Who are the people in your (Auckland) neighbourhood?

Auckland is one of the most diverse cities in the world, with more than 200 ethnicities represented and 160 languages spoken. However, those numbers represent diversity at the city-wide level. If ethnic groups are segregated into different ethnic enclaves, then the impression given by the overall numbers overestimate how much inter-ethnic interaction there is.

In a new working paper, my PhD student Mohana Mondal and I, along with Jacques Poot (University of Waikato and Vrije University Amsterdam) measure the cultural (ethnic) and socio-economic (age, income, education, occupation) residential segregation for Auckland city, using data from the Census for 1991 to 2013. Specifically, we wanted to see whether people were sorted more by their cultural characteristics or their economic characteristics.

We used an entropy-based measure of residential sorting, similar to the Theil Index (those interested in the technical detail can read about it in the paper, or in another forthcoming working paper, where we show that this measure is less biased by the size of the groups used to calculate it - more on that in a future post). The level of geography that we are working with is the area unit (approximately the size of a suburb).

The novelty of our paper is that we work with a more disaggregated level of ethnicity data than previous studies. [*] Previous studies have been limited to essentially four ethnic groups (New Zealand European, Māori, Pacific, Asian), whereas in our analysis we look at Samoan, Chinese, Indian, etc. This disaggregation of ethnic groups allows for a finer analysis of ethnic residential sorting than previous studies. [**] We then use similar numbers of groups by age, income, education and occupation in the calculation of residential sorting for those variables.

Overall, we show that:

...residential sorting is greater by cultural than by economic variables. At the area unit level, there is considerable spatial difference in ethnic diversity, but not so much in terms of economic characteristics.

In other words, Auckland suburbs are much more diverse in terms of age, income, education, and occupation, than they are by ethnicity. In other words, people of the same ethnic group tend to cluster together into the same suburb(s) much more than people of the same age, income, education, or occupation.

Looking at specific ethnic groups, we find that:

...the New Zealand European and the New Zealand Māori ethnic groups are consistently the least residentially sorted. At the other extreme, the African, Hispanic and Tokelauan ethnic groups are the most residentially sorted. We also observe growing residential sorting of the populations of Chinese and Indian ethnicity.

It's probably not surprising that small ethnic groups are more likely to cluster together, while the two largest ethnic groups are more spread out. When we look at ethnic sub-groups within the larger ethnic groups, we find that:

...Level 2 ethnic groups are increasingly sorting away from other Level 2 groups within the same Level 1 broad ethnic group. For instance, there are fewer suburbs that are generic Pacific Island communities, with Samoan, Tongan and other Pacific groups increasingly located separately from each other.

Again, this is not surprising. When an ethnic group is small, its members tend to co-locate with other similar ethnic groups. But when a groups gets large enough, it may be better for its members to find their own location within the city.

This is the first paper published from Mohana's PhD thesis. Ultimately, she is looking at using a prototype spatial microsimulation model to project Auckland's future ethnic diversity. I look forward to blogging on that in the future.

Read more:

• How segregated is Auckland, and New Zealand?

*****

[*] We use Level 2 of Statistics New Zealand's standard classification of ethnicities. Previous studies use Level 1 of the classification, or a limited number of the ethnic groups from Level 1.

[**] However, working with Level 2 ethnicity data comes at a cost. It means we look at the area unit (suburb) level, whereas previous studies have used data at the meshblock (neighbourhood) level.