Questions and answers about
the economy.

Will coronavirus cause a big city exodus?

Big cities thrive because of the economic and social benefits of proximity – but proximity also helps to spread Covid-19. Does this mean an end to the big city revival of recent years? Much will depend on how quickly we exit the pandemic – and how far the forced experiments of lockdown translate into new norms.

The coronavirus pandemic has generated a passionate debate about the future of big cities. Some predict that Covid-19 will create a tipping point that triggers the end of expensive, commuter-driven conurbations (Heath, 2020). Others are more positive about urban resilience (Financial Times, 2020): if the ‘death of distance’ didn’t finish off big cities in the 1980s, why should this time be any different?

The UK is now moving three ways at once:

  • First, a majority of workers are now commuting at least once a week, rather than only working from home (see Figure 1).
  • Second, big cities are different: as urban offices remain closed, city centre footfall – especially in London and major cities like Manchester and Birmingham – remains well down on pre-pandemic levels (see Figure 2).
  • Third, within big cities like London, footfall is shifting from cores to suburbs (Anderson et al, 2020).

Figure 1: Where are people working?

Source: ONS Opinions and Lifestyle Survey. Base = GB

Figure 2: Footfall recovery by city size

Source: Centre for Cities High Street Recovery Tracker, Locomizer.

The longer-term future of big cities – in the UK, at least – will partly depend on how durable and widespread the shift to working from home turns out to be. Optimists hope that this shift, plus wider changes in housing demand and shopping patterns, will help to rebalance the UK economy (O’Connor, 2020). Pessimists worry that this will undermine the UK’s better performing local economies, especially London (Thompson, 2020).

There are several big unknowns. One key uncertainty involves how we will exit this pandemic. We also lack evidence on how working from home might evolve, whether the density and proximity of big cities make them inherently vulnerable to globalised pandemics such as Covid-19, and how cities could adapt to those risks.

In this article, we start by reviewing what we know about working from home and its impacts on productivity and innovation. We next identify three scenarios for whether and how we exit the pandemic. We use these to work through how big cities might evolve.

Working from home and productivity

New technologies – and cheaper travel – have made remote working increasingly feasible, including for complex tasks (Clancy, 2020). Nevertheless, in the UK and most other countries, only a minority of jobs can be done from home given the tasks involved (Dingel and Neiman, 2020).

Dingel and Neiman estimate that 43.5% of UK jobs could be done from home. Before Covid-19, just 5% of workers regularly did so, and only 27% of workers had ever worked at home (Office for National Statistics, ONS, 2020). That ‘working from home gap’ has now closed: a recent survey suggests that regular remote working had increased from 6% before the pandemic to around 43% by the end of June. In big cities with large professional and business services sectors usually based in offices, that shift will be even larger.

Working from home is popular: in a recent survey, over 88% wanted to continue some form of it in the future (Felstead and Reuschke, 2020). In August, a BBC survey of 50 major employers, most based in big cities, found that none planned to bring all their staff back ‘in the near future’. While 20 had already re-opened offices for those unable to work from home, 24 had no plans to do so.

Related question: What has coronavirus taught us about working from home?

Even if working from home is feasible, that does not make it more productive. We have little hard evidence on the productivity effects of the increase in remote working during lockdown. Newly remote urban office workers have experienced longer working days, with more, shorter meetings and more email than before (DeFillipis et al, 2020). While 29% of UK workers in Felstead and Reuschke’s survey reported higher output under lockdown, 30% reported that it had fallen.

Related question: Who can work from home and how does it affect their productivity?

Unfortunately, there are few robust studies of the productivity effects of remote working before Covid-19: people who strongly value working from home tend to select into it, so simple comparisons are misleading (Mas and Pallais, 2017). For existing organisations, a static shift to some form of remote working can raise productivity, albeit in very specific cases like patent examining or call centres (Choudhury et al, 2020; Bloom et al, 2015).

Working from home and innovation

Crucially, most of this evidence focuses on how existing teams adjust to working from home (the ‘static effects’) but says less about the longer term (the ‘dynamic effects’). Evidence points to dynamic learning benefits of physical workplaces (for example, Sandvik et al, 2020; Catalini, 2017; Cornelissen et al, 2017; What Works Centre for Local Economic Growth, 2018). Co-located workers can learn more easily from each other, and develop and test new ideas.

These learning benefits also extend beyond the firm, with a large body of evidence linking physical proximity in cities to higher levels of innovation. As big cities have shifted towards more knowledge-intensive economic structures, these links have strengthened (Duranton and Puga, 2020). The learning effects are especially important for younger workers, who experience bigger wage benefits from moving to big cities (De La Roca and Puga, 2017).

Dense urban areas seem to be especially good at generating new and unconventional ideas (Berkes and Gaetani, 2019). Inventors moving into clusters substantially increase both their patent counts and citations (Moretti, 2019). Even as new technologies diffuse, the local hubs that generate them hold onto disproportionate shares of employment in those technologies, particularly for the higher skilled jobs (Bloom et al, 2020).

The innovation benefits of proximity are important for urban economic growth and for economic growth, more generally. So a collective reorganisation of work could have negative implications for urban and national economies. Even if shifting to remote working makes individual firms more productive, cities may lose the collective benefits of between-firm and worker interaction.

In principle, we might develop online or remote workarounds for these face-to-face dynamics (Page, 2020). But it is hard to predict when adaptations will emerge and in what form.

Scenario 1: Early exit

If an effective vaccine or treatment is developed quickly and we see an early exit from the pandemic – say in early 2021 – we will be less likely to see major long-run impacts on big cities.

Related question: How will the economic effects of coronavirus vary across areas of the UK?

An early vaccine means an earlier end to the disruptions caused by lockdowns and social distancing. Unfortunately, temporary economic shocks – such as lockdown restrictions – may have permanent effects on local economies, and there is conflicting evidence on how long these ‘scarring’ effects may persist.

What studies we have of ‘resilience’ suggest that southern and eastern regions in the UK bounce back strongly (for example, Martin et al, 2016), and that a population with high average levels of education and skills helps areas to respond (for example, Lee, 2014).

Forced experimentation and behaviour change

We should be wary of assuming a rapid return to pre-Covid-19 norms, however. There are at least two ways in which the pandemic might lead to longer-term behaviour changes, even if we exit early. One possibility is that Covid-19 changes expectations about the likelihood of future pandemics, and this changes preferences for big city living.

Another is that the current forced experiment of mass home working may push firms to realise they can organise work differently. If senior decision-makers see that remote working ‘works’, for example, this might sweep away pre-Covid-19 ‘presenteeism’. We are likely to see many more experiments in reorganising office work in the coming months and years.

Both these shifts – of preferences away from big city living and from office to home working – are more likely to occur if exit from the pandemic happens later. This is the scenario to which we now turn.

Scenario 2: Exit in three to four years

Vaccines usually take around four years to develop, although a working Covid-19 vaccine is likely to be available sooner (Lomas, 2020). We can also speculate that Covid-19 might become less infectious, or we might develop ‘herd immunity’, over roughly the same time scale (Scudellari, 2020). This prolongs the current period of forced experimentation, meaning that more firms will shift ways of working away from pre-Covid-19 norms.

New norms, new economic geographies

If remote working becomes the norm for many office workers, at least some of the time, the implication for big cities, and ‘economic geography’ more generally, may be profound. Much of this depends on the effects on productivity.

If working from home does turn out to be an effective substitute for face-to-face interaction – and as we discussed above, this is debatable – then the more it will increase the spatial footprint of our more dynamic firms and cities in much the same way as commuting does currently, but at lower economic, social and environmental costs. Many would gain in this scenario.

One challenge to this rosy outlook is the possibility of an increase in ‘offshoring’ – firms moving work to other countries to take advantage of lower staff costs. If a task can be done effectively anywhere, why not outside the UK?

A second challenge is that history tells us that all profound changes to transport and telecommunication costs (‘transaction costs’) induce changes in preferred locations. This means that the geography of people and jobs is likely to change.

Changes to commuting and housing markets 

For households, if working from home means less frequent commuting, then the link between residential and work location weakens – or breaks completely in the case of full-time remote working. In either case, this allows non-employment considerations to play a larger role in people’s choice of residential location.

In the UK, housing shortages and unresponsive housing supply make large population adjustments unfeasible. Instead, any significant drop in the demand for big cities will end up reflected in higher house prices in more desirable non-urban locations. These price effects will reinforce existing patterns of segregation as wealthier city households move out.

The fact that higher-paid jobs are more amenable to remote working (Felstead and Reuschke, 2020; ONS, 2020) will exacerbate these effects. In big cities, falls in demand would ease pressure on prices. Unfortunately, overall affordability is unlikely to improve – as remote working plausibly increases the overall demand for space at home.

A similar story could play out within cities. In London, for example, high house prices relative to wages suggest urban amenity benefits are substantial (Gibbons et al, 2011). In this case, relative price changes are likely to hurt current commuter suburbs, where house prices are relatively high, and benefit those suburbs with worse links to existing employment centres. Again, overall constraints on housing supply mean that the resulting population changes will tend to be compositional – with wealthier households relocating to the more desirable suburbs.

The economic geography of retail and leisure

What about the location decisions of firms? We start with what economists call ‘non-tradable’ services, such as retail and leisure. Wealthy households moving to the suburbs increase demand for high-end shops, restaurants and gyms in those neighbourhoods. This could end up increasing commuting for workers providing these services but unable to live nearby.

A critical question is then whether we will see one-for-one shifts of activity from central city locations to suburbs. This will depend on the extent to which households substitute home production, local or online purchases for the services they used to consume while at work. For example, many more workers may make their own lunch when working from home; and while shopping might shift to local businesses, it could also move online.

Lockdown will have seen many families substitute online shopping for physical retail (Relihan et al, 2020), and these effects might persist. Offsetting this is reduced expenditure on commuting and office clothing; against that, there is higher spending on office equipment and energy at home.

The future of offices and office work 

In large cities, financial and business services are a big share of ‘tradable’ activity: goods and services consumed outside the city where they are produced. We focus on office work, as most tradable activity in big cities is currently organised this way.

For these firms, we can summarise the adjustment process in three stages:

  • At the national level, a sudden shift to remote working decreases the amount of work that is done in offices and thus overall demand for office space.
  • Firms reorganise individual workflows, a process that is already beginning to unfold.
  • In aggregate, these decisions will have major implications for the commercial property market, with winners and losers in different cities.

One key decision for a firm is what tasks and roles must still be performed in an office environment, under social distancing if required. City traders who need very fast, bespoke internet connections might be one example. Others might include many creative sector employees who place high value on face-to-face interactions.

Another set of decisions is how the wider workforce uses office space – specifically, what mixes of remote and office work are most suitable to meet firms’ goals. Understanding the effects of home working on productivity and innovation, among other outcomes, is crucial here: as we explained above, the existing evidence is thin.

As firms start to experiment, caution is needed in making predictions about the office market overall. For example, across firms we might see a lot of full-time remote working, or more mixed patterns (‘five days in ten’, with social distancing).

In the former case, total demand for office space will be lower than before Covid-19. In the latter case, total demand might be the same or higher, if most firms keep at least some staff in socially distanced offices, for at least some of the week. In line with this, US firms surveyed in June predicted zero change in future demand for space – rather than the large drops some might expect (Altig et al, 2020).

The wider economic geography of these shifts is also hard to predict. A wide range of theoretical models suggest that falling transaction costs may lead to more concentration in economic activity or to other important structural shifts.

For example, one study suggests that falling transaction costs help to facilitate the concentration of ‘front office’ activities in city centres while ‘back office’ activities move to the suburbs (Ota and Fujita, 1993). Another study makes parallel predictions that falling transaction costs facilitate functional specialisation – explaining why headquarters are increasingly concentrated in bigger cities, while smaller cities specialise in production (Duranton and Puga, 2005).

These theoretical predictions are supported by empirical evidence, including the fact that recent reductions in transaction costs have occurred at the same time as big city growth.

Scenario 3: No exit

In the absence of an effective vaccine, declining infectiousness or herd immunity, Covid-19 will stay endemic in the community, and will need to be managed through large-scale testing, track and trace, and where unavoidable, local lockdowns (Aleta et al, 2020). What are the prospects for cities in this long-term, more speculative, scenario?

There are at least three reasons why a more radical shift in economic geography becomes more likely as the time to exit increases, or if Covid-19 stays endemic:

  • First, a long crisis might strengthen beliefs about the likelihood of future pandemics. In turn, that would change the behaviour of firms and households.
  • Second, slow exit creates more time to experiment with and optimise existing remote working technology – and for new technologies to emerge. These ‘diffusion lags’ can be lengthy: in one of the best-known examples, two decades passed before factories took full advantage of electrification (David, 1990).
  • Third, large-scale shocks can lead to permanent shifts – even if the underlying fundamentals don’t change. Economists talk about such situations as having ‘multiple equilibria’. For example, home working might not work for one person when everyone else is in the office, but it works when everyone is out of the office. The argument is similar for firms. Urban structures may also exhibit multiple equilibria: for example, the shift from non-traded services in central business districts to local neighbourhoods that we are already seeing might persist even if we see some traded activity return to city centres.

Are big cities more vulnerable to pandemics?

A key question in this long-term scenario is whether big cities are more vulnerable to Covid-19 or similar viruses? Big cities have a long history of being hard hit by pandemics. Evidence on Covid-19 is more limited. Contact-tracing studies highlight how prolonged exposure in crowded, enclosed environments and events spreads the virus (Cevic, 2020; Peel and Burn-Murdoch, 2020). Thus, it is possible that big city size, density or connectivity makes them inherently more vulnerable to Covid-19 (Glaeser, 2020).

Related question: Why has coronavirus affected cities more than rural areas?

Consistent with this, in the UK and elsewhere, the biggest, densest cities were hit earlier by coronavirus. But allowing for timing, one US study finds no significant link between Covid-19 death rates and density (Carozzi et al, 2020). That is, other factors may offset the proximity disadvantage of big cities. For example, they tend to have younger than average populations and greater numbers of people who can reduce their exposure by changing their behaviour.

If these findings generalise, broader economic, physical or demographic characteristics – for example, reliance on public transport, crowded urban housing, physically and economically exposed work in frontline services, and concentrations of vulnerable groups - may better explain the spread of Covid-19. While these are features of many big cities, especially in more deprived neighbourhoods, they are also present in smaller cities and towns – including areas where Covid-19 may now be pervasive. It is essential, but challenging, to disentangle these competing explanations (Kolko, 2020; Nathan, 2020).

Longer-term urban adaptation

‘No exit’ may also lead to other adjustments in big cities. For example, for older households, more vulnerable to the disease, the benefits of home working will increase, while the amenity benefits of cities will weaken. As the young are less vulnerable, this suggests that the demography of urban cores may get even younger, while older households move to the suburbs or out of cities entirely.

Another big unknown is how big cities will adapt. Working from home is one way in which cities are already adapting. How will cities adjust in other ways? For example, can cities that rely on public transport find innovative commuting arrangements that both manage congestion, and allow employment to remain concentrated in the existing city centre? How might they manage this against a backdrop of falling tax revenues, if overall demand for office space falls and takes some local services with it?

History suggests that big cities have usually managed to adapt ways of working and living to new diseases (The Economist, 2020; Glaeser, 2020). If that is not possible this time, we may see more profound changes to urban systems, with activity moving out of big cities towards smaller cities and towns.


Big cities have gone through long phases of growth and decline in the past. How will they evolve after Covid-19? The longer-term impact on big cities is highly uncertain. We have pulled together theory and evidence to speculate about some possible urban futures.

Covid-19 has already altered patterns of urban work, shopping and social interaction: the question is whether – and how – these continue. In countries with effective public health responses, we are more likely to see voluntary change: as public confidence returns, firms and households can try out new ways of organising life and work. By contrast, in locations where the virus remains endemic, forced adaptation will continue.

At one extreme, we might see no significant long-term changes: big cities will bounce back just as they have done in the past, with a return to old norms. At the other extreme, we could see radical changes to ways of working and living, and to our economic geography – although exactly what these might look like is not yet clear.

It is important to remember that in the recent past, long-term falls in transaction costs – due to better technology and cheaper travel – have been associated with more big city growth, not less. Covid-19 could strengthen, rather than weaken these tendencies. Nevertheless, the longer the exit, the more plausible is a shift within the economic geography of big cities, with suburbs benefiting at the expense of city centres. Under a long-term no exit scenario, if big cities cannot effectively adapt their systems and infrastructures, we could see more fundamental shifts in urban systems.

Where can I find out more?

Who are experts on this question?

Authors: Max Nathan and Henry Overman
Thanks to Valentine Quinio at Centre for Cities for data. Thanks also to Diane Coyle, Adam Dennett, Jonathan Dingel, Meg Kaufman, Mark Kleinman, Jon Reades, Helen Simpson, Anna Valero and Romesh Vaitilingam for helpful comments on earlier versions.
Photo by TheOtherKev from Pixabay
Recent Questions
View all articles
Do you have a question surrounding any of these topics? Or are you an economist and have an answer?
Ask a Question
Submit Evidence