Does remittance affect on behaviour of households a case study of 12 rural provinces in vietnam

Remittance considered the stablecapital flow for economic growth to labor exporting countries or countries with hugemigration, helps reducing poverty through increasing income of househo

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

DOES REMITTANCE AFFECT ON BEHAVIOUR OF HOUSEHOLDS? A CASE STUDY OF TWELVE

RURAL PROVINCES IN VIETNAM

BY

NGUYỄN NGÔ PHƯƠNG DIỆP

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, October 2014

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

DOES REMITTANCE AFFECT ON BEHAVIOUR OF HOUSEHOLDS? A CASE STUDY OF TWELVE

RURAL PROVINCES IN VIETNAM

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

OF ARTS IN DEVELOPMENT ECONOMICS

This is to certify that the substance of this thesis entitled “Does remittanceaffected on behaviour of households?: A case study of twelve rural provinces inVietnam”, which is submitted by me in fulfillment of requirements for the degree ofMaster of Art in Development Economic to the Vietnam – The NetherlandsProgramme

I certify that to best of my knowledge, any help received in preparing this thesis,and all sources used, have been acknowledged in this thesis

Nguyễn Ngô Phương Diệp

First of all, I would like to express my greatest appreciation to my supervisor,

Dr Phạm Thị Thu Trà, for her advices, instructions and comments, patient guidance, aswell as encouragement with all her heart during the time of studying and doing thisthesis Without her guidance, I could not implement and complete my thesis

I also would like to offer my special thanks to Dr.Phạm Khánh Nam andLecturer Lê Anh Khang for the valuable suggestions and materials that help me todevelop this thesis

In addition, I would like to special thanks to all the lecturers at the Vietnam –Netherlands Program for their knowledge of all the courses during I studied at theprogram

Besides my mentors, I would like to thank my friends who are always beside meand support for my thesis

Last, but not least, I am very deeply grateful to my family Without their warmencouragement, support, and attention, I would not be possible to complete thisdissertation

Average Treatment Effect’

Average Treatment effect on the TreatedBalance of Payment Manual

Difference in DifferenceForeign Direct InvestmentFixed-Effect Model

International Monetary FundNonProfit Institutions Serving HouseholdOfficial Development Assistance

Ordinary Least SquarePropensity Score MatchingTotal Factor ProductivityUnited States DollarVietnam Household Living Standard SurveyVietnam Access to Resources Household SurveyWorld Bank

During the past decades, remittance becomes a new capital source in developingcountries This capital flow which is increasing every year in the developing countriesderived from origins such as exporting labors, or migration from countries with lowincome to countries with medium or high income Remittance considered the stablecapital flow for economic growth to labor exporting countries or countries with hugemigration, helps reducing poverty through increasing income of households Until

1986, in most cases remittance received by households living in Ho Chi Minh city inViet Nam, after 1986 remittance inflow is expanded to households living in rural areas

of Viet Nam by labor exporting policy Up to now, remittance is increasing more andmore, and it is also one of tools to alleviate poverty, and to enhance living condition ofViet Nam households So, the aim of this study will examine the effect of remittance

on behaviours of households receiving remittance in rural Vietnam Estimation of thiseffect based on impact evaluation method and data set of Vietnam Access to ResourcesHousehold Survey (VARHS) in 2006, 2008, and 2010 This data set was carried out byInstitute of Labor and Social Sciences, a subsidiary of Ministry of Labor-Invalids andSocial Affairs, in 12 rural provinces in North, Middle and South of Viet Nam such as

Ha Tay, Lao Cai, Phu Tho, Lai Chau, Dien Bien, Nghe An, Quang Nam, Khanh Hoa,Dak Lak, Dak Nong, Lam Dong, Long An

TABLE OF CONTENTS

DECLARATION 3

ACKNOWLEDGMENT 4

ABBREVIATIONS 5

ABSTRACT 6

TABLE OF CONTENTS 7

LIST OF TABLES 9

LIST OF FIGURES 10

1 1 P ROBLEM S TATEMENT 11

1 2 R ESEARCH O BJECTIVE 14

1 3 R ESEARCH Q UESTION 14

1 4 S COPE 14

1 5 D ATA 16

1 6 M ETHODOLOGY 16

1 7 T HESIS OUTLINE 16

CHAPTER II - THEORETICAL FRAMEWORK 18

2 1 R ELEVANT CONCEPTS AND BROADER ISSUES OF CONCERN 18

2 1.1 Remittance 18

2.1.2 Literature of migration and remittances 19

2.1.3 Theory of Remittances 23

2.2 LITERATURE REVIEW : R EMITTANCES , I NCOME , S AVINGS , A SSET , I NSURANCE AND B ORROWING 26

2 3 S UMMARY OF LITERATURE REVIEW 33

2 4 C ONCEPTUAL FRAMEWORK 35

CHAPTER III - RESEARCH METHODOLOGY 36

3 1 P ROPENSITY SCORE M ATCHING 36

3.1.1 Assumptions of PSM method 37

3.1.2 Model Specification of Propensity Score Matching 38

3.1.3 Software 44

3 2 D IFFERENCE IN D IFFERENCE (DD) METHOD 44

3.2.1 Theory 44

3.2.2 Implementing DID 45

3.2.3 Model of this study 46

3.3 COMBINING PSM WITH DID METHODS 50

3.4 D ATA 51

CHAPTER IV - EMPIRICAL ANALYSIS 55

4 1 DESCRIPTIVE ANALYSIS OF THE SAMPLE 55

4 2 E MPIRICAL RESULTS 58

4 2.1 Propensity score matching (PSM) model 58

4 2.2 DID model 63

4 2.3 Combining PSM Method and DID Method 68

CHAPTER V - CONCLUSIONS AND RECOMMENDATIONS 72

5 1 C ONCLUSIONS 72

5 2 LIMITATIONS AND SUGGESTIONS FOR FURTHER RESEARCHES . 73

REFERENCES 74

APPENIX A REMITTANCES DEFINED IN BALANCE OF PAYMENT OF INTERNATIONAL MONETARY FUND 80

APPENDIX B QUESTIONNAIRE 84

APPENDIX C PROPENSITY SCORE MATCHING METHOD 91

APPENDIX D DIFFERENCE IN DIFFERENCE METHOD 127

APPENDIX E PSM AND DID METHODS 147

LIST OF TABLES

Table 3.1 Variables of Propensity score Matching 41

Table 3.2 Population in each province 42

Table 3.3 Weight of Population in VARHS 2006, VARHS 2008 and VARHS 2010 42

Table 3.4 Variables of Difference in Difference Method 49

Table 3.5 DID estimation between treatment group and control group. 50

Table 4 1 Number of HHs receiving remittance and receiving no remittance 55

Table 4 2 Summary of Households with and without remittance 56

Table 4.3 Expected sign in PSM Model 57

Table 4.4 Expected sign of variables in DD model 58

Table 4.5 Summary of Average Treatment Effect of on Treated Group 62

Table 4.6 Summary the impacts of remittance by PSM method 63

Table 4.7 Results of DID Method in 2006/2008 66

Table 4.8 Results of DID Method in 2008/2010 67

Table 4.9 Results of PSM - DID Method in 2006/ 2008 70

Table 4.10 Results of PSM – DID Method in 2008/2010 71

LIST OF FIGURES

Figure 1 1 Global Remittances Inflow 11

Figure 1 2 ODA, FDI and Remittance in Viet Nam from 2000 to 2011 13

Figure 2 1 Neo-classical mechanism leading to equilibrium 20

Figure 2 2 Motives for Remittances. 26

Figure 3 1 Example of Common Support 38

Figure 3 2 An example of Difference in Difference 46

DOES REMITTANCE AFFECT ON BEHAVIOUR

440 billion in 2010 as in Figure 1.1 This is an important source of external capitalcontributing to economic growth and development as well as reducing poverty ofhouseholds in developing countries Stemming from remittance, working papersstudied positive or negative effects of remittance on income, asset, borrowing andsaving as mentioned below supplying empirical evidences in developing countries

Figure 1 1 Global Remittances Inflow

Source: Migration Policy Institute, World Bank, 2011

Since the policy of “Doi moi” in 1986 and the even more radical market-orientedreform of 1989, Viet Nam has been considered the country with high economic growth.Average Gross Domestic Production growth in Vietnam is 7.6% during period 2000-

2007 (data of World Bank, 2013) The factors have contributed for economic growthincluding capital, labor, and Total Factor Productivity (TFP), in which capital is countfor more than fifty percent There are two kinds of capital such as foreign investmentand domestic investment Foreign investment includes Foreign Direct Investment(FDI) and Official Development Assistance (ODA), and domestic investment isderived from saving of households, enterprises, and government Household saving isthe major source of domestic investment and apart of household saving stems fromremittance

According to World Bank (2011), Vietnam is ranked in top 20 receiving countries in the world, and the second in East Asia & Pacific afterPhilippines Remittance flowing in Vietnam was increasing from 1.34 billion USD in

remittance-2000 to 8.6 billion USD in 2011 In comparison with Official Development Assistance(ODA) in Vietnam, remittance has been more increasing and higher than ODA since

2002 to now, and remittance was higher than Foreign Direct Investment (FDI) in 2010and 2011 (Figure 1.2)

Remittance is increasing year by year and becomes a stability finance for incomesmoothing, as well as economic growth in developing countries and Viet Namparticularly, during the past decades Remittance inflow in Viet Nam is derived fromtwo sources including Vietnamese immigrant and labor exporting Before 2000, VietNam received remittances from Vietnamese immigration to the United States between

1971 and 2001 (Wade D.P, 2008) Since 1994, remittances have been sent byVietnamese workers in developed countries Labor exporting in Viet Nam is policy ofGovernment; in particular, the Law and regulations encourage labor exporting forpurposes such as creating jobs, stabilizing social security, enhancing standard living,

and poverty reduction in provinces of Viet Nam The beginning is ConferenceResolution of Party Central Committee No.02/NQ-HNTW dated December 24th, 1996mentioned training skilled workers for industrial zones, export processing zones, takinginto account the demand for labor export The next is Law No 72/2006/QH11 datedNovember 29th, 2006 of Government promulgating Law on Vietnamese guest workers;and in particular Decision No.71/2009/QD-TTg date April 29th, 2009 approved theProject of supporting the poor districts for labor exporting to reduce sustainablepoverty in period 2009-2020 The latest is Instruction No.19-CT/TW dated November

5th, 2012 of Viet Nam Party Central Committee on enhancing the Party's leadership totrain vocation for rural workers

Figure 1 2 ODA, FDI and Remittance in Viet Nam from 2000 to 2011

Source: World bank (2012)

Remittance is increased rapidly by Labor exporting policy of Viet NamGovernment, however, the households having members attending labor exportingwhether changing being better than not attending, especially in rural provinces Based

on these literatures and empirical evidences, the aim of this study considers connection

of remittance and its influence on rural receiving-remittance households in Vietnam

1 2 Research Objective

Based on mentioned above concerns, the thesis was carried out for the purposeexploring whether or not there were any effect of remittance on behaviour ofhouseholds receiving-remittance in Vietnam rural, based on a comparison betweenhouseholds with and without remittances By applying research result to reality andbased on collected data, the thesis will then suggest the recommendation relating how

to use remittances to improve living condition in rural households

1 3 Research Question

By limiting the objective of this research to twelve provinces in rural Viet Nam,this thesis tried to outline the income receiving-remittance and receiving-no remittancerural households Specifically, we tried to find the answers to the question “How iseffect of remittance on behaviour of rural households receiving remittance inVietnam?”

1 4 Scope

For the recent decades, remittance inflow becomes diverse in Viet Nam, it is notonly focus on big cities as Ho Chi Minh city and Ha Noi city but also move to otherurban and rural areas, and origins of foreign remittances derived not only fromimmigrating relatives of households but also from exporting labor

The economic globalization and trade liberalization have led to free flows ofcapital, goods, and human resource from countries to countries The movement of

economic value as well as social benefits to developing countries and Vietnam, inparticular.

The national policy of Viet Nam always considers labor export as an importantfactor of external economic to create jobs for laborers and reduce unemployment,especially in rural provinces There are a few studies of relationship betweenremittance and welfare (Nguyen Viet Cuong, 2009), remittance and expenditures(Wade Donald Pfau and Long Thanh Giang, 2010) These studies were major based onVietnam Household Living Standard Survey (VHLSS) datasets of 1990s, having incomparison and extension for VHLSSs in 2002 and 2004 The question given “Is thereany effect of remittances on behaviour of remittance-receiving households in ruralprovinces of Viet Nam?” To my knowledge, the earlier studies of remittances are based

on VHLSS data for all provinces and cities in Vietnam, and papers in particular forimpacts of remittance on rural households in Vietnam based on VARHS are rare.Therefore, the purpose of this study presents changes income, borrowing, saving andasset of receiving-remittance households and receiving-no remittance households inrural provinces of Viet Nam To pursue this purpose, we chose data sets of VietnamAccess to Resources Household Survey (VARHS) in years of 2006, 2008 and 2010.VARHS has designed to provide the information collected by the General StatisticalOffice (GSP) under the umbrella of the Vietnam Household Living Standard Survey(VHLSS) every two year VARHS datasets were surveyed by Institute of Labor andSocial Sciences, subsidiary unit of Ministry of Labor - Invalids and Social Affairs, intwelve rural provinces of Viet Nam including: Ha Tay, Lao Cai, Phu Tho, Lai Chau,Dien Bien, Nghe An, Quang Nam, Khanh Hoa, Dak Lak, Dak Nong, Lam Dong, andLong An

1 5 Data

Due to limitation of our capacity in collecting data, and limited information onthis topic in Vietnamese context, we will use secondary data VARHSs collected byInstitute of Labor and Social Sciences to serve for the research, conclusion, andrecommendation in this thesis

1 6 Methodology

This study was implemented to find the effects of remittances on receivingremittance households by means of comparison with receiving no remittancehouseholds in provinces of North, Centre, and South of Viet Nam Descriptive analysiswas applied to explore receiving-remittance households surveyed in years of 2006,

2008, and 2010 We used models of Propensity Score Matching (PSM), Difference inDifference (DID), and PSM combining with Difference in Difference implemented byOLS regression and Fixed-effect model (FEM) to explain the different effects ofremittances on receiving-remittance and receiving-no remittance households Then, weapplied Stata software to process data collected and to approach the problem based onestimating a multivariate regression model, given the linkages among many ofexplanatory variables to be used

1 7 Thesis outline

The next, Chapter II will present theoretical links and literature review ofprevious studies relating to connection between remittance with saving, asset, loan,insurance, and income In this chapter also will focus on the main concern ofremittances and change in behaviour of receiving-remittance households Chapter IIIwill discuss specification of model used in analyzing the data Chapter IV will presentand interpret the empirical analysis results of the models Finally, Chapter V willpresent the conclusions and recommendations In this chapter, our remarks, findings,

and limitations raised during the survey process will be presented Finally, several recommendations for future research will be offered.

CHAPTER II - THEORETICAL FRAMEWORK

This chapter will present concepts, broad issues, and theory of remittance including:(i) Section 1 presents concepts and broader issues of remittance (ii) Section 2 showsliterature of linking between migration and remittance based on economic theories such

as Neo-classical theory, Keynesian theory, Developmentalist theory, Dual Labormarket theory, The New Economics of Labor migration, and Social Network Theory.(iii) Section 3 lists the empirical studies relating to remittance with income, saving, borrowing, and assets (iv) The last one is for presenting briefly literature review

2 1 Relevant concepts and broader issues of concern

2 1.1 Remittance

According to Maimbo and Ratha (2005) and Jeffrey H Cohen et al (2012),remittances are economic transfers to stem from an immigrant to their sending countryand households Richards H Adams (1991) mentioned international remittances asmoney and goods sent to families by individuals working abroad, and remittancesaffects significantly on the balance of payment in countries receiving remittances.Besides, Adams and Page (2003), Anne Harrison (2004) and Jens Reinke (2007)presented workers’ remittances as certain transactions derived from individuals living

or working abroad or related to their migration In Balance of payments Manual ofInternational Monetary Fund, the Fifth Edition (PBM5) in 1993, the term of remittanceinvolves Wokers’ remittances It is relative to current transfers from migrants includingpeople come and stay in foreign countries for more than a year or less than a year.Enclosed workers’ remittance is two supplementary items cover “compensation ofemployee” and “migrants’ transfers” Compensation of employee covers wages,salaries, and other benefits in terms of cash or in kind paid for people who work incountries being different their native country They considered as seasonal or other

short-term workers meaning less than a year Migrants’ transfers term is flows of goodsand changes in financial items stemmed from the migration (residence for more than ayear) In 2009, International Monetary Fund published Balance of Payments andInternational Investment Positional Manual – Six Edition (BPM6) to modify conceptuallimitations of remittance definitions in BPM5 At first, term of “workers’ remittance”replaced by “Personal transfers” The next, supplementary items related to remittancessuch as “Personal remittances”, ‘Total remittances” and “Total remittances andtransfers to Nonprofit Institutions Serving Households (NPISHs)” In there, Personaltransfers cover all current transfers in cash or in kind made or received by residenthouseholds to or from nonresident households Term of “personal remittances”includes personal transfers, compensation of employees, and capital transfers The next

is “Total remittances” comprising “Personal remittances and Social benefits” The lastterm is Total remittance and transfers to Nonprofit Institutions Serving Households(NPISHs) defined as a sum of “Total remittances, Current transfers to NPISHs, andCapital transfers to NPISHs” - Appendix A

2.1.2 Literature of migration and remittances

One of international capital flows interested in last decades is remittance This isnew transmission channel of capital taking place after migration phenomenon, itcontributes to economic growth in developing countries, and it is a tool fordevelopment in labor exporting countries The concept of remittance has been linked tomigration theory According to Yoko Niimi et al (2006), there are causes and linkagesbetween migration and remittances, migration are often linked to remittances (IMF,

2009), Elke Holst et al (2010) stated that “standard economic theory of remittance is based on the analysis of international labor migration Within this context, labor migration is usually motivated by labor income differences between the migrant’s destination and home country”, as well as Hein de Haas (2007) considered remittances

and migration are rarely separate together Due to the connection between remittancesand migration, there are many theoretical models built to explain this phenomenon.

Neoclassical theory: Macro and Micro Economics

The oldest theory of remittance and migration is Neo-classical theory This theorylooked remittance and migration in two aspects including macro and micro aspects.According to macroeconomics, migration happens in condition having non-equilibrium

in labor market, and the difference of wages Labors in regions with abundant laborand low wage will move to regions with scarce labor and high wage due The result ofthis movement is equilibrium of labor market The following movement of labor iscapital flows including remittances to go in opposite direction from capital-rich tocapital poor countries (Massey et al, 1993 and Roel Jennissen, 2004 & 2007) Themechanism leading to equilibrium Oberg (1995) is presented in Figure 2.1

In microeconomics aspect, Neo-classical theory developed the explanation ofmigration based on individual choice (Todaro, 1969) The high skill labors expect toreceive high real wages when moving to country with higher marginal productivity oflabor They have to consider carefully the benefit and the cost when they decide tomigrate The costs they pay including learning new language, costs of moving, lookingfor their jobs, adapting to new labor environment, and so on Massey et al (1993)presented Borjas’ point of view (1990) regarding to migration behavior that is theyexpected to get net return being higher when moving to alternative countries

Figure 2 1 Neo-classical mechanism leading to equilibrium

Migration

LabourLabour

Source: Oberg, 1995 Capital flows Capital

Developmentalist Theory (1950s and 1960s)

This theory showed that large-scale capital transfer and industrialization to poorcountries would tend these countries’ economic towards rapid economic developmentand modernization Developmentalist “migration optimists” were partial towards thinkthat migration leads to a North-South transfer of investment capital and accelerates theexposure of traditional communities to liberal, rational, and democratic ideas, modernknowledge, and education The policy of migration in developing countries started toencourage emigration, the reason is that return migrants are expected to bringremittances, skills, knowledge and experience to acquire abroad to contribute foreconomic development in the country of origin (Hein de Haas, 2007)

Dual Labor Market Theory

In Dual labor market theory (Piore, 1979 and Massey et al, 1993), labor market isseparated into two sectors such as primary sector and secondary sector The primarysector requires capital-intensive methods, and the secondary requires labor-intensivemethods As the results, there is a bifurcation of the labor force In capital-intensivesector, labors hold stable, skilled jobs with the best equipment and tool In labor-intensive sector, workers get unstable, unskilled jobs, meaning jobs at the bottom of thelabor market They can be laid off at any time with no subsidize from employers

Therefore, international or domestic migration becomes a measure of individual andhis family to cope with unemployment and poverty Along with the flow of labormoving among countries, there is flow of capital sent by share of migration workers’earning back to their families/countries called remittance Consequently, there is aconnecting migration and remittance.

In accordance with Dual labor market theory, we can be aware clearly linkedbetween migration and remittance in global economics context Due to there isshortage of labor in secondary sector in developed countries leading to moving laborfrom developing countries to developed countries Migration of labor motivates capitalflow from developed countries to developing countries that is remittances

The New Economics of Labor Migration (NELM)

The new economics of labor migration (Hein de Haas, 2007 and 2008) has theother vision in comparison with Neo-classical theory NELM replaced the behavior ofindividual migrants (in Neoclassical theory) based on the views of the family or thehousehold as the most appropriate decision-making unit The risk-sharing behavior ofhouseholds are implemented by diversify income resources to minimize income risks(Stark and Levhari, 1982) and maximum income According to Lucas and Stark (1985)and Hein de Haas (2007), households are conscious of migration to respond withincome risks due to migrant remittances serve as income insurance for households oforigin This can explicate theoretically the reason of migration in case having nosubstantial income differentials Beside, migration plays a vital role in providing apotential source of investment capital, which is especially important in the context ofthe imperfect credit (capital) and risk (insurance) markets that prevail in the mostdeveloping countries

Social Network Theory

This theory presented social role of remittance that is social relationship in thelives of the migrants and their households Remittances were mentioned as resources touse for exchanged between members of a social network Social network is defined as

a set of recurrent association between groups of people linked by occupational,familial, cultural, or effective ties Resources are seemed as displaced to the socialnetwork when a migrant sends remittances based on the following:

(i) Based on traditional or sentimental value, the migrant thought that theyaccumulate social obligation from the people to whom they remit for examplechildcare, sending goods

(ii) The migrant remitting, maybe, conforms to moral values learn as being a member

of the group

(iii) Remittances increase their social visibility in the sending and receiving countries,

in addition to avoiding the sanction by the social group if they do not remit

2.1.3 Theory of Remittances

Pure Altruism

Affection of family is always intangible ties to members in a family Regarding tothis relation, Lucas and Stark (1985 and 1988) explicated migrant’s motive forremitting which is altruism The migrants themselves feel to have their obligations totransfer remittances because of a simple reason, the migrants care about the well-being

of their families, and the other reason that is their families’ support during their formaleducation before and after leaving home Regarding to altruism model, Rapoport andDocquier (2005) also considered that the family sentiment is motive to impulsemigrants to remit their house Migrants’ income increases, then remittances tend toincrease and vice versa In addition, Glytsos (2001) stated that migrants’ remittingpurpose depends on length of time they reside abroad If the migrant is residentpermanently abroad, they send remittance for altruistic motive, and if migrants live

abroad in short time (temporary), they tend to remit for purpose of investment andfuture consumption Moreover, Rapoport and Docquier (2005) reinforced link betweenaltruism and remittances based on demonstration of Funkhouser (1995) for a model ofremittance behaviour as follows.

“(i) Emigrants with higher earnings potential remit more;

(ii) Low-income household receiving more;

(iii) Remittances should increase with both the degree of proximity between the migrant, and the remaining household members and the migrant’s intentions to return; and

(iv) Remittances from a given migrant should decrease with the number of other emigrants from the same household;

(v) The time profile of remittances should depend on the comparison between the migrants’ time-discount factor and their earnings profile abroad”.

Pure Self-Interest

The contrary of pure altruism is impure altruism (pure self-interet) Lucas andStark (1985) gave three reasons to explicate motive of remitting without altruism fromthe migrants to their families

Firstly, a migrant would send money home to increase their visibility henceeligible for inheritance, esteem or other resources in the community of origin

Secondly, migrants send remittances in order to reimburse the household for pastexpenditures such as schooling or the cost directly related to migration

Lastly, in the case of pure self-interest, migrants have intended to return theirorigin countries, this promotes to remit in order to purchase durable goods and invest inhousing, land, livestock or business at home

Tempered Altruism or Enlightened Self-Interest

Lucas and Stark (1985)’s theory of “tempered altruism” or “enlightened interest” was developed by El Mouhoub Mouhoud et al (2008), and Jessica Hagen-Zanker et al (2007) stem from three motivations of remitting by means of familyarrangements such as exchange, insurance and investment

self Exchange motive derived from migrants’ expectation for members in their family toreceive the better quality of welfare services (health, education, and so on) from their remittances

- Insurance motives are considered as intra-family arrangements between the migrantand their families to diversify the risk of income volatility in rural regions According

to Rapoport and Docquier (2005), families’ strategy is to dispose some members to goabroad or urban to insure income of family in case of poor harvest, and all cost relating

to migrants paid by their families These are motives the migrants to remit their family

- Investment motive: the migrants transfer their saving to invest in their home countrieswhen they calculate potential return in their home countries being higher than in thehost countries So, the remitting decision of migrants depends on the difference ofinterest rate between home and host countries

In general, relationship between migration and remittances is explicited byaspects including pure altruism, pure self-interest and tempered altruism Motive ofmigration is to remit to improve living conditions of members in a family, and threemotives for remittances summarized by Catalina Amuedo-Dorantes and Susan Pozo(2006) in Figure 2.2

Figure 2 2 Motives for Remittances.

Migrants’ Remittances

insurance (response of (nope response of risks (response of risks in

Source: Catalina Amuedo-Dorantes and Susan Pozo (2006)

2 2 Literature review: Remittances, Income, Savings, Asset, Insurance and

Borrowing

Theories mentioned above give reasons to explain why remittances sent frommigrants to their family in the home country At microeconomic level, the impacts anddeterminant of remittances on receiving-remittance countries depend on the aim ofusing remittances of households such as consumption, health, education, etc., and thenext question is how to use remittances in receiving-remittance households There aremany debates about effects of remittance on receiving-remittances households.According to Richard H Adams et al (2008), Adriana Castaldo and Barry Reilly(2007), remittances are considered based on three main observations At first,remittance is widely viewed as from any other source of income of households; it canincrease income of households or lower poverty incidence The next is a pessimisticoutlook that remittances can change behaviours of receiving-remittance households Itmeans that remittances received may reduce effects of increasing income in receiving-remittance households from other sources Based on this viewpoint, Chami et al

(2003), Yéro Baldé (2010) gives judgments including (i) remittance is major spent for

“status-oriented” consumption; (ii) a small part of remittance is for savings orrepayment of loans; (iii) and another way is for investment in land, housing or non-productive assets The last is argument of remittance regarding to other forms ofhouseholds’ income increasing investments in human and physical capital at themargin Khawaja Mamun and Hiranya K.Nath (2010) shown percentage distribution ofremittances by expenditure categories including Savings accounting for 3-7 percent ofremittances received by households; repayment of loans accounting for 10-19 percent

of the value of remittances spent; Health care accounting for 0-4 percent of remittancesspent; and food & clothing accounting for 20-36 percent of remittances spent Thearguing between remittance and insurance derives from volatility income and or wagerisks, this may be evidence of purchasing insurance for migrants as well as their family(Catia Batista & Janis Umblijs, 2014; and Dean Yang & HwaJung Choi, 2005)

Beside observations above, empirical studies as follows show the impacts ofremittances on income, assets, borrowing, insurance, and savings in householdreceiving remittances

 Relationship between Remittance and Income

There are studies the relation between remittance and income of households indeveloping countries such as Egypt, Small Island Developing States (SIDS), Vietnam,and the selected Asian and the Pacific countries as presented below

At microeconomic level - Richard H Adams Jr and John Page (2005) had an

investigation of relationship between remittances and alleviating poverty in 71developing countries with “low-income and middle-income” These countries include

18 countries in Sub-Saharan Africa and 53 remained countries in all regions of thedeveloping world The data set has 184 observations; each observation containsinformation of income, poverty, and inequality based on in time The methods appliedare instrumented and non-instrumented OLS estimates with the basic growth-poverty

model of Ravallion (1997), and Ravallion & Chen (1997) The finding of this paperimplied that an increase in per capita official remittances lead to decrease of peopleliving with less than $1.00 per person per day For instrumented OLS estimate, a 10percent increase in per capita official remittances correspond with 3.5 percent decrease

of people living in less than $1.00 per person per day For non-instrumented OLSestimate, a 10 percent increase in per capita official remittances was equivalent to 1.8percent decrease of people living in less than $1.00 per person per day

Empirical evidence in Juthathip Jongwanich’s working paper (2007) investigatedthe impact of workers’ remittances on economic growth as well as poverty reduction inthe 17 Asian and the Pacific countries The estimate technique is Generalized Method

of Moments (GMM) using panel data over the 1993 through 2003 period derived fromWorld Bank (2006) and IMF’s Balance of Payments Yearbook CD-ROM (2006) Theresults proved that remittances have a direct impact on increasing income ofhouseholds receiving remittances

Papers of Cuong Nguyen Viet (2008) and Wade Donald Pfau & Long Thanh Giang(2010) relied on datasets from 1992/93 to 2006 of Viet Nam Household LivingStandard Surveys (VHLSS) Methods including the impact evaluation method anddescriptive statistic were applied to study the relationship of international remittanceand per capita income in Vietnam The result of studies concluded that internationalremittance is an important factor in increasing income of receiving household

Moreover, the direct relation between remittances and income smoothing isconcreted by study of Catalina Amuedo-Dorantes and Susan Pozo, (2011) in case ofMexico The OLS regression model described linkage between dependent variable oftotal household income and independent variables of household remittances, varioushousehold-level characteristics in time-varying economic conditions at the state level.Data set apply household level data during the period 200-2008 of the MexicanEncuesta Nacional de Ingresos y Gastos de los Hogares (ENIGH) This paper showed

that remittance contributes to the stable of household income, in particular, remittancesmoothes income flows.

In Nigeria, the paper of Mmaduabuchukwu Mkpado (2012) based on secondarydata of Community Based Monitoring System (CBMS) Nigeria, library materials fromUniversity of Nigeria Nsukka (UNN) and Enugu State Agricultural DevelopmentProject and Central Bank of Nigeria Data set includes 5,000 rural households in EdemNsukka, Enugu State and Nigeria, and applied descriptive statistics method.Implication of this study was that reducing households’ income has relationship withvalue of decline in remittance in rural Nigeria

At macroeconomic level - Study of Bichaka Fayissa and Christian Nsiah (2008)

from 1980 to 2004 in 37 Africa countries examined relationship between remittancesand income The panel data was get from the World Bank Development Indicators(WDI, 2006) CD and the Freedom House’s Freedom in the World Country Ratings.They applied both fixed-effects model and random-effects model, including variablessuch as GDP per capita; Per capita workers’ remittances and compensation ofemployees, received; Gross enrollment rate; Gross fixed capital formation; Officialdevelopment assistance and official aid; Net barter terms of trade; Foreign directinvestment; Political rights index; and Initial per capita income The results showedthat a ten percent increase in remittances lead to a 0.3 percent increase in the GDP percapita income

Another paper of Bichaka Fayissa and Christian Nsiah (2010) also studiedrelationship between remittances and income in 18 Latin American Countries from

1980 to 2005 The panel data were from the World Bank development Indicators(WDI, 2006) CD, the fraser Institute’s Economic Freedom of the World Index, and theUNCTAD Handbook of Statistics Based on fixed-effects model and random-effectsmodel, the results found that a ten percent increase in remittances lead to a 0.15 percentincrease in the GDP per capita income

Linkage between remittance inflow and income was proved by study of SimonFeeny et al (2012) in Small Island Developing States (SIDS) In this study, datacollected from World Bank during the period 1971 – 2010, deriving from 136developing countries with 694 observations, including 25 SIDS with 133 observations.Regression model applied is the Ordinary Least Squares method The result of thisinvestigation indicated that per capita income growth would be lower in case withoutremittance inflows.

In conclusion, remittances is considered the stable financial source providingincome to the household poor, it also means that remittances help reducing povertythroughout increasing income and enhancing living conditions of households indeveloping countries

 Remittance and saving, asset, insurance & borrowing

Households receiving remittance have accumulated income sources leading toincreasing income, saving or in other manners such as furnishing assets in their houses,accumulating for saving, and alleviating borrowing The following studies showed thelinks between remittances with income, saving, borrowing, and assets

In 1995, John Foster published his research of relationship between remittancesand saving in Small Pacific Island States (Tongan and Western Samoa) The source oftime series data in this research was get from World Bank Tables during period 1985-

1995, and ordinary least squares (OLS) estimation of regression expression applied forcalculating The evidences found that migrants’ remittance is the key factor concerning

to saving, it also means that receiving remittance households’ income increases leading

to appearance of saving motive

In Nigeria, it was found that remittance help to increasing saving and asset inNigeria households Una Okonkwo Osili (2006) supplied this evidence based on self-reported data on the origin family in Nigeria Beside, method applied in this study

relies on matched sample of both international migrants and their origin households toinvestigate remittance flows The U.S.–Nigeria Migration Study is the first dataset toprovide a comprehensive picture of international migrants and their origin families.This unique dataset contains information on migrants' saving in the origin and hostcountry, as well as migrants' transfers to family members in the country of origin.Lisa A (2014) also found out a positive effect of remittances on consumer assetaccumulation in Ethiopia This study applied data from the newly collected ISAcademy: A World in Motion migration and development household survey Thissurvey included 1,282 randomly selected households with migrants abroad duringMarch and May 2011, and applying the propensity score matching method for study.According to Nguyen Viet Cuong (2008), remittance is not only improving income

of receiving-remittance household, but also increasing their saving and asset Datasetfor this study was get from Viet Nam Household Living Standard surveys (VHLSS)conducted by The General Statistic Office of Viet Nam (GSO) with technical supportfrom the World Bank in years 2002 and 2004 VHLSSs’ investigation relied on data setcovered 29,530 and 9,188 household respectively in provinces and cities of Viet Nam.This paper used impact evaluation method to uncover remittances having relationshipwith saving and assets accumulation

Besides, Rahila Munir et al (2011) had found that remittances have effects onPakistan households’ saving in short run and long run For this empirical analysis,authors collected time series data for the period 1973-2007 The source of data derivedfrom Economic Surveys of Pakistan, International Financial Statistics (IFS) and FiftyYear Handbook Statistics of Pakistan The study applied Autoregressive DistributedLag (ARDL) model with variables such as Private saving, Total real remittances byPakistanis emigrants, Foreign direct investment, Real deposit rate (%), and Real GDPper capita

Based on cross-sectional data in two China provinces with 1,500 householdssurveyed by the Chinese Academy of Social Sciences in 2006, Yu Zhu et al (2009)applied OLS estimation for linear regression equation with dependent variable beingsaving, and the result showed that saving in receiving-remittance household is higherthan saving of households with other income sources.

Regarding to remittance and borrowing, using survey data from the Matlab Healthand Socio-Economic Survey (MHSS), a rural region of Bangladesh, Khawaja A.Mamun et al (2011) applied a Tobit model of OLS regression and proved remittancereducing the borrowed amount of remittance-receiving households in rural Bangladesh.This is consistent with dissertation of Rafael A P (2004) for the relationship betweenremittance and borrowing He was based on a four-survey panel data set in period of1995-2001 in case of rural households in El Salvador

Income from remittance to promote household well-being is new evidence on therole of remittances on health care expenditures by Mexican households, according toAmuedo-Dorantes et al (2009), and another study of Catalina Amuedo-Dorantes et al(2002), they found that the decision to purchase insurance via the family versus viaprecautionary saving appears to be governed by different economic variables

In paper of Catia B., Janis U (2014), they explored a positive and significantrelationship between a purchase of self-insurance motive and remittance behaviour.This result derived from a unique survey in the Greater Dublin Area, Ireland, andapplied Tobit regression The survey was conducted among 1,500 inter-viewer aged 18years and older, who arrived in Ireland between ten years and six months

Consequently, remittance is leverage to precautionary saving, increasingaccumulation of assets and insurance for members of remittance-receiving households.Therefore, we are easy to see the relationship between remittances with saving,borrowings, assets, and insurances

In addition, relating to argument of pessimistic outlook, remittances can changebehaviours or receiving-remittance households The study of Aree J and Aphichat C.(2009) in Kanchanaburi province, Thailand shown that remittance received bymigrants’ households was not use for buying household assets The evidence stemmedfrom data set investigated by the Institute for Population and Social Research (IPSR),Mahidol University, in the Kanchanaburi province of Thailand, and linear regressionmodel on household assets in 2004 on a number of migrants sending money or goods.

On the other hand, Yu Zhu et al (2009) indicated that remittance received byhouseholds in rural China is not used for saving and asset accumulation in rural China,they used it for consumption purposes This result is found based on data of the New(Rural) Cooperative Medical System survey (NCMS) in Octorber 2006 conducted bythe Institute of Population and Labor economics, Chinese Academy of Social Sciencesand Nanjing Agricultural University, and applied Ordinary least squares (OLS) andTwo-stage Least Square (2SLS) estimations

The findings mentioned above are also in line with A Castaldo, B Reilly (2007) in

a study of linking between remittance and consumption patterns in Albanianhouseholds Model of Working-Leser applied in this study, and data derived from the

2002 Albania Living Standards Measurement Survey implemented by the AlbaniaNational Institute of Statistics combined with World Bank assistance

2 3 Summary of literature review

Deriving from Pure Altruism, Pure self-interest, and Enlightened Self-Interest,theories present why remittances sent by migrant to their families In general, threemain reasons are explicated including: (a) The migrants feel their obligation to takecare the well-being of their family because of their families’ support before and afterleaving home (b) The migrants remit to invest in housing, land, livestock, or business

at home to insure their future when coming back (c) Intra-family arrangementconsidered as insurance motives to diversity the risk of volatility income

In particular, how is remittances spent in migrants’ households? There are manypoints of views around this matter These arguments give that remittances is majorused for consumption, a smaller part of remittances for savings, or repayment of debts,the other part of remittances for investment, and buying insurance Together with theseviewpoints, empirical studies indicated that there is a close relationship between theremittances and income, saving, asset, borrowing as well as insurance Remittancescontributed on increasing income in receiving-remittances households in some ofdeveloping countries When income increasing, it will lead to either saving more orincreasing assets in most of households receiving remittances Besides, increasingincome also reduces the amount of borrowing in migrants’ households, and buyinginsurance is a method to protect the migrants and their families from income vitality.

CHAPTER III - RESEARCH METHODOLOGY

The main goal of this study is to analyse whether the impact of remittances onbehaviour of remittance-receiving households or not To evaluate this effect, we applythree methods including Propensity Score Matching (PSM), Difference in Difference(DID) or double differences, and combine of PSM and double differences Thesimilarities of these methods are (i) using panel data, (ii) parameters of ‘averagetreatment effect on the treated’ (ATT) or ‘average treatment effect’ (ATE) received themost attention in impact evaluation In each method, models and variables chosen to besatisfy the different assumptions This chapter consists of four sections The firstsection presents propensity score matching method, as well as model and variableschosen to comply with its assumptions The next section shows Difference inDifference method in company with assumptions, exogenous and endogenousvariables, and models applied The third section will combine Propensity ScoreMatching with Difference in Difference to check again the findings explored in eachmodel of PSM and DID methods The final section outlines data sample chosen foranalyzing effect of remittance on behaviour of households

3 1 Propensity Score Matching

We always wonder how to evaluate the changing behaviour of a household withand without remittance It is hard to have a credible answer, because we can onlyobserve separately outcome of receiving-remittance household and outcome ofreceiving-no remittance households There is no connection between two types ofhouseholds for evaluating and comparing effects of remittances on them as well as wecan not get two outcomes with and without remittance from the same household atsame time This problem considered as bias Rosenbaum and Rubin (1983) presentedmethod Propensity Score Matching (PSM) to reduce this problem The foundation idea

of this method is base on a group of treated households whose characteristics are thesame with a group of untreated households, and compare the differences of outcomes

of two groups When matching treated and control pairs, it creates distribution ofcovariates in control group being similar to the distribution of the treated group tolower sophistication in estimates In this study, the treated group is receiving-remittance households and the control group or untreated group is receiving-noremittance households The first condition needs to comply when applying PSM that issatisfying assumptions of Conditional Independence and Common Support as follows

3.1.1 Assumptions of PSM method

Based on Rosebaum and Rubin’s Theorems (1983), Marco Caliendo and SabineKopeinig (2005), Carolyn Heinrich et al (2010) and Shahidur R.Khandker et al (2010)presented briefly the assumptions for identification of the intervention effects betweentreated group and control group as follows

Assumption 1 – Conditional Independence Assumption (CIA)

There is a set of observed covariates X not affected by treatment, the potentialoutcomes Y are independent of treatment status T:

(Y 1 ,Y 0 )⊥T|X

This assumption is also known as unconfoundedness or selection on observablecharacteristics CIA is the main factor to identify exactly the impact of intervention aswell as to ensure the differences between treated group and control group to be reducedthe selection bias This is the reason of control group to be used to construct acounterfactual for the treatment group The counterfactual is unobserved outcome oftreated subjects

Assumption 2 – Common Support or Overlap Condition [Figure 3.1]

0<P(T=1|X)<1

Meaning of this expression is the proportion of treated and untreated groups must

be greater than zero for respectively possible values of X The requirement of overlapcondition or a common support is to ensure having sufficient overlap in thecharacteristics of the treated and control group to find adequate matches ATT or ATE

is specified in region of common support, as well as the treatment assignment isstrongly ignorable when assumptions as mentioned above are satisfied

Figure 3 1 Example of Common Support

Source: Shahidur R Khandker et al (2010)

3.1.2 Model Specification of Propensity Score Matching

There are several steps when applied propensity score matching includingpropensity score matching and determining region of common support to evaluate theimpacts of remittance on income, saving, insurance, borrowing and assets ofhouseholds with and without remittance in each year 2006, 2008 and 2010 as describedbelow

Step 1: Choosing Model and Variables for estimation of Propensity Score.

As Marco Caliendo and Sabine Kopeinig (2005), logit and probit models arepreferred than linear probability models in a PSM regression model because ofunlikelihood of these functions with result variables forecasted to be out of [0,1]

bounds of probability In comparison logit with probit model, the logit model is chosendue to the logit model is found on assumptions being stronger than probit model Onthe other hand, probit model is more burdensome when calculating.

The next of step 1 is variable choice This is very important if the majorvariables were omitted, the bias will increase in resulting estimates To get the bestone, the chosen variables must satisfy the assumption of Conditional IndependenceAssumption meaning they are independent of treatment condition on the propensityscore, and variables being satisfied CIA are fixed overtime For purpose matching, thevariables chosen are independent of Remittances and stemmed from the same source.This helps to improve the precision of the propensity score They are variables withcharacteristics of fixed over time put in regression model They conclude Membersliving in household; Gender of household head; Age of household head; Living area ofhousehold; Ethnicity code; Vietnamese language; Marital status; Accessing internet;Household’s total area land for cultivated, breed or for rent; Toilet type of household,Energy for cooking; Type of house; and 11 dummy variables of 12 provinces - [Table3.1]

Command of PSM determining the propensity score and satisfy the balancingproperty in Stata platform is “pscore” command

Beside chosen variables have already in VARHS dataset, weighting in propensityscore is calculated based on information of population in twelve provinces collected inStatistical Yearbook of Viet Nam in 2006, 2008 and 2010 in comparison withhouseholds investigated in twelve provinces after merging data of VARHS 2006,VARHS 2008 and VARHS 2010 – [Table 3.2 and Table 3.3]

Step 2: Determining Region of Common Support and Balancing Tests

In accordance with Shahidur R Khandker et al (2010), the common supportregion has to be specified in the propensity score distribution for treatment and controlgroup overlap

Balancing test is applied to check the average propensity score and mean of Xare the same or not between the treatment and control groups After matchingbalancing tests, the differences in covariate means between two groups dropped.

Step 3: Matching Treated Group and Control Group

As shown in studies of Marco Caliendo et al (2005), Shahidur R.K et al (2010)and Carolyn H et al (2010), there are many techniques to be applied matching treatedindividuals control individuals In this study, the following techniques are applied:

Nearest-neighbor matching (NNM): this technique is used to matching each

treatment unit with the comparison unit in terms of the closest propensity score (orthe case most similar in terms of observed characteristic) Variants of nearestneighbor matching are related to “with replacement” and “without replacement” Inthe former case, individuals from the control group can be used more than once as amatch, whereas in the later case, it is considered only once NN matching withreplacement faces to a trade-off between bias and variance, if we allowreplacement, the average quality of matching increases and bias decrease NNmatching without replacement is related to depending on the order in whichobservations get matched Thus, it should be sure the order is randomly done