Impact of working from home on European office rents and vacancy rates

One of the most visible impacts of the Covid-19 pandemic was the change in the daily routines of millions of office workers around the world. The lockdowns implemented in most countries resulted in a sudden shift of corporate activity from offices to employees’ homes, with office buildings either legally inaccessible or companies choosing this option in accordance with the official guidelines and to prioritise employee well-being. The unprecedented scale and the relative success of the global “working-from-home experiment” raised questions about a potential structural shift in the office sector. It has been argued that working from home may become a permanent solution for a large share of office occupiers, leading to collapsing demand for office space (Eavis and Haag 2021; Egan 2020). The discussion was further fuelled by announcements by some major companies regarding specific or potential plans to allow a large share of their employees to work from home on a regular basis (Deutsche Bank 2021; Kelly 2020; Microsoft 2020; Siemens 2020). While there is still a lot of scepticism about the long-term efficiency of homeworking solutions, if the trend continued, it could lead to a significant reduction of the demand for office space with the consequence of increasing vacancy, obsolescence and falling rents.

Amid the ongoing discussions, it is important to note that delivering office work from home is not a new idea. The phenomenon of shifting from a pure office environment to various modes of remote working has been ongoing for at least 20 years. Since corporate communication moved from paper to email, documents became instantly accessible online, and connection speeds and security improved, deep changes to the office working model have been inevitable. Even though the shift towards remote work was not as fast and revolutionary as some early predictions assumed, it was happening long before the pandemic.

It is still too early to indicate the direction of travel, and the outcome will depend on the evaluation of the lockdown experiences and subsequent corporate decisions in the coming years. However, even though an increase in working from home appears likely, its impact on the demand for office space is a separate question. Would office space occupied by a company decrease proportionally if a percentage of its staff worked from a different location? Recent discussions show that there is no consensus on this point. Some predictions, especially the early ones, indeed assume that offices will be used less, implying negative consequences for office markets (Cutter 2021; Nixey 2020). Others point out that the need for face-to-face communication remains, and it can be provided most efficiently in offices. Hence, even if some of the work was done from home, it would not have a major impact on the total amount of office space used by companies but merely change its character (Kröger 2021; Moran 2020; Schede 2021).

Against this background, the paper addresses the question of the sensitivity of office occupier markets to the share of employees working from home. However, the purpose is not to predict the outcome of the post-pandemic processes, but rather to look back and investigate, whether the past (slow) changes in the scale of homeworking had any measurable impact on the performance of office markets or allowed to explain international variation in market indicators. While the experience of the lockdowns was unique and may constitute a structural break that makes it difficult to project past experiences into the future, the historical evolution of markets can still contain some indications on the nature of the relationship between homeworking and aggregate office demand.

The issue can be approached from two distinct perspectives: the first one focusing on the evolution over time; the second one looking at cross-sectional differences between markets. Hence, two research questions arise:

The paper is organised as follows: Next section reviews the literature on the topic. It is followed by the discussion of pre-Covid trends with respect to working from home. Sect. 4 addresses the data and the methodology of the panel regression, while the results are presented in Sect. 5. The final section is devoted to the interpretation of the results and concluding remarks.

Literature on working from home is extensive and stretches over several decades. The early papers in the 1980s pointed out the possibilities of remote work that were arising from the introduction of personal computers to offices and homes (Olson and Primps 1984; Pratt 1984; Zuboff 1982). The subsequent main string of research focused on the work-life-balance and improvements to the well-being of employees. Numerous studies indicated that enabling workers to work from home could result in higher job satisfaction and improved productivity. This was, for example, the conclusion of the study by Bailyn (1989) conducted on a sample of UK computer firms. Numerous other studies reached similar conclusions, highlighting positive aspects of workplace flexibility (Bloom et al. 2013; Clancy 2020; Clark 2001; Duxbury et al. 1998; Fonner and Roloff 2010; Gajendran and Harrison 2007; Hill et al. 2003). On the other hand, various papers highlighted potential psychological issues arising from working from home, mostly related to the feeling of isolation and social detachment, lack of separation between work and private life, distraction and decreased commitment (Golden 2006; Harpaz 2002; Salomon and Salomon 1984). Consequently, an important research question was about the balance of advantages and disadvantages to individuals and companies and its dependence on various factors, in particular the characteristics of the individuals and their roles in the company (Allen et al. 2015; Ammons and Markham 2004; Bailey and Kurland 2002; Dockery and Bawa 2014; Golden and Veiga 2005; Harpaz 2002; Morganson et al. 2010; Olson 1983; Olson and Primps 1984). While most authors predicted a steady increase in working from home, the consensus was that it was a slow process that would stretch over years and decades, allowing sufficient time for demand and supply to adjust (Harris 2015).

Another, more recent, string of research focused on the assessments of the share of jobs that can be done from home. The widely cited paper by Dingel and Neiman (2020) uses the classification of occupations for that purpose; authors identify job characteristics that clearly rule out the possibility of working remotely. By combining it with US job statistics, they conclude that 37% of all jobs in the US could be done remotely; the share varies between 28% and 51% across US metropolitan areas. Another study by Adams-Prassl et al. (2020) goes further and investigates the tasks within occupations; it estimates the share of “remotable” jobs at 41% in the US and 39% in the UK. Other studies arrive to similar conclusions with the shares ranging mostly between 30% and 40% (Bartik et al. 2020; Cetrulo et al. 2020; Deng et al. 2020; Gottlieb et al. 2020; Holgersen et al. 2021; Lund et al. 2020).

Finally, a significant number of papers have been published in the recent months addressing the impact of the Covid-19 pandemic on working from home. Early studies were mainly descriptive, focussing on the scale of the phenomenon and the experiences of the employers and employees (Barrero et al. 2021; Brynjolfsson et al. 2020; Chung et al. 2020; Reuschke and Felstead 2020). The main research question was, however, if the “homeworking experiment” was a unique but temporary situation, or if working patterns were going to change permanently post-Covid. The vast majority of research appears to lean towards to latter thesis, with the consensus forming around a model with roughly 2 days per week spent at home and 3 days in the office (Alexander et al. 2021; Bloom 2020; Erdsiek 2021; Kamouri and Lister 2020; Kunze et al. 2020; Plößl and Just 2021; PwC 2021). This would be in line with the surveys conducted before the pandemic (Gallup 2017; Hood et al. 2018).

The other aspect addressed in this paper—performance of the rental markets for office space—received vast coverage in the literature in the past decades. From the occupier market perspective, (market) rent and vacancy rate are the key indicators determining the level of cashflows to owners. Numerous studies, both theoretical and empirical, demonstrated that these two indicators are closely linked: high vacancy creates a “tenant market” leading to weaker (negative) rent growth, while low vacancy is associated with a “landlord market” leading to stronger rent growth (Glascock et al. 1990; Hendershott et al. 1999; McDonald 2000; Wheaton and Torto 1988). Both rent growth and vacancy rate are therefore affected by similar supply and demand factors The latter include in particular GDP growth (D’Arcy et al. 1997; Gardiner and Henneberry 1989) and (office) employment (Dobson and Goddard 1992; Hendershott et al. 1999; Sivitanides 1997) as well as structural factors such as office space per employee (Miller 2014). It is through the employment variables, that a link between the performance of office markets and working from home can be suspected.

In the context of the potentially higher level of homeworking in the future, the long-term impact on the aggregate demand for office space and the performance of office markets continues to receive particular attention among practitioners. Most statements in the press so far reflect subjective views of real estate market participants, partially backed by anecdotal evidence. More rigorous analyses have been relatively few, focusing on the identification of office markets considered to be most at risk (DWS 2020) or on the design of a future office (Boland et al. 2020; JLL 2020). While several academic papers address the direct impact of the Covid-19 pandemic on commercial real estate values and returns (Balemi et al. 2021; Ling et al. 2020; Milcheva 2021; Voigtländer 2020), academic studies looking specifically into the long-term impact of increased working from home on office demand are still missing. This paper contributes to filling this gap.

As of 2021, it is still too early to judge what the future will bring for the office sector. However, looking back and analysing the past trends in homeworking can provide insights into the mechanisms involved in these processes. This section offers an overview of the state of homeworking in Europe in the decades preceding the Covid-19 pandemic.

Eurostat reports statistics on the share of employed persons working from home as a percentage of total employment (Eurostat 2021). This data is available annually since 1992 for 35 countries with some gaps especially in the early years. The dataset differentiates between those working from home usually, occasionally, and never. Based on this data, the share of employees (excluding self-employed) working occasionally from home in Europe began to increase around the turn of the century and has more than doubled from 3.3% in 2002 to around nearly 8% in 2020. At the same time, the share of employees working usually from home moved from 1.6% in 2002 to around 3% in 2019 and over 10% in 2020, reflecting a massive shift to working from home during the Covid-19 pandemic. These figures refer to all employees; the shares are obviously significantly higher for office workers compared to other jobs that require physical presence at the workplace. A global survey by GWA (Kamouri and Lister 2020) indicated that only 29% of office workers never worked from home. On the other hand, 9% worked from home 5 days a week, 22% worked from home 1–4 days a week, and 40% worked from home only occasionally. This mirrors the fact that most companies have been offering the option to work from home for some time now—more than 80% of companies surveyed in the Alternative Workplace Strategies Report (Hood et al. 2018) and virtually 100% of large enterprises. Hence, while full-time home working was rare, working from home on an occasional basis was reasonably widespread even before Covid-19.

However, there was huge variation across countries as depicted in Fig. 1. The share of occasional home workers in 2019 ranged between less than 3% in Italy or Spain, to over 25% in the Netherlands or Sweden (referring to all employees). The gap between the top and the bottom ranking countries has widened considerably during the past 20 years. While the share of home-workers remained largely unchanged in Italy and Spain, it grew from 6% to 13% in France, and 3% to 30% in Sweden.

The significant differences in the adoption of remote office work across Europe are not easy to explain. It seems reasonable to assume that the technical possibilities are largely comparable in all countries. Hence, the differences appear to be rooted in the local corporate policies, cultural factors and personal situations of the population. It is striking that the share of home-workers tends to be higher in northern Europe than in southern Europe. This may correlate with the more open and communicative culture of southern Europeans. Another reason might be the availability of suitable workspace at home. According to Eurostat, 66% of Spanish and 52% of Italians live in apartments rather than in houses; the respective share is 33% in France and 19% in the Netherlands. According to Euroconstruct, the average dwelling area per inhabitant is 33 sqm in Spain and 37 sqm in Italy compared with 43 sqm in the Netherlands and 45 sqm in Sweden (Pajakkala 2018). A combination of these factors possibly limits the ability and desirability to work from home in southern Europe.

The analysis consists of two parts, the first part focusing on temporal effects and the second part focussing on cross-sectional effects. Referring to the research questions formulated in Sect. 1, the parts address the first and the second question, respectively. Since the paper looks at two office market indicators, rent growth and vacancy rate, two variants of each research question result. Operationalising them yields following four questions to be answered with the panel regression: