Managerial (dis)preferences towards employees working from home: Post-pandemic experimental evidence

Work from home (WFH) has been a part of the professional landscape for over two decades, yet it was the COVID-19 pandemic that has substantially increased its prevalence. The impact of WFH on careers is rather ambiguous, and a question remains open about how this effect is manifested in the current times considering the recent extensive and widespread use of WFH during the pandemic. To answer these questions, this article investigates whether managerial preferences for promotion, salary increase and training allowance depend on employee engagement in WFH. We take into account the employee’s gender, parental status as well as the frequency of WFH. Furthermore, we examine whether managers’ experience with WFH and its prevalence in the team moderate the effect of WFH on careers. An online survey experiment was run on a sample of over 1,000 managers from the United Kingdom. The experiment was conducted between July and December 2022. The findings indicate that employees who WFH are less likely to be considered for promotion, salary increase and training than on-site workers. The pay and promotion penalties for WFH are particularly true for men (both fathers and non-fathers) and childless women, but not mothers. We also find that employees operating in teams with a higher prevalence of WFH do not experience negative career effects when working from home. Additionally, the more WFH experience the manager has, the lesser the career penalty for engaging in this mode of working. Our study not only provides evidence on WFH and career outcomes in the post-pandemic context but also furthers previous understanding of how WFH impacts careers by showing its effect across different groups of employees, highlighting the importance of familiarisation and social acceptance of flexible working arrangements in their impact on career outcomes.


Sex men women
Skills rank (1 very weak, 5 very strong) social 2, analytical 3 social 3, analytical 2 Age 40 years old 38 years old

Which employee would you give a promotion to?
Worker A Worker B

Which employee would you give a salary increase to?
Worker A Worker B

Which employee would you give training to?
Worker A Worker B Which employee do you consider to be more committed?
Worker A Worker B Which employee do you consider to be more competent?
Worker A Worker B Deviations from the pre-registered plan: Our study procedure and analysis plan were preregistered on the Open Science Framework https://osf.io/n2byz?view_only=7acb8a6da1c141fc8ca4f4391feea700.The following study pertains to the 'Component 1' section of the pre-registration.
In the pre-registration, we stated the following hypothesis: H1: The HBW effect is stronger for women than men (Gender effect) H2: The HBW effect is stronger among parents than non-parents (Parenthood effect) H3: The parenthood effect is stronger for male than female employees (Fatherhood vs motherhood effect) However, rather than analysing these effects in separate models (i.e. with the use of sub-groups) as previously indicated, we decided to run a triple interaction model in order to better account for the differences between these groups.In addition, we run all models only on the subsample of records where the performance rank was not revealed to the respondent (rather than in comparison with subsample with known performance rank).We decided to focus on the subsample with unknown performance in the first step to display the effects of WFH which closely resemble the 'real life' situation where managers are not fully aware of the productivity levels of their employers and may have certain pre-assumptions about those who WFH.We will compare the performance effect (known vs. unknown performance) in a second step as a sperate study in which we explore the role of the mechanisms behind the WFH effects observed in the 'real' world.Finally, 'Component 1' section in the pre-registration did not include information on the use of 'WFH prevalence' as a moderator in this study.This does not necessarily constitute a deviation from the original research plan as we stated that we may want to conduct exploratory analysis.However, we believe it is important to point out that the idea to use WFH prevalence as a moderator was born after the pre-registration.

Study questionnaire:
Participants first answered screening questions (sample restrictions are explained in the Data & Methods section).Then they were presented with information on the experiment and instructions (see p. 1 of the Appendix).Subsequently, they were tasked with selecting one employee from each pair for promotion, salary raise, training, and which employee they consider to be more competent and more committed to work.They then answered various questions about themselves and the company that they work within, which are presented in this section below.
Please, answer the following questions about the company that you work in.

Fig 1 .
Fig 1.The predicted probabilities for being chosen for promotion by the interaction of working mode, gender, parenthood status and the prevalence of WFH in the team: logit models.

Fig 3 .
Fig 3.The predicted probabilities for being chosen for training by the interaction of working mode, gender, parenthood status and the prevalence of WFH in the team: logit models.

Fig 4 .
Fig 4. The predicted probabilities for being chosen for promotion by the interaction of working mode, gender, parenthood status and the manager's frequency of WFH: logit models.

Fig 5 .Fig 6 .
Fig 5.The predicted probabilities for being chosen for salary raise by the interaction of working mode, gender, parenthood status and the manager's frequency of WFH: logit models.

Please, indicate which measures are provided in your company (Yes / No / Don't know): <1>
Support with childcare (e.g. company childcare facility, cooperation with childcare facilities, childcare during school holidays) or financial contributions towards childcare <2> Additional employer-funded childcare-related leave (e.g.maternity/paternity/parental, in case

Do you think the proportion of workers who currently work from home in your company will eventually: <1>
Decline to the levels observed from before the COVID-19 pandemic <2> Decline but will be higher than before the COVID-19 pandemic <3> Will remain as it is now <4> Will increaseDo you currently work from home at least from time to time?What is the major reason for which you currently work from home? <1> Work-life balance / Family obligations

indicate how much you agree with the following statements.
<1> Men should do as much housework and childcare as women.<2> A man's primary job is to earn money for the family, not to look after the children.<3> A woman's primary job is to look after the home and family.<4> All in all family life suffers when the mother has a full-time job.

Table 2 . The odds ratio of being chosen for promotion, salary increase, and training by the interaction of working mode and prevalence of WFH in the team: logit models.
The prevalence of WFH in the team is measured by the question 'How many of the workers under your supervision work from home at least one day a week on a regular basis?' with answers coded into three levels <40% (Low WFH), 40-79% (Moderate WFH) and 80+% (High WFH); SE in parentheses *** p<0.01, ** p<0.05, * p<0.1

Table 6 . The odds ratio of being chosen for promotion, salary increase, and training by the interaction of working mode, gender, parenthood status and manager's frequency of WFH: logit models.
Notes: The same as inTable 2 *** p<0.01, ** p<0.05, * p<0.1