Total operating costs could be split into variable and fixed costs. By definition, variable costs depend on the production volume and constitute more or less fixed proportion of the revenue. These are supplies, labor and utilities needed for production of product or performance of service. Fixed costs are non-volume related.
The variable-fixed costs split looks quite obvious in theory, but it is not that straightforward in practice. Let's take a closer look at variable costs (we will be dealing with fixed costs in the next chapter).
First, there is always a lag between a change in production volume and a corresponding change in variable costs. And such a lag is different for different components of variable costs. It takes time to adapt production facilities, supply chain and labor for a given production volume.
Take materials and parts, for example. Yes, you cannot use more or less of them for a particular production item, but as the level of productions declines or accelerates, the volume of inventories of materials and parts changes accordingly, even if only for a short period of time - before the company adjusts the rate of their supply to the rate of consumption. A change in inventories means change in the costs of keeping these inventories. An even bigger change in costs could be caused by changes in the price of materials and parts - they are usually quite sensitive to changes in supply and demand and the company is affecting the demand side of the balance.
The most inflexible part of the variable costs, not surprisingly, is labor. It is not that easy and fast to fire workers or to hire and train them. And any change in the number of the workforce has its "unit cost" implications, I.e. it affects the amount of variable costs needed for production of one unit of product, thus increasing the risk of under- or overestimating them in our projections.
Second, as already mentioned above, variable costs are more variable than what is implied by their definition: their proportion in total unit costs changes with any significant change in the production volume. It is the role of a good analyst to estimate how variable unit costs will change in the future - and forecasting future commodity prices is only part of the puzzle. For example, in a stable macroeconomic environment what will happen to unit variable costs if production doubles? The task of determining this is almost insurmountable to an outsider.
With all this in mind, the best alternative might be, surprisingly, an easy "way out" solution to assume that future variable costs will be the same (or increase at the rate of inflation) for a unit of production as they are now. This assumption, if applied universally to all companies we are valuing, at least will cause a similar error for each one of them and will not affect much relative valuation of the companies, which, at the end, in many cases is more important than "absolute" valuation of a company. By the way, there is nothing absolute in this world - everything is measured in relation to something else.