Interdependencies among projects in project portfolio management
a content analysis of techniques
Lappeenranta University of Technology, Lappeenranta, Finland
Humankind has been investigating interdependencies among alliances and companies for decades. This work investigates resource, technological, and market interdependencies among projects. Interdependencies are an important factor in selecting better projects into a project portfolio. They also help to increase the success rate in the field of project management, where high failure rates prevail. Empirics have shown that interdependency techniques influence the success rate and other aspects of productivity to different extents. Interdependencies are also deeply involved with research and development (R&D) and innovation. This paper presents a literature review of techniques addressing interdependencies. Although research exists on interdependency, prior publications have not given a thorough review of most available techniques in one paper. Also, the plurality of techniques indicates that there is no strong standpoint, and therefore researchers tend to go in different directions. This paper gives a compact overview of the techniques, with a short description of the techniques, and their advantages and disadvantages, to make the decision maker's choice easier in finding an appropriate method for the situation. This literature review is based on widely used electronic academic business databases. The titles and abstracts of articles have been examined, and relevant articles have been studied closely. The final review consists of some 200 articles.
Keywords: project portfolio management, interdependency techniques
Inter-unit interdependencies have been a widely used and important aspect of foundational organization and management studies, including works such as Thompson (2003) and Mintzberg (1983) (Sanchez & Worren, 2005, p. 5). This work investigates interdependencies among projects as modern subunits in companies today. Projects often have interdependencies with each other, and it has been found that managing these interdependencies systematically helps to increase the success rate of the projects (Rungi, 2010). Interdependencies are also deeply involved with research and development (R&D) and innovation (Rungi, 2009a). This is due to the fact that considering the interdependencies among projects helps in selecting better projects into the project portfolio (Archer & Ghasemzadeh, 1999a, p. 221). The interdependencies issue grows when the size of the project portfolio becomes larger. “Most project companies are seeking means to control and manage” multiproject management with interdependencies among projects (Arenius, Artto, Lahti, & Meklin, 2002, p. 291). Interdependencies are one of the major difficulties regarding a project's selection into a portfolio (Ghasemzadeh, Archer, & Iyogun, 1999, p. 745). Unfortunately, the academic literature has only rarely focused on the interdependency issue. Review articles are available that introduce the topic of interdependency and its techniques (e.g., Santhanam & Kyparisis, 1996; Staudenmayer, 1997; Archer & Ghasemzadeh, 1999a; Heidenberg & Stummer, 1999; Verma & Sinha, 2002; Cooper, Edgett, & Kleinschmidt, 2006), but they don't cover a full list of techniques and/or don't always reveal how the method(s) discussed deal with the interdependency issue. More importantly, most popular techniques for project selection neglect the interdependency issue totally (Santhanam & Kyparisis, 1996, p. 380), primarily due to the complexity and because their main research target has been project portfolio management (PPM) or anything else in general, not interdependencies per se. There is a gap in the literature (Schmidt, 1993). Also, the plurality of techniques indicates that there is no single strong standpoint and that researchers tend to go in different directions, and therefore a comprehensive review is usable. Some authors argue that knowledgeable techniques and other hard skills constitute 20% of project success (McKinlay, 2008); however, it is worthwhile investigating this percentage, considering the high failure rates of projects. It can be argued that if there were a full list of techniques, with a description of how the interdependency techniques can be applied, including the possible pros and cons of the techniques, it would help the decision maker to choose the one that fits the current situation of the company best. Empirics have shown that interdependency techniques influence success rates and other productivity aspects to different extents (Rungi & Kässi, 2009).
In general, empirics from companies about the use of techniques in practice are rare (Killen, Hunt, & Kleinschmidt, 2007, p. 25); this literature review serves also as a base for empirical research (Rungi, 2009b; Rungi & Kässi, 2009).
This paper is organized as follows. First, interdependency-related ontological issues are described. Then, an overview of the data gathering is given (i.e., how the literature review was performed). Thereafter, a short overview of project portfolio techniques is presented, which serves as a working context for interdependency techniques. Fourth, considerable prior work on different techniques of interdependency, including their pros and cons, are covered in next, two main sections, and final section concludes the paper.
Short Overview of Interdependency
Interdependency is defined as a contingent relationship between projects (modified from Thompson, 2003), therefore “interdependency” and “relationships” are used interchangeably in this article. According to Santhanam and Kyparisis (1996, p. 383), “the first ideas on project interdependencies were discussed by Weingartner and Reiter [in 1963], in the area of capital budgeting.”
The interdependency issue can be explained by many theories, among them PPM theory (e.g. Cooper, Edgett, & Kleinschmidt, 2006), PPM methods (e.g., Santhanam & Kyparisis, 1996), interdependency theory (e.g. Verma & Sinha, 2002; Staudenmayer, 1997), and interdependency methods and techniques (e.g. Heidenberger & Stummer, 1999).
The PPM process consists of two different parts: project selection and portfolio review. “The portfolio selection process is characterized by uncertain and changing information, dynamic opportunities, multiple goals and strategic considerations, interdependence among projects, and multiple decision makers and locations” (Cooper et al., 2006, p. 3). The critical factor is identifying the right projects for the portfolio. In detail, PPM has three main goals: 1) “maximizing the value of the portfolio,” 2) “balancing in the portfolio,” and 3) strategic alignment (Cooper, Edgett, & Kleinschmidt, 1999b, p. 29).
Cooper et al. (1999b, p. 28) emphasize that PPM and project selection are the most important strategic issues, especially, improvements in project selection (Scott, 2000). One survey (Archer & Ghasemzadeh, 2004, p. 239), indicates that 50% to 60% of new product development (NPD) projects fail. If the project selection is done with care, the success rate of the project portfolio is higher (Archer & Ghasemzadeh, 2004, p. 238). The project selection process consists of (1) idea gathering, (2) portfolio screening, (3) portfolio evaluation, and (4) portfolio prioritization (Koivuniemi, Piippo, Kärkkäinen, & Tuominen, 1999).
“Projects may be mutually exclusive” if companies explore competing technologies and products (Childs, Ott, & Triantis, 1998, p. 306). At the opposite end of the continuum, when projects are contingent, a “project can be chosen only when a second project is also selected” (Kleinmuntz, 2006, p. 7); for example, the development of an executive support system (ESS) is dependent on the development or existence of a basic accountant system. The modular structure of technical products, where modules serve as building blocks for several different systems, is another example of interdependencies between projects. “Other examples include situations where the success of one project may change the likelihood of success of another project” (Archer & Ghasemzadeh, 1999a, p. 221) or resource overlap (Ghasemzadeh et al., 1999, p. 749).
A more comprehensive review of interdependencies can be found in the following sources Staudenmayer (1997), and Sanchez and Worren (2005).
Procedure and Data of Bibliographic Study
There are many articles available, which deal with interdependency techniques among projects (Santhanam & Kyparisis, 1996; Archer & Ghasemzadeh, 1999a; Heidenberger & Stummer, 1999; Cooper et al., 2006), but they don't cover a full list of techniques and/or don't always reveal fully how the interdependency issue is dealt with by the techniques and in companies. There is a gap in the literature, mostly because interdependency has not been their main research target. For example, usually visual techniques are missing and/or a list of mathematical techniques is presented partially. Because interdependency, as such, has been found to be an important issue, a review article was considered to be relevant.
To fill the gap presented above, a search in the electronic online databases EBSCO, Emerald, IEEE Xplore, ProQuest, ScienceDirect/Elsevier, and JSTOR was performed in 2007/2008. These databases include articles from most well-known management journals distributed worldwide.
The bibliographic study was decided not to be limited by top journals, because according to Thiry and Deguire (2007, p. 650), project phenomenon is very rarely published in management literature. Only a few exceptions are available in top journals (e.g., Schwab & Miner, 2008).
The bibliographic research in the electronic databases was done by combining the search terms/keywords portfolio, multiproject, project, interdependency, interaction, interface, complexity, relationship, coordination1 and program/programme2, which in turn were used together with the name of the method, when a new interdependency method was found. The search terms were chosen on the basis of the linguistic similarity between the keywords, for example synonyms (e.g., portfolio and multiproject; and interaction, interdependency, and relationship), and on the basis of keywords found as a result of the first round of database searches (e.g. coordination). Following the bibliometric study of Artto, Martinsuo, Gemünden, and Murtoaro (2008, p. 3) “different keywords were combined to form one single keyword, for example: singular and plural forms of the same word” (e.g., interdependency and interdependencies), “two language versions of the same keyword” (e.g., integer optimization and integer optimisation), and “two ways to express the same issue” (e.g., design structure matrix and DSM, and multi-project and multiproject). The titles, abstracts, and keywords of the found articles were worked through and the most relevant ones were investigated deeply. As another data source for the review, the references were checked in the found articles. There were certain overlaps between the result sets received with different search terms, and these sources were investigated more deeply. Finally, some 200 sources were worked through.
Many kinds of research work have been conducted dedicated to interdependencies, such as interdependencies between companies/virtual organizations (Jarimo & Salo, 2006; Dabholkar & Neeley, 1998), between alliances and networks (Hoffmann, 2005), between the objectives of projects (Medaglia, Graves, & Ringuest, 2007), between project teams (Dietrich, 2007), or symbiotic interdependence (Pfeffer & Salancik, 1978), but these areas are not the subject of this article. This article deals only with interdependencies among projects.
Short Overview of Project Portfolio Techniques
“Over two hundred qualitative and quantitative [techniques]…exist in the literature for R&D project selection” (Coffin & Taylor, 1996, p. 208). A rather broad source of project portfolio selection techniques is available in Martino (1995). Hereby, the typology of the main portfolio selection techniques is drawn (Figure 1), categorized by goals of PPM (modified from Cooper et al. 1999b, p. 29; Archer & Ghasemzadeh, 1999a, pp. 211–220).
The effective use of each method is limited. Also, there is no consensus among academicians and practitioners regarding which method is the best (Koivuniemi & Edelman, 2003, p. 7). Most of the proposed techniques, including financial techniques, scoring models, the analytic hierarchy process, and ranking are claimed to neglect relationships among projects clearly (Santhanam & Kyparisis, 1996, p. 384; Archer & Ghasemzadeh, 1999b, p. 212; Lee & Kim, 2000, p. 369), but according to Cooper et al. (2006), these models are most popular among decision makers.
Overview of Techniques Addressing Interdependencies
A list of techniques addressing interdependencies among projects is given in Figure 2. More detailed descriptions of the techniques are presented below. The literature is full of descriptions of different techniques; however, there is a lack of discussion of the empirics (Killen et al., 2007, p. 1865) and examples in the literature.
There are some informal techniques for how interdependencies can be dealt with (e.g., “gut feeling,” “sacred cow,” formal/informal meetings, and (half-informal) group decision support systems (GDSS)). Informal techniques are rarely mentioned in the literature compared with the more formal techniques, although, there are indications that informal techniques are widely used (e.g., a survey by Rungi [2009b] found this to be the case). For meetings and their role to closely related strategy field, see Jarzabkowski and Seidl (2008). These techniques are easily understandable by common sense. Unfortunately, these techniques produce only roughly correct decisions. However, there is scientific evidence (experiments) that decisions based on gut feelings have produced better results than ones based on formal thinking (Aru & Bachmann, 2009).
The “sacred cow” is defined as a method where “the project is suggested by a senior and powerful official in the organization” (Meredith & Mantel, 1999, p. 141). “Gut feeling” refers to decisions made on the basis of intuition. Informal conversations and meetings (e.g., lobbying work, corridor discussions, and lunch table) are considered as very powerful and effective ways to influence and push one's will through.
GDSS are computerized techniques to support groups in their decision making (Kengpol & Tuominen, 2006). These interactive software techniques permit decision makers to work anonymously in parallel to find solutions for unstructured problems (ibid.; Kengpol, 2007, pp. 27–28).
How is Interdependency Handled?
Interdependencies are managed mostly in the mind. Project selection is based on intuition, projects are “driven more by personality and initiative, than by any explicit weighting of trade-offs among projects,” and the company perceives this personal involvement as a “strength” (Loch, Pich, Terwiesch, & Urbschat, 2001, p. 72). In the case of small portfolios, the responsible people are capable of considering most relationships among projects.
Mathematical Optimization Techniques
There are many models based on mathematical techniques (Coffin & Taylor, 1996, p. 208). The mathematical models usually optimize the objective function, with such constraints as resources, technology, strategy (Wang & Hwang, 2007, p. 248), and interdependencies. There are a number of modifications of integer optimization techniques, the division consists of “linear, nonlinear, integer, dynamic, goal, and stochastic mathematical programming” (ibid.). Among them, the “linear programming models assume that…no interdependencies exist” among the projects (Heidenberger & Stummer, 1999, p. 205); however, some nonlinear programming models can be equivalently formulated as linear programming models (see e.g. Anglani, Grieco, Guerriero, & Musmanno, 2005, p. 706). The method is called 0–1, “because of [its] discrete ‘select or not select' nature” (Ghasemzadeh et al., 1999, p. 746).
Integer Programming Models
In real life, there are many situations dealing with yes/no (i.e. go/no-go) decisions (Heidenberger & Stummer, 1999, p. 207). Real life offers “decision problems which are non-linear by nature” (ibid., p. 206).
How is interdependency handled?
Interdependencies among projects can be dealt with by solving the objective function (equation 1) (Ghasemzadeh et al., 1999, pp. 748–749): “
Z is a value function to be maximized, and
ai is the score of project i (for example, the AHP score)” (ibid., p. 748). “The decision variables are defined by:
for i = 1, …, N, where N is the total number of projects being considered, and
j = 1, …, T” (ibid., p. 748) where T is the total number of time periods in planning horizon with the following set of constraints for interdependency: “
“for i ∈ Pl, where Pl is the set of predeccor projects for a particular project l; l = 1, …, L” (ibid., p. 749). “Di is the duration of project i” (ibid., p. 749). These constraints guarantee that “the selection of precursor projects, once a project is selected” (equation 2), and “that all the precursor projects will be finished before the successor project starts” (ibid., p. 749) (equation 3).
It is difficult to do mathematic programming manually; there are software tools to carry out these techniques (see ibid., p. 751; Loch et al., 2001, p. 76).
Goal Programming Models
According to Heidenberger and Stummer (1999, p. 208), “goal setting is a common phenomenon” in real life. Therefore, techniques supporting it are preferred as well, and goal programming (GP) approaches are used quite often (ibid.). GP “can be used when more than one objective is explicitly identified and a clear priority exists among these goals. In goal programming, multidimensional goals are met in a sequential manner, where the decision maker's preferences are used to specify goals and priorities for each objective. The goal with highest priority is given the highest weight, followed by the second highest priority and so on” (Ghasemzadeh et al., 1999, pp. 745–746).
How is interdependency handled?
Lee and Kim (2000) have developed an analytic network process within a 0–1 goal programming model for project selection. The mathematical model of GP is similar to the non-linear programming model (see the section on integer programming, earlier in this paper, for further details). Mostly the goals are traditional PPM goals.
Fuzzy Mathematical Programming Models
In mathematics, values are usually taken as exact measurements (Coffin & Taylor, 1996, p. 210), but the surrounding environment doesn't usually provide precise data. In decision making, humans are reluctant “to make unqualified predictions” (Heidenberger & Stummer, 1999, p. 211), and here fuzzy logic can help. Fuzzy logic includes the “uncertainty [factor] into decision variables,” the objectives of the selection problem can be defined as a fuzzy set (Coffin & Taylor, 1996, pp. 210–211).
There are different interpretations of the fuzzy number. In project management literature, the trapezoidal fuzzy number is mostly used (Kuchta, 2001; Carlsson, Fuller, Heikkilä, & Majlender, 2007; Wang & Hwang, 2007). “The trapezoidal fuzzy number = (m1 m2, α, β) is defined as [equation 4]:
“where m1 and m2 are the left and right modal values and α and β are the left and right spreads” (Wang & Hwang, 2007, p. 249), where A represents the unknown quantity. The value of the function μA(x) is “the possibility distribution,” where x is a pregiven real number (Kuchta, 2001, p. 165).
Technological relationships, such as modularity, don't have such clear meaning in fuzzy mathematical programming, as in case of resource relationships. However, framework availability on time (technical relationship), has a meaning.
How is interdependency handled?
Kuchta (2001, p. 168) explains resource utilization through the fuzzy number approach: if there are S resources, then a fuzzy number represents “the [resource] utilization of the p-th resource by the i-th project.” This representation is a possibility distribution, which defines to what extent value x of the given resource “will be the actual utilization by the given project” (ibid.). Fuzzy numbers can then be used to model integer programming (see the integer programming chapter).
For an example of how fuzzy numbers are turned to integer programming, see Anglani et al. (2005).
In stochastic programming, “at least one input is uncertain” and can vary (Heidenberger & Stummer, 1999, p. 210). The works of Medaglia et al. (2007, p. 870) discuss stochastic parameter space investigation (PSI). Through stochastic PSI, it is possible to solve the selection problem with the constraint of the relationships.
How is interdependency handled?
The method proceeds in the following way (Medaglia et al., 2007, p. 870): (1) “project [resource] allocations are generated” on the basis of constraints: lower and upper bounds; (2) “if a project allocation does not satisfy the constraints on the resources, that configuration is discarded”; (3) if the allocation satisfies the constraints, then project objectives are evaluated through simulation; (4) from the simulation results, the statistics and deviation are found; and (5) if the project allocation is not a dominant one, then “it is preserved for…the decision maker at the end of the procedure (otherwise…discarded).” The model allows project relationships to be modeled by the objective functions (Medaglia et al., 2007, p. 870) (see the integer programming chapter for further details).
Dynamic programming models
Dynamic programming (DP) is a recursive technique (Heidenberger & Stummer, 1999, p. 209). If there is “a set of sequential decisions” and the optimal path needs to be found, then this technique is worth using (ibid.).
In portfolio selection literature, there are two different techniques with the same name: one is based on the idea of a decision tree and described here; the other one is based on the scoring model and is described below.
Real options (based on the decision tree approach) are similar to financial options in that they are based on the possibility “to choose for or against something” (Carlsson et al., 2007, p. 99), and “to take stepwise decisions” based on most up-to-date information (Gustafsson & Salo, 2005, p. 946).
This approach can be used in a situation in which the project has great potential to be successful in general, but is currently having problems, and the “decision maker has the opportunity to react…in different ways” (Perlitz, Peske, & Schrank, 1999, p. 256). The decision tree consists of decision-making points, and at each decision point the decision maker has to choose between possible actions (Gustafsson & Salo, 2005, p. 950). For making that decision, the decision maker has information on what state “prevails at this point” and what actions have been taken before (ibid.).
Simulation methods are also considered to be subpart of the real options methods (see also the section on the scoring model, later in this paper). One of the most known simulation methods is the Monte Carlo method. According to Martino (1995) simulation is appropriate to be used for resource dependencies.
How is interdependency handled?
The decision tree, together with the constraints of the relationships, can be modeled through an objective function, defined by a state tree (Gustafsson & Salo, 2005). After this has been created, the objective function of the decision tree can be cast in the form of integer programming (see the section on integer programming, earlier in this paper, for further details).
Scoring models use a list of criteria to rate projects. Usually there are not many decision criteria in use per one problem/situation (Archer & Ghasemzadeh, 1999a, p. 213). Each criterion is rated by evaluators (Cooper et al., 1999a, p. 107), most “typically on 1–5 or 0–10 [score] scales”. After the evaluating, the “scores are multiplied by the weightings [of the criteria], and summed across all criteria to yield a project score for each project” (op. cit., p. 107). Then the final scores are compared and projects with higher scores are selected.
How is interdependency handled?
There is some criticism about the suitability of scoring models—that is, from the mathematical point of view, the scoring model doesn't consider interdependency very well (Archer & Ghasemzadeh, 1999a, p. 214).
In the beginning, the model is similar to the traditional scoring model, but then the project selection decision is cast to the linear integer programming model (Meredith and Mantel, 1999, p. 151) (see the section on integer programming, earlier in this paper).
The unsuitability of the scoring model can be argued against. This could be the case from the mathematical point of view, and in any event it is more formal than informal techniques. The scoring model can lead to the same result as the widely used informal techniques, although Rungi and Kässi's (2009) results showed otherwise.
Real Options Reasoning (Based On The Scoring Model)
Real options are a “method for assessing uncertain projects that approximates option value through scoring a series of statements” (McGrath & MacMillian, 2000, p. 35). Method uses rated statements, which allow comparing different projects by comparing the scores. The Strategic Technology Assessment Review (STAR) statements have been developed to evaluate projects (ibid.).
How is interdependency handled?
The model contains statements, which are divided into several groups. For instance, the group “spillover” consists of statements for market and other interdependencies. The spillover group has statements whose target is to achieve cost advantages (McGrath & MacMillian, 2000, p. 43): “We may be able to use the evolving technology to reduce the cost of existing products;” “this project will create know-how and build skills that will be useful even after the project is over;” and “there is a significant danger that the product will cannibalize existing products.” These statements are rated to get the overall score for the project, under assessment.
Analytic Hierarchy Process (AHP) and Analytic Network Process (ANP)
The analytic hierarchy process (AHP) was first introduced by Saaty in the 1970s. AHP “is a theory of measurement through pairwise comparisons and relies on the judgments of experts to derive priority scales” (Saaty, 2008, p. 83). It combines mathematics and human psychology, it cleverly uses the idea that psychologically it is easier to compare pairs than all choices at a time. It divides a problem hierarchically into smaller parts/criteria, then for each criteria, a weight is assigned according to its importance. After that, possible alternatives (e.g., projects) are evaluated against each other in pairs in the light of the criteria, to see which one is more preferred (these evaluations are later on converted to numerical values). On the basis of the given evaluations and weights, the model calculates rankings for alternatives.
AHP assumes that the criteria are uncorrelated and unidirectional (Kengpol, 2007, pp. 28–29), and therefore Saaty introduced the analytic network process (ANP), which doesn't require hierarchical dividing from general to detailed, but is based on a network with feedback loops instead (ibid.; Kengpol & Tuominen, 2006, pp. 160–161).
How is interdependency handled?
In the project selection problem, one of the criteria can be the interdependency issue. Then the projects are compared with each other to see which one is more interrelated—for example, how much other projects depend on the project or how much the project can benefit from other projects. This can be based on a subjective judgment or predicted calculated figures. Later, the given estimations for this and other criteria are used to calculate the rankings for the projects. See the section on the scoring model, earlier in this paper, as well.
Dickinson, Thornton, and Graves (2001, p. 523) have created a dependency matrix.
How is interdependency handled?
The “dependency matrix is a square matrix” (Dickinson et al., 2001, p. 523), where each project represents one column and one row. Each element of the matrix dij represents the dependency that project i has on project j. The value must be between 0 and 1. A value 0 means that the “project i is entirely independent of project j” (ibid.). A value of 1 means the opposite—that is, “that project i is entirely dependent on project j” (ibid.). In the latter case, it is good to combine these two projects for the next portfolio selection meeting. The matrix makes it possible to calculate how much of a project's revenue comes from its interdependencies, as well as the percentage of the dependency of project i on project j.
For this purpose, input variable “minimum benefit level” (Mi) is used: “Mi is the percent of revenue expected if project is funded without funding the projects it depends on” (Dickinson et al., 2001, p. 523). “The remaining benefit of project i (1- Mi) is allocated to its dependent projects,” and the value of “W ij is the percentage dependency of project i on project j” (equation 5): “
where np is number of projects (ibid.).
Visual methods visualize relationships between projects, and there are different approaches with different notification systems (i.e., elements of how the project is presented on a diagram). According to Cooper et al. (1999a: p. 33), “many portfolio models yield information overload.” Decision makers want easily understandable solutions, and visual techniques have been popular for years. There are few visual techniques concerning interdependencies. Few of them are described here.
Design Structure Matrix
The design structure matrix (DSM) is similar to the dependency matrix discussed earlier. DSM is a square matrix, where each element (e.g., project, module) is represented by one column and one row. DSM was originally constructed by Warfield in the 1970s and Steward in the 1980s (MIT DSM Research Group).
How is interdependency handled?
The relationship between projects is marked (MIT DSM Research Group). For example, if project i gets input from projects j and k, then the corresponding matrix element (row i, column j and row i, column k) are marked. If the same project i gives output to project s, then the matrix element (column i, row s) are marked. If there is no relationship between two projects, then the corresponding place is left empty.
DSM can be used to represent static (clustering algorithm) and time-based (sequencing algorithm) relationships (Danilovic & Browning, 2007, pp. 302–304). The clustering algorithm is useful for “grouping highly related” modules as technological components of projects (Danilovic & Browning, 2007, p. 302) “into larger ‘modules'” (Hellström, 2005, p. 38), the sequencing algorithm can be used to “maximize the feed-forward flow of provided information and materials” (Danilovic & Browning, 2007, p. 302). As a result of the sequencing algorithm, it becomes clear which activities can be run simultaneously, plus it allows feedback loops to be removed (ibid., p. 302).
See also the section on the dependency matrix, earlier in this paper.
Thiry (2004, p. 275) proposes the trivial arrow diagramming method (ADM).
How is interdependency handled?
In this visual method the interdependencies are dealt with by “activities-on-arrow or a milestone chart” (Thiry, 2004, p. 273). The relationships are visualized by one-headed arrows. The arrow points to the direction of relationship.
Nested Options Model
Bardhan, Bagchi, and Sougstad (2004, pp. 33–34) have developed a model to value and prioritize projects.
How is interdependency handled?
The model of Bardhan et al. (2004) consists of visual and mathematical parts, so it is not a pure visual method. The mathematical part is a binomial model, which can be cast by the integer programming model (Bardhan et al., 2004) (see the section on the integer programming model, earlier in this paper). In the visual part, the relationships are again visualized by one-headed arrows, and projects are visualized as boxes.
Groenveld (1997) has developed a trivial method, which is based on “precedence relationships between projects” (Bardhan, 2004, p. 35). In addition, it links projects with objectives and technologies (ibid.).
How is interdependency handled?
Groenveld's approach is not a pure interdependency approach. Roadmapping integrates business objectives and technology (Groenveld, 1997, p. 48), and relationships between them are presented by arrows across time.
Vaidya and Kumar (2003, pp. 1099–1100) introduce a classification of interdependency as: (1) “extrovert dependency” (how many components depend on any specific component); (2) “introvert dependency” (how much one component depends on other components); and (3) “self dependency” (how much one component depends on itself).
How is interdependency handled?
The model of Vaidya and Kumar allows computing the introduced dependencies, and finding most critical and least important components within the system. They mention that the “higher the value of the intra-dependency index, [the] more sensitive will be the system to failure/partial-failure of the constituents; and hence, there is a greater need to design the system with great concern for better reliability” (Vaidya & Kumar 2003, pp. 1099–1100).
Main Pros and Cons of the Techniques
No model is perfect, and even the best one has its limits. The appendix presents the pros and cons of the techniques from the literature.
In addition, empirical evidence was collected from northern Europe to find out main problems (Rungi, 2009b) and the extent of advantages from techniques (Rungi & Kässi, 2009). With regard to the techniques, the following general problems were reported: not enough time to implement corresponding procedures, lack of knowledge, and that techniques can consider only limited number of relationships (Rungi, 2009b). Also, the influence from the use of techniques to the success rate and resource reduction was analyzed. Best success rate values were “achieved by mathematical methods, dependency matrix, and some visual methods …. [T]he usage of informal methods ends with worse results. Very surprisingly, the same can be claimed for popular scoring model.…Predictably, the usage of mathematical methods resulted in shortest length of projects among the users of interdependency…[S]urprisingly, the usage of scoring model increases the length of project for users of interdependency” (Rungi & Kässi, 2009).
This literature review has focused on giving a comprehensive overview of portfolio selection techniques concerning interdependencies among projects.
There are a great many interdependency techniques available, most of which are theoretically strong and complex, but there are considerably fewer visual techniques, which have shown to be very usable and popular among decision makers. In addition, descriptions of these techniques are dispersed over many different sources, and not in one place, making it difficult for the decision maker to choose the right one. Although there is much research on interdependency, prior publications have not given a full review of all available techniques in one paper. The main contribution of this paper is a compact overview of the main relationship techniques, including their advantages and disadvantages, and descriptions of how interdependency is handled. In addition, the classification of techniques has been developed further and some criticism given in addressing the techniques.
This bibliometric study also has limitations. First, no research can be absolute, and as this research is based only on articles searched from online databases, with a given set of search keywords and within certain time period, there is a chance that not all techniques are covered. However, on the basis of performed study, it can be claimed that the most influential solutions are covered. Secondly, during the study, greater emphasis was placed on describing the techniques in general, not introducing certain software packages where relationship techniques are implemented (e.g., enterprise resource planning [ERP]).
Future research will pursue empirical work to test the existing theories in practice and find out what concerns companies most when dealing with the interdependency issue today.
Anglani, A., Grieco, A., Guerriero, E., & Musmanno, R. (2005). Discrete optimization: Robust scheduling of parallel machines with sequence-dependent set-up costs. European Journal of Operational Research, 161, 704–720.
Archer, N., & Ghasemzadeh, F. (1999a). Project portfolio selection techniques: A review and a suggested integrated approach. In L. D. Dyer & J. S. Pennypacker, (Eds.), Project Portfolio Management (pp. 207–238). West Chester, PA: Center of Business Practices.
Archer, N., & Ghasemzadeh, F. (1999b). An integrated framework for project portfolio selection. International Journal of Project Management, 17(4), 207–216.
Archer, N., & Ghasemzadeh, F. (2004). Project portfolio selection and management. In P. W. G. Morris, & J. K. Pinto (Eds.), The Wiley guide to managing projects (pp. 237–256). New York: John Wiley & Sons.
Arenius, M., Artto, K., Lahti, M., & Meklin, J. (2002). Project companies and the multi-project paradigm: A new management approach. The Frontiers of Project Management Research. Newtown Square, PA: Project Management Institute.
Artto, K., Martinsuo, M., Gemünden, H. G., & Murtoaro, J. (2008). Foundations of program management: A bibliometric view. International Journal of Project Management, 27, 1–18.
Aru, J., & Bachmann, T. (2009). Tähelepanu ja teadvus. Tallinn, Estonia: Tänapäev.
Bardhan, I., Bagchi, S., & Sougstad, R. (2004). Prioritizing a portfolio of information technology investment projects. Journal of Management Information Systems, 21, 33–60.
Carlsson, C., Fuller, R., Heikkilä, M., & Majlender, P. (2007). A fuzzy approach to R&D project portfolio selection. International Journal of Approximate Reasoning, 44, 93–105.
Childs, P. D., Ott, S. H., & Triantis, A. J. (1998). Capital budgeting for interrelated projects: A real options approach. Journal of Financial and Quantitative Analysis, 33(3), 305–334.
Coffin, M. A., & Taylor, B. W. III (1996). Multiple criteria R&D project selection and scheduling using fuzzy logic. Computers Ops Res, 23(3), 207–220.
Cooper, R. G. (1997). Winning at new products accelerating the process from idea to launch. Reading, MA: Addison-Wesley.
Cooper, R. G., Edgett, S. J., & Kleinschmidt, E. J. (1998). Portfolio management for new products. Upper Saddle River, NJ: Addison-Wesley.
Cooper, R. G., Edgett, S. J., & Kleinschmidt, E. J. (1999a). Portfolio management in new product development: Lessons from the leaders, Phase I. In L. D. Dyer & J. S. Pennypacker (Eds.), Project Portfolio Management (pp. 97–116). West Chester, PA: Center of Business Practices.
Cooper, R. G., Edgett, S. J., & Kleinschmidt, E. J. (1999b). Portfolio management in new product development: lessons from the leaders, Phase II. In L. D. Dyer & J. S. Pennypacker (Eds.), Project Portfolio Management (pp. 23–38). West Chester, PA: Center of Business Practices.
Cooper, R. G., Edgett, S. J., & Kleinschmidt, E. J. (2006). Portfolio management for new product development: Results of an industry. Working Paper No. 13, The Product Development Institute. Retrieved February 17, 2008, from http://www.stage-gate.com/downloads/Results_of_an_Industry_Practices_Study.pdf
Dabholkar, P. A., & Neeley, S. M. (1998). Managing interdependency: A taxonomy for business-to-business relationships. Journal of Business & Industrial Marketing, 13(6), 439–460.
Danilovic, M., & Browning, T. R. (2007). Managing complex product development projects with design structure matrices and domain mapping matrices. International Journal of Project Management, 25, 300–314.
Dickinson, M. W., Thornton, A. C., & Graves, S. (2001). Technology portfolio management: optimizing interdependent projects over multiple time periods. IEEE Transactions on Engineering Management, 48(4), 518–527.
Dietrich, P. (2007). Coordination strategies in organizational development programs (doctoral dissertation). Helsinki University of Technology, Finland.
Durrani, T. S., Forbes, S. M., & McKinven, A. T. (2001). Management tools for real options techniques and task analysis. Paper presented at the IEEE IEMC 2001, Albany, NY.
Ghasemzadeh, F., Archer, N., & Iyogun, P. (1999). A zero-one model for project portfolio selection and scheduling. Journal of the Operational Research Society, 50, 745–755.
Groenveld, P. (1997). Roadmapping integrates business and technology. Research Technology Management, 40(5), 48–55.
Gustafsson, J., & Salo, A. (2005). Contingent portfolio programming for the management of risky projects. Operations Research, 53(6), 946–956.
Heidenberger, K., & Stummer, C. (1999, June). Research and development project selection and resource allocation: A review of quantitative modeling approaches. International Journal of Management Reviews, 197–224.
Hellström, M. (2005). Business concepts based on modularity a clinical inquiry into the business of delivering projects (doctoral dissertation). Åbo Akademi University Press, Finland.
Hoffmann, W. H. (2005). How to manage a portfolio of alliances. Long Range Planning, 38, 121–143.
Iamratanakul, S., Patanakul, P., & Milosevic, D. (2008). Project portfolio selection: From past to present. Paper presented at the IEEE ICMIT 2008.
Jarimo, T., & Salo, A. (2006). Optimal partner selection in virtual organisations with capacity risk and network interdependencies, Helsinki: VTT and HUT. Retrieved January 28, 2007, from http://www.sal.hut.fi/Publications/pdf-files/mjar06.pdf
Jarzabkowski, P., & Seidl, D. (2008). The role of meetings in the social practice of strategy. Organization Studies, 29(11), 1391–1426.
Kleinmuntz, D. N. (2006). Resource allocation decisions. In H. W. Richardson, P. Gordon, & J. E. II Moore (Eds.), The economic impacts of terrorist attacks. University of Southern California. Retrieved January 28, 2007, from http://www.usc.edu/dept/create/assets/002/50860.pdf
Kengpol, A. (2007). An evaluation of ICTs using decision support systems: Case applications from distributor's and end user's perspective group decision (doctoral dissertation). Lappeenranta University of Technology, Finland.
Kengpol, A., & Tuominen, M. (2006) A framework for group decision support systems: An application in the evaluation of information technology for logistics firms. International Journal Production Economics, 101, 159–171.
Killen, C. P., Hunt, R. A., & Kleinschmidt, E. J. (2007). Managing the new product development project portfolio: A review of the literature and empirical evidence. Paper presented at the PICMET Conference (pp. 1864–1874), Portland, OR.
Koivuniemi, J., & Edelmann, J. (2003). R&D project selection methods and systems in innovation management: A process based approach. Paper presented at the 12th International Conference on Management of Technology, IAMOT, Nancy, France.
Koivuniemi, J., Piippo, P., Kärkkäinen, H., & Tuominen, M. (1999). A groupware tool for R&D project selection for distributed company environment. Paper presented at the PICMET Conference, Portland, OR.
Kuchta, D. (2001). A fuzzy model for R&D project selection with benefit, outcome and resource interactions. The Engineering Economist, 46(3), 164–180.
Lee, J. W., & Kim, S. H. (2000). Using analytic network process and goal programming for interdependent information system project selection. Computers & Operations Research, 27, 367–382.
Loch, C. H., Pich, M. T., Terwiesch, C., & Urbschat M. (2001). Selecting R&D projects at BMW: A case study of adopting mathematical programming models. IEEE Transactions on Engineering Management, 48(1), 70–80.
Martino, J. P. (1995). Research and development project selection. New York: John Wiley & Sons.
McGrath, R. G., & MacMillan, I. C. (2000). Assessing technology projects using real options reasoning. Research Technology Management, 43(4), 35–49.
McKinlay, M. (2008). Project management to run but where's it going? Paper presented at the 22nd IPMA World Congress, Rome, Italy.
Medaglia, A. L., Graves, S. B., & Ringuest, J. L. (2007). A multiobjective evolutionary approach for linearly constrained project selection under uncertainty. European Journal of Operational Research, 179, 869–894.
Meredith, J. R., & Mantel, S. J. (1999). Project selection. In L. D. Dyer, & J. S. Pennypacker (Eds.), Project Portfolio Management (pp. 135–168). West Chester, PA: Center of Business Practices.
Mintzberg, H. (1983). Structure in fives: Designing effective organizations. Englewood Cliffs, NJ: Prentice-Hall.
MIT DSM Research Group (2008) MIT DSM website. Retrieved February 23, 2008, from http://www.dsmweb.org/
Perlitz, M., Peske, T., & Schrank, R. (1999). Real options valuation: The new frontier in R&D project evaluation? R&D Management, 29(3), 255–269.
Pfeffer, J., & Salancik, G. R. (1978). The external control of organizations. New York: Harper and Row.
Piippo, P., Kärkkäinen, H., Ojanen, V., & Tuominen, M. (1999). Problems and promotion of R&D project selection in Finnish high-tech manufacturing companies. Paper presented at the PICMET Conference, Portland, OR.
Rungi, M. (2009a). Interdependency Management in Project Portfolio Management: Bibliometric Citation and Clustering Analyses. In M. Torkkeli & O.-P. Hilmola. (Eds.). Exploring new research avenues of industrial management (pp. 111–136). Lappeenranta, Finland: Lappeenranta University of Technology.
Rungi, M. (2009b). Interdependency management: Survey comparison between small, medium and large companies. Paper presented at the 9th EURAM, Liverpool, UK.
Rungi, M. (2010). Success rate and resource consumption from project interdependencies. Industrial Management & Data Ssystems, 110(1), 93–110.
Rungi, M., & Kässi, T. (2009). Productivity from using interdependency methods in PPM. Paper presented at the 14th ICPQR, Alexandria, Egypt.
Saaty, T. L. (2008). Decision making with the analytic hierarchy process. International Journal of Services Sciences, 1(1), 83–98.
Sanchez, R., & Worren, N. (2005, July 18). A process design approach to managing interdependencies in complex organizations. Paper presented at the Academy of Management Meeting, Honolulu, HI.
Santhanam, R., & Kyparisis, G. J. (1996). A decision model for interdependent information system project selection. European Journal of Operational Research, 89, 380–399.
Schmidt, R. L. (1993). A model for R&D project selection with combined benefit, outcome and resource interactions. IEEE Transactions on Engineering Management, 40(4), 403–410.
Schwab, A., & Miner, A. S. (2008). Learning in hybrid-project systems: The effects of project performance on repeated collaboration. Academy of Management Journal, 51(6), 1117–1149.
Scott, G. M. (2000). Critical technology management issues of new product development in high-tech companies. Journal of Product Innovation Management, 17, 57–77.
Staudenmayer, N. A. (1997). Managing multiple interdependencies in large scale software development projects (doctoral dissertation). Massachusetts Institute of Technology, Cambridge, MA.
Thiry, M. (2004). Program management: A strategic decision management process. In P. W. G. Morris, & J. K. Pinto (Eds.), The Wiley guide to managing projects (pp. 257–287). New York: John Wiley & Sons.
Thiry, M., & Deguire, M. (2007). Recent developments in project-based organizations. International Journal of Project Management, 25, 649–658.
Thompson, J. D. (2003). Organizations in action: Social science bases of administrative theory. New Brunswick, NJ: Transaction Publishers.
Vaidya, O. S., & Kumar, S. (2003). Dependency and its predictions for systems and its components. International Journal of Quality & Reliability Management, 20(9), 1096–1116.
Verma, D., & Sinha, K. K. (2002). Toward a theory of project interdependencies in high tech R&D environments, Journal of Operations Management, 20, 451–468.
Wang, J., & Hwang, W.-L. (2007). A fuzzy set approach for R&D portfolio selection using a real options valuation model. Omega, 35, 247–257.
Appendix: Main Pros and Cons of Relationship Techniques
|General techniques|| || |
|Informal methods (IM)|| || |
|Group Decision Support System (GDSS)|| || |
|Scoring model (SM) || || |
|Analytic Hierarchy Process (AHP)|| || |
|Analytic Network Process (ANP)|| || |
|Dependency matrix (DM)|| || |
|Mathematical techniques (MTs)|| |
|Nonlinear integer programming (NLIP)|| || |
|Goal programming (GP)|| || |
|Fuzzy mathematical programming (FMP)|| || |
|Stochastic programming (SP)|| || |
|Dynamic programming (DP)|| || |
|- Real options (based on the decision tree approach)|| || |
|Visual methods|| || |
|Design structure matrix (DSM)|| || |
|Program-Level Network (Thiry, 2004)|| |
|Nested Options Model (Bardhan et al., 2004)|| || |
|Intra-dependency Index (Vaidya & Kumar, 2003)|| |
|Overall remarks|| |
(Koivuniemi & Edelmann, 2003, pp. 1–2).
(Piippo et al., 1999).
1 Interdependency management is also called “coordination” (Dietrich, 2007, p. 20).
2 The bibliometric study of Artto, Martinsuo, Gemünden, and Murtoaro (2008, p. 11) did not find “program” and “multiproject” to be related in business literature, however “program” was not left out due to close lingual similarity to “multiproject.”
© 2010 Project Management Institute