A Soft Computing Methodology based on Fuzzy Measures and Integrals for Ranking Workers Informing Labour Hiring Policies

Effective policy-making and design for labour nationalization programmes requires a deep understanding of the factors underpinning firms’ decisions as regards the hiring of workers across different sectors of the economy, and crucially, how these factors interact in terms of either synergies or redundancies in the overall decision-making process. There is the need to develop a method that predictively determines the stability of employer-employee matches by ranking prospective and employed workers by combining information on firms, workers, and market or institutional variables. The objective of this paper is to present a methodology for transforming criteria in matched employer-employee data into a form expressing directly the variable importance for each match, that can then be used to estimate a fuzzy measure and corresponding Sugeno Fuzzy Integral to create an interpretable regression model that is able to predict the hiring patterns of firms given a pool of applicants. The SFI is explained and compared against three well-known benchmark regression methods in matched employer-employee data from the Kingdom of Saudi Arabia and shown to outperform them. Results on calculating the variable importance with the Shapley Value derived from the estimated fuzzy measures for two selected jobs are also presented, within the scope of a larger intervention model which can be used to aid policy-makers in both designing policies and evaluating their outcome.