Employee intentions and employer expectations: a mixed-methods systematic review of “post-COVID” intentions to work from home

ABSTRACT

The COVID-19 pandemic accelerated cultural and organisational acceptance of remote working. For a portion of the commuting workforce, working from home (WFH) is now possible. Of great interest is whether increased WFH will diminish actual mobility, and thereby reduce the transport task of cities. To understand this possibility, we must know how much WFH will be sustained into the future. Using a bespoke approach combining scholarly and grey literature, this review develops a tangible record of employee desires and intentions to WFH, in the context of the expectations of employers. Its contribution is a novel and rigorous appraisal of recent practices and sentiments. Results confirm that there is a strong underlying demand to WFH. Many studies, however, estimate unrealistically high rates of WFH which cannot be projected onto the wider working population. Further, we find there is a conflict between employee preferences and their expectations to WFH, with estimations of preferences far greater than estimates of expectations. This finding is confirmed by the analysis of employer sentiments. Employers broadly realise that accommodating WFH reflects a best-practice approach, yet favour predictable routines where specific days of on-site attendance are mandated. We conclude with reflections on the impact of our findings on the transport system. We propose that the impact of WFH on commuter decision-making depends on the degree to which employers mandate on-site attendance. Finally, we emphasise the need to acknowledge the wider political, economic and social milieu in which work is performed as shaping future WFH practice.

KEYWORDS:

• Work from home
• Covid-19
• Telecommute
• Mixed-methods review
• Journey to work

Introduction

COVID-19 is the largest global disruption to urban economic and social activity in our lifetime. One of the most significant shifts has been to ways of working, particularly the practice of telecommuting, remote working or working from home (WFH) (Aksoy et al., 2022; Chan et al., 2023). Provoked by the need to sustain productivity through periods of physical lock-down, WFH as a reality has now been established as a valuable and acceptable practice (Smite et al., 2023).

Preliminary studies during the pandemic suggest that in the long-term, many workers wish to continue some degree of WFH after the conclusion of lockdown periods (Aksoy et al., 2022). In this complex and emergent context, however, routines of commuting versus WFH are nascent and variable. In terms of the temporal and spatial distribution of work, for example, options range from full-time on-site working, to working onsite some days of the week, month or year, to WFH 100% of the time (Bloom et al., 2022). This is further complicated by different degrees of mandate and regulation imposed by employers, who are often overlooked as key actors in the way WFH will feature into the future. Some workers have full autonomy, choosing when and where they work; others are bound by on-site mandates on specific days for specific hours; still others must work on-site, either because of the nature of their job or their employer (Chatterjee et al., 2022).

Of great interest is whether increased WFH has the potential to diminish actual mobility and thereby reduce the transport task of cities. If transport planners are to build these new realities into long-term, strategic transport models, we must have an informed understanding of where, and to what extent, WFH will reduce the commute burden. Furthermore, we need to consider how WFH will reconfigure other daily practices and the travel associated with their performance.

Past research has explored the barriers and enablers to WFH (see for example Bailey & Kurland, 2002; Mokhtarian & Salomon, 1997), and the flow-on effects its practice might have on transport systems (Mokhtarian, 1990; Mokhtarian, 2002; Mokhtarian, 2009). After the shock of the pandemic years, however, we do not know how much, who, and where increases in telework will be sustained into the future. While the data may not be available to answer these questions directly, the time is right to take stock of existing understandings.

Using a bespoke approach combining scholarly and grey literature, this review develops a tangible record of employee desires and intentions to WFH, in the context of the policies and expectations of employers. Its contribution is an evidence-based understanding of how prevalent WFH might be into the future. Heading recent calls for transport scholars to better understand the practices facilitated by transport (Kent, 2021) (in this case, work), we propose this knowledge is the first step in the complex task of assessing the transport and land-use implications of this historically significant shift.

This paper first details our mixed-methods approach to reviewing relevant literature, including a discussion of the quality of data used in the scholarly publications included. We progress to discuss the quantified estimates of future WFH provided in the literature, as well as the characteristics and considerations of employers and employees with access to WFH. We conclude with conjectures as to the implications of WFH on the transport task of cities, as well as recommendations for future research.

Methods

This review paper aims to record employee desires and intentions to continue WFH, in the context of the policies and expectations of employers. We have approached our research question from the perspective of both employees and employees in recognition that employers have the final say on where work is performed, and their approach is a key determinant of the degree to which WFH will continue. We use a three-step, mixed-methods approach, combining a systematic review of peer reviewed literature with an exploratory narrative review of grey literature, which informed a meta-analysis of grey survey data.

Mixed method reviews are warranted in circumstances where the scholarly literature demonstrates gaps which can be filled through reference to other sources, particularly those unsuitable for systematic analysis (Sutton et al., 2019). The approach can be particularly useful when the subject under examination is unstable and shifting. In this case, the present-day volatility of global conditions warrants an exploration of research and sentiment from sources outside of the scholarly literature, characterised as it is by protracted publication timeframes. This approach has been applied recently in a review of the impact of COVID-19 on passenger transport (Shortall et al., 2022).

Each component of the review is explained in detail below. The review was conducted sequentially, so that the findings of the systematic review could be explored in more detail by the exploratory review of grey literature. The results of the systematic review are presented in tabular format and also narrated through themes organised to accomplish the aim of the research. Where possible, this narrative component is augmented by reference to insights from the grey literature review. When discussing scholarly literature, we refer to “papers” and “studies” and when referring to grey sources we use the term “grey sources”. A meta-analysis of grey survey data located through the review of grey sources is included in the narration.

Systematic review of peer reviewed literature

The objective of the systematic search was to identify papers exploring future intentions to WFH “after” the COVID-19 pandemic. In the scholarly literature, this was variously defined as “when the virus is no longer a threat” (Conway et al., 2020, 3), “after the virus is gone” (Jain et al., 2022, 56) or “after everyone has been vaccinated” (Dianat et al., 2022, p. 299).

Scopus and Web of Science were utilised with the date range from 2020 through to the date the search was conducted (24 August 2022). The language was restricted to English. The following search terms were applied:

TITLE-ABS-KEY(covid* OR coronavirus OR pandemic)

AND

TITLE-ABS-KEY(work* AND from AND home OR telecommut* OR telework*)

AND

TITLE-ABS-KEY(transport* OR travel OR commut*)

Subject field restrictions were used to reduce irrelevant papers, as shown in Table 1. After excluding irrelevant subjects and papers that were not in English, 795 papers were exported into Covidence. Figure 1 presents the PRISMA diagram illustrating the screening process.

Figure 1. PRISMA diagram of systematic review process.

Table 1. Field restrictions utilised in systematic search.

	Initial search terms	In English only	Excluded subject fields	Results
Scopus	764	743	Medicine, Biochemistry, Nursing, Veterinary medicine, health professions	392
Web of Science (Core Collection)	455	444	Medicine, Nursing	403
Studies imported to Covidence				795

After removing 198 duplicates, the remaining 597 papers were screened based on title and abstract. Many of these papers looked at topics such as the impact of WFH on disease exposure, the experience of essential workers or other necessary commuters, telehealth or the downstream impacts of WFH on pollution or energy consumption.

The remaining 114 papers were sourced for full-text screening. Of these, 40% studied WFH during the pandemic years, but did not estimate the extent that WFH might continue in the future. Other papers were excluded because they were a literature review or conceptual model (with no original data presented, n = 18), they did not actually study WFH (n = 14), the paper was not available (n = 4), or they studied WFH before COVID-19 (n = 3). One record was an open-access dataset with no analysis included.

After full-text screening, 28 papers were identified as relevant to the systematic review. Examining the reference lists of these papers uncovered one additional study (Moens et al., 2022). We specifically chose not to screen further based on study quality, as the quality of study reporting ended up becoming one focus of our findings.

We note that as this is a rapidly-expanding field, additional studies have been published after the initial search date, but we restrict our focus to the systematic search findings.

Exploratory review of grey literature

Although some studies uncovered in the systematic search attempted to assess employer intentions towards WFH (for example, waves 2 and 3 of the Beck and Hensher research 2022a and 2022b), none of the studies in the systematic search focusses specifically on employer sentiment in the era post-lockdown. In recognition of this gap, we used the ProQuest ABI Inform database to source media articles and reports on employer intentions, sentiment and policies.

ProQuest’s ABI Inform Collection is a business research database combining content from ABI/INFORM Global, ABI/INFORM Dateline, and ABI/INFORM Trade and Industry. The content covers full-text business journals and periodicals, trade publications, market analyses and industry reports. It is thus a useful tool to source material relevant to employment and work.

The following search string was applied:

(“work from home” OR “hybrid work*” OR “telecommut*” OR “remote work”)

AND

“employer”

AND

(“policy” OR “policies”)

Results were limited to sources where this string appeared in fields other than full-text (for example, title, summary, key words).

We were interested in a snapshot of employer sentiment and policies in times when the option to return to work on-site was available. Results published prior to 2022 were therefore excluded. This resulted in 74 sources for review, which were used to develop a series of themes complementary to those extracted from the systematic phase of the search. In addition, 15 surveys of employer intentions were found through this process, and these surveys were accessed online. Eight surveys were deemed unsuitable because they did not publish key variables of interest, resulting in eight surveys of employer intentions for analysis, which is included in the narrative review.

Results of the systematic search

Twenty-nine papers were found through the search (see Table 2). Studies from Australia had the highest representation with nine papers; seven studies took place in North America and seven in Europe. The remaining studies took place in Vietnam, Saudi Arabia, Nigeria, Latin America, South Africa and “worldwide” through a global survey of transport experts. Two studies (Balbontin et al., 2021; Zhang et al., 2021) compared results from the populations of more than one country.

Table 2. Systematic search results.

Paper	Location	Journal IF quartile	Data collection date	Sample size	Sample population	Recruitment method	Excludes non-teleworkers?	Weighted data?	Sample demographics shown?	Sample compared to population?	Survey name
Ahmed & Khalil (2021)	Saudi Arabia	None	Not stated	300	Office workers who can WFH	Not stated	yes	no	yes	no	n/a
Appel-Meulenbroek et al. (2022)	Not stated	Q1	Not stated	218	Office workers of a commercial real estate company	Not stated	not stated	no	yes	no	n/a
Balbontin et al. (2021)	eight countries	Q1	2020 Aug–Nov	4,628	Employed persons	Mix of panel and convenience	no	no	yes	no	Includes Beck & Hensher
Barbour et al. (2021)	USA	None	2020 Jul–Aug	1,275	General population	Random population sample	no	no	yes	no	n/a
Beck & Hensher (2022b)	Australia	Q1	2020 May–Oct	2,413	General population (231 were employers)	Random population sample	no	no	yes	no	Beck & Hensher survey
Beck & Hensher (2022a)	Australia	Q1	2020 May–Oct	2,413	General population (231 were employers)	Random population sample	no	no	yes	no	Beck & Hensher survey
Beck & Hensher (2020)	Australia	Q1	2020 May–Jun	1,457	General population (some were employers)	Random population sample	no	no	yes	no	Beck & Hensher survey
Beck et al. (2020)	Australia	Q1	2020 Mar–Jun	1,180	Working population	Random population sample	no	no	yes	yes	Beck & Hensher survey
Ceccato et al. (2022)	Italy	Q1	2020 Jul–Sept	1,243	Employees of University of Padova	Convenience sample	no	no	yes	no	n/a
Conway et al. (2020)	USA	none	2020 Apr–Jun	1,595	General population	Mix of panel and convenience	no	no	yes	yes	COVID Future survey
Currie et al. (2021)	Melbourne	Q1	2020 Jun–Aug	2,163	Employed persons	Random population sample	no	yes	yes	no	C-19 Long Term Transport Impact Study
de Abreu e Silva (2022)	Lisbon	none	2021 Apr–Jul	532	Employed persons	Convenience sample	no	no	yes	yes	n/a
de Haas et al. (2020)	Netherlands	none	2020 Mar–Apr	2,800	All residents	Pre-existing longitudinal panel study	no	no	yes	yes	Netherlands Mobility Panel (MNP)
Dianat et al. (2022)	Toronto	Q1	2020 Jul	985	General population	Random population sample	no	yes	yes	yes	n/a
Hensher et al. (2021a)	Sydney	Q1	2020 May–Jun	409	Working population	Random population sample	no	no	yes	no	Beck & Hensher survey
Hensher et al. (2021b)	Sydney	Q1	2020 Mar–Jun	200	Commuters	Random population sample	no	no	yes	yes	Beck & Hensher survey
Hiselius & Arnfalk (2021)	Sweden	Q2	2020 May	719	Office workers of public agencies	Convenience sample	yes	no	yes	no	n/a
Jain et al. (2022)	Melbourne	Q1	2020 Jun–Aug	1,364	Employed persons	Random population sample	no	no	yes	no	C-19 Long Term Transport Impact Study
Kong et al. (2022)	Seattle	Q3	2020 Jun–Jul	874	Working population	Convenience sample	no	no	yes	no	UW COVID-19 travel behavior survey
Moens et al. (2022)	Flanders, Belgium	Q1	2020 Mar	2,673	Teleworkers	Post-stratification representative sampling	yes	yes^a	yes	no	n/a
Mogaji (2022)	Nigeria	Q1	2021 Jul–Aug	72	Employed persons	Not stated	no	n/a	yes	no	n/a
Mohammadi et al. (2022)	USA	Q3	2020 Apr- 2021 May	583	Employed persons who WFH	Mix of panel and convenience	yes	no	yes	no	COVID Future survey
Nayak & Pandit (2021)	India	Q1	2020 Mar and 2021 Mar	431	Employed persons	Convenience sample	yes	no	yes	no	n/a
Nguyen (2021)	Hanoi	Q1	2020 Mar–Apr	355	Teleworkers	Convenience sample	yes	no	yes	no	Nguyen survey
Nguyen & Armoogum (2021)	Hanoi	Q2	2020 March–Apr	355	Teleworkers	Convenience sample	yes	no	yes	no	Nguyen survey
Olde Kalter et al. (2021)	Netherlands	None	2020 Apr	1,515	Employed persons who did not WFH full-time before COVID	Pre-existing longitudinal panel study	no	no	yes	yes	Dutch Travel Panel (NVP)
Salon et al. (2021)	USA	Q1	2020 July–Oct	4,554	Employed adults	Mix of panel and convenience	no	yes	no	no	COVID Future survey
Ton et al. (2022)	Netherlands	Q1	2020 Apr and Jun	30,632	Employed persons	Pre-existing panel of train users	yes^b	yes	yes	yes	Dutch National Railways
Zhang et al. 2021)	Worldwide	Q1	2020 Apr–May	284	Transport experts	Convenience sample	n/a	no	no	no	n/a

^a Weighting was applied via post-stratification sampling.

^b Initial results included all respondents but most of the analyses only included teleworkers.

Almost without exception, surveys were conducted through online questionnaires. Only one study employed qualitative methods through interviews (Mogaji, 2022). Almost every study focussed on the perspective of the general public, employed persons or teleworkers. The two exceptions to this is a study of “transport experts” by Zhang et al. (2021), and three of the Beck & Hensher papers which included a sub-set of responses by employers/managers (Beck & Hensher, 2020; Beck & Hensher, 2022a; Beck & Hensher, 2022b). Nine papers excluded or did not analyse results for people who did not or could not WFH. This is understandable, as many papers were interested in factors influencing WFH. However, this means that any estimates of future WFH cannot be extrapolated to the working population unless we know what proportion of the population is able to WFH.

It was common for multiple papers to draw from the same dataset. Seven papers used a repeating cross-sectional panel survey overseen by Beck & Hensher of the University of Sydney (Australia). Three papers drew from the COVID Future Survey, jointly overseen by Arizona State University and University of Illinois Chicago with support from the National Science Foundation; these data are publicly available for download (Salon, 2023). Two papers used the C-19 Long Term Transport Impact Study, overseen by Currie from Monash University (Melbourne, Australia), and two studies drew from the same survey overseen by Nguyen from University of Transport and Communications, Hanoi, Vietnam.

Data quality and implications

We explicitly did not screen out studies based on the quality of the journal. Table 2 provides the journal impact factor quartile based on the JCR journal citation report.¹ In addition, key components of the survey methodology, such as sample size, data collection date and location, recruitment method and population of interest are collated in Table 2. In five papers, the authors referred readers to other sources for this information (Barbour et al., 2021; Hensher et al., 2021a, 2021b; Kong et al., 2022; Salon et al., 2021), which we did when possible. Two studies were missing several components of their basic methodology, such as data collection date, location, sample size or recruitment method (Ahmed & Khalil, 2021; Appel-Meulenbroek et al., 2022). Two studies did not provide any information about the demographics of their respondents (Salon et al., 2021; Zhang et al., 2021). In addition, Ahmed and Khalil (2021) did not show the wording of survey scale variables. Interestingly, all but one of those papers was published in a first-quartile (Q1) journal. This inconsistent reporting of basic survey information perhaps reflects the disruption of the pandemic and the rush to record its experience and impacts.

In particular, the representativeness of the survey samples varied significantly across studies. We discuss these issues because the use of non-representative samples undermines any ability to estimate city-wide impacts of WFH. Beck and Hensher (2022b, p. 273) state that the impact of COVID-19 was so widespread that “no demographic can escape the disruption [it] caused”. This may be true; however, the extent of these impacts differed across populations and cities.

Of the papers that reported their recruitment method, 13 used some kind of systematic recruitment method (such as via a recruitment panel), eight used convenience sampling and four used a combination of panel recruitment and convenience sampling. Ten papers used surveys based on systematic recruitment via market research panels (e.g. Barbour et al., 2021; Beck et al., 2020; Beck & Hensher, 2020; Beck & Hensher, 2022a, 2022b; Currie et al., 2021; Dianat et al., 2022; Hensher et al., 2021a, 2021b; Jain et al., 2022) whereas three (all from the Netherlands) drew from existing longitudinal population studies (e.g. de Haas et al., 2020; Olde Kalter et al., 2021; Ton et al., 2022). These studies used quota sampling and/or population-based weighting to capture a representative sample of the working population.²

The papers using convenience sampling drew responses from social media channels (e.g. Conway et al., 2020; de Abreu e Silva, 2022; Kong et al., 2022; Nayak & Pandit, 2021; Nguyen, 2021) or organisational email lists (e.g. Ceccato et al., 2022; Hiselius & Arnfalk, 2021; Nguyen, 2021). These studies generally had a smaller sample size (in the hundreds rather than thousands) and did not weight their data. Some of these studies compared their sample to the relevant population group and discussed the implications of recruiting a non-representative sample (de Abreu e Silva, 2022, p. 273). Yet 14 papers, just under half of the papers in this review, did not compare their sample to the relevant population they sampled from. Even without showing population demographics, it is clear that most studies over-represented highly-educated, high-income individuals, and some are very explicit about this limitation (Conway et al., 2020). Other studies were extremely niche in scope, such as surveying university employees (Ceccato et al., 2022) or public servants (Hiselius & Arnfalk, 2021).

Four papers, from two data sources, used a combination of recruitment approaches. The COVID Future Survey was used in three papers (Conway et al., 2020; Mohammadi et al., 2022; Salon et al., 2021). This study used recruitment companies, social media and snowballing to expand the sample size across the United States, although this was not effective in making the sample more representative of the population. For example, Mohammadi et al. (2022) found that 59% of their respondents had a graduate degree compared to only 13% of the US population. In another study, Balbontin et al. (2021) looked at commute mode choice across eight countries. Two of the countries (South Africa and Australia, which drew from the Beck & Hensher data) were random population samples; the other six countries employed convenience sampling.

Recruitment techniques have significant implications for the representativeness of the survey samples as well as the interpretation of the findings, particularly if the intent is to understand the extent of WFH across the general population. The implications of this are discussed in the next section.

Estimates of future work from home

Perspectives from academic literature

In the systematic search, we attempted to extract from each paper the anticipated increase in WFH “after COVID” (however that was operationalised in the survey, as discussed previously).

This was possible for 26 out of 29 papers; the remaining papers reported model outputs without providing the baseline increase in estimated WFH. Among those papers that provided some estimate of future WFH, some were reported as generic statements such as wanting to “work from home more in the future” (Beck & Hensher, 2022b; de Haas et al., 2020; Ton et al., 2022) or that WFH “should be further developed” in the future (Nguyen & Armoogum, 2021). In a survey of “transport experts” asking about lifestyle changes due to COVID, the statement that attracted the most agreement was that “Online working will become popular” (Zhang et al., 2021). These statements provide useful context of general opinion, and they were usually used as input or outputs in models estimating factors that influenced intent to WFH. On their own, however, they are not specific enough to provide meaningful outcomes for future transport planning.

Other studies reported more specific estimates of the quantum of future WFH. However, many were in formats that could not be applied to a “general population” or “working population”, often by focussing only on people who were already teleworking before or during COVID. For example, a survey of teleworkers in India found that 30% stated they will telework full-time in the future (Nayak & Pandit, 2021) and a survey of people who WFH in the US found that 27% anticipate WFH full-time after the pandemic (Mohammadi et al., 2022). Yet without a means to estimate the proportion of the working population that can WFH to begin with, these estimates only apply to an unknown proportion of a city’s workers. Other studies surveyed very niche populations; for example, a survey of University of Padova employees found that only 5% stated they would never WFH after COVID whereas 40% said they would never come into the office (Ceccato et al., 2022). And another set of studies included all employed people, but drew their estimates from convenience samples that were not weighted to the general population. For example, an American study found that only 27% of the sample would not WFH in the future (Conway et al., 2020) and a study from Lisbon found that only 8.7% had no intention to telework (de Abreu e Silva, 2022). Both of those surveys over-sampled highly-educated, high-income individuals and are therefore likely to over-estimate the quantum of WFH.

If we focus our discussion on studies that used population-weighted data or random population sampling, and considered all working adults, we are likely to get a more accurate estimate of future intent to WFH. Nine studies fit these criteria, but they drew from a very limited number of data sources and sometimes provided overlapping estimates (see Table 3). One set of studies quantified how much people expect to WFH in the future, the other quantified how much people want to WFH in the future. There was a significant difference between expectations and desires.

Table 3. Estimates of future work from home intention among studies using random sampling of the working population.

Source	Location	Data source	Estimate of future WFH
Studies quantifying how much people expect to WFH “after COVID”
Salon et al. (2021)	USA	COVID Future survey	Percent of workers who expect to WFH at least a few times a week increased from 13% to 26%
Currie et al. (2021), Jain et al. (2022)	Melbourne, Australia	C-19 Long Term Transport Impact Study	75% increase in the quantum of WFH; 123% for people who work in the city centre. Calculated a 6% reduction in commute trips; 20% for commutes into city centre.
Studies quantifying how much people want to WFH “after COVID”
Beck et al. (2020), Beck & Hensher (2020, 2022a)	Australia	Beck & Hensher survey	About 38% do not want to WFH at all, about 32% want to WFH at least a few days/week and 30% want to WFH most or all the time.
Hensher et al. (2021a, 2021b)	Sydney, Australia	Beck & Hensher survey	Average desire is to WFH 2 days/week.
Balbontin et al. (2021)	Major cities of Australia	Beck & Hensher survey	1.8 days/week WFH
	South Africa	Random sample + face to face	3.2 days/week WFH
	Argentina	Convenience sample	3.0 days/week WFH
	Brazil	Convenience sample	3.1 days/week WFH
	Chile	Convenience sample	2.9 days/week WFH
	Colombia	Convenience sample	3.3 days/week WFH
	Ecuador	Convenience sample	3.2 days/week WFH
	Peru	Convenience sample	3.5 days/week WFH

Two sets of studies measured future expectations of WFH. One of these, Salon et al. (2021), drew from the COVID Future survey, which drew from both panel and convenience sampling, but applied population-representative weights to their estimate. They found that the proportion of employed adults who expect to WFH “a few times a week or more” doubled from 13% before COVID to 26% “post-COVID”. However it’s worth noting that over half of their sample cannot work remotely at all and the remaining quarter could WFH but will do so rarely or not at all (Salon et al., 2021).

The other survey measuring expectations is the C-19 Long Term Transport Impact Study from Melbourne, Australia. In this survey the authors projected the change in WFH onto the total quantum of commute trips for the working population of Melbourne (Currie et al., 2021; Jain et al., 2022). They projected a 75% increase in WFH over a pre-COVID baseline; this increased to 123% for people who worked in the city centre. However, given the low baseline rates of WFH, the reduction in commuting trips was only 6% overall, although this jumped to 20% for trips into the city centre (Currie et al., 2021).

The remaining studies did not measure expectations. Rather, they asked respondents how many days per week they want to WFH in the “post-COVID” future. Three studies drawing from the Beck & Hensher data provided estimates for Australia as a whole (Beck et al., 2020; Beck & Hensher, 2020; Beck & Hensher, 2022a), and two focussed on Sydney, Australia, specifically (Hensher et al., 2021a, 2021b). These estimates vary depending on which survey wave is analysed in a given paper, but on the whole, around 38% of Australians surveyed did not want to WFH at all, 32% want to WFH at least a few days a week and 30% want to WFH full-time. The Sydney sub-sample of this survey (Hensher et al., 2021a, 2021b) found that on average people wanted to WFH two days a week.

Finally, Balbontin et al. (2022) was unique in comparing survey data from eight different countries. Their data from Australia was drawn from Beck & Hensher, data from South Africa was drawn from a market research panel supplemented by face to face recruitment, and the remaining samples from South America were convenience samples through social media and personal contacts. Estimates of desire to WFH varied from 1.8 days/week in Australia to 3.5 days/week in Peru. It would be misleading to interpret preferences for WFH as future projections.

Although most studies in the systematic review tried to measure some future projection of WFH, few provided these estimates in a way that could be quantified and compared to a general working population. Furthermore, most studies failed to consider that employee preferences are moderated by workplace constraints. Balbontin et al. (2022), for example, found that only 9–18% of respondents said that future decisions about WFH were entirely their choice, and between 20% and 44% said that WFH was not at all possible once COVID-19 restrictions end. Our review now turns to consider employer perspectives as a way to provide some insights into the extent to which WFH will be accommodated by employers.

Perspectives from grey literature

The grey literature on employer sentiment often focused on acceptance of “hybrid” working arrangements and ways to encourage workers back to on-site working. This was typically characterised as a hybrid working week with specific days of mandated attendance on-site.

Table 4 presents results of eight surveys identified through the grey literature review on employer intentions. The hybrid way of working dominates. While two surveys did not break down intentions for hybrid versus offsite working, most of those that did, indicated the most common mode of working will be a hybrid arrangement. Only one survey estimated that over 50% of their workforce will be onsite full-time.

Table 4. Grey literature surveys capturing employers’ intentions towards future WFH.

Survey name	Location	Date	N	Instigator	100% onsite	Hybrid	100% offsite
Embedding New Ways of Working	United Kingdom	NS	1046	Chartered Institute of Personnel and Development UK (2023)	41	37	22
What executives are saying about the future of hybrid work	Asia, Europe, Latin America, United States	Jan-21	100	McKinsey & Company (2021)	10	87	3
Deloitte Hybrid Work Survey	Slovakia	Mar-21	47	Deloitte (2021b)	21	79	0
Deloitte 2021 Return to Workplaces Survey (Deloitte)	United States	Apr-21	275	Deloitte (2021a)	21	68	11
Robert Half Future of Hybrid Work Survey	Canada	Jul-21	800	Robert Half Pty Ltd (2021)	56	44	0
PwC US Pulse Survey	United States	Aug-21	752	Price Waterhouse Coopers (2021)	19	NS	NS
Global Survey of Working Arrangements	Global	Aug-22	23,849	Working From Home Research 2022	30	NS	NS
2022 Gartner Hybrid Collaboration Survey (Gartner)	Global	Oct-22	3,524	Gartner (2022)	29	59	12
				Mean	28	62	8
				Standard deviation	14.4	19.5	8.6

These results need to be treated with caution. First, the conceptualisations of “hybrid” was not standardised and rarely stated. Hybrid working can mean many things, from a few days at home per year to near-full-time at home with infrequent visits to the office. In addition, most surveys failed to disclose details about respondents, and lacked specific information on the workforce under examination. There are likely many industries, cities and workers that will have been missed. In addition, although the surveys are global in scope, they are primarily instigated by consulting firms interested in specific industries located in specific jurisdictions. In short, while these data suggest a clear trend towards hybrid working, they don’t provide clear insights into the overall percentage of workers likely to WFH.

Insights on employer sentiment are also embedded in the ways they are accommodating, and investing in, the assumption that their workforce will by hybrid. In a series of interviews with senior management at transnational corporations, all participants envisioned a significant overhaul of their company’s real estate footprints, citing average reductions of 40% (Trevor & Holweg, 2023). This reduction wasn’t necessarily reflective of the intention to have less bodies physically present in workplaces because many employers mandate specific days of full teams working on-site. Instead, employers are reconfiguring spaces so they can accommodate team working, with very little provision for individual working spaces (Tansey, 2023). Reductions in office size and reconfigurations of workplaces are relatively permanent moves, providing yet another indication that hybrid working will continue into the future.

Employer factors influencing future WFH

The nature of work

The biggest determinant of employee intentions and employer policies for WFH is the nature of the actual work being conducted. It goes without saying that not all jobs can be done remotely. For example, it is unlikely that work in agriculture, restaurants and entertainment can be done remotely on a regular basis.

There have been attempts to calculate the percentage of occupations that can be done off-site. In the United States, around 63% of jobs require 100% onsite attendance (Dingel & Neiman, 2020). This obviously limits the uptake of WFH, regardless of the advances attained in its acceptance throughout the pandemic. An employee is unlikely to intend to WFH, and an employer is unlikely to allow it, if the actual job cannot be performed away from the workplace.

Whether or not a job can be performed from home will also differ between cities and nations. For example, access to basic infrastructure is likely to be a continuing barrier in lower-income countries. Mogaji (2022), for example, found that Nigerians have a strong desire to WFH more, but many lacked reliable internet access, power connections or appropriate technology.

Employer policy

When it comes to where work is performed, employers have the final say (Culley, 2022). Since the lifting of lock-downs, there have been several examples of case law, from a variety of different jurisdictions, which have established that the WFH model is a privilege, not a right. An employer can, as part of a contract, deem onsite attendance as a condition of employment, even if the job to be performed can be done at home (Samy, 2022).

Many grey sources focussed on different degrees of tension between employees wanting flexibility to WFH, and employers wanting to mandate certain days in the office (Brown et al., 2023; Burroughs, 2022; Golden, 2022b;). These mandates are often enshrined in specific policies which establish quotas for in-office attendance (Lewis, 2022; O'Brien, 2022; Singleton & Johnson, 2022).

Although policies and practices vary, the most common approach discussed is a combination of mandated on-site attendance with some days of WFH. Specific days for each are either stipulated by management or agreed to in teams. As Nicholas Bloom of Stanford University advises, “get folks in on two, three days a week … and be really tough about having them come in” (Chui, 2023).

There have been several high-profile instances where employees have attempted to push back on the model of mandated on-site attendance. For example, executives at the USA’s second largest employer – Amazon – made international headlines when they petitioned against a company mandate of four days on-site working per week (Barrabi, 2023) as did employees at entertainment company Disney (Telford, 2023).

The debate often centres on the fine line employers tread between keeping their staff happy and allaying internal fears that too much off-site work will negatively impact the company (Trevor & Holweg, 2023). This is further complicated by the way power shifts between employers and employees. As the world emerged from lockdown the consensus was generally that employees had the upper hand in demanding flexible working provisions because of worker shortages (Burroughs, 2022; Colvin, 2022; Samy, 2022). The impacts of higher inflation and slowed economic growth, however, have turned the balance of power towards employers in some sectors (Donner, 2022).

While eroding company culture and productivity dominated employer concerns during the pandemic, decreased innovation has emerged as an increasing concern. One article described metrics for decreases in innovation during the extended period of WFH, attributing this decrease to a lack of incidental face to face connections (Trevor & Holweg, 2023). Any slow etching away of advancements in systems, products, services and paradigms is not likely to be evident for a few years. Forward-thinking employers are trying to prevent impacts before they happen, and mandating onsite attendance is one way to do this (Golden, 2022a).

External dynamics

The review of grey sources highlighted a series of global political and economic factors that are likely to shape the immediate future of WFH. While these factors are transitory and situated, the fact they are occurring while the world is establishing new norms for WFH grants them particular relevance.

Geo-political unrest, primarily related to conflict in Eastern Europe, is having indirect effects on the global economy through fuel prices and inflation. In response, some companies (particularly in Europe) are looking for ways to reduce energy expenditure, and closing worksites for specific days of the week is one way to do this. The local authority in Milan, for example, now closes its offices on Fridays and all employees are expected to WFH (Stancati, 2022). Banking firm ING has pursued a similar approach in Amsterdam (Koc, 2022). These policies are motivated purely by cost-savings, but informed by the knowledge developed through the pandemic that WFH is possible.

In addition, many are predicting a global recession. Firms across an array of industries are seeking to downsize or innovate to escape the impacts of what seems to be an inevitable downturn in revenue (De Avila, 2023). This environment gives employees less power to negotiate preferences, including to WFH, than was perhaps first the case when lock down periods ended (Constantz, 2022). Employers therefore have more power to develop and enforce WFH arrangements that suit the company rather than respond to employee preferences.

The culture, approach to management and historical legacies embedded in a company can also impact WFH offerings. For some companies, providing well-appointed office space has been a hallmark of their contribution to employee well-being, and one they are not willing to relinquish easily (Blumgart et al., 2022). One source discussed the way law firms, steeped in traditions on hierarchy and face to face mentoring, were less likely to support off-site work arrangements (Taylor, 2022). Others have noted that accounting firms and even stock exchanges have been slow to take up WFH as a serious offering (Franklin & Moise, 2022).

Employee factors influencing future WFH

Although employers have the final say in the extent of WFH, when employees do have the option, demographics and attitudes determine whether they take it up.

Personal and household demographics

Much of the academic literature considered how personal and household characteristics influenced employee intent to WFH. Higher income and education levels were both associated with a greater intent to WFH, regardless of whether the study considered the entire population, or only sub-samples who were able to telecommute (e.g. Dianat et al., 2022 Nguyen, 2021).

In many studies, women and households with children were more likely to prefer to WFH in the future (Appel-Meulenbroek et al., 2022; Dianat et al., 2022; Moens et al., 2022; Nguyen, 2021; Nguyen & Armoogum, 2021). Quite a few studies, both during COVID-19 (Kroesen, 2022) and before COVID-19 (Hilbrecht et al., 2008), have found that women benefit more from telework, experiencing a greater increase in job performance and smaller impacts on work-life balance than men (Moens et al., 2022).

Grey sources suggest employers recognise that WFH enables better work-life balance, and that it appeals to those with caring responsibilities, primarily parents (Caffera et al., 2022). One employer recruiting staff post-pandemic, for example, reported a 30 percent increase in candidates from underrepresented demographics, particularly women, which they attribute to their highly flexible working environment (Trevor & Holweg, 2023). Even so, employers remain hesitant to provide carers with exemptions from mandated days on-site, even though they recognise that catering to the needs of carers is a hall mark of a flexible workplace (Burroughs, 2022).

Attitudes and preferences

Attitudes amongst employees towards WFH varied considerably. Papers using data from developed countries generally found that WFH was a positive experience (de Haas et al., 2020; Moens et al., 2022; Ton et al., 2022). In most studies, respondents believed they were as productive or more productive when WFH compared to days they commute into work (Balbontin et al., 2021). Some studies used factor analyses or segmentation to identify which clusters of employees enjoyed WFH and which strongly preferred to work on-site (Appel-Meulenbroek et al., 2022; Conway et al., 2020; Kong et al., 2022).

Some studies explored how attitudes, norms, and preferences influence intent to WFH in the future. Some studies conducted factor analyses of general attitudes towards WFH, preferences for in-person work, or other attitudes to work. These attitude measures generally behaved as expected, for example people who enjoy their workplace are more likely to want to work in the office (Nguyen & Armoogum, 2021).

Two studies used a validated behavioural model, the Theory of Planned Behaviour, to structure their research, both using structural equation modelling. Ahmed and Khalil (2021) found that attitudes, norms and self-efficacy were all related to intent to WFH, whereas Jain et al. (2022) found strong relationships between subjective norms and perceived behavioural control and intent to WFH, but small or non-significant relationships between attitude and intent.

Health and biosecurity concerns were a common theme. People who were afraid of infection were more likely to WFH and to avoid commuting by public transport (Ceccato et al., 2022). An American study found that people least concerned with COVID-19 were also the least likely to WFH before and after the pandemic (Conway et al., 2020). Relating to the finding that women are more likely to prefer to WFH, women are also more likely to fear infection from COVID-19 (Conway et al., 2020; Mohammadi et al., 2022).

Implications of changes to working from home

Any attempt to assess the implications of WFH will depend on an understanding of the likely prevalence of WFH into the future. Our intention has been to provide this understanding, rather than develop tangible insights into how WFH will change the structure and operations of cities. Through the review process, however, we did note some direct predictions and suppositions about transport and land-use in the context of the uptake of WFH which are addressed in this final section of our review.

In the scholarly literature, predictions about the implications of WFH on the transport system (and therefore the environmental impacts) varied considerably, depending on the study location and future scenarios considered. The most common forecast was for a significant shift away from public transport with estimated declines ranging between 10% and 40% (Currie et al., 2021; Olde Kalter et al., 2021; Salon et al., 2021). Some of this decline was forgone trips (i.e. replaced with WFH) whereas some of the was from a shift towards either car commuting (for longer commutes or car-dominated contexts) (Ceccato et al., 2022; Currie et al., 2021) or walking and cycling (for shorter commutes and less car-dominated contexts) (Ceccato et al., 2022; Olde Kalter et al., 2021). Other studies predicted an overall decline in car kilometres travelled for commuting, ranging from 12% to 15% (Olde Kalter et al., 2021; Salon et al., 2021). Conversely, a study of university employees found that even with a decrease in commuting, a shift towards car use on commuting days will increase emissions overall (Ceccato et al., 2022). An international study of experts found that 82% of experts in North America predicted increased car dependence, compared to between 48 and 54% of experts in Japan or Europe (Zhang et al., 2021). These findings echo the research prior to the pandemic that suggests that telecommuting may increase, not decrease, overall travel demand, particularly travel by private car (De Abreu e Silva & Melo, 2018; Zhu et al., 2018).

Our review of employer intentions suggests that the journey to work will remain a determinant of the transport task of cities, however with important caveats. We found that a common provision is for managers and teams to agree on two to three specific days in the office, with the other days WFH. This is important for transport planners because the more regularly and predictably the employee is required to work on-site, the more relevant the commute remains in decision-making around housing, schooling and access to other services. Predictability of routines also favours sustainable transport modes for workers, who can plan their trips around timetables and fixed routes rather than respond on a day-to-day basis using the private car.

The systematic review also found that changes to WFH and commuting are likely to play out in different ways across different spatial contexts. Predicted increases in WFH are more acute for people who work in a city centre relative to people who work in a suburb (Currie et al., 2021). In almost every study, people who lived farther from their workplace or had longer commutes were more likely to want to WFH, whereas people living in urban areas or closer to work were less likely (Appel-Meulenbroek et al., 2022; de Abreu e Silva, 2022; Dianat et al., 2022; Nguyen, 2021;). Again, these findings align with a long history of research on telecommuting that predates the pandemic (De Abreu e Silva & Melo, 2018). Econometric modelling suggests that the value of time that people place on their commute trip has, if anything, increased in response to COVID-19, suggesting that this relationship will continue into the future (Hensher et al., 2021a).

On the subject of adaptations in response to WFH, experts are less sure of the long-term impact of remote working on where people will choose to live. Only 10-36% of transport experts believe that people will choose to live farther from the city centre and 13-28% believe people will migrate outside of populated cities (Zhang et al., 2021). This may be linked to the complex factors impacting residential self-selection. In the words of de Abreu e Silva (2022, p. 157), “telework does not lead directly to sprawl, but it could allow individuals to transform their residential location preferences into actual residential locations.”

Discussion and conclusion

In summary, our review has confirmed that there is a strong underlying demand for access to WFH, and that it will, in some way, continue to be a feature of working practices. Many studies, however, draw mostly from non-representative samples to estimate unrealistically high rates of WFH which cannot be projected onto the wider working population. The most systematic estimate from academic research suggests that over 60% of workers want to WFH at least some of the time (Beck et al., 2020; Beck & Hensher, 2020; Beck & Hensher, 2022b). However, the most systematic estimates of expected WFH are quite modest. For example, the proportion of workers who expect WFH “at least a few times a week” may only increase from 13% before COVID to 26% “post-COVID” (Salon et al., 2021). Others have suggested this will result in only a 6% reduction in commute trips (Currie et al., 2021; Jain et al., 2022). The grey literature could not be used to estimate workforce-wide incidence of WFH. It does, however, indicate that employers favour hybrid working arrangements, and are making provisions for their employees to work off-site at least some of the time. Notably, there are differences between employer expectations and employee preferences, with the hopes of the latter far exceeding the alacrity of the former.

From the perspective of the characteristics of employees WFH, this paper confirmed existing understandings. WFH preferences are positively associated with higher income and education levels, and being a woman or a household with children (Appel-Meulenbroek et al., 2022; Dianat et al., 2022; Moens et al., 2022; Nguyen, 2021; Nguyen & Armoogum, 2021). Positive attitudes, supportive social norms and control over WFH choices were also associated with greater intent to WFH (Jain et al., 2022). From the perspective of workplaces, the nature of work, company policies, and wider global trends all act to limit how much of the latent employee demand for WFH will be satisfied.

This review has used a mixed-methods approach which has been useful to examine the complex and ever-changing phenomenon of WFH. Our analysis has several limitations. Some of these are related to the quality and focus of the studies available for review. Perhaps by virtue of the rapidly-moving nature of the topic, several papers from the scholarly literature were not subject to a rigorous peer-review process, even when published in first-quartile (Q1) journals. For example some studies lacked information on data collection dates, survey location, sample size, sample demographics, survey scale wording or recruitment method (Ahmed & Khalil, 2021; Appel-Meulenbroek et al., 2022; Salon et al., 2021; Zhang et al., 2021). Many drew from non-representative samples or sub-populations of workers who were already capable of WFH, limiting their relevance to the wider context. Most of the studies from the systematic search collected data in 2020, when much uncertainty remained about the extent and impacts of COVID-19. We acknowledge that neither the grey literature, nor the scholarly papers reviewed, aimed explicitly to estimate future WFH for the entire working population. The literature generally relates to the global north, meaning that the nuance of context relevant to the rest of the working world is lost. Furthermore, the dataset for the systematic review is dominated by three specific surveys. Although these datasets were analysed from different angles, their repetitive publication means that the intentions and practices of some people have been considered more than once. Further, we acknowledge that the literature on WFH in the COVID-19 context is voluminous and ever-growing. We have, no doubt, missed literature that falls outside of the scope of our focus, and literature that has been published since our review date. Despite the non-representativeness of our data and the articles reviewed, their contribution is vital to our understanding. Without this information, we are largely “flying blind” into a future where hybrid working is more common, but the extent of its impacts are largely unknown.

Future research

While estimates for preferences and allowances to WFH are somewhat tenuous, it is clear that some increase in WFH in hybrid form will remain. Yet we still know little about the availability and impact of WFH on the workforce as a whole. If we are to design specific policies within a given city context, we need more information about the extent of the practice of WFH. We need to know both the origin and destination of commute trips that are likely to be replaced by WFH, as well as their temporal rhythms. We also need better understandings of the way WFH will shift other household decisions, and their associated transport practices. Further, studies into the future need to be executed with rigour, including a specific focus on representativeness and generalisability – across the workforce and from multiple spatial contexts. Studies should also focus on the entire working population, rather than those with access to the ability to WFH. Although COVID “variants of concern” continue to emerge, we are likely now living a “COVID normal” future. As we move through 2023, it may be time to move away from a focus on “future estimates” of WFH, and shift focus to present patterns. This should include revealed preference studies to provide breadth, complemented by qualitative approaches to deepen appreciations of change across various life domains.

Publications in this space are still emerging, almost on a week-by-week basis. This knowledge base will inevitably grow to develop our knowledge about the transport impact of this intriguing and unprecedented shift.

Acknowledgements

The authors would like to acknowledge Professor Patricia Mokhtarian, whose long-term contribution to the literature on telecommuting, telecommunications and derived demand is foundational to this field. Dr Delbosc conducted the systematic search and Dr Kent conducted the grey literature search. Both authors contributed equally to the paper conception, writing and editing.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 https://jcr.clarivate.com/jcr/home

2 One interesting counter-example was Moens et al., 2022, which placed a print advertisement in the newspaper, then used post-stratification sample selection to choose a sub-set of responses that were representative of the population of Flanders, Belgium.