Sales forecasting and procurement considering supply and demand fluctuations: an application to the manufacturing industry under public emergency

The impacts of a public emergency on the manufacturing industry are reflected in limited supplies of raw materials and rapidly changing demands of the final products. It is difficult for the traditional sales forecasting models to capture the non-stationary simultaneous supply and demand fluctuations in sales data. To help manufacturers effectively grasp the sudden changes in raw material procurement, production and sales, this work proposes an extended Bass model to consider the simultaneous supply and demand fluctuations caused by public emergencies, and develops a dynamic framework of interactive sales forecasting and bi-objective optimization for procurement decisions. This study makes contributions by integrating sales forecasting and procurement into a coordinated problem and developing predictive and prescriptive models with supply and demand fluctuations. The real-world data of the automotive manufacturing industry in China during the COVID-19 pandemic are collected, and a semi-synthetic dataset is generated for the computational experiments. The extended Bass model considering the simultaneous supply and demand fluctuations outperforms the standard Bass model and some other benchmarks in sales forecasting. Using the sales forecasting results as inputs, the raw material procurement decisions under the static and rolling optimization determined by the solution of the bi-objective optimization model are also investigated.