Calculating GDP dispersion (regional inequality) within Europe and European nations.
I've written two blog posts about this.
- A blog post about why I think GDP Dispersion is the best single measure of regional inequality.
- A blog post with hot takes on how regional inequality has changed in Europe since 2000.
And also,
This is the result you're probably looking for. It's also on the blog posts above, which explain it.
.
What is GDP dispersion?
GDP dispersion is calculated as defined by Eurostat using the following formula.
Why create a UK* measure?
Regional inequality measures can be distorted if any of the regions used are smaller than functional economic areas, specifically where there is significant commuting between different regions. The methods I have created use NUTS2 regions, since these are as small as possible while almost never being smaller than a functional economic area.
There is a single major exception to this rule. London.
France refuses to split the NUTS1 region of Île-de-France (Greater Paris) into NUTS2 regions. Spain refuses to split the NUTS1 region of Comunidad de Madrid (Greater Madrid) into NUTS2 regions. But the UK does split the NUTS1 region of London (Greater London) into NUTS2 regions. This reduces the comparability of data on GDP dispersion across nations.
To fix this we create the UK* measure. This calculates GDP dispersion using NUTS2 regions for the UK except for London, where we remove all the NUTS2 regions for London and instead use the single NUTS1 region. It is arguable that parts of South-East England and East England should also be added to this zone. I have implemented such corrections and you can see them in the source code. I do not think that their inclusion is merited, and doing so would require a full reconsideration of other geographies within Europe which may show the same problem.
What about the additional measures like Theil Index and Income Dispersion?
I have included formulas to calculate additional measures including Theil Index and Income Dispersion. These have not been checked and are not safe to use. If you can check and fix these that would be great! My understanding of the mathematics of the Theil Index is limited, specifically I am not yet convinced that the index can be usefully compared across countries and within the same countries if internal boundaries are redrawn.
What happens around 2014/2015
Poland redefined its regions in 2014. France and The Netherlands did the same in 2015. Eurostat do not provide back-calculated data to the modern geographies. Therefore at these points the geographies of each country changes and there are discontinueties. There may be other such changes in geography that I have not noticed.
What are the EA12, EU15, EU19, and EU27 measures?
All combined measures exclude Ireland because its GDP figures are so distorted by its status as a tax haven. They all use the UK* correction method. The EA12 grouping is the founding member of the € plus Greece. The EU15, EU19, and EU27 are as defined and reflect various expansions of the EU.
Licence
- The raw data here comes from Eurostat and is available under The Eurostat Data Licence.
- The code used for processing the raw data is shared under The MIT License.
- The processed data is released under the most permissive license possible, and is probably restricted by The Eurostat Data Licence above. This is an open and permissive licence, so you're almost certainly safe to use it for whatever you like.