Dynamic Effects of Co-Ethnic Networks on Immigrants’ Economic Success

Michele Battisti, Giovanni Peri, and Agnese Romiti

The Economic Journal, Volume 132, Issue 641 (2022), Pages 58–88 



This paper estimates the causal effect of co-ethnic networks on the economic success of immigrants in Germany 

The analysis draws on longitudinal data of immigrants in Germany from the IAB-SOEP Migration Sample, a yearly survey of immigrants in Germany beginning in 2013, for individuals who are linked to IEB (Integrierten Erwerbsbiografien), the German social security archive that includes information on immigrants’ labor market history after arrival in Germany.  

To address the likely endogeneity of immigrants’ decisions about where to settle (i.e. immigrants choosing locations based on characteristics that are likely to improve their economic success, and/or because of the presence of co-ethnic networks), the authors: (a) control for a range of location and individual characteristics, including those characteristics affecting immigrants’ initial location choices; and (b) exploit the quasi-experimental variation in initial location for refugees subject to dispersal policies that distribute refugees across locations independently of most of their characteristics. 

Main findings: 

  • Immigrants arriving in districts with larger co-ethnic networks are significantly more likely to find employment within three years of arrival; this advantage fades over time and disappears within four to six years. An increase of one standard deviation (SD) in the initial co-ethnic network leads to an increase in employment of 9.3 percentage points in the first year after migration. This positive effect dissipates within three years of migration. 
  • There is a significant positive correlation between initial network size and the likelihood that the first job in Germany was found through personal contacts. A one SD increase in the co-ethnic network size at arrival corresponds to a 9.2 percentage point greater likelihood of having found a first job through personal contacts.  
  • The initial network size negatively affects the likelihood of being in school/training during the first six years following arrival. Immigrants first arriving in districts with co-ethnic networks that are one SD larger are 3.1 percentage points less likely to be in school/training in their first three years after migration. This negative effect slightly declines (2.4 percentage points) but persists until six years after arrival.  
  • Large co-ethnic network locations reduce time in school/college education rather than time invested in vocational training. The negative effect of the network size at arrival on school/formal education is long lasting; a one-SD increase in network size upon arrival translates into a one-percentage-point reduction in the probability of attending school and college, even in the long run. The effects on vocational training are smaller, and shorter lived. 
  • Larger initial networks tend to be associated with lower current language proficiency, especially for speaking, which is consistent with the idea that co-ethnic networks might reduce opportunities to speak German. The negative effects of network size on language proficiency are attenuated for people who have a better initial knowledge of German and are more severe for individuals that have lower pre-migration proficiency. 
  • Effects are stronger for immigrants with lower levels of education. For lower-educated immigrants, a one SD increase in the network size increases the probability of finding a job through personal contacts by 10 percentage points. For immigrants with tertiary education, the size of the initial network does not seem to affect economic outcomes. The relationship between network size and human capital investment is also stronger (more negative) for individuals with low and medium levels of education. 
  • For refugees and ethnic Germans subject to dispersal policies, effects on employment and human capital investments are similar and sometimes a bit larger. For a one SD increase in the network size, the probability of being employed increases by around 13 percentage points in the first three years of migration. The effect remains positive and significant in the medium term (4–6 years after migration), slightly decreasing to 11 percentage points. There is weak evidence that part of the positive effect may be persistent in the long run for this sample, equal to around six percentage points of employment. This group also exhibit a larger decline in human capital investment in the first three years, corresponding to a 4.3 percentage points reduction in the probability of investing in human capital. After six years this difference has disappeared. 

The results show that immigrants initially located in districts with larger co-ethnic networks are more likely to be employed soon after arrival. However, they are also less likely to invest in human capital, especially in the form of schooling and college education. Consequently, the employment advantage fades after four years, as migrants located in places with smaller co-ethnic networks catch up due to greater human capital investments. These effects appear stronger for lower-skilled immigrants, as well as for refugees and ethnic Germans. This appears to suggest that refugees experience a particularly strong network effect both increasing their employment and decreasing their schooling/training in the three years after arrival. The authors conclude that the benefits of a dense co-ethnic network are short lived, in terms of employment, and an unintended consequence of encouraging settlement in co-ethnic enclaves may be that new immigrants have fewer incentives to obtain more education and training in the long run.