Description
Supply chain management involves the selection, coordination and motivation of independently operated suppliers. The central planner perspective in operations management translates poorly to vertically separated chains, where suppliers recurrently seem to object to benevolent information sharing and centralized decision rights.
Risks and Coordination Mechanisms in Telecom Supply
Chains
Per J. Agrell, Robert Lindroth, and Andreas Norrman
Abstract. Supply chain management involves the selection, coordination and
motivation of independently operated suppliers. The central planner’s perspec-
tive in operations management translates poorly to vertically separated chains,
where suppliers recurrently seem to object to benevolent information sharing
and centralized decision rights. Seen from the supplier’s perspective, such
resistance may very well be rational. A downstream assembly line disclosing
reliable information on actual and forecasted sales puts itself at a disadvantage
when bargaining on share of chain pro…ts. In this paper, we use a minimal
agency model to contrast known optimal mechanisms with the actual prac-
tice in the telecommunications industry. A three-stage supply chain under
stochastic demand and varying coordination and information asymmetry is
modeled. A two-period investment-production game addresses the informa-
tion sharing and speci…c investment problem in the telecom industry. The
observed price-quantity contracts under limited commitment are shown to be
inadequte under realistic asymmetric information assumptions. More a result
of gradually evolving changes in bargaining power than coordination e¤orts,
the upstream urge to coordinate may further deteriorate performance in terms
of our model.
1. Introduction
Supply Chain Management (SCM) has been increasingly discussed in academia
since the mid-1980s (e.g. Houlihan 1985; Jones & Riley 1985). There have been
numerous de…nitions of SCM (see e.g. Mentzer et al, 2001) but most include the
selection, coordination and motivation of independently operating suppliers in at
least a three-tier chain. Although one core concept of SCM is close relationships, in-
dustrial experience has showed that it is di¢cult to work across inter-organizational
borders. A key component for making this happen is to share both risks and re-
wards among the members of the supply chain. Especially in industries moving
towards longer supply chains (e.g. due to outsourcing) and facing increasingly un-
certain demand as well as supply, the issue of supply chain risk handling and risk
sharing along the supply chain is a topic of growing importance.
In the …eld of business logistics, these important risk-sharing issues are often
mentioned but not further elaborated on (e.g. Cooper & Ellram 1993; Cooper,
Lambert & Pagh 1997; Motwani et al. 1998; Skjoett-Larsen 1999; Mentzer et al
2001). Far more e¤orts have been spent on risks and di¤erent types of supply chain
Key words and phrases. Supply chain, coordination, capacity, contracts.
1
2 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
contracts in the quantitative and modeling oriented research stream of SCM. Al-
though many interesting risk sharing instruments (especially contracts) have been
analyzed, the context has often been rather stylized, often involving two-stage sup-
ply chains with a single product, single buyer and single supplier. Reviewing recent
advances in the …eld, Tsay et al. (1998) conclude that the design of contracts and
analysis of supply chain behavior and performance is still an important and chal-
lenging …eld of research. They also argue that the scope of issues that the literature
has addressed so far has been limited, neglecting dynamics, strategic interaction and
production economics.
One complication for supply chain risk sharing is that the companies involved
often have di¤erent business logic, e.g. in terms of how their revenues and costs
are generated, and the size, speci…city and life span of investments. The …rms have
individual relations between revenue and product life cycles, ”clockspeed” (Fine,
2000), and the design of products and processes. The di¤erences in business logic
and clockspeed might reduce truthful information sharing and introduce production
ine¢ciency as well as risks of technological ine¢ciency. In other words, although
the need for joint coordination and risk sharing might be larger, the increased
asymmetry of information and business logic would probably result in the separate
companies of a supply chain trying to myopically reduce their own risk. A lack of
e¤ective incentive structure to induce global supply chain optimization may pro-
mote the opportunistic and myopic behavior of the chain …rms. This behavior is
Pareto-dominated, i.e. it implies a suboptimal overall chain pro…t and it may even
threaten the long-term viability of the chain. Telecommunications is one example
of a heterogeneous industry where the clockspeed has increased considerably and
where the business logic has changed during the past decade. Among the factors
that contribute to an increased need for improved coordination mechanisms in the
telecom industry are an increased use of outsourcing of manufacturing and R&D to
suppliers, shorter product life cycles and compressed time-to-market (i.e. increased
clockspeed). Furthermore, capacity limitation of key components, heavy ramp-up
of demand early in product life cycles and an increased demand for on-time deliv-
eries in shorter time frames, and generally shorter lead times all increase the risks
in the supply chain.
The purpose of this paper is to critically analyze two key coordination challenges
in the telecom industry, in order to …nd theoretical as well as practical arguments for
the development of new coordination mechanisms. The methodology will draw on
common approaches of industrial organization and agency theory, using a dynamic
three-stage supply chain model under stochastic demand.
The paper is structured as follows: In the next section we review previous work
on supply chain risk sharing. Thereafter the telecom industry and some key features
are discussed. In the fourth section the model is de…ned and some initial properties
are derived. A numerical illustration is provided in section …ve. The paper closes
with some conclusions and policy implications.
2. Previous work
When reviewing the literature on supply chain risk management, it is apparent
that two major research streams are present. One contains conceptual exploratory
research, and the other is mathematical modeling. In the …rst stream researchers
try to grasp and structure managerial issues related to supply chain risks, while the
RISKS AND COORDINATION IN TELECOM 3
other stream is focused on modeling to explore or devise decision rules for clearly
de…ned risk-sharing instruments, e.g. contractual mechanisms.
2.1. Conceptual Exploratory Research. Examples of conceptual exploratory
research are Zsidisin and Ellram (1999), Zsidisin et al. (2000) and Hall (1999).
Zsidisin et al. (2000) look at the purchasing organizations’ involvement in risk
assessment and risk management. They list a number of key supply risks: business
risk (e.g. supplier’s …nancial stability); supplier capacity constraints (e.g. lack of
equipment, people or input to handle demand ‡uctuations); quality; production
technology changes (e.g. current technology becomes obsolete), product design
changes (e.g. because of dynamic customer demands); and disasters. The conclu-
sion from this study is that although many companies were aware of risks associated
with supply, many did not take the requisite actions to mitigate them. Although
using a dyadic focus, the study gives voice to more general supply chain problems.
A normative guide for handling purchasing risks is given in Zsidisin and Ellram
(1999).
2.2. Mathematical Modeling. Tsay et al. (1998) classify the supply chain
modeling literature into eight streams based on contract clauses: speci…cation of
decision rights; pricing, minimum purchase commitments; quantity ‡exibility; buy-
back or returns policies; allocation rules; lead times; and quality.
In the pricing stream, Jeuland and Shugan (1983) show how quantity discounts
can in‡uence and improve channel (dyad) coordination, but also point out imple-
mentation barriers. Moorthy (1987) argues that a two-part tari¤ (a …xed franchise
fee + quantity dependent cost) is superior to quantity discount: it is simpler and
it separates the problems of coordination from the problem of pro…t sharing. The
mechanism design results are con…rmed in the general coordination framework by
La¤ont and Tirole (1993). Li and Kourvelis (1999) model risk-sharing supply con-
tracts by incorporating explicit risk premiums. They focus on the buyer in a context
of deterministic demand and uncertain price. Within the price window the buyer
pays the agreed price, but outside of this the …rms share (in a way agreed upon)
additional costs or bene…ts.
Eppen and Iyer (1997) study inventory commitment and buy-back agreements
in the fashion industry. The buyer commits to a volume and contracts on a prede-
termined capacity before the sales period, with deliveries in two periods. Penalties
are applied in case the buyer does not fully exercise the delivery option.
Quantity ‡exibility issues are important tools for handling and sharing sup-
ply chain risks. Such contracts can de…ne ‡exibility regarding quantity, time or
capacity.
Riordan (1984) considers a two-level system with asymmetric information on
demand at the retailer and private information on manufacturing costs at the sup-
plier. The decentralization mechanism proposed is a payment schedule that uses
information revealed by the managers to achieve optimal coordination. Porteus and
Whang (1991) address the intra-…rm agency problem between e¤ort-averse market-
ing managers and a manufacturing process allocating the realization of a stochastic
capacity. Their optimal incentive scheme translates into a franchising model with a
…xed fee per unit ex post capacity delivered in return for an ex ante …xed franchise
fee paid to the owner. A second contract makes the marketing function the resid-
ual claimant for realized capacity by delegating allocation decisions to a secondary
4 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
market. Porteus and Whang (1991) show that the latter incentive plan is costly for
the owner, although it achieves an optimal allocation of capacity and managerial
e¤ort.
van Mieghem (1999) investigates the e¤ectiveness of three contract types, …xed
price, bargaining and state-dependent contracts, for a single-period two-level co-
ordination and capacity investment problem. Although production coordination
can be achieved, only the state-dependent bargaining contract achieves full coor-
dination, but, unknown ex ante, the distribution of the surplus relies on veri…able
investments by the parties.
The coordination problem of production and inventory holding in a multi-
echelon system is studied by Lee and Whang (1999), who present an optimal im-
plementable mechanism assuming stationary parameters and perfect information.
The mechanism implements an order-up-to system, using a state-dependent con-
tract of consignment and non-linear backlog and shortage penalties. Cachon and
Zipkin (1999) give a set of linear contracts for a similar production-inventory multi-
echelon system. Their mechanism is based on ex post echelon and stage inventory
and backorder levels.
Cachon and Lariviere (1999) look at the so-called turn-and-earn allocation rule
among multiple suppliers. The allocation rule is investigated in a non-cooperative
setting, showing rent distortions for most capacity-constrained cases towards the
supplier. The retailers are competing for capacity even when capacity utilization is
low, which distorts sales to assure future availability.
Our model extends the existing research by explicitly addressing a three-stage
model under asymmetric information and limited commitment. By o¤ering more
realistic assumptions for the telecom industry, where full integration is infeasible or
unattractive, we can o¤er more detailed policy advice to the industry. We also ad-
dress the implications for supply chain policy of evolving partnership/coordination
roles in the chain. Further, by modeling the ex ante investment decision and ex
post information revelation problem jointly, we focus at aggregate e¤ects on supply
chain performance.
3. The Telecommunications Industry
The telecom supply chain has undergone major shifts during the past decade,
and is still turbulent. Because of demand uncertainty in both level and timing,
the roles and responsibilities in the supply chain are changing, often accelerated by
outsourcing, leading to initially unclear interfaces. There is also strong growth and
consolidation among suppliers, leading to shifts in the power balance, and di¤erent
business logic and clockspeed among the players. Our focus is on an OEM (Original
Equipment Manufacturer) delivering infrastructure such as networks and radio base
stations and his two closest tiers of suppliers.
Telecom supply chains in general involve many di¤erent types of players and
deliver di¤erent outputs such as physical products, immaterial capacity, and instal-
lations as well as personal services (Figure 1). A major EMS (Electronics Manu-
facturing Services Provider, e.g. Flextronics or Solectron), may support many of
the OEMs (e.g. Nokia, Ericsson or Lucent) (Weinberg 2001). Global operators
(e.g. Vodafone or Telefonica) usually choose di¤erent OEMs to supply di¤erent
countries, and quite frequently they use more than one supplier within a country.
Furthermore, some 2nd and 3rd tier suppliers are supplying not only the next step
RISKS AND COORDINATION IN TELECOM 5
Companies in the model
OEM EMS Operator Consumer
2
nd
tier
supplier
3
rd
tier
supplier
OEM EMS Operator Consumer
2
nd
tier
supplier
3
rd
tier
supplier
Coverage, Capacity
& Services
Systems & Services Products Components Material
Companies in the model
OEM EMS Operator Consumer
2
nd
tier
supplier
3
rd
tier
supplier
OEM EMS Operator Consumer
2
nd
tier
supplier
3
rd
tier
supplier
Coverage, Capacity
& Services
Systems & Services Products Components Material
Coverage, Capacity
& Services
Systems & Services Products Components Material
Figure 1. Simpli…ed telecom supply chain with model scope in
center grey zone.
in the supply chain, but also companies further downstream. However, in this pa-
per we focus primarily on three of the players: 2nd tier supplier, EMS and OEM.
3.1. Uncertain Demand. A key feature of the telecom industry is the un-
certain demand. The uncertainty on the telecom market, and the short product
life cycles, make it very di¢cult to produce reliable forecasts of required supply
chain capacity. There are many reasons for the uncertainty in the underlying de-
mand: new operators emerge, and the introduction of the new 3G telecom systems
is subject to governmental regulations regarding timing and coverage, causing a
very fast ramp-up of volumes. Large orders must be delivered simultaneously to
remote areas in di¤erent countries.
New consumers are continuously added, both geographically (new countries)
and in markets (e.g. transmission of data), in consumer segments (e.g. teenagers,
children and senior citizens) and in technology (shift from 2nd generation, e.g. GSM
or TDMA, to 3rd generation technology, such as WCDMA). On the whole, the un-
derlying demand for coverage, capacity and services has been growing profoundly,
and was expected to explode, making it extremely di¢cult to forecast demand ac-
curately. However, in the past year another side of this dynamic industry has been
revealed that further underscores the risks faced by the supply chain. Despite a
promising future, major telecom players have postponed many 3G plans. Pro…t
warnings and adjusted market forecasts have triggered massive restructuring pro-
grams, lay-o¤s and of course reduced demand for equipment. Suppliers planning
for a seemingly prosperous future may have been building capacity that will not be
needed for years.
OEMs have increased the service and software part of their o¤ering. There
is also an observed tendency in the leading edge of the industry to rede…ne core
services — away from hardware (switches and antennas) into pure technology de-
velopment (see, e.g., Feder, 2001).
3.2. Outsourcing and Unclear Interfaces. Another distinguishing feature
of the industry is its extensive use of outsourcing. Apart from traditional logistics
and internal support activities, even manufacturing activities have been outsourced.
In recent years the EMSs have increased their share of manufacturing and assembly.
In many cases this has been done by acquiring plants from the OEMs. An immense
growth of contract manufacturing can be seen: Flextronics increased their net sales
eight times from 1997-2001 (Flextronics 2001). Part of this growth can be explained
by consolidation. EMSs also add new services (e.g. logistics) to their services
6 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
$
Volume, y
Cost
2
nd
tier supplier
Revenue
$
Volume, y
Cost
EMS
Revenue
$
Volume, y
Cost
OEM
Revenue
$
Volume, y
Cost
2
nd
tier supplier
Revenue
$
Volume, y
Cost
2
nd
tier supplier
Revenue
$
Volume, y
Cost
EMS
Revenue
$
Volume, y
Cost
EMS
Revenue
$
Volume, y
Cost
OEM
Revenue
$
Volume, y
Cost
OEM
Revenue
Figure 2. Schematic cost and revenue functions in an empirical
telecom supply chain.
o¤ered, often by buying companies that possess the desired capabilities. These
changes have gradually changed the bargaining position of the EMSs relative to
the OEMs.
Thus it often happens that 2nd tier suppliers deliver and contract standard
products. However, for strategic or customized components, the interrelation with
the OEM could be quite strong both during the development phase and also during
the production phase. While forecasts and planning information are normally sent
from the OEM to the EMS, and from the EMS to the 2nd tier supplier, there
are situations where the OEM and the 2nd tier supplier shortcut this loop with
medium-term and long-term forecasts to counteract adverse supply chain e¤ects,
e.g. bullwhip e¤ect. For the most strategic components the OEM might do this
to secure the supply, sometimes also by …nancially helping the 2nd tier supplier
with speci…c assets, investments and raw material purchase. Being able to trust
the information supplied and the commitment given to coordination is important
for the smooth operation of a dynamic supply chain.
3.3. Heterogeneous Business Logic. As the telecom supply chain evolves,
the business logic of the companies within the supply chain di¤ers and changes over
time. Looking at the three tiers in our model, the cost and revenue functions are
di¤erent, as is illustrated in Figure 2.
The 2nd tier supplier invests in manufacturing capacity that involves building
new production lines or plants (Figure 2). Production is capital intensive and
relies heavily on …xed assets and less on more ‡exible resources such as manpower.
Investment decisions have to be made years before the …nal products reach the
market. Once the investment is made, the production yield is maximized. New
investments to increase capacity can make them temporarily unpro…table until a
new break-even volume is reached. One major risk for a 2nd tier supplier is low
utilization of production capacity; another is that technology or market changes
downstream will make its products and processes obsolete.
In many cases the EMS is not taking much risk regarding volume or speci…c
investments (Figure 2). The major activity is labor-intensive product assembly.
Typically, the OEM reimburses the EMS by a cost-plus contract: cost reimburse-
ment and an additional markup for pro…t. If the EMS has acquired speci…c facilities
from the OEM, the OEM usually guarantees a certain demand level for the …rst
years - the OEM carries the demand risk despite the outsourcing arrangement. A
key performance indicator for many EMSs is inventory turnover rate, thus they
are reluctant to carry inventory of supplies, regardless of who assumes the actual
holding cost.
RISKS AND COORDINATION IN TELECOM 7
The OEM, …nally, has huge investments in technology and product develop-
ment. In order to reach pro…table volumes it is critical to hit the market window
with accurate timing. The market window tends to be rather tight, so payback
time on the investments must be short. The costs of the OEM tend to be rather
volume-independent: once the product development is done, the variable costs are
quite insigni…cant. The OEM faces the risk of inadequate production capacity at
the supplier level. In early 2000, this was the case for many electronic components
in the industry, leading to allocations. Since late 2001 the situation is the opposite:
overcapacity plagues the business. The business logic implies that the OEM desires
to satisfy any realized demand in the narrow market window. He must be able to
deliver as much as possible when the demand hits his most positive forecast to get
payback on early investments. The upstream tiers facing limited capacity and vari-
able costs will not have the same leverage on volume. From the upstream (myopic)
perspective, a conservative capacity assessment lowers the risk and increases pro…t
in case of low demand.
From a decision-making perspective, we may distinguish three stages in the
development of the supply chain: the centralized stage, the transition stage, and
the decentralized stage.
Before the entrance of the EMS, the OEM produced and assembled in-house.
He enjoyed considerable bargaining power toward the upstream supplier, by virtue
of size and privileged information.
As the OEM outsources the assembly process to the EMS, the roles change
gradually. Initially, the OEM gains an exceptional advantage by retaining full
cost information on the EMS, yielding bargaining power but rejecting the actual
production responsibility.
In the third stage, the OEM’s cost information on the upstream tiers is weak-
ened. This naturally a¤ects its authority to coordinate and allocate supply chain
rents. Depending on the upstream economies of scale and product di¤erentiation,
the relative power positions between the OEM and the EMS may alter. Decision
making in this stage is conducted in a decentralized manner, where each member
of the supply chain independently reviews coordination initiatives.
The organizational and informational evolution in the industry has not been
accompanied by a corresponding development of the contractual coordination mech-
anisms. Our research attempts to highlight the implications for supply chain per-
formance of such approaches.
4. Model
The model structure has three participants, an OEM (or retailer) R serving
the …nal market, an EMS factory F (or …rst tier supplier) assembling the product
on order from the OEM, and a (second tier) supplier S providing components on
…xed orders from the factory. Production and sales take place during two periods
t = 1; 2: The modeled chain re‡ects the current institutional structure in the ver-
tically separated chains in the telecommunications industry, where manufacturing
operations are frequently outsourced and the organizational distance to the …rst-
tier suppliers is often long. Only a single product and a single …rm in each echelon
are considered; thus there is no possibility to implement tournaments or yardstick
regimes. The OEM is a price taker with a price p on a competitive market with
stochastic demand D. The EMS unit price w
t
= (1 +®) v
t
in period t is based on
8 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
•Decentralized
coordination.
•Inherited centralized
coordination.
•Centralized coordination.
•OEM loses contact with
2
nd
tier supplier.
•OEM loses visibility of 2
nd
tier supplier.
•OEM has information of
2
nd
tier’s cost structure.
•OEM has obsolete
information of EMS’s cost
structure.
•Processes, management
and resources the same.
àOEM knows EMS’s cost
structure.
•OEM knows cost
structure of factory.
•EMS getting stronger. •OEM still strong. •OEM strong vs. supplier.
•EMS establishing new
processes, management and
resources.
•Assembly/factory
outsourced.
•Assembly/factory in-house.
•Decentralized
coordination.
•Inherited centralized
coordination.
•Centralized coordination.
•OEM loses contact with
2
nd
tier supplier.
•OEM loses visibility of 2
nd
tier supplier.
•OEM has information of
2
nd
tier’s cost structure.
•OEM has obsolete
information of EMS’s cost
structure.
•Processes, management
and resources the same.
àOEM knows EMS’s cost
structure.
•OEM knows cost
structure of factory.
•EMS getting stronger. •OEM still strong. •OEM strong vs. supplier.
•EMS establishing new
processes, management and
resources.
•Assembly/factory
outsourced.
•Assembly/factory in-house.
~1995 ~2000
S
2
nd
tier
F
EMS
R
OEM
S
2
nd
tier
S
2
nd
tier
F
EMS
R
OEM
OEM
F R
Figure 3. Summary of organizational and information evolutions
in the telecom industry between 1995 and 2000.
a veri…able and observable input cost v
t
and a negotiated markup ®. The supplier
produces at cost function C(Q) and charges the factory a unit price v
t
for the
common order size Q at period t.
Below, we will elaborate on the informational assumptions regarding the three
…rms.
The continuously di¤erentiable, invertible and monotonous probability distrib-
ution function F
) of the demand, with distribution function f ) and the price p
are common knowledge.
The OEM may transmit a forecast s to signal the expected demand for the
upcoming period. Under any informational assumptions, the OEM sequentially
decides upon the order quantities for the two periods, Q
1
and Q
2
. The forecast is
observable but not veri…able in court. The OEM maximizes the two-period echelon
pro…t U
R
;
U
R
) = (p ¡w
1
) Q
1
+ (p ¡w
2
) Q
2
¡p
Ã
Z
Q1
0
F (x
1
) dx
1
+
Z
Q2
0
F (x
2
) dx
2
!
For notational convenience, denote by G(Q) =
R
Q
0
F (x) dx the expected num-
ber of units not sold at production level Q: The EMS factory maximizes the horizon
objective function U
F
) =
P
2
t=1
®v
t
Q
t
. We assume that the …xed costs for F are
normalized to zero: Later, we will introduce an additional informational assumption
on behalf of the factory that will render its role more interesting.
The supplier S has private information about its cost function C(Q) and in-
vestment opportunities. At the beginning of the …rst period, the supplier learns the
RISKS AND COORDINATION IN TELECOM 9
Figure 4. Timeline for the investment-production game with
OEM coordination and asymmetric information.
amount A it would take in order to achieve a decrease of marginal cost c
0
Q in the
next period. The initial investment amount is a draw from a uniform distribution
with support on
£
A; A
¤
: The supplier has the …rst period cost function
C
1
(Q
1
; ±
1
) =
1
2
cQ
2
1
+±
1
A
where Q
1
is the actual production ordered in the …rst period, and ±
1
= 1 if an
investment of size A is undertaken in period 1 and ±
1
= 0 otherwise. The second
period cost function C
2
(Q
2
; ±
1
) is
C
2
(Q
2
; ±
1
) =
1
2
(c ¡±
1
c
0
) Q
2
2
where ±
1
is the indicator of previous period investments and Q
2
is the second
period’s realized production. The supplier maximizes the horizon echelon pro…t
U
S
;
U
S
) = v
1
Q
1
+v
2
Q
2
¡C
1
(Q
1
; ±
1
) ¡C
2
(Q
2
; ±
1
)
Fixed costs at any stage are ignored, as the focus is on the pro…t contribution
of a particular decision.
4.1. Production Game. The decisions, production and information exchange
are modeled as a two-period game, which is structured below. We ignore discount-
ing e¤ects, assuming that the values are given in real terms or that the interval is
limited. However, this implies no loss of generality. The decisions in the model,
illustrated in Figure 4 are made in the following order, which has importance with
respect to the contracts derived.
(1) S draws and observes initial investment A from
£
A; A
¤
:
(2) R proposes a mechanism M
1
(w
1
; v
1
; Q
1
; X
1
) to S and F.
10 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
(3) S accepts or rejects M
1
(4) F accepts or rejects M
1
(5) (If M
1
accepted) R may announce (non-veri…ably) a demand forecast s
for period 2.
(6) (If M
1
accepted) S may announce (non-veri…ably) a sunk investment of
H
(7) (If M
1
accepted) S decides on investment ±
1
2 f0; 1g
(8) (If M
1
accepted) S and F produce Q
1
(9) (If M
1
accepted) Demand is revealed x
1
(10) (If M
1
accepted) Payouts to all for period 1 according to M
1
(w
1
; v
1
; Q
1
; x
1
)
(11) R proposes a mechanism M
2
(w
2
; v
2
; Q
2
; X
2
) to S and F.
(12) S accepts or rejects M
2
(13) F accepts or rejects M
2
(14) (If M
2
accepted) S and F produce Q
2
(15) (If M
2
accepted) Demand is revealed x
2
(16) (If M
2
accepted) Payouts to all for period 2 according to M
2
(w
2
; v
2
; Q
2
; x
2
)
Next, we will model the decentralized scenario as a function of (i) bargaining
power (or coordination role) and (ii) information access/asymmetry. In the …rst
case, the single-period price-quantity contract under full and OEM coordination
will be studied as a benchmark. The second case adds asymmetric information on
the supplier cost/capacity decision, still under OEM coordination. The third case
shifts the coordination role to the EMS, while maintaining the private information
on the supplier side and the single-period price-only contracts. The problems are
so chosen as to illustrate the outline of organizational development in the telecom
industry, where the roles and information access have been gradually evolving.
4.2. Centralized model. In the centralized solution, the three …rms coordi-
nate to achieve chain pro…t maximization. Note that the parameter ® = 0 in the
integrated scenario is set to avoid double marginalization. Instead, the joint pro…t
contribution is distributed ex post in some arbitrary way that we represent with a
triple (¼
R
; ¼
F
; ¼
S
) later in the propositions: The joint investment is the result of
the horizon control problem
max
Q1;Q2;±1
E[¦(Q; ±
1
)] =
2
X
t=1
[p (Q
t
¡G(Q
t
)) ¡C
t
(Q
t
; ±
1
)]
which for the investment case simpli…es to a two-period newsboy problem
E[¦; ±)] = p (Q
1
¡G(Q
1
)) ¡
1
2
cQ
2
1
¡A±
1
+p (Q
1
¡G(Q
2
)) ¡
1
2
(c ¡±
1
c
0
) Q
2
2
= p (E[X
1
jQ
1
] +E[X
2
jQ
2
])
¡
1
2
c
¡
Q
2
1
+Q
2
2
¢
+
µ
1
2
c
0
Q
2
2
¡A
¶
±
1
RISKS AND COORDINATION IN TELECOM 11
with the …rst order conditions
¦
Q
1
) = p ¡pF (Q
1
) ¡cQ
1
= 0
¦
Q2
; ±
1
) = p (1 ¡F (Q
2
)) ¡(c ¡±
1
c
0
) Q
2
= 0
¦
±1
) =
1
2
c
0
Q
2
2
¡A
and second order conditions
¡pf (Q
1
) ¡c < 0
¡pf (Q
2
) ¡(c ¡±
1
c
0
) < 0
which indicate a concave function in Q.
Thus, the optimal policy devolves from C (Q
¤¤
t
; ±
¤¤
1
) with the implicitly given
order quantities
F (Q
¤¤
1
)
Q
¤¤
1
=
c
p
F (Q
¤¤
1
(±
¤¤
1
))
Q
¤¤
1
(±
¤¤
1
)
=
c ¡±
¤¤
1
c
0
p
where F (x) = 1 ¡F (x). Obviously, Q
¤¤
2
(0) = Q
¤¤
1
; thus the argument can be
suppressed from Q
¤¤
2
= Q
¤¤
2
(1) : To avoid trivial or atypical solutions in the game,
we introduce the assumption of a non-trivial investment policy:
1
2
c
0
(Q
¤¤
2
)
2
< A.
Let
e
A
¤¤
=
1
2
c
0
(Q
¤¤
2
)
2
denote the highest acceptable investment cost under order
quantity Q
¤¤
2
: The investment decision ±
¤¤
1
= 1 is now governed by the decision rule
A ·
e
A
¤¤
; else investment is not undertaken, ±
¤¤
1
= 0: The overall integrated channel
pro…t contribution SC
¤¤
is calculated as
SC
¤¤
= p (Q
¤¤
1
¡G(Q
¤¤
1
)) ¡
1
2
cQ
¤¤2
1
+
Z e
A
¤¤
A
½
p (Q
¤¤
2
(1) ¡G(Q
¤¤
2
(1))) ¡
1
2
(c ¡c
0
) Q
¤¤2
2
(1) ¡a
¾
f (a) da
+
Z
A
e
A
¤¤
½
p (Q
¤¤
2
(0) ¡G(Q
¤¤
2
(0))) ¡
1
2
cQ
¤¤2
2
(0)
¾
f (a) da
=
·
p (Q
¤¤
1
¡G(Q
¤¤
1
)) ¡
1
2
cQ
¤¤2
1
¸
Ã
2A¡
e
A
¤¤
¡A
A¡A
!
+
·
p (Q
¤¤
2
¡G(Q
¤¤
2
)) ¡
1
2
(c ¡c
0
) Q
¤¤2
2
¡
1
2
³
e
A
¤¤
+A
´
¸
Ã
e
A
¤¤
¡A
A¡A
!
In the centralized model, the forecast s and the investment information H are
truthfully disseminated to all players and the potential expected pro…t of higher
accuracy would be added to the expected payo¤.
4.3. OEM Coordination with Full Information. We now turn our at-
tention to the decentralized case where the decision makers behave strategically
with respect to their information and decision opportunities. A key element in the
modeling is which player, if any, has the leading role. In the particular case of
telecommunications, we can empirically observe that the OEM has the direct link
12 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
to the market, the product de…nition and a prehistory as a vertically integrated
…rm. Thus, it is natural to allocate the decision initiative to the OEM. Initially, we
pass through the results assuming perfect information. For this model, all signaling
options are ignored.
Although the OEM observes ex post the investment decision ±, the perfor-
mance of the supply chain will heavily depend on its ability to contract ex ante on
the revealed information. We …rst state the most favorable case, where the OEM
makes the …rst-period contract contingent on the ex post investment observation.
The leverage for the supplier is now limited to rejection of single-period contracts,
which forces the OEM to fully internalize the investment. In e¤ect, this extreme
coordination achieves integration with some limits to the pro…t sharing mechanism.
Proposition 1. Under perfect information and reservation utilities normalized
at zero for S and EMS, the OEM may implement the …rst-best solution Q
¤¤
t
by
o¤ering the price-quantity single period contracts contingent on ±
(i) The investment is imposed ¡
¤
= 1 i¤
1
2
c
0
(Q
¤¤
2
(1))
2
¸ A else ¡
¤
= 0
(ii) The …rst-best production Q
¤¤
1
= F
¡1
³
p¡c
p
´
; Q
¤¤
2
(¡
¤
) = F
¡1
³
p¡c+c
0
¡
¤
p
´
(iii) No EMS rent ®
¤
= 0,
(iv) Full reimbursement of investment
v
¤
t
(¡; ±) =
8
>
<
>
:
1
Q
¤
1
C
1
(Q
¤¤
1
; ¡) for t = 1, ± = ¡
1
Q
¤
2
(¡)
C
2
(Q
¤¤
2
(¡) ; ¡) for t = 2, ± = ¡
1
Q
¤
1
C
1
(Q
¤¤
1
; 0) otherwise
9
>
=
>
;
(v) Supply chain surplus SC
¤¤
is extracted by the OEM without distortion,
(SC
¤¤
; 0; 0) :
Proof. The control variables of the OEM are v
t
(±) ; ®; Q
t
: The reaction func-
tion for the supplier depends on the net transfer o¤ered, v
t
(±) Q
t
in each period
t;
max
±
fmax fv
1
(±) Q
1
¡C
1
(Q
1
; ±) ; 0g +max fv
2
(±) Q
2
¡C
2
(Q
2
; ±) ; 0gg
A hold-up strategy by the OEM would not be an equilibrium, since the out-
come is a rejected investment in the …rst period and a rejected contract in the
second. Thus, a strategy that postpones the reward to the second period is not im-
plementable due to lack of commitment power. A hit-and-run deviation from the
supplier, to collect the investment premium without undertaking the investment
and then rejecting the second period contract, falls on the observability. The EMS
has no decision and subsequently no rent in this problem. Hence, the policy where
the OEM orders …rst-best, sinks an optimal investment in the …rst period, and the
two other tiers comply with the optimal policy, is a (weak) NE. ¤
Removing the contractability of the investment observation decreases the out-
come space in the problem. The OEM is now constrained to a single-period con-
tract, although the possible investment is observable for the second-period contract.
RISKS AND COORDINATION IN TELECOM 13
Proposition 2. Under perfect information and reservation utilities normalized
at zero for S and EMS, the OEM may implement the third-best solution Q
¤
t
by
o¤ering the price-quantity single period contracts
(i) No investment is undertaken ±
1
= 0
(ii) Production Q
¤
= Q
¤
1
= Q
¤
2
= F
¡1
³
p¡c
p
´
(iii) No EMS rent ®
¤
= 0,
(iv) Full reimbursement of investment
v
¤
t
(±) =
(
1
Q
¤
1
C
1
(Q
¤
1
; 0) for t = 1; 2 ± = 0
1
Q
¤
2
(1)
C
2
(Q
¤¤
2
(1) ; 1) for t = 2, ± = 1
)
(v) Supply chain surplus SC
1
= 2
¡
p(Q
¤
¡G(Q
¤
)) ¡
1
2
cQ
¤2
¢
that is extracted
by the OEM,
¡
SC
1
; 0; 0
¢
:
Proof. The control variables of the OEM are v
t
; ®; Q
t
: The reaction func-
tion stays as above, but the …rst-period revenue is no longer contingent on the
investment,
max
±
fmax fv
1
Q
1
¡C
1
(Q
1
; ±) ; 0g +max fv
2
(±) Q
2
¡C
2
(Q
2
; ±) ; 0gg
The …rst-best solution is generally not implementable, since the supplier would
hit-and-run in the case of ¡ = 1. The OEM could not credibly provide incentives
for the supplier to sink the investment A; since the hold-up strategy dominates
strictly in the second period, where the investment is observed. Thus, the only
NE is where no investment is undertaken, irrespective of cost and the quantities
Q
¤
= Q
¤
1
= Q
¤
2
= F
¡1
³
p¡c
p
´
: The OEM has no incentive to deviate in the absence
of investment. The supplier has no incentive to internalize an investment, nor any
reason to reject the contract.
The OEM collects a constant pro…t contribution of
SC
1
= ¦
1
R
= 2
µ
p (Q
¤
¡G(Q
¤
)) ¡
1
2
cQ
¤2
¶
¤
The results in Proposition 2 are hardly surprising, as perfect information en-
ables OEM to observe, but not necessarily promote, chain coordination. However,
this e¤ect is a general artifact of limited commitment, which, under dominant
strategy
1
, could be addressed by repeated contracting or reputation. In the context
of this presentation, the result merely gives a reference point for the cases under
asymmetric information.
4.4. OEM Coordination with Asymmetric Information. We now intro-
duce asymmetric information on the supplier capacity investment, such that OEM
cannot observe either the outcome of the cost A; or whether the investment has
been undertaken, i.e. ±
1
: In this scenario, the OEM needs either to commit to future
payments, or to accept that the supplier enjoys an information rent on its private
knowledge about the information.
1
In this context, we are not considering mixed strategies for the bargaining game.
14 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
Proposition 3. Under asymmetric information on the investment, single pe-
riod contracts and reservation utilities normalized at zero for S and EMS, the OEM
may implement the second-best solution Q
¤
t
by o¤ering a price only-contract
(i) Production Q
¤
1
= F
¡1
³
p¡c
p
´
and
Q
¤
2
(H) =
8
<
:
F
¡1
³
p¡c
p
´
for H = 0
F
¡1
³
p¡c+c
0
p
´
for H > 0
9
=
;
(iii) No EMS rent ®
¤
= 0,
(iii) No reimbursement of investment v
¤
1
=
1
Q
¤
1
C
1
(Q
¤
1
; 0) ; and
v
¤
t
(H) =
½
v
¤
1
for H = 0
1
Q
¤
2
(1)
C
2
(Q
¤
2
(1) ; 1) for H > 0
¾
(iv) Secret investment H
¤
) = 0
(v) Distorted investment is undertaken
±
¤
1
(A) =
½
1 i¤ A ·
1
2
c
0
(Q
¤
)
2
=
e
A
¤
<
e
A
¤¤
0 else
¾
(vi) Supply chain pro…t sharing
¡
¦
2
R
= SC
1
; 0; ¦
2
S
¢
under incomplete rent ex-
traction,
SC
2
= SC
1
+
1
2
h
c
0
Q
¤2
¡
e
A
¤
¡A
i
Ã
e
A
¤
¡A
A ¡A
!
= ¦
2
R
+ ¦
2
S
Proof. The EMS is still without leverage and cannot demand rents, ®
¤
= 0.
The reaction function of the supplier now re‡ects the two decision variables ±; H,
where only the second period contract possibly contingent on H, and the private
information leverage
max
±;H
fv
1
Q
1
+v
2
(H) Q
2
(H) ¡C
1
(Q
1
; ±) ¡C
2
(Q
2
(H) ; ±) ; 0g
The …rst-best solution Q
¤¤
1
; Q
¤¤
2
(±) is not an equilibrium under truthful signal-
ing, since the supplier’s hit-and-run strategy strictly dominates investment. Since
the signal H precedes the second period contract, the supplier has limited use if
it. On the other hand, the OEM may safely conclude ex post that H > 0 implies
a sunk investment to be exploited. This recursively reduces the value of the signal
to nil. The equilibrium outcome is Q
¤
= Q
¤
1
= Q
¤
2
but with higher integrated rent,
since the supplier safely invests under the distorted criterion
A ·
1
2
c
0
(Q
¤
1
)
2
=
e
A
¤
and signal H = 0 in the …rst period. Signalling H > 0 would in the case ± = 1
yield a second period contract Q
¤¤
2
(1) with no pro…t contribution and in the case
± = 0 a rejected contracts and stalled production, Q
2
= 0.
The OEM has no incentive to deviate by o¤ering higher compensation in the
…rst period, since it only transfers rent without warranting future capacity. De-
viation by o¤ering lower contracts in the second period would risk rejection. The
analysis is analogous for the OEM order quantities. The OEM surplus stays at
level ¦
1
R
= SC
1
RISKS AND COORDINATION IN TELECOM 15
The expected supplier surplus follows from the distribution of A
¦
2
S
=
Z e
A
¤
A
µ
1
2
c
0
(Q
¤
)
2
¡a
¶
f (a) da
=
³
e
A
¤
¡A
´
c
0
(Q
¤
)
2
¡
³
e
A
2
¡A
2
´
2
¡
A¡A
¢
¤
We notice here that the OEM with bargaining power limited to the EMS stage
cannot extract all rents from the chain. However, as opposed to the situation with
incomplete contracts under full information, the private information actually allows
the supplier to internalize the investment. Hence, the supply chain performance
actually increases when the OEM accepts the limitations of the contract-horizon
setup and abstains from investment veri…cation. This interpretation follows the
intuition that an overly leveraged party may gain from refraining ex ante of certain
instruments or ex post renegotiation. A regulatory body may increase social welfare
under risks of political capture by using incomplete instruments that limit the
possibility of post contractual opportunism.
A small remark can be made on the timing of the investment signal. If the
supplier could present non-veri…able information on the investment cost before the
…rst-period contract is signed and the investment is made, how would it change
the results? First, the OEM would now face the additional option of using a …rst-
period transfer v
¤¤
1
Q
¤¤
1
+H; followed by a second-period contract v
¤¤
2
Q
¤¤
2
(1) ; which
is …rst-best if H = A ·
e
A
¤¤
: However, the supplier would deviate by reporting H =
0 or A for all A 2
h
A;
e
A
¤
i
and H =
e
A
¤¤
for A 2
h
e
A
¤
; A
i
: For the …rst interval,
the supplier makes a secret investment and internalizes the pro…t. If a private
investment is not pro…table, the supplier signals the maximum public investment
and makes a hit-and-run, rejecting the second-period contract. Thus, changing the
information timing does not bring higher value to the chain, it merely changes the
nature of the distortion. In any case, the signal does not convey any meaningful
information.
4.5. EMS Coordination with Asymmetric Information. To re‡ect the
ongoing changes in the leverage between actors in the telecommunications supply
chain, we study the case with asymmetric information under which the OEM orders,
S invests and EMS holds the bargaining power to suggest and coordinate contracts.
One may perceive this solution as the result of a far-reaching decentralization, from
which the EMS factory has gained more stable or important revenue streams than
the OEM.
We state the properties in a proposition.
Proposition 4. Under asymmetric information on the investment, single pe-
riod contracts and reservation utilities normalized at zero for S and OEM, the EMS
may implement the third-best solution Q
E
t
by o¤ering a price only-contract
(i) Production Q
E
(w
¤
) = F
¡1
³
p¡w
¤
p
´
16 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
(ii) No reimbursement of investment
v
E
t
(H) =
(
1
Q
E
(w
¤
)
C
1
¡
Q
E
(w
¤
) ; 0
¢
for H = 0
1
Q
E
(w
¤
)
C
2
¡
Q
E
(w
¤
) ; 1
¢
for H > 0
)
(iii) Secret investment H
E
= 0
(iv) Distorted investment is undertaken
±
E
1
(A) =
½
1 i¤ A ·
1
2
c
0
¡
Q
E
¢
2
=
e
A
E
<
e
A
¤
0 else
¾
(v) Double marginalization ®
E
t
(w
¤
) =
w
¤
v
E
t
(w
¤
)
¡1 > 0
(vi) Supply chain surplus division
¡
¦
3
R
; ¦
3
F
; ¦
3
S
¢
under incomplete rent extrac-
tion,
SC
3
= 2p
µ
Q
E
¡
w
E
¢
¡G
¡
Q
E
¡
w
E
¢¢
¡
1
2
cQ
E2
¶
+
1
2
³
c
0
Q
E2
¡
e
A
E
¡A
´
Ã
e
A
E
¡A
A¡A
!
Proof. The EMS decision variables are ®; w
t
) and v
t
) : The OEM acts a
double pricetaker towards the market and the supply, optimizing the order quanti-
ties Q
t
over the two periods
max
Qt
E[U
R
(Q)] =
2
X
t=1
Ã
(p ¡w
t
)) Q
t
¡p
Z
Q1
0
F (x) dx
!
which yields a convex problem with the solutions
Q
E
t
(w
t
) = F
¡1
µ
p ¡w
t
p
¶
The supplier follows the reaction function under asymmetric information, using the
signal H; the investment ± and the veto right,
max
±;H
©
v
1
Q
E
1
(w
1
) +v
2
(H) Q
E
2
(w
2
(H)) ¡C
1
¡
Q
E
1
(w
1
) ; ±
¢
¡C
2
¡
Q
E
2
(w
2
(H)) ; ±
¢
; 0
ª
The EMS now faces the control problem in w
t
(H) ; since v
t
(w
t
; H) is implicitly
given from the participation constraint as v
1
(w
1
) =
1
2
cQ
1
(w
1
) and v
2
(w
2
; H) =
1
2
(c ¡c
0
) Q
2
(w
2
(H)) : However, as in the OEM coordination case with private in-
formation, the equilibrium does not involve any signaling. Instead, the supplier un-
dertakes the investment with the distorted decision rule ±
E
1
= 1 i¤ A ·
1
2
c
0
¡
Q
E
¢
2
=
e
A
E
. Ignoring the signal H and suppressing it as an argument for clarity, we con-
centrate on the case where the EMS applies a uniform policy and lets the supplier
internalize the investment. Simplifying the price signal w = (1 +®) v to the OEM
enables the markup factor ® to fall out of the participation constraint for the sup-
plier,
®
E
(w) =
w
v
1
(w)
¡1
The EMS solves the following program in w for the non-imperative investment case
(4.1) max
w
U
F
(w) = 2
£
wQ
E
(w) ¡C
1
¡
Q
E
(w) ; 0
¢¤
with the …rst order conditions (suppressing the superscript E in the implicit re-
sponse function Q
E
(w) )
Q
¡
w
E
¢
+
¡
w
E
¡cQ
¡
w
E
¢¢
Q
0
¡
w
E
¢
= 0
RISKS AND COORDINATION IN TELECOM 17
and second order condition
2Q
0
¡
w
E
¢
+Q
00
¡
w
E
¢ ¡
w
E
¡cQ
¡
w
E
¢¢
¡c
¡
Q
0
¡
w
E
¢¢
2
< 0
Due to monotonicity of F there exists a unique solution w
¤
: The OEM retains the
expected surplus
¦
3
R
= 2p
¡
Q
E
¡
w
E
¢
¡G
¡
Q
E
¡
w
E
¢¢¢
¡2
¡
1 +®
E
¡
w
E
¢¢
v
¡
w
E
¢
Q
E
¡
w
E
¢
where the EMS supplies the OEM with the optimal marginal cost and keeps the
surplus between the average cost v
E
and the marginal cost w
E
: The supplier surplus
follows from its internalized investment
¦
3
S
=
Z e
A
E
A
µ
1
2
c
0
¡
Q
E
2
¡
w
E
¢¢
2
¡a
¶
f (a) da
=
³
e
A
E
¡A
´
c
0
¡
Q
E
2
¡
w
E
¢¢
2
¡
³
e
A
E2
¡A
2
´
2
¡
A¡A
¢
The EMS retains the pro…t contribution
¦
3
F
= 2®
E
¡
w
E
¢
C
1
¡
Q
E
(w
¤
) ; 0
¢
¤
The EMS coordination implies a further lowering of supply chain performance
due to the superposing of asymmetric information at the supplier stage and double
marginalization at the OEM stage. Limited by the contractual structure, the EMS
is forced to extract the di¤erential between the average and the marginal supplier
cost by a mark-up. However, this impediment serves to discipline the EMS in its
relation with the OEM. The downstreams agent OEM implicitly allocates part of
the supply risk to be anticipated by the EMS, compensated by a lion’s share of the
chain pro…ts.
5. Illustration
To numerically illustrate the model and the …ndings, the three stages are pa-
rameterized as in Table 1. The results are summarized in Table 2, where the
production quantities, the pro…t contribution shares, the total chain pro…t and the
investment cut-o¤ limit are given for the base case and the three other cases. Note
that the EMS coordination performs worse than the no-investment case 1. The dif-
ferences are, however, fairly low with the given parameters: ¡3:4%; ¡1:5%; ¡3:5%
coordination loss relative to optimal coordination, respectively. This should not
be interpreted with respect to the actual telecom scenario, as the relative share of
variable to capacity cost parameters for the simpli…ed scenario has not yet been
validated. Finally, with respect to the investment, one may notice that contracting
problems preclude 50% and 54% of the optimal investments in cases 2 and 3, respec-
tively. This …nding has some anecdotal evidence in the industry, where suppliers
have recently demanded more active participation to undertake speci…c invesments
under uncertainty.
18 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
Parameter
p 500 D N (1000; 200)
c 1.00 A 55000
c
0
0.25 A 10000
Table 1. Parameters for the numerical illustration.
Centralized case Case 1 Case 2 Case 3
Q
¤¤
1
497.02 Q
¤
497.02 Q
¤
497.02 Q
E
481.15
Q
¤¤
2
642.14 Q
¤
2
(H > 0) 642.14 Q
¤
2
(H > 0) 642.14 Q
E
2
(H > 0) 575.24
¦
1
R
249,608 ¦
2
R
249,608 ¦
3
R
1,982
¦
1
F
0 ¦
2
F
0 ¦
3
F
247,364
¦
1
S
0 ¦
2
S
4,844 ¦
3
S
3,985
SC
¤¤
258,285 SC
1
249,608 SC
2
254,452 SC
3
249,347
e
A
¤¤
51,544
e
A 0
e
A
¤
30,879
e
A
E
28,938
Table 2. Numerical results for the centralized case and the cases
1, 2, 3.
6. Conclusion
We have proposed a three-level, two-period supply chain model as a vehicle to
model incentive con‡icts and coordinating contracts in the telecom supply chain.
Basing our structure on the telecom sector, some early results from the model
suggest that the shifting positions of bargaining strength may undermine chain
performance when the contractual structure is misaligned. Especially in the case of
previously integrated …rms that hold bargaining power, but su¤er from asymmet-
ric information and limited contractual commitment, the …ndings would suggest
reallocation of bargaining power and chain risks upstream. From an applied per-
spective, the current analysis points out the shortcomings of price-quantity-only
coordination in the absence of long contracts. However, the bounds on rent extrac-
tion that result from a reallocation of bargaining power to the EMS factory give
rise to some interesting speculations. Our conclusions for the scenarios are sum-
marized in Table 3. Although full coordination can theoretically be obtained with
simple price-quantity contracts, asymmetric information, even in limited three-tier
setting, will constrain the potential performance. A shift of bargaining power from
the OEM to the EMS is, in our model, not an improvement of coordination.
The recent changes in scope and pro…tability in the industry may be not only
the result of market outcomes on the retail or OEM level, but also an a¤ect of
the existing imperfect contractual instruments. Although not covered in the formal
analysis, some other bene…ts in terms of forecast accuracy and investment e¢ciency
may also spring from this seemingly inadequate structure. Advanced contractual
instruments that contract on forecasts and outcomes may, in the perspective of
asymmetric information, force the OEM to internalize investment costs based on
incomplete information. This may imply a downward distraction of forecasts that
may a¤ect supply chain performance. Simple contracts that allocate inventory risks
upstream are not subject to the same biases.
RISKS AND COORDINATION IN TELECOM 19
Information Coordinator Contractual relations Findings
Perfect OEM Single-period, v; w; Q contracts No investment
information Investment observable Even production
Losses of renegotiation
Asymmetric OEM Single-period, v; w; Q contracts Second-best chain pro…t
information S decides privately on ± S internalizes investment
Secret investment
Losses due to renegotiation
Asymmetric EMS Single-period, v; w contracts Third-best chain pro…t
information S decides privately on ± S internalizes investment
OEM decides on Q Losses due to renegotiation
OEM decides on quantity Double marginalization
Table 3. Model structure, relations and some …ndings.
Whether this structure could be improved by more advanced contractual in-
struments, such as two-part tari¤s and minimum quantity contracts, is the subject
of the next analysis stage. In any case, the dynamics and complexity of the un-
derlying business logic in the industry may very well be re‡ected in the optimal
contractual structure. Thus, to pursue the pervasive target of optimal mechanisms
under the true contractual conditions might be largely a futile exercise, whereas a
thorough analysis of stable and implementable mechanisms always conveys value to
industry and academia as well. In all modesty, perhaps this provides a new angle
on integrated supply chain management research.
References
[1] Cachon, G. and M. Lariviere. 1999. Capacity Choice and Allocation: Strategic Behavior and
Supply Chain Performance. Management Science, 45 (8), 1091-1108.
[2] Cachon, G.P. and P.H. Zipkin (1999) Competitive and Cooperative Inventory Policies in a
Two-Stage Supply Chain. Management Science, 45 (7), 936-953.
[3] Cooper, M.C. and L.M. Ellram (1993) Characteristics of Supply Chain Management and the
Implications for Purchasing and Logistics Strategy. The International Journal of Logistics
Management, 4 (2), 13-24.
[4] Cooper, M.C., D.M. Lambert and J.D. Pagh (1997) Supply chain management: More Than
a New Name for Logistics. International Journal of Logistics Management, 8 (1), 1-14.
[5] Eppen, G.D. and A.V. Iyer (1997) Backup agreements in fashion buying - the value of up-
stream ‡exibility. Management Science, 43, 1469-1484.
[6] Feder, B. J. (2001) Motorola to Sell Inner Workings of Cell Phones to Rivals, New York
Times, July 23.
[7] Fine, C.H. (2000) Clockspeed-based Strategies for Supply Chain Design. Production and
Operations Management, 9 (3), 213-221.
[8] Flextronics (2001) Flextronics 2001 annual report.
[9] Hall, R. (1999) Rearranging Risks and Rewards in a Supply Chain. Journal of General Man-
agement, 24 (3), 22-32.
[10] Houlihan, J. B. (1985) International Supply Chain Management. International Journal of
Physical Distribution and Materials Management, 15 (1), 22-38.
[11] Jeuland, A.P. and S.M. Shugan (1983) Managing Channel Pro…t. Marketing Science, 2 (3),
239-272.
[12] Jones, T. and D.W. Riley (1985) Using Inventory for Competitive Supply Chain Management.
International Journal of Physical Distribution and Materials Management, 15 (5), 16-26
20 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
[13] La¤ont, J.-J. and J. Tirole (1993) A Theory of Incentives in Procurement and Regulation.
MIT Press.
[14] Lee, H.L. and S. Whang (1999) Decentralized Multi-echelon Supply Chains: Incentives and
Information. Management Science, 45 (5), 633-640.
[15] Li, C-L. and P. Kouvelis (1999) Flexible and Risk-Sharing Supply Contracts Under Price
Uncertainty. Management Science, 45 (10), 1378-1398.
[16] Mentzer, J.T., W. DeWitt, J.S. Keebler, S. Min, N.W. Nix, C.D. Smith and Z.G. Zacharia
(2001) De…ning Supply Chain Management. Journal of Business Logistics, 22 (2), 1-25.
[17] Moorthy, K.S. (1987) Managing Channel Pro…ts: Comment. Marketing Sciences, 6 (4), 375-
379.
[18] Motwani, J., L. Larson and S. Ahuja (1998) Managing a Global Supply Chain Partnership.
Logistics Information Management, 11 (6), 349-354.
[19] Pasternack, B.A. (1985) Optimal Pricing and Returns Policies for Perishable Commodities.
Marketing Science, 4, 166-176.
[20] Porteus, E. and S. Whang (1991) On Manufacturing/Marketing Incentives. Management
Science, 37 (9), 1166-1181.
[21] Riordan, M. (1984) Uncertainty, Asymmetric Information and Bilateral Contracts. Review of
Economic Studies, 51, 82-93.
[22] Skjoett-Larsen, T. (1999) Supply Chain Management: A New Challenge for Researchers and
Managers in Logistics. International Journal of Logistics Management, 10 (2), 41-53.
[23] Tsay, A., A. S. Nahmias and N. Agrawal (1998) Modelling Supply Chain Contracts: A
Review. Chapter 10 in S. Tayur, R. Ganeshan and M. Magazine (Eds.) (1998) Quantitative
Models for Supply Chain Management. Kluwer Academic Publishers.
[24] van Mieghem, J. A. (1999) Coordinating Investment, Production, and Subcontracting. Man-
agement Science, 45, 954-971.
[25] Weinberg, A. (2001) Flextronics seizes an Opportunity, The Industry Standard, Feb 14.
[26] Zsidisin, G. and L. Ellram (1999) Supply Risk Assessment Analysis. Practix, Best Practices
in Purchasing and Supply Chain Management, June, 9-12.
[27] Zsidisin, G., A. Panelli and R. Upton (2000) Purchasing Organisation Involvement in Risk
Assessments, Contingency Plans, and Risk Management: An Exploratory Study. Supply
Chain Management: An International Journal, 5 (4), 187-197.
(Per Agrell) IAG School of Management, Catholic University of Louvain, 1 Place
des Doyens, B-1348 Louvain-la-neuve, BELGIUM
E-mail address, Per Agrell: [email protected]
URL: http://www.poms.ucl.ac.be.
(Robert Lindroth and Andreas Norrman) Engineering Logistics, Department of Indus-
trial Management and Logistics, Lund University, P.O.B. 118, S-221 00 Lund, SWEDEN
E-mail address, Robert Lindroth: [email protected]
E-mail address, Andreas Norrman: [email protected]
URL: http://www.tlog.lth.se
doc_428416647.pdf
Supply chain management involves the selection, coordination and motivation of independently operated suppliers. The central planner perspective in operations management translates poorly to vertically separated chains, where suppliers recurrently seem to object to benevolent information sharing and centralized decision rights.
Risks and Coordination Mechanisms in Telecom Supply
Chains
Per J. Agrell, Robert Lindroth, and Andreas Norrman
Abstract. Supply chain management involves the selection, coordination and
motivation of independently operated suppliers. The central planner’s perspec-
tive in operations management translates poorly to vertically separated chains,
where suppliers recurrently seem to object to benevolent information sharing
and centralized decision rights. Seen from the supplier’s perspective, such
resistance may very well be rational. A downstream assembly line disclosing
reliable information on actual and forecasted sales puts itself at a disadvantage
when bargaining on share of chain pro…ts. In this paper, we use a minimal
agency model to contrast known optimal mechanisms with the actual prac-
tice in the telecommunications industry. A three-stage supply chain under
stochastic demand and varying coordination and information asymmetry is
modeled. A two-period investment-production game addresses the informa-
tion sharing and speci…c investment problem in the telecom industry. The
observed price-quantity contracts under limited commitment are shown to be
inadequte under realistic asymmetric information assumptions. More a result
of gradually evolving changes in bargaining power than coordination e¤orts,
the upstream urge to coordinate may further deteriorate performance in terms
of our model.
1. Introduction
Supply Chain Management (SCM) has been increasingly discussed in academia
since the mid-1980s (e.g. Houlihan 1985; Jones & Riley 1985). There have been
numerous de…nitions of SCM (see e.g. Mentzer et al, 2001) but most include the
selection, coordination and motivation of independently operating suppliers in at
least a three-tier chain. Although one core concept of SCM is close relationships, in-
dustrial experience has showed that it is di¢cult to work across inter-organizational
borders. A key component for making this happen is to share both risks and re-
wards among the members of the supply chain. Especially in industries moving
towards longer supply chains (e.g. due to outsourcing) and facing increasingly un-
certain demand as well as supply, the issue of supply chain risk handling and risk
sharing along the supply chain is a topic of growing importance.
In the …eld of business logistics, these important risk-sharing issues are often
mentioned but not further elaborated on (e.g. Cooper & Ellram 1993; Cooper,
Lambert & Pagh 1997; Motwani et al. 1998; Skjoett-Larsen 1999; Mentzer et al
2001). Far more e¤orts have been spent on risks and di¤erent types of supply chain
Key words and phrases. Supply chain, coordination, capacity, contracts.
1
2 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
contracts in the quantitative and modeling oriented research stream of SCM. Al-
though many interesting risk sharing instruments (especially contracts) have been
analyzed, the context has often been rather stylized, often involving two-stage sup-
ply chains with a single product, single buyer and single supplier. Reviewing recent
advances in the …eld, Tsay et al. (1998) conclude that the design of contracts and
analysis of supply chain behavior and performance is still an important and chal-
lenging …eld of research. They also argue that the scope of issues that the literature
has addressed so far has been limited, neglecting dynamics, strategic interaction and
production economics.
One complication for supply chain risk sharing is that the companies involved
often have di¤erent business logic, e.g. in terms of how their revenues and costs
are generated, and the size, speci…city and life span of investments. The …rms have
individual relations between revenue and product life cycles, ”clockspeed” (Fine,
2000), and the design of products and processes. The di¤erences in business logic
and clockspeed might reduce truthful information sharing and introduce production
ine¢ciency as well as risks of technological ine¢ciency. In other words, although
the need for joint coordination and risk sharing might be larger, the increased
asymmetry of information and business logic would probably result in the separate
companies of a supply chain trying to myopically reduce their own risk. A lack of
e¤ective incentive structure to induce global supply chain optimization may pro-
mote the opportunistic and myopic behavior of the chain …rms. This behavior is
Pareto-dominated, i.e. it implies a suboptimal overall chain pro…t and it may even
threaten the long-term viability of the chain. Telecommunications is one example
of a heterogeneous industry where the clockspeed has increased considerably and
where the business logic has changed during the past decade. Among the factors
that contribute to an increased need for improved coordination mechanisms in the
telecom industry are an increased use of outsourcing of manufacturing and R&D to
suppliers, shorter product life cycles and compressed time-to-market (i.e. increased
clockspeed). Furthermore, capacity limitation of key components, heavy ramp-up
of demand early in product life cycles and an increased demand for on-time deliv-
eries in shorter time frames, and generally shorter lead times all increase the risks
in the supply chain.
The purpose of this paper is to critically analyze two key coordination challenges
in the telecom industry, in order to …nd theoretical as well as practical arguments for
the development of new coordination mechanisms. The methodology will draw on
common approaches of industrial organization and agency theory, using a dynamic
three-stage supply chain model under stochastic demand.
The paper is structured as follows: In the next section we review previous work
on supply chain risk sharing. Thereafter the telecom industry and some key features
are discussed. In the fourth section the model is de…ned and some initial properties
are derived. A numerical illustration is provided in section …ve. The paper closes
with some conclusions and policy implications.
2. Previous work
When reviewing the literature on supply chain risk management, it is apparent
that two major research streams are present. One contains conceptual exploratory
research, and the other is mathematical modeling. In the …rst stream researchers
try to grasp and structure managerial issues related to supply chain risks, while the
RISKS AND COORDINATION IN TELECOM 3
other stream is focused on modeling to explore or devise decision rules for clearly
de…ned risk-sharing instruments, e.g. contractual mechanisms.
2.1. Conceptual Exploratory Research. Examples of conceptual exploratory
research are Zsidisin and Ellram (1999), Zsidisin et al. (2000) and Hall (1999).
Zsidisin et al. (2000) look at the purchasing organizations’ involvement in risk
assessment and risk management. They list a number of key supply risks: business
risk (e.g. supplier’s …nancial stability); supplier capacity constraints (e.g. lack of
equipment, people or input to handle demand ‡uctuations); quality; production
technology changes (e.g. current technology becomes obsolete), product design
changes (e.g. because of dynamic customer demands); and disasters. The conclu-
sion from this study is that although many companies were aware of risks associated
with supply, many did not take the requisite actions to mitigate them. Although
using a dyadic focus, the study gives voice to more general supply chain problems.
A normative guide for handling purchasing risks is given in Zsidisin and Ellram
(1999).
2.2. Mathematical Modeling. Tsay et al. (1998) classify the supply chain
modeling literature into eight streams based on contract clauses: speci…cation of
decision rights; pricing, minimum purchase commitments; quantity ‡exibility; buy-
back or returns policies; allocation rules; lead times; and quality.
In the pricing stream, Jeuland and Shugan (1983) show how quantity discounts
can in‡uence and improve channel (dyad) coordination, but also point out imple-
mentation barriers. Moorthy (1987) argues that a two-part tari¤ (a …xed franchise
fee + quantity dependent cost) is superior to quantity discount: it is simpler and
it separates the problems of coordination from the problem of pro…t sharing. The
mechanism design results are con…rmed in the general coordination framework by
La¤ont and Tirole (1993). Li and Kourvelis (1999) model risk-sharing supply con-
tracts by incorporating explicit risk premiums. They focus on the buyer in a context
of deterministic demand and uncertain price. Within the price window the buyer
pays the agreed price, but outside of this the …rms share (in a way agreed upon)
additional costs or bene…ts.
Eppen and Iyer (1997) study inventory commitment and buy-back agreements
in the fashion industry. The buyer commits to a volume and contracts on a prede-
termined capacity before the sales period, with deliveries in two periods. Penalties
are applied in case the buyer does not fully exercise the delivery option.
Quantity ‡exibility issues are important tools for handling and sharing sup-
ply chain risks. Such contracts can de…ne ‡exibility regarding quantity, time or
capacity.
Riordan (1984) considers a two-level system with asymmetric information on
demand at the retailer and private information on manufacturing costs at the sup-
plier. The decentralization mechanism proposed is a payment schedule that uses
information revealed by the managers to achieve optimal coordination. Porteus and
Whang (1991) address the intra-…rm agency problem between e¤ort-averse market-
ing managers and a manufacturing process allocating the realization of a stochastic
capacity. Their optimal incentive scheme translates into a franchising model with a
…xed fee per unit ex post capacity delivered in return for an ex ante …xed franchise
fee paid to the owner. A second contract makes the marketing function the resid-
ual claimant for realized capacity by delegating allocation decisions to a secondary
4 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
market. Porteus and Whang (1991) show that the latter incentive plan is costly for
the owner, although it achieves an optimal allocation of capacity and managerial
e¤ort.
van Mieghem (1999) investigates the e¤ectiveness of three contract types, …xed
price, bargaining and state-dependent contracts, for a single-period two-level co-
ordination and capacity investment problem. Although production coordination
can be achieved, only the state-dependent bargaining contract achieves full coor-
dination, but, unknown ex ante, the distribution of the surplus relies on veri…able
investments by the parties.
The coordination problem of production and inventory holding in a multi-
echelon system is studied by Lee and Whang (1999), who present an optimal im-
plementable mechanism assuming stationary parameters and perfect information.
The mechanism implements an order-up-to system, using a state-dependent con-
tract of consignment and non-linear backlog and shortage penalties. Cachon and
Zipkin (1999) give a set of linear contracts for a similar production-inventory multi-
echelon system. Their mechanism is based on ex post echelon and stage inventory
and backorder levels.
Cachon and Lariviere (1999) look at the so-called turn-and-earn allocation rule
among multiple suppliers. The allocation rule is investigated in a non-cooperative
setting, showing rent distortions for most capacity-constrained cases towards the
supplier. The retailers are competing for capacity even when capacity utilization is
low, which distorts sales to assure future availability.
Our model extends the existing research by explicitly addressing a three-stage
model under asymmetric information and limited commitment. By o¤ering more
realistic assumptions for the telecom industry, where full integration is infeasible or
unattractive, we can o¤er more detailed policy advice to the industry. We also ad-
dress the implications for supply chain policy of evolving partnership/coordination
roles in the chain. Further, by modeling the ex ante investment decision and ex
post information revelation problem jointly, we focus at aggregate e¤ects on supply
chain performance.
3. The Telecommunications Industry
The telecom supply chain has undergone major shifts during the past decade,
and is still turbulent. Because of demand uncertainty in both level and timing,
the roles and responsibilities in the supply chain are changing, often accelerated by
outsourcing, leading to initially unclear interfaces. There is also strong growth and
consolidation among suppliers, leading to shifts in the power balance, and di¤erent
business logic and clockspeed among the players. Our focus is on an OEM (Original
Equipment Manufacturer) delivering infrastructure such as networks and radio base
stations and his two closest tiers of suppliers.
Telecom supply chains in general involve many di¤erent types of players and
deliver di¤erent outputs such as physical products, immaterial capacity, and instal-
lations as well as personal services (Figure 1). A major EMS (Electronics Manu-
facturing Services Provider, e.g. Flextronics or Solectron), may support many of
the OEMs (e.g. Nokia, Ericsson or Lucent) (Weinberg 2001). Global operators
(e.g. Vodafone or Telefonica) usually choose di¤erent OEMs to supply di¤erent
countries, and quite frequently they use more than one supplier within a country.
Furthermore, some 2nd and 3rd tier suppliers are supplying not only the next step
RISKS AND COORDINATION IN TELECOM 5
Companies in the model
OEM EMS Operator Consumer
2
nd
tier
supplier
3
rd
tier
supplier
OEM EMS Operator Consumer
2
nd
tier
supplier
3
rd
tier
supplier
Coverage, Capacity
& Services
Systems & Services Products Components Material
Companies in the model
OEM EMS Operator Consumer
2
nd
tier
supplier
3
rd
tier
supplier
OEM EMS Operator Consumer
2
nd
tier
supplier
3
rd
tier
supplier
Coverage, Capacity
& Services
Systems & Services Products Components Material
Coverage, Capacity
& Services
Systems & Services Products Components Material
Figure 1. Simpli…ed telecom supply chain with model scope in
center grey zone.
in the supply chain, but also companies further downstream. However, in this pa-
per we focus primarily on three of the players: 2nd tier supplier, EMS and OEM.
3.1. Uncertain Demand. A key feature of the telecom industry is the un-
certain demand. The uncertainty on the telecom market, and the short product
life cycles, make it very di¢cult to produce reliable forecasts of required supply
chain capacity. There are many reasons for the uncertainty in the underlying de-
mand: new operators emerge, and the introduction of the new 3G telecom systems
is subject to governmental regulations regarding timing and coverage, causing a
very fast ramp-up of volumes. Large orders must be delivered simultaneously to
remote areas in di¤erent countries.
New consumers are continuously added, both geographically (new countries)
and in markets (e.g. transmission of data), in consumer segments (e.g. teenagers,
children and senior citizens) and in technology (shift from 2nd generation, e.g. GSM
or TDMA, to 3rd generation technology, such as WCDMA). On the whole, the un-
derlying demand for coverage, capacity and services has been growing profoundly,
and was expected to explode, making it extremely di¢cult to forecast demand ac-
curately. However, in the past year another side of this dynamic industry has been
revealed that further underscores the risks faced by the supply chain. Despite a
promising future, major telecom players have postponed many 3G plans. Pro…t
warnings and adjusted market forecasts have triggered massive restructuring pro-
grams, lay-o¤s and of course reduced demand for equipment. Suppliers planning
for a seemingly prosperous future may have been building capacity that will not be
needed for years.
OEMs have increased the service and software part of their o¤ering. There
is also an observed tendency in the leading edge of the industry to rede…ne core
services — away from hardware (switches and antennas) into pure technology de-
velopment (see, e.g., Feder, 2001).
3.2. Outsourcing and Unclear Interfaces. Another distinguishing feature
of the industry is its extensive use of outsourcing. Apart from traditional logistics
and internal support activities, even manufacturing activities have been outsourced.
In recent years the EMSs have increased their share of manufacturing and assembly.
In many cases this has been done by acquiring plants from the OEMs. An immense
growth of contract manufacturing can be seen: Flextronics increased their net sales
eight times from 1997-2001 (Flextronics 2001). Part of this growth can be explained
by consolidation. EMSs also add new services (e.g. logistics) to their services
6 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
$
Volume, y
Cost
2
nd
tier supplier
Revenue
$
Volume, y
Cost
EMS
Revenue
$
Volume, y
Cost
OEM
Revenue
$
Volume, y
Cost
2
nd
tier supplier
Revenue
$
Volume, y
Cost
2
nd
tier supplier
Revenue
$
Volume, y
Cost
EMS
Revenue
$
Volume, y
Cost
EMS
Revenue
$
Volume, y
Cost
OEM
Revenue
$
Volume, y
Cost
OEM
Revenue
Figure 2. Schematic cost and revenue functions in an empirical
telecom supply chain.
o¤ered, often by buying companies that possess the desired capabilities. These
changes have gradually changed the bargaining position of the EMSs relative to
the OEMs.
Thus it often happens that 2nd tier suppliers deliver and contract standard
products. However, for strategic or customized components, the interrelation with
the OEM could be quite strong both during the development phase and also during
the production phase. While forecasts and planning information are normally sent
from the OEM to the EMS, and from the EMS to the 2nd tier supplier, there
are situations where the OEM and the 2nd tier supplier shortcut this loop with
medium-term and long-term forecasts to counteract adverse supply chain e¤ects,
e.g. bullwhip e¤ect. For the most strategic components the OEM might do this
to secure the supply, sometimes also by …nancially helping the 2nd tier supplier
with speci…c assets, investments and raw material purchase. Being able to trust
the information supplied and the commitment given to coordination is important
for the smooth operation of a dynamic supply chain.
3.3. Heterogeneous Business Logic. As the telecom supply chain evolves,
the business logic of the companies within the supply chain di¤ers and changes over
time. Looking at the three tiers in our model, the cost and revenue functions are
di¤erent, as is illustrated in Figure 2.
The 2nd tier supplier invests in manufacturing capacity that involves building
new production lines or plants (Figure 2). Production is capital intensive and
relies heavily on …xed assets and less on more ‡exible resources such as manpower.
Investment decisions have to be made years before the …nal products reach the
market. Once the investment is made, the production yield is maximized. New
investments to increase capacity can make them temporarily unpro…table until a
new break-even volume is reached. One major risk for a 2nd tier supplier is low
utilization of production capacity; another is that technology or market changes
downstream will make its products and processes obsolete.
In many cases the EMS is not taking much risk regarding volume or speci…c
investments (Figure 2). The major activity is labor-intensive product assembly.
Typically, the OEM reimburses the EMS by a cost-plus contract: cost reimburse-
ment and an additional markup for pro…t. If the EMS has acquired speci…c facilities
from the OEM, the OEM usually guarantees a certain demand level for the …rst
years - the OEM carries the demand risk despite the outsourcing arrangement. A
key performance indicator for many EMSs is inventory turnover rate, thus they
are reluctant to carry inventory of supplies, regardless of who assumes the actual
holding cost.
RISKS AND COORDINATION IN TELECOM 7
The OEM, …nally, has huge investments in technology and product develop-
ment. In order to reach pro…table volumes it is critical to hit the market window
with accurate timing. The market window tends to be rather tight, so payback
time on the investments must be short. The costs of the OEM tend to be rather
volume-independent: once the product development is done, the variable costs are
quite insigni…cant. The OEM faces the risk of inadequate production capacity at
the supplier level. In early 2000, this was the case for many electronic components
in the industry, leading to allocations. Since late 2001 the situation is the opposite:
overcapacity plagues the business. The business logic implies that the OEM desires
to satisfy any realized demand in the narrow market window. He must be able to
deliver as much as possible when the demand hits his most positive forecast to get
payback on early investments. The upstream tiers facing limited capacity and vari-
able costs will not have the same leverage on volume. From the upstream (myopic)
perspective, a conservative capacity assessment lowers the risk and increases pro…t
in case of low demand.
From a decision-making perspective, we may distinguish three stages in the
development of the supply chain: the centralized stage, the transition stage, and
the decentralized stage.
Before the entrance of the EMS, the OEM produced and assembled in-house.
He enjoyed considerable bargaining power toward the upstream supplier, by virtue
of size and privileged information.
As the OEM outsources the assembly process to the EMS, the roles change
gradually. Initially, the OEM gains an exceptional advantage by retaining full
cost information on the EMS, yielding bargaining power but rejecting the actual
production responsibility.
In the third stage, the OEM’s cost information on the upstream tiers is weak-
ened. This naturally a¤ects its authority to coordinate and allocate supply chain
rents. Depending on the upstream economies of scale and product di¤erentiation,
the relative power positions between the OEM and the EMS may alter. Decision
making in this stage is conducted in a decentralized manner, where each member
of the supply chain independently reviews coordination initiatives.
The organizational and informational evolution in the industry has not been
accompanied by a corresponding development of the contractual coordination mech-
anisms. Our research attempts to highlight the implications for supply chain per-
formance of such approaches.
4. Model
The model structure has three participants, an OEM (or retailer) R serving
the …nal market, an EMS factory F (or …rst tier supplier) assembling the product
on order from the OEM, and a (second tier) supplier S providing components on
…xed orders from the factory. Production and sales take place during two periods
t = 1; 2: The modeled chain re‡ects the current institutional structure in the ver-
tically separated chains in the telecommunications industry, where manufacturing
operations are frequently outsourced and the organizational distance to the …rst-
tier suppliers is often long. Only a single product and a single …rm in each echelon
are considered; thus there is no possibility to implement tournaments or yardstick
regimes. The OEM is a price taker with a price p on a competitive market with
stochastic demand D. The EMS unit price w
t
= (1 +®) v
t
in period t is based on
8 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
•Decentralized
coordination.
•Inherited centralized
coordination.
•Centralized coordination.
•OEM loses contact with
2
nd
tier supplier.
•OEM loses visibility of 2
nd
tier supplier.
•OEM has information of
2
nd
tier’s cost structure.
•OEM has obsolete
information of EMS’s cost
structure.
•Processes, management
and resources the same.
àOEM knows EMS’s cost
structure.
•OEM knows cost
structure of factory.
•EMS getting stronger. •OEM still strong. •OEM strong vs. supplier.
•EMS establishing new
processes, management and
resources.
•Assembly/factory
outsourced.
•Assembly/factory in-house.
•Decentralized
coordination.
•Inherited centralized
coordination.
•Centralized coordination.
•OEM loses contact with
2
nd
tier supplier.
•OEM loses visibility of 2
nd
tier supplier.
•OEM has information of
2
nd
tier’s cost structure.
•OEM has obsolete
information of EMS’s cost
structure.
•Processes, management
and resources the same.
àOEM knows EMS’s cost
structure.
•OEM knows cost
structure of factory.
•EMS getting stronger. •OEM still strong. •OEM strong vs. supplier.
•EMS establishing new
processes, management and
resources.
•Assembly/factory
outsourced.
•Assembly/factory in-house.
~1995 ~2000
S
2
nd
tier
F
EMS
R
OEM
S
2
nd
tier
S
2
nd
tier
F
EMS
R
OEM
OEM
F R
Figure 3. Summary of organizational and information evolutions
in the telecom industry between 1995 and 2000.
a veri…able and observable input cost v
t
and a negotiated markup ®. The supplier
produces at cost function C(Q) and charges the factory a unit price v
t
for the
common order size Q at period t.
Below, we will elaborate on the informational assumptions regarding the three
…rms.
The continuously di¤erentiable, invertible and monotonous probability distrib-
ution function F
) of the demand, with distribution function f ) and the price pare common knowledge.
The OEM may transmit a forecast s to signal the expected demand for the
upcoming period. Under any informational assumptions, the OEM sequentially
decides upon the order quantities for the two periods, Q
1
and Q
2
. The forecast is
observable but not veri…able in court. The OEM maximizes the two-period echelon
pro…t U
R
;
U
R
) = (p ¡w1
) Q
1
+ (p ¡w
2
) Q
2
¡p
Ã
Z
Q1
0
F (x
1
) dx
1
+
Z
Q2
0
F (x
2
) dx
2
!
For notational convenience, denote by G(Q) =
R
Q
0
F (x) dx the expected num-
ber of units not sold at production level Q: The EMS factory maximizes the horizon
objective function U
F
) =P
2
t=1
®v
t
Q
t
. We assume that the …xed costs for F are
normalized to zero: Later, we will introduce an additional informational assumption
on behalf of the factory that will render its role more interesting.
The supplier S has private information about its cost function C(Q) and in-
vestment opportunities. At the beginning of the …rst period, the supplier learns the
RISKS AND COORDINATION IN TELECOM 9
Figure 4. Timeline for the investment-production game with
OEM coordination and asymmetric information.
amount A it would take in order to achieve a decrease of marginal cost c
0
Q in the
next period. The initial investment amount is a draw from a uniform distribution
with support on
£
A; A
¤
: The supplier has the …rst period cost function
C
1
(Q
1
; ±
1
) =
1
2
cQ
2
1
+±
1
A
where Q
1
is the actual production ordered in the …rst period, and ±
1
= 1 if an
investment of size A is undertaken in period 1 and ±
1
= 0 otherwise. The second
period cost function C
2
(Q
2
; ±
1
) is
C
2
(Q
2
; ±
1
) =
1
2
(c ¡±
1
c
0
) Q
2
2
where ±
1
is the indicator of previous period investments and Q
2
is the second
period’s realized production. The supplier maximizes the horizon echelon pro…t
U
S
;
U
S
) = v1
Q
1
+v
2
Q
2
¡C
1
(Q
1
; ±
1
) ¡C
2
(Q
2
; ±
1
)
Fixed costs at any stage are ignored, as the focus is on the pro…t contribution
of a particular decision.
4.1. Production Game. The decisions, production and information exchange
are modeled as a two-period game, which is structured below. We ignore discount-
ing e¤ects, assuming that the values are given in real terms or that the interval is
limited. However, this implies no loss of generality. The decisions in the model,
illustrated in Figure 4 are made in the following order, which has importance with
respect to the contracts derived.
(1) S draws and observes initial investment A from
£
A; A
¤
:
(2) R proposes a mechanism M
1
(w
1
; v
1
; Q
1
; X
1
) to S and F.
10 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
(3) S accepts or rejects M
1
(4) F accepts or rejects M
1
(5) (If M
1
accepted) R may announce (non-veri…ably) a demand forecast s
for period 2.
(6) (If M
1
accepted) S may announce (non-veri…ably) a sunk investment of
H
(7) (If M
1
accepted) S decides on investment ±
1
2 f0; 1g
(8) (If M
1
accepted) S and F produce Q
1
(9) (If M
1
accepted) Demand is revealed x
1
(10) (If M
1
accepted) Payouts to all for period 1 according to M
1
(w
1
; v
1
; Q
1
; x
1
)
(11) R proposes a mechanism M
2
(w
2
; v
2
; Q
2
; X
2
) to S and F.
(12) S accepts or rejects M
2
(13) F accepts or rejects M
2
(14) (If M
2
accepted) S and F produce Q
2
(15) (If M
2
accepted) Demand is revealed x
2
(16) (If M
2
accepted) Payouts to all for period 2 according to M
2
(w
2
; v
2
; Q
2
; x
2
)
Next, we will model the decentralized scenario as a function of (i) bargaining
power (or coordination role) and (ii) information access/asymmetry. In the …rst
case, the single-period price-quantity contract under full and OEM coordination
will be studied as a benchmark. The second case adds asymmetric information on
the supplier cost/capacity decision, still under OEM coordination. The third case
shifts the coordination role to the EMS, while maintaining the private information
on the supplier side and the single-period price-only contracts. The problems are
so chosen as to illustrate the outline of organizational development in the telecom
industry, where the roles and information access have been gradually evolving.
4.2. Centralized model. In the centralized solution, the three …rms coordi-
nate to achieve chain pro…t maximization. Note that the parameter ® = 0 in the
integrated scenario is set to avoid double marginalization. Instead, the joint pro…t
contribution is distributed ex post in some arbitrary way that we represent with a
triple (¼
R
; ¼
F
; ¼
S
) later in the propositions: The joint investment is the result of
the horizon control problem
max
Q1;Q2;±1
E[¦(Q; ±
1
)] =
2
X
t=1
[p (Q
t
¡G(Q
t
)) ¡C
t
(Q
t
; ±
1
)]
which for the investment case simpli…es to a two-period newsboy problem
E[¦
; ±)] = p (Q1
¡G(Q
1
)) ¡
1
2
cQ
2
1
¡A±
1
+p (Q
1
¡G(Q
2
)) ¡
1
2
(c ¡±
1
c
0
) Q
2
2
= p (E[X
1
jQ
1
] +E[X
2
jQ
2
])
¡
1
2
c
¡
Q
2
1
+Q
2
2
¢
+
µ
1
2
c
0
Q
2
2
¡A
¶
±
1
RISKS AND COORDINATION IN TELECOM 11
with the …rst order conditions
¦
Q
1
) = p ¡pF (Q1
) ¡cQ
1
= 0
¦
Q2
; ±1
) = p (1 ¡F (Q
2
)) ¡(c ¡±
1
c
0
) Q
2
= 0
¦
±1
) =1
2
c
0
Q
2
2
¡A
and second order conditions
¡pf (Q
1
) ¡c < 0
¡pf (Q
2
) ¡(c ¡±
1
c
0
) < 0
which indicate a concave function in Q.
Thus, the optimal policy devolves from C (Q
¤¤
t
; ±
¤¤
1
) with the implicitly given
order quantities
F (Q
¤¤
1
)
Q
¤¤
1
=
c
p
F (Q
¤¤
1
(±
¤¤
1
))
Q
¤¤
1
(±
¤¤
1
)
=
c ¡±
¤¤
1
c
0
p
where F (x) = 1 ¡F (x). Obviously, Q
¤¤
2
(0) = Q
¤¤
1
; thus the argument can be
suppressed from Q
¤¤
2
= Q
¤¤
2
(1) : To avoid trivial or atypical solutions in the game,
we introduce the assumption of a non-trivial investment policy:
1
2
c
0
(Q
¤¤
2
)
2
< A.
Let
e
A
¤¤
=
1
2
c
0
(Q
¤¤
2
)
2
denote the highest acceptable investment cost under order
quantity Q
¤¤
2
: The investment decision ±
¤¤
1
= 1 is now governed by the decision rule
A ·
e
A
¤¤
; else investment is not undertaken, ±
¤¤
1
= 0: The overall integrated channel
pro…t contribution SC
¤¤
is calculated as
SC
¤¤
= p (Q
¤¤
1
¡G(Q
¤¤
1
)) ¡
1
2
cQ
¤¤2
1
+
Z e
A
¤¤
A
½
p (Q
¤¤
2
(1) ¡G(Q
¤¤
2
(1))) ¡
1
2
(c ¡c
0
) Q
¤¤2
2
(1) ¡a
¾
f (a) da
+
Z
A
e
A
¤¤
½
p (Q
¤¤
2
(0) ¡G(Q
¤¤
2
(0))) ¡
1
2
cQ
¤¤2
2
(0)
¾
f (a) da
=
·
p (Q
¤¤
1
¡G(Q
¤¤
1
)) ¡
1
2
cQ
¤¤2
1
¸
Ã
2A¡
e
A
¤¤
¡A
A¡A
!
+
·
p (Q
¤¤
2
¡G(Q
¤¤
2
)) ¡
1
2
(c ¡c
0
) Q
¤¤2
2
¡
1
2
³
e
A
¤¤
+A
´
¸
Ã
e
A
¤¤
¡A
A¡A
!
In the centralized model, the forecast s and the investment information H are
truthfully disseminated to all players and the potential expected pro…t of higher
accuracy would be added to the expected payo¤.
4.3. OEM Coordination with Full Information. We now turn our at-
tention to the decentralized case where the decision makers behave strategically
with respect to their information and decision opportunities. A key element in the
modeling is which player, if any, has the leading role. In the particular case of
telecommunications, we can empirically observe that the OEM has the direct link
12 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
to the market, the product de…nition and a prehistory as a vertically integrated
…rm. Thus, it is natural to allocate the decision initiative to the OEM. Initially, we
pass through the results assuming perfect information. For this model, all signaling
options are ignored.
Although the OEM observes ex post the investment decision ±, the perfor-
mance of the supply chain will heavily depend on its ability to contract ex ante on
the revealed information. We …rst state the most favorable case, where the OEM
makes the …rst-period contract contingent on the ex post investment observation.
The leverage for the supplier is now limited to rejection of single-period contracts,
which forces the OEM to fully internalize the investment. In e¤ect, this extreme
coordination achieves integration with some limits to the pro…t sharing mechanism.
Proposition 1. Under perfect information and reservation utilities normalized
at zero for S and EMS, the OEM may implement the …rst-best solution Q
¤¤
t
by
o¤ering the price-quantity single period contracts contingent on ±
(i) The investment is imposed ¡
¤
= 1 i¤
1
2
c
0
(Q
¤¤
2
(1))
2
¸ A else ¡
¤
= 0
(ii) The …rst-best production Q
¤¤
1
= F
¡1
³
p¡c
p
´
; Q
¤¤
2
(¡
¤
) = F
¡1
³
p¡c+c
0
¡
¤
p
´
(iii) No EMS rent ®
¤
= 0,
(iv) Full reimbursement of investment
v
¤
t
(¡; ±) =
8
>
<
>
:
1
Q
¤
1
C
1
(Q
¤¤
1
; ¡) for t = 1, ± = ¡
1
Q
¤
2
(¡)
C
2
(Q
¤¤
2
(¡) ; ¡) for t = 2, ± = ¡
1
Q
¤
1
C
1
(Q
¤¤
1
; 0) otherwise
9
>
=
>
;
(v) Supply chain surplus SC
¤¤
is extracted by the OEM without distortion,
(SC
¤¤
; 0; 0) :
Proof. The control variables of the OEM are v
t
(±) ; ®; Q
t
: The reaction func-
tion for the supplier depends on the net transfer o¤ered, v
t
(±) Q
t
in each period
t;
max
±
fmax fv
1
(±) Q
1
¡C
1
(Q
1
; ±) ; 0g +max fv
2
(±) Q
2
¡C
2
(Q
2
; ±) ; 0gg
A hold-up strategy by the OEM would not be an equilibrium, since the out-
come is a rejected investment in the …rst period and a rejected contract in the
second. Thus, a strategy that postpones the reward to the second period is not im-
plementable due to lack of commitment power. A hit-and-run deviation from the
supplier, to collect the investment premium without undertaking the investment
and then rejecting the second period contract, falls on the observability. The EMS
has no decision and subsequently no rent in this problem. Hence, the policy where
the OEM orders …rst-best, sinks an optimal investment in the …rst period, and the
two other tiers comply with the optimal policy, is a (weak) NE. ¤
Removing the contractability of the investment observation decreases the out-
come space in the problem. The OEM is now constrained to a single-period con-
tract, although the possible investment is observable for the second-period contract.
RISKS AND COORDINATION IN TELECOM 13
Proposition 2. Under perfect information and reservation utilities normalized
at zero for S and EMS, the OEM may implement the third-best solution Q
¤
t
by
o¤ering the price-quantity single period contracts
(i) No investment is undertaken ±
1
= 0
(ii) Production Q
¤
= Q
¤
1
= Q
¤
2
= F
¡1
³
p¡c
p
´
(iii) No EMS rent ®
¤
= 0,
(iv) Full reimbursement of investment
v
¤
t
(±) =
(
1
Q
¤
1
C
1
(Q
¤
1
; 0) for t = 1; 2 ± = 0
1
Q
¤
2
(1)
C
2
(Q
¤¤
2
(1) ; 1) for t = 2, ± = 1
)
(v) Supply chain surplus SC
1
= 2
¡
p(Q
¤
¡G(Q
¤
)) ¡
1
2
cQ
¤2
¢
that is extracted
by the OEM,
¡
SC
1
; 0; 0
¢
:
Proof. The control variables of the OEM are v
t
; ®; Q
t
: The reaction func-
tion stays as above, but the …rst-period revenue is no longer contingent on the
investment,
max
±
fmax fv
1
Q
1
¡C
1
(Q
1
; ±) ; 0g +max fv
2
(±) Q
2
¡C
2
(Q
2
; ±) ; 0gg
The …rst-best solution is generally not implementable, since the supplier would
hit-and-run in the case of ¡ = 1. The OEM could not credibly provide incentives
for the supplier to sink the investment A; since the hold-up strategy dominates
strictly in the second period, where the investment is observed. Thus, the only
NE is where no investment is undertaken, irrespective of cost and the quantities
Q
¤
= Q
¤
1
= Q
¤
2
= F
¡1
³
p¡c
p
´
: The OEM has no incentive to deviate in the absence
of investment. The supplier has no incentive to internalize an investment, nor any
reason to reject the contract.
The OEM collects a constant pro…t contribution of
SC
1
= ¦
1
R
= 2
µ
p (Q
¤
¡G(Q
¤
)) ¡
1
2
cQ
¤2
¶
¤
The results in Proposition 2 are hardly surprising, as perfect information en-
ables OEM to observe, but not necessarily promote, chain coordination. However,
this e¤ect is a general artifact of limited commitment, which, under dominant
strategy
1
, could be addressed by repeated contracting or reputation. In the context
of this presentation, the result merely gives a reference point for the cases under
asymmetric information.
4.4. OEM Coordination with Asymmetric Information. We now intro-
duce asymmetric information on the supplier capacity investment, such that OEM
cannot observe either the outcome of the cost A; or whether the investment has
been undertaken, i.e. ±
1
: In this scenario, the OEM needs either to commit to future
payments, or to accept that the supplier enjoys an information rent on its private
knowledge about the information.
1
In this context, we are not considering mixed strategies for the bargaining game.
14 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
Proposition 3. Under asymmetric information on the investment, single pe-
riod contracts and reservation utilities normalized at zero for S and EMS, the OEM
may implement the second-best solution Q
¤
t
by o¤ering a price only-contract
(i) Production Q
¤
1
= F
¡1
³
p¡c
p
´
and
Q
¤
2
(H) =
8
<
:
F
¡1
³
p¡c
p
´
for H = 0
F
¡1
³
p¡c+c
0
p
´
for H > 0
9
=
;
(iii) No EMS rent ®
¤
= 0,
(iii) No reimbursement of investment v
¤
1
=
1
Q
¤
1
C
1
(Q
¤
1
; 0) ; and
v
¤
t
(H) =
½
v
¤
1
for H = 0
1
Q
¤
2
(1)
C
2
(Q
¤
2
(1) ; 1) for H > 0
¾
(iv) Secret investment H
¤
) = 0(v) Distorted investment is undertaken
±
¤
1
(A) =
½
1 i¤ A ·
1
2
c
0
(Q
¤
)
2
=
e
A
¤
<
e
A
¤¤
0 else
¾
(vi) Supply chain pro…t sharing
¡
¦
2
R
= SC
1
; 0; ¦
2
S
¢
under incomplete rent ex-
traction,
SC
2
= SC
1
+
1
2
h
c
0
Q
¤2
¡
e
A
¤
¡A
i
Ã
e
A
¤
¡A
A ¡A
!
= ¦
2
R
+ ¦
2
S
Proof. The EMS is still without leverage and cannot demand rents, ®
¤
= 0.
The reaction function of the supplier now re‡ects the two decision variables ±; H,
where only the second period contract possibly contingent on H, and the private
information leverage
max
±;H
fv
1
Q
1
+v
2
(H) Q
2
(H) ¡C
1
(Q
1
; ±) ¡C
2
(Q
2
(H) ; ±) ; 0g
The …rst-best solution Q
¤¤
1
; Q
¤¤
2
(±) is not an equilibrium under truthful signal-
ing, since the supplier’s hit-and-run strategy strictly dominates investment. Since
the signal H precedes the second period contract, the supplier has limited use if
it. On the other hand, the OEM may safely conclude ex post that H > 0 implies
a sunk investment to be exploited. This recursively reduces the value of the signal
to nil. The equilibrium outcome is Q
¤
= Q
¤
1
= Q
¤
2
but with higher integrated rent,
since the supplier safely invests under the distorted criterion
A ·
1
2
c
0
(Q
¤
1
)
2
=
e
A
¤
and signal H = 0 in the …rst period. Signalling H > 0 would in the case ± = 1
yield a second period contract Q
¤¤
2
(1) with no pro…t contribution and in the case
± = 0 a rejected contracts and stalled production, Q
2
= 0.
The OEM has no incentive to deviate by o¤ering higher compensation in the
…rst period, since it only transfers rent without warranting future capacity. De-
viation by o¤ering lower contracts in the second period would risk rejection. The
analysis is analogous for the OEM order quantities. The OEM surplus stays at
level ¦
1
R
= SC
1
RISKS AND COORDINATION IN TELECOM 15
The expected supplier surplus follows from the distribution of A
¦
2
S
=
Z e
A
¤
A
µ
1
2
c
0
(Q
¤
)
2
¡a
¶
f (a) da
=
³
e
A
¤
¡A
´
c
0
(Q
¤
)
2
¡
³
e
A
2
¡A
2
´
2
¡
A¡A
¢
¤
We notice here that the OEM with bargaining power limited to the EMS stage
cannot extract all rents from the chain. However, as opposed to the situation with
incomplete contracts under full information, the private information actually allows
the supplier to internalize the investment. Hence, the supply chain performance
actually increases when the OEM accepts the limitations of the contract-horizon
setup and abstains from investment veri…cation. This interpretation follows the
intuition that an overly leveraged party may gain from refraining ex ante of certain
instruments or ex post renegotiation. A regulatory body may increase social welfare
under risks of political capture by using incomplete instruments that limit the
possibility of post contractual opportunism.
A small remark can be made on the timing of the investment signal. If the
supplier could present non-veri…able information on the investment cost before the
…rst-period contract is signed and the investment is made, how would it change
the results? First, the OEM would now face the additional option of using a …rst-
period transfer v
¤¤
1
Q
¤¤
1
+H; followed by a second-period contract v
¤¤
2
Q
¤¤
2
(1) ; which
is …rst-best if H = A ·
e
A
¤¤
: However, the supplier would deviate by reporting H =
0 or A for all A 2
h
A;
e
A
¤
i
and H =
e
A
¤¤
for A 2
h
e
A
¤
; A
i
: For the …rst interval,
the supplier makes a secret investment and internalizes the pro…t. If a private
investment is not pro…table, the supplier signals the maximum public investment
and makes a hit-and-run, rejecting the second-period contract. Thus, changing the
information timing does not bring higher value to the chain, it merely changes the
nature of the distortion. In any case, the signal does not convey any meaningful
information.
4.5. EMS Coordination with Asymmetric Information. To re‡ect the
ongoing changes in the leverage between actors in the telecommunications supply
chain, we study the case with asymmetric information under which the OEM orders,
S invests and EMS holds the bargaining power to suggest and coordinate contracts.
One may perceive this solution as the result of a far-reaching decentralization, from
which the EMS factory has gained more stable or important revenue streams than
the OEM.
We state the properties in a proposition.
Proposition 4. Under asymmetric information on the investment, single pe-
riod contracts and reservation utilities normalized at zero for S and OEM, the EMS
may implement the third-best solution Q
E
t
by o¤ering a price only-contract
(i) Production Q
E
(w
¤
) = F
¡1
³
p¡w
¤
p
´
16 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
(ii) No reimbursement of investment
v
E
t
(H) =
(
1
Q
E
(w
¤
)
C
1
¡
Q
E
(w
¤
) ; 0
¢
for H = 0
1
Q
E
(w
¤
)
C
2
¡
Q
E
(w
¤
) ; 1
¢
for H > 0
)
(iii) Secret investment H
E
= 0
(iv) Distorted investment is undertaken
±
E
1
(A) =
½
1 i¤ A ·
1
2
c
0
¡
Q
E
¢
2
=
e
A
E
<
e
A
¤
0 else
¾
(v) Double marginalization ®
E
t
(w
¤
) =
w
¤
v
E
t
(w
¤
)
¡1 > 0
(vi) Supply chain surplus division
¡
¦
3
R
; ¦
3
F
; ¦
3
S
¢
under incomplete rent extrac-
tion,
SC
3
= 2p
µ
Q
E
¡
w
E
¢
¡G
¡
Q
E
¡
w
E
¢¢
¡
1
2
cQ
E2
¶
+
1
2
³
c
0
Q
E2
¡
e
A
E
¡A
´
Ã
e
A
E
¡A
A¡A
!
Proof. The EMS decision variables are ®; w
t
) and vt
) : The OEM acts adouble pricetaker towards the market and the supply, optimizing the order quanti-
ties Q
t
over the two periods
max
Qt
E[U
R
(Q)] =
2
X
t=1
Ã
(p ¡w
t
)) Qt
¡p
Z
Q1
0
F (x) dx
!
which yields a convex problem with the solutions
Q
E
t
(w
t
) = F
¡1
µ
p ¡w
t
p
¶
The supplier follows the reaction function under asymmetric information, using the
signal H; the investment ± and the veto right,
max
±;H
©
v
1
Q
E
1
(w
1
) +v
2
(H) Q
E
2
(w
2
(H)) ¡C
1
¡
Q
E
1
(w
1
) ; ±
¢
¡C
2
¡
Q
E
2
(w
2
(H)) ; ±
¢
; 0
ª
The EMS now faces the control problem in w
t
(H) ; since v
t
(w
t
; H) is implicitly
given from the participation constraint as v
1
(w
1
) =
1
2
cQ
1
(w
1
) and v
2
(w
2
; H) =
1
2
(c ¡c
0
) Q
2
(w
2
(H)) : However, as in the OEM coordination case with private in-
formation, the equilibrium does not involve any signaling. Instead, the supplier un-
dertakes the investment with the distorted decision rule ±
E
1
= 1 i¤ A ·
1
2
c
0
¡
Q
E
¢
2
=
e
A
E
. Ignoring the signal H and suppressing it as an argument for clarity, we con-
centrate on the case where the EMS applies a uniform policy and lets the supplier
internalize the investment. Simplifying the price signal w = (1 +®) v to the OEM
enables the markup factor ® to fall out of the participation constraint for the sup-
plier,
®
E
(w) =
w
v
1
(w)
¡1
The EMS solves the following program in w for the non-imperative investment case
(4.1) max
w
U
F
(w) = 2
£
wQ
E
(w) ¡C
1
¡
Q
E
(w) ; 0
¢¤
with the …rst order conditions (suppressing the superscript E in the implicit re-
sponse function Q
E
(w) )
Q
¡
w
E
¢
+
¡
w
E
¡cQ
¡
w
E
¢¢
Q
0
¡
w
E
¢
= 0
RISKS AND COORDINATION IN TELECOM 17
and second order condition
2Q
0
¡
w
E
¢
+Q
00
¡
w
E
¢ ¡
w
E
¡cQ
¡
w
E
¢¢
¡c
¡
Q
0
¡
w
E
¢¢
2
< 0
Due to monotonicity of F there exists a unique solution w
¤
: The OEM retains the
expected surplus
¦
3
R
= 2p
¡
Q
E
¡
w
E
¢
¡G
¡
Q
E
¡
w
E
¢¢¢
¡2
¡
1 +®
E
¡
w
E
¢¢
v
¡
w
E
¢
Q
E
¡
w
E
¢
where the EMS supplies the OEM with the optimal marginal cost and keeps the
surplus between the average cost v
E
and the marginal cost w
E
: The supplier surplus
follows from its internalized investment
¦
3
S
=
Z e
A
E
A
µ
1
2
c
0
¡
Q
E
2
¡
w
E
¢¢
2
¡a
¶
f (a) da
=
³
e
A
E
¡A
´
c
0
¡
Q
E
2
¡
w
E
¢¢
2
¡
³
e
A
E2
¡A
2
´
2
¡
A¡A
¢
The EMS retains the pro…t contribution
¦
3
F
= 2®
E
¡
w
E
¢
C
1
¡
Q
E
(w
¤
) ; 0
¢
¤
The EMS coordination implies a further lowering of supply chain performance
due to the superposing of asymmetric information at the supplier stage and double
marginalization at the OEM stage. Limited by the contractual structure, the EMS
is forced to extract the di¤erential between the average and the marginal supplier
cost by a mark-up. However, this impediment serves to discipline the EMS in its
relation with the OEM. The downstreams agent OEM implicitly allocates part of
the supply risk to be anticipated by the EMS, compensated by a lion’s share of the
chain pro…ts.
5. Illustration
To numerically illustrate the model and the …ndings, the three stages are pa-
rameterized as in Table 1. The results are summarized in Table 2, where the
production quantities, the pro…t contribution shares, the total chain pro…t and the
investment cut-o¤ limit are given for the base case and the three other cases. Note
that the EMS coordination performs worse than the no-investment case 1. The dif-
ferences are, however, fairly low with the given parameters: ¡3:4%; ¡1:5%; ¡3:5%
coordination loss relative to optimal coordination, respectively. This should not
be interpreted with respect to the actual telecom scenario, as the relative share of
variable to capacity cost parameters for the simpli…ed scenario has not yet been
validated. Finally, with respect to the investment, one may notice that contracting
problems preclude 50% and 54% of the optimal investments in cases 2 and 3, respec-
tively. This …nding has some anecdotal evidence in the industry, where suppliers
have recently demanded more active participation to undertake speci…c invesments
under uncertainty.
18 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
Parameter
p 500 D N (1000; 200)
c 1.00 A 55000
c
0
0.25 A 10000
Table 1. Parameters for the numerical illustration.
Centralized case Case 1 Case 2 Case 3
Q
¤¤
1
497.02 Q
¤
497.02 Q
¤
497.02 Q
E
481.15
Q
¤¤
2
642.14 Q
¤
2
(H > 0) 642.14 Q
¤
2
(H > 0) 642.14 Q
E
2
(H > 0) 575.24
¦
1
R
249,608 ¦
2
R
249,608 ¦
3
R
1,982
¦
1
F
0 ¦
2
F
0 ¦
3
F
247,364
¦
1
S
0 ¦
2
S
4,844 ¦
3
S
3,985
SC
¤¤
258,285 SC
1
249,608 SC
2
254,452 SC
3
249,347
e
A
¤¤
51,544
e
A 0
e
A
¤
30,879
e
A
E
28,938
Table 2. Numerical results for the centralized case and the cases
1, 2, 3.
6. Conclusion
We have proposed a three-level, two-period supply chain model as a vehicle to
model incentive con‡icts and coordinating contracts in the telecom supply chain.
Basing our structure on the telecom sector, some early results from the model
suggest that the shifting positions of bargaining strength may undermine chain
performance when the contractual structure is misaligned. Especially in the case of
previously integrated …rms that hold bargaining power, but su¤er from asymmet-
ric information and limited contractual commitment, the …ndings would suggest
reallocation of bargaining power and chain risks upstream. From an applied per-
spective, the current analysis points out the shortcomings of price-quantity-only
coordination in the absence of long contracts. However, the bounds on rent extrac-
tion that result from a reallocation of bargaining power to the EMS factory give
rise to some interesting speculations. Our conclusions for the scenarios are sum-
marized in Table 3. Although full coordination can theoretically be obtained with
simple price-quantity contracts, asymmetric information, even in limited three-tier
setting, will constrain the potential performance. A shift of bargaining power from
the OEM to the EMS is, in our model, not an improvement of coordination.
The recent changes in scope and pro…tability in the industry may be not only
the result of market outcomes on the retail or OEM level, but also an a¤ect of
the existing imperfect contractual instruments. Although not covered in the formal
analysis, some other bene…ts in terms of forecast accuracy and investment e¢ciency
may also spring from this seemingly inadequate structure. Advanced contractual
instruments that contract on forecasts and outcomes may, in the perspective of
asymmetric information, force the OEM to internalize investment costs based on
incomplete information. This may imply a downward distraction of forecasts that
may a¤ect supply chain performance. Simple contracts that allocate inventory risks
upstream are not subject to the same biases.
RISKS AND COORDINATION IN TELECOM 19
Information Coordinator Contractual relations Findings
Perfect OEM Single-period, v; w; Q contracts No investment
information Investment observable Even production
Losses of renegotiation
Asymmetric OEM Single-period, v; w; Q contracts Second-best chain pro…t
information S decides privately on ± S internalizes investment
Secret investment
Losses due to renegotiation
Asymmetric EMS Single-period, v; w contracts Third-best chain pro…t
information S decides privately on ± S internalizes investment
OEM decides on Q Losses due to renegotiation
OEM decides on quantity Double marginalization
Table 3. Model structure, relations and some …ndings.
Whether this structure could be improved by more advanced contractual in-
struments, such as two-part tari¤s and minimum quantity contracts, is the subject
of the next analysis stage. In any case, the dynamics and complexity of the un-
derlying business logic in the industry may very well be re‡ected in the optimal
contractual structure. Thus, to pursue the pervasive target of optimal mechanisms
under the true contractual conditions might be largely a futile exercise, whereas a
thorough analysis of stable and implementable mechanisms always conveys value to
industry and academia as well. In all modesty, perhaps this provides a new angle
on integrated supply chain management research.
References
[1] Cachon, G. and M. Lariviere. 1999. Capacity Choice and Allocation: Strategic Behavior and
Supply Chain Performance. Management Science, 45 (8), 1091-1108.
[2] Cachon, G.P. and P.H. Zipkin (1999) Competitive and Cooperative Inventory Policies in a
Two-Stage Supply Chain. Management Science, 45 (7), 936-953.
[3] Cooper, M.C. and L.M. Ellram (1993) Characteristics of Supply Chain Management and the
Implications for Purchasing and Logistics Strategy. The International Journal of Logistics
Management, 4 (2), 13-24.
[4] Cooper, M.C., D.M. Lambert and J.D. Pagh (1997) Supply chain management: More Than
a New Name for Logistics. International Journal of Logistics Management, 8 (1), 1-14.
[5] Eppen, G.D. and A.V. Iyer (1997) Backup agreements in fashion buying - the value of up-
stream ‡exibility. Management Science, 43, 1469-1484.
[6] Feder, B. J. (2001) Motorola to Sell Inner Workings of Cell Phones to Rivals, New York
Times, July 23.
[7] Fine, C.H. (2000) Clockspeed-based Strategies for Supply Chain Design. Production and
Operations Management, 9 (3), 213-221.
[8] Flextronics (2001) Flextronics 2001 annual report.
[9] Hall, R. (1999) Rearranging Risks and Rewards in a Supply Chain. Journal of General Man-
agement, 24 (3), 22-32.
[10] Houlihan, J. B. (1985) International Supply Chain Management. International Journal of
Physical Distribution and Materials Management, 15 (1), 22-38.
[11] Jeuland, A.P. and S.M. Shugan (1983) Managing Channel Pro…t. Marketing Science, 2 (3),
239-272.
[12] Jones, T. and D.W. Riley (1985) Using Inventory for Competitive Supply Chain Management.
International Journal of Physical Distribution and Materials Management, 15 (5), 16-26
20 PER J. AGRELL, ROBERT LINDROTH, AND ANDREAS NORRMAN
[13] La¤ont, J.-J. and J. Tirole (1993) A Theory of Incentives in Procurement and Regulation.
MIT Press.
[14] Lee, H.L. and S. Whang (1999) Decentralized Multi-echelon Supply Chains: Incentives and
Information. Management Science, 45 (5), 633-640.
[15] Li, C-L. and P. Kouvelis (1999) Flexible and Risk-Sharing Supply Contracts Under Price
Uncertainty. Management Science, 45 (10), 1378-1398.
[16] Mentzer, J.T., W. DeWitt, J.S. Keebler, S. Min, N.W. Nix, C.D. Smith and Z.G. Zacharia
(2001) De…ning Supply Chain Management. Journal of Business Logistics, 22 (2), 1-25.
[17] Moorthy, K.S. (1987) Managing Channel Pro…ts: Comment. Marketing Sciences, 6 (4), 375-
379.
[18] Motwani, J., L. Larson and S. Ahuja (1998) Managing a Global Supply Chain Partnership.
Logistics Information Management, 11 (6), 349-354.
[19] Pasternack, B.A. (1985) Optimal Pricing and Returns Policies for Perishable Commodities.
Marketing Science, 4, 166-176.
[20] Porteus, E. and S. Whang (1991) On Manufacturing/Marketing Incentives. Management
Science, 37 (9), 1166-1181.
[21] Riordan, M. (1984) Uncertainty, Asymmetric Information and Bilateral Contracts. Review of
Economic Studies, 51, 82-93.
[22] Skjoett-Larsen, T. (1999) Supply Chain Management: A New Challenge for Researchers and
Managers in Logistics. International Journal of Logistics Management, 10 (2), 41-53.
[23] Tsay, A., A. S. Nahmias and N. Agrawal (1998) Modelling Supply Chain Contracts: A
Review. Chapter 10 in S. Tayur, R. Ganeshan and M. Magazine (Eds.) (1998) Quantitative
Models for Supply Chain Management. Kluwer Academic Publishers.
[24] van Mieghem, J. A. (1999) Coordinating Investment, Production, and Subcontracting. Man-
agement Science, 45, 954-971.
[25] Weinberg, A. (2001) Flextronics seizes an Opportunity, The Industry Standard, Feb 14.
[26] Zsidisin, G. and L. Ellram (1999) Supply Risk Assessment Analysis. Practix, Best Practices
in Purchasing and Supply Chain Management, June, 9-12.
[27] Zsidisin, G., A. Panelli and R. Upton (2000) Purchasing Organisation Involvement in Risk
Assessments, Contingency Plans, and Risk Management: An Exploratory Study. Supply
Chain Management: An International Journal, 5 (4), 187-197.
(Per Agrell) IAG School of Management, Catholic University of Louvain, 1 Place
des Doyens, B-1348 Louvain-la-neuve, BELGIUM
E-mail address, Per Agrell: [email protected]
URL: http://www.poms.ucl.ac.be.
(Robert Lindroth and Andreas Norrman) Engineering Logistics, Department of Indus-
trial Management and Logistics, Lund University, P.O.B. 118, S-221 00 Lund, SWEDEN
E-mail address, Robert Lindroth: [email protected]
E-mail address, Andreas Norrman: [email protected]
URL: http://www.tlog.lth.se
doc_428416647.pdf