11.08.2013 Views

Coordination by Option Contracts in a Retailer-Led Supply Chain with

Coordination by Option Contracts in a Retailer-Led Supply Chain with

Coordination by Option Contracts in a Retailer-Led Supply Chain with

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

TSINGHUA SCIENCE AND TECHNOLOGY<br />

ISSN 1007-0214 22/22 pp570-580<br />

Volume 13, Number 4, August 2008<br />

<strong>Coord<strong>in</strong>ation</strong> <strong>by</strong> <strong>Option</strong> <strong>Contracts</strong> <strong>in</strong> a <strong>Retailer</strong>-<strong>Led</strong> <strong>Supply</strong> Cha<strong>in</strong><br />

<strong>with</strong> Demand Update *<br />

WANG Xiaolong (王小龙), LIU Liwen (刘丽文) **<br />

School of Economics and Management, Ts<strong>in</strong>ghua University, Beij<strong>in</strong>g 100084, Ch<strong>in</strong>a<br />

Abstract: This research exam<strong>in</strong>es how to use an option contract to coord<strong>in</strong>ate a retailer-led supply cha<strong>in</strong><br />

where the market <strong>in</strong>formation can be updated. Based on Stackelberg game theory, we build a mode <strong>with</strong> one<br />

supplier and one retailer <strong>in</strong> which the retailer designs contracts to coord<strong>in</strong>ate the supplier’s production <strong>in</strong> a<br />

two-mode production environment. This focuses on an option contract that consists of two option prices and<br />

one exercise price. By theoretical analysis and numerical example, we f<strong>in</strong>d that such a contract can coord<strong>in</strong>ate<br />

the supplier and retailer to act <strong>in</strong> the best <strong>in</strong>terest of the channel. The optimal pric<strong>in</strong>g conditions are<br />

given as follows: First, option prices should be negatively correlated to the exercise price and should be <strong>in</strong> a<br />

relevant range. Second, the first-period option price should be no greater than the second-period price and<br />

should be l<strong>in</strong>early correlated to the second-period option price when the latter is beyond some threshold. The<br />

results show that such option contracts can arbitrarily allocate the extra system profit between the two parties<br />

so that each party is <strong>in</strong> a w<strong>in</strong>-w<strong>in</strong> situation.<br />

Key words: supply cha<strong>in</strong> coord<strong>in</strong>ation; option contracts; retailer-led; demand update<br />

Introduction<br />

Contemporary bus<strong>in</strong>ess puts more and more emphasis<br />

on channels, shift<strong>in</strong>g power from supplier to retailer at<br />

the downstream end of a supply cha<strong>in</strong>, which is the<br />

closest to the customers. Mess<strong>in</strong>ger and Narasimhan [1] ,<br />

Ailawadi [2] , and Bloom and Perry [3] offer empirical<br />

evidence on how power has transferred <strong>in</strong> several <strong>in</strong>ternational<br />

retail<strong>in</strong>g markets. Such shift<strong>in</strong>g enables the<br />

retailers to lead the supply cha<strong>in</strong> <strong>in</strong> some sense. They<br />

then beg<strong>in</strong> to raise str<strong>in</strong>gent requirements on the suppliers’<br />

production, compared <strong>with</strong> the traditional case<br />

where the supplier dom<strong>in</strong>ates the supply cha<strong>in</strong> and coord<strong>in</strong>ates<br />

the retailer’s order<strong>in</strong>g behavior. Another<br />

Received: 2006-10-24<br />

* Supported <strong>by</strong> the National Natural Science Foundation of Ch<strong>in</strong>a<br />

(Nos. 70532004 and 70621061)<br />

** To whom correspondence should be addressed.<br />

E-mail: liulw@sem.ts<strong>in</strong>ghua.edu.cn<br />

Tel: 86-10-62783553; Fax: 86-10-62785876<br />

aspect is that the market situation often changes greatly<br />

long after the retailer has set an order. Some retailers,<br />

like Wal-Mart, deal <strong>with</strong> such problems <strong>by</strong> order<strong>in</strong>g<br />

less but more frequently. However, this volatile market<br />

puts great pressure on the suppliers who must make<br />

wise production/capacity build<strong>in</strong>g decisions to cope<br />

<strong>with</strong> the irregular orders which mirror a volatile market.<br />

Unfortunately, quite often suppliers fail to do that. In<br />

an effort to reduce mismatches between supply and<br />

demand, many <strong>in</strong>dustries, like the fashion <strong>in</strong>dustry (as<br />

reported <strong>by</strong> Fisher et al. [4] ) and the consumer electronics<br />

<strong>in</strong>dustry, are mov<strong>in</strong>g to tighten coord<strong>in</strong>ation across<br />

their supply cha<strong>in</strong>s. One important <strong>in</strong>itiative for suppliers<br />

is to develop faster, but typically more expensive,<br />

production modes that are capable of produc<strong>in</strong>g a second<br />

run of products closer to the sell<strong>in</strong>g season. This<br />

mode allows both sides to take advantage of updated<br />

demand forecasts. Also, the retailer can better manipulate<br />

the supplier’s production so that channel coord<strong>in</strong>ation<br />

is reached. This phenomenon has become more


WANG Xiaolong (王小龙) et al:<strong>Coord<strong>in</strong>ation</strong> <strong>by</strong> <strong>Option</strong> <strong>Contracts</strong> <strong>in</strong> a <strong>Retailer</strong>-<strong>Led</strong> …<br />

common <strong>with</strong> advances <strong>in</strong> computer-based manufactur<strong>in</strong>g<br />

technologies which significantly reduce production<br />

run times and product changeover times.<br />

However, the addition of another production mode<br />

is not enough to reach channel coord<strong>in</strong>ation due to<br />

double marg<strong>in</strong>alization which refers to the fact that a<br />

party’s relative cost structure becomes distorted when<br />

a transfer price is <strong>in</strong>troduced <strong>with</strong><strong>in</strong> a supply channel.<br />

For example, <strong>in</strong>troduc<strong>in</strong>g a wholesale price <strong>in</strong>to a s<strong>in</strong>gle-stage<br />

newsvendor model causes the ratio of the<br />

buyer’s underage and overage costs to fall short of the<br />

channel ratio. As a result, the buyer orders less than the<br />

channel optimal quantity. In a two-stage production<br />

environment, double marg<strong>in</strong>alization threatens to impact<br />

not only the total production quantity but also<br />

proper allocation of production between periods. The<br />

double marg<strong>in</strong>alization concept was first <strong>in</strong>troduced <strong>in</strong><br />

the economics literature <strong>by</strong> Spengler [5] , who raised the<br />

issue <strong>in</strong> terms of <strong>in</strong>appropriate prices rather than <strong>in</strong>appropriate<br />

quantities. Tirole [6] provided further details<br />

on the phenomenon. Many works <strong>in</strong> the operations<br />

management field, like those of Lariviere and<br />

Porteus [7] , have shown that price-only contracts generally<br />

fail to coord<strong>in</strong>ate the supply cha<strong>in</strong> due to double<br />

marg<strong>in</strong>alization. However, other contracts have proven<br />

to efficiently coord<strong>in</strong>ate the supply cha<strong>in</strong>. These contracts<br />

<strong>in</strong>clude buyback contracts [8] , revenue shar<strong>in</strong>g<br />

contracts [9] , sales rebate contracts [10] , quantity discount<br />

contracts [11] , and quantity flexibility contracts [12] .<br />

The goal of this paper is to provide guidance <strong>in</strong> design<strong>in</strong>g<br />

an option contract <strong>in</strong> a retailer-led supply cha<strong>in</strong>,<br />

<strong>in</strong> which the supplier’s production needs coord<strong>in</strong>ate<br />

<strong>with</strong> the existence of a market demand <strong>in</strong>formation<br />

update. The researches on option contracts <strong>in</strong> supply<br />

cha<strong>in</strong> management are ma<strong>in</strong>ly divided <strong>in</strong>to economic<br />

and operational areas. Typical economic works <strong>in</strong>clude<br />

Newbery [13] and Wu et al. [14] Typical operational<br />

works <strong>in</strong>clude Donohue [15] and Barnes-Schuster et<br />

al. [16] Our contract pric<strong>in</strong>g scheme builds most closely<br />

upon the work of Barnes-Schuster et al. [16] , which considers<br />

a comb<strong>in</strong>ation of committed orders and options.<br />

Barnes-Schuster et al. [16] analyze a two-period model<br />

<strong>with</strong> correlated demand. The present analysis studies a<br />

two-period model <strong>with</strong> demand <strong>in</strong>formation update <strong>in</strong><br />

which the market demand does not realize until the end<br />

of the second period.<br />

Specifically, consider one supplier and one retailer.<br />

571<br />

At the beg<strong>in</strong>n<strong>in</strong>g of the first period both parties anticipate<br />

a demand <strong>in</strong>formation update that will appear at<br />

the beg<strong>in</strong>n<strong>in</strong>g of the second period and they take this<br />

<strong>in</strong>to account <strong>in</strong> their first-period decisions. The coor-<br />

d<strong>in</strong>at<strong>in</strong>g contract is designed as follows: The option<br />

contract has three parameters for two option prices 1 o<br />

and o 2 and one exercise price e. The option prices<br />

serve as the unit compensation for the supplier’s production<br />

quantities. When the realized demand is beyond<br />

the retailer’s total order, the retailer will purchase<br />

the supplier’s excess <strong>in</strong>ventory, if any, to meet the<br />

market demand at the unit price e. With this contract,<br />

the retailer’s decision is to determ<strong>in</strong>e orders at the beg<strong>in</strong>n<strong>in</strong>g<br />

of each period and the supplier’s decision is to<br />

determ<strong>in</strong>e the production allocation between the two<br />

periods. This project aims to f<strong>in</strong>d an optimal set of<br />

contracts to coord<strong>in</strong>ate the channel and to realize a<br />

proper profit allocation.<br />

1 Two-Period Model<br />

The two-period model depicts the dynamics of a retailer<br />

who purchases from one ma<strong>in</strong> supplier and markets<br />

these products to a number of retail outlets, resell<strong>in</strong>g<br />

at retail price p for a s<strong>in</strong>gle sell<strong>in</strong>g season.<br />

Unlike traditional coord<strong>in</strong>ation models, the retailer <strong>in</strong><br />

the model leads the supply cha<strong>in</strong> and designs a proper<br />

contract to coord<strong>in</strong>ate the supplier’s production quantities.<br />

The specific contract form will be <strong>in</strong>vestigated.<br />

The retailer utilizes two periods of demand <strong>in</strong>formation<br />

to determ<strong>in</strong>e the order quantities. Dur<strong>in</strong>g the first period,<br />

the demand prediction is relatively less accurate<br />

<strong>with</strong> a distribution function denoted as F(D). At the<br />

beg<strong>in</strong>n<strong>in</strong>g of the second period (market demand not yet<br />

realized), more current market <strong>in</strong>formation is available<br />

which helps the retailer f<strong>in</strong>e-tune the forecast where an<br />

effective market signal will be observed depend<strong>in</strong>g on<br />

specific <strong>in</strong>dustry situations and the lengths of the two<br />

periods which may not necessarily be equal. This updated<br />

<strong>in</strong>formation is the market signal ε. Let G(·) denote<br />

its estimated cumulative distribution function and<br />

let F(·|ε) denote the demand distribution function conditional<br />

on the market signal. Assume that ε is expressed<br />

<strong>in</strong> the same units as the total demand and is<br />

also non-negative. Market demand is stochastically<br />

<strong>in</strong>creas<strong>in</strong>g <strong>in</strong> the market signal, i.e., given x, F( x| ε h ) <<br />

F( x| ε l ) for all ε h > εl<br />

(ε h and ε l are two arbitrary


572<br />

market signals). Assume that all distributions are cont<strong>in</strong>uous,<br />

<strong>in</strong>vertible, doubly differentiable, and <strong>in</strong>dependent<br />

of the wholesale and retail prices. Furthermore,<br />

they are all common knowledge to both parties from<br />

the beg<strong>in</strong>n<strong>in</strong>g of period 1.<br />

At the beg<strong>in</strong>n<strong>in</strong>g of period 1, the retailer orders 1 d<br />

units at a unit wholesale price w 1 based on a rough<br />

knowledge of the demand distribution. Once ε is<br />

observed, the retailer will adjust the order quantities <strong>by</strong><br />

order<strong>in</strong>g another d2 units at a higher unit wholesale<br />

price w 2 . Note that the retailer’s second order is cont<strong>in</strong>gent<br />

on the specific market signal. If the signal exhibits<br />

a rather weak demand <strong>in</strong> the future, the retailer<br />

will not set another order, i.e., d 2 = 0 .<br />

The supplier produces over two time periods <strong>in</strong> response<br />

to the retailer’s orders, <strong>with</strong> orders filled <strong>in</strong> one<br />

shipment before the sell<strong>in</strong>g season beg<strong>in</strong>s. The supplier<br />

has two different production modes. One is the cheap<br />

mode which <strong>in</strong>curs a unit production cost c 1 , and the<br />

other is the expensive mode which <strong>in</strong>curs a unit production<br />

cost c2 > c1.<br />

Assume that production costs<br />

<strong>in</strong>clude the cost of hold<strong>in</strong>g <strong>in</strong>ventory until delivery and<br />

the cost of delivery itself. Other cost parameters <strong>in</strong>clude<br />

a shortage penalty s paid <strong>by</strong> the retailer to the<br />

customers for each unit of unfilled demand, and a salvage<br />

value v received for each unit of <strong>in</strong>ventory rema<strong>in</strong><strong>in</strong>g<br />

at the end of the season no matter to which<br />

side it belongs. Suppose that all these prices and costs<br />

are exogenous and the analysis is limited to the most<br />

relevant and <strong>in</strong>terest<strong>in</strong>g case where p > w2 > w1 > c2<br />

><br />

c1 > v.<br />

Next assume that the lead time for the cheap<br />

mode (and the tim<strong>in</strong>g of the market signal) is such that,<br />

to ensure completion before the start of the sell<strong>in</strong>g<br />

season, the supplier must beg<strong>in</strong> production before observ<strong>in</strong>g<br />

the market signal. In contrast, the lead time for<br />

the expensive mode is short enough to allow production<br />

after the market signal is observed. Therefore, the<br />

supplier’s problem is to allocate production quantities<br />

between the two modes. Dur<strong>in</strong>g the first period, the<br />

supplier needs to determ<strong>in</strong>e how many products, denoted<br />

as q 1 , would be produced among the cheap<br />

mode. In the second period after a market signal is ob-<br />

served, the supplier has to determ<strong>in</strong>e how many products,<br />

denoted as q 2 , will be produced among the expensive<br />

mode, if necessary. Denote d = d1+ d2<br />

and<br />

Ts<strong>in</strong>ghua Science and Technology, August 2008, 13(4): 570-580<br />

q= q1+ q2.<br />

1.1 Centralized system performance:<br />

A benchmark<br />

To provide a benchmark, first consider the problem<br />

where the supplier and the retailer are each owned <strong>by</strong> a<br />

risk-neutral entity. The owner aims to maximize his<br />

own expected profit <strong>by</strong> choos<strong>in</strong>g first and second period<br />

production quantities ( q1, q 2)<br />

which solve the<br />

follow<strong>in</strong>g two-period optimization problem.<br />

∞<br />

c c<br />

max Π ( q1) =− cq 1 1+∫Π ( ε, q1, q2)d G(<br />

ε)<br />

(1)<br />

q10<br />

0<br />

c c<br />

where Π ( ε, q1, q2) = max π ( ε,<br />

q1, q2)<br />

and<br />

q20<br />

c<br />

π ( ε , q , q ) = − c q + pEm<strong>in</strong>{ D, q + q } + v( q + q −<br />

1 2 2 2 1 2 1 2<br />

Em<strong>in</strong>{ D, q1+ q2}) −sE[ D− m<strong>in</strong>{ D, q1+ q2}]<br />

(2)<br />

c<br />

Here, Π ( ε , q1, q2)<br />

represents the second-stage<br />

problem after the market signal is observed and the<br />

demand forecast is updated accord<strong>in</strong>gly. To express<br />

the second-stage problem <strong>in</strong>dependently of q 1 , substitute<br />

q for q1+ q2<br />

and rearrange the terms. This<br />

new formulation yields the same objective solution as<br />

the problem <strong>in</strong> Eq. (1).<br />

∞<br />

c c<br />

max Π ( q1) = ( c2 − c1) q1+∫Π ( ε, q1)d G(<br />

ε)<br />

(3)<br />

q10<br />

0<br />

c c<br />

where Π ( ε, q1) = max π ( ε,<br />

q)<br />

and<br />

qq c<br />

π ε 2<br />

ε<br />

0<br />

1<br />

q<br />

( , q) = ( p− c + s) q−( p− v+ s) F( x| )dx−sED<br />

(4)<br />

Note that given q 1 , the owner faces a newsvendor<br />

problem <strong>in</strong> period 2 but <strong>with</strong> an <strong>in</strong>itial <strong>in</strong>ventory and<br />

c c<br />

an adjusted demand distribution. Let ( q1, q ) denote<br />

the optimal solution to the central problem. The solution<br />

structure is given <strong>by</strong> Lemma 1.<br />

Lemma 1 The argument of the central first period<br />

c<br />

problem, Π ( q1)<br />

, is concave <strong>in</strong> the <strong>in</strong>itial production<br />

quantity q 1 . The optimal total production quantity is<br />

c c<br />

⎧q1,<br />

if ε ε(<br />

q1);<br />

c ⎪<br />

q = ⎨ −1<br />

⎛ p− c2+ s ⎞<br />

⎪F<br />

⎜ ε ⎟,<br />

otherwise,<br />

⎩ ⎝ p− v+ s ⎠<br />

c ⎧ c p− c2+ s⎫<br />

where ε ( q1) = sup ⎨ε : F( q1<br />

| ε)<br />

⎬ and the<br />

⎩ p− v+ s ⎭<br />

first-period optimal production quantity is implied <strong>by</strong><br />

c<br />

ε ( q1<br />

)<br />

c<br />

2 1 ∫ 0<br />

2 1<br />

c − c + [ p− c + s−( p− v+ s) F( q | ε)]d G(<br />

ε)<br />

= 0.<br />


WANG Xiaolong (王小龙) et al:<strong>Coord<strong>in</strong>ation</strong> <strong>by</strong> <strong>Option</strong> <strong>Contracts</strong> <strong>in</strong> a <strong>Retailer</strong>-<strong>Led</strong> …<br />

c<br />

Proof First, prove the concavity of Π ( q1)<br />

. This<br />

c c<br />

only needs to prove that Π ( ε, q1) = max π ( ε,<br />

q)<br />

is<br />

qq1 concave <strong>in</strong> q 1 .<br />

Note that<br />

c * *<br />

⎧⎪ π ( ε,<br />

q ), if q c<br />

1 q ;<br />

Π ( ε,<br />

q1)<br />

= ⎨ c<br />

⎪⎩ π ( ε,<br />

q1),<br />

otherwise,<br />

*<br />

c c<br />

where q maximizes π ( ε , q).<br />

Now π ( ε , q)<br />

is obvi-<br />

2 c<br />

∂ π ( ε,<br />

q)<br />

ously concave <strong>in</strong> q because =−( p− v+ s)<br />

⋅<br />

2<br />

∂q<br />

c<br />

f( q| ε ) 0. Therefore, Π ( q1)<br />

is also concave <strong>in</strong> q 1 .<br />

The optimal total production quantity follows immediately<br />

from the fact that the second-period problem<br />

is a simple newsvendor problem <strong>with</strong> <strong>in</strong>itial <strong>in</strong>ventory<br />

q 1.<br />

Intuitively this quantity is positively related to the<br />

market signal. When the signal is not strong enough,<br />

the central planner will not produce a second sum. This<br />

idea is exhibited <strong>in</strong> Lemma 1 <strong>with</strong> a threshold value for<br />

c ⎧ c p− c2+ s<br />

the market signal: ε ( q1 ) = sup<br />

⎫<br />

⎨ε : Fq ( 1 | ε ) ⎬ .<br />

⎩<br />

p− v+ s ⎭<br />

This is the strongest signal that will still result <strong>in</strong> no<br />

second-period production. With this notation, suppose<br />

that the central planner’s expected profit is maximized<br />

at q , and then rewrite the problem as<br />

c<br />

1<br />

c c<br />

1 2 1<br />

c<br />

1<br />

c<br />

ε ( q1<br />

)<br />

∫ 0<br />

c c<br />

1<br />

∞<br />

c c<br />

∫ π ( ε, q )d G(<br />

ε),<br />

c<br />

ε ( q1<br />

)<br />

Π ( q ) = ( c − c ) q + π ( ε, q )d G(<br />

ε)<br />

+<br />

c 1 p c2 s<br />

where q F<br />

p v s ε<br />

− ⎛ − + ⎞<br />

= ⎜ ⎟ . Because<br />

⎝ − + ⎠<br />

c<br />

Π ( q1)<br />

is<br />

concave <strong>in</strong> q 1 , the first order condition works, i.e., an<br />

optimal first-period production quantity is implied <strong>by</strong><br />

c<br />

ε ( q1<br />

)<br />

c<br />

2 1<br />

0<br />

2 1<br />

∫<br />

c − c + [ p− c + s−( p− v+ s) F( q | ε)]d G(<br />

ε)<br />

= 0.<br />

□<br />

c<br />

Lemma 1 <strong>in</strong>troduces the concept of ε ( q1<br />

) which<br />

partitions the market signals <strong>in</strong>to two sets. If<br />

c<br />

ε ε(<br />

q1<br />

) , then the second-period production is not<br />

necessarily positive. Otherwise, it is exactly positive.<br />

1.2 Decentralized system performance <strong>with</strong>out<br />

coord<strong>in</strong>ation<br />

Here model the decentralized system as a Stackelberg<br />

game. The retailer sets orders <strong>in</strong> each period as mentioned<br />

above. The supplier <strong>in</strong> turn arranges production<br />

573<br />

based on the orders.<br />

Let d i denote the retailer’s order <strong>in</strong> period i. The<br />

problem is to choose d 1 and d2 = d − d1<br />

to maximize<br />

the expected profit. The problem structure is analogous<br />

to that <strong>in</strong> the centralized system.<br />

∞<br />

max ΠR( d1) =− wd 1 1+∫ ΠR( ε, d1, d2) d G(<br />

ε),<br />

d10<br />

0<br />

where ΠR( ε, d1, d2) = max πR( ε,<br />

d1, d2)<br />

and<br />

d20<br />

πR( ε , d1, d2) = − w2d2 + pEm<strong>in</strong>{ D, d1+ d2}<br />

+<br />

vd ( 1+ d2 − Em<strong>in</strong>{ Dd , 1+ d2})<br />

−<br />

sE[ D − m<strong>in</strong>{ D, d + d }].<br />

Rearrang<strong>in</strong>g,<br />

1 2<br />

max ΠR( d1) = ( w2 − w1) d1+∫ΠR( ε, d1) d G(<br />

ε)<br />

d10<br />

0<br />

where ΠR( ε, d1) = max πR( ε,<br />

d)<br />

and<br />

<br />

R 2<br />

d d1<br />

π ( ε, d) = ( p− w + s) d−( p− v+ s) F( x| ε)d<br />

x−sED. * *<br />

Let ( d1, d ) denote the optimal solution to the retailer’s<br />

problem. The solution structure is given <strong>in</strong><br />

Lemma 2.<br />

Lemma 2 The argument of the retailer’s first period<br />

problem, Π R( d1)<br />

, is concave <strong>in</strong> the <strong>in</strong>itial order<br />

quantity d 1 . The optimal total order is<br />

* *<br />

⎧d1,<br />

if ε ε(<br />

d1);<br />

* ⎪<br />

d = ⎨ −1<br />

⎛ p− w2+ s ⎞<br />

⎪F<br />

⎜ ε ⎟,<br />

otherwise,<br />

⎩ ⎝ p− v+ s ⎠<br />

where<br />

* ⎧ * p− w2+ s⎫<br />

ε( d1) = sup ⎨ε : F( d1<br />

| ε)<br />

⎬ and the first-<br />

⎩ p− v+ s ⎭<br />

period optimal order quantity is implied <strong>by</strong><br />

*<br />

ε ( d1<br />

)<br />

*<br />

2 1<br />

0<br />

2 1<br />

∫ <br />

w − w + [ p− w + s−( p− v+ s) F( d | ε)]d G(<br />

ε)<br />

= 0.<br />

The proof is quite similar to that of Lemma 1. Here,<br />

*<br />

ε ( d1<br />

) also partitions the market signals <strong>in</strong>to two sets.<br />

*<br />

If ε ε(<br />

d1<br />

) , then the retailer will not set a second<br />

order. Note the differences from the notations def<strong>in</strong>ed<br />

<strong>in</strong> Section 1.1. Furthermore, for a given x, ε ( x) < ε ( x)<br />

,<br />

which co<strong>in</strong>cides <strong>with</strong> the phenomenon that double<br />

marg<strong>in</strong>alization makes the buyer conservative. Compared<br />

<strong>with</strong> the central planner, given the same <strong>in</strong>ventory<br />

level, the buyer <strong>in</strong> the decentralized system needs<br />

a stronger market signal, which <strong>in</strong>duces a more<br />

optimistic demand estimate, to build up to that level.<br />

The supplier’s problem is to choose production<br />

∞<br />

∫<br />

0<br />

d


574<br />

quantities ( q1, q 2)<br />

that maximize his own expected<br />

profit, subject to the retailer’s order<strong>in</strong>g behavior. His<br />

decision on the second-stage production is relatively<br />

straightforward, i.e.,<br />

* * *<br />

q2 = max{ d − q1,0}<br />

.<br />

The difficulty lies <strong>in</strong> his decision on q 1 . Obviously<br />

he should at least produce d 1 , but should he produce<br />

more? The answer depends on his estimate on the likelihood<br />

of the retailer’s second order. For such a decision<br />

his problem is<br />

ε ( d1)<br />

max Π ( q ) = wd − cq + v( q − d )d G(<br />

ε)<br />

+<br />

q1d1 S 1 1 1 1 1<br />

0<br />

1 1<br />

ε ( q1<br />

)<br />

∫ ε ( d1)<br />

∞<br />

wd 2 2 vq1 d G<br />

∫<br />

ε ( q1<br />

)<br />

∫<br />

[ + ( − )]d ( ε ) +<br />

[ wd −c( d−q)]d G(<br />

ε ).<br />

2 2 2 1<br />

The solution to this problem is given <strong>by</strong> Lemma 3.<br />

Lemma 3 Given the retailer’s order<strong>in</strong>g behavior,<br />

the supplier will maximize his expected profit <strong>by</strong><br />

sett<strong>in</strong>g<br />

* *<br />

q1 d1 q′ 1<br />

* * *<br />

= max{ , } and q2 = max{ d − q1,0},<br />

where 1 q′ is implied <strong>by</strong> c2 − c1<br />

G( ε ( q′<br />

1))<br />

= .<br />

c2−v Because this problem has a newsvendor structure,<br />

the first order condition works. The proof is straightforward<br />

so the details are omitted. These problems<br />

have reviewed the benchmark and decentralized uncoord<strong>in</strong>ated<br />

system performance. Although the precise<br />

relationships between * *<br />

d 1 , q 1,<br />

and c<br />

q 1 depend<strong>in</strong>g on<br />

factors like the ratio of the two production costs, the<br />

ratio of the two wholesale prices, and the estimate of<br />

the market signal cannot be determ<strong>in</strong>ed, the channel is<br />

certa<strong>in</strong>ly not coord<strong>in</strong>ated. This is implied <strong>by</strong> Lemma 2<br />

which states that for a given market signal the retailer’s<br />

total order quantity is less than the optimal total <strong>in</strong>ventory<br />

required <strong>in</strong> a centralized system. The next section<br />

focuses on how an option mechanism coord<strong>in</strong>ates<br />

the supply cha<strong>in</strong> and realizes profit allocation.<br />

2 Channel <strong>Coord<strong>in</strong>ation</strong><br />

To coord<strong>in</strong>ate the channel, the retailer must choose the<br />

proper contract parameters so that the supplier’s production<br />

quantities <strong>in</strong> each period are equal to the opti-<br />

* c<br />

mal solutions <strong>in</strong> the centralized system, i.e., q = q<br />

and<br />

1 1<br />

* c<br />

q = q . To do so the retailer must give <strong>in</strong>centives<br />

Ts<strong>in</strong>ghua Science and Technology, August 2008, 13(4): 570-580<br />

to the supplier to overproduce <strong>in</strong> each period. Recall<br />

that q1d1 and q2 max{ d − q1,0}<br />

. Thus d 1 and<br />

max{ d − q1,0}<br />

are the m<strong>in</strong>imum production number<br />

for the supplier’s production <strong>in</strong> each period. The coord<strong>in</strong>at<strong>in</strong>g<br />

contract is def<strong>in</strong>ed as follows: The option contract<br />

has three parameters of two option prices o 1 and<br />

o 2 and one exercise price e . The option prices are<br />

the unit compensation for the supplier’s production<br />

quantities beyond the m<strong>in</strong>imums <strong>in</strong> each period. When<br />

the realized demand exceeds the retailer’s total order,<br />

he will purchase the supplier’s excess <strong>in</strong>ventory, if any,<br />

to meet the market demand at a unit price e .<br />

The sequence of events is summarized as follows:<br />

(1) At the beg<strong>in</strong>n<strong>in</strong>g of period 1, the retailer sets a<br />

firm order d 1 . The retailer also announces an option<br />

contract <strong>with</strong> option prices o 1 and o 2 and exercise<br />

price e. The supplier, <strong>in</strong> turn, decides to produce<br />

q1 d1<br />

of goods <strong>in</strong> the cheap mode which will produce<br />

an <strong>in</strong>come of o1( q1− d1)<br />

.<br />

(2) At the beg<strong>in</strong>n<strong>in</strong>g of period 2, after the market<br />

signal is observed, the retailer orders d 2 . The supplier<br />

selects a second-period production, q2 max{ d− q1,0}<br />

,<br />

that <strong>in</strong>volves the expensive mode and produces an <strong>in</strong>come<br />

of o2( q2 −max{ d − q1,0})<br />

.<br />

(3) At the beg<strong>in</strong>n<strong>in</strong>g of the sell<strong>in</strong>g season, after the<br />

demand is realized, the retailer purchases additionally<br />

m= m<strong>in</strong>{max{ D−d,0}, q− d}<br />

number of goods at<br />

unit price e . The supplier delivers all that the retailer<br />

wants. The problem assumes that the supplier is obligated<br />

to fill the retailer’s entire order and can neither<br />

sell directly to the f<strong>in</strong>al customer (i.e., the retailer’s<br />

customer) nor supply <strong>in</strong>ventory <strong>in</strong> excess of the retailer’s<br />

total order quantity.<br />

With the option contract, the supplier’s problem<br />

becomes<br />

max Π ( q ) = −( c − o ) q + ( w − o ) d +<br />

q1d1 S 1 1 1 1 1 1 1<br />

∫<br />

0<br />

∞<br />

Π ( ε, q , q )d G(<br />

ε),<br />

S 1 2<br />

where ΠS( ε, q1, q2) = max πS( ε,<br />

q1, q2)<br />

and<br />

q2max{ d−q1,0} π ( ε , q , q ) = − c q + o ( q −max{ d − q ,0}) +<br />

S 1 2 2 2 2 2 1<br />

d<br />

wd + vq ( + q − d)d F( x|<br />

ε ) +<br />

∫<br />

2 2<br />

0<br />

1 2<br />

q1+ q2<br />

d<br />

∫<br />

[( ex− d) + vq ( − x)]d Fx ( | ε ) +


WANG Xiaolong (王小龙) et al:<strong>Coord<strong>in</strong>ation</strong> <strong>by</strong> <strong>Option</strong> <strong>Contracts</strong> <strong>in</strong> a <strong>Retailer</strong>-<strong>Led</strong> …<br />

∞<br />

∫ eq ( 1+ q2 −d)d<br />

F( x| ε ).<br />

q1+ q2 Here, ΠS( ε , q1, q2)<br />

represents his second-stage problem.<br />

After mak<strong>in</strong>g the supplier’s second-stage problem<br />

<strong>in</strong>dependent of q 1 and after some algebra, the sup-<br />

plier’s problem is<br />

max Π ( q ) = ( c −c − o + o ) q + ( w − o ) d +<br />

q1d1 S 1 2 1 2 1 1 1 1 1<br />

∫<br />

0<br />

∞<br />

Π ( ε, q )d G(<br />

ε),<br />

S 1<br />

where ΠS( ε, q1) = max πS( ε,<br />

q)<br />

and<br />

qmax{ d, q1}<br />

π S( ε,<br />

q) = ( e− c2+ o2) q+ w2d2−ed−o2max{ d−q1,0} −<br />

q<br />

( e−v) ∫ F( x| ε )d x .<br />

d<br />

N N<br />

Let ( q , q 1 ) denote the solution to this problem, and<br />

then we have Proposition 1.<br />

Proposition 1 Given an option contract specified<br />

<strong>by</strong> the retailer, the supplier’s optimal total production<br />

quantity is<br />

N ⎧ N −1⎛e−<br />

c2 + o2<br />

⎞⎫<br />

q = max ⎨q1, F ⎜ ε ⎟⎬<br />

⎩ ⎝ e−v ⎠⎭<br />

.<br />

The proof is relatively straight forward s<strong>in</strong>ce the supplier’s<br />

problem follows a newsvendor structure.<br />

N<br />

Proposition 1 shows clearly that to coord<strong>in</strong>ate q =<br />

c<br />

e− c2 + o2 p− c2 + s<br />

q , it is necessary to set = , i.e.,<br />

e−v p− v+ s<br />

p− v+ s<br />

e= p+ s− o2<br />

(5)<br />

c2−v Based on Eq. (5), e v e+ o > c to make<br />

> and 2 2<br />

−1⎛ e− c2+ o2<br />

⎞<br />

F ⎜ | ε ⎟ mean<strong>in</strong>gful. Economically these two<br />

⎝ e−v ⎠<br />

conditions prevent the supplier from suffer<strong>in</strong>g losses.<br />

e+ o2is<br />

all the revenue that the supplier can get from<br />

one unit of product if it can be sold after the market<br />

demand is realized. If e+ o2 < c2,<br />

the supplier has no<br />

<strong>in</strong>centive to overproduce <strong>in</strong> the second period, and thus<br />

channel coord<strong>in</strong>ation is never reached. S<strong>in</strong>ce e= p+ s−<br />

p− v+ s<br />

o2<br />

, the two conditions are equivalent <strong>with</strong><br />

c2−v o2 < ( c2 − v)<br />

. Therefore, o2 < ( c2 − v)<br />

from now on.<br />

Next, consider the retailer’s problem. In all the fol-<br />

p− v+ s<br />

low<strong>in</strong>g parts, assume that e= p+ s− o2holds<br />

c2−v except where specified.<br />

The retailer’s problem is<br />

575<br />

∞<br />

max ΠR( d1) =−o1( q1−d1) − wd 1 1+∫ ΠR( ε, d1) d G(<br />

ε),<br />

d10<br />

0<br />

where ΠR( ε, d1) = max πR( ε,<br />

d)<br />

and<br />

dd1 π R( ε , d) = −w2d2 −o2( q2 −max{ d − q1,0})<br />

+<br />

pE m<strong>in</strong>{ D, q} + v( d −Em<strong>in</strong>{ D, d})<br />

−<br />

eE m<strong>in</strong>{max{ D −d,0}, q −d} −<br />

sED ( − Em<strong>in</strong>{ Dq , }).<br />

Rearrang<strong>in</strong>g,<br />

max Π ( d ) = ( o − o ) q + ( w − w + o ) d +<br />

where<br />

d10<br />

R 1 2 1 1 2 1 1 1<br />

∫<br />

∞<br />

Π<br />

0<br />

R( ε, d1)d G(<br />

ε),<br />

ΠR( ε, d1) = max πR( ε,<br />

d)<br />

and<br />

dd R 2<br />

1<br />

π ( ε, d) = ( e−w ) d −( e− v) F( x| ε)dx+<br />

o max{ d − q ,0} + ( p+ s−e−o ) q−<br />

2 1 2<br />

∫<br />

( p+ s−e) F( x| ε )dx−sED.<br />

0<br />

q<br />

N N<br />

Let ( 1 , d d ) denote the optimal solution to the retailer’s<br />

problem. Proposition 2 describes the retailer’s<br />

order<strong>in</strong>g behavior.<br />

e−w2 e− w2 + o2<br />

Proposition 2 Denote α = , β = ,<br />

e−v e−v N −1<br />

N<br />

and γ ( ε) = max{ d1, F (max{ α,0}| ε)}.<br />

Def<strong>in</strong>e ˆ( ε d1<br />

) =<br />

N<br />

N<br />

sup{ ε: Fd ( 1 | ε) max{ α,0}}<br />

and allow ˆ( ε d1<br />

) = +∞<br />

if ε max is <strong>in</strong>f<strong>in</strong>ite. For a given option contract, the<br />

optimal order<strong>in</strong>g behavior for the retailer has the follow<strong>in</strong>g<br />

structure:<br />

(1) After the market signal ε has appeared, the optimal<br />

total order quantity is determ<strong>in</strong>ed as follows:<br />

(i) If Fq ( 1 | ε ) β ,<br />

N<br />

d = γ ( ε ) ;<br />

(ii) Otherwise,<br />

−1<br />

⎧ γε ( ), if πR( ε, F (max{ α,0}| ε))<br />

><br />

⎪ N −1<br />

d = ⎨ πR( ε, F ( β | ε));<br />

⎪⎩ −1<br />

F ( β | ε),<br />

otherwise.<br />

(2) The optimal first-period order quantity is given<br />

<strong>by</strong><br />

N ˆ( ε d1<br />

)<br />

N<br />

2 1 1<br />

0<br />

2 1<br />

∫<br />

w − w + o + [ e−w −( e− v) F( d | ε)]d G(<br />

ε)<br />

= 0<br />

when e+ o1 > w1.<br />

Otherwise, N<br />

d1 ≡ 0 .<br />

Corollary For some second-period option price(s)<br />

−1<br />

o 2 such that β > 0 and set A= { ε : πR( ε, F (max{ α,<br />

−1<br />

0}| ε )) π ( ε, F ( β | ε))}<br />

is not empty, there exists a<br />

R<br />

∫<br />

0<br />

d


576<br />

signal T ( 2)<br />

o ε such that<br />

N −1<br />

only if T ( 2)<br />

o ε ε<br />

greater than 1 q and T ( 2) max o ε ε<br />

d = F ( β | ε)<br />

> q1if<br />

and<br />

<br />

N<br />

. Otherwise, d will never be<br />

= for such cases.<br />

Proof First, prove part (1). Omitt<strong>in</strong>g all the irrelevant<br />

parts, the retailer’s second-period problem can be<br />

rewritten as<br />

d<br />

⎧( e−w2) d −( e−v) F( x| ε )d x,<br />

⎪ ∫ 0<br />

⎪ if dq1; max πR( ε,<br />

d)<br />

= ⎨<br />

dd d<br />

1 ⎪( e− w2 + o2) d −( e−v) F( x| ε )d x,<br />

0<br />

⎪<br />

∫<br />

⎪⎩ otherwise.<br />

Then, the unconditional optimal N<br />

d may not be<br />

unique but can be either<br />

−1<br />

F (max{ α,0}| ε ) or<br />

−1<br />

N −1<br />

F (max{ β,0}| ε ) or both. However, if d = F ( β | ε ) ,<br />

−1<br />

there must be F ( β | ε)<br />

> q1,<br />

i.e., 0 < Fq ( 1 | ε ) < β , so<br />

that max{ d − q1,0} = d − q1<br />

makes sense. This completes<br />

the proof of part (i) when Fq ( 1 | ε ) β and<br />

N<br />

d = γ ( ε ) .<br />

Now consider 0 < Fq ( 1 | ε ) < β . The criterion is sim-<br />

N<br />

ple but clear. Whether d = γ ( ε ) or N 1 −<br />

d = F ( β | ε )<br />

depends on which can br<strong>in</strong>g more expected profit, i.e.,<br />

−1<br />

⎧ γε ( ), if πR( ε, F (max{ α,0}| ε))<br />

><br />

⎪ N −1<br />

d = ⎨ πR( ε, F ( β | ε));<br />

⎪⎩ −1<br />

F ( β | ε),<br />

otherwise.<br />

This completes the proof of part (1) <strong>in</strong> Proposition 2.<br />

Next, show the l<strong>in</strong>k between the retailer’s second-period<br />

expected profit and the market signal to<br />

estimate the optimal total order from the beg<strong>in</strong>n<strong>in</strong>g of<br />

period 1 based on the market signal conditions, which<br />

is the core idea of the corollary.<br />

−1<br />

When β > 0, let π ( ε, F (max{ α,0}| ε)) = π ( ε,<br />

−1<br />

F ( β | ε )) . Note that<br />

R R<br />

d ≡ d when β 0 because<br />

N N<br />

1<br />

β α . Suppose that there exists a solution T ( 2)<br />

o ε ,<br />

and then for T ( 2)<br />

o ε the retailer is <strong>in</strong>different towards<br />

the choice of<br />

1 −<br />

−1<br />

F (max{ α,0}| ε ) and F ( β | ε ) . Obviously,<br />

for stronger signals T ( 2),<br />

o ε ε ><br />

N −1<br />

d = F ( β | ε ) ><br />

q 1 . Otherwise, if such a T ( 2)<br />

o ε does not exist, or<br />

−1 −1<br />

equivalently, πR( ε, F (max{ α,0}| ε)) > πR( ε,<br />

F ( β | ε ))<br />

for any signals, def<strong>in</strong>e T ( 2) max o<br />

N<br />

ε = ε . Thus, d =<br />

−1<br />

F ( βε | ) > q1is<br />

impossible because the market signal<br />

cannot be greater than its maximum.<br />

Ts<strong>in</strong>ghua Science and Technology, August 2008, 13(4): 570-580<br />

Next consider part (2) which depicts the first-period<br />

order<strong>in</strong>g behavior. If o2=0, then the retailer’s second-period<br />

problem is given <strong>by</strong> the follow<strong>in</strong>g:<br />

For any α , similar to the reason<strong>in</strong>g <strong>in</strong> proof of<br />

Lemma 1,<br />

d is implied <strong>by</strong><br />

2 1 1<br />

N<br />

1<br />

N ˆ( ε d1<br />

)<br />

∫ 0<br />

2<br />

N<br />

1<br />

α , i.e., e w2<br />

N<br />

1<br />

w − w + o + [ e−w −( e− v) F( d | ε)]d G(<br />

ε)<br />

= 0.<br />

Particularly, when 0 , <strong>in</strong>equality<br />

Fd (<br />

N<br />

| ε ) 0 holds for any signal ε , then ˆ( ε d1<br />

)<br />

must be equal to ε max , which is <strong>in</strong>f<strong>in</strong>ite <strong>in</strong> our model.<br />

Under such cases, we can rewrite the above equation<br />

as<br />

+∞<br />

N<br />

2 1 1<br />

0<br />

2 1<br />

∫<br />

w − w + o + [ e−w −( e− v) F( d | ε)]d G(<br />

ε)<br />

= 0.<br />

Rearrang<strong>in</strong>g,<br />

+∞<br />

N<br />

1 1<br />

0<br />

1<br />

∫<br />

e− w + o −( e− v) F( d | ε)d G(<br />

ε)<br />

= 0.<br />

Note that 0<br />

∞<br />

N<br />

∫ Fd (<br />

0<br />

1 | ε)d G(<br />

ε)<br />

1<br />

special cases when e+ o1 w1<br />

, then<br />

. There are some<br />

N N<br />

d = d1=<br />

0.<br />

Note that e− w1+ o1 e−v cannot succeed because<br />

it requires o1 w1−v, which would <strong>in</strong>centivize the<br />

supplier to produce <strong>in</strong>f<strong>in</strong>itely <strong>in</strong> period 1 and, thus, is<br />

ridiculous.<br />

In summary, the optimal first-period order quantity<br />

d is given <strong>by</strong><br />

N<br />

1<br />

N ˆ( ε d1<br />

)<br />

N<br />

2 1 1<br />

0<br />

2 1<br />

∫<br />

w − w + o + [ e−w −( e− v) F( d | ε)]d G(<br />

ε)<br />

= 0<br />

N<br />

when e+ o1 > w1.<br />

Otherwise, d1 ≡ 0 . □<br />

Proposition 2 shows that the retailer’s order<strong>in</strong>g behavior<br />

is closely l<strong>in</strong>ked to the option price o 2 and the<br />

market signal ε . Different option prices result <strong>in</strong> different<br />

order<strong>in</strong>g behavior. Under some circumstances,<br />

the optimal total order may not even be unique. Signals<br />

such as ˆε and ε T can be used to categorize the retailer’s<br />

orders. These mark<strong>in</strong>g signals are important <strong>in</strong><br />

models concerned <strong>with</strong> demand <strong>in</strong>formation update.<br />

The conclusion <strong>in</strong> the corollary is mean<strong>in</strong>gful for disclos<strong>in</strong>g<br />

the relationship between the retailer’s second-period<br />

profit curve and the market signal that optimal<br />

total order more than q 1 is possible if and only<br />

if T ( 2) max o ε ε < exists. T ( 2)<br />

o ε must be always greater<br />

than ε ( q1)<br />

for a given q 1 because the conservative<br />

retailer has <strong>in</strong>centive to choose an optimal total order<br />

greater than q 1 <strong>with</strong> T ( 2)<br />

o ε that he will never do


WANG Xiaolong (王小龙) et al:<strong>Coord<strong>in</strong>ation</strong> <strong>by</strong> <strong>Option</strong> <strong>Contracts</strong> <strong>in</strong> a <strong>Retailer</strong>-<strong>Led</strong> …<br />

N * c<br />

<strong>with</strong> ε ( q1)<br />

because d d < q . A f<strong>in</strong>al <strong>in</strong>terest<strong>in</strong>g<br />

N<br />

issue is that d is not cont<strong>in</strong>uous <strong>in</strong> ε for a given<br />

contract and q 1 is a k<strong>in</strong>k po<strong>in</strong>t of the retailer’s expected<br />

profit function. No matter what the realized<br />

market signal is, the optimal total order never equals<br />

1 . q This may occur <strong>with</strong> the mechanism which sets the<br />

second-period compensation <strong>in</strong>come to be max{ d− q1,<br />

0}.Thus, the derivative of π R ( ε,<br />

d)<br />

will never be equal<br />

to zero at q 1 .<br />

Now, consider the supplier’s problem. Know<strong>in</strong>g the<br />

retailer’s order<strong>in</strong>g behavior, the supplier will set his<br />

first-period production quantity accord<strong>in</strong>gly. When the<br />

retailer knows the supplier’s first-period production<br />

decision (which is related to the retailer’s order<strong>in</strong>g be-<br />

c<br />

havior), the retailer coord<strong>in</strong>ates it to be equal to q 1 <strong>by</strong><br />

modify<strong>in</strong>g o 1 . This is the standard Stackelberg game.<br />

Comb<strong>in</strong><strong>in</strong>g Eq. (5), the overall f<strong>in</strong>d<strong>in</strong>gs for coord<strong>in</strong>ation<br />

are summarized <strong>in</strong> Proposition 3.<br />

e−w2 Proposition 3 Denote α = and β =<br />

e−v e− w2 + o2<br />

. To coord<strong>in</strong>ate the channel, ensure that<br />

e−v o2 < c2 − v and ma<strong>in</strong>ta<strong>in</strong> some specific functional relationships<br />

between the contract parameters of 1 o , o 2 ,<br />

and e .<br />

p− v+ s<br />

e= p+ s− o2and<br />

c2−v ⎡ c2 − c1⎤<br />

o1 =<br />

⎢<br />

G( εT<br />

( o2)) − o2<br />

c2−v ⎥ ,<br />

⎣ ⎦<br />

where T ( 2)<br />

o ε is determ<strong>in</strong>ed <strong>by</strong><br />

−1<br />

πR( εT( o2), F (max{ α,0}| εT(<br />

o2)))<br />

=<br />

−1<br />

πR( εT( o2), F ( β | εT(<br />

o2)))<br />

when β > 0 and<br />

−1<br />

{ εT( o2) : πR( εT( o2), F (max{ α,0}| εT( o2))) = πR( εT(<br />

o2),<br />

−1<br />

F ( β | εT(<br />

o2)))}<br />

≠∅. Otherwise, def<strong>in</strong>e T ( 2) max o ε = ε .<br />

Proof The necessity of equation e= p+ s−<br />

p− v+ s<br />

o2<br />

was proved <strong>in</strong> Proposition 1. Next, show<br />

c2−v that coord<strong>in</strong>ation still needs another condition,<br />

⎡ c2 − c1⎤<br />

o1 =<br />

⎢<br />

G( εT<br />

( o2)) − o2<br />

c2−v ⎥ .<br />

⎣ ⎦<br />

Recall that <strong>in</strong> this model coord<strong>in</strong>ation means<br />

N c<br />

N c<br />

q = q and q1 = q1<br />

. Based on the corollary <strong>in</strong><br />

Proposition 2, write the first-order condition for the<br />

optimal<br />

N<br />

q 1 as<br />

2 1 2 1<br />

ε ( q1<br />

)<br />

0<br />

2 2 1<br />

∫<br />

577<br />

c − c − o + o + [ e− c + o −( e− v) F( q | ε )]d G(<br />

ε ) +<br />

∫<br />

∞<br />

o<br />

(<br />

2d<br />

G(<br />

ε ) = 0.<br />

εT<br />

o2 )<br />

p− v+ s<br />

Substitute e= p+ s− o2<strong>in</strong>to<br />

this equation<br />

c2−v and rearrange<br />

⎛ o ( 1)<br />

2 ⎞ ε q<br />

c2 − c1− o2 + o1+ ⎜1 − [ p− c<br />

0<br />

2 + s−<br />

c2−v ⎟<br />

⎝ ⎠ ∫<br />

∞<br />

( p− v+ s) F( q | ε)]d G( ε) + o d G(<br />

ε)<br />

= 0.<br />

Compar<strong>in</strong>g <strong>with</strong><br />

1<br />

T ( 2 )<br />

2<br />

o ε<br />

c<br />

ε ( q1<br />

)<br />

2 1<br />

0<br />

2<br />

∫<br />

∫<br />

c − c + [ p− c + s−( p− v+<br />

c<br />

N c<br />

sFq ) ( 1 | ε)]d G(<br />

ε ) = 0,<br />

to let q1 = q1<br />

, set o 1 so that<br />

the follow<strong>in</strong>g equality holds:<br />

c o ε ( q1<br />

)<br />

2<br />

− o2+ o1− [ p− c<br />

0<br />

2+ s−( p− v+ s) F( q1| ε )]d G(<br />

ε ) +<br />

c −v∫ Note that<br />

2<br />

∫<br />

∞<br />

o d G(<br />

ε ) = 0.<br />

o<br />

ε (<br />

2<br />

T 2 )<br />

c<br />

ε ( q1<br />

)<br />

c<br />

2− 1+ − 2+ − − + 1<br />

0<br />

c c ∫ [ p c s ( p v s) F( q | ε )]d G(<br />

ε ) = 0<br />

and rearrang<strong>in</strong>g gives<br />

⎡ c2 − c1⎤<br />

o1 =<br />

⎢<br />

G( εT<br />

( o2)) − o2<br />

c2−v ⎥ . □<br />

⎣ ⎦<br />

There are the pric<strong>in</strong>g equations which assure channel<br />

profit maximization. However, an important issue<br />

to be considered before implement<strong>in</strong>g any new pric<strong>in</strong>g<br />

scheme is whether the scheme is Pareto improv<strong>in</strong>g<br />

<strong>with</strong> respect to an exist<strong>in</strong>g policy. In other words, will<br />

both the retailer and the supplier be better off <strong>with</strong> the<br />

proposed policy? To shed light on this question, exam<strong>in</strong>e<br />

how the retailer’s and supplier’s expected profits<br />

change as the contract parameters vary.<br />

Unfortunately, due to the model complexity, closed<br />

form solutions are not possible for some of the key<br />

variables; thus, quantitative comparisons are almost<br />

impossible. We cannot even prove that the monotonicity<br />

of each side’s expected profit function as o 2 varies<br />

because it is related to the specific forms of the distribution<br />

functions.<br />

3 Example<br />

Suppose that F and G are both uniform distribution<br />

functions <strong>with</strong> F~ U( a, b ) and G~ U(0, t ) . The critical


578<br />

factor <strong>in</strong> models consider<strong>in</strong>g demand update is how the<br />

market signal updates the demand <strong>in</strong>formation, i.e.,<br />

how to def<strong>in</strong>e F( x| ε ) . Make the follow<strong>in</strong>g assumptions<br />

related to the market signal:<br />

(1) The market signal does not change the demand<br />

distribution pattern, i.e., F( x| ε ) is still uniform.<br />

(2) The market signal only affects the mean of the<br />

demand distribution, not the variance. As the signal<br />

becomes stronger the mean shifts up accord<strong>in</strong>gly. For<br />

uniform distributions, this implies that if F( x| ε ) ~<br />

U( f1( ε ), f2(<br />

ε )), then f2( ε) − f1( ε)<br />

≡b−a and f1 ( ε ) and<br />

f2 ( ε ) are both <strong>in</strong>creas<strong>in</strong>g <strong>in</strong> ε .<br />

(3) A market signal as mathematically expected<br />

ma<strong>in</strong>ta<strong>in</strong>s the orig<strong>in</strong>al demand distribution function.<br />

That is, if ε = 0.5t , then F( x| ε ) = F( x)<br />

. Moreover,<br />

assume<br />

F( x| ε )~ U( a−4(0.5 t−ε), b−4(0.5 t−<br />

ε))<br />

.<br />

To avoid un<strong>in</strong>terest<strong>in</strong>g cases, def<strong>in</strong>e a> 2t.<br />

Consider the follow<strong>in</strong>g numerical example. Suppose<br />

the orig<strong>in</strong>al demand distribution is uniform <strong>with</strong><br />

a = 1000 and b = 2000 . The market signal is uniformly<br />

distributed between zero and t = 150.<br />

The cost<br />

parameters are c1= 50, c2= 65, w1= 60, w2= 80, v=<br />

20,<br />

s = 30,<br />

and p = 100 .<br />

For any o 2 ,<br />

p− v+ s<br />

e= p+ s− o .<br />

(1) Calculate e accord<strong>in</strong>g to<br />

(2) Solve for T ( 2)<br />

c2−v 2<br />

o ε .<br />

If β > 0 , εT( o2)<br />

= m<strong>in</strong>{150, ε′ T},<br />

where T<br />

ε′ is determ<strong>in</strong>ed<br />

<strong>by</strong><br />

−1 −1<br />

π R( ε′ T, F (max{ α,0}| ε′ T)) = πR( ε′ T, F ( β | ε′<br />

T))<br />

;<br />

If β 0 , T ( 2)<br />

150. o ε =<br />

⎡<br />

(3) Calculate o 1 accord<strong>in</strong>g to o1 = ⎢G(<br />

εT<br />

( o2))<br />

−<br />

⎣<br />

c2 − c1 ⎤<br />

o2<br />

c2−v ⎥ .<br />

⎦<br />

N<br />

(4) Solve for d 1 .<br />

N<br />

If e+ o1 w1,<br />

d1 ≡ 0 ;<br />

If e+ o1 > w1,<br />

substitute<br />

N<br />

N d1−[ a− 2 t+ ( b−a)max{ α,0}]<br />

ˆ( ε d1<br />

) = (which is<br />

4<br />

N N<br />

derived from Fd ( | ˆ ε( d )) = max{ α,0}<br />

) <strong>in</strong>to<br />

1 1<br />

N ˆ( ε d1<br />

)<br />

N<br />

2 1 1 ∫ 0<br />

2 1<br />

w − w + o + [ e−w −( e− v) F( d | ε)]d G(<br />

ε)<br />

= 0.<br />

Ts<strong>in</strong>ghua Science and Technology, August 2008, 13(4): 570-580<br />

If ˆ( ε d ) < 150,<br />

the results can be directly used.<br />

If ˆ( ε d ) 150,<br />

then<br />

N<br />

1<br />

N<br />

1<br />

N<br />

d 1 is implied <strong>by</strong><br />

150<br />

N<br />

2 1 1<br />

0<br />

2 1<br />

∫<br />

w − w + o + [ e−w −( e− v) F( d | ε)]d G(<br />

ε)<br />

= 0.<br />

(5) Calculate each party’s expected profit.<br />

Figure 1 shows how the retailer’s first-period order<br />

changes as o 2 varies. As expected, the retailer sets<br />

N<br />

d ≡ 0 when 1 1 , e o w<br />

o <br />

1<br />

p− w1+ s<br />

c2v p− c1+ s<br />

+ or equivalently, 2<br />

( − ) = 39.375.<br />

Such an order<strong>in</strong>g behavior<br />

directly <strong>in</strong>fluences each party’s expected profit curve<br />

shown <strong>in</strong> Fig. 2.<br />

Fig. 1 <strong>Retailer</strong>’s optimal first-period order quantity<br />

Figure 2 compares each party’s expected profit after<br />

coord<strong>in</strong>ation <strong>with</strong> that before coord<strong>in</strong>ation which is<br />

shown <strong>by</strong> the benchmark curve. The powerful retailer,<br />

who acts as a contract designer, will never be worse off,<br />

but the results are not so optimistic. The supplier’s expected<br />

profit even becomes negative for some cases!<br />

Notably, <strong>with</strong> these sett<strong>in</strong>gs, the first-best expected<br />

c<br />

profit Π = 221 209.<br />

Before coord<strong>in</strong>ation, however,<br />

Π R = 175 960 and Π S = 16 437. This leaves an improvement<br />

of 28 812 which is almost twice the supplier’s<br />

expected profit before coord<strong>in</strong>ation.<br />

Fig. 2 Expected profit for each party before and after<br />

coord<strong>in</strong>ation


WANG Xiaolong (王小龙) et al:<strong>Coord<strong>in</strong>ation</strong> <strong>by</strong> <strong>Option</strong> <strong>Contracts</strong> <strong>in</strong> a <strong>Retailer</strong>-<strong>Led</strong> …<br />

Table 1 illustrates the profit allocation issues <strong>in</strong> detail.<br />

For these calculations, the new contract is feasible<br />

only when 34.8o2 43.8.<br />

For low o 2 <strong>in</strong> this range,<br />

each party becomes strictly better off. In particular,<br />

consider the best profit allocation situations for which<br />

each side gets a double-digit share of the extra profit.<br />

When o 2 is around 40, the supplier will get more<br />

than 60% of the extra system profit which is double<br />

what he earned before coord<strong>in</strong>ation. Such a rosy profit<br />

improvement surely offers a strong <strong>in</strong>centive for the<br />

supplier to accept the new contract. Most <strong>in</strong>terest<strong>in</strong>gly,<br />

such profit allocations correspond to o2 39.6 which<br />

N N<br />

implies e+ o1 w1<br />

and d = d1=<br />

0. Thus, order<br />

postponement (until the demand uncerta<strong>in</strong>ty is resolved)<br />

can be beneficial to each party. This is contrast<br />

to the traditional thought that suppliers always welcome<br />

as-early-as-possible orders. Our op<strong>in</strong>ion is that<br />

suppliers hate late orders mostly because they make<br />

the work of arrang<strong>in</strong>g production schedules more difficult.<br />

However, a properly designed <strong>in</strong>centive mechanism,<br />

which both clarifies the quantity that the supplier<br />

should produce and br<strong>in</strong>gs extra profit, elim<strong>in</strong>ates<br />

these negative concerns. Such a result, however, needs<br />

further proof from theory and practice.<br />

Table 1 Profit allocations between the two parties<br />

Extra profit allocation on supplier<br />

Second-period option price<br />

(% of improvement scope)<br />

0 100<br />

34.8 1<br />

… …<br />

39.6 67<br />

39.9 63<br />

40.2 58<br />

40.5 53<br />

40.8 48<br />

41.1 44<br />

41.4 39<br />

41.7 35<br />

42.0 30<br />

42.3 26<br />

… …<br />

43.8 3<br />

Note: For simplicity, we state here only the allocation status for the supplier.<br />

Its counterpart for the retailer can be calculated s<strong>in</strong>ce the sum is equal<br />

to 1.<br />

4 Conclusions<br />

579<br />

This analysis considers the coord<strong>in</strong>ation of option contracts<br />

<strong>in</strong> a retailer-led supply cha<strong>in</strong> where the retailer<br />

designs the contract and the supplier’s production<br />

needs to be coord<strong>in</strong>ated <strong>with</strong> the contract. Consider the<br />

case where the supplier operates two modes of production<br />

and the retailer has the ability to update the demand<br />

forecast as the sell<strong>in</strong>g season approaches. With<strong>in</strong><br />

this contract class, a set of pric<strong>in</strong>g conditions are def<strong>in</strong>ed<br />

that align the supplier and retailer to act <strong>in</strong> the<br />

best <strong>in</strong>terests of the channel. The analysis leads to<br />

some <strong>in</strong>terest<strong>in</strong>g results.<br />

First, option prices should be negatively correlated<br />

to the price and should be <strong>in</strong> a relevant range. This<br />

gives a trade-off between the higher unit compensation<br />

the supplier requires and the lower repurchas<strong>in</strong>g price<br />

he will get. To assure that the option mechanism is an<br />

effective <strong>in</strong>centive, the option prices should be <strong>with</strong><strong>in</strong> a<br />

relevant range. The second-period option price should<br />

be no greater than the difference between the second-<br />

period unit production costs and the salvage value.<br />

Otherwise, the supplier will earn little due to the rapidly<br />

decreas<strong>in</strong>g exercise price and the coord<strong>in</strong>ation will<br />

fail.<br />

Second, the first-period option price should be no<br />

greater than the second-period price and should be<br />

l<strong>in</strong>early correlated to the second-period option price<br />

when the latter is beyond some threshold. This result is<br />

true regardless of the specific functional form of the<br />

market demand or market signal.<br />

Third, the retailer’s order<strong>in</strong>g behavior is <strong>in</strong>fluenced<br />

<strong>by</strong> the dual effect of the predictive power of the new<br />

market <strong>in</strong>formation and option prices. Some extreme<br />

pric<strong>in</strong>g cases may result <strong>in</strong> the order<strong>in</strong>g behavior be<strong>in</strong>g<br />

<strong>in</strong>dependent of the new market <strong>in</strong>formation. The <strong>in</strong>fluence<br />

pattern is so complex that there is no closed<br />

form solution to the retailer’s first-period optimal order<br />

quantity. Fortunately, this does not prevent explor<strong>in</strong>g<br />

the optimal pric<strong>in</strong>g conditions which are stated <strong>in</strong><br />

Proposition 3. A numerical example is given for the<br />

profit allocation issues. An <strong>in</strong>terest<strong>in</strong>g result is that<br />

order postponement may be beneficial to both parties,<br />

though this needs further theoretical proof.


580<br />

References<br />

[1] Mess<strong>in</strong>ger P, Narasimhan C. Has power shifted <strong>in</strong> the<br />

grocery channel. Market<strong>in</strong>g Science, 1995, 14(2): 189-223.<br />

[2] Ailawadi K. The retail power-performance conundrum:<br />

What have we learned. Journal of Retail<strong>in</strong>g, 2001, 77:<br />

299-318.<br />

[3] Bloom P, Perry V. <strong>Retailer</strong> power and supplier welfare:<br />

The case of Wal-Mart. Journal of Retail<strong>in</strong>g, 2001, 77:<br />

379-396.<br />

[4] Fisher M, Hammond J, Obermeyer W, et al. Mak<strong>in</strong>g supply<br />

meet demand <strong>in</strong> an uncerta<strong>in</strong> world. Harvard Bus<strong>in</strong>ess<br />

Review, 1994, 72(3): 83-93.<br />

[5] Spengler J. Vertical <strong>in</strong>tegration and antitrust policy. Journal<br />

of Political Economy, 1950: 347-352.<br />

[6] Tirole J. The Theory of Industrial Organization. Cambridge,<br />

Boston, USA: The MIT Press, 1988.<br />

[7] Lariviere M, Porteus E. Sell<strong>in</strong>g to the newsvendor: An<br />

analysis of price-only contracts. Manufactur<strong>in</strong>g and Service<br />

Operations Management, 2001, 3(4): 293-305.<br />

[8] Pasternack B. Optimal pric<strong>in</strong>g and returns policies for<br />

perishable commodities. Market<strong>in</strong>g Science, 1985, 4(2):<br />

166-176.<br />

[9] Cachon G, Lariviere M. <strong>Supply</strong> cha<strong>in</strong> coord<strong>in</strong>ation <strong>with</strong><br />

Ts<strong>in</strong>ghua Science and Technology, August 2008, 13(4): 570-580<br />

revenue-shar<strong>in</strong>g contracts: Strengths and limitations. Management<br />

Science, 2005, 51(1): 30-44.<br />

[10] Taylor T. <strong>Supply</strong> cha<strong>in</strong> coord<strong>in</strong>ation under channel rebates<br />

<strong>with</strong> sales effort effects. Management Science, 2002, 48(8):<br />

992-1007.<br />

[11] Moorthy K S. Manag<strong>in</strong>g channel profits: Comment. Market<strong>in</strong>g<br />

Science, 1987, 6: 375-379.<br />

[12] Tsay A. Quantity-flexibility contract and supplier-customer<br />

<strong>in</strong>centives. Management Science, 1999, 45(10):<br />

1339-1358.<br />

[13] Newbery G. Competition, contracts, and entry <strong>in</strong> the electricity<br />

spot market. RAND Journal of Economics, 1998, 29:<br />

726-749.<br />

[14] Wu D J, Kle<strong>in</strong>dorfer P R, Zhang J E. Optimal bidd<strong>in</strong>g and<br />

contract<strong>in</strong>g strategies for capital-<strong>in</strong>tensive goods. European<br />

Journal of Operational Research, 2002, 137:<br />

657-676.<br />

[15] Donohue K. Efficient supply contracts for fashion goods<br />

<strong>with</strong> forecast updat<strong>in</strong>g and two production modes. Management<br />

Science, 2000, 46(11): 1397-1411.<br />

[16] Barnes-Schuster D, Bassok Y, Anup<strong>in</strong>di R. <strong>Coord<strong>in</strong>ation</strong><br />

and flexibility <strong>in</strong> supply contracts <strong>with</strong> options. Manufactur<strong>in</strong>g<br />

& Service Operations Management, 2002, 4(3):<br />

171-207.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!