The appeal of cities may not wane due to the COVID-19 pandemic and remote working

The study addresses whether the COVID-19 pandemic and the rise of remote working have changed residential preferences and the appeal of living in a large global city, specifically Tokyo Metropolis. Early in the pandemic, highly connected large cities experienced disproportionate infections and deaths, fueling speculation about urban decline and prompting increases in suburban and rural real estate transactions and altered internal migration patterns. Potential drivers include infection risks in dense areas, anxiety about crowded spaces, closures of schools and amenities, social distancing, telework adoption, limited space for home offices, and housing affordability. While some scholars argue cities are resilient and outmigration trends may reverse, evidence has been mixed in Japan, with increased moves from central Tokyo to suburbs and slowed migration to major cities. Given limited mid- to long-term evidence and the need for timely insights for planning, this study uses a post-pandemic resident survey with experimental scenarios (pandemic re-occurrence reminders and remote-work options) to test whether preferences for living in Tokyo have shifted.

Prior work focuses largely on before-versus-during pandemic comparisons. Studies in Japan found shifting preferences among movers to suburbs in Tokyo and Osaka (greater preference for detached homes and more space, especially among families with school-aged children), and that proximity to work lost salience relative to community and environmental factors. In Poland, priorities shifted during the post-pandemic period toward proximity to natural and urban green spaces, with less emphasis on owning plots outside the residence or having balconies/terraces. In Korea (2020), many urban residents expressed concerns about city living; Seoul residents, in particular, reported higher inclination to consider moving to suburban/rural areas, though commuters’ expressed concerns did not translate into strong intentions to leave cities. Other migration research noted increases in moves away from inner-city Stockholm and temporary shifts in several countries (Australia, Germany, Sweden), with suggestions that some effects reversed after restrictions eased. In Japan, analyses reported rising moves from central Tokyo to suburbs and negative net inter-prefectural migration in major prefectures in 2020, with increased interest in lower-density areas. Overall, the literature highlights mixed, context-specific changes and a lack of evidence on enduring impacts on preferences for big-city living.

Study area and design: The study focuses on Tokyo Metropolis (excluding island regions). Neighborhoods were categorized into: (1) inner city (five central wards: Chiyoda, Chuo, Minato, Shibuya, Shinjuku), (2) within the 23 wards (other than the inner five), and (3) within Tokyo but outside the 23 wards (Tama region). Data collection: An online resident survey (Qualtrics) was conducted June 28–29, 2022, using Rakuten Insight panels. Stratified random sampling gave equal weight to each of the 23 wards, and the Tama region was weighted double due to its population and area. Of 1871 participants, after exclusions (outside Tokyo Metropolis: 24; no consent: 162; incomplete responses: 354), 1333 valid respondents remained (71% response rate). Experimental setup: The survey had two components: (1) a Randomized Controlled Trial (RCT) scenario followed by seven conjoint choice tasks; (2) socio-demographic and pandemic-experience questions. A 2×2 RCT assigned respondents to one of four groups: Control–Control, Control–Reminder (pandemic re-occurrence reminder), Remote–Control (remote work option), and Remote–Reminder. The Reminder presented information that pandemic likelihood could double in coming decades; the Remote scenario presented extensive remote-working options versus office-based work in control. Conjoint experiment: In each of seven choice sets, respondents chose between two hypothetical residence profiles or staying at their current residence for the next five years (free-choice). A forced-choice version (choosing between the two profiles) was also analyzed as a baseline. Each profile comprised eight randomized attributes with multiple levels: - Neighborhood: inner city (5 wards); Tokyo 23 wards; outside 23 wards within Tokyo; outside Tokyo - Housing type: single/semi-detached; terrace house; apartment/mansion - Facilities: convenience store (combini); combini & supermarket; combini, supermarket & medical services - Urban green space: local; large park - Community support: no; yes - Public transport availability: none; bus stop; train station - Commute mode: walking; private car/moped; cycling; public transport - Noise: not so quiet in both day and night; not so quiet in day but quiet at night; always quiet Infeasible combinations (no public transport availability with commute by public transport) were removed. The full factorial produced 4752 feasible profiles; with 1333 respondents × seven tasks, 18,662 profile evaluations were obtained. Analysis: Average Marginal Component Effects (AMCEs) were estimated via OLS with respondent-clustered standard errors to assess attribute-level effects on the probability of choosing a profile (free-choice: choosing profile over current residence; forced-choice: choosing one profile over the other). Balance checks via t-tests compared RCT groups; observed imbalances (e.g., housing type, commute time, mode change, COVID-19 experience) were included as covariates in subgroup AMCE models. Heterogeneity analyses stratified by income, age, and self-reported post-pandemic mental status. Acceptance rates (intention to relocate) across RCT groups were compared using mean differences with standard errors.

- Baseline relocation tendency: In 40% of cases, respondents opted to relocate from their current residence when presented with alternatives. - Attribute effects (overall): • Green space and community support had slight/significant effects mainly in forced-choice (+1.2 percentage points, p.p.). • Public transport: In forced-choice, train stations were preferred over bus stops by +6.1 p.p., and both were preferred to no public transport; however, these did not generally motivate relocation in free-choice. • Housing type: Terrace housing was disfavored in both settings (−1.9 to −4.0 p.p. vs single/semi-detached). Apartments/mansions were slightly favored (≈+0.9 to +1.4 p.p.) in decisions between two alternatives. • Neighborhood: Respondents preferred inner-city neighborhoods. In forced-choice, out-of-23-wards within Tokyo and outside Tokyo reduced choice probabilities by −11.5 p.p. and −16.6 p.p., respectively (vs inner city), with little difference between inner city and 23 wards. In free-choice, relative to inner city, Tokyo 23 wards: −2.0 p.p.; out of 23 wards within Tokyo: −5.8 p.p.; outside Tokyo: −4.7 p.p. • Commute mode: Public transport was comparable to walking; cycling and private car/moped were less favorable. In forced-choice, private car/moped: −6.4 p.p. vs walking; under free-choice among movers, this penalty attenuated (+3.9 p.p. relative shift). - RCT effects on acceptance rates (free-choice, intention to relocate): • Control–Control mean acceptance: 38.8%. • Control–Reminder: −0.5 p.p. vs control (ns), indicating reminders of future pandemics did not affect relocation intent. • Remote–Control: +3.8 p.p. vs control (p<0.01), indicating remote-work availability increased relocation intent. • Remote–Reminder: +1.6 p.p. vs control (p<0.1). - RCT subgroup AMCEs: • Terrace houses: Remote–Control group had a significant negative preference (−4.2 p.p. vs detached). • Neighborhood ranking: Remote–Control showed Inner city ≈ 23 wards > out of 23 within Tokyo > outside Tokyo (−6.1 p.p. for outside Tokyo). Reminder groups preferred staying within Tokyo; their aversion to outside Tokyo was smaller (e.g., −3.7 p.p.) than Remote–Control. • Facilities mattered mainly for Remote–Reminder. • Remote–Control showed reduced concern for public transport availability, consistent with less commuting need. • Noise: Control–Reminder and Remote–Reminder significantly preferred quieter profiles: not so quiet day/quiet night (+4.4 and +4.6 p.p., respectively); always quiet (+4.3 and +6.1 p.p.). - Heterogeneity by income: • Housing type: High-income preferred apartments/mansions in Control–Control and Control–Remote but shifted to detached when reminded of a future pandemic; middle-income showed the opposite pattern (reminder → apartments; remote → detached). • Neighborhood: Low-income in Control–Control preferred 23 wards vs inner city (+6.7 p.p.), but this preference vanished with reminder. High-income disliked “out of 23 within Tokyo” (−10.3 p.p.) and “outside Tokyo” (−7.5 p.p.) in Control–Control; aversion to outside Tokyo strengthened with remote (−9.8 p.p.) but attenuated with reminder; remote+reminder made outside Tokyo relatively more desirable than 23 wards or out-of-23 within Tokyo. • Public transport availability: Only low-income in Control–Control significantly preferred bus stops (+5.6 p.p.) and train stations (+11 p.p.); with reminder or remote+reminder, they became indifferent. • Commute mode: Middle-income in Control–Control disliked private car/moped (−8.6 p.p.) and public transport (−6.1 p.p.); no clear preferences under reminder/combined scenarios. • Noise: High-income in Remote–Reminder strongly preferred quieter settings (+6.8 and +10.9 p.p.). - Heterogeneity by age: • Age >60 were generally less reluctant toward residences outside Tokyo across RCT groups; they cared more about urban green space only in Remote–Reminder. • Age 20–39 were strongly reluctant to relocate outside Tokyo under remote (−11.9 p.p.); this reluctance decreased when reminded of another pandemic. Younger respondents valued bus/train access under Control–Remote, and facilities/community support under Remote–Reminder; they preferred quieter residences except when offered remote work. - Mental health heterogeneity: Respondents with worsened mental health after the pandemic were more likely to disapprove residences outside the 23 wards, especially with reminders, and showed stronger preferences for quieter environments (e.g., not so quiet day/quiet night). Overall: Despite remote work increasing relocation intent modestly, respondents still preferred to remain within Tokyo (inner city or 23 wards), with clear disfavor for outside Tokyo.

The findings indicate that, even after experiencing the pandemic and considering remote-work options, the appeal of Tokyo—particularly its inner city and 23 wards—remains strong. Remote work increased the intention to move slightly, yet preferences still favored staying within Tokyo boundaries and avoiding outside-Tokyo locations. This contrasts with reported hollowing-out in some global cities and suggests context-specific dynamics: in Japan, remote work adoption remains relatively low and organizational/cultural norms favor office-based, face-to-face work, which may dampen incentives to disperse. Results also highlight nuanced attribute effects (e.g., disfavor for terrace housing, limited role of green space/community support in motivating moves, reduced salience of public transport among teleworkers) and substantial heterogeneity by income, age, and mental health status. These patterns imply that teleworking does not necessarily induce urban decline and that policies should be tailored to demographic subgroups and local contexts. The sustained desirability of central Tokyo supports planning for compact, resilient urban development, with attention to amenity access and noise environments that some subgroups prioritize, especially under pandemic reminders.

The study provides experimental evidence that the allure of Tokyo has not diminished post-pandemic. While the option to work remotely modestly increases intentions to relocate, most preferences still favor staying within Tokyo’s inner city and 23 wards, with marked aversion to outside-Tokyo locations. Younger cohorts show particularly strong reluctance to move outside Tokyo. These insights can alleviate concerns about large-scale urban outmigration and support policies promoting well-designed compact development to enhance resilience and sustainability. Future research should analyze updated population and migration data to validate stated preferences against revealed behaviors and to assess whether observed patterns persist as work practices evolve.

- Causal inference is constrained: the survey was fielded post-pandemic, so direct causal effects of the pandemic on preferences cannot be established; the ‘reminder’ treatment reinforces salience but does not identify pandemic impacts per se. - Stated-preference design: conjoint responses may not match actual behavior due to hypothetical bias or imperfect self-awareness. - Some RCT covariates were imbalanced across groups (e.g., housing type, commute time, mode changes, COVID-19 experience); models controlled for these, but residual confounding is possible. - The study focuses on Tokyo Metropolis and may not generalize to other cities or national contexts. - Remote work policies and pandemic perceptions are evolving; findings capture preferences at a specific time (June 2022).