Abstract: Under on uncertain environment, using risk measure to resist the future uncertainties is an important way to avoid risks in newsvendor problem. Second order Stochastic Dominance (SSD) can be used as a robust risk measure. In this paper, we propose a risk aversion multi-product newsvendor model (SSD model) with SSD constraints and capability constraint. Moreover, the sample average approximation (SAA) method is used to approximate the problem, and the convergence analysis of the SAA problem is studied. Finally, in the numerical experiments section, we use the cutting plane method to solve SAA problem, at the same time it is compared with the reference models based on risk neutral (without risk constraints) and risk aversion (with variance as risk constraints). The numerical results show that the SSD model can avoid risks better and get higher return than the reference models under out-of-sample forecast.

Key words: newsvendor problem, second order stochastic dominance, capability, cutting plane method, sample average approximation