Citation for published version: Davies, S 2008, 'Essays on remittances in rural Malawi', Ph.D., University of Bath.

Transcription

1 Citation for publihed verion: Davie, S 2008, 'Eay on remittance in rural Malawi', Ph.D., Univerity of Bath. Publication date: 2008 Link to publication Univerity of Bath General right Copyright and moral right for the publication made acceible in the public portal are retained by the author and/or other copyright owner and it i a condition of acceing publication that uer recognie and abide by the legal requirement aociated with thee right. Take down policy If you believe that thi document breache copyright pleae contact u providing detail, and we will remove acce to the work immediately and invetigate your claim. Download date: 15. Nov. 2018

2 ESSAYS ON REMITTANCES IN RURAL MALAWI Simon Davie A thei ubmitted for the degree of Doctor of Philoophy in Economic Department of Economic and International Development Univerity of Bath September, 2008 COPYRIGHT Attention i drawn to the fact that copyright of thi thei ret with it author. Thi copy of the thei ha been upplied on condition that anyone who conult it i undertood to recognie that it copyright ret with it author and that no quotation from the thei and no information derived from it may be publihed without the prior written conent of the author. RESTRICTIONS ON USE Thi thei may be made available for conultation within the Univerity Library and may be photocopied or lent to other librarie for the purpoe of conultation. Signed: Simon Davie i

3 TABLE OF CONTENTS Table of content.. ii Lit of table...iv Lit of figure v Acknowledgement...vi Summary. vii Abbreviation. viii 1 CONTEXT AND BACKGROUND INTRODUCTION BACKGROUND INFORMATION ON REMITTANCES Evolution and importance of international remittance flow Microeconomic impact THE MALAWIAN CONTEXT Geography Hitory and ociety The economy A comparion of Malawi and her neighbour WHY STUDY REMITTANCES IN THE MALAWIAN CONTEXT? OBJECTIVES OF THE STUDY OUTLINE OF THE STUDY LITERATURE REVIEW INTRODUCTION THE THEORY OF REMITTANCES Summary Modelling altruim Modelling inurance Payment for ervice Mental accounting link MOTIVATIONS FOR REMITTING MENTAL ACCOUNTING AND HOW IT RELATES TO REMITTANCES Theoretical background Pychological extenion and evidence REMITTANCES AS INSURANCE Remittance a an ex-pot coping trategy Conumption moothing model Review of empirical evidence of hock and conumption moothing CONCLUSION APPENDIX DATA DESCRIPTION AND EXPLORATION INTRODUCTION MALAWI FAMILY TRANSFER PROJECT Characteritic Remittance flow MALAWI INTEGRATED HOUSEHOLD SURVEY Houehold characteritic Income Conumption Aet holding MALAWI COMPLEMENTARY PANEL SURVEY A COMPARISON OF THE DATA SETS CONCLUSION METHODOLOGY INTRODUCTION ii

4 4.2 ORDINARY LEAST SQUARES Correction PANEL DATA ANALYSIS PROBIT MODELS Random effect probit TOBIT MODELS Conumption function and Tobit model CONCLUSION WHY DO PEOPLE REMIT? MOTIVATIONS FOR REMITTING IN RURAL MALAWI INTRODUCTION METHODOLOGY Dicriminating hypothee Model pecification RESULTS AND ANALYSIS Repondent and their parent Repondent and their children Repondent and their ibling Comparing and contrating motivation AN EXTENSION: CHILD REMITTANCES BY GENDER AND TRIBE CONCLUSIONS AND FURTHER EXTENSIONS APPENDIX MENTAL ACCOUNTING AND REMITTANCES INTRODUCTION METHODOLOGY RESULTS AND ANALYSIS Evidence for mental accounting How are remittance ued? Are remittance behaviour changing? CONCLUSIONS APPENDIX REMITTANCES AS INSURANCE FOR HOUSEHOLD AND COMMUNITY SHOCKS INTRODUCTION METHODOLOGY RESULTS AND ANALYSIS Shock and likelihood of receiving remittance Do remittance inure conumption? A Focu on health hock AN EXTENSION: DO SHOCKS HAVE A PERSISTENT IMPACT ON CONSUMPTION? CONCLUSION AND FURTHER EXTENSIONS SUMMARY AND CONCLUSIONS SUMMARY OF RESEARCH Non-empirical finding Empirical finding Limitation of the tudy CONCLUDING REMARKS BIBLIOGRAPHY iii

5 LIST OF TABLES Table 1.1: Reported Ue of Remittance by Repondent of 2006 Malawi Migration Baeline Survey... 7 Table 1.2: Selected national indicator Malawi and urrounding countrie, Table 2.1: Summary of key finding from tudie on motivation to remit Table 2.2: Summary of key finding from tudie of the rik haring hypothei Table 3.1: Repondent characteritic Table 3.2: Parent characteritic Table 3.3: Children characteritic Table 3.4: Sibling characteritic Table 3.5: Incidence and value of remittance flow Table 3.6: Selected decriptive tatitic from Integrated Houehold Survey Table 3.7: Selected decriptive tatitic from Complementary Panel Survey Table 3.8: Percentage of houehold receiving remittance from different ource Table 3.9: Percentage of houehold uffering from hock Table 3.10: Correlation between hock and remittance receipt Table 5.1: Motivation for remitting for the Repondent-Parent dyad Table 5.2: Motivation for remitting for the Repondent-Child dyad Table 5.3: Motivation for remitting for the Repondent-Sibling dyad Table 5.4: Summary tatitic for Repondent-Sibling remittance flow Table 5.5: Summary tatitic for Repondent-Parent remittance flow Table 5.6: Summary tatitic for Repondent-Child remittance flow Table 5.7: Repondent-Parent remittance flow Table 5.8: Repondent-Children remittance flow Table 5.9: Repondent-Sibling remittance flow Table 5.10: A focu on Child to Repondent remittance flow Table 6.1: Summary tatitic for variable ued included in model Table 6.2: Robut OLS model. Dependent Variable: Total Conumption PAE Table 6.3: Conumption of different categorie of good from different income ource Male head Table 6.4: Conumption of different categorie of good from different income ource Female head Table 6.5: Conumption of different categorie of good from total income Male remittance receiver Table 6.6: Conumption of different categorie of good from total income Male non remittance receiver Table 6.7: Conumption of different categorie of good from total income Female remittance receiver Table 6.8: Conumption of different categorie of good from total income Female non remittance receiver Table 6.9: Marginal propenitie to conume out of different ource of income, derived from Tobit model Table 7.1: Summary tatitic for variable included in model Table 7.2: Potential econometric iue and correction or robutne tet undertaken Table 7.3: Percentage of houehold uffering health hock Table 7.4: Correlation between health hock and remittance receipt Table 7.5: Probit model etimating probability of receiving remittance from different ource203 Table 7.6: Robut OLS (pooled panel). Dependent variable: Change in log per capita conumption Table 7.7: Predicting health hock and excluding predictable hock Table 7.8: Food veru non-food conumption. Dependent variable: Change in log of per capita conumption iv

6 Table 7.9: Fixed Effect regreion. Dependent variable: Change in log of per capita conumption Table 7.10: Excluding the extreme poor. Dependent variable: Change in log of per capita conumption Table 7.11: Impact of ickne of different houehold member. Dependent variable: Change in log of per capita conumption Table 7.12: Health hock in male and female headed houehold. Dependent variable: Change in log of per capita conumption Table 7.13: The Peritence of Shock. Dependent Variable: Log Per Capita Conumption LIST OF FIGURES Figure 1-1: Map of Malawi Figure 1-2: Map of outhern Africa Figure 2-1: Graphical repreentation of difference between traditional and behavioural lifecycle model Figure 3-1: Potential remittance flow Figure 7-1: Mean Conumption Over Round (1) Figure 7-2: Mean Conumption Over Round (2) v

7 ACKNOWLEDGEMENTS I would like to thank my upervior, Johy Eaaw and Atanu Ghohray for their upport throughout my thei. In addition, I would like to thank other academic and non-academic member of taff at the Univerity of Bath who have upported me academically and peronally over the lat few year. In particular, Jame Copetake, Pete Dawon, Joe Divine, Tim Hink, John Hudon, Suan Johnon, Ajit Mihra and Bruce Morley have all helped me to broaden and challenge my idea, and provided guidance. Michelle Hick, Elaine Irvine, David Ramey and Maggie Smith have aited me in adminitrative matter during my time at the Univerity of Bath. Both my internal and external examiner, Colin Lawon and Julie Litchfield made uggetion during the viva voce which have improved the quality of thi thei. I alo owe a debt of gratitude to previou univeritie and lecturer. In particularly, Catia Montagna at the Univerity of Dundee and Matthieu Crozet at the Univerity of Pari 1 Panthéon-Sorbonne were of great inpiration during my time at thee intitution and have continued to upport me ince. Thi thei ha alo benefited from inight from Willie Kachaka of the Malawian National Statitic Office in Zomba, Malawi, Suan Watkin at the Malawi Social Network Centre at Univerity of Pennylvania, United State and participant at numerou eminar at the Univerity of Bath and Chancellor College, Univerity of Malawi. Lat but not leat, I am grateful to numerou friend and colleague with whom intellectual dicuion have helped to timulate idea, reolve problem and generally ave the world throughout the lat few year. At the Univerity of Bath, Lavanya Anneboina, Loui Boakye-Yiadom, Antonio Savoia and Harry Walton deerve thank. Outide of Bath, thank go to Godefroy Clair, Greg Davie (Dad), Sue Davie (Mum), Hela Gharbi, Karen Giffaut, Hari Irhad, Richard Record, Myriem Remal, Luca Ruoi, Hugo Thenint and Sylvain Zini. vi

8 SUMMARY After dicuing the Malawian context and ummariing the remittance reearch, I focu on remittance in rural Malawi. I follow remittance from the giver motivation through to the receiver view of remittance and how the receiver ue them and finally to their impact a a mean of moderating the effect of negative hock on the receiving houehold. Reult how that parent remit to repondent for altruitic purpoe, or for inurance motivation (e.g. to help out the repondent if they are ick). Repondent remit to parent for altruitic motivation and inheritance. There i trong bi-directionality in the remittance flow. Children remit to repondent a an inurance premium, and for inheritance motivation. Altruim motivate repondent to give to their children. There i trong evidence of co-inurance between repondent and their ibling with both inurance payout and premium being paid. Repondent and their ibling alo remit to each other for altruitic motivation. There i trong evidence of mental accounting amongt both male and female headed houehold. Remittance exhibit a much lower MPC than alary and farming income. Male and female headed houehold differ in their ue of income from different ource, however one reult i conitent: remittance are ued for education. Probit model indicate that houehold are more likely to receive remittance from local area if omeone in the houehold i ick (local remittance inure a health hock). Houehold that uffer from drought are more likely to receive remittance from more ditant area (other ditrict, a city, abroad). Drought ha a major negative impact on conumption level but ditant remittance inure affected houehold who uffer from thee. Local remittance, which make up mot remittance flow, are unable to inure thee community hock. Only around 10 per cent of houehold receive remittance from outide their home ditrict however. Remittance help to inure houehold conumption againt health hock, but only food conumption i inured. vii

9 2SLS ADMARC CIA cdf CLT CPS CSR df(c) FDI FE FEWS(Net) FHH FTP GDP HIPC IFPRI IHS98 IRA IV LC-PIH LDC LPM LSDV MBS MDHS MLE MHH MK MPC MSE NSO OLS PAE PCA pdf PIH QUANGO RBM RE RemDitant RemLocal RER SEM STI SUR UNDP US$ VIF WDI ABBREVIATIONS Two Stage Leat Square Agricultural Development and Marketing Corporation (Malawian paratatal, imilar to the old Milk Marketing Board in the UK) Central Intelligence Agency Cumulative Ditribution Function (of a normal ditribution) Central Limit Theorem Complementary Panel Survey Malawian Center for Social Reearch Degree of Freedom (Degree of Cluter Freedom) Foreign Direct Invetment Fixed Effect (Regreion) Famine Early Warning Sytem (Network) Female Headed Houehold Family Tranfer Project Gro Dometic Product Heavily Indebted Poor Country International Food Policy Reearch Intitute Malawian 1998 Integrated Houehold Survey Individual Retirement Account Intrumental Variable (Regreion) Life-Cycle Permanent Income Hypothei Le Developed Countrie Linear Probability Model Leat Square Dummy Variable Malawian Migration Baeline Survey Malawi Demographic and Health Survey Maximum Likelihood Etimation Male Headed Houehold Malawi Kwacha (the unit of currency in Malawi) Marginal Propenity to Conume Micro and Small Enterprie Malawian National Statitical Office Ordinary Leat Square Per Adult Equivalent Principle Component Analyi Probability Ditribution Function (of a normal ditribution) Permanent Income Hypothei Quai-Autonomou Non-Governmental Organiation Reerve Bank of Malawi Random Effect (Regreion) Ditant Remittance (from outide a houehold' home ditrict) Local Remittance (from a houehold' home village or ditrict) Real Exchange Rate Simultaneou Equation Model Sexually Tranmitted Infection Seemingly Unrelated Regreion United Nation Development Programme United State Dollar Variance Inflation Factor World Development Indicator viii

10 1 CONTEXT AND BACKGROUND 1.1 INTRODUCTION Remittance have become an increaingly important ource of income for developing countrie and their population making the undertanding of the role that they play at national and houehold level an ever more relevant iue. In Malawi, remittance are a key component of both income and expenditure at the houehold level. They fulfil not only a ocial role, but alo have eential economic purpoe uch a inuring weather hock vital in an agricultural economy facing increaing climactic uncertainty. Uing Malawian data, thi tudy look at remittance from three eparate but related angle. Firtly, motivation to remit are tudied in Chapter 5. In Chapter 6, the ue of remittance, and how they are viewed by the receiver i tudied, and in Chapter 7, the importance of remittance a a mean of inuring negative hock i dicued. In all cae, exiting work i extended in a timely and relevant way. Prior to analying the data in chapter 5 to 7, an in-depth literature review i provided in Chapter 2, decriptive tatitic and a general overview of the data are given in Chapter 3, and the empirical methodologie ued are dicued in Chapter 4. Thi preent chapter fulfil the important role of giving background information on both remittance and the Malawian context. In doing o, it help to ituate thi thei in relation to other tudie and to draw parallel with macro level tudie. It give important information regarding the Malawian economic and ocial context, which i relevant for the thei. In particular, health i given it own ection due to it importance in both ocial and economic term in Malawi, a well a the link between health and everal element of thi tudy. Chapter 5 include health hock a a potential motivation for remitting, and Chapter 7 focue on the role of remittance (regardle of motivation) in inuring hock (including health). For imilar reaon, while dicuing the economic background and environment in Malawi, coniderable weight i given to dicuing agriculture. Agriculture i the maintay of the Malawian economy in term of employment, export and contribution to GDP. In 1

11 addition, the agricultural environment i particularly relevant for thi thei, ince all empirical chapter analye rural Malawi. 1.2 BACKGROUND INFORMATION ON REMITTANCES Thi thei tudie remittance on a micro (houehold or individual) level. In order to place the tudy within it wider context however, thi ection give background information on remittance. Firtly the macroeconomic and international trend and impact are ummaried before focuing on the microeconomic level. Thi help to draw parallel between houehold and international tudie of remittance flow including their motivation and impact, and help view thi thei from a more global perpective Evolution and importance of international remittance flow Elbadawi and Rocher (1992) in El-Sakka and McNabb (1999) calculate that by 1990 remittance flow had increaed fivefold ince 1970 to US$33.8bn. Gammeltoft (2002) compare flow of remittance, aid, FDI and other official and financial flow from the official tatitic of developing countrie from Official remittance doubled to US$65bn between 1991 and 1999 and including etimate of unofficial flow uch a thoe made through hawala-type ytem thee flow could increae to over US$100bn. Over the ame period, aid fell by 16 per cent to US$41.6bn and the value of flow of FDI ha increaed continually. In 1991, aid wa the larget ingle component of reource flow compriing 32 per cent, but by 1999 FDI accounted for over half of all financial flow to developing countrie. Ratha (2004) calculate that remittance receipt for developing countrie were worth 1.3 per cent of their GDP or around US$72.3bn in Thi amount to 42 per cent of the value of FDI in thi year. He find that remittance to leat developed countrie (LDC) are larger a a hare of GDP than thoe to middle-income countrie (1.9 per cent veru 0.8 per cent). Ratha (2007) how that 2006 remittance flow were double their 2001 level at over US$200bn, and believe them to be greater than FDI when informal tranfer are included. Several factor have contributed to thi increae with increaed migration being particularly reponible. El-Sakka and NcNabb (1999) alo dicu active policy by the Egyptian Government to encourage remittance inflow ince the The Pakitani Government alo operate policie to encourage remittance including offering privilege 2

12 uch a pecial lounge in airport, free inurance, exemption from certain import dutie and emergency renewal of paport for thoe who remit over a certain amount. The Mexican Government ubidie remittance ued to invet in buinee (Orozco, 2002b). In addition, improved and expanded data collection ha helped to increae apparent remittance flow (De Luna Martinez, 2005). All of thee tudie are however, limited by the lack of information on tranfer. For example, informal and largely unrecorded tranfer made through hawala/hundi 1 ytem are believed to amount to a ignificant proportion of international remittance flow (e.g. Anarfi et al., 2000). In addition, official remittance flow are ignificantly under-reported ince different countrie collect different information. For example, ome countrie collect data on tranfer made through the Pot Office a well a bank, but other collect data only on thoe made through bank. Bank will often allow a cutomer to poe two debit card, one of which can be ent to a remittance receiver, who i able to withdraw money directly from cah machine. The value of thee flow i etimated in ome place, but not in other (De Luna Martinez, 2005). Adam and Page (2005) found that ub- Saharan African countrie have the highet proportion of unrecorded remittance equal to 73 per cent of recorded flow. If Malawi i typical, the true value of international remittance will be cloer to two per cent of GDP rather than the one per cent currently etimated (WDI, 2006). Depite the data limitation, a number of characteritic and impact of international remittance can be dicerned. For example, they tend to be table or counter-cyclical (Gammeltoft, 2002), and increaed in value during the 1995 Mexican crii, the Aian currency crii during the late 1990 and during natural diater uch a hurricane in central America (Clarke and Wallten, 2003) the Aian Tunami (Ratha, 2007) and the northern Pakitani earthquake in 2005 (Savage and Harvey, 2007). Ahmed (2000) alo note a teady increae in remittance toward Somalia during conflict year. In addition, El-Sakka and McNabb (1999) find that inflation in the home country (Egypt) ha a poitive 1 Thee ytem allow the ender to enter a dealer office (often located in their own home) and requet that fund be ent to a recipient. S/he then pay the dealer, who then contact another dealer cloe to the home of the intended recipient (uually by telephone). In rural area, the local recipient will often be able to identify the recipient by name and deliver the fund directly to their houe. Dealer keep record of flow in each direction, and any flow which are not netted out are ettled through the occaional tranfer of fund between dealer (in the form of cah, gold, jewellery or other mean). Thee ytem are widepread and able to function at very low cot (around 0.5 to 1% of the value of the tranfer) and the low et-up cot mean that the ytem i highly pervaive, even in remote area. 3

13 impact on remittance concluding that ender increae flow in order to upport familie during time of riing price level. Although thee are meaured on a macro level, thee characteritic indicate that remittance are, at leat in part, driven by altruim and/or inurance. In contrat to El-Sakka and McNabb (1999), Higgin, Hyenbegai and Pozo (2004) find that inflation in a migrant country of origin ha a negative impact on remittance flow in Latin America whilt increaed inflation in their hot country erve to increae remittance a migrant eek to protect their purchaing power. A nominal depreciation of the receiving country (migrant country of origin) exchange rate increae remittance uggeting that migrant take advantage of potentially higher return to invetment. An anticipated depreciation however caue remittance to fall temporarily a they are delayed. Motivation to remit on a micro level are teted for in Chapter 5. Amuedo-Dorante and Pozo (2004) look at the impact of remittance receipt on the real exchange rate (RER) of 13 Latin American and Caribbean countrie. They control for technical progre differential, government pending, external term of trade, interet rate and foreign aid and account for the endogeneity of remittance and potential endogeneity of foreign aid. They find that a doubling in the value of per capita remittance caue the RER of the receiving country to appreciate by 22 per cent, and conclude that remittance are potentially aociated with the rik of Dutch Dieae. The ame tudy find that foreign aid doe not have a ignificant impact on the RER, perhap becaue during the period tudied ( ) much aid wa tied to purchae from the donor country. Rik of Dutch Dieae and/or Balaa-Samuelon effect i trongly linked to the microeconomic ue of remittance income. El-Sakka and McNabb (1999) find that the income elaticity of import from remittance i high while the price elaticity i low. Thi may indicate that remittance in kind (defined, in Egypt, to include only the value of import funded directly by remittance) are predominately luxury good. Finance may thu be diverted away from invetment and toward conpicuou conumption. The ue of remittance are tudied in Chapter 6. Page and Plaza (2006) note that remittance, the extent of a country overea diapora, and invetment are trongly linked. In 2000, 45 per cent of the total US$41bn of inward 4

14 FDI for China came from the Chinee diapora. Gould (1994) find that a 10 per cent increae in immigrant to the US i aociated with a 4.7 per cent increae in US export to immigrant country of origin. Head and Rie (1998) ue a Tobit pecification of an augmented gravity model to find that a 10 per cent increae in the tock of immigrant increae export by per cent and import by per cent. Remittance can have an important impact on economic growth and poverty. Summariing previou literature, Chami et al. (2003) report that poitive Keyneian multiplier can be identified for remittance. Short-term multiplier between 1.24 and 3.2 are reported (Adelman and Taylor, 1990; Davie and Davey, 2008). Modelling remittance, Chami et al. (2003) meaure per capita economic growth a a function of worker remittance. Under all pecification of their theoretical model, they find a negative relationhip. Their reult are robut even when intrumenting to correct the problem of endogeneity; the main caue of growth might alo influence remittance. They conclude that [n]ot only do remittance in low-growth countrie tend to be higher, but higher remittance within a country are aociated with lower growth (p.74). Adam and Page (2003) look at the impact of inward remittance on poverty. They find that a 10 per cent increae in remittance a a percentage of GDP decreae the hare of people living in poverty by 1.6 per cent. Furthermore, the pooret tend to be the larget beneficiarie, with the poverty gap alo decreaing in remittance. Thi paper ha the weakne of auming that remittance are exogenou. It doe not conider a houehold income under the alternative ituation that the remitting migrant had tayed and worked at home. Yang and Martinez (2005) ue data on houehold level remittance receipt in the Philippine from before and after the 1997 Aian currency crii and find that that an increae in the migrant houehold remittance receipt equivalent to 10 per cent of precrii houehold income reduce the poverty rate among uch houehold by 2.8 percentage point. Additionally, they find that non-migrant houehold benefit from increaed gift, reduction in poverty rate and poverty gap thank to the trickle-down effect of remittance income. 5

15 Gutafon and Makonnen (1993) imulate a model to etimate the change in poverty rate in Leotho a country in which remittance (motly from South African mine) made up around 40 per cent of GNP at the time tudied (1986/87) if worker were forced to return home. In their mot advanced model which conider the labour of returned miner, an additional 11 to 14 per cent of houehold would be claified a poor Microeconomic impact Studie have hown that remittance can impact on conumption and aving behaviour (tudied in Chapter 6) including invetment in education (dicued in Chapter 6) and buine, labour market behaviour, and can act a inurance for hock (tudied in Chapter 7). The impact of remittance can be determined by the motivation for ending them (tudied in Chapter 5) Conumption behaviour In an analyi of 200 emi-urban Salvadorian houehold, Benavide et al. (2003) potulate that houehold with and without migrant will exhibit different conumption pattern. They plit their ample accordingly and find that houehold with migrant exhibit a higher aving rate (22.5 per cent veru 3.9 per cent of income) and invet over three time more in health, education and other categorie than non-migrant houehold. Furthermore chool-age children in migrant houehold are more likely to be enrolled in education than thoe in other houehold. Looking at poverty, thee ame author find that abolute poverty i lower for houehold with migrant than thoe without (72.5 per cent veru 94.4 per cent). However, income and expenditure in migrant houehold are around 50 per cent higher than in non-migrant houehold leaving open the quetion of cauality. Taylor and Mora (2006) attempt to anwer the quetion Doe Migration Rehape Expenditure in Rural Houehold?. They ugget that thi may occur through remittance but chooe to eparate their (Mexican) ample into houehold with a migrant and thoe without whilt controlling for any potential election bia uing Invere Mill Ratio. They find that houehold with international migrant dedicate a larger marginal budget hare to invetment than non-migrant houehold (0.21 compared with 0.10). Houehold with US migrant alo pend more at the margin on conumer durable than other houehold (0.22 againt 0.12) and more on ervice than non-migrant houehold (0.23 veru 0.16). 6

16 However, houehold with internal migrant exhibit a lower marginal propenity to invet than non-migrant houehold (0.06 compared with 0.10). The 2006 Malawian Migration Baeline Survey (MBS) interviewed 9,546 repondent of whom 736 were migrant. It looked at the link between migration and remittance and found that remittance contributed an average of ix per cent to total houehold income with farming produce adding an additional 31 per cent, caual labour (known a ganyu ) 27 per cent and wage employment 18 per cent. 27 per cent of houehold only ent (and did not receive) remittance, with the MBS report auming thi alway went to a worker away from home. 15 per cent of houehold received remittance and 17 per cent both ent and received remittance. The finding alo indicate that male migrant are more likely to remit than female migrant with two third of male remitting a oppoed to one third of female. Thi i likely to be a reflection of the reaon for migrating. Men are more likely than female to migrate for employment and remit home. Although difficult to ae due to the fungibility of aet, the MBS aked repondent about how they pend the remittance they receive. Repondent reported that remittance were primarily ued to fund the purchae of food, with 75 per cent of urban houehold, and 46 per cent of rural houehold reporting uing their remittance receipt to purchae food. Remittance were alo important to pay for water and medical bill. Thee ue how that remittance are ued for neceitie (perhap following a negative hock), and the medical expenditure ugget inurance or altruitic motivation for remitting. Table 1.1: Reported Ue of Remittance by Repondent of 2006 Malawi Migration Baeline Survey Rural Urban Food 46.0% 75.0% Water 25.6% 5.4% Medicine 20.1% 13.4% Source: Adapted from NSO (2006) Education Yang (2005) and Yang and Martinez (2005) find that increae in remittance flow in the Philippine reulting from the 1997 exchange rate hock led to improved chooling expenditure; a decreae in child labour; an increae in durable good ownerhip and increaed invetment in capital intenive entrepreneurial activity. 7

17 Cox Edward and Ureta (2003) focu on the ue of remittance for invetment in education in El Salvador. They examine the effect of remittance on chool dropout rate amongt familie which receive remittance and thoe which do not. They find that the hazard of dropping out decreae by 0.4 per cent a houehold income increae by 100 colone, although in rural area thi impact only after the 6 th grade. The receipt of 875 colone (the median remittance) decreae the hazard of dropping out before the 6 th grade by 54 per cent and by 27 per cent from the 7 th to 12 th grade. The impact of remittance at it mallet i 10 time the ize of the effect of any other income ource. The author offer two explanation for thi. Firtly they ugget that remittance are more table than other ource of income and o offer a better proxy for permanent income than other ource of income (although, ince they ue cro-ectional data, they offer no proof of thi in thi intance). The econd i that there i a higher propenity to pend on children chooling out of remittance than out of other ource. They explain thi reult by uggeting that uch pending i a condition of continuing to receive remittance payment. Chapter 6 focue on MPC out of remittance and find that remittance are indeed ued to fund education. However, Chapter 6 offer an alternative explanation. Kozel and Alderman (1990) in Chami et al. (2003) report that the receipt of remittance ha a negative effect on the probability of labour market participation amongt male adult in Pakitani houehold. The full long-term impact of thi i difficult to ae given that we do not know whether time wa pent in education or leiure. Bokoi (2001) note that labour i already underemployed in Malawi. Luca (1987) found improvement in agricultural practice due to invetment from remittance income. However, he found that the aociated gain were offet by the effect of labour withdrawal from agriculture toward mine. More importantly, Azam and Gubert (2004) find that the lower crop production i not imply a reult of the withdrawn labour, but a behavioural (moral hazard) repone to the receipt of remittance Buine Amuedo-Dorante and Pozo (2006a) analye the link between remittance and buine ownerhip. They model remittance and entrepreneurhip a being imultaneouly determined ince ownerhip of a buine may influence receipt of remittance while receipt of remittance may influence entrepreneurhip. They find that buine 8

18 ownerhip raie the probability of receiving remittance (even from extended family) by 20 percentage point. However, receiving remittance lower the probability of owning a buine. They ugget that while remittance may looen the budget contraint faced by ome houehold when it come to buine ownerhip, thee monetary inflow alo induce an income effect that raie the reervation wage of thoe houehold member (p.950). Orozco (2003) indicate that remittance tend to be aved or inveted in education, health or wealth generation. Citing Woodruf and Zenteno (2001), he note that remittance were reponible for 27 per cent of the capital inveted in micro-enterprie in Mexico. Orozco (2002) find in a tudy of ix developing countrie plu Portugal that at leat 10 per cent of recipient houehold ue their money for a aving or invetment cheme, and Ikanda (2002) in Orozco (2002) tate that 70 per cent of Moroccan living abroad have inveted in Morocco, compared with only 30 per cent who have inveted in their hot countrie. Maey and Parrado (1998) ue repeated cro-ectional data from and from 1987 to 1992 to ae the link between international remittance and buine formation in Mexico. Uing a probit model, they find that receipt of remittance increae the likelihood that a houehold will form a buine and increae productive invetment, although the latter wa more influenced by peronal, houehold and community characteritic than remittance or migration a uch. They alo find that remittance received during a houehold migrant abence had little impact on buine formation, but rather, following the migrant return, remittance increaed buine invetment. Local market condition alo play a role, with ome area being more entrepreneurial than other. In a report baed on the Micro and Small Enterprie (MSE) Survey undertaken in Malawi in 2000, Ebony Conulting International Ltd, Kadale Conultant and the National Statitic Office (2000) find that the principal ource of MSE tart-up capital i remittance income in around 12 to 19 per cent of cae between 1998 and Furthermore, the percentage of buine tart-up uing thi ource of income a their major tart-up capital i increaing over time. Thi i compared with an average of two per cent which 9

19 ued credit intitution, even per cent who ued loan from family or friend and 61 per cent who ued peronal aving. McCormick and Wahba (2001) ue a dataet of 1,526 return migrant to Egypt. They note that 10 per cent of returnee invet in economic project. They find that 19 per cent become employer compared with 10 per cent pre-migration and one third of new entrepreneur judged kill acquired abroad to be ueful. Saving accumulated overea were alo important. Furthermore, the private ector i more favoured pot migration with jut nine per cent returning to public enterprie againt 44 per cent before migration. The indutrial pattern of employment remain imilar however with a modet hift away from agriculture and production. Uing a probit model to etimate the probability of becoming an entrepreneur, McCormick and Wahba (2001) find that the aving acquired abroad are mot important in etting up a buine for thoe with lower education who perhap find it more difficult to acce credit in Egypt, while overea experience (kill gained) matter mot for the better educated. Ahmed (2000) note that 15 per cent of remittance receiver in pot-war Somaliland ued their remittance to et up buinee or acquire land or houing. He alo reported other main ue of remittance a being conumption, celebration (uch a marriage) and blood money (payable under Ilamic haria law in lieu of execution following a murder). Woodruff and Zenteno (2001) ue a probit model to etimate the probability of an individual being a buine owner, controlling for peronal characteritic. They ue the migration rate of the tate in which the individual i living and the tate in which /he wa born a a proxy for acce to remittance. The reult are compared with Bank of Mexico remittance data and reult are hown to be imilar. They find that both higher migration rate and higher remittance level are generally aociated with higher rate of enterprie ownerhip, with the effect being tronger for female and internal migrant than for male and thoe who migrate abroad. The paper continue to etimate the log of capital inveted in buine enterprie (but eem not to control for potential election bia) to find that inveted capital i higher in tate with more migration to the US and a one tandard deviation increae in the migration rate i aociated with an increae in inveted capital of about 10 per cent. 10

20 Remittance mechanim and the impact of technology International remittance can be ent formally or informally. Formal remittance will tend to be in the form of cah ent through bank or money tranfer agencie. Different countrie count thee differently. For example ome include thoe ent through the pot office and other do not. In addition, ome bank allow account holder to have everal cah card. One of thee can be ent abroad and the remittance receiver can withdraw a certain amount per month. Some countrie make effort to calculate the value of thee emi-formal remittance, whilt other do not. In thi cae it i difficult to know which flow are remittance and which are imply the remitter viiting family in hi/her home country. Both internal and international informal tranfer are made in everal way. The remitter might ave money and then carry it in cah to the recipient. Alternatively cah might be given to a friend or family member to take to the receiver. Internally, or acro land border, cah might be given to thoe who make a trip regularly uch a bu driver. In ome area, hindu or hawala ytem are alo ued. Finally, the remitter might make purchae directly for the receiver. Thi might include paying chool fee directly, or giving gift. Over large (or even mall) ditance, and where technology i not eaily available, the cot of remittance can be prohibitive. It can take a day or longer to make even hort journey in rural Africa, and large firm are often unable to reach the rural poor. The recent development of mobile phone network which pervade even the remotet rural African area are giving African acce to national and international network which can be ued to tranfer fund. Many phone companie allow remitter to text fund to receiver, who are then able to cah in their balance at local tore. Thi i greatly reducing the cot of remitting to rural area, and, although little reearch ha been undertaken to date, thi innovation ha the potential to greatly reduce tranaction cot. The impact of thi ha yet to be oberved. Finally, both government and the World Bank 2 are making effort to publicie the cot of ending international remittance in an effort to encourage competition amongt 2 See: 11

21 formal money tranferor and reduce cot. Thi ha had additional impetu ince 9/11 ince wetern government value the importance of formal remittance flow a oppoed to informal finance, which can be ued to finance terrorim. 1.3 THE MALAWIAN CONTEXT Geography Malawi i a mall country ituated in outh-eat Africa meauring approximately 900km from north to outh and km from eat to wet with Lake Malawi covering approximately a fifth of it total area. It i denely populated with a population of around 13 million. Although Lake Malawi i the third larget frehwater lake in Africa, the country i landlocked and border Tanzania to the north, Mozambique to the wet, outh and eat and Zambia to the eat. There have been recent move to allow medium ized ocean going veel direct acce to Malawian port through canal and river in Mozambique. Figure 1-1: Map of Malawi Malawi i a largely rural country with around 15 per cent of the population living in citie. The four main citie are Lilongwe, the capital in the centre of the country, Mzuzu in the north and Blantyre (the commercial capital) and Zomba (the former capital) in the outh. Figure 1-2: Map of outhern Africa Source: oaiafrica.com Source: CIA World Factbook 12

22 1.3.2 Hitory and ociety The Malawian population i largely decended from everal wave of migration. Bantu tribe began to migrate into the area from central Africa from around 200AD bringing with them Iron Age agricultural technologie. During the ixteenth century Maravi Bantu tribe began to ettle in Malawi and the majority of modern Malawian including the Chewa who are dominant in the outh and central region and the Tumbuka who are dominant in the north are decended from thi wave of migration. During the nineteenth century Yao tribe entered the outhern region from Mozambique partly diplacing the Maravi from the area and Ngoni Zulu migrated northward from South Africa, eventually ettling in central and northern Malawi. Arab lave and ivory trader, who had been active along the eat coat of Africa for a millennium began moving inland during the nineteenth century bringing with them Arabic culture and Ilam. Arabic influence remain trong in ome area of the outh. Around the ame time Britih miionarie and cientit were arriving in Malawi bringing with them Chritianity and commerce. Modern Malawian chool book look favourably upon the achievement of the Mulim and the Chritian miionary group (Standard 6: p.80). Today around 80 per cent of the country i Chritian of variou denomination; around 15 per cent i Mulim; and 5 per cent hold traditional religiou belief of worhip of a ingle God through ancetor (Mkandawire, 1999: p.38; Phiri, 2004: p.29). There i little (but growing) religiou tenion and a majority Chritian electorate elected a outhern Mulim preident, Bakili Muluzi, in Although Malawi i generally a harmoniou place whoe people are proud to be a peaceful nation, the country ha trong regional diviion. Today thee diviion manifet themelve in the religiou make-up of the region (Mulim are concentrated in the South along Arab-Swahili lave trading route); language (Tumbuka i poken in the North and Yao in ome outhern area while Chichewa i ued in mot other region); voting (political partie have trong regional biae (Mkandawire, 1999: pp.20-21)); and inheritance and marriage cutom (the north i largely patrilineal and patrilocal whilt the outh and centre i more matrilineal and matrilocal). 13

23 In addition, the colonial legacy ha left vat diparitie in education and indutrial capacity. Chritian miionarie tended to et up chool in the north of Malawi, while indutrialit et up production facilitie in the outh. The reult i that upon independence, 55 per cent of northern children attended chool compared with around a third in the ret of the country (McCracken, 2002). Indutry i concentrated in the outh where the indutrialit tended to ettle. Today, literacy rate are 82.1 per cent of the population for men and 70.7 per cent for women, but both literacy and chool attendance are higher in the north (NSO, 2006b: p.37). Furthermore, in ome northern region, female literacy rate are higher than thoe of men, while the greatet gender diparitie exit in the outhern tip (Sena-peaking area) of Malawi with the male rate of literacy being over 22 percentage point higher than that for female. School enrolment rate amongt 6 to 13 year old are on average 65.7 per cent but thi increae to over 85 per cent in many northern area, and fall to under 55 per cent in ome outhern area. Education level are greater on average in the north, with 7-9 year of education on average, while many area in the outh have le than 4 year of formal education. Educational achievement i higher in urban area than rural one, with le than 2.5 per cent of rural people achieving their Malawi High School Certificate, compared with over 10 per cent in urban area (NSO, 1998) Health Thi ection will how that health i an important iue in Malawi and can have a large impact on the both the houehold and the wider economy. Dicuion of health i bet done by claifying iue into thoe related to HIV/AIDS and other iue. Thi ection provide eential information for undertanding the context of the tudy, and alo help to jutify tudying health hock in the context of motivation for remitting in Chapter 5 and remittance a inurance for health hock in Chapter 7. An overview of Malawi health ituation and finding regarding pecific iue uch a malaria and malnutrition are provided. A dedicated ection deal eparately with HIV/AIDS. Malawi ha three central hopital and over 40 regional one. However thee are not pread evenly with ome ditrict not having any medical facilitie (King and King, 2000). Acce to thee hopital remain difficult with average walk of over 12 kilometre to reach a hopital and between 2 and 5km to reach a health centre (Conroy et al., 2006: p.105). Health care pending i around 3 per patient compared with 700 per patient in 14

24 the UK (King and King, 2000: p.19); one urgeon per one million people compared with one per 30,000 in the UK and one doctor per 30,000 compared with one per 600 in the UK (King and King, 2000: p.30). The health ervice relie on medical aitant and clinical officer. Clinical officer are often confued with doctor but do not have the ame level of training. Thi i partly a calculated move by the Government which ceae a doctor training before doctor tatu i reached, enuring taff do not migrate to developed countrie. Malawi lack important medical taff with vacancy rate of between 42 and 87.8 per cent for poition of different grade in both government and Chritian Health Aociation of Malawi (CHAM) 3 clinic and hopital (Conroy et al., 2006: p.38). Thi remain the cae depite the Britih Department for International Development (DfID) paying to double medical worker alarie in 2006 and enuring the fund are available to pay any new medical taff. Of the 26 ditrict in the country, ten have no Minitry of Health doctor and four ditrict are without any doctor (Conroy et al., 2006: pp.43-44). Uing Demographic and Houehold urvey data, Devereux et al. (2006) report that around 48 per cent of children in Malawi are tunted. Furthermore, there i a trong aociation between child tunting and the mother education, with incidence of tunting falling ignificantly after 4 year of education. The problem i mot evere in urban area (NSO, 2006b: p.12). Malnutrition i evere in Malawi becaue the great majority of the population cannot afford to eat enough even in normal [crop yield] year (Conroy et al., 2006: p.41). The average Malawian conume 1818 calorie per day, and poor houehold which account for 65 percent of the population conume only 1428 calorie per day compared to the 2000 to 2500 calorie needed to utain the average adult male (Conroy et al., 2006: pp.39-41). The impact of malaria i alo evere for houehold and the national economy in Malawi. 40 per cent of outpatient viit are due to malaria and malaria account for 18 per cent of the death of under-five (NSO, 2006b: p.17; Conroy et al., 2006: p.34). Furthermore, malaria rik i highet during the agricultural eaon; the time at which houehold have the greatet need of their labour. Malaria dramatically decreae a houehold 3 CHAM provide around 37 per cent of Malawi healthcare (Conroy et al., 2006: p.146; The Nation, 20/10/2006). 15

25 productivity ince miing key date for planting, harveting or weeding ha a evere negative impact on crop yield. Maize yield can be between almot nothing to around 4,000kg per hectare depending upon fertilier ue and the weeding regime. The total cot of malaria to the average Malawian houehold i etimated to be around 22 per cent of annual houehold income (Conroy et al., 2006: pp and p.90) A focu on HIV/AIDS Malawi ha the tenth highet HIV prevalence in the world, with over half a million Malawian having died a a reult of the dieae and around half a million children having lot at leat one parent to HIV/AIDS. AIDS ha alo had a trong impact on life expectancy with it having dropped to le than 39 year of age due to the dieae. Without AIDS, it i etimated to have been around 56 year (Morah, 2007; Conroy et al., 2006: p.65). Official rate of HIV are around per cent and have remained relatively teady ince the turn of the 21 t century. There i alo ome indication that official rate are actually over-etimate and that prevalence ha fallen in recent year with the 2004 Malawi Demographic and Health Survey (MDHS) finding an infection rate of around 12 per cent (11 per cent in rural area and 17 per cent in urban area) (Morah, 2007). Morah (2007) analye attitude to the pread of HIV in Malawi following the dicovery in a national urvey conducted in 2003, that 94 per cent of the adult population agree with the propoition that HIV i pread by people who know they are HIV-poitive, but cannot or will not change their behaviour. Morah (2007) ue a ample of 743 repondent of whom 327 were openly living with HIV and 416 individual who were unaware of their HIV tatu. He find that the HIV poitive have better knowledge of the dieae and are more much more likely to engage in afer exual behaviour. Although they are more likely to practice afer ex, 56.6 per cent of HIV poitive people did not report alway practicing afer ex. Reaon for not practicing afer ex amongt thoe who did not know their ero-tatu varied but a full 12.6 per cent of thoe who did not know the ero-tatu of their partner reported not knowing where to purchae condom or not knowing how to ue them. Other reaon included fatalim: with 35.3 per cent of HIV poitive and 30.7 per cent of ero-tatu unaware reporting that they believed that becoming infected wa pre- 16