Safety Stock: Period Coverages or Fixed Quantities?

Planners usually set targets as dynamic period coverages or fixed quantities. As described in this article, both techniques have their pro and cons. In both cases, you will have to rely on a solid automated quantitative model to refresh your targets regularly and focus on exception and alert management.

Before discussing how to express safety stocks, let’s recap some critical points. Many planners and practitioners confuse safety stocks, reorder points, and order up-to level. These three concepts are intertwined but are different. Let’s review them one by one:

Orion Nebula (credit: my father, picture taken from his garden in Brussels, Belgium)

• Safety stock. Inventory to protect yourself against demand and supply variability over the risk-horizon. (I prefer the term risk-horizon, which depicts the total lead time and the review period. I coined it in my book Inventory Optimization: Models and Simulations.)
• Reorder point. In a reorder point policy, you will make an order when your inventory level reaches the reorder point. The reorder point is computed as the expected demand (or forecast) over the risk-horizon plus your safety stocks.
• Up-to-level. In an up-to-level ordering policy, each time you make an order, you have to order enough units to reach the up-to-level. The up-to-level is computed as the expected demand over the risk-horizon plus safety stocks.
  For example: if you have a lead time of four weeks and make weekly orders, your up-to-level should be five weeks of demand (technically, your five-week cumulative forecast) plus any safety stock.

Planners usually set inventory targets either as period coverages (“2 weeks of safety stocks”) or as fixed quantities (“20 units of safety stocks”). In both cases, you initially get your target based on the usual safety stock formula (or more advanced models). This formula will give you a quantity that you can then translate into a coverage period. Initially, the period coverage and the fixed quantity would be the same (the difference would be artificial). But over time, as the forecast changes, a disparitywill appear. So, which one should you use? Let’s review the pros and…