Forecasting demand variation when there are stockouts

• Published in: The Journal of the Operational Research Society Vol. 51; no. 3; pp. 358 - 363
• Main Author: Bell, P C
• Format: Journal Article
• Language: English
• Published: London Taylor & Francis 01.03.2000; Macmillan Press; Palgrave Macmillan UK; Palgrave; Taylor & Francis Ltd
• Subjects: Analytical forecasting; Applied sciences; Arithmetic mean; Business and Management; Data smoothing; Demand; Exact sciences and technology; exponential smoothing; Forecasting; Inventory control; Management; Maximum likelihood estimation; Operational research and scientific management; Operational research. Management science; Operations Research/Decision Theory; Optimization; Planning. Forecasting; Retail outlets; Retail stores; Sales forecasting; Standard deviation; Statistical variance; stochastic inventory problem; Stochastic models; Stock sales; stockouts; Studies; Technical Note; Technical Notes; stochastic inventory problem; exponential smoothing; stockouts; forecasting; Demand; Exponential smoothing; Storage; Demand forecasting; Distribution; Optimal policy; Variations; Stockout; Forecasting
• ISSN: 0160-5682; 1476-9360
• DOI: 10.1057/palgrave.jors.2600877

This paper addresses the common problem of forecasting demand when there are a large number of stockouts. The well-known single period stochastic inventory (or 'newsboy') problem provides the optimum, single period, stocking level for a product subject to stochastic demand. There are many situations where repetitive 'newsboy' solutions are implemented to guide stocking of repeat, but related, products, such as newspapers, magazines, or perishable groceries. Implementation of the 'newsboy' solution requires forecasts of the distribution of demand, although there are many plausible cost parameters that lead to optimum stocking policies where there is a high probability of a stockout. The company is, therefore, faced with the problem of attempting to forecast demand when a high percentage of the available sales data reflects the stock available for sale, rather than the true demand. A procedure has been developed 1 to improve estimates of the mean and variance of the distribution of demand when there are stockouts, but this procedure fails when the percentage of stockouts increases above 50%. A modified stockout adjustment procedure is presented in this paper, and it is shown that use of this new procedure can lead to greatly improved estimates of demand parameters, and greatly improved profitability, when there are a high percentage of stockouts.