Determinants and socioeconomic impacts of migrant remittances: a study of rural Bangladeshi migrants in Italy

Transcription

1 Southern Cross University Theses 2015 Determinants and socioeconomic impacts of migrant remittances: a study of rural Bangladeshi migrants in Italy Abdul Mannan Kazi Southern Cross University Publication details Kazi, AM 2015, 'Determinants and socioeconomic impacts of migrant remittances: a study of rural Bangladeshi migrants in Italy', DBA thesis, Southern Cross University, Lismore, NSW. Copyright AM Kazi 2015 epublications@scu is an electronic repository administered by Southern Cross University Library. Its goal is to capture and preserve the intellectual output of Southern Cross University authors and researchers, and to increase visibility and impact through open access to researchers around the world. For further information please contact epubs@scu.edu.au.

2 Determinants and socioeconomic impacts of migrant remittances: A study of rural Bangladeshi migrants in Italy Submitted in partial fulfilment of the requirements for the award of Doctor of Business Administration Southern Cross University, Australia Presented by candidate: Kazi Abdul Mannan Submitted: July 2015

3 DECLARATION I certify that the work presented in this thesis is, to the best of my knowledge and belief, original, except as acknowledged in the text, and that the material has not been submitted, either in whole or in part, for a degree at this or any other university. I acknowledge that I have read and understood the University's rules, requirements, procedures and policy relating to my higher degree research award and to my thesis. I certify that I have complied with the rules, requirements, procedures and policies of the University. SIGNED: DATED: 04 July 2017 Candidate: Kazi Abdul Mannan ii

4 ACKNOWLEDGEMENTS A special thankyou goes to my thesis supervisors, Mr. Paul Weeks and Professor Dr L. J. Fredericks, for their invaluable insights, suggestions, encouragement, guidance and time spent in reading and correcting my manuscript. I also want to thank Dr. Talukdar Golam Rabby who helped me to source reference books from the University of Malaya and journal articles from internet databases. I am also indebted to the respondents, household members and local people at Vogeshore Union, where the field survey was undertaken. They were very friendly and helpful during my census and survey period. They shared their precious time, thoughts and concerns with me. Last but not least, I thank my wife (Dr. Khandaker Mursheda Farhana, Assistant Professor, Shanto-Mariam University of Creative Technology, Head of Department, Sociology and Anthropology) for her support, patience and understanding as I did not spend enough time with her during the course of my study and thesis preparation. iii

5 ABSTRACT International migration has major socioeconomic consequences, one of which is the flow of remittances from migrant workers in different countries to their places of origin. The flow of remittances is a complex phenomenon with national and global economic, social, legal and cultural implications. In the literature, some impacts of remittances have been studied, especially for certain developing countries. However, the socio-economic impacts of international remittances from developed countries on rural households in lesser developed countries is still poorly understood. Thus this research focuses on the factors affecting the flow and socioeconomic impacts of remittances sent by migrant workers in Italy to rural households in Bangladesh. It draws on migration literature, including the New Economics of Labour Model (NELM), which simultaneously considers migration determinants, remittance behaviour and impacts. A conceptual model is proposed which captures the social, micro-economic and sociodemographic factors affecting remittance flow, as well as the economic and social impacts on the workers households. The model is operationalised using existing measures from the literature. To test the model, a three-stage research design was used. Firstly, a census of the 4013 households in ten villages in the Shariatpur district of Bangladesh was conducted, in order to identify the village household members working in Italy. Of these, 338 households received remittances from household members working in Italy. Secondly, a pilot survey was conducted on 50 households within the sampling frame. Thirdly, a full scale survey using the pilot-tested questionnaire was undertaken with a sample of 300 (excluded 50 pilot sample) rural households. The survey data collected was tested using univariate and multivariate estimations (log linear regressions, and chi-square and gamma tests). The descriptive survey data provides insights into the socio-demographic characteristics of the migrant household members and heads of household. For instance, the majority of migrants are between 21 and 35 years old (59%), have educational attainments of less than higher secondary (98%), are mostly married (71%). Most have been migrants for between five and nine years (56 per cent), most had the required legal documentation (83 per cent) and had visited their original homes two to five times since they had emigrated (61 per cent). Most heads of households were aged between 51 and75 years (57 per cent), 53 per cent were male, iv

6 most were married (83 per cent), most had educational levels lower than secondary school (92 per cent) and the most common occupation was that of housewife (47 per cent). Households in the study area came from two religious backgrounds: Muslim (68 per cent) and Hindu (32 per cent). Most of the households had between two and five members (51 per cent), excluding migrant members during the study period. The largest cohort of household properties had single ownership titles in the name of the land migrant (47 per cent), and the majority of household properties were liable to be inherited by two to three members (83 per cent). The results of the multi-variate analysis supported the relationships posited in the conceptual model. Seven hypotheses were developed and tested. It was found that there were three dominant factors: altruism, investment and kinship. Household size, annual income, assets and inheritance were non-significant predictors for remittances flow. The findings also show that the micro-economic determinants of remittance were influenced by three characteristics: migrant, household head and household composition. Remittance flow was associated with migrant characteristics such as age and number of visits by the migrant. The results showed that younger people were more likely to migrate, and that the remittances sent by younger emigrants tended to be higher. Younger emigrants remittances were found to be significantly linked to the motivation to maintain strong kinship relations with household members in Bangladesh. Other significant influential determinants of remittances were the household head s age, gender and marital status, the household head s relationship to the migrant, composition of the household s investment and household living expenses. There was a highly significant relationship between remittances and attributes such as the migrant s age, education and legal status. In addition, there was also a correlation between remittances and the household head s relationship to the migrant, and between remittances and household characteristics such as religion, title of land, amount of land, investment in housing development, business, welfare and loan repayments. The research presented in this thesis suggests that the size and frequency of remittances and the factors affecting socioeconomic impacts differ at the destination, community and household levels. It contributes to knowledge on the dynamics of international remittances by establishing a backward elimination regression process to analyse the cross-sectional survey data. The econometric model is developed to observe the association between socioeconomic impact and the exploratory determinant variables. Thus, this implies that the impacts of v

7 diverse remittance determinants are crucial in designing the policy for specific categories of remain-behind village households. vi

8 LIST OF ABBREVIATIONS ADB AMC ATM BB BBS BMET BoP BHRS CBA CIA CMA CMD DR EAP EC FAO GBS GCIM GDI GDP GI GNP GoB HH HIES ILO IMF IOM LCM LDC LSMS Asian Development Bank Asia Migration Centre automatic teller machine Bangladesh Bank Bangladesh Bureau of Statistics Bureau of Manpower Employment and Training balance of payments Bangladesh Household Remittance Survey cost-benefit analysis Central Intelligence Agency Center for Migrants Advocacy Centre for Migration and Development dependency ratio economically active population European Community Food and Agriculture Organisation Guyana Bureau of Statistics Global Commission on International Migration gross development index gross development product Gini Index gross national product Government of Bangladesh Household head Household Income and Expenditure Survey International Labour Organisation International Monetary Fund International Organization for Migration less concentration of migrant households less developed country Living Standard Measurement Survey vii

9 MCM MFA MLM MNC MTO RMMRU NGO OECD PLSS RUM SAMReN SEARC SMC SPSS UN WES more concentration of migrant households Migrant Forum in Asia multi-level marketing company multinational corporation money transfer operator Refugee and Migratory Movements Research Unit non-governmental organisation Organization for Economic Co-operation and Development Peruvian Living Standards Survey rural urban migration South Asian Migration Research Network Southeast Asia Research Centre Scalabrini Migration Centre Statistical Package for Social Sciences United Nations Wage Earner s Scheme Equivalence 1.0 USD (United States Dollars) BDT (Bangladesh Taka) Source: Universal currency converter, 14th September viii

10 TABLE OF CONTENTS Title Declaration Acknowledgements Abstract List of Abbreviations Table of contents List of tables List of figures i Ii Iii Iv Vii Ix Xvi Xx CHAPTER ONE: INTRODUCTION 1.1 International Migration and Remittances Global and Regional Migration and Remittances Migration and Remittances: a Bangladeshi perspective Research objectives: resreach problem, research questions and hypotheses Justification for the research and contribution Methodological implications Theoretical implications Policy implications Theoretical and Conceptual framework The Research methodology Delimitations Definitions of key terms Thesis outline Conclusion 13 ix

11 CHAPTER TWO: LITERATURE REVIEW 2.1 Introduction The research setting involving Bangladesh and Italy Bangladesh as a remittance receiving country Italy as a migrant-receiving and remittance-sending country Bangladeshi migrants in Italy Theories of migration Macro theories of migration The neoclassical theory Mabogunje s (1970) Migration as a system model Historical-structural theory Micro theories of migration The New Economics of Labour Migration (NELM) Migration and remittances Amplifications to the motivation to remit Modelling the motivation to remit Empirical evidence of the determinants and impact of remittances Macroeconomic determinants and impacts on developing countries Poverty and inequality impact Health and education impact Investment and entrepreneurial activity Labour supply and participation impact Economic growth impacts Micro determinants of remittances and impacts Empirical literature on migration and remittance Remittances at the national level (Bangladesh) International Migration Micro studies of remittances The nexus between socio-demographic characteristics and remittances and 50 impacts Remittance sender 50 x

12 2.6.2 Household head Characteristics of household Summary of the extant literature Research gap Research problem Research questions Research hypotheses Theoretical framework Conclusion 63 CHAPTER: THREE METHODOLOGY 3.1 Introduction Study paradigm Paradigms in social science research Research Methodology Selection of survey village and field survey Study Design STAGE 1: The households census in the 10 survey villages Construction of questionnaire and interview schedule Sample selection Sampling procedures and data sources Sampling technique and plan Sample Size Coding and measurement scales STAGE 2: Pre-testing and questionnaire modification STAGE 3: Field experience Data quality control Reliability of the instruments Data limitations Limitations of the survey Data analysis 77 xi

13 3.3.1 Data processing procedures and statistical program Univariate analysis Bivariate Analysis Gamma test Multivariate analysis Research measurement implications Research quality Objectivity and conformity Generalisability Reliability and dependability Validity Internal validity, credibility, and authenticity External validity, transferability and fit Errors Ethical considerations Conclusion 86 CHAPTER FOUR: DATA ANALYSIS 4.1 Introduction Assumption testing Normality distribution Socio-demographic profile Remittance sender characteristics Age and remittances Educational level and remittances Marital status and remittances Year of migration and remittances Legal status and remittances Number of visit and remittances Characteristics of household heads Age and remittances 93 xii

14 Gender and remittances Marital status and remittances Educational level and remittances Employment status and remittances Relationship of household head to migrants and remittances Household characteristics Household religious status and remittances Household size and remittances Household land ownership and remittances Number of inheritors in household and remittances Factors influencing remittance flows Econometric model building and multivariate analysis Remittance determinants of the rural households Empirical results Whole sample Unit analysis of the age of migrants The age group The under 30 age group The 31 to 35 age group The age group The above 40 age group Unit analysis of the age of rural household head The under 35 age group The age group The above 60 age group Comparative analysis of gender of the household head Analysis of marital status of the household head Married household head Unmarried household head Unit analysis of the household relation to the migrant Father as a household head Wife as a household head 128 xiii

15 Brother as a household head Mother as a household head Socioeconomic impact of remittances on the households Remittances impact on households Uses of remittances Changes in household socioeconomic status Changes in family status Relationship between household socio-demographic characteristics and socio-economic impact of remittances Construction impact variable Empirical econometric model to analyse of socioeconomic impact Empirical results of regressions Socioeconomic impact and migrant socio-demographic characteristics Age Educational level Legal status Socioeconomic impact and socio-demographic characteristics of household head Household head relation to migrants Socioeconomic impact and household socio-demographic characteristics Religion Conclusion 147 CHAPTER FIVE: DISCUSSION 5.1 Introduction Responses to the research questions Research question Socio-demographic profiles of migrant workers sending remittances to originating households 149 xiv

16 Household head profile Household composition Research question Migrant characteristics Characteristics of the head of household Household compositions Research question Household level Community-level impacts Uses of remittances at the family level Research question Conclusions about the research hypotheses Conclusions about the research problem 174 CHAPTER SIX: CONTRIBUTIONS AND LIMITATIONS OF THE STUDY 6.1 Introduction Contribution Knowledge supported by this research Knowledge extended by this research Policy recommendations Research limitations Future research Conclusion 183 xv

17 REFERENCES 185 APPENDIX A : Data Normality Test 209 APPENDIX B : The Survey Questionnaire 223 APPENDIX C : Survey Invitation 228 APPENDIX D : Information Sheet 231 APPENDIX E : Human Research Ethics Committee (HREC) Notification 235 LIST OF TABLES CHAPTER: TWO Table 2.1 : Outflow of Bangladeshi migrants to Italy, Table 2.2 : Official inflows of remittances from Italy to Bangladesh Table 2.3 : Summary of the International Migration literature 39 Table 2.4 : Summary of the Micro studies of remittances literature 43 Table 2.5 : Summary of Expected Direction of Regression Coefficients by Hypotheses 58 CHAPTER: THREE Table 3.1 : Allocation of 10 village HH among the field workers 68 Table 3.2 : Population and sample of household of the study area 72 Table 3.3 : Specification of variables for multivariate analysis 74 CHAPTER: FOUR Table 4.1 Table 4.2 : Distribution of household annual remittance received and migrant s age : Distribution of household annual remittance received and education of migrants xvi

18 Table 4.3 Table 4.4 Table 4.5 Table 4.6 Table 4.7 Table 4.8 Table 4.9 Table 4.10 Table 4.11 Table 4.12 Table 4.13 Table 4.14 Table 4.15 Table 4.16 : Distribution of household annual remittance received and marital status of migrants : Distribution of household annual remittance received and period of migration : Distribution of household annual remittance received and legal status of migrants : Distribution of household annual remittance received and number of visits by migrants : Distribution of household annual remittance received and age of household heads : Distribution of household annual remittance received by gender of household heads : Distribution of household annual remittance received and marital status of household heads : Distribution of household annual remittance received and education of household heads : Distribution of household annual remittance received and employment of household heads : Distribution of household annual remittance received and migrant s kinship relation to household heads : Distribution of annual remittances received and religion of household heads : Distribution of household annual remittance received and household size : Distribution of household annual remittance received and household land ownership : Distribution of household annual remittance received and household number of heirs Table 4.17 : Regression results of the full model 105 Table 4.18 : Regression results of stage II 106 Table 4.19 : Stage II regression model summary 107 Table 4.20 : Regression results of stage-iii 108 xvii

19 Table 4.21 : Regression results of stage-iii 109 Table 4.22 : Stage-III regression model summary 109 Table 4.23 : Final stage regression results 110 Table 4.24 Table 4.25 Table 4.26 Table 4.27 Table 4.28 Table 4.29 Table 4.30 Table 4.31 Table 4.32 Table 4.33 : Determinants of average household remittance: Log linear regression results of the 10 rural villages, 2013 : Log linear regression results of age group 26 to 30 years old migrants in the 10 rural villages, 2013 : Log linear regression results of below 30 years old migrants in the 10 rural villages, 2013 : Log linear regression results of age group 31 to 35 years old migrants in the 10 rural villages, 2013 : Log linear regression results of below years old migrants in the 10 rural villages, 2013 : Log linear regression results for the above 40 year-old migrants in the 10 rural villages, 2013 : Linear regression results for household heads younger than thirty five years in the 10 rural villages, 2013 : Linear regression results for the fifty six to sixty year-old household heads in the 10 rural villages, 2013 : Long-Short linear regression results for the above sixty year-old household heads in the 10 rural villages, 2013 : Log linear regression results of gender of the head of household in the 10 rural villages, Table 4.34 : Log linear regression results of married household head in the 10 rural villages, Table 4.35 Table 4.36 Table 4.37 : Log linear regression results of unmarried household heads in the 10 rural villages, 2013 : Long-Short linear regression results of father as a household head in the 10 rural villages, 2013 : Short linear regression results of wife as a household head in the 10 rural villages, Table 4.38 : Short linear regression results of brother as a household head in 130 xviii

20 Table 4.39 the 10 rural villages, 2013 : Short linear regression results of mother as a household head in the 10 rural villages, Table 4:40 : Uses of remittances by household members 133 Table 4.41 Table 4.42 : Sector-wise financial investment by remittance receiving rural households : Respondent s perceptions of the socioeconomic impact of remittances Table 4.43 : Socioeconomic impact of remittances on family status 137 Table 4.44 Full model regression results 140 Table 4.45 : Model summary 141 Table 4.46 : Significant coefficients of the best fit model results 142 Table 4.47 : Age of migrants and socioeconomic impact from remittances 143 Table 4.48 : Education of migrant and socioeconomic impact of remittances 144 Table 4.49 : Socioeconomic impact from remittances and legal status 145 Table 4.50 : Kinship and socioeconomic impact of remittances 146 Table 4.51 : Religion of household head and socioeconomic impact of remittances CHAPTER: FIVE 147 Table 5.1 : Socio-demographic profile of migrant household members 149 Table 5.2 : Socio-demographic profile of migrant household heads 152 Table 5.3 : Socio-demographic profile of migrant household 154 Table 5.4 : Factors influencing remittance flow to rural households 157 Table 5.5 : Socioeconomic impact from remittance at the household level 162 Table 5.6 : Socioeconomic impact from remittance at the community level 164 Table 5.7 : Significant relationship between determinants of remittances and of socioeconomic impact 168 Table 5.8 : Summary of research hypotheses 171 Table 5.9 : Investigation of key research issues 175 xix

21 LIST OF FIGURES CHAPTER: TWO Figure 2.1 : Remittances flows are large, and growing 24 Figure 2.2 : Concept map of this section 53 Figure 2.3 : Conceptual framework 60 CHAPTER: THREE Figure 3.1 : Research process 66 CHAPTER: FOUR Figure 4.13 : Distribution of sector wise remittance uses 134 xx

22 CHAPTER ONE: INTRODUCTION 1.1 International Migration and Remittances International migration brings a huge inflow of financial remittances for migrant origin countries and provides opportunities for domestic unemployed workers who may represent significant problems for their households, society and the country. As a result, policy makers in many less-developed countries support international migration in order to reduce unemployment problems and acquire valuable foreign currency, thereby allowing greater economic development in the country of origin. In recent decades, the numbers of migrants have been increasing all over the world. Most of these migrants come from developing countries Global and Regional Migration and Remittances According to newly available census data, the stock of international migrants is estimated at 247 million in 2013, significantly larger than the previous estimate of 232 million (UN 2014). This number has been growing at an annual rate of 1.6 per cent, and will surpass 250 million in 2015 (World Bank 2014). The largest five migrant receiving countries remain the United States, Saudi Arabia, Germany, the Russian Federation, and the United Arab Emirates. In 2013, the largest migration corridor in the world is accounted for 13 million migrants from Mexico to the United States (UN 2014). Russia to Ukraine is the second largest, followed by Bangladesh to India, and Ukraine to Russia. The latter three are South-South corridors according to UN classification. Indeed, South-South migration stood at 37 per cent of the global migrant stock, larger than South-North migration at 35 per cent (World Bank 2014). At the same time, South-South remittances estimated for 34 per cent of global remittance flows. Formal estimated remittance flows to developing nations are accounted to have touched $436 billion in 2014, a raised of 4.4 per cent over a year ago (World Bank 2014). Remittance flows to developing nations are projected to slow down to 0.9 per cent growth in 2015 (to $440 billion), owing to a weak economic outlook in remittance source countries in Europe and Russia. Remittance flows are supposed to accelerate in 2016, and touch $479 billion by 2017 because of the more assured world economic standpoint (World Bank 2014). Global remittance receipts, including by both 1

23 developing and high-income countries, are estimated at $583 billion in 2014, and could rise to $586 billion in 2015 and $636 billion in 2017 (World Bank 2014). Remittances stand a vital source of earnings for developing nations, so far surpassing official growth facilitation and surprisingly foreign direct investment (except China). Recently, the World Bank Global Economic Prospects 2014 reported that remittances are also less volatile than formal aid inflows. Annual remittances are also larger than, or equal to, foreign exchange reserves in many small countries. Even in large emerging markets, such as India, remittances are equivalent to at least a quarter of total foreign exchange reserves. India, China, Philippines and Mexico retained their position as the top recipients of migrant remittances in Remittances as a share of GDP are larger in small economies, particularly in Central Asian countries and Pacific islands e.g., about 49 per cent of GDP in Tajikistan and a quarter of GDP in Tonga (UN 2014). This high dependency on remittances increases these countries vulnerability to shocks from remittance-sending countries. In this context, it is noteworthy that recently there has been a worldwide discussion about the stability of Bangladesh s economy and its future. According to a United Nations (2013) report, a number of least developed countries, including Bangladesh, saw solid investment and consumption in 2012, supported by sustained inflows of workers remittances. External demand from the overseas countries opened up the scope for a large number of Bangladeshis (mostly semi-skilled or less skilled) to go abroad as migrant workers. The migration of Bangladeshis has taken place as a consequence of several pull (e.g., demand for labour, kith and kinship associations with other Bangladeshis living abroad, etc.) and push factors (e.g., poverty, unemployment/underemployment/disguised unemployment, lack of development in rural areas, etc.) Migration and Remittances: a Bangladeshi perspective This section describes migration and remittances from a Bangladeshi perspective. The Bureau of Manpower Employment and Training (BMET) is the only government source of migration data in Bangladesh. Thus, it has been the main source of information for this section. Labour migration from Bangladesh has been taking place since the 1970s. Data from the last 10 years clearly show that about four-fifths of all Bangladeshi labour migration was to countries in the Middle East. The majority of Bangladeshi migrant workers are employed in Saudi Arabia, the United Arab Emirates, Kuwait, Oman, Malaysia, and Singapore. In 2

24 addition, people are also going to Bahrain, Qatar, Jordan, Lebanon, the Republic of Korea, Brunei, Mauritius, the United Kingdom, Italy and Ireland. While Middle Eastern countries may still be the principal destinations for Bangladeshi migrants, an analysis of destination countries over time reveals significant changes (BMET 2014). In 2001, about 72 per cent of the Bangladeshi workers who journeyed abroad that year went to Saudi Arabia. But this proportion fell down to only 3 per cent in 2011 (BMET 2014). In contrast, in 2001, about 9 per cent of the total migrant labour pool went to the United Arab Emirates, but by 2011 that figure had reached 49 per cent (BMET 2014). During the same period, the number of migrant workers from Bangladesh heading to Oman increased by a factor of twelve. In 2001, only about 4 per cent of Bangladeshi migrant workers went to countries outside the Middle East, but, by 2011, destinations outside that region attracted 13 per cent of departing migrant workers (BMET 2014). If one only considers the period from 2007 through 2011, one will see that Bangladeshi labour migration to Saudi Arabia, Kuwait, and Malaysia dropped by 92.6, 99.3, and 99.7 per cent respectively over that five year period (BMET 2014). But, in most cases, the continuation of this migration movement could not be maintainable and came to a halt. In the fiscal year, the remittance sent home to Bangladesh by migrant workers was US$14.46 billion (BB 2014). Remittance during was about US$12.84 billion, a figure that was 10 per cent larger than that of the previous fiscal year (BB 2014). The flow of remittance into Bangladesh has almost always been on an increasing trend. Since , the yearly remittance flow has soared, and is now nearly 295 times greater that it was 36 years ago (from US$49 million to US$14,460 million) (BB 2014). In 2012, Bangladesh ranked sixth among the top 10 remittance-receiving countries. Bangladesh is the second largest remittance receiver in South Asia. The Sixth Five Year Plan (FY2011 FY2015) of the Bangladesh Government accordingly acknowledges the importance of migrant workers remittances for the development of Bangladesh (Bangladesh Bank 2014). This research was designed to identify the micro-determinants of remittances and their socioeconomic impacts on the originating households. More importantly, this research examined the ways in which remittances were utilised by the migrants originating households and explored positive or negative socioeconomic changes in these households. For the assessment of the impact of remittances, the technique of comparing cross sectional survey data was used. Likewise, correlations between various independent variables (age, 3

25 marital status, education, religion, legal status etc.) and dependent variables (changes in household socioeconomic status, changes in family status etc.) were established to investigate the impacts of remittances. 1.2 Research objectives: research problem, research questions and hypotheses Research objectives The research objectives clarify the specific purposes of the study (Zikmund et al. 2010). Based on the research questions enumerated above, this study has four objectives: 1. To examine the socio-demographic profiles of the foreign migrant worker and the heads of Bangladeshi households receiving money from their family members working in Italy. 2. To explore the factors influencing the amount of money sent by household members working in Italy to their homes in Bangladesh. 3. To investigate the socioeconomic impact of such remittances on the households in Bangladesh. 4. To trace the relationships between the socio-demographic profiles of the Bangladeshi households receiving remittances from Italy with their determinants and socio-economic impact. As will be detailed in Chapter 2, a review of the relevant research identified specific gaps in understanding the connection between the micro-determinants of international migrant remittances and their impact on rural Bangladesh households and communities. The research problem, therefore, requires analytical insights into the determinants of migrant remittances, including household composition, social and economic conditions and their impacts on rural households and communities. Accordingly, the research problem can be stated as follows: What are the key determinants of remittance inflows from migrant workers in Italy and their socioeconomic impacts on the originating households in rural Bangladesh? To address the research problem based on the research gaps identified in the literature, four research questions were developed: 4

26 1. What are the socio-demographic profiles of the foreign migrant workers and of the heads of the Bangladeshi households receiving remittances from their family members working in Italy? 2. What are the determinants of the remittances sent by household members working in Italy to their households in Bangladesh? 3. What is the socioeconomic impact of such migrant worker remittances on the originating households in Bangladesh? 4. What are the links between the socio-demographic profiles of the Bangladeshi households receiving remittances from Italy with the determinants and socioeconomic impact of such remittances? The following seven research hypotheses were developed for this study: H1 The higher the household living expenditures, the higher the likelihood of a migrant s sending remittances. H2 The higher the number of dependents in the household, the higher the likelihood of a migrant s sending remittances. H3 The lower the household s income, the higher the likelihood of a migrant s sending remittances. H4 The lower the household s assets, the higher the amount of remittances from migrants. H5 The higher the prospects for future inheritance from household members, the higher the propensity to remit. H6 The more favourable the investing opportunities in the origin community, the higher the propensity to remit. H7 The stronger a migrant's links to the community, the higher the likelihood of migrant s sending remittances. 5

27 1.3. Justification for the research and contribution The available literature at the national level on socio-demographic characteristics of international migrants (Murshid et al. 2002; Siddiqui 2004; Bangladesh Bank 2011) explores overall migration destination. Micro-economic level literature on international remittances reveals that there are some important factors influencing remittances sent to the home country (Mishra 2011). These include individual migrant characteristics such as age, marital status, education, year of migration, legal status and number of visits to home country. Household members attributes including household head age, gender, educational attainments, marital status, employment status and relation to migrant as well as household size and religion, also are also influences. The literature of international migration and on the impact of remittances at the macro and micro levels explores that remittances have macro-level impacts as well as micro-level impacts on the home country economy (De Haas 2006). The national- and international-level literature on migration and remittances, and of the literature on the macroand micro-level impacts of remittances, indicate a gap in the understanding the determinants of remittances and their socioeconomic impacts particularly non-ec rural household members who migrate to the EC, specifically in Italy. The present research is also justified based on the methodological, theoretical and policy implications of the determinants of migrant worker remittances and their socioeconomic impacts on remittance-receiving households in a developing country Methodological implications From an economic perspective, the analysis in this study helps in the examination of the factors influencing the flow of foreign worker remittances and their impact on the originating households which perceive international migration as a livelihood diversification strategy. It follows that, given the significance of these remittances to Bangladesh from a macroeconomic perspective, and to the sustainability of the originating (and largely poor) households, policy makers in Bangladesh must consider them in designing policies to support such households. 6

28 1.3.2 Theoretical implications This study has reviewed the major theories and literature on international migration and remittances to explore the dynamics of this socioeconomic phenomenon as they relate to its causes, consequences and impacts in the originating households and communities. The study examines the links between the socio-demographic attributes of the head of household, attributes of the migrant, household composition and the international remittances and remittance inflows and socioeconomic impacts at the household and community levels. The association between socio-demographic characteristics, remittance inflow and socioeconomic impact is found to be logical, and remittance is a rational strategy adopted by rural households as a livelihood diversification strategy Policy implications For a poor country like Bangladesh with its large and growing population base and relatively scarce resources in a climatically-afflicted environment, sustainable economic growth is as much a complex challenge as an incontrovertible prerequisite. It is anticipated that the findings from this study will contribute to a deeper understanding of this complex phenomenon, its theoretical underpinnings and the design of policies to ensure higher economic benefits from such foreign earnings for the originating households in emerging countries such as Bangladesh. It is undeniable that foreign remittances by migrants play a vital role in the Bangladesh economy; thus, a more insightful appreciation of the determinants of remittance inflows would enable the government to design policies and strategies to optimise their socioeconomic benefits for rural households and stimulate the growth of foreign remittances in Bangladesh. It is foreseen that the outcomes from this research will contribute to the understanding of its theoretical underpinnings and the formulation of relevant policies to improve the use of such remittances to improve the livelihoods of the originating households in emerging countries. 1.4 Theoretical and conceptual framework The early neo-classical literature on migration did not explore the role of foreign remittances on households as it focused more on the migration decision of the individual or on migrant income maximisation strategies (Sjaastad 1962; Todaro 1969; Harris & Todaro 1970). Others perceive that the decision to emigrate more as a household decision-making strategy to 7

29 minimise risk and diversify livelihoods (Stark & Taylor 1986). According to the New Economics of Labour Migration (NELM), decisions about whether one member of a household should emigrate are made jointly by households (Stark & Bloom 1985). Therefore, the issue of remittances and their determinants has become a key consideration in a number of theoretical and empirical studies. A distinguishing attribute of the NELM model is its simultaneous consideration of migration determinants, remittance behaviour and impacts, making NELM models relatively demanding in estimation methods and data requirements. Another key insight of NELM is that households allocate members to improve not only absolute but relative incomes and alleviate their deprivation against a reference cluster such as the village community (Stark & Bloom 1985; Stark et al. 1986; Stark & Taylor 1989; Stark 1991). Micro-level studies of remittance determinants and their impacts on the household at the rural level are based on either household primary surveys (Gubert 2002), or surveys of the individual migrants at the originating or destination country (Amuedo-Dorantes & Pozo 2006; Holst & Schrooten 2006). Against such a theoretical background, a broad stream of studies has identified the determinants of remittances and their socioeconomic impact on countries, communities, and households that have sent workers abroad (Borja 2012). Two broad perspectives have been adopted: macro determinants and the impact of foreign remittances on originating countries and their micro determinants and consequences. The focus in this study will be on the latter as the study area is the Shariatpur District of Bangladesh which has traditionally sent migrant workers to Italy. In this study, variables are divided into three categories namely background (sociodemographic), intervening independent and dependent variables. Background variables include three major categories such as migrant (age, marital status, education, year of migration, number of visit), household head (age, gender, marital status, education, religion, employment status, kinship) and household composition (household size, annual income, land ownership title, living expenses, welfare expenses, investment in house development, investment in financial sector, investment in business and loan repayment). All these background variables are playing important role in determining the determination of remittance and the impact of remain-behind households. The dependent variables are also directly affected by the background variables thus these intangible variables exist intrinsically and are amongst factors in determining the determinants of remittances and socioeconomic impacts. 8

30 Intervening independent variables are manipulated, influential experimental factors. In this study, these include social, economic and household conditions. Dependent variables are the variables that are measured in a research project. The variables may change as the independent variable is influenced. The term dependent variables is used as they depend on what happens to the respondents in the examination in terms of intervening and background variables. The dependent variables in this study are annual remittance received by the household and socioeconomic impact. 1.5 The research methodology This study on the determinants of foreign remittances to Bangladeshi households and their linkages adopts the positivist paradigm and uses a quantitative approach. The phenomena being examined are social constructions informed by previous research on the international migration and development nexus, on foreign remittances sent through formal and informal channels by temporary transnational migrants, on the nature and determinants of the remittance flows and the socioeconomic structure of the receiving households, and on the relationships between the latter two factors. The study methodology used to gather the primary data required to inform conceptual framework is a survey. Details of the survey design are provided in Chapter 3. The first step is to identify the research area in Bangladesh, which is the Shariatpur District, which has traditionally sent migrant workers to Italy. In the 10 survey villages identified, a complete census of all households was conducted to ascertain which households were receiving remittances from household members working in Italy. After this, a two-stage procedure a pilot study and a face-to-face questionnaire survey was conducted to collect the primary field data from the sample 300 households. In the first stage of the data collection process, a representative sample of 300 households from 10 villages in Vogeshore sub-sub-district (Union) in the Naria sub-district (upazila) in Shariatpur district of Dhaka division in rural Bangladesh was selected to examine the household population receiving remittances from household members working in Italy. A structured questionnaire was used in the primary survey to ensure comparability of the data, increase speed and accuracy of recording, and facilitate data processing (Malhotra 2010, p. 335) using a four-point Likert attitude rating scale. The questionnaire, administered by the researcher, was formulated in English before being translated into the Bangladeshi language (the researcher is a Bangladeshi) using the most familiar and locally known words. Prior to 9

31 the actual questionnaire survey, the questionnaire was tested in a pilot study to improve its validity. The primary data gathered was analysed by applying suitable quantitative techniques to clarify the research problem and provide responses to the research questions. The data was analysed by applying tabular (cross tabulation) and graphical formats for the individual and household characteristics, and univariate and multivariate regression models to examine the nature of associations hypothesised. In this way, the socio-demographic profile of households, the determinants of the remittance flows and their relationships and impacts were explored. 1.6 Delimitations This study did not consider household-level characteristics like dependency ratios, number of daughters, number of school-going children, quality of land and accessibility to agricultural facilities were also not considered as remittance determinants for different types of households. For an in-depth understanding of the female household head s vulnerabilities, research is required on social exclusion and gender relationships. This study, however, did not delve into this topic as it predominantly focuses on the economic perspectives of remittances. This research as based on household survey data from rural Bangladesh. Migrant information was limited by the non-presence of the migrants. Thus, important variables that could not be included in this study included migrant income, expenditure, employment status and living status in Italy. The questionnaire survey participants were located mainly in the rural sub-subdistrict of Shariatpur. The study findings, thus, may not be generalisable for the whole of Bangladesh. This study also did not look into: the tensions between individualisation and obligations and pressures originating in the home country that shape the development and circulation of social remittances; the interface between social remittances and social capital; the complex ways in which the physical and social distance between senders and recipients affect the circulation of social remittances, or their impact on migrants communities of origin. 10

32 1.7 Definitions of key terms Household A group of persons, related or unrelated, living together and taking food from the same kitchen (BBS 2011, p.2). Migration Migration is defined in this survey as the movement of persons who change their place of residence, except for marriage, for a period of six months or more. The time period for the migration due to marriage is not fixed (BBS 2011, p.5). Migrant household A migrant household is defined as having at least one of its members living in Italy during the survey period between July to December, MCM and LCM districts There are 64 districts in Bangladesh. Districts that had the same proportion of migrant households as the national average or higher were defined as MCM districts (More Concentration of Migrant Households). Thus, LCM districts (Less Concentration of Migrant Households) were the districts having smaller proportions of migrant households than the national average. Migrant data Since migrants were living abroad, data about them was obtained from another member of the household chosen as the respondent for the survey questionnaire. Thus, information reported in this survey about migrants should not be understood as information provided directly by the migrants themselves. Respondent A respondent was defined as a head of the household or a senior member of a migrant household who answered the questionnaire in the interview about the household and its migrants. The respondent was usually the household head. In cases when the head of 11

33 household was not available to be interviewed, another adult household member able to answer the survey questionnaire was interviewed. Proxy respondents The respondent was usually the household head. In cases when the head of the household was not available to be interviewed, a proxy respondent was used. This was another adult member of the household. Rural area A farming area or an area related to farming. Rural areas can be agricultural in character. Remittance Remittance refers to financial transfers that are sent by a migrant worker in a foreign country to his/her household. 1.8 Thesis outline This thesis has a unified structure based on six chapters. Chapter One sets the scene for the research. It describes the background of this research, the justification for conducting it, its contributions to knowledge, and the research questions. The research methodology, ethical issues, definition of key terms, study and thesis chapter structure are briefly outlined. Chapter Two reviews the relevant literature on the background and focal theories (Phillips & Pugh 2005) to build a theoretical foundation for the research. The research issues and gaps in the existing body of knowledge were identified, followed by the development of research questions for subsequent testing. Chapter Three discusses the methodology for the research. It examines the various research paradigms and justifies the use of a quantitative study methodology, together with the study population and sample extraction. The structured questionnaire is described followed by the data analysis techniques and the pertinent ethical issues. Chapter Four reveals the collected primary data from the questionnaire survey in the context of the four research questions. This chapter examines the results from the respondents in the sample data set, and is set out using a number of numerical tables and explanatory text. 12

34 Chapter Five presents a comparative analysis of the research findings with those of the literature reviewed in Chapter Two. Data from the sample is collated and analysed through a series of tables, with related analytical discussion. Chapter Six discuss the implications of the research for theory development, policy formulation and practice and recommendations for policy makers together with suggestions for future research. 1.9 Conclusion This chapter has introduced the research topic, the research problem and research questions. The research project was justified in terms of its utility and contribution to scholarship, policy and practice. The chapter also outlined the research approach, research design and the data collection instrument followed by its limitations and ethical concerns. On these foundations, the thesis can proceed with a detailed description of the research. 13

35 CHAPTER TWO: LITERATURE REVIEW 2.1 Introduction This chapter provides a review of the existing knowledge of the topic under examination to enable its conceptualistion with accuracy and clarity (Neuman 2006). As per the advice of a number of authors (Perry 2002; Marshall & Rossman 1999; Baker 2000; Sekaran 2003; Neuman 2006), the review of literature in this research thesis is undertaken with the following goals: a. to indicate the significance of the research problem b. to demonstrate the key research issues and the emerging themes affecting the problem c. to identify the research gaps as a basis for formulating the research questions. d. to develop a theoretical framework for the study This study required the examination of the literature on such critical issues as why people migrate, whether domestically or internationally (the migration-livelihood-development nexus), who are the people most likely to migrate (their socio-demographic profiles), the remittances sent to their originating households (determinants of remittances), the sociodemographic attributes of the originating households, in particular the heads of households, and, finally, what the impacts of the migrant remittances are on the originating households. The reviewed literature was drawn from global sources and studies covering the beforementioned issues. The various bodies of theory are linked and help contribute to meeting the chapter objectives. The background (or parent) theories are the Bangladeshi contextual setting, migration, international remittance theories and models, and trends in the analysis of rural households. The focal (or immediate) theory of the research is the New Economics of Labour Migration (NELM). Section 2.1 introduces the literature review, the purpose of the literature review, the concept map and the structure of Chapter Two. Section 2.2 provides a review of the contextual information about the study area as well as the destination country. 14

36 Section 2.3 reviews migration and development theories such as neo-classical theory, Mabogunje s (1970) migration as a system model, historical-structural theory and the New Economics of Labour Migration (NELM). Section 2.4 discusses the determinants of remittances, motivation to remit, amplifications of the motivation to remit, modelling the motivation to remit, empirical evidence of the determinants and the impact of remittances. Section 2.5 examines the empirical evidence on migration and remittance from both national (Bangladeshi) and international perspectives to identify relevant theories, models and research gaps. Section 2.6 explains the nexus between socio-demographic characteristics and remittances. It also examines the determinants of remittances in the context of remittance sender, household head and household composition. Section 2.7 summarises the review literature and identifies the research gaps, the research problems and formulates the four research questions. Section 2.8 discusses the theoretical framework for this study. Section 2.9 concludes the chapter. 2.2 The research setting involving Bangladesh and Italy This section provides a review of the contextual information about the study area as well as the destination country, as discussed below Bangladesh as a remittance receiving country Bangladesh has a relatively small land mass of 147,570 sq. km (about 0.03 per cent of the world s land surface) and a population of million people, making it the eighth-most densely populated country globally (976 person km 2 ) (Bangladesh Bureau of Statistics (BBS), 2012). Of its population, per cent and per cent live in the urban and rural areas, respectively, with average household sizes of 4.36 and 4.46 people respectively. Agriculture is Bangladesh s main source of livelihood and 25.7 per cent of the work force is involved in agriculture, forestry and fisheries. 15

37 Globalisation as a phenomenon not only involves the movement of goods and services worldwide but also has resulted in major flows of workers across countries in search of livelihoods that are not available in the originating countries. The United Nations (UN 2015) reported in 2015 that, of the 244 million international migrants worldwide, 104 million (43 per cent), were born in Asia. Europe was the birthplace of the second-largest number (62 million or 25 per cent), followed by Latin America and the Caribbean (37 million or 15 per cent) and Africa (34 million or 14 per cent). There are currently an estimated 97,036 Bangladeshi migrants in Italy (IOM 2015). Bangladesh has been recording migration data officially since 1976 for workers on contract employment in the Middle East. After independence in 1971, migration was accelerated for political reasons and spread beyond the Middle East to neighbouring countries and the EU. International migration has become a livelihood strategy, reflecting the unstable political and socioeconomic situation in Bangladesh. The pattern of migration from Bangladesh to other destinations is dominated by short-term and unskilled or semi-skilled workers, mostly to the Gulf region and Southeast Asia (Chowdhury 2011) although comprehensive and reliable migration data is lacking. A steady increase in remittances from abroad is clearly seen for all the countries for the period International migration has major socioeconomic consequences, one of which is the flow of remittances from the migrant workers in different countries to their villages of origin. This constitutes a complex phenomenon whose dynamics have national and global economic, social, legal and cultural implications. Thus, this study focuses on the factors affecting the flow and socioeconomic impacts of international remittances sent by male immigrants in Italy to their originating rural households in the Shariatpur district of Bangladesh. 16

38 2.2.2 Italy as a migrant-receiving and remittance-sending country European Union countries, especially Italy, Germany, France, Spain, Portugal and Greece, have been livelihood migration destinations for East European and Southeast Asian migrants since early 1970s (King et al. 2000; King 2001; Anthias & Lazaridis, 2000; Bonifazi et al. 2008). According to Zontini (2010), during the period , the stock of immigrants doubled; it rose from about 650,000 to almost 1,300,000. Other studies estimate that between 1986 and 2004, the foreign origin resident population increased from under 300,000 to 2.6 million (Chaloff 2006). The number of immigrants in Italy reached more than 4.8 million in 2009 (Cesareo 2009). With the decrease in work permits and an increase in family reunion opportunities (SOPEMI 2010), Italy has indeed shifted from being an international labour source to being a host country for migrant workers. Italy has been one of the major destinations for Bangladeshi migrants. For instance, Bangladeshi immigrants living in Italy numbered around 84,000 in 2009 (Blangiardo 2009). The number of documented Bangladeshi migrants is projected to reach 118,000 in 2015, 158,000 in 2020, and 232,000 in 2030 (Blangiardo 2009). In addition to regular immigrants, Italy also hosts a large number of irregular immigrants from developed and developing nations (King 2001). The number of irregular Bangladeshi emigrants was reported to have been 11,000 in 2009 (Blangiardo 2009). However, another source suggests that the number of irregular Bangladeshi emigrants was nearly 74,000 in During its first phase ( ) as an immigrant receiving country, Italy had no comprehensive legal framework for providing legal, social and economic status. In 1986, Italy introduced legislation (Law No. 943 of 1986, Regulations concerning the Placement and Treatment of Immigrant Extra-Community Workers and Against Clandestine Immigration ) which promoted a regularisation and integration process. It provided migrants equal rights and privileges to access jobs and social security benefits. However, this legislation was never implemented and the first operative legislation dates back to 1990 (Rusconi 2010). The first provision on migration, adopted in 1990, was the so-called Legge Martelli (Law no. 39/1990). The law included international protection provisions, and required visas for most of the originating countries. It also revised the deportation regulations for undocumented migrants, penalties for human traffickers and the integration policy for 17

39 economic migrants. This act also extended the right to seek asylum to migrants and abolished the previous geographic restrictions (Rusconi 2010). The fundamental constitutional amendment (2001) passed by the centre-left alliance delegated many policy decisions to the Italian regions, and prohibited the central government from intervening (Keating & Wilson, 2010). Legislation approved by centre-left governments (the Bassanini Laws, the Bindi reform) and centre-right governments (the La Loggia Law) have supported these amendments and provided regions with the executive power to follow up their recent activities (Rusconi, 2010). The centre-right alliance, which has since come to power, highlighted immigration issues for voters (Keating and Wilson, 2010). The new government enacted the new Bossi Fini Law (n. 189/2002), which amended the previous migration law. The aim of this new law was to reduce the number of undocumented migrants entering in the country to improve the existing control system at points of illegal entry and at the same time develop an integration system to link migration to national labour market trends. During the period , the short-lived centre-left alliance was not able to make any fundamental changes to immigration policy. Therefore, Italy adopted the so-called security package (Pacchetto Sicurezza), comprising Law no. 125 of 2008 and Law no. 94 of The new laws regularised migration in keeping with public security concerns. The new legal framework implies a wide range of reductions in immigrants rights and privileges. Thus, the combination of these two laws is in conflict with fundamental rights protected by the constitution (Rusconi 2010). Bangladeshi migration was largely irregular and was a response to frequent regularisation drives and immigration law reforms in Italy. Migration through irregular channels usually involved a combination of air and land routes, or air and sea routes, and sometimes all three. In the early 1980s, most migrants used air and land routes while, in the 1990s and the 2000s, land and sea routes emerged as the dominant channels for irregular Bangladeshi migration to Italy. 18

40 2.2.3 Bangladeshi migrants in Italy Table 2.1 shows the outflows of Bangladeshi migrants to Italy. This figure shows only official migrants. Table 2.1: Outflow of Bangladeshi migrants to Italy, Sources: BMET, 2013 The Bureau of Manpower, Employment and Training (BMET) is a government body in Bangladesh that has been recording and monitoring migration to Italy from Bangladesh since That migration was only for the plantation and agriculture sector and was a form of temporary and contract migration. According to the official data of BMET (2013), the total migration outflow reached 49,822 migrants from 2002 to Table 2.4 above shows the annual Bangladeshi migration to Italy. However, the central bank of Bangladesh, namely the Bangladesh Bank (BB) has been recording separately remittance inflows from Italy since The BB estimated the inflow of remittance from Italy to Bangladesh as US$ million during 2007 to Table 2.2 shows the annual distribution of these remittances. It is to be noted that the form of remittance inflows in Bangladesh not only depends on formal channels but also informal channels. 19

41 Table 2.2: Official inflows of remittances from Italy to Bangladesh Theories of migration Source: Bangladesh Bank, 2014 The movement of people, whether on a short-term or permanent basis, whether domestically or internationally, is termed migration. Although there are many motivations to migrate, in the economics literature, the main impetus is to seek a better life or better employment opportunities that the original environment cannot provide or sustain. Hence, the migrationdevelopment nexus is a powerful theme in migration theories although the interactions are fundamentally heterogeneous (de Haas 2007) Macro theories of migration The neoclassical theory The neoclassical migration model rests on spatial-economic equilibrium of labour supplies and wages reflecting geographical or regional disparities in labour supply and demand. The earliest models were conceptualised by Lewis (1954) and Ranis and Fei (1961) for ruralurban labour flows away from agricultural to urban industrialised employment opportunities, assuming a perfect market. Both push and pull factors are primarily (and wrongly) assumed to be economic and linked to employment opportunities or their scarcity. Todaro (1969) and Harris and Todaro (1970) refined this model to account for considerable urban unemployment, which they explained by the unrealised expectation of a real and positive income differentials between the risks and costs of rural-urban migration. Amplification for this phenomenon was provided by Peter and Piore s (1970) dual labour market model for developed countries with a primary sector (with good wages) and a secondary and separate market attracting temporary and unskilled migrants. Human capital theory (Becker 1962; 20

42 Sjaastad 1962, cited in de Haas 2007) can be seen as an enlargement of the Harris-Todaro model in that migrants have diverse demographic attributes reflecting, to a degree, investments in human capital; these determinants affect the returns from migration and migrant selectivity Mabogunje s (1970) Migration as a system model This model views rural-urban migration in emerging countries as a dynamic spatial process operating through a rural control subsystem which regulates labour outflows and an urban control subsystem affecting labour inflows moderated by a feedback system and the background socio-political environment. As all the subsystems are constantly changing, the whole migration system model is a dynamic process Historical-structural theory The historical-structural approach (Frank 1966; Wallerstein 1974) views the international migration of labour as the outcome of a world of nations with unequal political and economic strengths and access to resources. Thus, the underdeveloped nations are economically dependent on the richer, more developed and capitalistic economies whose growth creates disruptions and upheavals to their traditional societies. Domestic and international migration is the result which, in Wallerstein (1974) world systems theory, flows from the peripheral nations to the capitalist core countries. Somewhat linked to the macro theories of migration is Zelinsky (1971) mobility transition model in which the movement of people domestically and internationally is inherent in social change and development processes, especially the link between the stages of modernisation and demographic transition Micro theories of migration Micro migration explanations focus on the determinants of migration (or the decision to migrate) at the individual or household level. Among these models is Lee s (1966) push-pull framework built on Ravenstein s 19 th -century spatial laws of migration. As noted by de Haas (2007), the push to migrate is affected by rural population growth exceeding resource availability balanced by the pull of better economic conditions and employment opportunities in urban centres and industrialised countries. 21

43 The New Economics of Labour Migration (NELM) NELM theory, popularised in the 1980s and 1990s, rejected the neo-classical theory of migration as being too individualistic and simplistic to explain the complexities of the migration-development nexus. Seen in a broader societal context, the decision to migrate is more a household than an individual decision and is made to diversify and sustain the household s sources of livelihood in a strategy not primarily aimed at income maximisation but minimising and spreading risks. Such risks are not only associated with income variability in rural or agricultural communities but are also inherent in institutional imperfections such as weak credit, capital and insurance markets, domination by local elites, and situations in which access to income is difficult or costly. As a migration theory, NELM provides significant insights into the current study as it highlights issues involved in the decision for rural Bangladeshi household members to migrate internationally. To Hagen-Zanker (2008), NELM highlights that the decision to migrate reflects rational household behaviour in maximising joint (or household) income and status while minimising risks. The migrant worker does not solely concentrate on maximising his/her income as the migration decision is based on maximising household or family income. As noted by Stark (1991, the decision by the household to send a family member abroad is motivated by potentially higher absolute incomes and also relative incomes in comparison to some self-selected reference group. The latter goal, if achieved, enhances the household s status in the community. The third goal is risk minimisation as the community environment suffers from such institutional weaknesses as poor public support or protection infrastructure, and weak and imperfect capital, credit and insurance markets. Migration is then viewed as a rational strategy to overcome low and variable income and, if the household is unable to increase its income within its rural or agricultural context, the decision to diversify rural incomes into non-traditional alternatives such as migration is explored on a household basis. In addition, as stressed by Hagen-Zanker (2008), this establishes the link between migration and remittances, while to de Haas (2007), remittances represent one of the most fundamental stimuli for migrating. The formulation of this essential relationship has opened up research on the determinants and impacts of worker remittances behaviour at the household level, which is the focus of this study. 22

44 Migration can also be viewed as a viable, non-traditional, non-farm and alternative income generating source for households seeking to diversify and sustain livelihoods and to overcome social, economic and institutional development constraints in the place of origin (de Haas 2007 p. 53). Livelihoods basically refers to the utilisation of assets and capabilities to secure and sustain a living for an individual or household. Carney (1998), identifies five such assets, namely natural, social, human, physical and financial, the utilisation and management of which is shaped by the wider social, economic and political environments of the individual and households. De Haas (2007, p.30) reiterates that, while economic motives impel domestic and international migration, a framework that situates it within a broader paradigm shift in social theory towards approaches attempting to harmonise actor and structure approaches has more refined explanatory power as it focuses on the role of nation states, geographical proximity, institutions, social networks, and cultural and historical factors in creating new migration patterns (De Haas 2007, p. 57). Thus, migration, whether domestic or international, shortterm, permanent or circular, is a complex social phenomenon incorporating multiple attributes, characteristics and dimensions. One such dimension, directly associating migration with development, is the sending and receiving of migrant worker remittances. 2.4 Migration and remittances Figure 2.1 shows that, among the external resource flows into the developing countries, remittances sharply rose over the period and are expected to maintain their momentum in the future, far exceeding ODA (official development assistance) flows while being more stable than private debt and equity. Hence, migrant remittances have become a pivotal source of exogenous capital for emerging nations generated by their own labour resources. 23

45 Figure 2.1: Remittances flows are large, and growing Source: World Bank: Migration and Development Brief, April 11, 2014 Remittance costs (and the availability and range of financial instruments for migrant workers to remit money to their countries and households) are also a factor in the migrationdevelopment relationship not often discussed in the literature. For the World Bank, it is one of the four major concerns of its Global Remittances Agenda (World Bank 2011). The global average remittance costs fell to 8.4 per cent in 2014 compared to 9.3 per cent a year earlier. Orozco and Ellis (2014) provide useful dimensions of remittances in developing countries: a. A migrant economic link to the origin country is reflected by the transfer of remittances, demand for such services as telecommunications, consumer goods and travel, capital investments and charitable donations to the home communities. b. Remittances have to be transferred through a marketplace. Remittance amounts vary in size, frequency, length of time remittances are sent, whether they are sent via licensed channels (banks, remittance agents) or informally, and whether they involve a transfer cost influenced by financial infrastructure, regulatory framework, economies of scale of transfers, interdependence between home and host countries, and private sector competition. The significance of international labour migration in a globalised world has particular relevance for the emerging economies as the remittance flows from their migrant workers impact significantly on their economic development. From the policymaker s perspective, 24

46 then, there is much value that can be obtained by investigating the determinants of international migrant remittances so as to optimise their flows and their impact at the household, community and national levels. Two dimensions of this phenomenon are examined in this thesis: its micro (individual or household level) determinants and impacts and its macro (national level) determinants and impacts. This study of the determinants of worker remittances was inspired by the seminal work of Lucas and Stark (1985), which postulates the primary theoretical motives shaping such behaviour by migrant workers. Their empirical examination is enabled by the interpretation of the impact of the remittances on the receiving households (the micro level) and the specific country concerned (the macro level). In relation to this thesis, therefore, the motivations underpinning remittances and their impacts are different facets of the same phenomenon. Apart from the motivation to remit, the literature has also examined the sociodemographic attributes of the migrant workers and, for good measure, the heads of the sending households to ascertain, to the degree possible, whether specific combinations of socio-demographic attributes of both sender and recipient can be associated with the quantity, flow and frequency of such remittances. These perspectives will form the substance of the analysis of the literature review leading to the formulation of the conceptual model underpinning this thesis. The initial research interest in the determinants or motivations of worker remittances at the family or household level was initiated in Botswana by Lucas and Stark (1985). They postulated the existence of two extremes in the continuum of such motivations: pure altruism in the care and consideration for the welfare of household members and pure selfinterest. However, the reality of life and human behaviour is such that motivations to remit money to the originating households approximate a more complex economic behaviour termed enlightened self-interest or tempered altruism. The household jointly makes a decision to send one of its members to work elsewhere (domestically or overseas). This represents a self-enforcing contract to enable risk sharing or investment in potentially higher income streams between the household and the migrant. Lucas and Stark (1985) argue that the selfish motivations for remitting are: the desire to inherit, if necessary consolidated by higher remittances; the desire to invest in assets and their maintenance; and, finally, to enhance social capital or standing in the community by purchasing assets for the family. The tempered altruism or enlightened self-interest motive 25

47 provides greater amplification of the motivation to remit in the form of a self-enforcing agreement between household and the migrant encompassing investment and risk. In terms of risk, two aspects emerge: weaknesses in the institutional infrastructure of the community (access to capital and credit, lack of insurance against rural product price fluctuations and weather, poor public support facilities, etc.) compel rural and agricultural households to diversify into non-agricultural pursuits such as migration in order to spread the risks of sustaining incomes and livelihoods Amplifications to the motivation to remit According to Rapoport and Doquier (2005), the altruistic motivation of migrants to remit to their originating households may be tempered or moderated by more selfish motives such as (a) the incentive to provide services to maintain the migrant s assets and left-behind relatives, (b) the repayment of the investment (financed by informal loans at high interest rates) in migration costs and even previous investments in the migrant s education, and (c) the desire to level the volatility of household incomes, especially in rural and climaticallydisadvantaged areas through a co-insurance instrument. Thus, tempered or enlightened altruism combines several components: altruism, loan repayment, co-insurance and exchange of services in an implicit and to some degree enforceable contracts between the migrant and household members (see also Solimano 2003). Rapoport and Doquier (2005) also attribute a strategic motive (first hypothesised by Stark 1995 and Stark & Wang 2002) for remittances leading to migrant self-selection biased towards those who are more skilled or educated. This can serve to discourage the migration of household members who are relatively less skilled as it would negatively affect the average wage to be earned in a foreign labour market Modelling the motivation to remit The above hypothesis of the migrant s remittance behaviour is difficult to test empirically because of the non-exclusive nature or the relationships between the motivations. Funkhouser (1995) has formulated a behavioural model of altruism that predicts the following: a. migrants with higher income probabilities remit more b. households with lower incomes receive more c. the degree of proximity between household member and the migrant s intention to return will increase remittances. 26

48 Hagen-Zanker and Siegel (2007) have also hypothesised the motivations to remit as follows: a. Adverse conditions in the receiving household and remittances are positively correlated. b. Remittances should increase with migrant income and decrease with recipient income. c. In a three-generation scenario, the motivation to remit to care for his parents has an inherent demonstration effect to ensure that his own children will look after him in his declining years. d. In the tempered altruism motive, both migrant and the originating household benefit from the migration decision in an implicit and self-enforcing contract for the purposes of co-insurance (mutual benefit), loan repayment (human capital investment and migration costs) and exchange of services (child care, maintaining assets) Empirical evidence of the determinants and impact of remittances Since Lucas and Stark s (1985) seminal research, a vast body of research has been published. In analysing the empirical evidence of the determinants and impact of workers remittances, the major perspectives that are related to the current study are its macro and micro dimensions embracing countrywide and household settings. These are analysed below Macroeconomic determinants and impacts on developing countries In studying the macroeconomic impacts of the inflow of workers remittances from the destination to the originating country, diverse datasets have been analysed. Adams (2011) conducted a meta-analysis of the economic impacts of international remittances in 50 empirical studies using household survey data. The value of such analysis is clearly dependent on the quality of the data and their comparability. Thus, Adams (2011) stresses the methodological issues which limit the findings, including: (a) endogeneity, which makes the migration-remittances link problematical as many other household decisions are made simultaneously; (b) reverse causality in the sense that international remittances can reduce poverty while the incidence of poverty can affect remittances; (c) migrant selectivity or the difficulty of comparing more educated or skilled migrants with those less qualified in determining remittances sent home; and (d) the impact of unobservable characteristics which affect remittances such as the quality of risk aversion in different sending households. The 27

49 main impacts identified by Adams (2011) are discussed below, amplified by other related research Poverty and inequality impact The impact on income inequality is negative as international migration is expensive and favours the better-off households. Adams (2007) review of four household survey studies applying different methodologies found that all indicated a positive linkage between international remittances and the reduction of poverty level and depth although to varying degrees. Rapoport and Doquier (2005), however, found conflicting outcomes from four household surveys. Stark et al. s (1988) investigations in two Mexican villages found the level of income inequality was dependent on the trickle-down diffusion of migration-related information to other households such that when more households avail themselves of the opportunity to migrate, poverty levels will be reduced. However, a similar study by Milanovic (1987) investigating panel data over a decade reported the reverse, as did Adams (1989) in Egypt. Dissimilar findings are reported in studies by McKenzie and Rapoport (2004), Barham and Boucher (1998), Taylor and Wyatt (1996) and Taylor (1992). De Haas (2007) observes that the selectivity of foreign migration ensures that remittances will not benefit the poorest communities or the poorest countries. The major beneficiaries are the lower and middle-income countries like the Philippines, Morocco, Mexico and island nations in the Caribbean, the Pacific and the Atlantic. For the poor people and communities, remittances constitute an important share of income (De Haas 2007). The impacts of migrants remittances on income inequality have spatio-temporal aspects which vary for diverse migration types and periods of migration. Migration tends to be selective, but tends to be less so in subsequent migration movements because of the network and demonstration effect initiated by the diffusion of information to less knowledgeable communities and households. De Haas (2007) argues that international migration from poorer countries may coexist with increasing and decreasing spatial inequalities. Thus, de Haas (2014, p. 14) concludes that international remittances do not inevitably result in increasing or diminishing spatial inequalities of income and that there are ambiguities involved in attaching relative weights to distributional versus absolute income objectives. 28

50 Health and education impact The findings from a wide range of studies are mixed. Most studies show the impact to be positive on infant mortality and child health because of increased incomes and health knowledge. For education, the impact is controversial while school retention rates are shown to improve in some studies, others show declining school attendance rates for teenagers because of the absence of parents Investment and entrepreneurial activity Some studies in the literature have found that remittances are used for consumption goods, generating little knock-on economic activity; others have shown increased expenditure in education and housing. The findings are mixed insofar as inducing entrepreneurial activity is concerned with some attesting to enhanced entrepreneurial activity while others arrive at opposite results. Rapoport and Docquier (2005) report on several related studies. The clearest findings are in Woodruff and Zenteno s (2001) study on remittances and entrepreneurial activity in urban Mexico using three datasets. Their research indicates that remittances are an important source of investment in micro enterprises and they show that although the migrants in these cases were from rural areas, their entrepreneurial activities were located in the urban sector. De Haas cites many studies that indicate that migration, while having short-term negative impacts on productivity because of the loss of labour, eventually has a positive effect through the remittances sent to the originating household. It also has a temporal dimension consistent with a pattern of migration stages involving pioneers, early adopters and late adopters, implying that the full impact can only be measured in the second and third stages when most of the community has been involved in migration for decades. Thus, de Haas (2007) concludes that remittances impact differently on diverse sending countries and communities migrants sometimes invest in land and cattle, or in housing and land and even in urban areas where investment opportunities are more evident. Also, the apparent use of remittances on consumption expenditure may conceal the indirect impact on non-migrants such as feasts and funerals. De Haas (2007, pp ) thus states, Criticism of migrants unproductive or irrational expenditure behaviour also unveils an apparent inability to comprehend difficult social, economic, legal and 29

51 political conditions that often prevail in migrant-sending countries the distinction between consumption and investment is, therefore, often blurred and the outcome of evaluations on the relative merits of such expenditure partly depends on value judgements. Oroszco and Ellis (2014) have found increased rates of domestic investment and a reduction in growth volatility both as a direct result of remittances and indirectly through a diffusion of the influence of external and macroeconomic shocks on the economy. Remittances increase domestic savings and function as credit and insurance mechanisms, reflecting institutional weaknesses in the home country. However, when domestic financial systems are strong, remittances transmit less important effects on growth and may even be channelled into activities that do not promote growth. Yang (2011) states that many studies have found that international migrant remittances are seldom channelled into productive investments, while other research indicates investment in small businesses, agricultural land, agricultural ventures and schooling. On account of methodological concerns, Yang and Martinez (2005) and Yang (2008) undertook a natural experiment involving Filipino migrants in many different countries suffering foreign exchange crises because of the 1997 Asian Financial Crisis. Households benefiting from foreign exchange shocks used such resources productively for education or entrepreneurial activities rather than for consumption. Yang (2011) reviewed many studies which showed the insurance motive behind migrants remittances and their impact on risk-pooling arrangements. In such cases, remittances build up savings in good times and are drawn upon in difficult conditions, to increase labour supplies or diversify crops. Yang and Choi (2007) and Yang (2008) examined the impact of negative weather shocks on remittances: there is an inverse link between local negative weather shocks and remittances from migrants, while hurricane damage in the least developed nations leads to large inflows of remittances Labour supply and participation impact In general, the empirical studies reviewed show that international remittances tend to reduce the supply of household labour and workforce participation although the effect of gender cannot be discounted. This impact is explained by the likelihood that household members have higher wages, discouraging them from entering the labour force. 30

52 Economic growth impacts Relationship between economic growth (GDP) and the impacts of remittances is controversial, as research findings have been both positive and negative, partially due to difficulties in specifying the links between remittances and GDP, and difficulties in identifying the long-term impacts on human capital formation. Bettin, Presbitero and Spatafora (2014) analysed bilateral remittances from 103 Italian provinces to 107 developing countries over the period from the perspectives of structural characteristics, macroeconomic conditions and adverse shocks in both home and host countries using a gravity model. Their findings indicate that remittances are: negatively linked with the business cycles in recipient countries; increase in response to averse economic shocks such as natural disasters or declines in the terms of trade; are positively linked to economic conditions in the source Italian province; and are countercyclical with respect to the recipient country when both home and host countries suffer negative shocks. Analysis undertaken by the African Development Bank Group (Ncube & Brixiova 2013) also show that migrants remittances are affected by higher incomes in the host countries but are deterred by unstable economic conditions in the home countries, thus reflecting the investment motive. El-Sakka and McNabb (1999) examine the data for Egypt on remittances for and imports financed by remittances for the period The authors found that exchange and interest rate differentials significantly influence official remittance flows. However, imports funded by remittances have high income elasticity, suggesting that they are consumer or luxury goods or are purchased by higher income groups. Orozco and Ellis s (2014) review of the literature indicates that remittances and economic growth (GDP) are positively linked, and that remittances affect exchange and interest rates positively, although if higher import rates ensue, they may be reversed. Remittances are countercyclical in nature, with increasing remittances during downturns although if more severe in the host country, this counter cyclicality may be dissipated. Adams (2007) constructed a data set from IMF statistics on official remittance flows for 115 developing countries and derived the following findings. Based on a gravity model and groupings of countries based on per capita GDP, the cost of international migration is highlighted as a determinant of emigration linked to ability to afford migration costs and the distance between the originating and host countries. It was established that distance is 31

53 negatively related to international remittances. This implies that countries closer to most remittance-sending nations (like the Latin American countries and the United States) receive more international remittances. Developing nations with either lower or higher per capita incomes receive less international remittances than middle-income countries. This implies that low-income countries (like those in Sub-Saharan Africa) cannot afford the migration costs involved, while people in high-income countries have no incentive to seek employment overseas. It also implies that poverty levels in a country do not affect the quantum of remittances received. According to Hagen-Zanker and Siegel (2007), the primary determinants of migrants remittances are the stock of migrants and the state of the economies in the sending and receiving countries. However, Coulibaly s (2009) study of 14 Latin American and Caribbean countries for shows that the economic conditions of the host, the United States, significantly explained remittance dynamics but the home economic situation did not. A study by Schiopuand Siegfried (2006) focused on the determinants of remittances from a sample from 21 European countries to seven neighbouring countries. Income differences between home and host countries increased remittances, implying the altruism motive. Interest rate differentials did not have a significant impact, implying a weak motive to invest, remittances. As migrants skill levels increase, so do their remittances, and, lower remittance costs induce higher remittance flows provided the countries are sufficiently far apart Micro determinants of remittances and impacts In the above review of the macro determinants and impacts of international migrants remittances, three broad data sets were used: panel data covering several years, specific data sets drawn from panel data, and meta-analysis of topical reviews primarily based on household surveys. The spatial coverage has been global in nature although for this study the focus is on developing countries. What follows below is a selective survey of migrant household studies gathering primary data although, where relevant, reviews of household surveys conducted elsewhere will also be cited. The primary aim of the next part is to examine the socio-demographic attributes of the migrant worker and the head of household at the originating country and to identify the determinants and impacts on the receiving household from remittances. 32

54 2.5 Empirical literature on migration and remittance Remittances at the national level (Bangladesh) After its independence in 1971, Bangladesh tried to survive using overseas income from remittances. This involved political migration to Germany and economic migration to the Middle East. Therefore, migration and remittance literature has been growing since the early 1980s in Bangladesh. The oldest study of remittance in Bangladesh is Habibullah (1980). This micro-level study (sample size 554) which also involved a macro-level analysis, examined the issues of manpower export promotion, remittance maximisation, and the productive use of remittances. Habibullah suggests to the policy maker that taxes can be imposed on conditional terms depending upon the mode of use of the remittance amount and the setting up of specialised organisations to look after investment opportunities for the remittances. Stahl and Habib (1989) investigated the impact of migrant workers remittances on indigenous industries in Bangladesh. They used secondary time series data for remittance inflow in Bangladesh from the World Bank for the period 1976 to 1988, with a forty-seven sector wise input-output coefficient matrix. Their macro econometric analysis revealed that only small proportions of remittances are used straight way on investment goods and nonproductive uses. They found strong linkages between remittances and other aspects of the country s economy. Moreover, at this stage, Bangladesh was a new country, and migration and remittances were not key economic factors while in the current economy they are major considerations. A comprehensive study by Murshid et al. (2002) for the International Organisation for Migration (IOM), estimated the impact of remittances sent by migrants and presented socioeconomic profiles of three categories of households: current migrants, returnees and would-be migrants. A questionnaire was used to survey 191 households from eight rural villages in four major districts in Bangladesh. Their study provides descriptive statistics about the variations in socioeconomic characteristics within the region and between households. This socioeconomic impact can be analysed in depth statistically to derive an understanding of the most influential determinants of remittances. 33

55 Siddiqui (2004) investigated Bangladesh s experience of migrants remittance management in terms of migration outflow, age, employment status, educational level and gender distribution of migrants. Her study provides primary and secondary data through descriptive and exploratory methods. She found evidence of socioeconomic development in most of migrant households. She showed that Bangladeshi workers are predominantly men, the migrants are young and their ages are generally between 15 and 30 years old, when they first go abroad. Most are either illiterate or have a secondary level of education. Others (4.40 per cent) are expatriates, 31 per cent are skilled, 16 per cent are semi-skilled and 47 per cent are unskilled workers. These descriptive statistics may vary in different countries and regions. Mendola (2004) focused on the relationship between migration and remittance from a household perspective, and examined the impact of remittances on the left-behind household members. Mendola s analysis of household survey data from two rural areas of Bangladesh shows that international labour migration mostly relates to poverty. While it has some investment aspects, poverty might form a motivation to migrate as well as an obligation to do so. In addition, socio-demographic characteristics of migrants such as average age of international migrant 31.96, average education 6.15 years of schooling, and married per cent. Household characteristics show that mean household size is 8.81, mean education of household head 3.53 years schooling and mean household land ownership, 1.81 acres. This study provides different value of migrant characteristics than Siddiqui (2004) and also more information about the household assets and properties. Beaudouin (2006) analysed the impacts of migration and remittance on origin countries, and analyses the indirect and direct impacts of migration and remittance on migrant household income. Beaudouin also analysed data from a rural household survey in Bangladesh using a Three Stage Least Squares estimator to determine and compute the net impact of remittances on household earnings. The author found the determinants of remittances included the village norms and households sources of income. Theoretically, the determinants of remittances depend on various characteristics including types of migrants, types of households, and economic conditions in the origin and destination countries. Rahman (2007) analysed the basic socio-demographic characteristics of migrant households based on the primary survey data, first in 2001 and for a second time in 2006 in Gurail, Bangladesh. He found that people became more involved in the international migration process; as a result, there was an increased number of migrant households in the 2006 survey. 34

56 In 64 per cent of households, one member had been an immigrant for five to seven years. In 34 per cent of households, one member had been working overseas for seven to nine years. Rahman found that 86 per cent of migrants were married, 68 per cent were members of nuclear families, and 72 per cent of migrant households had no male members involved in local economic activities. Moreover, he showed that, for 94 per cent of the households studied, the aggregate effect of migration and remittance on migrant was positive. In addition, the author found that remittances were used for productive sectors, consumption, family maintenance, children s educational expenses and health care, and that portions of remittances were used for conspicuous consumption. These socioeconomic characteristics of migrants and households exemplify the migration pattern between two Asian countries. The migration and remittance patterns between European Economic Area and non-european Economic Area (EEA) countries could be different. Barua et al. (2007) examined the determinants of migrants remittances in Bangladesh. The time series panel data was analysed from remittances received from 10 main destinations by Bangladesh from 1993 to 2005 using a feasible Generalised Least Square method to account for heteroskedasticity across cross sections. The authors found that the earning differences between origin and destination locations were positively related to the inflow of remittances. In addition, their results found altruistic and investment motives for remittances. IOM (2009) undertook The Bangladesh Household Remittance Survey (BHRS) in order to obtain comprehensive understanding of remittances and their utilisation patterns at the household level. The survey examined the characteristics of remittance senders and receivers, remittance transfers to individual households, average transfer sizes, frequencies, transfer mechanisms, utilisation and the socioeconomic impacts on rural households. Their examination of household characteristics showed that 82.5 per cent of dwellings had tin roofs, 16.1 per cent had cement or concrete roofs and 1.2 per cent had thatched roofs. Their examination of water and sanitation revealed that 86.9 per cent sourced their drinking water from tube wells, 9.1 per cent from pipe water, 99.1 had a latrine and 73.7 per cent had improved toilet facilities. Their examination of overall assets showed that 94.9 per cent owned jewellery, 91.4 per cent owned chairs, 90.3 per cent owned tables, 89.9 per cent owned cell phones, 84.7 per cent owned watches or clocks, 78.8 per cent owned a cupboard 51.9 per cent owned televisions, 29.4 per cent owned bicycles, 20.5 per cent owned DVD players, 18.1 per cent owned refrigerators, and 13.6 per cent had deep/shallow tube wells. The above descriptive statistics indicate the uses of remittances at the household level all over 35

57 Bangladesh. The percentages will be different for specific rural households and different EEA countries. Furthermore, IOM (2009) found that 97.8 per cent of migrant households owned land. They found that 98.3 per cent of migrants were male, their mean age was 32, 60.1 per cent were married, and 50.4 per cent had a secondary level of education. Finally, this study showed that 89 per cent of migrant households experienced improvements in the consumption of food, 83 per cent showed improvements in educational opportunities, 59 per cent increased their ability to procure medical services, and overall there was a 20 per cent increase in household income. The reported outcomes vary in different studies (Siddiqui 2004; Mendola 2004; Rahman 2007). Thus, socioeconomic profiles will also vary. Raihan (2010) explored the socio-demographic profile of remittance receiving households, remittance senders and their impacts at the household level during the global financial crisis. Primary survey questionnaires were conducted on a random sample of 217 migrant households and two focus group discussions (FGDs) were conducted with three Upazillas in three Zillas in Bangladesh. The descriptive statistics show 41 per cent of household heads were illiterate, 34.4 per cent were domestic workers and 32.1 per cent worked in agriculture. In addition, 92.9 per cent of migrants were male, their average age was 32.6, 54.6 per cent were husbands of household heads, 67.3 per cent were married and 28.5 per cent had a junior secondary level of education. The destination countries were 69.2 per cent in the Middle East, 12.3 per cent in East Asia and 16.7 per cent in Europe. The Bangladesh Bank (2011), the Central Bank in Bangladesh conducted a study on the importance of remittance to the national economy, migrant profiles, the uses of remittances, and a prospective plan for the period until Their report was entitled The use of remittance in Bangladesh: Future direction, which in Bengali is: Bangladeshe Remittance ar Babohar: Vobishat Dik-nirdeshana. Secondary data was gathered from 47 scheduled banks and also from BMET on remittance policies, remittance flows, remittance receivers, and personal information on remittance senders. Primary data was obtained from 3010 remittance receiving household members through 56 branches of scheduled banks. The study (Bangladesh Bank 2011) explored migrant profiles and found 29 per cent had a bachelor/masters level of education and 41 per cent had paid their emigration costs by selling land. The study found that 75 per cent of households used remittance money for food consumption, 42 per cent for loan payments, 65 per cent for education, 57 per cent medical 36

58 and health, 40 per cent for marriage and 4 per cent for legal expenses. In addition, the households invested their remittance in the following ways: 42 per cent land/flats, 13 per cent saving certificates, 21 per cent fixed deposits, 1 per cent loans on interest, 12 per cent sole proprietor businesses, 2 per cent joint stock businesses, 18 per cent send household members abroad, 39 per cent built/renovated houses, 5 per cent purchased car/motor cycle, 8 per cent invested in the capital market and 3 per cent arranged household member employment opportunities for participant household members. Rahman and Belanger (2012) surveyed Bangladeshi male and female migrants and their households in the United Arab Emirates and rural Bangladesh. The study found inconsistent results in terms of remittances received and their uses. These differences are due to different gender perspectives. The head of the household wanted to enjoy special privileges as the recipient of remittances; married migrants tended to prefer remitting to their parents rather than in-laws or even spouses; female migrants tended to remit to sisters rather than brothers and fathers or mothers, whereas male migrants preferred remitting to their brothers rather than sisters and fathers or mothers; and more female migrants than male migrants tended to remit to male household members. A substantial proportion of male migrants used remittances for housing, whereas female migrants did not, and a small proportion of male migrants used remittances for loan repayments while the majority of female migrants used remittances for this purpose. This study was based on short-term temporary labour migration to the United Arab Emirates. The patterns and processes of migration to Italy are different, and thus the current research provided different outcomes. In the context of Bangladesh, all the studies mentioned above have more or less identified either the determinants of remittances or their impacts. Theoretically and empirically, the determinants of remittances are influenced by a wide range of variables including the characteristics of the individual migrant, household, and community. Remittances can be analysed statistically using several methods such as bivariate or multivariate analysis to find more accurate variables. At the same time, the impacts of remittances at the household and community level are influenced by migrant characteristics including age, income and education. Data on these variables can be analysed using a new generation statistical methods. 37

59 2.5.2 International Migration Table 2.3 presents an overview of this section. It also shows how the various empirical international migration studies of remittances are linked, and this contributes to meeting the chapter objectives. The background (or parent) theories are the Bangladeshi contextual setting, migration, international remittance theories and models, and trends in the analysis of rural households. As this study uses a multivariate approach, it is useful to examine other similar approaches to investigating remittances from foreign countries to home countries, as well as their socioeconomic impacts. 38

60 Table 2.3: Summary of the International Migration literature Source: Author developed for this study Davis and Stecklov (2002) examined the impact of rural migration upon the Mexican economy by employing data from the rural areas of Mexico. Their hypothesis was that people from rural areas of Mexico could migrate within Mexico to seek jobs in the agricultural and other sectors, as well as to the United States. The authors show that the determinants of migrants to the United States and non-agricultural Mexico are homogeneous. The level of 39

61 education of Mexican agricultural migrants was low, they were indigenous, and they lived in isolation from mainstream society. Garip (2006) formulates a model to capture the migration decisions of individuals and their remittance behaviour and finds that remittances and migration are interrelated actions. The author combines the determinants of migration and remittances within a single framework and also analyses two sets (1984 and 1994) of household census data for twenty-two villages in Thailand. The author provides a theoretical model of migration and remittance and uses Probit and Multinomial Logit estimation to analyse the data. The empirical results show that the determinants of migration and remittance are gender, level of education, employment status and migration experience. In addition, the author confirms that the altruism motive is an important motivator for remittance to their origin household and finds that the desire to secure an inheritance is not a significant motive. Finally, the author argues that migration and remittance can be explained under a new theoretical framework for labour migration (Garip, 2006). Poveda (2007) analysed the determinants of migration in Mexico. Poveda used a multinomial logistic (three-level) model to examine the personal and household characteristics of the migrants. The author shows that the determinants of migration to traditional destinations are related to certain personal and family characteristics, and to local conditions. The sociodemographic characteristics of the migrants influence the probability of migration. Young, single men are more likely to migrate to nearby places. Household variables play a major role in the migration process to border areas. The growth rate of the population is the main factor that drives migration. This led to permanent resettlement and migrants did not return to their place of origin. Ullah (2007) explored the migration of Bangladeshis to Malaysia and Hong Kong. The researcher surveyed primary data from Bangladeshi migrants in Malaysia and Hong Kong. He employed both descriptive and inferential statistical analysis tools such as chi-square, t- test and multiple regression analysis. He examined socio-demographic and socioeconomic characteristics of Bangladeshi migrants in both countries as well as patterns of remittances. The researcher compared the socio-demographic characteristics of the migrants between the two countries and shows that there are small variations in gender ratio, educational attainment and religion. However, the overall findings are that migrants are mainly male dominate, their average age 34 and most are unmarried. The average size of the migrant s original household 40

62 was six. Migrants to Hong Kong stayed there for three to six years while migrants to Malaysia stayed ten to twelve years. Regarding migrants legal status, the study found that most were undocumented. Most migrants were Muslim and most had a secondary school education. The researcher found two types of remittances: cash and in-kind. In-kind remittances include radios, watches, televisions, blankets, cassette players, gold, bags/briefcases, VCPs/VCRs, videos/dvds, cameras, mobile phones and computers. Cash remittances were used by the left-behind household members in the following areas: education, release of mortgaged land, homestead land purchase, becoming mortgagee of land, repayment of loan, furniture, sending relatives for pilgrimages, home construction and renovation, agricultural land purchase, gifts/donations to relatives, sending family member abroad, medical treatment, investment in business, savings/fixed deposits, and social ceremonies. The study found that remittances had a positive relationship with household housing development. Atamanov and Berg (2012) described the heterogeneous impact of international migration and remittances in the Kyrgyz Republic. The authors analysed data from the household survey on remittances and poverty conducted by the Kyrgyz Republic in early The survey divided its sample of 3997 households into three strata namely capital, urban and rural areas. The authors provide a modified theoretical model (Taylor et al. 2003) to examine the impact of migration and remittances, using regressions analysis. They found that larger household size tends to be associated with migration and permanent migration was associated with highly educated adults. Moreover, the authors found that labour migration had a significant and negative impact on agriculture. Rahman and Kabir (2012) explored the migration patterns and integration of Bangladeshi migrants to Italy as well as the impact of remittances on the origin country. The authors analysed the cluster data from the survey Bangladesh Household Remittance Survey (BHRS) 2009 conducted by International Organisation for Migration (IOM) in Bangladesh. The authors explain that they employed descriptive statistics and explanatory methods. The authors identified three distinctive types of migration: opportunistic, recruitment (formal and informal) and family reunion. They found that Bangladeshi migrants to Italy are mainly male, and females migrate mainly under family reunion schemes or as dependent spouses. Migration costs to Italy average approximately US$1000 per person. 41

63 Rahman and Kabir found that most migrants were married and that their spouses were living in the home country. Most migrants had a secondary level education although the percentage of graduates and postgraduates was higher than for other Asian and Middle Eastern migrants. Most (55 per cent) stayed in Italy for six to twenty-one years or more and others (45 per cent) stayed for one to six years. Average remittance frequencies were three to four times a year and the remittance amount was an estimated USD1900. Remittances were used by the household members for numerous purposes such as household living expenses (food and clothing mainly), religious festivals (Eid-ul-Fitre and Eid-ul-Adha), medical treatments, education, and repayment of loans. Sprenger (2013) examined the determinants of international migration and the impacts of migration on twenty-one EU and OECD member countries. The author analysed migration flows for the period 1998 to The author provides several theoretical models for each context and employs a statistically ordinary least square approach to analyse the data. The study found that some demographic characteristics of migrants were significantly correlated with migration. Being young was positively associated with the likelihood of migrating while education and gender showed negative relationships to migration flows. The degree of cultural difference between origin and destination countries had a variable effect on migration while having support networks of migrants in the host country had a strong positive effect on migration Micro studies of remittances Table 2.4 presents an overview of this section. It also shows how the various empirical micro studies of remittances are linked, and this contributes to meeting the chapter objectives. The background (or parent) theories are the Bangladeshi contextual setting, migration, international remittance theories and models, and trends in the analysis of rural households. As this study uses a multivariate approach, it is useful to examine other similar approaches to investigating remittances from foreign countries to home countries, as well as their socioeconomic impacts. 42

64 Table 2.4: Summary of the Micro studies of remittances literature Source: Author developed for this study 43

65 Adams (1989) examined household survey census data on 1000 households in three rural villages in Egypt in In 339 of these households, at least one member was abroad. The author used three econometric models to identify the determinants of remittances. The author reveals that age, marital status, employment status, size of land and gender in terms of number of household male members above 13 years old were the determinants of migration and remittances. Adams found that remittances were linked positively to upper level villagers. Socio-demographic characteristics were the determinants of migration, and remittances in Egypt could be compared with those in the current study. Hoddinott (1994) analysed rural household survey data from a single geographic location of 215 participants consisting of 126 (58.9 per cent) migrant households and 89 (41.1 per cent) non-migrant households in Karateng East sub-location between January and June 1988 in western Kenya. The author points out the advantage of using a single rural location to examine migration and remittance determinants and impact as the effects of community level factors such as distance to urban centres, ethnicity, amenities and land distribution are controlled for (Hoddinott 1994, p.465). He provides econometric models for migration and conducts multivariate analyses to identify explanatory variables such as household and migrant characteristics. The author found that household and migrant characteristics influence migration decisions and therefore remittance flows partly depend on household land ownership. The study shows that the determinants of remittances may vary from household to household, migrant to migrant, and location to location. Thus, a single study area was selected for this study, and a multivariate regression model was used to examine the determinants of remittances and their socioeconomic impacts. Itzigsohn (1995) explored the relationship between remittance and urban low-income households. The author surveys migrant households of four countries are namely Jamaica, Haiti, the Dominican Republic and Guatemala in The author analysed primary survey data on the characteristics of households, household members and migrants and employs multivariate analysis techniques like ordinary least square and logistic regression to determine the remittance determinants and their impacts. The study found that remittances not only play a vital role in macroeconomic factors but also make a large contribution in micro-economies in terms of urban middle and low income households. Moreover, the author found differences between the countries in terms of household remittance determinants and impacts. The present study took place in Bangladesh, where the determinants for remittances, 44

66 and the impacts the impacts of remittances, were different. Thus the current study could provide new outcomes for the same variables. Brown (1997) surveyed 982 households from Tonga and Western Samoa in Sydney to test the remittance decay hypothesis, which predicts that remittances will decrease over time. Brown used the Tobit model of regression analysis to show that there was no validity to remittance decay hypothesis in the remittance behaviour of these households. Moreover, the author found that motivational variables such as altruism, intention to return, inheritance, household land, business investment, level of education of household head and implicit coinsurance significantly influenced remittance to the household in the country of origin. The level of significance of each variable differed for each country of origin. The motivational factors in the present study could be compared with the motivational factors impacting migrants from the South Pacific. Cox et al. (1998) explored remittance receiving households in Peru. They used socioeconomic survey data from 5,109 households in 1985 and1986. The authors used an econometric model to identify remittance determinants and employed Tobit and Trobit regression estimation. They found that the strongest remittance determinant is altruism and that older migrants were more likely to remit than young migrants, and parents sent remittances more than other relatives while education of household members was not a significant determinant of remittances. These findings provide a basis for selecting the study area and for building hypotheses to explore the research questions in depth. Agrawal and Horowitz (2002) analysed data for the period of 1992 to 1993 from the Living Standard Measurement (LSMS) survey and the Household Income and Expenditure Survey (HIES). Both surveys were conducted by the Guyana Bureau of Statistics (GBS). The authors studied 270 households of 524 migrants. One hundred of the households were receiving no remittances. A theoretical remittance model and multivariate analysis were used in the study. The empirical results indicated a significant difference between the remittances received by households with single and multiple migrant members. Moreover, they found strong altruism in terms of remittance behaviour. This study used a multivariate regression model to test the altruism hypothesis. 45

67 Simati and Gibson (2001) explored Tuvaluan migrants remittance patterns and behaviour in New Zealand. The authors conducted a random-sample survey of 100 participants from West Auckland, Pukekohe, Hamilton and Wellington in To analyse the survey data, the authors used the econometric model of Brown (1997), even though all data do not similar and lacking of availability of few variables, especially Tuvalu co-family land ownership complicating pattern. The empirical findings were consistent with findings for Samoan and Tongan families in Sydney (Brown 1997, 1998) and also do not support the remittance decay hypothesis. Moreover, Simati and Gibson found that the duration of migration had no statistical relationship with remittances to the country of origin. This duration of migration can be analysed in-depth statistically to derive an understanding of the most influential determinants of remittances. De la Briere et al. (2002) surveyed 20 randomly selected Dominican Sierra communities. In total 400 households participated in the study. The authors used an econometric model of remittance determinants and employ multivariate statistical analysis to identify the micro determinants of household remittances. The empirical results provide data on the heterogeneous characteristics of migrants and households such as destination, household composition and gender. These characteristics are affecting the remittances sent to household members in Dominican Sierra. In addition, the econometric results show that, for female migrants, the main motive for remittance is insurance. In contrast, the motives of investment and inheritance are gender neutral. The motivational factors of insurance, investment and inheritance had different impacts in Dominican Sierra communities and rural households in Bangladesh. Black et al. (2003) examined the relationships between migration, remittance, development and poverty alleviation and the lifestyle of migrants when they return to Ghana, their country of origin. The authors conducted a survey questionnaire of a representative sample of 152 participants selected by snowball sampling. Descriptive statistical methods were employed to describe the socioeconomic characteristics and reintegration processes in the home country after migration. The authors show that 33 per cent sent remittances occasionally while 34 per cent sent savings back to their households in Ghana. Returnees were likely to invest in personal services, retailing, light industries, construction, restaurant, businesses and agriculture. The results of the present study could be compared with Black et al. s (2003) results. 46

68 Blue (2004) focused the characteristics of migrants who had been sending remittances to household members in Cuba and the contribution these remittances made to changes of economic conditions in the origin country. The research team conducted interviews of 334 household heads about the migrant member of the household in terms of socio-demographic characteristics including household all members socioeconomic information. According to the descriptive analysis, the relationships of the migrants to the household heads were strong remittance determinants and these relationships were mainly between parents, children and siblings. In addition, Blue found that gender is also a determinant of remittance as female migrants are likely to remit more than male migrants. Therefore, the gender of the head of the household is one of the key variables in the present study. De Haas (2006) explored the economic development impact of migration and remittance in the origin country of Morocco. The author conducted a socioeconomic household survey of a total 507 households comprising 237 international migrants and internal and non-migrants in six villages. The author analysed the data using descriptive statistics and a Gini index. The researcher found that international migration and remittances positively influenced the home country s economic development and improved household standards of living and increased the freedom of dominant cultural groups. In addition, the study shows that the international remittances were likely to be invested by the household in the rural economy housing, agriculture and small industries. Household standards of living and investment in housing and agriculture can therefore be analysed in-depth for the study area in Bangladesh. Markova and Reilly (2007) examined the factors influencing Bulgarian migrant remittances from Spain. The author conducted two surveys of 198 Bulgarian migrants in Spain in and The authors used an econometric model to analyse micro-economic determinants of remittances and employed a Tobit estimation. The study found a strong relationship between migrant legal status and remittances as irregular migrants remittances are substantially less than those of documented migrants. In addition, the authors highlight gender, age and marital status as having an inevitable impact on migrant remittances. The legal status of the Bangladeshi migrants in destination countries is one of the key variables that has not previously been studied. 47

69 Azhar (2008) described the socio-cultural and economic impacts of international remittances on left-behind rural household members in Pakistan. The study used a survey questionnaire in rural villages on a total of 300 left-behind households. The researcher analysed the survey data using univariate analysis in the form of frequency distribution; bivariate analysis in the form of chi-square tests, gamma tests and correlations; and multivariate analysis in the form of multiple linear regression. In addition, the author constructed index variables to determine the relationship between socio-demographic characteristics of household members as well as socioeconomic impacts. The author found that originating households standards of living and economic conditions substantially improved. Migrants age, education, pre-migration conditions, relationship to household head were significantly correlated to amount of remittance and socioeconomic impact. The remittances were mostly used for household consumption and durable purchases. This study can be used to provide guidance in the selection of the study area, sample selection and data analysis procedures. Kelly and Solomon (2009) examined the influence of religion on the remittance motives of new legal migrants in the United States. The authors analysed secondary data from The Pilot for the New Immigrant Survey (NIS-P), a panel survey of a random sample of 985 their religious categories were: Catholic, Protestant, Muslim, Jewish, Buddhist, Hindu, Orthodox, other religion and no religion. They found that religion was an important factor in remittances to the country of origin. Variations occurred between religions as well as within different forms of one religion. They compared the remittance behaviour of Catholics, Protestants and no religion and their statistical result showed that Catholics sent more remittances than those in the no religion category, while Protestants were more likely to remit than Catholics. The influence of religion on the remittance motives can be analysed in-depth statistically to derive an understanding of the most influential determinants of remittances and their impacts. Sackey (2010) examined the microeconomic determinants of remittance inflows to household members in the origin country of Ghana. The researcher used a self-administrated survey questionnaire of 75 participants selected using random sampling techniques. Sackey obtained only 60 completed questionnaires to use for analysis of the data due to insufficient information. The researcher provides a theoretical and econometric model of remittance determinants and ran logit estimation for statistical analysis of the survey data. The research shows that the exchange and insurance motive significantly influenced remittances to the originating households. In addition, the study found that the period of migration and household size at the origin country were statistically significant influences. The insurance 48

70 motive, the duration of migration and household size are variables that could be considered in order to better understand remittance behaviour. Mishra (2011) examined the determinants of the remittance behaviour of Nepalese migrants in various destination countries. The researcher analysed the data of the Chitwan Valley Family Study (CVFS) which surveyed 151 household and individual participants from 1976 to The researcher employed descriptive and probit estimates to assess the significance of the influence of explanatory variables on remittances. The results show that inheritance and intention to return home are highly significant. In addition, age of migrant was associated with a decline in remittances, males were likely to remit than females, household size affected the remittance flow and social norms also influenced remittances. The socioeconomic features of migrants and households characterise the remittance motives in two neighbouring Asian developing countries. The migration patterns and remittance motives in EEA and non-eea countries could be different, and are therefore worthy of investigation. McDonald and Valenzuela (2012) examined household remittance determinants in the Philippines. The authors analysed household income and expenditure survey data from overseas Filipinos. The author proposed a theoretical model of household remittance behaviour and employed ordinary least square and probit estimates to analyse the survey data. The empirical results show that remittance strongly relates to the origin and destination countries and the most influential factor was the altruism motive. In addition, gender and employment status also influenced remittance but these depended on their overseas earning opportunities and capacities. The study demonstrates that gender and employment status can influence remittances, together with the characteristics of the originating households, such as household income. Busetta et al. (2013) explored individual characteristics of migrants in Italy to determine the factors that influence remittances to household members in the home country. The authors analysed data from the Istat 2009 Income and living conditions survey, which comprised migrant socio-demographic characteristics, attitudes, remittance behaviour, plans to return to, home country and the composition of households in the home and residence countries. The study examined 6000 households of 120 nationalities including Bangladesh. The authors used the two-part model (Agrawal and Horowitz, 2002) and ran logit estimates to analyse the data set. The empirical results show that duration of migration and household composition are key determinants of remittances. Household composition in the destination country and duration 49

71 of migration are key determinants of remittances. These variables can therefore be added to those related to the originating country. 2.6 The nexus between socio-demographic characteristics and remittances and impacts Remittance sender The NELM theory postulates that remittance size is dependent on the socioeconomic and socio-demographic characteristics of individual migrants (Lucas and Stark, 1985). It also postulates that these characteristics play a vital role in terms of remittance flow at the origin household and community level. There is a relationship between a migrant s age and the duration of migration at the destination (Menjivar et al. 1998; Rodriguez 1996). Often, the incomes of migrants increase over time, and as a result the money they have available to send in remittances also increases. Larger remittances are sent by younger migrants (de la Briere et al. 1997). Lerch et al. (2006) found a relationship between the age of migrant and the likelihood to send the money to the destination and the length of stay additionally correlates with the age of migrants. They also found a linkage between age and family relationships which influences remittance to the household in the country of origin. Several micro level studies in Bangladesh found that most of the migrants are young especially when they first migrate (Siddiqui & Abrar, 2003; Siddiqui 2004; Afsar et al. 2002; Murshid et al. 2002). McDonald and Valenzuela (2012) found that the higher the educational attainment of the migrant, the higher the propensity to send remittances. Rapoport and Docquier (2005) explain that the educational level of the migrant does not have a significant influence on the altruistic and exchange motives. The investment motive views remittances as a repayment made by the migrant (Garip 2012), in terms of loans to cover costs of education or migration at the destination (Cox 1987; Poirine 1997). Therefore, remittances increase with migrants education. This supports the findings of several empirical studies (Ahlburg & Brown 1998; Cai 2003; Cox et al. 1998; Hoddinott 1994; Lillard & Willis 1997; Lucas & Stark 1985; Regmi & Tisdell 2002).There is also evidence that educational level and remittance behaviour varies depending on the migrant s country of origin (Hagen-Zanker & Siegel 2008). 50

72 Several studies show that the marital status of the migrant and the residency patterns of family members are significant factors influencing remittance motivation (Johnson & Whitelaw 1974; Menjivar et al. 1998; Vanwey 2004; Casale & Posel 2006; Luke 2010; Alba & Sugui 2009). According to Sahu and Das (2009), single migrants and married heads living alone at the destination are likely to remit more than married heads living with their spouses and children. However, Collier et al. (2011) found that migrants marital status does not influence remittance motivation. Furthermore, Sorenson (2004a, 2004b, 2005) and Atekmangoh (2011) reveal that marital status is a key determinant of remittance behaviour and remittances also vary with gender when migrant changes their marital status after migration. According to Siddiqui and Abrar (2003), a higher proportion of married than unmarried migrants send remittances to their rural household members in Bangladesh. Legal status and remittances are closely linked (Holst et al. 2008, 2010, 2011; Bettin & Lucchetti 2012). Amuedo-Dorantes and Pozo (2006) have found undocumented migrants are more likely to send monies than documented migrants. Markova and Reilly (2007) have found a strong positive relationship between remittances and the legal status of migrants. Similar findings were also reported by other studies (Collier et al. 2011), which indicate that the legal status of migrants affects their earnings and therefore remittance flows fluctuate according to legal status. At the micro-economic level studies of remittance motives show different effects of the length of stay of migrants. Some studies have found negative impacts (Holst & Schrooten 2006; Vargas-Silva 2006; Fairchild & Simpson 2004; Menjivar et al. 1998), while others find that the length of stay has no significant effect (Merkle & Zimmermann 1992). The evidence of empirical studies indicates that the impact of the period of stay in the host country on remittance motivation varies (Banerjee 1984). In addition, the remittance flow is higher in the beginning stages of migration and decreases with time (Lucas & Stark 1985; Banarjee 1984; Vete 1995). Likewise, Collier et al. (2011) have found that migrants who have intention to return their origin country are more likely to remit. Mostly, these remittances are used for investment purposes and the flow of remittance increases over the period of migration. Many empirical studies have found that the number of trips to the original household influences remittance behaviour (Lerch et al. 2006; Garip 2012; Roberts & Morris 2003). During the visit to their original household, migrants bring gifts for their household and family members as well as their community networks (Goldring 1998). Therefore, 51

73 remittances have direct and indirect impacts in the home country in cash and in kind. Migrants who make frequent visits to their country of origin not only sustain a community connection, but also have an economic impact (Kemper 1991). However, Grabel (2008) have found that a huge percentage of remittances are carried by hand by migrants during their trips home (Garip2012; Roberts & Morris 2003). In contrast, Holst and Schrooten (2006) have found that personal trips to the origin country do not increase the flow of remittances. Furthermore, migrants are not a homogenous group in regard to their remittance motivations Household head The head of household s age is one important determinant which plays a vital role in remittance behaviour and the influence of age varies from country to country (DeVoretz & Vadean 2008). The nexus between age of the household head and gender behaviour influences the remittance motivation. For example, male household heads are less likely to receive remittances (McDonald & Valenzuela 2012). However, Germenji et al. (2001) show that older household heads receive more remittances than younger household heads. This indicates that adult children care for their old parents and grandparents. Moreover, Walewski (2009) finds the reverse: that younger household heads tend to receive remittances more and that there is a strong correlation between household head age and remittance flow. Male heads of households use remittances differently from female heads, which affects households resource allocation (Pfeiffer et al. 2008). When migrants leave behind their wives in the country of origin, the women in the origin household have increased responsibilities including the household budget, spending the remittance income and educating children. Numerous empirical studies find that the marital status of the head of the household is a major influence on remittance motivation. For instance, Pfau and Giang (2008) have found that the female household heads are likely to receive higher remittances than male household heads when both are married. In addition, the authors have found that widowed female household heads experience lower frequencies of remittances. McDonald and Valenzuela (2012) imply that higher educational attainment on the part of the head of household probably means better earning opportunities and economic resources in the origin household and so less dependence on remittances. Therefore, the level of education of the head of household is not strongly correlated to the remittance flow and this evidence would indicate an altruistic motive for remittances. 52

74 The households head s occupation and employment status are linked to migration decisions and remittance motivation. Quartey (2006) have found that the connection between the occupation of the head of household head and socioeconomic impacts varies depending on the home country. Quartey found that the employment in the government sector and private sector are associated with different types of socioeconomic development Characteristics of household There have not been many studies on the influence of religion on migration and remittances but moral economics shows the nexus between remittance and religion (Stevanovic (2012). Cadge and Ecklund (2006, 2007) found a link between religion and pro-social characteristics on the one hand, and remittance sender and originating household on the other. Ellison (1991) have found a strong positive relationship between religious practice by the migrant and remittances. Similarly, Kelly and Solomon (2009) have found a link between religion and altruistic motivation to remit. Household size is one of the factors influencing migration and remittance behaviour. The empirical study Atamanov and Berg (2012) shows larger household size tends to make migration more likely. Sackey (2010) found that household size is significantly correlated with the exchange and insurance motives to remit. Mishra (2011) also reports that household size affects remittance inflows in Nepal. Ullah (2007) shows that, in Bangladesh, the average size of households with migrants is six when the migrant member is included Summary of the extant literature Figure 2.2 Concept map of this section Source: Author developed for this study 53

75 2.7.1 Research gap The literature review revealed there is adequate information on the historical background and fundamental concepts associated with the parent disciplines of international migration and remittances, including remittance determinants. There was literature on the micro-economic impact of such remittances at the household level in regions of countries other than Bangladesh. Hence, the extant general literature provides a good foundation for the particular focus of this study. However, a review of the pertinent literature has identified four gaps in our knowledge regarding nexus between socio-demographic characteristics and impacts in international migration and remittances. There is limited literature on the international migration and remittance at both the macro and micro levels in Bangladesh. The available literature at the national level on socio-demographic characteristics of international migrants does not apply to any particular region or country (Murshid et al. 2002; Siddiqui 2004; Bangladesh Bank 2011), therefore, the first research gap is: 1. A gap in the understanding of the socio-demographic characteristics of non-eea rural household members who migrate to the EEA, specifically Bangladeshi in Italy. A review of micro-economic level literature on international remittances revealed that there are some important factors influencing remittances sent to the home country. These include individual migrant characteristics such as age, marital status, education, year of migration, legal status and number of visits to home country. Household members attributes including household head age, gender, educational attainments, marital status, employment status and relation to migrant as well as household size and religion, also are also influences. However, owing to a lack of national level literature, a search gap exists. Therefore, the second research gap is: 2. A gap in the understanding of what factors influence the amount of remittance sent by Bangladeshi migrants to their household members in non-eea, developing countries. The review of the literature of international migration, and on the impact of remittances at the macro and micro levels revealed that remittances have macro-level impacts as well as microlevel impacts on the home country economy. However, there is a gap in the literature relating to specific EU countries such as Italy and the least developed countries like Bangladesh. Hence, the third research gap is: 54

76 3. A gap in the understanding of the socioeconomic impacts of such international remittances on the micro-economies of non-eea, developing country rural households. The review of the national- and international-level literature on migration and remittances, and of the literature on the macro- and micro-level impacts of remittances, revealed a gap in the understanding of significant relationships between the determinants of remittances and their socioeconomic impacts. Thus, the final research gap is: 4. A gap in the understanding of the relationship between the socio-demographic characteristics of remittance receiving household members, including migrant members, the determinants of remittance flow and the socioeconomic impacts of the remittances on non- EEA, developing country rural households Research problem As discussed above, the literature review identified four specific gaps in our knowledge in connection with the micro-determinants of international remittances and their impacts in rural Bangladesh. The research problem is therefore that not enough is known about the determinants of remittances, including household composition, social conditions, economic conditions and impact on the rural households and also the originating community. Thus, this study seeks to fill these gaps in the literature. As discussed in Sections 2.4 and 2.6.1, the particular context for the research into remittances from an EEA country to non- EEA developing country rural households are Italy and Bangladesh respectively. Accordingly, the research problem is stated as: What are the key determinants of remittance inflows from migrant workers in Italy and their socioeconomic impacts on the originating households in rural Bangladesh? Research questions According to Punch (2005, p. 34), Empirical research is driven by research questions. Research questions define an investigation, set boundaries, provide direction, and act as a frame of reference for assessing the research work (O Leary 2005, p. 33). To address this research problem, and based on the before-mentioned research gaps, four research questions are pursued in this thesis research. 55

77 The literature review shows that the socio-demographic characteristics of international migrants, and of household members, are not available, particularly in regard to migrants from less-developed Asian countries in European nations. This problem underpins the first research question: 1. What are the socio-demographic profiles of the foreign migrant workers and of the heads of the Bangladeshi households receiving remittances from their family members working in Italy? Previous studies show that some important factors such as the characteristics of migrants, household members and household composition, influence the remittances sent to the home country. However, owing to a lack of national level literature, a research gap exists. Therefore, the second research question is: 2. What are the determinants of the remittances sent by household members working in Italy to their households in Bangladesh? The review of the literature on national- and international-level migration and remittances, and of the literature on the macro- and micro-level impacts of remittances, revealed that remittances have micro-level impacts on the home country economy. However, there is a gap in the literature relating to specific EU countries such as Italy, and the least-developed countries like Bangladesh. Hence, the third research question is: 3. What is the socioeconomic impact of such migrant worker remittances on the originating households in Bangladesh? The review of the literature on international migration, and on the impact of remittances at the macro and micro levels, revealed a gap in the understanding about significant relationships between the determinants of remittances and their socioeconomic impacts. Thus, the final research question is: 4. What are the links between the socio-demographic profiles of the Bangladeshi households receiving remittances from Italy with the determinants and socioeconomic impact of such remittances? 56

78 2.7.4 Research hypotheses Quantitative research is usually based on a positivist paradigm and generally involves hypothesis testing (Sobh & Perry 2006; Creswell & Clark 2007). Some critics of the positivist paradigm argue that positivism interprets the participants responses using statistical formulas which may lack sensitivity to the research context (Neuman 2006). Key insights into the new economics of migration are that households send members overseas not only to increase absolute income, but also to improve their relative income (Stark & Bloom 1985). In spite of the important socioeconomic role of international remittances for many households and communities in developing nations, the literature on migrants remittances includes only two distinct motivations: risk sharing, and altruism (Stark 1991; Stark & Levhari s 1982). Agarwal and Horowitz (2002) argued that, if the risk-sharing hypothesis holds, migrants remittances should act as insurance against low wages and unemployment. The authors concluded that their findings indicate the existence of an altruistic motive to remit. In addition to risk sharing or altruism, remittances might also facilitate exchanges of favours in a household. Therefore, one might well expect that the characteristics of the migrant s originating household, such as the number of dependants and family structure, influence the flow of remittances received. Moreover, Hoddinott (1994) argued that remittances might have an inheritance motive. In light of the above, the flow of remittances received may depend on the investment opportunities in the countries where the migrant works. Remittances from international migrants are sensitive to interest rate differentials between origin and destination countries (Foster 1995; Durand, Kandel et al 1996). While investment motives might be a significant factor, Brown (1997) argued that remittances also reflect migrants kinship relations in their originating countries, and how assimilated they are in their destination communities. Following the above discussion and literature review, this study argues that, if remittances are correlated to altruistic behaviour, the investment motive and kinship relations, the volume remitted by a migrant should not be significantly correlated with the size of the household, land, income and land ownership, or with risks to the migrant s income at the destination. Table 2.5 summarises the hypotheses for this study: 57

79 Table 2.5: Summary of Expected Direction of Regression Coefficients by Hypotheses 2.8 Theoretical framework The early neo-classical literature on migration did not explore the role of foreign remittances on households as it focused more on the migration decision of the individual or on migrant income maximisation strategies (Sjaastad 1962; Todaro 1969; Harris & Todaro 1970). Others perceive the decision to emigrate more as a household decision-making strategy to minimise risk and diversify livelihoods (Stark & Taylor 1986).When village markets are imperfect and households relocate their members to spatially diverse markets, they do so to spread the risks to household income. According to the New Economics of Labour Migration (NELM), decisions about whether one member of a household should emigrate are made jointly by households (Stark & Bloom 1985). Therefore, the issue of remittances and their determinants has become a key consideration in a number of theoretical and empirical studies. A distinguishing attribute of 58

80 the NELM model is its simultaneous consideration of migration determinants, remittance behaviour and impacts, making NELM models relatively demanding in estimation methods and data requirements. Another key insight of NELM is that households allocate members to improve not only absolute but relative incomes and alleviate their deprivation against a reference cluster such as the village community (Stark & Bloom 1985; Stark et al. 1986; Stark & Taylor 1989; Stark 1991). Micro-level studies of remittance determinants and their impacts on the household at the rural level are based on either household primary surveys (Gubert 2002) or surveys of the individual migrants at the originating or destination country (Amuedo-Dorantes & Pozo 2006; Holst & Schrooten 2006). Against such a theoretical background, a broad stream of studies has identified the determinants of remittances and their socioeconomic impact on countries, communities, and households that have sent workers abroad (Borja 2012). Two broad perspectives have been adopted: macro determinants and the impact of foreign remittances on originating countries and their micro determinants and consequences. According to BMET (2013), there were 64,834 remittance-receiving households in Shariatpur District, and 49,882 remittance-sending household members in Italy (BMET 2013) although this figure might have been around 84,000 in 2009 (Blangiardo 2009). The districts represented most strongly in Italy were Dhaka, Faridpur, Comilla and Noakhali (Knights 1996). Shariatpur presently belongs in the Dhaka division and the greater Faridpur district. Administratively, the Shariatpur district is divided into seven upazilas (Thana) and many people from the Naria upazila have migrated to Italy. However, on account of a lack of official data, there is no specific remittance figure at the district level. Usually, a high incidence of emigration from an area suggests it is an area that receives a large number of remittances. The focus in this study will be on the latter as the study area is the Shariatpur District of Bangladesh, which has traditionally sent migrant workers to Italy. Figure 2.3 shows three characteristics, namely the socio-demographic characteristics of the migrant, the characteristics of the head of the household and the household composition in relation to the twenty-three independent variables for the multiple regression models where remittance is the dependent variable. Figure 2.3 also reflects how the determinants of remittance, socioeconomic impact and motivational factors are interlinked with sociodemographic characteristics. 59

81 Figure 2.3: Conceptual framework Source: Developed for this study The consequences of working abroad on the sending household are economic (Azam & Gubert 2006) and social (Rahman 2007) and are discernible in household income, the smoothing of consumption patterns, investment in new business enterprises, accessibility to financial systems and accumulation of capital in land, machines, technology, building construction and renovation (Yang 2008; Adams 2006; Taylor & Wyatt 1996; Lucas 1987; Woodruff & Zenteno 2001). Remittances can reduce unexpected health hazards, encourage long-term investment in education (Ratha et al. 2011), conserve cash resources by avoiding the sale of assets during calamities (Mohapatra et al. 2009), or even serve as co-insurers for the originating family (Ponsot & Obegi 2010). 60

82 Remittances are a beneficent transfer to the household members and are motivated by kindness and emotional attachment (Lucas & Stark, 1985). Such altruism affects households economic situations, strength of household kinships and contractual household arrangements. The set of variables that can affect the influence of altruism on remittances include marital status, household income, gender, age, education level, the number of dependents and the length of the worker s stay abroad (Barua et al. 2007). Self-interest relates to the motives of exchange, risk sharing, inheritance seeking, investment and loan repayments. Migrants often send remittances for the purchase of goods and services to improve the welfare of those remain-behind, or to pay for the maintenance of property belonging to the migrant or household members (Cox 1987; Lucas & Stark 1985). This motive can affect remittance size (Rapoport & Docquier 2005). Predominantly remittances are used for maintaining the welfare of very old or very young family members (Cox et al. 1998). Migration not only maximises income and minimises economic risk but also diversifies household income sources. The altruism model predicts that remittances will slowly decrease over time and distance (the remittance decay hypothesis) with the weakening of the altruistic motive, while the insurance model hypothesises that remittance size depends on the degree of loyalty to the family the migrant has remain-behind (Rapoport & Docquier 2005). NELM claims that migration becomes a household strategy of risk diversification because of market failure at the point of origin (Wouterse 2006). During natural calamities, affected households receive more remittances than during normal seasons (Lucas & Stark 1985) and, similarly, negative income shocks significantly increase foreign remittances (Pleitez-Chavez 2004). Inheritance, seen as an enforcement mechanism, increases remittances and household income and decreases risk. The inflow of remittances increases with: (a) household properties and income, (b) the member inheriting, and (c) the migrant s assets and income (de la Baiere et al cited in Rapoport & Docquier 2005). The loan repayment hypothesis suggests that remittances are sent to repay funds provided for the migrant s education or for financing the migration process (Brown 1997; Cox et al. 1998, Glytsos 1988, Poirine 1997). However, studies have yielded inconsistent evidence about workers remittances being sent as repayments for migration costs. Remittances can be channelled into investments in the sending community in the form of property, land, cattle and other assets (Durand et al. 1996; Blue 2004), although such remittances also reflect family ties. When the altruism motive is strong and the destination 61

83 countries incomes are higher, more remittance will be sent (Brown 1997; Blue 2004; Ahlburg & Brown, 1998; Vadean 2007; Dakila & Claveria 2007; Lueth & Ruiz-Arranz 2007), thus creating direct and indirect impacts on the household and community, respectively. In reality, however, remittances are not motivated by just one factor (Brown 2011), but by a mixed range which varies over time and place (Rapoport & Docquier 2005). Although motivations differ among different individuals, diverse motives to send remittances may coexist in one individual (Lucas & Stark 1985; Cox et al. 1998; Feinerman & Seiler 2002). Remittances sent home by migrants form a significant part of the household s strategy to maintain its economic viability and meet its social obligations. However, they must offset migration costs including those for the agent, transportation, documentation, food and clothing. Remittance flows perpetuate social networks across space (Caces et al. 1985) and strengthen and influence future migration decision strategies (Fawcett & Arnold 1987). Thus, the unit of analysis is the household (the sampling unit) in in Naria Upazila, Shariatpur District in Bangladesh. In this study, variables are divided into three categories, namely background (sociodemographic), intervening independent and dependent variables. Background variables includes three major categories such as migrant (age, marital status, education, year of migration, number of visit), household head (age, gender, marital status, education, religion, employment status, kinship) and household composition (household size, annual income, land ownership title, living expenses, welfare expenses, investment in house development, investment in financial sector, investment in business and loan repayment). All these background variables are playing important role in determining the determination of remittance and the impact of remain-behind households. The dependent variables are also directly affected by the background variables and so these intangible variables exist intrinsically and are amongst the determinants of remittances and socioeconomic impacts. Intervening independent variables constitute manipulated, influential experimental factors. In this study these include social, economic and household conditions. Dependent variables are the variables that are measured in an examination. The variables may change as the independent variable is influenced. The term dependent variables is used as they depend on what happens to the respondents in the examination in terms of intervening and background 62

84 variables. The dependent variables in this study are annual remittance received by the household and socioeconomic impact. 2.9 Conclusion This chapter has reviewed the literature relating to this research. The literature on international migration remittances was explored. This literature examined the rural microeconomic level of household remittance determinants and their impacts on households and communities. The researcher identified research gaps in the current literature and in doing so formulated the research problem as well as the research objectives. Eventually, four research questions were identified. The above literature review has provided an ample theoretical foundation for this research. This foundation is used by the researcher as a frame work to analyse and discuss the research outcomes. Chapter 3 describes the research methodology for this study. It was further concluded from the literature review that there was a paucity of literature about the determinants of remittances in Bangladesh. The majority of studies at the national level used descriptive statistics to determine the impact of remittances. The literature at the international level was available in abundance. These studies identified the economic determinants of remittances and migration including the microeconomic determinants. It was concluded from the literature survey that remittances were influenced by a number of factors. The next chapter discusses the research methodology of this study, and gives details of the study s research design. 63

85 CHAPTER THREE: METHODOLOGY 3.1 Introduction The design of the research is discussed in this chapter. Among the main areas covered are the study s paradigmatic orientation, design of the field study to gather the primary data including the sample villages and sample respondents, questionnaire formulation and pilot testing, data analytical techniques and their implications, and, finally, the ethical dimensions that were taken into account. This chapter describes the methodology for this research. It comprises the research approach, the study design, the data collection instruments, the research quality and the ethical issues Study paradigm Neuman (2006, p.8 1) defines a paradigm as a basic orientation to theory and research affecting how and why an individual conducts a research study. Paradigm implies a system of beliefs, a worldview, an application, and a phenomenological perspective of social reality (Guba & Lincoln 1994; Heron & Reason 1997; Perry et al. 1999; Ticehurst & Veal 2000). It involves an overall approach or methodology underpinning a researchable problem (Deshpande 1983), and. to Creswell (2003), the fundamental assumptions made by an investigator about how and what can be learned during an inquiry. Research paradigms commonly include, among others, the positivist and phenomenological (Easterby-Smith, Thorpe & Lowe 1991), rationalist and interpretivist (Seymour& Rooke 1995), qualitative and quantitative (Bonoma 1985), inductive and deductive (Parkhe 1993), and feminist and postmodern (Neuman 2006) Paradigms in social science research Social science research has been conducted following four main (and contested) philosophical stances and paradigms (Guba & Lincoln 1994; Crotty 1998; Perry, Riege & Brown 1999) which view reality as: singular or objective (positivism); individually, collectively, or subjectively and culturally constructed (constructivism); individually, collectively, or subjectively and historically contingent (critical theory);a singular reality (realism). Some researchers contend that there are two fundamental worldviews shaping any investigation: positivism and constructivism (Gummesson 2000). The former, based on science and the scientific method, is concerned with theory testing and confirmation 64

86 (Deshpande 1983), while the latter is focused more upon developing theory and describing phenomena (Lincoln & Guba 1985), which can be described as an inductive and a qualitative method in contrast to the former deductive and quantitative orientation. The paradigmatic approach of the investigator strongly shapes the choice of the inquiry methodology and research instrument. The nature and understanding of reality is what constitutes or ontology and the approaches to acquiring an appreciation of reality and phenomena constitute epistemology. The selection of the mode of inquiry and process constitutes the study methodology or design (Guba & Lincoln 1994; Perry, Riege & Brown 1999; Sandelowski 2000; Creswell 2003). Based on the preceding discussion, this study on the determinants of foreign remittances to Bangladeshi households and their linkages adopts the positivist paradigm and uses a quantitative approach. The phenomena being examined are social constructions informed by previous research on the international migration and development nexus, on foreign remittances sent through formal and informal channels by temporary transnational migrants, on the nature and determinants of the remittance flows and the socioeconomic structure of the receiving households, and on the relationships between the latter two factors. An implied study objective is to test the credibility of the extant models, theories and frameworks underlying the various phenomena being studied and their associations. 3.2 Research Methodology Research methodology refers to the thinking and planning involved in deciding what methods or approaches will be used for gathering data to address the research problem and respond to the study questions. It relates to the data collection process, analytical tools, discussion and interpretation. Thus, the study methodology constitutes the design or blueprint to clarify the issues linked to the research objectives, questions and hypotheses (Nachmias & Nachmias 1992). The main objective of this chapter, therefore, is to explain the choice of the various tools and techniques used in this thesis to address the research problem. 65

87 3.2.1 Selection of survey villages and field survey In this research, the study area was selected on the basis of its high concentration of emigrants, especially emigrants to Italy. Figure 3.1 depicts the whole research process. Figure 3.1 Research process Source: Author develop for this research The sample households were chosen based on the prevalence of the receipt of remittances at the lowest administrative unit (sub-sub-district called Union Parisad). According to BMET (2013), there were 59,662 remittance-receiving households in Shariatpur District and 49,882 remittance-sending household members in Italy (BMET 2013). A study by Knights (1996) estimated that the districts in Bangladesh that showed a strong representation of Bangladeshi migrant workers were Dhaka, Faridpur, Comilla and Noakhali, while the number of undocumented Bangladeshi migrants was reported to be about 11,000 in 2009 (Blangiardo 2009). Shariatpur was formed as an individual district in 1982 and it presently belongs to the Dhaka division. Formerly it was part of the greater Faridpur District. The Shariatpur district consist seven sub-districts (Upazila). Among the sub-districts, Naria Upazila has the largest number of household members who have migrated to Italy. The Vogeshore Union, one of the subsub-districts of Naria Upazila, was selected randomly for the study due to a lack of data on household migration to Italy. Bangladeshi migration to Italy is predominantly a rural and 66

88 male phenomenon (Rahman & Kabir 2012). The fieldwork was conducted in villages with high concentrations of household members who had migrated to Italy (as determined by the initial step of identifying which households had received remittances from household members who had temporarily migrated to Italy to work). Shariatpur is about 70 km away from Dhaka, the capital of Bangladesh. As a Bangladeshi, the researcher could interact easily with the villagers and, after visiting several village communities and discussing the research with households from different levels of the village communities, he selected the Vogeshore Union which comprises 10 villages: Kalyan Patti, Achura, Dulukhanda, Sujasar, Anakhanda, Upashi, Masura, Pachak, Chandaniand Kalikata. Each village has an estimated 400 households consisting of 30 to 40 families. The number of households sending members to work in Italy was so widespread that almost every family had at least one such member. The main crops the villagers cultivate are paddy (only one cropper year) and seasonal vegetables together with small scale fish farming Study design The survey data was collected from the migrant households of 10 rural villages in the Vogeshore Union of Shariatpur District in Bangladesh, as the researcher is familiar with the study area language, culture and geography. In the first phase, first-hand knowledge of the migrant-sending households was obtained by asking a single question regarding whether the household had members who had worked in Italy or not. The question was asked of each of the 4013 households in the 10 study villages. Thereafter, a structured questionnaire was prepared comprising several open-ended and closed questions relevant to the research objectives. The respondents were the heads of households or senior household members STAGE 1: The household census in the 10 survey villages To identify the village household members working in Italy, the researcher appointed four field workers who had a higher secondary level of education and experience collecting population census survey data for the Bangladesh Bureau of Statistics in These field workers conducted a survey under the supervision of the researcher to ascertain the actual population of remittance-receiving households in the 10 study sample villages. After ascertaining the names and addresses of household heads from the Union Parisad (the lowest administrative unit of local government), the four field workers were directed to only ask a single question (whether the household had members, at the time of the survey, who worked 67

89 in Italy or not) of each of the 4013 households in the 10 study villages. This enabled the identification of the population from which the sample of households which had received remittances from household members working in Italy would be drawn. The survey took one month (15 July to 15 August 2013). The allocation of sample village households among the four field workers is given in Table 3.2 below. Table 3.2: Allocation of 10 village HH among the field workers Source: Author developed for this study Construction of questionnaire and interview schedule Survey questionnaires are interview instruments for collecting relatively structured data from respondents based on a questionnaire formally designed with questions in a pre-determined order (Ticehurst & Veal 2000; De Vaus 2002). The questionnaire design affects the structure and content of the questionnaire and determines the attributes of the information gathered especially its accuracy and relevance (Sekaran 2003; Zikmund 2003; Burns & Bush 2006). The concepts and constructs involved, and the relationships to be examined in the study, must therefore be clear and guide the questionnaire design process (Ticehurst & Veal 2000). Tanur (1992) and Sekaran (2003) have provided a useful checklist for designing a research questionnaire. The checklist comprises: what to look for and find, simple and clear questions, selection of closed and open-ended questions, avoiding leading questions, question arrangement, pre-test or pilot-test the questionnaire. Researchers should avoid: leading and biased questions, double-barrelled questions addressing multiple issues, and questions challenging the respondent s memory. 68

90 In designing the survey questionnaire used in this study, the researcher made a considerable effort to formulate its contents, format, structure and sequence. Two guiding principles were followed: avoiding confusion and maintaining the perspective of the respondent at all times. To ensure as far as possible that the information needed would be given by the respondents, maximum attention was given to the formulation of the questions. At the questionnaire design stage, appropriate care was exercised when formulating questions related to personal or household economic conditions and to include them in the questionnaire after an interval during which it could be expected that the respondent and interviewer would have developed a measure of mutual trust. The researcher also gave much attention to the order in which questions were presented with a view to encouraging the participants to give the most appropriate answers. The following information was obtained (for the complete questionnaire see Appendix B) in the frame of a standardised interview schedule: a) all household members: age, gender, education, occupation, marriage status b) personal characteristics of the migrant before or during the migration period: educational level, employment status, remittances c) family structure: joint or nuclear d) investment in the agricultural sector: assets, value of land, investment in livestock e) household income, expenditure and revenue: composition of household income, agricultural revenue from renting out, detailed break-down of the expenditures (including marriage expenditure) f) land assets, purchase of land, property of houses and additional buildings. To control and check the respondents answers, several experts (Union parisad Chairman, a local college and school teacher, a farmer and a land officer) in each village were asked the following: general information about the village or Union Parisad, the land tenure system, agriculture yields and typical cropping patterns. The questionnaire used for the field interviews is presented in Appendix B. 69

91 3.2.5 Sample selection Sampling procedures and data sources A sample means a smaller set of cases a researcher selects from a larger pool and generalizes to the population (Neuman 2006, p. 219). In other words, sampling is a process of selecting a sufficient number of elements from the population so that by studying the sample, and understanding the properties or characteristics of the sample subjects, it would then be possible to generalise the properties or characteristics to the population elements (Sekaran 2000, p. 268). Strauss and Corbin (1998, p. 202) emphasise that the aim of theoretical sampling is to maximise opportunities to compare [different situations] to determine how a category varies in terms of its properties and dimensions. That is, a good sample must be chosen at random, and it must be large and unbiased (Hussey & Hussey 1997). The sampling process in this study involved several steps: defining the population, selecting the sample frame and unit, choosing the sampling technique, deciding on the sample plan, and determining the sample size (Luck & Rubin 1987; Kinnear & Taylor 1996; Churchill1999; Zikmund 2003; Neuman 2006). The population is the entire group of households that the investigator must examine, while the element is a single member of that population (Zikmund 2003; Bryman 2004; Davis 2005; Burns & Bush 2006; Neuman 2006). The target population is the specific pool of cases that needs to be investigated (Zikmund 2003; Neuman 2006). The sampling frame includes a specific list that closely approximates all the elements in the population from which the sample may be drawn, that is, the working population (Sekaran 2003; Zikmund 2003; Davis 2005; Burns & Bush 2006; Neuman 2006) Sampling technique and plan Surveys of samples are conducted to achieve a representative data on the target population by contacting individuals and respondents (Bryman 2004). This can be a method of primary data collection based on communication with representative sample of individuals (Zikmund 1997, p. 202). The choice of sampling technique is conditioned by necessity, effectiveness, time and cost limitations (Saunder, Lewis & Thornhill 2000; Sekaran 2003; Sarantakos 2005). Sampling techniques are divided into probability sampling and non-probability sampling techniques (Cooper & Schindler 2003; Sekaran 2003; Zikmund 2003; Davis 2005; Burns & Bush 2006; Neuman 2006). Probability sampling focuses on the techniques that produce a highly representative sample (Neuman 2006) to produce accurate generalisations. 70

92 Examples include: simple random, systematic, stratified and cluster sampling (Saunders, Lewis & Thornhill 2000; Zikmund 2003; Berg 2004). Probability sampling applies statistical means to select the sample in order to be more technically acceptable and to reduce sampling bias and error (Sekaran 2003; Zikmund 2003). Non-probability sampling involves choosing the size of the sample, or cases or units to describe the phenomena of interest (Neuman 2006) and focuses on specific cases, events or actions to deepen the contextual understanding of the phenomenon being studied. Non-probability sampling techniques include convenience, quota, purposive, snowball, deviant case, sequential and theoretical sampling (Saunders, Lewis & Thornhill 2000; Zikmund 2003; Berg 2004). In non-probability sampling, the judgment of the researcher affects the selection of sampling units (Sekaran 2003; Zikmund 2003) Sample size The multiple attributes of a sample are linked to its size relative to the true population, the number of constructs, the level of analytical detail and precision of results, and resource constraints (Ticehurst & Veal 2000; Tabachnick & Fidell 2001). Hair et al. (2006, p.12) stipulate that the minimum sample size must be least five times... the number of variables to be analysed, and the more acceptable sample size would have a 10:1 ratio. Roscoe (1975) argues that the optimum sample size is between 30and 500 while Green (1991) argues that a major determinant is the number of independent latent variables in the conceptual model. Bartlett et al. (2001) recommend that the sample size should be five to ten times more than the number of independent constructs for multivariate research. Hoelter (1983) recommends that the respondent sample size be between 100 and 200 when a quantitative method is used (Bagozzi 1997; Chin & Newsted 1999; Hair et al. 2006). Based on model complexity, missing data and error variance of questions and items, a sample size range of is suggested (Hair et al. 2006; Manning & Munro 2007). The recommendation of Krejcie and Morgan (1970) was followed in selecting a representative sample size for this research; the households who had at least one member in Italy were identified. In the random sample selection process 300 households were selected (Table 3.2), with the share in each village corresponding to its proportion of the total population (the remittance- receiving households). In the selection process, during the first stage of the survey, each household was separated, coded and recorded on a piece of paper. Then, each paper was folded to maintain confidentiality and picked up from each village randomly. Subsequently, and to overcome systematic sample error, all papers were mixed up 71

93 to ensure there was an equal probability of choosing each household. The researcher himself selected the sample units by picking up the folded papers one by one. Afterwards the stack of folded papers was blended and a new person was selected to choose a new selection. The selection process was continued till the target of 300 sample households was attained. Finally, the researcher himself conducted interviews in the sample households with either the head of the household or a senior member using a structured questionnaire. The recommendation of Krejcie and Morgan (1970) was followed in selecting a sample size for this research. The households that had at least one member in Italy were identified. There were a total of 4,013 households, 18,240 family members and 1,344 remittance-sending migrants in Italy. In the random sample selection process a total of 1,344 remittance-sending migrants in Italy were selected as the total population for this study. Three hundred households were selected according to the average aggregate weight of each village population as shown in the Table 3.3. Table 3.3: Population and sample of household of the study area Coding and measurement scales The survey instrument questions and items captured quantifiable data to enable their statistical analysis using the statistical software package SPSS 16.0 to ascertain whether predicted associations and patterns are observable, and to draw the appropriate conclusions. The questionnaire used multiple-item measurement scales to find out whether the overall 72

94 observed score correctly reflected the true population score, and at what confidence levels. The measurement of the study constructs was based on nominal, ordinal, interval, and ratio scales with the least and most precise measurement quality given by the first and last scales (Cooper & Schindler 2003; Zikmund 2003; Davis 2005; Burns & Bush 2006; Hair et al. 2006; Manning & Munro 2006; Neuman 2006). A quantitative approach enables the use of various measurement scales in a survey questionnaire to capture intensity, direction, level and potency of constructs along a continuum (Neuman 2006). In designing the survey questionnaire for the current study, nominal, ordinal, and interval scale-level measurements were used to measure the objective and subjective attributes of the respondents. Rating and attitudinal scales are also used in this thesis. The former evaluates a phenomenon along a continuum or in a category, while the latter measures a respondent s orientation to a specific phenomenon but does not allow for refinement. A Likert scale, which can be ordinal or interval-based, is a rating scale enabling respondents to indicate their reactions to specific constructs in ordinal-level categories ranked along a continuum (Ticehurts & Veal 2000). A Likert scale has the advantage that the variability of the scores of many respondents increases the spread of variance of responses and the ease of construction and administration (Malhorta2004; Burns & Bush 2006). Its disadvantages are the tendency for the aggregate total scores of respondents to be identical, and the time to complete the questionnaire is longer than for itemised rating scales (Malhotra 2004). This research study adopted a four-point Likert scale as the measurement scale in the survey questionnaire because it is simple to administer and code, offers more options for respondents (with less skewed distribution), and is adaptable to varied statistical analysis (Burns & Bush 2006; Manning & Munro 2006). Questions were pre-coded during the survey questionnaire, data processing and analysis. The data were subsequently entered into SPSS version 16.0 for analysis. The definition of variables in the regression equation operationalize is given in Table 3.4 below. 73

95 Table 3.4: Specification of variables for multivariate analysis STAGE 2: Pre-testing and questionnaire modification Pre-testing a survey questionnaire can provide useful feedback on the structure and clarity of the questions, uncover any biased or ambiguous questions and items that can affect its the adequacy, reliability, and validity of the research (Saunders et al. 2003; Sekaran 2003; Zikmund 2003). Zikmund (1997, p.108) defines the pre-testing process as a collective term for any small scale exploratory research technique that uses sampling but does not apply rigorous standards. Latent variables can be validated and the effectiveness of scales applied can be assessed. Feedback from the pre-testing process assists in the revision and improvement of the draft survey questionnaire (Ticehurst & Veal 2000). The group chosen 74

96 for pre-testing the questionnaire should not be different from the actual participants (Zikmund 2003) and must be excluded from the main research. With regard to the pilot study sample size, Hunt et al. (1982) recommend 12 to 30 respondents but Emory and Cooper (1991) and Cooper and Schindler (2003) suggest 25 to 100. In this study, 50 households from the area were chosen for the pre-testing of the survey questionnaire, which was administered by the researcher himself. Appropriate modifications were subsequently made to the draft questionnaire to improve its quality. According to Lancaster et al. (2002, p. 331), the pilot study sample should not be included in the main analysis, because then the decision to proceed with the main study would not be made independently of the results of the pilot study. Therefore, the pilot study sample was excluded from the full-scale sample (300 participants) for analysis STAGE 3: Field experience The data were collected over six months from July to December 2013 by the researcher who resided in the survey area for virtually the whole six months. As transportation is poor in the study area, the researcher conducted his field work by walking and cycling. The survey procedure suffered from a number of problems. Lengthy and detailed interviews were perceived by the respondents as useless. Giving an interview was often considered a waste of time, particularly if it entailed high opportunity costs for the respondent. In addition, travelling long distances consumed the researcher s time and energy while the nonavailability of respondents during his first and second visits negatively affected the data collection schedule. Much time was required for preparing the sampling frame and for the organisation of fieldwork. In addition, data collection was hampered because survey respondents were reluctant to provide information about such topics as migrant attitudes and income. However, the patience and persistence of the researcher eventually resulted in satisfactory responses Data quality control Checking all the completed questionnaires for errors, omissions and discrepancies soon after the interview is necessary to ensure high data quality. The researcher checked and verified each questionnaire on the day of the interview. The data cleansing process was done after the completion of the entry to tally the entered data with a coding sheet or the original questionnaire to eliminate mismatched codes, inconsistencies and incompatibilities. On 75

97 completion, the data was stored on hard discs and data flash drives for further analysis. Afterwards, a backup master copy was also made for safekeeping Reliability of the instruments Any researcher conducting a field survey must ensure that respondents provide accurate information; apart from the interviewer s competence, the survey instrument itself can influence the collection of good quality and reliable data. A reliable questionnaire yields consistently good quality information over repeated measurements and subjects; Bryman and Cramer (2000) state that a reliable measurement has stability (i.e. it is invariant over time) and equivalence (i.e. it is invariant over different respondents). Reliability can be expected when a variable based on summated scales (interrelated items reflecting the underlying construct) is used as a predictor in a model. However, although reliability of data does not ensure its validity and an unreliable instrument cannot produce valid results, in the present research the correlation coefficients for the selected variables were calculated by using SPSS version 16.0 using Microsoft Access software. Office editing was carried out prior to and post-entry. Random checks on data entry disclosed few errors, while the stratification/status and confidence intervals for cluster sampling were assessed Data limitations The data do not account for country- or regional-level factors in explaining the determinants or impacts of foreign remittances, as the research is a micro-level investigation of the survey individuals and households. Although this will influence the generalisability of the analysis and findings, it should be noted that there is a wealth of research on the same topic in multiple countries and regions, including Bangladesh, thereby enabling comparisons and divergences, and hence perceptions of its generality Limitations of the survey The manpower brokers who recruited individuals to work as migrants in Italy and their subagents attempted to persuade the participant household members not to cooperate with the researcher as they claimed that the study would benefit neither household members nor migrant members in Italy but only the researcher. Moreover, potential participants were concerned that undocumented migrants in Italy might be affected by this study. As a result, the researcher took some measures to overcome these limitations: he conducted meetings with influential people in the villages at different points to make the purpose of this research 76

98 clear. He also met the local people, teachers and villagers to convince them about the value of this study as academic research for the purpose of obtaining a higher degree. 3.3 Data analysis The data analysis process involves organising, categorising, tabulating, and examining raw data and transforming it into aggregations for testing hypotheses (Emory & Cooper 1991; Zikmund 2000; Davis 2005; Burns & Bush 2006). The empirical basis for this research study is the data gathered from 300 respondent households which can be initially analysed using descriptive statistics complemented by more advanced techniques suggested by past studies and the research questions Data processing procedures and statistical program Raw data gathered from afield survey requires editing, sorting, coding and error checking before inputting into the computer for statistical manipulation (Cooper & Schindler 2003; Zikmund 2003). The investigator checks the completeness of the survey questionnaires and the eligibility of respondents. Then, the data coding process identifies and classifies each response with numerical scores and symbols followed by cleansing and screening of the data, and checking for consistency and missing values (Tabachnick & Fidell 2001; Hair et al. 2006; Manning & Munro 2006). The screened and checked data is then inputted into the computer using the statistical software program (SPSS 16.0) to generate descriptive and inferential statistical analysis, summarise information and data, and examine the research questions and hypothesised conceptual model (Barclay et al. 1995; Tabachnick & Fidell 2001; Manning & Munro 2006). Various statistical techniques and tests such as chi-square, Gamma, bivariate analysis and multiple linear regression models were used to process the data for examining the study questions. These are briefly described below Univariate analysis The basic patterns in the sample data are identified by descriptive statistics depicted in frequency distributions of specific variables (Render & Stair 1994; Petrie & Sabin 2000; Tabachnick & Fidell 2001). Further, cross-tabulation allows tabular arrangement by determining the frequency of responses and classifying the data against other data sets (Cooper & Schindler 2003). 77

99 3.3.3 Bivariate analysis Inferential statistics refers to the level of confidence of a particular parameter in the population. It involves using sample data to test hypotheses and conceptual models about the relationships in the population on manifest and latent variables (Gay & Diehl 1992; Ticehurst & Veal 2000; Sekaran 2003). In addition, the selected statistical test is based on the format of data, level of measurements, and the numbers of constructs and variables to evaluate the reliability measured questions and items (Sekaran 2003). The Chi-square distribution is the probability distribution of the Chi-square statistic for an infinite number of random samples of the same size, drawn from populations where two variables are independent of each other (Argyrous 2000 p.435). The statistic used to evaluate these differences in Chi-square (χ 2 ) is defined as: where F o = observed cell frequencies F e = expected cell frequencies Gamma test The strength of the relationship between the independent and dependent variables is measured by Goodman and Kruskal s Gamma ( ). This is an asymmetrical statistic, based on the same order pairs (Ns) and a number of different ordered pairs (Nd) calculated by using the formula; If the same order pair predominates, the value of predominate, it is negative. is positive; if different order pairs Multivariate analysis A first-generation statistical technique is a regression-based approach (multiple, discriminant, logistic, conjoint analysis, canonical correlation analysis and analysis of variance) or a factorbased approach (exploratory factor analysis, confirmatory factor analysis and cluster analysis) to either identify or confirm theoretical hypotheses (Haenlein & Kaplan 2004). 78

100 First-generation statistical tests for data analysis are based on testing outlier assumptions and normality distributions, evaluating reliability (Cronbach s alpha) and validity (homogeneity and internal consistency), conducting factor analysis, looking for cause-and-effect relationships between latent variables and testing hypothesised conceptual model relationships. This research required the assessment of the effect of independent variables on a dependent variable. Several methods are available to describe the relationship between a dependent variable Y and several explanatory variables,x 1, X 2, X n,one of which is the multiple linear regression expressed by the following equation: Where the x s are the independent variables, a is the intercept, the b s are the regression coefficients (extent of change in the dependent variable to a unit change in the independent variable), and e is an error term (the proportion of unexplained variance in the dependent variables). To compare the independent variables, standardised regression coefficients (or beta weights) are calculated by multiplying the regression coefficients derived from the different units of measurement by the standard deviation of the relevant independent variables (Azhar 2008). Standardised regression coefficients in a regression equation can be used to compare the significance of two or more independent variables in the explained variance in the dependent variable. The independent variables are highly correlated if the regression coefficient is more than The significance of each independent variable is examined by stepwise analysis while the aggregate influence of all independent variables on the dependent variable is given by the adjusted R 2 value (coefficient of determination) (Johnson & Wichem 2002). In this study, the theoretical perspective is influenced by the structural approach that establishes a relationship between remittances and the socio-demographic characteristics of the receiving households. Such an approach has been empirically tested in analyses of the nature and impact of foreign remittances on Bangladeshi households (Ali 2012; Hasan et al. 2012; Beaudouin 2006, to name a few). The determinants of remittances in the 10 study villages include: (1) socio-demographic characteristics of migrant; (2) socio-demographic characteristics of household heads including household members; (3) asset endowment of household; and (4) expenditure 79

101 pattern of the household. In this study, multivariate analysis was used to assess the relative contributions made by the factors affecting household remittances. Estimation signs and the significance of the determinants will indicate the relative influence of a range of possible determinants of the size of and variations in household remittances. To operationalise the Regression Equation 3.4, a list of variables is required. As a result of the literature review, the determinants of remittances in the 10 study villages were placed in the following categories: (a) individual characteristics (migrant, household head, household members) (b) household characteristics (c) asset inventory of the household (d) expenditure and welfare activities of the household. Each of these categories comprises multiple variables. Therefore, to obtain effective outcomes, the most important variables referred to in the literature were considered for the analysis. After identification of the variables (Table 3.4), the model emerged as follows: Here, e 1 is error term. Since the migration status and household characteristics are heterogeneous in terms of remittance inflows, the magnitude of remittance determinants must vary among different categories of households. To identify the important remittance determinants, all the variables from the survey data have been included in Equation 3.4. However, it is essential to clarify whether the variables can generate the maximum precision of the model. Since the model has not been tested by any previous studies, clarification of the best fit model criteria is required. 80

102 Although the variables are logically included in the model, all the exogenous variables may not be statistically significant. A range of variations in statistical non-significance may be observed for some variables. Some variables may need to be removed, if they are, the highest non-significant variable will be deleted first and the process is iterated until the best model fit is obtained. In this connection, the model building procedure has been conducted in such a way that the highest degree of model robustness incorporates the largest number of explanatory variables. Then, the established model was considered in the data analysis. This model building procedure has been discussed in section on multivariate analysis (Section 4.4.1). 3.4 Research measurement implications Research quality Research quality is tested by the reliability and validity of the research instrument (Creswell 2003) to establish credibility and believability (Ticehurts & Veal 2000; Zikmund 2003). Zikmund (2003) argued that a research instrument is valid if it measures what it is supposed to measure. During the development phase of the survey questionnaire in this research study, specific consideration was given to issues such as reliability and validity, among other design measures, to ensure the quality of the research study outcomes. These research quality measurements, including objectivity, generalisability, reliability, validity, and possible existing errors, are explored in the next section Objectivity and conformity LeCompte and Goetz (1982) argue that objectivity and conformability are external reliabilities with the emphasis on the replicability of the research study by others. Objectivity should also be based on facts resulting from the findings from the actual data and is affected by the use of imperfect methodologies (Lincoln & Guba 1985; Sekaran 2003). In this study, objectivity was established by the actual data outcomes of the advanced statistical analysis with care given to the process of research design and methodology. 81

103 3.4.3 Generalisability Generalisability refers to the degree of applicability of the research study findings and outcomes from one organisational setting to another (Sekaran 2003) and requires making connections to unstudied parts of the original case or to other cases (Maxwell 2002). There are three levels of generalisation: from sample to population, analytic (i.e. theory-connected), and case-to-case transfer (Firestone 1993). Noblit and Hare (1988) note that the generalising process is not mechanical and is more like translating, refuting or synthesising two or more studies of similar phenomena with careful interpretation. To be able to generalise about regularities in human and social behaviour, an appropriate sample size needs to be selected (Easterby-Smith et al. 1991) Reliability and dependability A significant requirement is whether the research process is dependable and consistent, and reasonably static during the period and across investigators (Lincoln & Guba 1985; Davis & Cosenza 1993; Guba & Lincoln 1994; Miles & Huberman 1994; Yin 1994; Ticehurst & Veal 2000; Cooper & Schindler 2003; Sekaran 2003; Zikmund 2003; Davis 2005; Burns & Bush 2006; Neuman 2006). Reliability should minimise errors and biases (Hussey & Hussey 1997; Yin 2002; Zikmund 2003) and demonstrate that data collection procedures under identical or similar conditions can be repeated and achieve the same results (Easterby-Smithet al. 1991; Neuman 2006). Miller and Kirk (1986) argue that the stability of observations over time is different from the stability in the same time frame and warned about cases in which multiple respondents provide somewhat similar responses. According to Bryman (2004), reliability can be internal (i.e., degree to which items making up a scale is consistent) or external (i.e., stability of variables/test-retest). There are three types of reliability: stability (across time), representative (across sub-populations) and equivalence (across multiple measures). Neuman (2006) argues that reliability is rarely perfect and can be improved by conceptualising the latent variables and using precise measurement levels, multiple items and pilot tests. Reliability can be also be tested by testretest or testing the same respondents at two different times; checking half the results of scaled items against the other half; and equivalent-form checking or measuring the correlation between alternative instruments to the same respondents (Zikmund 2003). In this study, the internal reliability was established by an extensive literature review, construct conceptualisation and measurement, survey questionnaire design with multiple items, 82

104 questionnaire pre-testing, and coefficient alpha and composite reliability evaluation (Tabachnick & Fidell 2001; Davis 2005; Manning & Munro 2007). It was not possible to conduct tests of external reliability of the same respondents at different times because of time and cost constraints Validity Validity is the ability of a scale or instrument to measure what it is required to measure (Davis & Cosenza 1993; Zikmund 2003) and it depends on the strength of the formulation of the constructs (Bryman 2004). Validity can be assessed by content or face validity (how acceptable the dimensions of the concept are for measuring what it intends to measure); construct validity (confirmation by measurements and representations of theory-based hypotheses); and criterion validity (correlation of items/measures with other items/measures with other relevant criteria) (Sekaran 2003; Cooper & Schindler 2003; Zikmund 2003; Davis 2005; Burns & Bush 2006; Hair et al. 2006). Validity can also be internal or external as discussed in the next section Internal validity, credibility and authenticity The study findings must be credible and make sense to informed readers (Miles & Huberman 1994). Any change, cause-and-effect relationships to the dependent latent variable can only be attributed to the manipulation of the independent latent variable (Lincoln & Guba 1985; Yin 1994; Zikmund 2003). Internal validity is the central means and degree of confidence for ascertaining the true value of how things really are and whether they really work given the research study situation (Guba & Lincoln 1994; Sekaran 2003; Davis 2005; Neuman 2006). Kvale (1989) views validity as a process of checking, questioning and theorising; in reality, validation becomes the issue of choosing among competing and falsifiable explanations of matters (Miles & Huberman 1994, p.279). According to Maxwell (2002), theoretical validity relates to the existence of a more abstract explanation of the phenomena being studied, while internal validity ensures that the survey questionnaire covers the research study issues (Neuman 2006). In this study, internal and content validity were established by the extant literature review, feedback from experts and questionnaire pre-testing and modification (Dillman 2007; Warner 2008). Construct validity was ensured by the appropriate definition of concepts and the application of statistical tests (Westen & Rosenthal 2003). The conceptual model validity was 83

105 tested by the similarity of alternative measurements of all items of a single primary construct while discriminant validity was secured by different measurements of each construct (Steenkamp & Van Trijp 1991; Bagozzi 1994). Criterion validity was substantiated by measures made at different times and for different respondents (Sekaran 2003) External validity, transferability and fit Whether the research conclusions are broadly generalisable in different contexts and time periods is of importance in the form of statistical confidence limits (Lincoln & Guba Miles & Huberman 1994; Zikmund 2003; Davis 2005). Hence, the survey questionnaire response rate is vital (Sekaran 2003). In this study, the external validity was secured by having a moderately large sample size with a high response rate to ensure a high probability of drawing valid findings (Neuman 2006) Errors Statistical errors can occur if there is a difference between the value of a sample statistic of interest and the value of the corresponding population parameter (Zikmund 2003, p.376). They arise from random sampling error (statistical fluctuations between sample results and data collection using identical procedures), systematic error (non-sampling factors in the research study design and execution), sampling frame error (not considering potential respondents or including non-listed respondents in the population), and non-response error (Zikmund 2003; Burns & Bush 2006). In this study, error was minimised by careful attention to the study design, sampling, survey questionnaire questions and statistical analysis. 3.5 Ethical considerations Research ethics relates to a code of behaviour appropriate to academics and the conduct of research (Wells 1994, p.284) that is followed to ensure that research activities do not harm participants or have any adverse consequences (Cooper &Schindler 2006; Ticehurst & Veal 1999; Zikmund et al. 2010). There is no single approach to ethics as ethical values are influenced by an individual s moral standards and broader social norms (Wells 1994; Zikmundet al. 2010). Diener and Crandall (1978) view research ethics as comprising the right to privacy, informed consent, the absence of deception and causing no harm to participants. 84

106 For this research, ethical issues concerned three parties the researcher, the participants and Southern Cross University, with each party having certain rights and obligations to the other parties. The researcher maintained objectivity, presented honest and true research findings and obtained the approval of the Human Research Ethics Committee of Southern Cross University (Approval Number ECN ), as shown in Appendix E, before commencing the primary data collection activities. This research was conducted in compliance with the National Statement on Ethical Conduct in Human Research (2007) and the ethical issues included: Informed consent: A discussion guide and information sheets were sent to the survey participants to inform them of the discussion topics and the nature and purpose of the research, its benefits and possible risk to those involved, and estimated time taken. The receipt of the completed questionnaires implied the respondents consent to being involved in the study. Voluntary participation: Participation of the survey respondents was voluntary and without coercion and inducement. The participants could withdraw at any time without any questions being asked. Privacy: The right of privacy of the participants was respected and not to be intruded upon. Participants had the right to refuse to participate or to refuse answering any question in the interview or questionnaire. Confidentiality: The data collected was kept confidential and this included participants identity which could not be disclosed to third parties. After the mandatory storage period, the data will be destroyed. In the report, the data will be presented in an aggregate form and no individual data will be shown. Anonymity: The participants will not be named in the report. Deception: Participants will not be misled in any way. 85

107 3.6 Conclusion This chapter has reviewed and justified the broad methodological parameters of the thesis including the study paradigm, the selection of study area, the population frame and sample households, the construction of the survey questionnaire and pilot testing, techniques of data analysis, technical aspects of research measurement and the ethical considerations. The following chapter provides the empirical analysis of the survey data. 86

108 CHAPTER FOUR: DATA ANALYSIS 4.1 Introduction The primary data were gathered according to the study methodology described in Chapter 4 and analysed using SPSS version 16.0 to address the four research questions derived from the literature reviewed in Chapter 2. This chapter includes: the tests of the assumptions and data normality; worker remittance flows; the socioeconomic profile of the migrants and household heads; household composition; and an econometric examination of the micro-determinants of the remittances and their impact on the originating households. Finally, the index variables were created to assess the nexus between the determinants of the remittances and the sociodemographic characteristics of the remittance senders. 4.2 Assumption testing Statistical tests use different sets of assumptions (e.g., identification of univariate and multivariate outliers and normality of distributions) to determine if the sample respondents and scores belong to the same or different populations and if they are normally distributed (Manning & Munro 2006; Neuman 2006) Normality distribution The normality of distributions of latent variables with interval and ratio scales is tested by examining the shape of the distribution of the data collected with a normal outcome defined by its bell shape (Neuman 2006). Data distributions can be too positively or negatively skewed, too flattened (platykurtic) or too peaked (leptokurtic). One solution is to transform the data before further analysis by using transformed scores (Manning & Munro 2006) or by using non-parametric tests (Petrie & Sabin 2000). The normality of data distributions are tested by calculating the scores of kurtosis and skewness for the frequency of values for the explanatory variables. In the presenter search, the P-P (Proportion-Proportion) probability plot is used as a normal distribution method in which the observed cumulative proportion is plotted against the expected cumulative proportion if the data were a sample from a specific distribution. The experimental results for all variables are shown in Appendix A. The P-P plots indicate that the variables are normally distributed about their means. 87

109 4.3 Socio-demographic profile The socio-demographic profile of both the migrant workers and the heads of the originating households were analysed to determine their attributes. All calculations in the following section and subsection are the author s own Remittance sender characteristics Age and remittances The cross tabulation of the ages and amount of remittances received by the heads of the households is shown in Table 4.1. The majority (59 per cent) of the migrant workers are between 21 and 35 years, with 55 per cent concentrated in the year cohort sending BDT 100, ,000 annually to their originating households. It is clear that the older migrant workers, particularly those 45 years and above, are less in demand, probably because of the readily availability of the younger (and more productive and hard-working) migrants. Table 4.1: Distribution of household annual remittance received and migrant s age 77 Bangladesh Taka=1US$. 88

110 Educational level and remittances Table 4.2 indicates that 57 per cent of the migrant workers had attained an educational level of between 9-10 years of secondary education, 20 per cent had completed higher secondary, 10 per cent had completed between 6 and 8 years at secondary level, 7 per cent had completed between 1-5 years primary level, 2 per cent had bachelor s degrees while 4 per cent had received adult education. The data thus reveals that the migrant workers in Italy are relatively less educated workers who are more likely to seek employment in Europe. Table 4.2: Distribution of household annual remittance received and education of migrants 77 Bangladesh Taka=1US$ Marital status and remittances Table 4.3 provides a cross-tabulation of the marital status of the migrant workers and the amount of remittances sent to their originating households. The majority (71 per cent) of the migrant workers were married, thus indicating that marriage (and possibly immediate family dependents) is a strong factor in making the migration decision. 89

111 Table 4.3: Distribution of household annual remittance received and marital status of migrants 77 Bangladesh Taka=1US$ Year of migration and remittances Table 4.4 shows that the length of time spent in Italy by the Bangladeshi workers ranged from 1 to 24 years with 94 per cent having worked less than14 years in Italy and 56 per cent in the 5-9 years range. In general, the longer the migrants worked overseas, the higher the amount of annual remittances to the originating households in Bangladesh. 90

112 Table 4.4: Distribution of household annual remittance received and period of migration 77 Bangladesh Taka=1US Legal status and remittances The survey results presented in Table 4.5 show that, of the sample respondents who sent remittances to their households in Bangladesh, 83 per cent were documented (and could therefore travel freely to their point of origin and back) while 17 per cent were not (and were therefore unable to travel freely to their point of origin). A migrant s legal status affected his remittance flows and enabled frequent visits home, bringing in-kind remittances such as goldjewellery, electronics and clothes for their household members as well as friends and other relatives. 91

113 Table 4.5: Distribution of household annual remittance received and legal status of migrants 77 Bangladesh Taka=1US Number of visits and remittances The cross tabulation of the 300 household participant survey responses regarding the number of visits to Bangladesh and the amount of remittance sent is shown in Table 4.6. The frequency of visits is 0-14 times during the migration period with the highest percentage frequency of visits (21 per cent) being four times and the lowest (1 per cent) being 14 times. About 66 per cent of the migrant workers visited between one and six times while 18 per cent had made no visits at all. However, the number of visits is also related to the migrant s legal status and the kinship relationship with the household members at the point of origin. The results also show that the remittance flows vary from BDT 1,400,001-1,500,000 and BDT 100, ,000 annually. 92

114 Table 4.6: Distribution of household annual remittance received and number of visits by migrants 77 Bangladesh Taka=1US Characteristics of household heads Age and remittances Table 4.7 indicates that the largest group (23 per cent) of the respondent household heads in the originating households in Bangladesh who were between 56 and 60 years old received the highest range of remittances (BDT 700, ,000) while the largest group in the age range received BDT 1,400,001-1,500,000. Although the age of household head may not theoretically link directly with the remittances received, further analysis is required to study this. 93

115 Table 4.7: Distribution of household annual remittance received and age of household heads 77 Bangladesh Taka=1US Gender and remittances The respondents were asked about the gender of household heads who received remittances from Italy. Table 4.8 shows that 57 per cent were male and 43 per cent were female. While the male household heads received remittances of BDT 900,001-1,000,000, female household heads received BDT 1,400,001-1,500,000. However, there were no major differences in the lowest ranges of remittances for male and female household heads. 94

116 Table 4.8: Distribution of household annual remittance received by gender of household heads 77 Bangladesh Taka=1US Marital status and remittances The cross tabulation of the heads of households marital status and the size of remittances received is given in the Table 4.9. The results show that the majority (86.0 per cent) of household heads were married and the remainder were not married. Both married and unmarried household heads received various ranges of remittances. Among the married household heads, 3 per cent received the highest range of remittances of between BDT and 1,500,000 annually compared to 3 per cent of unmarried household heads who received between BDT 1,100,001 to 1,200,000. For all the ranges of remittances received, the married household heads received higher remittances annually. 95

117 Table 4.9: Distribution of household annual remittance received and marital status of household heads 77 Bangladesh Taka=1US Educational level and remittances The respondents were asked about the educational level of household heads who received remittances from their household migrant members in Italy. Table 4.10 shows that 34 per cent of the household heads had attained an educational level of between 9 and 10 years of secondary-level education, 1 per cent had attained the bachelor s degree level, 7 per cent higher secondary level, 23 per cent between 6-8 years secondary level, 21 per cent between 1-5 years primary level, 8 per cent adult education and 6 per cent ere uneducated. The data reveals that the migrants with a secondary level of education were more likely to migrate to Italy and send the maximum yearly remittance flows of between BDT 500,000 and 800,

118 Table 4.10: Distribution of household annual remittance received and education of household head 77 Bangladesh Taka=1US Employment status and remittances The respondents were asked about the employment status of household heads who received remittances from Italy. As shown in Table 4.11, 47 per cent of the household heads were housewives, 40 per cent were businessmen, 6 per cent were farmers, 4 per cent were retired persons and 3 per cent were in private service. Thus, almost half of the household heads were housewives receiving a maximum yearly remittance flow of between BDT 1,400,000 and 1,500,

119 Table 4.11: Distribution of household annual remittance received and employment of household heads 77 Bangladesh Taka=1US Relationship of household head to migrants and remittances The cross tabulation showing the relationship of the household head to the migrant and remittance received is shown in Table The majority (43.0 per cent) of household heads were fathers, 32 per cent were wives of migrants, 15 per cent were mothers and 10 per cent were brothers. The highest range of remittances of between BDT 1,400,001 to 1,500,000 were received by 3 per cent of the wives compared to 3 per cent of mothers who received the highest range of remittances of between BDT 1,100,001-1,200,000, 3 per cent of the fathers received BDT 900,001-1,000,000, and 6 per cent of the brothers BDT 700,001 to 800,000. The data also shows that fathers received the highest percentage of remittances. 98

120 Table 4.12: Distribution of household annual remittance received and migrant s kinship relation to household head 77 Bangladesh Taka=1US Household characteristics The household characteristics surveyed were religion, household size, land ownership and nature of inheritance of the land Household religious status and remittances The cross tabulation showing the correlation between household religious status and remittances is shown in Table Most of the household heads (68 per cent) were Muslim and the remainder were Hindu. Three per cent of Muslim household heads received the highest range of remittances between BDT 1,400,001 to 1,500,000 and 3 per cent of the Hindu household heads received remittances in the range BDT 1,100,001 to 1,200,000. However, the religious status may not link directly with the remittances received, and further analysis is required to study this. 99

121 Table 4.13: Distribution of annual remittances received and religion of household heads 77 Bangladesh Taka=1US Household size and remittances The cross tabulation showing the correlation between household size and remittance is given in Table Household sizes ranged from 3-10 members; 11 per cent of households had 3 members while 1 per cent had 10 members. The largest cohort (30.0 per cent) had five members (excluding the migrant member in Italy). The largest remittances of BDT 1,400,001 to 1,500,000 were sent to households with four members. This group comprised 3 per cent of the sample households. 100

122 Table 4.14: Distribution of household annual remittance received and household size 77 Bangladesh Taka=1US Household land ownership and remittances The respondents were asked about the household land ownership status which was cross tabulated with the remittances received. Table 4.15 shows that 47 per cent of land ownership titles were in the name of the migrant member of the household, 36 per cent were in the name of the household head and the remaining 17 per cent were jointly in the names of two or more family members. The largest remittances (in the BDT 1,400,001 to 1,500,000 range) were sent to households in which the migrant had title to the land. Further analysis is needed to determine whether there is a link between land ownership and the motivation for sending remittances. 101

123 Table 4.15: Distribution of household annual remittance received and household land ownership 77 Bangladesh Taka=1US Number of inheritors in household and remittances The participants were asked about the number of household members who stood to inherit household property and who received remittances from Italy. Table 4.16 shows that the number of inheritors among the survey participants was one to eight members including the migrant member. Forty-six per cent of the households contained two inheritors. The maximum level of remittances (BDT 1,400,001 to 1,500,000) were sent to households in which there were only three inheritors. The number of inheritors varies according to religion, gender and relationship to the household head. Inheritance sometimes influences remittances. The size of remittances also varies from country to country, and is also affected by other factors. 102

124 Table 4.16 Distribution of household annual remittance received and household number of heirs 77 Bangladesh Taka=1US 4. 4 Factors influencing remittance flows Based on the literature review in Chapter 2, the key variables determining the flow of foreign migrants remittances to their originating households are analysed in this section based on the econometric model derived in Chapter Econometric model building and multivariate analysis Based on the literature review in Chapter 2, the key variables determining the flow of foreign migrants remittances to their originating households are analysed. The econometric model developed in Chapter 3 in Equation 3.4 was used to assess the relationships between annual household remittances received from migrant members and the exploratory determinant variables. The econometric model developed in Chapter 3 in Equation 3.4 was tested to assess the relationships between annual household remittances received from migrant members and the exploratory determinant variables. To enable a best fit regressions model, variable reduction was undertaken following a method of backward elimination that begins with the inclusion of all expected independent variables, measuring their statistical significance individually and discarding those which are highly non-significant. The backward regression method was 103

125 conducted in three steps to establish a good fit model that assesses the most influential factors related to migrant workers remittances and to examine the hypothesised relationships in this study. This complies with the explanation of data analysis of Slatten, Svensson and Svaeri (2011) and Juhdia, Pa wanb and Hansaramb (2013). Before beginning the backward elimination steps, the collected data must be checked to test for certain basic statistical considerations to account for the implementation of the good fit regression model (Hocking 1976). In this study, there were three major considerations: the data normality test multicollinearity and autocorrelation as Gujarati (2003) suggests that it is not necessary to carry out all the available assumption tests as some were not relevant to this study. To build up a best fit regression model, twenty three potential explanatory variables were selected for the piloted model given in Equation 3.4 as follows: For the first regression (Equation 4.1), outcomes of the full model are provided in Table The empirical results indicate that some of the potential variables are statistically nonsignificant. For the equation taken as a whole, the R 2 (0.611), F value (13.294) and a p value nearest to zero. Thus, the results postulate that all independent (explanatory) variables in aggregate affect the annual remittance received (the dependent variable) by the household. In the statistical assumption, multicollinearity provides a condition of linear relationships among either all predictor variables or few of them in the regressions equation. The whole procedure usually appears when either all or few of the explanatory variables in the regressions model are strongly significant to each other. As a consequence, the multicollinearity test is very important for selecting the best fit model of regression. Hence, the researcher used the multicollinearity test for the present study. 104

126 Table 4.17: Regression results of the full model The regression results are shown in Table 4.18.The multicollinearity for the high R 2 =0.611 and 10 variables (duration of migration (YMIGm), legal status of the migrant (LEGSm), household land ownership title (HLOWtitle), household size (HHsize), educational level of migrant (EDUm), log household income (Ln_HH_Incom), religion (RELhh), education of household head (EDUhh), log household welfare (Ln_Welf) and loan repayment (Loan_Rep)), are not statistically significant in the initial regressions model of 13 variables. 105

127 Since the classical symptoms of multicollinearity high R 2 but few significant t ratios are found in the first model, clarification is needed of the statistical problem by observing the variance and covariance of the regression estimators. As Gujarati (2003, p.350) states, the OLS estimators and standard error can be sensitive to even the smallest change in the data. The increase of variance and covariance of coefficients are falsified and this can be observed with variance-inflating factor (VIF) and tolerance (TOL) also in Table The rule-of-thumb states that the closer the value of TOL and VIF is to 1, the greater the evidence that one explanatory variable is not collinear with the other explanatory variable (Gujarati 2003). The values of Tolerance (TOL) and VIF in Table 4.18 indicate that there is no multicollinearity existing among the explanatory variables. Table 4.18: Regression results of stage II As stated earlier, the variables are considered for removal sequentially based on their statistically non-significant p value in the equations. For instance, the regression outcomes of the first model (Equation 4.1: long regression) in Table 4.17 shows that R 2 =0.611, and 106

128 adjusted R 2 =0.565 with an acceptable value of d= The elimination process was begun by discarding the variable year of migration (YMIGm) which had the highest p value (0.909), from the first model. This procedure was continued until a best fit model for the explanatory variables was found. The result of the whole backward elimination process is given in Table The ultimate outcome is the first best fit model as represented in the following equation: Table 4.19: Stage II regression model summary However, the results obtained using Equation 4.3 are shown in Table They show that R 2 is slightly decreased (0.596) compared to the first model (0.611) with 13 explanatory variables. This was expected as increasing the number of variables increases the value of R 2 and vice versa. In this stage, the p value of the two explanatory variables, investment in business (Invest_Busi) (0.763) and investment in financial sector (Invest_Fin_Sec) (0.199), were statistically insignificant. Therefore, a further backward elimination process was taken to arrive at the best fit model. 107

129 Table 4.20: Regression results of stage-iii This procedure was continued until a best fit model for the explanatory variables was obtained. The results of the whole backward elimination process are given in Table

130 Table 4.21: Regression results of stage-iii The ultimate outcome of the best fit model is represented in the following equation: Table 4.22: Stage-III regression model summary The best fit model shown Table 4.23 has only 11 explanatory variables with statistical significance levels in the range of 1 per cent to 5 per cent. Both regressions, long and short, provide the degree of the direction and strength of causality between the dependent and 109

131 explanatory variables, which are the yearly household remittance received and the explanatory variables or remittance determinants identified in the literature. Table 4.23: Final stage regression results Remittance determinants of the rural households The statistically developed best fit models, in both short and long equation forms, are regression equation models applied in the data analysis as the aggregate sample data has been broken into sub-samples with specific attributes (e.g., migrant: age and legal status; the head of the household: age, gender, marital status and kinship relation to the migrant) to generate a clear picture of the factors influencing the remittances of those specific study households. Therefore, the regressions are: 110

132 Long regression and Short regression The model variable identification is given in Table 3.3. The error terms e 1 and 1 that satisfy the assumptions of : (i) zero mean, E(e 1 )=0; E( 1 )=0 (ii) constant variance, E(e 1 ) 2 = e 2 ; E( 1 ) 2 = 2 (iii) no autocorrelation exist in the error e 1 and 1 ; E(e 1j )=0 and E( 1j )=0; where 1 j Empirical results The cross-sectional data collected from 300 households through a one-off primary survey are used to estimate the two regression models. To identify the variation in the strength of the remittance determinants, the expected nexus to the dependent and explanatory variables and to quantify those relationships with the maximum information, the analysis comprises the following six parts: (a) analysis of the whole sample in the 10 villages (b) unit analysis of the age of migrants (c) comparative analysis of legal status of the migrants 111

133 (d) comparative analysis of gender of the heads of household (e) analysis of marital status of the heads of household (f) unit analysis of the kinship relations to the migrants. The results of these analyses follow below Whole sample The results for the whole sample consisting of 300 households show the explanatory power of both long and short regressions measured by the adjusted R 2 values, which are statistically significant and high. Table 4.24 indicates that R 2 values for long and short regression equations are and respectively. The test results for overall significance, the F- test, are also statistically highly significant at the level of 1 per cent in both regression equations. Table 4.24 shows that 11 variables are statistically insignificant in the long regression whereas the significance levels of the remaining 11 variables varied between 1 per cent to 5 per cent levels in long regressions. Finally, the marital status of migrant (MARSm) and Land (Ln_Land) variables were removed for the second stage (short regressions) of the model owing to their being statistically insignificant. Thus, the significant influential variables are age of migrant (AGEm), number of visits (NVISTm), age of the head of household (AGEhh), gender of the head of household (GENhh), marital status of the head of household (MARShh), the head of household relation to migrant (RELMhh), development in housing sector (Invest_Hous_Dev) and household standard of living expenditure (Ln_Live_Exp). The following analysis will investigate in more depth the relationships of individual variables. 112

134 Table 4.24: Determinants of average household remittance: Log linear regression results of the 10 rural villages,

135 Unit analysis of the age of migrants The full regression model showed significance at the.001 level while the other two models were shown to be significant at the.000 level of confidence. Therefore, the three models all found a strong relationship between the age of migrant and remittances inflows. The study investigated the relationship between age group and other variables. Hence, the sample was broken into different sub-samples as follows: The age group Table 4.25 indicates that the R 2 values for short regression equations are about and respectively. The results for the test of overall significance, the F-test, are also statistically highly significant at the level of 1 per cent in the short regression equation. On account of statistical limitations, the long regression could not be run in SPSS at this age group while the short regression result in Table 4.25 showed strongly significant correlations with other variables such as the migrant s: number of trips (NVISTm); household head; age (AGEhh), marital status (MARShh), employment status (EMPShh), kinship relation to the migrant (RELMhh); and household development in housing sector (Invest_Hous_Dev). 114

136 Table 4.25: Log linear regression results of age group 26 to 30 years old migrants in the 10 rural villages, The under 30 age group Table 4.26 indicates that R 2 values for long and short regression equations are about and respectively. The results for the test of overall significance, the F-test, are also statistically highly significant at the level of 1 per cent in both regression equations. This under-30 age group indicated a strongly significant correlation with almost all variables except log land of the household. In contrast, the short regression found a limited number of variables were strongly significant such as: number of trips (NVISTm); household head: age (AGEhh), marital status (MARShh), kinship relation to the migrant (RELMhh); and household development in housing sector (Invest_Hous_Dev). 115

137 Table 4.26: Log linear regression results of below 30 years old migrants in the 10 rural villages, The 31 to 35 age group Table 4.27 indicates that R 2 values for long and short regression equations are about 0.99 and respectively. The results for the test of overall significance, the F-test, are also statistically highly significant at the level of 1 per cent in both regression equations. In addition, the unit analysis of the age group showed (Table 4.29) that none of variables was significant at the full model of regression, while the short regression model showed a 116

138 significant correlation with the number of trips by the migrant (NVISTm) and log of household land (Ln_Land). Table 4.27: Log linear regression results of age group 31 to 35 years old migrants in the 10 rural villages,

139 The age group Table 4.28 indicates that R 2 values for long and short regression equations are about and respectively. The results for the test of overall significance, the F-test, are also statistically highly significant at the level of 1 per cent in both regression equations. Table 4.30 explored the results of the age group. The empirical results of the long regression indicated a strong relationship with the legal status of the migrant (LEGSm), the marital status of the head of household (MARShh), educational level of the head of household (EDUhh), employment status of the head of household (EMPShh), land ownership title (HLOWNtitle), investment in financial sectors (Invest_Fin_Sec), development in housing sector (Invest_Hous_Dev), log of household land (Ln_Land), log of household standard of living expenditure (Ln_Live_Exp) and log of household welfare expenses (Ln_WelF). On the other hand, the short regression results showed the strongly significant correlations, with the number of visits by the migrant (NVISTm), development in housing sector (Invest_Hous_Dev), log household land (Ln_Land) and log household standard of living expenditure (Ln_Live_Exp). 118

140 Table 4.28: Log linear regression results of below years old migrants in the 10 rural villages, The above 40 age group Table 4.29 indicates that an R 2 values for short regression equations is about The results for the test of overall significance, the F-test, are also statistically highly significant at the level of 1 per cent in short regression equation. Finally, the above 40 age group results are explored in the Table This age group also could not be used to run full model of Regression Equations 4.1, while Equation 4.3 indicated strong relationships with the number of trips by the migrant (NVISTm), employment status of the head of household (EMPShh) and log household standard of living expenditure (Ln_Live_Exp). 119

141 Table 4.29: Log linear regression results for the above 40 year-old migrants in the 10 rural villages, Unit analysis of the age of rural household head The study intended to examine the different age group with other variables. Hence, the sample was broken into different sub-samples as follows: The under 35 age group Table 4.30 indicates that an R 2 values for short regression equations was about The results for the test of overall significance, the F-test, are also statistically highly significant at the level of 1 per cent in the short regression equation. On account of statistical limitations, the long regression could not be run SPSS for this group while short regression results in Table 4.32 showed that strongly significant relationships with migrant characteristics including Age (AGEm), number of visits (NVISTm); household compositions as development in housing sector (Invest_Hous_Dev), log household land (Ln_Land) and log household standard of living expenditure (Ln_Live_Exp). 120

142 Table 4.30: Linear regression results for household heads younger than thirty-five in the 10 rural villages, The age group Table 4.32 indicates that R 2 values for the short regression equation is The results for the test of overall significance, the F-test, are also statistically highly significant at the level of 1 per cent in the short regression equation. As for the previous two age groups, a statistical problem for the old age group prevented the application of the long regressions model while short regressions are explored in Table The empirical results for this age group indicated strongly significant relationships with marital status (MARSm); household head as marital status (MARShh) and relation to migrant (RELMhh); and none of the determinants found a significant correlation with household composition. 121

143 Table 4.31: Linear regression results for the fifty-six to sixty year-old household heads in the 10 rural villages, The above 60 age group Table 4.33 indicates that R 2 values for long and short regression equations are about and respectively. The results for the test of overall significance, the F-test, are also statistically highly significant at the level of 1 per cent in both regression equations. The over-60 age group long and short regressions results are shown in Table The empirical results of the long regression indicated a strongly significant relationship with the migrant s age (AGEm), level of education (EDUm), marital status (MARSm), legal status (LEGSm) and number of visit (NVISTm). Moreover, the household head as age (AGEhh), marital status (MARShh), educational attainment (EDUhh), employment status (EMPShh) and religion (RELhh). Furthermore, there were strong correlations with household size (HHsize), investment in financial sector (Invest_Fin_Sec), development in housing sector (Invest_Hous_Dev), log household land (Ln_Land), log household standard of living expenditure (Ln_Live_Exp), log household income (Ln_HH_Incom), household business investment (Invest_Busi) and loan repayment (Loan_Rep). The short regression showed strongly significant correlations with age of migrant (AGEm) and number of visits 122

144 (NVISTm); household head employment status (EMPShh); and household compositions as, log household land (Ln_Land) and log household living expenses (Ln_Live_Exp). Table 4.32: Long-Short linear regression results for the above sixty year-old household heads in the 10 rural villages,

145 Comparative analysis of gender of the household head The short regression (model fit) in Table 4.33 shows that the significant determinants of remittance size for households with male household heads were: the migrant s age, marital status and number of trips; the household head s age, employment status and relation to migrant; household composition; investment in housing development and land; and household living expenditure. In contrast, significant determinants of female household heads were the migrant s: age, marital status, number of trips; household head: age; and household composition: investment in housing development, household land and household living expenditure. R 2 and adjusted R 2 square values for male show (0.527 and 0.498) and female (0.840 and 0.829). Table 4.33: Log linear regression results of gender of the head of household in the 10 rural villages,

146 Analysis of marital status of the household head : Married household head Table 4.34 indicates the regression results for the sub-sample of married household heads who have been receiving remittances from Italy to their rural villages in Bangladesh. Of the aggregate 300 households, 258 were identified as having married household heads for whom the results of both long and short regressions are given in Table The R 2 values in the long regression are and short regression and both are statistically highly significant. The. F-test, was found to be significant at the 1 per cent level of significance. 125

147 Table 4.34: Log linear regression results of married household head in the 10 rural villages,

148 Unmarried household head Table 4.35 explores the regression results for the sub-sample of unmarried household heads who have been receiving remittances from Italy in rural Bangladesh. Of the aggregate 300 households, 42 were identified as having unmarried migrants for whom the results of both long and short regressions are given in Table The R 2 values (0.776 and in the long and short regressions, respectively) are statistically highly significant and the joint test of significance, the F-test, was found to be significant at the 1 per cent level of significance. Table 4.35: Log linear regression results of unmarried household heads in the 10 rural villages, Unit analysis of the household relation to the migrant Father as a household head Table 4.36 indicates that R 2 values for long and short regression equations are about and respectively. The results for the test of overall significance, the F-test, are also statistically highly significant at the level of 1 per cent in both regression equations. The long and short regression results are explored in Table The long regression results indicated the strong significance of the correlations between having the father of the migrant as household head and the migrant s: age (AGEm), number of trips (NVISTm); household head: age (AGEhh); and household composition: log household living expenses (Ln_Live_Exp) 127

149 and log of household income (Ln_HH_Incom). In contrast, the short regression found a limited number of variables to be strongly significant such as migrant: age (AGEm), number of trips (NVISTm); household head: age (AGEhh); household composition: log living expenses (Ln_Live_Exp). Table 4.36: Long-Short linear regression results of father as a household head in the 10 rural villages, Wife as a household head Table 4.37 indicates that the R 2 value for short regression equations is about The results for the test of overall significance, the F-test, are also statistically highly significant at the level of 1 per cent in the short regression equation. On account of statistical limitations, the long regression could not run in SPSS for wife as household head, while the short 128

150 regression results in Table 4.37 show a significant correlation with other variables such the age of migrant (AGEm), number of trips by the migrant (NVISTm), age of the head of household AGEhh), household investment in housing development (Invest_Hous_Dev), log household land (Ln_Land) and log living expenses (Ln_Live_Exp). Table 4.37: Short linear regression results of wife as a household head in the 10 rural villages, Brother as a household head Table 4.38 indicates that the R 2 value for short regression equations was about The results for the test of overall significance, the F-test, are also statistically highly significant at the level of 1 per cent in the short regression equation. For brother as a household head, a statistical problem in the long regression model arose that prevented the analysis. However, for the short regression model, Table 4.38 showed a significant correlation with the age of migrant (AGEm), number of visit by the migrant (NVISTm) and household investment in housing development (Invest_Hous_Dev). 129

151 Table 4.38: Short linear regression results of brother as a household head in the 10 rural villages, Mother as household head Table 4.39 indicates that an R 2 value for the short regression equation was about The results for the test of overall significance, the F-test, are also statistically highly significant at the level of 1 per cent in the short regression equation. However, the empirical results (Table 4.39) of the short regression indicated a strong relationship with: the marital status of the migrant (MARSm), number of visit by the migrant (NVISTm) and household investment in housing development (Invest_Hous_Dev). 130

152 Table 4.39: Short linear regression results of mother as a household head in the 10 rural villages, Socioeconomic impact of remittances on the households In this section, data will be analysed for the socioeconomic impact of remittances and the relationships among the variables Impact of remittances on households This study was designed to analyse the socioeconomic impacts of remittances on the originating households in Bangladesh. Thus, following section will discuss in detail: Use of remittances This section discusses how the rural household members spend or use the remittances they receive. By focusing on the head of household, a more detailed and heterogeneous picture of the dynamics of remittance uses emerges. In general, the earnings of migrants may be consumed, invested or saved in the country of destination or they may be saved in the home country for future consumption. As already observed, the proportion of the income that migrants remit depends on a number of factors. 131

153 The data presented in Table 4.40 show that remittances from household migrant members working in Italy are an essential source of income for households in rural Bangladesh for their social and economic activities such as consumption, healthcare, loan repayments and house construction. All the survey respondents used foreign remittances for their daily living expenditure, healthcare, gifts and donations to family members and society, socio-cultural ceremonies, durables, telecommunication and religious activities. The data also reveal that 99 per cent of households extended home stay facilities or renovated old houses while 69 per cent built new houses. As households had taken out loans to cover migration costs and for other purposes, 97 per cent of the household respondents reported the use of foreign remittances to repay migration loans and 98 per cent repaid other loans. Some 89 per cent of the households reported the use of remittances for household members educational expenditure and also to purchase gold jewellery; 55 per cent of household members used remittances for tea, betel leaves and tobacco for themselves and also for entertaining their family guests. No household member reported the ownership of a motor car, while 50 per cent bought motor cycles using the remittances and 4 per cent used the remittances for legal expenses. The main industries in the study area are agriculture and local trading. Agrarian-based industries have not yet been commercially developed in the study area. Nevertheless, 44 per cent of the households invested in fisheries, 15 per cent in farm equipment and 13 per cent in the small livestock sector. Among the land investments, 12 per cent of the households used the remittances to purchase agricultural land, 7 per cent non-agricultural land, 4 per cent commercial land and another 4 per cent bought homestead land and 4 per cent took mortgages on land for other either agricultural or commercial uses. The use of remittances to buy housing properties and to release mortgaged land were confirmed by 2 per cent of the families. While few of the male household heads were shopkeepers, 50 per cent of them used remittances for their working capital, 53 per cent for seasonal businesses and 8 per cent invested in multi-level companies in order to obtain better returns. The data showed that 66 per cent of the households invested in long-term and short-term fixed deposits in local government banks or private banks. In addition, 40 per cent of all respondents invested in the share market, 45 per cent purchased insurance for education or for life coverage. Besides gifts, donations and social ceremonies, the household used remittances for other social and 132

154 welfare activities. For example, 25 per cent of respondents reported using remittances to help send household members abroad), while 1 per cent contributed to dowries for poor relatives. Table 4.40: Uses of remittances by household members However, the above listed individual uses of remittances by the behind household are classified into12 broad categories. Figure 4.1 show the details of these categories of uses of remittances by the household members in rural Bangladesh. The mean percentages of uses were: 28 per cent luxurious life style, 13 per cent welfare sector, 11 per cent loan repayment, 9 per cent housing, 9 per cent financial sector, 7 per cent business sector, 6 per cent healthcare, 6 per cent consumption, 5 per cent children education, 4 per cent agro-based sector, 2 per cent land buying and 29 per cent for other. 133

155 Figure 4.1 Distribution of sector wise remittance uses Table 4.41 explores the financial investment categories in which the recipient rural households invested remittances from Italy. The majority invested mainly in four sectors: financial (62 per cent), fixed deposits (66 per cent), insurance (45 per cent) and the capital market (40 per cent). A few households invested in specialised investment categories set up for migrant households such as five-year term wage earners development bonds (7 per cent), three-year term USD premium bonds (4 per cent), three-year term USD investment bonds (12 per cent), ICB mutual funds (10 per cent), non-residence deposit schemes (8 per cent), nonresidence saving schemes (4 per cent), remittance cards (18.33 per cent) and other nonresidence welfare schemes (31 per cent). 134

156 Table 4.41: Sector-wise financial investment by remittance receiving rural households Thus, foreign remittances play an important role in addressing the socioeconomic needs of the household members remain-behind. The data also suggests that the migration of household members can enhance the socioeconomic status of households Changes in household socioeconomic status The socioeconomic impact of the remittance-receiving households was examined in thisstudy. The highest cohort of the respondents confirmed an increase in both indicators of remittances benefits, namely improvement of children s education and improvement of housing conditions. Table 4.42 shows that 56.0 per cent and 41.0 per cent reported an improvement in housing and children s education to a great extent respectively, and 16.3 per cent reported an increase in the size of social networks and 13.0 per cent reported an increase in the size of family networks. In addition, 6.7 per cent reported a general increase in living standards, 6 per cent reported improvements in household members employment opportunities and 4.0 per cent reported significant increases in their savings. 135

157 Table 4.42: Respondent s perceptions of the socioeconomic impact of remittances Changes in family status Table 4.43 shows that the migrants and their household members remain-behind in Bangladesh had certain ambitions for their household members. There was strong agreement that remittances had helped to pay for assisting family members (4 per cent), saving of money for future happiness (2 per cent), charitable donations (1 per cent) and children s education (1 per cent) Most of the respondents (87 per cent) agreed that there was an increase in social status, 79.0 per cent reported saving money for future happiness, 73.0 per cent reported using remittance money to help relatives, 70.0 per cent donated to religious organisations and 49.0 per cent investment in business to increase income. The table shows that all the respondents agreed that the remittances prevented the sale of inheritance property. At the same time, 96.0 per cent reported that remittance money enabled children s higher education. However, remittances may have a positive impact on elementary education while higher education may not depend on remittances. Higher education might depend on academic success and enrolment opportunities as Bangladesh has a lack of higher educational opportunities. 136

158 Table 4.43: Socioeconomic impact of remittances on family status 4.6 Relationship between household socio-demographic characteristics and socio-economic impact of remittances Construction impact variable The measuring instrument for socioeconomic impact consisted of questions answered by employing the four point Likert scale in the questionnaire. The socioeconomic impact variable comprises more than one item and various respondent categories. From the descriptive or univariate analysis, each statement is observed by tabulating a frequency table and computing the percentages of the respondents answers that were in each category. In bivariate analysis, every single item in the matrix question is cross-tabulated with the respective answer variable which may be long and not significant. To overcome these challenges, an index variable is commonly created to study the associated outcomes of whole statements in forecasting the answer variable. In this context, all answers (to a great extent, to a reasonable extent, to some extent and no change) in the matrix questions are compiled simultaneously to construct the index variable. As the index variable is developed, consistency within all the items in the question matrix has to be ascertained by a reliability check using Cronbach s Alpha, obtainable using SPSS As long as the value of Cronbach s Alpha is between 0.7 and 1.0, all of the statements in the question matrix can be used to create the index variable, while statements with a value below 0.7 are discarded. Once the exercise is satisfactorily completed, the maximum and minimum 137

159 outputs are ascertained by sub-menu descriptive statistics together with their respective frequencies. The minimum scores are deducted from the maximum scores, and the residuals are divided by the number of categories in the index variable. A reliability test provided the value of Cronbach s Alpha (0.842) that confirmed the availability of consistency components in all the statements. Therefore, they can be combined to create a single assertion describing benefits obtained after remittances were received. In this context, the score of all statements was summed up and the minimum score attributed in the matrix question was 5 whereas the maximum score was 15. The score range from 5 to 15 was divided into three ranges of answers. Scores of 5 to 8 is grouped as low outcomes, range 9 to 12 were labelled as medium, and 13 to 16 as strong. The three categories (low, medium and strong) reveal the strength of the socioeconomic benefits from the remittances. Descriptive statistics were used to ascertain the mean score and the standard deviation, which were and respectively Empirical econometric model to analyse of socioeconomic impact Finally, an econometric model was developed to assess the association between socioeconomic impact and the exploratory variables. To analyse the data, a twelve-stage backward elimination process was undertaken to build up a best fit model of a regression equation. The elimination processes were started by discarding one by one all statistically insignificant variables. Thus, the model determined the relationship with three categories such as migrant household head and household composition. Before beginning the backward elimination steps, the collected data was checked to ensure that it satisfied certain basic statistical requirements to account for the implementation of the good fit regressions model (Hocking 1976). In this study, the cross-sectional data used for that purpose were analysed using three main tests the data normality test, multicollinearity and autocorrelation. This is because Gujarati (2003) suggests that it is not necessary to test all assumptions, while it should be applicable depending on the types of data. In constructing a tentative complete model, 24 variables were included (Equation 4.4). The complete model is as follows: 138

160 The ultimate outcome is the best fit model which is represented in the following equation: Here, e 1 is error term. The explanation of variables in the regression equation above is provided in Chapter 3, Table Empirical results of regressions The cross-sectional data collected from 300 households through a one-off primary survey were used to estimate the results in the regression models. To identify the variation in the strength of the socioeconomic impact of remittances, the expected association between socioeconomic (dependent) variable and explanatory variables, and to quantify those relationships with the maximum information, the analysis was conducted as follows: Table 4.44 shows the results of the regressions for Equation 4.4. The analysis found significant relationships with socio-demographic characteristics such as migrant s legal status, and aspects of household composition including religion, size of household, investment in banks, insurance or other financial services,, business, welfare and loan repayment. 139

161 Table 4.44: Full model regression results Table 4.45 presents the results of the best fit model of Regressions Equation 4.5 after the twelve step elimination process. Each stage of the elimination process significantly decreased the values of R and R 2 and simultaneously increased residual and F values. Thus, the model built up the statistical best fit to determine the level of significance to the other variables. 140

162 Table 4.45: Model summary Finally, Table 4.46 below explores the results of the best fit model of Regression Equation 4.5. The empirical results indicate that there is a significant relationship with remittances and also other socio-demographic attributes of the age of migrant, educational attainment legal status; household s head relation to the migrant; and aspects of household composition including religion, title of land, amount of land, investment in housing development, business, welfare and loan repayment. 141

163 Table 4.46: Significant coefficients of the best fit model results Socioeconomic impact and migrant socio-demographic characteristics Age Table 4.47 shows a linkage between the age of the migrant and the socioeconomic impacts on the rural household from remittances. It reveals that the migrants were aged 21 to 25 years and that their remittances had a strong socioeconomic impact in 50 per cent of were associated with a high socioeconomic impact on households due to remittances. The cross tabulation further shows that 27.8 per cent of the 51 to 55 age group acquired the lower value on the index variable. Thus, the age of migrants is strongly associated with the extent of socioeconomic impacts of remittances. Younger migrants had strong socioeconomic impacts while older migrants had lower socioeconomic impacts. This was expected as the younger migrants were likely to send substantial remittances for the improvement of socioeconomic conditions in the household, and the older migrants were likely to send less. The earning capacity of the younger migrants may also have been higher than the older migrants. Furthermore, the chi-square test result of P<0.005 and also the Gamma test confirmed the significant association between the age of migrants and the socioeconomic impacts of their remittances. Thus, this research suggests that the younger age of the migrants, the higher is the socioeconomic impacts of their remittances. 142

164 Table 4.47: Age of migrants and socioeconomic impact from remittances Educational level Table 4.48 indicates an association between the educational level of migrants and the socioeconomic impacts of their remittances. The data shows that, for migrants with only 1 to 5 years of primary education, their remittances had a strong socioeconomic impact in 57.1 per cent of cases and there was a weak socioeconomic impact in 4.8 per cent of cases. Table 4.51 below shows that, when the migrant had a bachelor s degree, in 16.7 per cent of cases the respondents reported a low socioeconomic impact from remittances, and 50.0 per cent of respondents reported a strong socioeconomic impact from remittances. Hence, the migrants level of education is significantly linked to the socioeconomic impacts of the remittances. Remittances from migrants with lower levels of education (primary level) had strong socioeconomic impacts on the originating households more often than did remittances from migrants with higher levels of education. This might be explained by the fact that the migrants with a lower educational level had experienced hard work before migration and could therefore earn more in the destination and send higher remittances than the more educated migrants. The chi-square test value P<0.005 and the Gamma test show the strong association between the educational attainment of migrants and the socioeconomic impacts of their remittances. Thus, this research suggests that the lower the educational level of the migrants, the stronger the socioeconomic impacts of their remittances. 143

165 Table 4.48: Education of migrant and socioeconomic impact of remittances Legal status Table 4.49 below indicates the connection between migrant legal status at the destination and socioeconomic impact of remittances on the originating household. The results show that remittances sent by undocumented migrants had a strong socioeconomic impact in 82.4 per cent of cases. Remittances sent by documented migrants a strong socioeconomic impact in 16.1 per cent of cases and a low impact in 7.2 per cent of cases. The chi-square test result P< and the Gamma test confirmed the significant association between legal status of the migrant and socioeconomic impact of their remittances on the receiving household. This outcome may be because migrants with legal documents tend to settle in the host country and undocumented migrants are more likely to maintain a connection to members of their household who remain in Bangladesh. 144

166 Table 4.49: Socioeconomic impact from remittances and legal status Socioeconomic impact and socio-demographic characteristics of household head Household head relation to migrants Table 4.50 indicates the association between the relationship of the household head to the migrant and the socioeconomic impacts from remittances. When the household head was the mother of the migrant, remittances had a strong socioeconomic impact in 40.0 per cent of cases, whereas they had a low socioeconomic impact in 4.4 per cent of cases. The results also show that, when the household head was the migrant s brother, remittances had a strong socioeconomic impact in 23.3 per cent of cases and a low impact in 3.3 per cent of cases. Hence, the household head relation to migrants is strongly associated with socioeconomic impacts. When the household head was the migrant s mother, the socioeconomic impact was more likely to be strong and when the household head was the migrant s brother, the socioeconomic impact as less likely to be strong. This may be because mothers as household heads are more than trusted than brothers. The chi-square test result states P< 0.1 and the Gamma test confirmed the strong association between socioeconomic impact and the kinship relation of the migrant to the household head. 145

167 Table 4.50: Kinship and socioeconomic impact of remittances Socioeconomic impact and household socio-demographic characteristics Religion Table 4.51 shows the association between religion of household heads and socioeconomic impact of remittances. The results reveal that when a Muslim was the head of the household, remittances had a strong socioeconomic impact in 31.4 per cent of cases. The table also shows that when the head of the household was a Hindu, remittances had a low impact in 4.2 per cent of cases and a strong impact in 18.8 per cent of cases. Hence, the religious background of household is strongly associated with socioeconomic impact. Muslim household heads were more likely to be associated with strong socioeconomic impacts, while Hindu household heads were more likely to be associated with low impacts. This may be because Muslim household heads and the household are permanently settled in the study area, and therefore they are more economically active, whereas Hindu households have extended households in India. The chi-square test result P< and the Gamma confirmed that there was a strong association between religious background of the household and the socioeconomic impacts of remittances. Thus, the study suggests that, when a Muslim is the household head, the socioeconomic impacts from remittances are likely to be higher. 146

168 Table 4.51: Religion of household head and socioeconomic impact of remittances 4.7 Conclusion This chapter presented and analysed the primary data collected from the questionnaire survey. The quantitative data from the questionnaire survey were analysed using the SPSS statistical software package. The findings permitted the testing of the four research questions. The next and final chapter discusses the conclusions that can be drawn from the testing of the research questions. It then discusses the implications for the research problem, and for policy and practice. 147

169 CHAPTER FIVE: ANALYSIS AND DISCUSSION 5.1 Introduction In Chapter 5, the primary data gathered was analysed by applying suitable quantitative techniques to clarify the research problem and provide responses to the research questions. The data was analysed by applying tabular (cross tabulation) and graphical formats for the individual and household characteristics, and univariate and multivariate regression models to examine the nature of associations hypothesised. In this way, the socio-demographic profile of households, the determinants of the remittance flows and their relationships and impacts were explored. This research investigated the determinants of international migrant remittance inflows from Italy and their socioeconomic impacts on the originating households in rural Bangladesh. This chapter presents the findings of this research. The objective of Chapter 5 is to build on the previous chapters, particularly Chapters 2 to 4. Chapter 5 describes the impact of the study on the basis of the evidence emerging from the primary data analysis process detailed in Chapter 4 and the findings of the literature review in Chapter 2 The study has four sections that are responses to each of the four research questions. Responses are framed by the literature review and are based on the analysis of the primary data gathered. Chapter 5 summarises the findings emerging from the testing of the research hypotheses. It also reviews and highlights the research issues. 5.2 Responses to the research questions In this section, responses for each of the four research questions are given based on the analysis of the primary data gathered and framed by the literature review Research Question 1 The first research question is: What are the socio-demographic profiles of the foreign migrant workers and the heads of the Bangladeshi households receiving remittances from their family members working in Italy? 148

170 Socio-demographic profiles of migrant workers sending remittances to originating households Several questions about the Socio-demographic characteristics in the questionnaire examined of the remittances to recipient households and remittance senders. Details are discussed below. Table 5.1 Socio-demographic profile of migrant household members *European Community, Non- European Community Source: Developed for this study Age In line with the literature on Bangladeshi migrant workers in diverse destination countries, this study shows that the predominant group of migrant workers in Italy (87 per cent) are under 45 years, while about 59 per cent belong to the year age cohort (see Table 4.3). These figures are similar to those in studies conducted by Siddiqui (2004), Ullah (2007) and Rahman and Kabir (2012). Given that the international migration by Bangladeshi workers started in the early 1990s, this study confirmed that the younger cohorts are now more likely to migrate compared to the older household members. This trend is linked to the unemployment rates among the younger cohorts in rural Bangladesh, which are currently higher than they have been in the past. These younger workers may be attracted by the better income opportunities in Italy. 149

171 Education Related empirical studies indicate that the educational attainment of migrants is associated with their income and is one of the major determinants of the size of remittances (Agrawal & Horowitz 2002; Fonchamnyo 2012; Lucas& Stark 1985). However, most of the related research in Bangladesh revealed that for Bangladeshi emigrants, different destinations tended to be associated with different educational levels. Most of the research shows that migrants had no education only primary-level schooling. However, Rahman and Kabir (2012) found a comparatively higher literacy rate among Bangladeshi migrants in Singapore. The present study found that the majority (98 per cent) of the migrants in Italy had lower-secondary levels of education, similar to most other studies on Bangladeshi migrant workers, possibly because this study was conducted in a rural area, and in Bangladesh literacy and educational levels are lower in rural areas. Marital status The literature review revealed diverse findings in the many studies on this attribute. Married migrants who lived with their spouses at the destination are less motivated to remit money to their home country than migrant workers who have a spouse who has remained behind in the originating country (Carling et al. 2008). In the context of Bangladesh, studies have revealed that most migrants were married (Siddiqui and Abrar 2001; Raihan 2010). This study also found that the largest group of migrant workers in Italy (71 per cent) were married. The same study also confirmed that the married people are more likely to migrate than unmarried household members. This trend is linked to the current economic conditions in rural households, the struggle to survive, and the desire to provide children with better educational opportunities. All of these factors motivate married migrants to seek better income opportunities in Italy. Legal status The legal status of migrants is a complicating factor in the study of determinants and the impact of foreign remittances on the originating households. Legal status depends on the receiving country s immigration rules, regulations and policies. Bangladeshi emigrants are predominantly unskilled temporary workers in the Middle East and South East Asia. Ullah (2007) found that most migrants in Malaysia and Hong Kong came with either visitors passes or crossed the border via a third country. Literature on the consequences of such 150

172 labour migration has revealed that many migrants comprise a vulnerable group (TIP 2013). However, this study identified two categories of migrants: documented migrants (who can travel freely between their home country and the destination country), and undocumented migrants who are not eligible to return to the destination country if they return to their country of origin. This study found that most migrants were documented (83 per cent) while other studies found their migration entry process mostly undocumented (for example: Rahman & Kabir 2012). This trend is linked to the liberal legalisation and integration policies in Italy for migrants from Bangladesh. Length of stay The literature review revealed diverse findings about the connection between the determinants of remittances and duration of stay in the host country. Some studies find negative effects on remittance inflows (Holst & Schrooten 2006; Vargas-Silva 2006; Fairchild & Simpson 2004; Menjivar et al. 1998), while others find that the length of stay has no significant impact on the size and frequency of remittances (Merkle & Zimmermann 1992). However, country-level studies indicate that the average duration of migration depends on the choice of destination country. The duration period of most migrants in this study (5-9 years) is different from the migration of Bangladeshi workers to the Middle East and other South East Asian countries, but similar to the findings in Rahman and Kabir s (2012) study. Number of trips The literature review revealed that the number of visits by the migrants to their home countries depended on: the migrants legal status in the destination country, duration of migration, economic conditions in the country of origin, and the relationships to the household members in the origin country. When visiting the home country, migrants normally bring gifts for their household members, extended family and friends to maintain their community membership and ties (Goldring 1998). Thus, there are direct and indirect forms of remittances. Migrants who make frequent visits to their homeland do so not only to sustain community relationships but also to maintain critical economic linkages (Kemper 1991). In this study, the documented Bangladeshi migrants in Italy visited Bangladesh between 1 and 11 times while undocumented migrant workers could only travel home once. This finding differs from other research about migrants to other destination countries but closely resembles the work done by Rahman and Kabir (2012). 151

173 Household head profile The head of the household age The head of household age is a significant determinant of remittance behaviour, although the age factor varies in different destination countries (DeVoretz & Vadean 2008). Age of the head of household, and gender, influence the motivations for sending remittances. For instance, male heads of the households are less likely to receive remittances than female heads of households (McDonald & Valenzuela 2012). This study shows that the majority (57 per cent) of household heads are in the age bracket followed by the year old group (32 per cent). This outcome suggests household heads are mostly parents and spouses in the home country; this finding is similar output to other country-level studies. Table 5.2 Socio-demographic profile of migrant household heads *European Community, Non- European Community Source: developed for this study. Gender of household heads The literature shows that the gender of the household head is critical in the migration decision and remittance flows to the originating countries. Gender of the household head must be given special attention as in households where the head is male, the motivations for sending remittances, and the use of those remittances, are different from female-headed households (Pfeiffer et al. 2008). As for wives left in the originating country, they are forced to bear greater responsibilities over household budgets, remittance income use and children s education. In the current study, most of the household heads were males, in the older age groups and fathers, similar to the findings of other remittance studies in Bangladesh. 152

174 Marital status of household heads Empirical studies of remittance motivation have shown that the marital status of the household head also determines the nature of the remittances received. Thus, Pfau and Giang (2008) found that married household heads are likely to receive less annual remittance flows, while remittance flows to widows and unmarried heads tend to be higher. Also, female heads of households, particularly those who are married, tend to receive more remittances. The current study shows the majority (86 per cent) of household heads were married and were mostly either parents or spouses, similar to related studies in Bangladesh. Education of household heads The empirical investigation of McDonald and Valenzuela (2012) concluded that highly educated household heads tended to be in households with better economic resources and income opportunities that were less dependent on international remittances. Thus the size of remittances was not correlated with such household heads. This lends some support for the altruism motive for sending remittances home. However, in this study, the majority (92 per cent) of the household heads in Bangladesh had not progressed to secondary school, a level of education similar to that found in other rural-based migration studies in Bangladesh. Occupation of household heads Occupation of household heads has also been found to be linked with the decision to migrate and the motivation of migrants to send remittances. For instance, Quartey (2006) explored the different levels of remittance linked to the occupation of household heads. National-level studies of international migration and remittances have indicated that most of the household heads are female and the situation was similar in this study. In rural Bangladesh, female household heads are usually not engaged in employment except in household work. They take care of household members by cooking, cleaning, nursing and are sometimes involved in the children s education Household composition Religion of households The literature review revealed links between religious faith and remittances sent by migrants to their household members in their home countries (Cadge & Ecklund 2007). In the study 153

175 area, the predominant religion is Islam followed by Hinduism and this is reflected in the migrant households as well. Table 5.3 Socio-demographic profile of migrant household *European Community, Non- European Community Source: developed for this study. Household size In addition, the literature review shows that household size, migration decision and remittance behaviour are interrelated (Sackey 2010; Mishra 2011). Larger households are more likely to have household members who migrate, although household economic conditions must also be considered. In the current study, the average household size was between three and five members while 51 per cent of the households have 4 of 5 members, including the migrant household member. This finding differs slightly from those of Ullah (2007), but closely resembles the work undertaken by Atamanov and Berg (2012). The average household size is lower than in both Ullah (2007) and Atamanov and Berg (2012), possibly because this study was conducted in a specific rural area, and in Bangladesh household size varies from region to region. Land ownership The ownership of land and remittance behaviour has been found to be associated with related investments and also the inheritance motive (Brown 1997; Blue 2004; Ahlburg & Brown, 1998; Vadean 2007; Dakila & Claveria 2007; Lueth & Ruiz-Arranz 2007). The questionnaire had one question about ownership of the land by the migrant s household in Bangladesh and one about the ways in which remittances were invested in the land sector. The second 154

176 question was more relevant to the research topic and the results will be discussed in the section on the impact of remittances. This study revealed that, in 47 per cent of households, the household land as owned by the migrant. These findings are similar to those in studies conducted by Brown (1997) as Tongan and Western Samoan migrants are motivated to remit for reasons of self-interest, and in particular for asset accumulation and investment in their countries of origin. This suggests that most of the migrants were household heads or bought land using their remittances. Number of heirs The number of heirs to the household assets is one determinant of remittance behaviour because it influences inheritance motivation (Brown et al. 2014). However, it should be noted that the inheritance concept varies across countries, regions and religions. The area in which the current was conducted has two main religions and each religion has its own cultural framework in respect of the inheritance of household property and assets. For instance, in Muslim households, the inheritance pattern and practice varies with gender and is completely different for male and female household members. According to the Muslim legal framework, sons inherit double the father s property or assets compared to daughters, and the cultural norm is that daughters do not take any property from the father while they are married (Sarwar et al. 2007). In the current study, most (83 per cent) of the households have two to three heirs Research Question 2 This section summarises the findings emerging from Research Question 2: What are the determinants of the remittances sent by household members working in Italy to their households in Bangladesh? In this study, the researcher considered 23 variables from the prior international and national literature reviewed. The elimination processes were started by discarding one by one all statistically insignificant variables. Thus, the regression model examined the influence of variables such as migrant household head and household composition. Statistically developed best-fit regression models, in both short and long forms, were used to analyse the survey data. The aggregate sample data was broken into sub-samples with specific attributes to generate a clear picture of the factors influencing the remittances of the study households. The crosssectional data collected from 300 households through a one-off primary survey were used to 155

177 develop the two regression models. Thus the models were used to identify variations in the strength of the remittance determinants, the expected significance of correlations between annual remittances flow and explanatory variables, and to identify which relationships were the most significant. The results for the whole sample consisting of 300 households show the explanatory power of both long and short regressions measured by the adjusted R 2 values which were found to be statistically significant. The results of the test of overall significance, the F-test, are also statistically highly significant at the level of 1 per cent in both regression equations. Eleven variables were shown to be statistically insignificant in the long regression whereas the remaining 11 variables ere significant at either the 1 per cent or 5 per cent levels in the long regressions. The marital status of migrant and Log Land variables were removed at the second stage (short regressions) of the model because they were statistically insignificant. Thus, the significant variables are migrant age, frequency of visits at the home country; head of the household s age, gender, marital status and kinship relation with migrant; investment in household development and household living expenditure. Table 5.4 below shows that the above eight variables were highly significant. 156

178 Table 5.4 Factors influencing remittance flow to rural households Migrant characteristics Source: developed for this study *European Community, Non- European Community The literature review revealed that the age of migrant was one of the key determinants influencing remittance behaviour (Osili 2007). Empirical studies have found that larger remittances were sent by younger migrants (de la Briere et al. 1997; Lerch et al. 2006). In addition, several micro-level studies in Bangladesh have found that most of the migrants were young especially when they first migrate (Siddiqi & Abrar 2003; Siddiqi 2004; Afsar et al. 2000; Murshid et al. 2002) and that the remittances inflows from young migrants were also higher. The current research also found that the younger age groups have a higher propensity for migration and that their remittance flows are higher. Numerous empirical studies have investigated whether the numbers of trips by migrant workers to their originating households affected their remittance behaviour (Lerch et al. 2006; Garip 2012). When visiting their households, migrants invariably bring gifts for their household members, extended families, relatives and friends to maintain community ties and networks (Goldring 1998); hence, remittances can be direct or indirect, in cash or in kind. In contrast, a small number of journeys home by migrants is associated with a lower likelihood that they will send remittances, either as cash or in-kind. Migrants who make frequent visits to their originating communities not only sustain community linkages and networks, they also 157

179 consolidate critical economic linkages (Kemper 1991). This study also revealed that the number of visits by the migrants were significantly linked to remittances inflows, thus showing a motivation to maintain strong kinship relations with the left-behind household members. The empirical micro-level studies have established links between the education level of migrants and the income of the migrant (Agrawal & Horowitz 2002). McDonald and Valenzuela (2012) have found that the more educated the migrant, the higher the level of remittances, while Hagen-Zanker and Siegel (2008) found evidence that the relationship between education level and remittances behaviour varies with the migrant s country of origin, even when the same destination is shared. In this study, the descriptive statistics showed variations between the different levels of educational attainment of the migrants and remittance inflows while the three stage regressions result did not find a strongly significant connection between educational attainment and remittances inflows. This suggests that the employment status and income in the destination country may not relevant to the level of education attained by the migrant at the point of migration. Sahu and Das (2009) found that single migrants and married household heads living alone at the destination were more likely to remit than married household heads and migrants staying with their spouses and children. However, Collier et al. (2011) found that the migrant s marital status did not influence remittance motivation. The present study found that marital status of the migrant did influence remittance flows but the three stage regression results did not find a significant relationship between marital status and remittances flows. This suggests that marital status and income in the destination country do not influence the size and frequency of remittances. Micro-economic level studies of remittance motivation have yielded a range of findings about the effect on remittances of the length of stay of the migrants: some studies have found that the length of stay has a negative effect on remittance inflows (Holst & Schrooten 2006; Vargas-Silva 2006; Fairchild & Simpson 2004; Menjivar et al. 1998), while others have found that the length of stay has no significant impact (Merkle and Zimmermann 1992).This study found that duration of migration influenced migrants remittance inflows but the three stage regression process did not find a significant relationship between the duration of migration and remittance inflows. This suggests that duration of migration and income in the destination country may have a significant impact on remittance size and frequency. 158

180 Some empirical studies have found a link between the legal status of the migrant and remittance flows (Holst et al. 2008, 2010, 2011; Bettin & Lucchetti 2012). Markova and Reilly (2007) have found a positive significant correlation between the legal status at the destination country and remittance inflow in the country of origin. Other studies have arrived at similar findings (Collier et al. 2011) and that the migrant s legal status affects their earnings in the destination country and therefore remittance flows fluctuate according to legal status. This study found (see Table 4.7) a correlation between the legal status of the migrants and remittance inflows but the three-stage regressions result did not find a strong relationship Characteristics of the head of household The household head age is another important determinant of the remittance behaviour of migrant workers but its influence varies from country to country (DeVoretz & Vadean 2008). Germenji et al. (2001) have found that older household heads received more remittances than the younger household heads, suggesting an altruistic motive in adult children caring for their old parents as well as their grandparents. This research confirms Germenji et al. s (2001) finding that older household heads (56-60 years) tend to receive larger remittances than the younger household heads. The gender of the household head can similarly be linked to the remittance behaviour of the migrant worker and deserves special attention as there is evidence that remittance behaviour varies with the gender of the household head and affects how households allocate their resources (Pfeiffer et al. 2008). In addition, when a wife is left at home in the country of origin, the women in the household experience changes and increased responsibilities in managing the household budget and remittance income as well children s education. This study also found that gender had a significant impact on remittances flows. This research showed that male household heads were more likely to receive remittances than female household heads. The findings of this study were consistent with the results of international and national studies. The literature has also found a link between the marital status of the household head and remittances received by household members. Married household heads are likely to receive smaller amounts in annual remittances, while remittance flows to widows and unmarried household heads tend to be higher. In general, however, when the head of the household is female the household is likely to receive higher remittances especially when the head of the household is the wife of the migrant (Pfau & Giang 2008). The current research finds that 159

181 married household heads received more remittances than household heads with other marital statuses. This was the finding in both the descriptive statistics and in all three stages of the regressions. The theories of remittances explaining the different dimensions of the remittances concentrate upon roles of the household and the relationships between household members (Stark & Bloom 1985). In this study, the descriptive statistics showed that most household heads were either the fathers or wives of the migrants. The regressions also found a significant relationship between the relationship of household head to the migrant and remittances. This is consistent with other national and international studies. The literature has generally shown that the relationship between educational attainment of the household head and remittances inflows can produce positive effects, negative effects or no effects. For instance, McDonald and Valenzuela (2012) found that the higher education levels of the household heads may reflect better household resources and income opportunities and so less economic need for overseas income: thus, the educational attainment of the household head has no significant association with the remittance amount showing some support for the altruism motive. The current study confirms the findings of McDonald and Valenzuela (2012) Household compositions A number of studies have found significant relationships between housing development either in renovation or new construction and this is confirmed in the current study (Siddique 2004; Ullah 2007; Rahman 2012). In the literature, when remittances were mostly used for household living expenses for the originating household, the expenditure was in areas such as household food consumption, education and health care. In this study, the descriptive statistics showed that almost every household spent remittances for their living expenses and the three stages of the regression strongly confirmed this. Such remittance behaviour on the part of the migrant worker strongly reflects an altruistic motive. Previous empirical studies have suggested a nexus between the religion of the migrants and their family members in the originating country, and remittance flows (Cadge & Ecklund 2007). The current study has found that, in the descriptive statistics, there were variations in remittances depending on the religion of the sample households (Muslim and Hindu). However, the three stage regression analysis did not find a significant relationship. 160

182 Previous empirical studies have suggested different outcomes and relationships between the migrant household size and the remittance behaviour of the migrant worker. For instance, Atamanov and Berg (2012) showed that larger households are more likely to send migrant workers abroad. In contrast, Sackey (2010) found a significant relationship between household size and insurance motive of migrant workers to remit to their home country. Mishra (2011) has also found that household size affected remittance inflows in Nepal. In the present study, the descriptive statistics found a connection between the size at the origin household and remittance inflows but the three stage regression did not find a strong relationship Research Question 3 This section summarises the findings emerging from Research Question 3 What is the socioeconomic impact of such migrant worker remittances on the originating households in Bangladesh? Household level Several questions were included in the questionnaire to examine the socioeconomic impacts of remittances sent by the migrant member to their households in Bangladesh (Table 5.5). The impact of the remittances received included improving children s education, improving housing conditions, increasing household members employment opportunities, increasing living standards, enhancing financial stability, and saving money for future happiness. These socioeconomic impacts are discussed in this section. 161

183 Table 5.5 Socioeconomic impact of remittances at the household level *European Community, Non- European Community Source: developed for this study In the literature, it has been found that remittances have a positive impact on the education of children in the originating households, and this was strongly confirmed in the present research (IOM 2009; Azhar 2008). This implies that the remittance-receiving households could not adequately finance their children s education. Expenditure on education included buying books and study materials, paying tuition fees and tutor fees, and sending children to better schools. In supporting the results of past studies at the national and international levels, the present research also revealed that the originating households used remittances on housing development (Siddique 2004; Ullah 2007; Rahman 2012), strongly attesting to the fact that rural housing is generally poor. In the families of unmarried migrants, household heads normally give priority to settling previous loans before undertaking housing improvements, partially to improve the family s socioeconomic situation before marriage. With regard to the impact of remittances on the employment opportunities of members of the originating households, empirical research has revealed both positive and negative impacts (International Organisation for Migration (IOM), 2009). In the current study, the majority of household members employment opportunities in the home country were largely unchanged. This is consistent with related studies at the national (Bangladesh Bank 2011) and international levels (Azhar 2008). Nevertheless, a few of the sample households reported that 162

184 their migrant members helped other household members to migrate to Italy or other destinations on account of the poor employment opportunities in Bangladesh. With respect to the impact of remittances on household members standards of living, the literature reveals mixed outcomes. However, in the current study, the members of the recipient households experienced reasonably higher but not significantly higher living standards, as some other studies have reported (Brown 1997). Such outcomes suggest that the migration period in Italy may not be sufficiently long for living standards to improve appreciably as other priorities for the use of remittances may appear such as the repayment of migration loans, education, housing development and marriage expenses, which are usually high in Bangladesh. With regard to the impact of remittances on the financial stability of the originating family, this study, like past research, found that most of the households experienced reasonable improvements in their financial stability (Ranis 2008). This suggests that household members primarily used remittances for basic needs, repayment of loans, household infrastructure development and this increased financial stability. The empirical studies on remittances have found mixed savings behaviour in the left-behind household members. The present study revealed that the majority household member save money for their future happiness. This result was strongly consistent with the findings of national and international studies (IOM 2009) and suggests that the households were trying to save some amounts for future happiness and probably bought insurance for household members. In some past empirical studies on remittances and educational level, a distinction was made between elementary and higher education (Zweig 2008). Some studies have divided the education sector into elementary and higher levels of education (Rains 2008). The empirical studies have found different impacts at both levels of education. This study was strongly consistent with the other studies in finding that the lower levels of education of the household members improved significantly but that remittances had a minimal impact on higher levels of education (De Haas 2006). This result suggests that household members spent their money only lower levels of education. At the same time, higher levels of education in the home country and also abroad are very expensive, which may have discouraged migrant households from investing in them. 163

185 The literature review revealed that the savings by household members followed various patterns which also varied from country to country, rural to urban and so on (Bangladesh Bank 2011). In this study, savings were identified in only few sectors like banks and the capital market. Most households indicated that they put their savings into long-term fixed deposits in banks and the capital market. This result was strongly consistent with other studies (IOM 2009), although usually rural level household are not involved in the capital market. This study found that the households also invested their saving in the secondary capital market (buying shares and stocks) owing to the strong influence of capital market manipulation just before the survey period. Thus this study suggests that households who invested in the capital market may lose their money Community-level impacts Several questions in the questionnaire examined the socioeconomic impact at the community level of the remittances to recipient households. Remittance expenditure in this area focused on increasing social status, helping relatives, donating to religious organisations, preventing the sale of inheritance property, extending family networks, extending social networks, and investing in business to increase income. Details are discussed below. Table 5.6 Socioeconomic impact from remittance at the community level *European Community, Non- European Communit Source: developed for this study Past studies have shown that remittances increased the social status of the recipient households in their communities in the origin country (Ullah 2007). This was confirmed in the current research although the extent of the increase was modest. Such a result appears to 164

186 suggest that, in the study area, social status depended not only on income level of the household but also on other factors such as cultural norms, family status, educational attainment and employment status. The empirical studies of international migration and remittances have found that the remittance receiving household members usually help their relatives directly or indirectly (Rahman & Kabir 2012); this was supported by the current research although the responses varied in the extent of assistance provided. Thus, households assist their relatives to different degrees dependent on such factors as amount of remittances inflow, kinship relationship and economic conditions. Several studies have also revealed that religion is a determinant of the remittance motive (Kelly & Solomon 2009) but this was not a significant factor in this study, although the descriptive statistics showed a moderate impact (Table 4.43). However, many of the respondent households contributed to their religious institutions irrespective of whether the households were Muslim or Hindu. Both religions have their own religious festivals, practices and institutions to support. In the literature, migrants who sent remittances were sometimes motivated by the desire to protect their inheritance property rights (Cox 1987; Lucas & Stark 1985), although variations have been observed among home countries affected by different legal, religious and cultural practices. This study did not find an inheritance motive in any of the three stages regression or in the descriptive statistics. This appears to suggest that, in the study area, social status neither depended on property level nor on number of inheritors of such property in the household but other factors such as migrant income, cultural norms, family status, educational attainment and employment status. Micro-level studies on remittances have found evidence of their impact on extended household family networks (Lerch et al. 2006; Garip 2012). The present study also found evidence of such impact. This suggests that the remittance receiving households are conscious of their family networks and are motivated to assist their poorer relatives to draw attention to their family status or for altruistic reasons. In addition, and in line with the evidence in other studies that workers remittances increased household members social networks at the community and cultural levels, this study also confirmed that the majority of the households reported that their social networks had extended. This suggests that the remain-behind household members participated socially or contributed economically to their 165

187 communities. Empirical evidence also shows that the remittance-receiving households invested remittances to improve and increase their income from regular or seasonal businesses dependent on such factors as the household structure, the household head s employment status and so on Uses of remittances at the family level The uses of remittances are particularly important in developing insights into the migrationdevelopment nexus. The current study examined the major uses of remittances by each migrant household as shown in Table The researcher could not quantify the amount of cash used in each end use for two reasons: (i) migrant families usually did not maintain any records; and (ii) there was a risk of gathering inflated data. Thus, the analysis in this study is confined to the major uses of the remittances received. The study showed that remittances from household migrant members in Italy were an essential source for the originating households in rural Bangladesh to undertake such activities as household consumption, education, healthcare, repaying loans, house renovation or construction, gifts and donations to family members and the community, socio-cultural ceremonies, and the purchase of consumer durables and telecommunication devices. The findings of this study are similar to those found in studies conducted by Siddiqui (2004), Ullah (2007), Bangladesh Bank (2011) and Rahman and Kabir (2012). In the study area, the main occupations are in agriculture and fishing and in local businesses linked to the fishing, farm equipment and livestock sub-sectors. Also, the households invested in land for a variety of purposes. The few male household heads who were shopkeepers used remittances to invest in their businesses while others invested in seasonal businesses and in multi-level marketing companies to get better returns. Investments were also made in short- and long-term bank deposits, the stock market, and insurance policies for education or life coverage. Households also invested in social, cultural and religious activities and supported poorer relatives. The study findings demonstrate that migrants remittances play a pivotal role in addressing the socioeconomic and cultural needs of the originating households and their members while enhancing the households socioeconomic status in their villages. 166

188 5.2.4 Research Question 4 This section summarises the findings emerging from Research Question 4: What are the links between the socio-demographic profiles of the Bangladeshi households receiving remittances from Italy and the determinants and socioeconomic impacts of such remittances? In this study, the researcher considered 24 socioeconomic factors affecting variables which included the remittances themselves, individual migrants, household heads and household composition as independent variables based on the prior international and national literature reviewed. The socioeconomic impact variable (details in Chapter Four, Section 4.7.1) was created from the four-point Likert scale in the questionnaire by employing univariate analysis. A reliability test yielded a Cronbach s Alpha of 0.842, which confirmed the availability of unity components in all the statements. Therefore, they can be combined to create a single variable describing the benefits obtained after remittances were received. Finally, an econometric model was developed to observe the association between socioeconomic impact and the exploratory variables. To analyse the data, a twelve-stage backward elimination process was undertaken to build up a best-fit regression equation. The elimination processes were started by discarding one-by-one all statistically insignificant variables. Thus, the model determined the relationship of remittances with three factors: migrant, household head and household composition (details in Chapter Four, Section 4.7.2). The cross-sectional data collected from 300 households through a one-off primary survey were used to estimate the regression models. Furthermore, these regression coefficients estimate the strength of the relationships between predictor and response variables in order to verify the associations emerging from the Chi-square and Gamma tests. The values of coefficients linked to the study variables are given in Table

189 Table 5.7 Significant relationship between determinants of remittances and of socioeconomic impact Source: developed for this study Consistent with similar studies, this research found (see Section ) that younger household members are now more likely to migrate than older household members. This is linked to the current higher unemployment rates among the younger cohorts in rural Bangladesh, and the attraction of the better income opportunities in Italy. The literature review also revealed that the age of the migrant was one of the key determinants of remittance behaviour (Siddiqui & Abrar, 2003; Siddiqui 2004; Afsar et al. 2002; Murshid et al. 2002). The current research yielded similar results (Section 5.2.2). Younger age groups were more likely to migrate than older age groups and their remittance flows are higher. This research further investigated the relationship between age of the migrant and the socioeconomic impact on the originating household. The study found a significant relationship between age of migrant and socioeconomic impact. Thus, this study suggests indepth research into the connection between age of migrant and socioeconomic impact in the home country. The current research revealed (Section ) that 98 per cent of the migrants in Italy had not progressed beyond lower secondary school. This is similar to most other studies on Bangladeshi migrant workers (Ullah 2007; Bangladesh Bank 2011; Rahman & Kabir 2012). The education levels may be low because, in Bangladesh, the literacy and higher education rates are lower in the rural villages than in urban areas. The descriptive statistics showed variations in associations between different levels of educational attainment of the migrants 168

190 on remittance inflows but the three stage regression results (Section 5.2.2) did not find a significant connection between educational attainment and remittance inflows. This suggests that the employment status and income in the destination country may not be relevant to the educational level attained by the migrant at the point of migration. However, the researcher intended to investigate in-depth the relationship between the educational background of the migrant and the socioeconomic impact of remittances on the originating household. The study found a significant relationship. Thus, the study suggests that future research on the link between educational attainment and other factors which affect remittances inflows as well as socioeconomic impacts on the remain-behind household. Legal status depends on the receiving country s immigration rules, regulations and policies. This study classified migrants into two categories: documented migrants (who could travel to and from their country of origin), and those who were not eligible to travel back to their destination country if they ever returned to their country of origin. This study revealed (Section ) that most migrants were documented (83 per cent). This trend is linked to the liberal legalisation and integration at the destination country of Italy. The descriptive statistics showed (Table 4.7) a variation between the legal status of the migrants and remittance inflows, but the three-stage regression results did not find a strong relationship. The legal status of the migrant is a complicating factor in the study of the determinants and impacts of foreign remittances on the originating households (Holst et al. 2008, 2010, 2011; Bettin & Lucchetti 2012). However, the researcher only analysed the legal status of the migrant member in the destination country based on information from the originating household head during the survey period July to December, The results (Table 4.49) show that there is a significant correlation between socioeconomic impact at the originating household and the legal status of the migrant. Hence, the study suggests more research at the destination in regard to their regulatory format in terms of migration status and economic conditions. In the study area, the predominant religion is Islam, followed by Hinduism, and this is also reflected in the migrant households. The present study has found that, in the descriptive statistics, there was a correlation between religion (Muslim or Hindu) of the sample households and the size and frequency of remittances (Section ). However, the threestage regression analysis did not find a significant relationship. Many of the respondent households contributed to their religious institutions irrespective of whether the households were Muslim or Hindu. The researcher found a significant relationship between religion and 169

191 socioeconomic impact in the study area. Religion is one of the key elements in the rural household in terms of inheritance of the property which is directly linked household economics. Therefore, the study suggests more in-depth research on the basis of religious background of household in the light of moral economics theories. In this study, the descriptive statistics showed (Section ) that most of the household heads were either fathers or wives of the migrant. The regressions found a significant relationship a significant relationship between remittances and what relation (father, wife, mother, etc.) the head of household was to the migrant confirming other national and international empirical work (Section 5.2.2). The theories of remittances explaining the different dimensions of the remittances concentrate upon the roles of the household and relationships of household members (Stark & Bloom, 1985). Therefore, the researcher analysed further and found a significant relationship between the household head s relation to migrant and socioeconomic impact of remittances at the originating rural household. Thus, this study suggests further investigation of all household members relationships to the migrant and socioeconomic impact on the remain-behind household. 5.3 Conclusions about the research hypotheses This section summarises the findings emerging from Research Hypothesis. In this study, the researcher considered 23 variables from the prior international and national literature reviewed. The statistical estimation process was discussed section 4.5. The models were used to identify variations in the strength of the remittance determinants, the expected significance of correlations between annual remittances flow and explanatory variables, and to identify which relationships were the most significant. The results for the whole sample, consisting of 300 households, show the explanatory power of both long and short regressions measured by the adjusted R2 values which were found to be statistically significant. The results of the test of overall significance, the F-test, are also statistically highly significant at the level of 1 per cent in both regression equations. Eleven variables were shown to be statistically insignificant in the long regression whereas the remaining 11 variables were significant at either the 1 per cent or 5 per cent levels in the long regressions. Thus, the significant variables are migrant age, frequency of visits at the home country; head of the household s age, gender, marital status and kinship relation with migrant; investment in household development and household living expenditure. Table 5.4 above shows that the above eight variables were highly significant. The research hypotheses are discussed below. 170

192 Table 5.8 Summary of research hypotheses Hypothesis H1 The higher the household living expenditures, the higher the likelihood of a migrant s sending remittances. In the literature, when remittances were mostly used for household living expenses for the originating household, the expenditure was is areas such as household food consumption, education and health care. In this study, the descriptive statistics showed that almost every household spent remittances for their living expenses and the three stages of the regression strongly confirmed this hypothesis. Such remittance behaviour on the part of the migrant worker strongly reflects an altruistic motive. The altruism model predicts that remittances will slowly decrease over time and distance (the remittance decay hypothesis) with the weakening of the altruistic motive (Rapoport & Docquier 2005). NELM claims that migration becomes a household strategy of risk diversification because of market failure at the point of origin (Wouterse 2006). During natural calamities, affected households receive 171

193 more remittances than during normal seasons (Lucas & Stark 1985) and, similarly, negative income shocks significantly increase foreign remittances (Pleitez-Chavez 2004). H2 The higher the number of dependents in the household, the higher the likelihood of a migrant s sending remittances. Previous empirical studies have suggested different outcomes and relationships between the migrant household size and the remittance behaviour of the migrant worker. For instance, Atamanov and Berg (2012) showed that larger households are more likely to send migrant workers abroad. In contrast, Sackey (2010) found a significant relationship between household size and the insurance motive of migrant workers to remit to their home country. Mishra (2011) has also found that household size affected remittance inflows in Nepal. In the present study, the descriptive statistics found a connection between the size at the origin household and remittance inflows but the three stage regression did not find a strong relationship. H3 The lower the household s income, the higher the likelihood of a migrant s sending remittances. Migration not only maximises income and minimises economic risk, but also diversifies household income sources. The consequences of working abroad on the sending household are economic (Azam & Gubert 2006) and social (Rahman 2007) and are discernible in household income, the smoothing of consumption patterns, investment in new business enterprises, accessibility to financial systems and accumulation of capital in land, machines, technology, building construction and renovation (Yang 2008; Adams 2006; Taylor & Wyatt 1996; Lucas 1987; Woodruff & Zenteno 2001). However, this research did not find significant relationship between household income and remittance flow. H4 The lower the household s assets, the higher the amount of remittances from migrants. Remittances can be motivated in the sending community in the form of property, land, cattle and other assets (Durand et al. 1996; Blue 2004), although such remittances also reflect family ties. When the altruism motive is strong and the destination countries incomes are higher, more remittance will be sent (Brown 1997; Blue 2004; Ahlburg & Brown, 1998; Vadean 2007; Dakila & Claveria 2007; Lueth & Ruiz-Arranz 2007), thus creating direct and 172

194 indirect impacts on the household and community, respectively. However, this research did not find significant relationship between household assets and remittance flow. H5 The higher the prospects for future inheritance from household members, the higher the propensity to remit. Inheritance, seen as an enforcement mechanism, increases remittances and household income and decreases risk. The inflow of remittances increases with (a) household properties and income, (b) the member inheriting and (c) the migrant s assets and income (de la Baiere et al cited in Rapoport & Docquier 2005). This research, however, did not find significant relationship between household income and remittance flow. H6 The more favourable the investing opportunities in the origin community, the higher the propensity to remit. A number of studies have found significant relationships between housing development either in renovation or new construction and this is confirmed in the current study (Siddique 2004; Ullah 2007; Rahman 2012). In this study, the descriptive statistics showed that almost every household spent remittances for their living expenses and the three stages of the regression strongly confirmed this hypothesis. H7 The stronger a migrant's links to the community, the higher the likelihood of migrant s sending remittances. Many empirical studies find that the number of trips to the original household influences remittance behaviour (Lerch et al. 2006; Garip 2012; Roberts & Morris 2003). During the visit to their original household, migrants bring gifts for their household and family members as well as their community networks (Goldring 1998). Therefore, remittances have direct and indirect impacts in the home country in cash and in kind. Migrants who make frequent visits to their country of origin not only sustain a community connection, but also have an economic impact (Kemper 1991). However, Grabel (2008) found that a huge percentage of remittances were carried by hand by migrants during their trips home (Garip2012; Roberts & Morris 2003). In contrast, Holst and Schrooten (2006) found that personal trips to the origin country did not increase the flow of remittances. In this study, the descriptive statistics showed that almost every household spent remittances for their living expenses. The three stages of the regression strongly confirmed this hypothesis. 173

195 In reality, however, remittances are not motivated by just one factor (Brown 2011) but by a mixed range which varies over time and place (Rapoport & Docquier 2005). Although motivations differ among different individuals, diverse motives to send remittances may coexist in one individual (Lucas & Stark 1985; Cox et al. 1998; Feinerman & Seiler 2002). 5.4 Conclusions about the research problem The findings emerging from this study relate to the research problem based on the literature review and the highlighted the research issues and the pertinent theories that underpinned its parent and immediate disciplines. The findings have revealed that: The socio-demographic characteristics found in this study are more or less similar to those found in other studies of households with at least one migrant member in Italy. Some differences with other studies are discussed in Section The factors influencing remittances such as age of migrant, number of visits by the migrant, age of household head, gender of household head, marital status of household head, and household head relation to migrant were the key sociodemographic determinants. The socioeconomic impacts of remittances showed positive effects at the household and community levels, including significant impacts on the household children s education and housing development. The research gaps which formed the basis for the development of the research questions that were tested in this research. Quantitative data were collected through a questionnaire survey. The analysis of the literature review and the quantitative primary research data yielded comprehensive and useful insights on the research gaps as set out in Table 5.9 and underlined the relevance of the research problem stated as: What are the key determinants of remittance inflows from migrant workers in Italy and their socioeconomic impacts on the originating households in rural Bangladesh? 174

196 These findings indicate that: The factors that determined remittance flows were not the same as the factors that determined the impact of remittances. However, some socio-demographic and socioeconomic characteristics were both influential determinants of remittance flows and socioeconomic impacts. These included migrant age, kinship relation between household head to the remittance sender and development at the originating household. The factors significantly influencing remittance flows were frequency of migrant visit at the home country, household head age, gender and marital and household living expenses while factors significantly influencing the impact of remittances included education of migrant, legal status of the migrant, household religion, land ownership title, investment in business, welfare and loan repayments. Table 5.9: Investigation of key research issues Source: calculations by the author from the survey data The research findings enabled the research gaps to be bridged. This permitted the reaffirmation of the relevance of the research problem stated as: What are the key determinants of remittance inflows from migrant workers in Italy and their socioeconomic impacts on the originating households in rural Bangladesh? 175

197 CHAPTER SIX: CONTRIBUTIONS AND LIMITATIONS OF THE STUDY 6.1 Introduction This research investigated determinants of international migrant remittances inflows from Italy and their socioeconomic impacts on the originating households in rural Bangladesh. This final chapter presents the research contribution, research limitations, possible future research direction and conclusions. It commences with an overview of the earlier chapters. Chapter 1 introduced the Research Problem: What are the key determinants of remittance inflows from migrant workers in Italy and their socioeconomic impacts on the originating households in rural Bangladesh? It presented the justification, research methodology, defined the key terms and set out the limitations of this research. Chapter 2 reviewed the literature on the international migration and remittances with the preliminary objectives of developing the theoretical framework for the research and identifying the research gaps. Chapter 3 discussed the research paradigm, research design and justified the adoption of the quantitative methodology for the research. It also discussed questionnaire survey data collection procedures and the measures taken to address the concerns of validity, reliability and ethical issues. Chapter 4 presented data analysis of the primary questionnaire survey data collected. The econometric model was built up in order to answer the four research questions and test the seven hypotheses. Chapter 5 described as responses for each of the four research questions are given based on the analysis of the primary data gathered framed by the literature review; summarises the findings emerging from research hypothesis; reviews and the highlights the research issues. The objective of this chapter is to highlight the contributions and limitations of the research. It focuses on the study s contributions from the theoretical and practical perspectives, and discusses the research s limitations. The chapter concludes with propositions for future research. 176

198 6.2 Contribution This research has contributed to three fields of knowledge as follows: Knowledge supported by this research In this study, the household members socio-demographic, socioeconomic and household composition were analysed using a quantitative approach with primary survey data. This analysis provides an understanding of which factors affect international remittance inflows in rural household as well as how the households socioeconomic development linked with such remittances. The study developed an econometric model and applied it in explaining the connection between international remittances determinants and impacts on the rural remainbehind household. The household unit of analysis in this research contributes a direction of sub-sample analysis to find the depth in each variable. For instance, the age of migrant is one of key significant determinants of remittance as a single unit. This unit (age of migrant) was divided into five parts to observe the strength among the age group in terms of remittance inflow. The present study finds that the each age group of the migrants was linked to different factors which also influenced remittance flow. This form of analysis can be undertaken in terms of socioeconomic impact at the rural level. The best fit model also found a significant relationship between characteristics of the head of household and remittance flow. The study also segmented this (age of household head) variable into four parts and the empirical indicated that the different age groups of the head of the households were correlated with other variables. This unit (age of household head) can be analysed in terms of socioeconomic impact to find out which age segments have the most positive and negative impacts on the rural household. The current study analysed the marital status of the head of household divided into married and unmarried classifications. The empirical results indicate that the unmarried head of the household were correlated with migrant: age, number of visits and marital status while married head of the household with kinship relation to the migrant. This unit (marital status of the head of household) can be further analysed in terms of gender, employment status and so on as well as socioeconomic impact at the household and also at the community level. 177

199 This study also examined the influence of the gender of the head of the household on remittance flows. Both gender stated few common variables which influenced remittance flow while there was also discrimination on account of the gender behaviour with their covariables. This unit (gender of the head of the household) can be further analysed in terms of kinship relation with the migrant and also age of migrants and heads of the households. Furthermore, gender can be analysed to find the connection between socioeconomic impact and other factors influencing remittances. This research also looked at the influence of four kinship relations namely father, mother, wife and brother. The empirical results also show that the head of household s relationship to the migrant had a significant impact. Four kinship relationships explored were correlated with migrant: age, marital status and number of visits, which in turn influenced remittance flows. The cross-sectional data collected from 300 households through a one-off primary survey were used to develop two (long and short) regression models. To identify the variations in the strength of the remittance determinants, the significance of the relationship between the dependent and explanatory variables as examined, and to quantify those relationships with the maximum information, the analysis examined independent variable separately. From the economic perspective, the analysis helps in the examination of what factors can enhance remittances received from foreign migrant members and what can lead to their deterioration. Finally, the measuring instrument of socioeconomic impact used a Likert scale in the questionnaire answers. The socioeconomic impact variable comprises more than one item and various respondent categories. From the descriptive or univariate analysis, each statement is observed by tabulating a frequency table and calculating ratio of the answers. In bivariate analysis, every single item in the matrix question is cross-tabulated with the respective answer variable, which may be lengthy and not significant. To overcome these challenges, an index variable is created to study the associated outcomes of all the statements in forecasting the response variable. In this context, all answers (to a great extent, to a reasonable extent, to some extent and no change) in the matrix questions are compiled simultaneously to construct the index variable. Then, the econometric model is developed to observe the association between socioeconomic impact and the exploratory determinant variables. This implies that the impacts of diverse remittance determinants are crucial in designing the policy for specific categories of remain-behind village households. 178

200 6.2.2 Knowledge extended by this research This study has reviewed the major theories and literature on migration (national and international) and remittances to explore the dynamics of remittances as they relate to the causes, consequences, and impacts of remittances on the remain-behind households and communities. Migration and remittances may be investigated from several theoretical perspectives balanced by a mixed methodology approach as well as a single subjective theoretical approach. The literature review identified and discussed the theoretical framework of this study (Section 2.7). The New Economics of Labour Migration argues that households allocate members to migrate in order to improve not only absolute but relative income and alleviate their deprivation against a reference cluster as the village community (Stark & Bloom 1985; Stark et al. 1986; Stark & Taylor 1989; Stark 1991). This study found that, for the household to send a family member to Italy, this represents a self-enforcing contract to enable risk sharing or investment in potentially higher income streams between the household and the migrant. The results of this study support those of other empirical studies (Lucas & Stark 1985; Rapoport & Doquier 2005; Solimano 2003; Hagen-Zanker & Siegel 2007). Remittances are a beneficent transfer to the household members and are motivated by kindness and emotional attachment (Lucas & Stark, 1985). Such altruism affects a household s economic situation, the strength of family ties and contractual family arrangements. The set of variables that can affect the altruism determinant of remittances include: marital status, household income, gender, age, education level, the number of dependents and the length of the worker s stay abroad (Barua et al. 2008). The findings of this study support the findings of previous empirical studies such as that of Lucas and Stark (1985) and Barua et al. (2008). The study contributes to the research on the dynamics of international remittances by establishing a backward elimination regression process to analyse the cross-sectional survey data. In particular, this study examines the links between the socio-demographic attributes of the head of household, attributes of the migrant, household composition and the international remittances and remittance inflows and socioeconomic impacts at the household and community levels. The association between socio-demographic characteristics, remittance inflow and socioeconomic impact has been found to be logical, and remittance is a rational strategy adopted by rural households as a livelihood diversification strategy. 179

201 6.2.3 Policy recommendations The government policies regulating foreign migration of Bangladesh workers are clearly pivotal to protecting the social welfare of its citizens in various destination countries, ensuring that the processes governing the outflows and inflows of workers are not exploited for financial gain to the detriment of the migrants, and maximising the flows and uses of remittances for migrants, their originating households, communities and the country as a whole. Within the constraints on resources and time, this research study has provided important insights about the socioeconomic impact of remittances in rural Bangladesh. The following recommendations are offered to enhance the socioeconomic impacts of foreign remittances: 1. Bangladesh is an over-populated country with severe unemployment and underemployment problems. The niche markets for labour in international destinations must be carefully studied and analysed in support of proactive labour policies that will reduce unemployment in Bangladesh, all the while maximising the flow of remittances into the country. 2. Given the very low saving rates in Bangladesh on account of widespread poverty, the inflow of foreign remittances can be managed to mobilise savings in attractive in-country saving schemes. 3. National policies should be formulated to promote a healthy and attractive investment climate while also providing incentives and inducements to migrants and their households to invest in viable commercial and entrepreneurial opportunities in Bangladesh. For instance, facilities for the migrants to import machinery and equipment at concessionary duties and taxes could be offered domestically. 4. Guidance services should be provided to the migrants about the available investment opportunities, central bank saving facilities and capital market advisory services by the Bangladesh Bank and other financial institutions. Business and consultancy services can also be offered to the migrant workers and their households to establish viable businesses in productive sectors. 5. Minimal levels of education of migrants prevents them from securing attractive jobs in the destination country. Therefore, the government should consider establishing 180

202 vocational/professional training institutes in collaboration with the destination countries to provide recognised curriculums in occupations in high demand. 6. Unplanned investment or using agricultural land for housing sector investment has created enormous problems in the remain-behind households. The government should consider adopting or formulating better plans for housing developments for the remain-behind households so that they can acquire property and increase security levels in the migrant community. 7. Consideration or privilege should be given to the abolition or redemption of import duties and other taxes on capital goods to set up small and medium industries in Bangladesh and also to export to destination countries. 8. The government should consider establishing counselling centres to educate and advise youths in left-behind households to minimise the adverse effects of the absence of household heads. In rural Bangladesh, the policy recommendations given above may not be effective without instituting a decentralised government system and good governance balanced by economic and social development strategies with the participation and involvement of the rural households and migrants. 6.3 Research limitations At the questionnaire survey stage, the manpower brokers and their sub-agents tried to persuade the participant household members not to cooperate with the researcher as they claimed that the study would benefit neither household members nor migrants in Italy and would only benefit the researcher. Moreover, they claimed that those migrant workers in Italy who are undocumented could be affected by this study. The researcher, however, took measures to overcome these limitations by conducting meetings with influential people at different points to make the purpose of this research clear. The researcher also met the local people, teachers and villagers to convince them of the value of this study. This study did not consider variations in the age ranges of individuals like migrants, household heads and other household members to provide in-depth explanations about the determinants of remittances among different age cohorts. Other household-level characteristics like dependency ratios, number of daughters, number of school-going 181

203 children, quality of land and accessibility to agricultural facilities were also not considered as remittance determinants for different types of households. For an in-depth understanding of the female household head s vulnerabilities, research is required on social exclusion and gender relationships. This study, however, did not delve into this topic as it predominantly focuses on the economic perspectives of remittances. This research was based on household survey data from rural Bangladesh. Migrant information was limited by the non-presence of the migrants. Thus, important variables that could not be included in this study included migrant income, expenditure, employment status and living status in Italy. 6.4 Future research The research into remittances is vast and multi-dimensional. A number of critical issues identified in the course of research project could not be dealt with in greater depth because they were beyond the scope of this work and resource limitations. It is recommended that the following research issues be pursued: 1. The importance, role of financial institutions such as banks, insurance companies, and the Investment Corporation of Bangladesh in linking international remittances to economic development of the rural households and village communities. 2. The potential for using migrants savings in the form of ICB mutual or US dollar investment funds for the betterment and development of their communities. 3. The informal and formal methods and channels used in both countries for the flow of remittances could be analysed and improvements made to them as a result. 4. The macroeconomic determinants and impacts of remittances may be estimated and analysed as this is important information for Bangladesh as a whole. 5. Bangladesh government policies in terms present incentive schemes for the migrant households to invest remittances in productive sectors uses could be analysed. 6. The question of brain drain in Bangladesh owing to migration may be considered. This question is a vital issue for the future. 7. In-depth research should explore the gender perspectives of left-behind household members in rural Bangladesh. A gender-sensitive approach may be followed to understand 182

204 the socio-cultural impacts of remittances on female household heads or spouses remainbehind. 8. Research is needed to explore links between migration, remittances and rural development strategies and policies introduced by the government of Bangladesh. Thus, the study suggests more in-depth research is needed into dependent variables, such as determinants of remittance and socioeconomic impacts. 6.5 Conclusion International migration has major socioeconomic consequences, one of which is the flow of remittances from the migrant workers in different countries to their villages of origin. This constitutes a complex phenomenon whose dynamics have national and global economic, social, legal and cultural implications from both macro and micro perspectives. As a result, this study has focused on the factors affecting the flow and socioeconomic impacts of remittances sent by migrant workers in Italy to their originating rural households in the Shariatpur district of Bangladesh. This research explores the micro-economic determinants of international remittances and their socioeconomic impacts on their households in rural Bangladesh. A primary survey involving 300 households that had received remittances from Italy was undertaken between July and December 2013 in Bangladesh. This study developed an econometric model and tested the survey data using univariate and multivariate estimation (log linear regressions, Chi-square and Gamma test). The descriptive results demonstrate the socio-demographic characteristics of the head of household as well as migrant members in Italy. The survey output showed that the majority (59 per cent) of migrants are between 21 and 35 years old, their educational attainment is less than higher secondary (98 per cent), most are married (71 per cent), duration of migration is mostly between 5 and 9 years (56 per cent), most migrants were documented (83 per cent) and visited their original household two to five times (61 per cent). The heads of the households were mainly aged between 51 and 75 years old (57 per cent), 53 per cent were males, most were married (83 per cent), and 47 per cent of household heads were housewives. The household religions were Islam (68 per cent) and Hinduism (32 per cent), most of the households had between two and five members (51 per cent) excluding migrant members, household land was owned solely by the migrant himself in 47 per cent of cases, 183

205 and the majority of household property was to be inherited by 2 to 3 members in 83 per cent of cases. Key factors influencing remittance flow include migrant characteristics such as age and number of visits home by the migrant. The study found that younger household members have a higher propensity to migrate and that their remittance flows are higher, thus demonstrating a strong motivation to maintain kinship relations with the remainbehindhousehold members. Other significant influential determinants of remittance flow are household head age, gender and marital status; kinship relation of the head of household to the migrant; and household composition, namely development in the housing sector and standard of living expenditure. The study suggests that the determinants of remittances and the factors affecting socioeconomic impacts differ at the destination, community and household levels. 184

206 REFERENCES Adams, R 1989, Worker remittance and inequality in rural Egypt, Economic Development and Cultural Change, vol. 38, no. 1, pp Adams, R 2006, International remittances and the household: Analysis and review of global evidence, Journal of African Economies, vol. 15, no. 2, pp Adams, R 2007, International remittances and the household: analysis and review of global evidence, Policy Research Working Paper Series 4116, The World Bank. Adams, R 2011, Evaluating the economic impact of international remittances on developing countries using household surveys: a literature review, Journal of Development Studies, vol. 47, no. 6, pp Afsar, R, Yunus, M & Islam S 2002, Are migrants chasing after the golden deer?, International Organization for Migration (IOM), Regional Office South Asia. Agrawal, R & Horowitz, AW 2002, Are international remittances altruism or insurance? Evidence from Guyana using multiple-migrant households, World Development, vol. 30, no.11, pp Ahlburg, D & Brown, R 1998 Migrants intentions to return home and capital transfers: A study of Tongans and Samoans in Australia, Journal of Development Studies, vol. 35 no. 2, pp Alba, M & Sugui, Jessaine SC 2009, Motives and giving norms behind remittances the case of Filipino overseas workers and their recipient households UP School of Economics, Discussion Paper No Ali, BM, 2012, Macroeconomic factors influencing foreign remittances: the case of Bangladesh, International Journal of Academic Research in Accounting, Finance and Management Sciences, vol. 2, no. 3, pp Amuedo-Dorantes, C & Pozo, S 2006, Remittance receipt and business ownership in the Dominican Republic, World Economy, vol. 29, no. 7, pp Ansari AYDY 2014, Innovation practices as a path to business growth performance: a study of small and medium sized firms in the emerging UAE market, PhD Thesis, Southern Cross University, Australia. 185

207 Anthias, F & Lazaridis, G (eds) 2000, Gender and migration in southern Europe, Berg, New York. Argyrous, A 2000, Statistics for health and social research, with a guide to SPSS, Sage, London. Atekmangoh, C 2011, Expectations abound-family obligations and remittance flow amongst Cameroonian-busfallers-in Sweden-Agender insight, Masters Thesis, Lund University, Graduate school of social science. Atamanov, A & Berg, MV 2012, Heterogeneous effects of international migration and remittances on crop income: evidence from the Kyrgyz Republic, World Development, vol. 40, no. 3, pp Azhar, KIA 2008, Overseas migration and its socio-economic impacts on the families left behind in Pakistan: a case study in the province of Punjab, Pakistan, PhD Thesis, Faculty of Economics, University of Kassel. Azam, JP & Gubert, F 2006, Migrants remittances and the household in Africa: A review of evidence, Journal of African Economies, vol. 15, no. 2, pp Bagozzi, RP 1994, Principles of marketing research, Blackwell Publishing, Cambridge. Bagozzi, RP 1997, Advance methods of marketing research, Blackwell Publishing, Oxford. Baker, MJ 2000, Writing a literature review, The Marketing Review, vol. 1, no. 2, pp Bamberger IM 2000, Integrating quantitative and qualitative research in development projects, Washington DC, World Bank. Banerjee, B 1984, The probability, size and uses of remittances from urban to rural areas in India, Journal of Development Economics, vol. 16, no. 3, pp Bangladesh Bank 2014, Economic data, Central Bank of Bangladesh, the People's Republic of Bangladesh, Official link: Bangladesh Bank 2011, Bangladeshe Remittance ar Babohar: Vobishat Diknirdeshana,Bangladesh Bank, The People s Republic of Bangladesh. Barham, B & Boucher, S 1998, Migration, remittances and inequality: estimating the net effects of migration on income distribution, Journal of Development Economics, vol. 55, no.3, pp

208 Barclay, D, Higgins, C & Thompson, R 1995, The partial least squares (PLS) approach to causal modeling: Personal computer adoption and use as an illustration, Technology Studies, vol. 2, no. 2, pp Bartlett, JE, Kotrlik, JW & Higgins, CC 2001, Organizational research: Determining appropriate sample size in survey research, Information Technology, Learning and Performance Journal, vol. 19, no. 1, pp Barua, S, Majumder, A & Akhtaruzzaman, M 2007, Determinants of workers remittances in Bangladesh: an empirical study, Bangladesh Bank, Dhaka. A Policy Analysis Unit Working Paper Series No. WP BBS 2012, Community report, Shariatpur Zila, Bangladesh Bureau of Statistics, Government of the People s Republic of Bangladesh, June. Becker, GS 1962, Investment in Human Capital: A Theoretical Analysis, Journal of Political Economy, vol. 70, no. 5, pp Beaudouin, P 2006, Economic impact of migration on a rural area in Bangladesh, Center deconimie de la Sorbonne, Working Paper. Berg, BL 2004, Qualitative research methods, Pearson Education, Boston. Bettin, G & Lucchetti, R 2012, Intertemporal remittance behaviour by immigrants in Germany, MOFIR Working Paper No. 75, October. Bettin, G, Presbitero, AF & Spatafora, N 2014, Remittances and vulnerability in developing countries, Mo.Fi.R. Working Papers 93, Money and Finance Research group (Mo.Fi.R.) - Univ. Politecnica Marche - Dept. Economic and Social Sciences. Black, R, King, R & Tiemoko, R 2003, Migration, return and small enterprise development in Ghana: a route out of poverty?,international Workshop on Migration and Poverty in West Africa, March 13-14, University of Sussex. Blangiardo, GC 2009, Foreigners presence in Italy: from a reference framework to future development scenarios in Vincenzo Cesareo (ed), The fifteenth Italian report on migrations, Polimetrica, Milan. Blue, SA 2004, State policy, economic crisis, gender, and family ties: Determinants of family remittances to Cuba, Economic Geography, vol. 80 no. 1, pp

209 BMET 2013, Statistics, Bureau of Manpower, Employment and Training, Government of the People's Republic of Bangladesh, Official link: Bonoma, TV 1985, Case research in marketing: opportunities, problems, and a process,journal of Marketing Research, vol. 22, no. 2, pp Bonifazi, C, Okolski, M, Schoorl, J & Simon, P (eds) 2008, International Migration in Europe: New trends and new methods of Analysis, Amsterdam University Press, Amsterdam. Borja, K 2012, What drives remittances to Latin America? A review of the literature, International Journal of Business and Social Science, vol. 3, no. 1, pp Brown, R, Connell, J & Jimenez-Soto, EV 2014, migrants' remittances, poverty and social protection in the South Pacific: Fiji and Tonga, Population, Space and Place, vol. 20, no. 5, pp Brown, RPC 2011, Subjectively-assessed welfare and international remittances: evidence from Tonga, The Journal of Development Studies, vol. 47, no. 6, pp Brown, RPC 1997, Estimating remittance functions for Pacific Island migrants, World Development, vol. 25, no. 4, pp Brown, RPC 1998, Do migrants remittances decline over time-evidence from Tongans and Western Samoans in Australia? The Contemporary Pacific, vol. 10, no. 1, pp Bryman, A 2004, Social research methods, Oxford University Press, New York. Bryman, A & Cramer, D 2000, Qualitative data analysis for social scientists, London, Routledge. Burns, AC & Bush, RF (eds) 2006, Marketing research, Pearson Education, Upper Saddle River. Busetta, A, Cetorelli, V & Stranges, M 2013, Remittance behaviours of foreigners in Italy, XXVII IUSSP International Population Conference, Caces, F, Arnold, F, Fawcett, JT, & Gardner, RW 1985, Shadow households and competing auspices: Migration behavior in the Philippines, Journal of Development Economics, vol. 17, no. 1, pp

210 Cadge, W & Ecklund, E 2006, Religious service attendance among immigrants: evidence from the new immigrant survey-pilot, American Behavioral Scientist, vol. 49, no. 11, pp Cadge, W & Ecklund, E 2007, Immigration and religion, Annual Review of Sociology, vol. 33, pp Cai, Q 2003, Migrant remittances and family ties: a case study in China, International Journal of Population Geography, vol. 9, no. 6, pp Carling, C, Bedi, AS & Sparrow, R 2008, The determinants of migrant remittances, World Development, vol. 37, no. 6, pp Carney, D 1998, Sustainable livelihood. What difference can we make?,paper presented at the Department of International Development s Natural Resources Advisers Conference, July, London. Casale, D & Dori, P 2006, Migration and remittances in South Africa: Background document on migration and first set of draft questions for inclusion in the National Income Dynamics Study, South African Labour and Development Research Unit, University of Cape Town. Cavana, R, Delahaya, B & Sekaran, U 2001, Applied business research: qualitative and quantitative methods, Wiley, Milton. Chaloff, J 2006, Innovating in the supply of services to meet the needs of immigrants in Italy in OCED, From immigration to integration: local solutions to a global challenge, Organization for Economic Co-operation and Development, Paris. Chowdhury, KR, Hamid, F & Chatterjee, DD 2010 Remittance as a tool of economic development: Bangladesh perspective, Bangladesh Research Publications Journal, vol. 4, no. 3, p Chowdhury, MB 2011, Remittances flow and financial development in Bangladesh, Economic Modelling, vol. 28, no. 6, pp Chin, WW & Newsted, PR 1999, Structural equation modeling analysis with small samples using partial least squares, in R Hoyle, ed. Statistical strategies for small sample research, Sage Publications, Thousand Oaks, pp

211 Churchill, GA (eds) 1999, Marketing research: methodological foundations, Dryden Press, Forth Worth. Collier, W Piracha, M & Randazzo, T 2011, Remittances and return migration, Discussion Paper Series, Forschungsinstitut zur Zukunft der Arbeit, No Coulibaly, D 2009, Macroeconomic determinants of migrants remittances: new evidence from a panel VAR, Documents de travail du Centre d Economie de la Sorbonne 09007, ISSN X. Cox, D 1987, Motives for private transfers, Journal of Political Economy, vol. 95, no. 3, pp Cox, D, Eser, Z & Jimenez, E 1998, Motives for private transfers over the life cycle: An analytical framework and evidence for Peru, Journal of Development Economics, vol. 55, no. 1, pp Cooper, D & Schindler, P (eds) 2003, Business research methods, McGraw-Hill, New York. Cooper, D & Schindler, P (eds) 2006, Business research methods, 9th edn, McGraw-Hill, Boston. Crotty, M 1998, The foundations of social research: meaning and perspective in the research process, Sage Publications, London. Creswell, JW (eds) 2003, Research design, Sage Publications, Thousand Oaks. Creswell, JW & Clark, VLP 2007, Designing and conducting mixed methods research, Sage Publications, California. Dakila, F & Claveria, R 2007, Identifying the determinants of overseas Filipinos remittances: Which exchange rate measure is most relevant?, Center for Monetary and Financial Policy. BSP Working Paper Series No Davis, B & Stecklov, G 2002, Domestic and international migration from rural Mexico: Disaggregating the effects of network structure and composition, Population Studies, vol. 56, no. 3, pp Davis, D (eds) 2005, Business research for decision making, Thomson South-Western, Mason. Davis, D & Cosenza, RM 1993, Business research for decision making, Wadsworth, Belmont. 190

212 de Haas, H 2007, Remittances, migration and social development, a conceptual review of the literature, United Nations Research Institute for Social Development, Social Policy and Development, Programme Paper Number 34. de Haas, H 2006, Migration, remittances and regional development in Southern Morocco, Geoforum, vol. 37, no. 4, pp de Vaus, D 2002, Analyzing social science data, Sage Publications, Thousand Oaks. de Jong, GF, Root, BD & Abad, RG 1986, Family reunification and Philippine migration to the United States: The immigrants' perspective, International Migration Review, vol. 20, no. 3, pp de la Briere, B, Janvry, A, Lambert, S & Sadoulet, E 1997, Why do migrants remit? An analysis for Dominican Sierra, FCND Discussion Paper No.37. de la Brière, B, Sadoulet, E, Janvry, A & Lambert, S 2002, The roles of destination, gender, and household composition in explaining remittances: An analysis for the Dominican Sierra, Journal of Development Economics, vol. 68, no. 2, pp Denzin, NK 1978, The research act: A theoretical introduction to sociological methods, 2nd edn, McGraw-Hill, New York. Deshpande, R 1983, Paradigms lost: On theory and method in research in marketing, Journal of Marketing, vol. 47, no. 4, pp DeVoretz, DJ & Vadean, FP 2008, Cultural differences in the remittance behaviour of households: evidence from Canadian micro data, University of Kent, Department of Economics Discussion Papers, November, KDPE Dinerman, IR 1978, Patterns of adaptation among households of US-bound migrants from Michoacan, Mexico, International Migration Review, vol. 12, no. 4, pp Diener, E & Crandall, R 1978, Ethics in social and behavioral research, University of Chicago Press, Chicago. Dillman, DA (eds) 2007, Mail and internet survey: the tailored design method, Wiley, Hoboken. Durand, J, Kandel, W, Parrado, EA & Massey, DS 1996, International migration and development in Mexican communities, Demography, vol. 33, no. 2, pp

213 Easterby-Smith, M, Thorpe, R & Lowe, A 1991, Management research: an introduction, Sage, London. Ellison, CG 1991, Religious involvement and subjective well-being, Journal of Health and Social Behavior, vol. 32, no. 1, pp El-Sakka, MIT & Mcnabb, R 1999, The macroeconomic determinants of emigrant remittances, World Development, vol. 27, no. 8, pp Emory, CW & Cooper, DR 1991, Business Research Methods, Irwin, Boston. Escobar, A, GonzÃilez, M, & Roberts, B 1987, Migration, labour markets, and the international economy: Jalisco, Mexico, and the United States, in Eades, J Migrants, workers, and the social order,association of Social Anthropologists [ASA], 42-64, London, United Kingdom. Fairchild, ST & Simpson, NB 2004, Mexican migration to the United States Pacific Northwest, Population Research and Policy Review, vol. 23, no. 3, pp Fawcett, JT & Fred A 1987, Explaining diversity: Asian and Pacific immigration systems, in Fawcett, J and Carino, B (eds) in Pacific bridges: the new immigration from Asia and the Pacific Islands. Staten Island, NY: Center for Migration Studies. pp Feinerman, E & Seiler, EJ 2002, Private transfers with incomplete information: A contribution to the "Altruism-exchange motivation for transfers" debate, Journal of Population Economics, vol. 15, no. 4, pp Firestone, WA 1993, Alternative arguments for generalising from data as applied to qualitative research, Educational Researcher, vol. 22, no. 4, pp Fonchamnyo, DC 2012, The altruistic motive of remittances a panel data analysis of economies in Sub Saharan Africa, International Journal of Economics and Finance, vol. 4, no. 10, pp Frank, AG The development of underdevelopment, Monthly Review, vol. 18, September, pp Funkhouser, E 1995, Remittances from international migration: a comparison of El Salvador and Nicaragua, The Review of Economics and Statistics, vol. 77, no. 1, pp

214 Gerber, A, Gruber, J & Hungerman, D 2008, Does church attendance cause people to vote? Using Blue Laws repeal to estimate the effect of religiosity on voter turnout, NBER Working Paper Series. Working Paper Garip, F 2006, Social and economic determinants of migration and remittances: An analysis of 22 Thai villages, Presented at the Annual Meeting of the Population Association of America, Los Angeles, California, March 30- April 1. Garip, F 2012, An Integrated Analysis of migration and remittances-modeling migration as a mechanism for selection, Population Research & Policy Review, vol. 31, no. 5, pp Gay, LR & Diehl, PL 1992, Research methods for business and management, Palgrave MacMillan, New York. Germenji, E, Beka, I & Sarris, A 2001, Estimating remittance functions for rural-based Albanian emigrants, Working paper, ACE research project, P R. Glytsos, NP 1988, Remittances in temporary migration: a theoretical model and its testing with the Greek-German Experience, Weltwirtschaftliches Archiv, Bd. 124, H. 3, pp Goldring, L 1998, The power and status of transnational social fields, Comparative Urban and Community Research, vol. 6, pp Grabel, GK 2008, The political economy of remittances -what do we know-what we need to know, ZEI working paper, No. B Green, SB 1991, How many subjects does it take to do a regression analysis? Multivariate Behavioural Research, vol. 26, no. 3, pp Guba, EG, & Lincoln, YS 1994, Competing paradigms in qualitative research, in NK Denzin & YS Lincoln (eds) Handbook of Qualitative Research, Sage, Thousand Oaks, pp Gubert, F 2002, Do migrants insure those who stay behind? Evidence from the Kayes Area (Western Mali), Oxford Development Studies, vol. 30, no. 3, pp Gujarati, DN 2003, Basic econometrics, 4th edn, New York, McGraw-Hill. Gummesson, E (ed) 2000, Qualitative methods in management research, Sage, Thousand Oaks. 193

215 Habibullah, AK 1980, World Bank survey on the export of manpower from Bangladesh to Middle East, World Bank, March. Hagen-Zanker, J & Siegel, M 2007, The determinants of remittances: A review of the literature, Working Paper MGSoG/2007/WP003. Hagen-Zanker, J & Siegel, M 2008, A critical discussion of the motivation to remit in Albania and Maldova, CMR Working Papers No. 30/88. Haenlein, M & Kaplan, A 2004, A beginner s guide to partial least squares analysis, Understanding Statistics, vol. 3, no. 4, pp Harris, JR & Todaro, MP 1970 Migration, unemployment and development: a two-sector analysis, American Economic Review, vol. 60, no. 1, pp Harbison, SF 1981, Family structure and family strategy in migration decision making, In G. de Jong, & R Gardner (eds) Migration decision making: Multidisciplinary approaches to microlevel studies in developed and developing countries, Pergamon Policy Studies, pp Hair, JF, Black, WC, Babin, BJ, Anderson, RE & Tatham, RL (eds) 2006, Multivariate data analysis, Pearson Education, Upper Saddle River. Hasan, G, Shamim, S & Mohammed, B 2012, Nonlinear growth effect of remittances in recipient countries an econometric analysis of remittances-growth nexus in Bangladesh, Paper presented at 53rd Annual NZAE Conference, Palmerston North Conference Centre, Palmerston North, June. Hay, MJ 1980, A structural equations model of migration in Tunisia, Economic Development and Cultural Change, vol. 28, no. 2, pp Heron, J & Reason, P 1997, A participatory inquiry paradigm, Qualitative Inquiry, vol. 3, no. 3, pp Hocking, RR 1976, The analysis and selection of variables in linear regression, Biometrics, vol. 32, no. 1, pp Hoelter, JW 1983, The analysis of covariance structures: Goodness-of-fit indices, Sociological Methods and Research, vol. 11, no. 3, pp

216 Holst, E & Schrooten, M 2006, Migration and money: what determines remittances? Evidence from Germany, Discussion Paper Series A No.477, The Institute of Economic Research, Hitotsubashi University. Holst, E, Schäfer, A & Schrooten, M 2008, Gender, migration, remittances-evidence from Germany, Discussion papers, German Institute for Economic Research, No Holst, E, Schäfer, A & Schrooten, M 2010, Gender, transnational networks and remittances: evidence from Germany, Discussion papers, German Institute for Economic Research, No Holst, E, Schäfer, A & Schrooten, M 2011, Remittances and gender: theoretical considerations and empirical evidence, IZA DP No Hoddinott, J 1994, A model of migration and remittances applied to western Kenya, Oxford Economic Papers, New Series, vol. 46, no. 3, pp Hussey, J & Hussey, R 1997, Business research: a practical guide for undergraduate andpostgraduate students, Palgrave MacMillan, London. Hunt, S, Sparkman, R & Wilcox, J 1982, The pre-test in survey research: Issues and preliminary findings, Journal of Marketing Research, vol. 19, no. 2, pp Itzigsohn, J 1995, Migrant remittances, labour markets, and household strategies: a comparative analysis of low-income household strategies in the Caribbean Basin, Social Forces, vol. 72, no. 2, pp IOM 2009, The Bangladesh household remittance survey 2009, Regional Office for South Asia, International Organization for Migration (IOM). Jack, JC 1916, The economic life of a Bengal district, London, UK, Oxford Clarendon Press. Johnson G, & Whitelaw 1974, Urban-rural income transfers in Kenya: an estimated remittances function, Economic Development and Cultural Change, vol. 22, no. 3, pp Johnson, RA & Wichem, DW 2002, Applied multivariate statistical analysis, Upper Saddle River, Prentice-Hall. Juhdia, N, Fatimah, PF & Hansaramb, RMK 2013, HR practices and turnover intention: the mediating roles of organizational commitment and organizational engagement in a 195

217 selected region in Malaysia, The International Journal of Human Resource Management, vol. 24, no. 15, pp Kanbur, R & Rapoport, H 2005, Migration selectivity and the evolution of spatial inequality, Journal of Economic Geography, vol. 5, no. 1, pp Kalzuny, RL 1975, Determinants of household migration: a comparative study by race and poverty level, Review of Economics and Statistics, vol. 57, no. 3, pp Kapur, D 2003, Remittances: the new development mantra?, United Nations Conference on Trade and Development, Geneva. Kelly, CS & Solomon B 2009, The influence of religion on remittances sent to relatives and friends back home, Journal of Business & Economics Research, vol. 7, no. 1, pp Kemper, RV 1991, Obstacles and Opportunities: Household Economics of Tzin-tzuntean migrants in Mexico City, Urban Anthropology, vol. 10, no. 3, pp Keating, M, & Wilson, A 2010, Federalism and Decentralisation in Italy, PSA Conference, Edinburgh, March-April. King, R (Ed.) 2001, Mediterranean passage: migration and new cultural encounters in Southern Europe, Liverpool University Press, London. King, R, Lazaridis, G & Tsardanidis, C (eds) 2000, Eldorado or fortress?, Migration in Southern Europe, Macmillan Press Ltd, New York. Kinnear, TC & Taylor JR (eds) 1996, Marketing research: an applied approach, McGraw- Hill, New York. Knights, M 1996, Bangladeshi immigrants in Italy: from geopolitics to micropolitics, Transactions of the Institute of British Geographers, New Series, vol. 21, no. 1, pp Krejcie, RV & Morgan, DW 1970, Determining sample size for research activities, Educational and Psychological Measurement, vol. 30, no. 3, pp Kvale, S 1989, Issues of validity in qualitative research, Studentlitteratur, Lund, Sweden. Lancaster, GA, Dodd, S & Williamson, PR 2002, Design and analysis of pilot studies: recommendations for good practice, Journal of Evaluation in Clinical Practice, vol. 10, no. 2, pp

218 LeCompte, M & Goetz, J 1982, Ethnographic data collection in evaluation research, Educational Evaluation and Policy Analysis, vol. 4, no. 3, pp Lee, ES 1966, A theory of migration, Demography, vol. 3, no. 1, pp Lerch, M, Dahinden, J & Wanner, P 2006, Remittance behaviour of Serbian migrants living in Switzerland a survey, State Secretariat for Economic Affairs (SECO) of the Government of Switzerland. Lewis, AW 1954, Economic development with unlimited supplies of labour, Manchester School of Economic and Social Studies, vol. 22, pp Lillard, LA & Wills, RJ 1997, Motives for intergenerational transfers-evidence from Malaysia, Demography, vol. 34, no. 1, pp Lincoln, YS & Guba EG 1985, Naturalistic inquiry, Sage, Newbury Park. Luke, N 2010, Migrants' competing commitments-sexual partners in urban Africa and remittances to the rural origin, American Journal of Sociology, vol. 115, no. 5, pp Lucas, REB & Stark, O 1985, Motivations to remit: evidence from Botswana, Journal of Political Economy, vol. 93, no. 5, pp Lucas, REB 1987, Models of business cycles, Oxford, Basil Blackwell. Lueth, E & Ruiz-Arranz, M 2007, Are workers' remittances a hedge against macroeconomic shocks? The case of Sri Lanka, IMF Working Paper No Luck, DJ & Rubin, RS (eds) 1987, Marketing research, Prentice Hall International, London. Mabogunje, AL 1970, Systems approach to a theory of rural-urban migration Geographical Analysis, vol. 2, no. 1, pp Malhorta, NK (eds) 2004, Marketing research: an applied orientation, Pearson Prentice-Hall, Upper Saddle River, New Jersey. Malhotra, NK 2010, Marketing research: an applied orientation, 6th edn, Pearson Education, Upper Saddle River, New Jersey. Mannan, KA & Farhana, K 2014 Legal status, remittances and socio-economic impacts on rural household in Bangladesh: An empirical study of Bangladeshi migrants in Italy 197

219 International Journal of Management Sciences and Business Research, vol. 3, no. 8, pp Mannan, KA, Farhana, KM & Chowdhury, TA 2014 Rural household contribution to the financial and capital market in Bangladesh: a micro level study of remittances from Italy, International Journal of Management Sciences and Business Research, vol. 3, no. 10, pp Manning, M & Munro D (eds) 2006, The survey researcher s SPSS cookbook, Pearson Education Australia, Sydney. Manning, M & Munro D (eds) 2007, The survey researcher s SPSS cookbook, Pearson Education Australia, Sydney. Markova, E & Reilly, B 2007, Bulgarian migrant remittances and legal status some micro level evidence from Madrid, South-Eastern Europe Journal of Economics, vol. 5, no. 1 pp Marshall, C & Rossman, GB 1999, Designing qualitative research, 3rd edn, Sage Publications, Thousand Oaks, California. Maxwell, J 2002, Understanding and validity in qualitative research, in AM Huberman and MB Miles (eds) The qualitative research companion, Sage, Thousand Oaks, pp McDonald, J & Valenzuela, R 2012, Why Filipino migrants remit evidence from a homehost country matched sample, Monash University, Department of Economics, Discussion Papers No. 9/12. Mckenzie, D & Rapoport, H 2004, Network effects and the dynamics of migration and inequality: theory and evidence from Mexico, Working Papers , Bar-Ilan University, Department of Economics. Merkle, L & Zimmermann, K 1992, Savings, remittances, and return migration, Economic Letters, January, vol. 38, no. 1, pp Mendola, M 2004, Migration and technological change in rural households: complements or substitutes? Development Studies Working Paper, No. 195, University of Milan- Bicocca and Centro Studi Luca d Agliano. 198

220 Menjivar, C, DaVanzo, J, Greenwell, L & Valdez, B R 1998, Remittance behavior among Salvadoran and Filipino immigrants in Los Angeles, International Migration Review, vol. 32, no. 1, pp Miller, DC 1991, Handbook of research design and social measurement, 5th edn, Sage, London. Miles, MB & Huberman, AM (eds) 1994, Qualitative data analysis, Sage, Thousand Oaks. Miller, M & Kirk, J 1986, Reliability and validity in qualitative research, Sage, Newbury Park. Milanovic, B 1987, Patterns of regional growth in Yugoslavia, , Journal of Development Economics, vol. 25, no. 1, pp Mishra, P 2011, Motivations to remit-evidence from Chitwan, Nepal, Working Paper OPR , Office of Population Research, Princeton University. Mohapatra, S, Joseph, G & Ratha, D 2012, Remittances and natural disasters: ex-post response and contribution to ex-ante preparedness, Environment, Development and Sustainability, vol. 14, no. 3, pp Murshid, KSA, Iqbal, K & Ahmed, M 2002, A study on remittances inflow and utilization, International Organization for Migration (IOM), Regional Office for South Asia. Nachmias, D & Nachmias, CF 1992, Research methods in the social sciences, Volume 1, St. Martin's Press, Indiana University. Nakosteen, RA & Zimmer, M 1980, Migration and income: the question of self-selection, Southern Economic Journal, vol. 46, no. 3, pp Navratil, FJ & Doyle, J 1977 The socio-economic determinants of migration and the level of aggregation, Southern Economic Journal, vol. 43, no. 4, pp Ncube, M & Brixiova, Z 2013, Remittances and their macroeconomic impact: evidence from Africa, Working Paper 188, Working Paper Series 996, African Development Bank. Neuman, WL (eds) 2006, Social research methods, Pearson Education, Boston. Noblit, GW & Hare, RD 1988, Meta-ethnography: synthesizing qualitative studies, Sage, Newbury Park. 199

221 O Leary, Z 2005, Researchingreal world problems: a guide to methods of inquiry, Sage, London. Orozco, M & Ellis, CG 2014 Impact of Remittances in Developing Countries in Anich, R, Crush, J, Melde, S & Oucho, JO, eds. A new perspective on human mobility in the south, Global Migration Issues, vol. 3, pp Osili, UO 2007 Remittances and savings from international migration: theory and evidence using a matched sample, Journal of Development Economics, vol. 83, no. 2, pp Parkhe, A 1993, Messy research, methodological predispositions, and theory development in international joint ventures, Academy of Management Review, vol. 18, no. 2, pp Pfeiffer, L, Richter, S, Fletcher, P & Taylor, JE 2008, Gender and the Determinants of International Migration from Rural Mexico over Time, Chapter 3 in A.R. Morrison, M. Schiff and M. Sjöblom, eds. The International Migration of Women, pp Pfau, WD & Giang, LT 2008, Determinants and impacts of international remittances on household welfare in Vietnam, International Social Science Journal, vol. 60, no. 194, pp Perry, C 1995, A five chapter PhD thesis, Research Workshop Handout, Southern Cross University, Lismore, Australia. Perry, C 2002, A structured approach to presenting thesis: Notes for students and their supervisors, Study Guide, Southern Cross University, Lismore. Perry, C, Riege, A & Brown, L 1999, Realism s role among scientific paradigms in marketing research, Irish Marketing Review, vol. 12, no. 2, pp Pessar, PR 1982, The role of households in international migration and the case of US-bound migration from the Dominican Republic, International Migration Review, vol. 16, no. 2, pp Peter, DB & Piore, MJ 1970, Internal Labor Markets and Manpower Analysis, Manpower Administration (DOL), Washington, D.C, Office of Manpower Research, Harvard Univ, Cambridge, Mass, Massachusetts Inst. of Tech, Cambridge, May. Petrie, A & Sabin, C 2000, Medical statistics at a glance, Blackwell Science, Osney Mead. 200

222 Phillips, EM & Pugh, DS 2005, How to get a PhD: a handbook for students and their supervisors, 4th edn, Open University Press, McGraw-Hill. Pleitez-Chavez, RA 2004, Remittances as a strategy to cope with systemic risk: panel results from rural households in El Salvador, PhD, Colombus: Economics, Ohio State University. Poirine, B 1997, A Theory of Remittances as an Implicit Family Loan Arrangement, World Development, vol. 25, no. 5, pp Piper, N 2005, Gender and migration, A paper prepared for the Policy Analysis and Research Programme, of the Global Commission on International Migration, September. Ponsot, F & Obegi, B 2010, Etude de capitalisation des initiatives et mécanismes en matire de transferts de fonds au Mali, Study conducted for Centre de Information et de Gestion des Migrations (CIGEM), Mali. Poveda, AR 2007, Determinants and consequences of internal and international migration: The case of rural populations in the south of Veracruz, Mexico, Demographic Research, vol. 16, no. 10, pp Punch, KF 2005, Introduction to social research: quantitative and qualitative approaches, Sage, London. Quartey, P 2006, The impact of migrant remittances on household welfare in Ghana, African Economic Research Consortium, AERC No.158, Nairobi, August. Rahman, MM & Kabir, MA 2012, Moving to Europe Bangladeshi migration to Italy, ISAS Working Paper, No Rahman MM 2011, Does labour migration bring about economic advantage? A case of Bangladeshi Migrants in Saudi Arabia, ISAS Working Paper, No Rahman, MM 2007, Migration and social development: a family perspective, Working Paper, Asia Research Institute, National University of Singapore. Rapoport, H & Doquier, F 2005, The economics of migrants- remittances, IZA Discussion paper series, No Raihan, S 2010, Impact of the global crisis in migration and remittances in Bangladesh a survey based analysis, MPRA Paper No

223 Ranis, G & Fei, JCH 1961, A theory of economic development, American Economic Review, vol. 51. no. 4, pp Ranis, G 2008, Relationships between migration and development, SSRC Migration & Development Conference Paper No. 3, 28 February-1 March, New York. Rapoport, H & Docquier, F 2005, the economics of migrants' remittances, IZA Discussion Paper Series No Ratha, D, Mohapatra, S & Scheja, E 2011, Impact of migration on economic and social development: a review of evidence and emerging issues, World Bank Policy Research Working Paper Series. Regmi, G & Tisdell, C 2002, Remitting behaviour of Nepalese rural-to-urban migrants: implications for theory and policy, The Journal of Development Studies, vol. 38, no. 3, pp Render, B & Stair, RM (eds) 1994, Quantitative analysis for management, Prentice-Hall, Englewood Cliffs. Roberts, KD & Morris, MDS 2003, Fortune, risk, and remittances an application of option theory to participation in village based migration networks, International Migration Review, vol. 37, no. 4, pp Rodriguez, ER 1996, Migrants remittances in the Philippines, The Canadian Journal of Economics / Revue canadienne d Economique, vol. 29, no. 2, pp Roscoe, JT (eds) 1975, Fundamental research statistics for the behavioural sciences, Holt Rinehart & Winston, New York. Rusconi, S 2010 Italy s migration experiences, Network Migration in Europe E.V. viewed 1 January 20014, Sackey, FG 2010, Microeconomic determinants of private inward remittances to households in Ghana: a case study, Journal of Economics and Sustainable Development, vol. 2, no. 2, pp Sahu, GB & Das, B 2009, How does remittance market function? Evidence from Surat city, 11th Annual Conference on Money and Finance in the Indian Economy, held at the Indira Gandhi Institute of Development Research, (IGIDR), Mumbai, 23rd to 24th January. 202

224 Sandelowski, M 2000, Focus on research methods: Combining qualitative and quantitative sampling, data collection, and analysis techniques in mixed-method studies, Research in Nursing and Health, vol. 23, no. 3, pp Sarwar, GM, Islam, R & Monzoor, S 2007, Women s right to land in Bangladesh: roles, limitations and transformation, Unnayan Onneshan (The Innovators), Dhanmondi, Dhaka, August. Sarantakos, S 2005, Social research, Palgrave Macmillan, New York. Saunders, M, Lewis, P & Thornhill, A 2000, Research methods for business students, Pearson Education, London. Schiopu, L & Siegfried N 2006, Determinants of workers remittances: evidence from the European neighboring region, Working Paper Series No. 688, European Central Bank. Schmink, M 1984, Household economic strategies: review and research agenda, Latin American Research Review, vol. 19, no. 3, pp Sekaran, U (ed) 2000, Research methods for business: a skill-building approach, Wiley, New York. Sekaran, U (ed) 2003, Research methods for business: a skill-building approach, Wiley, New York. Seymour, D & Rooke, J 1995, The culture of the industry and the culture of research, Construction Management and Economics, vol. 13, no. 6, pp Shariatpur 2014, Shariatpur district, the People Republic of Bangladesh, Official link: viewed 2nd February 20012, Siddiqui, T & Abrar, CR 2003, Migrant workers remittances and micro-finance in Bangladesh, Social Finance Programme, International Labour Office, September, Working paper No. 38. Siddiqui, T 2004, Efficiency of migrant workers remittances the Bangladesh case, Asian Development Bank, Manila. Simati, AM & Gibson,J 2001, Do remittances decay? Evidence from Tuvaluan Migrants in New Zealand, Pacific Economic Bulletin, vol. 16, no. 1, pp

225 Singh, RJ, Lee, K & Haacker, M 2009, Determinants and macroeconomic impact of remittances in Sub-Saharan Africa, IMF Working Papers 09/216, International Monetary Fund. Sjaastad, LA 1962, The costs and returns of human migration, Journal of Political Economy, vol. 70, no. 5, pp Slatten T, Svensson, G & Svaeri, S 2011, Service quality and turnover intentions as perceived by employees: Antecedents and consequences, Personnel Review, vol. 40 no. 2, pp Sobh, R & Perry, C 2006, Research design and data analysis in realism research, European Journal of Marketing, vol. 40, no. 11, pp Solimano, A 2003, Remittances by emigrants-evidence and Issues, United Nations, October, SOPEMI 2010, International migration outlook, OECD, Paris. Sorenson, NN 2004a, The development dimension of migrants transfers, Danish Institute for International Studies, DIIS Working Paper no 2004/16. Sorenson, NN 2004b, Migrant transfers development tool Morocco, Danish Institute for International Studies, DIIS Working Paper 2004/17. Sorenson, NN 2005, The development dimension of migrants remittances towards a gendered typology, international forum on remittances IDB Conference, Centre June 28-30, Washington DC. Sprenger, E 2013, The determinants of international migration in the European Union: an empirical analysis, IOS Working Papers, No Stahl, CW & Habib, A 1989, The impact of overseas workers remittances on indigenous industries evidence from Bangladesh, Developing Economics, vol. 27, no. 3, pp Stevanovic, N 2012, Remittances and moral economies of Bangladeshi New York immigrants in light of the economic crisis, PhD thesis, Columbia University. Stark, O 1991, The migration of labor, Oxford, Basil Blackwell. 204

226 Stark, O 1995, Return and dynamics: the path of labor migration when workers differ in their skills and information is asymmetric, Scandinavian Journal of Economics, Wiley Blackwell, vol. 97, no. 1, pp Stark, O & Bloom, E 1985, The new economics of labour migration, American Economic Review, vol. 75, no. 2, pp Stark, O & Lucas, REB 1988, Migration remittances family, Economic Development and Cultural Change, vol. 36, no. 3, pp Stark, O, Taylor, J. E & S. Yitzhaki 1986, Remittances and inequality, Economic Journal, vol. 96, no. 383, pp Stark, O & Taylor 1989, Relative deprivation and international migration, Demography, vol. 26, no. 1, pp Stark, O & Wang, YQ 2002, Migration dynamics, Economics Letters, vol. 76, no. 2, pp Steenkamp, JB & Van Trijp, HC 1991, The use of LISREL in validating marketing constructs, International Journal of Research in Marketing, vol. 8, no. 4, pp Strauss, A & Corbin, J (eds) 1998, Basic of qualitative research, techniques, and procedures for developing grounded theory, Sage, Thousand Oaks, California. Tanur, JM (ed.) 1992, Questions about questions: inquiries into the cognitive bases of surveys, Sage Foundation, New York. Tashakkori, A & Teddlie, C 1998, Mixed methodology: combining qualitative and quantitative approaches, Sage, Thousand Oaks, California. Tabachnick, BG & Fidell, LS (eds) 2001, Using multivariate statistics, Allyn & Bacon, Needham Heights. Taylor, JE 1992, Remittances and inequality reconsidered: Direct, indirect, and intertemporal effects, Journal of Policy Modeling, vol. 14, no. 2, pp Taylor, JE & Stark, O 1986, Differential migration, networks, information and risk, In O. Stark, (ed), Migration theory, human capital and development, Greenwich, CT JAI Press, pp

227 Taylor, E, Rozelle, S & de Brauw, A 2003, Migration and incomes in source communities: a new economics of migration perspective from China, Economic Development and Cultural Change, vol. 52, no. 1, pp Taylor, JE & Wyatt, TJ 1996, The shadow value of migrant remittances, income and inequality in a household and farm economy, Journal of Development Studies, vol. 32, no.6, pp Ticehurst, GW & Veal, AJ 1999, Business research methods: a managerial approach, Addison Wesley Longman, South Melbourne, Australia. Ticehurst, GW & Veal, AJ 2000, Business research methods: a managerial approach, Pearson Education, Frenchs Forest. TIP 2013, Trafficking in persons report 2013, U.S Department of State, viewed 1 March 20014, Todaro, MP 1969, A model of labor migration and urban unemployment in less developed countries, American Economic Review, vol. 59, no. 1, pp Ullah, AKM Ahsan 2007, Rationalizing migration: Bangladeshi migrant workers in Hong Kong and Malaysia, PhD Thesis, Department of Asian and International Studies, City University of Hong Kong. UN 2013, World migration in figures, UN-DESA and the OECD to the United Nations, High-Level Dialogue on Migration and Development, 3-4 October. Vadean, F 2007, Citizenship status and remittances: evidence from German micro data,asia Research Institute Working Paper No. 89. Vargas-Silva, Carlos 2006, The determinants of international remittances: evidence from the legalized population, Academy of Economics and Finance Papers and Proceedings no. 30, pp Vanwey, LK 2004, Altruistic and contractual remittances between male and female migrants and households in rural Thailand, Demography, vol. 41, no. 4, pp Vete MF 2005, The determinants of remittances amongst Tongans in Auckland, Asian and Pacific Migration Journal, vol. 4, no. 1, pp Walewski, M 2009, The influence of large remittance inflows on the economic behaviour of rural households. the case of Moldova, CASE Network E-briefs, 7/2009, June. 206

228 Wallerstein, I 1974, The modern world system: capitalist agriculture and the origins of the European world-economy in the sixteenth century, New York, Academic Press. Warner, RM 2008, Applied statistics: from bivariate through multivariate techniques, Sage, Thousand Oaks, California. Wells, P 1994, Ethics in business and management research, in VJ Wass & PE Wells (eds), Principles and practice in business and management research, Aldershot, Dartmouth, pp Westen, D & Rosenthal, R 2003, Quantifying construct validity: two simple measures, Journal of Personality and Social Psychology, vol. 84, vol. 3, pp World Bank 2014, Database, The World Bank, viewed 15 February 20014, Official link: Woodruff, C & Zenteno, R 2001, Remittances and microenterprises in Mexico, UCSD, Graduate School of International Relations and Pacific Studies Working Paper. Wouterse, F 2006, Survival or accumulation: migration and rural households in Burkina Faso, Wageningen University and Research Centre, Wageningen. Yang, D 2011, Migrant Remittances, Journal of Economic Perspectives, vol. 25, no. 3, pp Yang, D 2008, International migration, human capital, and entrepreneurship: evidence from Philippine migrants' exchange rate shocks, The Economic Journal, vol. 118, no. 528, pp Yang, D & Choi, H 2007, Are remittances insurance? Evidence from rainfall shocks in the Philippines, World Bank Economic Review, vol. 21, no. 2, pp Yang, D & Martinez, CA 2005, Remittance and poverty in migrants home area: evidence from the Philippines, in C. Ozden and M. Schiff (eds), International, migration, remittances, and the brain drain, World Bank, Washington, DC, pp Yeoh, BSA, Elspeth, G & Paul JB 2002, Migrations and family relations in the Asia Pacific Region, Asian and Pacific Migration Journal, vol. 11, no. 1, pp Yin, RK (ed) 1989, Case study research: design and methods, Sage, Newbury Park. Yin, RK (ed) 1994, Case study research: design and methods, Sage, Thousand Oaks. 207

229 Yin, RK (ed) 2002, Case study research: design and methods, Sage, Thousand Oaks, California. Zelinsky, W 1971 The Hypothesis of the Mobility Transition, Geographical Review, Vol. 61, No. 2, pp Zikmund, WG (ed) 1997, Business research methods, Dryden Press, Fort Worth. Zikmund, WG (eds) 2000, Business research methods, Dryden Press, Chicago. Zikmund, WG (eds) 2003, Business research methods, Thomson/South-Western, Cincinnati. Zikmund, WG, Babin, BJ, Carr, JC & Griffin M 2010, Business research methods, 8 th edn, South-Western. Zontini, E 2010, Transnational families, migration and gender: Moroccan and Filipino women in Bologna and Barcelona, Berghahn Books, Oxford. Zweig, D 2008, Returnees, diasporas, and failure: can governments benefit from skilled out migration?,ssrc Migration & Development, Conference Paper No. 5, 28 February 1 March, New York. 208

230 APPENDIX A Data normality test 209

231 1. Remittance log 2. Age of migrant 210

232 3. Education of migrant 4. Marital status of migrant 5. Year of migration 211

233 6. Legal status of migrant 212

234 7. Number of visits by the migrant 8. Age of household head 213

235 9. Gender of household head 10. Marital status of household head 214

236 11. Education level of household head 12. Religion of household head 215

237 13. Employment status of household head 14. Household head relation to migrant 216

238 15. Household size 16. Household land ownership 217

239 17. Remittance investment in financial sectors 18. Remittance investment in house development 218

240 19. Log remittance in household welfare expenses 20. Loghousehold total land 219

241 21. Log household yearly living expenditure 22. Remittance investment in livestock 220

242 23. Log household yearly income 24. Remittance investment in business 221

243 25. Loan repayment from remittance 26. Remittance investment in buying land 222

244 APPENDIX-B THE SURVEY QUESTIONNAIRE Determinants and socioeconomic impact of foreign remittances sending from Italy on rural Bangladeshi households: A study of Naria Upazila, Shariatpur District. (To be filled by household head respondent) Survey Questionnaire 1. Please complete the following table? 223

245 Coding for table-1 224

246 2. Could you please explain your household expenditure at following table? 225

247 226

248 Thank you very much 227

249 APPENDIX-C SURVEY INVITATION 228

250 229

251 230

252 APPENDIX-D INFORMATION SHEET 231

253 232

254 233

255 234

256 235

257 236

258 237

259 238