03The past four decades of economic complexity (1981-2019)

The puzzle of why big cities are complex but not especially productive can be answered by looking back through history. Using data from the Census in 1981 to compare economic complexity in the same cities across an almost forty-year period provides insights into how their economies and the national economy have or have not changed.

Cities that are complex today were usually complex in 1981 too

The majority of cities that were the most complex in 1981 are also the most complex today. As Figure 6 shows, cities and large towns in the Greater South East had a higher likelihood of attracting knowledge-intensive workers and businesses. Box 3 looks in more detail at London’s development over this period which goes well beyond the general characterisation of the impact of the ‘Big Bang’.

Figure 6: Most Greater South East cities remained complex between 1981 and 2019, while big cities became more complex

Source: ONS; Census, 1981. Centre for Cities’ calculations.

At the opposite end of the spectrum, most cities and large towns in the North and Midlands remained ‘trapped’ in low complexity economies. With comparatively little knowledge in 1981, these cities have since struggled to attract new innovative businesses, which would have consequently increased their productivity.

Box 3: London’s growth and the role of the financial sector

The rise of London over the past 40 years is generally associated with the ‘Big Bang’, a set of financial deregulations in the mid-80s that lead to the expansion of the financial sector. However, London’s success cannot be entirely explained by the rise of finance. London did not just simply replicate its existing strengths in finance. The capital was also able to attract talent, innovate and diversify its economic base across a number of sectors.

These changes are in line with how the economic complexity theory describes development, as complex places are not just highly specialised – they also find it easier to successfully transition into new high-value activity and industries.

The  growth of finance happened within the context of specific changes in the industry itself, combined with globalisation and an overall economic shift towards the service sector. When compared with other knowledge-based services, London’s growth of finance-related jobs was dwarfed by sectors like programming, design, advertising, and research. Figure 7 shows that in 2019, other knowledge-based services (e.g. design, computer programming, advertising) accounted for 47.2 per cent of all exporting jobs, above the 26.5 per cent from Finance and Insurance related activities.

Figure 7: Finance-related employment rose but not as much as knowledge intensive services

Source: ONS; Census, 1981.

Big cities are an exception, and have become more complex over time

Most big cities were able to break out of the ‘low complexity’ trap and substantially increase their complexity. The relatively complex nature of the economies of large cities today is the result of four decades of improvement. This has meant that as a group, Britain’s large cities (excluding London) have shifted from having below-average levels of complexity in 1981 to being substantially more complex than the urban average today, as shown in Figure 8.

Figure 8: Most of the largest cities have become substantially more complex in the last four decades

Source: ONS; Census, 1981. Centre for Cities’ calculations. Urban ECI computed at the Local Authority level including all local authorities. City’s ECI computed at the PUA level, including urban areas only. Largest cities measured by total employment and ECI scores are a weighted average considering each PUA’s size.
Note: The cities considered as largest are the following: Birmingham; Bristol; Glasgow; Liverpool; Leeds; Manchester; Newcastle; Nottingham; Sheffield.

From the 63 cities under analysis, Manchester is the city that improved its economic complexity the most, in relative terms, followed by Glasgow.21 Manchester was the 52nd most complex economy in Britain in 1981 – today it is 18th. Likewise, Glasgow was the 38th most complex city, and today it is 7th. Leeds went from being the 40th to the 13th most complex economy, and  Liverpool moved from 36th to 17th.

The re-emergence of the largest cities is not a phenomenon specific to the UK, but is a part of a global shift that larger cities have been able to benefit from the most.22 But across the largest cities, Birmingham and Sheffield are the two exceptions: they had comparatively low complexity economies in 1981 and saw their complexity decline in relative terms.

As Box 4 discusses, big cities were unusual in that they were to attract complex businesses without having a large share of jobs in related activities by 1981. Big cities have begun to prosper because of this change.

Box 4: Cities can develop industries without previous knowledge in similar areas - especially large cities

In recent years, some of Britain’s largest cities were able to become relatively more complex by specialising in new high-knowledge activities. Data at the occupational level suggests that in some cases – often in Britain’s biggest cities – the observed improvements were unlikely to be a result of the existing economic structure in 1981.

Computer-related activities

In 2019, economies with a strong IT-related sector were generally specialised in electronics-related occupations in 1981 with both sectors being considered complex in 2019 and 1981 respectively.23 Half of the cities with a competitive advantage in the IT sector – such as Reading, Slough, London or Brighton – were specialised in electronics activities in 1981, as shown in Figure 9.

Nevertheless, Leeds and Nottingham were able to specialise in IT-related activities by 2019, without having an electronics’ legacy from 1981; Nottingham ranked 37th out of 62 cities in terms of being specialised in electronics and is now 14th in the country for IT.

A similar trend can be found in a number of other activities. Liverpool was able to build a competitive advantage in research-related activities, one of the most complex sectors the city has today, despite being ranked 49th  out of 63 cities in Research and Development activities in 1981.24 Additionally, cities like Manchester and Nottingham were some of the least specialised economies in telecommunications in 1981; but had developed a competitive advantage in wireless telecommunications activities by 2019.

Economic complexity theory suggests that places develop specialisms based on previous specialisms. At the city level, this analysis suggests that having a previous specialism may not be a necessary condition to develop one today. If cities can attract talent, therefore increase their accumulated knowledge, complex economic activities are likely to emerge.

Figure 9: Economies focused on electronics were more likely to move towards IT-related occupations but there are some notable exceptions

Source: ONS; Census, 1981.

The big picture is that the geography of knowledge in 2019 is less Southern than it was in 1981, because big cities outside London in the Midlands, North of England, and Scotland have become much more important places of knowledge creation.

Big cities have benefitted from how the economy has changed, not from policy bias

The turnaround of most of Britain’s large cities goes somewhat against economic complexity theory, which says that the evolution of an economy is related to its past specialisms.

One explanation in the public debate is that large cities have been explicitly favoured by policy in recent decades. This has sucked jobs into cities, so the argument goes, at the cost of their surrounding areas.

But finding evidence for this view is difficult. First, there is no consistent pattern of pro-urban policy over this period. For instance, the Local Government Act 1986 abolished the Greater London Council and the metropolitan county councils, and these were not reversed until 2000 and the 2010s respectively.

Second, there have been city-specific policies, such as Michael Heseltine’s City Challenge or City Deals under the Cameron-led government.25 But as Box 5 illustrates, in the wide gamut of local growth policies that have been put in place in the last 40 years, very few have been city focussed.

Furthermore, all local growth policies tend to be relatively marginal compared to other policies or spending decisions by national government. This affects how the nationally economy adapts and changes, with consequences for local economies too.

Box 5: A timeline of sub-national policies

While not exhaustive, the below sets out a long list of local growth initiatives that have been put in place since the 1980s. Only three have had an explicit city focus – City Challenge, City Deals and Mayoral Devolution Deals. Meanwhile, there have been a number of initiatives that have been more explicit in not having a city focus, such as the creation of the Coalfield Regeneration Trust, Coastal Communities Fund and the recently announced Towns Fund. The dates below are the year the policy was introduced.

1981 – Enterprise Zones

1991 – City Challenge

1994 – Single Regeneration Budget

1998 – Regional Development Agencies

1998 – New Deal for Communities

1999 – Coalfield Regeneration Trust

2002 – Neighbourhood Management Pathfinder Programme

2006 – Local Enterprise Growth Initiative

2011 – Local Enterprise Partnerships

2011 – City Deals

2012 – Coastal Communities Fund

2012 – Enterprise Zones

2014 – Local Growth Fund

2014 – Mayoral Devolution Deals

2019 – Towns Fund

2019 – Future High Streets Fund

2021 – Levelling Up Fund

Another explanation for why large cities have improved over the past forty years is that the national and global economy has moved from manufacturing towards knowledge-based services. The inherent benefits that large cities offer has meant they have been well-positioned to make the most of this change.

And, cities have benefitted from non-spatial policies that were introduced by successive governments over the past 40 years. For instance, the expansion of higher education has seen the growth of universities that are largely city-based and immigration policy has also benefited London in particular. But this has not been the result of an explicit policy focus towards cities, but rather one that implicitly benefits cities due to their role in driving economic growth.

In other words, the big cities are stepping into a new role that the global economy has demanded of them. As their relationship to knowledge has changed and improved so much since 1981, their current productivity underperformance suggests this process is unfinished. Further improvements to their economic performance are possible.

‘Building on your strengths’ is unlikely to turn around struggling economies

Yet not every urban area has seen its relative economic complexity improve. When looking at the cities and large towns that have struggled in Figure 6, the data suggests policymakers should be cautious about economic strategies that encourage struggling places to ‘build on their strengths’ and ‘smart specialisation’. These cities have either failed to transition from activities that were complex in 1981 to new industries that are complex in 2019, or they have replaced one set of low-knowledge activities with another set of low-knowledge activities. They fall broadly into three groups:

Complex cities that did not develop new specialisms are less complex today

In contrast to the improving complexity of many large cities, there is a group of cities and large towns where the opposite is the case, and which have become less complex. Cities and large towns found in the bottom right of Figure 6 – such as Aberdeen, Blackpool and Swansea – have taken a backward step in the last 40 years. They had above-average levels of complexity in 1981 but have below-average levels in 2019.

A common characteristic of these places is that they continued to specialise in the same activities as they did in 1981 while the global economy has shifted towards knowledge-based services, as Table 2 shows. Aberdeen has not moved beyond oil. Blackpool continues to specialise in aerospace manufacture. And steel still dominates in Swansea.

Table 2: Those places that continued to specialise in the same industries became less complex

PUA 1981 2019 Complexity (1981-2019)
Edinburgh Radio/electronic capital goods (8.2%) Computer programming, consultancy and related activities (19.0%) Remained high
London Banking/bill-discounting (8.4%) Computer programming, consultancy and related activities (16.8%) Remained high
Reading Electronic data processing equipment (4.8%) Computer programming, consultancy and related activities (37.4%) Remained high
Aberdeen Extraction: mineral oil/natural gas (24.5%) Extraction: mineral oil/natural gas (28.3%) Deteriorated
Blackpool Aerospace manufacture/repairing (20.6%) Aerospace manufacture/repairing (26.7%) Deteriorated
Swansea Iron and Steel industry (12.1%) Manufacture of basic iron and steel and of ferro-alloys (13.6%) Deteriorated

Source: ONS; Census, 1981.

This is in contrast to those cities and large towns that remained complex throughout the period, such as Reading and Edinburgh. These cities changed their main specialisation patterns from electronics to IT-related activities, allowing them to maintain their comparatively high productivity levels.

Cities dominated by a single industry

Cities with low complexity today tended to be more specialised in a single activity in 1981, having on average 18.8 per cent of their exporting jobs in a single occupation, compared to the urban average of 11.1 per cent. From the 15 cities that were dominated by a single industry (20 per cent or more of their exporting jobs) in 1981, only three are complex today.26

This is likely to be because overreliance on a small number of activities has limited the ability of cities to innovate and move to new economic activities.27 The observed dynamic has similarities with the economic concept of the ‘resource curse’, which argues that high reliance on few economic activities, associated with natural resources hinders long-term growth.28

Cities that moved from one low complexity activity to another

The final set of cities are those that have been ‘trapped’ in low complexity activities of different types over the past four decades, shown in the bottom left of Figure 6.

These cities offer a counterexample to the experience of most large cities. Like the large cities, their development also appears to have been guided by the inherent benefits that they offer to businesses, rather than their past industrial structure. The problem is that the benefits they have offered, such as large pools of labour and cheap land or geographic location, have appealed to lower productivity and lower wage activities (e.g. warehousing, distribution or food manufacturing).

Of the 22 cities and large towns that are ‘trapped’ in low complex activities, six of them specialised in coal mining in 1981. Today, these urban areas have moved away from coal, and are specialised in other activities, as Table 3 shows. None of these sectors are related to coal mining, suggesting their earlier specialism driven by the location of coal seams had no direct impact on the subsequent specialisms they developed.

Table 3: Most prevalent exporting occupation as share of exporting jobs, 1981-2019

PUA 1981 2019
Barnsley Deep coal mines (40.8%) Warehousing and storage (18.8%)
Doncaster Deep coal mines (34.9%) Warehousing and storage (20.7%)
Mansfield Deep coal mines (26.6%) Manufacturing of plastic products (12.3%)
Sunderland Deep coal mines (10.8%) Manufacturing of motor vehicles (17.7%)
Wakefield Deep coal mines (29.1%) Warehousing and storage (29.5%)
Wigan Deep coal mines (8.1%) Manufacturing of food products (15.8%)

Source: ONS; Census, 1981.

But all are lower skilled activities, suggesting that it is the inherent benefits – often driven by indirect legacy of previous activities (e.g. contamination affecting land values) – that these places offer to businesses that has driven their development, rather than the specific accumulated knowledge from past industrial structure. Box 6 gives an example of how policy that aims to increase economic growth can sometimes inadvertently reinforce these qualities, and how replicating the existing geography of knowledge does not always mean retaining the same industrial structure.

Box 6: Industrial structure did not attract Nissan to Sunderland

In 1984, the British government and Nissan reached an agreement to open a car plant in Sunderland.29 Before the car plant, Sunderland’s economy was mostly dominated by coal mining and shipbuilding.30 There is little evidence that the accumulated knowledge derived from such activities was the reason why Nissan located in Sunderland.

If Nissan had located in Sunderland mainly as a result of the specific capabilities built from previous industries, we should expect to see a strong relationship between coal mining and shipbuilding with car manufacturing in other British cities.

Figure 10 shows that there is no relation between Sunderland’s previous advantages and the likelihood of making vehicles. Cities that mined lots of coal and built lots of ships are now unlikely to build lots of cars. Likewise, cities that today are centres of automobile production are very unlikely to have been centres of coal and ship production in the past.

Urban economies with some degree of specialisation in car manufacturing today did not share Sunderland’s economic features in 1981.  Unlike Sunderland, Coventry and Birmingham transitioned from cycle to car manufacturing in the previous century.31

At the same time, places focused on mining and shipbuilding like Barnsley or Doncaster did not move their economic structure towards car manufacturing. Sunderland was able to attract Nissan due to other benefits, such as public subsidies, that were not directly related to the accumulated knowledge and capabilities from its previous industrial structure.

Instead, it was the characteristics of Sunderland’s labour and real estate markets –cheap labour and land– that indirectly supported Nissan’s choice of Sunderland. These were associated with Sunderland’s economy as a centre of shipbuilding and coal mining, but could also be found in other cities across the country in the 1980s.

Figure 10: Cities and large towns with car manufacturing plants today were not similar to Sunderland in the early 80s

Source: ONS; Census, 1981.
Methodology: Top five car manufacturing cities in 2019 include Oxford, Coventry, Luton, Liverpool and Birmingham. *The most similar cities to Sunderland in 1981 are Barnsley, Plymouth, Doncaster, Mansfield, Portsmouth and Newcastle, based on their percentage of 1981 jobs in the following sectors: deep coal mining; Shipbuilding and repairing; Other glass products; Mechanical lifting/handling equipment; and active components/sub-assemblies.


  • 21 See Figure 12 (Appendix 1) for further details.
  • 22 An economic trend described by Moretti as ‘the Great Divergence’ where the geographical clustering of the most productive companies disproportionally benefits a small number of cities.
  • 23 IT-related occupations include “Computer programming, consultancy and related activities” and “Data processing, hosting and related activities; web portals”; and Electronics-related occupations include “Electronic data processing equipment” and “Radio/electronic capital goods”.
  • 24 Cities ranked by sector’s job prevalence, as a share of all exporting jobs.
  • 25 Despite its name, ‘City Challenge’ also covered local authorities that the Centre for Cities does not consider urban areas such as Sefton and Hartlepool. In other cases, the policy targeted peripheral urban local authorities (e.g. Sandwell and Walsall).
  • 26 Those cities are Oxford with 24.9 per cent of exporting jobs in Motor vehicle bodies; Crawley with 23.8 per cent of exporting jobs in Air transport; and Peterborough with 25.1 per cent of exporting jobs in Internal combustion engines. The activities they specialized in do not seem to be reason why these cities were able to remain complex as some low complexity cities had competitive advantages in the same sectors (e.g. Luton).
  • 27 This is echoed in existing research. Moretti’s ‘The New Geography of Jobs’ (2012) explains the problems caused by the heavily dependence on manufacturing in cities like Detroit (page 75), and shows how the most sophisticated technologies become more common and less value with time (page 82).
  • 28 Commander S (2018), One-company towns: Scale and consequences, IZA World of Labour; The term resource curse was first used by Richard Auty in 1993 to describe how resource-rich countries tended to be relatively poor. Today, the IMF considers a nation ‘resource-rich’ if at least 20 per cent of exports or fiscal revenue is derived from ‘non-renewable natural resources’. Aunty R (1993), Economic and Political Reform in Developing Countries: Economic Development and the Resource Curse Thesis (page 58 to 80).
  • 29 See Centre for Cities’ blog ‘Does Nissan provide a model for levelling up?’ at https://www.centreforcities.org/blog/does-nissan-provide-a-model-for-levelling-up/
  • 30 Followed by manufacturing activities like other glass products; mechanical lifting/handling equipment and active components/sub-assemblies.
  • 31 Clayton N & Mandair R, (2014) Cities Outlook 1901, London: Centre for Cities