The doughnut effect: how cities are changing as we know it
Two Stanford economists have observed a ‘hollowing out’ of city centres across the United States, in favour of suburbia. Their findings help explain how COVID-19 can reshape real estate; the consequences of which reach out farther and wider than simply where people want to live – but to the fundamental way in which we define cities.
COVID-19 has changed how and where we work. But aside from creating makeshift desk space at the kitchen table, it has also had a longer-lasting impact on where we work from in a more geographical sense.
“Historically there has been a lock-tie between where you live and where you work,” notes Arjun Ramani of Stanford University. “But that has been broken.”
Ramani and Nicholas Bloom authored a brief for the Stanford Institute for Economic Policy Research on how the pandemic is shaping the real estate market. The pandemic has plunged millions of workers, confined to their homes for a year, into an existential crisis. Residential properties have been the setting for working and socialising, instead of the office and other public settings. As such, many people are now much more acutely aware of every square metre of space that they inhabit. Their findings go some way to proving the existence of this self-realisation.
In the centres of the largest cities in the United States, such as New York City and San Francisco, they observed a slump in real estate demand since the start of the pandemic – a 10-15% drop in residential rents – in comparison to the suburbs. They also observed a similar trend in purchase prices, negating any pretence that this is a short-term effect.
One can also extrapolate a similar theory from UK house prices. The Office for National Statistics collected data on house price growth in 2020 by regions in England. The data indicated that the region with the lowest annual house price change, at just under 4 percentage points, was London. In comparison, the South West, East Midlands, North West and Yorkshire all saw prices skyrocket by more than 10%. While this doesn’t tell an identical story, preferences may indicate a desire for a similar relocation.

Searches for homes with gardens – by definition properties which need more space and consequently are more likely to be located in lower density areas – on property search website Rightmove rose by 84% year-on-year for renters and 42% for buyers in May 2020.
Why is this happening? Firstly, the requirement to work from home might be a driver in housing demand. Ramani and Bloom’s research found a correlation between population density and the share of jobs that can be done from home – and as such, a negative correlation between the areas with a higher share of work-from-home jobs and the percentage change in house prices. Not only do people generally want a larger space in which to work from home – particularly if they are currently living in a smaller, more expensive city centre residential space with no dedicated office or desk – but there is also no longer a requirement for them to live right next to where they work. The pandemic has also reduced the value of living in a city centre, separate to a shorter commute; amenities such as theatres and restaurants are currently inaccessible.
“You’re seeing the centre of the city hollowing out,” says Ramani. “We’re calling this phenomenon the donut effect.”
(For the sake of consistency, I’m going to use the British English ‘doughnut’ from now on, but if you’re looking for Ramani and Bloom’s paper, you’re better off searching for ‘donut’.)
So we have a shift in workers’ preferences to live in the suburbs – the ‘doughnut’, moving away from the central ‘hole’. But while these factors might dissipate once vaccination programmes have rolled out – a city’s financial district is unlikely to continue its resemblance of a ghost town – earlier research from Bloom argues that increases in working from home will be persistent. Leesman data supports this theory; of 48,000 respondents, 85% expressed a preference to work from home a minimum of two days per week.
Perhaps individuals are taking the opportunity they have been given – away from the office – to relocate elsewhere in the United States? Not so; Ramani and Bloom note that housing demand hasn’t shifted significantly from these larger conurbations to less expensive cities. Ramani believes this desire for a hybrid working strategy, with a pull to the office at least once per week, is enabling a geographical shift to local areas, rather than a reallocation across the country. As Ramani explains: “What hybrid working doesn’t enable you to do is pick up your bags and move across the country. Instead, people are now willing to live an hour away in exchange for a bigger house, because they don’t have to commute in every day.”
Consequently, the research indicates a surge in house prices in suburbia, and a fall in prices in the city centres. What sort of knock-on consequences could this effect have on the economy of those cities – and at a societal level?
Ramani makes some poignant remarks about decisions policymakers will need to consider should the doughnut effect stick. Will cities need to rethink their transport network and infrastructure? Theory dictates fewer, or less-frequent, trips on the NYC Subway or the London Underground. Certainly, rush hour would involve fewer workers crammed tightly inside carriages.

But commuter trains will likely need to travel longer distances to reach the city centre – suburban rail lines may need be extended out of the traditional commuter belt.
Transport franchises, such as Transport for London, will need to factor in lower revenues. Larger cities in the U.S. will need to factor in a loss in sales tax revenues. Ramani thinks that in the short-term, governments will continue to support cities while they try to recover from the pandemic, but eventually these cities will need to balance their budgets. Developers will also be wary of falling commercial rents. Says Ramani: “While there is less variation in commercial office space prices between larger and smaller cities, there appears to be a drop of 10-15% across the board.”
There are also long-term implications for consumer spending in hospitality and retail. Ramani explains: “[Bloom] has a paper that projects about a 10% drop in consumer spending in America’s largest metro areas. If you’re commuting to the office less frequently and moving away from the city centre towards a suburb, you’ll see a shift in consumer spending in the same direction.”
A lot of businesses in the city centres, particularly service businesses, are going to suffer – and a failure to return to pre-pandemic levels raises potential concerns of an increase in inequality, says Ramani.

“A lot of the service workers, on a typically lower wage, are going to be hurt, while the higher wage workers who have the ability to work from home are going to have the flexibility to do what they want.”
Dramatic stuff. Could a city’s central nervous system essentially collapse if people just decide not to go there? Is it a delicate house of cards which needs delicately rebuilding at the foundations? This is unlikely. In fact, historical data on population density helps to explain the value of a sprawling, bustling, vibrant city.
They weren’t always seen as immense, towering centres of work and living. “Back in the 1800s,” Ramani notes, “working from home was quite common. Workers would go into a city or to a market to get raw materials and goods and return home to work – for example
in the manufacture of clothes.”
Early data shows that in the 1800s, 40% of people were working from home.
“As late as 1900, a third of people were working from home in France. It’s only later in the 20th Century where we have this type of arrangement where people go to work in a different physical location.”

Ramani points to the concept of the ’superstar city’ – extremely dense, metro areas, especially those with central business districts being a defining area. Cities which are examples of economics of agglomeration, a concept within urban economics related to the idea of economies of scale, in which the costs of production may significantly fall as a result of clusters of businesses and a pool of workers in close proximity to each other.
After the development of computers and the internet in the 1970s, 1980s and 1990s, city centres experienced an even bigger concentration of skilled workers, with cities growing faster than before.
“The question is why?” Why, asks, Ramani, with increasingly faster connectivity and increasingly more affordable technology, did this pool of people flock to cities?
Last year, four US-based economists posed a similar question.
Why, in large cities in which living costs are much higher, do a disproportionate number of workers who could work remotely – and live anywhere – decide to pay such an exorbitant price of living?
Their research finds that the denser a city, the more work can be done remotely. The economists called this oddity ‘the City Paradox’.
The answers to these questions are to do with increasing returns to collaboration. When people live in a city where a lot of other people in
their industry also live, there’s a spillover of ideas; collaboration becomes much easier.
“Crucially, this also happens between firms,” he says. “One interesting finding is innovation actually increases in these dense, superstar cities, because it becomes easier to exchange ideas, and perhaps come up with a new organisation.” This type of innovation due to clustering is a key mechanism of why superstar cities reap significant economic benefits.
In a virtuous circle for these cities, recruiting also becomes easier. It becomes much easier to recruit skilled workers, such as software engineers in San Francisco, if all of them are in the same location. And by growing these common labour pools and clusters, that drives innovation on. This economic agglomeration and innovation spiral is also a reason why real estate prices tend to be so high in these cities – the skilled workers flocking to innovate together tend to be higher income workers. And it is this virtuous circle which is also driving the doughnut effect Ramani has observed: the higher income workers are a mechanism for real estate demand.
The future is fascinating. What sort of impact might different hybrid working models have on the doughnut size? Could one take organisations who decide to work five days per week in the office, and compare those employees with those working at firms working just one day in the office? Would we see a larger ‘hole’ for those who will be spending more time at home? As much as economists love to theorise with rational decision-making, it’s wise to add a healthy pinch of salt. Bloom ran an experiment in 2015, which found considerable variation across different workers. “About half of them don’t even like working from home,” explains Ramani. “They prefer to be in the office. For all these phenomena, there’s a lot of heterogeneity.”
A mix of preferences is just one factor why the workforce won’t be entirely dispersed. Furthermore, there is great value placed on learning from others and what the office provides that the home cannot, and cities are unlikely to ignore those returns on collaboration that served them so well in decades past.
What next for Ramani? “We’ve got this global experiment that’s lasted for more than a year – some of the stigma attached to working from home seems to have been reduced. We’ll keep on running the same types of analyses to see how real estate demand continues to change.”
And if the doughnut effect has a lasting impact? Not only might cities undergo a redistribution of its population, but cities might become more affordable for many of those that have been previously priced out. The ripple effects of lower city centre prices – and fewer five-days-per-week commuters – on consumer spending, long-term transport infrastructure plans and other less obvious correlations, are likely to be felt for many years to come.