The geography of inequality
A case study on Paris
tl;dr if you’re only interested in figuring out how you rank in terms of income in your neighborhood in Paris and its adjacent cities go directly here.
The source code for everything is available here.
If you want more details, read on!
Inequality is one of the hottest topics these days and one particularly dear to my heart. As a French who lived in the UK for a year and in California for 5, I always find fascinating to try to understand how societies evolve and what values and social contracts they establish.
Income (and wealth) inequality acceptance is obviously very different between France and the USA. France overall is very attached to a notion of social justice, and this was made clear by the recent protests that happened in the last year (the Yellow Vest movement, the protest against the retirement reform). A common feeling in discussions with French people is that for someone to make more, someone else has to be making less, and this leads to a fairly binary opposition of “the rich” vs “the poor” (I tried to debunk these terms in a previous post highlighting that these notions are hard to define).
In the US the perception of wealth is very different. The whole concept of the American Dream encourages everyone to believe they can make it big. Being a successful company founder is seen as creating jobs and driving the economy forward, and overall the idea of trickle-down seems more widely accepted.
However, the thing that struck me the most when I arrived in San Francisco (and I can say that this was also the case for many if not most of the European visitors I discussed with) was the homelessness situation. I found it incredible that you could see tech workers happily scooting to work next to a heroin user screaming in the middle of the streets.
Over time I realized that poverty in San Francisco is hyper localized. From one block to the next you can go from very affluent to dodgy. By contrast Paris seems to have much smoother gradients (and less glaring homelessness issues). So I decided to use data to see what income inequality looks like in my hometown.
The first step was to get access to granular data about income. I found just this on the website of the French national statistics institute (INSEE). This data provides the decile breakdown of declared income by consumption unit for every IRIS (a geographic unit created by the institute) as of 2015. It is probably interesting at this point to clarify what the consumption unit concept is. As defined on their website:
To compare the standards of living of households of different sizes or compositions, we use a measurement of income corrected by the consumption unit (CU) using an equivalence scale.
The most widely used scale at present (known as the OECD scale) uses the following weighting :
1 CU for the first adult in the household ;
0.5 CU for the other persons aged 14 years or older ;
0.3 CU for the children under 14 years.
After a bit of digging I then found the geographic definition of the IRIS used in this data on the national geographic institute website (IGN).
Armed with those two elements I was able to start plotting what each decile of income looked like for Paris. So let’s start our dive into the data with the plot of the median income by consumption unit for every IRIS area in Paris.
This picture will probably not surprise many Parisians as it is well-known that more affluent neighborhoods tend to be located in the West of the city and the poorer areas can be found around the 18th, 19th and 20th districts. Interestingly we can see that the highest median income (Euro 66,116 per CU) can be found in the area just next to the Eiffel tower which is probably justified by the view you get if you live there!
Interestingly this seemed to confirm my initial thought that Paris is indeed more homogeneous in terms of income. Of course there are poorer neighborhoods but the gradient is not so sharp. For instance the ratio between the highest and the lowest IRIS area is 7x in terms of median income.
Now the second thing that I always like looking at is the extreme values. It is where inequality is the most obvious. The most “extreme” data point we have for every IRIS is the 9th decile of income. This means the income you need per CU to be in the top 10% of earners in a given IRIS area.
And as expected, this does not disappoint. When we still had a fairly homogeneous map for the median income, here we start seeing that the very high earners are localized in specific areas. Once again, the Eiffel Tower attracts top earners and you will need to make Euro 287,276 per CU to be in the top 10% earners of the most affluent area in Paris, North of the Eiffel Tower.
It is very clear that the 7th, 8th, 16th, and the 17th to a lesser extent are where top earners live in Paris, whereas you will “only” need to make Euro 23,000 per CU to be in the top 10% of the northernmost areas of the 18th.
On this graph, the ratio between the highest and the lowest IRIS area jumps to 12x.
Finally, I looked not only at Paris but also the neighboring cities (called the Petite Couronne).
Here we see that the differences in income observed within the city actually extend outwards with richer Western cities and poorer areas in the North of Paris (in particular Seine Saint Denis).
It is particularly interesting to me to see the actual economic status of areas of Paris as it highlights something that might be a counter-intuitive to Americans used to inner cities issues. Paris’ city center is concentrating high-income earners and the poorer areas tend to be pushed to the outskirts of the city. There are of course very affluent suburbs but overall inner Paris has a high income level compared to its urban area.
Now for the data geeks reading this and wondering “wait, can I explore this data?”, the answer is yes! I created an interactive map of this data and it is available here!
Using the layer selector one can choose the decile of interest. Basically Dn shows the required income per CU to belong to the top (100 -10*n)% earners of the IRIS area. So D8 corresponds to the required income per CU to belong to the top 20% earners of a given area, D3 the income to belong to the top 70% etc.
Please feel free to share interesting insights you discovered or suggestions to improve or go further in the comments!