Local Ethnic Composition and Natives’ and Immigrants’ Geographic Mobility in France, 1982–1999
Abstract: This article provides empirical results on patterns of native and immigrant geographic mobility in France. Using longitudinal data, we measure mobility from one French municipality (commune) to another over time and estimate the effect of the initial municipality’s ethnic composition on the probability of moving out. These data allow us to use panel techniques to correct for biases related to selection based on geographic and individual unobservables. Our findings tend to discredit the hypothesis of a “white flight” pattern in residential mobility dynamics in France. Some evidence does show ethnic avoidance mechanisms in natives’ relocating. We also find a strong negative and highly robust effect of co-ethnics’ presence on immigrants’ geographic mobility.
Summary (3 min read)
- Studies of ethnicity have lacked scientific and political legitimacy for decades in France, but French urban sociology is increasingly concerned with this issue.
- During the 2005 riots, black and Arab youth in the French suburbs were primarily depicted in violent images.
- This debate has lacked evidence regarding the extent to which neighborhood ethnic characteristics are driving geographic mobility.
- The authors seek to describe natives’ and immigrants’2 geographic mobility in France, as well as how these groups react to neighborhoods’ ethnic compositions.
- Geographic mobility, white flight, ethnic clustering, ethnic preferences, contextual variables individual and geographic unobserved characteristics on mobility, thus enhancing confidence in their estimates.
- Findings show very little support for “French white flight” in outmigration but some support for avoidance patterns in relocating.
- On the other hand, the authors find the ethnic clustering pattern to be highly robust.
IN SEARCh oF WhItE FLIGht
- Classical sociologists depict geographic mobility as the channel through which ethnic segregation can lose ground (Duncan and Lieberson 1959; Park and Burgess 1921).
- Much work supports the idea that, after Jim Crow and with the upheaval of anti-discrimination laws, segregation has been sustained by whites’ unwillingness to remain in neighborhoods with large and growing ethnic minority populations.
- The dominant discourse about the lower level of ethnic and racial segregation in France compared to the United States builds on a general acceptance that U.S. society generates more social inequality due to lower redistribution (Brandolini and Smeeding 2006).
- It is not strictly correct to speak of white flight when using French data, because some ethnic minority populations are natives and cannot be distinguished from “non-ethnic French” in the census.
- The authors data were extracted from a large French longitudinal database called Echantillon Démographique Permanent (EDP).
- The EDP is constructed through simple individual sampling: it includes individuals born on certain days of the year (4 out of 365 days, around 1 percent of the population) and for whom a census form or civil status certificate issued upon a major demographic event in the individual’s life (e.g., birth, marriage, death, or childbirth) is available.
- Nevertheless, use of the municipality level is relevant to analysis of contextual effects on geographic mobility for several reasons.
- Second, immigrants’ share is positively correlated with commune size: the proportion of immigrants is more than three times higher in communes of more than 10,000 inhabitants compared to communes of fewer than 10,000 inhabitants.
- The most noticeable disparities concern the total proportion of table 1. Description of French Communes in 1990 Unit: Commune Unit: Individual All Sizes Less than 10,000 inh.
Isolating the Causal Effect of Local Ethnic Composition
- A frequent concern regarding the use of aggregated contextual variables is related to nonrandom sorting into geographic units.
- This selection may lead to considerable estimation biases, among which the authors distinguish two sources.
- Some unobserved determinants may still affect mobility.
- The binary outcome Y is equal to one if the individual moved during the period.
- The authors include other covariates to control for observable and unobservable heterogeneity: Z represents local contextual variables; X individual variables; and μ individual, time, and geographic effects.
- There are three reasons why measurement error may be an issue in their study.
- First, the main covariates of interest are proportions of immigrants, and some municipalities may have few immigrants.
- Second, the authors computed these proportions using a one-fourth extract of the censuses, because detailed information on immigrants’ country of origin is not available on an exhaustive basis.
- Finally, French municipalities are rather small (more than 20,000 of the 36,600 municipalities have fewer than 500 inhabitants).
- Small municipalities’ coefficients are likely to be biased downward, but their sign can still be informative.
Autocorrelation within Units
- Estimating effects of aggregate variables on micro-units may lead to severely biased results.
- Moulton (1990) stresses that, when a multi-level analysis is carried out, one must account for the cluster structure of the variancecovariance.
- Omitting the relevant cluster structure will likely lead to downward-biased standard errors for the coefficient relating to contextual covariates: estimates will too often appear significantly different from zero when they are not.
- This issue is dealt with by relaxing the assumption that the error terms of two observations belonging to the same municipality are not correlated.
- Adjusting the variance-covariance matrix to account for this cluster structure is enough to recover unbiased inference.
- A binary logit model shows a significant effect of local proportions of immigrants for both natives and immigrants (see Table 2).
- Undeclared without children Models 1 and 2 introduce only geographic heterogeneity, yet their data lend themselves to the additional control for individual heterogeneity.
Differences across Groups?
- Table 4’s first panel reports interaction effects for the share of immigrants and the share of co-ethnics with individual occupation (managers, blue collar, and other); interactions with age are displayed in the second panel (younger than 55 years and older than 55).
- Each interaction effect is incorporated successively into Model 1 (Table 3).
- They also suggest that the effect of ethnic preferences on the decision to move is activated only when family or labor-related constraints are reduced.
- The authors test for different effects across immigrant groups and they consistently find a nonsignificant white flight and a sizeable ethnic clustering effect in all regressions.
Geographic Definition of Residential Mobility
- In the analyses presented earlier, the effect of municipalities’ ethnic composition is measured only for individual moves between municipalities.
- The second column uses a tobit model in which the first equation models the propensity to move out of the municipality, and the second equation models the distance between the origin and destination municipalities for the individual who moved.
- It remains possible, however, that some native flight dynamics are at play at a smaller contextual scale, namely for moves within communes.
- On the other hand, all models in Table 5 show that the ethnic clustering effect is very robust to changes in the definition of mobility.
- This suggests that immigrants’ within-commune moves tend to be frequent in communes where the share of their coethnics is large, most likely because immigrants relocate even closer to their co-ethnics.
Ethnic Avoidance in Relocation Decisions
- The white flight paradigm supposes that a neighborhood’s ethnic composition acts like a push factor for the white population’s mobility.
- In that case, the authors should observe a very small or even nonsignificant ethnic composition effect on the probability of fleeing, whereas the same effect would be determinant upon relocating.
- To test for a potential relocating effect, the authors ran an aggregate model11 that counts, for each commune, the number of natives and the number of immigrants entering the commune in 1990 and 1999.
- Nonetheless, these findings suffer from their aggregated nature and do not lend themselves to interpretation in terms of individual choice.
DISCuSSIoN AND CoNCLuSIoNS
- This article is one of the first to measure effects of local ethnic composition on native and immigrant geographic mobility in France.
- Anti-immigrant sentiments experienced a steady rise during the period covered by their data.
- These results put into perspective attempts to empirically detect white flight in quantitative data on individuals’ place of residence.
- Glikman and Semyonov’s (2012) comparative study shows, for example, that although preferences for ethnic neighbors explain a considerable share of Asians’ concentration in European cities, perceived discrimination is a more influential factor underlying Muslim and African immigrants’ segregation.
- Segregation of nonEuropean immigrants in public housing has increased to such an extent that only in such neighborhoods can one find substantially more immigrants than natives (Barou 2002).
Did you find this useful? Give us your feedback
Cites background from "Local Ethnic Composition and Native..."
...The outward migration of the U.S. middle class is justifiably termed “white flight” because it departs minority-laden urban cores to settle in white havens (Rathelot and Safi 2014)....
Related Papers (5)
Frequently Asked Questions (2)
Q1. What are the contributions in "Local ethnic composition and natives’ and immigrants’ geographic mobility in france, 1982–1999" ?
This article provides empirical results on patterns of native and immigrant geographic mobility in France.
Q2. What are the future works in "Local ethnic composition and natives’ and immigrants’ geographic mobility in france, 1982–1999" ?
Geographic categorization issues may also be at stake: because this study relies on data at the municipality level, the authors can not dismiss the possibility that some native flight dynamics might be at play at a smaller contextual scale. Their analyses suggest otherwise, however, given that their findings are not sensitive to the definition of residential mobility. Some qualitative studies suggest that subsidized housing agencies practiced ethnoracial profiling of tenants, which may partly explain the increasing pattern of ethnic segregation within public housing ( Tissot 2005 ). Middleclass natives may be able to reject the first housing offer ( partly motivated by a location ’ s ethnic composition ), but immigrant families are more likely to be desperately in need of a place to live, and thus inclined to take the first offer even if it is in the least desirable neighborhood.