The prime minister promised to supercharge Britain by reducing regional inequalities. Two years on, an exclusive analysis by Bloomberg News shows things are going backwards.

In 2019, Boris Johnson led the Conservative Party to a resounding general election win, pledging to revive large parts of the UK left behind during the era of globalization that made London one of the world’s richest cities.

Johnson’s rise was driven by his successful campaign to pull Britain out of the European Union. The so-called “levelling up” agenda was designed to turn that into tangible benefits by 2030, especially for the working class Brexit voters who abandoned the opposition Labour party to hand Johnson his party’s biggest majority since the 1980s.

More than two years on, in a period dominated by the coronavirus pandemic, most of the places that lagged behind London and the South East of England when Johnson came to power have seen little sign of better times. In fact, as a new Bloomberg News analysis shows, they’re more likely to be falling further behind.

Note: For information on how we determined an overall category for each constituency, see the methodology section

To understand how levelling up is progressing, we analyzed 12 key socioeconomic metrics across every one of the UK’s 650 parliamentary constituencies to measure whether the gap has changed—one way or another—since 2019.

The data we used are based on priorities outlined in the government’s official levelling up policy paper and were compiled in consultation with Bloomberg Economics. Where data at the constituency level was unavailable, we used data for higher-level geographies and matched them to the relevant parliamentary seats.

Our analysis shows that the salary gap is widening in nine out of 10 constituencies, that home affordability is getting worse nearly everywhere, and that public spending per head has fallen behind the capital in every region of England.

Note: For information on how we calculated constituencies’ performance on individual metrics, see the detailed example here

Only on a few metrics has the gap narrowed for much of the UK—including life expectancy and the share of people receiving Universal Credit benefits—and in both those cases it’s because the situation in London and the South East has worsened. As a result of Covid-19 the death rate is up and more people are claiming welfare benefits. This is not the kind of levelling up Boris Johnson was looking for.

“These findings highlight the vital importance and urgency of levelling up across the country as we recover from the unprecedented Covid-19 pandemic,” the UK’s Department for Levelling Up, Housing and Communities said in an e-mailed statement. “Our Levelling Up white paper sets out a long-term plan for spreading opportunity and reversing inequality by working across central and local government and the public and private sector.”

To see how each political constituency performed on all 12 metrics and its overall category designation, you can search in the table below. Details on how we determined each constituency’s overall category can be found in the methodology at the bottom of the story.

The results of Bloomberg’s analysis—which show many parts of Britain are slipping further behind relative to 2019—are a blow to Johnson, but they are also a setback for global efforts to hit on a formula that can rein in inequality.

Tension lingers in the US after the storming of the US Capitol in January 2021; extremists made major gains in last month’s French presidential election; the cost-of-living crisis is helping to fuel resentment in large parts of the UK. That’s led major investors to worry that the uneven distribution of the fruits of capitalism is starting to fray the social fabric on which so much of its success depends.

The UK is a test case, with greater regional inequality than almost any other developed nation, according to research by the Bank of England and the University of Sheffield, and a prime minister who has staked his reputation on reducing it.

As Johnson’s flagship policy develops, Bloomberg News will be using our Levelling Up Scorecard as part of an ongoing project to track its progress.

The premier himself is sober about the scale of the challenge. His government’s plan—devised by former Bank of England chief economist Andy Haldane and Levelling Up Minister Michael Gove—is to move more money and decision-making power away from London, while holding itself to account through a series of levelling up targets to be achieved by 2030. The full legislation for the plan was published Wednesday.

Still, there’s already some skepticism. Johnson’s government has been criticized for not matching EU regional funding lost since Brexit. As Bloomberg’s analysis shows, there’s also been slow progress on initiatives like spreading civil service jobs around the UK and increasing government spending on public services and transport infrastructure.

“We’ve had three years of broken promises and empty slogans,” said Lisa Nandy, the opposition Labour Party’s shadow secretary for levelling up, in a statement. “Levelling up might not happen overnight but under the Conservatives it is going backwards. For all the big promises, we’re seeing yet more managed decline for places outside London and the South East.”

Note: Net metrics is the difference between how many metrics a constituency was ahead on versus behind on in 2019; if it was ahead on 3 but behind on 9, that’s a net metric score of -6. Constituencies split evenly (net score of 0) were assigned to overall “Behind” or “Ahead” in 2019 based on which category has a plurality of 6 or more metrics. If there’s no such plurality, then constituencies were assigned to overall “Behind” or “Ahead” based on the median 2019 gap relative to London and the South East across all 12 metrics.

The UK Treasury has been reluctant to dedicate large new pots of money to the levelling-up cause, citing the need to repair the public finances post-Covid. The roughly 12 billion pounds of funding announced so far amounts to about 3% of total government departmental spending in the 2019 fiscal year. Haldane argues this shouldn’t be a particularly limiting factor because tilting more of existing government budgets away from London and the South East will spur levelling up alone.

And the public money that has been dedicated to levelling up hasn’t always gone to the areas that need it most. Of the 100 most deprived areas in England, only 38% of councils won at least some of the Levelling Up Fund money they requested, 34% didn’t participate at all and 28% had all their bids rejected, BBC Panorama reported this week.

“There’s no evidence yet of that incredibly serious, complete re-orientation,” said Eleanor Shearer, a researcher at the Institute for Government, a think-tank. “We haven’t yet seen the solutions in a lot of detail.”

Watch the Quicktake video:

Five Reasons Why Levelling Up Has Stalled →

To be sure, Johnson’s administration has been buffeted by the pandemic, which aggravated deprivation in the poorest parts of the UK, consumed much of the government’s energy and shattered the public finances—as well as nearly killing the prime minister. He’s also been blown off course by threats to his leadership over law-breaking during lockdown, Russia’s war in Ukraine and soaring inflation.

“Delivery is looking really difficult right now,” said Adam Hawksbee, deputy director of the Onward think tank, which conducts focus groups and publishes research on the levelling up agenda. “There have been a number of major distractions.”’

Johnson is also trying to deliver levelling up against economic headwinds caused by a global surge in inflation as well as by Brexit, the cause that propelled him to 10 Downing Street. Britain’s split with the EU has imposed new costs on traders and sparked a steep decline in the number of companies exporting to the UK’s biggest trading partner. The independent Office for Budget Responsibility forecasts a 4% reduction in the size of the British economy due to Brexit over the next 15 years.

“Brexit was a fundamental shock to the regional trade system,” said Philip McCann, chair of urban and regional economics at Sheffield University Management School and author of multiple books and papers on the economic geography of the UK. “The implementation of the new UK-EU agreement marks a crucial break point and emphasizes the importance of scaling up the whole levelling up agenda so that the economically weaker regions are not left even further behind.”

And even then, reversing decades of economic divergence is complicated and will require coordinated action on multiple fronts, Hawksbee said. It won’t be enough to only create better transport links or increase skills or improve the local council in a deprived area—all of those would need to happen at once to see meaningful progress, he said.

“The problem historically is that you’ve had short-term, disjointed interventions by one or two departments as opposed to by the whole of government,” he said. “Now is the moment the urgency needs to be stepped up.”

How We Measure Levelling Up

For each of our 12 metrics, we calculated the gap between each constituency (or available higher-level geography) and a population-weighted average of Greater London and the South East of England in 2019 and in the latest period with available data. We then assessed how each constituency’s performance against those baseline averages changed between the two periods. Constituencies were grouped into four categories based on their relative standing in 2019 and how that’s changed since. Here is an example using the salaries metric:

The data presented here has been structured to allow us to assess how each parliamentary constituency is faring. Where possible, we’re using datasets that directly report at the constituency level. But where such granularity wasn’t available, or in order to get updated figures past 2019, we used higher geographic levels such as local authorities or regions and assigned the data to all related constituencies. Bloomberg’s inclusion of aggregated data in this way helps us paint the most complete picture of levelling up across the widest possible set of socioeconomic indicators.

Several of the levelling up priority areas laid out by the government proved difficult to consistently measure across all parts of the United Kingdom, most notably education, which involves different testing standards in several of the devolved nations, and where data collection has been badly impacted by the Covid pandemic. In other cases, we used alternative metrics to those preferred by the government if they provided greater geographic granularity or were more recently updated.

While the overall analysis provides a valuable assessment of how different parts of the UK are performing broadly in Bloomberg’s analysis, digging into each of the individual socioeconomic metrics reveals important insights, considerations and data limitations. General details about population calculations and how higher-level data are assigned to individual political constituencies can be found in the main methodology section at the bottom of the story.

The overall category assigned to each constituency in the top two graphics is based on the underlying 12 metrics. Where a constituency has a plurality (or outright majority) of 6 or more metrics in a single category, say “Behind in 2019 but levelling up”, that’s what is used for the overall category. This covers about 44% of seats. For the remaining constituencies, we took the 2019 relative gap with London and the South East for all 12 indicators and used the median to determine whether the area was “Behind in 2019” or “Ahead in 2019”. For all constituencies, the median of the 12 metrics’ change in the relative gap since 2019 is used to determine whether the given area was “levelling up” (or “gaining”, if ahead in 2019) or “falling behind”.

The searchable table below details which method was used for each constituency:

For metrics where data is reported at local authority or higher geographic levels, we assigned values down to political constituencies using Office for National Statistics Geography correspondence tables and a two-step analysis of boundary overlaps:

We matched constituencies with the geographic unit within which a majority of the population or—where recent population estimates were not available–most of the land area falls.

For higher-level geographic units that were left unmatched, we determined whether they’re fully contained within a constituency or which constituency the majority of its population—or land area—falls, thus creating geography “aggregations” for a number of seats.

For constituencies that were matched to a single higher-level geographic unit, we assigned them that one unit’s data. For constituencies where a geography “aggregation” was used, we assigned them the average value of their matched geographic units’ data.

The ONS Geography data we used was as of end of 2019 or equivalent for the reference period and as of end of 2021 or equivalent for the latest period with data. Geographic units might have been created or abolished between periods. However the source data is not always capturing those changes. When a geographic unit of a particular vintage appeared in the source data when it shouldn’t—for example, if a local authority created in 2020 doesn’tn’t appear in 2021 data, but its predecessor does—we matched constituencies with that unit’s predecessor(s) or successor(s).

For metrics that are reported in per capita or similar terms (such as per 100,000 people), the 2019 reference values were calculated using mid-2019 population estimates from the Office for National Statistics except for NUTS3 units, where populations as of January 2019 were used. The latest period data were calculated using mid-2020 population estimates, which are the latest available as of publication, except for NUTS3 units, where populations as of January 2020 were used.

In the metric sections, the qualifiers “a little” and “a lot” related to how much a given area is falling behind or levelling up or gaining is based on a normalized measure of variance, which takes the change in the gap with London and the South East compared to 2019 and divides that by the average change value for the relevant metric. This measure is then adjusted so when it’s greater than zero it means the area has improved relative to London and the South East while negative means it’s gotten worse. “A little” refers to any value between -0.75 and 0.75, which represent roughly half of the positive and negative distributions, respectively, with “a lot” referring to any values less than -0.75 or greater than 0.75.

With assistance from Dan Hanson, Andrew Atkinson, Sam Dodge, Alex McIntryre, Mathieu Benhamou and Cedric Sam

Edited by Ben Sills, Adam Blenford and Jeremy Scott Diamond