Hybrid approach to modeling labor productivity factors: Synthesis of randomized controlled experiments and causal Bayesian networks

Solving the problem of effective management of labor productivity of company employees, taking into account many heterogeneous factors, often of a stochastic, non-stationary and non-linear nature, embedded in complex chains of cause-and-effect interactions in the context of digital transformation of the economy presents certain difficulties. The paper proposes a technology that ensures, with a high degree of certainty, the establishment of causal relationship between the implementation of alternative management decisions and the productivity of company employees, and is designed to select solutions based on an assessment of the effect of their impact on labor productivity. The novelty of the proposed technology is based on a hybrid approach to modeling the object of study and combines two models. First model – a structural model built on the basis of a priori knowledge of the laws of functioning and development and providing a causal understanding of the object and capable of predicting the effect of factors (explicit and indirect). Second model – a model based on data, which is tuned (adapted) taking into account empirical data obtained as a result of observation (measurement) of an object. The developed technology uses heterogeneous research methods —a randomized controlled experiment to obtain information about the tested activities, statistical data analysis —descriptive data analysis, correlation and regression analysis, the difference-difference method to establish a causal relationship between the implemented event and the growth of labor productivity, a Bayesian network of causality for building and analyzing a structural model of an object and explaining the causal relationships of explicit and hidden factors that affect labor productivity in the context of the implementation of measures. Of practical significance are the results of testing the proposed theoretical provisions, methods and technologies on actual data on the activities of a food service company. The results obtained will contribute to the effective use of the developed technology aimed at ensuring the growth of labor productivity under uncertainty in the external and internal environment and will contribute to the sustainable development of companies and the growth of its profitability.