Holding cost is what it costs, over a year, to keep one unit in stock. It is expressed as a percentage of the item’s value and aggregates very different natures: tied-up capital, obsolescence, space, handling and insurance. It is the quietest parameter in the domain and one of the most influential, since it serves as the denominator of almost every sizing trade-off.
Why it matters
Almost every decision in this domain rests on comparing what it costs to hold stock with what it costs not to have it. The economic order quantity divides by this cost, discount profitability leans on it, the trade-off between pooling and proximity assumes it known, the return threshold of an assortment derives from it. It is therefore the common denominator of the whole argument.
Yet it is also the figure most often inherited without examination. A rate of twenty or twenty-five percent lingers in configurations, taken from a textbook or an old go-live, without anyone knowing where it came from or what it covers. The second panel of the signature shows the consequence: that number silently fixes every lot size in the portfolio, hence a major share of cycle stock.
The breakdown above explains why a single rate is almost always wrong. Capital is only one component among several, and rarely the heaviest. Obsolescence weighs enormously on a short-cycle product and almost nothing on a metal spare part. Space counts for a great deal on bulky items and for nothing on electronic components. These natures do not move together.
This is what makes a single rate applied across a whole catalogue economically indefensible. Two references of equal value, one perishable and one inert, do not cost the same to hold, and therefore have no reason to be ordered in comparable lots.
The good news is that precision is not required. Since the economic lot varies with the square root of the rate, an error of a third on the rate produces only about fifteen percent difference on lot size, and total cost is flat around the optimum. The point is not to compute a rate to the decimal, but to avoid being wrong by an order of magnitude.
The bad news is that order-of-magnitude errors are common, and almost always in the same direction. Inherited rates capture capital and space, which are invoiced, and omit obsolescence and markdowns, which are not. On fast-renewing catalogues, that omission halves the true rate, leading to oversized lots that then feed exactly the dead stock everyone will later deplore.
The expert lesson is to differentiate by broad family rather than refine an average rate. Three or four rates, calibrated on the catalogue’s risk profiles, deliver far more value than one carefully computed rate. And running the exercise has its own virtue: it forces you to quantify obsolescence, which few organisations have ever done.
The mechanism
The financial component comes first and computes easily: it is the company’s cost of capital applied to the value tied up. It is the only one finance knows in advance, and often the only one retained, which explains the general underestimation.
Obsolescence and markdowns form the second component, and the most variable. It corresponds to the value an item loses while it waits, through technical dating, seasonal change, expiry or plain markdown. It is measured by observing the catalogue’s own history, not by postulating it.
Then come the physical components: space occupied, with its rent and charges, then handling, internal movements and counting. These relate to volume rather than value, which explains why a bulky, inexpensive item can show a holding rate far above what its value would suggest.
Insurance and taxes close the list, usually for a modest share. Together they give an annual rate which, applied to unit value, yields the holding cost used in every lot calculation. One caution applies to fixed costs: a warehouse already paid for becomes a holding cost only if the space released can genuinely be monetised, otherwise you are adding a charge that does not vary.
The traps
It is the easiest component to obtain, and it rarely represents half the total. A rate built on capital alone heavily understates the real cost on fast-renewing catalogues, and leads to oversized lots that later feed dead stock.
An inert metal part and a fashion item do not cost the same to hold, even at equal value. A single average rate oversizes the lots of risky references and undersizes those of stable ones.
Loading the full cost of an already-paid, non-monetisable warehouse inflates the rate artificially and pushes toward lots that are too small, hence needless ordering costs. Only genuinely avoidable charges belong in the calculation.
The rollout
Get the applicable rate from finance and set it as the base. It is the one component nobody disputes, and it gives the floor.
Measure the value loss actually observed, markdowns and destructions included, by broad family. It is the heaviest component and the one most often absent.
Include space, handling and insurance, retaining only charges that would genuinely vary if stock fell.
Establish three or four rates rather than one, calibrated on perishability and bulk profiles. This differentiation returns more than precision in the calculation.
Recompute lot sizes with the new rates and measure the gap with the quantities in force. That gap is the immediate value of the exercise.
Neighboring concepts
From knowledge to action
An inherited percentage silently fixes the cycle stock of the whole portfolio. Our Inventory & distribution file breaks it down, quantifies obsolescence and differentiates it by risk family.