Why Do We Need Safety Stock?

As I am in the process of writing a series of articles introducing a new formula to improve upon the existing safety stock calculations, an old recurring question surfaces again: why do we even care about safety stock?

I get this question a lot in response to articles or in training sessions on the topic. In this article I will answer it.

What is Safety Stock?

Ask five people this question and chances are you get five different answers. If you google for definitions you may find any of the following:

• "inventory that is carried to prevent stock outs"
• "inventory to buffer for forecast error"
• "inventory to buffer for demand variability"
• "inventory to buffer for demand uncertainty and supply variability"
• "inventory to buffer for demand uncertainty and supply lead time variability"

Then there are those who consider only the exact formula used in their ERP or MRP system of choice to be called safety stock, and anything else to not be, even if it has the same purpose. There are also people who differentiate between raw materials (RM), work-in-progress (WIP), and finished goods (FG), saying it is only safety stock if it is held for WIP (for example), but call it something else for RM or FG. Finally, often a distinction is made between safety stock, buffers, and decoupling points.

I pose that all are merely flavors of the same thing. Depending on its use the formula may need to be different, but they all intend to achieve part or all of the same thing. I use a definition that is both broader and more specific than most:

Safety stock is a minimum level of inventory for individuals items in individual locations targeted during planning, with the objective to minimize the risk of stock out caused by demand uncertainty and supply uncertainty.

Stated simply, safety stock is an extra amount of product you plan to hold on hand just in case demand is higher or supply is lower than expected, so that you do not run out. For a more complete overview in simple terms of common inventory concepts and how safety stock fits into the big picture you may want to check out the article "Inventory Management for Dummies".

Let's dig into the above definition, one bold term at a time. Safety stock is NOT inventory itself, or part thereof. It is a target or minimum level that we aim to keep the real inventory above. We use it when we plan future purchases, production and distribution. I often see or hear the statement that Inventory = Safety Stock + Cycle Stock + (sometimes) Strategic/Seasonal Stock. This erroneously suggests safety stock is part of inventory, rather than a target. See for example "Everyone Is Doing It, But Nobody Knows." for some common situations that cannot be explained using that perspective. I will also re-use one of the graphics from that article below.

Safety stock needs to be determined for individual items at individual locations. There is no sense setting safety levels for groups of items or across multiple locations. If you do so, it is guaranteed some items within the group or some locations within a group will suffer loss of service whilst others in the same group will have excessive inventory consuming cash and space for no benefit. Also, since safety stock is not real inventory it should not be aggregated into groups for summary reporting. The resulting values will often be deceiving, leading to bad planning decisions. Note that "items" can be any tangible object or fluid, it could be raw materials, packing materials, labels, WIP, finished goods, gray stock. Similarly "locations" can be physical locations such as warehouses, factories, distribution centers, but also spaces within those, such as bins, tanks, shelves, pallets in a corner, a cupboard, or logical groups of these, such as groups of co-located shelves or tanks.

Between the time of planning and the time of execution a lot of unknowns will occur. At time of execution safety stock has little relevance, since the amount to execute is not related to the unknowns that safety stock buffers for. For execution the distinction of a buffer or decoupling point becomes important. For planning it is only relevant insofar that it may change the formula to calculate the proper level.

The objective of safety stocks is to protect against stock outs. In other words it is NOT meant to protect against over-stocks, product expiration or waste. Our plans can aim too high or too low. Safety stocks are there to protect the supply chain in those cases where the aim is too low and, not where it is too high. Also, there is an exponential cost impact of trying to prevent the last few stock outs. It becomes prohibitively expensive to prevent all stock outs, so we can only aim to minimize them and their impact. In practice, a safety stock level is chosen that balances the risk of stock out with the cost to do so. Some stock outs are thus still expected to occur.

Safety stock does not aim to prevent stock outs from all possible causes, only those caused by the uncertainties in demand and supply under typical circumstances. Demand uncertainty contains two parts: the precision of an unbiased demand forecast and the natural variability of the demand. The term "forecast" is used freely here, encompassing any form of expected future demand. For example, some approaches use last year's or last month's demand, or average demand rates, which are just other kinds of forecast. Notice that forecast accuracy, or its inverse forecast error, are NOT part of this definition. These are often used as input into safety stock formulas, but they are bad proxies to the real factors of importance and their use leads to big inaccuracies in safety stock levels (see for example Safety Stock and the Hazard of the Fitted Forecast Error). Supply uncertainty should include delivery delays, incomplete deliveries, and cancellations by the supplier. In practice, this is usually reduced to just covering for lead time delays, which is fine if the assumption holds that all undelivered quantities will be delivered in a later shipment, in time to cover any backlogs caused by the shortfall.

Why was Safety Stock Conceived?

Safety stock is a by-product of oversimplified forecasting and planning approaches. Whilst supply chains are rife with uncertainty, forecasts and plans are created using exact numbers. Instead of saying future demand within the next replenishment cycle will have a distribution of possible values, we simply state what the average expected quantity is. Instead of saying raw material supply is expected according to some distribution of days after ordering we provide some generous exact number of days of expected lead time. And the plans assume the full requested quantity will be delivered at such time. Internally, the same simplifications are made: manufacturing, inventory, and distribution plans are similarly expressed in exact numbers, typically historical averages.

If we plan using average demand then it should be obvious that roughly half the time actual demand will be lower than that average and roughly half the time it will be higher (simplified for clarity, more accurate behavior will be explained in subsequent articles). If for one item in one location across a certain replenishment cycle demand happens to be lower than the average there will be excess inventory. And unless it expires before it can be consumed that is not a big problem; merely taking up more space and locking down cash. But if for an item in a location across a replenishment cycle demand happens to be greater than the average, there will be a real problem: we run out of inventory and cannot satisfy the demand of the customer or downstream processes. Similarly, if supply arrives sooner than we need it we will have inflated stock, and if it arrives later we will stock out.

There are two simple approaches one could take to reduce stock outs:

1. Overestimate the demand quantities and the supply lead times
2. Introduce safety stock

Some downsides of option 1 are that it introduces dramatic amounts of bias and it becomes impossible to measure how accurate demand forecasts are and how reliable suppliers are. Another problem is determining how large the overestimation needs to be. Both potential demand quantities and supply lead times are unbounded. It is impossible to pick a "maximum" value, since it does not exist. So an arbitrarily high value needs to be used. The simple alternative is to introduce safety stocks, allowing demand and supply to remain unbiased and separately measure the accuracy of demand forecasts, reliability of suppliers, and effectiveness of the safety stock. Whilst most companies have realized this and use safety stocks to buffer for the demand uncertainty, surprisingly, for supply lead times companies pervasively use vastly bloated values. This is a costly mistake. In the next series of articles introducing a new safety stock formula I will explain both why this mistake is made, and suggest some ways it can be avoided. For now, it is assumed your safety stocks protect against uncertainties of both demand and supply.

Why do we Need Safety Stock?

IF demand and supply were NOT uncertain, the most cost-effective way to drive inventory would be to order resupply exactly the known lead time before stock runs out. This point, a lead time prior to planned delivery, is known as the reorder point. At the exact moment you run out of inventory a new delivery arrives and inventory is plentiful again. This process repeats indefinitely, and a projection of inventory across time will exhibit a typical saw-tooth pattern, where the size of each order (and thus the height of the saw tooth) will be some economic reorder quantity:

If the pattern above would be what occurred you would have a 100% customer service level - since there is always stock available when needed - and at the minimum inventory cost - since you never carry more than the economic order quantity. In this situation the blue inventory line is commonly referred to as cycle stock: the stock you carry just to cover for expected demand during a single replenishment cycle.

But if demand were greater than expected the inventory projection on the left below occurs, and if supply is delayed the inventory projection on the right below occurs:

On the left, due to higher demand than expected the slope of the inventory decrease is steeper after the reorder point is reached. On the right, due to longer lead time than expected the slope continues longer before the resupply arrives. Both cases lead to stock outs and in the above examples roughly 80% customer service level each. Notice that I did not project what the stock level will look like after the stock out, because that will be very much dependent on the percentage of the backlog that will still be accepted by the customers. This will depend on many factors, such as industry, the product, and how long the customer will need to wait.

The remedy that safety stocks provide are to target a higher level of remaining inventory at the expected time of delivery of supply:

Now if demand is higher than expected or resupply takes longer than expected in most cases you would still have enough inventory:

With the same higher demand than expected on the left, or longer than expected lead time on the right, stock outs are prevented. Service levels in this cycle are 100% and the customers get all the product they ordered.

At the cost of carrying extra inventory, revenue and customer satisfaction are preserved. This is the value of safety stock

Conclusion

In this article I have hopefully demonstrated why any supply chain company would need safety stock - or inventory buffers in general, by any other name. In practice, determining how much safety stock to target is the difficult part. There is a balance where incremental costs of inventory and risks of waste or discounting start to exceed the benefits such as increased revenue and customer satisfaction. Since not all costs and benefits can easily be translated to hard currency, finding the right balance is part art or strategy, rather than pure science. Each company will place that balance differently.

In the upcoming series of articles I hope to demonstrate both why the common formulas for safety stock are wrong and how to fix them.

Find all my articles by category here. Also listing outstanding articles by other authors.