Companies Still Paying a High Price for Flawed Inventory Policies

Optimizing inventory management policies can achieve dramatic reductions in operational costs, yet many companies leave substantial savings on the table because they fail to develop the most efficient management policy for their business.

“Finding the best policy to control inventory systems is a complex problem, but companies can develop solutions that are much closer to the optimum by starting with a simplified case and taking a more flexible approach to the challenge,” says Dr. JianJun Xu, PhD Summer Academy Program Director and Assistant Professor at Zaragoza Logistics Center, Zaragoza, Spain.

Xu is studying optimal inventory policies using analytical inventory models. Although these models have been studied for several decades, “there is still huge room for improvement because it is not uncommon for companies to be using inefficient or inappropriate inventory policies,” he says.

There are a number of reasons why companies still struggle to devise the most efficient approaches to inventory management. Examples include the use of inaccurate data, poor coordination between supply chain partners, a lack of expertise, and difficulties in implementing optimal inventory policies.

A key reason is that analyzing the various policy options is a major challenge when multiple SKUs and various warehouse and manufacturing facilities are involved. When there are many variables the sheer number of possibilities makes it very difficult to develop and implement the most efficient algorithms for managing inventory. Even when software is used to compare the options, the calculations can be complex and take a lot of time. In many cases, companies settle for inferior solutions because the best policy is too difficult to implement, or they are unaware that their solutions are sub-optimal.

System flexibility is critically important yet often ignored by companies, says Xu. Examples are the use of interchangeable products – a supermarket that can stock different brands of the same product or a manufacturer that is able to substitute a component for a similar part – and the ability to make more than one product in a production facility. Xu is studying the optimal structural properties that are part of flexible production systems. By gaining a better understanding of these properties in relation to system flexibility, he can help companies to simplify inventory policy computations.

Consider, for example, a relatively straightforward manufacturing scenario where there are two finished products and two manufacturing facilities. The base inventory policy is that each facility makes one product. Next, we develop two heuristic polices for two cases with different production costs. In Case 1, each facility can cost-effectively make its own product but incurs excessive cost when producing the other product. In Case 2 one facility can cost-effectively manufacture both products, but the second facility incurs high costs when making the two products. The base case is the worst performer. The two heuristic cases out-perform the base case, but Case 1 is closer to the optimum solution.

Creating heuristics in this way enables companies to develop inventory policies more systematically, and simplifies the calculations.

The approach does have limitations, especially where there are very large numbers of SKUs and manufacturing combinations, and the market is extremely dynamic. Also, introducing flexibility such as plants that can make more than one product requires some investment, and companies have to evaluate the trade-off between the additional investment required and the cost gains achieved by improved inventory policies.

Even so, “if a company is finding it difficult to develop an optimal inventory management policy, I believe that by starting from a simple case we can help them to improve and at least get much closer to the best solution,” says Xu.

Simplifying the problem in this way also helps to identify other factors that are impacting inventory performance. Say a company wants to improve its customer service level by increasing product fill rates. Its inventory management practices must be considered, but other variables such as the efficacy of its demand forecasting system and production efficiency might be affecting fill rate performance.

Going forward, Xu wants to apply his models to more complex scenarios that include many products and facilities. “The basic logic will be the same, but I want to extend the model to bigger and more complex supply chains,” he says.

There is also potential for applying a similar methodology in other areas. For example, in the development of easy-to-implement solutions to production substitution and inventory transshipment problems.

For more information on the inventory policy research contact Dr. JianJun Xu.