Words matter for job applicants
13 Apr 2022
What jobs do women apply for in urban India, and how does the wording of adverts influence their choices?
Women continue to earn less than men with similar qualifications in the labour market, writes Zahra Siddique of the University of Bristol. In a new research paper, we examine whether this is because women apply to jobs which offer low wages. This may happen if employers looking to fill low-wage jobs also exhibit a preference for hiring women rather than men. We also look at the wording of job adverts, to see how this influences the pattern of applications.
The analysis is described in the paper “Words Matter: Gender, Jobs and Applicant Behaviour”, by Sugat Chaturvedi, Kanika Mahajan, and Zahra Siddique, to be presented at the annual conference of the Royal Economic Society. The investigation draws on 6.45 million applications made by male and female job seekers, in response to approximately 160,000 job advertisements posted on an online job portal in India, between July 2018 and February 2020.
Approximately 8% of these job ads include an explicit preference for hiring a male or female. Job ads which include an explicit female preference offer the lowest wages, but attract the highest share of female applicants.
The job portal we use primarily caters to young university graduates in the Indian urban labour market. Jobs advertised on this portal are high-skill jobs with posted wages which are, on average, 21% higher than wages earned by a nationally-representative and comparable sample of employed Indian workers.
Consistent with low female labour force participation rates in India, there are only half as many female as male applicants who search for jobs using the portal. However, these female applicants are better educated than the male applicants. We find that female applicants apply to jobs with lower advertised wages than similarly qualified men, so there is a gender wage gap in applications. Further, we find that this gap can be partly explained by employers expressing explicit gender requests or preferences.
To investigate what drives these relationships, we use methods from machine learning to retrieve words contained in job ads which are predictive of an explicit gender preference. We refer to these as “gendered” words. We assign gendered words to the categories of hard and soft-skills, personality traits, and job flexibility. We find that, where skills-related gendered words in job ads are predictive of an explicit female preference by the employer, these ads are associated with a lower offered wage, but attract more female applicants.
At the same time, where flexibility-related gendered words in job ads are predictive of an explicit male preference by the employer - these words indicate decreased flexibility, such as frequent travel or unusual working hours – these ads are associated with a higher offered wage but result in a smaller share of female applicants. This contributes to the gender wage gap in applications.
We also identify words in job ads which attract a higher share of female applicants. We find a positive correlation of such words with gendered words within the flexibility and hard-skills categories. However, there are zero and negative correlations of such words with gendered words within the soft-skills and personality categories respectively. So, while words in job ads such as “punctual”, ”smile”, and “pleasant” are highly predictive of an employer's female preference, we find that they reduce the share of women in the applicant pool.
Our research shows the willingness of women to pay for job flexibility in high skill jobs within a developing country. Given the lack of matched employer-employee data in developing country settings, our work also illustrates how applications data from job portals can be useful to study job search behaviour.
Finally, given that early career events or disruptions (“shocks”) have important cumulative consequences for future labour market returns, our results - using primarily entry-level job ads data - indicate that gender stereotypes are likely to be an important source of persistent gender disparities in the labour market.
Notes to Editors:
The press release is highlighting research papers presented at the RES Annual Conference 2022 (#RES20220) for further information, please contact firstname.lastname@example.org on 07970 201456 if you want the link to the full paper.
The Royal Economic Society’s popular 2022 virtual annual conference with in-person conference hubs, provides a forum for research, debate, and networking. The RES provides resources for economists and support for education and the training of students, teachers and researchers.
The Royal Economic Society’s purpose is to promote the study of economic science. With over 6,500 members worldwide, RES is one of the oldest economic associations in the world. RES are a registered charity and membership is open to anyone who shares our aims and objectives.
Please do tag RES on Twitter @RoyalEconSoc using #RES2022