02What is the problem?

There is variation across the country on a range of indicators

Looking at maps of a whole range of indicators shows why there is a political desire to level up the country. Figure 1 picks out three – life expectancy, the share of people with no formal qualifications, and productivity – and shows that there is variation across the country in all three.

Figure 1: The North lags the South on a range of indicators

Source: ONS

For example:

  • Health – the average male resident in Westminster lives 10 years more than someone in Glasgow local authority.
  • Skills – In 2019, 30 per cent of working-age people did not have five good GCSEs or equivalent in Barrow-on-Furness, and 22 per cent had degree. This compares to 9 per cent and 59 per cent respectively in St Albans.
  • Productivity – on average, a worker in Milton Keynes produced in three days what a worker in Blackburn takes five days to produce in 2018.

These are stark differences. And the likely immediate reaction of any policymaker will be a desire to reduce these differences. But they must be circumspect in the ability of policy to be able to do this.

In principle, a policy of levelling up can even out the maps of life expectancy and skills. This is because there is no fundamental reason why someone living in one part of the country is more likely to have no formal qualifications than in another part, for example.2 There is no reason why geography should be a driver of these patterns.

To do so, the Government could set a floor target to pull every local authority above in terms of life expectancy and skills. It could then pull a number of direct levers to achieve this. For example, it could commit more funds to further education in places where the share of people with no formal qualifications is particularly high and it could commit more resources (and/or introduce reform) to NHS services where life expectancy is low, along with more money to fund day-to-day public services to support this.

While we should not accept the current variation in productivity, we should both expect to see, and be comfortable with, differences in it

We should, however, expect to see variation in the productivity map because different locations in the country play different roles in the national economy. This means that looking at a map of productivity is misleading because it does not compare like with like. Instead we should be comparing and contrasting urban with urban and rural with rural, rather than expecting them to all be equal.

The roles different places play results from the inherent benefits (and costs) that they offer to businesses. Where businesses locate depends on their trade-off of costs and benefits that a location offers.3

Broadly, cities offer access – access to workers, access to customers and access to knowledge through the face-to-face interactions that cities and city centres in particular encourage. But costs tend to be higher too, for example through rents, congestion and pollution. More knowledge-based service activities tend to favour these benefits over the costs because the benefits make them more productive. This is why the 62 largest cities and towns in Britain account for 9 per cent of land, but 59 per cent of jobs and 71 per cent of knowledge-based services jobs. In contrast, 49 per cent of jobs in production firms, which prioritise cheaper land and access to road transport, are in these 62 places 4

This pattern results from what is known as agglomeration, which is discussed in more detail in Box 1, and explains why such a disproportionate share of the economy is found in city centres in particular.

Box 1: How agglomeration affects the location of businesses within cities

Agglomeration is the process by which concentrating economic activity in one place increases the productivity of that activity. Benefits are characterised into three types: learning, which reflects the ability to share ideas and information; sharing, the sharing of inputs such as roads and broadband; and matching, the matching of workers to jobs and jobs to workers.5

These benefits of agglomeration play out over very different geographies.

  • The labour pool that businesses have access to stretches well beyond its boundaries. Although this is likely to vary depending on geography, previous research suggests that this effect extends up to a drive time of 80 minutes from a British city, with the effect becoming weaker as distance from a city increases.6
  • The ability to exchange ideas and information, known as ‘knowledge spillovers’ tends to operate over very small geographies. For example, for the advertising industry in Manhattan it has been estimated that these knowledge spillovers operate over a distance of just over 750 metres, while other research finds that these agglomeration effects are strongest over a distance of one mile.7

On the latter, this is why we see much activity – and high-skilled activity in particular – locate within city centres. In 2015, city centres in Britain collectively accounted for 0.1 per cent of all land. But they accounted for 14 per cent of all jobs and 25 per cent of all jobs in more productive services businesses.8

The firms most influenced by agglomeration are ‘exporting’ businesses – those that sell to regional, national and international markets. Because they sell to so many markets, they are more likely to choose their location based on the benefits and costs set out above.

The location of local services businesses, on the other hand (such as hairdressers and restaurants), is instead governed by where their customer base is located. Their location decisions are much less likely to be directly influenced by agglomeration, and more by centres of population.

It is exporter businesses, and high-skilled ones in particular, that are crucial for determining productivity because of their ability to absorb new innovations.9 That means that its ability to attract or grow its exporting base determines the overall productivity performance of a city.

Crucially for levelling up, while the Government can introduce policies to enhance these existing benefits, fundamentally it cannot intervene to make every place have these benefits (beyond building a new large city in Cornwall, say). This limits the power of policy to influence where firms locate, despite politicians frequently claiming the opposite.10

In making the most of these benefits (and offsetting higher costs), the theory of agglomeration says that city-based businesses should be more productive than firms elsewhere. Manchester should be far more productive than Cumbria.

In the UK though this theory only holds in the Greater South East. Cities in this part of the country are more productive than their non-urban neighbours, as Figure 2 shows. But this is not the case everywhere in the country. While Cities and large towns elsewhere are more productive than more remote areas, but not more productive than their immediate neighbours. Surprisingly, these cities (such as Bradford and Newcastle) are less productive on average than even some of the more rural parts of the Greater South East (such as King’s Lynn and West Norfolk and West Suffolk).

Figure 2: Cities elsewhere in Britain lag behind those in the Greater South East

Source: ONS

Note: Four local authorities that border a city have been classed as more rural because of their size. These are County Durham, Northumberland, Scottish Borders and Powys.

Big cities sit at the heart of regional productivity woes

The main cause of this gap in performance is the lagging performance of the biggest cities. Previous Centre for Cities’ research shows that the main difference between urban Britain and cities in France, Germany and the US is that as a group, productivity in larger cities in Britain lags the national average, rather than leading it.11

This is not because smaller cities in Britain are unusually productive. It is because Birmingham, Glasgow and Manchester in particular are considerably less productive than international counterparts such as Brussels, Gothenburg and Munich. For example, in 2011 Manchester was 30 per cent less productive than Marseille, and 63 per cent less productive than Munich.12

These big cities are considerably below the ‘productivity potential’ that their size would suggest

Conservative estimates based on Figure 3 suggest that Manchester is furthest from its productivity potential, represented by the dotted line, of any city in the UK at £15 billion, followed by Birmingham (£11 billion) and Glasgow (£7 billion). For the eight largest cities outside London combined, this gap is £47 billion.13

Figure 3: Big cities punch well below their weight

Source: ONS, Regional gross domestic product (GDP) reference tables; ONS, Census 2011

Note: City size is adjusted to take into account the number of people in commutable distance, rather than just people that live within a city’s boundaries, in accordance with the distance that matching element of agglomeration plays out over (see Box 1). This better captures the benefits that cities offer to businesses. Data is on a log scale.

Underperformance is not just reserved for large cities. Figure 3 shows that there are 25 other cities outside the Greater South East that also underperform. The gap between actual and potential productivity for these places combined is an estimated £19 billion.

There are of course non-urban areas that underperform too. But because of their lower potential to improve productivity (resulting from the more restricted benefits they offer to businesses), while closing this gap could be important locally, it would have a very small impact on the wider regional or national economy. Improving the performance of the 76 lagging non-urban authorities across the country would add an estimated £16.1 billion to the national economy – just slightly more than closing the output gap of Manchester alone.14

Using this breakdown, these conservative estimates suggest that the British economy is £83 billion, or 4 per cent, smaller per year than it should be because of the underperformance of places outside the Greater South East. Big cities are the biggest contributor to this gap. Despite covering 7 per cent of land in underperforming areas in the rest of Britain, they account for 57 per cent of this underperformance (see Figure 4). Box 2 discusses the role of large city centres in this.

Figure 4: Big cities account for most of Britain’s lost output

Source: ONS

Box 2: The underperformance of big city centres

While productivity data does not yet exist at the city centre level, proxy measures of productivity suggest there is a much stronger relationship between city centre size and productivity than the one seen at the city level. Figure 5 shows that the larger the city centre, the larger the share of jobs in that city centre that is high skilled. As agglomeration would predict, this suggests that the ‘knowledge spillover’ benefits in city centres increase with the amount of economic activity within them, and so become more attractive to high-skilled activities. Agglomeration within English and Welsh city centres is plain to see.

Despite this relationship, this chart suggests the underperformance of Birmingham and Manchester city centres in particular. Not only is the gap between central London and the next largest city centres very wide, but Manchester and Birmingham city centres are no larger than those in Bristol and Leeds, despite the city populations of the former being more than three times larger than the latter.

This means that, while Manchester and Birmingham city centres in their own right are very high-skilled economies and have gone through a period of rejuvenation in the last 30 years,15 they currently are too small. The main focus of helping these cities achieve their productivity potential should be to enlarge their successful city centre economies.

Figure 5: Larger city centres have higher shares of high skilled jobs

Source: ONS, Census 2011

Note: Data for Scotland is not available. The number of jobs is on a log scale.

In terms of levelling up productivity, the aim of policy should be to help places to achieve their productivity potential, which will differ from place to place, rather than trying to get every place to achieve the same level of productivity. In other words, the goal should not to make all authorities in the productivity map in Figure 1 the same shade of green. We should expect big productivity differences between Manchester and Cumbria. But we should not expect them between Manchester and Bristol. Currently neither of these things is true.

In particular, policy will not level up if it does not deal with the underperformance of these biggest cities (and in particular their city centres). The much stronger performance of the European counterparts shows that closing this gap is realistic. The size of this gap shows that there is considerable gains to be made. Manchester needs to perform more like Munich than Mansfield. And Birmingham needs to be more like Brussels than Birkenhead.

Towns are unlikely to prosper without prospering neighbouring cities

It is important to underline here that this is not an argument for cities versus towns, as the debate has sometimes been presented. Policy should be making interventions designed to improve productivity in smaller places where their productivity is lagging. There are two points to consider within this.

The first is that previous work by Centre for Cities has shown that the performance of towns is in part dependent on the performance of nearby cities.16 Most towns that perform well – and there are many of them – tend to be located close to a successful city. And strongly performing towns near poorly performing cities are rare. Box 3 discusses this in more detail, while Box 4 shows why we should not expect working patterns post Covid-19 to affect this either.

Box 3: The relationship between cities and towns

Much commentary over recent years has pitted cities against towns, and has claimed that city growth has come at the expense of towns. This is not supported on two counts. First, the notion that cities have grown and are strong performers does not hold in both the data presented in this paper and when looking at data on long-term change.17 Second, research by Centre for Cities suggests that the fortunes of towns are tied to a great extent to the fortunes of their neighbouring cities.

Not all towns have been ‘left behind’, with some performing very well. But there is a geography to this. Towns closer to cities (109 of the 164 towns sized between 30,000 and 135,000) have better employment outcomes for their residents than towns further away. It is rare to find a town far from a city that performs well on this measure. But being close to a city is not sufficient – those towns closer to successful cities, on average, have better employment outcomes than those close to less successful ones. This is not just seen in employment outcomes – they tend to have stronger economies in their own right, with larger shares of high-skilled exporting businesses in their economies.

Given this, pitting towns against cities is not helpful for the towns agenda. It will be hard to improve the performance of towns without improving the performance of the city neighbours.

The second is that the number of people living in each area will determine the size of the impact of any policy intervention. Hartlepool fits into Manchester 27 times over. Manchester’s output gap is many times larger again. This difference in scale means that levelling up – politically sensible as it may be – cannot be about intervening in towns alone if it is to bring noticeable improvements in economic prosperity to many millions of people living across the UK. Once again, policy will not level up if it does not deal with the underperformance of these biggest cities.

Box 4: Will work from home change all of this?

Many column inches have been devoted since the onset of Covid-19 to the idea that remote working will make geography irrelevant, with the argument being that workers will able to work anywhere.

This argument is not new – it was first made at the turn of the century as virtual technologies developed. But since then the opposite has happened, with jobs concentrating in successful city centres in particular.18

This occurred because of the value of face to face interaction and the ‘knowledge spillovers’ that occur from this interaction, as discussed in Box 1. Given that the UK is likely to continue to specialise in knowledge-based activity that benefits most from this interaction, location is likely to continue to be important. The continued clustering of people in East Asian cities in the last 20 years, despite the epidemics that they have been challenged by over the last two decades, further supports this.19

For those jobs that can be done remotely in principle, it may be the case that some days a week are spent out of the office. But the requirement for that face-to-face interaction will mean that city centres will likely have an ever more significant role in the economy in the coming years.


  • 2 Noting that sorting effects, with people self-selecting into certain areas, will have some influence on the maps.
  • 3 Different places offer different benefits to people too. City centres tend to be disproportionately to younger people, while suburbs and areas around cities are more appealing to older people. See Thomas E, Serwicka I and Swinney P (2015), Urban Demographics: Why People Live Where They Do, London: Centre for Cities.
  • 4 While manufacturing is more likely to be based out of town, it is interesting to observe splits in activity in this sector between knowledge and production. For example, while Siemens makes its wind turbine blades in Hull because of the direct access to the North Sea, its UK headquarters is in Manchester. This spilt is seen within service firms too. For example, Barclay’s investment banking functions are in Canary Wharf, but its administration activities are in Liverpool and Sunderland.
  • 5 For example, see Rosenthal S and Strange W (2004), Chapter 49 – Evidence on the Nature and Sources of Agglomeration Economies, Handbook of Regional and Urban Economics, Volume 4, Pages 2119-2171
  • 6 Rice P, Venables AJ and Patacchini E (2006), Spatial Determinants of Productivity: Analysis for the Regions of Great Britain, Regional Science and Urban Economics 36 (6), 727-752. A study on the largest US cities suggests that agglomeration has an effect over a 60 minute drive time, with the majority of the gains concentrated in the first 20 minutes. See Melo P, Graham D, Levinson D and Aarabi S (2015) Agglomeration, accessibility and productivity: Evidence for large metropolitan areas in the US, Urban Studies
  • 7 Arzaghi M & Henderson J (2008) Networking Off Madison Avenue, Review of Economic Studies (October 2008), pp. 1011-1038; Rosenthal S & Strange W (2003) Geography, Industrial Organization, and Agglomeration, Review of Economics and Statistics (May 2003), pp. 377-393
  • 8 Swinney P and Serwicka I (2016), Trading Places: Why firms locate where they do, London: Centre for Cities
  • 9 Swinney P (2018): The wrong tail? London: Centre for Cities
  • 10 For example, Shipman S, ‘Boris Johnson: I’ll stop brain drain to the cities’, The Times 9 May 2021
  • 11 Swinney P and Enenkel K (2020), Big cities and levelling up, London: Centre for Cities
  • 12 See Centre for Cities European Cities Datatool at https://www.centreforcities.org/data-tool/dataset/european-cities#graph=table&indicator=gva-per-worker\\single\\2011&city=birmingham&city=glasgow&city=manchester&city=brussels&city=marseille&city=munich&city=gteborg&sortOrder=high
  • 13 Bristol is not included in this list as it is the only large city that is meeting its productivity potential.
  • 14 This assumes that the productivity potential of a lagging non-urban area is the same as the productivity potential of the smallest city using the methodology in Swinney P and Enenkel K (2020), Big cities and levelling up, London: Centre for Cities. This is a generous benchmark – more rural authorities in particular should have a level of productivity potential below that of the smallest city.
  • 15 Swinney P and Sivaev D (2013) Beyond the high street: Why our city centres really matter, London: Centre for Cities
  • 16 Swinney P, McDonald R and Ramuni L (2018), Talk of the Town: The economic links between cities and towns, London: Centre for Cities
  • 17 Workplace jobs data from the 1951, 1971, 1991 and 2011 censuses show large cities in the North and Midlands to be the poorest performers across England and Wales, and had fewer jobs in 2011 than they did 60 years earlier.
  • 18 For example, see Swinney P and Sivaev D (2013) Beyond the high street: Why our city centres really matter, London: Centre for Cities
  • 19 Centre for Cities (2021), Cities Outlook 2021, London: Centre for Cities