Technical Note—Data-Driven Profit Estimation Error in the Newsvendor Model
研究发现数据驱动的报童模型中，预期利润的自然估计存在严格正偏，偏差可达真实预期利润的50%以上，并设计了一种调整方法得到无偏估计，帮助管理者准确预测未来利润。
An unbiased forecast of profit is important in most business environments. Typically, forecasts are generated from data. However, in “Technical Note—Data-Driven Profit Estimation Error in the newsvendor model,” Siegel and Wagner identify a strictly positive bias in a natural estimation of expected profit in a data-driven newsvendor model, where managers will expect more profit than will actually be realized, on average. This bias can reach significant proportions (in some cases 50%+) of the true expected profit and could therefore have undesired and damaging effects in the real world. Siegel and Wagner then design a data-driven adjustment that results in an unbiased estimator of expected profit, so that managers may have an accurate forecast of future profit that is free of systematic bias.