INTERNATIONAL JOINT VENTURES: A SPECIAL MULTIPRINCIPAL ...

Global Business & Economics Review Vol.2, No.1 INTERNATIONAL JOINT VENTURES: A COMMON AGENCY PROBLEM

Dr. Ursula F. Ott, Loughborough University, UK Visiting London School of Economics, UK

ABSTRACT

In an international joint venture, two parent companies in two countries called the local and the foreign firm found a new enterprise or form a third player - the IJV-management or the board of directors of the IJV. The triangle structure of this particular enterprise creates a special information problem: the common agency (several principals have one agent) situation. The two parent firms contribute their expertise to the joint enterprise in order to develop a new product or production process, to get market access or to avoid risks. Since the third player acts as their agent in the IJV, both principals are interested in gaining knowledge about the effort of the IJV-management. The agent knows more about the actual costs, the production process, market potential and/or other factors of the management process. Thus, the parents offer incentives to get truthful reports about the ability or effort levels of the management by applying the revelation principle in order to disclose the hidden information or action. The paper studies the translation of the particular IJV information structure into the terms of common agency theory as a special form of the game theoretical application to a three-player mechanism design problem.

I. INTRODUCTION

In the past two decades, key issues like partner characteristics, formation motives, performance and control have dominated the theoretical and empirical research of international joint ventures (IJV) in a variety of academic disciplines (Parkhe, 1996). Especially, the international business literature contains empirical research on these topics. Nevertheless, the theoretical aspects and the application of theoretical tools to the special organizational form of an IJV still seem to be a challenging endeavour. Geringer and Woodcock (1995) developed a principal-agent contractual model (Jensen and Meckling, 1976; Holmstrom, 1979; Myerson, 1982; Guesnerie and Laffont, 1984) for IJVs. They used agency theory in order to combine a contractual approach with a socio-psychological perspective. This was the first time that agency theory was applied to analyse joint ventures and to study cross-cultural behaviour in the international business literature. The authors examined whether the total number of partners participating in a joint venture or the cross-cultural divergence among these partners have an impact on the subsequent performance of the IJV. The principal-agent model was developed to show the efficient structure of joint ventures. Although this paper generated insights about the inherent structure of the IJV, the authors missed to derive the nature of this specific business endeavour. Geringer and Woodcock stressed the presence of two contracts and two principals, yet their focus was on the nature of the contracts involved instead of the multiple hidden information and communication problems. In general, the relationship between the parties of an IJV is described as a ‘contract’ which reflects the efficient organization of information and risk bearing according to Jensen and Meckling (1976). The role model of a principalagent relationship shows that one party delegates work to the other who performs that work. The focus of the theory is on the contingent entity influenced by the principal and agent’s self-interest, bounded rationality and risk aversion. Furthermore, factors like goal conflicts and information asymmetry have an impact on the efficiency and effectiveness of the contract. Geringer and Woodcock distinguished between behaviourally based and outcome-based contracts depending on the capability to monitor the behaviour of the agent and on the measurability of the outcome. As far as the authors introduced agency theory to the IJV issue, an important tool was applied to analyse hidden information, although the refinement into multi-principal agent or common agency problems was not developed. The special common agency problem can be found in situations where an agent serves several principals (Bernheim and Whinston, 1986; Gal-Or, 1991; Martimort, 1990, 1992, 1995, 1996; Stole, 1990). Martimort (1990) and Stole (1990) have developed a theory of common agency or a multi-principal agent theory. This means that each principal offers a contract to the agent. The agent announces his type to the principal, given the other principal’s contract and the agent’s optimal reaction to contract offers, which maximizes her expected payoff. With two principals, if the agent does not announce her type truthfully to principal i, she may also lie to principal j, and perhaps in a different way. Considering an international joint venture as a game of three players with different tasks, capacities and strengths, the focus of this paper lies on showing the theoretical background of this particular common agency problem. As far as two parent firms located in two countries with special regulations are concerned, there exists at least one agent on the other side. The agent has hidden information of the joint venture enterprise. To reveal this information, incentive

schemes and reporting methods could be applied. The theory of multi-principals describes how different incentive mechanisms compete with each other. Regarding the set of relationships between several parent firms and the joint venture, they regulate a set of competing contracts in which the control of the process is shared among these different participants. The design of control and communication channels is crucial to the underlying problem. If the agent (management of the joint venture) knows something relevant to the performance, the contract of the principals (parent firms) needs to specify the terms of the contract to reduce informational disadvantages. Since the general assumption considers the IJV – an independent legal and business entity - as the agent, it is important to stress that the agent does not necessarily be a single person like in the conventional principal agent theory. Thus, the underlying situation deals as well with problems of coalitions. For this reason, in common agency games we can find elements of co-operative and non-cooperative game theory. To start with the inherent problems of an IJV, the following questions are raised: How can hidden knowledge affect common agency problems in IJVs? Are there any possibilities of applying the revelation principle to the management of international joint ventures? The paper is organized as follows. The first section deals with the three archetypes of control and strategy in IJVs. The author applies the principal-agent notions to these frameworks and develops hidden knowledge as well as hidden action scenarios for the underlying cases. Furthermore, the order of the play and the timing of the adverse selection and moral hazard problems show the inherent game theoretical reasoning. The second section considers a formal framework by applying Bayesian games according to Myerson (1991).

II. COMMON AGENCY IN THE IJV

Assume we have two principals or parent firms from two countries to be called local player Plocal and foreign player Pforeign. Furthermore, there exist an agent who is the management of the international joint venture AIJV. This management has special knowledge about the cost structure and other managerial details in the IJV. For this reason, the problem of asymmetric information between the parents and the IJV arises and the common agency framework provides the basis for incentive schemes. To get revelation of hidden information, the principals offer incentives to receive truthful reports from the management of the IJV about its efforts. The three archetypes of IJVs developed by Killing (1982) are the dominant parent venture, the shared management enterprise and the independent joint venture. These categories were used to describe control. Datta (1988) and Lorange and Probst (1987) advanced frameworks for IJVs using these archetypes to show different strategic decisions being based on the control types. The dominant parent strategy is connected to the concept of a wholly owned subsidiary with the dominant partner. The executives from each parent compose the board of directors which plays a subordinate role, whereas the executives of the dominant partner are responsible for the IJVs strategic and operational decisions. This is an appropriate response to the pressures from the host government or to a passive partner seeking solely a financial investment with an acceptable rate of return. Lorange and Probst (1987) stated that this strategy type shows a self-organising system consisting of a dominant parent together with a joint venture adapting to its parent future strategic interests. Splitting the value-creating functions between the dominant partner and the IJV organisation, it often may be difficult to establish a sufficiently simple division of labour. Accepting changes in roles between the two players should lead to a redefinition of the execution of the IJVs’ strategies. The concept of a dominant parent strategy can be found in IJVs having a parents from a developing country and the other one from a developed country. Because of the legal restriction in the host country it might only be possible to have major control rights in the hands of the local partner, whereas the foreign firm will contribute technological and/or management expertise or financial resources.

Figure 1: Dominant parent venture

Principal local (parent firm in country A)

Principal foreign (parent firm in country B)

Agent (IJV-management)

The shared management venture is more common in manufacturing processes where one partner supplies the technological know how and the other firm provides the enterprise with knowledge of the local market. Both partners play an active role (Datta, 1988). Each of the parents continues to have an active interest in the same type of business. This leads to the second archetype of having two inter-related self-organising systems as far as the relationship between firm A and the IJV and between firm B and the IJV are concerned. Focussing on the value-creating process, both parents and the IJV are involved and play an active role by contributing their resources. Given potentially diverging

perspectives, this may lead to conflicts, co-operation or compromises. If one parent, for instance, attempts to shift over more autonomy to the joint venture organization, while the other tries to hold back, the strength of the parents and the bargaining power have an impact on the control scenario of the IJV (Lorange and Probst, 1987). Figure 2: Shared management Principal local (parent firm in country A)

Principal foreign (parent firm in country B)

Agent (IJV-management)

The independent joint venture is relatively free from parental interference and is relatively rare. For Lorange and Probst (1987) the free standing joint venture shows the situation that none of the parents have a strong direct operating future strategic interest in the strategy that the joint venture is pursuing. The structuring of the various tasks in the value-creating process can be done like in an independent company. It is up to the parents to equip the joint venture with a degree of redundancy for having the necessary capacity to be flexible. Datta (1988) points out that numerous interdependencies inevitably exist between the joint venture and the parent enterprises. Furthermore, parents are generally reluctant to give joint ventures the autonomy and freedom to develop independently. Figure 3: Independent joint venture

Principal local (parent firm in country A)

Principal foreign (parent firm in country B)

Agent (IJV-management)

Adverse Selection and Common Agency

The adverse selection situation in an IJV occurs because of incomplete information about the agent’s ability. Figure 4 shows a four stage model of adverse selection in which nature chooses the agent’s ability to perform in the management process of the IJV. In the second step, the parent firms choose their actions by offering their contracts. In the third stage, the agent may accept or reject the contracts and finally in the forth stage he plays the mechanism provided by the contracts. Additionally, a signalling procedure might occur between stage one and two. Figure 4: Adverse selection in the common agency of the IJV Stage 1

Stage 2

Stage 3

ability high

contract foreign AIJV learns type

Nature

one or both contracts

Pforeign

accept Alocal reject

Plocal ability low

Stage 4

contract local

To apply game theoretic reasoning, the structure of a dominant parent IJV shows the importance of providing a mediation plan for the agent from two sources with different bargaining power. Firstly, the local firm does not know whether the agent has the ability to contribute to the value adding process in a research laboratory, production plant or marketing distribution unit (Buckley and Casson, 1996). This means that, depending on the local parent’s contribution to the venture, the agent has knowledge about his own capacity of fulfilling the tasks. For this reason, the mechanism design has the purpose to give the agent the appropriate incentive in order to work for the principals. As far as the local parent can use his local knowledge offering a respective contract under this condition, the foreign parent has difficulties

in judging the agent’s ability. Besides different bargaining power and objective functions, the two principals will offer the agent two contracts according to their special needs. Concerning the contracting situation in a shared management joint venture, both parents add complementary skills to the manufacturing process in the IJV. The agent’s ability to contribute to the success of the enterprise can be seen as hidden knowledge. Solutions may be to make the quality of the product contractible, testing, and monitoring of the production process. Furthermore, special ways of signalling the capability of the management might be applied. Both principals will offer a contract to the agent in order to get the revelation of his type. Different options concerning the acceptability of the contracts are possible such as accepting both, rejecting one, and accepting the other, as well as rejecting both. The same situation will occur in an independent joint venture where both parents have only an interest in receiving returns of their investment. Figure 5: Order of the play – adverse selection

Players: Nature (Dummy player – important for the randomization of the types) Local firm (Plocal) Foreign firm (Pforeign) IJV-management (AIJV) Order of the play: (1) Nature chooses the ability of the agent, unobserved by the principals or by one of them. (2) The agent learns his type. (3) The principals offer the agent one or more contracts. (4) The agent accepts or rejects either one or all contracts. (5) Nature chooses a value for the state of the world according to the distribution G(θ). Payoffs: Depending on the risk aversion of the players and their role in the IJV, the payoff can be profits or return on investment or return on equity. With respect to the IJV-management’s payoff, the utility function has to be an appropriate payoff as far as personal goals matter too.

Moral Hazard and Common Agency

Chi (1996) pointed out the double moral hazard nature of joint ventures and stated that the lack of performance verifiability could lead to a shirking problem. Dealing with the contributions of the parent firms, like technology of the foreign parent and local expertise of the local partner, the output of the venture is dependent on the actions of both firms involved. For this reason, Chi focussed on a situation where both parents of such a contractual relationship are tempted to shirk their responsibilities. The author showed furthermore that a contract under which each party shares a fraction of the output is superior to a contract under which one of them is paid only a flat fee. Thus, the mechanism design of an IJV seems to be important for the actual management process and the behaviour of all parties involved. Another approach to tackle the moral hazard problem in joint ventures was developed by Dasgupta and Tao (1999). The authors showed the importance of sharing equity instead of output because of its ex ante verifiability in order to avoid contractual incompleteness. Like Chi the two authors pointed out the double moral hazard nature of this special business unit. The linkage between the parents and the IJV-management as a control structure contains several variations of the moral hazard problem, and for this reason it is important to show the actual hidden action occurring in IJVs. As far as three players are involved in this game, the different systems of bargaining power have an impact on the actual shirking problem. According to the literature, the concept of the dominant parent joint venture shows the opportunities of embezzling in the following way. The local firm and the IJV board pursue the set of strategies Clocal and CIJV in the management of the joint enterprise. As far as the local partner contributes technological, production or marketing skills into the joint project, the problem occurs firstly in the possibility that the agent (AIJV) may have an incentive to contribute low or high efforts in the value creating process. Since the local firm has access to the joint venture through the board of directors and the actual opportunity to monitor the agent, the incentive scheme may have a design different from the contract the foreign firm will offer the agent. The foreign firm intends to contribute either capital endowment or technology and is interested in a more passive role in the IJV. Nevertheless, the parent firm wants to get returns on the investment. For this reason, the foreign firm will offer the agent a contract to prevent him from shirking. The effort level may be

connected to the output of the joint venture. In the sense of a selling the store agreement, a flat fee contract will leave him to be the residual claimant and a risk taker. From the agent’s perspective, different mechanisms concerning the reporting process and the revelation of his type may be appropriate. Especially, in connection with the different interests and contributions of the principals, the agent may accept or reject the offered contracts. The conflict bearing structure of this hybrid can be seen in the presence of various forms of agreements.

Figure 6: Moral hazard problems in the common agency of the IJV

Stage 1

Stage 2

Stage 3

Stage 4

high

Pforeign contract

accept

Alocal

effort

Nature low

Alocal reject Plocal

The shared management IJV comprises the complementary resources of the parents in the production process. Both partner firms have a strong bargaining power and offer the agent contracts to prevent him from embezzling. The incentive schemes may be related to the parent’s contributions and particular interests. As far as there exists incomplete information about the effort of the agent, the principals can design their contracts in connection with the verified output. Royalties and bonuses in addition to a basic salary may help to create incentives to put the optimal level of effort into the joint venture. Finally, the third archetype called the independent IJV reflects a ‘minor’ interest of the parents in the actual manufacturing or value creating process in comparison with the other two forms. Nevertheless, both parents have an interest in getting a return out of the joint enterprise. For this reason, the principals offer contracts guaranteeing that the agent avoids embezzling. As far as independent IJV is concerned, an equity-sharing contract might be appropriate, since both parents and the IJV are not interrelated in a day-to-day business context. The following order of the play shows the general game theoretical structure of the above-mentioned archetypes, and the plan of actions, and the payoffs on an abstract level. Since the different cases have special features, the conceptualisation of the IJV common agency problem provides a useful tool of analysis. Figure 7: Order of the play – moral hazard

Players: Local firm (Plocal) Foreign firm (Pforeign) IJV-management (AIJV) Nature (Dummy player - person/machine for the randomization of types) Order of the play: (1) The principals offer their contracts to the agent. (2) The agent may either accept or reject one of the contracts or both. (3) The agent will choose his effort level: either ‘shirk’ or ‘do not shirk’. (4) Nature picks the state of the world to be success or failure with a certain probability. Payoffs: Depending on the risk aversion of the players and their role in the IJV, the payoff could be profits or return on investment or return on equity.

III. MECHANISM DESIGN AND COMMON AGENCY IN IJV

The Basic Framework The intention of this paper is to present a general framework for a common agency game of IJV players. When designing a mechanism for the special management process in an IJV, it is necessary to show the rules of this particular game in which at least two principals have to deal with one single agent. The two principals shown above are the local and the foreign firm providing the international joint venture management with expertise through their appointed agents. Thus, the board of directors consists of representatives of both parent firms. It would be much more complicated to consider the management as a coalition of various players, for this reason we assume that the incomplete information arises because of the constellation of the agent, while using just one player – the IJV-management. Problems of hidden action and information occur because of the particular objectives and strategies chosen by the players. Suppose that we have a set of players, N = {Plocal, Pforeign, AIJV}. These players have sets of strategies Ci, communication strategies Bi or a nonempty set of possible social options and outcomes D. The sets of actions consist of, for instance, offering incentive schemes by both partners to the agent and reporting of the agent to the principals. The payoffs ui - the third core element of a game – are dependent on the decisions of the players and the type of the agent. In the context of an IJV it only matters whether the agent is telling the truth or lying. The principals’ types are not important. Thus, only the set of the agent’s type TIJV is required, since the principals’ type is not relevant in this game. Compared to the general concept developed by Myerson (1991), this has an impact on the mediator’s recommendation that will be dealt with in the following paragraphs. Furthermore, the probability function pi maps the randomization of the types. The mediation plan looks like a Bayesian collective choice game or Bayesian game with communication depending whether it is an adverse selection or a moral hazard process. We can use the above-mentioned games and apply the structure to the particular form of an IJV management process.

Adverse Selection To define a mechanism or collective-choice plan as any function µ : T → ∆(D) for every choice d in D and every possible profile of types t in T, the mediation plan µ specifies the probability µ (d | t) that d would be the chosen outcome if t were the combination of types reported by the players. The expected payoff for type ti of player i if all players report honestly is dependent on the players’ payoff, the mechanism and the prior probability function of the true type Ui ( µ | ti) =

∑ ∑

t − i∈T − i

pi (t-i | ti)

µ

(d | t) ui (d,t).

(1)

d ∈D

For the underlying case of an IJV, the agent’s expected utility is based on his true type and on the mechanism in connection with his reported type. His utility function is dependent on the offered mechanism of the principals and his type. As far as the expected utility is a function of the probability, the mechanism and the agent’s payoff dependent on his type and the decisions of all players. UIJV ( µ | tIJV) =

∑ ∑

t − i∈T − i

pIJV (t-IJV | tIJV)

µ

(d | t) uIJV (d,t).

(2)

d ∈D

Whereas the principals’ expected utility is a function of the agent’s type, the offered contracts and utility functions. Ulocal ( µ | tIJV) =

∑ ∑

t − i∈T − i

Uforeign ( µ | tIJV) =

∑ ∑

t − i∈T − i

plocal (t-IJV | tIJV)

µ

(d | t) ulocal (dlocal,t)

(3)

µ

(4)

d ∈D

d ∈D

pforeign (t-IJV | tIJV)

(d | t) uforeign (dforeign,t)

The mechanism is incentive compatible iff honest reporting by all players is a Bayesian equilibrium of the game induced by µ . That is, µ is incentive compatible iff it satisfies the following informational incentive constraints: UIJV ( µ | tIJV) ≥ UIJV* ( µ , sIJV | tIJV)

∀i ∈ N, ∀ ti ∈ Ti, ∀ si ∈ Ti.

(5)

The agent is better off to report his type honestly than to lie which has the following expected payoff UIJV* ( µ , sIJV | tIJV) =

∑ ∑

t − i∈T − i

pIJV (t-IJV | tIJV)

µ

(d | t-IJV, sIJV) uIJV (d,t).

(6)

d ∈D

Compared to the single principal-agent case, the IJV management can either reveal his true type to none, one or both of the principals and thus accept zero, one or two contracts.

Moral Hazard

Defining the following Bayesian game with communication as a general game theoretical concept for the underlying situation

Γµb = ( N, (Bi)i∈N, (Ti)i∈N, (pi)i∈N, ( u i)i∈N).

(7)

Each player is asked to confidentially report his type to the mediator, who is according to Myerson (1991) a person or machine that can help the players communicate and share information, recommends an action to the player after receiving these reports. The recommendations depend on the players’ reports in either a deterministic or random way. For this reason, for any c = (ci) i∈N and any t = (ti) i∈N let µ (c | t) denote the conditional probability that the mediator would recommend to each player i that he should use action ci, if each player j reported his type to be tj. Obviously, µ (c | t) must satisfy the following probability constraints

∑

µ

(c | t) = 1 and

µ

(d | t)

≥ 0 ∀d ∈ C , ∀t ∈ T .

(8)

c∈C

Any such function µ : T → ∆ (C) is called a mechanism for the game Γb with communication. If every player reports his type honestly to the mediator and obeys the recommendations, then the expected utility for type ti of player i from the plan µ would be

Ui ( µ | ti ) =

∑ ∑

t − i∈T − i

pi (t-i | ti)

µ

(c | t) ui (c, t)

(9)

c∈C

where T-i = Χ j∈N-1 Tj and t =(t-i, ti). For the IJV game the expected utility function of the agent would be UIJV ( µ | tIJV ) =

∑ ∑

t − IJV ∈T − IJV c∈C

pIJV (t-IJV | tIJV)

µ

(c | t) uIJV (c, t).

(10)

The agent’s expected utility is dependent on the prior probability function of his true type, on the mechanism reflecting the recommended action under the condition of his type and his utility dependent on his action and choice. The principals’ utility functions are dependent on their offered contracts as well as on the agent’s type:

∑ ∑

Ulocal ( µ | tIJV ) =

pIJV (t-IJV | tIJV)

µ

(c | t) ulocal (clocal, t)

(11)

t − IJV ∈T − IJV c∈C

Uforeign ( µ | tIJV ) =

∑ ∑

pIJV (t-IJV | tIJV)

µ

(c | t) uforeign (cforeign, t).

(12)

t − IJV ∈T − IJV c∈C

As far as the agent could lie about his type or disobey the mediator’s recommendation, it can be assumed that the players’ types are not verifiable by the mediator and that the choice of an action in Ci can be controlled only by player i. For this reason, a mediation plan µ induces a communication game in which each player must select his type report and his plan for choosing an action in Ci as a function of the mediator’s recommendation. This Bayesian game can be explained by the following components of (7), where for each player i,

Bi = {(si,

and

δ

i)

| si ∈ Ti,

δ

i:

→ Ci},

Ci

(13)

u i: (Χ j∈N Bj) × T → R is defined by the equation u i((sj, δ j) j∈N, t) =

∑

µ

(c | (sj)j∈N) ui (( δ j (cj)j∈N, t).

(14)

c∈C

The strategy (si, δ i) in Bi represents a plan for player i to report si and then to choose his action Ci as a function of the mediator’s recommendation according to δ i; so he should choose δ i (ci) if the mediator recommended ci. To consider a Bayesian equilibrium of the induced communication game, it can be stated that a mediation plan µ is incentive compatible iff it is a Bayesian equilibrium for all players to report their types honestly and to obey the mediator’s recommendations. Using the mediation plan µ which is incentive compatible iff it satisfies the following general incentive constraint Ui ( µ | ti ) ∀i∈N,

≥ Ui* ( µ , δ i, si | ti ), ∀ti ∈ Ti,

∀si ∈Ti,

(15) ∀ δ i: Ci

→ Ci.

In general, there could be many different Bayesian equilibria of a communication game. Assuming that the players communicate with each other through a mediator who first asks each player to reveal all of his private information and who then reveals to each player only the minimum information needed to direct his action, in such a situation no player has an incentive to lie or disobey. This can be called the revelation principle (Myerson, 1982). Given a general communication system v:R → ∆(M) and communication strategy set (Bi)i∈N, a Bayesian equilibrium would be a vector σ that specifies, for each i in N, each (ri, δ i) in Bi, and each ti in Ti, a number σi (ri, δ i| ti ), that represents the probability that i would report ri and choose his final action according to δ i (as a function of the message that he receives) if his actual type were ti. If σ is such a Bayesian equilibrium of the communication game induced by the communication system v, then we can construct an equivalent incentive-compatible mediation plan µ by letting

µ

(c | t) =

∑

∑ (∏

( r ,δ )∈B m∈δ −1( c )

σi (ri,

δ

i| ti ))

v(m| r),

∀c∈C,

∀t ∈ T

(16)

i∈N

The above-mentioned revelation principle was applied to the IJV case by considering the disclosure of the agent’s type to the mediator and the option to lie or disobey with respect to the other players - the principals. The strategies are dependent on the IJV’s type revelation. The formal conceptualizations of moral hazard and adverse selection show the

mechanism design approach in connection with a mediator who collects the reports of the players. The agent can either report honestly to one or both principals or even to none of them. This means that either zero, one or two contracts are accepted. To focus on the payoffs of the players, it should be considered that two principals are involved with different incentive Ilocal and Iforeign. The principals, Plocal and Pforeign, are interested in decisions clocal and cforeign leading to the following utilities

ulocal = Vlocal (clocal, t) – Ilocal

(17)

uforeign = Vforeign (cforeign, t) – Iforeign.

(18)

The agent AIJV has the utility uAIJV = VAIJV (clocal, cforeign, t) + Ilocal + Iforeign.

(19)

A Nash equilibrium in contracts is a pair of incentives dependent on the decision of the principals which can be related to the outputs estimates either in sales or technology or produced quantities of the common good. Since the principals base their decisions on their performance measure with respect to the agent’s ability and effort, the mechanisms are dependent on the revelation of the agent’s type. For this reason, equation (21) considers the types.

Nash Equilibrium of contracts = {Ilocal (clocal), Iforeign (cf oreign)}

(20)

or dependent on the type of the agent

Nash Equilibrium of contracts = {(Ilocal (tIJV),clocal(tIJV)), (Iforeign (tIJV), cforeign(tIJV)) }

(21)

where tIJV is the agent’s announcement of type to the principals, such that each principal maximizes the expected payoff, given the other principal’s contract and the agent’s optimal reaction to contract offers. Principal i observes only the report tIJV or the decision ci meant for him.

IV. CONCLUSION

The intention of this paper was to translate IJV characteristics of multi-person decision making into a common agency situation. Firstly, the problem of different parent strategies in combination with the IJV management strategies was derived from the archetypes of control and strategies. Secondly, the game theoretical features were developed and connected to the order of the play. Thirdly, the formalization of the problem showed the game theoretical nature of an IJV and the three-player model of a common agency mechanism. The common agency frameworks were developed to provide a tool for future research in the field of information asymmetries in IJVs. Although there are lots of open research questions, it was important to start with a conceptualization in this respect. Further research has to be considered in the sense of appropriate incentive schemes and refinements of the problem. (a) Financial support from the Austrian Science Fund (grants J1510-OEK and H139-OEK) and Austrian Forschungsgemeinschaft is gratefully acknowledged.

REFERENCES Bernheim, B.D. and Whinston, M.D, ‘Common agency’, Econometrica, Vol. 54, No. 4, 1986, pp. 923-942. Buckley, P. and Casson, M., ‘An economic model of international joint venture strategy’, Journal of International Business Studies, Special Issue, 1996, pp. 849-903. Chi, T., ‘Performance verifiability and output sharing in collaborative ventures’, Management Science, Vol.42, 1996, pp. 93-109. Daspupta, S. and Tao, Z., ‘Contractual incompleteness and the optimality of equity joint ventures’, Journal of Economic Behaviour & Organisation, Vol. 37, 1999, pp. 391-413. Datta, D.K., ‘International joint ventures: a framework for analysis’, Journal of General Management, Vol. 14, No. 2, 1988, pp. 78-91. Fudenberg, D. and Tirole, J., Game Theory (MIT Press, 1991). Gal-Or, E., ‘A common agency with incomplete information’, RAND Journal of Economics, Vol. 22, No. 2, 1991, pp. 274-286. Geringer, J.M. and Woodcock, C.P., ‘Agency costs and the structure and performance of international joint ventures’, Group Decision and Negotiation, Vol. 4, 1995, pp. 453-467. Guesnerie, R. and Laffont, J.-J., ‘A complete solution to a class of principal-agent problems with an application to the control of a self-managed firm’, Journal of Public Economics, Vol. 25, 1984, pp. 329-369. Holmstrom, B., ‘Moral hazard and observability’, Bell Journal of Economics, Vol. 10, 1979, pp. 74-91. Killing, J.P., ‘How to make a global joint venture work’, Harvard Business Review, Vol. 61, No. 3, 1982, pp. 120-127. Lorange, P. and Probst, G.J.B., ‘Joint ventures as self-organizing systems: A key to successful joint venture design and implementation’, Columbia Journal of World Business, Vol. 22, 1987, pp. 71-77. Macho-Stadler, I. and Perez-Castrillo, J.D., An Introduction to the Economics of Information - Incentives and Contracts, (Oxford University Press, 1997). Martimort, D., ‘Multiprincipaux en economie de l’information’, Ph.D. diss, 1992, Toulouse. Martimort, D., ‘Multiprincipal regulatory charter as a safeguard against opportunism’, Working Paper, 1995a, Toulouse. Martimort, D., ‘The Multiprincipal nature of government’, Working Paper, 1995b, Toulouse. Martimort, D., ‘Exclusive dealing, common agency, and multiprincipal incentive theory’, RAND Journal of Economics, Vol. 27, No. 1, 1996, pp. 1-31. Myerson, R.B., ‘Optimal coordination mechanisms in generalized principal-agent problems’, Journal of Mathematical Economics, Vol. 10, 1982, pp. 67-81. Myerson, R.B., Game Theory: Analysis of Conflict (Harvard University Press, 1991). Parkhe, A., ‘International joint ventures’, in B.J. Punnett & O. Shenkar (Eds.), Handbook for International Management Research, pp. 429-459 (Cambridge, MA: Blackwell, 1996). Rasmusen, E., Games and Information (Blackwell, 1995).