Is the UK coast poor?
Using data and maps to answer a difficult simple question.
For months I've been trying to answer that simple question by talking with people, assembling data, and writing algorithms.
The question is beautifully imprecise. Any introduction of precision is controversial. What is a coast? What does poor mean? All answers are contested, even the definition of the UK. But most have value.
So let us make some outrageous definitions, take a broad definition of the coastline, and start exploring.
Since it is a line on a map, a coast cannot be rich or poor. People, households, and communities can be rich or poor, and we can reasonably define them as coastal if they are within a certain distance of the coast. Let's say 10 kilometres.
Now we have to define rich and poor. Does a higher income make you rich? Does more accumulated wealth make you rich? Does a strong local economy make you rich? Does higher life satisfaction, sense of purpose, and happiness make you rich? There is no good answer, so let's say that income is what matters and get started.
A very hard technical question
This is already a very hard technical question to answer and starting with the whole of the UK would be a bad idea. So I started with the island of Ireland. Just the main island, which is called Ireland.
We start with the World Vector Shoreline which is freely available under an open licence (LGPL).
Working just with the island of Ireland we add a 10km buffer to the line of the coast.

Next we download boundary files for the two smallest geographies for which Ireland and Northern Ireland make income data available. These are super output areas for Northern Ireland and electoral districts for Ireland.

We calculate for each super output area of Northern Ireland and each electoral district of Ireland whether it touches the 10km buffer from the coast of the island of Ireland and retain just those.

Next we need to match incomes from separate small area income tables to the coastal geographies and create a single map.

We can also produce summary statistics from the tables behind these maps. For example, we can say that across the whole island of Ireland coastal areas are more less likely to have middling incomes, and more likely to be either very rich (income in the top fifth of their country) or very poor (income in the bottom fifth of their country).

Why?
Returning to the first question, is the UK's coast poor? There are so many possible answers that being able to explore more of them more quickly is valuable. Is 5km from the sea better than 10km as a definition of coastal? How far up a river or a loch or a ship canal is still the coast? Should we exclude cities from our analysis of coastal economies? Can we look at accessibility to jobs and services on the coast? Should we split out income from pensions and from work in Cornwall?
These are all great questions and because all of the maps and graphs I've shown so far are generated programmatically, I can change variables and add new datasets easily. This lets us have a discussion that can be challenged and directed by data much more easily.
And since my code doesn't care about countries we can extend our analysis to more countries even if they have different geographies and different types of data. Which is what I've done for seven different countries - Ireland, Northern Ireland, Wales, Scotland, England, France, and Belgium.

And this finally lets us answer the first deeply flawed question. Is the UK coast poor?

Yes. But only because of England.
But what about?
That's enough maps and graphs for now. There are lots of issues with all of this work. Here's just a few,
- Income for small areas is measured and reported differently in different countries and thus can't be compared. For example Northern Ireland only publishes income deprivation rankings for its SOAs, Ireland only publishes median income, while England and Wales only publish equivalised mean income.
- Small areas vary significantly in size and population, both across countries and within countries.
- Methods for comparing income across countries, especially ones with different governments, tax and benefit systems, price levels, and currencies, is very difficult.
- Methods for comparing income within countries is not much easier.
- Some small areas in France and Belgium have such small populations that income data is not published.
My decision to work with income quintiles reduces these problems, but many still remain. When I published high quality maps clearly explaining the challenges I have found it impossible to stop people from clipping the maps to remove those explanations and adding their own incorrect interpretations to the maps. Examples of this are,
- "Look at how much richer Cardiff is than Bristol" (this is not true - Cardiff is a rich place for Wales and Bristol is a high-middle-income place for England, but England has higher average incomes than Wales. Average incomes in Bristol are higher than in Cardiff).
- Look at how poor France is (this is not true, it's just France's richer places tend to be in and near cities whereas England and Scotland's richer places tend to be suburban or rural. This makes the map of France look redder and the maps of Scotland and England greener).
But there are more.
At this stage the best thing for me to do is share the code that I've written (as a GitHub Gist for now) and the data that it has produced. I have overcome dozens of GIS problems to create the maps and data. I am sure that these methods can help others, but also sure that they can be improved and expanded.