Our research shows that the most frequently covered topic in OM courses is Inventory Management (Ch.12). In that chapter, we do discuss the importance of record accuracy, cycle counting, and shrinkage. But what we do not discuss is the use of the numerous inventory models if inventory is only “partially observed”.
In today’s issue of Decision Line (Oct., 2010), an excellent article by Prof. Suresh Sethi, at U.Texas-Dallas, goes into the reasons for partial observation of inventory levels and then discusses how these impact modeling efforts. Here are 5 causes Suresh details:
1. Sales recorded wrong (eg, a clerk scans an item twice, when there were actually 2 different flavors of soup).
2. Misplaced inventory (eg, when items are stored dynamically, not in a fixed location). Suresh tells of a top retailer who discovered 16% of its items were misplaced.
3. Spoilage (eg, when customers tear open a package to look at the item inside, spill drinks on clothes, or scratch a car they test drive).
4. Product quality and yield (eg, when some items coming into the warehouse are unknowingly damaged).
5. Theft (eg, break-ins, employee pilferage, and customer shoplifting). The Limited, eg., recorded an inventory discrepancy of $142 million a few years ago—the equivalence of 21,000 ocean containers!
Suresh concludes, “By now it should be clear that the incomplete inventory information (i3) problem is quite common in practice, that policies in current use are neither optimal nor applicable”. He finishes the article by discussing 4 ways to classify i3 problems.
The real point worth making in class is that the models we discuss in Ch.12 depend on accurate record keeping, which may be impossible in a variety of real world companies.