Inventory Policies and Lot-Sizing Decisions

Introduction

Master production scheduling (MPS) translates aggregate production plans into time-phased decisions for specific products. A central MPS decision is lot sizing: determining how much to produce or purchase when an order or replenishment is scheduled. The decision is not merely a technical exercise. It affects capacity loading, machine utilisation, inventory investment, customer service, obsolescence risk, supplier performance and cash flow.

A small lot may reduce inventory and obsolescence but create frequent setups, purchasing transactions and schedule instability. A large lot may reduce setup and ordering frequency but increase working capital, storage requirements and exposure to demand uncertainty. Effective production management therefore seeks a policy that is economically and operationally appropriate rather than universally minimising one cost.

This chapter evaluates lot-for-lot, fixed-order quantity, economic order quantity, periodic ordering and dynamic lot-sizing policies. It also considers shortages, service levels, working capital, product obsolescence and the practical role of enterprise resource planning (ERP), material requirements planning (MRP) and advanced planning systems.

Learning Objectives

By the end of this chapter, the learner should be able to:

1. Explain the purpose and logic of inventory lot-sizing policies.
2. Distinguish independent-demand and dependent-demand inventory decisions.
3. Calculate and interpret economic order quantity and related reorder parameters.
4. Compare lot-for-lot, fixed-order quantity, periodic ordering and dynamic lot-sizing methods.
5. Evaluate setup, holding, shortage, obsolescence and working-capital consequences.
6. Select and govern an appropriate policy in an MPS environment.

1. Inventory Policy in the MPS Context

Inventory policy specifies how a firm responds to demand over time. It normally defines the replenishment quantity, replenishment timing, safety-stock logic, review frequency and treatment of exceptions. In MPS, these decisions interact with the production calendar, lead times, available capacity and the bill of materials.

Demand may be:

• Independent demand: demand for finished goods or service parts, usually forecast or observed from the market.
• Dependent demand: demand for components and raw materials calculated from the production plan, bill of materials and planned order releases.

For independent demand, statistical inventory policies such as reorder-point and periodic-review systems are common. For dependent demand, MRP often calculates time-phased requirements and uses lot-sizing rules to consolidate them. The same item can require different policies in different contexts: a finished spare part may use a reorder point, whereas a component made exclusively for a scheduled product may use lot-for-lot.

Important policy parameters include:

2. The Economics of Lot Size

The principal cost categories are ordering or setup cost, cycle-stock holding cost, shortage cost, obsolescence cost and the opportunity cost of capital.

2.1 Ordering and setup cost

A fixed ordering cost is incurred whenever a purchase order is placed. In manufacturing, the analogous cost is a setup cost: scheduling, tooling, cleaning, first-piece approval and lost production time. If annual demand is divided among many small orders, the number of setups rises.

2.2 Holding cost

Holding cost includes warehouse space, insurance, handling, deterioration, taxes and the financial cost of inventory. A common annual holding-cost representation is:

where H is annual holding cost per unit, i is the annual holding-rate fraction and C is unit value. For example, if C = $80 per unit and i = 0.25 per year, H = $20 per unit per year. The formula assumes that the holding rate appropriately captures relevant proportional costs and that unit value is stable.

With constant demand and instantaneous replenishment, average cycle inventory is approximately Q ÷ 2, where Q is order quantity. Thus annual cycle-stock holding cost is:

This excludes safety stock, pipeline inventory and any seasonal or anticipation inventory.

2.3 Shortage and obsolescence cost

Shortage cost may include lost contribution margin, expediting, contractual penalties, production stoppage and loss of customer goodwill. It is often difficult to estimate and is not necessarily linear. A one-hour shortage of a critical semiconductor may be much more costly than a one-day shortage of a routine consumable.

Obsolescence is especially important for electronics, fashion, pharmaceuticals, food and customised items. A nominally low holding-cost policy may be inferior if large lots leave material unusable after engineering change or market decline.

2.4 Working capital

Inventory ties up cash before revenue is received. A useful approximation is:

The cash consequence depends on payment terms, production conversion, demand speed and recoverability. Two policies with equal accounting cost can have materially different effects on liquidity and return on invested capital.

3. Economic Order Quantity

The classical economic order quantity (EOQ) balances ordering cost and holding cost. Under the basic assumptions of constant known demand, constant lead time, instantaneous replenishment, no shortages, no quantity discounts and a constant unit price, annual relevant cost is:

where:

• TC(Q) = annual ordering plus cycle-stock holding cost;
• D = annual demand in units per year;
• Q = order quantity in units;
• S = ordering or setup cost per order;
• H = annual holding cost per unit.

The first term is annual ordering cost and the second is annual cycle-stock holding cost. Differentiating with respect to Q and setting the derivative to zero gives:

The result is a minimum because total relevant cost decreases before Q and increases after Q. The formula is dimensionally consistent: DS has units of currency-units per year, division by H (currency per unit per year) gives units², and the square root gives units.

Worked example: basic EOQ

Annual demand for a standard component is 24,000 units. Ordering cost is $75 per order. Unit cost is $40 and the annual holding rate is 20%.

First calculate holding cost:

H = 0.20 × $40 = $8 per unit per year

Then:

Q* = √((2 × 24,000 × 75) ÷ 8) = √450,000 ≈ 671 units

The approximate number of orders per year is:

24,000 ÷ 671 ≈ 35.8 orders

Average cycle stock is approximately 335.5 units, excluding safety stock. EOQ is a benchmark, not an automatic release quantity. Packaging multiples, supplier minimums, capacity and shelf life may require adjustment.

Reorder point and safety stock

For a continuous-review system with demand during lead time and uncertainty, the reorder point is commonly represented as:

where μL is expected demand during lead time and SS is safety stock. If average demand per period is d and lead time is L periods, μL = dL. Under a normal-demand approximation with constant lead time:

where z is the standard-normal service factor and σL is the standard deviation of demand during lead time. This formula assumes that lead-time demand is approximately normal and that the selected service measure corresponds to the chosen z value. If both demand and lead time vary independently, a commonly used variance expression is:

where σd is standard deviation of demand per period and σL is standard deviation of lead time in periods. Correlation, non-normal demand and intermittent demand require more appropriate models.

4. Core Lot-Sizing Policies

4.1 Lot-for-lot

Lot-for-lot (L4L) sets the planned order quantity equal to the net requirement in each period. If requirements are 400, 250 and 600 units in three periods, L4L plans those same quantities in their respective release periods, adjusted for lead time.

L4L is attractive when:

• inventory carrying cost or obsolescence risk is high;
• setup cost is low or setups are rapid;
• demand is irregular or highly time-phased;
• the product is customised;
• MRP visibility is strong.

Its disadvantages are frequent setups, procurement transactions and possible loss of scale economies. L4L may also create an unstable schedule when requirements change frequently.

4.2 Fixed-order quantity

A fixed-order-quantity policy releases a predetermined quantity, often called Q, whenever inventory reaches a trigger or whenever MRP generates a requirement. The quantity may be based on EOQ, a supplier pack, a truckload, a production batch or management convention.

The method is operationally simple and useful where processes have stable setups and replenishment is repetitive. However, fixed quantities may generate excess inventory when demand falls and shortages when demand rises. In MRP, a fixed quantity can also create residual inventory after the immediate requirement is satisfied.

4.3 Periodic ordering

Under periodic review, inventory position is checked every T periods. The order raises the inventory position to a target level S. A general order expression is:

where Q_t is the order at review time t and IP_t is inventory position, defined as on-hand inventory plus on-order inventory minus backorders. The target must cover demand over the protection interval T + L and the desired safety buffer. Under average demand d:

Periodic ordering is convenient for route deliveries, supplier schedules and low-value items. Because no order is placed between review dates, it requires more protection against demand and lead-time uncertainty than continuous review. It can also synchronise orders and reduce administrative effort.

4.4 Dynamic lot sizing

Dynamic lot-sizing methods respond to requirements that vary by period. They are particularly relevant to MRP, where demand is time-phased rather than uniform. The objective is generally to minimise setup and holding costs over a finite horizon while meeting all requirements.

The part-period balancing heuristic selects a lot so that cumulative holding cost for future requirements is approximately comparable with one setup cost. A part-period is one unit held for one period. If a setup costs $500 and holding cost is $2 per unit per period, the target is approximately 250 part-periods. This is a heuristic, not a guaranteed optimum.

The Wagner–Whitin dynamic programming model can identify an optimal finite-horizon lot-sizing plan under restrictive assumptions: deterministic time-phased demand, fixed setup costs, linear holding costs, no shortages, known lead times and unlimited replenishment capacity. Let D_t be demand in period t, S be setup cost and h be per-unit per-period holding cost. A recurrence can be expressed as:

where F_t is minimum cost through period t and k is the period in which the lot covering periods k through t is released. The summation charges holding cost for units produced in k and used later. Although exact dynamic programming is valuable for analysis, practical MPS may need to add capacity, safety stock, minimum lots, multiple items and uncertainty.

4.5 Other practical rules

Common MRP lot-sizing rules include:

• Minimum order quantity: enforces a supplier or process minimum.
• Fixed-period requirements: combines requirements over a specified number of periods.
• Least total cost: extends a lot while added holding cost remains economically justified relative to setup cost.
• Maximum quantity: limits exposure to capacity, shelf life or cash constraints.
• Silver-meal heuristic: seeks a lot with low average cost per period, especially under changing requirements.

No heuristic should be selected solely because it is embedded in software. Its cost logic, responsiveness and operational consequences must be tested against actual conditions.

5. Comparative Evaluation

Policy choice should consider the full system. A large batch can be rational at one work centre but harmful downstream if it creates congestion. Similarly, purchasing discounts may reduce unit price while increasing obsolescence and storage costs.

6. Shortages, Service and Risk

Inventory policy is a risk-management decision. A firm should specify the service measure being protected. Cycle service level is the probability of no stockout in a replenishment cycle. Fill rate is the fraction of demand supplied immediately from available stock. These are not interchangeable: a policy can achieve a high cycle service level while still having a poor fill rate if occasional shortages are large.

Shortage treatment may include backorders, partial shipment, substitution, expediting or lost sales. The appropriate response depends on customer criticality and contractual obligations. Safety stock should not be used to conceal unreliable suppliers, inaccurate bills of material, unstable processes or poor forecasts. Root-cause correction is often cheaper than permanent buffer growth.

7. Obsolescence and Working-Capital Governance

Traditional lot-sizing models treat inventory as usable indefinitely. In practice, inventory has a life-cycle value. A useful managerial decision is to compare the expected benefit of a larger lot with the expected cost of excess and obsolete stock:

This is an expected-value approximation and should include salvage value where relevant. It is especially important when engineering changes, regulatory expiry or rapid technology substitution is likely.

Working-capital governance requires more than minimising average inventory. Managers should monitor inventory turns, days of supply, aged inventory, excess-and-obsolete value, service level, schedule adherence and cash conversion. A basic inventory-turns measure is:

The measure is meaningful only when valuation methods and time periods are consistent. Very high turns may indicate efficiency, but may also signal understocking and lost sales.

8. Technology, Data and AI

ERP and MRP systems apply lot-sizing rules to time-phased requirements. Their quality depends on master data: lead times, setup costs, order multiples, safety stocks, scrap factors, yields and bills of material. Incorrect parameters can produce mathematically correct but operationally poor plans.

Advanced planning systems can simulate capacity, supplier constraints and alternative scenarios. Machine-learning forecasts may improve demand estimates, but forecasting accuracy does not eliminate the need for policy judgement. AI-generated recommendations should be governed through data lineage, explainability, approval thresholds, exception monitoring and human accountability. A model that recommends large lots without considering shelf life or cash limits is not a sound production-management system.

9. Illustrative Case: Meridian Medical Devices

Background

Meridian Medical Devices produces a disposable diagnostic cartridge. Demand is seasonal and uncertain. Each production setup costs $1,200 because the line must be cleaned, calibrated and validated. The cartridge has a six-month approved shelf life. Unit manufacturing cost is $18, and the organisation uses a 24% annual holding rate. The MPS horizon is six months, with monthly requirements of 1,000, 1,800, 900, 2,600, 1,200 and 700 units.

Problem

The plant currently uses a fixed monthly batch of 2,500 units. This reduces setups but creates substantial inventory before low-demand months. Quality management has also identified a risk that inventory produced early may approach expiry.

Analysis

The approximate annual holding cost per unit is:

H = 0.24 × $18 = $4.32 per unit per year

A basic EOQ using annualised demand of 16,400 units gives:

Q* = √((2 × 16,400 × 1,200) ÷ 4.32) ≈ 3,019 units

However, the EOQ assumptions are weak here. Demand is seasonal, shelf life is material, and validation rules may restrict the timing of production. A 3,019-unit batch could create expiry exposure and would not align naturally with the time-phased requirements.

Approach and solution

The production team compared three policies: fixed 2,500-unit batches, L4L, and a dynamic plan that consolidated requirements only when the projected setup saving exceeded incremental holding and expiry risk. The selected plan produced 2,800 units in month 1 to cover months 1 and 2, 2,600 units in month 3 to cover month 4, and 1,900 units in month 5 to cover months 5 and 6. The plan was subject to a final expiry check and a capacity limit of 3,000 units per run.

The team also introduced weekly review of projected expiry, formal approval for lots exceeding three months of forecast demand, and a supplier agreement for emergency validated production.

Outcome

The dynamic plan reduced the number of setups compared with L4L, while avoiding the repeated 2,500-unit batches that had created aged inventory. It also made the trade-off transparent: a setup was accepted when the holding and expiry cost of consolidation exceeded the setup saving. The case illustrates that EOQ can be a useful reference but not a sufficient decision rule.

Learning points

1. Time-phased demand can make dynamic lot sizing more suitable than a fixed quantity.
2. Shelf life converts inventory quantity into a time-dependent risk.
3. Capacity, validation and quality constraints must be incorporated into MPS decisions.
4. Policy parameters require periodic review as demand and process economics change.

10. Best-Practice Decision Framework

A disciplined lot-sizing process should:

1. Classify the item by demand type, value, criticality, variability and life cycle.
2. Validate demand, lead time, setup, holding, shelf-life and capacity data.
3. Identify feasible policies rather than comparing theoretical policies only.
4. Calculate relevant cost, including shortage and obsolescence exposure.
5. Test service, capacity and cash-flow outcomes through scenarios.
6. Establish exception rules for demand changes, engineering changes and supplier disruption.
7. Review parameters using actual performance data.
8. Coordinate purchasing, production, finance, quality and customer-service objectives.

The best policy is therefore contingent. Stable, low-value items may justify periodic or fixed-quantity replenishment. Volatile, expensive or perishable items may require L4L, dynamic sizing or postponement. The policy should be simple enough to execute but sophisticated enough to reflect the principal economic risks.

Challenges and Limitations

All lot-sizing models simplify reality. Costs may be nonlinear, capacity may be finite, demand may be intermittent, lead times may be unreliable and multiple products may compete for the same resource. EOQ and related formulas can give false precision when cost estimates are poor. Dynamic programming can become computationally difficult when capacity, multiple levels of bills of material and uncertainty are added. Heuristics are practical but not necessarily optimal.

There are also behavioural limitations. Departments may inflate forecasts or order quantities to protect local performance. Buyers may prioritise unit-price discounts, while production prioritises long runs and finance prioritises low inventory. Effective governance therefore requires common measures, visible assumptions and cross-functional review.

Chapter Summary

Lot sizing determines the quantity and timing of replenishment in the master production schedule. Lot-for-lot minimises planned cycle inventory but may create frequent setups. Fixed-order quantity offers operational simplicity, while EOQ balances ordering and holding costs under restrictive assumptions. Periodic ordering groups decisions at review intervals but needs protection across the review and lead-time interval. Dynamic lot-sizing methods use time-phased requirements and are especially relevant to MRP.

Sound decisions include setup, ordering, holding, shortage, obsolescence and working-capital effects. Service level must be defined carefully, because cycle service and fill rate measure different outcomes. EOQ, reorder points and safety-stock equations are useful analytical tools, but real decisions must incorporate capacity, shelf life, quality, supplier constraints and demand uncertainty. ERP and AI can support policy execution, yet reliable master data, governance and human accountability remain essential.

Key Terms

Lot sizing: Determining the quantity released or ordered at a replenishment event.

Lot-for-lot: A rule that sets each planned order equal to the net requirement in that period.

Fixed-order quantity: A policy that uses a predetermined replenishment quantity.

Economic order quantity: The quantity that balances ordering and cycle-stock holding cost under stated assumptions.

Ordering cost: Cost incurred each time a purchase order is placed.

Setup cost: Cost and lost time associated with preparing a production process.

Holding cost: Cost of carrying inventory over time.

Inventory position: On-hand plus on-order inventory minus backorders.

Reorder point: Inventory position at which replenishment is triggered.

Safety stock: Inventory held to protect against uncertainty.

Periodic review: A system in which inventory is evaluated at fixed time intervals.

Dynamic lot sizing: Adjusting lot quantities to time-phased requirements.

Independent demand: Demand arising from external markets or customers.

Dependent demand: Demand calculated from parent-item plans and bills of material.

Cycle service level: Probability of avoiding a stockout during a replenishment cycle.

Fill rate: Proportion of demand supplied immediately from available inventory.

Obsolescence: Loss of usable or economic value due to age, change or expiry.

Working capital: Funds committed to operating assets such as inventory.

Part-period: One unit held for one time period, used in lot-sizing heuristics.

Review Questions

Short Answer Questions

1. What is the difference between independent and dependent demand?
2. State two assumptions of the basic EOQ model.
3. Why does lot-for-lot often increase setup frequency?
4. Define inventory position.
5. Distinguish cycle service level from fill rate.

Long Answer Questions

1. Explain how setup, holding, shortage and obsolescence costs influence lot-sizing decisions.
2. Compare lot-for-lot, fixed-order quantity, EOQ and periodic ordering policies.
3. Derive the basic EOQ expression and explain its managerial interpretation.
4. Discuss why dynamic lot sizing is particularly relevant to MRP.
5. Explain how working-capital objectives can conflict with production efficiency.

Analytical / Application Questions

1. Annual demand is 30,000 units, ordering cost is $90 per order and annual holding cost is $6 per unit. Calculate EOQ and explain two reasons why the actual order quantity might differ.
2. A firm reviews inventory every two weeks, has a four-week lead time and average demand of 500 units per week. Explain the demand interval that the target stock must cover and identify the data needed to set safety stock.
3. A product has high setup cost but a three-month shelf life and highly uncertain demand. Recommend a lot-sizing approach and justify it using cost, service, obsolescence and governance considerations.

Multiple Choice Questions with Answers

Question 1

Which policy sets the planned order quantity equal to the net requirement in each period?

A. Lot-for-lot

B. Fixed-order quantity

C. Periodic review

D. Economic order quantity

Correct Answer: A. Lot-for-lot

Question 2

Under the basic EOQ model, which cost is balanced against ordering cost?

A. Shortage cost only

B. Cycle-stock holding cost

C. Expediting cost only

D. Labour turnover cost

Correct Answer: B. Cycle-stock holding cost

Question 3

Annual demand is 24,000 units, ordering cost is $75, and annual holding cost is $8 per unit. The EOQ is approximately:

A. 150 units

B. 335 units

C. 671 units

D. 1,500 units

Correct Answer: C. 671 units

Question 4

Which expression represents a basic reorder point with safety stock?

A. ROP = setup cost + holding cost

B. ROP = demand during lead time + safety stock

C. ROP = annual demand ÷ order quantity

D. ROP = unit cost × holding rate

Correct Answer: B. ROP = demand during lead time + safety stock

Question 5

In a periodic-review system, the protection interval normally includes:

A. Only the review interval

B. Only supplier lead time

C. Review interval plus lead time

D. Annual demand divided by unit cost

Correct Answer: C. Review interval plus lead time

Question 6

Which condition most strongly supports lot-for-lot sizing?

A. Very high setup costs and stable demand

B. High obsolescence risk and low setup cost

C. Large quantity discounts and unlimited shelf life

D. Constant demand with no time-phased requirements

Correct Answer: B. High obsolescence risk and low setup cost

Question 7

What does inventory position normally include?

A. Only on-hand inventory

B. On-hand plus on-order inventory minus backorders

C. Annual demand minus safety stock

D. Finished goods value only

Correct Answer: B. On-hand plus on-order inventory minus backorders

Question 8

A part-period in lot-sizing analysis represents:

A. One order per supplier period

B. One unit held for one period

C. One setup per production shift

D. One shortage per customer

Correct Answer: B. One unit held for one period

Question 9

Which is a limitation of the basic EOQ model?

A. It requires no demand information

B. It assumes demand and lead time are known and stable

C. It automatically models expiry risk

D. It always incorporates finite capacity

Correct Answer: B. It assumes demand and lead time are known and stable

Question 10

Which measure expresses the proportion of demand supplied immediately from available stock?

A. Fill rate

B. Inventory turns

C. Setup rate

D. Cycle time

Correct Answer: A. Fill rate