Productivity & Innovation

Bringing work home: Flexible work arrangements as ‘gateway jobs’ for women

  • Blog Post Date 24 May, 2024
  • Articles
  • Print Page
Author Image

Lisa Ho

Massachusetts Institute of Technology

Author Image

Suhani Jalota

Stanford University

Author Image

Anahita Karandikar

University of British Columbia

Millions of women stay out of the workforce despite having a desire for paid work, often because available opportunities are incompatible with traditional norms of household roles. Based on an experiment in West Bengal, this article shows that flexible work arrangements significantly raise women’s take-up of jobs. Although home-based work reduces worker productivity, once women have some experience with flexible work, they are more likely to accept future outside-the-home jobs.  

Women have a large unmet desire for jobs: it is estimated that several hundred million women across the world are outside of the labour force despite exhibiting a preference for paid work (Gallup and International Labour Organization, 2017). We hypothesise that one important reason is a mismatch between the supply of jobs available and the jobs women can or want to do given existing norms of appropriate behaviour for women. This mismatch between available jobs and gender norms can be addressed through two possible strategies: first, change norms so that women incur smaller costs when taking up existing jobs; or second, change jobs to make them more compatible with existing norms. Previous studies show that it can be difficult to directly influence attitudes towards women’s participation in paid work (for example, Dean and Jayachandran (2019)), but we know less about how modifying jobs to be more compatible with current norms would affect women’s employment and related outcomes. Creating work opportunities that are more aligned with current norms might allow more women to experience paid work – even if only short term – which, in turn, might affect household members’ beliefs about women’s abilities as well as attitudes about appropriate gender roles, potentially shaping women’s future labour supply decisions about working from home and outside the home. 

Using a field experiment in West Bengal, we test the effects of offering jobs that are relatively more compatible with existing norms of women’s behaviour (Ho, Jalota and Karandikar 2024). In our setting, the widespread expectation is that the women are primarily responsible for household chores and caregiving. Motivated by the growth of online gig work opportunities in developing countries (Datta et al. 2023), our experiment includes piece-rate data entry jobs on an online platform that allow women to work from home, multitask work with childcare and/or choose their work hours. Our study answers three main questions: First, does offering flexible work arrangements increase female labour force participation, and if so, which dimensions of flexibility are important? Second, what are the effects of flexibility on job performance and the types of workers drawn into these jobs, and what does this suggest for the viability of firms introducing flexible arrangements? Third, since flexibility may still not be compatible with many types of jobs, can flexible jobs act as a gateway to less flexible jobs for women in more traditional households who may initially only be allowed to work from home? 

The study

We conduct a randomised controlled trial with 1,670 women from lower-middle-income households in a mix of urban, peri-urban, and rural areas in and around Kolkata. Recruitment is conducted house-to-house and through meetings conducted by our partner NGO, the Calcutta Foundation. Our participants are mostly married women with little prior work experience. We exclude women who are currently in the labour force, and we avoid selecting the sample based on interest in work by not mentioning a potential job opportunity during study recruitment. We randomise participants to either receive a job offer (the ‘treatment’ group) or into a ‘control’ group with no offer. The offers vary along three dimensions of flexibility: (i) the ability to choose work hours each day, (ii) the ability to multitask childcare while working, and (iii) the ability to work from home. All job offers are part-time, last for one month, and are offered in partnership with Project Karya at Microsoft Research. The work involves completing tasks that contribute to a Bangla or Hindi speech dataset that can help train language models. 

To understand how introducing flexible work arrangements impacts employers, we estimate effects on worker performance, separating effects that operate via ‘selection’ (changes in worker composition due to enhanced flexibility, as certain types of workers may be more likely to be drawn to flexible work) and those that are taking place on account of the intervention. With regard to selection, we randomly choose half of the treated participants who accepted a less flexible job to be ‘upgraded’ to the most flexible job  – where there are no restrictions on work time, participants can combine work with childcare, and work from home (as in Karlan and Zinman (2009)). To measure the effect of selection, we hold actual job flexibility constant and compare the performance of workers who were willing to do less flexible jobs with the performance of workers who were only willing to do the most flexible job. 

To measure the effect of work flexibility, we hold ‘worker type’ constant – looking only at women who were initially assigned to and who accepted an inflexible job. Within this group, we compare the performance of those who were randomly selected for an upgrade to the most flexible job, with those who were not made this offer and hence stayed in the inflexible job. We measure the effects of work experience on gender attitudes once the jobs are completed, and two to three months later, we measure take-up of different work opportunities, which include non-digital jobs that involve working outside the home. 

Key findings

We find three sets of main results. First, flexible work arrangements more than triple women’s job take-up, from 15% for the office job to 48% for our maximally flexible job which accommodates working from home, choosing work hours flexibility, and multitasking work with childcare. We present the effects of varying each dimension of flexibility in Figure 1. Work from home, even without multitasking work with childcare, doubles job take-up from 14.6% to 29.2%. The ability to multitask work with childcare is also important, increasing take up by 59%. Thus, of the 33 percentage point effect of job flexibility, about half (16 percentage points) comes from the ability to work from home and the other half (17 percentage points) comes from the ability to multitask work with childcare. 

Figure 1. Effects of flexible work arrangements on women’s job take-up

Notes: (i) This figure plots the take-up rate for each of the five jobs during the initial intervention. (ii) The left panel plots job take-up for 1,250 treatment group participations, each of whom receives one job offer. (iii) Take-up is measured as a binary variable equal to one if the participant starts work (that is, submits completed tasks to the employer). (iv) The vertical lines indicate 90% and 95% confidence intervals from a regression of job take-up on ‘dummy variables’ for each of the four jobs other than “Flex”. A confidence interval is a way of expressing uncertainty about estimated effects – specifically, it means that if the study was repeated over and over with new samples, 90% and 95% of the time, respectively, the calculated confidence interval would contain the true effect. (v) The right panel describes how the five jobs turn on the ability to choose work hours flexibly, multitask work with childcare, and work from home. 

Our estimates show that offering jobs with greater flexibility would dramatically increase a firm’s pool of potential workers – so why do employers not offer flexible work arrangements more often? Beyond fixed costs (such as investing in systems and software to allow workers to work remotely) and feasibility, employers would likely want to understand how offering flexible work arrangements affects job performance, both in terms of the type of workers drawn into the firm as well as effects on performance of those who were willing to work under less flexible conditions. We shed light on both these effects using the random upgrades feature of our design and three measures of job performance: (i) retention – whether a worker who accepted the job actually starts work and stays through the end of the job, (ii) quality – the accuracy with which the worker completes tasks and the number of tasks completed, and (iii) efficiency – the speed with which workers complete tasks. Among workers who would accept the job irrespective of flexibility, the ability to work from home improves reliability by reducing the fraction of workers who accept the job but never start work. However, working from home decreases accuracy and efficiency: home-based workers spend more time completing the same amount of work, and they make more mistakes. In terms of worker selection, we find that women who would only accept flexible jobs are equally reliable in terms of starting work after accepting the job and staying through the end of the job but work more slowly than women whose job take-up is not marginal to job flexibility. Thus, we find negative effects of home-based work on performance. 

Finally, given that the transition from being an unpaid homemaker to working outside the home may be a large leap for women and their households, we ask if short-term, flexible jobs offered in the study could act as stepping-stones to work outside the home. We hypothesise that flexible jobs could allow previously excluded women to take multiple smaller steps into the labour market. To test this ‘gateway jobs’ hypothesis, we offer all study participants another job two to three months after the endline survey. Once again, jobs vary in their flexibility as well as the type of work, including both the same online gig-work and other forms of non-digital piece rate work. We find evidence in support of the gateway jobs hypothesis – women who were assigned flexible work in the initial offer round are 5 percentage points more likely to take up a less flexible job assignment during the second round of offers, compared to women who did not get a similar gateway experience. This effect is concentrated on women who had no previous work experience prior to participation in the experiment. One channel that is consistent with this gateway effect is that work experience causes a change in attitudes about appropriate behaviour for women. Indeed, we find that job flexibility makes the biggest difference to labour supply for women with more traditional gender attitudes at baseline, and in turn, work experience changes the attitudes of more traditional women to become less traditional. This result highlights that there is a mutually reinforcing relationship between women’s employment and gender attitudes that support women’s work. 

Policy implications and conclusion

One policy implication of our findings is that offering flexible work arrangements could be an effective way to both recruit and retain women workers. However, flexible work arrangements might be underprovided for at least two reasons. One is that firms may underestimate the effects of job flexibility on labour supply or overestimate negative impacts of work from home on worker performance. A second is that even if workers are more likely to accept office-based jobs after having some experience with flexible, home-based work, this positive externality is unlikely to be considered by employers. To overcome these forces that could drive under-provision of flexible work arrangements, temporary subsidies or assistance for firms transitioning jobs to be more flexibility-friendly could have long-term positive effects on women’s employment. 

One important point of caution, however, is to prevent the gateway approach from ‘trapping’ women in at-home jobs that may also be lower paying – to ensure this, it may be important that these intermediate jobs are temporary in nature. We identify this issue, as well as the labour-demand side, as an avenue for future research: what are the implications of flexible work arrangements for the gender segregation of labour markets? And are firms well-informed about the potential benefits of job flexibility, and how can they introduce greater flexibility while maintaining the benefits of work-from-home with the fewest sacrifices to worker performance? 

I4I is now on  Substack. Please click here (@Ideas for India) to subscribe to our channel for quick updates on our content

Further Reading

No comments yet
Join the conversation
Captcha Captcha Reload

Comments will be held for moderation. Your contact information will not be made public.

Related content

Sign up to our newsletter