Management control in multi-project organizations: a study of R&D companies Paper submitted to Accounting, Organizations and Society July 30, 2003 Jörgen Dahlgren, PhD School of Management Linköping University SE-581 83 LINKÖPING SWEDEN Corresponding author: Jonas Söderlund, PhD School of Management Linköping University SE-581 83 LINKÖPING SWEDEN email: jonso@eki.liu.se Tel. +46 13 284065 Fax. +46 13 281873 Management control in multi-project organizations: a study of R&D companies Abstract There is a widespread acceptance that organizing by projects is on the increase. Research has developed considerably in recent years and various perspectives have been launched, and elaborated on, in order to increase our understanding of projects and project management. The focus has traditionally, however, predominantly been on project-centric approaches, downplaying organization- or company-wide matters in firms whose financial performance largely depends on the success of projects. We would argue that the understanding of projects must also be directed towards such company-wide matters. In the present paper we focus on the management control and especially the formal control strategies and control mechanisms utilized by firms operating by projects. The present paper focuses on how multi-project organizations solve the problem of linking its activities to each other and to the organization as a whole by the use of various formal control mechanisms. Based on a multiple-case study of four R&D units/companies, we propose a typology with four different kinds of multi-project settings. The typology is based on two dimensions, dependency between projects and project uncertainty. Theoretically, the paper contributes to the literature on multi-project management, project portfolio management and management control in project-intensive companies. Keywords: projects, project management, matrix organization, multi-project organization, management control, adhocracy, R&D. 2 1. Introduction In many industries and sectors, projects are on the increase as a way of improving corporate efficiency and effectiveness in terms of innovation, action orientation and motivation of employees (Ekstedt et al, 1999; Whittington et al, 1999). For many companies, projects even represent the basic commercial activity. A number of researchers have taken this development as a starting point for launching various research initiatives in order to increase our knowledge about project management. Research about project management has been reported in many of the leading management and organization research journals (see e.g. Lindkvist et al, 1998; Lundin & Söderholm, 1995; Shenhar & Dvir, 1996. This research has covered many aspects of project management but, so far, the principal thrust has been on focusing the individual project, its characteristics and its management and organization. Although the study of how to organize and manage an individual project is generally relevant and important, for those organizations that run several projects simultaneously as a multi-project operation, the wider and fundamental issues of organization-wide coordination and control between projects need to be addressed. With a set of concurrently ongoing projects, these issues need to be vertically linked to the strategy of the organization as well as horizontally to each other. The context for the latter coordination can further be expected to vary from situations where projects are operationally completely unrelated to cases where projects are mutually dependent upon each other, thereby creating operational dependency. Based on this observation, several researchers have argued for the need to study management control and company-wide aspects of project-based companies (Engwall, 2003; Hobday, 2000; Midler, 1995). These authors, it seems, advocate a widened perspective for the analysis of project-based operations in general. Such a widening of perspective, we submit, needs to link research on project management to a traditional organization and management studies perspective. We also believe that such a widened 3 perspective contributes to the general understanding of management control in knowledge-based project-intensive firms by illuminating both the relationships between projects and the management control activities in professional work. In recent contributions in the area of management control emphasis has been put into the control activities of single projects (Davila, 2000; Nandhakumar & Jones, 2001). These contributions point to the general importance of the study of management control in project environments. However, limited attention has been given to the relationship between projects, a relationship that we believe is equally as important as the study of the single project. As will be demonstrated in our empirical studies, it also seems like the management control activities and the dependency between projects represent a fundamental challenge for many R&D companies. The present paper contributes to our understanding of management control in multi-project organizations (MPOs). Our focus is thus primarily on formal control strategies (Simons, 1995) in R&D companies with a high degree of knowledge development and engineering capacity. However, we give special consideration to the management control strategies for handling the activities and relationships between projects. In doing so, we adopt a contingency approach in comparing the different control strategies in MPOs (cf. Chenhall, 2003; Gerdin & Greve, 2003). The paper is structured in the following way. Firstly, we will present a general discussion and analysis of management control issues and problematics in MPOs including a summary of previous research. Secondly, we describe the methodology used in the empirical studies, which are presented in the third section. In the fourth section, the cases are analyzed and a typology is offered with respect to mechanisms of control used in the companies. The paper ends with the main conclusions of this paper. 4 2. Previous research on MPOs Multi-project management normally means the management of a set of many projects (Cusumano & Nobeoka, 1998). In some firms it could be the management of as few as three projects, whereas in other firms it could be hundreds of projects. Closely related to the concept of multi-project management is what has been referred to as program management, platform management and project-based management (Söderlund, 2003). The literature is not always clear about the relationship between these terms, but it should be noted that the term multi-project management was introduced at a later stage than, for instance, program management. Program management was introduced at about the same time period as project management (Archibald, 1976). Program management has to do with how a set of closely connected projects was handled. The first cases of program management were reported from the US defense industry, most notably NASA, and the aircraft industry (Sayles & Chandler, 1971). It is thus worth noting that in the early days of management research, the development of project management was closely related to the management activities and the integration processes of the set of projects. In recent years, research on multi-project management has witnessed a renewed interest from both researchers within the field of project management (e.g. Anell, 2000; Engwall & Jerbrant, 2003), and from researchers on product development (Cusumano & Nobeoka, 1998). The growing interest is presumably explained by the fact that the success or failure of a single project is in many cases explained its relationship to other projects. Consequently, success cannot be studied and explained only by studying the single-project level (Pinto, 2002). There are, however, some additional management trends that can be observed empirically within this area. One is the development of “platforms” within many product development contexts (e.g. automotive industry), which makes the project-level increasingly much more complicated. The dependencies between on-going projects are expanded, creating a need of coordination on a 5 higher organizational level. Similar observations are found in other industries (see e.g. Meyer & Lehnerd, 1997). The matrix organization is one organizational solution that has been suggested to deal with multiproject contexts. However, the number of studies on matrix organization in real-life is limited (cf. Knight, 1976; Ford & Randolph, 1992). Research often points to the advantages of a “balance” in the matrix (Katz & Allen, 1985). In the present paper, we will not discuss the various balances in matrix organizations, but instead zero in on what we argue are other important parameters for understanding multi-project organizations. Mintzberg (1983), for instance, in his analysis of the adhocracy claims that we know relatively little about project-intensive companies in R&D settings. His analysis is predominantly on a general level discussing various types of coordination mechanisms compared to other types of organizations/configurations. However, the analysis of the differences between adhocracies (cf. multi-project organizations in R&D) is very limited. For instance, the need for creative and innovative work is considered to be the key in adhocracies, but although this generally holds true also in the type of companies that we have studied, the need for control and standardization is also a fundamental condition. Another important aspect is the project dimension, which Mintzberg highlights by referring to a number of classic project management texts. He does not, however, take into account the multi-project aspects and problems that confront the management of R&D companies. Mintzberg refers to “mutual adjustment” as the key coordination mechanism in project-intensive R&D companies. This particular control strategy seems nevertheless too underspecified to grasp the differences between multi-project organizations in development and complex contexts. Closely related to the work by Mintzberg is the seminal work of Simons (1995) who points to various control strategies in contemporary organizations. He specifically illustrates the relationship between management control and the project management systems. He does not, however, focus on the various logics and strategies in handling the operative dilemmas of multi-project organizations. 6 As was pointed out in the introduction, most of the research on control in relation to projects is focused on intra-project control in terms of work breakdown structures, budgets, deadlines etc (see Packendorff, 1995, Söderlund, 2003, for reviews). This is obviously an important area for project management and for the control of the single project. But just as control of individual activities is not sufficient in an organization, intra-project control must be supplemented by interproject and organizational control. As pointed out by Van der Merwe (1997:223), there is almost no research on projects from business and company-wide perspectives. The traditional research on project control, we submit, must be widened to also include aspects and concerns outside the individual projects. We thus argue that research must be directed towards the relationships between projects, how R&D companies solve the dependencies between projects and what major control strategies that seem to dominate in such companies. 3. Control problems in multi-project settings In terms of control, every organizational management faces five fundamental problems: to define organizational purpose, to create a division of labor, to coordinate the activities, to motivate members of the organization and to develop and adapt the organization and its activities in face of a changing environment. These are the same problems regardless of whether the organization is project-based or not. This paper does not focus on the strategic, motivational or the learning issues, but concentrates instead on how problems related to division of labor and coordination in an environment characterized by varying levels of uncertainty can be solved. Division of labor and coordination represent two sides of a coin. The traditional organizational solution to the problem of balancing 7 differentiation and integration is the hierarchy (Lawrence & Lorsch, 1969). The organizational hierarchy expresses both the division of labor and the way cooperation should take place both laterally and vertically. Although hierarchy will still be used in every organization, it also seems reasonable to assume that in MPOs this mechanism has to be supplemented by others. It also seems reasonable to assume that what other mechanisms and control instruments that are used is contingent upon the relations both between projects and between projects and the organization. From a control perspective, projects in the companies that we have studied present some additional characteristics that influence the way the control strategy and systems are designed. We believe the following ones to be the most important ones for an analysis of the management control strategies: the focus on a specialized task, often in close cooperation with a customer, makes direct control less applicable and instead requires decentralized or delegated control, the fact that participation in projects is limited in time, often also part-time, makes leadership, authority and responsibility more complicated, constantly upsetting any attempt to a more permanent structure, the need to maintain and develop specialized competence must be reflected in organizational responsibilities, and projects vary in terms of uncertainty in relation to both appropriate activities and expected outcome. Accordingly, a contingency framework for the management control in multi-project organizations must take these issues into account. Although some researchers have reached the conclusion that “classical management cannot apply” (Turner & Keegan 2001), we believe that nothing suggests that radically new mechanisms or instruments are required, but that the traditional instruments 8 must be adapted and configured in accordance with the specific context. Instead, we would argue, a contingency framework must rest upon the understanding put forward by classic organization and management studies and modified to fit the specific challenges facing multiproject organizations in R&D settings. 3.1 Management control and the principle of departmentalization The two classical models for departmentalization are structuring by function (sales-productionpurchase) or by purpose (product/customer/geographical area) (Galbraith, 1973). The respective advantages of these two forms are well known. Functional structures optimize attention to specialized knowledge while purpose structures focus the needs of the market. The fundamental business process of many project-based companies is to develop and apply specialized knowledge in different projects. Consequently, the need for a primary focus on such specialized knowledge would seem to lead to a U-form as the preferred organizational structure (Williamson 1997). The departmentalization of the organization will then be based on the different types of specialized knowledge that is required in each project. In a U-form structure projects are established by combining resources from different functional specialties forming a temporary project team. The participation of each member in the project group is often part time and basic knowledge development occurs outside the project. The need for specialized knowledge might be a fundamental characteristic for certain projects. In general, however, organizational activities in the form of projects by themselves can represent “self-contained” activities with specific purposes, specific clients, specific time limits and within specific economic frames. If then the required degree of specialized knowledge is more limited, 9 departmentalization need instead be based on an M-form structure where the projects make up the self-contained components typical for that structure. 3.2 Management control and dependencies Departmentalization leads to differentiation, which introduces the problem of “keeping the act together” and consequently creates a need for organization-wide integrative mechanisms and instruments. Examples of dependencies are that projects sometimes draw on the same resource base in the line organization and that organizational profits are the combined results from each and every project. A loss on one project certainly can potentially impact negatively upon the financial resources available for other projects. The main result from the Lawrence and Lorsch study was that high performing organizations use different integrative devices in order to cope with different levels of differentiation. In a project context, one such category is made up of companies who, although needing to adapt to each and every client, basically sell the same standardized program with the same standardized implementation process. To this category belong many software and consulting organizations. At the other end of the scale are companies with large and unique projects, involving development activities and with a high degree dependency in terms of resources, e.g. people and knowledge. If variation were great between projects, one would expect integration through special integrative units or permanent cross-project functions. If on the other hand all projects are basically the same, integration could be created through hierarchy or standardized systems. 10 In order to understand the link between control structure and dependency, one needs to further explore the nature of dependency. However, far apart projects in an organization may be in terms clients, resources, knowledge, scale etc, since they jointly generate the profits for the organization, there is always a pooled dependency which emanates from being part of the same organization. This idea of a pooled dependency also applies to shared resources like management, development laboratories, people and systems. According to Thompson (1967), the basic control mechanism in this case would be standardization. Each project must follow certain rules of behavior when it comes to planning, budgets, follow-ups, hiring and firing, reporting etc. Apart from pooled dependency, the relation between projects can also exhibit the characteristics of mutual or reciprocal dependency. Such a situation occurs when one project feeds into another project, which in turn provides input to the first mentioned project. This happens, for instance, in commercial projects that include development activities. The necessary knowledge for the development could well be derived from a separate project, which in turn then needs to know the conditions for the commercial project. Such a situation calls for control by mutual adjustment. Finally, there could also be sequential dependencies, meaning that one project produces the input for another project and it needs to be completed before the second project can start. Traditional intra-project planning models like CPM and PERT were developed to model sequential dependencies. In general, this type of dependency calls for control by plans to coordinate activities. Relating this discussion to management control in MPOs one would, given the fact that projects more than ongoing operations are characterized by a high degree of inter-project independence (own budget, own objective etc), first of all recognize the pooled dependency aspect and consequently the need for standardization. It is to be expected that multi-project organizations 11 have developed rules of conduct to guarantee standardized behavior in the projects, e.g. standard project management models, such as PRINCE and PROPS. The link between pooled dependency and standard operating procedures has also been demonstrated by Macintosh & Daft (1987). The situation here is similar to what Scott (1965) denotes as “heteronomous organizations”, which display a high degree of independence on the vertical axis. In such organizations, delegation is used to handle the independence, but since delegation means a local right to take decisions according to centrally laid down principles, this in fact is a type of standardization. The independence of projects, coupled with uncertainty, presumably also leads to an increased importance of control through “beliefs systems” (Simons 1995). Finally, if mutual dependencies also are at hand, these are likely to lead to the establishment of “cross-project interaction” and information exchange and sequential dependencies are handle through plans, e.g. budgets and statistical reports (Macintosh & Daft, 1987). 4. Research questions The previous sections have intended to stress firstly the need for better knowledge of control in multi-project organizations, and secondly to link the strategies of control in MPOs to traditional organizational control theory. Our argument is that in order to increase our understanding of different types of multi-project contexts both conceptual development and additional empirical research is required The aim of this paper is therefore to elaborate on a model or typology based on both insights into traditional organization theory and some key traits of R&D projects. Following this, the two research questions addressed in this study are: 1. How are multi-project organizations controlled in order to achieve integration and resolve dependencies? 12 2. How and why do control mechanisms vary between different multi-project contexts? The model offered will be illustrated by examples from the case studies presented below. 5. Methodology The methodology applied for the studies reported in the present paper could best be described as a multiple-case study. Such an approach is most applicable in situations where the researcher wants to both get a closer in-depth understanding of particular social phenomena, and be able to make comparisons between various contexts. As a matter of fact, the main reason for including more cases than one in theory development or refinement projects, is that the researcher expects a variations between different cases in different situations. Furthermore, the use of multiple cases makes it possible to develop typologies (Normann, 1976). Eisenhardt (1989) points to the importance of logic behind the choice of different cases. In the present study, the logic has mainly been that we wanted to cover different industrial settings: in our case R&D companies/units within larger corporate groups. We also wanted to cover settings that varied along the following lines: number of projects, linkage between projects (e.g. dependency of the same resources, engineers, etc), functional department or project focus. The study started with a number of interviews at each company in order to get a first understanding of the context, strategy and overall organization of each of the firms. After that we made a first analysis and tried to categorize the companies included in our study. Subsequently, we did a few (2-3) interviews with project managers or project directors in order to focus on the 13 formal control mechanisms applied in each of the firms and also get a better understanding of the various types of projects in the firms. Approximately, four or five persons were interviewed in each company. The first draft of each case study was distributed to one or two key actors in each firm. We wanted to use several quotes from the interviews in order to get a real-life story and also to increase the trustworthiness of the case studies presented in the paper. During the whole research process we studied various types of literatures that we believed were of interest in a multi-project management context. We therefore included fairly broad organization theory literature on matrix organization and adhocracy, but, of course, also the literature focusing entirely on multi-project management. Our basic assumption was that the two strands of research are quite closely related but much research in the project management stream does not fully recognize the linkage to traditional organization theory. We thus tried to balance these streams of research and set out to integrate them and focus specifically on formal control mechanisms. In terms of theory development, it is important to understand what it is that is generalized. Although case studies are an excellent vehicle for producing “rich stories”, it is not the specifics of the case that are generalized, but rather “the general in the specific” (Baxter & Chua, 1998:80). Our purpose is thus to theorize about “chronic behavioral and social issues that are exemplified in and by our case studies”. An essential prerequisite for finding such chronic behavioral/social issues is theory. Without theory, any identification of general themes is likely to lead to a fragile construction where local conditions are allowed to influence conclusions in a manner that gives results that are at best interesting, but without any general significance. The cases in our study come from different settings. Three come from the mobile telephone industry and one from the aerospace industry. In the aerospace industry, the R&D is close to 14 basic research, whereas the other three can be characterized as applied research, or development. The question might then be asked whether the similarity is sufficient for allowing generalizations. Since all projects are development projects, only limited comparison can be made with other project types, e.g. construction projects or consulting projects. First, one must keep in mind that we are not setting out to generalize one concept or one model to all types of organizations. On the contrary we are looking for variations between organizations in order to map that variation in our model. The variations are, so to speak, a basic platform for us, not something that impedes or threatens generalizations. Secondly, our model clearly recognizes the differences between, for example basic research and applied research, in terms of the amount of uncertainty involved. On a general level, there may of course be many other differences between the industries, organizations and R&D activities, but to refer such differences to the role of background is, of course, at the heart of any model-building approach. Thirdly, and finally, we believe that by using general theoretical concepts from organization theory, such as uncertainty and dependency, we believe that we are addressing themes that are relevant to all our cases. 6. Case studies Below we present four case studies. We give an overall contextual description of each of the companies. A summary of the cases and some additional information will be given in a summarizing table of the cases. 15 6.1 Case 1: Saab Aerospace Future Products Saab Aerospace develops aircraft systems and sub-systems primarily for military markets. They also work as sub-supplier and manufacturer for large civil aircrafts. The Future Products division is one of three business areas within Saab Aerospace. Within this business area most work is oriented towards developing defense technology for the future. It is thus a very high-tech environment and the level of engineering skills of the division is well known and recognized. Along side the long-term development work, the division also develops systems and products that support the current operations of the Saab Group – from manufacturing process to customer support. The business has a clear orientation towards “network defense” in the sense that it develops defense systems that integrate management, information and operations to function powerfully in a network. The division works not only with other parts of the Saab Group but also with a number of international partners and in various types of consortia. The number of employees amounts to just above 300. The production activities of division are organized into the following four departments: Product Development, System Technology, Operations Analysis and Advanced Design. In these departments the development of the products is carried out. In addition, there is also a department for the marketing and sales of the products which reports directly to the top management of the division. Future Products is organized as a matrix organization. The four departments make up one dimension and five program areas the other. The departments supply knowledge and resources to the program areas, which are the organization’s operative units, i.e. the programs are responsible for getting and completing the contracts. Each of the departments consists of a department head and a number of project managers. The project managers report both to the department head 16 and to the program manager. There is one program manager for each of the program areas, i.e. five program managers in total. Today the line managers have more power than the project managers. The project manager is responsible for when and what should be done; the line manager is responsible for appointing the right person for the job, for determining how the job should be done and what methods and technologies that are most appropriate for the job at hand. “The line manager often thinks that he should control the whole work package and that the project manager only should look after the financial figures. This creates an imbalance in the work, which is explained by the fact that the line managers are not clear about their roles in the organization. /…/ There is a lot of work still to do in order to get this to function.” (Project Manager) A major reason for conflict is the fact that the line managers’ responsibility is not clearly defined. Another frequently conflict situation often appears in smaller projects due to fact that in these projects, the line manager often also is the manager of both the resource owner and the project manager. Such situations, often create problems for the project manager to carry out his job successfully. The development cycles for Future Products are very long, which is typical for the industry as a whole. The development lead-time is often more than five years. “It is possible to look at the replacement need for our main customer. It is thus possible to estimate what is needed and required in the future and when new 17 products will be needed…it is fairly uncomplicated to give a prognosis on the replacement need.” (Director of Technology) The difficulty is instead considered to be deciding on when to start development work in order to reach the market in time. Furthermore, consideration has to be given to the changing environment, such as political conflicts and new defense alliances. According to the management team the fundamental way of controlling projects lies in choosing the right ones. The first phase of each project is the market analysis in order to be able to set the correct target for each of the projects. In the second phase, focus is put on potential business and time estimates for the various opportunities. The resource estimates are done on a fairly comprehensive level and are based on standardized areas of competence. Suggestions for projects come from several sources - clients, supra-projects or engineers in the company. “These are different sources that we try to assemble a few times each year. We evaluate each of the suggestions and let a group of people take a closer look at each suggestion. It is also this group who is responsible for prioritizing between the projects and suggest which project that we should launch in the nearest future.” The company recognizes that this process is key for competition, which has led management to devise more clearly developed processes and routines for handling the suggestions. However, they also recognize that the process cannot, and should not be, overly standardized. 18 The valuation of the projects is not only based on a product level, but is based on other aspects related to the business plan. The strategies are developed based on five focus areas which make up a kind of Balanced Scorecard. The five focus areas are: profit, customer, cooperation, operations and co-worker. Most of the work in Future Products is carried out in various types of projects. In total more than 80 percent of the employees work in different types of development projects. The unit applies Saab’s project model, PSM (Project Steering Model), and the project process is relatively standardized. A project is always initiated by a project owner, even though the suggestion might come from other parts of the firm. It is also the project owner who appoints project manager. 6.2 Case 2: Ericsson BSC Ericsson BSC (henceforth BSC) is a unit within the Ericsson Group. It is responsible for developing products and services within the area of base station control. The unit has a strong line organization and a number of projects that cut across the organization. The projects are organized and managed by a project manager, an assistant project manager. The rest of the people in the project are recruited from the line organization. Within BSC, there are four different departments. The responsibility for making priorities and determining which projects that are to be launched falls on the systems department. The integration and verification department is responsible for verifying that the developed product works according to the specification and that the product is reliable. Apart from these two departments, two so called project offices are also part of the organization. One of the offices has the main responsibility for developing the product from development to product delivery. Formally the assignments from 19 the strategic product management are directed to this project office. The second project office is responsible for support, i.e. customer support. Both project offices borrow resources from the line organization and from design and support units around the world. The design units within the BSC organization are geographically dispersed to Dublin, Linköping, Dallas and London. The design units are separate companies but fully owned by the Ericsson Group. These companies have different competence profiles and are responsible for developing the functionalities that are to be part of the project. The assembly and final test of the product is done in the BSC unit in Linköping. The design companies also support and work on other projects within the Ericsson Radio Systems company. The BSC organization also includes a steering committee. Every manager is a member of this committee and the local design units’ managers are also represented. The committee makes decisions on the projects’ strategies, directives and changes. The BSC unit has three parallel projects that are so-called main projects. They are primarily software projects. The three projects are in different phases of the project lifecycle. One is in the pre-study phase, one in implementation and one in the conclusion phase. Apart from these three main projects, a number of other smaller projects are also carried out. They follow a more simplified process compared to the main projects. These projects are normally minor changes on the current platform or additional services that are possible to separate from the main projects and do not fit the overall time plan of the main projects. “A few years ago we designed whole packages, an entire release with design, testing and verification. That could take up to three or four years. Now we try to divide the package into smaller parts which makes it possible to release the first package after, for instance, one year.” (Assistant Project Manager) 20 The main projects are the principle agenda setters for the management team. “You can look upon the main projects as “software trains” – when the first train is fully loaded with the right functionality, you can release it.” (Manager, BSC) The coordination between the projects is sometimes extensive. “There is somewhat of a dream that you can build projects of stand alone modules. This is, however, not possible in real life. There are a number of interactions that occur under way which creates a number of dependencies on various parts of the projects, not only on the technology side.” (Manager, BSC) 6.3 Case 3: Ericsson SRF Ericsson SRF (henceforth SRF) is a unit within Ericsson Radio Systems. The unit works with development of radio base stations and is one of the most important parts for the Ericsson Group. The projects are primarily software projects. The unit also has a strategic product management with responsibility for the entire product range within the unit. A centralized forum called the Radio Development Board makes all the Tollgate decisions for all the main new development projects within the entire unit. It is thus within this board that all the business decisions concerning each project are made. The board consists of the manager for SR (the business area), the manager for the strategic product management and each of the sub- 21 product units of which SRF is one. Each of the sub-product units has an operative steering committee with responsibility for each project. The operative steering committee consists of the line managers who “own” the resources, the project office manager, the manager for product management and the project managers. From time to time line managers from other line units (i.e. from other units within SR) are also invited if there are aspects of a project that affects the entire organization. The unit has decided not to have an overall project office. The project management offices are decentralized and one is thus located in the SRF organization. The project management office is responsible for the operative project planning and the manager for the office is responsible for all the project managers within SRF. He is also the one who appoints project managers to the projects within SRF and for coordination between the various projects. He is also in charge of the unit’s project portfolio. The organization also includes a special unit with the responsibility for the many sub-projects in SRF. Such sub-projects are frequently geographically dispersed to different parts of Sweden and to design units in Europe. These design units work on assignments from the SRF unit. In SRF four projects are carried out simultaneously. Normally each project lasts for approximately three years and the projects are partly overlapping. Every project employs more than fifty people. Apart from these three-year projects, a number of smaller projects (longevity, man hours and people employed) are carried out. Several projects within the SRF unit do not only have strong dependencies to other projects in the SR unit but also with other projects in the Ericsson Radio Systems organization as a whole. “When you initially develop a product, you only have one large project which you focus entirely on. However, when you have an established product portfolio, you 22 have a completely different situation. You must be able to handle a variety of products and a multitude of projects.” (Manager, Strategic Product Management) According to the same manager, the unit has a number of projects that they would like to start or proceed with. “It is normally a matter of man hours. We have a certain budget and we cannot hire 1000 engineers more if we would like. We have limited pool of resources.” (Manager, Strategic Product Management) The SR unit uses the standard Ericsson PROPS model for controlling projects. This model is also used when it comes to making priorities between projects. Every new product development project has to be decided upon by the Radio Development Board. After such a decision (the TG 2 decision) the responsibility is handed over to the various steering committees in the unit. It is the respective steering committee that is responsible for making the rest of the TG decisions throughout project implementation. Before every TG decision a risk analysis has to be made. The product management is responsible for setting the requirement agenda and collecting information from the customers, the other units within SR and the Ericsson Group worldwide. Profitability is always key in the discussion when it comes to making priorities between projects: Normally each unit has one large project that has the highest priority. It is also this project which sets much of the constraints for the other projects when it comes to resources and attention. Of great importance is also whether a project is in line with the unit’s strategy and operative targets. However, once in a while, a project, which is outside the main strategy, is given priority in order 23 to develop new markets or new technologies. Still, every project has to be based on a clear-cut business case. “The critical issue is to understand the dependency that exists between the different projects and before we start a new project be able to make an estimation on how that project will affect the other ongoing projects. What are the critical resources in the project and how will the resources be affected by the new project? (Manager, Strategic Product Management) The management also emphasizes the role of communication in managing multiple projects. “The best way is to talk to the project managers about the critical resources in the projects. However, we are currently looking at a new methodology to handle this more formally. I hope that this will improve the management of multiple projects.” (Project Office Manager) The review of each of the projects is normally done in the operative steering committee. The decisions on whether a project should proceed or be terminated is made by the RDB. “It is very painful to terminate a project that has passed the TG 2 decision.” 24 6.4 Case 4:Telia Mobile Telia Mobile is a company within the Telia Group (a Swedish telecom operator). It was founded in 1997 as a pure technology company with responsibility for the development and maintenance of the mobile telephone infrastructure. The Telia Mobile line organization consists of two business areas - the “Net Business” and the “Final Customer Business”. Under each unit, four departments hold the resources that work on the projects. Additionally, a development department, R&D, organizes sixteen project managers in a project management pool. The project mangers are not only managing projects within the R&D unit, but also for the business units Net and Final Customer. Sometimes the need for project managers exceeds the available number of project managers. In these situations, consultants are recruited to assume the responsibility for managing the project. Telia Mobile has a number of staff functions one being the project management office. This office is responsible for supporting the project management model and for project coaching. Telia Mobile has three permanent steering committees, one for each of the business units and one that is responsible for the common projects, that involve both the Net and the Customer organization. In each of the business units’ steering committee, a project coordinator has the role as contact person for the project managers and the rest of the organization. The project coordinator is also the connecting link between the projects and is the one who suggests priorities to the steering committee. The projects in Telia Mobile are of many different types: new platforms to build services on, new platforms for making maintenance more efficient and regulatory projects, i.e. projects that has to be carried out due to legal requirements. 25 In the organization, it is not clear what really constitutes a project. “When you live in an organization like us, with many projects, it is very hard to determine how many projects you have going on at the same time.” (Project Manager) The projects are carried out in parallel and their size vary both in terms of time and of engineers employed. Many of the service development projects have longevity between three and fifteen months (ten months on average). The project teams vary between five and thirty engineers (ten persons on average). The net-development projects have a lifetime between seven months and two years (on average fourteen months). Every project is put on the project web, which looks very much like a structured web site. The project web is also used for control and follow-up purposes. Many projects have a high dependence and there are many projects going on. “We have too many projects at the moment. If we would be able to take a few of them away, things would be much better. We have a problem today.” (Manager, Project Office) However, the company has worked intensely to improve the project selection process. “We have so many good ideas. The tricky thing is to decide which of the ideas to go for.” (Project Manager) 26 Priorities between projects have been assigned to a group called NASA (an abbreviation based on the Swedish words for Net and Customer Businesses). This group has made a priority list divided with four categories. The time factor and delivery dates are normally the ones driving the prioritization. “We have this type of funnel where we initially have somewhat of a wide and open idea generation phase. After that we have decision points where we screen the projects and determine which of the various ideas that should move on to the prestudy phase. Following that, we have a decision phase where we decide on which of the projects that should enter project implementation. However, it requires tough management to tell committed individuals to terminate a project which they have been working on for a number of months. I still believe we could do better on the entire funnel-process.” (Project Manager). 7. Analysis The analysis section is structured as follows. First, we will give a brief summary of the case studies presented and do a cross-case analysis based on some important facets of multi-project organizations. The parameters chosen are the ones that have emerged as key traits from our empirical study and discussions with the people directly involved in the multi-project management of each of the companies. Second, we will elaborate on a typology which explains much of the control strategies observed in our cases. 27 Some of the characteristics of the cases described in the previous section can be summarized in the following table: /Insert Table 1 here/ 28 All the projects described in the previous section are development projects and for all case companies such development projects are crucial for the firms’ competition since all companies compete in markets where technological advances are required for success. Projects are consequently of vital importance for company survival. The relationship between projects and the organizational levels above is one of pooled dependency. According to Thompson (1967), this can be handled by standardization. In all of our case companies we find the use of a standardized project model. In fact all the companies use the same model since Saab’s PSM is actually an adaptation of PROPS. By requiring the use of such a model the management of the organization is able to standardize behavior in otherwise decentralized operations. In Ericsson BSC this standardization is further emphasized by the use of planning constants and stressing the importance of “global thinking”. The use of devices like planning constants is of course tied to the similarity between projects in space and time. The projects within Saab Future products are so different and with such a long project time that planning constants might not be relevant. The organizational structure chosen by the four companies varies. Two companies use matrix structures, one functional and one a business unit structure. The matrix structures seem however not to be identical since the structure within Saab is dominated by the line organization. This presumably reflects a strong emphasis on specialized knowledge as the primary success factor in product development. This knowledge has always been the hallmark of Saab in their development of civil and military airplanes and continues to dominate the organization. One might therefore question whether the structure is not in fact more functional than matrix. As for inter-project dependencies, this seems to be strongest in Ericsson BSC and weakest in Saab. Mutual dependencies are best handled by creating arenas where projects can meet. In 29 Ericsson BSC some members work in different projects simultaneously which could lead to information being channeled from one project to the next. This is further promoted through monthly progress meetings and a common steering committee. Because of the interdependencies created by the need for a common resource, some resources are not allocated to the projects but remain unallocated and are supplied after special decisions. In the Saab case, with no real mutual or sequential dependencies, no common arena is called for. 7.1 Control strategies in MPOs Based on our empirical studies we are now prepared to suggest a control typology for multiproject organizations. The typology is based on two variables - dependency and uncertainty. Dependency is here taken to denote the extent to which the success of one project is contingent upon the result of activities in other projects (inter-dependency). This dependency could be either pooled, sequential or reciprocal (Thompson 1967). In a situation of pooled dependency no direct transactions occur between projects, but instead projects are related through marketing or financial relations. Projects in the construction industry can be an example of this - each project is independent, but if one project fails it can effect the customer valuation of other projects and/or lead to financial burdens for them. The sequential dependency occurs when one project’s output is input for another project. In our case this happens e.g. with projects in Telia Mobile where projects in Final Customer Business are sequentially dependent on projects within Net Business. Finally, the reciprocal dependency means that projects are involved in an inter-project exchange. An example of this would be the relationship between development projects within Saab Future Projects, where technical know-how and design in one area has to be adapted 30 knowledge and design in other areas and vice versa. All projects within an MPO are interdependent, but different contexts produce different degrees of dependency. The uncertainty of projects is the degree of precision with which the variation in outcome, resources and work process of projects can be forecasted. Every project, almost by definition, carries with it a certain degree of uncertainty. Projects are in this sense more influenced by environmental changes. However, the degree of uncertainty varies. In development projects close to basic research, such as within the pharmaceutical industry or the aircraft industry, the uncertainty is so great that it also includes the question of whether a project will lead to any outcome whatsoever. On the other end of the scale we find projects like the implementation of a known software system, where basically a standard procedure can be applied with more or less well-known result. As is clear from the discussion above, no company can easily be grouped according to one situation. Instead, we need to draw on ideal type reasoning in order to explain the variances observed in our cases. We would argue that the companies have some aspect of dependency and uncertainty, but that they differ in some respect along these lines. Combining the dependency - uncertainty dimensions gives rise to the following matrix: /Insert Figure 1 here/ The matrix above gives us four types of situations. Given the characteristics of each situation we can now link each such situation to a suggested appropriate control strategy. Since all MPOs are thought to contain dependencies and uncertainty, the resulting typology can best be thought of as a staircase, where the control instruments used in the ‘simplest’ situation (I) are also to be found in the most complex (IV). 31 In situation I we find a situation where projects show a high degree of independence and low uncertainty. Empirically we find these MPOs involved in rather standardized projects with little resource sharing or exchange of services. To say that the projects are standardized does not necessarily mean each project needs to be identical to other projects. In the construction industry not all buildings are alike, but they do rely on a common knowledge base and common methodology. From a control perspective the limited inter-project dependency poses no real problem. The control problem for the organization is instead to maintain the focus on the routines and standard operating procedures. In MPOs the strongest instrument for this is the project model. Each project is required to pass through certain phases which end with a tollgate decision and include pre-specified activities that should produce the basis for the tollgate decision. All our case companies use the same project model, PROPS (or variants thereof). Other means of standardization can be the use of planning constants (Ericsson). The control strategy for this type of situation can be called Routine-based Control. Situation II differs from the previous in the sense that although the projects still display a low degree of uncertainty; they are dependent on each other. Very often this dependency takes the form of one project being dependent on activities in another project being completed before the first project can start or to progress e. g. a certain development project requires for its completion a module which is processed by another project. Since uncertainty is low, there is in this situation a possibility to use plans as a coordinating mechanism. An example of this can be found in Ericsson where all the functionality and the dependencies between the functions are described in what is known as anatomy plans. The plan consists of rings where each ring denotes a certain functionality and where each ring is dependent on the functionality defined by the neighbouring ring closer to the centre. In case of lack of time, the outermost rings can be excluded. In the case 32 that not all coordination can be solved by plans, MPOs in this quadrant turn to simple coordinating devices such as coordinating committees. Because of the reliance on plans, the control strategy for this situation is termed Planning-based Control. In contexts with high uncertainty (III and IV), plans can no longer be relied on as the main control mechanism since plans require a certain level of stability. Because of the uncertainty controlling requires close attention to the extent that the organization’s management does not have the ability to monitor the projects personally. The uncertainty therefore typically requires a decentralized control where project management is vital for the success of a project. If projects are rather independent from each other (III), no special inter-project coordination needs to be established and the control issues therefore centers more or less on the question of controlling a portfolio of projects each with a high degree of uncertainty. Empirically we believe that such a situation can be found in the pharmaceutical or bio- technical industry where each development project has rather uncertain prospects, but where each such project is basically self-contained. From a control point of view, standards could be set and plans be made but they are of limited use since the intensity of the activities and the results thereof tend to follow their own logic and not some pre-determined plan. Rather than with output and/or process, the basic control mechanism is instead to be found on the input side. There are two fundamental resource decisions that lay the ground for control in this case. These are the choice of project manager (management) and the resources allocated to the projects. Since uncertainty precludes the reliance on plans, the problem of controlling the project becomes the problem of choosing a project manager which to the best of his abilities acts in the way that the management of the organization would have done had they had the time and competence. The focus here will be very much on the belief system of the project management team. We call this control strategy Resource-based Control. 33 Finally, when dependencies are high and uncertainty great, the resulting situation (IV) is the most exacting from a control perspective. The decentralized control through project managers from the previous situation must be retained, but on top of that because of inter-project dependencies, some means of coordinating this dependency must be found. Empirically we might expect to find these situations where projects contain elements close to basic research and where development projects within different functional areas display reciprocal dependencies like in the aircraft industry. In cases like this the solution lies in introducing what Thompson calls an “over-arching second-order group” (p. 59). Such a group can also be found at Saab FP that has created the function of program management whose task it is to coordinate between products. The control strategy is therefore named Program-based control. The following figure and table summarizes the discussion. /Insert Figure 2 here/ /Insert Table 2 here/ 8. Conclusions In many contemporary firms, projects are playing an increasingly important role. The understanding of the specific management control activities of project-intensive companies, or multi-project organizations, are however limited. Some researchers have even gone argued that the management procedures have to be re-conceptualized from their roots in order to fit the multi-project organizations. In this paper we build on classic contingency theories and recent findings in multi-project organizing in order to develop a conceptual framework for analyzing one layer of the management control in such environments. Our framework points specifically to the importance of project uncertainty and inter-project dependency in explaining the control 34 strategies applied by R&D companies. Our study thus gives valuable insights into a research theme to which earlier research has paid scant attention. In much contingency writings the importance of mutual adjustment is emphasized. The control logic is however underspecified and in order to handle the management control problematics illustrated in this paper, we pointed to a much broader variety of control strategies, including the role of routines, planning and input control. The paper also demonstrates and explains the role of program management, a management control function that seems to be utilized in many contemporary multi-project organizations. In this sense, the paper contributes to the understanding of the relationship between project management and program management. The paper presented a typology identifying four different control settings. We argued that the basic logic for management control in multi-project organizations, would be the routine-based control. In many multi-project organizations, however, due to uncertainty or dependency, additional control strategies are required. In the paper we offered three additional control strategies to complement the routine-based, which was observed in all of the studied companies. The second control strategy found in situations characterized by a higher degree of dependency, we argued that a planning-based control strategy would be most likely. Here we stated that the observed interdependency between the projects would be possible to handle with increased planning efforts due to the limited uncertainty associated with each of the projects. When uncertainty was introduced, more sophisticated control strategies were called for. In situations characterized by high uncertainty, but low degree of dependency, it would be possible to assume that a resource-based control strategy would be suitable due to the limited need of coordinating between projects. In the final cell, we argued, a more interactive control strategy seems appropriate due to the mutual handling of uncertainty and dependency. We suggest to label this control strategy, program-based, to stress the additional layer of control frequently observed in such situations. 35 Our findings and the presented model are, of course, limited in several respects. Future studies, we would argue, should try to further the discussion on various types of multi-project organizational settings in order to create possibilities for establishing a contingency perspective on control in such contexts. The proposed framework, we would argue, might be utilized in future studies that take a broader approach towards the control strategies in R&D and professional companies. Our framework especially illustrates the importance of not only studying management control on the project-centric level (cf. Davila, 2000) and company-wide aspects of project management systems (Simons, 1995:114). Our framework thus contributes to the present understanding by illuminating the intersection between the control of single projects and the organization of project management systems in R&D companies. 36 References: Anell, B. (2000). Managing project portfolios. In R. A. Lundin & F. Hartman (eds.), Projects as business constituents and guiding motives. Boston: Kluwer Academic Publishers. Archibald, R. (1976). Managing high-technology programs and projects. New York: Wiley. Baxter, J. and W. F. Chua. 1998. Doing Field Research: Practice and Meta-Theory in Counterpoint. Journal of Management Accounting Research, 10, 69-87. Chenhall, R. H. (2003). Management control systems design within its organizational context: findings from contingency-based research and directions for the future. Accounting, Organizations and Society, Vol. 28, 127-168. Cusumano, M. & K. Nobeoka (1998). Thinking beyond lean: how multi-project management is transforming product development at Toyota and other companies. New York: Free Press. Davila, T. (2000). An empirical study on the drivers of management control systems’ design in new product development. Accounting, Organizations and Society, Vol. 25, 383-409. Eisenhardt, K. M. (1989). Building theories from case study research. Academy of Management Review, Vol. 14:532-550. Engwall, M. (2003). No project is an island: linking projects to history and context. Research Policy, Vol. 32, No. 5, 789-808. Engwall, M. & A. Jerbrant (2003). The resource allocation syndrome: the prime challenge of multi-project management? International Journal of Project Management, Vol. 21, 403-409. Ford, R. & W. A. Randolph (1992). Cross-functional structures: a review and integration of matrix organization and project management. Journal of Management, Vol. 18, No. 2: 267-294. Galbraith, J. R. (1973). Designing complex organizations. Reading: Addison-Wesley. Gerdin, J. & J. Greve (2003). Forms of contingency fit in management accounting research: a critical review. Accounting, Organizations and Society. Forthcoming. Hobday, M. (2000). The project-based organization: an ideal form for management of complex products and systems? Research Policy, Vol. 29: 871-893. Katz, R. & T. J. Allen (1985). Project performance and the locus of influence in the R&D matrix. Academy of Management Journal, Vol. 28, No. 1: 67-87. Keating, P. J. (1995). A framework for classifying and evaluating the theoretical contribution of case research in Management Accounting. Journal of Management Accounting Research, Vol. 7, 6686. Knight, K. (1976). Matrix organization: a review. Journal of Management Studies, Vol. 13, No. 2, 111130. Lawrence, P. R. & J. W. Lorsch (1967). Organization and Environment: Managing Differentiation and Integration. Boston: Harvard University Division of Research. Leonard-Barton, D. (1990). A dual methodology for case studies: synergistic use of longitudinal single site with replicated multiple sites. Organization Science, Vol. 1, No. 3, 248-263. Lindkvist, L., J. Söderlund & F. Tell (1998). Managing Product Development Projects: On the Significance of Fountains and Deadlines. Organization Studies, Vol. 19, No. 6, 931-951. Lundin, R. A. & A. Söderholm (1995). A Theory of the temporary organization. Scandinavian Journal of Management, Vol. 11, No. 4, 437-455. Macintosh, N. B. & R. L. Daft (1987). Management Control Systems and Departmental Interdependencies: An empirical study. Accounting, Organizations and Society, Vol. 12, No. 1, 4961. Meyer, M. & A. Lehnerd (1997). The power of product platforms. New York: Free Press. Midler, C. (1995). Projectification’ of the firm: the Renault case. Scandinavian Journal of Management, Vol. 11, No. 4, 363-376. Mintzberg, H. (1983). Structure in fives. New York: Prentice-Hall. 37 Nandhakumar, J. & M. Jones (2001). Accounting for time: managing time in project-based teamworking. Accounting, Organizations and Society, Vol. 26: 193-214. Normann, R. (1976). In Search of a Methodology. Stockholm: SIAR. Packendorff, J. (1995). Inquiring into the Temporary Organization: New Directions for Project Management Research. Scandinavian Journal of Management, Vol. 11, No. 4, 319-334. Pinto, J. (2002). Project management 2002. Research Technology Management, No. 2, 22-37. Scott, W. R. (1965). Reactions to Supervision in a Heteronomous Professional Organization. Administrative Science Quarterly, Vol. 10, 65-81. Shenhar, A. & D. Dvir (1996). Toward a typological theory of project management. Research Policy, Vol. 25, 607-632. Simons, R. (1995). Levers of control. Boston: HBS Press. Söderlund, J. (2003). On the broadening scope of the research on projects: a model and a review, International Journal of Project Management. Forthcoming. Thompson, J D. (1967). Organizations in Action. New York: McGraw-Hill. Turner, J R. & A. Keegan (2001). Mechanisms of Governance in Project-based Organization: Roles of the Broker and Steward. European Management Journal, Vol. 19, No. 3, 254-267. Van der Merwe, A P. (1997). Multi-project Management - organizational structure and control. International Journal of Project Management, Vol. 15, No. 4, 223-233. Williamson, O E. (1997): The Modern Corporation. In D. Pugh (Ed.), Organization Theory, London: Penguin Business. Whittington, R., A. Pettigrew, S. Peck, E. Fenton, M. Conyon (1999). Change and complementarities in the new competitive landscape: a European panel study. Organization Science, Vol. 10, No. 5, 583-600. 38 Typical length of project Type of dependence between projects Organization al structure Standardized project model Key traits of multi-project organization Saab Future Products 5 years Ericsson BSC 1-2 years Ericsson SRF 3 years Telia Mobile 1 year Limited Mutual + sequential Mutual Mutual Matrix Matrix Functional Purpose Yes (PSM) Yes (PROPS) Yes (PROPS) Yes (PROPS) Project manager reports to department head and program manager Management team considers the fundamental way of control-ling projects as “selecting” the right ones Resource estimation is based on standardized areas Balanced score-card has been implemented PSM is implemented and used in most cases Systems department in charge of launching projects Two project offices are part of the organization Geographically dispersed units effect the control One steering committee with several members Steering committee makes decisions on the projects’ strategies, directives and changes Smaller projects have increased in number in recent years Strategic Product Management responsible for product range Radio Development Board, centralized forum, making all major decisions in projects Each sub-unit has an operative steering committee Project management office at decentralized level Operative planning resides at the project management office Manager of project office responsible for each unit’s project portfolio Sub-projects often geographically dispersed High dependencies both within and outside the SRF unit Main logic of control Line manager dominates and assigns resources Program Manager coordinates between projects Steering committee Planning constants Critical resources are not allocated Table 1. A cross-case comparison 39 Manager for project office (staff function) Sixteen project managers in a project management pool Project management office, one of several staff functions Each business unit has a steering committee A project coordinator is responsible for contacts between projects Many projects of many different types Project web used for control and follow-up purposes Much work has been put into the project selection process NASA: a group for making priorities between projects Priority list divided into four categories Steering committee at business unit level Special coordination teams created when interproject dependence is high Routine-based control Control setting Low dependency between projects. Low project uncertainty. Primary control Routines, project mechanism management models. Key function Project controllers Planning-based control High dependency between projects. Low project uncertainty. Planning Resource-based control Low dependency between projects. High project uncertainty. Self-contained projects, input control Project planners, Project managers project coordinators Table 2. Summary of findings: four control strategies 40 Program-based control High dependency between projects. High project uncertainty. Continuous dialogue, program management. Program managers Project uncertainty Low High Low I III II IV Dependency between projects High Figure 1. Four control settings: dependency and uncertainty 41 Project uncertainty Low Low Dependency between projects High High Routine-based control Resource-based control Planningbased control Program-based control Figure 2. Control strategies in multi-project organizations 42