Language and Integration of Immigrants

Aslan Zorlu, University of Amsterdam

This study examines the causal effects of the Dutch language proficiency of immigrants on their labour market and social integration outcomes. Firstly, we estimate determinants of the proficiency of Dutch language using rich surveys. Subsequently, we examine the role of language proficiency in shaping the self-assessed integration outcome and feeling-Dutch in addition to two major structural determinants of integration, employment and income. The analysis shows that endogeneity of language skills masks a substantial part of language effects. Once accounted for endogeneity, effects of the Dutch language proficiency on social and economic integration of immigrants are more than double of the estimates ignoring endogeneity.

Proficiency in local dominant language is an essential first step toward an upward socioeconomic mobility of immigrants in the host countries. Language skills are not only a form of human capital to enhance individual productivity in economic actions (Chiswick and Miller 1995) but also a gateway for immigrants entering the receiving society. Language skills create a foothold for immigrants to explore opportunities and to integrate in the host society.

This study examines the causal effects of the Dutch language proficiency of immigrants on their labour market and social integration outcomes. Comparing two different (linguistic) groups of immigrants, i.e. immigrants from Turkey and Morocco (Mediterranean) and immigrants from Suriname (a former colony of the Netherlands) and Dutch Antilles, the impact of a small language, Dutch, is assessed. Different than big languages such as English, Spanish, Frans and German, Dutch is a small language with a limited geographical coverage. This implies that Mediterranean immigrants should learn Dutch after immigration and their incentives for learning Dutch would not be high, in particular when they intend to leave the Netherlands within a period. In contrast, a vast majority of Surinamese and Antillean immigrants already speaks Dutch due to the colonial history. We explore this contrast to uncover the effects of language , using instruments affecting acquisition of language, such as age at migration, homeland education and homeland media orientation.

Since language skills have been conceptualized as a form of host-country specific human capital in the economic literature. Effects of language acquisition have been extensively studied on labour market performance of immigrants (see Chiswick and Miller 2014 for an overview, Carliner, 1981; McManus et al., 1983). Most empirical studies have predominantly examined the impact of English language proficiency on earnings of male immigrants in traditional immigration countries (Chiswick and Miller, 1995; Dustmann and Fabbri, 2003; Bleakley and Chin, 2004; Miranda and Zhu, 2013a,b). Yet, few studies examines effects of German (Dustmann, 1994; Dustmann and van Soest, 2001, 2002), Hebrew (Chiswick, 1998), Spanish (Budra and Swedberg, 2012; Isphording 2013), Catalan (Di Paolo and Raymond, 2012) and Dutch (Chiswick and Wang 2016; Yao and van Ours 2015). A few recent studies studies examined the effect of second language on social outcome of immigrants, such as health and demographic outcomes, children’s education and residential choice (Bleakley and Chin 2010; Chen 2013; Guven and Islam 2015).

Immigrants’ acquisition of second language is a function of their pre- and post-migration characteristics and in turn, language skills determine their socioeconomic outcomes. There is a broad consensus that language skills and socioeconomic outcomes are mutually determined for immigrants. Language skills are likely correlated with unobserved characteristics of immigrants , such as motivation, innate, financial and cultural resources, as well as some contextual factors in the host country such as, availability and quality of local language courses and density of encounters with native speakers. These unobserved factors are one of main sources of endogeneity. Another source is potential measurement errors in self-assessed language proficiency (Dustmann an van Soest 2002).

We use three commonly used identification strategies: instrumental variable approach to estimate the causal effects of Dutch language proficiency (Chiswick and Miller 1995; Yao and van Ours 2015; Guven and Islam 2015). We basically use three instruments: age at arrival, homeland education and having satellite antenna at home. The first variable is a well-established instrument, which is justified from a linguistic evidence on a critical period of second language acquisition (Bleakley and Chin 2004, 2010; Chiswick and Miller 2008; Yao and van Ours 2015). We prove the role of age at migration in determining Dutch language proficiency and therewith reinforce its validity as an instrument. The other variables, homeland education and having satellite antenna at home, have not been used as instruments for language acquisition before, to the best of our knowledge. Homeland education is strongly correlated with post-immigration investment (Van Tubergen and Van de Werfhorst 2007). Language is likely a first part of this investment. The acquisition of host country language is more necessary for higher educated immigrants to be able to perform their original occupation which more likely needs language skills, compared to a low skilled immigrant. Pre-migration education functions as an indicator of efficiency, lower opportunity costs and higher motivation to learn host country language. The variable having satellite antenna is associated with watching TV channels from home land in mother tongue and captures a high intensity of mother tongue at home, possibly at the expense of host country language. This variable largely captures a high degree of home country orientation of immigrants and a little inclination to learn host country language, assuming that media preferences reflect language preferences

Presented in Session 1145: Economics, Human Capital, and Labour Markets