Seminar Series

2023 - 2024

Speaker: Reeju Guha (IE, Madrid).

Title: “Gig workers’ learning and task allocation: Evidence

from an on-demand grocery platform ”.

joint with Daniel Corsten (IE Business School).

Date: Tuesday, February 27th at 12h30 (Paris Time).

Abstract: On-demand platforms operating on a gig-contractor model rely on self-scheduling gig workers to meet realtime demand, while maintaining desired levels of productivity and service quality. Such platforms either adopt a broadcast mechanism, where workers have the autonomy to select orders, or a dispatch mechanism, where the platform assigns orders to workers. In most platforms, tasks are randomly allocated, or assigned without considering gig workers’ behaviors and skills. Commonly used allocation mechanisms such as experiencebased or ratings-based allocations are not useful in new geographies where gig workers might not have sufficient experience or ratings. Drawing from theories of learning, and flow, we investigate whether withinday learning can positively impact gig workers’ performance, and thus be utilized as a ranking parameter in task allocations. Utilizing data from an on-demand grocery platform, we develop an econometric model to analyze workers’ productivity and service quality based on their within-day experience, while accounting for sample selection and endogeneity. Our findings reveal that as same-day experience increases, productivity and service quality improve. Since our objective is to allocate work based on task characteristics, we next examine the role of learning in the presence of task batching and complexity, on performance.We observe that when workers batch orders, same-day experience reduces delays, but also reduces picking productivity and item substitutions. We further observe, for complex tasks, higher same-day experience is beneficial up to a threshold, beyond which it improves picking productivity but reduces stockout-based substitutions. Utilizing results from our analysis, we rank gig workers based on prior and same-day experience, and develop a task allocation algorithm to allocate tasks based on workers’ rank, and task complexity. We predict improvements in productivity and service quality from the new allocations and demonstrate that allocating higher-ranked workers to complex tasks while limiting order batching leads to improvements in performance. 

Here is, a link to the speaker’s and the FEELab website:

https://reejuguha.com/

https://sites.google.com/view/feelabtbs/

You are cordially invited to participate in the seminar, which will take place in Room 321, Lascrosses building. 

For more information, please contact: Pierre Mella-Barral p.mella-barral@tbs-education.fr