Newsvendor Conditional Value-at-Risk Minimisation with a Non-Parametric Approach
share
In the classical Newsvendor problem, one must determine the order quantity that maximises the expected profit. Some recent works have proposed an alternative approach, in which the goal is to minimise the conditional value-at-risk (CVaR), a very popular risk measure in financial risk management. Unfortunately, CVaR estimation involves considering observations with extreme values, which poses problems for both parametric and non-parametric methods. Indeed, parametric methods often underestimate the downside risk, which leads to significant losses in extreme cases. The existing non-parametric methods, on the other hand, are extremely computationally expensive for large instances. In this paper, we propose an alternative non-parametric approach to CVaR minimisation that uses only a small proportion of the data. Using both simulation and real-life case studies, we show that the proposed method can be very useful in practice, allowing the decision makers to suffer less downside loss in extreme cases while requiring reasonable computing effort.
READ FULL TEXT