As migration flows to developed countries have increased since the mid-1980s, so has the diversity of origins of new migrants. With this, it is likely that linguistic and cultural differences play an increasing role in migration decisions. Differences in language may create barriers that prevent the full realisation of the potential economic gains from international mobility as people choose to move to more culturally similar countries.
An extensive literature shows that both fluency in the destination language and the ability to learn it quickly are key to the successful transfer of existing human capital to the destination countries’ labour markets (e.g. Bleakley and Chin 2004, Dustmann and Fabbri 2003).
Although the role of language in determining the direction of international migration is clearly important, previous evidence has mostly been limited to including a control for sharing a common language (see Chiswick and Miller 2014 for a review). Only two recent studies employed more sophisticated linguistic measures to show that cultural barriers explain patterns of migration flows better than traditional economic variables, and then only in samples limited to developed countries (Belot and Hatton 2012, Belot and Ederveen 2012).
Our work is the first that disentangles this relationship from multiple angles by studying the role of linguistic proximity, widely spoken languages, linguistic communities and language-based policy requirements at a destination (Adsera and Pytlikova 2015).
The role of linguistic distance in migration
We study the role of language in determining international migration flows between 30 OECD countries and all world countries for the period from 1980 to 2010. We use information on each country’s official and major languages and their position within the linguistic tree of a language encyclopaedia called Ethnologue to construct a new set of refined indicators of the linguistic proximity between each pair of countries – one for the distance between first official languages; a second for the major language; and a third for any of the multiple official languages and two major languages in each country.
Each linguistic proximity index ranges from 0 to 1 depending on how many levels of the linguistic family tree the languages of both the destination and the source country share. They provide better adjusted and smoother indicators of proximity than the standard dummy for common language used in most of the literature.
Figure 1 presents detailed information about the distribution of migration flows for different levels of linguistic proximity between origin and destination countries. When we look at the distance between first official languages, the majority of migrant flows occur between linguistically distant countries. From the approximately 110 million people migrating to OECD countries during 1980-2010 in our data, about 14.6 million people migrated to countries that share the same first official language and about 40 million migrated to countries with a first official language that did not have any level of the linguistic tree in common with that of their country of origin.
Figure 1. Migration flows to OECD destinations (1980-2010) by the linguistic proximity between source and destination language
Source: Data from unbalanced panel of 30 OECD nations from 223 source countries for the period 1980-2010 in Adserà and Pytlikova (2015). Higher levels indicate greater linguistic proximity between either the first official language in both destination and origin or between the closest among any multiple official or two major languages in each countries.
Not surprisingly, flows are somewhat larger at high levels of linguistic proximity when we use instead the minimum distance among all multiple official and the two most widely spoken languages in each country rather than just the first official, since the language of the former colonial power is often one of many official languages in former colonies.
We find that migration rates are higher between countries whose languages are more similar. Migration flows to a country with the same first official language as that in the origin country are around 20% higher than those to a destination with the most distant language, even after taking account of differences in other socio-economic conditions between origin and destination countries.
For example, migration rates to France from Benin (where French is the first official language) should be around 18% higher than those from Zambia (whose language shares only one level of the linguistic tree with French) but only 6% higher than those from São Tomé and Príncipe (whose language shares up to four levels with French).
This finding is robust to the inclusion of a number of variables that capture cultural, historical and trade ties between countries, such as genetic distance, dummies for common historical past and common borders, distance, and bilateral trade ties.
In the context of other traditional determinants of migration, our study finds that the impact of linguistic proximity on migration flows is lower than that of ethnic networks or income per capita in the destination country, but much stronger than that of differences in unemployment rates.
The results are highly robust to the use of two alternative continuous measures of proximity developed by linguists: the Levenshtein distance, which measures the phonetic dissimilarity between two languages of a core set of 40 common words and is available for all languages; and the Dyen index among Indo-European languages, which is also based on similarities between samples of words.
Migration rates to countries with similar languages are 19-35% higher than those with no linguistic connection. This holds even when the first official languages are replaced in analyses with the proximity between the most commonly used language in each country (which sometimes is not an official language), or with the minimum distance between any of the official and major languages in both countries.
Furthermore, we find that source countries with higher tertiary enrolment rates have larger migration outflows. This accords with the prediction of the human capital investment model that more-educated individuals are more mobile. In addition, the relevance of linguistic proximity in explaining the direction of migration flows is greater for origin countries with more educated workers, probably because of the greater need for skill transferability in the destination labour market.
Widely spoken languages as an additional pull factor in migration
A few languages are widely used across the world. Among them, English is clearly the most popular. A widely spoken destination language can constitute an immigration pull factor on its own.
To investigate the role of English, we estimate separate coefficients of our indicators of linguistic proximity for English and non-English speaking destinations and we show that linguistic proximity is more relevant for explaining migration flows to non-English-speaking destinations than to English-speaking ones. English seems to constitute less of a barrier to migrants than other languages. This may occur for a set of different reasons:
- First, English is widely used in international transactions and media, and it is taught in many countries as a second language. Pre-migration exposure to English by the average migrant probably weakens the linguistic barriers to migrate and lowers the cost associated with transferring his or her skills to the new market.
- Second, English is an asset in the labour market across the world. The hope of improving one’s English proficiency may also increase the appeal of English-speaking destinations, even for temporary migrants who expect to use this skill upon returning home.
We find that migrants are significantly attracted to destinations that already host large communities (likely, linguistic enclaves) with their same linguistic background, where the pressure to learn the local language immediately after arrival is likely to be lower and where they can find psychological support and practical information.1
Our estimates reveal that linguistic proximity between a migrant’s mother tongue and that of the destination country matters less in the presence of a large share of individuals with a first language similar to that of the migrant in the destination country.
But such linguistic or cultural enclaves – think Chinatown or Little Italy – might constitute a mixed blessing for migrants since they may slow down their (and most importantly, their children’s) socio-economic adaptation to their new country of residence.
Language-based immigration policy requirements
The relevance of linguistic proximity in determining the direction and strength of migration flows is also likely mediated by immigration policies that affect the selection of immigrants across destinations. For example, immigration policies in Australia, Canada and New Zealand emphasise candidates’ skills in their application processes for permanent resident visas, awarding points for English language proficiency (and French in Canada), educational attainment and age at migration.
To test whether immigration and naturalisation policies with strict language proficiency requirements may deter migration flows and affect the composition of migrants, we code the existence of both formal (tests) and informal language requirements for naturalisation in 30 OECD destinations for the years 1980-2010.2
Our results show that migration flows to countries with stricter language requirements are smaller. But even when these are taken into account, the linguistic proximity between origin and destination still matters.
Adserà, A, and M Pytliková (2015) “The role of languages in shaping international migration”, Economic Journal, 125(586): F49–F81.
Belot, M, and S Ederveen (2012) “Cultural and institutional barriers in migration between OECD countries”, Journal of Population Economics, 25(3): 1077–1105.
Belot, M, and T J Hatton (2012) “Skill selection and immigration in OECD countries”, Scandinavian Journal of Economics, 114(4): 681–730.
Bleakley, H, and A Chin (2004) “Language skills and earnings: Evidence from childhood immigrants”, Review of Economics and Statistics, 84(2): 481–496.
Chiswick, B R, and P W Miller (2014) “International migration and the economics of language”, in B R Chiswick and P W Miller (eds), Handbook on the Economics of International Migration, Amsterdam, Netherlands: Elsevier, 2014; pp. 211–270.
Dustmann, C, and F Fabbri (2003) “Language proficiency and labour market performance of immigrants in the UK”, Economic Journal, 113(489): 695–717.
1We use two indicators that measure the size of the linguistic networks – the total stock of migrants that share the same level of the linguistic tree (at least either level 3 or level 4).
2 Given the heterogeneity of immigration schemes across countries (skilled and unskilled workers; economic, spouse or student visas, among others), we followed the advice of experts who suggested naturalisation policy requirements would be easier to measure in a homogeneous way than general immigration policy.