Informal Insurance Arrangements in Ghanaian Migrants’ Transnational Networks: The Role of Reverse Remittances and Geographic Proximity

Informal Insurance Arrangements in Ghanaian Migrants’ Transnational Networks: The Role of Reverse Remittances and Geographic Proximity

World Development Vol. 37, No. 6, pp. 1105–1115, 2009 Ó 2008 Elsevier Ltd. All rights reserved 0305-750X/$ - see front matter www.elsevier.com/locate/...

184KB Sizes 1 Downloads 34 Views

World Development Vol. 37, No. 6, pp. 1105–1115, 2009 Ó 2008 Elsevier Ltd. All rights reserved 0305-750X/$ - see front matter www.elsevier.com/locate/worlddev

doi:10.1016/j.worlddev.2008.11.001

Informal Insurance Arrangements in Ghanaian Migrants’ Transnational Networks: The Role of Reverse Remittances and Geographic ProximityI VALENTINA MAZZUCATO * University of Amsterdam, The Netherlands University of Maastricht, The Netherlands Summary. — Risk pooling literature argues the need for geographic proximity to ensure the functioning of informal insurance arrangements. This paper investigates whether these arrangements exist between migrants and their network members back home and, if so, how they work in the absence of geographic proximity. Analysis of a simultaneous matched sample of migrants in the Netherlands and network members in Ghana reveals the existence of reverse remittances. These remittances show that there is risk pooling between migrants and network members. The paper elaborates on the institutional arrangements that make such a system possible. Ó 2008 Elsevier Ltd. All rights reserved. Key words — migration, remittances, insurance, transnational networks, Africa, Ghana

attesting to some degree of insurance taking place (Doss, 2001; Grimard, 1997; Rosenzweig & Stark, 1989). While these empirical studies focus on outcomes, the theoretical literature concentrates on how such informal insurance arrangements work. The key to this is effective monitoring and enforcement mechanisms (Arnott & Stiglitz, 1991; Stiglitz, 1990). These mechanisms ensure that people do not engage in riskier behavior just because they are part of such an arrangement (moral hazard), and that they will not opt out of a contract once they have received help (free riding). Platteau (1997) explains that monitoring and enforcement in informal arrangements can best take place in small, homogeneous groups, which share customs, language, and religion, thereby ensuring that there is a common understanding by all participants of the rules of the game. Indeed, empirical studies confirm that geographic proximity is fundamental for effective informal insurance arrangements. Most works focus on the household or village level where groups and communities tend to be homogeneous and can easily monitor and enforce their rights and duties (Arnott & Stiglitz, 1991; Coate & Ravaillon, 1993; Platteau, 1991,

1. INTRODUCTION This paper reports an investigation of how informal insurance arrangements work between overseas migrants and their network of friends, family, and business partners back home. Informal insurance arrangements have predominantly been conceived at the village level (Ligon, Thomas, & Worrall, 2002; Townsend, 1994) where geographic proximity allows insurance partners to monitor and enforce informal contracts effectively. But the question arises, how do informal insurance arrangements work over large distances? A few empirical studies have looked at informal insurance arrangements beyond the village boundary (Grimard, 1997; Rosenzweig & Stark, 1989). This explorative study builds on this literature, but, rather than focus on consumption outcomes as existing empirical studies have done, it analyzes the institutional arrangements that make such a system possible where there is no geographic proximity to ensure effective monitoring and enforcement. In so doing, the paper brings together literature on risk pooling in developing countries with literature on the new economics of labor migration (NELM). In many developing countries, formal insurance markets are rare owing to the difficulty of writing and enforcing market contracts, low educational levels, and high communication costs due to poor infrastructure. These factors together with a risky environment lead to high premiums that most of the local population cannot pay (Besley, 1995). Insurance in such societies is provided through mutual help relationships, otherwise referred to as informal insurance arrangements, where people pool their risks together so that when one or several members incur a shock, other members will help them to recover from the shock. Typical shocks that are investigated are loss of labor due to illness and death and crop failure due to adverse climatic conditions. These arrangements are informal because the costs paid for being insured in times of need (called premiums) and the help ultimately received are not strictly defined. Empirical studies of such informal arrangements analyze the consumption outcomes of these arrangements. In general, they find that those who are ensured have smoother consumption streams than those who are not

q This paper reports the results of a collaborative research program between the University of Amsterdam (AMIDSt), Vrije Universiteit Amsterdam (AOE), Amsterdam Institute for International Development (AIID), and African Studies Centre Leiden, the Netherlands, and the Institute of Statistical Social and Economic Research (ISSER), in Ghana. The research program is entitled ‘‘Transnational networks and the creation of local economies: Economic principles and institutions of Ghanaian migrants at home and abroad” www2.fmg.uva.nl/ghanatransnet (Netherlands Organization for Scientific Research (NWO) Grant Number 410-13-010P). * The author would like to thank Jan Willem Gunning, Marleen Dekker participants at the FADO seminars at the Vrije Universiteit Amsterdam, and participants at the CSAE conference ‘‘Reducing poverty and inequality: How can Africa be included” held in Oxford in 2006 for helpful comments received on earlier version of this paper. Final revision accepted: November 14, 2008.

1105

1106

WORLD DEVELOPMENT

1997). Furthermore, studies that look at what happens when people move away from their villages find that they do not use informal insurance arrangements anymore. For example, Gluckman (1960) and more recently Ferguson (1999) have found that laborers who leave their villages to mine in the Copperbelt of Zambia disrupt their support networks and therefore cannot as easily rely on them in times of crisis. Dekker (2004) shows that, in resettlement villages in Zimbabwe where households from across the country were brought together to form new villages (that is to say, where there is no history of geographic proximity), villagers tend to use informal insurance arrangements less than they do in pre-existing villages where the population has been living together for a long time. All of the above-mentioned studies take the household or the village as their unit of analysis. Recent studies have used other units, such as networks of people who depend on each other for help (De Weerdt & Dercon, 2006). However, here too, data are collected from one village and thus predominantly relate to intra-village networks. These studies thus leave open the question of what happens to risk pooling when members of a household or village migrate. Research that looks at risk pooling beyond village boundaries, such as at informal insurance arrangements between villages in the Philippines (Fafchamps & Gubert, 2007), also concludes that geographic proximity is the major determinant of risk-sharing networks. Furthermore, recent work attests to the fact that moral hazard can be a problem in insurance arrangements when members are geographically far from each other. Azam and Gubert (2006) found in the Kayes area of Mali that those households in a reliable insurance arrangement with a migrant were agriculturally less productive than households without a migrant. Chami, Fullenkamp, and Jahjah (2003) argue that moral hazard explains why they find that remittances are negatively correlated with economic growth. Thus empirical studies and theory suggest that geographic proximity is fundamental for effective informal insurance arrangements. A few studies that have looked at insurance on a larger scale, nationally (Rosenzweig & Stark, 1989) or ethnically (Grimard, 1997), indicate that informal insurance arrangements seem to work even at these higher levels. These studies conclude that insurance exists, at least to some degree, between members of groups with ethnic or marital ties who live geographically dispersed, because their consumption is partially influenced by the income of the group as a whole. This conclusion suggests that people who are socially proximate (i.e., linked by ethnic or marital ties) are able to monitor and enforce informal contracts. 1 Nevertheless, because they concentrate on consumption outcomes, these studies leave unanswered how such monitoring and enforcement work over great distances. This paper puts forward an explanation in a similar vein to Greif’s (1993) analysis of another form of long-distance contract: how informal insurance works within social networks of geographically dispersed people. Another branch of literature, that is relevant for this study, is the economic studies of migration. These studies often describe migration as an insurance strategy for households. Even before the introduction of NELM, economists studying migration argued that family welfare was a main motivator to remit (Johnson & Whitelaw, 1974), in contrast with neo-classical economic models that explain migration as a reflection of wage differentials between countries. NELM theory developed this idea further to argue that the decision to migrate is taken at the household-level as part of a household-level strategy for

dealing with risk (Lucas & Stark, 1985; Stark & Bloom, 1985; Stark & Levhari, 1982; Taylor, 1999). The family helps the migrant move to a place where income opportunities are expected to be better. The migrant then sends remittances either as delayed payment for the initial investment the family has made in the migration, or as insurance to the family in times of shock (Gubert, 2002; Stark & Lucas, 1988). In this latter case, remittances help households deal with imperfect or non-existing insurance markets by way of providing households with an external source of income not prone to the same risks as locally earned income. Thus by conceptualizing the decision to migrate as a household-level decision, NELM theory explains remittances as the outcome of a self-enforcing contractual arrangement between migrants and their families from which both parties expect to gain. Indeed, Lucas and Stark (1985) find evidence in Botswana that migrants remit more to their households after they have experienced a shock, attesting to remittances fulfilling an insurance function for the rural household. There is a growing body of literature investigating the effects of remittances on income and consumption streams of the receiving household (Adams, 2004; Lucas, 2005), but more needs to be known about how informal insurance works between the migrants and their families, as Doss (2001) also concludes. NELM literature discusses the question, what motivates migrants to abide by the contract as once they have migrated they could easily cut ties with the family and use their earnings for themselves. Three motives are discussed: altruistic feelings toward the family, the desire to be eligible for family inheritance, usually land or cattle, and co-insurance, that is to say, that the migrant is also insured by the contract (De la Brie`re, Sadoulet, de Janvry, & Lambert, 2002; Hoddinott, 1992; Lucas & Stark, 1985). While the first two motives are investigated empirically, the last motive has not been part of empirical analysis largely due to the absence of data on migrants. Very few studies include migrants in their samples. Hoddinott (1994) uses one of the few economic datasets where some migrants are also interviewed. Yet there is a growing body of literature on transnational migration that highlights that migrants are linked to their families back home not only because it is easier to do so due to modern information, communication, and transportation technologies, but also because in many cases their insecure position in the host country society makes it important for migrants to maintain linkages with their support networks in case of need or unforeseen crises (Basch, Glick Schiller, & Szanton Blanc, 1994; Levitt, 2001; Mazzucato, 2007; Vertovec, 2001). A major contribution of transnational studies has been to recognize the individual migrant as a member of a larger whole that extends beyond geographical boundaries. As such, these studies focus on units such as transnational networks, communities, circuits, or villagers (Guarnizo, Portes, & Haller, 2003; Levitt, 2001; Rouse, 1992; Smith, 2001). However, most studies are small-scale and do not collect data systematically (Mazzucato, 2008; Portes, 2001). They thus remain separate of risk pooling and of NELM literature discussed above. 2 This study uses transnational networks as the unit of analysis by collecting data from a matched sample of Ghanaian migrants living in the Netherlands and at least a subset of their social network members based in Ghana to contribute to an understanding of how informal insurance works. While NELM empirical studies such as those mentioned above deduce migrant motivations from the data collected from the receivers of remittances, we investigate these motives by collecting data from migrants. Are migrants really concerned about inheritance? The only evidence in the literature is that

INFORMAL INSURANCE ARRANGEMENTS IN GHANAIAN MIGRANTS’ TRANSNATIONAL NETWORKS

more prosperous families tend to receive more remittances than poorer families, all else being equal (Lucas & Stark, 1985; Schrieder & Knerr, 2000). In Ghana, where there is a land market and fertile land is still available in the central and southern parts of the country, results from in-depth interviews of migrants living in the Netherlands show that inheritance does not seem to be a top priority for migrants. The altruism motive stipulated by NELM theory also leaves some questions unanswered. Why would migrants accept conditions of adverse selection in which they enter a contract with people who, once migrants receive their new income stream in their new countries of residence, are all worse off than they are and will constantly be calling on them for help? Furthermore, altruism is bounded when confronted too often with moral hazard and thus requires some monitoring and enforcement of the family back home. A frequent complaint of migrants interviewed during this research was that their families were too ready to make financial calls on them: ‘‘They think that money grows on trees here in Europe!” How does a migrant monitor and enforce, without geographic proximity, the contract with her family? Finally, both risk pooling and NELM literature most often take the household as the relevant unit of analysis. NELM models conceive migration as a contract that emerges from negotiations between migrants and the households from which they come, represented by the heads of household. This conception assumes that the migrant knows what conditions will be met in the new destination and is capable of negotiating terms for a contract with the family. It is also assumed that the family consists of a household, located in one place, whose interests and objectives for remittances coincide. This paper investigates these assumptions. Are those who receive insurance remittances primarily household members? If not, who are the recipients? The paper is organized as follows: Section 2 presents the setting and explains the multi-sited research design and methodologies used to collect data at the transnational network level. In Section 3, we define insurance events as they are incurred by network members in Ghana, their characteristics, and then look at whether those receiving insurance are members of the migrant’s household in Ghana. An analysis of the way in which insurance works at the transnational level is reported in Section 4. In Section 5 we describe how monitoring and enforcement are conducted in the absence of geographic proximity. The paper ends with a conclusion section. 2. BACKGROUND AND METHODOLOGY Ghana is a good case in which to study transnational phenomena. The roughly 1.5 million Ghanaians living abroad (Twum-Baah, 2005) sent an estimated US$1 billion in remittances in 2003 (Addison, 2005). Total remittances are probably much greater since, as Mazzucato, van der Boom, and Nsowah-Nuamah (2008) estimate, 65% of remittances are sent through unofficial channels and escape formal statistics. Consequently, Ghana is probably comparable to large emigration and remittance-receiving countries such as Mexico and the Philippines. The main Western destinations for Ghanaian migrants are the United States and Canada, United Kingdom, Germany, Italy, and the Netherlands. The majority of migrants to the West have historically been Akan, because they benefited from the cocoa boom of the 1970s; however, recent data show that other ethnicities from central and southern Ghana are also migrating to the West (Nimako, 2000; Twum-Baah, Nabila,

1107

& Aryee, 1995). The Akan are a matrilineal ethnic group. This study focuses on Akan migrants to the Netherlands. There were 18,000 officially registered, first generation Ghanaian migrants in the Netherlands in 2003, 3 the second largest group from sub-Saharan Africa (CBS, 2008); however, unofficial estimates in 2000 were of the order of 40,000 attesting to a large undocumented population (Mazzucato, 2005). In general, migrants from Ghana are both highly skilled and unskilled; the Netherlands tends to attract a majority of unskilled migrants, since the highly skilled prefer Anglophone countries (Carrington & Detragiache, 1998). Data were collected using a multi-sited research design. First, social networks were elicited using a network survey based on 17 name-generator questions which was conducted among 106 Ghanaians living in Amsterdam. No baseline survey of Ghanaian migrants in the Netherlands exists. The survey was therefore conducted by randomly selecting migrants encountered through 11 different gateways (two churches, one cultural project, two prominent people in the Ghanaian community in Amsterdam, one workplace, three hometown associations, chance encounters, and initial contact with migrants’ families in Ghana). Snowball sampling, often used in migration studies where there is no baseline survey, was not used so as to avoid the risk of only obtaining access to certain types of migrants. Given the diversity of gateways, the sample was constructed as randomly as possible in the absence of a baseline survey. The name-generator questionnaire is a tool used in quantitative social network analysis (Burt, 1984; Campbell & Lee, 1991). In this study, an exchange approach to networks was used in which questions were asked with respect to the exchange of emotional and material supportive content between ego and alters (McAllister & Fischer, 1978), 4 both positive relations (such as friendships) and negative relations (such as people one argues with) and strong and weak ties (Granovetter, 1973) were asked. Respondents were asked to mention as many names as they could think of for each question and along with the names, also the location of the person and the relationship with the respondent (kin/non-kin and for each, specifying what kind of relationship, such as ‘‘business partner” or ‘‘mother’s sister”). The tool was tested for cultural relevance of questions and saturation. 5 Second, a sub-sample of 29 6 ego-centered networks was selected, based on (a) the individual characteristics of the migrant (sex, age, income, education, and length of migration period) and (b) network characteristics (size and density). Table 1 shows the descriptive statistics of the 106 networks and compares these with those of the 29 networks selected for in-depth study. It shows that the characteristics of the 29 networks do not differ significantly from the larger sample. The reason for opting for a small number of cases (networks), albeit larger than in anthropological studies, is because of the study’s explorative nature. Since our interest was in understanding how insurance arrangements work and in identifying the mechanisms involved in monitoring and enforcement over large geographical distances, we considered that fewer cases would allow more visits per case to collect quantitative and observational data. This study can serve as a pilot for future quantitative studies seeking to measure the relative importance and efficacy of the mechanisms identified. Four research projects were carried out to study the network members based in the main nodes of migrants’ networks: migrants in Amsterdam and network members in Accra, Kumasi, and rural towns in the Ashanti Region. The result was a matched dataset of migrants and a subset of their network members, 7 totaling 131 respondents. Each of the respondents

1108

WORLD DEVELOPMENT Table 1. Network characteristics (2003) b

Size

Kin

Density

Network members in Ghana

Kin in Ghana – based network

N = 106

N = 29

N = 106

N = 29

N = 106 (%)

N = 29 (%)

N = 106 (%)

N = 29 (%)

N = 106 (%)

N = 29 (%)

Malea Female

21 19

24 21

0.52 0.55

0.53 0.56

51 54

50 49

33 43

36 43

73 71

69 66

Total

21

23

0.53

0.55

52

49

38

40

72

68

Source: Network study 2003. a The means for each column characteristic (size, density, etc.) were compared between the two samples using a t-test. None of the means in the same row, per characteristic, differ significantly between the two groups (N = 106 and N = 29) at p < 0.05. b Density is a measure of network structure used in social network analysis indicating how many alters know each other. A density of 1 means all network members know each other; a density of 0 means no one knows each other. The formula for density is k/(n(n  1)/2), where k is the number of lines present (line = 1 if a dyad knows each other or 0 if they do not know each other) and n is the maximum possible number of lines connecting alters within a network.

was followed over a one-year period (July 2003–June 2004); the same research tools were used in the four locations to collect simultaneous and comparable data for all members of the network. This approach allowed us to collect data from both senders and receivers of remittances. In this sense the methodology approximates Massey’s (1987) ethnosurvey methodology for studying United States–Mexico migration, although here data are collected simultaneously and equal emphasis is given to both sides of the matched sample. 8 Various methods were employed to collect in-depth data from the 29 networks. A network survey was conducted with the selected members of each migrant’s network based on the same name-generator tool used for migrants. A transaction study was conducted for 115 of the 131 respondents on a monthly basis over the one-year period, resulting in 16,735 recorded transactions. Collecting information from both sides (migrant and network members back home) shows the intentions with which remittances are sent and how they are ultimately spent. The transaction study concentrated on eight domains of life important for Ghanaians: housing; business (including farming); church; funerals; education; healthcare; community development; remittances for general sustenance. In-depth interviews were conducted on the eight domains and life histories were collected on the role of network members in a person’s life. Finally, informal conversations and observation were used at various social occasions attended during June 2002–February 2005 in Amsterdam and during May 2003–August 2004 in Ghana. 3. INSURANCE EVENTS IN GHANA AND THEIR BENEFICIARIES An insurance event must involve a degree of unpredictability; otherwise people could plan for it and would not need to rely on an insurance arrangement if it were known whether and when the event would occur. Such an event must also involve a substantial financial cost, otherwise people would be able to pay for the event on their own (Dercon, 2000). Looking at the list of things remittances were sent for, as reported by the 29 migrants in a one-year period (Table 2), healthcare and funerals are considered shocks, because they are unpredictable and costly, making them domains for which insurance is needed. These same events were found to constitute important shocks for households in Africa (De Weerdt & Dercon, 2006). Although these domains do not comprise a large share of the total remittances sent (approximately 6% of all remittances), they are nevertheless of vital importance to those receiving them. In the case of illness, death may be avoided;

Table 2. Purposes of remittances sent from the Netherlands to Ghana (2003–04) Purpose

Average yearly amount sent (%)

Business Housing Help Subsistence Education Family funerals Ceremonies, other Health Community projects Costs of papers Other

33 16 41 20 11 6 2 1 1 4 6

Total

100

Source: Transaction study 2003–04.

in the case of funerals, disrepute will not be incurred, as we explain below. Illnesses in a developing country, where diseases are many and healthcare is scarce, are contracted by people of all ages and can quickly and unpredictably lead to death. In such cases, sudden and large requests will be made of migrants to help meet the costs of hospital fees and medication. Funerals are also unexpected and costly events. The death of a relative cannot be predicted. Furthermore, funerals are of great cultural importance not only for the Akan, but also for most ethnicities in Ghana (Aborampah, 1999; Arhin, 1994; De Witte, 2004; Lentz, 1994). A large funeral is one of the most powerful ways for matrilineal members of an extended family to gain prestige and respect in their community. The cost of a funeral can easily run into thousands of euros. If there is a migrant in the family, the funeral will most often be financed by him or her (Mazzucato, Kabki, & Smith, 2006), leading to the popular perception in Ghana that migrant-financed funerals are the most grand (Kabki, Mazzucato, & Appiah, 2004). During the one-year’s collection of quantitative data, we encountered 31 cases of illness or funerals concerning a close family member of a migrant, averaging approximately 1.5 cases per migrant per year (Table 1). Remittances for these purposes were made in a bulk sum at the time of the request and not paid from a regular stream of remittances that the recipient may have received in the past. Finally, according to migrants, no direct return payment is expected. Some degree of insurance can therefore be said to be taking place, since

INFORMAL INSURANCE ARRANGEMENTS IN GHANAIAN MIGRANTS’ TRANSNATIONAL NETWORKS

1109

Table 3. Remittances for healthcare and funerals (2003–04)

Illness Funeral

N = no. of events

Average no. of events/migrant

Total no. of receivers

20 11

1.5 1.4

20 32

Average yearly amount sent by migrant per event

Average share of total yearly remittances

Share of monthly income

(€)

SD

(%)

SD

(%)

SD

104 1,128

108 2,371

3 17

3 16

15 148

15 295

Source: Transaction study 2003–04.

migrants send remittances in times of crisis. Table 3 shows the range of costs that healthcare and funerals entail. NELM conceptualizes migration and insurance as a contract between migrant and household. Here, we investigated the insurance events summarized in Table 3, to see who the receivers of insurance were. Are they indeed members of the migrant’s household? Kin relationships seem to be the guiding factor determining whose claims the migrant will honor (Table 4). The majority of remittance receivers were kin-based relations (80% for health and 75% for funerals), especially from the matriliny. This can be explained in the case of funerals, since matrilineal family members of the deceased are customarily responsible for organizing and paying for the funeral. NELM theory stipulates that these kin relations that are insured by the migrant are part of his or her household. We wanted to see whether this was indeed the case; however, to do this, a problem needed to be solved: it is impossible to establish what the migrants’ household is in Ghana. This is because a migrant often has moved various times within Ghana before arriving to the Netherlands. A stylized case resulting from life histories we collected amongst the 29 migrants shows that a town or village dweller will move to Kumasi and then to Accra, or directly to Accra, before leaving for Europe. The sojourn in these cities can last between a couple of weeks to many years. During this period, the migrant lives with parts of his or her extended family or non-kin friends. Which of these units should be considered a migrant’s household is not specified in the NELM theory. Thus in those cases where a migrant provides insurance to one person, we cannot determine whether the person being insured represents the migrant’s household in Ghana. However, a way to solve this issue was to look at cases where a migrant provided insurance for two or more people in one year. NELM theory would predict that both recipients be part of the same household: that of the migrant. Table 5 shows that in all such cases, people were not from the same household. Insurance receivers came predominantly from the same region, but not the same village, town, or city. The case of Grace illustrates why migrants might insure different network members spanning different households. Table 4. Share of receivers of insurance by type of relationship to migrant (2003–04) Health (%) Maternal family (incl. siblings) Paternal family Child In-law Non-kin Self Total Source: Transaction study 2003–04.

Funeral (%)

76 3 0 0 21 0

48 13 6 3 26 3

100

100

Table 5. Location of members helped in an insurance event (2003–04) % When more than one event occurred where a blood or marital relation was insured: Share of events in one household Share of events in two households Share of events in more than two households Total Are households located in the same village/town/city? Share of households in one village/town/city Share of households in more than one village/town/city Total Are households located in the same region? Share of households in one region Share of households in more than one region Total

0 50 50 100

25 75 100

75 25 100

Source: Transaction study 2003–04.

Grace was undocumented at the time of research and working only 15 h/week. She at times had to reduce her calorie intake to make ends meet at the end of the month. When her mother died, her brother paid for her ticket to Ghana and a close friend in the United States helped her finance the funeral costs. Although both of these sums were given as gifts and no financial repayment was expected, Grace felt she could not ask them again for help when the wife died in the family that was taking care of Grace’s only daughter. The husband of the family was Grace’s distant kin. Her non-participation in the funeral might have jeopardized the care her daughter was receiving. Grace thus mobilized other friends in the Netherlands to be able to make a financial contribution to that family and thus ended up helping two households in one year.

While these results do not deny that migrants may be insuring their households back in Ghana, Tables 4 and 5 and qualitative evidence collected from the 29 cases of which Grace was one illustration indicate that kin relations at a higher scale than the household are also at play when receiving insurance from migrants. This concurs with the results obtained by Grimard (1997). 4. MUTUAL INSURANCE OVER LARGE DISTANCES There are various reasons why a migrant provides insurance to kin relations and these relations go beyond the household. One reason is the emotional ties to kin. No one wants to see a close relative die through lack of healthcare when one is able to pay for it. This is the altruism motive referred to in the NELM literature (Agarwal & Horowitz, 2002; Stark & Lucas, 1988). Another reason is that goods and land are inherited via the matriliny. Thus, one’s inheritance is determined by how

1110

WORLD DEVELOPMENT

one acts vis-a`-vis matrilineal members of the family. Indeed, the primary beneficiaries of remittances for healthcare and funerals are matrilineal members of the extended family (Table 4). The PNDC Inheritance Law of 1997 makes a person’s nuclear family the direct inheritors of one’s goods and land; in practice, however, inheritance continues to pass through the matriliny. A final reason why migrants tend to help kin relations is that their objective is to return to Ghana and ultimately be buried there with a grand funeral attesting to their lifelong achievements. Since matrilineal extended family members are responsible for organizing one’s funeral, the expectation of its appropriate arrangement is also an incentive to help matrilineal family members when they are in need. However, if the above-mentioned reasons were the only reasons for a migrant to provide in times of need, they would not amount to a form of informal insurance, because the migrant would just be doing something to obtain another objective, say inheritance, and this something happens to have an insurance effect. A mutual insurance arrangement requires a pooling of risks; that is to say, all participants have risks that they insure through migration and, when an event occurs, participants contribute to the alleviation of the income shock. Our findings suggest that such pooling of risks within a transnational network does occur. This phenomenon is often overlooked in migration literature that tends to concentrate on a one-way flow of goods and money from the migrant to the home community. However, the migrant also receives insurance from members of the home country network in what we call reverse remittances. A migrant’s life in a developed country is full of unpredictable risks that call for insurance (Mazzucato, 2005). A migrant may encounter difficulties while trying to obtain a residence permit, be deceived in a marriage transaction, lose a job, or be caught without appropriate documentation and be sent to prison. The migrant cannot foresee any of these situations, all of which require help in terms of finance and services. Table 6 gives for each migrant-incurred insurance event how many respondents encountered each one, an indication of how unpredictable and urgent these events are, and who is called upon for help. These events are described in the next section. Interviews conducted in Ghana with members of migrants’ social networks revealed that people usually have a realistic picture of what migrants’ lives in the Netherlands are like.

Potential migrants should therefore be able to foresee what may happen to them when they migrate. Nevertheless, different aspects make these events unpredictable. First, one has no certainty that they will occur. For example, some people are lucky and are never caught without a residence permit, or they encounter honest marriage partners. The chance that things can go well makes each potential migrant, even though aware of the risks, believe all will indeed be well. Second, migration policy in the Netherlands and in Europe has changed quite dramatically and frequently over the past 20 years, becoming increasingly stringent toward migrants. These changes have increased the likelihood of getting outdated information about obtaining documents in the Netherlands that the migrant only finds is no longer valid on arrival there. These events for which a migrant needs to insure occur at a specific phase in the migration trajectory. This phase is becoming of increasingly long duration. Life histories have shown that the migration trajectories of Ghanaian migrants in the Netherlands have three phases (Table 7): first, the preparation phase in Ghana to acquire the money and necessary documents for the trip; second, the installation phase in the Netherlands during which migrants regularize their stay by obtaining the necessary documents; and third, the settled phase in which they have succeeded in obtaining their documents and are free to build a new life in the Netherlands. The phase of the migration trajectory when a migrant is most likely to encounter an insurance event is the second. Migrants are remaining in phase II for longer periods (Mazzucato, 2007), thereby lengthening the time during which they need assistance from their network members based in Ghana. This situation reflects the increasingly stringent visa policies in the Netherlands (as in the rest of Europe), especially since the mid 1990s. ðaÞ Migrant experienced insurance events All the insurance events migrants experience entail substantial financial costs; however, the role of people in Ghana is primarily that of providing a service needed for resolving the migrant’s insurance event. Below follows a description of the events and the kinds of services that are needed. Respondents without papers first try the cheapest route of obtaining papers: via the official channels. This approach entails a lot of work in

Table 6. Insurance events for migrants in the Netherlands (2003–04) Crisis

Un-predictabilitya Urgencyb

No. of cases

Service

Getting a staying permit legally Getting a staying permit on the black market A marriage deal gone wrong

18 8

+ ++

+ ++

4

Paperwork in Ghana Finding a candidate candidate paperwork Arbitration

++

+++

Losing a job

5

Getting access to a new job

+++

+++

Being put in prison

2

Getting you out of prison

+++

+++

Childcare in Ghana gone wrong

2

Finding a new child care arrangement

+++

+++

From whom Family and friends in Ghana Family and friends in Ghana and the Netherlands Family and friends in Ghana and the Netherlands Family and friends in the Netherlands Family and friends in Ghana and the Netherlands Family and friends in Ghana

Source: Life histories (2003–04), N = 29. a Unpredictability indicates how difficult it is to obtain accurate information regarding the probability that such an event will occur, from (+) least difficult to (+++) most difficult to obtain information. b Urgency is used as a proxy for financial cost, because the help that is needed for these events is primarily in the form of services like obtaining correct documentation, arbitration, and obtaining accurate information. These services are not quantifiable in terms of financial costs, but are needed with some degree of urgency. They have therefore been ranked from most urgent (+++) indicating that the consequences of not solving this problem promptly will have financial repercussions greater than €10,000 to (++) indicating financial repercussions between €1,000 and €9,999 to (+) with financial repercussions of less than €999. In the case of childcare that has not proceeded as planned, the financial repercussions are difficult to estimate (how to value a missing child, or a child who is not being given proper care?). We have, however, valued this state of affairs as most urgent (+++) given the emotional costs for a parent discovering that childcare has gone amiss.

INFORMAL INSURANCE ARRANGEMENTS IN GHANAIAN MIGRANTS’ TRANSNATIONAL NETWORKS Table 7. Migration phases and the occurrence of insurance events (2003–04) Flow of help remittances Ghana to the Netherlandsa The Netherlands to Ghanab

Migration phases

Total

Phase I

Phase II

Phase III

2 0

35 10

2 21

39 31

Source: Life histories and transaction study. Refers to the events reported in Table 4. b Refers to the events reported in Table 1. a

Ghana in collecting the necessary documents and informing all those who might be interviewed by the embassy controllers what answers to give. This system is necessary, because at the time of research the Dutch embassy employed a strict documentation verification procedure that is virtually impossible for Ghanaians older than 30 to satisfy, because the kinds of documents and the consistent documentation system needed to satisfy embassy requirements did not exist 30 years ago. 9 Migrants need to have documents made, school records changed, hospital birth records forged, and the extended family informed about the ‘‘official” answers to be given to the controllers who come around to one’s village to verify all the information put on the application form. A friend or family member in Ghana is entrusted with collecting forms at the embassy, making sure all records give consistent information, informing extended family members of the ‘‘correct” answers to give embassy controllers, all of which entails many hours of waiting in line, traveling, and coordinating. If these services are not done or are done incorrectly, no resident permit will be obtained, which in turn has heavy financial consequences for the migrant. These services are needed within a short time span and are unpredictable for the migrant and so they qualify as an insurance event. Obtaining documents on the black market means in most cases that the migrant needs to find a marriage partner with legal documents. After three years, the migrant can divorce and has the right to a resident permit. Again here, migrants make use of friends or family in the Netherlands or in Ghana to put them in contact with a potential candidate. Migrants are in an extremely vulnerable position vis-a`-vis marriage partners, because the marriage costs need to be paid upfront. Costs of these transactions in 2004 were between €10,000 and €15,000. Thus the friend or family member who can find and guarantee the trustworthiness of the marriage candidate is of vital importance. Should something go wrong in the marriage arrangement, the family and friends who know the candidate can exert pressure for compliance with the transaction through the candidate’s social network. Again, not finding a good candidate leads to heavy monetary loss and having to find a candidate is an unpredictable event. The more quickly a candidate can be found, the fewer are the risks a migrant has of being caught and put in prison or deported. We therefore consider finding a good candidate to be an insurance event. Losing a job is also a possible event incurred in this phase since immigration police controls are commonly held at the place of work. Migrants who escape these controls by running away when police arrive are forced to leave their jobs for fear of being caught. We also came across cases of employers who knew migrants were working with false documents and fired them after one month, without pay. This is a way for malafide employers to obtain free labor. In such cases, migrants need to find a new job within a short time to be able to pay for their living costs, otherwise they will be forced to leave their apartments and reduce the amount they eat. This too is an insurance event, because it is unpredictable and needs a speedy solution.

1111

Migrants who are caught and put in prison rely on friends and families in the Netherlands to provide various services for them while there, such as buying phone cards to use in prison, contacting employers (undocumented migrants often have jobs on the black market or work with someone else’s documents) to give an explanation for absence in order not to lose the job, hiring a lawyer, and taking care of any bills that need to be paid. A two-month stay in prison resulted in costs of €4,000 for one of our respondents, including costs of foregone employment, a migration lawyer, and international telephone calls. The family in Ghana may also be called upon. In the case of our respondent, the family in Ghana provided help by going for two weeks to a special prayer camp known for the pastors’ powers to resolve ‘‘paper problems” (Van Dijk, 1997). This action involved financial costs for the family back home for travel and donations made at the prayer camp and the loss of their labor during the two-week stay at the prayer camp. Again, being put in prison is an unexpected event with high financial costs, and being able to mobilize one’s network to be released from prison leads to a smaller income shock. We consider this, too, to be an insurance event. Finally, it is common for Ghanaian migrants to leave their children in Ghana in the care of one of their network members. This plan does not always turn out as the migrant envisaged. In two cases of childcare-related crises we encountered, one network member lost track of a migrant’s child and another network member died, leaving the minor without care. In both cases, an immediate solution was needed to find the missing child in one case and to find a new carer in the other. In both cases, network members in Ghana were mobilized to solve the crisis. The unpredictability as well as the urgency with which a solution needs to be found are high and therefore this type of event qualifies as an insurance event.

5. MONITORING AND ENFORCEMENT WITHIN TRANSNATIONAL NETWORKS As explained in the introduction, monitoring and enforcement are key principles for making informal insurance contracts work. The literature on insurance explains that geographical proximity is fundamental for enabling the monitoring and enforcement of contracts. Geographical proximity allows people to see what others are doing, that someone making a claim is indeed experiencing a difficulty, and that when the claim is abided by, the help obtained is indeed used to alleviate the difficulty and not for some other purpose. Geographical proximity also means that participants in an informal insurance arrangement share similar norms and values so that the ‘‘rules of the game” are generally agreed and accepted. However, international migration provides an interesting paradox, because while some degree of risk pooling is obtained within a transnational network, there is no geographical proximity to ensure monitoring and enforcement. Table 8 shows the different possible moral hazards and free riding scenarios that could result due to the lack of geographic proximity. Below we describe the monitoring and enforcement mechanisms in place that reduce the likelihood of these scenarios taking place. The cases in quadrants II and III are not relevant for the case being analyzed. The case in quadrant II is not relevant for insurance events as the migrant receives services that are specific to the insurance event, such as help in negotiating a marriage deal gone wrong. Therefore asking for help when it is not needed is of no use to the migrant as it cannot be used for other purposes.

1112

WORLD DEVELOPMENT Table 8. Moral hazard and free riding scenarios in the absence of geographic proximity Network members in Ghana

Migrant in the Netherlands

Moral hazard

I. Family asks too much of migrant by appealing to him for all sorts of help without first trying to solve the problem themselves. For example, not using their own resources to pay for a medical bill and rather requesting the migrant to pay the bill.

II. Migrant asks for help to family in Ghana when in reality she does not need it or could resolve the problem herself. For example, the migrant pretends she has had a problem with a marriage deal gone wrong when in reality this did not happen.

Free riding

III. Family does not help when a migrant asks for it. For example, when a migrant is in prison and asks her family in Ghana to go to a prayer camp, they do not go.

IV. Migrant does not help a network member in Ghana in need. For example, when a request is made by a network member to help pay a medical bill, the migrant in reality has the financial means to help but says he does not.

Notes: Shaded quadrants are the most relevant for the case of risk pooling between network members and migrants overseas.

The case in quadrant III is also not relevant for insurance events and indeed no such case was encountered amongst the research population. This is because a migrant’s ability to ensure her living overseas is a prerequisite for being able to help network members back home. Thus aside from altruism, network members also have a vested interest to assist a migrant in need of help. Another reason is the repeated nature of the insurance events that network members in Ghana incur: funerals and health care. If a network member refuses to help a migrant, according to custom, this disqualifies him from receiving help the next time he is in need. As funds for funerals and health care are frequently needed, network members are unlikely to engage in behavior that will disqualify them from such help. The cases in quadrants I and IV instead are more relevant for risk pooling between network members in Ghana and migrants in the Netherlands. The case presented in quadrant I manifests itself when network members in Ghana engage in riskier behavior because they know a migrant will insure them. Azam and Gubert (2006), for example, find that network members tend to reduce their agricultural productivity when they have a reliable migrant who sends remittances in times of need. However, in the insurance events that we analyze here, funerals and health care, moral hazard is unlikely, as more risky behavior would result in death of the network member (Dercon, De Weerdt, Bold, & Pankhurst, 2006). But another version of moral hazard, such as the example given in the table, is possible so the migrant needs to make sure network members really cannot solve the problems themselves. Migrants were found to keep track of how a network member spent the remittance as well as that no large purchases are made, indicating that the network member could have otherwise spent the money on resolving the insurance event. The mechanism migrants use for this is to have a trusted network member in Ghana who can keep an eye on remittance disbursal and expenditure. In-depth interviews of 29 migrants revealed that friends were often entrusted with the management of migrants’ remittances to make sure that they reached the desired beneficiary and that the remittances were indeed used for the purposes for which they were intended (Mazzucato, 2003; Smith & Mazzucato, 2003). These services of monitoring and enforcement provided by the friend also apply to insurance events. This we can illustrate with an example from our case studies: Abena needed to send money to Ghana to pay for the visa and passport for her youngest daughter to come and join her in the Netherlands. Abena had a son in his 20s conveniently living in Accra,

where all the documentation services are located. She wanted to ask him to take care of obtaining the documents, although she also knew the great temptation the money she remitted would represent for this young man, living in a capital city where most consumer products are readily available. She resolved this problem by sending the money through a male friend of hers in Accra. She explained: ‘‘If I give the money to him [the friend], my son will know that he is being checked and will not use the money for something else.”

Why do people use friends for monitoring and enforcement? Interviews with migrants indicated that migrants have more sanctioning power over a friend who misbehaves than over a family member, since custom makes it difficult to sever relationships with kin (Mazzucato, 2003). Also, kin members are more easily influenced by other kin members and can be persuaded to misreport information back to the migrant. Looking at the characteristics of the friend showed that friends are not in financial need and are therefore less likely to be tempted to divert some of the remittances for their own use. For the most part, they had also migrated at some point in their lives, and so knew that money was not easy to come by overseas. Finally, migrants in the third phase of the migration trajectory travel to and from Ghana on regular and frequent visits. These provide them with the opportunity to monitor and enforce informal insurance contracts themselves by, for example, bringing money for a funeral to Ghana and disbursing it themselves. There are various enforcement mechanisms that help decrease the possibility of the case in quadrant IV from happening. First is that information does flow back to villages and towns where migrants’ network members are located. This happens largely through other migrants who come from the same villages and towns as the network members. In the case of Ghanaian migrants in the Netherlands, they predominantly live in the three major Dutch cities where they have an active associational life, both in terms of churches and associations to which they belong. Gatherings are frequent to celebrate funerals, baptisms, and marriages. These gatherings and association meetings (including church gatherings) provide members of the Ghanaian community the opportunity to obtain and observe information about each other. During community members’ trips back to Ghana, this information gets communicated. The case of Mercy is telling: When she first came to Ghana, Mercy started dating a Ghanaian man who was part of the ‘‘wrong crowd”, as Mercy later confessed to me. This information quickly spread throughout the Ghanaian community

INFORMAL INSURANCE ARRANGEMENTS IN GHANAIAN MIGRANTS’ TRANSNATIONAL NETWORKS in Amsterdam, and members from Mercy’s hometown related this to Mercy’s mother in Ghana. This led to a great rift between Mercy and her mother and they did not speak to each other for 10 years. Mercy did not send her mother any remittances in this period nor did her mother help her with ‘‘paper problems” that she encountered.

Mercy’s story shows that it is possible for network members to obtain information about a migrant overseas even when the migrant does not want that information to reach her community back home. Although this is only possible when a large enough number of migrants from the same hometown are living in close proximity of each other overseas. This is not a foolproof method of information gathering as some migrants in our sample expressly moved away from the neighborhood in Amsterdam where many Ghanaians live in order to escape from the gossip circuits. A second reason particular to funerals, as to why migrants would not shirk from helping out, is that a meager funeral for a family member reflects badly on the whole family, as well as on the migrant. We found that in smaller villages, elders instituted a funeral card system since Ghanaians started migrating overseas in large numbers in the mid-1980s (Mazzucato et al., 2006). These cards keep track of each villager’s (including migrants’) contribution to funerals of fellow-villagers. If one is seen to be shirking on one’s contributions, that person and his family will be prevented from burying their family members in village territory: a great source of shame in all Ghanaian communities. For all the above-mentioned enforcement mechanisms to work, there is a need for social proximity to ensure that people have similar norms, are driven by common objectives, and generally agree to and abide by the rules of the game (Platteau, 1997). Indeed, nearly all the members of the 29 networks are Ghanaians and, of these, virtually all are from the same ethnic group: the Akan. Thus while geographic proximity is not a necessary condition for informal insurance arrangements to work, social proximity is. 6. CONCLUSION We have found informal insurance arrangements between migrants and their network members back home to exist. Migrants send remittances to network members back home, mostly to the matrilineal kin group, when insurance events occur such as health emergencies or funerals. While the analysis does not exclude altruism or inheritance as possible motivators, we found evidence of risk pooling between migrants and network members, adding to the motivations for migrants to help their network members in times of need. However, risk pooling literature argues that geographic proximity is necessary for such arrangements to work, while economic studies of migration mention risk pooling as a possible motivator to remit, but they focus predominantly on how migrants insure their households back home leaving a gap in knowledge as to how migrants arrange for risk pooling at their end. Some authors writing on risk pooling acknowledge the gap in knowledge that results from not having data on migrants (Fafchamps & Gubert, 2007; Udry, 1994). In this paper, we have used a matched dataset and have found that: (a) migrants also face risks and their strategies for dealing with these risks involve services from family and friends back home and (b) even in a transnational context there are mechanisms that make monitoring and enforcement effective over large distances. Below we expand on these two findings. Migrants not only send remittances; they also receive them, what we call reverse remittances. These are mainly in the form

1113

of services undertaken for migrants during the second (installation) phase of the migration process. Although once migrants reach the third (settlement) phase, insurance of risks is less relevant for them, preliminary indications are that the second phase is becoming longer for labor migrants such as Ghanaians in the Netherlands, who face increasingly exclusionary migration policies. While the services they receive are difficult to quantify, they are fundamental for a migrant to be able to resolve the crisis on the occasion of an insurance event. This finding is of relevance for both insurance literature and economic migration studies. For insurance literature, reverse remittances attest to the fact that there is a pooling of risks between a migrant and network members back home. That is, migrants not only provide insurance, but also receive it. In the migration literature, the existence of reverse remittances has largely passed undetected, because data are collected from household members in migrants’ home countries and data collection centers on what households receive from migrants. Having a matched sample allowed us to detect the hardships that migrants incurred while abroad and the ways in which people back home were involved in resolving these hardships. Future studies could quantify reverse remittances (for example, in terms of time spent on migrant-related services) by designing migration surveys that ask migrant network members specifically about what they send to migrants, not only what they receive. It is important to try to estimate the volume of reverse remittances when evaluating the net effects of migration in developing countries because reverse remittances are a cost (especially in terms of time taken away from productive and reproductive activities) that network members incur that directly stem from the migration process. This study found that mutual insurance exists even over large geographic distances separating migrants from the people in their networks. Whether geographic proximity is indeed a necessary characteristic of risk pooling networks is thus called into question. Studies of risk pooling behavior seldom look beyond village boundaries or groups of villages and therefore leave links with migrants overseas out of their analyses. Two specific aspects regarding geographic proximity were investigated. First, we find that the members back home insured by the migrants are not necessarily from the same household. These members were dispersed in various villages, but mainly within the same administrative region. This draws into question, at least for the case of Ghanaian migration, the assumption made by NELM studies that migration is a household-level strategy. NELM studies postulate that because the decision to migrate is taken at the household-level, remittance behavior can be explained as a form of insurance or payback mechanism made to the household for the initial investment made to allow the migrant to move. However, the results of our analysis suggest that risk pooling occurs within a wider network of extended family members than just the household raising the question, what does this imply about the decision to migrate? Are others involved and could this be the reason why remittances are sent to extended family members within different households? Alternatively, as migrants also receive insurance, their remittance behavior may not be related to who decided on and financed their migration, but rather to their desire to insure themselves against the risks that they incur while abroad. This finding implies that insurance effects may be better captured by collecting data at the network rather than the household-level, as is currently done. Second, we find that various institutional arrangements enable the monitoring and enforcement of informal insurance contracts over great distances. Besides the repeated nature of

1114

WORLD DEVELOPMENT

insurance events and social proximity already identified in the literature, this study has found that non-kin network members play an important part in transnational insurance arrangements by providing the necessary monitoring and enforcement needed to avoid problems associated with moral hazard. How well these mechanisms work to produce effective risk pooling

arrangements remains a matter for empirical investigation. This paper has suggested ways in which to collect data in order to be able to investigate empirically the effectiveness of such risk pooling arrangements by collecting data from matched samples of migrants and people back home and using this transnational network as the unit of analysis.

NOTES 1. Geographical literature has indicated that social proximity is important when defining space and that such proximity is possible to maintain over large distances (Massey, 2005), although not in specific reference to insurance arrangements between migrants and network members back home. 2. De la Cruz (1995), for example, conducted a case study of five Mexican families and their migrants in the United States. Klaver (1997) also studied Mexican families and their migrants in Los Angeles. An important exception is the quantitative work conducted by Osili (2004, 2007) on migration, savings and investments although it does not treat insurance. 3. In 2007 this increased to 19,500. 4. An example of a question for an emotional support tie is ‘‘If you experience problems with raising your children, whom would you go to, to seek advice?”. An example of a question getting at a material tie is ‘‘who does the household chores such as food shopping, cleaning, cooking, child minding?”.

5. Saturation refers to eliciting as complete a network with as few questions as possible. The complete questionnaire can be found on www2.fmg.uva.nl/ghanatransnet. 6. An original 33 networks were selected but in four cases the migrant did not want to take part in the research thus only his network members were interviewed. 7. For practical purposes, only those network members based in one of the three research locations in Ghana were interviewed. 8. See Mazzucato (2008) for a detailed discussion of a simultaneousmatched-sample methodology and the effects of collecting data simultaneously. 9. This procedure was judged unconstitutional by the Dutch Supreme Court and was suspended at the end of 2004. However, obtaining a residency permit has remained a difficult procedure for Ghanaian citizens making the analysis presented here still relevant today.

REFERENCES Aborampah, O. M. (1999). Women’s roles in the mourning rituals of the Akan of Ghana. Ethnology, 38(3), 257–271. Adams, R. (2004). Remittances and poverty in Guatemala (Policy Research Working Paper 3418). Washington, DC: The World Bank. Addison, E. (2005). The macroeconomic impact of remittances. In T. Manuh (Ed.), At home in the world: International migration and development in contemporary Ghana and West Africa (pp. 118–138). Accra: Sub-Saharan Publishers. Agarwal, R., & Horowitz, A. (2002). Are international remittances altruism or insurance? Evidence from Guyana using multiple-migrant households. World Development, 30(11), 2033–2044. Arhin, K. (1994). The economic implications of transformations in Akan funeral rites. Africa, 64(3), 307–322. Arnott, R., & Stiglitz, J. (1991). Moral hazard and non market institutions: Dysfunctional crowding out or peer monitoring. American Economic Review, 81(1), 179–190. Azam, J. P., & Gubert, F. (2006). Migrants’ remittances and the household in Africa: A review of evidence. Journal of African Economies, 15(Supplement 2), 426–462. Basch, L., Glick Schiller, N., & Szanton Blanc, C. (Eds.) (1994). Nations unbound. Transnational Projects, postcolonial predicaments and deterritorialized nation-states. London: Routledge. Besley, T. (1995). Nonmarket institutions for credit and risk sharing in low-income countries. Journal of Economic Perspectives, 9(3), 115–127. Burt, R. (1984). Network items in the general social survey. Social Networks, 6(4), 293–339. Campbell, K. E., & Lee, B. A. (1991). Name generators in surveys of personal networks. Social Networks, 13(4), 203–221. Carrington, W., & Detragiache, E. (1998). How big is the brain drain? (IMF Working Paper WP/98/102). Washington, DC: IMF. CBS (2008). StatLine database. . Accessed 31.10.08. Chami, R., Fullenkamp, C., & Jahjah, S. (2003). Are immigrant remittance flows a source of capital for development? (IMF Working Paper WP/03/ 189). Washington, DC: IMF.

Coate, S., & Ravaillon, M. (1993). Reciprocity without commitment: Characterization and performance of informal risk-sharing arrangements. Journal of Development Economics, 40(1), 1–24. De la Brie`re, B., Sadoulet, E., de Janvry, A., & Lambert, S. (2002). The roles of destination, gender and household composition in explaining remittances: An analysis for the Dominican Sierra. Journal of Development Economics, 68(2), 309–328. De la Cruz, B. (1995). The socioeconomic dimensions of remittances: Case studies of five Mexican families. The Berkeley McNair Journal, 3, 1–10. De Weerdt, J., & Dercon, S. (2006). Risk-sharing networks and insurance against illness. Journal of Development Economics, 81(2006), 337–356. De Witte, M. (2004). Money and death. Funeral Business in Asante, Ghana. Africa, 73(4), 531–559. Dekker, M. (2004). Risk resettlement and relations: Social security in rural Zimbabwe (Tinbergen Institute Research Series No. 331). Amsterdam: Tinbergen Institute. Dercon, S. (2000). Income risk, coping strategies and safety nets (CSAE Working Paper WPS/2000.26). Oxford: Center for the Study of African Economies. Dercon, S., De Weerdt, J., Bold, T., & Pankhurst, A. (2006). Group-based funeral insurance in Ethiopia and Tanzania. World Development, 34(4), 685–703. Doss, C. (2001). Is risk fully pooled within the household? Evidence from Ghana. Economic Development and Cultural Change, 50(1), 101–130. Fafchamps, M., & Gubert, F. (2007). Risk sharing and network formation. American Economic Review Papers and Proceedings, 97, 75–79. Ferguson, J. (1999). Expectations of modernity: Myths and meanings of Urban live on the Zambian copperbelt. Berkeley: University of California Press. Gluckman (1960). Tribalism in modern British Central Africa. Cahiers d’Etudes Africaines, 1, 55–70. Granovetter, M. (1973). The strength of weak ties. American Journal of Sociology, 78(6), 1360–1380. Greif, A. (1993). Contract enforceability and economic institutions in early trade: The Maghribi traders’ coalition. American Economic Review, 83(3), 525–554.

INFORMAL INSURANCE ARRANGEMENTS IN GHANAIAN MIGRANTS’ TRANSNATIONAL NETWORKS Grimard, F. (1997). Household consumption smoothing through ethnic ties: Evidence from Cote d’Ivoire. Journal of Development Economics, 53(2), 391–422. Guarnizo, L., Portes, A., & Haller, W. (2003). Assimilation and transnationalism: Determinants of transnational political action among contemporary migrants. American Journal of Sociology, 108(6), 1211–1248. Gubert, F. (2002). Do migrants insure those who stay behind? Evidence from the Kayes Area (Western Mali). Oxford Development Studies, 30(3), 267–287. Hoddinott, J. (1992). Rotten kids or manipulative parents: Are children old age security in Kenya?. Economic Development and Cultural Change, 40(3), 545–565. Hoddinott, J. (1994). A model of migration and remittances applied to western Kenya. Oxford Economic Papers, 46(3), 459–476. Johnson, G., & Whitelaw, W. (1974). Urban-rural income transfers in Kenya: An estimated-remittances function. Economic Development and Cultural Change, 22(3), 470–473. Kabki, M., Mazzucato, V., & Appiah, E. (2004). Wo benane a eye bebree: The economic impact of remittances of Netherlands-based Ghanaian migrants on rural Ashanti. Population, Space and Place, 10(2), 85–97. Klaver, J. (1997). From the land of the sun to the city of angels (Netherlands Geographic Studies No. 228). Amsterdam: Netherlands Geographic Institute. Lentz, C. (1994). Home, death and leadership: Discourses of an educated elite from north-western Ghana. Social Anthropology, 2(2), 149–169. Levitt, P. (2001). The transnational villagers. Berkeley: University of California Press. Ligon, E., Thomas, J., & Worrall, T. (2002). Informal insurance arrangements with limited commitment: Theory and evidence from village economies. Review of Economic Studies, 69(1), 209–244. Lucas, R. (2005). International migration and economic development: Lessons from low-income countries. Stockholm: Expert Group on Development Issues. Lucas, R., & Stark, O. (1985). Motivations to remit: Evidence from Botswana. The Journal of Political Economy, 93(5), 901–918. Massey, D. (1987). The ethnosurvey in theory and practice. International Migration Review, 21(4), 1498–1522. Massey, D. (2005). For space. London: Sage. Mazzucato, V. (2003). Asante transnational relations: Historical renegotiation or changing structure of social relationships? In Paper presented at CERES Summer School, June 23–26, 2003, Amsterdam. Mazzucato, V. (2005). Ghanaian migrants’ double engagement: A transnational view of development and integration policies (Global Migration Perspectives No. 48). Geneva: Global Commission on International Migration. Mazzucato, V. (2007). The role of transnational networks and legal status in securing a living: Ghanaian migrants in The Netherlands (COMPAS Working Paper 43). Oxford: University of Oxford. Mazzucato, V. (2008). Simultaneity and networks in transnational migration: Lessons learned from a simultaneous matched sample methodology. In J. DeWind, & J. Holdaway (Eds.), Migration and development within and across borders. Geneva: International Organization for Migration. Mazzucato, V., Kabki, M., & Smith, L. (2006). Transnational migration and the economy of funerals: Changing practices in Ghana. Development and Change, 37(5), 1047–1072. Mazzucato, V., van der Boom, B., & Nsowah-Nuamah, N. N. N. (2008). Remittances in Ghana: Origin, destination and issues of measurement. International Migration, 46(1), 103–122. McAllister, L., & Fischer, C. S. (1978). A procedure for surveying personal networks. Sociological Methods and Research, 7(2), 131–148. Nimako, K. (2000). De ghanese gemeenschap: Van migranten tot etnische minderheid [The Ghanaian community: From migrants to ethnic minority]. In I. Van Kessel, & N. Tellegen (Eds.), Afrikanen in Nederland (pp. 165–181). Amsterdam: KIT.

1115

Osili, U. O. (2004). Migrants and housing investments: Theory and evidence from Nigeria. Economic Development and Cultural Change, 52(4), 821–849. Osili, U. O. (2007). Remittances and savings from international migration: Theory and evidence using a matched sample. Journal of Development Economics, 83(2007), 446–465. Platteau, J. P. (1991). Traditional systems of social security and hunger insurance: Some lessons from the evidence pertaining to Third World village societies. In E. Ahmad, J. Dre`ze, J. Hills, & A. Sen (Eds.), Social security in developing countries. Oxford: Oxford University Press. Platteau, J. P. (1997). Mutual insurance as an elusive concept in traditional rural communities. Journal of Development Studies, 33(6), 764–796. Portes, A. (2001). Introduction: The debates and significance of immigrant transnationalism. Global Networks, 1(3), 181–193. Rosenzweig, M., & Stark, O. (1989). Consumption smoothing, migration, and marriage: Evidence from rural India. The Journal of Political Economy, 97(4), 905–926. Rouse, R. (1992). Making sense of settlement: Class transformation, cultural struggle, and transnationalism among Mexican migrants in the United States. In N. Glick Schiller, L. Basch, & C. Szanton Blanc (Eds.), Towards a transnational perspective on migration: Race, class, ethnicity, and nationalism reconsidered (pp. 25–52). New York: New York Academy of Sciences. Schrieder, G., & Knerr, B. (2000). Labour migration as a social security mechanism for smallholder households in sub-saharan Africa: The case of Cameroon. Oxford Development Studies, 28(2), 223–236. Smith, M. P. (2001). Transnational urbanism: Locating globalization. Oxford: Blackwell Publishers. Smith, L., & Mazzucato, V. (2003). Houses, businesses and urban livelihoods: The influence of transnational networks on economic activities in Accra, Ghana. In Paper presented at international workshop on migration and poverty in West Africa, March 13–14, 2003, University of Sussex, UK. Stark, O., & Bloom, D. (1985). The new economics of labor migration. American Economic Review, 75(2), 173–178. Stark, O., & Levhari, D. (1982). On migration and risk in LDCs. Economic Development and Cultural Change, 31(1), 191–196. Stark, O., & Lucas, R. (1988). Migration, remittances and the family. Economic Development and Cultural Change, 36(3), 465–481. Stiglitz, J. (1990). Peer monitoring and credit markets. World Bank Economic Review, 4(3), 351–366. Taylor, E. (1999). The new economics of labour migration and the role of remittances in the migration process. International Migration, 37(1), 63–88. Townsend, R. (1994). Risk and insurance in village India. Econometrica, 62(3), 539–591. Twum-Baah, K. A., Nabila, J. S., & Aryee, A. F. (Eds.) (1995). Migration research study in Ghana. International migration (Vol. 2). Accra: Ghana Statistical Service. Twum-Baah, K. A. (2005). Volume and characteristics of international Ghanaian migration. In T. Manuh (Ed.), At home in the world? International migration and development in contemporary Ghana and West Africa (pp. 55–77). Accra: Sub-Saharan Africa Press. Udry, C. (1994). Risk and insurance in a rural credit market: An empirical investigation in northern Nigeria. Review of Economic Studies, 61(3), 495–526. Van Dijk, R. (1997). From camp to encompassment: Discourses of transsubjectivity in the Ghanaian Pentecostal diaspora. Journal of Religion in Africa, 27(2), 135–169. Vertovec, S. (2001). Transnationalism and identity. Journal of Ethnic and Migration Studies, 27(4), 573–582.

Available online at www.sciencedirect.com