Study On Supply And Demand Management In MTS And MTO Firms From The Perspective Of Revenue Management

Revenue management is a set of strategies and techniques firms use to maximizetheir revenue by setting optimal prices and optimal capacity control for limited capacitybased on market segment and demand forecast. In order to match the limited supply andthe varying demand, and maximize revenue, this paper applies the strategies andtechniques of dynamic pricing and capacity control from revenue management intomake-to-stock (MTS) and make-to-order (MTO) firms. Production and pricingintegrating policies are used to control the supply and demand in MTS firms, orderacceptance policy is employed to manage the supply in MTO firms, and pricing anddelivery date quotation are adopted to control the demand in MTO firms.For the supply and demand management in MTS firms, we first study the pricingand production policies based on linear demand. Assume demand is a linear function ofprices, and demand only depends on the current prices. The function of demand isknown, but with unknown demand parameters. A demand learning approach based onleast square method is applied to learn the unknown demand parameters. We considerthe optimal pricing and production problem without and with competition:(1) Singlefirm. We model the firm's optimal control problem without considering competition,and obtain the optimal pricing and production policy. We also analyze the sensitivitiesof the maximum profit on the parameters;(2) Oligopolistic competition. There areseveral firms competing with each other. We show that all firms' optimal controlproblem yields a generalized differential Nash equilibrium problem, which isrepresented as a differential variational inequality. We prove the equivalence betweenthe two and a generalized-differential-Nash equilibrium exists. Through the numericalexamples, two dynamic pricing policies are compared, and we obtain that the policybased on demand learning method can generate higher profit than the one based onfixed demand parameters. We also discuss the impact of competitor's price sensitivityon the optimal policies and profit;(3) Stackelberg competition. There are one leader andmultiple followers. We prove that a Stackelberg-generalized-differential-Nashequilibrium exists. The numerical examples show the optimal prices and production policies of the leader and the followers. And we compare the profit of each firm underoligopolistic and Stackelberg competition, and find out that a firm can yield higherprofit to be the leader in the Stackelberg game, however achieve lower profit to be thefollower in the Stackelberg game, than in oligopolistic game.Next, we consider the pricing and production problem based on demanddynamics. Instead of assuming demand is a linear function of prices, we considerdemand dynamics. Demand depends not only on the current prices, but also on theprices of the past periods. We express the demand dynamics based on evolutionarygame theory with known function and unknown demand parameters. A demand learningapproach based on Markov chain Monte Carlo method is used to predict the unknowndemand parameters. We also consider the problem under the three cases: single firm,oligopolistic and Stackelberg competition. The optimal pricing and production policiesare obtained under each case. For the single firm, we analyze the sensitivities of themaximum profit on the parameters. For the oligopolistic game, through the numericalexample, we compared two different demand learning methods, the one based on leastsquare method and the one based on Markov chain Monte Carlo method. The resultshows that the latter demand learning method can generate higher profit. For theStackelberg game, we show the impact of the leader's price sensitivity on the optimalpolices and the profits of the leader and the follower.For the supply and demand management in MTO firms, we first study the supplymanagement. We consider the order acceptance policy under the single resource case.Expected Marginal Seat Revenue-a and Expected Marginal Seat Revenue-b (EMSR-aand EMSR-b) are used to calculate the two different booking limits for different classes.Then according to the booking limit, we decide to accept or reject an order. Theexample results show that the order acceptance policies based on EMSR-a and EMSR-bperform better than the First-Come-First-Served (FCFS) policy. Then we address theorder acceptance and allocating problem under multiple resources case considering theperishability of the machine resource. The problem is modeled by dynamicprogramming approach, and solved by reverse recursive method. Through an example,we anaylze the effect of the parameters on the maximum expected profit and find outour optimal policy outperforms the FCFS policy.Then we study the demand management. First, we consider the pricing policy when a customer never cancel order after he places it. The probability that the customerwill place an order depends on price. We obtain the optimal price and the maximumexpected profit, and analyze the sensitivities of the two on all parameters. Then weaddress the pricing problem when the customer may cancel the order after he places it.We also obtain the optimal price and the maximum expected profit, and analyze thesensitivities of the two on all parameters, and discuss the impact of ignoring ordercancellation on the optimal profit, when customers indeed cancel orders with aprobability. At last we consider the pricing and other factors integrating policies:(1)Pricing and delivery date quotation problem. The probability that customer will placeorder is a function of price and the quoted delivery date. With the objective ofmaximizing a single order's profit, we obtain the optimal prices and the quoted deliverydate.(2) Pricing and production scheduling problem. In this case, customer determinesthe delivery date, and the probability that he will place order depends on price.According to the customer's expected delivery date, we obtain the optimal pricing andproduction scheduling policy.(3) Pricing and order acceptance problem. Consider aMTO firm with contractual customers and fill-in customers. We use the dynamicprogramming to model the order acceptance problem for fill-in customers, and gain theoptimal order acceptance policy. And the optimal pricing policy for contractualcustomers is obtained by using genetic algorithm. The result shows that the optimalorder acceptance policy is better than the FCFS order acceptance policy.