How to size the safety stock?

A couple of years ago I wrote a post about the five functions of inventories, which is the logistic function that I most identify with. And among the 5 five functions of stocks, the one that fascinates me the most is the safety stock. This is because, while the other four are the result of concrete and deterministic variables and decisions, the safety stock is based on probabilities and uncertainties, being the most complex to dimension.

The dimensioning of the safety stock is a result of the level of uncertainty of the operation and the level of service to be delivered. The more uncertainty and the greater the desired availability, the greater the safety stock should be. Intuitively, this sentence makes a lot of sense, the difficult thing is to measure the uncertainty, decide the optimal service level and, consequently, determine the ideal level of safety stock.

Operational uncertainties are basically derived from demand and supply, as these are the result of actions by external agents to the company: customers and suppliers, respectively. To dimension the necessary safety stock, it is necessary to understand very well the behavior of these uncertainties. It is necessary to know how demand behaves on average, how is its variation, how is its seasonality, what is the accuracy level of the demand forecast, if there is any kind of bias in the forecast and what is the variation of the error. And a one-off mapping is not enough, it needs to be done constantly, as the behavior of demand changes according to economic cycles, competition strategies, launches of substitute products or services, etc... It is also necessary to know the behavior of supply through of the lead times expected delays, frequencies and magnitudes of delays, and variations in quantity or quality received. And all these behaviors change when we change suppliers or transport modes, being important variables for the hiring decision.

The ideal service level will depend on the company's positioning in the market, its excess inventory costs and its shortage costs. The more expensive it is for the company to lose sales due to stockouts, the higher the level of service it will want to deliver to its customer. Companies with placements premium Market share and large margins tend to deliver higher service levels because losing sales to them results in a large cost of shortages. The cost of the lack can be the loss of the product's margin, the loss of the company's image, the loss of market share, a contractual fine for unavailability of the product, among others. The cost of excess represents the cost of capital placed in inventory, the cost of losing products due to perishability or obsolescence, and the cost of moving and storing the product. When these costs are very relevant or if the company has a small margin and seeks to minimize its operating costs, the ideal service level tends to be lower, as sustaining high availability through safety stock can be expensive. Figure 1 illustrates the variables that influence the optimal level of safety stock.

Figure 1: Variables that impact the sizing of the safety stock

Figure 1 : Variables that impact the sizing of the safety stock

Source: ILOS

It is possible to arrive at the ideal level of safety stock through simulation or through deterministic formulas. The simulation replicates the behavior of uncertainties in a controlled environment. In this case, the simulation input variables would be the behaviors of uncertainties (actual demand, forecast error, lead time, delays, among others) and shortage costs (product margin, image, market share or contractual fines) and excess costs (capital cost, storage cost or loss cost). This allows you to experiment with different levels of safety stock in a context that mimics a real system and look at the results in terms of cost, stock level and service level to decide how much to keep in stock.

Deterministic models, on the other hand, start from the premise that the uncertainty variables have a known variation that follows some kind of distribution, usually the normal distribution. Thus, based on the standard deviation of demand or the forecast error and the lead time, it is possible to dimension the safety stock for a certain desired non-rupture percentage. Deterministic models are easy to apply, where the parameters can be found from an equation, however, they may not be applicable in highly complex systems, as they require simplifications of the real system.

The market already offers a series of tools that measure the ideal safety stock levels for each product or supplier based on the historical behavior of uncertainties and the company's current costs. However, it is important to understand the variables that impact this sizing and the effect of this impact, as any change in demand variation, supplier response time, forecast error or resupply reliability can have significant impacts on availability or necessary levels of safety stock, being a function of the company with great potential for opportunities for better results.

If you are interested in this topic, be sure to check out the ILOS courses Inventory Management Update e Inventory Management Online Course. These courses bring simple methods and tools to size inventories.

Sources:

'https://ilos.com.br/web/gerenciando-incertezas-no-planejamento-logistico-o-papel-do-estoque-de-seguranca/

'https://ilos.com.br/web/as-cinco-funcoes-dos-estoques/