Family-level determinants for residential mobility

The results come from a series of nested logistic models (Table A.1. in Appendix) of residential relocation within Switzerland with a special focus on family structures and transitions, including interaction terms addressing these effects on mobility separately for natives and immigrants. As seen from Table 2, immigrants and natives differ with respect to their age structure, a variable that strongly conditions family events (Clark 2013). As such, all of the following models control for this variable. In addition, because nativity interacts with family statuses and events, we synthesize the main effects and interaction effects in the form of predicted probabilities in the following figures.

Figure 1: Predicted probabilities of residential mobility by birthplace

Note: Predicted probabilities are computed at the mean values of other covariates, meaning that the probabilities are only generalizable to individuals having those attributes. Predicted probabilities on the left-hand side come from Model 1 of Table A.1. in the appendix, which only includes age and birthplace as covariates. Predicted probabilities on the right-hand side account for demographic, socioeconomic, housing and contextual attributes (Model 7 from Table A.1. in appendix).

Starting with the baseline probability for residential mobility in Figure 1 (on the left-hand side), we can already see some differences across birthplaces. The yearly probability for residential mobility is about 11 per cent for the Swiss-born population, which is one per cent lower compared to the EU group (12%) and one per cent higher than the non-EU group (10%). However, following the inclusion of a set of covariates (in particular the inclusion of the time since immigration in Model 4), the EU group appears less residentially mobile (on the right-hand side). Indeed, a large share of EU immigrants recently arrived on the Swiss territory, a situation that often necessitates housing adjustment. As such, all other things being equal, the most residentially mobile

0,080 0,085 0,090 0,095 0,100 0,105 0,110 0,115 0,120 0,125 0,130

Swiss EU/EFTA Non-EU Swiss EU/EFTA Non-EU

LE PARCOURS DE VIE DES IMMIGRÉS EN SUISSE

122

group is the Swiss-born, followed by the EU and non-EU groups, although the differences are relatively small.

In line with the elevated mobility hypothesis, the models confirm a higher mobility propensity for the Swiss-born population at the time of childbirth: the yearly probability is positive and almost stable across logistic models, with and without control for other predictors of residential mobility (Model 2 to Model 7 in Table A.1.). The transition to parenthood, regardless of birth parity, comes with a 3 per cent change in the probability of relocation in the Swiss-born population (Figure 2). As expected, the results show evidence of an immigrant status effect: childbirth increased the one-year probability of residential mobility of non-EU immigrants by only one per cent and even decreased that probability among the EU immigrants by one per cent.

Figure 2: Change in the probability of residential mobility at the time of childbirth, by birthplace

Note: We used the contrast postestimation command from Stata 14 to jointly test for the effect of childbirth across birthplaces following Model 7. Lecture note: The difference in the probability to change residence between a Swiss having a newborn child (13%) and a Swiss who did not (10%) is 3%. This effect is statistically different from the non-EU group whose change in probability between individuals who have a child (9%) and those who do not (8%) is only 1%.

-0,030 -0,020 -0,010 0,000 0,010 0,020 0,030 0,040

Swiss EU/EFTA Non-EU

FAMILY-LEVEL SOURCES OF INTERNAL MIGRATION FOR NATIVES AND IMMIGRANTS

123

Persisting differences in the synchronicity of childbirth and residential relocation after control for important predictors of residential mobility calls for a closer look at multiplicative (interaction) effects. Doing so, we found specific mobility patterns for immigrants and natives at different levels of income – an important determinant of housing opportunities. Figure 3 presents the change in the probabilities of residential mobility – by birthplace and income level – between individuals having a newborn child and individual who did not experience this transition.

Figure 3: Change in the probability of residential mobility at the time of childbirth, by birthplace and household income

Unsurprisingly, housing adjustment at the time of childbirth is more likely when household economic resources are higher. Although this is true for all groups, this effect is more pronounced for the two immigrant groups (positive interaction effect, model available upon request). For the native-born, the mobility propensity with childbirth remains stable among the low-income category and increases by 4 and 6 per cent, respectively, among medium- and high-income categories. For immigrants, the gap between income categories is greater, especially for EU immigrants. In fact, the immigrant status effect only stands for the low-income category: medium- and high-income groups are very similar in their mobility behaviours. Conversely, all low-high-income groups (immigrants and natives) have the lowest mobility at the time of childbirth, but

-0,100 -0,080 -0,060 -0,040 -0,020 0,000 0,020 0,040 0,060 0,080 0,100 0,120

Low Med High Low Med High Low Med High

Swiss EU/EFTA Non-EU

LE PARCOURS DE VIE DES IMMIGRÉS EN SUISSE

124

with a clear distinction for the EU groups whose propensity for residential mobility reduced by 7 per cent.

The results also confirm the simultaneity in the processes of marriage and residential mobility. When considering all marriages (without consideration for pre-marital cohabitation status), migration propensity steeply rises in the year interval where a transition from unmarried to married takes place. As expected, this effect is stronger among non-EU immigrants, whose probability of residential mobility increases by 16 per cent during marrying intervals (left hand side of Figure 4). In comparison, the probability only increases by about 7 per cent for natives and EU immigrants, supporting the differentiated mobility hypothesis.

Figure 4: Change in the probability of residential mobility at the time of marriage, by cohabitation status and birthplace

However, moves related to marriage events have different implications depending on whether or not couples were already cohabiting before getting married.

Following the inclusion of the cohabitation status in Model 3, we account for specific mobility patterns by origin (interaction between birthplace and marriage) and by cohabitation status (interaction between cohabitation status and marriage).

Unsurprisingly, newlyweds are more mobile if they were not cohabiting before getting married (middle panel of Figure 4), regardless of their origin. In this situation, couples

-0,050 0,000 0,050 0,100 0,150 0,200 0,250 0,300

Swiss EU/EFTA Non-EU Swiss EU/EFTA Non-EU Swiss EU/EFTA Non-EU All marriage Non-cohabiting partners Cohabiting partners

FAMILY-LEVEL SOURCES OF INTERNAL MIGRATION FOR NATIVES AND IMMIGRANTS

125

have a 20% increase in the probability to change residence in the given year. For cohabiting partners (right-hand side of Figure 4), by contrast, we found no evidence of housing adjustment: changes in probability are close to zero among all origin groups.

In sum, the results confirm that the higher mobility for the non-EU group is due to a difference in pre-marital cohabitation behaviours: EU immigrants are less likely to be living with an unmarried partner at the time of marriage and residential relocation is more likely among non-cohabiting partners.

Figure 5: Change in the probability of residential mobility by birthplace and marital status (Ref: singles)

As mentioned above, the impact of divorce on mobility is not as direct as for other family transitions, and may imply sizeable delays. Nevertheless, the effect size of this variable is fairly large: the occurrence of divorce increases the probability of relocation by about 10 per cent for all groups (not shown). As expected, the long-term status implications of divorce for mobility are important, as residential mobility remains elevated for a long period of time after the occurrence of a divorce in all three groups (elevated mobility hypothesis), especially among EU immigrants, supporting the differentiated mobility hypothesis. Compared to their single counterparts, divorced Swiss-native are 3 per cent more likely to change residence each year, whereas EU and non-EU immigrants show respectively about 5 and 4 per cent change in the

-0,04 -0,02 0 0,02 0,04 0,06 0,08 0,1 0,12 0,14

Married Divorced Widowed Married Divorced Widowed Married Divorced Widowed

Swiss EU/EFTA Non-EU

LE PARCOURS DE VIE DES IMMIGRÉS EN SUISSE

126

probability of relocation. By contrast, married Swiss-born displayed a slight negative change in mobility propensity whereas all immigrants maintain the same level of mobility compared to their single counterparts.