Close
Yujung Hwang, Johns Hopkins University and Christopher Udry, Northwestern University
(Project no. 1001968)
Social integration of minorities is very important. High racial segregation in a country, for example ghettoization of new immigrants can breed racial hatred and discrimination. For example, the 19th arrondisseement in Paris has been mentioned as a contributing factor to the November 2015 terrorist attack in Paris, which killed hundreds. Despite its importance, this research topic has been relatively understudied in economics.
Economic theories on social integration have diverged into two clear literatures: those of "neighborhood sorting" and those of "preference evolution". The purpose of this research is to link these two separate literatures and establish a new theory of social integration. I will provide empirical evidence from the experience of South Asian immigrants in the UK, drawn from ONS LS data. Specifically, I propose to understand social integration as a joint process of preference evolution and selective neighborhood sorting. So far, literature on neighborhood segregation has explained it by homophily and preference differences across racial groups but never explained where the preference differences have come from in the first place, (for instance, Bayer et al.(2004)). In addition, to the best of my knowledge, literature on preference evolution did not consider the effect of neighborhood segregation as a reason why cultural differences persist among immigrants (for instance, Bowles(1998), Giavazzi et al. (2014)). I hypothesize that these two processes are closely linked and therefore, they must be understood together in a single framework. For instance, the reason why immigrants rarely change their cultural attitude is because they choose to live in an ethnic enclave and rarely interact with other racial groups.
The experience of South Asian immigrants in the United Kingdom is an ideal case to study a new social integration theory for several reasons. First of all, the UK has collected rich interdisciplinary measures about ethnic minority groups on the issues of discrimination, racial prejudice, and cultural attitude. Such datasets include the Fourth National Survey of Ethnic Minorities (FNSEM), the Understanding Society Panel (UKHLS), the Citizenship Survey, and the Ethnic Minority Psychiatric Illness Rates (EMPIRIC), which can be used jointly with the ONS Longitudinal Study to produce a final research outcome. This contrasts with the U.S., who has a long history of immigration but many interdisciplinary datasets, such as the General Social Survey, do not include enough sample on ethnic minorities. Moreover, the US does not have a large immigration group from a single country (excluding Mexico), which makes it hard to define a narrow homogeneous immigrant group with a distinct culture. On the contrary, the majority of UK immigrants are from only a few South Asian countries (India, Pakistan, Bangladesh), which makes it feasible to analyze a homogenous cultural group with enough sample size. Second, the UK economy experienced an intense structural change in the late 70s and early 80s, which provides a useful source of variation for statistical analysis. Many South Asian immigrants were working in the manufacturing sector but after the industry declined, they had to move to another industry, which led to much internal migration. I am planning to exploit this regional variation in industrial composition to estimate an economic model. Finally, other social science studies have provided enough evidence that social interaction played an important role in terms of cultural attitude change and migration decision. For instance, McGarrigle (2009, p.162p) illustrates through qualitative interview results, the recent desire among South Asian Muslim females to move out from ethnic enclaves to become free from strict "purdah" norms (a social norm among Muslims defining proper gender roles). This evidence supports the hypothesis in this proposal that cultural attitude change would relate to migration decisions.
The ONS Longitudinal Study is critical to carry out my current project because the dataset includes large ethnic minority group sample. For this reason, many ethnographic studies have already used the ONS LS dataset to study ethnic minorities (for instance, Simpson and Finney (2009), Robinson (1990)). Moreover, my preliminary data analysis based on FNSEM and EMPIRIC shows that Indians/Pakistanis/Bangladeshis have culturally assimilated at different speeds, which raises a need to analyse these groups separately. However, this demands a large dataset, such as the ONS LS, to include enough sample members from each country. Finally, the main source of economic variation to be used in the statistical analysis has happened in late 1970s. Comparison between the Census 1971 file and the Census afterward in ONS LS would allow this empirical strategy. Without the ONS LS, this empirical strategy would be impossible because most other datasets (for instance, FNSEM/UKHLS) have started in the 1990s.
To protect confidentiality, I will ask the research team to merge necessary information (local ethnic minority population density, regional racial prejudice index, regional hate crime statistics, various local labor market statistics) based on lower-level geographic identifier variables, and I am going to use only the deidentified data. This will considerably lower the risk of losing confidentiality.
Finally, some of the descriptive statistics that I am going to find may seem redundant compared to previous projects done using ONS LS datasets. However, it is necessary to compute these statistics again with refined definitions to keep consistency with an economic model. Specifically, I will classify sample members based on country of origin, ethnicity report in Census 2011 and immigration year to maintain cultural similarity within each cell and compute migration flows and labor market statistics. I will need a full aggregate statistics table to estimate an economic model in the end. In addition, I may need to perform a few basic regression analysis using micro data.