How Does Geographic Distance Affect the Magnitude of the “Migrant Mortality Advantage”?
Hill Kulu, University of St Andrews
Matthew Wallace, Ined
Rachel Stuchbury, Celsius, UCL
The migrant mortality advantage (empirically observed lowermortality among migrants relative to the host population)1 is one of the most pervasive findings in the social scienceliterature. Despite the regularity with which it is observed, the factorsinfluencing the magnitude of the advantage remain poorly understood. One suchfactor is the distance between origin and host country. In the literature, agreater distance is proposed to increase moving costs, so individuals with moreunobserved human capital, and higher levels of health, should beover-represented among people who move further2,3. Conversley the cost of moving from countries nearby is lower. So,those with less unobserved human capital, and lower levels of health, should beover-represented among people who migrate shorter distances4. In this study we aim to test this assertion. Based upon theliterature, we would expect the size of the migrant mortality advantageto increase as distance between the origin and host country increases.
Data and Methods
We use unlinked, national-level mortality and population countinformation provided by the Office for National Statistics (ONS). Thepopulation count data is from the 2011 Census and is broken down by country ofbirth, sex, and age (starting at age 20-24 up to the open-ended 85+). Themortality data relates to a 3-year period from 2010 to 2012 and is broken downin an identical way to that of the population counts. We combine the data bymultiplying the population counts by three to cover the mortality period. Thenwe fit two Poisson regressions (one for males, one for females) which adjustscountry of birth (with England and Wales-born as the ref) and age (with age 80as the ref). Finally, we take these mortality ratios and present them in ascatter against the distance of the country of origin from England and Wales.We also include the linear regression line and R-value (with associatedP-value) to identify the strength and direction of the linear relationshipbetween the size of migrant mortality advantage and the distance of thecountry of origin from England and Wales. Distance, in this instance, isdefined as straight line distance between London and the other origin countrycapital cities.
From Fig.1, we can see that the R-value (bottom right of each panel)shows a weak-to-moderate negative linear relationship between the size of themigrant mortality advantage and the distance of the country of origin fromEngland and Wales, significant to p<0.05 level. In short, the further thecountry of origin is from England and Wales, the larger the migrantmortality advantage tends to be. There also appears to be some kind ofdistance threshold (at around 1000 miles) in which a mortality disadvantageis possible among individuals who travel the shortest distances. Above thisthreshold a migrant mortality advantage is near-systematic. Fig.1 alsotells us there must be additional factors which influence the size of the migrantmortality advantage, such as ease of access and duration of stay.
Figure1. The relationship between the magnitude of themigrant mortality advantage and distance of the origin country from England andWales.
We plan to move beyond aggregate-level data to an individual-leveldataset (the ONS Longitudinal Study) which links census with life event data (suchas mortality). This will allow us to investigate further the relationshipbetween the migrant mortality advantage and distance from the country oforigin. We will experiment with different distance measures (such as the culturaland socioeconomic distance) to observe how this affects the relationshipbetween the migrant mortality advantage and distance from the country oforigin. Additionally, given that we have linked data at the individual-level wewill be able to take into consideration other important factors such as theduration of stay and age at arrival. Such factors will help provide a morenuanced investigation of the relationship, because they will allow us to calculatemortality among migrants shortly after they have arrived in the host country,when exposure in the host country has had little time to influence their mortalityrisk.
2 van Tubergen, F., Maas, I. &Flap, H. The Economic Incorporation of Immigrants in 18 Western Societies:Origin, Destination, and Community Effects. American Sociological Review69, 704-727, doi:10.1177/000312240406900505 (2004).
4 Jasso, G., Massey, D.S., Rosenzweig, M. R. & Smith, J. P. in Critical perspectives on racialand ethnic differences in health in late life. (eds N.B. Anderson, R.A.Bulatao, & B. Cohen) Ch. 227-266, (National Research Council, 2004).