8 January 2025
This is the second post in our series featuring work submitted to the ECB’s 2024 Young Economist Prize. Morgane Richard was selected as one of the finalists with the research highlighted in this post. Applications for the 2025 Prize will be open from 13 January to 12 February 2025. For more details, go to the dedicated webpage.
The rise of working-from-home during the pandemic dramatically changed the way we organise work. And teleworking is transforming more than just our professional lives. The ECB Blog looks at how this shift is affecting the housing market and inequality.
The COVID-19 lockdowns induced a structural change in the organisation of work: working from home (WFH), previously an option used by only a fraction of the population, became a widespread practice. Adapting to this new environment, workers have been changing what they look for in a house, increasing demand for larger properties located farther away from city centres. Simultaneously, WFH has introduced a new division within the workforce, between those who can work from home and those who can’t. In this post, I show that this change in work patterns has had significant effects on the housing market, as well as increased inequality among workers. The London housing market serves as my case study.
Changes in the London housing market
In a first step, I looked at raw house price and rent data for the 2018 –2022 period. I found that, since the onset of the COVID-19 pandemic and the rise in WFH, properties located further away from the city centre and larger homes saw the fastest increases.[1]
Chart 1
Growth in properties’ value as a function of distance to the centre (London)
House prices (left), rents (right)
Percentage change
Notes: Each dot represents one of London’s local authorities (e.g. Camden or Hackney). The x-axis plots changes in average house prices and rents between the year before COVID, and the last year of data available (July 2021 to June 2022 for house prices, and January to December 2021 for rents). The y-axis plots the logarithm of local authorities’ average distances to the Bank of England (in metres). I excluded the top 1% in house prices, rents, and size (in square metres) to remove outliers. A linearly fitted line is added to the plots.
Chart 1 displays changes in house prices (left panel), and rents (right panel) as a function of distance to the city centre.[2] Each dot represents one of London’s local authorities (Camden, Hackney, etc.). We observe a clear positive relationship between house price and rent growth from 2019 to 2022, on the one hand, and distance from the city centre, on the other. For example, the average price of properties located in central London decreased by 1%, while prices for properties located in the periphery increased by an average of 13%.
Chart 2
House prices and rents by size of property (London)
House prices (left), rents (right)
Index
Notes: Properties are split by number of rooms. I excluded the top 1% in house prices, rents, and size (in square metres) in order to remove outliers.
Chart 2 displays house price (left panel) and rent (right panel) indexes by property size. The reference period is February 2020, right before the onset of the pandemic. Properties are split according to number of rooms. This indicates that larger properties have appreciated faster since the rise in remote work. For instance, between February 2020 and June 2022, the average price of large houses (5 rooms or more) increased by 20%, while that of small ones (studio or one-room) dropped by 1%.[3]
Have larger, suburban properties really gotten more expensive?
Now, how do we know that the changes we see in the raw data are indeed related to the location and size of houses, and not to other neighbourhood and property characteristics that we did not control for? To estimate the isolated impact of property size and proximity to the city centre on house prices and rents, I use the hedonic pricing approach.[4] The idea behind this method is that a house is made up of many characteristics, such as type of property and energy efficiency, all of which may affect its value. Hedonic pricing models are used to estimate the marginal contribution of these characteristics. The regression estimates give the implicit prices of each characteristic.
The results show that the premium for space has increased since February 2020. In the home-ownership market, for example, the space premium has grown by 5%. Upgrading from an 86 m2 house (the statistical average) to a 102 m2 house (i.e. the 75th percentile) comes at a size premium of £79,000 before February 2020, and £83,000 after.
Additionally, the penalty for being farther from the city centre, what we call the commuting penalty, has decreased by 6%. The distance penalty associated with the average house in the suburbs (i.e. beyond Zone 2 of the London Underground) compared to the centre (i.e. Zones 1 and 2) was £107,000 before the rise in WFH and £100,000 after.
Did housing preferences change because of WFH?
So we can see that the documented changes in house prices occurred at the same time as the rise in WFH. But how do we know that these developments were causally linked?
The limited data available do not allow us to establish a causal link between WFH and the change in house prices empirically. What is more, the data provide detailed characteristics of the properties, but we don't know anything about the households that live in them.
I used a model to get around this problem, a fake economy that reproduces key observed features of the data to use as a laboratory to explore the consequences of shocks and public policies. More precisely, I used a dynamic spatial heterogeneous agent model. In the model, households decide where to live in the city, whether to own or rent their home, and – if they are employed in a job compatible with WFH – how to divide their time between WFH and working in the office. House prices and rents are determined in general equilibrium in each of the city’s locations.
The shock I introduce in the model is a permanent change in workers' preference for remote work. We can think of it as a change in attitude towards WFH in the sense that there used to be some social stigma attached to it, which has now decreased once many of us had to try it during the pandemic.
I simulated the model after the WFH shock and found that equilibrium house prices rose everywhere in the city, and that the increase was greater in the suburbs. This highlights the direct link between WFH and the change in housing demand, particularly the rise in demand for space and the decline in the commuting penalty observed in the data.
The impact of WFH on inequality
This change in housing preferences also has direct consequences for workers in occupations that do not allow for remote work. These households, who are typically at the lower end of the income and wealth distribution[5], tend to rent or own cheaper properties in the suburbs. In the model, they are priced out of home ownership and pushed into renting due to rising suburban prices. Their home ownership rate decreased by four percentage points, with the decline concentrated in the suburbs.
Another result we see in the model is that the boom in WFH leads to what could be called a tele-premium: an additional benefit for those working in occupations where WFH is feasible. Remote work exacerbates inequality between occupations across several dimensions, including income, consumption, housing, and liquid wealth. For example, in the model’s baseline economy the housing wealth of those able to regularly WFH was approximately twice that of those who cannot, increasing to 2.5 times after the rise in remote work. Finally, the workers who cannot WFH experience welfare losses averaging nearly 3% of current consumption.
What can we do about it?
There are measures that could mitigate the impact of WFH on the housing market and its effects on inequality. For example, I used the model to explore a hypothetical policy that would facilitate office-to-apartment conversions in the city centre, thereby increasing land availability for residential use. In the model, the policy led to a decrease in house prices and rents in the city centre, while significantly dampening price increases in the suburbs. This resulted in a significant reduction in welfare losses for those who cannot work from home. The model I developed can thus can also help to find solutions to the problems it identifies.
The views expressed in each blog entry are those of the author(s) and do not necessarily represent the views of the European Central Bank and the Eurosystem.
Check out The ECB Blog and subscribe for future posts.
For topics relating to banking supervision, why not have a look at The Supervision Blog?
The real estate data used for this project are at the property level, and provide a mapping between house prices and rents and detailed dwelling characteristics. They capture the universe of residential properties sold in the United Kingdom since 1995, as well as properties available for rent on the Zoopla website between 2012 and 2021 for England and Wales.
The city centre in this analysis is the Bank of England building located at the heart of the City of London. The results for Chancery Lane were similar.
The drop in rents that occurred in 2020 is most likely due to the lockdowns and people leaving their central London apartments during the peak of the pandemic. This trend began to reverse in early 2021.
Controlling for rich property characteristics (such as property type, whether the house is newly built, whether the house is a leasehold, energy efficiency, etc.) as well as month and neighbourhood fixed effects.
I estimate occupation-specific income processes with microdata from the Annual Survey of Hours and Earnings (ASHE). In this dataset, workers employed in occupations where WFH is not feasible have lower earnings than workers in WFH-friendly occupations.