Integrating Dynamic Pricing and Inventory Control for Fresh Agriproduct under Multinomial Logit Choice

Wang, Xiong-zhi; Zhou, Wenliang

doi:https://doi.org/10.1155/2018/5936971

Mathematical Problems in Engineering

On this page

Abstract Introduction Literature Review Model Data Availability Disclosure Conflicts of Interest Acknowledgments References Copyright Related Articles

Research Article | Open Access

Volume 2018 | Article ID 5936971 | https://doi.org/10.1155/2018/5936971

Integrating Dynamic Pricing and Inventory Control for Fresh Agriproduct under Multinomial Logit Choice

Xiong-zhi Wang¹and Wenliang Zhou¹

Academic Editor: Mustapha Nourelfath

Received16 May 2018

Revised01 Sept 2018

Accepted29 Nov 2018

Published25 Dec 2018

Abstract

In this article, we investigate a joint pricing and inventory problem for a retailer selling fresh agriproducts (FAPs) with two-period shelf lifetime in a dynamic stochastic setting, where new and old FAPs are on sale simultaneously. At the beginning of each period the retailer makes ordering decision for new FAP and sets regular and discount price for new and old inventories, respectively. After demand realization, the expired leftover is disposed and unexpired inventory is carried to the next period, continuing selling. Unmet demand of all FAPs is backordered. The objective is to maximize the total expected discount profit over the whole planning horizon. We present a price-dependent, stochastic dynamic programming model taking into account zero lead time, linear ordering costs, inventory holding, and backlogging costs, as well as disposal cost. Considering the influence of the perishability, we integrate a Multinomial Logit (MNL) choice model to describe the consumer behavior on purchasing fresh or nonfresh product. By way of the inverse of the price vector, the original formulation can be transferred to be jointly concave and tractable. Finally we characterize the optimal policy and develop effective methods to solve the problem and conduct a simple numerical illustration.

1. Introduction

Motivated by local fresh agriproducts (FAPs) store retailing issues in its operations management, this paper investigates a revenue problem in a stochastic setting where a retailer considers a jointly pricing and inventory control issue under consumer choice. In retailing practice, FAPs, such as fresh fruits or vegetables, are usually delivered to the store or market every day in the morning based on the retailer’s order, submitted in the previous evening, and to be selling at a regular price in the following one or two more days. As the perishability nature FAPs are always deteriorating along time going, gradually being perceived nonfresh and less attractive to customers, which weakens the demand and results in the loss of revenue. To reduce the amount of mismatch between supply and demand, the retailer always sets a discount price for the nonfresh FAPs at a certain time point, which otherwise are deposed if they remain unsold at the end of selling. In the meantime the store shelf is replenished with a new batch of FAPs, ordered by the seller and charged a regular price on selling simultaneously with those unsold nonfresh products. While the regular price and discount price are determined centrally on a long-term basis, some fundamental decisions are at the store manager’s discretion in such an environment, including how much new FAPs to order and what pricing policies to follow for selling. It is not a trivial problem for the seller to sell deteriorated items on discount paralleled with fresh products with regular price. As every coin has two sides, on one hand, the price cutting sales can generate some revenue for the otherwise disposed items since offering a price discount on those items can induce purchasing, while, on the other hand, in codisplay selling simultaneously mode, discount policy results in competition among consumers to purchase either fresh or nonfresh products on the basis of its utility from combination of quality and price. Some of the consumers who can afford the full price may be attracted by the price discount to buy nonfresh products, which has a negative impact on the revenue from selling at the full price.

In some form or other, jointly pricing and inventory control problem arises in a lot of instances. For example, Stater Bros, a well-known grocery store in America, always sells nonfresh vegetables or meat at a discount along with fresh ones and disposes those unsold items at the end of their shelf lifetime. That is, the Stater Brothers store’s manager should make optimal order policy for fresh vegetables or meat and charge for both fresh and nonfresh products differently every day so that the inventory and demand can be extremely matched. In essence, these examples exhibit one homogeneous competition issue which is an important concern and common in retailing practice. Meanwhile, pricing is also a double-edged sword in that pricing can not only influence the demand but also have great impact on the sellers’ profit, especially a poor pricing policy, which can easily damage the seller’s benefit. For example, Tesco, the world’s third-biggest supermarket chain, failed to revive its sales performance despite spending £500 million on price cuts in Eurozone (http://www.ft.com/intl/cms/s/0/dee4d20c-2172-11e1-a1d8-00144feabdc0.html#axzz3Iua2R9jr). On the contrary, Stater Bros. Markets reported its modest profit gains in appropriate price strategy in fiscal 2011 despite higher commodity costs and competitive pressures (http://www.pe.com/articles/million-634066-store-year.html). Therefore, instead of simple pricing of changing prices dramatically, sellers should be more prudent to adjust the price in a dynamic fashion, which enable price change in line with the availability of inventory and products residual shelf-life time. Nevertheless, as consumers may pick any product available and selects his preferred choice based on the utility of an appealing quality-price combination due to the codisplaying mode on sale, managers need to carefully take account of consumer behavior when coordinating discount sales with inventory ordering decisions.

In this study, we formulate a novel model in which the retailer makes simultaneous decisions dynamically on charging appropriate price and optimal inventory policy for perishable products, in the presence of homogeneously selling cannibalization and demand uncertainty. The retail time of any batch of FAPs is divided into two periods, a regular sale period and a discount pricing period, which implies that FAPs have two periods of shelf lifetime. At every period the retailer starts with ordering new FAPs and charging price policy for two different ages of FAPs which compete among customers in their attributes of quality and price and ends with disposing all unsold leftovers with zero shelf lifetime and carrying the leftover inventory of unexpired products to the next period for continuing selling. Generally, new FAPs are perceived to have a higher quality than the old ones. The utility for customers is mainly determined by quality-price combination. Demand in each period is random and depends on the current price, which takes a form of price function plus an additive random perturbation. The seller’s objective is to maximize the total expected discount profit over the whole planning horizon. We formulate a stochastic, price-dependent demand, dynamic programming model taking into account zero lead time, linear ordering costs, inventory holding, and backlogging costs. Our model assumes a population of homogeneous (statistically identical and independent) customers making mutually exclusive choices from two different ages of FAPs. Meanwhile, we integrate a Multinomial Logit Choice model to describe dynamic consumer behavior on choosing new or old products available in inventory and prices adjustment. The retailer, who should consider both the influence of perishable nature of fresh agricultural products and the homogeneous retailing cannibalization, integrates inventory control and dynamic pricing policy to maximize his profit on the whole planning horizon. To the best of our knowledge, this is the first stochastic and dynamic inventory model on perishable fresh agricultural products, taking account of consumer behavior with joint ordering and discount decisions. The problem is extremely complicated even if the products are not perishable. Unlike standard inventory models of general products with same attributes, the perishable property plays an important role in driving the dynamic pricing adjustments which affects the demand share of new and old FAPs. Traditional dynamic inventory models cannot be suitable to our problem in that even the single period profit function is neither jointly convex nor concave in the decision variables, therefore intractable. In addition, the homogeneous competition results in interdependence across the planning horizon, which leads to complicated dynamics in how inventory is controlled over from one period to another. To deal with the complexity, we work with the transformation technique and exploiting specific properties by way of the inverse of the price vector. We also conduct numerical studies to further characterize the optimal policy.

The remainder of this paper is organized as follows. In the next section, we provide a literature review and introduce the model and formulate the problems a dynamic program in Section 3. In Sections 4 and 5, we perform the structural analysis and present an optimal pricing policy coordinated with inventory control strategies, respectively. A numerical study is conducted to show the policy efficiency in Section 6. We conclude our work in Section 7.

2. Literature Review

Our research is mostly related to jointly pricing and inventory control on perishable products, including three streams of literatures: (1) dynamic inventory control for perishable products, (2) combined dynamic pricing and inventory control, and (3) customer behavior in the context of dynamic pricing.

Firstly, our work is closely related to dynamic inventory control for perishables. As far as we know, earliest work dates back to Ghare & Schrader [1] who first considered inventory control of exponential decay product and generalized an EOQ model by assuming the lifetime of each product is exponentially distributed. Other forms of random lifetime models can be referred to Weibull distributed deterioration [2], Gamma distributed decay [3], generalized exponential decay [4], etc. These works mainly focused on static inventory control policies, instead of dynamic strategies. In retailing, the perishable product lifetime was always known deterministically. Fixed lifetime models are also concerned and can be traced to Nahmias and Pierskalls [5] who first explored the problem in a two-period lifetime setting with zero lead time and demand uncertainty. On the basis of their research framework, Nahmias [6] and Fries [7] extended the work on the case with multiperiod lifetime in periodic-review system, where only the excess inventory, expired at the end of the current period, is disposed of. They probed the characteristic of optimal policy and pointed out that the optimal order quantity was a decreasing function of on-hand inventory of different ages. And a series of papers thereafter focused on the similar topic, see Nahmias [8], Cohen [9], Nandakumar and Morton [10], Liu and Lian [11], Lian and Liu [12], Gurler and Ozkaya [13], Berk and Gurler [14], etc. Note that a suitable model might vary even for the same problem, analysis of which was lengthy and difficult to be generalized. Due to the fact that complexity, the literature thenceforward concentrates more on developing heuristics except for model analysis. Nahmias [15, 16] and Karaesmen et al. [17] have reviewed the early and lately developments. Recently, Xue et al. [18], Li et al. 2013, and Li and Yu [19] studied a perishable inventory model in a secondary market setting where the excess inventory can be cleared with certain salvage value. They provided some structure properties and then brought about an effective heuristic. Recently, Chao et al. [20], Zhang et al. [21], and Chao et al. [22] developed approximation algorithms for perishable inventory systems. Our study generalizes this literature to allow for consumer choice on two different ages of FAPs in periodic-review system. And we do focus on exploring optimal policy instead of effective algorithms.

Secondly, in terms of jointly pricing and inventory management, to the best of our knowledge, a mountain of work contributes to model on nonperishable products in recent decades, under the environment of from deterministic periodic-review inventory and pricing models to stochastic models with different cost structures, distinguishing between single period and multiperiod. See, for example, Wagner and Whitin [23], Kunreuther and Schrage [24], Zabel [25], Thomas [26], Gilbert [27], Petruzzi and Dada [28], Federgruen and Heching [29], Feng and Chen [30], Geunes et al. [31], Chen and Simchi-Levi [32, 33], Raz and Porteus [34], Huh and Janakiraman [35], Chen et al. ([36], 2010), Song et al. [37], and Simchi-Levi [38]. Among them, there are some recent surveys on different operations research and management science perspective, which refer to Eliashberg and Steinberg [39], Elmaghraby and Keskinocak [40], Yano and Gilbert [41] and Chan et al. [42], and Chen and Simchi-Levi [43]. This substantial body of literature illustrates that the retailer can benefit great from the structures of optimal policies and effective algorithms of integrated pricing and inventory strategies. Dynamic pricing, supposed to be an effective lever to manage the profitability of perishables retailing, was firstly investigated together with inventory control by Rajan et al. [4] who studied dynamic pricing and ordering decisions for a certain product that experiences a general exponential decaying under deterministic demand, whereafter Abad [44] generalized this work by allowing for partially backlogged demand with the same objective of maximizing the long-run average profit. As Nahmias [15] pointed out, developing pricing policies for perishable items is very important in a demand uncertainty context. There are a few studies pertaining to the case on perishables selling. Among them, Ferguson and Koenigsberg [45] considered a joint pricing and inventory control problem in a two-period setting, addressing the competition impact between new and old products inventory. They presented a stylized model and concentrated on deriving managerial insights. With the assumption of inventory depleted in a first-in-first-out (FIFO) sequence, Li et al. [46] explored a dynamic joint pricing and inventory control for a two-period lifetime perishable product over an infinite horizon, taking into account linear price-response demand, backlogging, and zero lead time. And they also extended the problem to a stationary system with multiperiod lifetime and developed a base-stock/list-price (BSLP) heuristic policy. Then, Li et al. [47] continued analyzing the infinite-horizon lost-sales case where new and old inventory can not be simultaneously sold and sellers can decide whether to dispose of or carry all ending inventory until it expires at the end of each period. They proposed a stationary structural policy consisting of an inventory order-up-to level, state-dependent price, and inventory clearing decisions and developed a fractional programming algorithm to obtain the optimal policy among the class of proposed structural policies. In another relevant work, Sainathan [48] studied the two-period lifetime case where new and old inventory can also be selling simultaneously, and the seller makes decisions on price polices and how much to order at the beginning of each period. Recognizing the influence of perishability on demand, Sainathan modeled the consumer choice and characterized the structure of the optimal policy under different demand context, including deterministic demand, two-point demand, and substitution demand. Extending the analysis and results to the case with arbitrary period lifetime, Chen et al. [49] presented a significant generalization of these papers by allowing for positive lead time, both backlogging and lost-sales cases, and unrestricted ordering decisions. They developed the structural results of L^#-concavity and generalize the regularity conditions of demand functions for lost-sales inventory-pricing models of setting a single price for inventories of different ages. Wu et al. [50] study a problem that the seller dynamically makes the joint pricing and inventory replenishment decisions over multiple periods, where each periods consist of two stages for ordering and markdown pricing, respectively. Chen & Chu [51] and Chen & Cong (2018) integrated a strategic customer’s behavior into joint pricing and inventory model and derived an optimal joint pricing, delivery, and inventory policy. Recently, Peng Hu et al. (2016) formulated a model on a firm’s dynamic inventory and markdown decisions for perishable goods where every period consists of regular sales phase and clearance phase under strategic consumer behavior. They showed that the seller should either put the leftover inventory on discount or dispose all of them, and the choice depends on the amount of leftover inventory from the previous period.

Finally, our work involves modeling of consumer choice behavior which is an important phenomenon within revenue management system. The earliest study in the literature related to this research is pioneered by Kincaid and Darling [52], who described the probability of the charging price, by which the expected demand can be calculated. This general model has been extended by other authors, Bitran and Mondschein [53], who concentrated on applications of the model to the sale of fashion goods. van Ryzin and Mahajan [54] firstly emploied the MNL model to describe the normal demand distribution for substitute products. Aydin and Ryan [55] used the MNL model to study pricing problem assuming the firm satisfies demand by make-to-order. Recent papers related stream of literature include Aydin and Porteus (2005), Zhang and Cooper [56], Xu and Hopp [57], Perakis and Sood [58], Lin and Sibdari [59], Song and Xue [60], and Martínez-de-Albéniz and Talluri [61]. In all those papers, customers can decide which product to buy, based on prices and inventory levels at time of purchase, with the MNL model for product categories, assuming homogeneous product groups. Our work is consistent with those characteristics above, demand uncertainty, two-period shelf lifetime, codisplaying selling, and homogeneous cannibalization. We use MNL model to describe consumer discrete choice and present a jointly dynamic pricing and inventory control policy to maximize seller’s profit on the whole planning horizon.

In summary, our contribution to the literature is as follows. Firstly, we generalize the perishable joint pricing and inventory models for perishable products of two-period lifetime. After characterizing the structure of model, we present an optimal solution to the dynamic pricing of perishable products with consumers incurring dynamic substitutions throughout the selling season. Second, we provide sorts of insights on the behavior of the optimal dynamic prices and highlight the complex interplay of inventory scarcity and product quality difference. Thirdly, we generalize the regularity conditions of demand functions for integration of inventory-pricing models, using the concept of concavity, and characterize the optimal policy and develop effective methods.

3. The Problem Formulation and Model

3.1. Problem Formulation

Consider a periodic-review single type of FAP inventory system over a finite planning horizon of T periods. The product is perishable with two-period shelf lifetime, fresh in the first period and nonfresh in the next period. At the beginning of each period selling, the retailer reviews the current inventories and decides simultaneously on an order quantity for new FAP as well as charging prices for both new delivery and old inventory. During the sale, customers arrive at the store or market and make purchase choices based on FAPs attributes of quality and price, which causes a drop in the inventory level. The demand for the current inventory depends on the newly charged prices, and it is always met to the maximum extent with on-hand inventories. At the end of the period, the leftover inventory with one period lifetime is carried forward to next period, becoming old products to sell while the unsold products with zero shelf-life time have to be discarded. Then, the sale moves to the next period and the above processes repeat.

For convenience and tractability, we assume that the replenishment order, placed by the retailer, arrives immediately before the demand unfolds, that is to say, replenishment is instantaneous with zero lead time. Meanwhile, there could have chance to replenish fresh products during the sale from second market since in retailing practice backorder policy is prominent, which means unsatisfied demands of all products during selling are to be backlogged. And there is neither fixed ordering cost nor constraints on the supply capacity. Demand at each period is uncertain and independent across all periods, generally decreasing on current prices. In line with Petruzzi & Dada [28] and Chen & Simchi-Levi [32] we take a form of a function of prices charged at that period multiplies a stochastic variable plus a random disturbance. Nevertheless, customers are assumed to be homogeneous and decide purchasing based on current prices and product attributes alone, instead of acting strategically by adjusting their buying behavior in response to the firm’s pricing policy. In the whole horizon planning, although the retailer gets revenue from selling new and old products, he bears some costs incurred in the operations, such as order decision for wholesale each order incurs a variable cost c for per new FAP. And three more other costs are incurred: leftover inventory carrying costs over from one period to the next, backlogged costs for unmet demand from on-hand inventory, and disposal costs of leftovers with zero shelf-life time. The objective is to dynamically determine ordering and pricing decisions in all periods so as to maximize the total expected discounted profit over the whole planning horizon.

3.2. Demand Model

During sales, the new and old products compete among customers by combination of their attributes and price. And each customer selects his preferred choice based on his utility from purchasing a unit of product, new or old, which is given by , where i=1 for the fresh product and i=2 for the nonfresh product, respectively. In the formula, and p_i are defined as the average perceptive value and the charged price on product i, respectively, while is a certain customer’s deviation from , which is influenced by unobservable characteristics. The coefficient, , is denoted as a customer’s quality sensitivity. Note that, all consumers are statistically homogeneous and have the same sensitivity to price. Generally, any customer who visits the retailer has three choices: select one unit of new or old FAP, or do nothing. Here we set a virtual product and let i=3 with price zero for denoting no purchase choice. Let be a certain customer’s deviation from . Obviously = 0 and = 0. Therefore, under the assumption of rationality, a customer will always purchase one product with the highest utility among these three options, that is, a customer chooses to buy product i, where . Suppose random variable be i.i.d. distributed with a double exponential distribution, that is, , , with mean zero and variance . See Guadagni and Little [62] and Anderson et al. [63]. Define to be the probability that an arriving customer in a period selects fresh or nonfresh agriproduct in response to price . It can be shown that a consumer’s choice probability has a simple form:which is an MNL model, often used to study a set of products differentiated in either quality or style, a term coined in the literature of MNL models (see [63]). Let be the charged prices at any period . On the basis of analysis above, we denote the market share of product i at a certain period t, which is similar to equation (1) and expressed as follows:Note that the total market share .

Denote to be the amount of arriving customers to buy the products at period t, which is generally uncertain and independently distributed with finite mean of and variance across all time periods. Note that the expected demand of new or old product is proportional to the market share . So, provided with price as well as a demand disturbance on product i at period t, , we can write out the demand function of product i, denoted as below, an additive form which is commonly used in the literature (see, e.g., [28, 32]).where is assumed to be identically distributed over time with zero mean and finite variance as well as c.d.f. and p.d.f. . Denote as the expected demand level, then . The selling price is generally restricted to an interval and .

Note that the monotonicity of the expected demand function implies a one-to-one correspondence between the selling price and the expected demand. And we observe that if we use the inverse of the market share function to express the price variables as a function of the market share, then the revenue function becomes jointly concave under some certain condition, allowing us to analyze the structure of model. Accordingly, we have the following.

Proposition 1. Under the MNL market share model, the price function at period t isThe corresponding market share space can be expressed aswhich is convex and compact.

Proof. Equation (4) comes from (2) by transforming and taking operation on both sides. The range is derived from the fact that the prices should be nonnegative. Consider any product, new or old FAP. If we charge the lowest price, , for one product and the highest price, , for the other, then this product should occupy the highest possible market share, .

3.3. Dynamic Programming Formulation

Based on the description aforementioned, we define some other parameters with respect to period t below. initial inventory level of old FAP before ordering at the beginning of period t. order quantity of new FAP at the beginning of period t. unit procurement cost of new FAP. unit holding cost for the excess inventory. unit disposal cost of remaining inventory with zero shelf-life time. backordered cost for unit shortage of new and old FAP, . profit discount factor, .

Here, we do not factor dynamic demand substitution that the customer may buy the other product type when his preferred one is out of stock. At the beginning of period , the retailer reviews the current inventory of old FAP which comes from the unsold inventory of new products at the previous period and places replenishment order for new FAP, as well as setting product prices, &, for new and old FAPs, respectively. Obviously, under zero lead time assumption, the inventory of new FAPs is instantaneously replenished to be . After demand realization and the retailer satisfies the demand to the maximum extent, the remaining inventory levels of the new and old products are and , respectively. The of leftover that remains one period of shelf-life time will be carried to next period for sale as old FAP, incurring holding cost and becoming the initial inventory of nonfresh product, , at period t+1, while the of old products will be disposed of on account of zero shelf-life time, incurring disposal cost. Otherwise the relevant backordered amount of new FAPs and old FAPs for period t is and , respectively, when demand is greater than on-hand inventories, where and .

At period t, the retailer revenue, , can be calculated as follows:Substitute the price function (5) into (6), we can obtain the expected revenue, .After demand realization at period t, some more costs incurs, backordered cost of or holding cost of for new FAPs, and backlogged cost of or disposal cost of for old FAPs, totally denoted as .And the corresponding expected function is .

Let be the profit of period t; we haveAccordingly, the expected profit function is as below.Denote to be the maximum total expected profit-to-go function from period t to T in state, , with a discount factor . We have the recursive equation . With transfer function, , by defining ,we have the following dynamic programming model:where (13) is the boundary condition since all leftovers will be disposed of at the end of horizon planning. That is to say, there are no second markets where some unexpired inventory can be cleared with some salvage value. On the contrary, the firm needs to pay on deposing expired inventories, even unexpired leftovers. Our analysis can be readily extended to address this situation.

4. Analysis of Model Structure

To analyze the structure of the optimal policy, we need to know the concavity property of which is strictly relative to revenue function and inventory cost function .

Lemma 2. is continuously, twice differentiable, and jointly concave in .

Proof. The continuity and twice differentiability of is quite intuitive since it is expressed by some elementary operations of and . To show the concavity, we need to prove the negative semidefinite property of its Hessian Matrix. Taking the second-order derivative with respect to yields . Thus is concave on . Taking the second-order derivative with respect to yields , which implies the concavity of on . And taking the cross-partial derivative with respect to and yields . Pairing up the above terms leads to Hessian Matrix,And the corresponding determinant is as follows:which implies that matrix is seminegative definite. On the basis of that, is jointly concave in .

Lemma 3. is continuously twice differentiable and jointly convex in and .

Proof. We observe that is continuously and twice differentiable for any realization of and . And the expectation, , is easily justified to be continuously twice differentiable using the canonical argument (Durrett, 2010).
To show the convexity, we define some column vectors, expressed by boldface lowercase letters, such as , , and . Then the demand and cost function can be expressed of vector form as follows:Define a new vector variable , and we can get , which is convex on (Sundaranm, R. K. 1996). Fix the random part to its realization and consider any pair and , with the convexity of ; we havewhere . Therefore, is jointly convex in and , so does the expectation function, .

Theorem 4. At any time period t with initial inventory , the expected revenue function is continuously twice differentiable and jointly concave in and .

Proof. With Lemmas 2 and 3, we can observe that is continuously twice differentiable. We first observe that, for any t, the three terms of in (10) are jointly concave. The first term is concave in according to Lemma 2. The second term is jointly concave in and (see Lemma 3), and the third term is linear in , which implies the concavity. With the property that a nonnegative weighted sum of concave functions is concave (Stephen Boyd, Lieven Vandenberghe, 2004), is continuously twice differentiable and jointly concave in and .

On the basis of the concavity of , we can have Theorem 5.

Theorem 5. Consider any time period t with initial inventory ; the function is continuously twice differentiable and jointly concave in , and the function is concave and nonincreasing in .

Proof. By induction, first, we prove that is continuously twice differentiable. According to Lemma 2, , as a part of , is thus continuously twice differentiable, which implies that is continuously and twice differentiable. Assuming now that is continuously twice differentiable for , we prove that is also continuously twice differentiable according to Theorem 4. Let , where and is the p.d.f. of random variable and , respectively, which are generally assumed to be independent. For any realization of , such as , taking the second-order derivative of with respective to yields , which implies that is convex in . Taking the second-order derivative with respect to yields , which implies the concavity of in . And taking the cross-partial derivative with respect to and yields . Pairing up the above three derivative terms leads to Hessian Matrix,And the corresponding determinant, , which implies matrix is semipositive definite. On the basis of analysis above, we can conclude that is jointly convex in . Meanwhile, the last term, , is obviously justified to be jointly concave in . According to the concavity property (Stephen Boyd, Lieven Vandenberghe, 2004), is jointly concave in and . is easily verified to be concave as well, and it is obviously nonincreasing.
Assume now that is jointly concave for some and that is also concave and nonincreasing. Then, is jointly concave: joint concavity of the first term of (10) is verified for the case above. For any pair of points and with given value of , we haveby the concavity of , which implies that is jointly concave in and . With above argument, the desired result holds.

5. Optimal Policy

Following the structure property of model, we can obtain the following result.

Theorem 6. Consider any period t, with the initial inventory , there is a unique maximizer, , to maximize the constrained expected periodical profit function,

Proof. It can be easily proved on the basis of Theorem 5.

On the basis of the dynamic programming analysis, including the concavity property, we can construct an optimization problem for any period t with constrained conditions as follows:The Lagrangian and the Karush-Kuhn-Tucker(KTT) conditions provide an efficient solution approach to this optimization problem. Denote , , , , , and . We construct for each period t the Lagrangian of the optimization problem as where the Lagrange multipliers are . Let be the optimizing solution; we can construct the KKT conditions as follows:where the first equation represents F.O.C. of Lagrangian and the latter four equations represent the conditions of complementary slackness.

With solution of the KKT conditions, , and (4), we can get the optimization pricing and ordering policy of at period t, where is as follows:Note that depends on state variable, , and parameters of and as well as , which implies that the retailer decides the optimal policy on the basis of these parameters at every period t. With backward induction, we get the optimal strategy, a sequence of .

6. Numerical Experiment

We now conduct a numerical example to illustrate our results and select the parameters as follows. The demand follows the multiplicative form with i.i.d. terms for all periods. We assume , since fresh product has a higher perceptive value than nonfresh ones. The cost structure is symmetric: , , and . For simplicity to compute the optimal prices, we consider a finite horizon with , the discount rate is .9, and the initial inventory level . Using Matlab to program the dynamic model and get the following results with KKT condition method, we can obtain a sequence of order quantity and optimal prices, such as , , , , , at which the maximum profit occurs.

In Table 1, we compare the performance between dynamic pricing policy and a general policy under one period lifetime setting where all products are sold at a regular price at any period of time. Similarly, we assume in dynamic pricing policy while in general policy. Note that in general setting the optimal pricing and ordering policy can be obtained by way of general dynamic programming method. Three examples of different finite horizon time (,10,20) are conducted with the same cost structure and discount rate mentioned above. Accordingly, the optimal solutions, such as pricing and order quantity, as well as dynamic inventory, are shown in Table 1, where superscript g and d represent regular policy and dynamic policy, respectively.

A summary of interesting observations is as follows. The dynamic policy is obviously better than general pricing and inventory policy, with which the retailer’s profit increases by 15-20%. Further, the demand is also matched to a more certain extent by the supply since the order quantity is almost more than that under general policy while the. Based on the results, we can conclude that dynamic pricing and inventory policy is beneficial for retailer’s revenue management in selling fresh agriculture products. Furthermore, from the perspective of agriculture industrial development, it also can maximally reduce the operating costs, such as holding costs or lost costs, since the optimal match between products supply and demand.

7. Concluding Remark

In this paper, we address a joint pricing and inventory control problem for stochastic fresh agriproduct inventory systems under consumer choice over a finite horizon in the backlogging case. In the model the products are perishable along time and the customers are sensitive to the ages of inventories, which result in consumer choice behavior based on product freshness. By using MNL model to describe consumer discrete choice we construct a stochastic and dynamic inventory model facing strategic customer behavior with joint ordering and discounting decisions. We are able to characterize the firm’s optimal policy with the use of a transformation technique and employing the concept of the concavity, and finally develop a Lagrangian and the Karush-Kuhn-Tucker (KTT) conditions to provide an efficient solution approach to this optimization problem.

A limitation in our model is that we assume that the lifetime of the product is only two periods. Although this is true for some fresh agriproducts in the market, there are other products that are perishable but yet have longer lifetimes. To extend our model to the products with more than two periods lifetimes, the state space has to be increased more to specify the amounts of inventory that are going to expire at different points of time in the future. This definitely makes the problem quite hard. Another issue is to incorporate an operating cost for discounting pricing practice, which, however, will introduce additional technical complexity to the dynamic program. There are several future research directions. One is to extend the model to the environment where all the demand can be lost or only non-FAP demand can be lost while the fresh FAP can be backlogged instead. Another is to consider the case where the product has a fixed finite lifetime of exactly periods or random lifetime. Finally, it would be interesting to account for stock-out based substitutions.

Data Availability

The data used to support the findings of this study are included within the article.

Disclosure

This paper was previously presented as an abstract in 2016 International Symposium for Facing New Conditions and Challenges: the Agricultural Development in USA and China.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Acknowledgments

This work is supported by fund of Guandong Province Humanities and Social Science Base for Colleges and Universities Foundation Grant 11JDXM63004 “Collaborative Proposal: Agri-Business: Supply Chain Contract and Coordination Mechanism”. It is also funded by National Natural Science Foundation of China (71501051) and Philosophy and Social Science Foundation of Guangdong Province (GD15XGL42). I should like to acknowledge the helpful comments on an earlier draft.

References

P. Ghare M and G. F. Schrader, “An inventory model for deteriorating item for exponentially deteriorating items,” Journal of Industrial Engineering, vol. 14, pp. 238–243, 1963.
View at: Google Scholar
R. P. Covert and G. C. Philip, “An EOQ model for items with Weibull distribution deterioration,” IIE Transactions, vol. 5, no. 4, pp. 323–326, 1973.
View at: Publisher Site | Google Scholar
P. R. Tadikamalla, “An eoq inventory model for items with gamma distributed deterioration,” IIE Transactions, vol. 10, no. 1, pp. 100–103, 1978.
View at: Publisher Site | Google Scholar
A. Rajan, Rakesh, and R. Steinberg, “Dynamic Pricing and Ordering Decisions by a Monopolist,” Management Science, vol. 38, no. 2, pp. 240–262, 1992.
View at: Publisher Site | Google Scholar
S. Nahmias and W. P. Pierskalla, “Optimal ordering policies for a product that perishes in two periods subject to stochastic demand,” Naval Research Logistics Quarterly, vol. 20, pp. 207–229, 1973.
View at: Publisher Site | Google Scholar | MathSciNet
S. Nahmias, “Optimal Ordering Policies for Perishable Inventory—II,” Operations Research, vol. 23, no. 4, pp. 735–749, 1975.
View at: Publisher Site | Google Scholar
B. E. Fries, “Optimal ordering policy for a perishable commodity with fixed lifetime,” Operations Research, vol. 23, no. 1, pp. 46–61, 1975.
View at: Publisher Site | Google Scholar | MathSciNet
S. Nahmias, “Myopic approximations for the perishable inventory problem,” Management Science, vol. 22, no. 9, pp. 1002–1008, 1975.
View at: Publisher Site | Google Scholar | MathSciNet
M. A. Cohen, “Analysis of single critical number ordering policies for perishable inventories,” Operations Research, vol. 24, no. 4, pp. 726–741, 1976.
View at: Publisher Site | Google Scholar
P. Nadakumar and T. E. Morton, “Near myopic heuristics for the fixed-life perishability problem,” Management Science, vol. 39, no. 12, pp. 1490–1498, 1993.
View at: Publisher Site | Google Scholar
L. Liu and Z. Lian, “(s, S) continuous review models for products with fixed lifetimes,” Operations Research, vol. 47, no. 1, pp. 150–158, 1999.
View at: Publisher Site | Google Scholar
Z. Lian and L. Liu, “Continuous review perishable inventory systems: Models and heuristics,” Institute of Industrial Engineers (IIE). IIE Transactions, vol. 33, no. 9, pp. 809–822, 2001.
View at: Google Scholar
Ü. Gürler and Yüksel Ö. B., “A note on “Continuous review perishable inventory systems: models and heuristics”,” IIE Transactions, vol. 35, no. 3, pp. 321–323, 2003.
View at: Publisher Site | Google Scholar
E. Berk and Ü. Gürler, “Analysis of the (Q, r) inventory model for perishables with positive lead times and lost sales,” Operations Research, vol. 56, no. 5, pp. 1238–1246, 2008.
View at: Publisher Site | Google Scholar | MathSciNet
S. Nahmias, “Perishable inventory theory: a review,” Operations Research, vol. 30, no. 4, pp. 680–708, 1982.
View at: Publisher Site | Google Scholar
S. Nahmias, Perishable inventory systems (International series in operations research and management science), 2011.
I. Z. Karaesmen, A. Scheller-Wolf, and B. Deniz, “Managing perishable and aging inventories: Review and future research directions,” International Series in Operations Research and Management Science, vol. 151, pp. 393–436, 2011.
View at: Publisher Site | Google Scholar
Z. Xue, M. Ettl, and D. Yao, Managing freshness inventory: Optimal policy, bounds and heuristics, Working paper, IEOR Department, Columbia University, NY, 2012.
Q. Li and P. Yu, “Multimodularity and Its Applications in Three Stochastic Dynamic Inventory Problems,” Manufacturing & Service Operations Management, vol. 16, no. 3, pp. 455–463, 2014.
View at: Publisher Site | Google Scholar
X. Chao, X. Gong, C. Shi, and H. Zhang, “Approximation algorithms for perishable inventory systems,” Operations Research, vol. 63, no. 3, pp. 585–601, 2015.
View at: Publisher Site | Google Scholar | MathSciNet
H. Zhang, C. Shi, and X. Chao, “Technical note---approximation algorithms for perishable inventory systems with setup costs,” Operations Research, vol. 64, no. 2, pp. 432–440, 2016.
View at: Publisher Site | Google Scholar | MathSciNet
X. Chao, X. Gong, C. Shi, C. Yang, H. Zhang, and S. X. Zhou, “Approximation Algorithms for Capacitated Perishable Inventory Systems with Positive Lead Times,” Management Science, vol. 64, no. 11, pp. 5038–5061, 2018.
View at: Publisher Site | Google Scholar
H. M. Wagner and T. M. Whitin, “Dynamic version of the economic lot size model,” Management Science, vol. 5, no. 1, pp. 89–96, 1958.
View at: Publisher Site | Google Scholar | MathSciNet
H. Kunreuther and L. Schrage, “Joint Pricing and Inventory Decisions for Constant Priced Items,” Management Science, vol. 19, no. 7, pp. 732–738, 1973.
View at: Publisher Site | Google Scholar
E. Zabel, “Multiperiod monopoly under uncertainty,” Journal of Economic Theory, vol. 5, no. 3, pp. 524–536, 1972.
View at: Publisher Site | Google Scholar | MathSciNet
J. Thomas, “Price and Production Decisions with Random Demand,” Operations Research, vol. 22, no. 3, pp. 513–518, 1974.
View at: Google Scholar
S. M. Gilbert, “Coordination of pricing and multi-period production for constant priced goods,” European Journal of Operational Research, vol. 114, no. 2, pp. 330–337, 1999.
View at: Publisher Site | Google Scholar
N. C. Petruzzi and M. Dada, “Pricing and the newsvendor problem: a review with extensions,” Operations Research, vol. 47, no. 2, pp. 183–194, 1999.
View at: Publisher Site | Google Scholar
A. Federgruen and A. Heching, “Combined pricing and inventory control under uncertainty,” Operations Research, vol. 47, no. 3, pp. 454–475, 1999.
View at: Publisher Site | Google Scholar | MathSciNet
Y. Feng and F. Y. Chen, “Joint pricing and inventory control with setup costs and demand uncertainty,” The Chinese University of Hong Kong, 2003.
View at: Google Scholar
J. Geunes, H. E. Romeijn, and K. Taaffe, “Requirements planning with pricing and order selection flexibility,” Operations Research, vol. 54, no. 2, pp. 394–401, 2006.
View at: Publisher Site | Google Scholar
X. Chen and D. Simchi-Levi, “Coordinating inventory control and pricing strategies with random demand and fixed ordering cost: the finite horizon case,” Operations Research, vol. 52, no. 6, pp. 887–896, 2004.
View at: Publisher Site | Google Scholar | MathSciNet
X. Chen and D. Simchi-Levi, “Coordinating inventory control and pricing strategies: the continuous review model,” Operations Research Letters, vol. 34, no. 3, pp. 323–332, 2006.
View at: Publisher Site | Google Scholar | MathSciNet
G. Raz and E. L. Porteus, “A fractiles perspective to the joint price/quantity newsvendor model,” Management Science, vol. 52, no. 11, pp. 1764–1777, 2006.
View at: Publisher Site | Google Scholar
W. T. Huh and G. Janakiraman, “(s,S) optimality in joint inventory-pricing control: an alternate approach,” Operations Research, vol. 56, no. 3, pp. 783–790, 2008.
View at: Publisher Site | Google Scholar | MathSciNet
X. Chen, “Inventory centralization games with price-dependent demand and quantity discount,” Operations Research, vol. 57, no. 6, pp. 1394–1406, 2009.
View at: Publisher Site | Google Scholar | MathSciNet
Y. Song, S. Ray, and T. Boyaci, “Optimal dynamic joint inventory-pricing control for multiplicative demand with fixed order costs and lost sales,” Operations Research, vol. 57, no. 1, pp. 245–250, 2009.
View at: Publisher Site | Google Scholar
D. Simchi-Levi, X. Chen, and J. Bramel, The Logic of Logistics: Theory, Algorithms, and Applications for Logistics Management, Springer, New York, 3rd edition, 2014.
J. Eliashberg and R. Steinberg, “Competitive Strategies for Two Firms with Asymmetric Production Cost Structures,” Management Science, vol. 37, no. 11, pp. 1452–1473, 1991.
View at: Publisher Site | Google Scholar
W. Elmaghraby and P. Keskinocak, “Dynamic pricing in the presence of inventory considerations: research overview, current practices, and future directions,” Management Science, vol. 49, no. 10, pp. 1287–1309, 2003.
View at: Publisher Site | Google Scholar
A. Yano and S. M. Gilbert, “Coordinated Pricing and Production/Procurement Decisions: A Review,” in Managing Business Interfaces: Marketing, Engineering, and Manufacturing Perspectives, J. Eliashberg and A. Chakravarty, Eds., Kluwer, Norwell, MA, 2003.
View at: Google Scholar
L. M. Chan, Z. J. Shen, D. Simchi-Levi, and J. L. Swann, “Coordination of Pricing and Inventory Decisions: A Survey and Classification,” in Handbook of Quantitative Supply Chain Analysis, vol. 74 of International Series in Operations Research & Management Science, pp. 335–392, Springer US, Boston, MA, 2004.
View at: Publisher Site | Google Scholar
X. Chen and D. Simchi-Levi, “Pricing and Inventory Management,” The Oxford Handbook of Pricing Management, pp. 784–822, 2012.
View at: Google Scholar
P. L. Abad, “Optimal pricing and lot-sizing under conditions of perishability and partial backordering,” Management Science, vol. 42, no. 8, pp. 1093–1104, 1996.
View at: Publisher Site | Google Scholar
M. E. Ferguson and O. Koenigsberg, “How should a firm manage deteriorating inventory?” Production Engineering Research and Development, vol. 16, no. 3, pp. 306–321, 2007.
View at: Google Scholar
Y. Li, A. Lim, and B. Rodrigues, “Pricing and inventory control for a perishable product,” Manufacturing and Service Operations Management, vol. 11, no. 3, pp. 538–542, 2009.
View at: Publisher Site | Google Scholar
Y. Li, B. Cheang, and A. Lim, “Grocery perishables management,” Production Engineering Research and Development, vol. 21, no. 3, pp. 504–517, 2012.
View at: Publisher Site | Google Scholar
A. Sainathan, “Pricing and replenishment of competing perishable product variants under dynamic demand substitution,” Production Engineering Research and Development, vol. 22, no. 5, pp. 1157–1181, 2013.
View at: Publisher Site | Google Scholar
X. Chen, Z. Pang, and L. Pan, “Coordinating inventory control and pricing strategies for perishable products,” Operations Research, vol. 62, no. 2, pp. 284–300, 2014.
View at: Publisher Site | Google Scholar | MathSciNet
S. Wu, Q. Liu, and R. Q. Zhang, “The reference effects on a retailer's dynamic pricing and inventory strategies with strategic consumers,” Operations Research, vol. 63, no. 6, pp. 1320–1335, 2015.
View at: Publisher Site | Google Scholar | MathSciNet
Y. Chen and L. Y. Chu, Synchronizing pricing and replenishment to serve forward-looking customers with lost sales. Working Paper, University of Southern California, Los Angeles, CA, 2016.
View at: Publisher Site
W. M. Kincaid and D. A. Darling, “An inventory pricing problem,” Journal of Mathematical Analysis and Applications, vol. 7, pp. 183–208, 1963.
View at: Publisher Site | Google Scholar | MathSciNet
G. R. Bitran and S. V. Mondschein, “Periodic pricing of seasonal products in retailing,” Management Science, vol. 43, no. 1, pp. 64–79, 1997.
View at: Publisher Site | Google Scholar
G. Van Ryzin and S. Mahajan, “On the relationship between inventory costs and variety benefits in retail assortments,” Management Science, vol. 45, no. 11, pp. 1496–1509, 1999.
View at: Publisher Site | Google Scholar
A. Goker and K. Jennifer, “Product line selection and pricing under the multinomial logit choice model,” in Proceedings of the 2000 MSOM conference, 2000.
View at: Google Scholar
D. Zhang and W. L. Cooper, “Revenue management for parallel flights with customer-choice behavior,” Operations Research, vol. 53, no. 3, pp. 415–431, 2005.
View at: Publisher Site | Google Scholar | MathSciNet
X. Xu and W. J. Hopp, “A monopolistic and oligopolistic stochastic flow revenue management model,” Operations Research, vol. 54, no. 6, pp. 1098–1109, 2006.
View at: Publisher Site | Google Scholar
G. Perakis and A. Sood, “Competitive multi-period pricing for perishable products: a robust optimization approach,” Mathematical Programming, vol. 107, no. 1-2, Ser. B, pp. 295–335, 2006.
View at: Publisher Site | Google Scholar | MathSciNet
K. Y. Lin and S. Y. Sibdari, “Dynamic price competition with discrete customer choices,” European Journal of Operational Research, vol. 197, no. 3, pp. 969–980, 2009.
View at: Publisher Site | Google Scholar | MathSciNet
S. Jing-Sheng and X. Zhengliang, Demand management and inventory control for substitutable products, Working paper, 2007.
V. Martínez-De-Albéniz and K. Talluri, “Dynamic price competition with fixed capacities,” Management Science, vol. 57, no. 6, pp. 1078–1093, 2011.
View at: Publisher Site | Google Scholar
P. M. Guadagni and J. D. Little, “A Logit Model of Brand Choice Calibrated on Scanner Data,” Marketing Science, vol. 2, no. 3, pp. 203–238, 1983.
View at: Publisher Site | Google Scholar
S. P. Anderson, A. de Palma, and J.-F. Thisse, Discrete Choice Theory of Product Differentiation, MIT Press, Cambridge, Mass, USA, 1992.
View at: MathSciNet

Copyright

Copyright © 2018 Xiong-zhi Wang and Wenliang Zhou. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

PDF Download Citation

Download other formats

Order printed copies

Views

1246

Downloads

922

Citations