Profit Implications of Judgmental Adjustments to Forecast Inputs: Evidence from a Large-Scale Field Experiment

In this paper, we report the results from a large-scale field experiment at a spare parts retail chain that considers whether allowing merchants to override forecast inputs to an inventory algorithm improves profits. Although the judgmental forecasting literature has studied extensively whether judgmental adjustments improve forecast performance, causal empirical evidence is missing in regard to whether judgmental adjustments improve bottom-line profits. Our results show that judgmental adjustments to the forecast input increase profitability by 4.92% on average compared with relying on automation without human intervention. We find that the well-established motivation-opportunity-ability framework provides clear insight into when judgmental adjustments improve profits, by examining heterogeneity in our data regarding stock-keeping unit margin, lifecycle, and size of supplier. Our data set also allows for examining both forecast accuracy and profits. We empirically support the wisdom from the judgmental forecasting literature that forecast performance need not translate to profit performance, calling attention to the need to consider operational performance beyond forecast accuracy as an end in itself. This paper was accepted by Felipe Caro, Special Issue on the Human-Algorithm Connection. Supplemental Material: The online appendices and data files are available at https://doi.org/10.1287/mnsc.2024.06321 .