research article the complex dynamics of bertrand...
TRANSCRIPT
-
Research ArticleThe Complex Dynamics of Bertrand-Stackelberg PricingModels in a Risk-Averse Supply Chain
Junhai Ma and Qiuxiang Li
College of Management and Economics, Tianjin University, Tianjin 300072, China
Correspondence should be addressed to Qiuxiang Li; [email protected]
Received 22 September 2013; Accepted 17 February 2014; Published 18 May 2014
Academic Editor: Mingshu Peng
Copyright Β© 2014 J. Ma and Q. Li. 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.
We construct dynamic Bertrand-Stackelberg pricing models including two manufacturers and a common retailer in a risk-aversesupply chain with the uncertain demand. The risk-averse supply chain follows these strategies: Bertrand game between the twomanufacturers and Stackelberg game between the manufacturer and the retailer. We study the effect of the price adjustment speed,the risk preference, and the uncertain demand on the stability of the risk-averse supply chain using bifurcation, power spectrum,attractor, and so forth. It is observed that there exists slip bifurcation when the price adjustment speed across some critical value,the stable region, and total profit of the risk-averse supply chain will increase with increase of π
π1and decrease with increase of π.
The profit of the supply chain and the two manufacturers will decrease and the weaker (retailer) is a beneficiary when the supplychain is in chaos. The fluctuation in the supply chain can be gradually controlled by the control of the price adjustment speed.
1. Introduction
With the development of economic globalization, the rela-tionship among the supply chain members becomes moreand more complex under the different environment. Anenterprise which is involved in the middle of multiple supplychains has all kinds of complicated relationship when theparameter of the upstream and downstream enterprises ischanged, such as market uncertainty and risk preferencefactor; the enterpriseβs decision-making behaviors becomemore complicated and hard to predict.
We all know that the price is always a sensitive topic,which can affect customer needs and wants, distribution ofthe products and services among the supply chains. Scholarsat home and abroad have done a lot of research on thisaspect. Wei et al. [1] studied pricing decisions in a supplychain with two manufacturers and one common retailerand constructed five pricing models under decentralizeddecision cases with consideration of different market powerstructures. Mukhopadhyay et al. [2] considered two separatefirms, which had private forecast information about marketuncertainties and offered complement goods in a leader-follower type, and devised a βsimple to implementβ informa-tion sharing scheme under which both firms and the total
system are better off.These literatures analyzed and comparedthe optimal solution under differentmarket power structures,but they did not consider the effect of risk reference of theparticipants on the optimal solution.
There are many literatures taking risk preference intoaccount. Caliskan-Demirag et al. [3] constructed modelsof the supply chain with a risk-averse retailer by adoptingthe conditional-value-at-risk (CVaR) decision criterion. Luoand Huang [4] explored the impact of risk preference onthe strategies of the supply chain by taking the differentattention of high profit and low profit as retailers riskmeasurement. These literatures only consider the unilateralrisk, but, in the uncertainty environment, there exists bilateralrisk among the participants which corresponds to the actualsituation. In this paper, we will consider a supply chainunder bilateral risk with two risk-averse manufacturers anda risk-averse retailer, which make the supply chain morecomplex.
Many literatures study the complexity of supply chain.Huang and Chen [5] studied the sale-surety contract optionof supply chain with effort dependent demand and riskpreference and got some meaningful conclusions. Guan andZhou [6] researched the integrated optimization problem
Hindawi Publishing CorporationDiscrete Dynamics in Nature and SocietyVolume 2014, Article ID 749769, 14 pageshttp://dx.doi.org/10.1155/2014/749769
-
2 Discrete Dynamics in Nature and Society
of three-level supply chain consisting of suppliers, distrib-utors, and retailers under decision-makers having differentrisk attitude. Huang and Yang [7] studied a two-echelonsupply chain model with one supplier and one retailerin a newsvendor problem; the supplier with different riskattitude has great influence on the retailerβs optimal orderquantity; the operation efficiency of the supply chain will beunderperformed when the supplier is much too risk-averse.These literatures have studied the participantβs behaviors ofthe risk supply chain, but they did not present the dynamiccomplex features of the risk supply chain.
Research on dynamical complexity of a system has beenof concern to scholars. Puu [8] found that the Cournot threeoligopoly model could appear strange attractors with fractaldimension, and he studied the situation of the duopoly game.Many researchers improved the classical Cournot modeland found that certain dynamical behaviors of the systemoccurred in the course of repeated games with three orfour duopolies. Many experts have also studied this fieldunder different conditions, such as different expectations andincomplete information select it Ma and Sun [9] establisheda decentralized pricing game model and studied its complexdynamic characteristics of triopoly under different decision-making rule; the result showed that the process of gamewould tend to a Nash equilibrium at a lower price adjustmentspeed, and, with the increase of the value of adjustmentspeed, the system would appear to be unstable and graduallygone into a chaos state. Ma and Bangura [10] studied thedynamic complexity of financial and economic system underthe condition of three parameters changing.
In recent years, many experts apply the dynamical com-plexity to study the dynamic change process of supply chain.Hwarng and Xie [11] found that there existed the chaoticenlargement phenomenon among the members of the supplychain which enriched the connotation of the bullwhip effect.J. Wang and X. Wang [12] established nonlinear supplychain inventory systemmodels with forbidden returning andlimited supply capacity; numerable simulations showed thatthe supply chain inventory system had complex dynamicbehaviors under certain parameter settings; they gave somesuggestions to eliminate the complexity of the dynamicsupply chain. Ma and Feng [13] presented investigationsimulations of retailerβs demand and stock; the behaviorsof the system exhibited deterministic chaos with consid-eration of system constraints. These literatures researchedthe dynamic complexity of the supply chain but did notconsider the influence of the decision-makerβs risk behav-iors on the supply chain management. In this paper, wewill study the dynamic complexity of a risk-averse supplychain with two manufacturers and a common retailer underuncertain demand. Considering the change of parameters inthe dynamic risk supply chain, such as the price adjustmentspeed, risk preference, and uncertain demand, we can studythe influence of parameters on the price and stable region ofthe two manufacturers and retailer.
The remainder of this paper is organized as follows.In Section 2, we describe the supply chain problem, makeassumptions of the system model, and discuss the systemmodel. In Section 3, we construct a Bertrand-Stackelberg
dynamic pricing model which consists of two manufacturesand one retailer with risk-averse attitude. Analysis is madeunder different variable conditions in Section 4. In Section 5,the variable feedback control method will be used to con-trol chaos in the system. In Section 6, we outline someconclusions and hence relevant recommendations for futureresearch.
2. Model
2.1. Description of the Problem. In this section, we constructa dynamic pricing game model in a risk-averse supply chainwhich consists of twomanufacturers (π
1andπ
2) and a com-
mon retailer (π ). The two manufacturers are competitive andsell respective products to the common retailer, and the com-mon retailer sells two kinds of products to consumers directly.The customersβ demand is stochastic. We consider the supplychain following these strategies: Bertrand game betweenthe two manufacturers and Stackelberg game between themanufacturer and the retailer. In these strategies, the twomanufacturers and the retailer make their own decisions,respectively, for maximizing their profit; the decision processis as follows: the twomanufacturers, as the Stackelberg leader,determine the respective wholesale price (π€
π) (π = 1, 2);
the retailer as the follower sets his own optimal retail price(ππ) (π = 1, 2) based on the manufacturerβs decisions.Furthermore, in order to capture the uncertain demand
which is affected by the change of economic and businessconditions and prediction errors, we assume the marketdemand random variable π is as follows: π = π + π,where π is the primary demand level and π follows a normaldistribution such as πΈ(π) = 0, Var(π) = π2. However,the normality assumption has been used extensively in theliterature (e.g., Gal-Or [14]; Raju and Roy [15]; Vives [16]).The twomanufacturers and the retailer know the distributionof the uncertain demand and determine their behaviors,respectively.
Because the customer demand is stochastic, there isfinancial risk to the two manufacturers and the retailer.Therefore, we should consider the effect of the risk preferenceof the twomanufacturers and the retailer on pricing decision.The preference theory provides the framework which incor-porates the participatorsβ financial risk preference into theirdecision process. The valuation measure we use is knownas the certainty equivalent in the preference theory and isdefined as certain value that a participator is just willing toaccept an uncertain event (Kunstman [17]).
One form of the utility function in both theoreticaland applied work in areas of decision theory and financeis the exponential utility function which can be expressedas (ππ) = βπ
βππ/π π (π = π
1,π2, π ), where π
πis the risk
tolerance level of the two manufacturers and retailer, ππis
the profit, and π is the exponential constant. When π π< β,
it implies that the decision-maker has risk-averse behavior,and π
πapproaches β which implies the decision-maker
is risk-neutral (Walls [18]). If the decision-maker is risk-averse,π follows a normal distribution, and expected utility is
-
Discrete Dynamics in Nature and Society 3
πΈ(π) = πΈ(π) β (Var(π)/2π ), where πΈ(π) is the mean of π andVar(π) is the variance of π.
2.2. Assumption of the System
(1) Customer demand is always satisfied, demand func-tion is linear, and the two manufacturers and theretailer make decentralized decision.
(2) We consider twopartly substitutable products comingfrom a competitive market in which consumers canbuy any one of them.
(3) The consumer demand is stochastic; the two manu-facturers and the retailer are all risk aversion.
(4) πΆπ1
is marginal cost of π1; πΆπ2
is marginal cost ofπ2.
2.3. Revenue Function of the System. In this study, π€πand
ππ(π = 1, 2) are decision variables and other variables are
exogenous variables. As is known in the supply chain, weassume that π
πβ₯ π€π(π = 1, 2); this inequality ensures that
each participant can obtain a positive profit.Extend the demand function in Banker et al. [19]. We
assume that the primary demand function in this paper isdecided by π
π(π = 1, 2) as follows:
π·1= π β π
1π1+ π1π2,
π·2= π β π
2π2+ π2π1,
(1)
where π represent the base demand level of the product,ππ(π = 1, 2) are price sensitive coefficient of demand,
the cross-price sensitive coefficient ππ(π = 1, 2) reflects
the substitution degree of the products, and π, ππ, ππ> 0.
We can obtain the expected utility functions of the twomanufacturers and the retailer as follows:
πΈ (ππ ) = πΈ (π
π ) β
Var (ππ )
2π π
= (π1β π€1) (π β π
1π1+ π1π2) + (π
2β π€2)
Γ (π β π2π2+ π2π1) β
(π1+ π2β π€1β π€2)2π2
2π π
,
(2)
πΈ (ππ1
) = πΈ (ππ1
) β
Var (ππ1
)
2π π1
= (π€1β ππ1
) (π β π1π1+ π1π2) β
(π€1β ππ1
)2
π2
2π π1
,
(3)
πΈ (ππ2
) = πΈ (ππ2
) β
Var (ππ2
)
2π π2
= (π€2β ππ2
) (π β π2π2+ π2π1)
β
(π€2β ππ2
)2
π2
2π π2
.
(4)
From formulas (2), (3), and (4), we obtain the revenuefunctions of the two manufacturers and the retailer whichis more in accordance with the actual situation using theutility function. When a manufacturer changes the value ofparameter, how to adjust value of parameter and what impactit will have on the othermanufacturers and the retailer are themain innovation points of this paper.
3. Bertrand-Stackelberg Model
Suppose that the two manufacturers and the retailer haveprincipal and subordinate relationship, the two manufactur-ers are Stackelberg leaders, the retailer is follower, and thereis Bertrand competition between the two manufacturers.Then, the manufacturers and the retailer process sequentialdynamic game; the game equilibrium is Stackelberg equilib-rium. In this game, the two manufacturers make decisionsfor wholesale price according to the market information; theretailer makes decisions according to the two manufacturers.Using backward induction, we first find the response func-tions of the second stage from the game model. The optimalmarginal utility of the retailer can be obtained by the first-order conditions of formula (2); the calculation results are asfollows:
ππΈ (ππ )
ππ1
= π + π1π€1β π2π€2+(π€1+ π€2) π2
π π
β (π2
π π
+ 2π1)π1+ (π1+ π2βπ2
π π
)π2,
ππΈ (ππ )
ππ2
= π + π2π€2β π1π€1+(π€1+ π€2) π2
π π
β (π2
π π
+ 2π2)π2+ (π1+ π2βπ2
π π
)π1.
(5)
The retailerβs reaction functions are as follows by solvingformula (5):
πβ
1=π΄πΈ + π΅πΆ
π·πΈ β πΆ2,
πβ
2=π΄πΆ + π΅π·
π·πΈ β πΆ2,
(6)
-
4 Discrete Dynamics in Nature and Society
where π΄ = π + π1π€1β π2π€2+ ((π€
1+ π€2)π2/π π ), π΅ = π +
π2π€2β π1π€1+ ((π€1+ π€2)π2/π π ), πΆ = π
1+ π2β (π2/π π ), π· =
(π2/π π ) + 2π
1, πΈ = (π
2/π π ) + 2π
2.
Formula (6) is the optimal decisionmaking of the retaileron the premise of π€
1, π€2; the retailer can obtain the decision
after it observes the manufacturerβs behavior. Substituteformula (6) into formulas (3) and (4); the optimal wholesaleprice of the two manufacturers can be obtained by the first-order conditions of formulas (3) and (4):
ππΈ (ππ1
)
ππ€1
= π β π1π1+ π1π2β
(π€1β ππ1
) π2
π π1
β
(π€1β ππ1
)
(π2/π π + 2π1) (π2/π
π + 2π2) β (π
1+ π2β π2/π
π )2
Γ {π1[(
π2
π π
+ 2π2)(
π2
π π
+ π1)
+ (π1+ π2βπ2
π π
)(βπ1+π2
π π
)]
β π1[(π1+ π2βπ2
π π
) (π2
π π
+ π1)
+ (π2
π π
+ 2π1)(βπ
1+π2
π π
)]} ,
ππΈ (ππ2
)
ππ€2
= π β π2π2+ π2π1β
(π€2β ππ2
) π2
π π2
β
(π€2β ππ2
)
(π2/π π + 2π1) (π2/π
π + 2π2) β (π
1+ π2β π2/π
π )2
Γ {π2[(
π2
π π
+ 2π2)(
π2
π π
β π2)
+ (π1+ π2βπ2
π π
)(π2+π2
π π
)]
β π2[(π1+ π2βπ2
π π
)(π2
π π
β π2)
+ (π2
π π
+ 2π1)(π2+π2
π π
)]} .
(7)
Let ππΈ(ππ1
)/ππ€1= 0 and ππΈ(π
π2
)/ππ€2= 0; we can find
out the equilibrium solutions of the two manufacturers tothe retailer at some stage; the equilibrium solutions expressoptimal decision of the twomanufacturers in various possiblesituations in a game stage.
In the actual decision process, the economic behavior ofthe decision-maker always shows limited rational character-istics, such as risk-averse behavior. Then, decision results aredifferent from the one which is perfectly rational. Optimalsolution is an optimal state when other parameters in thesystem change. Although it is very much rare for the systemto go into an optimal state, how to determine the adjustmentspeed and pricing orientation is the focus of this research. Inthis paper, the two manufacturers make decisions based onlimited rational expectations; they adjust the game processon the basis of last period marginal utilities. If the marginalutilities in period π‘ are positive, they will continue theiroutput adjustment strategy in period π‘ + 1. The process canbe modeled as follows:
π€1 (π‘ + 1) = π€1 (π‘) + π1π€1 (π‘)
ππΈ (ππ1
)
ππ€1
,
π€2 (π‘ + 1) = π€2 (π‘) + π2π€2 (π‘)
ππΈ (ππ2
)
ππ€2
,
π1 (π‘) = (([π + π1π€1 (π‘) β π2π€2 (π‘)
+(π€1 (π‘) + π€2 (π‘)) π
2
π π
](π2
π π
+ 2π2)
+ [π + π2π€2 (π‘) β π1π€1 (π‘)
+(π€1 (π‘) + π€2 (π‘)) π
2
π π
](π1+ π2βπ2
π π
))
Γ ((π2
π π
+ 2π1)(
π2
π π
+ 2π2)
β(π1+ π2βπ2
π π
)
2
)
β1
) ,
π2 (π‘) = (([π + π1π€1 (π‘) β π2π€2 (π‘)
+(π€1 (π‘) + π€2 (π‘)) π
2
π π
](π1+ π2βπ2
π π
)
+ [π + π2π€2 (π‘) β π1π€1 (π‘)
+(π€1 (π‘) + π€2 (π‘)) π
2
π π
](π2
π π
+ 2π1))
Γ ((π2
π π
+ 2π1)(
π2
π π
+ 2π2)
β(π1+ π2βπ2
π π
)
2
)
β1
) ,
(8)
-
Discrete Dynamics in Nature and Society 5
0 1 20
5
10
15
20
25
30
35
40
45
X: 0.00102Y: 31.22
k1
w
X: 0.00103Y: 26.33
Γ10β30.5 1.5
(a)
k1Γ10β3
28
30
32
34
36
38
40
42
44
0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5
X: 0.00102Y: 37.37
p X: 0.00102Y: 36.59
(b)
k1Γ10β3
Lyapun
ov
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6β1.2
β1
β0.8
β0.6
β0.4
β0.2
0
0.2
0.4
(c)
Figure 1: Price bifurcation and Lyapunov of two manufacturers and retailer with change of π1when π
2= 0.001: (a) wholesale price; (b) retail
price; (c) Lyapunov.
where π1is the adjustment coefficient of π€
1and π
2is the
adjustment coefficient of π€2. According to the dynamic
adjustment process, we can see that the wholesale price of themanufacturer is related to the price adjustment speed, retailprice, the mean and variance of the base demand level, andrisk tolerance level ofmanufacturers. Similarly, the retail priceof the retailer is related to the adjustment speed coefficient,wholesale price, the mean and variance of the base demandlevel, and risk tolerance level of the retailer.
4. The Complex Dynamic Behavior
The ultimate goal of the supply chain is to pursue profitmaximization for each of the participants and to achieveoptimum overall. Therefore, they should adjust price basedon their marginal profit of last period.
4.1. The Fixed Point. In system (8), letting π€π(π‘ + 1) =
π€π(π‘) (π = 1, 2), we can get the fixed points of sys-
tem (8). Before we solve the fixed points of system (8),we first assign some parameters considering the actualcompetition: π = 1000, π
1= 1.2, π
2= 1, π
1=
0.8, π2
= 0.7, π = 60, π π
= 60, π π1
= π π2
= 60,
πΆπ1
= 15, and πΆπ2
= 10. We will calculate all the fixedpoints and only consider the Nash equilibrium point (π€
1=
31.22, π€2= 26.33, π
1= 37.37, and π
2= 36.59).
Jacobian matrix of (8) in the Nash equilibrium point is
π½ = (1 + 91895.38π
1146744.3π
1
202024.66π2
1 + 119860.3π2
) . (9)
Characteristic polynomial of (9) isπ(π) = π2βπ΄π+π΅, whereπ΄ = 91895.38π
1+ 119860.3π
2+ 2 and π΅ = 91895.38π
1+
119860.3π2+ 10718148142.27π
1π2+ 1.
-
6 Discrete Dynamics in Nature and Society
0
5
10
15
20
25
30
35
40
45
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
k1 and k2
X: 0.00098Y: 31.22
X: 0.00102Y: 26.33
Γ10β3
w
(a)
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
k1 and k2 Γ10β3
10
15
20
25
30
35
40
45
50
X: 0.001Y: 37.37
X: 0.001Y: 36.59
p
(b)
Figure 2: Price bifurcation of two manufacturers and retailer with change of π1and π
2: (a) wholesale price; (b) retail price.
25 3035 40
45 50
2530
3540
45502530
35
40
45
50
X: 31.22Y: 37.37Z: 36.59
w1p1
p2
(a)
w1p1
p2
010
2030
4050
020
40600
10
20
30
40
50
(b)
Figure 3: The chaos attractor of the supply chain system: (a) π1= π2= 0.001; (b) π
1= π2= 0.0015.
0
5
10
15
20
25
30
35
40
45
X: 0.00102Y: 31.18
k1
X: 0.00104Y: 26.28
Γ10β30 0.5 1 1.5 2
w
(a)
k1Γ10β3
0 0.5 1 1.5 220
25
30
35
40
45
50
55
X : 0.00102Y : 45.09
X : 0.00102Y : 45
p
(b)
Figure 4: Bifurcation diagram of π€1, π€2, π1, and π
2with change of π
1when π
2= 0.001, π
π = 120: (a) π€
1, π€2; (b) π
1, π2.
-
Discrete Dynamics in Nature and Society 7
0
10
20
30
40
50
60
70
k1Γ10β3
0 0.5 1 1.5 2
X: 0.00129Y: 46.73
X: 0.00134Y: 26.4
w
(a)
k1Γ10β3
0 0.5 1 1.5 215
20
25
30
35
40
45
50
55
60
X: 0.00129Y: 48.89
X: 0.0013Y: 40.6p
(b)
Figure 5: Bifurcation diagram of π€1, π€2, π1, and π
2with change of π
1when π
2= 0.001, π
π1= 120: (a) π€
1, π€2; (b) π
1, π2.
4 5 6 7 8 9 10 11 12 13
14
16
18
20
22
24
26
28
30
32
X: 0.00085Y: 26.98
X: 0.00087Y: 22.06
Γ10β4k1
w
(a)
Γ10β4k1
4 5 6 7 8 9 10 11 12 13
24
26
28
30
32
34
36
X: 0.00085Y: 31.14
X: 0.00085Y: 29.99
p
(b)
Figure 6: Bifurcation diagram of π€1, π€2, π1, and π
2with change of π
1when π
2= 0.001, π = 70: (a) π€
1, π€2; (b) π
1, π2.
The local stability of Nash equilibrium can be gainedaccording to Routh-Hurwizβs condition:
π (1) = 1 β π΄ + π΅ > 0,
πΉ (β1) = 1 + π΄ + π΅ > 0,
π΅ β 1 < 0.
(10)
Condition (10) gives the necessary and sufficient conditionsof stable region of the Nash equilibrium point. Economicmeaning of the stable region is that, whatever initial price ischosen by the two manufacturers in local stable region, theywill eventually achieve Nash equilibrium price after finitegames. It is important to notice that the two manufacturersmay accelerate the price adjustment speed in order to increasetheir profit. Price adjustment parameter does not change
Nash equilibriumpoint. Once onemanufacturer adjusts pricetoo fast and pushes π
1, π2out of the stable region, the system
tends to become unstable and falls into chaos. On the basis ofthe parameters assigned above, we use numerical simulationto describe the dynamic behaviors of system (8).
4.2. The Effect of Price Adjustment Speed on the System
(1)The InfluenceWhich the Price Adjustment SpeedHas on theBehaviors of the TwoManufacturers and the Retailer. Since thetwo manufacturersβ behaviors are similar, we only discuss theinfluence on system behaviors when the parameter of π
1is
changed.First, we can get the price trajectory diagrams of the
two manufacturers and retailer with change of π1when
-
8 Discrete Dynamics in Nature and Society
0 50 100 150 2000
5
10
15
20
25
30
35
40
45
50
Price
t
Figure 7: Power spectrum of π€1, π€2, π1, and π
2when π
1= π2=
0.0005.
π2
= 0.001, π€1
= 20, and π€2
= 15, as shown inFigure 1(a) and Figure 1(b). We can obtain that π€
2is less
affected by change of π1and π€
1, π€2, π1, and π
2change from
the stable period, period-doubling bifurcation to the chaosin three trajectories. When π
1β [0, 0.00102], π€
1, π€2, π1,
and π2are stable. When π
1= 0.00102, the first bifurcation
appears in π€1, π€2, π1, and π
2, and the price of the two
manufacturers and retailer vibrates in two points; after thatthe second bifurcation appears in the system; finally, thesystem goes into chaos, the price behaves more disorderly,and the market behaviors become unpredictable. Figure 1(c)shows corresponding change of the Lyapunov exponent. Thepositive Lyapunov exponent is used to mark the chaos; thebigger the Lyapunov exponent, the stronger the chaos. Thesystem is in chaos when most of the Lyapunov exponentsare positive. Figure 2 shows the price trajectory diagrams ofthe two manufacturers and retailer with the change of π
1and
π2simultaneously. We can observe that the stable regions of
π€1, π€2are smaller than the one with change of only π
1or only
π2and the π€
1, π1, and π
2change from the stable period and
period-doubling bifurcation to the chaos in four trajectories;the dynamic characteristics of π€
2in particular do not follow
the period-doubling bifurcation. Because π1and π2change at
the same time, the retailerβs behaviors appear in complicatedcharacteristics.
Figure 3(a) shows an attractor of the supply chain whichis in stable state when π
1= π2= 0.001. Figure 3(b) shows
a chaos attractor of the supply chain which appears in acomplex state when π
1= π2= 0.0015; it is another chaos
characteristic of the variables.
Proposition 1. Thevalue of price adjustment speed determineswhether the supply chain is stable or not; when we determinethis value, each manufacturer should consider market reactionof the competitors and retailer. Only upstream and downstreamenterprises keep stable, to ensure the stability of the supply chainand to maximize the enterpriseβs profit.
(2) The Influence Which Change of Risk Preference Has onthe Behaviors of the Two Manufacturers and the Retailer. Letπ2= 0.001, π
π = 120; the value of other parameters is
the same as in the previous assumption; we can obtain theprice bifurcation diagrams of the two manufacturers and theretailer with change of π
1, as shown in Figure 4. We can see
that the stable region is not changed, wholesale price of thetwo manufacturers declines, and the retail price goes up withchange of π
π . Figure 5 shows the price bifurcation diagrams
of the two manufacturers and the retailer with change of π1
when π2= 0.001, π
π1
= 120, the first bifurcation pointis π1= 0.00129, and the equilibrium value is (46.73, 26.4,
48.89, and 40.6). We can make a conclusion that the riskpreference of the manufacturer can affect the stable region ofsystem and change the equilibrium point of system, and therisk preference of the retailer cannot affect the stable regionof system and change the equilibrium point of system.
(3) The Influence Which the Uncertain Demand Has on theSystem Variable Behaviors. Because the demand forecastingexists errors, the customer demand is always uncertain. Nextwe will observe the price change of the two manufacturersand retailer with change of π
1when π = 70, π
2= 0.001, and
it is shown in Figure 6. We can see that, with π increasing,the first bifurcation point of the system is π
1= 0.00105, and
π€1, π€2, π1, and π
2are all falling.
Proposition 2. Uncertain demand makes the supply chainaccess chaos quickly, and the wholesale price and the retail pricedecline. It is uncertain about their respective profit; thereby thecompetitiveness of the supply chain becomes weak.
(4) The Power Spectrum of Variables and the Sensitive Depen-dence on Initial Conditions.We adopt a cycle diagrammethodto estimate the power spectrum of variables. Next, we willobserve the price change of the two manufacturers andretailer when π
1= π2= 0.001 and the initial values are
π€1= 20, π€
2= 15; it is shown in Figure 7. We know that the
supply chain is stable when π1= π2= 0.001, so the power
spectrum of the variables is straight lines which conformto the attractor in Figure 3(a). When π
1= π2= 0.0015,
the supply chain is in chaos. From Figure 8, we can see thatπ€1, π€2, π1, and π
2vibrate with a frequency, π€
2changes with
approximate periodic motion, and range of movement of π1
is bigger than the one of π2.
Then, we will observe the price change of the twomanufacturers and the retailer when π
1= π2= 0.0015
and (π€1, π€2) = (20.01, 15) and (20, 15) which has smaller
change inπ€1and no change inπ€
2; they are shown in Figure 9.
We can see that the price of the two manufacturers andretailer has distinct change. The sensitive dependence oninitial conditions is another important characteristic of thechaotic system as it fully manifests the sensitive dependenceon the initial conditions of the system (8).
Through the above analysis, we know that the risk-aversesupply chain has been in chaos. At this state, the price of thetwo manufacturers and the retailer changes disorderly fromthe beginning of the Nash equilibrium.
-
Discrete Dynamics in Nature and Society 9
0 50 100 150 2000
5
10
15
20
25
30
35
40
45
50
0 50 100 150 2000
5
10
15
20
25
30
35
40
45
50w1
w2
p1
p2
t t
tt
0 50 100 150 2000
5
10
15
20
25
30
35
40
45
50
0 50 100 150 2000
5
10
15
20
25
30
35
40
45
50
Figure 8: Power spectrum of π€1, π€2, π1, and π
2when π
1= π2= 0.0015.
(5) The Effect of Parameter Change on the Profit. Figure 10shows the profit bifurcation of the two manufacturers andretailer with changes of π
1. Obviously, the profit bifurcation is
similar to price bifurcation including period-doubling bifur-cation, four-period-doubling bifurcation, and chaos state.Figure 11 shows the profit of the two manufacturers andretailer in 50 games when the system is in a stationary period,two-period-doubling bifurcation, and a chaotic period. In thedifferent period, the fluctuation range of price ofπ
1is larger
than that ofπ2. Tables 1, 2, and 3 give the profit data of the two
manufacturers and retailer, respectively, in different periodswith change of π
1, π π1
, and π.
Proposition 3. First, the total profit of the system in the chaoticperiod is less than that in other periods. Second, the total profitof the system and respective profit of the twomanufacturers andthe retailer will increase with increase ofπ
π1
and decrease withincrease ofπ.Third, when the system is in chaos, the profit of thetwo manufacturers will decrease, but the profit of retailer willincrease. Namely, when the system goes into chaos, the retaileris a beneficiary.This is why some participants set out contract to
Table 1:The profit of the twomanufacturers and retailer in differentperiods when π
1= 0.001.
Differentperiods Stable period
Two-period-doublingbifurcation
Chaotic period
ππ
8159.76 8144.99 8175.85ππ1
8074.74 6126.11 2403.59ππ2
8128.54 8149.99 8054.7βπ 24363.04 22421.09 18634.14
avoid disordered competition in some situations and why someparticipants prefer chaotic market in some cases.
5. Chaos Control
Competitive manufacturers will certainly want to achievemaximum profit in the existing supply chain. Through theabove analysis, we can see that the change of π
1, π2, and
π often causes disorder behaviors in the market which are
-
10 Discrete Dynamics in Nature and Society
0 20 40 60 80 100
0
10
20
30
40w1
w2
p1
p2
t t
t t
β10
β20
β30
β400 20 40 60 80 100
0
1
β0.5
β1
β1.5
0.5
1.5
0 20 40 60 80 100
0
10
20
30
β10
β20
β30
X: 49Y: 25.74
X: 36Y: β25.78
0 20 40 60 80 100
0
2
4
6
8
10
X: 49Y: 8.783
X: 36Y: β8.895
β2
β4
β6
β8
β10
Figure 9: Different change of price of the two manufacturers and the retailer with increase of game at π1= π2= 0.0015.
Table 2:The profit of the twomanufacturers and retailer in differentperiods when π
π1= 120.
Differentperiods Stable period
Two-period-doublingbifurcation
Chaotic period
ππ
8182.74 8190.65 8203.86ππ1
15539.95 14331.78 12443.96ππ2
8144.84 8136.82 8125.35βπ 31867.63 30659.25 28773.17
of disadvantage to the stability of the supply chain and thedevelopment of the enterprise. However, the participantsoften maximize their own profit by any kind of means inthe process of marketization. So the market will be out oforder and finally falls into chaos. It is particularly importantthat each participant should make rational strategic decisiontimely formaking the system return to the stable equilibrium.
Parameter adjustment and feedback control method ofthe system variable will be used to control the chaos of
Table 3:The profit of the twomanufacturers and retailer in differentperiods when π = 120.
Differentperiods Stable period
Twoperiod-doubling
bifurcationChaotic period
ππ
6001.45 6009.92 6013.69ππ1
5918.76 4346.43 3573.6ππ2
5972.61 5968.58 5967.1βπ 17892.81 16324.92 15554.38
the system (8). It is often used at the chaos control ofgeneral discrete dynamic system. We will analyze systemchaos control based on the influence of π
1on the stability of
the risk-averse supply chain. Assume system (8) isπ€π(π‘ +1) =
ππ(π€1(π‘), π€2(π‘)); the system under control is as follows:
π€1 (π‘ + π) = (1 β π’) π
π
1(π€1 (π‘) , π€2 (π‘)) + π’π€1 (π‘) ,
π€2 (π‘ + π) = (1 β π’) π
π
2(π€1 (π‘) , π€2 (π‘)) + π’π€1 (π‘) ,
-
Discrete Dynamics in Nature and Society 11
0 10
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000X: 0.00104Y: 8072
1.50.5
k1
2Γ10β3
M1
0 1 1.50.5k1
2Γ10β3
0 1 1.50.5k1
2Γ10β3
M2
8020
8040
8060
8080
8100
8120
8140
8160
8180
8200
8220
8100
8150
8200
8250
8300
8350
8400
R1
Figure 10: The profit bifurcation with change of π1when π
2= 0.001.
π1 (π‘ + π) = π3 (π€1 (π‘) , π€2 (π‘)) ,
π2 (π‘ + π) = π4 (π€1 (π‘) , π€2 (π‘)) .
(11)
When π’ = 0, the controlled system is the original system;thus, the controlled system has the same periodic orbit athe original system. When π = 1, each step iteration ofthe fixed point is in control. When π = 2, 4, and so forth,namely, to control the second cycle orbit, four cycles orbit. Aslong as we select an appropriate value for π’, we can ensuredelayed bifurcation with π
1at fixed point and keep the supply
chain being stable within large scope of π1. In addition, with
appropriate adjustment value of π, we can realize stabilitycontrol of higher cycle orbits for the chaotic attractor. Then,we discuss the stability control in the supply chain whenπ = 1 and try to stabilize the price on the fixed point. Thecontrolled system can be represented as
π€1 (π‘ + 1)=(1 β π’)(π€1 (π‘) + π1π€1 (π‘)
ππΈ (ππ1
)
ππ€1
) + π’π€1 (π‘) ,
π€2 (π‘ + 1)=(1 β π’)(π€2 (π‘) + π2π€2 (π‘)
ππΈ (ππ2
)
ππ€2
) + π’π€2 (π‘) ,
π1 (π‘)
= (([π + π1π€1 (π‘) β π2π€2 (π‘)
+(π€1 (π‘) + π€2 (π‘)) π
2
π π
](π2
π π
+ 2π2)
+ [π + π2π€2 (π‘) β π1π€1 (π‘)
+(π€1 (π‘) + π€2 (π‘)) π
2
π π
] (π1+ π2βπ2
π π
))
Γ ((π2
π π
+ 2π1)(
π2
π π
+ 2π2)
β(π1+ π2βπ2
π π
)
2
)
β1
) ,
-
12 Discrete Dynamics in Nature and Society
30 35 40 45 50
6200
6400
6600
6800
7000
7200
7400
7600
7800
8000
8200
Profi
t
t
M1M2
R
(a)
M1M2
R
t
0 10 20 30 40 504000
4500
5000
5500
6000
6500
7000
7500
8000
8500
Profi
t(b)
M1M2
R
t
0 10 20 30 40 500
1000
2000
3000
4000
5000
6000
7000
8000
9000
Profi
t
(c)
Figure 11: The profit of the two manufacturers and retailer π2= 0.001: (a) stable period; (b) two-period-doubling bifurcation; (c) chaotic
period.
π2 (π‘)
= (([π + π1π€1 (π‘) β π2π€2 (π‘)
+(π€1 (π‘) + π€2 (π‘)) π
2
π π
](π1+ π2βπ2
π π
)
+ [π + π2π€2 (π‘) β π1π€1 (π‘)
+(π€1 (π‘) + π€2 (π‘)) π
2
π π
](π2
π π
+ 2π1))
Γ ((π2
π π
+ 2π1)(
π2
π π
+ 2π2)
β (π1+ π2βπ2
π π
)
2
)
β1
) .
(12)
Figures 12 and 13 show that the chaos system can be controlledgradually from the four-period-doubling bifurcation andtwo-period-doubling bifurcation to the fixed point with the
-
Discrete Dynamics in Nature and Society 13
20
22
24
26
28
30
32
34
36
38
k1Γ10β3
w
0 0.5 1 1.5 2 2.5
(a)
w
k1Γ10β3
0 0.5 1 1.5 2 2.523
24
25
26
27
28
29
30
31
32
(b)
Figure 12: Control bifurcation diagram of π€1and π€
2with change of π
1when π
2= 0.001: (a) π’ = 0.5; (b) π’ = 0.7.
0 10
5
10
15
20
25
30
35
40
45
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
w
u
(a)
p
u
0 110
15
20
25
30
35
40
45
50
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
(b)
Figure 13: Control bifurcation diagram of π€1, π€2, π1and π
2with change of π’ when π
1= 0.002, π
2= 0.001: (a) π€
1and π€
2; (b) π
1and π
2.
control parameter π’ increasing. When π’ > 0.4825, thecontrolled system can be stabilized at the Nash equilibriumpoint. In a real market, we can consider π’ as the regulationon the price adjustment speed which could avoid marketchaos whenmanufacturers pursue their maximum profit.Wecan also consider π’ as the learning ability or adaptabilityof the market. For instance, the manufacturer will adjustprice according to the information in the past. Due to thecomplexity, disorder, and randomness in chaotic state, themanufacturer should have a clear forecast of chaos control.After adjusting parameter π’, we can make the periodic orbitstable at the expected point.
6. Conclusions
Considering the randomness of customerβs requirements andsupplyβs risk preference, we study the supply chain in themar-ket which consists of two manufacturers and a retailer. We
construct a dynamic Stackelberg and Bertrand pricing gamemodel and find that bifurcation, chaos, and other complexphenomena occur when the price adjustment speed, the levelof risk preference, and predict error change. When the chaosoccurs, the stability of the whole supply system is broken, andthe market becomes abnormal, irregular, and unpredictable.It is important to note that, when determining the valueof the parameter, each manufacturer should consider themarket reaction of competitors and retailer to ensure that theupstream and downstream enterprises keep stable and ensurethat the supply chain keeps stable for their maximum profit.Finally, we use the parameter adjustment method to controlthe supply system.Then, we have obtained some conclusionsfrom our research.
(1) The twomanufacturers play the Bertrand game; whenπ1changes its price adjustment speed, its wholesale
price presents deterministic characteristics of chaos,
-
14 Discrete Dynamics in Nature and Society
and the wholesale price of π2is influenced less.
The price of retailer is influenced largely. When priceadjustment speed of the two manufacturers changesat the same time, it has no effect on the optimalpricing strategy, the stable region of the supply chaingets smaller, and the two manufacturers all showcomplicated behavior characteristics.
(2) Increasing the risk tolerance level of the retailer doesnot impact stable region, but retail price will increaseandwholesale price of the twomanufacturers will fall.If π1increases his risk tolerance level, the stability
betweenπ1and the retailer is enhanced, and the two
manufacturersβ wholesale price and productsβ retailprice go up. Uncertain demand makes the whole-sale prices and retailer prices access chaos quickly,two manufacturersβ wholesale price and retail pricedecline, and total profit and their respective profitdecrease.
(3) The total profit of the system in the chaotic period isless than that in other periods; it and respective profitof the twomanufacturers and the retailer will increasewith increase of π
π1
and decrease with increase ofπ. When the system is in chaos, the profit of the twomanufacturers will decrease, but the profit of retailerwill increase. Namely, when the system goes into thechaos, the weaker (retailer) is a beneficiary.
This paper is a realistic guide for the risk-averse supply chainto formulate its parameter adjustment strategies to avoid theloss of the respective profit and the total profit. It is alsoa realistic reference for the managers to formulate relevantpolicies onmacroeconomic control.Themanagers can adjustparameters to make the supply chain in good operatingcondition based on the operation condition of the supplychain; for example, when the variance of customer demandgets bigger, the manager should reduce the price adjustmentspeed for the stability of the supply chain.
There are several possible directions for the futurestudy. First, one can study three-echelon supply chain withcomplexity behaviors under the different game strategiesconsidering the delay time, such as the Stackelberg gamebetween the two manufacturers. Second, one can adopt adifferent form of demand function and, finally, other controlmethods may be applied in order to achieve different results.
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper.
Acknowledgments
The authors thank the reviewers for their careful reading andproviding some pertinent suggestions. The research was sup-ported by the National Natural Science Foundation of China(No. 61273231), Doctoral Fund of Ministry of Education ofChina (Grant No. 20130032110073) and supported by TianjinUniversity Innovation Fund.
References
[1] J. Wei, J. Zhao, and Y. Li, βPricing decisions for complementaryproducts with firmsβ differentmarket powers,β European Journalof Operational Research, vol. 224, no. 3, pp. 507β519, 2013.
[2] S. K. Mukhopadhyay, X. Yue, and X. Zhu, βA Stackelbergmodel of pricing of complementary goods under informationasymmetry,β International Journal of Production Economics, vol.134, no. 2, pp. 424β433, 2011.
[3] O. Caliskan-Demirag, Y. Chen, and J. Li, βCustomer and retailerrebates under risk aversion,β International Journal of ProductionEconomics, vol. 133, no. 2, pp. 736β750, 2011.
[4] C. Luo and J. Huang, βPricing and ordering strategies in supplychain with different risk preference,β Computer Integrated Man-ufacturing Systems, vol. 18, no. 4, pp. 859β866, 2012.
[5] B. Hu and X. Chen, βSupply chain sale-surety option contractwith effort dependent demand and risk preference,β Journal ofIndustrial Engineering: Engineering Management, vol. 26, no. 3,pp. 184β190, 2012.
[6] Z. Guan and B. Zhou, βIntegrated optimization of three-levelsupply chain under different risk attitude,β Operations Researchand Management Science, vol. 20, no. 6, pp. 19β24, 2011.
[7] S. Huang and C. Yang, βSupply chain coordination modelconsidering risk attitudes of agents with VaR constraints,βJournal of Industrial Engineering: Engineering Management, vol.25, no. 2, pp. 136β141, 2012.
[8] T. Puu, βThe chaotic duopolists revisited,β Journal of EconomicBehavior and Organization, vol. 33, no. 3-4, pp. 385β394, 1998.
[9] J. Ma and Z. Sun, βThe research on price game model and itscomplex characteristics of triopoly in different decision-makingrule,β Nonlinear Dynamics, vol. 71, no. 1-2, pp. 35β53, 2013.
[10] J. Ma and H. I. Bangura, βComplexity analysis research offinancial and economic system under the condition of threeparametersβ change circumstances,β Nonlinear Dynamics, vol.70, no. 4, pp. 2313β2326, 2012.
[11] H. B. Hwarng and N. Xie, βUnderstanding supply chaindynamics: a chaos perspective,β European Journal of OperationalResearch, vol. 184, no. 3, pp. 1163β1178, 2008.
[12] J. Wang and X. Wang, βComplex dynamic behaviors of con-strained supply chain systems,β Systems Engineering-Theory &Practice, vol. 32, no. 4, pp. 746β751, 2012.
[13] J.Ma andY. Feng, βThe study of the chaotic behavior in retailerβsdemand model,β Discrete Dynamics in Nature and Society, vol.2008, Article ID 792031, 12 pages, 2008.
[14] E. Gal-Or, βInformation sharing in oligopoly,β Econometrica,vol. 53, no. 2, pp. 329β343, 1985.
[15] J. S. Raju and A. Roy, βMarket information and firm perfor-mance,β Management Science, vol. 46, no. 8, pp. 1075β1084,2000.
[16] X. Vives, βDuopoly information equilibrium: cournot andBertrand,β Journal of Economic Theory, vol. 34, no. 1, pp. 71β94,1984.
[17] A.Kunstman, βDecisionmaking under uncertainty:models andchoices,β European Journal of Operational Research, vol. 8, no. 4,pp. 401β402, 1981.
[18] M. R. Walls, βCombining decision analysis and portfolio man-agement to improve project selection in the exploration andproduction firm,β Journal of Petroleum Science and Engineering,vol. 44, no. 1-2, pp. 55β65, 2004.
[19] R. D. Banker, I. Khosla, and K. K. Sinha, βQuality and competi-tion,βManagement Science, vol. 44, no. 9, pp. 1179β1192, 1998.
-
Submit your manuscripts athttp://www.hindawi.com
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttp://www.hindawi.com
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
CombinatoricsHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
International Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com
Volume 2014 Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Stochastic AnalysisInternational Journal of