Managing inventory with two suppliers under yield uncertainty and risk aversion
Introduction
Many inventory models have been developed based on the assumption that items are replenished from a single supplier. However, in practice, the sole supplier often fails to meet the retailer's demand due to various reasons. These include insufficient supply of raw materials, production of poor quality products, machine breakdown, workers strike, and so on. Business organizations use a secondary supplier or multiple suppliers today to maintain a desirable service level or to reduce customer service time or even to reduce costs. Dual or multiple sourcing strategy is particularly very useful to retailers for newly launched products which undergo several changes and updates during their early stage of life cycle.
In the supply chain literature, mainly two forms of supply uncertainty have been considered—supply disruption and yield uncertainty. Supply disruption refers to the complete inoperativeness of a portion of the supply chain whereas yield uncertainty refers to a form of supply uncertainty in which the quantity produced or received differs from the quantity ordered by a random amount. Supply disruption models have been studied extensively both for single supplier systems (Parlar and Berkin, 1991, Moinzadeh and Aggarwal, 1997, Arreola-Risa and DeCroix, 1998) and two-supplier systems (Parlar and Perry, 1996, Gurler and Parlar, 1997, Tomlin, 2006). But the majority of yield uncertainty models have been developed for single supplier systems, see Yano and Lee (1995) for a comprehensive review of yield uncertainty literature. Dual sourcing in the context of yield uncertainty has attracted the attention of only a few researchers. Gerchak and Parlar (1990) investigate a second sourcing option in an EOQ (Economic Order Quantity) setting to reduce the effective yield randomness of firm's purchase quantity and deduce conditions under which double sourcing (with distinct yield distributions to two suppliers) is preferable to single sourcing. Parlar and Wang (1993) compare single and double sourcing alternatives in the newsvendor model assuming that actual incoming quantities are a function of random yield. Agrawal and Nahmias (1997) consider a single period supplier selection and order allocation problem with normally distributed supply and show that for two non-identical suppliers, the expected profit function is concave in the number of suppliers. Anupindi and Akella (1993) address the operational issue of quantity allocation between two uncertain suppliers and its effects on the inventory policies of the buyer. Gurnani et al. (2000) simultaneously determine ordering and production decisions for a two component assembly system with random yield from two suppliers, each providing a distinct component. Chopra et al. (2007) develop a single period model integrating two types of supply uncertainty. One supplier is subject to both recurrent and disruption uncertainties and the other one is perfectly reliable. They show that bundling the two uncertainties results in an over-utilization of the unreliable supplier and under-utilization of the reliable supplier.
The above works focus on characterizing the replenishment decisions which optimize the expected cost or profit. That is, the problems are studied from the point of view of risk-neutral decision makers. However, risks due to market fluctuation, high degree of uncertainties in demand and supply, etc. may have a significant impact on cost. For this reason, inventory managers sometimes accept a reasonably higher expected cost in order to reduce the variability of cost. So, there is a need to incorporate risk into the managerial decision making. Optimal decisions of a risk-averse retailer in the single-item single-period (newsvendor) problem setting have been extensively studied in the literature, see Eeckhoudt et al. (1995), Agrawal and Seshadri (2000), Keren and Pliskin (2006), Chen et al. (2007), and Chopra et al. (2007); multi-item single period setting (Gotoh and Takano, 2007, Borgonovo and Peccati, 2009); single-item multi-period setting (Bouakiz and Sobel, 1992, Ahmed et al., 2007). The above references utilize mean-variance criterion or expected utility theory to develop models under risk. The exceptions being Ahmed et al. (2007), Gotoh and Takano (2007) and Borgonovo and Peccati (2009) which are based on coherent risk measures.
In this paper, we consider a single-period inventory model for a short-life product which is supplied by two suppliers, one is unreliable and cheaper and other one is reliable but more expensive. The underlying problem scenario is very close to Chopra et al. (2007) in which the delivery quantity from the unreliable supplier is assumed to follow a probability distribution having mean and variance independent of the order quantity. The authors develop the model for a risk-neutral retailer. In this paper, we consider an extended newsvendor model assuming that the mean and variance of random yield are dependent on the order quantity. We develop the model and derive the optimal dual sourcing strategy from the point of view of a risk-averse retailer. We investigate, by numerical experiments, how the resulting dual sourcing strategies differ from those obtained in the standard (risk neutral) mean cost analysis.
Section snippets
Notation
The notation used in this paper is as follows: X random variable representing the yield probability density function for normal distribution with mean variance probability density function for standard normal distribution cumulative distribution function for standard normal distribution , the standard normal loss function where z is the standard normal variate d known demand over one period s order quantity from Supplier 1 R reserved quantity from Supplier 2 r
Definition of the problem
Suppose that a retail firm is served by two suppliers—Supplier 1 (primary) and Supplier 2 (secondary). Supplier 1 is cheaper but unreliable i.e. there exists a positive probability that the marginal quantity delivered by Supplier 1 is typically less than the quantity ordered. Supplier 2 is more expensive than Supplier 1 but reliable as (s)he always delivers exactly what is ordered. The responsiveness of Supplier 2 allows to place order after observing the response from Supplier 1. However, the
Utility function and risk measure
Von Neumann and Morgenstern (1944) developed a model that describes how decision makers choose between uncertain prospects. If a decision maker is able to choose consistently between potential random losses L, then there exists a utility function to apprise the wealth W such that the decisions (s)he makes are exactly the same as those resulting from comparing the losses L based on the expectation E[u(W−L)]. Although it is impossible to determine a decision maker's utility function exactly,
Model formulation and analysis
We first show that the random yield from Supplier 1 can be assumed to follow approximately a normal distribution with mean and variance dependent on the order quantity. Let us suppose that an individual unit ordered from Supplier 1 is delivered to the retailer with probability . Therefore, the probability that out of s ordered units, x units are delivered is binomially distributed with mean and variance . We further assume that the probability is neither close to 0 nor close to 1
Numerical illustrations
We take the values of the parameters involved in the model as d=100, cu= 12, co = 8, w1 = 5, w2 = 7, r = 1, in appropriate units. For , the convex behavior of the objective function U(s, R) is observed for wide ranges of values of the parameters. One instance is reflected in Fig. 1 which shows that the surface generated by U(s,R) for k = 0.005 is convex. When the probability increases from 0.1 to 0.9, both the order quantity and reserved quantity decrease drastically, see Fig. 2. The
Discussion and conclusions
This paper has considered a single-product single-period inventory model in which the risk-averse retailer faces yield uncertainty from the primary supplier; the secondary supplier being reliable though capacity constrained. The random yield has been modeled with a probability distribution having mean and variance dependent on the order quantity and the risk aversion has been modeled via an exponential utility function. From the numerical study it has been observed that the risk aversion can
Acknowledgements
The author would like to thank the two anonymous referees and the Guest Editor for helpful suggestions and comments on the earlier version of the manuscript. The author is indebted to Prof. S. Minner, Department of Business Administration, University of Vienna, for many insightful comments and suggestions. The author also gratefully acknowledges the financial support provided by the Alexander von Humboldt Foundation. This work was carried out when the author visited Mannheim University, Germany
References (25)
- et al.
Coherent risk measures in inventory problems
European Journal of Operational Research
(2007) - et al.
Financial management in inventory problems: risk averse vs. risk neutral policies
International Journal of Production Economics
(2009) - et al.
Newsvendor solutions via conditional value-at-risk minimization
European Journal of Operational Research
(2007) - et al.
A benchmark solution for the risk-averse newsvendor problem
European Journal of Operational Research
(2006) - et al.
Diversification under yield randomness in inventory models
European Journal of Operational Research
(1993) - et al.
Conditional value-at-risk for general distributions
Journal of Banking & Finance
(2002) - et al.
Rationalization of the supplier base in the presence of yield uncertainty
Production and Operations Management
(1997) - et al.
Impact of uncertainty and risk aversion on price and order quantity in the newsvendor problem
Manufacturing & Service Operations Management
(2000) - et al.
Diversification under supply uncertainty
Management Science
(1993) - et al.
Inventory management under random supply disruptions and partial backlogging
Naval Research Logistics
(1998)
Risk aversion in maintenance: overmaintenance and the principal-agent problem
IMA Journal of Management Mathematics
Inventory control with an expected utility criterion
Operations Research
Cited by (84)
Newsvendor models with random supply capacity and backup sourcing
2022, European Journal of Operational ResearchCitation Excerpt :However, as we explain in the end of Section 4, their arguments for characterizing the optimal solution are incomplete. To the best of our knowledge, all other relevant models in the literature, that is, models that employ the backup strategy used in this paper, deal mainly with primary suppliers that are subject to random disruptions and/or random yield (see Tomlin, 2009; Saghafian & Van Oyen, 2012; Huang & Xu, 2015; Köle & Bakal, 2017 for random disruptions models, Giri, 2011 for a model with a random yield supplier, and Chopra, Reinhardt, & Mohan, 2007; Guo, Zhao, & Xu, 2016; Giri & Bardhan, 2015; Hou, Zeng, & Sun, 2017 for suppliers that may experience both disruptions and random yield). There also exist models in the literature that are related to ours in the sense that they involve suppliers with random capacity and some form of capacity reservation (Jain & Hazra, 2016; Serel, 2007; 2017).
Two-stage robust optimization for perishable inventory management with order modification
2022, Expert Systems with ApplicationsA service system where junior servers approach a senior server on behalf of customers
2022, International Journal of Production EconomicsTypology and literature review on multiple supplier inventory control models
2021, European Journal of Operational ResearchRisk-averse supplier selection and order allocation in the centralized supply chains under disruption risks
2021, Expert Systems with Applications