1 – Occupational and residential choice as interdependent fields of decisions

6Migration choices will depend first on the individuals’ needs, which are most significantly expressed in two fields — that of the occupation and that of the residence. In the first instance, individuals mostly consider the local characteristics of the labour market — trends in the supply of jobs and the level of unemployment, characteristics of the available jobs and required level of skills. In the second instance, they will consider the characteristics of housing (type of dwelling, size of the living area, quality of appliances, etc.) and real estate (e.g. area of the adjacent plot of land) as well as the general quality of the environment. The latter refers to a complex set of characteristics opposing the attractions of nature that are more specifically rural, such as the landscape or the low density of population, to more specifically urban traits such as better public services and equipment.

7As suggested by Jayet (1996), all those factors should be taken into consideration for a good understanding of migration decisions. Our hypothesis is that professional and residential motivations are combined (Détang-Dessendre and Molho, 1999) and that one or the other will carry the most weight because of the individual’s personal characteristics (age, sex, social origin, marital status, educational level, etc.) (Gordon and Vickerman, 1982). Some characteristics such as marital status or educational level may change over time and thoroughly modify the individuals’ needs and/or the constraints they must face. This position is close to that of Clark and Hunter (1992).

2 – Historical dimension of the migratory process

8A migration decision taken at time t will depend on the individual’s situation as of t, but also on his or her past. In other words, what is at stake here is that the personal history must be taken into account in analysing migration. The introduction of the individuals’ characteristics at time t, and particularly of their position in the life cycle, provides an insight into the evolution of their needs and expectations (Molho, 1986) that will help explain their choices of residence as well as their migration history.

9We advance the hypothesis, therefore, that individuals in the earlier stages of their life cycle will reach the decision to migrate in order to satisfy their needs of occupational and social integration. Credentials will play a significant role in the migratory processes of these individuals. In a second phase, after a family has been started and gets larger, the satisfaction of residential needs (which are centred on housing characteristics) becomes the deciding factor in the decision to move. But there are other factors at work since occupational considerations are present throughout working life. Finally, as the time of retirement draws near, the individuals’ needs change again. The end of active life signals the ability to express a preference for either the pleasures of country living or the superior amenities of the city.

10Beyond the stage in their life cycle, the personal history of individuals also influences the choices made at t. For example, previous migration experience plays a role. Individuals with experience in exploring new places, seeking out information, planning their moves and fitting into a new space may be more willing to consider migration as a possibility (Gordon and Molho, 1995). In our analysis, we will try to include variables that are related to the individuals’ previous migration experience.

3 – Accounting for local opportunities

11Obviously the choice made by individuals of a place to establish their residence cannot be dissociated from the opportunities, both occupational and residential, available to them locally (Gordon and Vickerman, 1982). From this point of view, urban, peri-urban and rural areas differ radically. For example, accommodation offered in rural or peri-urban areas consists of individual houses rather than flats in apartment blocks; it is also larger and cheaper than in urban centres, mostly because of the cost of land (INRA and INSEE, 1998). Conversely, the local access to shops, services and equipment is not as good in rural as in urban areas, owing to the dispersion of the population over the territory (Firlej and Hilal, 2000). Similarly, there are fewer opportunities for employment in rural areas, because of low density as well as the high turnover of rural jobs (Blanc et al., 1999).

12It is not enough, however, to take into account just the characteristics of places, as is usual in analyses of migratory flows (Jayet, 1996). Studies of the new urban economy have shown that accessibility of the working place (Goffette-Nagot, 1996) and of sources of supplies, which condition transportation costs, must also be taken into consideration.

13In this article, our purpose is to analyse individual decisions with regard to internal migration from the different types of areas, urban or rural, within a microeconomic framework. Migration choices result from the weighing of needs of goods and services, localised or not, in the face of constraints of prices and access costs. Those needs vary depending on the individual’s own characteristics, and they change in the course of the life cycle. Whereas the young seek access to the labour market first and foremost, residence-related concerns grow in importance with age, and eventually they are the only ones remaining after retirement.

14Housing or job opportunities, moreover, are unevenly distributed in space, and play an important part in distinguishing urban from rural areas. On the one hand, the characteristics of the labour market and the housing stock are drastically different. On the other hand, urban areas provide the gamut of amenities and services that the households need, whereas the natural environment in rural areas exerts an attraction that has tended to become increasingly popular in recent decades. Because the endowment of local goods and their prices vary in space, while the needs and constraints of individuals vary in time, the latter are led to change their place of residence.

1 – Occupational and residential factors jointly motivate the migration of 15-24 year-olds, particularly those living in rural and peri-urban areas

27In the younger group, occupational and residential motivations affect the migration decisions. First of all, family characteristics and their evolution, which are assumed to enter into the definition of residential needs, play an important role here. Individuals who contract unions or give birth to children are more likely to migrate than those whose family make-up has remained unchanged. However, while the probability that young people living in peri-urban and rural communes in 1982 will migrate is particularly high during the early stages of family formation (at the start of the union and at the birth of a first child), for those residing in an urban core area in 1982 it is higher after subsequent births. Such a result may be explained by the supply of housing in the relevant areas. Starting to live together as a couple often implies renting a place, and this may force the residents of rural and peri-urban communes to migrate, because of the scarcity of leases in their areas of origin. In contrast, in the urban cores, the new couple does not need to move to a new area because the rental market is well developed. When the family continues to increase after the first birth, the probability of migration will increase more markedly, because the need for more living space is confronted with the high cost of urban real estate.

28The characteristics of housing in 1982 affect the decisions to migrate of the younger group only to a small extent. Only the young living in urban core areas were motivated to leave if their accommodations were small (one or two rooms). That previous accommodation should have so little influence may seem surprising at first, but it is largely due to the fact that the characteristics listed are usually those of the parents’ home [9] or of the young person when he or she was still living alone. The recorded contrast then cannot lead to the conclusion that there were differences in behaviour.

29It should also be noted that the difficulty of access to services, measured by the distance to a services area [10], increases the probability of migration for young people who in 1982 were living away from such an area and were residing in peri-urban or rural communes. This can be seen as a sign of the part played by residential motivations in the migration of these young people.

30In order to capture the role of occupational status, we can use the individual’s occupational category in 1990 and the educational level reached in 1990 [11]. Two well-known mechanisms are at work here: the young migrate both to get training, and because they have been trained (Détang-Dessendre and Perrier-Cornet, 1996). The influence of the educational level on migration appears to be more significant among the young who were residing in rural areas in 1982 than among others. This is consistent with the fact that rural labour markets offer few highly skilled jobs (Jayet, 2000), and this incites the young who live there to migrate to places where the labour market offers more opportunities, i.e. to urban core areas. The role of the occupational category is also consistent with the results of other studies (INRA and INSEE, 1998): young executives move most, blue collar workers and farmers least, this being particularly true in peri-urban and rural areas. Occupational mobility, which is measured here by a variable called “upward social mobility”, plays no significant role among the young who have entered the labour market only recently.

31The educational level and the type of job are not independent. To control for this link, we have re-estimated the models by introducing an interaction term between the occupational category and the level of education (see Table 4 [12]). The results are particularly meaningful for this age group. Among young people who were living in rural areas in 1982, being an executive in 1990 increases the probability of having migrated very markedly, whether they had a university degree or not. The position determines migration. For the intermediate occupations and for the clerical and industrial workers, there is a two-way gradation. Industrial workers are least mobile, then clerical workers, then the intermediate occupations. And within each category, the more educated tend to migrate more than the less educated. Among young city-dwellers, the two-way gradation applies to the entire population, including the executives. The respective labour market structures mentioned above are undoubtedly the cause of the contrast in behaviour between urban and rural young.

Table 4

Determinants of migration between 1982 and 1990. Interaction terms between occupational categories and educational level in the logit models

Ocupational and educational categories Commune of residence in 1982 Urban core Peri-urban commune Predominantly rural area Age in 1982 Age in 1982 Age in 1982 15-24 25-44 15-24 25-44 15-24 25-44 Farmer 0.79 n.s. – 2.07 n.s. – 1.36 – 0.95 Craftsman, shopkeeper, company manager 0.59 0.35 n.s. n.s. n.s. n.s. Executive with higher education 1.27 0.19 1.77 0.46 1.67 n.s. Executive with no higher education 0.48 0.34 2.17 0.62 1.63 0.65 Intermediate occupation with higher education 0.94 n.s. 0.98 0.56 1.63 0.33 Intermediate occupation with no higher education 0.57 0.25 0.52 n.s. 0.77 0.29 Clerical worker with no primary or technical education n.s. n.s. n.s. n.s. 0.34 n.s. Clerical worker with primary or technical education n.s. 0.22 n.s. n.s. 0.42 n.s. Clerical worker with secondary education or higher 0.59 0.26 n.s. 0.69 0.73 n.s. Industrial worker with no primary or technical education – 0.26 – 0.18 n.s. n.s. – 0.33 – 0.35 Industrial worker with primary or technical education Ref. Ref. Ref. Ref. Ref. Ref. Industrial worker with secondary education or higher 0.78 n.s. n.s. n.s. n.s. n.s. Economically inactive 0.46 0.22 n.s. n.s. 0.27 n.s. n.s. = non significant at 5% level; Ref. = reference category. Interpretation: All other things being equal, among the 15-24 year-olds in 1982, executives with a higher education degree in 1990, of whatever type of area of origin, are more likely to have migrated between 1982 and 1990 than industrial workers with a primary or technical education (the reference category).

Determinants of migration between 1982 and 1990. Interaction terms between occupational categories and educational level in the logit models

32Analysing the impact of distance to a city with a population of more than 200,000 inhabitants shows that on the one hand, the urban young are less likely to leave a large city than a small one, and on the other hand, that the young who reside in peri-urban communes or in a rural area close to a large city are less likely to migrate than others. It would seem that the proximity of a large labour market decreases the probability of migration, with access to institutions of higher education playing probably a complementary role. Finally, the coefficients linked to the various levels of unemployment have the expected sign for the young living in urban cores, who are more likely to leave cities where the rate of unemployment is high than cities with a less stressful labour market. Conversely, among other youth, the coefficients are not significantly different from zero. This result should be compared with those of other studies, particularly those by Descours and Jacquot (1992), that bring out the complex relation between unemployment rates and migration.

33The individuals’ own characteristics may encourage them to migrate, and this is the case for accumulated migration experience. An individual who has already migrated before 1982 is far more likely to change residence between 1982 and 1990. Various hypotheses may be advanced to account for this. According to Becker’s theory (1975), migration is part and parcel of human capital. Individuals who have migrated previously have accumulated knowledge that will enable them to migrate at a lower cost. Gordon and Molho (1995) have emphasised the part played by having roots in a territory in curbing migration. It appears that individuals who have previously migrated are more likely to have fewer ties to where they reside than those who have spent a large portion of their lives there. They will therefore find it easier to migrate.

34Finally, and in conformity with our hypotheses, occupational integration motivates a large part of young people’s migration decisions, and this is demonstrated by the influence of the educational level, of the occupational category, and of the distance to a large urban core on the migration probability. All three characteristics have an impact on the mobility of young people, particularly rural ones. Residential motivations are added on: the family situation and its evolution modify greatly the housing requirements of the young, and this occasions greater migratory streams among the population of the peri-urban communes and the predominantly rural areas than among the population of the urban cores. Thus, the migration decisions of young people can be understood as the joint outcome of occupational concerns and residential motivations.

2 – Occupational motivations give way to residential ones in encouraging migration among the 25-44 year-olds, particularly in urban areas

35The mechanisms described so far for the young with respect to the influence of the start of union and the birth of children, also operate among the adults who were 25 to 44 years old in 1982. Whereas the influence of such family events tends to decrease slightly among the residents of peri-urban communes and predominantly rural areas, the formation of single-parent families (as the result of the birth of a child living with a single parent, of separation, or of the death of a member of the couple) gains in importance and increases their probability of migration very significantly. The same applies to persons who are left alone after a divorce or the death of a spouse; the effect is now observed also among people who came from urban core areas, though to a lesser extent. It may be that such family events, which affect a small portion of the population under study (see Table A2 in the Appendix), cause people to switch housing and return to urban centres.

36The characteristics of their accommodation in 1982 do not seem to play a more significant part among the 25-44 year-olds than in the younger population. However, the fact of owning a home in 1982 causes the probability of migration to decrease very markedly, whatever the nature of the place of residence. Undoubtedly, two kinds of processes are operating here: on the one hand, ownership reinforces the attachment to the place of residence, and on the other hand, aspiration to the ownership of a house, which has often been presented as one key element of contemporary social behaviour (Goffette-Nagot, 1996), contributes to the mobility of tenants who aspire to ownership. It should be noted that ownership as a curb on mobility is observed less among residents of urban cores than among residents of rural and peri-urban communes.

37The distance to a service centre plays no part at all among residents of peri-urban communes, and only a small part among those of rural communes. Only those rural people who live over 6 km away from a service centre have an incentive to migrate, probably so as to get near the relevant facilities. Conversely, our results show an unexpected effect for those who lived in urban cores as of 1982: the probability of migration decreases at the same time as the distance from a service centre. There are few communes that belong to an urban core and are not service centres, and they are typically residential suburbs. We assume that this result reveals not so much the role of distance to the facilities as the assets of these particular communities: detached houses, low density, environment, etc. Unfortunately there is no way to test this hypothesis.

38Occupational characteristics play a definitely less important role in the migration of adults aged 25 to 44 than in those of younger people. In rural areas, the lesser mobility of farmers and industrial workers and the higher mobility of executives and the members of intermediate occupations are observed again, though to a lesser degree. The influence of the occupational category is very low among individuals who were living in peri-urban communes in 1982 (only executives are a little more mobile), and the behaviour of executives, intermediate occupations and clerical workers living in urban cores is very uniform. Moreover, the explanatory power of the educational level is definitely lower, and testing for the interaction of education and occupational category (Table 4) does not bring out the two-way gradation that was observed in the younger age group.

39Including the “rise in social status” of individuals in the model, as captured by comparing the socio-economic category in 1982 and 1990, shows that upward mobility has a positive effect, whatever the area of origin. It appears that the major portion of migration for occupational reasons took place during the first part of working life, but that a drastic change in socio-economic status may still lead to later migration.

40As was the case among the younger people, the individuals who were aged 25 to 44 in 1982 are less likely to leave large than small cities. Conversely, the proximity of a large city does not decrease mobility from peri-urban and rural areas.

41As to the level of unemployment, it appears to play a role that is more often significant among the 25-44 year-olds than among the younger cohort, but this effect is limited in scope. Whatever the type of commune of residence, high unemployment in the employment area will cause the 25-44 year-olds to migrate, whereas among the young, this was mostly true for the residents of urban cores. Could that mean that the older individuals have a greater awareness of the local economic situation?

42Just as among the younger cohort, the experience of a move between 1968 and 1982 increases very markedly the probability of migrating between 1982 and 1990. Moreover, those persons who in 1982 were living in another département than the one where they were born are more likely to migrate than others. Could this mean that this is the beginning of “going back home” migration?

43Finally, changes in family structure play a strong role in the migration decisions of individuals when the latter are in a phase of full activity. This does not mean, however, that occupational concerns have stopped influencing migrations. They have less influence, and are expressed in different ways at adult ages.

3 – Migrations among the 45-64 year-olds: the weight of retirement and of changes in family structure

44The effect of socio-economic status on migration probabilities that appeared attenuated among persons aged 25 to 44 in 1982, is marked again among the 45-64 year-olds. As a matter of fact, a number of individuals who were still working in 1982 had retired between 1982 and 1990. We hypothesize that the differences in behaviour of the occupational categories reflect in fact a range of responses to retirement. Executives, members of the intermediate occupations, and craftsmen, shopkeepers and company managers are more inclined to migrate than others, and we believe this is more an effect of household income than of the previous occupational status.

45It should also be noted that retirement before 1982 (which involved mostly the oldest individuals in our sample) was an incentive to migrate between 1982 and 1990 only for residents of peri-urban communes. Our hypothesis here is that this effect reflects the changing needs for services, which result concretely in the return to urban areas of elderly persons who had settled in peri-urban communes in a previous stage of their life cycle. Conversely, complementary estimates based on their occupational category in 1990 (rather than in 1982) show that those who were retired at that time were most likely to have migrated if they had been living in an urban area in 1982. This can be interpreted as a sign of people “deserting” the city when they retire.

46With respect to changes in family structure, couple formation (a marginal event that affects only one per cent of the population at those ages) as well as dissolution (by separation or death) always increases the probability of migration. This latter effect is more marked for persons living in peri-urban or rural communes. We infer in this instance that moves are mostly urban-bound and are motivated by reduced incomes after the couple has split up or one of the partners has died; as a result, the individual seeks to move closer to services and/or to the company of family or friends. Conversely, the departure of children from their parents’ home does not seem to stimulate migration among peri-urban or rural residents, but it has a significant and positive impact on the probability of migration of the residents of urban core areas. Is this a matter of moving to homes that are better suited to the living space needs of a smaller family? Do people move to more pleasant areas, where what used to be their second home is located? Or do they return to the place where they grew up?

47The inference that the latter type of migration exists seems to be supported by the fact that probability of migration is higher among persons who were not residing in the département where they were born, whatever the category of the commune where they were living in 1982. Moreover, as for the younger age groups, a previous experience of migration reinforces the tendency to move, whatever the place of residence in 1982.

48Last, we note that living away from a service area, for persons who were residing in predominantly rural areas in 1982, increases the probability of migration significantly. This confirms the importance of the proximity of health and social services in particular.

49On the whole, the nature of the determinants of migratory behaviour among the 45 to 64 year-olds has changed in comparison with those of younger cohorts. Fairly clear differences emerge, depending on whether the individuals initially resided in urban or non-urban areas. For example, the migratory behaviour of the residents of urban cores is more sensitive to changes in family structure: their probability of migration increases not only when the couple splits, but also when the children leave the home. Retirement between 1982 and 1990 also makes departures from urban core areas more likely, particularly among the more privileged social categories. Moreover, it seems that a “fourth age” effect has emerged among persons residing in peri-urban communes, with the oldest old having a higher probability of leaving that type of area.