- We think that the enforced remote-working experiment of recent months will cause a dramatic demand shift in the office sector, with as many as 50% of office-based employees working from home at least once a week. Even with a heroic supply response through substantial conversions and demolitions, we expect vacancy to rise markedly in the next five years and still be elevated in 2030. A combination of reduced occupancy and lower rental values will mean that income returns in an average office portfolio are likely to be down by more than 20% by 2025, only some of which will have been regained by 2030.
- Employment that has historically been viewed as office-based will take a hit in the short-term, but its fall will be smaller than for the whole economy and in the years post-COVID, we expect it to grow at a faster rate. Compared to 2019, we expect office-based jobs to be up by over 10% in the US, 7% in the UK and 4% in the euro-zone by 2030. But the problem is, much of this growth will not take place in the office.
- Prior to COVID, the share of remote-working in office-based sectors was close to 20%. That figure has spiked temporarily to around 70% this year. We think that by 2025, up to 50% of office workers will work remotely at least once a week.
- The implied fall in the occupied stock over the next five years will be dramatic, even assuming no further decline in average space per worker, with firms scrambling to adjust their physical footprints as soon as lease events allow. In the US, we expect occupied stock to be down by roughly 7%-8% by 2025.
- This will drive vacancy sharply higher by 2025. Indeed, even if there is a strong supply response, with large swathes of obsolete office inventory being converted or demolished in the second half of the decade, vacancy would still be above its 2019 level by the end of the decade.
- Rising vacancy will hit rents. In the US, we could see asking rents 10% lower by the end of 2025 and net effectives down by 15%, before staging a decent recovery thereafter. Despite the improvement later in the period, asking rents would only just return to their 2019 levels at the end of the decade.
- The combined effect of higher vacancy and lower rents will be to thump portfolio income returns, which could be down by 20%-25% by 2025. Not all of this would be recouped by 2030, leaving office investors looking at a rough decade.
- While our approach has been US-focused, we would expect office rents in the UK and mainland Europe to be similarly affected, though the precise magnitude of this impact will vary according to individual market dynamics. We will explore this further in future research.
- This is the first publication in our Future of Property series, considering the long-term outlook for real estate. If you would like to learn more about the series, please contact your account manager or sales executive.
Remote working to hit office income by 20-25%
The property press and the wider financial press have been full of stories questioning whether COVID spells the end of the office as we know it. There have been similar apparent watershed moments before. Indeed, the end of the office and a switch to remote and flexible working was a major theme following the Dotcom crash, but, so far at least, that has not come true. Could COVID-19 finally be the straw that breaks the camel’s back?
In this Focus we explore the outlook for office space demand in the context of the sharp rise in remote-working brought about by COVID, as well as how and where firms will occupy space in the longer-term. To that end, we assume that the threat of COVID-19 has dispersed within the next couple of years. We also draw on various sources from the major commercial property markets that we cover. However, our modelling focuses mostly on the US market, where generally the best time series data exist. Nevertheless, we expect office markets in the UK and mainland Europe to be similarly affected.
Our approach is to put firms’ office space requirements into the context of jobs growth, which will enable us to estimate how much actual space could be needed over the next decade.
Property markets are, of course, dynamic, which means that any change, or expected change, to demand should prompt a supply response. We will consider what this could look like and what these factors, when combined, could mean for office vacancy rates and rental growth rates in the next decade.
What is the outlook for office-based-employment?
To predict how much office space will be needed in future, the level of office-based employment is integral. In fact, in the past, service job growth has been strong enough to off-set more efficient and effective use of space, driving the demand for office space higher in most locations. But this may not necessarily be the case in future.
Our recent Update detailed our office-based employment projections for the UK, US and euro-zone. In the short-run, we expect a sharp drop in total employment, particularly in the US, where there is less fiscal support in place to prevent sharp increases in job losses. However, the nature of this downturn means that, for once, we think that office-based employment will outperform in the downturn. After all, the nature of the virus containment measures mean that the biggest economic hits will be to the retail and hospitality sectors, whereas many office-based industries are continuing at close to normal capacity as they are able to work from home.
Further ahead and post-COVID, we expect the continued rise of the service sector in developed economies to mean that office-based employment outgrows total employment, albeit by a smaller margin than in the past. Nevertheless, this means that in 10 years’ time, the number of office-type jobs will be larger than in 2019 in all three major markets. Indeed, we expect the number of US jobs to be 10% higher by 2030. (See Chart 1.)
Chart 1: Cumulative Change in Employment Compared to 2019 (%)
Source: Capital Economics
The key question then is, will this be enough to keep demand for office space growing?
How had office use changed prior to the virus?
Many offices today tend to be open-plan. This does vary by sector – for example, lawyers and accountants will tend to need more privacy and so will maintain a higher proportion of personal offices.
However, in the last 50 years or so, offices have changed considerably. Those changes and their timings vary by market, but can be broadly characterised as follows:
- Mostly personal cellular offices, promoting privacy and quiet working environments.
- Gradual switch to open-plan dominated space, often with six, eight or more desks grouped together, with the intention of supporting collaboration and team-based working.
- The latest trend in office space use is known as activity-based working. The idea being to provide flexible space, offering workers space appropriate to their needs at that time. For example, quiet reading spaces, collaborative space to support idea sharing and separate rooms for private calls or meetings.
The recent historical trend has been reinforced by a move towards more flexibility in working practices, with remote working and use of hot-desking growing. Even without the virus outbreak, it is likely that this trend would have continued. But the key question for our analysis is whether the virus may accelerate this trend or change its maximum level.
In particular, by demonstrating that office-based sectors can work remotely with relative success, lockdowns have highlighted that, along with flexibility of working arrangements within an office, there is also the possibility of much greater flexibility of location.
Yet, pre-COVID, a relatively low proportion of office workers spent most or all of their time working remotely, even though the technology has been in place for some time. For example, the share of employed people in the European Union who usually worked from home had been fairly stable at around 5% since 2008. That said, the share who sometimes work from home has trended upward over this period to 11% in 2019.
In the US, there has been a trend of increasing time spent working from home, with Census data showing that 9.5% of all workers worked from home at least one day a week by 2010. (See Chart 2.)
Chart 2: Share of US Workers Who Work From Home At Least One Day Per Week (%)
Source: US Census Bureau
While the latest Census is being conducted this year, we also have more timely data from the American Community Survey, which is conducted annually. This survey asks for the main location of work in the week prior to the survey. The proportion answering “work from home” has shown steady growth over time. (See Chart 3.)
Chart 3: Share of Workers Whose Primary Location of Work Was Their Home (%)
Source: American Community Survey
Although this series shows a lower level of home-working than the Census data, it clearly points to a continuation of the upward trend. Nevertheless, even in a world without COVID, assuming the Census data had continued to trend up as in the 2005-10 period, only 12%-13% would be working from home. In short, this suggests that only a small proportion of workers were consistently working from home prior to the virus.
What’s happened in 2020 and will it continue?
The immediate impact of the virus changed this. European surveys show a dramatic shift in the share of those working from home. This shift is particularly pronounced for “white-collar” jobs, which is a good proxy for office-based roles, with the majority of such workers in many of Europe’s biggest economies working from home on a regular basis in the Spring and Summer months. (See Chart 4.) This isn’t surprising. A March 2020 survey by YouGov in the UK found a far higher proportion of workers in office-based sectors felt they could work from home if needed against a UK-wide average of 40% of workers who believed this. For example, 73% of those working in finance and 78% of those in IT said they could work from home if needed.
Chart 4: Survey of White Collar Workers (June 2020, % of Respondents)
Sources: Cass Business School, IESE Business School, SD Worx
Clearly these have been extraordinary times. But home working also aligns with some employee and corporate preferences. Indeed, large tech companies including Twitter and Facebook have said they will allow some of their employees to work from home ‘forever’ if they choose. But perhaps more significant is that other non-tech corporates are also reviewing their office portfolios. And most surveys suggest that, for now at least, working from home has not materially impeded productivity.
There are some potential mitigating factors though. For one, not everyone can work from home. For example, in certain office-based industries, data security requires work to only be undertaken in one location, while for others, proximity to exchanges is vital. Indeed, some finance firms have referred to concerns about compliance monitoring for remote workers. A central space also provides benefits for collaboration and training and for functions which require more direct client interaction.
Indeed, research from CBRE on New York City shows that over the past three years, even companies which function mostly online, such as distance learning or cloud computing, have still been leasing office space. This is consistent with the 2020 CBRE Global Occupier Survey, which highlighted that around 60% of employers prefer to adopt a hybrid working approach between home and the office, compared to just 25% preferring a full-time remote model. As a result, we think that most companies will continue to require at least some office space.
In addition, not all workers will want to work from home. Remote working may not be practical for those with limited space or with other family members at home to effectively balance work and life. And a 2013 study by Bloom, Liang, Roberts and Ying found that, after a nine-month period in which workers at a major Chinese travel agency worked from home, around 50% chose to go back to the office. That was despite working from home increasing productivity by 13% (and therefore earning bigger bonuses) and generating greater happiness (measured by a lower rate of employee attrition). Clearly there are reasons why workers want to spend time in the office.
Finally, more remote working does not necessarily translate into lower demand. If more firms pursue an activity-based working set-up, they may require more space per worker, to provide the range and quantity of space required for meetings, collaboration and quiet work.
That said, remote working also means that there are more choices about where offices could be located. Recent survey evidence suggests that many companies are considering alternative portfolio make-ups, including using flexible office space and re-considering whether all of their space necessarily needs to be in the CBD. (See box on next page.)
Half of the office-based workforce could be working remotely by 2030
What does this all mean for the future? Most of the pros and cons of working from home were already known, at least in theory. But we think the last six months have made many who were formerly doubtful and were real estate decision-makers believe that at least some degree of remote working could be a net benefit to both their business and their employees.
Therefore, while we don’t expect all workers who are currently at home to stay at home forever, we think it’s reasonable to assume that we will not go back to the level of remote working seen in 2019. In short, the last six months have been a live experiment that has broken down some of the barriers to a more remote workforce. Indeed, CBRE’s Future of Work survey shows a sharp uptick in firms likely to be offering remote work in future. (See Chart 5.)
Chart 5: Past versus Future of Remote Work (%)
Based on survey data from across Europe showing that 60%-70% of workers were working remotely at the height of the crisis, we have assumed a similar rate in the US this year. However, we expect this to drop back when conditions become safe to do so. But, even by the end of 2022, we think that it is possible a total of 50% of workers will be working remotely in some form, with a little over 20% working at home full-time and 30% doing so for at least one day per week. There is then only a slow rise in those working at home one day a week over the remainder of the decade, whereas we think that 20% could be the ceiling for those working at home full-time, at least for the next decade. This reflects the benefits of the office environment, although we acknowledge there is huge uncertainty around this split. (See Chart 6.)
Chart 6: Share of US Office-Based Workers Working At Home (%)
Sources: US Census, Capital Economics
In terms of our modelling, that split is important. While full-time working from home should require no office space per worker, those who work from home part-time still require a proportion of their usual space. Indeed, if workers are at home just one day a week, firms may not seek to reduce their physical footprints. But for those working more than one day a week at home, firms are able to use hot-desking to increase the staff ratio per desk to above 1 (more staff than physical desks), perhaps between 1.2-1.5 per desk, depending on how much remote working the firm adopts.
Could out-of-town offices return to fashion?
There has been a fair amount of discussion recently around office-based firms moving to more of a “hub and spoke” model. This would presumably take the form of a smaller-than-at-present main, centrally located office (the hub), along with a handful of smaller satellites in more suburban locations (the spokes).
Survey-based evidence suggests that around a quarter of firms are considering this at present, presumably spurred on by concerns around commuting into densely populated cities by mass transit. But the evidence also suggests that more firms are considering the use of flexible space (which could equally be centrally located).
We will be considering the potential of such models in further research, but it is important to note that many of the arguments against out-of-town locations haven’t changed. These locations often have poor amenities and a rather mundane environment for workers. What’s more, office park locations are often heavily car-oriented. Indeed, as remote working has grown in recent years, such a model has fallen out of favour. After all, those working remotely often do so because they prefer the flexibility and comfort of working at home rather than having to travel to an office location and complete their work at fixed hours. I.e. This model seems to work for occupiers (it tends to be cheaper), but not for staff (car-reliant with limited lunchtime amenity).
Of course, although the trigger for considering a “hub and spoke” model currently is the concerns brought about by the virus, that’s not the only consideration. After all, firms may recognise that they can save on central space – and therefore cost – and still provide effective workspace for their employees in suburban locations. And the cost-saving incentives should never be ignored.
However, whether there is a sufficient case to then pay for suburban space or whether working from home better ticks the boxes for both employers and their staff remains to be seen. For now though, we don’t think it will materially affect the results of this Focus, as we are more interested in gross office space demand. Look out for an upcoming Update on this topic.
What does this mean for office space?
To recap, office-based employment is set to grow over the next decade, albeit at a slower rate than in previous cycles. At the same time, we expect to see a substantial increase in working from home. What might the net effect of these countervailing forces be?
To answer that question, we need to estimate the likely space per worker and how it may change over the coming decade. The level of space per worker varies by country and city, but for the purposes of this analysis we focus on the geography for which we have the best time series for both office-based workers and occupied floorspace. That is the US, specifically the major six cities that we forecast – Boston, Chicago, Washington D.C., Los Angeles, New York City and San Francisco.
Chart 7 shows movements in space per worker in both those six cities and at the national level.
Chart 7: Occupied Stock per Office Worker in the Major Six US Office Markets (sq ft)
Source: Capital Economics
It’s notable that the two series have substantial differences in their levels. However, this reflects the fact that our occupied space data for the US is technically not a national figure, as it includes 82 US markets rather than every single square foot of office space, so there is a mismatch with the office-based jobs figure we use. Nevertheless, the direction of travel is at least consistent in the last decade.
Given this, the more interesting and relevant take-ways are:
- Floorspace per worker has risen in all three downturns shown – the early 1990s, early 2000s and late 2000s. That’s not a surprise. Employment falls relatively fast during a downturn, while firms either hoard space as they expect to need it or are constrained in any attempt to reduce space by ongoing leases.
- Floorspace per worker fell in the 1990s. This is likely to reflect more efficient space use, with the shift to more open-plan offices in this period.
- There was a more pronounced fall in occupied space per worker between 2009 and 2019 than in the previous cycle. While this is likely to be partly reversed by the current recession, this decline seems to be driven by more than just the employment cycle.
We think there are probably two key factors that drove the pronounced fall in the last decade. First, the gradual growth in remote working would have enabled some firms to reduce their occupied space, either because more workers simply worked from home or because of a switch to hot-desking, which enabled the greater utilisation of space. Second, more recently the rapid expansion of flexible office space is likely to have played a role simply because that space tends to be more densely occupied than existing CBD office space. Reports cite a space per worker of between 60 and 100 sq ft in some flexible space.
Given Chart 7 also incorporates the effect of additional remote working and hot-desking, it’s likely that physical space per desk is higher than these results. For our modelling, we have estimated that a physical desk occupies around 200 sq ft, roughly the level seen in the big six cities in the first decade of this century. Translating that to our national series implies a figure of around 120 sq ft to be used alongside our estimates on workers.
As noted previously, our best guess is that there will be a substantial rise in working from home in the coming years (albeit a fall from the immediate COVID-driven levels). In the US, by 2030, the figures would equate to 8 million office-based workers working at home full-time and a further 11.6 million doing so part-time.
To convert that increase in working from home to occupier demand, we need to make a further assumption with our space per worker estimates. For those working permanently from home, we assume a zero allocation of space. For those working from home part-time, we assume that the space is used more intensively, so the figure of 120 sq ft per worker is reduced by 20%. It is normal for firms to keep capacity above its implied minimum as it is unlikely that occupancy will be uniform across the days of the week or indeed weeks of the year. Therefore that 20% reduction is less than the mean number of days spent working from home.
Our estimates for the required volume of office space are shown in Chart 8.
Chart 8: Physically Required Floorspace Over Time
Source: Capital Economics
Note that these are not the same as the actual quantity of space that will be leased and theoretically occupied at any point in time. On the face of it, these projections out to 2030 imply that the quantity of required floorspace will be 5.5%, or 184 million sq ft, lower than what was required in 2019.
What could impact the speed of adjustment?
But the 184 million sq ft will not disappear overnight. For a start, tenants cannot necessarily vacate their office space as soon as they wish to. Real estate’s leasing structures mean that adjustment would take time. After all, we have previously noted that the biggest risk to office occupancy this cycle isn’t from bankruptcies, but is instead from an increase in remote-working. (See our US Commercial Property Outlook.) That means that tenants will only be able to vacate or reduce space when they have a lease event, such as a break or a lease expiry. Those lease lengths vary by country. (See Chart 9.)
Chart 9: Average Office Lease Lengths (Weighted by Rental Value)
Sources: CBRE, MSCI, Capital Economics
So while roughly one-tenth of office space in the UK has a lease expiry in any given year, the equivalent figure is more like one-eighth in the US, meaning that we should expect a slightly more gradual adjustment in the UK. Of course, break clauses also need to be factored in. With close to 45% of leases in the UK having a break clause (and these more common in longer and larger leases), we expect that lease events could apply to a fifth of space each year, meaning that the majority of any adjustment required should be complete by 2025. In the US, if a similar proportion of leases feature break clauses, it is likely to happen faster.
A second factor that could affect the adjustment, is how long new leases are signed for. Chart 10 shows estimates for the different amounts of space required when tenants commit to leases. Space per worker tends to be smaller for shorter leases as there is no need to plan for future expansion.
Chart 10: Average Space Leased per Worker in the US by Length of Lease (Sq Ft)
Source: Norm G. Miller, ERES 2013
And we know that in a downturn, lease lengths tend to fall as firms are uncertain about the future. (See Chart 11.) This means that firms are more likely to choose those shorter lease options when negotiating their next lease.
Chart 11: Average Unweighted US Office Lease Term (Four-Quarter Moving Average)*
Source: CBRE *Shaded areas represent US recessions
This effect could be especially pronounced in this cycle because of the substantial increase in flexible office space. Before the outbreak, the traditional office was the main location for most workers. There had been some growth in the use of flexible office space, but this space still only accounted for a small share of total office stock, even in the fastest growing markets. In Europe for example, the highest share of flexible space is in London and Amsterdam, but even in those cities, it accounts for only 5-6% of total office space. What’s more, its growth had started to slow in recent years and the big providers had clearly overstretched.
As noted, the 5.5% reduction in required floorspace that we have identified will not happen overnight. Incorporating the two factors noted above, we have spread out the major structural shift in our modelled US figures through to 2025 and reduced the excess space held by firms in this cycle. (See Chart 12.)
Chart 12: Required and Occupied Floorspace in the US (Billion Sq Ft)
Sources: REIS, Capital Economics
Could there be a supply response?
In order to examine the potential effect this might have on vacancy, we have to make assumptions about supply. For now, we input our latest forecasts for inventory growth through to 2024 and then assume a gradual trend higher in stock, of 0.5% per year from 2025 to 2030. On this basis, our demand projections from Chart 12 would point to a reduction in leased space of 7.7%. This would push up the vacancy rate from 16.8% at the end of 2019 to 26% in 2025. There would then be a gradual trickle higher by 2030, to just over 28%, a new record high.
Given that the likelihood of developing profitably will be much lower if vacancy rises as we are predicting, this would imply a lower level of supply growth than initially projected. Indeed, at these kinds of vacancy levels, it’s likely that development activity would come to a halt, at least in some markets. Indeed, we have previously noted that office space under construction has been more subdued in recent years than in previous cycles. (See our Update.) That being said, there would only be a small reduction to our already-lowered completions forecasts for the next few years, where much of the work is already underway.
What’s more, over time, as the probability of certain types of space finding tenants falls, more and more office stock is likely to become obsolete. Indeed, the rate of obsolescence will be driven not just by a reduction in demand, but also by any new supply that is built absorbing any demand growth in the market. After all, we know that developers’ concern is whether they can attract tenants to their developments and generate an appropriate profit for their perceived risk. Instead, the owners of older buildings or poorly located buildings will be the ones who lose out, as it may not be possible to modernise or re-develop profitably when tenants leave.
As a result, as the decade progresses, we would expect to see a wave of conversions of old office space to alternative uses or perhaps even outright demolitions. Of course, it’s difficult at this stage to know what the most viable (i.e. profitable) alternative uses might be, given the potential impact of the rise in remote working on the other property sectors. For example, if more households move to the suburbs thanks to greater working from home, there may not be much demand to convert offices to apartments in the cities. (The next publication in our Future of Property series considers the outlook for the residential sector.)
For an idea of how fast that change could occur, the recent experience of Amsterdam in the Netherlands could help. The vacancy rate rose from around 5% in 2000 to 18% at the height of the GFC, but was as low as 3.7% at the end of 2019. This was partly driven by a temporary ban on new development from 2010 to 2017, but the greater factor was the municipal support for conversions to alternative uses. As a result of these changes of use, office inventory fell by 9% in the seven-year period between 2012 and 2019.
It’s unlikely that we would see such a substantial fall across a whole country given the same circumstances, but if our predictions for the demand side of the equation are even in the right ballpark, the problems facing office landlords are astronomical. An adjustment of similar magnitude would be possible in a similarly land-constrained market, but most of those in the US are simply not constrained in the same way. Nevertheless, we have pencilled in an outright fall in office inventory in the 2024-2030 period of 3%. (See Chart 13.)
Chart 13: Change in US Office Inventory (% y/y)
Sources: REIS, Capital Economics
Chart 14 pulls together the demand and supply sides of our equation and their impact on vacancy.
Chart 14: US Office Demand and Supply
Sources: REIS, Capital Economics
Rents are set to decline markedly in the next five years
In Chart 14, the modelled increase in vacancy is from just under 17% in 2019 to nearly 25% in 2025. In magnitude, this would be close to that seen following the Dotcom crash in the early 2000s. However, it is important to note that while the rise in vacancy in the early 2000s was due to huge numbers of firms going bust and vacating space in a very short period, this time firms are, on the whole, operating reasonably well. The rate of change in this case will be limited by the leasing structures in place, as noted earlier. Therefore, the rise in vacancy and the related fall in rents is likely to be more gradual, with vacancy rising in each of the next five years and rents falling for even longer. (See Chart 15.)
Chart 15: US Office Vacancy and Asking Rents
Sources: REIS, Capital Economics
Although not shown explicitly in this chart, the rise in obsolescence that we have previously noted would tend to mean a higher level of structural vacancy. So we would expect rental growth to kick in at a higher level of vacancy than in the past.
The rental decline shown in Chart 15 totals 10% and would take average asking rents in the US down to a level last seen in early 2015. Only in 2030 would rents again reach their 2019 levels. The picture is bleaker still for net effective rents, which we expect to fall by a cumulative 15% and still be below their 2019 level at the end of the decade. (See Chart 16.)
Putting this in context, zero change in asking rents in the 2019-2030 period would compare with over 2% p.a. in the 2010-2019 period. Indeed, even in a period spanning two huge office market crashes, from 2000 to 2019, asking rents grew by 1.3% p.a.
Chart 16: US Office Rents (Index, End 2019 = 100)
Sources: REIS, Capital Economics
The assumptions that underlie the forecasts in this Focus face a high degree of uncertainty, particularly as they span a 10-year period and are based on predicting changes in employer and employee behaviour, which is always a tough task. However, we think that on balance, they provide a fair picture of the likely path of office demand and supply in the US and the implications of those on occupancy and rental levels. And we believe that these findings can be applied more widely to all developed markets.
While one takeaway is that asking rents should return to their 2019 levels by the end of the period, this would be at a lower level of market occupancy, meaning that the average landlords’ office portfolio will still have seen a fall in its net operating income, which could still be 5%-10% lower, even by 2030. This reflects the fact that there will be winners and losers. As noted earlier in the text, some developers will continue to develop profitably, even while inventory is contracting. On the other hand, owners of poorly located or badly configured space, are likely to lose out.
Due to data availability, the specific conclusions of this piece relate to the US market. And it is important to note that although we expect a similar change in office demand in other developed markets, the assumptions made will not necessarily hold in exactly the same way in other markets. What’s more, the starting points of different markets will vary, both in terms of their existing level of remote-working, the potential growth in remote-working (which can be affected by a city’s industrial structure) and the general demand and supply balance of those markets. We will be exploring further the outlook for different cities in all of the commercial real estate markets that we cover in upcoming work.
Kiran Raichura, Senior Property Economist, +44 (0)7739 932 077, email@example.com
Prohad Khan, Property Economist, firstname.lastname@example.org
Amy Wood, Property Economist, email@example.com
James Yeatman, Research Assistant, firstname.lastname@example.org