Training in call centres

The intervention included a week-long training intervention targeted at call agents in the largest department of the call centre. The aim of the training programme was to increase the efficiency of agents answering customer calls. The management decided to organise the training in order to decrease the average time needed for handling calls, because the call centre performed below the targets set.

2009-03-09

2009-04-10

Worker productivity (as measured by average handling time)

Worker productivity is defined as the inverse of average handling time: y=100/(average handling time in week t measured in seconds)

Out of the 157 individuals working in the department at the beginning of our observation period (week 50/2008), 86 were selected for the experiment. Conditional on being selected for participation in the field experiment, 37 out of the 86 agents were randomly selected for participation in the treatment group. The remaining 49 agents who were assigned to the control group were trained after the post-experiment period.

Agents assigned to participate in the fi eld experiment were randomly assigned to be treated during the treatment period, or to be treated after the end of the experiment. Due to the restriction that agents should be trained with other agents from the same team, half of the teams were randomly assigned to the treatment group, whereas the other half was assigned to the control group. Each team was then randomly split up in different training groups, due to size constraints of the training centre.

Randomization was done on the PI's office computer

Main randomization unit: team; secondly: agents of teams were randomly assigned to training groups

Yes

10 teams

86 agents selected for the experiment

3 treatment; 3 controls

This paper analyses the effects of work-related training on worker productivity. To identify the causal effects from training, we combine a field experiment that randomly assigns workers to treatment and control groups with panel data on individual worker performance before and after training. We find that participation in the training programme leads to a 10 percent increase in performance. Moreover, we provide experimental evidence for externalities from treated workers on their untreated teammates: An increase of 10 percentage points in the share of treated peers leads to a performance increase of 0.51 percent. We provide evidence that the estimated effects are causal and not the result of employee selection into and out of training. Furthermore, we find that the performance increase in not due to lower quality provided by the worker.

Andries De Grip, Jan Sauermann (2012), "The Effects of Training on Own and Co-Worker Productivity: Evidence from a Field Experiment", Economic Journal, 122(560), pp. 376-399.

https://ideas.repec.org/p/iza/izadps/dp5976.html

Do reciprocal workers have higher returns to employer-sponsored training? Using a field experiment with random assignment to training combined with survey information on workers' reciprocal inclinations, the results show that reciprocal workers reciprocate employers' training investments by higher posttraining performance. This result, which is robust to controlling for observed personality traits and worker fixed effects, suggests that individuals reciprocate the firm's human capital investment with higher effort, in line with theoretical models on gift exchange in the workplace. This finding provides an alternative rationale to explain firm training investments even with the risk of poaching.

Sauermann, J. (2023). Worker reciprocity and the returns to training: Evidence from a field experiment. Journal of Economics & Management Strategy, 32, 543–557.

https://doi.org/10.1111/jems.12419

We assess selection bias in estimated returns to workplace training by exploiting a field experiment with random assignment of workers to a one-week training program. We compare experimental estimates of this program with non-experimental estimates that are estimated by using a sample of agents who were selected by management not to participate in the experiment. Our results show that non-experimental estimates are biased, yielding returns about twice as large as the causal effect. When controlling for pre-treatment performance or individual fixed effects, only about one tenth of this bias remains and is even further reduced when applying common support restrictions.

Sauermann, Jan, and Anders Stenberg, "Assessing Selection Bias in Non-experimental Estimates of the Returns to Workplace Training", IZA DP No. 13789]

https://www.iza.org/publications/dp/13789/assessing-selection-bias-in-non-experimental-estimates-of-the-returns-to-workplace-training