Hybrid working from home improves retention without damaging performance

Working from home (WFH) surged after the COVID-19 pandemic, with university-graduate employees typically WFH for one to two days a week during 2023. Previous causal research on WFH has focused on employees who are fully remote, usually working on independent tasks in call-centre, data-entry and helpdesk roles. This literature has found that the effects of fully remote working on productivity are often negative, which has resulted in calls to curtail WFH. However, there are two challenges when it comes to interpreting this literature. First, more than 70% of employees WFH globally are on a hybrid schedule. This group comprises more than 100 million individuals, with the most common working pattern being three days a week in the office and two days a week at home. Second, most employees who are regularly WFH are university graduates in creative team jobs that are important in science, law, finance, information technology (IT) and other industries, rather than performing repetitive data-entry or call processing tasks.

This paper addresses the gap in previous studies in two key ways. First, it uses a randomized control trial to examine the causal effect of a hybrid schedule in which employees are allowed to WFH two days per week. Second, it focuses on university-graduate employees in software engineering, marketing, accounting and finance, whose activities are mainly creative team tasks.

Our study describes a randomized control trial from August 2021 to January 2022, which involved 1,612 graduate employees in the Airfare and IT divisions of a large Chinese travel technology multinational called Trip.com. Employees were randomized by even or odd birthdays into the option to WFH on Wednesday and Friday and come into the office on the other three days, or to come into the office on all five days.

We found that in the hybrid WFH (‘treatment’) group, attrition rates dropped by one-third (meancontrol = 7.20, meantreat = 4.80, t(1610) = 2.02, P = 0.043) and work satisfaction scores improved (meancontrol = 7.84, meantreat = 8.19, t(1343) = 4.17, P < 0.001). Employees reported that WFH saved on commuting time and costs and afforded them the flexibility to attend to occasional personal tasks during the day (and catch up in the evenings or weekends). These effects on reduced attrition were significant for non-managerial employees (meancontrol = 8.59, meantreat = 5.33, t(1215) = 2.23, P = 0.026), female employees (meancontrol = 9.19, meantreat = 4.18, t(568) = 2.40, P = 0.017) and those with long (above-median) commutes (meancontrol = 6.00, meantreat = 2.89, t(609) = 1.87, P = 0.062).

At the same time, we found no evidence of a significant effect on employees’ performance reviews, on the basis of null equivalence tests, and no evidence of a difference in promotion rates over periods of up to two years (‘Null results’ section of the Methods). We did find significant differences in pre-experiment beliefs about the effects of WFH on productivity between non-managers and managers. Before the experiment, managers tended to have more negative views, reporting that hybrid WFH would be likely to affect productivity by −2.6%, whereas non-managers had more positive views (+0.7%) (t(1313) = −4.56, P < 0.001). After the experiment, the views of managers increased to +1.0%, converging towards non-managers’ views (meannon-manager = 1.62, meanmanager = 1.05, t(1343) = −0.945, P = 0.345). This highlights how the experience of hybrid working leads to a more positive assessment of its effect on productivity—consistent with the overall experience in Asia, the Americas and Europe throughout the pandemic, where perceptions of WFH improved considerably.