|
Article Excerpt
1. Summary and Literature Review
Commonality strategies assemble different products from at least one common component and one other product-specific component. Analytic studies of commonality traditionally compare two distinct models: a no-commonality model, where each product requires two product-dedicated components, and a pure commonality model, where each product requires one dedicated and one common component. The value of commonality is then explained in terms of the risk-pooling benefit in the pure commonality model. Clearly, this benefit comes at a cost, as the common component is more expensive than the dedicated ones that it replaces. The traditional requirement to adopt commonality is that the optimal value of the commonality model exceeds that of the no-commonality model, yet no simple conditions have been available.
This note analyzes a single unified model with five inputs and two products, as shown in Figure 1, that captures these two models as special cases. Under the no-commonality strategy, product i [member of] {1, 2} uses dedicated components i and i + 3, while under the pure commonality strategy it uses common component 3 and its other input i + 3. In general, however, the model allows for stocking both dedicated and common components and for alternate assembly allocations by substituting components 1 or 2 for the common component 3. Hence, we will refer to inputs 1 and 2 as substitutable inputs, while inputs 4 and 5 may be called always-dedicated, or simply "other," inputs. The required presence of other inputs in addition to a common input is the defining feature of commonality that distinguishes it from substitution flexibility. At their core, capacity flexibility, inventory substitution, and dual sourcing refer to having a second option to fill a product demand; they typically do not require the assembly of multiple inputs. In other words, those strategies only have inputs 1, 2, and 3 in a two-product setting. (Among many, Bassok et al. 1999, Tibben-Lembke and Bassok 2002, and Van Mieghem 1998 (hereafter abbreviated as VM98), analyze substitution flexibility.)
This note shows that the distinguishing feature of commonality, i.e., the required presence of the other components 4 and 5, is mathematically inconsequential in the sense that the unified commonality problem is equivalent to the substitution flexibility problem of VM98. This equivalence is first established in a single-period model and then extended to a multiperiod setting with i.i.d. demand and negligible leadtimes. The equivalence between a five- and three-variables problem is mathematically surprising and significantly simplifies the analysis of commonality. It also establishes a precise relationship between different literatures that intuitively are related, yet have been developed in isolation. The key advantage of the unified model is in allowing the optimal strategy, service levels, and commonality adoption conditions to emerge through direct optimization. This note presents the first (1) general, closed-form condition for when commonality should be adopted, and identifies its driving factors. In addition, contrary to earlier statements in the literature, (2) commonality can be valuable even with perfect correlation when risk pooling is impossible. The explanation is found in the concept of a revenue-maximization option that is introduced as another benefit of commonality, independent of risk pooling. The note ends by describing the dual sourcing, externalities, and operational hedging features of optimal commonality strategies.
[FIGURE 1 OMITTED]
Commonality falls within the broader supply chain operations umbrella of assemble-to-order systems, which are reviewed by Song and Zipkin (2003) and combine elements of assembly and distribution systems. Commonality research comes in three forms: parsimonious analytic studies, detailed mathematical programming formulations, and product-design studies....
|
