty-six developing countries over the period

1970 – 2005 with system GMM in tandem with adding a new representative ofthe development of financial system to gauge the banking performanceefficiency in remittance-receiving countries, they find that remittance inflowscontribute to the economic growth in countries whose banking systems operatewell

In this first approach studies group, nevertheless, the financial developmentdegree is given and the direct impact of remittance inflows on financialdevelopment in the recipient economy is neglected Hence, researches havebeen investigated influences of remittances on the development of financialsectors more directly as remittances have shown their gradually increasingimportant effects on the development of economic

Taking into account the limitations of indirect approach in the first group ofstudies, the second strand of empirical studies has been investigated moredirect impact of remittances on financial development in the recipient countrywith mix results in the recent years The outcome that appears quite frequently

in recent papers is that remittances promote financial development in manycountries For instance, Gupta et al (2009) shows this when examining the

remittances – financial development linkage and financial development –poverty nexus by employing a panel data of 44 Sub-Saharan Africa countriesover the 1975-2004 periods with three stage least square, fixed effects andrandom effects models of regression along with six years period averages Theresults show that remittances in this area, which are still small in comparisonwith other aid flows, but with a stable and gradually increasing volume duringthe study period, have a significant positive impact on financial developmentand direct effect on poverty alleviation These findings remain unchangedafter dealing with reverse causality which may exists when assessing therelationship between remittances and financial development or poverty in thisarea.

Demirguc-Kunt et al (2010) present evidences of a direct positive impact ofremittance inflows on the banking sector’s breadth and depth by analyzingcounty-level data for households and banking sectors comprising the number

of bank branches, the number of bank accounts and value of bank depositsand credit in Mexico in 2000 In detail, the study shows that the higher theproportion of the population has remittances in a county, the higher thenumber of bank branches, the number of bank accounts and value of bankdeposits exist in that county after taking into consideration of the potentialendogeneity When the portions of households who receive remittances grow

by one percentage, the proportion of households utilizing financial productsand services will be increased by nearly 0.16 to over 0.19 percentages

Aggarwal et al (2011) obtain the analogous outcome when examining thenexus between remittance inflows and financial systems development inrecipient economies by utilizing a large sample of 109 developing countriesover the periods from 1975 to 2007 with various research methods comprisingfixed effects estimations, dynamic system GMM, and instrumental variables(IV) estimations As bank credit and deposit to private sector expressed inproportion of GDP represented for the performance of financial sectors, this

study finds that worker remittances have a direct and positive significantinfluence on the aggregate bank credit and deposit to private sector indeveloping countries after considering for variable omission, reverse causalityand measurement errors effects.

Cooray (2012) also provide evidence for the association of remittances andfinancial sector when examining the annual dataset of 94 non-OECD countriesover the period from 1990 to 2010 with pooled OLS and system GMM Thestudy analyzed the effects of remittance inflows on the size and efficiency offinancial systems, and the results show that remittances increase these twodimensions of financial systems The study, furthermore, examines thecorrelation of remittances on above two measurements of financial sectors viatheir interaction with the government ownership of remittance-receivingcountries’ banks The outcomes indicate that the lower the government ownership of bank a country has, the higher the impact of migrant remittances on theenlargement of the size of financial sector in countries Nevertheless, thehigher increase in the efficiency of financial sectors required the highergovernment ownership of banks

With regard to Asia, Chowdhury (2011) investigates whether remittance flowshave positive contribution to the development of the financial sector inBangladesh with annual dataset from 1971 to 2008 and Cointegration andVector Error Correction Model method The study provides evidence forsignificant and positive effects of remittance inflows on widening anddeepening the financial system in this country The results also indicate thatremittance inflows and indicators of financial development do not havereverse causation when applying test for endogeneity bias

Even though most of studies favor the association between remittances andfinancial development as presented in above studies, other researches still findevidences for different results as remittances appeared to have no or negative

effects on the development of financial sectors Brown et al (2011), forinstance, show evidence that migrants’ remittances have no effect or impedefinancial development when examining this nexus on both macro as well asmicro level On macro level, researchers collected a dataset of 138 countries ofworld for the period of 1970 to 2005 and incorporated fixed effects model ofpanel data and Probit model in exploring financial development andremittances On the other hand, at micro level, data are collected from 3,899households from Azerbaijan and 3,995 households from Kyrgyzstan Thestudy finds that remittances seem to have no effect or even have deleteriouseffect on financial development at macro level At micro level, the study find

an inverse impact of remittances from both perspectives households as well ascommunity in Azerbaijan while there appears a direct impact of remittanceswith both sources households and community in Kyrgyzstan

2.3 Other determinants of financial development

Beside remittances, former papers have indicated main macroeconomic andopenness variables which have effect on financial development comprisingcountry size, GDP per capita, inflation, financial openness and trade openness.With regard to country size, Goldsmith (1969) and Gurley and Shaw (1967)find that the growth of an economy leads to the increase in the needs forfinancial products and services; and financial sector accommodate itself tothose growth requirements Such increment in demand will foster moresophisticated and advanced financial intermediaries in order to meet new needfor their products and services (Yartey, 2008)

Concerning the economic development and quality of country legalinstitutions, it is supposed that higher income may encourage higher rate ofsavings leading to the increment of supplementary financial instruments whichassists channeling these flows into better investment (Kamar and Ben Naceur,2007; Yartey, 2008) On the other hand, it is wisely founded that bad

institutions is one of major hindrances to the development of an economy Acountry with low quality of institutions may induce an impediment of financialdevelopment due to fear increase in the stability of savings and investments inthis sector Furthermore, as founded in the research of La Porta et al (1997),variations in preservations for the rights of investors and creditors or in thelevel of law and regulations enforcement may illuminate why financialsystems in some countries are more advanced than others.

With respect to inflation, that high inflation deteriorates financial propertiesand raise the encouragement to invest in other sectors rather deposit money infinancial sector Consequently, high inflation may be harmful to thedevelopment of financial sectors The negative relationship of inflation andperformance of financial sector have been evidenced by many recent studies,for instance, in the paper of Boyd et al (2001) and Naceur and Ghazouani(2008)

Regarding financial openness, according to the findings of Chinn and Ito(2002) and Baltagi et al (2009), the capital account openness enhancesfinancial sector accountability and transparency, hence augmenting the accessand utilization of financial products and services

Finally, as for trade openness, country with higher integration degree to worldeconomy may have more incentives to push for the development of moreadvanced financial systems in order to assisting trade transactions Findings onthe study of Rajan and Zingales (2003) indicate that in a country with lowlevel of trade openness, industrial incumbents might hinder the growth offinancial sector

In general, despite the positive impact of remittances on financial developmentappeared to overwhelm in previous studies, the inconsistent results are deriveddue to different dataset and approaches involve in this relationship Moreover,despite the positive impact of country size, GDP per capita, financial openness

and trade openness and negative impact of inflations on financial development

in most of former studies, the influences of these determinants on financialdevelopment in Asia and different income groups of countries in this area arestill an open question Consequently, there is a need to examine separately therole of remittances and other factors on the development of financial sector inselected Asian countries in either general or in different income groups

CHAPTER 3: MODEL SPECIFICATION AND DATA

This chapter consists of three main parts The first part presents appliedempirical model based on previous studies The second part introduces anddiscusses empirical models of the study in order to answer research questions.The last part describes indicators for financial development, remittances andother determinants of financial development and data sources

3.1 Model specification

When analyzing the relationship between remittances and financialdevelopment, potential endogeneity due to measurement error, variableomissions and reverse causality should be taken in to account (Aggarwal et al.,2011) Remittances are believed to be recorded with measurement errors asinformal transfer channels through friends, relatives and other Hawala-typeorganizations Causal relationship is the one of the most concerns inexamining the impact of remittances on financial development since higherlevel of financial development might result in higher recorded remittanceseither since financial development stimulates remittance inflows or because ahigher proportion of remittances are recorded as those remittance inflows aretransferred via banks or other formal financial organizations Moreover,financial development might provide lower cost of transferring remittances,hence resulting in such inflows increment Finally, the omissions of essentialfactors which can affect either the growth of remittances or financial sectorperformance also lead to biases when estimating influences of remittances onthe development of financial sectors

In an effort to construct a model to investigate the impact of remittances onfinancial development, this study relies on previous papers to select anappropriate model and various variables Following the papers of Aggarwal et

al (2006) and Gupta et al (2009), dynamic panel data model is applied in

order to address the problems mentioned above The general form of dynamicpanel data model is demonstrated as below:

Yit=α+γYi,t-1+β1Xit+ni+ εt+uit (3.1)where i represents for individual 1,2,…,n and t represents for time 1,2, ,T; Yit

is the indicators of financial development; Yit-1 is the lagged values ofindicators of financial development; Xit is a set of explanatory variablescomprising remittances and other determinants of financial development.; ni isunobserved country specific effect; εt is a time – specific effect; uit is the errorterm

Basing on former studies concerning financial development, besidesremittance indicators is the main explanatory proxy, other factors which areevidenced to have influence on financial development are also comprised inthe model as follows:

FDi,t = β1FDi,t-1 + β2REMITi,t + β3GDPPCi,t + β4LNGDPi,t + β5INFi,t +

β6TRADEOPENNESSi,t + β7FINANCIALOPPENNESSi,t + β0 +

as a share of GDP (CREDIT) and the ratio of broad money to GDP (M2).Domestic credit to private sector by banks measured as a share of GDP isadopted in order to measure the development and the efficiency in allocatingresources of banking sector (Beck et al., 2000; Aggarwal et al., 2006) As thisindicator represents for the amount of money transfer to private sector, thehigher this ratio is, the higher the domestic investment in economy is, therebythe higher the development of financial sector Besides, King and Levie(1993), Beck et al (2000) utilize broad money to GDP to access the size of

financial system Aggarwa et al (2006) made argument that the growth ofprivate financial assets, applying broad money (represented by M2 or M3),shows the liquidity position of the financial system and analyzes the degree ofmonetization and the development of financial market On the other hand,broad money expressed in percentage of GDP (M2/GDP) was considered as

an indicator of financial deepening (Brown et al., 2013)

REMIT is personal remittances expressed in proportion of GDP, whichcomprises of two parts: personal transfers and compensation of employeesaccording to the new definition in the sixth edition of the IMF Balance ofPayments and International Investment Position Manual (BPM6) Inparticular, personal remittances comprise two parts: personal transfers andcompensation of employees Personal transfers is broader than worker’sremittances which is utilized in many previous studies when representing forremittances as personal transfers are independent with the source of earnings

of migrants and the relationship between the recipients Compensation ofemployees comprise three part: wages and salaries in cash, in kind and socialcontributions by the employers involving border, seasonal, and other short-term workers who are employed in an economy where they are not residentand of residents employed by nonresident entities (World Bank, 2016)

In all models, this study controls for macroeconomic and openness variableswhich have been indicated by literature to have effect on financialdevelopment GDPPC representing for economic development and the quality

of institutions is GDP per capita in thousands of constant 2010 US$ LNGDPpresents for country size by taking log of GDP while GDP’s data are inconstant 2010 U.S dollars INF is inflation as measured by the annual growthrate of the GDP implicit deflator shows the rate of price change in theeconomy as a whole TRADEOPENESS is trade openness presented by thesum of exports and imports of goods and services measured as a share of gross

by KAOPEN index calculated by Chinn and Ito (2006) This index isconstructed from four binary dummy variables that codify restrictions oncross-border financial transactions that are reported in the IMF's AnnualReport on Exchange Arrangements and Exchange Restrictions Thesevariables indicate:

 the existence of multiple exchange rates;

 restrictions on current account transactions;

 restrictions on capital account transactions; and

 the requirement of the surrender of export proceeds

Chinn and Ito reverse these binary variables—so that they are equal to unitywhen capital account restrictions are non-existent—and derive the firstprincipal component, which is their summary measure (KAOPEN)

3.2 Data sources

In an effort to identify how remittances impact financial development, thisstudy utilizes eight variables comprising two financial development measures,personal remittances, GDP in thousands of constant 2010 US$, GDP percapita in thousands of constant 2010 US$, trade openness and financialopenness

Most of data except for financial openness were gathered from WorldDevelopment Indicators (WDI) Financial openness is presented by KAOPENindex, which is calculated by Chinn and Ito (2006)

This study originally intended to cover the data for all countries in Asia;nevertheless, many countries do not have commensurate data for all variables,leading to examinations on unbalanced dataset Hence, this study includes data

of thirty-seven Asian countries covering the period 1990-2014, based on theavailability of data

Table 3.1: Definition and expected sign of variables

Relationship

Financial

development

indicators

CREDIT Domestic credit to private

sector by banks expressed as

a percentage of GDP

proportion of GDP

comprising personal transfersand compensation ofemployees expressed as apercentage of GDP

Country size LNGDP Log of GDP in constant 2005

Trade Openness TRADE

-OPENESS

Sum of exports and imports

of goods and servicesexpressed as a share of GDP

Positive

Financial

Openness

FINANCIAL- OPENNESS

KAOPEN Index calculated

by Chinn and Ito

Positive

Moreover, in order to estimate impact of remittance inflows in high, middleand low income groups of countries, countries are defined as high, middle orlow income countries according to the income classifications of World Bank

in each year Next, this study utilizes two dummy variables HIGH andMIDDLE, where:

 HIGH equal 1 if that country is classified as high income country whileHIGH equal 0 if that country is classified as middle or low income country

 MIDDLE equal 1 if that country is classified as middle income countrywhile MIDDLE equal 0 if that country is classified as high or low incomecountry

In total of 704 observations, 121 observations are classified as high incomecountry, 365 observations are classified as middle income country and theremaining are classified as low income country

3.3 Estimation methods

In an effort to address the problems mentioned above, this paper utilizesseveral empirical models to investigate the impact of remittances on financialdevelopment First, Pooled OLS model, fixed effects model (FEM) andrandom effects model (REM) are applied to estimate panel data for a overlook

of remittances and financial development relationship Second, systemGeneral Method of Moments (GMM) is conducted in order to solve theproblems of causal relationship

As the lagged dependent variable added in dynamic panel data model as anexplanatory variable leads to biased in the results of Pooled OLS model, FEMand REM, models which are excluded lagged dependent variable are applied

in order to analyze these above regressions Moreover, in order to estimateimpact of remittance inflows in different income groups of countries, theinteractive terms of remittance and dummy variable of income group are

included in the model Separate equations for two indicators of financialdevelopment are depicted as follows:

CREDITi,t = β1REMITi,t + β2GDPPCi,t + β3LNGDPi,t + β4INFi,t +

β5TRADEOPENNESSi,t + β6FINANCIALOPPENNESSi,t + β0 +

CREDITi,t = β1REMITi,t + β2REMIT_HIGHi,t + β3REMIT_MIDDLEi,t +

β4GDPPCi,t + β5LNGDPi,t + β6INFi,t + β7TRADEOPENNESSi,t +

β8FINANCIALOPPENNESSi,t + β0 + ui,t (2)M2i,t = β1REMITi,t + β2GDPPCi,t + β3LNGDPi,t + β4INFi,t +

β5TRADEOPENNESSi,t + β6FINANCIALOPPENNESSi,t + β0 +

M2i,t = β1REMITi,t + β2REMIT_HIGHi,t + β3REMIT_MIDDLEi,t +

β4GDPPCi,t + β5LNGDPi,t + β6INFi,t + β7TRADEOPENNESSi,t +

β8FINANCIALOPPENNESSi,t + β0 + ui,t (4)

3.3.1 Pooled OLS model

Pooled OLS model is one of the most common employed methods withgeneral form as below:

Yit=β0+β1Xit+uit (3.3)where i represents for individual 1,2,…,n and t represents for time 1,2, ,T �

is dependent variable X is independent variable β1 is coefficient for Xit, β0is

constant intercept, and uit is the error term With the assumption that there are

no differences among individuals in estimated cross-section, the Pooled OLSmethod predicts common constant β for all individuals in cross-section

However, in order to adopt this method, many strict assumptions must besatisfied such as the parameters in model must be linear, and the error term 𝑢𝑖𝑡

is assumed to be independently and identically distributed (iid), 𝑢 ~ iid(0,σ2).Moreover, for each individual, the error term must have no correlation with

explanatory variables for all the time In addition, the variance of uit must behomoscedastic, and there is no autocorrelation between ui and uj (i ≠ j).

In comparison with FEM and REM, the results derived from the regression ofthe Pool OLS the constant intercept and slope coefficients

3.3.2 Fixed effect model

Fixed effects regression is utilized to estimate the effects of the explanatoryvariables on explained variable depending on the changes in the variables overtime with general form as below:

Where i represents for individual 1,2,…,N and t represents for time 1,2, ,T; αi

is the specific intercept for individual i; Y is explained variable; �

is explanatory variable; β is coefficient for Xit,; εit is error component

Fixed effects model expresses that the coefficient for explanatory variables areconstant and the individual specific effects do not change over time but variedamong individuals As FEM addresses individual specific effects throughintercept, the model permits for the existence of correlation between time-invariant individual specific effects and explanatory variables whileidiosyncratic error (εit) is still not allowed to have correlation with explanatoryvariables

As individual characteristics are assumed to have relationship withindependent variables or dependent variable, FEM can remove the influence

of those time-invariant characteristics and evaluate the net impact of thepredictor variables on the predicted variable However, those time-invariantcharacteristics must be peculiar to individual and do not related tocharacteristics of other individuals and the same requirement for thecorrelation among the error terms of individual If those conditions are notsatisfied, FEM will be inappropriate and other model like random effect modelshould be applied

3.3.3 Random effect model

In contrast with FEM, REM allows for variation among individualcharacteristics and assumes that those characteristics are random and do nothave impact on input variables or output variable consisted in the model

The common form of REM is:

Yit=β0+β1Xit+uit where uit=εit+vit (3.5)

Where i represents for individual 1,2,…,N and t represents for time 1,2, ,T; �

is regressand; X is regressor; is coefficient for Xit; β0 , β1are intercepts; uit iserror term which comprises two components, 𝜀𝑖𝑡 referring to error component

of cross-section, and vit, referring to error component for cross-section andtime series combination

In this model, country specific effect is a random variable and it is assumednot to correlate with regressors Compared with fixed effect model, invariantvariables can be added into random effect model However, if invariantvariables which are omitted correlate with regressors, estimators will be biasand inconsistent

3.3.4 Tests for choosing sufficient model

In an attempt to figure out the adequate model among three above model, thisstudy adopt three tests consist of F-test, Breusch–Pagan LM test, HausmanSpecification test The F – test is utilized to choose better model between FEMand Pooled OLS model while Breusch–Pagan LM test is applied to find moreappropriate model between REM and Pooled OLS model The HausmanSpecification test is employed to compare FEM and REM and choose moresufficient model between them

3.3.4.1 The F – test

Basing on goodness of fit, the F-test is applied to choose an appropriatetechnique between fixed effects method and Pooled OLS The regression

�

2

2

model of fix effect method is given as Yit=α+ μi+βXit+εit to test the hypothesis

H0: μ1 = μ2=… = μn-1 = 0 F-test is calculated as follows:

(RSS-URSS)/(N-1)F=

URSS/(NT-N-K) ~ FN-1,N(T-1)-KWhere RSS is the restricted residual sum of squares obtained from the PooledOLS model; URSS is the unrestricted residual sum of squares of FEM

If the hypothesis H0 is rejected, this means that at least one μi is different fromzero Consequently we can conclude that fixed effect model is favored overthe pooled OLS

3.3.4.2 Breusch – Pagan LM Test

Breusch – Pagan LM test is utilized to determine REM or the pooled OLSshould be employed in this study The regression model of random effectmethod is given as Yit=β0+β1Xit+uit where uit=εit+vit to test the hypothesis

H0: var(u)=0 The LM test is computed as:

𝑇

� ∑ �(∑𝑇 𝜀2 )

�� =

[2(𝑇 −

3.3.4.3 The Hausman Specification Test

The Hausman Specification Test is applied to choose the sufficient modelbetween FEM and REM by considering whether the time – invariantindividual effects ui are correlated with the regressors xit If ui are correlatedwith xi,t, FEM is consistent and efficient while REM is not consistent On thecontrary, if ui are uncorrelated with xi,t, REM will be consistent and efficientwhile FEM is inefficient

(𝛽�� − 𝛽𝑅 )′[𝑉𝑎𝑟(𝛽��) − 𝑉𝑎𝑟(𝛽𝑅 )]-1(𝛽�� − 𝛽𝑅 )~χ2

The null hypothesis Ho: ui is not correlated with the xit or the REM should beemployed

H1: ui is correlated with xit or the FEM should be employed

In summary, in order to choose the most sufficient model, this study utilizedseveral tests including F-test, LM test and Hausman test First, the F test isused for determined FEM or the pooled OLS Second, LM test is conducted tochoose between REM and the pooled OLS Finally, the Hausman test isapplied to decide the sufficient model between FEM and REM

3.3.5 The system generalized method of moment estimation

In attempt to investigating the nexus between remittance inflows and financialdevelopment with panel dataset, the positive impact of the past financialdevelopment on the current financial development should be put intoconsideration (Chinn and Ito, 2006, Baltagi et al., 2009) This mean that thelag values of financial development must be comprised in the model or adynamic model should be utilized

Yit=α+γYi,t-1+β1Xit+ni+ εt+uit (3.1)where i represents for individual 1,2,…,n and t represents for time 1,2, ,T;

�𝑖𝑡 is the indicators of financial development; Yit-1 is the lagged valuesof

indicators of financial development; Xit is a set of explanatory variablescomprising remittances and other determinants of financial development.; ni isunobserved country specific effect; εt is a time – specific effectuit; is the errorterm

To resolve the potential endogeneity due to reverse causality betweenremittances and financial development, relevant omitted factors andmeasurement errors, the study utilizes the Arrelano-Bover/Blundell-Bondsystem GMM estimation technique

Three vital assumptions should be taken into account when employing GMMtechnique First, the error terms are not serially correlated Second, theexplanatory variables are weakly exogenous Third, there is no correlationbetween the changes in explanatory variables and the country specific effects,

β0, then the following moment conditions can be applied to obtain unbiasedestimates of the regressors:

E [Yi,t-s(ui,t-ui,t-1)]=0 for s≥2;t=3,…,T

E [Xi,t-s(ui,t-ui,t-1)]=0 for s≥2;t=3,…,T

E [(Yi,t-s-Yi,t-s-1)(εt + ui,t)] =0 for s=1

E [(Xi,t-s-Xi,t-s-1)(εt + ui,t)] =0 for s=1The dynamic regression model utilized for examining the impact of remittanceinflows on financial development can be written as follows:

CREDITi,t = β1CREDITi,t-1 + β2REMITi,t + β3GDPPCi,t + β4LNGDPi,t +

β7FINANCIALOPPENNESSi,t + β0 + ui,t (5)CREDITi,t = β1CREDITi,t-1 + β2REMITi,t + β3REMIT_HIGHi,t +

β4REMIT_MIDDLEi,t + β5GDPPCi,t + β6LNGDPi,t + β7INFi,t +

β8TRADEOPENNESSi,t + β9FINANCIALOPPENNESSi,t + β0 +

M2i,t = β1M2i,t-1 + β2REMITi,t + β3GDPPCi,t + β4LNGDPi,t + β5INFi,t +

β6TRADEOPENNESSi,t + β7FINANCIALOPPENNESSi,t + β0 +

M2i,t = β1M2i,t-1 + β2REMITi,t + β3REMIT_HIGHi,t +

β4REMIT_MIDDLEi,t + β5GDPPCi,t + β6LNGDPi,t + β7INFi,t ++β8TRADEOPENNESSi,t + β9FINANCIALOPPENNESSi,t + β0 +

Despite the potential endogeneity due to measurement error, variableomissions and reverse causality which is mentioned in previous studies(Aggarwal et al., 2011), it is necessary to utilize various methods to analyzethe relationship between remittances and financial development Pooled OLSmethod, FEM and REM are employed together with the system GMM in thisstudy in attempt to figure more accurate on the impact of remittances onfinancial development in selected Asian countries.