1In both North and South, the role of international migrants in the development of their home country has become a central topic of debate. Discussion has focused on the way in which transfers of goods and money affect the economies of the countries of origin, and the living conditions of sending communities and households. Using data from a survey on international migration conducted among 1,141 Dakarian households in 2008, David Lessault, Cris Beauchemin and Papa Sakho measure the role of migrant remittances in the urban renewal that has taken place in Dakar despite the ongoing economic crisis and the withdrawal of the state from the housing sector. Their findings are mixed: Senegalese migrant remittances help to cover the routine expenses of their households of origin (bills, rent) and this is certainly a useful contribution, but migrants do not appear to play a key role – directly at least – in financing the purchase or renovation of property in the capital.
2Over the last two decades, the participation of migrants in the development of their home country has become an increasing focus of debate. The idea is central to the policies of codevelopment advocated both by Europe and its Member States, and by the countries of Western Africa. One question in this area centres more specifically on how such initiatives should be financed and how urban growth should be managed in a context where state intervention is decreasing due to a lack of capital resources. In Senegal, many scientific and political observers agree that international migrants play a major role in promoting urban renewal. Via direct investments in property, from a distance or upon their return, or via cash transfers to their family in the sending country, international migrants are among the most active players in the improvement of household living conditions, especially since the devaluation of the CFA franc in 1994 which doubled their purchasing power and boosted their investment potential.
3Long recognized as a transit hub for migrants leaving the provinces of Senegal for a foreign destination, the Dakar region is today the country’s main supplier of international migrants, and of those heading to Northern countries in particular (Lessault, 2008). It is also Dakar that attracts the majority of migrant property investments (Tall, 2000). Despite the effects of a lasting economic crisis, housing quality in Dakar has improved substantially. It is tempting, therefore, to assume that property construction and urban renewal in the city are largely funded from abroad. Yet the assumption that international migrants are responsible for the improvements in the urban habitat, perceived as self-evident in the local social and political context, is based primarily on qualitative studies, not supported by any appropriate statistical data. This article aims to shed new light on the problem using quantitative data collected in a household survey entitled Migrations between Africa and Europe (MAFE, 2008).  These data enable us to address the relationship between international migration and housing from an original angle by examining how international migrants help to satisfy the housing needs of Dakarian households. Households are thus the focus of our analysis.
4The hypothesis whereby international migrants are key drivers of the improvement in household living conditions in Dakar is based upon a theoretical approach which considers the causes and consequences of international migration in terms of collective rather than individual strategies. It is founded upon the theories of the new economics of migration, and of networks, which challenge the neoclassical vision of migration as an act decided by individuals themselves exclusively for their personal benefit. Under the first theory, migration is an act which involves groups (families, communities) more than individuals. Migration is seen as a means for the community to diversify income sources and to make up for an inadequate local supply of credit, insurance, etc. A migrant’s departure may, for example, be a strategy to earn money for buying a family home (Stark and Bloom, 1985; Katz and Stark, 1986). Under the second theory,  migration networks are viewed as a set of interpersonal ties between migrants, former migrants and non-migrants which may be based on kinship, on friendship or simply on membership of the same community of origin (Massey and Arango, 1993). It therefore attributes a major role to the family, as a decision-making unit, but also as an institution providing the necessary human, financial and material support for migration (Zlotnik, 2003). In return for the chance to migrate, the migrant’s “debt” to the family is refunded through the transfer of goods and cash (Guilmoto and Sandron, 2001). In the context of Senegal, the approach based on network theory provides an opportunity to contribute to the ongoing debate on the individualization of behaviours in Africa and on the erosion of reciprocity in African societies, notably in the cities (Vidal, 1994; Marie, 1997).
5With these theoretical considerations as a backdrop, our aim is to determine whether international migrants do in fact play a key role in improving the living conditions of Dakarian households. The first section highlights the paradox whereby housing conditions in Dakar are improving despite declining household incomes and dwindling government funding for housing development. It is this very paradox which prompted – and today substantiates – the hypothesis that international migrants are the main drivers of residential property development in the city. In the following section, we present our data source and our methodological approach for analysing the relationships between migration and property. Our findings are presented in the third section. We examine, in succession, the housing conditions of households with or without migrant members, the priority given to housing in the uses of cash sent home by migrants, and the role of international migration in access to home ownership among Dakarian households.
I – Background
1 – Despite the crisis, housing conditions are improving
6According to the National Agency for Statistics and Demography (Agence nationale de la statistique et de la démographie, ANSD), Dakar, the capital of Senegal, had 2.2 million inhabitants in 2002 (ANSD, 2006). The urban agglomeration now covers practically all the Cap Vert peninsula (550 sq.km), with a mean population density of 4,000 inhabitant per sq.km. Although expanding more slowly than in the past, its annual growth rate is still around 2.5 % (compared with 4 % in the 1980s). If it continues to grow at the current rate, its population will double in 25 years, increasing by 80,000 inhabitants annually. As a result, the city has changed considerably over the last two decades. In terms of urban dynamics, a dual process is under way. First, with the rapid disappearance of available building land in urban areas, the existing urban fabric is becoming more dense (new rooms added to homes, more occupants per room, addition of one or more storeys). Second, the city is continuing to spread outwards, and new residential districts are being developed. The changing urban landscapes bear witness to this process.
7The city is developing rapidly eastward into new inland areas (Salem, 1998; Lessault and Sakho, 2008). The town of Rufisque, some 30 kilometres from the city centre, was recently absorbed into the conurbation of Dakar (Dakar, Pikine, Rufisque, Thiès and Mbour). This situation is a direct consequence of current housing construction strategies. Multi-storey buildings and apartment blocks, and houses built on two or more levels are rare in Dakar, where the urbanization process has long been characterized by rapid horizontal expansion: “In the absence of social housing or of access to credit, the construction of small houses by the owners themselves, using the very limited resources at their disposal, is the approach adopted by most” (Dureau et al., 2000). The situation has been modified recently, however, by the emergence of building techniques that enable families to enlarge their homes progressively, if and when the necessary cash becomes available. It is now possible to add one or more floors in order to house new family members, to set up a business or to rent out rooms to tenants, for example. The growing number of houses built on two or more levels, and the spread of this practice to suburban areas have changed the urban landscape of Dakar. In fact, according to the latest census of the population and the living environment (2002), almost one-third of Dakarian homes are now “multi-storey”, reflecting the new process under way in the capital.
8Gainsaying the alarmist predictions of the United Nations,  the quality of housing in Dakar has also improved thanks to vigorous construction activity. Most new buildings are now “permanent” constructions, even in previously rural areas affected by urban encroachment (Lessault, 2005). Traditional huts and shacks have practically disappeared from the urban landscape, representing only 3.7 % and 6.4 %, respectively, of the built environment according to the forecasting and statistics department of the Ministry of the Economy, Finance and Planning (DPS, 1988), and the National Agency for Statistics and Demography (ANSD, 2006). Housing amenities have also improved. In 1989, 40 % of Dakarian households were connected to the water mains and 63 % to the mains power supply (Antoine et al., 1995; DPS, 1988), while in 2002, the proportions had reached 77 % and 86 %, respectively (ANSD, 2006). In short “the urban areas have spread outwards, linking the densely and continuously built-up agglomeration to periurban zones ever more distant from the city centre (…). In morphological and demographic terms, the built environment includes more permanent structures, has increased in density and height, and has acquired modern architectural contours and colours” (Piermay and Sarr, 2007). Occurring over a generally inauspicious period, this change for the better is an apparently surprising development.
9The Senegalese government, formerly interventionist in the field of housing, no longer has the financial means to invest in social housing construction. This situation dates back to the implementation of structural adjustment plans from the early 1980s which led to sweeping reforms in government housing policy. Under pressure from donors, the state withdrew from the sector and introduced a policy that shifted responsibility for housing onto individuals and households (Mbow, 1992). The private sector housing industry was also restructured. Numerous semi-public enterprises emerged in the 1980s, and facilities were granted to housing cooperatives (Hamo, Scat Urbam, etc.). Alongside their role in land development, the cooperatives were expected to capitalize on the experience of operations such as the “Parcelles Assainies” programmes,  thereby contributing to self-help construction initiatives. The scheme is continuing today on the edges of the city (Keur Massar), but with highly selective conditions of eligibility for beneficiaries. Most households who do not meet the necessary solvency criteria are thus excluded from the land and housing programmes run by these semi-public companies.
10In the capital, in a context of inadequate housing supply – in both qualitative and quantitative terms – and ever-increasing demand, many households are obliged to opt for “involuntary” cohabitation, which increases the housing occupancy rates, or to look for alternative solutions, sometimes outside the law, on the outer edges of the city (Lessault, 2005; Diagne and Lessault, 2007). The housing shortage also encourages speculation which make access to home ownership very difficult, especially at a time of economic crisis when households’ purchasing power and saving capacity are reduced. As a result, the majority of home-owning households acquire their dwelling either outside the circuit of planned housing developments and/or by building it themselves. Households are thus assuming increased responsibility for their “residential fate” in a difficult socioeconomic context marked by an upsurge in urban poverty (Antoine and Fall, 2009). According to a recent study based on the findings of the Jeremi survey (Antoine, 2002), pauperization in Dakar is increasing and “inequalities are aggravated by the strategies applied by some to keep poverty at bay, and the means adopted by others to survive in an urban environment where access to employment or housing remains a luxury for a large section of the population” (Antoine and Fall, 2009). In such a context, the general improvement in housing conditions in Dakar is a surprising phenomenon. How can this paradox be explained?
2 – The explanation for this paradox in public discourse
11It is generally acknowledged that housing construction and the resulting changes in the urban landscape are underpinned by the activity of international migrants, notably in the real estate sector. This has been shown by researchers, and is also reflected in recent discourse and policy practice. A similar message is conveyed by the media. Indeed, several qualitative studies have shown that international migrants from sub-Saharan Africa are generally believed to invest primarily in property. For some researchers, real estate is seen as the preferred form of investment among migrant Africans. They note that, for cultural reasons in particular, African migrants in Europe may purchase up to two or three houses in the home country over a period of 10-20 years (Smith and Mazzucato, 2003). For others, “via their investments in the city, migrant investors give new impetus to urbanization processes that are often held back by a lack of financial resources. (…) In this hurly-burly, emigrants invest massively in the housing sector” (Tall, 2000). Observers also unanimously highlight the interest shown by the so-called “Senegalese abroad” in a sector as dynamic as the housing industry. This interest is now no longer limited to the search for a family home, but also increasingly involves the creation of small and mediumsized construction businesses (Barro, 2008). Most of these recent studies highlight the “pivotal” or even “providential” role of migrants in the housing sector, thus encouraging governments to define or structure their urban development policies to make the most of this phenomenon.
12In fact, the political decision-makers appear to have progressively integrated this new component into Senegalese housing policy, and emigrants are currently perceived as a mainstay of local development: “it was for this reason that the government set up the Ministry of Senegalese Abroad, and then the CSSSE ” (Tall, 2008). According to the Minister of Senegalese Abroad, the government has decided to view emigration as a potential partner for development, with the aim of smoothing the way for migrants wishing to return home and invest. The authorities have introduced a series of measures to address the difficulties faced by migrants seeking a place to live in their home country. In collaboration with the Banque d’habitat du Sénégal (BHS), the aim is to develop a network of bank branches abroad to enable expatriates to invest in property at home (the 500 billion CFA francs in annual remittances are used above all for construction projects). Practical “guides” have also been published to help migrants with administrative aspects. Certain Senegalese abroad have also set up associations to facilitate migrant investment in housing. 
3 – The economic potential of international emigration from Dakar
13Estimates of the size of the Senegalese diaspora vary substantially between sources. According to the United Nations, 463,403 Senegalese were living abroad in 2005 (DRC, 2007), while for the Ministry of Senegalese Abroad, the total is above 2 million. The latest population and housing census of Senegal (ANSD, 2006) gives detailed figures of recent emigration flows. For a total estimated population of 9.8 million inhabitants, the number of Senegalese who had left the country since 1997 and were still living abroad in 2002 was 170,000. The main destination countries are, in decreasing order, Italy, France, Gambia, Mauritania and Spain. This ranking confirms both the success of European countries and the declining appeal of France. Between them, these five countries attracted more than two-thirds (64 %) of Senegalese migrants who left the country between 1997 and 2002. More than a quarter of these emigrants came from Dakar, which is by far the country’s major migrant sending region.  According to the census, 10 % of Dakarian households have past or current members who recently emigrated abroad. Emigrants from Dakar also stand out by their choice of destination, with a much higher proportion than elsewhere choosing the United States and southern Europe (Ebin, 1990 and 1996; Lessault, 2008). Given these preferences, and the large numbers of emigrants concerned, the Dakar region probably has a stronger migrant investment potential than the country’s other regions.
14While no information is available on the total volume of remittances received by the region of Dakar, recent estimates show that the practice has developed to an unprecedented extent in Senegal. According to World Bank data, Senegal, with total remittances of USD 0.9 billion, was among the ten main recipient countries of migrant remittances in sub-Saharan Africa in 2007, behind Nigeria (USD 3.3 billion), Kenya (USD 1.3 billion) and Sudan (USD 1.2 billion). These figures are corroborated by a study published in 2008 by the Senegalese forecasting and economic studies division (Direction de la prévision et des études économiques, DPEE, 2008), which gave a total of USD 0.944 billion in 2007 versus USD 0.843 billion in 2006 and USD 0.586 million in 2005. These remittances represent “7 % of GDP, 28 % of bank deposits and 25 % of the money supply, ranking Senegal in fourth place among the countries of sub-Saharan Africa in terms of absolute value, after Nigeria, Lesotho and Sudan” (Tall, 2009). Besides, these totals do not include “unofficial” transactions.
15Last, as observed in Senegal and elsewhere in West Africa (Bertrand, 1999, 2006 and 2009; Smith and Mazzucato, 2003), the figure of the international migrant as a “builder” and a “property developer” is very present in public and political discourse, in the media, and in the scientific community. However, in the absence of quantitative data, the questions of the true impact of international migrants on the housing sector and of the possible existence of other sources of funding remain unanswered (Diop, 2008).
II – Method
16The main purpose of this study is to test the hypothesis whereby international migrants play a key role in improving the living conditions of Dakarian households. The MAFE survey (Migrations between Africa and Europe) was designed to study the relationship between migration and housing via a range of statistical measures. The MAFE project aims to produce quantitative data on international migration to shed light on migration patterns (determinants of departure, return, circulatory migration practices, choice of destination, etc.) and to analyse the effects in the sending country (Beauchemin et al., 2007). The Senegalese part of the project  involved a quasi-simultaneous series of sociodemographic surveys conducted in 2008 in both Senegal and Europe, in the three main destination countries of Senegalese migrants (France, Spain, Italy). Two questionnaires were designed for these surveys: a household questionnaire (in Senegal only) and an individual questionnaire (in all four countries). The present study, which focuses on the housing conditions of Senegalese households, is based on data collected via the household questionnaire.
1 – The MAFE sample in Senegal
17In Senegal, MAFE data collection was limited to the Dakar region, where 1,141 households were interviewed. The sampling plan was designed to overrepresent households with international migrants while providing a representative sample of all households in the Dakar region (see Appendix). The results presented in this article are weighted to correct for over-representation and non-response. As MAFE is a sample survey and not an exhaustive data source such as the census, it cannot be used for morphological or spatial studies to identify the exact distribution of migrant investments across the city. Moreover, certain micro-districts with a high concentration of migrant investments may have been missed by the sampling procedure, despite our best efforts to represent zones with a high density of households with migrant members. This is the case, for example, for the “diaspora” homes recently built at the Liberté V bus terminus and for other emigrant neighbourhoods in the outskirts of Dakar (M’Bao and Keur Massar development zones). To ensure representativeness and to evaluate the true impact of international migrants at the level of the entire urbanized region, it is important to avoid “zoom” effects which tend to exaggerate the role of expatriates in the overall dynamics of the property market. In any event, we possess a representative sample of the population living in the Dakar region which can be used to determine the extent to which experience of migration influences the living conditions of Dakarian households. Note that only primary residences are taken into account, and not empty homes and dwellings occasionally occupied by migrants and their families during visits to the country. Consequently, investments by migrants for personal and occasional purposes, and not for households living permanently in Dakar, are excluded from our analyses. 
2 – Measuring the relationships between living environment and migration
Three complementary viewpoints
18The MAFE household survey can be used to study relationships between migration and living conditions from three different angles. The first concerns criteria of comfort (type of dwelling, occupant density, connections to mains water/power supplies, household amenities). The situations of households with and without migrant(s) are compared at the time of the survey (2008), taking account of their socioeconomic characteristics. If it is true that migrants contribute to improved living conditions in Dakarian homes, then households with migrant(s) should have more amenities than those without migrants, all other things being equal. If this finding were confirmed, however, it would not prove that migration is the cause of improved living conditions: migrants may simply come from better-off households. In other words, better living conditions might be the cause rather than the consequence of migration. Other measures are needed, therefore, to establish the direction of the relationship between living conditions and international migration.
19The second angle concerns the contribution of migrant remittances to spending on housing (routine expenses and/or property purchase). Two types of information are available in the survey. First, a questionnaire module is devoted to remittances received by households; it includes several questions on the way this money is spent, for housing in particular. Second, the module on housing conditions includes questions on how the purchase and transformation of the dwelling occupied at the time of the survey was paid for. By matching these two items of information, the share of migrant remittances spent on housing can be calculated, as well as the proportion of homes that are paid for in this way. These direct measures have a well-known limitation, however, due to the fungibility of migration remittances. By covering other routine expenses (food, utility bills, etc.), remittances may also enable households to save money towards a future home purchase, though this is not reported in the answers to the questionnaire. In other words, the money may not be invested directly in housing, but may free up other resources for this purpose.
20The third angle involves comparing households with and without migrant(s) not only at the time of the survey, but also at their time of entry into the dwelling. If it is indeed easier for households with migrant members to buy their own home, then the home ownership rate of households with migrant(s) at the time of entry into a dwelling should be higher than that of households without migrants. And since home ownership is governed by many factors in addition to household migration status (such as socioeconomic status of head of household, for example), we perform a multivariate analysis to measure the influence of household migration status on the probability of buying a house, all other things being equal.
21In short, these three angles of investigation combine two approaches. One uses households’ responses to determine whether housing is funded by migration (second angle). In this direct approach to the migration/housing relationship, the results depend largely on the way in which questions on money earned abroad are formulated and understood. The two other angles, on the other hand, rely on an indirect approach whereby the migration/housing relationship is diagnosed by observing statistical correlations between migrant status and housing conditions. In both cases, the results depend on the dividing line that is established between households with and without migrant(s). The following sections discuss the assumptions upon which the results of these direct and indirect methods are based.
What is a “household with migrants”?
22The definition of a household (Pilon and Vignikin, 2007) used in the MAFE survey is fairly standard: it is “a group of individuals who live together and who share all or part of their income to meet their primary needs (housing and food in particular)”.  The originality of our survey lies in the fact that household migrants, i.e. individuals who, by definition, are not members of the household (because they live elsewhere) but who have ties with it, are included in the questionnaire. Who are these individuals? The questionnaire records several categories of international migrants, i.e. of persons living outside Senegal:
- children of the head of household 
- spouse(s) of household members
- relatives of the head of household or of his/her spouse who have maintained regular contacts  with the household in the 12 months preceding the survey.
23Starting from this broad definition, our analyses seek to pinpoint the role of migration in urban renewal by dividing households with migrant(s) into different types:
- by individuals’ migrant status: beyond the ro le of current migrants (individuals living outside Senegal at the time of the survey), we will examine whether return migrants have a specific influence on housing conditions;
- by number of migrants: the idea is to test the existence of a potential cumulative effect whereby investment potential increases with the number of migrants, especially in the case of housing projects requiring large investments;
- by migrant destination: we will test the hypothesis whereby migrants living in Europe – who tend to have higher incomes and culturally different residential behaviours – contribute more to housing improvements in Dakar than their counterparts living in Africa;
- by the relationship between the migrants and the head of household: we will attempt to determine whether solidarity between migrants and Dakarian households is limited to close relatives (spouses and children of the head of household), as suggested by studies on the individualization of social behaviours (Marie, 1997), or involves broader social networks.
What is “money earned abroad”?
24To examine the influence of international migration on the living conditions of Dakarian households, one of our three methods involves analysing respondents’ answers to questions about money earned abroad. On this point, the MAFE household questionnaire provides two complementary types of information. First, an entire questionnaire module is devoted to cash remittances or goods received by households during the twelve months preceding the survey (module C).  Several questions concern the way money sent back by migrants is spent, notably for housing. A distinction is made between routine expenses (to pay the rent, bills, etc.) and investment spending (to buy a home, a plot of land, etc.). A second module includes questions on living conditions (size and type of dwelling, occupants per room, connection to mains power/water supply, goods owned, etc.) but also, for property owners, on how the occupied dwelling was purchased and modified (module E). For each type of investment (purchase of a plot or of an existing dwelling, construction or modification of a dwelling), two types of question are asked:
Has the plot [or other investment] been paid for, in part at least, with money earned abroad? (multiple answers possible): No; Yes with money earned in another African country; Yes, with money earned in Europe; Yes, with money earned in another country; Don’t know.
Has this money been earned by… (multiple answers possible): The household head; A person we already talked about (note the person’s ID); Somebody else.
26These questions enable us to measure the extent to which Dakarian households (homeowners, tenants or occupants housed for free) make use of money earned abroad to pay for their plot of land, dwelling construction or rebuilding/modification (adding a room on the ground floor, constructing an additional floor, building a fence, etc.). They also identify the source of this money (profile of migrant(s)).
27These two modules thus shed light on the migration/housing relationship from a dual perspective: the share of remittances spent on housing (module C), and the contribution of money earned abroad to housing expenses (module E). The information in module E gives no indication of the extent to which money from abroad contributed to total investment spending on the occupied dwelling. But it does go beyond certain limits inherent to the more standard questions on remittances which:
- are limited to the 12 months preceding the survey, while housing investments often involve a long initial period of capital accumulation;
- exclude money not directly received from abroad, such as money saved by migrants in the host country before returning to settle in their home country, for example.
III – Results
1 – Household living conditions vary according to their links with international migration
28Here, household living conditions are captured via the type of building (horizontal or vertical, i.e. multi-storey house or apartment) and various comfort indicators (occupants per room, connection to the mains power/water supply, sewerage, cable or satellite television, etc.). Via several series of logistic models, we examined the extent to which these various indicators are correlated with the household migrant status, independently of the household socioeconomic characteristics (Table 1). In a first series of models, migration status is first captured as simply as possible: the main explanatory variable of interest indicates whether the household includes migrants or not, whether they have returned or are still living outside Senegal (current migrants). In the following series (series 2-5), the models estimate the effect of more detailed migration statuses (distinction between current/return migrants; number of migrants; destination of migrants; migrants’ relationship with head of household).
29The first series of models shows that households with migrant(s) are, in practically all cases, more likely to be better housed than households without migrants. First, they are less likely to live in an overcrowded dwelling with a number of persons per room above the regional average. Second, households with migrant(s) are more likely to be connected to the sewerage system (+85 %), to have a mains power supply (more than twice as likely), and to have cable or satellite TV (almost three times more likely) than households without migrants. These results probably reflect a geographical localization effect. An analysis of Senegalese census data at infra-urban level shows that migration intensity is much higher in neighbourhoods close to the city centre than in the outer suburbs. We thus observe a general gradient of decreasing migration prevalence from the city centre to the suburbs, which coincides with a similar decrease in the household wealth index.  The fact that households with migrant(s) are relatively concentrated in the central neighbourhoods of the agglomeration (Dakar département) means that they are much more frequently connected to the mains power supply and sewerage system. Migration status is not correlated with all aspects of housing conditions, however. Exploratory analyses show, for example, that the presence of a tap inside the dwelling (rather than tap outside, well, standpipe, etc.) and internet access are not correlated with household migration status (migration status has no significant effect in any model; results not shown). Moreover, at first sight, there is no apparent link between building type (vertical or otherwise) and household link to current or return migration. The results given below, which detail the link between households and migration, qualify these preliminary findings.
30The second series of models refines this initial analysis and reveals notable differences between households who report current migrants and those with return migrants. For all indicators, the likelihood of living comfortably is higher when the household has at least one migrant living abroad. Return migrants do not have such a strong effect on the housing conditions of Dakarian households. The number of migrants (third series of models) has an effect only for households with current migrant(s): a household with at least three migrants abroad is four times more likely than the others to be connected to the sewerage system, to have a mains power supply and a cable or satellite TV. The migrants’ destination (fourth series of models) also influences households’ housing conditions. When migrants live outside Africa (primarily in Europe), the probability of living in a comfortable home is higher from all points of view. For households with at least one migrant living elsewhere in Africa, this is rarely the case: only for the number of occupants per room and cable or satellite TV are the effects positive and significant. Last, the fifth series of model reveals the strong influence on housing conditions of the migrant’s kinship tie with the household head. When the migrant is a spouse or child of the household head, households are much less likely to live in an overcrowded dwelling, they are twice as likely to be linked to the sewerage system, and up to five times more likely to have a mains power supply.
31Predictably, these initial findings show that at the time of the survey, households with migrant(s) were more likely overall to be better housed than those without. They also show that the effect of migration is highly selective. In practically all areas, it is strongest for households with numerous migrants, for those with migrants living outside Africa at the time of the survey, and for those with close kinship ties to the migrants (spouses and children of the household head). These intuitive results support the hypotheses formulated above. However, they do not establish a causal relationship between migration and housing. While the most comfortably housed households are also those linked to migration, we do not know whether they were better off in the first place, and therefore able to finance the emigration of household members to costly destinations, or whether their living conditions have improved thanks to migrant remittances.
2 – Use of migrant remittances: limited investment in housing
32To determine the effect of migration on housing conditions, we assessed the extent to which Dakarian households used migrant remittances to invest in housing. An initial analysis of remittances sent over the previous twelve months reveals that the money is used more to cover routine expenses that to invest in property: “Above all, money earned abroad provides a means for households to live – or simply survive – on a daily basis in Dakar. It serves mainly to cover daily household expenses, and is rarely saved or used to invest in property” (Diagne and Rakotoranivo, 2010). According to this study, only 1.7 % of households used migrant remittances received over the previous year primarily to buy a dwelling, and only 0.3 % of households gave priority to saving. By contrast, in more than one-third of cases, migrant remittances were used above all by households to pay rent and bills.  This unsurprising finding (Gubert, 2005; Ndione and Lalou, 2005) is linked to the very nature of the measure: the observation period is limited to just one year, so does not capture the long-term capital accumulation required for property investment. Direct questions on the way the dwelling was paid for give a more precise idea of the contribution of migrant remittances.
33All households who were homeowners, tenants or housed for free and who were able to reply (696 households, representing 61 % of the survey sample) were asked: “Has this dwelling (purchase or construction) been paid for, in part at least, with money earned abroad?” (…); “Have you modified this dwelling since you moved in? If so, have these modifications been paid for with money earned abroad?”
34Among these households, 14.2 % report using money earned abroad to purchase their dwelling (Table 2). Given the intensity and frequency of the links between Dakarian households and international migration (54 % of households have migrant members and 25 % of household heads have a spouse or child currently abroad), the proportion seems quite low. Among households who have modified or enlarged their dwelling,  16.8 % report using money earned abroad (Table 2). So the wealth accumulated by migrants abroad is not sufficient to explain the recent improvements in housing conditions observed in the Senegalese capital.
35That said, if we focus analysis on the type of modifications (additional floor(s) or room(s) on the ground floor) we nonetheless observe differences in household behaviour by migration status. These two types of modification reflect the recent tendencies – i.e. multi-storey construction and horizontal spread – responsible for the increase in housing density observed in Dakar. We see that among households with migrant(s) who have modified their home, 18 % (N = 386) have added another floor, versus just 8 % (N = 327) among households without migrants. Migration status may thus have a relative influence on the decision to build another storey. For modifications involving the addition of a room on the ground floor, the difference is smaller, but the proportions are higher: 27 % and 22 %, respectively. These findings suggest that households with migrant(s) build vertically more often than the others, but this is not the only solution adopted to enlarge their home.
36Among households who use money earned abroad to buy (N = 104) or to modify (N = 86) their home, we observe that the contributors differ by type of investment (Table 2). Purchases are financed mainly with money earned abroad by the household head him/herself (39.6 %) and by his/her spouse (25 %), while home improvements are financed mainly with money earned by the children of the household head (32.3 %). In addition, households where the migrant is a child or spouse of the household head use money earned abroad to buy or modify their home twice as often as those where the migrant is another relative (Table 2). Clearly, this shows that with regard to housing, solidarity between migrants and Dakarian households is strongest among closely related family members (spouses and children of the household head). This finding is consistent with the earlier observation (section III.1) whereby households whose migrants are children or spouses of the household head are more likely to occupy a comfortable dwelling.
Number of households using money earned abroad to purchase or modify a dwelling
Number of households using money earned abroad to purchase or modify a dwelling
37Almost two-thirds of households who have used money earned abroad to buy or modify their home report that it comes from Europe. But this proportion again conceals a structural effect linked to the geographical distribution of Dakarian international migrants who, as we have seen, have tended in recent years to migrate primarily towards European countries.
38Be it in terms of the way money earned abroad is used or the way home purchases or renovation are funded, we note that Dakarian households invest relatively little migrant money in property. These findings suggest that international migration is not a key driver of the general improvement in Dakarian housing conditions, including among households with migrant(s). Given the extent of international migration (more than one in two Dakarian households has at least one current or return migrant), the use of money earned abroad to finance, even partially, the purchase or renovation of their home is not especially frequent. Moreover, property investments mainly concern a very narrow circle of migrant contributors (children and spouse of household head), although a second more distant circle (in terms of kinship ties) may contribute to routine expenses and more modest home improvements. 
3 – International migration and home ownership
39To tease out the relations of cause and effect between international migration and living conditions, our last focus of attention is the transition to owner-occupier status. The aim is to determine whether international migration enables Dakarian households to acquire a more secure occupancy status. Strictly speaking, this question requires us to build a duration model to calculate the probability for a household of a transition from tenant to owner status, taking account of the time that elapses and of the change in the household’s position over time in socioeconomic terms (greater wealth or poverty), in sociodemographic terms (variations in the household size and composition) and in migratory terms (variation in the number of international migrants linked to the household, their destination, length of absence, their ties with the household). But such a detailed analysis is not possible with the MAFE household survey data. 
40The survey can be used, on the other hand, to determine whether the occupancy status (owner or tenant) of the household varies according to its migration status, not only at the time of the survey but also at the time of entry into the dwelling.  It thus enables us to determine the direction of the migration/ownership relationship since we know whether migration preceded or followed the transition to owner-occupier status. If the migrants reported by a household left after the dwelling was purchased, migration cannot be seen as an explanatory factor for home ownership. Conversely, if the dwelling is bought after the emigration of household members, this supports the hypothesis of a causal effect of migration on home ownership.
41To begin with, we compare the home ownership rates of households on two separate dates: at the time of the survey and at the time of entry into the dwelling. However, the definition of migrants in the household is more limited than in earlier sections of this article. By looking back in time (to the time of entry into the dwelling) we are obliged to discount all migrants who are not household heads, their children or spouses. In other words, the category of migrants who are related to the household head and/or spouse and who have been in contact with him/her in the previous year cannot be included because the twelve-month time criterion does not identify individuals who may have contributed to the household economy further back in the past. In this section, we thus define a household with migrant(s) at the time of entry into the dwelling as any household with at least one member (including household head, his/ her children or spouses) living outside Senegal or having previously lived abroad in the year of entry into the dwelling.
42At the time of the survey, the results on occupancy status mirror the relative advantage in terms of housing conditions of households with migrant(s) over those without, with the former having a much higher proportion of owner-occupiers than the latter: 52 % versus 36 % (Figure 1). However, at the time of entry into the dwelling, the proportions of owner-occupier households with migrant(s) and without migrants are very similar: 45 % and 46 %, respectively. Hence, at the time of entry, there does not appear to be a relation between migration status and occupancy status.
43This finding calls for a more detailed analysis, however, since home ownership clearly does not depend solely on household migration status. We therefore conducted a multivariate analysis of transition to home ownership (Table 3). A first logistic model was built to isolate the net effect of household migration status on the likelihood of becoming owner-occupiers (Model 1). The analysis covers all surveyed households in the Dakar region who entered their dwelling between 1970 and 2008, i.e. a total of 914 households, each forming one observation unit. The main variable of interest in this model is the household migration status, which comprises two categories: households with or without migrant(s) and/or return migrants at the time of entry into the dwelling. The model also includes various characteristics of the household head that may influence his/her occupancy status (level of education, place of birth, religion, sex, age).  The period of entry into the dwelling is also included in the model to identify any potential effects of devaluation on home ownership. Since the literature suggests that the depreciation of the CFA franc stimulated migrant property investments, we would expect households with migrant(s) to have a higher chance of becoming home owners after 1994. To test this hypothesis, two complementary models were built. One specifically concerns households without migrants at the time of entry into the dwelling (Model 2) and the other concerns households with migrant(s) (Model 3).
Factors influencing the “risk” of becoming a home owner for a Dakarian household at the time of entry into the current dwelling, 1970-2008 (odds ratio)
44The results of the first model confirm that the likelihood of becoming a home owner do not vary significantly by household migration status, whatever the household characteristics or the period of entry into the dwelling: the difference between households with and without migrant(s) is not significant.  The period of entry, on the other hand, appears to be a highly significant variable. For all households (Model 1), we observe that the likelihood of becoming a home owner has decreased substantially over time, and has done so for households both with and without migrant(s). For the former, the devaluation does not cause a break in the trend: the chance of becoming a home owner did not increase after 1994 (Model 3). What is the explanation for this finding? First, the gains in migrant purchasing power depend on the country where they live. As many migrants are living in other African countries, they may have been affected by devaluation just as much as their families who remained in Senegal. Second, persons whose purchasing power in Senegal increased as a result of devaluation may have invested in goods other than the dwelling occupied by their relatives or spouses, preferring to invest in rental property, for example. But these investments are not captured in this analysis since they concern households and not the migrants themselves. Last, the downward tendency observed for all households could be the result of two simultaneous processes: the spread of poverty in the Dakarian agglomeration and the increase in land and property prices.
45In addition, what are the characteristics of a household head which have a significant influence on their chance of becoming a home owner-occupier? Not surprisingly, the older they are, the more likely they are to own their dwelling. Being of Dakarian origin is also a determining factor. Natives of the Cap Vert peninsula (of which Dakar forms a part), who have privileged access to land, are much more likely to become owner-occupiers: among all households, Layenes  are almost six times more likely than the others to be home owners. The place of birth, on the other hand, has no impact on the probability of home ownership. By contrast with the criterion of being a “native”, being born in Dakar does not ensure easy access to building land. The finding for the sex of the household head is, at first sight, more surprising: women are almost twice as likely to be home owners as men. This result does not tally with the idea that access to land and property ownership is more difficult for women in Senegal. It probably reflects a data collection artefact: in home-owning households where the husband is absent because the main residence is elsewhere, it is the woman who acts as household head in the survey. She is thus taken to be the home owner, even though in many cases, the “real” owner is in fact the husband whose main residence is elsewhere. This may be the case for polygamous families in which the spouses occupy different dwellings, visited in turn by the husband. It may also be the case for households where the husband is an international migrant: he is not present in the household, so cannot be identified as the owner in the survey but is nonetheless the owner of the dwelling where his wife and children reside.
46Returning to the relationship between migration status and household home ownership, what conclusions should be drawn from this multivariate and retrospective approach? The findings obtained when household migration status is taken into account not at the time of the survey but at the time of entry into the dwelling challenge the idea that there is a causal link between the migration of household members and home ownership. A simple comparison of home ownership rates of households with and without migrant(s) does not reveal any difference between these two populations. Likewise, no significant differences are revealed by the multivariate analysis. The supposedly positive effect of migration on property ownership is not confirmed. It would appear, rather, that owner-occupier households are more likely than the others to have migrant members living outside Senegal. The causal link is reversed: in most cases, home ownership does not appear to be the result of migration, but rather – as an indicator of household well-being – a factor that favours migration.
47International migrants are generally seen as the new drivers of urban renewal in Dakar. They provide an explanation for the paradox whereby the quality of housing is improving despite economic crisis and state withdrawal from housing construction. After examining this hypothesis by means of several complementary approaches, our conclusions are more nuanced.
48Our analyses show that a large proportion of households are concerned by international migration in the Dakar region, and that close relationships are maintained between migrants and their reference household. Through the remittances they send back to their family, international migrants contribute to routine household expenses, helping tenant households to pay their rent and their utility bills. In this way, they improve their families’ living conditions and free them from certain economic constraints, sometimes enabling them to invest elsewhere. Yet the strong and intense links between migrants and their household of origin do not tell the whole story. Our results on the contribution of international migration to improvements in living conditions or to the acquisition of property do not fully support the generally accepted belief that migrants are key players in Dakarian urban renewal. First, the remittances sent by migrants are rarely used by the beneficiary households to build or renovate a dwelling. Second, analysis of home ownership shows that migrants have no influence on home purchases. It would thus appear that international migration plays no more than a relative role in the acquisition of property or in refurbishment by homeowners. So what is the explanation for the paradox of housing improvements in a depressed economic context?
49First, migrants may invest in property outside their reference household, and/or elsewhere than in Dakar. Our analysis is limited to households in the Dakar region, so the individual behaviour of international migrants in relation to property investment is only partially explored. Houses built by migrants may not be destined for the family but for the rental market. Although more in-depth analysis is not possible, we observe, for example, that a non-negligible share of tenant households (40 % of households able to reply: N = 242) report that the dwelling owner lives or lived abroad. It is possible, therefore, that they contribute to the dynamics of Dakarian urban renewal in other ways. They may also invest outside the Dakar region, in their town or village of origin, for example, or in other newly attractive geographical areas (tourist resorts on the Petite Côte, for example). Last, while the role of international migrants is less pivotal than expected, other new players are emerging within Senegal, and other private property developers are taking over from the state. This raises the question of how the land and property markets in Dakar are organized. To what extent can private home builders compensate for the withdrawal of the state and respond to growing demand for housing in both qualitative and quantitative terms? And what role can international migrants play among the emerging actors of the housing construction and rental markets?
AcknowledgementsThe survey was conducted with the financial support of INED, the Agence nationale de la recherche, the Île-de-France region and the FSP programme “ Migrations internationales, recompositions territoriales et développement dans les pays du Sud”.
Household sampling strategy in Senegal
50To draw a probabilistic representative sample of households in the region of Dakar, a three-stage stratified random sampling strategy was applied using the 2002 Population Census as a sampling frame. At the first stage, census districts, which include about 100 households in Senegalese urban areas, were randomly selected with varying probabilities. At the second stage, households were selected randomly in each of the selected primary sampling units. At the third stage, individuals were selected within the households.
First stage: selection of primary sampling units
51Sixty census districts were randomly selected. This number of primary sampling units ensures a balance between a large geographical dispersion of households (which decreases sampling errors) and a more concentrated sample (which reduces costs). The region of Dakar was divided into 10 strata of equal size, according to the percentage of migrant households within each of them (on average, 11.6 % of the households were “migrants”). Six census districts per stratum were drawn, with a probability proportional to the number of households within each census district. By doing this, districts with a large number of migrants were more likely to be selected than those with low numbers of migrants. This provides samples of return migrants and of households affected by migration that are large enough for statistical analyses.
52As important changes had taken place during the period between the Census in 2002 and the MAFE fieldwork in Dakar, especially in suburban areas, the listing of households in the 60 randomly selected census districts was updated before starting the second stage.
Second stage: selection of households
53Households were randomly selected from the updated list of households in the selected primary sampling units. Two strata were distinguished: households with migrant(s) and households without migrants. “Migrant households” could not exceed 50 % per district. Selected households that could not be reached (absence, refusals, etc.) were not replaced during the fieldwork. Replacement would distort the computation of sampling weights, and could also bias the sample. Instead, 22 households were selected to reach an effective sample size of 20 households per census district on average (the original target was 1,200 household questionnaires); in other words, we expected a potential drop of 10 %.
Third stage: selection of individuals
54Individuals were selected within households for the life history survey. In each household, individuals were classified into 3 non-overlapping strata:
- return migrants, who were aged 18 or over at their (first) departure (or whose age at departure is unknown);
- spouses/partners of migrants (if the spouse/partner is not a return migrant himself/herself);
- other people.
- up to 2 return migrants (random selection if more than two in the households, all the return migrants were selected if not more than two in the household);
- up to 2 spouses/partners of migrants (random selection if more than two in the household);
- other individual.
The Senegalese part of the MAFE project is coordinated by INED (Cris Beauchemin) in association with the Institut de population, développement et santé de la reproduction (IPDSR) of Cheikh Anta Diop University in Senegal (P. Sakho). The other partners are: Pompeu Fabra University (P. Baizan), Consejo Superior de Investigaciones Científicas (A. Gonzalez-Ferrer), and Forum Internazionale ed Europeo di Ricerche sull’Immigrazion (E. Castagnone). To find out more: http://www.mafeproject.com/
The social network theory, for its part, considers that all human actions and the ensuing social processes are embedded in social networks (Granovetter, 1990).
In her official speech in Dakar in 2005, the executive director of UN-Habitat pointed out “the need to bring an end to the proliferation of informal settlements in the major African cities, and in the Senegalese capital especially”. Senegal, one of the most urbanized countries of sub-Saharan Africa (with almost 50 % of urban dwellers), has a “very high prevalence rate of informal settlements (…). The case of Dakar, whose population increases by almost 120,000 people each year, is especially worrying, since most newcomers will head for neighbourhoods that are cruelly lacking in basic infrastructure and services”.
These operations led to the development and attribution of almost 10,500 building plots in the inner suburbs of Dakar in the 1980s.
Comité de suivi du symposium des Sénégalais de l’extérieur (follow-up committee of the symposium of Senegalese abroad).
One such example is Paris MBoko International (PMI), a mutual aid association based in France set up to find housing plots in Dakar for Senegalese emigrants.
The main emigration zones are the regions of Dakar (26 % of recent emigrants), Touba (7 %) and the départements of the Senegal River Valley, Matam and Podor (12 %). Almost half of all departures (45 %) are from these three main regions of Senegal.
The project includes two other parts, one on migration from Ghana, and the other on migration from the Democratic Republic of the Congo (DRC). For further information on the project, see http://www.mafeproject.com/
However, these investments are included in studies that make use of MAFE data at the individual migrant level (Mezger and Beauchemin, 2010).
This definition corresponds to the concept of njël in Wolof, ngaak in Serer, hirande in Pular, niakhamé in Soninke, fousil or sinkamé in Jola, sinkiro in Manding. To be considered as household members, individuals must have been living or must intend to live in the household for at least 6 months.
In fact, all the children of the head of household are recorded in the questionnaire, whatever their place of residence, be it the household itself, another household in Senegal or another household abroad.
Respondents were free to interpret this notion of regularity of contact, with regard to both type (phone calls, visits, letters, remittances, etc.) and frequency.
This criterion was excluded because the household can only be defined at the time of survey and not in the past. The household corresponds to a given list of individuals living together in one place and at a particular moment in time. Referring to past households introduces too much complexity, as changes of residence and changes in household composition must be taken into account. A household which has changed in some way is no longer the same household and cannot therefore serve as a reference.
The questionnaires are available online at http://mafeproject.site.ined.fr/en/
The results of these studies focusing specifically on intra-urban analysis of migration practices in Dakar are being developed in a project by D. Lessault and C. Mezger on territories of recent international emigration in Senegal.
For further details on the use of migrant remittances in Dakar, see the study by Alioune Diagne and Andonirina Rakotoranivo (2010).
i.e. 477 households who are homeowners, tenants or housed for free, 40 % report having refurbished their dwelling, 25 % report having added a room on the ground floor and 13 % report having built an additional floor.
This finding is logical in the highly urbanized context of Dakar where families are tending to “nuclearize”, even though complex households of extended families still exist in the city. This is a contextual effect that would probably not be found elsewhere, in rural areas especially.
To our knowledge, such an analysis is not possible using survey data. The main obstacle relates to the notion of household, which is a fundamentally cross-sectional concept. At the time of the survey it associates a place (dwelling) and a set of persons with diverse statuses (household head and others). Longitudinal observation of households is extremely complex: are we still talking about the same household when its members move to a new home, or when a new person becomes head of household? Observing changes in households over time is still a methodological challenge. This type of analysis has been performed at individual level, however (see Mezger and Beauchemin, 2010).
This is made possible by the fact that the date of entry into the dwelling (that can be assimilated to the date of purchase for owners) and the dates of first departure of household migrants are both known. In our sample, 21 % of households entered their dwelling before 1980, 22 % between 1980 and 1994, and 57 % after 1994.
This analysis is based on the assumption that the individual who was household head at the time of the survey was also household head at the time of entry into the dwelling. Variables relating to the marital status or type of occupation of the household head could not be used since the available information does not vary over time: only the characteristics at the time of the survey are available.
Other models were tested with more refined migration status variables, taking account of the location of migrants (returned to Senegal, elsewhere in Africa, in Europe or in the rest of the world), their number or the cumulative period spent outside Senegal (variable that aggregates periods spent abroad by all migrants). The detailed results are not presented here because they are no different from those of the simple model: migration experience at the time of entry into the dwelling, of whatever nature, never has a significant effect on the probability of being a home owner.
Layenes belong to a Muslim brotherhood whose members are mainly from the Cap Vert peninsula. Most of them are natives of the Dakar region. This gives them privileged access to land and to the status of owner-occupier if they have the means to build a home.