On learning process of a newsvendor with censored demand information

A behavioural study on the newsvendor problem without demand distribution knowledge or realized demands is conducted. We configure a censored‐information scenario, where only sales data in history are provided. With laboratory experiments, we analyse participants ordering levels and ordering oscillations over time. It is observed that participants perform a significant learning process in the censored‐information scenario. Moreover, we find that participants make orders by anchoring on the previous‐period sales and adjusting to an adapted inventory level. On the basis of the observations, we propose an exponential‐type learning (EXP) model to describe the behaviour of decision makers. Comparing with a chasing model that is popularly used in the full‐information scenario, the EXP model is more recommended in the censored‐information scenario.