|
Article Excerpt
1. Introduction
The engineering design problem most often examined in the literature on concurrency, uncertainty, and coordination is a two-stage operation where upstream operations provide elements of the project to downstream operations. Development activities in projects where operations in the two stages are overlapping and interdependent is concurrent design, and concurrent design involves the use of preliminary design information for coordination (Krishnan and Ulrich 2001). Concurrency in design operations is challenging to manage because interdependent design decisions may be ignored or may change over different stages of development. These interdependent design decisions often lead to rework--which we define as design change whose implementation alters work that was previously done upstream and downstream. Rework has a frequency dimension (number of change iterations) and a magnitude dimension (amount of change) relative to the original design. Rework can occur in one or more stages and this can lead to project delay, where project delay is defined as the time between planned and actual project completion.
Two aspects of concurrent design affecting project completion, through design rework, are uncertainty and coordination. Uncertainty and coordination are related, as greater uncertainty requires greater coordination to define and implement the design (Galbraith 1974). Analytical models indicate that project delays result from excessive iteration for rework related to uncertainty (Krishnan et al. 1997), and excessive communication for coordination (Loch and Terwiesch 1998). Early resolution of uncertainty, when design requirements are known and stable, mitigates the risk of project delay (Terwiesch and Loch 1999). However, uncertainty from fit novelty, where design exceptions increase with the organization's inexperience with the design's fit issues (Adler 1995), and unforeseeable uncertainty, where relevant variables and their functional relationships cannot be recognized (Sommer and Loch 2004), can both prohibit early resolution. Under these conditions, different forms of coordination--cooperative planning and dynamic adjustment--may be the most effective means of dealing with uncertainty.
Product design is defined as a transformation of a market opportunity and technology possibilities into a design solution (Krishnan and Ulrich 2001, Gerwin and Barrowman 2002), and the focus in the literature has been on the design of complex products such as automobiles, aerospace systems, software, and industrial equipment (Mihm et al. 2003). This design shares the same characteristics as IT-dependent process redesign projects in dynamic environments. Krishnan and Bhattacharya (2002) describe two-stage design as an upstream product definition stage where input data and information about customer needs and emerging technologies are used to finalize key specifications, and these specifications are used downstream in detailed design and prototyping. Similarly, business process (BP) redesign and information technology (IT) development involves gathering and redefining business rules and functional requirements together with emerging technology options upstream, and using these requirements and technology options to develop detailed IT design and implementation solutions downstream.
One means of controlling the amount of rework resulting in project delay is to mitigate the uncertainty associated with fit novelty. A primary concern in BP redesign and related IT development is the fit between IT capabilities and BP requirements--a form of fit novelty. Uncertainty surrounding the fit between BP and IT increases when BP design parameters are more sensitive to IT decisions (or vice versa), when an organization has less experience resolving problems related to BP-IT interdependence, and the more innovative the BP design or IT platform. Consequently, greater fit novelty requires greater coordination between upstream and downstream operations (Adler 1995).
In the early phases of a project (visioning and design), an effective means of coordination is the pre-communication between upstream and downstream operations (Loch and Terwiesch 1998). One form of pre-communication is cooperative planning, where the early resolution of uncertainty is contingent on the level of strategy coupling and cross-functional involvement (Mitchell and Zmud 1999). Strategy coupling facilitates the exchange of preliminary information necessary for early requirements specification, and cross-functional involvement in the planning process promotes information precision and stability in setting design specifications. The degree of cooperative planning is a coordination mechanism that management controls.
Prior concurrency research has highlighted the effects of uncertainty and coordination on the frequency of rework and project delay primarily though analytical modeling. We conduct a confirmatory field study at the micro-level that empirically tests these theoretical relationships to determine the extent to which, in practice, the magnitude of rework and project delay in concurrent design is under managerial control. Our research is guided by the following questions: (1) Does greater cooperative planning reduce the magnitude of upstream and downstream rework, and does it mediate the relationship between upstream and downstream rework? (2) Does greater uncertainty increase the magnitude of upstream and downstream rework? (3) Does the magnitude of rework affect the duration of project delay?
We study 120 BP redesign and IT development projects in the healthcare and telecommunications sectors where upstream BP design and downstream IT platform design are interdependent. Using a Partial Least Squares (PLS) model and a magnitude of rework scale we develop based on Henderson and Clark's (1990) types of design change, we find strong results about the decoupling of downstream delay from upstream uncertainty through cooperative planning, and about the impacts of uncertainty and complexity. Specifically, we find that greater cooperative planning reduces the magnitude of both upstream and downstream rework. We also find that greater uncertainty from a lack of prior experience with the design affects the magnitude of upstream rework but does not significantly affect the magnitude of downstream rework. Moreover, the PLS analysis indicates that cooperative planning mediates the relationship between upstream and downstream rework such that in the presence of cooperative planning the direct effect of upstream rework on downstream rework is not significant. In addition, we find that more extensive downstream rework leads to longer downstream delay, and these in turn lead to longer overall project delay. In contrast, more extensive upstream rework does not directly lead to longer project delay, but does lead to longer downstream delay. These results suggest that the duration of project delay is under managerial control, primarily through cooperative planning that reduces the magnitude of upstream and downstream rework as well as downstream delay.
The remainder of this paper proceeds as follows. First, we examine the literature on concurrent design, highlighting studies of uncertainty and of coordination. Next, we provide an example illustrating BP-IT interdependencies. Then, we describe our methodology with the research model, instrument development, and data collection. Subsequently, we present our PLS analysis and our results. We finish with our discussion and conclusion summarizing the results, outlining the limitations of our study, and describing the implications for future research and for practice.
2. The Literature on Concurrency, Uncertainty, and Coordination
In a classic study of the auto industry, Clark and Fujimoto (1989) describe the product development process as a set of information processing activities for problem solving. In an overlapping, or concurrent, approach upstream operations release preliminary information to downstream operations before the upstream information is finalized. This preliminary upstream information is used downstream to forecast final outcomes of upstream activity. Downstream activity based on this forecast starts before upstream output is complete and may provide feedback upstream. Inaccurate downstream forecasts of upstream output result in rework, whereby work previously done downstream based on preliminary design forecasts must be updated with the revised upstream output results. Hence, concurrency without appropriate coordination may reduce design quality and create project delay due to mismanagement of information and uncertainty between upstream and downstream activities.
Henderson and Clark (1990) delineate four categories of technological design change in product development--incremental, modular, architectural, and radical--whose progression is related to the magnitude of effort in altering the core design of system components and/or how those components are linked together. Borrowing these categories to define the magnitude of rework, we relate the degree of upstream and downstream rework to an incremental-radical continuum that captures the level of deviation from the intended process (or platform) design. Where incremental rework involves minor design change, modular, architectural, and radical rework involves major design changes to varying degrees. Modular rework is limited to major changes in process or platform components, architectural rework entails major changes to linkages, and radical rework involves major changes to both components and linkages. Henderson and Cockburn (1994) argue that due to complexities of system interdependence, design change involving linkages (architectural and radical) are more difficult to enact than those confined to components (modular).
The magnitude of rework reflects the nature of design uncertainties left unresolved, in part due to project complexity. In a model examining limits to concurrency and the causes of cycle time expansion, Hoedemaker et al. (1999) find that the usefulness of concurrency is limited by project complexity. In concurrent design, the communication burden increases in proportion to the degree of task subdivision, and a higher degree of task subdivision increases potential integration problems and the likelihood of rework. Joglekar et al. (2001) find that concurrent engineering need not be the optimal work strategy in many settings and management must consider information exchanges, rework issues, performance thresholds, and resource restrictions.
2.1. Uncertainty
When downstream operations start work before upstream operations have finalized their problem solving, uncertainty related to fit novelty can result from a lack of prior experience with the design task (Adler 1995). In this context,...
|
