23
The Join/Leave/Remain (JLR) decision in collaborative networked organizations Claudia-Melania Chituc a,1 , Shimon Y. Nof b, * a Electrical & Computer Engineering, University of Porto, and INESC Porto/FEUP, Portugal b PRISM Center and School of Industrial Engineering, Purdue University, West Lafayette, IN 47907-2023, USA Received 14 February 2006; received in revised form 10 March 2007; accepted 4 May 2007 Available online 13 May 2007 Abstract Collaborative networked organizations (CNOs) are a powerful mechanism to achieve competitiveness in today’s turbu- lent market conditions. The aim of the work reported in this article is to contribute towards CNO formal modeling and performance analysis from an economic perspective. A recently developed, combined three-dimensional approach for CNO performance assessment, based on the key metrics of cost, payoff and agility is presented, with emphasis on the Join/Leave/Remain (JLR) problem which has been defined in this project. Ó 2007 Published by Elsevier Ltd. Keywords: Collaboration; e-Work; Join/Leave/Remain (JLR) decision problem; Networked organizations; Performance assessment; 3D networked enterprise performance 1. Collaborative networked organizations and their collaborative–competitive environment 1.1. Introduction New forms of business collaboration have emerged as a result of globalization and the rapid progress in information and communication technologies and collaborative e-Work (e.g., Nof, 1994, 2003; Nof, Morel, Monostori, Molina, & Filip, 2006). Collaborative networked organizations (CNOs) consist of geographically distributed organizations that efficiently combine, for a certain period of time, the most suitable set of skills and resources in order to achieve a common goal (e.g., Chituc & Azevedo, 2005). CNOs require adequate tech- nologies and supporting infrastructures, proper management tools, and suitable ways to efficiently monitor and assess their performance. CNOs represent a powerful mechanism to achieve competitiveness in today’s turbulent market conditions by comprising various entities with diverse (complementary) competencies, and symbiotic interests. CNOs usually comprise geographically distributed entities with different cultures, working methods, and technolo- 0360-8352/$ - see front matter Ó 2007 Published by Elsevier Ltd. doi:10.1016/j.cie.2007.05.002 * Corresponding author. Tel.: +1 765 494 5427; fax: +1 765 494 1299. E-mail addresses: [email protected] (C.-M. Chituc), [email protected] (S.Y. Nof). 1 PRISM Center, Purdue University (visiting scholar, 2006). Computers & Industrial Engineering 53 (2007) 173–195 www.elsevier.com/locate/dsw

The Join/Leave/Remain (JLR) decision in collaborative networked organizations

Embed Size (px)

Citation preview

Computers & Industrial Engineering 53 (2007) 173–195

www.elsevier.com/locate/dsw

The Join/Leave/Remain (JLR) decision incollaborative networked organizations

Claudia-Melania Chituc a,1, Shimon Y. Nof b,*

a Electrical & Computer Engineering, University of Porto, and INESC Porto/FEUP, Portugalb PRISM Center and School of Industrial Engineering, Purdue University, West Lafayette, IN 47907-2023, USA

Received 14 February 2006; received in revised form 10 March 2007; accepted 4 May 2007Available online 13 May 2007

Abstract

Collaborative networked organizations (CNOs) are a powerful mechanism to achieve competitiveness in today’s turbu-lent market conditions. The aim of the work reported in this article is to contribute towards CNO formal modeling andperformance analysis from an economic perspective. A recently developed, combined three-dimensional approach forCNO performance assessment, based on the key metrics of cost, payoff and agility is presented, with emphasis on theJoin/Leave/Remain (JLR) problem which has been defined in this project.� 2007 Published by Elsevier Ltd.

Keywords: Collaboration; e-Work; Join/Leave/Remain (JLR) decision problem; Networked organizations; Performance assessment; 3Dnetworked enterprise performance

1. Collaborative networked organizations and their collaborative–competitive environment

1.1. Introduction

New forms of business collaboration have emerged as a result of globalization and the rapid progress ininformation and communication technologies and collaborative e-Work (e.g., Nof, 1994, 2003; Nof, Morel,Monostori, Molina, & Filip, 2006). Collaborative networked organizations (CNOs) consist of geographicallydistributed organizations that efficiently combine, for a certain period of time, the most suitable set of skillsand resources in order to achieve a common goal (e.g., Chituc & Azevedo, 2005). CNOs require adequate tech-nologies and supporting infrastructures, proper management tools, and suitable ways to efficiently monitorand assess their performance.

CNOs represent a powerful mechanism to achieve competitiveness in today’s turbulent market conditionsby comprising various entities with diverse (complementary) competencies, and symbiotic interests. CNOsusually comprise geographically distributed entities with different cultures, working methods, and technolo-

0360-8352/$ - see front matter � 2007 Published by Elsevier Ltd.doi:10.1016/j.cie.2007.05.002

* Corresponding author. Tel.: +1 765 494 5427; fax: +1 765 494 1299.E-mail addresses: [email protected] (C.-M. Chituc), [email protected] (S.Y. Nof).

1 PRISM Center, Purdue University (visiting scholar, 2006).

174 C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195

gies. Although CNO’s partners aim at achieving common business goals and following a common strategy,each member organization has its own goals and strategy, and all these elements complicate CNO coordina-tion and management, which assume a critical role in the CNO’s effectiveness (e.g., Chituc & Azevedo, 2005).

There have been a large number of successful research projects in the area of collaborative networks and thebenefits proclaimed. Most research projects have focused on the technological infrastructure supporting organi-zations’ inter- and intra-operations (e.g., Nof et al., 2006), and some have focused on partner selection in virtualenterprises (e.g., Wu & Su, 2005; Yigin, Taskin, Cedimoglu, & Topal, 2007). Yet analytical tools are still neededto establish: Are the benefits achieved by the whole CNO and by its individual member organizations greater thanthe associated costs? For how many instances and for which time periods is this statement true?

Important issues in CNO effectiveness are the motivation of a given organization to join a certain CNO;quantification of the advantages, or payoffs gained; and the costs associated with participation in a certainCNO. Also, a delicate aspect is the consideration of a given organization to leave its CNO and perhaps joinanother CNO: What are the costs associated with disengagement? How can the CNO be maintained and reor-ganized to yield a positive balance between the benefits (payoffs) and the costs incurred while achieving itsglobal goals and strategic objectives? The potential conflicts between the CNO’s global objectives and thoseof each individual organization are an ongoing source of tension. Hence, the JLR (Join/Leave/Remain) prob-lem is defined. The issues related to the decisions of joining a CNO, remaining in a CNO, or leaving a CNO isnamed in this article as the JLR decision problem.

The JLR problem can be analyzed at different levels: CNO level, individual organization’s level, and inter-organizational business processes level. (There can also be interest in sub-grouping of some organizations withinthe CNO, which is termed sub-CNO, and in multi-CNOs; these levels will be addressed in the future.) Each levelcan be analyzed for each phase of the CNO life: CNO creation, CNO activity, CNO dissolution, and CNO sup-port. It can be noted, however, that such performance analysis is phase- and multi-phase correlated.

So far, relatively scarce research has been pursued on CNO analytical modeling and CNO performancemeasurements, and no study has been published on the JLR decision problem from an economic perspective.Therefore, in this article we endeavor to develop an analytical approach for CNO performance assessment,with emphasis on the JLR decision.

The aim in this article is to develop CNO formal modeling useful for economic performance analysis. Anew combined, three-dimensional approach for CNO performance assessment, based on three key metrics:cost, payoff, and agility is presented, focusing on Join/Leave/Remain (JLR) decision problem.

Issues addressed pertain to:

• Theories and approaches synthesized for CNO representation and formal modeling.• Methods, frameworks, and methodologies synthesized for CNO performance measurement in the context

of a collaborative–competitive economic networked environment.• A three-dimensional model for CNO performance assessment along the following main axes: agility, cost,

and payoff, with attention to the JLR decision.

1.2. Background

A CNO is a collection of heterogeneous organizations with different, but complementary competencies, andsymbiotic interests. They join, effectively combining the most suitable set of skills and resources (e.g., knowl-edge, capital, assets) for a period of time in order to achieve a common objective. They use ICT to coordinate,nurture and support their activities. The term CNO is used, in a broad sense, for other emerging collaborativeorganizational forms with similar properties, such as virtual enterprises and extended enterprises. CNOs oper-ate in a collaborative–competitive economic environment (CCEE), in which various organizations exist andact, collaborate and compete, taking advantage of emerging opportunities under various restrictions. Whena business opportunity is identified by an organization acting as a broker, a subset of these organizations,which can be of any type, e.g., enterprise, government agency, association of enterprises, libraries, and soon, can join their resources to achieve a common objective. This subset forms a CNO. Several CNOs canco-exist simultaneously in a CCEE, and an organization can belong to several CNOs at the same time.

C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195 175

Different terms are used to specify organizations’ and CNOs’ environment, e.g., Virtual Breeding Environ-ment (VBE), Digital Business Ecosystem (DBE). A VBE represents an association or pool of organizationsand their related supporting institutions that have both the potential and the will to cooperate with each otherthrough the establishment of a long-term cooperation agreement and interoperable infrastructure (Camarin-ha-Matos & Afsarmanesh, 2004). DBE paradigm (http://www.digital-ecosystem.org) emerged worldwide asan approach supporting the adoption and development of information and communication technologiesamong SMEs. It aims at improving SMEs efficiency by creating tools that enable seamless connectivity andsupport SMEs’ operation, adding value, and helping them overcome current technical obstacles. Both VBEand DBE concepts have been modeled following biological living systems.

These terms portray individual organizations and CNOs’ business arena. Their focus, however, is different.While VBE and DBE concepts are targeting CNOs’ technical aspects (e.g., technologies, infrastructures), theCCEE term, as applied here, emphasizes the economic nature of organizations and CNOs’ relationships (e.g.,cooperation, competition, and payoffs), which are supported by information and communication technolo-gies. Therefore, the CCEE and CNO terms will be used hereafter where appropriate in the context of thisarticle.

The main research questions that have guided this project are:

(a) CNO formal modeling

1. Which are the most appropriate theories, models, frameworks or approaches that can be useful foreffective CNO formal modeling and performance measurement?

(b) The JLR decision problem for individual organizations

2. Why would a particular organization join a given CNO?3. What are the specific benefits a particular organization would gain by participating in a given CNO?4. For a particular organization, how much would it cost to participate in a given CNO?5. Why would a particular organization remain in a given CNO?

(c) The JLR decision problem for CNOs

6. Is there a significant positive balance between total potential benefits and the costs incurred (includingincreased coordination) relative to each of the member organizations?7. Which are the key dimensions to most succinctly characterize and assess the whole CNO and itsperformance?

A number of areas that need further investigation in this context have been identified by researchers, includ-ing: theoretical definitions for networked organizations, consistent specification paradigms, and formal mod-eling tools. Examples of topics of interest include: formal theories, graph theory, normative models, visualmodeling, theories of complexity, algebraic reference models, ontologies, multi-agent systems, game theory,emerging behavior in complex collaborative networks, modeling social aspects of collaborative networks:social actors networks, and models of integration/models interoperability.

Characteristics, models, and performance measures of collaborative e-Work organizations are describedby Nof (2003, 2006). As pointed out by Alfaro Saiz, Rodriguez, and Ortiz (2005) based on extensive lit-erature review, no framework fully satisfying these requirements is currently available. They have devel-oped a Performance Measurement System for Extended and Virtual Enterprises, built on the conceptsof trust and equity. The focus of the framework developed, however, is not on economic aspects; its basiccomponents are goals, objectives, strategies, plans, policies, critical success factors, and organization,resources, information and function. Wu and Su (2005) analyzed CNO reconfiguration considering man-ufacturing cost and time as decision factors. However, the study is focusing only of partner’s selection invirtual enterprises.

A study combining strategic, operational and financial performance in the virtual organization by Walters(2005) identifies recent contributions towards a planning, performance, and control model for evaluating dif-ferent alternatives and strategic issues. The focus is on the implications of integrating and coordinating theoperating and cash cycles of component firms of virtual business organizations, from operational and strategicperspectives. Although it is one of the few economic approaches for virtual enterprises, Walters’ interest is inperformance planning and management, mainly cash flows, management of working capital and fixed assets,and not on CNO performance assessment.

176 C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195

One European research project that has directly addressed the issue of CNO performance, attempting todevelop performance indicators for CNO, is ECOLEAD (2005). Preliminary results have been presented byCamarinha-Matos and Abreu (2005), with focus on social networks related to a business collaborative envi-ronment. The metrics proposed refer to ‘‘task level’’. An approach for defining performance indicators for pol-icy chains, which are understood as collaborative networks in the government sector that develop, implementand administer public bodies, is presented by Brost, Baaijens, and Meijer (2005). Their approach is based ontwo existing methods, which are tailored to networked organizations: one for defining performance indicatorsfor traditional, non-virtual public organizations, and another for network analysis. Only the methods are pre-sented, however, and the project is illustrated for terrorism defense organizations. Networks for policy chainsrepresent one relevant case of CNO.

The issues of performance appraisal and economic justification for the JLR problem for a CNO are ofutmost importance to designers and managers. The fact that there are difficulties to quantify investmentsin, or benefits from joining, leaving or remaining in a CNO does not mean that one should refrain from tryingto quantify them. On the contrary, it means that traditional methods need to be enhanced and new paradigms,methods, frameworks, and models aiming at CNO performance assessment need to be developed. Such par-adigms, methods, frameworks, and models may not be radically different from those of the existing organiza-tional paradigms and economics, since some measures do not necessarily negate any of the earlier traditionaleconomic measures. In this sense, it is required to synthesize traditional, existing paradigms, methods, frame-works, and models and apply or map them to CNOs, as research which has been pursued in this project with aspecial emphasis on the interesting JLR problem.

The rest of this paper is organized as follows. The next section presents the state of the art of the most rel-evant theories which can serve as a basis for CNO formal modeling and representation. Section three refers toways for performance measurements in the context of a networked environment. A three-dimensionalapproach for CNO performance assessment is then presented, with focus on the JLR decision problem.The paper concludes with a section addressing the needs for further research.

2. State of the art of CNO representation and formal modeling

An extensive literature survey has been conducted. The most relevant theories and approaches which cansupport CNO formal modeling and representation were identified and are described briefly.

2.1. Graph theory

A graph is a set of objects called vertices (or nodes) connected by links called edges (or arcs), which can bedirected (assigned a direction) or not. Any mathematical object involving points and connections betweenthem may be called a graph (Gross & Yellen, 2003). Networks have numerous applications in the practicalside of graph theory, such as network analysis (Novak, Gibbons, & van Rijsbergen, 1999). Structures thatcan be represented by graphs are ubiquitous. Many problems of practical interest can be represented bygraphs, and algorithms to handle graphs have been developed extensively. A graph structure can be extendedby assigning a weight to each edge, and graphs with edges can represent designs such as CNOs. A digraph withweighted edges is, indeed, called a network.

Formal Definition: A common definition of a graph (e.g., Gross & Yellen, 2003) is G = (V,E), where thesets V are vertices (or nodes), and E edges, respectively. Each edge has a set of one or two vertices associatedwith it, which are called its endpoints. An edge is said to join its endpoints.

Graph theory is relevant for an effective analysis of a networked organization and its environment. Themotivation is that a CNO is, in fact, a graph, where the nodes are the organizations, and the edges can bethe relationships among nodes, to which there can be associated different weights and directions.

2.2. Game theory

Game theory is a branch of applied mathematics that studies strategic situations where players choose dif-ferent actions in an attempt to maximize their returns. Similar and related to decision theory, game theory

C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195 177

studies decisions that are made in an environment where various players interact. The goal is choice analysis ofoptimal behavior when costs and benefits of each option are variable, depending upon the choices of otherindividuals.

The games studied by this theory are well defined mathematical objects, a game consisting of a set of play-ers, set of moves (or strategies) available to those players, and a specification of payoffs for each strategyprofile.

When it is presumed that each player acts simultaneously or, at least, without knowing the actions of oth-ers, the game is represented as normal (or strategic form) game, by a matrix illustrating the players, strategies,and payoffs. If players have some information about the choices of other players, the game is usually presentedin extensive form, as a tree, attempting to capture games with some important order. In an extensive repre-sentation, each vertex (or node) represents a point of choice for a player, and the players are specified by anumber listed by the vertex. Lines exiting a vertex represent a possible action for that player, and payoffsare specified at the bottom of the tree.

Significant work in game theory has focused on cooperative game theory, which analyzes optimal strategiesfor groups of individuals (‘‘coalitions’’), presuming that they can enforce agreements between them aboutproper strategies. In this case, the game is a competition between coalitions of players, rather than betweenindividual players. According to Aumann and Dreze (1975), a coalition structure in an n-person game is a par-tition of the set of players. A cooperative game allows the formation of coalitions: players join forces based ona binding agreement. This modeling approach is highly relevant to the concept of CNO.

A cooperative game is given by specifying a value for every (non-empty) coalition in the set V. Mathemat-ically, the game is a function:

V : 2N ! R ð1Þ

mapping from the gains of a set of N coalitions to a set of payments, R. This function describes how muchcollective payoff a set of players can gain by forming a coalition. The players are assumed to choose whichcoalition to form, according to their estimate of the way the payment will be divided among coalition mem-bers. It is assumed that the empty coalition gains nil.

A CNO is a coalition where organizations cooperate, aiming to achieve a common goal. Important prop-erties of game theory that can be useful for CNO analysis are: Monotonicity, super-additivity, and Shapley

value.

(1) According to the monotonicity property, larger coalitions gain more, which can be represented asfollows:

A � B) VðAÞ 6 VðBÞ: ð2Þ

where A, B � N are two disjoint coalitions.(2) If A and B are any two coalitions (e.g., two CNOs), then their joint value is no greater than the sum of

their values (super-additivity property):

VðA [ BÞ 6 VðAÞ þ VðBÞ; ð3Þ

where A and B are disjoint subsets of N (set of players). The interpretation is: if A is a coalition of players whoagree to cooperate, then V(A) describes the total expected gain from this cooperation, independent of what theactors outside of A do. The super additivity condition (of this coalition game V) expresses the fact that col-laboration can only help and can never hurt, depending, of course, on the effectiveness of collaboration. Thisis an important property, especially regarding the motivation of an organization or a group of organizationsto join a network, or to form a new collaborative network.

2.3. Business processes alignment

An approach for CNO analysis based on public (inter-organizational) and private (intra-organiza-tional) business processes has been proposed by Chituc and Azevedo (2005), aiming at business processes

178 C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195

alignment. Earlier, Park and Nof (2003a, 2003b) analyzed how shared processes and proprietary pro-cesses can impact the effectiveness of multi-enterprise collaboration. A business process is considered acollection of activities, whose goal (result) is the output of the business process, generated by using acertain set of resources. In the context of a business networking environment, a CNO can be describedas a network of business processes (Fig. 1) which can be identified, documented, controlled andimproved.

The process-oriented organization overcomes the problems of traditional, hierarchical organizationalstructures where the information must flow up the chain so that decisions can be made only at thetop. In fact, the goal of a process-oriented organization such as a CNO, is to facilitate an ‘‘optimized’’execution of processes. Such a business process oriented approach, however, requires an effective manage-ment methodology and framework supported through a precision performance system to measure effi-ciency (e.g., resources consumed within the business process), and effectiveness (e.g., quantified abilityof a business process to deliver products or services according to their specifications). Such a precisionmanagement system can assure business process alignment; the models analyzed by Park and Nof(2003) enable the design of such alignment. A key challenge is for any particular organization’s privatebusiness processes to be aligned effectively with the activities and public business processes performedby other organizations.

2.4. Petri nets

Petri net (also known as a place/transition net or P/T net) is one of several mathematical represen-tations of discrete distributed systems. As a modeling language, it graphically depicts the structure of adistributed system as a directed bipartite graph with annotations. A Petri net has place nodes, transi-tion nodes, and directed arcs connecting places with transactions (state transitions). Arcs run betweenplaces and transitions. Places may contain any number of tokens. Transitions can fire, that is, execute:when a transition fires, it consumes a token from each of its input places, and produces a token ineach of its output places. A transition is enabled if it can fire (e.g., there are tokens in every relatedinput place).

Formal definition: According to Murata (1989), a Petri net is a 5-tuple, PN = (P, T, F, W, Mo),where

P = {p1, p2, . . .,pm} is a finite set of places, T = {t1, t2, . . ., tn} is a finite set of transitions,F � (P · T) [ (T · P) is a set of arcs (flow relation), W: F! {1,2,3,. . .} is a weight function, Mo:P! {1,2,3,. . .} is the initial marking, P \ T = / and P [ T = /.

Fig. 1. CNO as a collection of public and private business processes (PuBPi, public business process; PvBPi, private business process)(Source: adapted after Chituc and Azevedo, 2005).

C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195 179

A CNO can be viewed as a Petri net, where organizations are place nodes; transition nodes can be, forexample, organizations initiating a certain business process, and the arcs associate places and transitions.For a business process aware CNO, Petri nets can be used to model and analyze business processes (e.g.,van Aalst & van Hee, 1995) and BP management (e.g, van Aalst, 2003; van Aalst, 2004). The mathematicalfoundation of Petri nets supports the precise and unambiguous description of the behavior of the business pro-cesses to be modeled (van Aalst & van Hee, 1995), and due to their graphical nature, Petri net models supportalso business process visual representation. Petri nets and their extensions have proved to be useful also in thecontext of logistics and production control (e.g., Nof, Whinston, & Bullers, 1980; Champagnat, Esteban,Pingaud, & Valette, 1998; van Aalst, 1994). However, Petri nets have certain limitations. For instance, execu-tion of Petri nets is nondeterministic, since multiple transactions can be enabled at the same time. Otherlimitations are the lack of compositionality, a serious deficiency which limits modularity; and lack of localitybecause input tokens of a transition disappear simultaneously, which reduces the model realism. On the otherhand, Petri nets are useful as relatively simple representation of complex network activities, and can serve wellfor certain CNO logical and topological analyses, including the powerful property of Reachability (Chen,2002; Chen & Nof, 2002).

2.5. Gaussian networks

Gaussian networks were initially developed for distributed systems (Hsu, Chung, & Hu, 1996), andapplied by Chen and Nof (2002) and Anussornnitisarn and Nof (2003) to model enterprise networks withtwo potential advantages: cost reduction and network scalability. In a Gaussian network, the nodes areused to represent a computer in a network or a CPU in a parallel computer. All nodes viewed from acomputing point of view are identical or similar, and capable of performing almost the same operations.When applied to CNO, this approach can model well general CNO behaviors under simplifyingassumptions.

Coordination relations, including vertical integration and horizontal integration in an enterprise network,can be modeled and analyzed with a Gaussian cube (Chen, 2002; Chen & Nof, 2002). A useful feature ofGaussian cube networks is that they inherit the property of scalability. According to Anussornnitisarn andNof (2003), by using Gaussian cubes, when a bottleneck is formed in an enterprise node, the enterprise net-work can be reconfigured by adding more edges to the Gaussian cube that the node represents. This propertymeans that the enterprise can request coordinated help from other enterprises in case of insufficient capacity.Based on pilot experiments, results show that under certain conditions, the coordination cost can be compen-sated by the profit earned from lower vulnerability cost.

2.6. Transaction cost economics

Transaction cost economics (TCE) was initiated by Williamson in the mid 1970s (1979,2005), and as analyticaltool is mostly used to explain economic problems where resource specificity plays a key role. Transaction costs donot refer only to the processes of selling and buying; they include other kinds of transaction costs which are rel-evant in the context of CCEE/CNO, such as: search and information costs, which are costs incurred in determin-ing if a product is available in the market; costs associated with the search for a partner in the CCEE which has adesired set of skills and expertise; bargaining costs required to reach an agreement among CNO members, draw-ing up such an agreement; policing and enforcement costs required to ensure that all CNO members adhere to theterms of contracts, and taking legal actions when they do not.

TCE is relevant for CNO economic analysis since it includes costs which are often neglected or under-eval-uated. Several additional costs that are unique to the CNO life-cycle must also be considered.

2.7. Social networks

Social network is a term first coined by Barnes (1954). It is a social structure comprised of actors who can beindividuals or organizations, indicating the way by which they are connected to various social familiarities, rang-ing from casual acquaintance to close familial bounds. Social network theory views social relationships in terms

180 C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195

of nodes (individual actors within the network), and ties (relationships between actors). In its simplest form, asocial network is a map of all the relevant ties between the nodes being studied. These concepts can be representedin a social network diagram, where nodes are the points, and ties are the lines.

The shape of the social network helps determine a network’s usefulness to its individuals. Smaller, tighternetworks can be less useful to their members than networks with relatively loose connection (weak ties) toindividuals outside the main network. More ‘‘open’’ networks, with many weak ties and social connections,are more likely to introduce new ideas and opportunities to their members than closed networks with manyredundant ties (Marr, Schiuma, & Neely, 2004).

Social networks have been applied to examine how companies interact, characterizing informal connectionsthat link executives together, and associations and connections between individual employees at different com-panies. These networks provide ways for companies to gather information, deter competition, and even col-lude in setting prices or policies. Social networks can provide general insight for relationships within CNOs,for instance, during the creation of a CNO.

2.8. Methods and approaches – summary

The above-mentioned theories, approaches, and modeling methods have certain advantages and limitationswhen applied to CNO economic analysis, depending on the purpose and focus. The definition, formal repre-sentation and elements, advantages and limitations, and the field of application for each of these theories andmethods, relative to CNOs are summarized in Table 1.

Of relevance can be also other areas. For instance, micro-economics for understanding how to solveorganization’s problems, including analysis of the market structure relative to the corporation; settingoptimal prices and levels of production. Monopolistic practices can be applied for understanding the net-work and role of certain organization member in the CNO, or acting in a collaborative business environ-ment. Sometimes, certain organization members of a CNO can have a monopolistic control over thenetwork due to the functions they perform within the CNO. With these considerations, a new analyticalapproach has been synthesized to better understand a CNO and its environment, by supporting CNOcharacterization and performance measurement.

A graph theoretic definition of CNO can serve as the basis for CNO understanding, and game theory isused to analyze payoff distribution in support of the JLR decision problem. The analysis is based on acontribution of transaction cost economics, and microeconomics, as illustrated in Fig. 2. The combinationenables integration of the inherent relative strengths of these techniques for answering the research ques-tions defined earlier.

The combined approach assumes that a given CNO has several properties, as follows. A CNO is regarded as adirected multi-graph, which is a finite set of nodes (vertices) and directed edges. The notation (CNO) n will be usedto represent an n-dimensional CNO, where n is the number of nodes (organizations) in the CNO (n finite andn P 2). This assumption can be represented as follows:

Definition 2.1. Collaborative Networked Organization (CNO)n

ðCNOÞn ¼ ðO;E;BP;T Þ ð4Þ

O = set of vertices or nodes, representing the organizations in the CNOE = set of edges, representing directed lines connecting 2 nodes of the CNO, representing the set of rela-tionships between 2 nodes (organizations)BP = set of public business processes of the CNOT = the time of analysisn = the dimension of the CNO, which is the number of organizations participating in the CNO

If n = 0, than (CNO)0 is a null graph, and if n = 1, than (CNO)1 contains only one organization, and a sin-gle organization can be assessed by traditional economic methods.

Table 1Approaches to CNO modeling and formal definition

Theory/model/approach

Definition and basic elements Main strengths forCNO modeling andformal representation

Main limitations for CNOmodeling and formalrepresentation

Main fields of application

1 Graphtheory

– Graph: G = (V, E)– V = nodes– E = edges

– A CNO can berepresented as agraph for effectiveanalyses

– Difficulty in definingtypes of relationshipsamong CNO memberorganizations

Network analysis (Novak et al.,1999), Mathematics andcomputer science (Gross &Yellen, 2003)

2 Gametheory

– A game is a function:V : 2N ! R

– Players; moves or strategies;specification of payoffs foreach strategy profile

– Organizations in aCNO are in factplayers

– Focus on player’spayoff

– Shapley value– Super-additivity

property– Partner selection– JLR problem

– Shapley value can beused for particularcases of CNO

– Not all outcomes canbe evaluated in moneyin the case of CNO

– Several axioms, prop-erties and theorems ingame theory assumethat coalitions consistin disjunctive sets

Economic systems (Aumann,1994), biology, philosophy,computer science system

3 Businessprocessesalignment

Public (inter-organizational) andprivate (intra-organizational)business processes

– A CNO is a collec-tion of public andprivate businessprocesses

– Can be modeled asworkflow processes

– Lack of managementmethodology orframework

– Lack of methodologyor framework to assurebusiness processesalignment

CNO (Chituc & Azevedo, 2005)

4 Petri nets – Tuple (S,T,F,M0,W,K); S =places; T = transitions; F =flow relation; M0 = initialmarking; W = arc weights;K = capacity restrictions

– A (business pro-cess aware) CNOcan be representedas a Petri net forcertain analyses

– Lack of locality– Lack of compos-

itionality

Computer science, networkanalysis (e.g., business processmodeling and analysis: van Aalstand van Hee, 1995); productioncontrol (e.g., Nof et al., 1980;Champagnat et al., 1998; vanAalst, 1994)

5 Gaussiannetworks

– Nodes– Coordination relationships

– Scalabilityconsiderations

– Cost reductionanalysis

– So far designed to ana-lyze distributed comput-ing and collaboratingsystems

Distributed computing systems(Hsu et al., 1996), enterprisenetworks (Chen, 2002;Anussornnitisarn & Nof, 2003),network analysis

6 Transactioncosteconomics

– Particular focus on transac-tion costs

– Strategic costs inCNO creation

– Several other costs areinvolved in a CNOand must be considered

Economic systems (Williamson,1979, 2005)

7 Socialnetworks

– Nodes– Ties

– Relationshipsamong CNOmembers

– Focusing mostly onsocial aspects amongmember organizations

Social science, Internet socialnetworks

C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195 181

Since CNOs are assumed to have a dynamic structure, their analysis and characterization should referto a specific point in time (e.g., a CNO can be composed of 8 organizations at its creation, but duringits evolution phase 2 organizations drop off, changing the CNO to now consist of 6 organizations): atT0 (corresponding to CNO creation): card (CNO) = 8, while at T1 (corresponding to CNO operation):card (CNO) = 6.

Definition 2.2. Collaborative–competitive economic environment (CCEE). A CCEE represents the environ-ment (with its restrictions and opportunities) in which a finite set of organizations (represented as Oi, where1 < i 6 m) collaborate and compete.

CCEE ¼[m

i¼1Oi

n o; 1 < i 6 m

n oð5Þ

CNO Performance Analysis

Graphtheory

Transaction CostEconomics

Game theory

Micro-economics specification

Fig. 2. The combined approach for CNO performance analysis.

182 C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195

where:

Oi = organization having the potential and the will to cooperate with other organizations, and mrepresents the number of organization in the CCEE, where m is finite; in other words: card(CCEE) = m.

The environment in which networked organizations collaborate and compete provides several advanta-ges (e.g., facilitates meetings) and constraints (e.g., marketing channel segmentation, legislation concerningCNO operation, tax payments). Typically, m > 1 because if m = 0, than CCEE is a null set; m = 1 is not anetwork.

These defined concepts are illustrated, in a simplified way, in Fig. 3. The CCEE is formed by ten organi-zations (m = 10) and three collaborative networked organizations. Each organization, member of one or moreCNOs, is characterized by specific relationships in each of the CNOs to which it belongs.

CNO1 CNO2 CNO3

cardðCNO1Þ ¼ 5 cardðCNO2Þ ¼ 2 cardðCNO3Þ ¼ 3ð6Þ

Each organization Oi can belong to more than one CNO (e.g., in Eq. (6) O2 belongs to both CNO1 andCNO2). The CCEE can contain organizations that do not currently belong to any CNO.

Different from game theory coalitions that comprise of disjunctive sets, this view on coalitions states thatany organization Oi can be member of more than one networked organization, and the membership state inone or more CNOs depends on the observation time, T.

O4

O2

O1

O5

O9

O10O3

O8O7

O6

CNO 1

CNO2

CNO3

CCEE(T)

Fig. 3. Illustration of a Collaborative–competitive economic environment at time T.

C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195 183

3. Performance assessment in a collaborative–competitive networked environment

CNOs have several attributes, some quantifiable, such as costs associated with a CNO and with individualmember organizations of the CNO, while others are difficult or fuzzy to quantify, e.g., trust. Literature oncollaboration identifies various motives and benefits associated with the inter-organizational collaboration.Relevant motivations are summarized by Chituc and Azevedo (2005): increased market share and asset utili-zation, enhanced customer service, reduced time to market, reduced time for product development, increasedquality of products, increased skills and knowledge, access to technology, reduced inventory, shared or min-imized risk of failure, increased flexibility, and greater access to resources (e.g., skills, knowledge, technology).All those are usually well beyond what a single, albeit very large organization can accomplish.

CNO successful creation and activity depends, however, on certain basic commonality among its members:common goals; interoperable ICT infrastructures and supporting services enabling communication amongpartners, and supporting their activities and real-time information sharing and flow; standards or commonviews on a number of areas, for instance for describing and orchestrating business process flows across multi-ple systems; mutual trust; a common value system and a common way of performing business. Achieving thesecommon challenges requires adequate reference models and extensive support technologies, a proper manage-rial and technological alignment of the distributed business processes, a proper alignment between partners’individual actions, goals and strategies, common strategic objectives of the collaborative network, and ade-quate tools for performance monitoring and measurement.

CNO effectiveness is the foundation for the JLR decision. An organization or network of organizationsdecides to join, leave, or remain in a CNO only based on the answer to questions such as: ‘‘If I (we) decideto join, remain, or leave the network, will the difference between the total costs and payoffs be in my (our)favor?’’ ‘‘How can such balance be evaluated?’’ ‘‘How is it possible to assess this CNO performance?’’ Toanswer these questions, three main ways for CNO performance assessment have been identified, which are pre-sented in the following sections: performance indicators, benchmarking method, and frameworks for perfor-mance measurement. They are combined in Section four as a three-dimensional approach.

3.1. Performance indicators

Performance indicators are quantitative measures of a specific characteristic or aspect related to CNOs,such as costs. They can serve as a basis for comparison by calculating various indices. Upstream metricsand downstream metrics can be developed. Indicators of CNO performance for individual CNO members’evaluation must consider the strategy and vision of the entire CNO and of the individual organization byitself.

A downstream metric, according to Goranson (1999), is the conventional one, related to benchmarking:Certain performance measures are defined, extracted, and monitored, and then compared to a large bodyof reference measurements. The characteristics of interest can be benchmarked relative to each other. Thistype of measures, however, does not convey knowledge about the internal workings of the entity being mea-sured, nor does it contain any information on how it is possible to achieve improvements. Moreover, sincedownstream metrics assume that the future will resemble the conventional past, they are limited wheneveradaptability to a new evolutionary or revolutionary context is concerned.

Upstream metrics of a process are based on its internal mechanisms, and the understanding of these metricsdepends on understanding the process. There is precedence for this type of metrics in manufacturing, based onengineering analogy applied to physical processes. This method also applies across all the elements in a virtualenterprise. Upstream metrics need not be confused with metrics of leading indicators. Leading indicators onlyprovide warning of impending change assuming the experience base is robust and conditions to be known.There is no indication of why and when significant changes will occur.

A general roadmap for defining performance indicators is presented by Wondergem (2004). This roadmapreflects different basic methods which are validated within a public body, namely, a Dutch police region.According to him, examples of methods for defining performance indicators are Balanced Scorecard method,strategy trees, process and system models, and methods that are based on customer value. Conditions to befulfilled for the definition of effective performance indicators are presented by Brost et al. (2005).

184 C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195

3.2. Benchmarking models

Benchmarking models can be applied to determine how well a CNO is performing compared to others. Allfour types of benchmarking methods can be applied: internal (i.e., benchmarking within the CNO), compet-itive (i.e., benchmarking performance of business processes with competitors); functional (i.e., benchmarkingsimilar business processes within an industry), and generic (i.e., comparing operations or business processesbetween general, unrelated industries).

Benchmarking assumes there is always an opportunity for improving the present situation, and also theexistence of a statistically meaningful sample size (Brewer, 2003; Kumar & Motwani, 1995). With certain indi-cators, however, ‘‘more’’ is not always ‘‘better’’. For example, consider the agility of a CNO (discussed below).Assuming that more units of agility are ‘‘best in class’’ may be wrong, since a CNO can be highly profitablewithout being agile.

The main limitation of benchmarking methods and also of the frameworks for performance measurement isthe total dependence on relevant measurement indicators as reference for analysis and comparison. Only whensuch relevant references do exist, the benchmarking models can be useful.

3.3. Frameworks for performance assessment

Several performance measurement and analysis systems are available for companies, e.g., Balanced Score-card, BSC (Kaplan & Norton, 1996; Murby & Gould, 2005), Performance Prism, PP (Neely, Adams, & Ken-nerley, 2002), Performance Pyramid System, PPS (Lynch & Cross, 1995). They describe variousimplementation processes and practice examples. Indeed, some of the existing frameworks can be appliedto assess CNO performance: The Balanced Scorecard is suggested by Chituc and Azevedo (2005) for CNOmanagement, strategic control, and public and private business processes alignment. Li and Williams(2000, 2002, 2003) explain how to manage networked organizations by the Purdue Enterprise ReferenceArchitecture, PERA.

The above frameworks and methodologies have several limitations when applied to evaluate CNOs. Forinstance, BSC considers the control model and strategy to be mainly top-down (Norreklit, 2000); interac-tions between criteria are not clearly considered (Kanji & Moura e Sa, 2002) and BSC focuses only onresults (Kanji & Moura e Sa, 2002). The BSC approach tends to disregard the influence of external com-petitors and technological advances. PERA and PPS follow a hierarchical approach, which requires eitherto have them complemented with other methods (e.g., cost-benefit analysis or economic value added forBSC), or to develop new frameworks especially designed for CNO performance assessment. In general,such frameworks and methodologies rely on the selection of a number of indicators, and these indicatorsmay have limited value for CNOs.

3.4. Performance assessment – summary

Alternative ways to assess CNO performance have briefly been presented, underlying the need to comple-ment them with new metrics and analytical approaches. The need for appropriate theoretical definitions andconsistent specific paradigms is described also by Camarinha-Matos and Afsarmanesh (2003). The most rel-evant quantitative measures for CNO performance assessment and for the JLR decision are performance indi-cators. Other methods, i.e., benchmarking and reference frameworks for performance assessment, relythemselves on the availability and selection of a set of indicators. Before defining metrics, it is important todefine the dimensions or criteria for CNO performance assessment and the JLR decision problem.

4. Three-dimensional approach for CNO performance analysis and JLR decision support

4.1. Overview

A CNO consists of heterogeneous organizations with symbiotic interests. Their inter-connected and collab-orative relationships can provide opportunities, but also constrains on their actions.

C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195 185

In the context of a collaborative–competitive economic environment, several questions arise: how is it pos-sible to motivate organizations to equally strive for the success of public and private business processes?Which are the most relevant criteria to measure CNO performance? Answers to such questions inform deci-sion makers who are deliberating the JLR problem.

The approach recently developed at the PRISM Center (in collaboration with FEUP) combines three axesfor CNO performance assessment: agility, cost and payoff, as illustrated in Fig. 4b. These three dimensions areanalyzed in more detail in the following paragraphs, with special emphasis on the JLR decision problem. Thefundamental assumption is that the CNO effectiveness determines an organization’s decision to join, leave, orremain in a CNO, or to participate in a certain business process.

CNO performance can be assessed at three levels: organization level, CNO level, and business process level(Fig. 4c). All three perspectives can be analyzed during each phase of a n-CNO life-cycle: CNO creation, oper-ation, and CNO dissolution (Fig. 4a). Performance at organizational level can be assessed by using the indi-cators and tools of ‘‘classic’’ financial economics, and it will not be addressed in this article. At theorganizational level, however, the balance of cost-payoff attained by each individual organization by partic-ipating in the CNO is a relevant component of the JLR problem analysis.

Cost, payoff, and agility are considered as the three key criteria for a JLR decision. The values obtained forthese three dimensions are considered together. If their values are lower than an acceptable threshold value(AL) which assures an accepted level of profitability for the organization or CNO making the JLR decision,that is (A, C, P) > AL(A, C, P), then the organization or CNO may decide to join or remain in the network (orparticipate in a certain business process). Otherwise, the decision is to leave the CNO (or not to participate in acertain public business process) as shown in Fig. 4b.

4.2. Agility

Agility, in the context of this research, is defined based on Goranson’s (1999) definition of an agile enter-prise. Considering that definition, and the aim here, an agile CNO is one that responds to, and ideally benefitsfrom unexpected change, keeping the time, quality and costs at a reasonable level.

Fig. 4. Views on n-CNO performance assessment, with emphasis on the JLR decision problem.

186 C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195

It is important to clarify the difference between a flexible CNO and an agile CNO. The agility of a CNO is afunction of its capacity to respond to an unexpected event u_ei, while the flexibility of a CNO is its ability torespond when an expected event ei occurs. This distinction means that a CNO can be flexible without beingagile. In this context, the agility (A) of a CNO is a measure that indicates how well this CNO can adjust itselfand obtain support from other CNOs or organizations to fulfill its objectives and perform its public businessprocesses in a profitable manner, when un unexpected event u_ei occurs. Agility can be measured at CNO,organizational, or business process level. It is possible for an organization to be agile only internally, whilenot being able to contribute to the overall agility of a CNO. In this case, an organization can be agile consid-ering its strategy, objectives and business processes, but without having any contribution to its CNO’s agility.A CNO can be profitable without necessarily being flexible or agile. However, unexpected events are likely tooccur in an environment where clients often change their demands and expectations, new opportunities con-tinuously arise (e.g., newly identified product niche), and a number of conditions can change, affecting any ofthe organizations comprising the CNO. An agile response, under these circumstances, is clearly a critical fac-tor of CNO’s success.

Agility and its measurement are not a new issue. Authors have discussed different definitions and context ofagility, and different ways to measure it (Dove, 2001; Metes, Gundry, & Bradish, 1998). The focus has beengenerally on enterprise and on system agility. A view of the PRISM Centre (Production Robotics and Inte-gration Software for Manufacturing and Management) on enterprise agility is defined by Huang and Nof(1999). They have further developed the analytic measurement of enterprise agility from two perspectives:business and organizational agility, and operational and logistics agility (Huang, Ceroni, & Nof, 2000). Arecent review of agility and complexity measurement by Areta and Giachetti (2004) reveals that there has beenlittle work on measuring agility in the context of networked organizations. Measuring agility in the context ofvirtual enterprises, Goranson (1999) proposes a two-part metric: characterizing the context in which the agilityis considered, and the response in terms of cost and time. He attempts to model and measure agility in thecontext of networked organizations, based on speech-act theory. Using a graph of entities, with arcs represent-ing speech-acts between those entities, two measures on the graph structure are proposed as agility metrics: thedistance metric based on the number of arcs connected to each node; and the time delay metric based on loopsin the same graph.

In the context of networked organizations, agility must be considered mainly during CNO creation phases(Fig. 4), when organizations unite their resources, and during CNO activity, when unexpected events are likelyto occur. For instance, the decision of a member organization to leave the network will require the CNO toreact with agility and overcome the disruptions to achieve its objectives. Agility may also be highly relevantduring CNO dissolution, for instance, in calculating penalties and compensations, respectively, for the orga-nizations that lose profit from the CNO break-up.

Focusing on public business processes, a metric for agility (A) in the context of a CNO can be defined. Thedefinition here follows the model of conflict resolution and error recovery defined by Huang and Nof (1999)for analyzing the agility of enterprises.

A ¼Pk

i¼1ð1� piÞ þPk

i¼1pi � ri

kþ b; ð7Þ

where:k = the number of public business processes a CNO performspi = the probability of unexpected disturbances to occur, that prevent the CNO from accomplishing its

objectivesri = the probability that the unexpected disturbance can be overcome (recovered from), so that the CNO’s

objectives are achievedb = the total number of supplementary business processes required to overcome (recover from) the unex-

pected disturbances occurred, in order to achieve the CNO’s objectives.In an ideal situation (pi = 0), then A ¼ 1� b

kþb. Correspondingly, if the CNO overcomes all unexpected dis-turbances (that is r = 1), then A ¼ 1� b

kþb.When analyzing agility, it can be viewed as the ‘‘insurance’’ assigned by the CNO to achieve its objectives

despite unexpected delays. The level of necessary agility depends on the probability of unexpected

C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195 187

disturbances to occur and the probability and associated costs to recover from them. Petri nets might be usefulto find the Reachability conditions under state space probabilities of agility and recovery.

For pi = p, and ri = r "i; k = 10 and b = 4, CNO agility, based on its business processes and under variableprobabilities, is plotted in Fig. 5.

The values obtained are specific to the numerical values assumed for the parameters. However, the graphsshow that the agility is increasing when r is increasing (for p constant); it decreases when p is increasing and r isconstant, and the slope is relatively larger. But the results must be interpreted with care since, as previouslymentioned, a higher agility does not necessarily mean that the CNO is more profitable. If b is increasing, A

is automatically decreasing. These experiments have been repeated considering other values for k and b (withp, r 2 {0,1; 0,2; 0,3; 0,4; 0,5; 0,6; 0,7; 0,8; 0,9}), and the results obtained display similar characteristics.

Agility is practically an ‘‘insurance’’ (e.g., Goranson, 1999), and it requires a certain investment, e.g., cost ofagility, Ca. Agility, both in terms of timely capacity and cost, must be considered in any JLR decision, espe-cially because unexpected disturbances are likely to occur. However, each CNO must define a certain level ofacceptability for its agility, A_al, its insurance to successfully achieve its objectives. When A > A_al, it is likelythat an organization or a sub-CNO may decide to leave the CNO, or the entire CNO may decode to dissolve.But the costs implied by setting a certain threshold value for its agility, Ca, should always be considered. Atthe same time, agility by itself is not an adequate measure for a complex JLR decision. Other key criteria mustalso be considered, two of which are analyzed next.

4.3. Costs

The costs incurred in a CNO are the costs associated with all activities and business processes performed bythe CNO during its life-cycle. The costs incurred, especially in comparison with the payoffs attained, representa useful parameter upon which JLR decisions can be based.

The costs associated with CNO creation (C_creation) are mainly the so-called strategic costs, includingcosts of acquiring new technology; costs related to employees’ training on how to use the new technology; search

and information costs (e.g., costs incurred in determining whether a certain product is available on the market);costs associated with identifying a partner in the CCEE who has a set of desired skills and expertise; bargainingcosts, which are costs required to reach an acceptable agreement among CNO members, drawing up appro-priate service agreements; policing and enforcement costs, which are costs of ensuring that all CNO membersfollow the terms of their contracts and agreements, and taking appropriate legal actions otherwise; and costsneeded to assure a desired level of agility.

Costs during CNO operation phase represent, in fact, costs associated with the inter-organizational busi-ness processes, further referred to as C_BP Æ C_BP is the sum of the costs associated with all inter-organiza-tional business processes, as illustrated in Eq. (8).

Fig. 5.for par

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Agility (r=0.1)

Agility (r=0.2)

Agility (r=0.3)

Agility (r=0.4)

Agility (r=0.5)

Agility (r=0.6)

Agility (r=0.7)

Agility (r=0.8)

Agility (r=0.9)

Agility

AL (A) for O1

AL (A) for O3

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 P

CNO agility (A) under variable values for unexpected disturbance (p) and recovery (r) probabilities, and the threshold value (AL)ticular cases O1; CNO3 (following Huang et al., 2000).

188 C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195

C BP ¼Xk

i¼1

C BPi ð8Þ

where k represents the total number of public business processes of the CNO being analyzed, and C_BPi rep-resents the costs associated with the i-th public business process.

The costs associated with CNO dissolution phase, further called C_dissolution refer, for instance, to taxpayments.

It is possible that even after CNO dissolution certain CNO partners will still perform certain functions orroles (e.g., maintenance, upgrades, recycling). These costs will be generically called ‘‘support costs’’ (C_sup-port). The costs of a CNO are, in fact, the sum of the costs mentioned above:

Costs CNO ¼ C creationþ C BPþ C dissolutionþ C support ð9Þ

Generally, C_creation, C_dissolution, C_support can be estimated by using statistical and financial methods.Some difficulties may arise in estimating C_BP due to the business processes complexity and their degree ofuncertainty.

Let pi be the probability of an unexpected event u_ei to occur, and which prevents BPi to be fulfilled. Let Cpi

be the cost generated in case pi is positive (disturbances do occur). Let ri be the probability that the event u_ei

can be corrected (recovered from), so that BPi will be fulfilled, and suppose Cri represents the additional costsof recovery. With these assumptions, the cost of a public business process i (C_BPi) can be modeled as illus-trated in Eq. (10), where Cop_BPi represents the total cost to complete the public business process i:

C BPi

Cop BP i; if pi ¼ 0

pi � Cpi; if 0 < pi 6 1 and ri ¼ 0

Cop BP i þ pi � ri � Cri ; if 0 < pi 6 1 and 0 < ri 6 1

8><>:

ð10Þ

The total cost of all business processes (C_BP) is then illustrated in Eq. (11):

C BP

Pki¼1Cop BPi if pi ¼ 0Pki¼1pi � Cpi

if 0 < pi 6 1 and ri ¼ 0

Pki¼1Cop BPi þ

Pki¼1

pi � ri � Cri if 0 < pi 6 1 and 0 < ri 6 1

8>>>><>>>>:

ð11Þ

where 1 6 i 6 k, and k is the total number of public business processes in the CNO.For pi = 0 (corresponding to the perfect situation when activities occur without the any disrupting situa-

tion), C_BPi = Cop_BPi. Suppose pi = p, ri = r, Cpi¼ Cp and Cri ¼ Cr, Cop_BPi = Cop_BP for "i, 1 6 i 6 k,

then Eq. (11) becomes:

C BP

k � Cop BP if p ¼ 0

k � p � Cp if 0 < p 6 1 and r ¼ 0

k � Cop BPþ k � p � r � Cr if 0 < p 6 1 and 0 < ri 6 1

8><>:

ð12Þ

The objective of any CNO and of any individual member organization is to minimize its costs.Variable values can be assigned to Cop BP , k, r, Cp, Cr, and graphically represented, considering, for

instance, p and r between 0.1 and 0.9. Assuming for illustration that Cop_BP = 200 cost units, k = 10 (corre-sponding to 10 business processes), Cp = 150 cost units, and Cr = 100 cost units, the results obtained aredepicted in Fig. 6.

The values of the costs illustrated in Fig. 6 are specific to the numerical values assumed for the parameters.The example illustrates the increase in costs of the business processes when p is increasing, for a given r. Unex-pected disrupting events determine the increased costs, and recovery from them requires additional costs. Ifunexpected events do not occur (p = 0), the costs have a fixed determined value. As illustrated in Eq. (12)C_BP is dependent on the cost of the public business processes, the cost of disturbance events, and the costof recovering from these disturbances. Obviously, when more disturbances occur and the cost to recover fromthem also increases, the pressure to abandon business processes also increases. In such circumstances, as a

1950

2150

2350

2550

2750

2950

C_BP (p=1) C_BP (r=0.1)C_BP (r=0.2)C_BP (r=0.3)C_BP (r=0.4)C_BP (r=0.5)C_BP (r=0.6)C_BP (r=0.7)C_BP (r=0.8)C_BP (r=0.9)C_BP (p=0)

C_BP

AL (C_BP) for O3

AL (C_BP) for O1

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 P

Fig. 6. The cost of business processes, (C_BP) under variable values for unexpected disturbance (p) and recovery (r) probabilities, and thethreshold value (AL) for particular cases O1; CNO3 (following Huang et al., 2000).

C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195 189

consequence, the pressure to remove an organization (or replace it), or even dissolve the CNO, will also begreater.

4.4. Payoff

To motivate organizations to join a CNO and remain in it implies agreement to share and maintain theirshared skills, resources, assets, and knowledge. Clearly, specific measurable benefits and advantages need to beoffered. The benefits can be of financial nature, e.g., a certain percentage of the profits obtained, royalties.Benefits can also consist of other advantages, such as increased market share, access to limited resources(e.g., skills, technology) well beyond what the organization may currently posses, greater flexibility, increasedagility, achieving together critical mass, patents, etc. It is important to note that the payoff to be generatedneeds to be quantifiable, e.g., by the specific value of royalties, trademarks, patents, licenses. Often, however,it is difficult to quantify certain benefits, e.g., potential process improvement, larger market share domination.The payoff can be expressed in direct financial terms, or in intangible terms, as in company spin-offs and tacitknowledge. The payoff in the context of this article refers to the benefits or reward received by an organizationbased, possibly, on its participation in a CNO.

The payoff can be established during CNO formation and stipulated in the business agreement (contract).An approach can be automated for the calculation of profit shares for participants in non-hierarchical pro-duction networks. Under consideration of incentive and sanction mechanisms, for instance, evaluation andbreak-up phases with calculations made for the maximization of the CNO utility are introduced by Jahn,Fischer, and Zimmermann (2005).

The CNO can be considered successful if its goal has been achieved and the benefits of each member orga-nization are maximized. A target value of payoff can be set, and it can be compared with the actual valueattained during different CNO phases. The benefits derived or anticipated, compared with the associated costs,are, in fact, the inputs for the JLR decision problem: Each organization is striving to maximize its own ben-efits, and the payoff received is the leading motivation to join, remain or leave a CNO.

The rewards of each organization are equal to the payoffs obtained after subtracting the total costs. Thereward of an organization Oi is:

Rewards Oi ¼ Payoff Oi � Costs Oi ð13Þ

A collaborative competitive economic environment (CCEE) is actually what in game theory is meant by acooperative game, in which players, organizations in a CNO, or a subgroup of them, have the right to reachcontractual agreements that are fully binding. For these reasons, game theory is used to analyze the distribu-tion of payoffs among CNO members.

190 C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195

Suppose a CCEE has N organizations, with N being finite. A CNO is simply a subset of N. Its memberorganizations combine their resources in accordance with a business agreement to achieve a common goal pre-sented by a business opportunity. Any subset of N can be formed to achieve a certain objective, hence there are2n possible CNOs including the CNO consisting of all N. The empty CNO consists of no one; and there are 2n-2 CNOs consisting of between 1 and N � 1 organizations. Additionally, different CNOs may join to form anew CNO.

For a CNO, the overall payoff can be defined as a set of payoff vectors that passes a minimal criterion ofacceptability (corresponding in game theory to the core type). A CNO is made up of individual organizations,and according to their mission, their aim is to maximize their profits. A CNO can be defined as a function ofpayoffs y associating the set of organizations O to the set of vectors of payoffs Fm.Definition 4.1. Payoff function in a CNO, y

y: O! Fm where y(Oi) = fi, Oi 2 O, fi 2 Fm, and 1 < i 6 mO ¼

Smi¼1Oi, 1 < i 6 m, where card (O) = m (O has m member organizations).

Fm represents the set of payoffs associated to the CNO, achievable by the organization members of theCNO.

This definition assumes that the payoff is achievable; each organization is expecting a payoff, and the payoffeach member organization obtains, fi is higher for all members than what each organization can obtain byitself. These conditions imply that participation in the CNO is advantageous. Thus, for each payoff fi obtainedby an organization, there is always a payoff f 0i so that f 0i < fi and fi = max (fik), where fik represents the set ofpayoffs achievable by Oi (k < i). This condition is observed in Corollary 4.1.Corollary 4.1. "fi 2 Fm, 9f 0i so that f 0i < fi

A threshold value can be defined for the payoff each organization obtains, fiL which represents the mini-mum value of the payoff acceptable by a member organization to be profitable. When its payoff is fi < fiL thenthe organization Oi may decide to leave the CNO.Definition 4.2. Threshold payoff value, fiL

The threshold payoff value fiL represents the minimum acceptable value for the payoff that enables the orga-nization or the CNO to reach an acceptable (desired) profitability level.

By comparing fi to fiL an organization or CNO can decide whether it is relatively more economical to join,leave, orremain in the CNO (for an organization’s decision), and to join an organization, let it stay in theCNO, or dissolve the CNO (for the CNO’s decision).

4.4.1. Profit allocation following the Shapley value

Assuming that a CCEE can be modeled as a super-additive game, and satisfying the four conditions ofShapley value, then Shapley value can guide the distribution of the payoff to each player (organization in thiscase). By considering Shapley value (Friedman, 1990), the payoff of each player is a weighted average of thecontributions that each organization makes to each of the coalitions to which it belongs. The weights dependon the number of players and the number of members in each coalition. It is a ‘‘fair’’ distribution in the sensethat it is the only distribution with certain desirable properties.

The Shapley value is denoted by U(v), where U(v) 2 Fn and Ui(v) is the Shapley value of the i-th organiza-tion (Eq. (14)).

UiðvÞ ¼XK�N

½vðKÞ � vðK� figÞ� � ðk� 1Þ! � ðn� kÞ!n!

ð14Þ

where card (CCEE) = n, K is a coalition, card (K) = k, and v is defined as in Function (1).Under the assumptions that all achievable outcomes can be divided among the CNO members and that the

set of payoffs represents a set of values, then Fm becomes F, and y: O! F. If each organization of the CCEE isa member of one and only one CNO, then the Shepley value can be applied for the fair distribution of payoffsamong the CNO members.

In practice, the utility (payoff) is not always measurable in financial terms. It is also possible that an orga-nization is a member of several CNOs. That means that the Shapley value must be applied with care when

Fig. 7. Threshold value for agility, cost and payoff, AL(A, C, P), and the JLR decision problem.

C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195 191

analyzing the CNO’s distribution of payoffs. Another limitation is that according to the monotonicity axiom,an organization never loses by participating in a CNO. But in reality, several additional costs are incurred(e.g., to join, to remain) According to the super-additivity axiom in game theory, the larger the coalition,the richer it becomes. But this situation is not always valid with CNOs due to the fact that a growingCNO increases in complexity, e.g., CNO management cost increases. A CNO consists of organizations withdifferent cultures, different geographical distributions, and different aims, all challenging its management andcoordination which are critical for its success. Game theory can, however, be applicable as a guide when select-ing a partner to join a CNO. For example, the organization that yields the highest contribution to a CNO’spayoff will be selected, or to motivate an organization to remain in the CNO, or to leave it.

Agility (A), costs (C) and payoffs (P) need to be considered simultaneously. They are highly inter-related ofCNO performance. Agility measures how the CNO can at all achieve its objectives; it has direct impact on thecosts of the CNO and indirect impact on the CNO payoffs. The capacity of a CNO to recover from unexpectedevents in a timely and cost-effective manner allows it accomplish its objectives. The costs and payoffs providethe basis for cost-benefit analysis. Together, these measures enable an organization and a CNO to determinethe JLR decision. As illustrated in Fig. 4b, in order to join, leave or remain in a given CNO, the set (A, C, P)must be compared with an accepted level value (threshold), AL, which assures the organization’s or CNO’slevel of acceptable profitability. If (A, C, P) < AL(A, C, P) than the organization decides whether to joinor remain in the CNO, or for the CNO, decides whether or not to remain as the same CNO. Otherwise,the JLR decision will have to be, concerning an organization, to leave the CNO, or for the CNO to reorganize,e.g., ask a weak organization to leave (Figs. 5 and 6). Note that usually each organization and CNO sets itsown AL(A, C, P) (Fig. 7), since what is acceptable for one organization or CNO may not be adequate forothers.

5. Conclusions and further research

In the modern economy, organizations, in order to survive and achieve their business goals, need to collab-orate by joining their valuable skills and resources. Collaborative Networked Organizations (CNOs) haveemerged as a valuable instrument to effectively achieve strategic objectives in a dynamically changing costand time constraints, with a high expected level of quality standards and service delivery.

Assessment of CNO performance metrics is a challenge in today’s collaborative–competitive economicenvironment. But such metrics are necessary to project current and required performance capabilities.

CNOs, by their nature, are highly dynamic, reconfigurable entities. These characteristics must be consid-ered when defining CNO metrics to be assessed. Agility, for example, represents the potential to adapt in atimely response to changes. Therefore, agility is considered as mostly a critical metric of dynamic performance,while cost and payoff can be considered as metrics of both static and dynamic performance in the CNO.

Performance indicators, benchmarking methods and assessment frameworks have been developed for CNOperformance measurement and monitoring. Benchmarking methods assume always a reference to ‘‘best inclass’’ results, which may not always portray the CNO reality. Besides, as the frameworks for assessment,

192 C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195

benchmarking methods rely on sets of performance indicators, which have to be developed for CNO assess-ment. Some efforts focused on social networks and task level, while this work has addressed economic aspectsof the collaborative business environment.

This article contributes to the solution of the JLR (Join/Leave/Remain) decision problem.Aiming to define metrics for CNO performance assessment, a brief survey has been presented with the main

theories and models for CNO formal modeling. Their strengths and limitations for representing networkedorganizations are compared in Table 1, including graph theory, game theory, business processes alignment,Petri nets, Gaussian networks, transaction cost economics, and social networks.

Based on this review and the research pursued so far, a three-dimensional approach aiming to define met-rics for CNO performance assessment has been developed. The three axes are: cost (C), payoff (P) and agility(A). They are analyzed considering the probability of unexpected disturbance events to occur, the probabilityto recover from them, and the inter-organizational business processes involved. The three dimensions repre-sent the foundation for the decisions of joining a CNO,remaining in a CNO, or leaving a CNO. This decisionproblem is defined in this article as the JLR decision problem. The solution presented is as follows. Agilitymeasures the ability of the CNO, or an individual organization, to adaptively respond to significant unex-pected events. It can be considered as an insurance investment made by the CNO (or an organization) toachieve its objectives in a profitable manner, which directly impacts on the costs and the profits of the memberorganization and the CNO. The gap between costs incurred and payoffs attained serves as basis for a cost-ben-efit analysis. As illustrated in Fig. 4b, in order to join, leave, or remain in a given CNO, the set of (A, C, P)must be compared with a threshold value, AL, which assures the organization and CNO a level of profitabil-ity. If (A, C, P) < AL(A, C, P) than the organization or CNO will decide to join or remain in the CNO, orretain the member in question in the CNO, respectively. Otherwise, the given organization, or the CNO, willdecide to dissolve their association. The set (A, C, P) can be calculated during each CNO phase, and theexpected values can be compared with the actual performance values.

The analysis results are illustrated with assumed numerical values for agility and costs, and must be inter-preted carefully. The resulting graphs (Fig. 5 and 6) lead to several conclusions. Agility increases when r isincreases (for a constant p). It decreases when p increases (given a constant r) but the slope is relatively larger.The relative costs of the business processes are substantially higher when p increases.

A solution proposed for the distribution of payoffs, based on game theory, can be applied as guideline but ithas some limitation. When applying game theory to model the payoff, transferable utility is considered inmonetary. Often, CNO’s payoffs are difficult to quantify although indirectly they have clear economic value.For example, tacit knowledge, human resources improvement, increased trust of partners. Also, in game the-ory, the Shapley value describes one approach to the fair allocation of gains obtained by cooperating actors.Since within a coalition some actors may contribute more to the coalition than others, the question arises howto fairly distribute the gains among the actors. In other words: how important is each actor to the overall oper-ation, and what payoff can they reasonably expect?

Besides presenting a three-dimensional model for CNO performance assessment, this work underlines theimportance of developing performance indicators for CNO. Benefits of such metrics are of importance for theJLR decision problem both for individual organization and at the CNO level. At CNO level, the performancemetrics are important especially for the coordination and management of the network, typically a challengingtask. Such metrics and decisions require analysis and monitoring of the CNO’s evolution and effectiveness,and its benefits. At the organization’s level, performance measurements are needed to analyze and monitorits performance and determine whether it is justified to participate, or continue participating in a givenCNO. For a CNO, such measures can serve as basis for its decision to eliminate an organization form the net-work. Such performance measures, at business process level are relevant to analyze and monitor inter-orga-nizational business processes execution, and to evaluate which business processes (or organizations) areinefficient and should be improved or removed. At a broker’s level, performance measurements can supportdecisions regarding partner’s selection, e.g., during CNO creation phase, based on its past performance.

Further research on the three-dimensional approach and JLR problem and its variations will be pursued tounderstand economic performance indicators for CNO assessment, and to extend the capabilities of the pres-ent method, including validation by real industry cases. For example, challenging and interesting questionsinclude:

C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195 193

1. Which CNO to join? When several CNOs are relevant for a given organization looking to join a CNO,which of them would be preferred?

2. CNOs joining each other: When a CNO wishes to expand by collaborating with one or more other CNOs,thus organizing a ‘‘super-CNO’’, which CNOs or their sub-sets would be the best candidates?

3. Joining two or more CNOs by an organization becoming the link that combines them for a ‘‘multi-CNO’’ –How can the economic measures be useful for this decision problem?

4. Profitability versus sustainability: In economic terms, can the dimension of Payoffs be replaced by Sustain-ability measures, especially for public sector and non-profit organizations?

5. Penalties: Agreements, contracts, and legal binding protocols with their associated penalties are common inCNO life-cycles. How do these penalties figure in addressing the above questions and even the basic JLRdecision problem defined and analyzed in this article?

6. Topology-based performance: Does the CNO economic performance depend on its network topology interms of looseness, redundancy, and contract-binding terms over time?

Answers to these and related questions about the JLR decision problem and the economic performancemeasurement in CNOs will continue to influence the design and planning of effective CNOs. These answerswill be as interesting and relevant for the member organizations of emerging CNOs.

Acknowledgements

This research continues the long-term enterprise networking and e-Work research sponsored by the PRISMCenter at Purdue University, and research on collaborative networked organizations by the Faculty of Engi-neering at the University of Porto and INESC Porto. The first author thanks Fondcau para a Ciencia e aTecnologia for PhD grant SFRH/BD/19751/2004, and Purdue University and the PRISM Center for guid-ance and support.

References

Alfaro Saiz, J. J., Rodriguez, R., Ortiz, A. B. (2005). A performance measurement system for virtual and extended enterprises. In:Collaborative Networks and Breeding Environments, Camarinha-Matos, L., Afsarmanesh, H., Ortiz, A. (Eds.), Proceedings of the 6thIFIP Working Conference on Virtual Enterprises PRO-VE’05, Springer, pp. 285–292.

Anussornnitisarn, P., & Nof, S. Y. (2003). e-Work: The challenge of the next generation ERP systems. Production Planning and Control,

14(8), 753–765.Areta, B. M., & Giachetti, R. E. (2004). A measure of agility as the complexity of enterprise system. Robotics and Computer-Integrated

Manufacturing, 20, 495–503.Aumann, R. (1994). The Shapley value. Game–theoretic methods in general equilibrium analysis. Dordrecht: Kluwer Academic Publishers.Aumann, R., & Dreze, J. (1975). Cooperative games with coalition structures. International Journal of Game Theory, 4, 217–237.Barnes, J. A. (1954). Class and committees in a Norwegian Island Parish. Human Relations.Brewer, P. (2003). Putting strategy into balanced scorecard, International Federation of Accountants (www.ifac.org).Brost, I., Baaijens, J., Meijer, G. R. (2005). Network analysis of terrorism defence organizations – a network approach for developing

performance indicators. In: Collaborative Networks and Breeding Environments, Camarinha-Matos, L. M., Afsarmanesh, H., Ortiz,A. (Eds.), Proceedings of the 6th IFIP Working Conference on Virtual Enterprises PRO-VE’05, Springer pp. 265–272.

Camarinha-Matos, L. M., & Abreu, A. (2005). Performance indicators based on collaboration benefits. In L. M. Camarinha-Matos et al.(Eds.), Collaborative Networks and Breeding Environments (pp. 273–282). Springer.

Camarinha-Matos, L. M., & Afsarmanesh, H. (2003). Elements of a VE base infrastructure. Computers in Industry, 51(2), 139–163.Camarinha-Matos, L. M., & Afsarmanesh, H. (2004). The emerging discipline of collaborative networks in enterprises and collaborative

networks. Kluwer Academic Publishers.Champagnat, R., Esteban, P., Pingaud, H., & Valette, R. (1998). Petri net based modeling of hybrid systems. Computers in Industry,

36(1–2), 139–146.Chen, J. (2002) Modeling and analysis of coordination for multi-enterprise networks. Ph.D. Dissertation, Purdue University, West

Lafayette, Indiana, USA.Chen, J., Nof, Y. S. (2002). Multi-enterprise networking. In: Proceedings of the International Conference on Manufacturing Systems:

Innovations for the 21st Century, Ann Arbor, MI.Chituc, C. M., & Azevedo, A. L. (2005). Multi-perspective challenges on collaborative networks business environment. In L. M.

Camarinha-Matos et al. (Eds.), Collaborative networks and breeding environments (pp. 25–32). Springer.Dove, R. (2001). Response ability: The language, structure, and culture of the agile enterprise. New York: Wiley.

194 C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195

ECOLEAD (2005) European Collaborative networked Organizations LEADership initiative, (http://www.ecolead.org),.Friedman, J. W. (1990). Game theory with applications to economics (2nd ed.). Oxford University Press.Goranson, H. T. (1999). The agile virtual enterprise: cases, metrics, tools. Quorum Books.Gross, J. L., & Yellen, J. (2003). Handbook of graph theory. CRC Press.Hsu, W. J., Chung, M. J., & Hu, Z. (1996). Gaussian networks for scalable distributed systems. Computer Journal(5), 417–426.Huang, C. Y., Ceroni, J. A., & Nof, S. Y. (2000). Agility of networked enterprises – parallelism, error recovery and conflict resolution.

Computers in Industry, 42, 275–287.Huang, C. Y., & Nof, S. Y. (1999). Enterprise agility: A view from the PRISM lab. International Journal of Agile Management Systems,

1(1), 51–59.Jahn, H., Fischer, M., & Zimmermann, M. (2005). An approach for the ascertainment of profit shares for network participants. In L. M.

Camarinha-Matos et al. (Eds.), Collaborative Networks and Breeding Environments (pp. 257–264). Springer.Kanji, G. K., & Moura e Sa, P. (2002). Kanji’s business scorecard. Total Quality Management, 13(1), 13–27.Kaplan, R., & Norton, D. (1996). The balanced scorecard – translating strategy into action. Boston: Harvard Business School Press.Kumar, A., & Motwani, J. (1995). A methodology for assessing time-based competitive advantage of manufacturing firms. International

Journal of Operation Production Management, 15(2), 36–53.Li, H., & Williams, T. J. (2000). The interconnected chain of enterprises as presented by the purdue enterprise reference architecture.

Computers in Industry, 42, 265–274.Li, H., & Williams, T. J. (2002). Management of complexity in enterprise integration projects by the PERA methodology. Journal of

Intelligent Manufacturing, 13(6), 417–427.Li, H., & Williams, T. J. (2003). Interface design for the purdue enterprise reference architecture (PERA) and methodology in e-Work.

Production Planning and Control, 14(8), 704–719.Lynch, R. L., & Cross, K. F. (1995). Measures up: Yardsticks for continuous improvement (2nd ed.). Cambridge: Blackwell Publishers Inc.Marr, B., Schiuma, G., & Neely, A. (2004). The dynamics of value creation – mapping your intellectual performance drivers. Journal of

Intellectual Capital, 5(4), 312–325.Metes, G., Gundry, J., & Bradish, P. (1998). Agile networking: competing through the Internet and Intranets. Englewood Cliffs, NJ: Prentice

Hall.Murata, T. (1989). Petri nets: Properties, analysis and applications. Proceedings of IEEE, 77, 541–580.Murby, L., & Gould, S. (2005). The Chartered Institute of Management Accounts: Effective Performance Management with the Balanced

Scorecard – Technical Report.Neely, A., Adams, C., & Kennerley, M. (2002). The performance prism: The scorecard for measuring and managing business success.

London: Financial Times Prentice Hall.Norreklit, H. (2000). The balance of the Balanced Scorecard – a critical analysis of some of its assumptions. Management Accounting

Research, 11, 65–88.Nof, S. Y., Whinston, A. B., & Bullers, W. I. (1980). Control and decision support in automatic manufacturing systems. AIIE

Transactions, 12(2), 156–165.Nof, S. Y. (Ed.). (1994). Information and Collaboration Models of Integration. Dordrecht: Kluwer Academic Publishers, NATO-ASI

Series.Nof, S. Y. (2003). Design of effective e-Work: Review of models, tools, and emerging challenges. Production Planning and Control, 14,

681–703.Nof, S. Y. (2006). Collaborative e-Work and e-Manufacturing: Challenges for production and logistics managers. J. Intelligent

Manufacturing, 17(6), 689–702.Nof, S. Y., Morel, G., Monostori, L., Molina, A., & Filip, F. (2006). From plant and logistics control to multi-enterprise collaboration.

Annual Reviews in Control, 30, 55–68.Novak, L., Gibbons, A., & van Rijsbergen, C. J. (1999). Hybrid graph theory and network analysis. Cambridge University Press, ISBN

052161170.Park, N., Nof, S. Y. (2003a), Collaboration and integration of business processes based on shared process and workflow control,

Proceedings of Annual Industrial Engineering Research Conference, IERC-2003’03, Portland, OR, May (CD-ROM).Park, N., Nof, S. Y. (2003b). Design of a transactional workflow controller for the integration of collaborative processes, Proceedings of

International conference on production research, ICPR-17, Blacksburg, VA, August (CD-ROM).van Aalst, W. M. P. (1994). Putting Petri nets to work in industry. Computers in Industry, 25(1), 45–54.van Aalst, W. M. P., van Hee, K. M. (1995). Framework for Business process Redesign. In: Collahan, J.R. (Ed.), Proceedings of the 4th

Workshop on Enabling Technologies: Infrastructures for Collaborative Enterprise (WETICE), pp. 36–45.van Aalst, W. M. P. (2003). Challenges in business process management: Verification of business processes using Petri nets. Bulletin of the

EATCS, 80, 174–198.van Aalst, W. M. P. (2004). Business process management demystified: A tutorial on models, systems and standards for workflow

management. In J. Densel, W. Reising, & G. Rozenberg (Eds.), Lectures on Concurrency and Petri nets. Lecture Notes on Computer

Science (vol. 3098, pp. 1–65). Berlin: Springer-Verlag.Walters, D. (2005). Combining strategic, operational and financial performance in the virtual organization. In: Collaborative Networks

and Breeding Environments, Camarinha-Matos, L.M., et al. (Eds.) Proceedings of the 6th IFIP working Conference on VirtualEnterprises PRO-VE’05, pp. 321–328.

Williamson, O. E. (1979). Transaction-cost economics: The governance of contractual relations. Journal of Law and Economics, 22(2),233–261.

C.-M. Chituc, S.Y. Nof / Computers & Industrial Engineering 53 (2007) 173–195 195

Williamson, O. E. (2005). Transaction cost economics. In Handbook of New Institutional Economics (pp. 41–65). Dordrecht and NewYork: Springer.

Wondergem, B. C. M. (2004). Doelgericht presteren met processen: indicatoren voor proces en resultaat, (LogicaCMG/Politie Noord- enOost Gelderland).

Wu, N., & Su, P. (2005). Selection of partners in virtual enterprises. In Robotics and Computer Integrated Manufacturing, 21, 119–131.Yigin, I. H., Taskin, H., Cedimoglu, I. H., & Topal, B. (2007). Supplier selection: An expert system approach. Production Planning and

Control, 18(1), 16–24.