The emergence of low-skill manufacturing sectors in developing countries can increase labor market opportunities and provide other economics benefits for women (Heath and Mobarak, 2015; Tanaka, 2017). But in light of the poor conditions that characterize many low-skill manufacturing sectors, some researchers have questioned whether manufacturing jobs are actually better for workers than their alternatives. (Blattman and Dercon, 2018; Blattman, Dercon, and Franklin, 2019). Recent research by co-PI’s Boudreau and Heath provides evidence consistent with a model of information frictions around working conditions.
In this project, it will build upon our previous research (and a pilot RCT conducted by co-PI Boudreau) by experimentally investigating to what extent information and search frictions in Bangladesh’s labor market contribute to inefficient matching between workers and firms, and how these frictions interact with gender. Specifically, it will implement a cluster randomized controlled trial (RCT) that provides information about job characteristics (wages and working conditions), job openings, or both, and then assess the impact of treatment on outcomes such as their beliefs about working conditions and wages in the garments sector, job search activity, and employment outcomes. Women likely both particularly value safe workplaces and differentially lack information about working conditions and job openings. This project will then indicate whether policies to alleviate information frictions can help to close gender gaps in labor outcomes and improve the working conditions faced by women, allowing them to enjoy the benefits of these jobs without risking their health, safety, or wellbeing.
The research team will begin by conducting a household survey of a residential sample of garment workers in a sample of neighborhoods in Savar, Gazipur, and possibly Narayanganj, which are peri-urban regions of Dhaka home to large clusters of garments factories and to working-age populations with particularly high rates of employment in the garments sector. After the initial survey, individuals with less than two years of experience will be eligible to participate in the experiment. Because of the possibility of spillovers, treatment will be assigned at the neighborhood level. Within neighborhood, they first stratify eligible individuals by their gender, their beliefs about their factory’s quality relative to other factories nearby, and their factory’s actual quality relative to other factories, and then randomly select individuals within each treatment. The selected experimental sample will be given a longer survey that asks more detailed questions about their work history, current information sources, and perceptions of other factories.
Both the information about working conditions and job openings will be given on pamphlets that enumerators will go through in detail with participants. The information on working conditions will include both workers‘ reports of working conditions from the household survey and information from a unique collection of measures from buyer and government safety and other initiatives that the team has assembled. The project team will pilot different approaches to providing workers with continued access to this information, such as “refresher” phone calls or text messages to workers and a toll-free phone number that workers can call to receive the same information by phone. To gather information on job openings, they will send enumerators to factories during important hiring periods (namely, the first 5-10 days of the month) to find out whether they are hiring. They will provide vacancy information every month for a period of eight months via text messages and/or automated phone calls.
To collect follow-up data over the eight months of the treatment, both treatment and control participants will receive short, monthly Interactive Voice Response (IVR) phone-based surveys for eight months. The team will compensate participants for completing these surveys with airtime credit and send enumerators to follow-up in person with difficult-to-reach participants. In each of these rounds of data collection, the team will collect data on the four key outcomes of the study: respondents’ beliefs about the quality of their current factory, reported job quality, wages, and recently mobility. They are powered to detect, after conservatively adjusting for multiple hypothesis testing and allowing for an attrition rate of 25 per cent, an 8 percentage point increase in the probability of changing factories (the outcome that the research team identify as most difficult to impact) on an assumed control group mean of 30 per cent mobility.
This research contributes to growing literature on information and search frictions in labor markets, in particular in developing countries. While several recent experimental studies highlight that demand-side information frictions about worker ability impede efficient matches from occurring between workers and employers in developing countries (Abebe et al., 2018; Bassi and Nansamba, 2019; Carranza et al., 2019), there has been less empirical work on supply-side information frictions about job attributes. There is also evidence that information about job opportunities does not easily diffuse in developing countries (Jensen, 2012; Oster and Steinberg, 2013; Dammert et al., 2013; Beam, 2016; Abebe et al., 2018), and that experimental interventions reducing search frictions particularly help female workers (Abebe et al 2018). Our project will study the relative importance of search and information frictions, and how these vary by gender, in the same setting. This study will also contribute to a small body of literature that aims to estimate the Value of a Statistical Life (VSL) for populations in developing countries, by using an information intervention to estimate a VSL, and explicitly including both men and women.