Impacts of Soft Skills Management Training on Productivity and Worker Retention and Welfare

We propose to study two soft skills training programs in garment factories in India, one for production line supervisors (managers overseeing frontline machine operators and helpers) and the other for upper-level managers (so-called “floor in-charges”). The training programs were designed by our research team in collaboration with Shahi Exports (garment firm partner) and O&S (Bengaluru-based personnel consultancy). Supervisors in the program attend training 1.5 hours per week for 25 weeks. The analogous program for upper-level managers will involve roughly half the total hours of training and will be administered over about 8 weeks. The training modules in both programs cover topics such as communication, self-esteem, gender sensitivity, problem solving, planning work, and preventing harassment at the workplace.
The goal of the training program is to improve the soft skills of line supervisors and thereby increase the productivity and retention of workers. The main hypothesized mechanisms for these effects are increases in effective manager-worker communication, time management and production planning improvements, greater sensitivity to gender issues, and an overall increase in the quality of the work environment for frontline workers.
Accordingly, data on training progress, attendance, completion, and pre and post testing will be matched to administrative data on workplace outcomes and comprehensive survey data. The administrative data include both worker and line-level productivity, worker attendance and tardiness, payroll and retention. The survey will involve a “360” evaluation of supervisors including measures of styles and practices from line supervisors themselves, both objective subjective assessments of managerial quality from upper-level managers overseeing supervisors, and incidences of abuse and harassment as well as job satisfaction measures from workers reporting to line supervisors.

2017-04-01

2018-07-30

management, productivity, attention, retention, workplace environment

workplace environment (measured by addressing abuse and harassment, communication, time management, and production planning); management (measured by management index using information on how the supervisor treats underperformers, star performers, and how supervisor retains talent, standardized monitoring index by the frequency of visits to the line by the supervisor, and standardized line problem identification index by how any problem is identified in the line.)

The trials will take place in 41 factories among about 1500 upper-level managers and 4000 supervisors overseeing around 70,000 workers.
The first stage of the trial is a randomized phase-in trial in which approximately half of supervisors within each Shahi factory are randomized (stratifying by upper-level managers’ predictions of gains for each supervisor at baseline) to the first phase of soft skills training, and the remainder to the second phase (post “endline” outcome measurement). Before randomization of supervisors, their managers (floor in-charges) were asked to rank supervisors in their team in terms of who would gain most from the training. After the intervention concludes we will study how well upper-level managers were able to identify supervisors who would gain from the training. Because of concerns about spillovers, we also randomized the share of supervisors on each production floor of each factory that received training (to either 30% or 70%), helping us identify these spillovers and come closer to a pure treatment effect (for the 17 factories we have 70 distinct factory-building-floor units for this randomization).
The second stage (the upper level management trial) is a cross-cutting randomized phase-in at the factory level of soft skills training for upper level managers of the supervisors that were part of the first stage trial (For midline - we have surveyed all floor in charges aka upper level managers, not restricted to those reported by supervisors). We are also concerned with the issue of spillovers for this group. In this case, we are not able to randomize the intensity of training (or proportion trained) at the factory floor level. Instead, we will assign upper-level managers to treatment groups at the factory level. This way the control group will not be affected by spillovers and impacts on workers under a particular upper-level manager will be interpreted as the impact of that manager’s training as well as the training of other managers in the same unit.

Randomization done by a statistical software on a computer

1. In the first stage of the experiment, the randomisation is at individual supervisor level.
2. In the second stage of the experiment, the randomization is at factory level with total number of clusters being equal to 41

Yes

1. In the first stage of the experiment, the randomization is at individual supervisor level
2. In the second stage of the experiment, the randomization is at factory level with total number of clusters being equal to 41

We will survey the Upper-level managers (500-600), all the supervisors (4000). We will also survey a sample of workers (4000) in the factories to understand Supervisor performance.

1. In the first stage of the experiment, the sample size will be 1900 supervisors.
2. In the second stage of the experiment, the experiment will cover 41 factories with each factory having on an average 12-15 managers.
Also in the second stage we will cover remainder of the supervisors. (new + old who weren't covered in baseline and are still working at Shahi)