Base-Stock Policy

The defining mechanism of a base-stock policy is the one-for-one replenishment rule. Every time a demand event occurs and a unit is consumed, a replenishment order for exactly one unit is immediately placed with the upstream supplier. This creates a direct, unbroken link between consumption and replenishment.

• Trigger: Each individual demand unit, not a reorder point threshold.
• Order Quantity: Always exactly one unit.
• Result: Inventory position remains constant at the base-stock level R.
• Contrast: Unlike (Q,R) policies, there is no batch ordering; unlike periodic review, there is no fixed review interval.

Under a pure base-stock policy, the inventory position—defined as on-hand inventory plus units on order minus backorders—is perpetually fixed at the base-stock level R. This invariance is the policy's central mathematical property and simplifies analysis significantly.

• On-hand stock fluctuates with demand.
• Pipeline stock fluctuates inversely to on-hand stock.
• Inventory position = On-hand + On-order - Backorders = R (constant).
• This property holds only when orders are placed immediately upon each demand unit.

The base-stock policy is the mathematically optimal control policy for items with Poisson or compound Poisson demand patterns, particularly when the cost of placing an order is negligible relative to holding and shortage costs.

• Ideal candidates: Expensive spare parts, aircraft components, specialized medical devices, high-end electronics.
• Why optimal: Eliminates batch-ordering inefficiencies; you never hold more inventory than the target R.
• Key assumption: Fixed ordering cost K ≈ 0. If K is significant, an (s,S) or (Q,R) policy may be superior.
• Service level: The base-stock level R directly determines the probability of no stockout during the replenishment lead time.

The optimal base-stock level R* is calculated by balancing the marginal cost of holding an additional unit against the marginal benefit of avoiding a stockout. For a Poisson demand process with rate λ and lead time L:

• Lead time demand is Poisson distributed with mean λL.
• R* is the smallest integer such that: P(Demand during L ≤ R) ≥ Critical Ratio.
• Critical Ratio = p / (p + h), where p is the unit shortage cost and h is the unit holding cost per lead time.
• This is a newsvendor-style critical fractile solution applied to the lead time demand distribution.

The classic base-stock policy assumes continuous review—every demand is observed instantly. In practice, many systems operate with periodic review intervals, leading to the (R,S) policy variant.

• Continuous review (Base-Stock): Order placed immediately upon each demand; inventory position always equals R.
• Periodic review (Order-Up-To): At fixed intervals, order enough to raise inventory position back to S. Inventory position drops between reviews.
• Key difference: Periodic review requires higher safety stock to cover the review period plus lead time.
• Hybrid approach: Modern demand sensing with real-time POS data enables near-continuous review even in traditionally periodic environments.

The inventory level that triggers a replenishment order. In a base-stock policy, the reorder point is effectively equal to the base-stock level minus one unit, since an order is placed immediately upon each demand. The ROP is calculated as:

• ROP = (d × LT) + SS
• Where d is average demand per period, LT is lead time, and SS is safety stock.
• For a pure base-stock system with unit demands, the ROP is a dynamic threshold that maintains the inventory position exactly at the base-stock level.

The maximum inventory position target used in periodic review systems. While a base-stock policy operates continuously, an order-up-to (OUT) policy reviews inventory at fixed intervals and places an order to raise the position back to the target.

• Base-stock is a special case of an order-up-to policy where the review period is instantaneous (continuous review).
• In practice, the OUT level equals: d × (R + LT) + SS, where R is the review interval.
• The key distinction: base-stock triggers orders per demand; OUT triggers orders per calendar interval.

The algorithmic calculation of buffer inventory required to absorb variability. In a base-stock policy, the base-stock level itself is the sum of expected lead-time demand plus safety stock:

• S = d × LT + z × σ × √LT
• Where z is the safety factor derived from the target service level, and σ is the standard deviation of demand.
• For high-value, slow-moving items governed by base-stock policies, safety stock is often calculated using a Poisson distribution rather than a normal distribution to accurately model discrete, low-volume demand patterns.

A holistic methodology that simultaneously optimizes stock levels across all supply chain nodes. A base-stock policy at a single echelon ignores upstream constraints, but MEIO models the network effect:

• A stockout at a warehouse delays replenishment to retail stores, dynamically increasing their effective lead time.
• Guaranteed Service Models (GSM) assume deterministic service times and place safety stock at strategic echelons.
• Stochastic Service Models (SSM) propagate real-time variability, requiring higher base-stock levels at downstream nodes when upstream service is unreliable.
• MEIO prevents the Bullwhip Effect by coordinating base-stock targets across the entire network.

Two distinct metrics that govern base-stock level calculation. Choosing the wrong metric leads to significant overstocking or understocking:

• Cycle Service Level (CSL): The probability of no stockout during a replenishment cycle. A 95% CSL means 5% of cycles experience a shortage, regardless of magnitude.
• Fill Rate (FR): The fraction of total demand units immediately fulfilled from stock. A 95% fill rate means 5% of units are backordered.
• For slow-moving items with a base-stock policy, fill rate is the more economically meaningful target, as CSL can be misleading when the number of demand events per cycle is very low.

The application of machine learning to short-term, high-frequency data signals to refine near-term forecasts. For base-stock policies governing slow-moving items, demand sensing improves the accuracy of the demand rate parameter (d):

• Traditional time-series models struggle with intermittent demand patterns.
• Croston's method and its variants decompose demand into size and interval components, providing a more stable estimate for base-stock calculation.
• Advanced techniques use Poisson and negative binomial distributions to model the discrete, sporadic nature of slow-moving demand, directly feeding the base-stock formula with a more accurate probability distribution.