Inventory Management in a Stochastic Demand Environment

Dec. 11, 2015
Consistent review and adjustment of inventory re-stocking parameters through a degree of human intervention, will have the greatest impact and allow for a well-managed inventory under stochastic demand.

In a steady and predictable manufacturing or service environment, determining the correct amount of inventory can be done using historical data and forecasted demand. In transit repair operations, or any other non-repetitive demand business, the correct amount of inventory will be more difficult to determine.

When a company’s production is unpredictable, the inventory management becomes challenging. Normal operational requirements must be met, and factors that could cause stochastic demand need to be considered. Stock-out situations are amplified since they usually occur during a crisis or equipment failure event. There is a way to adapt an existing inventory management database system to better serve the needs of a stochastic operation. During the best of times inventory management should always consider the “what-ifs”, however, in a stochastic demand environment, the “what-ifs” become more commonplace and inventory management must not only consider them, but plan for those factors which are variable, dynamic and difficult to control.

Research Theories

Research on inventory management relies heavily on mathematical and theoretical formulas. The theories appear sound, have much data, and conclusions are proven with tables and graphs. These theories are less than desirable when considering a stochastic environment. The research that relies heavily on math also relies heavily on assumptions to make the math work properly.

The assumptions made are for the purpose of creating a more stable, less stochastic situation on which to test the theory. For example, one article uses the assumptions of “constant annual demand rate” and “all customers are willing to wait” to make the formulas work. An article on Just-In-Time (JIT) inventory listed twelve assumptions smoothing out supply and demand. The problem is, none of these assumptions actually exist in a real-life stochastic demand operation.

Consistent Review

There is a clear link between regular review of the inventory parameters and successful outcomes in a non-repetitive demand shop. Monitoring erratic demands are necessary in order to develop a workable approach that balances service levels and financial concerns. Inventory’s response to stochastic demand must be influenced with a scheduled and systematic review of the inventory control parameters. This can fill the gap and create an involved control. Technology has created a situation where production cycles are shorter and inventory management becomes more challenging. Planning production in a job-shop system is complex and creates an equally complex situation for inventory management. There is an inability for traditional database management software to keep up with real-time events and adjust quickly enough to repair based environments.

Stochastic, non-repetitive demand can happen in many environments. Healthcare, humanitarian supply chain, job-shops, transportation and transit organizations, and other repair facilities. An output value, or signal, of a suggested replenishment given to a Material Planner or Analyst for processing demand is the important feature of most inventory management systems. The goal for stochastic inventory demand is a signal process “which represents the practice of dynamically estimating the demand and subsequently updating the parameters of inventory control policy." Consistent review of inventory reorder parameters is being shown as the preferred method to better control the inventory management in a stochastic environment. Gaining control of the parameters will improve stability of the entire supply chain.

Game theory gives a basis for setting up the process for consistent review and making adjustments. Game theory, although mathematical in nature, comes down to constant review of the process in real-time and interactive decision making. Game theory also shows that some of our decisions depend on the choices of others and that decision situations are interactive. Production and inventory planning in a job-shop environment has this type of intricacy. The decision making relies on shop flows with a tendency for change. Decisions are based on a variety of factors that require “continuous monitoring of the system’s actual state.” Stochastic demand has a need for human intervention. You must develop a hybrid inventory policy that incorporates continuous review of your actual state. One established inventory management system incorporates the idea of evaluation and revision. Reorder Point (ROP) systems continuously review the in-stock quantity against the reorder point quantity. If the activity changes, the system’s “dynamic adjustments of the replenishment system” will catch a surge in usage. However, even this more dynamic system catches the surge after the fact, and adjustments need to be made to both quantity and timing with independent involvement.

A company must identify the volume of the non-repetitive demand processes. Once it is determined that volatile demand exists there can be an acknowledgement that the operation, or part of the operation, is stochastic. If this acknowledgement does not occur, an organization will continue to use standard inventory management software and become frustrated that the program is not working well. If the true nature of the organization is honestly identified as stochastic, there can be the realization that no packaged database management system can do everything that is needed. More important than the realization, companies can now begin to develop the policies, models, and procedures required to become better in the execution of supply chain and inventory strategies. Stochastic inventory models must include a process to consistently review the inventory management parameters. There must be a method of monitoring and adjusting those parts of an inventory database management system that signals replenishment requirements to the user.

Establish Your Metrics

Whatever inventory system is used, the questions that must be answered by inventory management fall into the categories of demand and lead time. How many and when? Every modification made to create a hybrid system, should make it easier to answer those two basic inventory management questions.

Inventory management is complex. Answering the how many and when question is difficult. Discouraging the negative perceptions of stakeholders is daunting. As with any important process, metrics should be established. Developing metrics for the key performance indicators of the inventory management process will reduce the amount of negative perception associated with inventory control. The metrics should be focused toward a company’s specific and anticipated results. These metrics must be regularly reviewed to ensure the new processes are on track and providing the desired outcomes. There should be evidence that the consistent review and adjustments of inventory re-stocking parameters through a degree of human intervention are having a positive impact on the inventory management process.

Michelle Berry is a manager of inventory with the Greater Cleveland Regional Transit Authority.