Multiskilled personnel assignment problem under uncertain demand: A benchmarking analysis

The personnel assignment problem in different service industries aims to minimize the staff surplus/shortage costs. However, uncertainty in the staff demand challenges the accomplishment of that objective. This research studies the personnel assignment problem considering uncertain demand and multiskilled workforce configured through a 2-chaining strategy. We develop a two-stage stochastic optimization (TSSO) approach to calculate the multiskilling requirements that minimize the training costs and the expected costs of staff surplus/shortage. Later, we evaluate and compare the performance of the TSSO approach solutions with the solutions of two alternative optimization approaches under uncertainty - robust optimization (RO) and closed-form equation (CF). These two alternative approaches were published in Henao et al. [1] and Henao et al. [2], respectively. In addition, we compare the performance of the TSSO approach solutions with the solution of the deterministic (DT) approach and the solutions of myopic multiskilling approaches. To make performance comparisons between the different approaches, we used both real and simulated data derived from a retail store operating in Chile. The results show that, for different demand variability levels, TSSO, RO, and CF always belong to the set of approaches with the lowest average total cost. That is, in this group, there are no statistical differences from one approach to another, so these approaches are the most cost-effective. We also provide insights to retail decision-makers for addressing two key aspects. First, the methodology allows to address two fundamental multiskilling issues: how much multiskilling to add and how it should be added. Second, it is provided understanding on how to select the most suitable approach according to the balance between the conservatism and the reliability associated with the solutions delivered by each approach. Finally, we identify some methodological challenges for future research, such as the evaluation of k-chaining strategies with k≥2.