Leonhard Grabe

We investigate the dynamics of skill management in organizations through a natural field experiment with 2,582 service technicians. Employees in both treatment and control groups could self-assess skills and propose training measures to managers. An intervention restricting managers’ access to objective skill information reduced training intensity, work performance, and job satisfaction – indicating underinvestment in human capital, even with employee-initiated development. Using detailed personnel records and survey data, we show that the intervention dampened both employees’ efforts to identify training needs and managers’ engagement in skill management. In particular, high-skill employees received less training aimed at broadening their skill set, and performance declined due to longer completion times for complex assignments. These findings highlight the pivotal role managers play in fostering skill development and workforce productivity.

We investigate whether organizations can shape work habits through monetary incentives. In a field experiment with 829 service technicians in 15 firms, we randomly allocated half of the technicians in each firm to a treatment group receiving bonuses for regularly performing sales activities for 12 weeks. We find a significant increase in sales activities not only during but also after the incentive phase. Using data from a post-experimental survey, we compare different behavioral channels. We find no evidence for increased automaticity, human capital acquisition, or signals about task priorities, but strong evidence for the role of acquired taste: Technicians in the treatment group report higher levels of intrinsic motivation to perform sales activities even after the incentive has been discontinued.

This paper studies how holding employees accountable for their errors affects their future performance. While errors are ubiquitous, the behavioral consequences of being confronted with past errors by their managers remain unclear. Using proprietary performance and error records from more than 3,200 service technicians across 50 firms, this paper shows that such confrontations do not improve output quality and reduce customer satisfaction. Instead, output quantity increases by 5-7%, suggesting that employees are reallocating effort towards tasks that are more readily observable to their managers. These effects are driven by employees with shorter tenure and lower prior performance who have stronger incentives to restore their reputation. A back-of-the-envelope calculation suggests that the positive effect on output quantity is twice as large as the negative effect on customer satisfaction. These findings contribute to the field literature on multitasking and the broader literature on reputational concerns in firms.

Firms increasingly rely on technology to support employee skill development, raising the question of whether such technology complements or substitutes for the supervisor's role. We study this question in a nine-month field experiment with a large European service firm, randomly assigning 187 teams to a control condition, a complement condition that makes development conversations salient and prompts supervisors to hold them, or a substitute condition that lets supervisors schedule training directly without a development meeting. The complement condition increases trainings, development interactions, employee knowledge, job satisfaction, perceived supervisor support, retention, and performance on the most skill-intensive tasks. The substitute condition raises employee knowledge but moves none of the broader outcomes. Technology could substitute for the supervisor in producing the measurable skill gain, but not in producing the relational and operational benefits that flow through supervisor engagement. Our findings inform the design of management controls for skill development.