Family prestige as old-age security: Evidence from rural Senegal more

Family Prestige as Old-Age Security: Evidence from Rural Senegal Jean-Luc Demonsant1 May 2007 ABSTRACT The crux of this paper is to find out the motivations of migrants to support their ageing parents in the village. I conducted a household survey in the Senegal River Valley designed to take due account of the local socio-political structure. The empirical results suggest that while belonging to a prestigious family lowers the probability of migrating, it raises the probability of frequently remitting to the patriarch. Thus migration can either be seen as an individual decision to escape from a poor social position or, conversely, as a family decision to maintain its dominant position in the village. Keywords: Migration; Remittances; Social Stratification; Old-Age Support. JEL Classification: O12; D12; J62 ARQADE, Toulouse School of Economics, Manufacture des Tabacs - Aile Jean-Jacques Laffont, 21 allées de Brienne, 31000 Toulouse, France; jean-luc.demonsant@univ-tlse1.fr. For my research in Senegal, I benefited from financial support from both the ministry in charge of Higher Education (Aires Culturelles grant, 2001) and the ministry of Foreign Affairs (Lavoisier grant, 2003-04). First of all, I would like to thank my supervisor, Jean-Paul Azam for his encouraging support and insightful suggestions. I also thank seminar participants in Toulouse, Dakar, Paris and Bordeaux for providing useful comments. Last but not least, I am extremely grateful to Monica Ramirez and Anthony Stenton for doing a great job in tidying up stylistic mistakes on previous drafts of this paper. I am solely responsible for all remaining omissions or errors. 1 1 ‘When a young man considers migrating, he must keep in mind the day when he will come back.’2 INTRODUCTION In contrast to cases of elderly abuse reported in many developing countries (see WHO 2002), some migrants do make huge transfers to their elderly parents. What kind of enforcement mechanisms exist to make them abide by this sharing rule despite geographical distance? What do the older parents have to offer in return? My study does not pretend to draw universal conclusions on migration and remittances behaviour, but is rather rooted in the local cultural context of my fieldwork in rural Northern Senegal, although it could well be extended to other types of village societies featuring gerontocracy embedded in a social stratification. In a broad sense, it seems rather irrelevant to study rural household economics without taking due account of the socio-political structure of the village. For instance, too often emphasis has been placed on ethnic diversity, ignoring social divisions within ethnic groups. However, it is well documented that many Sahelian societies are based on a so-called ‘caste system’, which is still very pervasive in some areas. Strikingly, ethnic groups that historically make up the majority of Sub-Saharan migrants in France, originate from the same area -the Senegal River Valley- and are renowned for being very attached to their traditions, and in particular their officially abolished social stratification. Here is some insight into why this may not be purely fortuitous. On the one hand, the fact that high status families have greater needs in terms of prestige expenses to signal that they deserve their rank has surprisingly been overlooked when studying migration economics3. Without any doubt this holds true anywhere: our Western society’s consumption rites have great similarities with conspicuous spending which has been described by anthropologists in other non-Western societies (i.e. the ‘potlatch’...). Furthermore, this should even be stronger in highly hierarchical societies, where belonging to a high status family gives rise to substantial privileges. On the other hand, it has previously been argued that the aspiration to inherit may be a strong motive for remittances. If there are no potential valuable physical bequests, the family status becomes the desired symbolic bequest. For instance, in her Ph.D. thesis, Flore Gubert showed how among the Soninke, aristocratic families send out migrants to insure their pride (Gubert 2000). I find comparable but slightly different results studying old-age support among the neighbouring Haalpulaaren. These results yield different policy implications. Nevertheless, when studying old-age support, the only relevant case for economists arises when the child leaves his parents’ home. Transfers between separate households have the advantage of being observable whereas daily and continuous exchanges among coresidents are almost impossible to disentangle, or indeed to measure. In developing countries it is mostly the case that coresidence is the norm, except when migration enters the picture. One cannot conclude that non-migrants are not supporting their parents, but when leaving his parent’s home, the migrant may also turn his back on his family. So, the focus will be on migrants and their remitting behaviour. As in many village societies, gerontocracy is the rule within the local political power. However, this may lead to erroneous predictions as far as old-age support is concerned. Indeed, if elders drew their authority from strong peer and 2 3 . Pulaar saying, my translation. . With the notable exception of Lucas and Stark (1985). 2 face-to-face pressure in the village, distance should trigger rebellion. Results from my survey data show that both migration and remittance choices of a son are linked with the social ranking of his father4 within the traditional village society. Indeed, it appears that belonging to a low status family raises the probability of migrating, whereas the migrants from high status families are more likely to remit. Why are low status migrants more likely to turn their back on their family while high status migrants often constitute the main family support? In reality, two opposing motives for migrating seem to come into play, depending on the position of the migrant’s family in the social ladder of the village. Migration can be seen as a means either to escape from a poor social position, or conversely to maintain one’s dominant position. The first case is very similar to Luke and Munshi (2005), where the most disadvantaged group - in the South Indian context, low caste women - migrates to gain social mobility. By contrast, in the second case, the migrant intends to come back in the village to benefit from the advantageous position of his father. As a result, he will have to keep up his own rank by giving his parents the financial means to sustain their status. As in Gubert (2000), remittances are embedded in a broader family arrangement aimed at insuring the pride of the family. In this paper, I will focus on the enforcement mechanisms ageing parents have at hand to elicit the loyalty of their migrant son to the family contract. In the second section, I will position my framework in the microeconomics literature on transfer motives and, in particular, old-age support in developing countries. The third section draws on the cultural context of the fieldwork, i.e. the Haalpulaaren of the Senegal River Middle Valley, to place this study in perspective with other social sciences, and my own field experience. In the fourth section, I will briefly describe the theoretical framework, emphasizing the role of family prestige as an enforcement device of old-age support arrangements. The fifth section displays the empirical strategy - from the data collection process to the econometric specification - developed to highlight the relevance of this hypothesis. The sixth part will present the results based on my household survey, along with additional qualitative evidence that support this scenario. Finally, I will conclude with some policy implications on migration and development. REVIEW OF MICROECONOMIC LITERATURE Old-age support and social status Cases of elder abuse can be found in Africa despite mainstream opinion that respect towards elders is the norm (see Ferreira 2005). Miguel (2003), for instance, analyzed the extreme situation in which elderly widows were either killed or expelled at times of hardships. He highlighted the positive correlation between drought shocks and increases in numbers of witch killings. On the other (brighter!) end of the spectrum, one can reasonably wonder why migrants send substantial parts of their earnings to their ageing parents who stayed behind in the village, and do not seem to be in a position to reciprocate. Indeed, many studies point out how poor elders’ livelihood highly depends on family support in many developing countries as highlighted in Lloyd-Sherlock (2000). It has been argued that high fertility rates are due to dependence on children for old-age support (see for instance Nugent 1985).This is due to the absence of any suitable social security schemes, health insurance as well as any formal pension program. The elderly becomes 4 . I will indeed focus on father to son relationship as Haalpulaar is a patrilineal society. 3 dependent on either their asset stocks, or their own labour income, or else their children’s support. Nevertheless in many developing countries since both financial institutions and property rights are insecure, and the currency may be subject to inflation or nominal shocks such as devaluation, there exists no reliable form of asset accumulation whatsoever. Hence with low asset stocks, and when they are unable to work any longer, individuals reaching old-age do depend entirely on their children. The number and gender of children, along with the living arrangements (inter-generational coresidence or separate households) are important determinants of the welfare of the elders. These traditional family support systems for elderly people are commonly believed to be working relatively well due to specific cultural behaviour such as respect for elders, and the existence of strong family ties. Such analysis is based rather on intuition than on serious empirical research. Furthermore, economic and social change may gradually put more strain on these informal mechanisms. In semi-arid rural areas, incomes from agriculture are rather precarious, because of weather shocks (periodical times of drought during the dry season followed by floods during the rainy season), or parasite invasions such as recent locusts invasions in the Sahel. In these countries, malnutrition is very common, particularly during the lean season (soudure in French): the period when the harvest from the previous year has been exhausted and the next season’s harvest is not yet ripe5. The quasi-absence of both formal credit and insurance markets leads to a variety of informal coinsurance arrangements in village economies6. Given that positive correlations in weather outcomes diminish with distance7, households can allocate some members to different sectors via migration. Rosenzweig and Stark (1989) found, in rural India, that marriages of daughters in distant villages provided some kind of insurance mechanism for the family. Furthermore, despite the fact that urban and foreign jobs present some kind of risks too, these are not correlated with local activities. Indeed, the new economics of labour migration introduced by Stark (1991), views migration as an intrafamily coinsurance arrangement aimed at providing remittances in case of drops in rural incomes in exchange for assistance in case of unemployment, or covering the expensive costs of migrating (i.e. travelling ticket plus smuggling fees for illegal migrations). If this were the case, one should find a negative correlation between amounts transferred and the recipient’s pre-transfer income. It is noticeable that, in this regard, a perfect insurance mechanism yields the same result as the pure altruism case. What if remittances are not linked with the actual income of the family who stayed behind, but rather their social needs: family or religious expenses? Indeed in Stark’s (1991) framework, the social status of the family should not enter the picture. Nevertheless it has been long acknowledged that individuals value the esteem of others. Adam Smith wrote: ‘It is not wealth that men desire but the consideration and good opinions that wait upon riches’. Expressions just as the rat race or keeping up with the Joneses appear in many influential economists’ writings in different contexts (in both developed and developing countries). In . In the sample, over 45% of families had suffered days of forced fasting at least once in past 2 years. .Some authors suggest it may well be the other way around: efficient intra-familial arrangements prevent the formation of those markets (see Stark 1995 and Kranton 1996). 7 . In the words of Rosenzweig (1988). 6 5 4 hierarchical caste societies this becomes even more relevant, as belonging to a high status caste yields broader perspectives. As a consequence, high-status families are eager to keep up with what is expected from their status: they must face high conspicuous spending to signal that they deserve their rank. The survey was conducted among Haalpulaaren families who are known for still being very attached to the strict hierarchical organization of their society. Following Gubert (2000) I found that the family is not only insuring its income but also its pride: The risk against which the family seeks some insurance is not just the loss of utility due to a consumption shortfall entailed by a negative shock. The family is in fact protecting itself against the shame that would affect the whole clan if the family’s standard of living was to fall below some norm, corresponding to it social status (Azam & Gubert 2005). If distance provides some kind of risk diversification, it yields enforcement problems for both parties involved too. Azam and Gubert (2005) studied the moral hazard problem faced by the migrant: in the village, they showed how families who can rely on remittances, tend to shirk on agricultural activities, and underreport their incomes to their migrants. I will now turn carefully to the family issue: how can the family make sure that the migrant will remit? Indeed, when the migrant is a young son, and the family left behind primarily his old parents8 , they do not seem to be in a position to assist him in return as the mutual insurance contract stipulates, in case of unemployment for instance. Commitment without reciprocity In village economies, it is believed that low monitoring costs, along with concerns for reputation due to repeated interactions, perform a good enforcement device. Coate and Ravallion (1993) describe how enforcement can occur thanks to reciprocity. Nevertheless in their model, there is an a priori symmetry between agents coming into play. With an infinite horizon, at each period they randomly find themselves either in a position to help or in need of help from the other. Thus, if it were the case, one should observe transfers in both directions over time. ‘So, why individual with a high and secure revenue will share it with others while they do not ever benefit from the mutual assistance arrangements?’ in Ambec’s (2000) words. The same holds true in my framework, where the situation is unbalanced as the son cannot expect any support from his old parents who stayed behind in return for his remittances. Let us review all the mechanisms that ageing parents could possibly use to prevent their migrating son from defaulting. Firstly, all enforcement issues could be wiped out, if altruism or self-enforcing mechanism came into play. Altruism could indeed solve the puzzle, but all empirical studies show that the Pure Altruism model is never robust to data9. This underlines the fact that there is room for other types of motivation. Altruism also has some adverse side effects. First, the love the parents feel towards their son could make their punishment strategies less credible, as mentioned by Bernheim and Stark (1988). Moreover, parents could rely on their children’s altruism, knowing that in case of hardship they would remit more. As a result, parents would not even try to take measures aimed at insuring themselves against negative shocks. . Though we will see that in our sample survey, many migrants have also other dependants (wife, children) left in the household, so that there is room for some kind of exchange motive described later. 9 . See Laferrère and Wolff (2006) for a good review of the empirical literature on transfers' motives. 8 5 This is the Samaritan’s dilemma issue addressed by Coate (1995)10. So, apart from love and affection, what is the benefit to the son in taking good care of his old parents? Economists have tried since Samuelson (1958) to model self-enforcing inter-generational social contract. There is an obvious problem in the Children as Old-Age Security model: if parents, who invest in children to elicit their help when reaching old-age, are rational, what is the rationale behind the children paying back their parents? A detailed analysis of sustainable selfish transfers between three generations is the work of Cigno (1993). According to the so-called Family Constitution model, family transfers are seen as informal loan arrangements, adults being the only generation to earn a living. They lend to children and repay their aging parents. It is sustainable if the following family rule (a kind of collective punishment) is applied: One can deny support to one’s own parents if and only if they did not themselves take good care of their own parents. Thus the coresidence of grandchildren and grand-parents should be observed along with conspicuous spending by the son to take care of his old parents. As Guttman (2001) pointed out, this optimal rule cannot be implemented if changes in future economic conditions are foreseen (e.g. the introduction of a reliable pension system). This is most likely to be the case in developing countries. Consequently, this type of model does not seem suitable to our case, though denying the rights to future family solidarity, is a threat that could play some role, as we will see further on. Overall, this literature remains highly theoretical and has few testable empirical implications. Let us now turn to credible parent strategies which seem relevant to our case study. Some credible enforcement mechanisms Practically speaking, there are several mechanisms at play. All contribute to make this family arrangement work to some extent and it would be hard and even impossible to say that one plays a greater role than the others. Four credible strategies which parents could develop at several stages of the parent-child relationship will be briefly reviewed. Preference shaping Parents could invest in shaping their children’s propensity for being loyal. In his Nobel lecture, Becker (1993) argues that economists have excessively relied on altruism for the enforcement of inter-generational contracts and suggested instead accounting for the endogenous formation of preferences within the family (Becker 1993 and 1996). Parents try to instil a feeling of guilt if children do not respect social norms. In this view, investing in religious and traditional educations which inculcate family values would seem promising to shape a son’s loyalty. I will return to this later when describing traditional Koranic schools in my area of study. In Akerlof’s (1983) words, the Koranic school is a ‘loyalty filter’. However, manipulative attempts to inculcate loyalty seem likely to fail. To put it more simply: ‘Do as I say not as I do’ is notoriously ineffective. So, alternatively, parents could set the example by taking good care of their own parents. We shall see that the presence of own children in the household could induce the migrant to send remittances. Migrants’ selection . Coate recommends to send in-kind transfers, such as building irrigation schemes in drought-prone areas, instead of cash. Whereas it is implementable from an NGO's point of view, it seems rather unrealistic from the migrant's perspective... 10 6 The patriarch holds precise information on his sons’ character type. He could consequently select the most promising ones, i.e. those who combine both high earnings potential and degrees of loyalty. I try in my econometric implementation to account for this selection issue by estimating simultaneously migration and remitting behaviours. Exchange motive As in Lucas and Stark (1985), remittances can be seen as paying for services to the family who stayed behind to take care of children or assets of the migrant: cattle, land... In my survey sample, many migrants do have dependants (spouses, children) left in the household, so that there is room for some kind of exchange motive. Strategic bequest motive Strategic motives can also enter the picture: parents can threaten their children not to leave them any inheritance if they are not grateful. Thus children enter a competition to get the largest share of the bequest. Lucas and Stark (1985) had mentioned it, but the strategic bequest motive11 model is thoroughly described in Bernheim, Schleifer and Summers (1985). Remitting here is done to secure the bequest. Hoddinott (1994) tested this model on rural Kenyan data, where land was the inheritable part at stake. What happens then in a changing society when land ownership is no longer sought after? Furthermore, in most countries, inheritance division is settled by law. Islam for instance has very precise inheritance sharing rules: a son gets one share and a daughter gets half a share. My hypothesis for the enforcement of this old-age support arrangement derives from the Strategic Bequest Motive. It departs from their work by introducing a symbolic bequest rather than a physical one, namely: the transmission of the status of the family. The migrant values belonging to a prestigious family because it gives him access to the family’s network while on migration and when returning to the village he will reap the benefits of the status of the family by becoming a notable. Other mechanisms Nevertheless, apart from these parent strategies, if the migrant is planning on returning, reputation concerns may still exist but not as much with the parents as with the community. Lucas and Stark (1985) show that this plays an important role too. In their view, the migrant may remit ‘to enhance prestige or political influence’ for the days when he will return. Taking good care of one’s parents is a hallmark of reliability. We will see later, that this could be an incentive for migrants seeking wives within the community. The loss of prestige incurred by defaulting sons could also prevent their return for retirement, which is central to many African migrants. Last but not least, the desire to be buried at home is an additional motive to maintain the social status of the family. It is essential for many Muslim migrants, and in particular those from the Senegal River Valley. The main concern is more a deep attachment to homeland, than religious grounds. Many migrants associations were indeed created in order to collect money for the repatriation of dead bodies12. Let us now turn to some cultural features of the surveyed population to better grasp the intuitions underlying the study. THE HAALPULAAR SOCIETY OF MATAM . Not to be mistaken with the strategic motive for remittances specified by Stark (1995) chapter 4, who refers to high skilled migrants, which is not the case here. 12 . See Petit (2002) in Marseille, France. 11 7 Geographical position The Senegal River constitutes the border between Senegal and Mauritania. The Middle Valley is a semi-arid savanna sandwiched between arid zones of Ferlo and the Sahara desert. The harsh climate distinguishes two seasons. From June to August is the wet season with frequent showers and river floods. The rest of the year is the dry season with temperature rising up to 50°C before the first rain. On the fringe of the desert, the soil is poor so the most fertile lands lie by the river banks thanks to alluvial deposits due to the river’s natural flood cycles. The three main economic activities have always been fishing, cattle-herding and agriculture. There are 3 kinds of agriculture: the oldest being rain-fed cropping in the uplands (‘diery’) from June to August, and in the lowlands (‘walo’) cereals are planted after flood recession from September to December, and the most recent being irrigated cropping. Main crops are cereals (mostly millet, the traditional crop), rice (the main cash crop) and various vegetables (tomatoes, squash, sweet potatoes...). The introduction of water management facilities has reshaped the profits of these activities. During colonization, the French, and after independence, the state both invested in huge irrigation projects, and dams. The idea was that floods were unpredictable and sometimes devastating for villages on the river banks. Irrigation provides year-around crops and the dam is planned to provide electricity to the 3 neighbouring countries, i.e. Mali, Mauritania and Senegal. However the results are mitigated as the energy objective is not compatible with natural floods cycles: so far, floods are artificially maintained, but peasants fear it will not last. The dams have also been devastating for the fishing activity, along with the frequent droughts, reaching the point where most of the fish consumed in the valley come from the ocean, usually dried. Fishing has become a marginal secondary activity for fishermen who now engage in boat transportation which becomes a necessity when roads are flooded, as is usually the case for 3 to 4 months a year. Initially irrigation was very promising but the cost of oil and maintenance for motor-pumps, along with the cost of fertilizers and seeds make it profitable for rich farmers or efficient peasant cooperatives only. For the vast majority of the peasants of the valley it remains impossible to make a self-sustainable business out of it. As a consequence, nowadays, there are many schemes and water supplies that are badly-kept. In fact, conventional wisdom admits that it is rather difficult to make a decent living out of agricultural activities in the valley. In the meantime, migrants engage in conspicuous consumption either for religious celebrations or in housing, after just a few years abroad. Consequently, I collected data on remittance behaviour in the Senegal River middle valley where migration has long been considered a better strategy than working in irrigated schemes. In this context, the value of ‘physical’ bequests has become close to insignificant for a migrant son. Historical background of the Haalpulaaren13 The Haalpulaaren14 make up the majority of the inhabitants in the Middle Valley. Towards the river delta lay the former Walo empire of the Wolof whereas in the upper valley are the Soninke. Haalpulaaren can be found on both banks of the river, further North are the Moors . Based on two main sources: MAS (1959) and Wane (1969). . Meaning ‘those who speak Pulaar’, i.e. the language of the Fulbe (‘Peuls’ in French). They are also known as ‘Toucouleurs’ (or ‘Tukulors’) which sounds derogatory nowadays... 14 13 8 and more to the South in the Ferlo Fulani nomads can be found. Indeed, Haalpulaaren are believed to be descendant of Fulani nomads who settled in the valley and mixed with local populations (Serer, Wolof, Moore, Soninke...). In the Xth century, the region was referred to as Tekrur which is where the name Tukulors comes from15. It was first annexed by the Ghana Empire, and then the Mali Empire, until Koly Tengala Ba (a Denyanke clan of Fulani) in the XVIth century, freed the region from the Manding domination and founded the Futa Toro16. Islamization began as early as the IXth century in the area and spread until charismatic religious leaders took power in 1776. The ‘Torrobe’17 revolution brought an end to two centuries of Denyanke dynasty. In the XIXth century, El Hajj Futyu Umar Tall led a jihad recruiting among the Futankobe to resist the French army and founded the Macina Empire in modern Mali. The Haalpulaaren are responsible for spreading Islam all over West-Africa, and are highly respected for that in the sub-region. Koly Tenguela and El Hajj Umar are the icons of Haalpulaar pride, and descending from them along with other religious or political figureheads of the Futa is the most sought-after social recognition. Another very important aspect of the family’s fame is determined by which social category the family belongs to. The Haalpulaar caste system is rather complex, let us now consider its major features. Social stratification Traditional Haalpulaar society, like many Sahelian societies, (see Tamari, 1991) is based on a division into three main social categories: the ‘free born’, the artisans (musicians, metal workers...) and the (descendants of) slaves. To give an idea about the complexity of this system, Table 1 translated from Wane (1969) summarizes all categories18. The use of the term ‘caste’ is rather controversial, as it usually refers to the different artisan social sub-categories. In this paper, I refer to the caste system as the whole stratified social organization. Of course nowadays slavery has been abolished and artisans do not necessarily work in the branch of their group’s specialization, but still this classification has some meaning today. Indeed both the artisans and the descendants of slaves suffer from discrimination. Today, this caste-based discrimination mainly consists of marriage constraints, as well as denied access to land, and to political and religious leadership19. In the next section, we shall turn to a brief description of each category, and the main subcategories. The free-born (sometimes called ‘nobles’, ‘Rimbe’ in Pulaar) make up the majority and consequently are definitely the most heterogeneous group. The standard designation of the ‘free-born’ as ‘nobles’ may seem a little misleading, since this group includes even the most impoverished peasants. Indeed, the hereditary aristocratic lineages make up a minority within this category: they are the only families entitled to accede to political power (village chieftainship). Four main subgroups compose this first category depending on the main . According to Wane (1969). . Hence another name for the inhabitants of the region: the Futankobe (‘those of Futa (Toro)’). 17 . It means ‘beggars for alms’ in Pulaar, as they were studying Koran and therefore not economically productive. 18 . Though it is a benchmark in studies of Haalpulaar society, the main criticism of the Wane classification is that it omits the cattle-breeders, the Fulbe category, as part of the ‘free-born’. 19 . See for instance Sy (2002). I will also report different discrimination experiences towards those groups I witnessed during my stays. 16 15 9 activity they relate to. At the top of the hierarchy are the ‘Torrobe’ who are the religious leaders, along with some ‘Fulbe’ traditionally cattle breeders. Usually of lesser prestige than former sub-categories are the fishermen (‘Subalbe’) and the warriors (‘Sebe’). The artisans (‘Nyenbe’) are referred to as the ‘casted’ (‘les casté s’ in French) as this group matches best the idea of caste with job specialization (along with the usual endogamous feature it shares with the other categories). All the traditional crafts are represented from the blacksmith (‘Waylibe’) to the leather workers (‘Sakebe’) and the wood workers ( ‘Laobe’ ). Along with the craftsmen are the bards (‘Griots’ in French, ‘Aolube’ in Pulaar). Nowadays casted families do not necessarily practice in their specialized activity but they usually still have been transmitting the know-how of their job. They cannot have access to the local political power. At the bottom end of the pyramid are the former slaves (‘Maccube’), rather called captives. In fact at times of wars and ‘jihads’, prisoners ended up as slaves. As a result, many family names of slaves are of other ethnic group origins20, and as a consequence stigmatized, although many slaves took the name of their master. Although slave status was legally banned over a century ago21, their status has evolved slowly. Nowadays, in Haalpulaar society descendants of slaves are still expected to be deferential to nobles, to cook on festive occasions and sometimes repair their former master’s house22. They distinguish themselves by their willingness to work hard. An illustration of their position in the society can be found in ‘joking relationships’. Depending on one’s family name, one is allowed to tease people from another clan (one’s joking kin) without offending anybody. Usually the jokes are very humiliating and are centred on either greed for food or calling the other your slave (and the other reciprocates) During my fieldwork, I once had the opportunity to work with a man of slave origin who would systematically refuse to interview people belonging to his joking kinship. No matter how bright an educated he was, he explicitly told me how he felt uncomfortable with such jokes, and did not feel like reciprocating. The main form of discrimination is the fact that only the free born and the most important artisans may traditionally own the most valuable lands by the river banks (‘walo’ in Pulaar). Nowadays, cases can be found of slaves’ families that now own such lands after their former masters gave them away but it is still rather an exception. Regularly in Senegalese newspapers, there are articles about mixed marriages (‘inter-castes’) which prove that it is still a scandal, even in the capital city, or abroad23. In villages it is all the more difficult for an artisan or the descendant of slaves to marry a noble woman. As a consequence, before a marriage occurs, there is always a preliminary enquiry led by the inlaws, to make sure of the compatibility between both social categories of the groom and the bride. In reality, it is never the case that all these categories can be found in one village. Usually there are villages founded by fishermen others by warriors, or even slaves usually near to nobles’ villages. Nevertheless, there are some rules that always apply: the Imam will . …and usually of noble origins, as the artisans were not enslaved but kept their specialization. . See Klein (2005) for more details. 22 . Each year after the rainy season, traditional mud houses need to be repaired, especially the roof. I witnessed this in two different villages during my stay. See for instance Klein (2005) for similar accounts. 23 . It is interesting to look at the frequency of inter-caste marriages as the subject of many discussions in Senegalese forums (often of the diaspora). 21 20 10 always be chosen among the Torrobe, the village chief and notables among the dominant free-born category (fishermen, warriors, Torrobe or Fulbe). These important families have usually taken along their dependent families: their former slaves and artisans. It is clear that all these social categories were not fixed, but that they evolved as some commoners have strategically been conferred a title of nobility as a reward for good services, or new warriors recruited among former slaves24... It is important to keep in mind that these categories corresponded to a genuine need for a specific job allocation during the former era. The official abolition of slavery along with the emergence of the modern market economy render this classification outdated on practical grounds. Nevertheless it is striking how relevant this classification remains in contemporary Haalpulaar culture. There is for instance a set of implicit rules that each group must abide by. These rules are mainly driven by ‘the sense of honour’ which is specific to the nobles. Let us now turn to this aspect in details and in its implications. The Nobles’ Code of Honour There are many ways for higher status individuals to distinguish themselves from lower status ones. Bodily posture, where one sits and how one addresses others in public meetings are manners of displaying one’s rank. Typically, although not specific to Haalpulaar society, lengthy traditional greetings are relevant to the social status of protagonists. One aspect is the repetition of the family name, which is called ‘yetode’ in Pulaar and literally means ‘the name that honours’. Ly (1966) described the attitude of the Haalpulaar aristocrats very precisely in his thesis. According to Ly, they are characterized by their sense of honour which does not exist in other classes. At the core of this principle, is the importance of always being in the position of giving and never asking. It is by giving and being publicly generous that the noble will be able to demonstrate his worthiness. When belonging to such families it is an unbearable shame to have economic difficulties. There are great pressures to display generous behaviour. Maintaining a high social status is expensive as it means giving to their griots and many clients (former slaves, artisans...). The size of the clientele represents the power of the family. Along with the warm-glow effect of giving, griots directly raise the utility of the giver by advertising his generosity, which directly improves his reputation25. Important families ought to give to their former slaves and bards during baptisms, weddings and religious events. In this traditional view, your birth determines your personality (see Wane 1969 who depicts each caste with a specific personal trait). It is said for instance that one should not trust an artisan because he may just be trying to seduce oneself in order to receive gifts. Not knowing the whole genealogy of one’s family is also a problem: it is the case that former slaves in particular have lost track of their ancestors... Hence the similarity between the Pulaar word for ‘slave’, ‘maccube’ and ‘majjube’ meaning ‘somebody who lost his/her way’ (Wane, 1969). In a stratified society ruled as a gerontocracy, an old man may benefit from an advantageous position in local politics if and only if he belongs to a high ranked family. . The well-documented historical novel Les Peuls by Tierno Monénembo perfectly illustrates how mobile these social categories were over generations. 25 . See Azam (1995) for an interesting economic interpretation of such norm. 24 11 Only a few families have a say in choosing the village leader, and even less have a chance to become one. Hence, the family name becomes a valuable symbolic bequest for such important families. It is well known that nowadays, many former slaves, or bards take advantage of their position too, and like to remind their former masters of their duty to provide them with gifts. Thus respecting this social pressure to give requires high income. Migration appears as a solution to finance these conspicuous expenses. The Baraka motive I was told many times during my investigations (both in France among the Diaspora, and in Senegal) that a very strong motivation for remitting is to earn blessings and prayers for felicity. Here religion and traditional beliefs interfere. Indeed there are some mystical practices in Haalpulaar, Wolof and Serer traditions, consisting in calling the migrant to come back home (Sylla & M'Baye 1991). There are legends26 about migrants coming back home in a rush and losing memory of how this happened and then going insane. This is said to be adapted as a last resort, a retaliation on an unreliable son. Repeatedly I was told on my first visit in Senegal and many times afterwards also by migrants in France, that it was very important to be near your dying parents as they will transmit the baraka of the whole dynasty through their last blessing27. Indeed a dying old man will call his beloved sons, and give them his final blessing: the ‘duhahu’ (in Pulaar) which gives baraka (or ‘barka’ in Pulaar). This Islamic word could roughly be translated as Allah’s blessing, and it has strong influence throughout the whole Islamic world28. It is a rather complicated concept as many anthropologists are still trying to properly define it. It is important to bear in mind that the final blessing is strongly linked to good luck, wealth, happiness, health, fertility, and social success. Being rich and healthy is not a sufficient condition to possess baraka: Isn’t he old and abandoned today! Didn’t all his spouses go back to their family! He is so bad-tempered that only one of his grandsons agreed to stay with him. To take care of his cattle he must hire shepherds who cheat him while he is unable to catch them. Where do you see baraka here?29 reports a villager talking about his neighbour who is the biggest cattle owner of the area in a Fula region of Cameroon. It can be acquired hereditarily through genealogical link to saints, or their contacts (see Schmitz 2000), but also by a strict observance of the five pillars of Islam. It can be transmitted by blessings. Being old and a well known good Muslim puts oneself in a good position to have some. In my different stays in Northern Senegal I witnessed many blessings by older people: when giving a ride to an old man, he would make me open my palms, murmured Islamic blessings, and blowing in my hands, then I would lay my palms upon my face30. Since living abroad makes it rather difficult to be physically near your dying parents, . I was told many similar stories during my stay in the region of my survey. . The importance of the parent's last blessing is not specific to Islam. It is mentionned in many parts of the Bible, and particularly in the Genesis. Think of Jacob's story: he valued so much his father's blessing that he managed to get blessed instead of his brother Esaü. 28 . And beyond, as it can be linked to the Christian word grace or the Jewish word berakhot, or also the Polynesian manna, although these related words are comparable they are not completely similar to baraka. 29 . Dognin (1990), my translation. 30 .This seems quite related to a local Sufi practice, detailed in Schmitz (2000). 27 26 12 sending money appears as one practical alternative. Of course, the value of this baraka in the eyes of the migrant strongly depends on the notability of his patriarch and consequently on the status of the family. Obviously belonging to a prestigious family yields considerable returns in terms of prospects for marriage and political position in the village. While on migration, he could also benefit from the status of the family to rely on migrants from the same village. Besides, migrants from the same village usually live in neighbouring rooms in workers’ hostels (‘foyers de travailleurs’) reconstituting the village society. It has been repeatedly suggested that this is a way of reproducing hierarchic village society (for instance see Ba 1996: 207). As this is the anthropological aspect of my research, I will try to translate it into testable empirical evidence from the survey data I collected in the area. THEORETICAL FRAMEWORK Aware that it may not be the only mechanism at work, I will emphasize here the role of social status. Let us consider a patriarchal stratified society. Within the family the patriarch is the decision-maker. At the village level, there are two groups of families: the nobles and the others (non-nobles). Only the patriarchs from noble families can accede to influential positions. To deserve their advantageous dominant position, the nobles must pay nonnobles substantial gifts in monetary terms. The non-nobles who benefit from these transfers must in return renounce basic human rights: not choose a spouse among the nobles, give up their political ambitions, and admit the superiority of the nobles in community decisionmaking. There is an implicit family contract between the patriarch and his adult sons. The patriarch sends his most promising sons on migration, expecting them to remit. The motivation of migrant sons is to obtain the father’s position on their return. In this view, remittances should be observed for high ranked migrants only. The migration of the lower status migrants relates to other motivations: trading monetary gifts for social mobility. They will loosen ties with their community, and make their new life abroad, taking advantage of the anonymity of city life. Let us now turn to the data collected to test this scenario. DATA The Survey31 Of course, the link between the migrant and his relatives who stayed behind is central to understanding remittance behaviour. This is not an issue in my study, since I focused on the case of sons remitting to their ageing father. So, in the survey sample, all migrants have in common that their father is alive and lives in the village. Hence, remittances are interpreted as old-age support from the migrant sons. The survey was conducted just before the rainy season from April to June 2004, when weather conditions are particularly tough. It is a time of drought and food scarcity as the granaries are empty. The good thing is that since peasants are preparing the fields for the rain cultivations, all members of the household who are away, but within reach, come back to give a hand. The survey covers 149 patriarchs over 6 villages randomly chosen around Matam32. Three . I must thank Ousmane Ba, Moussa Ly, Abou Sow, Abou Thiam and Hamath Thiam for doing a great job in data collection, as well as providing useful assistance in the survey design. 32 . Diery villages are: Thiambe, Galoyabe Toucouleur, Hombo. Walo villages are: Bow, Thiemping and 31 13 are located by the Senegal river, in the so-called walo region, the other three are located approximately 10 km away from the river, in the diery region, one of them being right by the highway (the others being 4km away on a dirt track). Walo villages are isolated (at least) 3 months a year during the rainy season as roads become unfit for vehicles, which can only be reached by pirogues (2-3 hour instead of 20 minute ride during the dry season). Diery villagers do not own irrigated lands which are located near the river, but some noble families do own ‘walo’ lands by the river. It is an exhaustive survey, as all the men of the selected villages who met chosen criteria were selected. The criteria are that they must be born before 1939, and that their eldest living son should have been born before 1979. Information was gathered at 3 levels: • The patriarch level: Information on both his social status and his living conditions were gathered. To measure social status I gathered information on the family background such as which caste it belongs to, its tenure in the village, the political and religious role played by the patriarch and his father. Then much emphasis was put on where the patriarch gets support from, but also on his wealth (housing, fields, cattle, belongings...), health, nutrition, education, past migrations and economic activities. • The household level: I define a household as a unit of consumption and income sharing (sometimes several households can coreside in the same compound), which can be made of several nuclear family units33. I collected very precise family links between household members, and some basic information on each one, such as: education level, economic activity, sources of revenue... • Finally, information on each child has been collected. Special attention was given to children living outside the household, for which I gathered indirect information with spouses, mother, siblings, or the father. The main focus was on remittances sent by migrant sons and their old father, and the rest of the family. Information on news and visits given to the family were also collected. I put much effort on making sure to have at hand the most precise and reliable information on the social status of the families. For double-checking purposes, the status of the family was gathered twice. First, I obtained the information directly by asking the patriarch, then indirectly, on a second round, asking informants (mainly the village chief). As it can be a very sensitive subject, the question was asked towards the end of the survey, in order to prevent the patriarch from being offended or upset, or even from refusing to be surveyed. The question was asked among a series of question on historical background of the family, to make it sound like the family ‘social origin’. For the same reasons, I dared to ask the village chief about the status of my surveyed families, on a second round in September, so that we had already got to know each other better. Descriptive Statistics A detailed table of descriptive statistics for the initial sample, the migrants and the remitters is shown in table A in the appendix. Thially Souballo. All villagers and especially village chiefs must be gratefully acknowledged for their cooperation and kind hospitality. I would also like to thank Ousmane Diouf, and Bocar Ba’s family in Matam for their logistic assistance and gracious hospitality. 33 . The Pulaar word is ‘foyre’ (plural: ‘poye’). 14 Social composition of the sample A table of the caste composition of the different samples studied is displayed in table 2. Only the sons of the patriarchs above 16 years old were considered as potential old-age support providers. There are 470 adult sons in the initial sample. 45% are on migration. Among the migrants, 62% regularly remit. It is noteworthy that only 66% of the migrants belong to ‘free-born’ families, whereas they represent 71% of the remitters. Migrants’ destinations The destination distribution is very similar to previous studies of Haalpulaar migration: mainly domestic, the international migration is mainly African, and very few have left the continent. 44% of all migrations are domestic migrations, or migrations to the neighbouring countries, i.e. Mauritania, Mali, Gambia and Guinea. Dakar concentrates 77% of these ‘domestic’ (Senegal or neighbouring countries) destinations. However, the capital city is often just a temporary stop on the way to further destinations. 38% of migrants went to another African country. Most of African migration (73%) is directed towards Central Africa: Gabon, Congo, and Cameroon. There is a long tradition of Senegal River Valley migration in this area. There are accounts of ‘diamantaires’ who made quite a good fortune in these countries34. Côte d’Ivoire has also been a frequent choice of destination, but since 2000 because of the political instability, and xenophobic practices many have fled. In the end, only 18% of the migrants left the African continent (they represent 39 migrants): not surprisingly, most went to France (26 migrants). The USA is the second destination (9 migrants), then a few migrants are in Spain or Italy (5). The US is an increasingly attractive destination, but this is still a recent phenomenon, so that the network is still weak, not to mention the cost of migration. During my field investigations, I met returning or visiting migrants from Germany and Denmark, and also surprisingly distant destinations like Australia and even China! Even if this is not statistically significant, it indicates that diversification of destinations is used to circumvent tough migration policies. It is important to bear in mind that information about migrants is only second hand. I assumed the best information I could collect about them would be from the mother or the wife of the migrant. I also asked the brothers and the father if necessary. Empirical Strategy Since only those who have migrated can make transfers, I had to use a two-step Heckman model (generalized for logit models) to take into account the selection bias. The question is : Do the ones with a higher probability of migrating have a higher or lower probability of remitting? As pointed out in the next section, we do not find such effects. Furthermore, since I do have migrants from the same families (up to 7 brothers!), I incorporated the correlation inside the family using a robust estimator of standard errors with family clusters. The descriptive statistics evidence found in the previous section need to be checked ‘all things being equal’. Hence, let us now turn to the regression results. They are displayed in table 3. . Starting in the mid-fifties in Sierra-Leone, near both sides of the Congo rivers by the sixties, they have reached Angola and Zambia, striving to avoid being expelled, Bredeloup (1994). 34 15 When computing the probability of migrating, in line with Gubert (2000), I decided to take a snapshot of the family when each son migrated. I computed his age and the age of the patriarch when he left the household (‘age when migrating’, ‘age of the patriarch when son migrated’), along with the number of brothers already on migration at that time. In the case of the non-migrant, the age is left unchanged, so the ‘age when migrating’ corresponds in reality to the age at the time of the survey. The same holds for the age of the patriarch. Also, the ‘number of brothers already on migration’ simply corresponds to the total number of brothers in migration at the time of the survey. 16 RESULTS In reality I find that both decisions to migrate and then to remit are not correlated: rho is not significantly different from 0. Therefore, these two regressions can be studied separately. For both regressions, usual results in line with previous studies are found along with other different findings. Results are displayed in Table 3. Probability of migrating Overall this model correctly predicts 71% of migrating behaviour. Nevertheless only 53% of actual migrants are properly predicted by this model. This highlights the fact that there are in reality many different motives for migrating, which is not in contradiction with the theoretical framework developed here. The variables that appeared not to be significant are notably: years of formal schooling of the migrant35, and the number of migrating brothers. The fact that the number of years in formal schooling is not significant is relevant to the fact that migrants do not need to hold literacy skills for the job they will find in migration. One must keep in mind that most migrants work in the informal sector, in Dakar and Africa, and in unskilled jobs in industrialized countries (mainly in the hotel or catering trades). As for the number of migrating brothers, there are two opposite effects: a network and a saturation effect. Having a brother can facilitate migration: older brothers can finance their cadets, creating shifts of migrants36. On the other hand, when there are already migrants in the family, the patriarch may require other sons to stay to help him manage the household. Here it seems that both effects are neutralized. Turning to significant variables, the fact that his father spent many years on migration increases the chance of the son migrating. There are here 2 intertwined positive effects: a wealth and a network effect. All migrations are costly but especially for distant destinations: fares, bribing to get documents, or to get across the border safely. It usually represents far more than what a basic rural household can afford. Then there are some high costs of accommodation: even in Dakar renting a room has become very expensive.... Of patriarchs who have been on migration around 55% receive a pension. However the ones who did not get a formal job usually managed to invest in productive activities financing their own sons’ migration and settlement. Consequently, it appears that sons of economically better-off families have a higher chance to migrate. Along with this wealth effect, there is a network effect: fathers who have been on migration have been able to build networks in their destination and acquire accordingly easy entry on some job markets. The best illustration of this is that many first migrants of this area started working in the hotel trade in Dakar, many at the airport, then in France. Or even working as boy or cook for the French colonialists who then took them to France to work for them. This result is in line with other studies. Another interesting result is that sons that have received thorough Koranic studies have a higher chance of migrating. Indeed in the area most children, both boys and girls, attend at least the first years of Koranic schools in order to be able to recite basic prayers. As . I had to drop this variable from the regression in order to keep the ‘Koranic studies’ variable: indeed, only those without formal schooling may be enrolled in Koranic studies. 36 . In French this is called ‘noria’ (see Ba 1996: 30). This can no longer hold for destinations which have become very difficult to reach such as Europe... 35 17 mentioned before this central result illustrates the idea that Koranic schools are above all aimed at transmitting moral values and among them the respect for one’s own parents. However when going up the levels of Koranic schools the two school systems enter into competition, as they are rarely compatible. Parents, traditionally fathers, have to decide which education to give to their sons (girls usually do not pursue Koranic studies for many years). This must be put in parallel with the fact that having formal schooling does not improve the probability of migrating. Here a father who wants to send his son on migration will prefer to previously send him to a Koranic school instead of the official Francophone school system. Finally, the age of the father has a negative but decreasing effect, whereas the age of the son when migrating has no effect. Last but not least, sons of families belonging to the higher castes are less inclined to migrate. This result contradicts Gubert’s (2000) findings among Soninke: she found that only aristocrats migrate. It is necessary to turn to migrants remitting behaviour to understand if migrants from high and low status have the same behaviour once in migration. Probability of remitting Among the migrants, I turn now to their probability of remitting. This model has a good predictive power as overall nearly 80% of migrants are properly predicted to remit or not. First, the age of the migrant appears not to be significant for his remitting behaviour. The size of the household, as well as the presence of his own children or spouse in the household also has no impact on his probability of remitting. Some expected ‘wealth effects’ are found: data on the migrant’s earnings were not available, but there are some proxies. The time spent on migration appeared not to be significant at first, but I re-coded it into a dummy variable for migrants who have been on migration for over 3 years. This dummy has a positive impact on the probability of remitting, which is significant at the 10% level. This reflects the idea that it usually takes time for a migrant to obtain a secure earning, i.e. regular wages. Similarly, living in a developed country positively affects his probability of remitting. This is obviously linked to the higher earnings available there: he earns more than his own living expenses and can accordingly send back a part of his earning. Surprisingly, no monitoring effects are found here: having brothers living in the same place of migration has no significant effect on his remitting behaviour. I found that migrants who had followed religious schooling have a higher probability of remitting. Hence the idea of migrant’s selection by the patriarch. He will favour the migration of the son who has received a traditional Koranic education, being more confident about his loyalty. It is consistent with the preference shaping theory I mentioned earlier. There is a trigger effect from his brothers remitting behaviour too: the more brothers who are remitting there are, the more likely it is that he will remit, and the more who are not remitting, the more likely it is that he will not remit. Finally, central to the main point of this article: belonging to a politically powerful family in the village makes the migrant more likely to remit. Indeed, this status dummy variable is strongly significant at a 5% level (p-value of 3.1%). Here what seems to be central is the fact that the patriarch plays a significant role in the village belonging to the circle of influential notables. Indeed, belonging to the free born category is not precise enough, as 18 this is a heterogeneous category. For instance, newly arrived free born families may not have regained a dominant position in the village37. On the contrary, a few families from the artisans or former slaves groups may have acquired a political position being linked with a very prestigious family. One could suggest that those families are also more likely to possess the most valuable lands, and as a result this is in line with Hoddinott (1994). Indeed having received valuable land by bequest has no significant effect of the probability of remitting. Therefore it seems that the effect here of the political status variable is mainly a social status effect and not a wealth effect. Other Supportive Evidence The econometric evidence may seem a little weak as the survey does not cover a huge sample of migrants: there are only 213 migrants in the sample. Unfortunately the lack of social status variables in other larger statistical surveys makes it impossible to compare this result with other studies so far. As a complement to the empirical evidence some other qualitative pieces of evidence which confirm the hypothesis made in this paper are listed here. To gain respectability in the village the most efficient way is to contribute to the building of a mosque. There are several anecdotes I heard during my stay in the valley, about the boycott of mosques built by migrant slaves. The Imam who is necessarily a noble, simply refused to lead the prayers... I heard a similar story about a well built by a successful slave that the nobles boycotted, opting for the more distant well. These anecdotes show that it is rather inefficient for a slave to invest in order to raise his respectability in the village: he will always be considered as a slave and accordingly will be reminded to keep his lowprofile rank. The only way to get rid of these social stigmas is to migrate. In anonymous cities or foreign countries, he will more easily merge into society and be free in his acts. It is noticeable that during my repeated visits at different workers hostels, I met very few descendants of slaves or artisans38. This is consistent, with reported cases of continuing discrimination against slaves on migration. The pattern is that slaves are excluded from any decision-making position in the migrant’s association. Ba (1996) reports such a case in Cameroon (Ba 1996: 238). In Dakar, many people with Haalpulaaren names who do not speak Pulaar happen to be of slave or artisan origins. They seem to be the product of this melting pot. In many previous studies on migration among the Haalpulaaren it has been mentioned that the desire to escape from rigid social constraints were among the main motives for migration. As they usually refer to the position of younger brothers willing to escape the tyranny of older brothers (see Platteau 2000), it could easily be extended to lower status migrants. The study of Delaunay (1984) goes in the same direction as my results. He finds that any type of caste can be found among the migrants from the Senegal valley, but recognizes that the former slaves are more likely to settle definitively, whereas the nobles go on temporary migration returning to the village, and sending remittances. CONCLUDING REMARKS . It is the case of many Mauritanian refugees (from the 1989 unrest) who despite being Fulbe have rarely gained a position as notable in their new Senegalese settlement. 38 . except the descendants of slaves of the village chief's family, which would tend to prove that former slaves of important families have some kind of respectability too. 37 19 These empirical findings suggest that migration is available to any social category, but for different aims: it can either be seen as an individual decision to escape from a poor social position with dire prospects in the village or, conversely, as a family decision to maintain its dominant local position. As a consequence, it is obvious that the first type of migrants will not be interested in investing in the development of the village but will benefit from the urban melting pot to get rid of their social stigmas. The good side of this is that they are factors of change as they move from a network-based to a modern market economy. In order to prevent rural exodus and urban concentrations, an efficient policy would be to invest in intensive consciousness-raising campaigns to get rid of the obsolete caste system. This will probably take time and be hard to implement as I found out that this was still a highly taboo subject in contemporary Senegal. Central to political debates in France, republican values of racial integration and assimilation may differently appeal to both types of migrants previously described. France’s model of integration aims to assimilate immigrants by instilling French cultural values. As a result, keeping strong ties with the country of origin is not well received. Accordingly, any attempt to assimilate the migrants of noble origins will fail. Indeed, they see themselves as temporarily migrant and have sound plans to return to the village. When returning they will genuinely keep silent their shameful subordinate occupations39 while abroad and turn into powerful notables. Alternatively, the migrants from lower social categories will display strong motivations to fit in. The aim of this study was primarily to draw attention to the local institutions when trying to understand migration in a non-western environment. Economists and econometricians too often draw conclusions from surveys that were lacking the social context. As this was a limited study, it would be interesting to pursue on a broader range. The recent World Bank book Culture and Public Action (Rao & Walton, 2006) sends a good signal that things are changing and that more and more cooperation between economists and anthropologists will be expected in the near future. REFERENCES Ambec, Stefan. 2000. “Income Sharing Within Extended Families as a Social Norm.” Mimeo, University of Salerno. Azam, Jean-Paul. 1995. “L’Etat Autogéré en Afrique.” Revue d'Economie du Développement 4: 3-19. Azam, Jean-Paul and Flore Gubert. 2006. “Migrants’ Remittances and the Household in Africa: A Review of Evidence.” Journal of African Economies 15 (AERC sup. 2): 42662. Ba, Cheikh Oumar. 1996. “Dynamiques Migratoires et Changements Sociaux au Sein des Relations de Genre et des Rapports Jeunes/Vieux des Originaires de la Moyenne Vallée du Fleuve Sénégal.” PhD thesis, Université Cheikh Anta Diop de Dakar. Becker, Gary S. 1993. “Nobel Lecture: The Economic Way of Looking at Behavior.” Journal of Political Economy 101: 385-409. ----. 1996. Accounting for Tastes. Cambridge: Harvard University Press. . I can witness that it is absolutely taboo to ask a visiting migrant about his job in the host country (or in the capital city). As one might guess such work was not prestigious, it is generally accepted to keep it silent... or to tell it only to a total foreigner away from other villagers! 39 20 Bernheim, B. Douglas, Andrei Schleifer and Lawrence H. Summers. 1985. “The Strategic Bequest Motive.” Journal of Political Economy 93:1045-76. Bernheim, B. Douglas and Oded Stark. 1988. “Altruism within the Family Reconsidered: Do nice guys finish last?” American Economic Review 78: 1034-45. Bredeloup, Sylvie. 1994. “L’Aventure Contemporaine des Diamantaires Sénégalais” Politique Africaine 56: 77-93. Cigno, Alessandro. 1993. “Inter-generational Transfers without Altruism: Family, Market and State.” European Journal of Political Economy 9: 505-18. Coate, Stephen and Martin Ravallion. 1993. “Reciprocity without Commitment: Characterization and Performance of Informal Insurance Arrangements.” Journal of Development Economics 40: 1-24. Cox, Donald. 1987. “Motives for Private Income Transfers.” Journal of Political Economy 95: 508-46. Delaunay, Daniel. 1984. De la Captivité à l'Exil : Histoire et Démographie des Migrations Paysannes dans la Moyenne Vallée du Fleuve Sénégal. Paris: ORSTOM. Dognin, René. 1990. “L’Arbre Peul.” Cahiers des Sciences Humaines, 26 (4): 505-29. Ferreira, Monica. 2005. “Elder Abuse in Africa: What Policy and Legal Provisions Are There to Address the Violence?” Journal of Elder Abuse & Neglect 16, 2: 17-32. Gubert, Flore. 2000. “Migration et Gestion Collective des Risques. L'exemple de la région de Kayes.” PhD thesis, Université de Clermont-Ferrand I. Guttman, Joel M. 2001. “Self-Enforcing Reciprocity Norms and Inter-generational Transfers: Theory and Evidence.” Journal of Public Economics 81:117-51. Hoddinott, John. 1992. “Rotten Kids or Manipulative Parents: Are Children Old-Age Security in Western Kenya?” Economic Development and Cultural Change 40:545-65. Klein, Martin A. 2005. “The Concept of Honour and the Persistence of Servility in the Western Soudan.” Cahiers d'Etudes Africaines 45, 179-180: 831-51. Laferrère, Anne and François-Charles Wolff. 2006. “Microeconomic Model of Family Transfers.” In Handbook of the Economics of Giving, Altruism and Reciprocity, Vol. 2, edited by S.C. Kolm and J.M. Ythier, ch11. Amsterdam: North-Holland/Elsevier. Luke, Nancy and Kaivan Munshi. 2005. “Women as Agents of Change: Female Income, Social Affiliation and Household Decisions in South India.” Mimeo, Brown University. Ly, Boubakar. 1966. “Honneur et Valeurs Morales dans les Sociétés Ouolofs et Pular.” PhD thesis, Université de Paris. Miguel, Edward. 2003. “Poverty and Witch Killing.” BREAD Working Paper 41. Mission d’Aménagement du Fleuve Sénégal (MAS). 1959. Les Hommes du Fouta Toro. Saint-Louis: MAS. Monenembo, Tierno. 2005. Les Peuls. Paris: Seuil. Nugent, Jeffrey B. 1985. “The Old-Age Security Motive for Fertility.” Population and Development Review 11, 1: .75-97. Petit, Agathe. 2002. “L’Ultime Retour des Gens du Fleuve Sénégal.” Hommes et Migrations 1236: 44-52. Platteau, Jean-Philippe. 2000. Institutions, Social Norms and Economic Development. Amsterdam: Harwood Academic Publisher. Rao, Vijayendra and Michael Walton, eds. 2006. Culture and Public Action. Stanford: Stanford University Press. Rosenzweig, Mark R. 1988. “Risk, Implicit Contracts and the Family in Rural Areas of 21 Low-Income Countries.” Economic Journal 98, 393: 1148-70. Rosenzweig, Mark R. and Oded Stark. 1989. “Consumption Smoothing, Migration and Marriage: Evidence from rural India.” Journal of Political Economy 97: 905-26. Samuelson, Paul A. 1958. “An Exact Consumption Loan Model of Interest with or without the Social Contrivance of Money.” Journal of Political Economy 66: 467-82. Schmitz, Jean. 2000. “Le Souffle de la Parenté. Mariage et Transmission de la Baraka chez les Clercs Musulmans de la Vallée du Sénégal.” L'Homme 154-155: 241-77. Stark, Oded. 1991. The Migration of Labor. Oxford: Basil Blackwell. ----. 1995. Altruism and Beyond: An Economic Analysis of Transfers and Exchanges within Families and Groups. Cambridge University Press. Sylla, Omar and Mor Mbaye. 1991. “Psychopathologie et Migration: un Cas de Wootal.” Psychopathologie Africaine XXIII 3: 353-63. Wane, Yaya. 1969. Les Toucouleurs du Fouta Toro: Stratification Sociale et Structure Familiale. Dakar: IFAN. World Bank. 2005. Global Development Finance: Mobilizing Finance and Managing Vulnerability. Washington: World Bank Publications. World Health Organization (WHO). 2002. The Toronto Declaration on the Global Prevention of Elder Abuse. Geneva: WHO. 22 TABLE 1 The Haalpulaar complex caste system (translated from Wane 1969: 33). Social Category (order) Strata (class) Rimbe ardiibe (free) Caste Toorobbe Lawakoobe or fileteebe Jangube or ardiibe diine Miiskineebe Worgankoobe Kolyaabe Mbonaabe Awgalnaabe Helgalagel, etc. Occupation and Title Temporal power Spiritual power : ceerno Free men, farmers: demoowo Former monarchs: farba ceddo Koly’s soldiers : jagaraf Local variants of kolyaabe Courtiers and advisers: sooma Fishermen: jaaltaabe Weavers: jarno Blacksmiths: farba baylo Jewelers: farba baylo Shoemakers: foosiri Cobblers Pirogue builders: kalmbaan Kitchen tools’ makers: maalaw Potters and ceramicists Traditional guitar players Singers Griots genealogists : farba gawlo Free but servile Slaves and servants: jagodiin Rimbe Rimbe huunybe (courtiers) Sebbe Jaawambe Subalbe Fecciram golle (artisans) Maabube sanyoobe Wayilbe Baleebe Sayakoobe Alawbe Sakkeebe Wodeebe or garankoobe Laade Lawbe Nyeenybe Worworbe (bobi) Buurnaabe Wambaabe Naalankoobe Maabube suudu Paate (entertainers) or nyaagotoobe (praise Maabube jaawambe Lawbe gumbala (kontimpaaji) singers) Awlube Soottiibe (freed slaves) Maccube Jyaabe Halfaabe (dependants) Maccube 23 TABLE 2 Caste composition of samples of interest. Categories Free-born Artisans Slaves Patriarchs (%) (N=147) 70.1 12.9 17.0 All Sons (%) (N=470) 72.1 11.7 16.2 Migrants (%) (N=213) 65.7 16.9 17.4 Remitters (%) (N=133) 70.7 15.8 13.5 24 TABLE 3 Two-stage probit regression: probability of migrating and then of remitting. Est. Probability of migrating (N=470) Intercept Age when migrating Koranic studies Nb. of brothers already on migration Age of the patriarch when son migrated (age of the patriarch)² Nb. of years the father went on migration Family of higher casts Village dummies Probability of remitting (N=213: migrants) Intercept Ages Koranic studies On migration for over 3 years Brothers in region of migration Residing in a developed country Spouse left in the household Nb. of brothers remitting Nb. of brothers who don't remit Age of the patriarch Local power of the lineage Valuable agric land on bequest Number of dependants for patriarch Village dummies Rho 95% conf. Interval (1) Robust std. Errors (1) Prob. Values 40.358 -0.001 0.420 -0.065 -1.070 0.007 0.024 -0.459 Non sig. 9.843 0.010 0.177 0.061 0.270 0.002 0.008 0.216 .000 .955 .018 .287 .000 .000 .003 .034 *** ** *** *** *** ** -3.525 -0.002 0.435 0.510 0.270 1.311 0.389 0.140 -0.365 0.046 0.493 -0.318 0.011 Non sig. -0.554 1.532 0.017 0.251 0.278 0.263 0.413 0.246 0.085 0.120 0.021 0.228 0.250 0.034 .021 .926 .084 .066 .306 .002 .113 .098 .002 .029 .031 .203 .757 ** * * *** * *** ** ** -0.967 .664 adjusted for family clusters. ***1%, **5%, *10% significativity levels. 25 TABLE A Descriptive statistics for the initial sample, the migrants and the remitters. Initial Adult Sons (N=470) Mean Std. Min. Max. 45% 30.8 28.1 2.6 53% 36% 2.4 8.4 8.3 1.6 17 3 1 61 51 8 Migrants (N=213) Mean Std. Min. Max. 62% 33.2 27.4 2.4 58% 33% 2.4 6.1 54% 18% 41% 1.0 1.3 0.8 74.9 69.1 7.4 33% 13.9 66% 46% 61% 28% 15% 5% 15% 18% 18% Remitters (N=133) Mean Std. Min. Max. On migration Remitting Age Age when leaving the household Birth order Koranic studies Formal schooling Years of (formal) schooling Years on migration In migration for over 3 years Residing in a developed country Spouse (of the son) left in the household Nb. of brothers already on migration Nb. of brothers remitting Nb. of brothers who don't remit Age of the patriarch Age of the patriarch when son migrated Number of dependants of the patriarch Family in region of migration Nb. of years the father went on migration Family of higher castes Local power of the lineage Valuable agric land on bequest Thiempeng Bow Thially Galoyabe Hombo Thiambe 8.1 8.3 1.5 17 3 1 61 50 8 3.7 0 17 4.2 5.7 0 0 17 25 1.0 1.3 0 7 FAMILY 73.9 71.2 7.6 10.6 72% 42% 58% 31% 18% 5% 14% 16% 16% 6.3 7.5 3.8 12.9 66 42 2 0 91 89 17 47 1.3 1.5 1.0 6.4 8.4 3.7 14.2 0 0 0 66 42 2 0 6 6 4 91 89 16 47 34.5 27.8 2.3 65% 27% 1.8 6.9 66% 26% 52% 1.0 1.6 0.4 76.2 69.5 7.0 32% 15.0 71% 56% 56% 31% 15% 3% 16% 18% 17% 8.5 8.3 1.6 18 11 1 61 50 8 SON 3.6 5.8 0 0 13 25 1.4 1.7 0.9 6.3 8.4 3.7 15.1 0 0 0 66 42 2 0 5 5 4 91 89 16 47 VILLAGES 26 27