Multiple Sourcing Alternatives Using Nearly Optimal Programming
Taylor & Francis Ltd.
The Journal of Education for Business
A linear programming (LP) model representing multiple sourcing is investigated. The LP represents the allocation of order quantity for three potential vendors. The optimal solution uses two of three vendors, with 75% of the order being supplied by the second vendor and the remaining 25% by the third. Nearly optimal programming (NOP), a form of sensitivity analysis not generally covered in management science or operations research texts, is applied to the LP model. We show that with only minor modifications to the model, a variety of new solutions can be generated that represent distinct solutions from the optimal. These distinct solutions provide the manager and the business student with alternatives, which could be useful if there are problems with specific vendors.
Managers and business people frequently have to make decisions based on the results derived from quantitative analysis. There are various methods for different applications, but linear programming (LP) historically has been of great importance. For example, in a survey of Fortune 500 companies, 85% of the firms that responded used LP (Winston, 1991). In addition, Anderson, Sweeney, & Williams (1994) reported on surveys that show that 83.8% of management science practitioners and 74% of corporate executives have used LP. Because of this emphasis on LP in the corporate world, it is an integral part of business education at the college and university level. For example, it is taught in quantitative analysis, production and operations management, actuarial operations research, and managerial and cost accounting courses.
Our purpose in this article is to present an underused form of sensitivity analysis in LP, called nearly optimal programming (NOP). This is a method that is not generally covered in management science or operations research texts, but that a manager could easily use, and that can generate a variety of solutions that are significantly close to the optimal solution but present distinct business options. A small straightforward example for vendor allocation demonstrates the benefit of this technique.