The United Kingdom is one of the most geographically unequal nations in Europe and, since the onset of the Great Recession in 2007, spatial inequality has increased.
We might think we know where the rich and poor areas within the UK are. The South of England is rich whereas the North of England is poor and the Midlands are, well, somewhere in the middle.
But, actually, that is not quite correct. The South is not just one prosperous economy and nor is the North just one poor economy. Each region is made up of a mixture of relatively prosperous and relatively poor local economies.
You can see this in Figure 1 below, where I have shown the percentage of local economies that fall into each output per head quintile by NUTS1 region. (In this post, I define a local economy as a local authority district and not as an NUTS3 region for reasons described in the annex and, don’t worry, it does not affect the results.) As before, I will drop the “per head” suffix from here on out – all output figures refer to output per head.
There is a clear regional pattern – richer regions contain more prosperous local economies and vice versa. Unsurprisingly, London and the South East (as the nation’s richest regions) have the highest percentage of local economies in the top quintile (45% each). Around 70% of the UK’s local economies in the top decile are in London and the South East, which suggests that a significant portion of the top 10% of local economies that are pulling away from the rest of the UK are in these two regions (see annex for the equivalent NUTS3 figure). Wales, the nation’s 2nd poorest region, has the highest percentage of local economies in the bottom quintile (60%) and neither Wales, the North East nor Yorkshire and the Humber have any local economies in the top quintile.
However, around 10% of local economies in both London and the South East are in the bottom quintile. London and the South East may be the nation’s richest regions, but they do still contain poor local economies.
In Figure 2 below, I have constructed a map that shows which output decile each local economy falls into. Deeper blues indicate poorer local economies and deeper reds indicate richer ones, with yellow colours showing the local economies in the middle. London and the South East are made up of lots of prosperous local economies whereas Wales and Northern Ireland are full of penurious ones. But, again, these regions are not homogenous in terms of how prosperous the local economies within those regions are. Wales and Northern Ireland may be the 2nd and 3rd poorest regions in the United Kingdom but Belfast and Cardiff are the 52nd and 88th richest local economies in the United Kingdom (out of 391).
In Figure 3 below, I have enlarged London to show that the capital is not a homogenous blob of rich local economies.
Looking visually at the above maps, there also appears to be a relationship between how prosperous a local economy is and how prosperous its neighbours are. Local economies that are relatively prosperous tend to border other prosperous local economies (as in London and the South East) and poorer local economies tend to border other poorer local economies (as in Wales and Northern Ireland). As Tobler did not quite put it, “Everything is related to everything else, but near things appear to be more related than distant things”.
We can test whether there is a relationship between how prosperous a local economy is and how prosperous its neighbours are using a spatial similarity statistic known as Moran’s I measure of global spatial autocorrelation. I have calculated Moran’s I for the UK in Figure 4 below, categorising local economies by output per head percentiles and omitting the UK’s islands (for what I hope is the obvious reason that they don’t have neighbours).
Each dot represents a local economy (local authority district) and the relationship between its own output per head and that of its neighbours. The diagonal line running through the points measures the degree of national spatial similarity – the average relationship between the output of a local economy and the output of its neighbours for the whole of the UK. It is positive, meaning that prosperous local economies tend to have prosperous neighbours and vice versa.
The x-axis shows how prosperous a local economy is and the y-axis shows how prosperous a local economy’s’ neighbours are. A value that lies to the right of the red line on the x-axis indicates that a local economy has greater than average output per head (i.e. it is in the 50th percentile or above) and a value that is above the red line on the y-axis indicates that a local economy’s neighbours have greater than average output per head.
Figure 4: Global Moran’s I Scatterplot of UK Local Economies’ (LA Districts) Output per Head
This national relationship is highly significant and positive but the relationship is far from perfect. Generally speaking, but not always, local economies that produce more than average per person tend to also have neighbours that produce more than average and vice versa.
Now, let us dig down a little bit further. We can see from Figure 4 that each local economy can fall into one of 4 quadrants:
Table 1: Quadrants of Spatial Similarity Statistics for Output Per Head in UK Local Economies
Each local economy falls into one of those four quadrants and has its own local measure of spatial similarity. The local spatial similarity statistic is not, however, always statistically significant. In Figures 5 and 6 below, I have mapped where a local economy’s spatial similarity statistic is significant (i.e. where the p-value <0.05) and also shown which of the above four quadrants it belongs to.
The map shows us both clusters of local economies that are of similar prosperity and islands that are unlike the surrounding areas. In Figure 5 below, we see large areas of prosperity in London, the East and the South East and penurious areas in Wales, Northern Island and the South West. But there are prosperous local economies that are surrounded by poorer neighbours such as Cardiff, Belfast, Glasgow and Edinburgh. It is not a coincidence that these are all major cities (I will return to this point in a later post).
London shows a similar pattern as before, with a swathe of wealthy areas (surrounded by other wealthy areas) in the Centre and West of London, while the rest of London shows no significant association with the surrounding local economies.
There is a relationship between how prosperous a region is and the type of significant local spatial similarity within those regions. As shown in Table 2 below, richer regions tend to have prosperous local economies surrounded by other prosperous local economies (more High-High relationships) and the opposite is true for poorer regions (more Low-Low relationships). I have highlighted cells that have a value greater than or equal to 50% to show the pattern more clearly but, again, the pattern is not uniform and, in poorer regions, there is often a major city that is relatively prosperous and is surrounded by relatively poor neighbours.
Table 2: Locally Significant Spatial Associations by NUTS1 Region
In the next post, I am going to construct a thought experiment (based on the core periphery model) to describe why it could be that more prosperous local economies cluster together and begin testing how well this thought experiment holds up as an explanation for the UK’s spatial inequality. It was originally going to go in this post but, as you can see, this post has gotten rather long as it is.
Annex: Local Economy Definitions – NUTS3 Regions and Local Authority Districts
If you have read the last post, you would have seen that I used NUTS3 regions to examine local economy inequality whereas here I used local authority districts. I have used local authority districts here because it is more useful to use smaller geographic areas when analysing and describing spatial inequality. For example, the NUTS3 region of Essex Thames Gateway contains the Castle Point, Rochford and Basildon local authority districts which are the 5th, 41st and 213th poorest local authority districts respectively. There is a lot of spatial inequality that you would miss if you just looked at NUTS3 regions.
You may reasonably ask why, in that case, I did not examine local authority districts in the last post if I prefer smaller geographic areas. The simple reason is that, because of the way the data is constructed, the time series becomes less accurate the further back you go.
Now, there is a cost to examining relative prosperity at the local authority level. And that is that it can lead to a perceived bias in the output per head statistic. This is because output per head is calculated as the output for a local authority divided by the resident population. As people commute to work, it could be the case that the output per head statistics becomes biased as the denominator is lowered due to the difference between the working and resident population in each local economy.
I don’t really consider this to be a problem for two reasons. Firstly, when you think about whether you live in a rich or poor area, you think about the place where your home is and not where the place you work is. I used to live in Luton and work in Westminster and I did not think I lived in one of the richest parts of the country because I worked in Westminster. Secondly, as people prefer shorter commutes, large differences between the resident and working population show that an area is relatively more prosperous. I am sure many people who work in Westminster would like to live there but cannot due to its prosperity and associated higher living costs.
In any case, I have recalculated all the above analysis using NUTS3 data to show that the relationships displayed above are not being driven by this NUTS3/Local Authority difference in Figures A1 to A4 below and, as you can see, all the main results still hold. Annoyingly, I can only show this for Great Britain due to inconsistent shape data.
In addition, 75% of NUTS3 regions in the top decile are in London and the South East showing that that a significant portion of the top 10% of local economies that are pulling away from the rest are likely to be in these two NUTS1 Regions.
Finally, you might want to know what the difference between NUTS1 and NUTS3 regions actually are, given I have used these terms so liberally in this post, and so I have shown the definitions in the table below (from the Wikipedia page here). On an off topic side note, I adore Wikipedia and think it is an excellent resource and it is why I have used the table below. There is too much Wikipedia related snobbery in this world for my liking.
GVA per Head by Local Authority: https://www.ons.gov.uk/economy/grossvalueaddedgva/datasets/regionalgrossvalueaddedbalancedbylocalauthorityintheuk
GVA per Head by NUTS3: https://www.ons.gov.uk/economy/grossvalueaddedgva/datasets/regionalgvanuts3