Abstract
Despite the growing body of public data available on immigrant and migrant populations, quantitative studies on these populations are very rare in France. In particular, although the downgrading and poverty of new immigrants is now a classic research topic, little work is being done on how these new migrants can recover from their economic disadvantage. In this work we propose to assess the impact of two migration policy instruments that could facilitate the integration of immigrants: the recent language training provided under the contract for reception and integration and naturalisation as a step in the integration process. The methodological challenge is significant because naturalisation and language training are not random treatments: naturalised migrants are thus positively selected on the basis of characteristics that cannot be observed in econometrics. We successively use a discontinuity regression, a double difference estimate and a panel estimate to neutralise these selection effects. We find a slight impact of language training on migrants’ French language level, but we do not see any effects on broader outcomes such as the probability of leaving unemployment, income, the number of months worked or the willingness to settle permanently. With a fixed-effect model, we do not find a significant impact of naturalisation.