Antecedents and consequences of telework during the COVID-19 pandemic: a natural experiment in Japan

The study investigates why telework adoption was comparatively low in Japan during COVID-19 and examines both the antecedents facilitating its implementation and its socio-psychological consequences within Japanese workplaces. Given that much prior telework research is based on Western contexts, the authors explore how Japanese cultural and institutional features—such as life-long employment, seniority-based remuneration, mutual monitoring, micro-management, and high-context communication—might impede telework. Two research questions guide the work: which socio-psychological workplace factors predicted telework implementation (RQ1), and how did telework affect subsequent socio-psychological factors (RQ2)? The research leverages a natural experiment via a longitudinal panel spanning pre- and mid-pandemic waves to capture causal patterns in a Japanese cultural context.

Prior meta-analytic work in Western settings links telework to higher productivity, retention, commitment, and performance, but generalizability to non-Western contexts remains uncertain. During COVID-19, telework rates in Japan lagged behind Western countries and even some East Asian peers, despite adequate IT infrastructure, suggesting cultural-institutional barriers. Japanese employment traditions (life-long employment, seniority-based pay) foster low generalized trust and reliance on mutual monitoring, micro-management, and process-based evaluation. High-context communication and on-the-job training embed skills and interactions within organizations, emphasizing interdependence and teamwork; telework could disrupt these patterns by reducing in-person cues and proximity while increasing individual agency. Hierarchical structures based on rank, seniority, and tenure further shape telework feasibility, with initial evidence suggesting lower-ranking employees teleworked less. Prior surveys in Japan yielded mixed findings on telework’s psychological impacts, with some reporting isolation and stress, while others noted growing acceptance and desire to continue teleworking as the pandemic progressed. Most prior studies were cross-sectional or descriptive, limiting causal inference, hence the need for a longitudinal, natural experiment to test institutional, interactional, and individual-level factors (e.g., seniority vs. meritocracy systems, social vigilance, concern for intragroup relationships, independence-interdependence, clan vs. market culture, hierarchy mutability, sense of power, responsibility, organizational commitment, social isolation, and superior–subordinate disintegration).

Design: Longitudinal three-wave panel survey treating the COVID-19 pandemic as a natural experiment to assess antecedents (pre-pandemic workplace socio-psychological characteristics) and consequences (post-implementation changes) of telework in Japan. Sampling and participants: Employees in Japanese companies recruited via a research company using quota-based stratification in Wave 1 (February 2020; pre-pandemic in Japan). Initial N=1248 balanced by gender (50% male/female), age (equal thirds in 30s, 40s, 50s), and organizational rank (section/assistant manager, unit head/supervisor, staff). Wave 2 in February 2021; Wave 3 in February 2022. Wave 2 responses N=936; Wave 3 responses N=633. After excluding 71 who changed organizations and those failing attention checks, final analytic panel N=367 (184 males, 183 females); age and position distributions remained approximately balanced in Waves 2 and 3. Measures: - Telework implementation: In Waves 2 and 3, respondents reported workplace telework extent versus pre-COVID (0–100% in 10% increments). Wave 1 telework level was obtained retrospectively at Wave 3 for February 2020. Telework was binarized: 0% (no telework) vs. 10–100% (some telework) due to skew. Telework prevalence: 35.0% (Wave 1), 51.7% (Wave 2), 53.1% (Wave 3); correlations: Wave1–Wave2 r=0.53; Wave2–Wave3 r=0.80. - Socio-psychological variables (measured each wave; details in Table 1): institutional (seniority system, meritocracy), interactional (clan culture, market culture; perceived hierarchy mutability), and individual-level constructs (independence, interdependence, social vigilance, concern for intragroup relationships, sense of power, perceived responsibility) plus relationship quality indicators (organizational/affective commitment, social isolation, superior–subordinate disintegration). Most constructs used 1–2 items due to survey length; Wave 3 included fuller scales for independence, interdependence, clan, and market culture. - Demographics: Sex, age, residence, industry, job type, work position, company size, and client base (domestic/overseas). Analytic strategy: - RQ1 (antecedents): Logistic regressions predicting Wave 2 telework presence from each Wave 1 socio-psych variable (14 separate models), controlling for base telework (Wave 1) and demographics. Demographics were summarized via a composite telework probability using logistic Lasso on Wave 1 demographics (lambda via cross-validation) to adjust for known demographic influences on telework. - RQ2 (consequences): Linear regressions predicting each Wave 3 socio-psych outcome from Wave 2 telework presence, controlling for the same outcome at Wave 2, demographics (propensity score), and Wave 1 meritocracy. Inverse probability weighting with propensity scores (from Wave 1 demographics) addressed non-random telework implementation. Base models with controls were compared to full models; ΔR² reported. - Exploratory moderation by managership (manager vs. non-manager) assessed main effects and telework-by-managership interactions; only deviations from main analyses were highlighted. Ethics: Approved by Kyoto University IRB; informed consent obtained. Data sharing: raw data not shared due to consent limits; code and materials available via OSF.

- Telework prevalence and stability: Telework increased from 35.0% (Wave 1) to 51.7% (Wave 2) to 53.1% (Wave 3). Telework was moderately stable across waves (r=0.53 for Wave1–Wave2; r=0.80 for Wave2–Wave3). - RQ1 antecedents (logistic regressions, Wave 1 predictors → Wave 2 telework): Only meritocracy significantly predicted implementation (B=0.232, z=1.976, p=0.048), indicating higher likelihood of telework adoption in workplaces already perceived as meritocratic. Base model controls were significant (Wave 1 telework: z=5.038, p<0.001; demographic composite: z=7.761, p<0.001). Other socio-psychological predictors (e.g., seniority system, clan/market culture, independence/interdependence, vigilance, hierarchy mutability) were not significant. - RQ2 consequences (linear regressions, Wave 2 telework → Wave 3 outcomes): Telework predicted higher independent social orientation (B=0.162, t=2.332, p=0.020), higher perceived hierarchy mutability (B=0.245, t=2.074, p=0.039), and higher organizational commitment (B=0.178, t=2.058, p=0.041), controlling for Wave 2 levels and covariates. In follow-up analyses controlling for managership, the effects on independence (t=2.133, p=0.034) and organizational commitment (t=2.097, p=0.037) remained; hierarchy mutability became marginal (t=1.933, p=0.054). Telework also significantly reduced social isolation (t=-2.227, p=0.027) in follow-up. - Moderation by managership: Telework marginally increased perceived meritocracy among non-managers (t=1.724, p=0.088) but not among managers (t=-1.511, p=0.132); it decreased concern for ingroup relationships among non-managers (t=-2.033, p=0.045) but had little effect for managers (t=0.962, p=0.337); it increased sense of power among managers (t=2.743, p=0.007) with little effect among non-managers (t=-1.116, p=0.268). - No measurable negative effects: Across outcomes, implementing telework did not show detectable adverse impacts on relationship quality; instead, it was associated with increased organizational commitment and reduced social isolation.

Findings address the two research questions within a Japanese cultural-institutional context. First, organizations characterized by meritocracy before the pandemic were more likely to adopt telework, suggesting compatibility between performance-focused evaluation systems and remote work where monitoring of process and in-office presence is reduced. This partially supports the hypothesis that traditional Japanese systems (seniority, process orientation, micro-management) hinder telework, while meritocratic systems facilitate adoption. Second, telework produced positive socio-psychological changes one year later: greater independence, greater perceived hierarchy mutability, and higher organizational commitment, alongside reduced social isolation in follow-up analyses. These results suggest telework can catalyze cultural change in organizations by shifting interaction norms and individual orientations toward less traditional patterns without degrading relationship quality. Moderation results imply role-contingent impacts: non-managers perceived increasing meritocracy and reduced preoccupation with intragroup relationships, while managers experienced higher sense of power. Overall, telework and meritocracy may be mutually reinforcing, and the absence of negative effects counters concerns that telework inherently undermines cohesion in Japanese workplaces.

This natural experiment with three waves spanning pre- and mid-pandemic periods shows that meritocracy is a key antecedent for telework adoption in Japan. Telework, once implemented, fostered non-traditional work orientations and perceptions—greater independence, higher hierarchy mutability—and strengthened organizational commitment while reducing social isolation, with role-specific effects for managers and non-managers. The study contributes evidence that telework can align with and promote institutional change in Japanese organizations rather than harming workplace relationships. Future research should examine mechanisms linking telework to perceived meritocracy and hierarchy dynamics, differentiate voluntary versus mandatory telework intensity, employ multi-item and multi-informant or objective organizational measures of institutions, and test generalizability across East Asian contexts and sectors.

- Measurement brevity: Most constructs were assessed with one or two items due to survey length constraints, limiting reliability; fuller scales were included only in Wave 3 for select constructs. - Perceptual measures of institutions: Institutional variables (e.g., meritocracy) relied on individual perceptions rather than objective indicators or aggregated multi-respondent data, introducing potential measurement error. - Telework operationalization: Telework was binarized (0% vs. 10–100%), conflating intensity and voluntary versus forced telework; high-intensity or mandated telework may have different effects. - Potential survival bias: Organizations or individuals for whom telework failed may have discontinued it, possibly biasing observed consequences (e.g., reduced social isolation) among those continuing telework by later waves. - Attrition and sample changes: Although attention checks and exclusion of organizational movers enhance internal validity, attrition from Wave 1 to Wave 3 reduced the sample; generalizability may be affected. - Causal mechanisms: Proposed mechanisms (e.g., trust signals, shifts from process to outcome evaluation) were not directly measured and remain speculative.