Skip to content

Lesson 30: Inventory Control: Concepts and Cost Structure

30.1 Learning Outcomes

By the end of this hour a student will be able to:

  1. State the functions inventory performs and why holding stock can be rational.
  2. Identify and classify the four inventory costs, and know which are relevant to a decision.
  3. Define the decision variables and parameters of an inventory model and read its notation.
  4. Classify inventory systems (deterministic/probabilistic, continuous/periodic review).
  5. Perform an ABC (Pareto) analysis and set control policy by class.

30.2 Why Hold Inventory at All?

Inventory is idle capital. A firm holding \(₹5\) crore of stock has \(₹5\) crore not earning a return, occupying space, at risk of obsolescence. Yet every firm holds it. The reasons are specific and each one suggests a different remedy.

Function Reason for holding Reduce it by
Cycle stock Ordering costs make frequent small orders expensive Cutting the ordering cost (EDI, blanket contracts)
Safety stock Demand and lead time are uncertain Better forecasts, shorter lead times
Pipeline stock Goods in transit are owned but unavailable Faster or nearer supply
Anticipation stock Known seasonal peaks, planned shutdowns Level production, flexible capacity
Decoupling stock Prevents one stage's stoppage halting the next Improved reliability, smaller batches
Speculative stock Expected price rise or quantity discount — (a deliberate financial bet)

The central trade-off. Ordering (or setting up) frequently keeps average stock low but incurs the ordering cost often. Ordering rarely does the opposite. The optimal order quantity balances these two costs, and every model in Lessons 31–35 is a variation on that single idea.

30.3 The Four Costs

Ordering / setup cost C_o

Cost per order placed, independent of the size of the order: raising the requisition, inspecting the delivery, processing the invoice; in manufacturing, the setup cost of changing a machine over.

Only the variable part counts. The purchasing manager's salary is paid whether or not one more order is placed, so it is not part of \(C_o\). Including fixed overheads in \(C_o\) is the commonest costing error in this topic and it systematically inflates the recommended order quantity.

Holding / carrying cost C_h

Cost of holding one unit for one unit of time. Components:

Component Typical share of item value per year
Capital cost (opportunity cost of funds) 10–15 %
Storage and handling 2–5 %
Insurance and taxes 1–3 %
Obsolescence, deterioration, pilferage 4–10 %
Total \(i\) typically 20–30 % per year

Usually expressed as \(C_h = i\,C\) where \(C\) is the unit purchase price and \(i\) the carrying rate. This link matters in Lesson 35: a discount that lowers \(C\) also lowers \(C_h\).

Shortage / stock-out cost C_s

Cost of being one unit short for one unit of time (backorder case), or per unit short (lost-sales case). Includes expediting, penalties, idle production, and lost goodwill.

The hardest number in the model. Lost goodwill has no invoice. In practice one either estimates \(C_s\) crudely and reports the answer's sensitivity, or inverts the problem: given a chosen service level, infer the \(C_s\) that would justify it. The second approach is usually more persuasive.

Purchase / production cost

\(C\) per unit. If \(C\) is constant it does not affect the optimal \(Q\), because total purchase cost \(DC\) is the same however you split the orders. It becomes decision-relevant only when the price depends on quantity — which is exactly the quantity-discount problem of Lesson 35.

30.4 Notation and Classification

Symbol Meaning
\(D\) demand rate (units per unit time)
\(Q\) order quantity — the decision variable
\(C_o\) ordering/setup cost per order
\(C_h\) holding cost per unit per unit time
\(C_s\) shortage cost per unit per unit time
\(C\) purchase price per unit
\(t\) cycle length \(=Q/D\)
\(n\) number of orders per period \(=D/Q\)
\(L\) lead time
\(ROP\) reorder point

Classification of models

By demand:

  • Deterministic — demand known with certainty. Static (constant rate, Lessons 31–35) or dynamic (known but varying).
  • Probabilistic — demand is a random variable; safety stock and service levels enter.

By review:

  • Continuous review \((Q, ROP)\) — stock is monitored constantly; order a fixed \(Q\) whenever the level falls to \(ROP\). Order quantity fixed, order timing variable.
  • Periodic review \((T, S)\) — check every \(T\) periods and top up to a level \(S\). Order timing fixed, order quantity variable. Cheaper to administer, needs more safety stock.

The reorder point

With deterministic demand and lead time \(L\),

\[\boxed{\ ROP = D\times L\ }\tag{3.1}\]

and if \(L>t\) the formula must be applied modulo the cycle: with \(L\) spanning \(k\) whole cycles, \(ROP = D(L-kt)\), because \(k\) orders are already in the pipeline. Missing this is the classic slip.

30.5 ABC Analysis

Not every item deserves the same attention. Pareto's principle applies with unusual force to inventories: a small fraction of items accounts for most of the money.

ABC classification. Rank items by annual usage value (\(=\) annual demand \(\times\) unit price), then:

Class Share of items Share of value Control policy
A ~10–20 % ~70–80 % Tight: continuous review, accurate forecasts, frequent small orders, close supplier relations
B ~30 % ~15–20 % Moderate: periodic review, routine reordering
C ~50–60 % ~5–10 % Loose: large orders, two-bin system, minimal record-keeping

Annual usage value, not unit price, is the criterion. A cheap item consumed in huge volume can be an A item; an expensive item bought once a year is often a C item. Ranking by price alone is the standard mistake.

30.6 Solved Problems

Problem 1 — ABC classification

Classify these ten items.

Item Annual units Unit price (₹)
A-1 7000 5.00
A-2 1600 15.00
A-3 2400 8.00
A-4 100 60.00
A-5 6000 1.50
A-6 3800 0.40
A-7 300 90.00
A-8 1300 4.00
A-9 500 22.00
A-10 900 7.00

Solution.

Step 1 — annual usage value; Step 2 — sort descending; Step 3 — cumulate.

Rank Item Units Price Usage value % of total Cumulative % Class
1 A-1 7000 5.00 35 000 24.27 24.27 A
2 A-7 300 90.00 27 000 18.72 42.99 A
3 A-2 1600 15.00 24 000 16.64 59.63 A
4 A-3 2400 8.00 19 200 13.31 72.94 A
5 A-9 500 22.00 11 000 7.63 80.57 B
6 A-5 6000 1.50 9 000 6.24 86.81 B
7 A-10 900 7.00 6 300 4.37 91.18 B
8 A-4 100 60.00 6 000 4.16 95.34 C
9 A-8 1300 4.00 5 200 3.61 98.95 C
10 A-6 3800 0.40 1 520 1.05 100.00 C
Total ₹144 220
\[\boxed{\text{A: A-1, A-7, A-2, A-3 — 40\% of items, 72.9\% of value}}\]
\[\text{B: A-9, A-5, A-10 — 30\% of items, 18.2\% of value}\]
\[\text{C: A-4, A-8, A-6 — 30\% of items, 8.8\% of value}\]

Interpretation — the two instructive rows.

  • A-1 is the cheapest-but-one item at ₹5 and yet ranks first by value, because \(7000\) units are consumed. It deserves the tightest control in the store.
  • A-4 costs ₹60 — the second-highest price — yet lands in class C, because only \(100\) are used. Managing it intensively would waste effort.

Ranking by unit price would have put A-7 and A-4 at the top and buried A-1 in the middle: exactly backwards for two of the ten items.

Problem 2 — Which costs are decision-relevant?

A firm buys a component for \(₹200\). Its accountant lists: (a) purchasing department salaries \(₹40{,}000\)/yr; (b) \(₹500\) per order for inspection and paperwork; (c) warehouse rent \(₹1\) lakh/yr for a building that is half empty; (d) capital cost at \(12\,\%\) per year; (e) insurance \(2\,\%\) of stock value; (f) obsolescence \(6\,\%\) per year; (g) \(₹300\) per unit per year of production stoppage when short. Which enter the EOQ model, and with what values?

Solution.

Item In the model? Reason
(a) salaries ₹40,000/yr No Fixed; unchanged by ordering more or less often
(b) ₹500 per order Yes — \(C_o=₹500\) Varies directly with the number of orders
(c) rent for a half-empty warehouse No A sunk cost; the space is already paid for
(d) capital 12 % Yes Part of \(i\)
(e) insurance 2 % Yes Part of \(i\)
(f) obsolescence 6 % Yes Part of \(i\)
(g) ₹300/unit/yr short Yes — \(C_s=₹300\) The shortage cost
\[i = 12+2+6 = 20\,\%\text{ per year} \Rightarrow C_h = 0.20\times200 = \mathbf{₹40\text{ per unit per year}}\]
\[\boxed{C_o=₹500,\quad C_h=₹40/\text{unit/yr},\quad C_s=₹300/\text{unit/yr}}\]

Interpretation. The rent case deserves care. If the warehouse is half empty, extra stock costs nothing extra in space, so the rent is irrelevant. If the warehouse were full, so that more stock would require renting more space, the marginal rent would belong in \(C_h\). The test is always "does this cost change if I change \(Q\)?" — never "is this cost associated with inventory?"

Problem 3 — The reorder point, including the lead-time trap

Demand is \(2400\) units per year over \(300\) working days; the order quantity is \(400\) units. Find \(ROP\) when the lead time is (a) \(5\) days, (b) \(60\) days.

Solution. Daily demand \(d = 2400/300 = 8\) units. Cycle length \(t = Q/d = 400/8 = 50\) days.

(a) \(L = 5\) days. Since \(L<t\):

\[ROP = d\,L = 8(5)=\mathbf{40\text{ units}}\]

Reorder when stock falls to \(40\).

(b) \(L = 60\) days. Now \(L > t = 50\), so one order is always already in transit. Write \(L = kt + L'\) with \(k=1\) and \(L' = 60-50 = 10\) days:

\[ROP = d\,L' = 8(10) = \mathbf{80\text{ units}}\]

and the firm operates with one outstanding order at all times.

Interpretation. Applying \(ROP=dL=8(60)=480\) blindly would demand a reorder point of \(480\) when the maximum stock is only \(400\) — an impossible instruction. Whenever \(ROP\) exceeds \(Q\), check the lead time against the cycle length. The physically meaningful statement is always about the inventory position (on hand plus on order), not stock on hand.

Problem 4 — Continuous versus periodic review

A distributor stocks \(4000\) SKUs. Class A items (200 SKUs) are worth \(₹8\) crore a year; class C (2400 SKUs) are worth \(₹40\) lakh. Continuous review costs \(₹2000\) per SKU per year in systems and staff time; periodic review costs \(₹200\). Which policy for which class?

Solution.

Class A. Cost of continuous review: \(200\times₹2000 = ₹4\) lakh per year, i.e. \(0.5\,\%\) of the \(₹8\) crore of value under control. Continuous review typically cuts stock by \(10\)–\(15\,\%\); at a \(25\,\%\) carrying rate on an average stock of, say, \(₹1.5\) crore, a \(10\,\%\) reduction saves

\[0.10\times ₹1.5\text{ crore}\times0.25 = ₹3.75\text{ lakh per year}.\]

Comparable to the cost — and that is before counting the reduced risk of stocking out on the items that matter most. Use continuous review for A.

Class C. Continuous review would cost \(2400\times₹2000 = ₹48\) lakh per year to control \(₹40\) lakh of annual usage. The control system would cost more than the inventory. Periodic review at \(2400\times₹200 = ₹4.8\) lakh is still \(12\,\%\) of value — even that is questionable.

\[\boxed{\text{A: continuous review. C: two-bin system, order large quantities, minimal records.}}\]

Interpretation. The two-bin system is the classical C-item answer: keep the stock in two containers, and when the first empties, order a replenishment while consuming the second. It requires no records, no forecasts and no computers, and for a ₹15 item that is exactly right. Control effort should be proportional to value at stake — the entire point of ABC.

Problem 5 — Reading an inventory situation

A hospital pharmacy stocks: (i) saline drips, steady use, cheap, ₹20 each, 50 000 a year; (ii) a cardiac drug, ₹4000 a vial, unpredictable demand, 100 a year, fatal if unavailable; (iii) seasonal flu vaccine, ₹300 a dose, all demand in three months. Classify each and state a policy.

Solution.

Item Usage value ABC Demand type Policy
(i) saline ₹10 lakh A deterministic, steady Classic EOQ with continuous review; frequent deliveries; the textbook case
(ii) cardiac drug ₹4 lakh A probabilistic, lumpy EOQ is inappropriate. Set stock by service level, not by cost minimisation
(iii) vaccine ₹?? depends deterministic but seasonal Neither EOQ nor safety stock: this is a single-period (newsvendor) problem

Interpretation — the important part of this problem.

  • Item (i) is the only one the EOQ model of Lesson 31 actually fits.
  • Item (ii) has an effectively infinite \(C_s\) — no manager will write down the rupee value of a patient's death. The right formulation is: hold enough to meet demand with probability \(99.9\,\%\), and let the model report what that costs. Cost minimisation is the wrong objective when one of the costs is not commensurable with money.
  • Item (iii) is perishable with a single selling season. Order too much and it is destroyed; order too little and demand is lost. The optimum balances overage against underage cost — a different model entirely.

The lesson: choosing the model is a bigger decision than solving it. Lessons 31–35 develop the deterministic family thoroughly, but part of mastering them is knowing where they stop applying.

30.7 Practice Problems

  1. Five items have annual usages ₹90 000, ₹8 000, ₹52 000, ₹3 000, ₹1 500. Classify them. (Answer: total ₹154 500. A: ₹90 000 (58.25 %); B: ₹52 000 (cumulative 91.91 %); C: the remaining three, 8.09 %)
  2. An item costs ₹800; capital \(14\,\%\), storage \(3\,\%\), insurance \(1\,\%\), obsolescence \(7\,\%\). Find \(C_h\). (Answer: \(i=25\,\%\), \(C_h=₹200\) per unit per year)
  3. \(D=7200\)/yr over \(360\) days, \(Q=600\), lead time \(25\) days. Find the cycle length and \(ROP\). (Answer: \(t=30\) days, \(ROP=500\) units)
  4. Repeat (3) with a lead time of \(75\) days. (Answer: \(L=2t+15\), so two orders are outstanding and \(ROP = 20(15)=300\) units)
  5. A firm includes its ₹18 lakh purchasing-department salary bill in \(C_o\) by dividing it across \(600\) orders per year (₹3000 per order) and adds this to a genuine ₹400 variable cost. Show that this increases the computed EOQ, and say in which direction the firm's real costs then move.

30.8 Summary

  • Inventory exists to serve identifiable functions — cycle, safety, pipeline, anticipation, decoupling, speculation. Each is reduced by a different remedy.
  • Four costs: ordering \(C_o\), holding \(C_h=iC\) (typically \(i=20\)–\(30\,\%\)/yr), shortage \(C_s\), purchase \(C\). A constant \(C\) does not affect the optimal \(Q\).
  • Include a cost only if it changes when \(Q\) changes. Fixed salaries and sunk rent do not belong in \(C_o\) or \(C_h\); including them inflates the EOQ.
  • \(C_s\) is the least reliable input. Either report sensitivity or invert the model to infer the \(C_s\) implied by a chosen service level.
  • Classify models by demand (deterministic/probabilistic) and review (continuous \((Q,ROP)\) or periodic \((T,S)\)).
  • \(ROP = D\times L\), but when \(L>t\) subtract whole cycles: orders are already in the pipeline.
  • ABC analysis ranks items by annual usage value, not unit price. Concentrate control effort on the A items; use a two-bin system for C items.
  • Choosing the right model matters more than solving it. EOQ fits steady deterministic demand only.

Next lesson: L31 — The EOQ model with no shortages: derivation, total cost and sensitivity.

Practise now: unit3-queue-inventory-lab.html → ABC Analyser: paste an item list and get the sorted Pareto table and chart, with class boundaries you can drag.