Two Types Of Newsvendor Problem Under The Random Environment

The newsvendor problem is a common problem in economic life,involving many areas such as supply chain management,inventory management,and risk management.It mainly studies the ordering and sales of products with short life cycle,unsuitable storage and corrosives,and has a wide range of applications.The classic newsvendor problem is a single-product single-stage inventory problem,which is important in supply chain because of the simplicity and is easy to solve.The multi-product multi-stage newsvendor problem is a dynamic process of continuous ordering and sales,which is instructive for reality because it is in good agreement with real economic activities.There are many uncertainties in the market environment,such as product demand,traffic conditions and so on.Scholars use random environments to portray these uncertainties.Market activity in a random environment is risky which needs to be quantified.This paper mainly uses risk measurement and multi-stage decision-making to avoid risks separately,then to study the multi-product newsvendor problem in random environment.(1)Given the development of risk theory,and the economic model with risk mea-surement can better help decision makers to deal with the risks brought by uncertainty.The first type of newsvendor problem studied in this paper considers the market demand as random environment,and incorporates the risk measure.Firstly,under the constraint of ordering ability,the risk aversion newsvendor model with second-order stochastic dom-inance(SSD)constraint is established.Among them,SSD as a risk-averse risk measure can describe the behavior of risk-averse investors,which is more stable than other risk measures.Secondly,the sample average approximation(SAA)method is used to approx-imate the problem,and the convergence of SAA problem is analyzed.Finally,in the numerical experiments,the tangent plane method is used to solve the SAA problem,and at the same time it is compared with the risk neutral(no risk constraint)and the risk aversion model with variance as the risk measure.The numerical results show that the SSD model can better avoid risks and obtain higher returns than the reference models.(2)The randomness of market environment,including logistics network is increasing,which promotes the evolution of manufacturers' production and sales behavior towards multi-varieties,small batches and multi-stages.The second type of newsvendor problem studied in this thesis is that in a complex transportation network,the transportation status of each path is a random variable that changes with time,and the price of the product is affected by the random environment of the market;In the interest of the man-ufacturer,a multi-product multi-stage newsboy model in the transportation environment was proposed.The model is then solved using the Stochastic Dual Dynamic Program-ming(SDDP)algorithm.In the numerical experiments section,the two-stage model is compared with the multi-stage model.The numerical results show that as the number of stages increases,the ultimate benefit of the multi-stage model will gradually increase.