A 5/2 Guarantee for Echelon Stock (s,S) Policies in One-Warehouse Multi-Retailer Systems

We study a periodic-review warehouse serving multiple heterogeneous retailers with stochastic demand and fixed costs and lead times at both echelons. We construct a deterministic echelon stock $(s,S)$ policy whose long-run expected average cost is at most $5/2$ times the optimum over all nonanticipative policies. The guarantee holds when the common backlog penalty and retailer holding rates are at least the warehouse holding rate and retailer setup costs are at most $3/2$ times the supplier's. The retailer-specific setup-cost condition extends beyond supplier setup-cost dominance: a retailer dispatch may cost up to 50 percent more than a supplier order. Demand vectors are i.i.d. over periods with finite positive means, and demands of different retailers may be correlated. Each retailer selects the full cycle of its single-location plan or a shifted cycle of half its width, and the warehouse serves requests in arrival order, allowing partial shipments. The upper bound on the policy's cost includes each retailer's mirror cost, which prices the warehouse imbalance created by its batching. The upper and lower bounds share a pooled supplier problem and add one problem per retailer, giving a factor independent of all lead times and the number of retailers. For arbitrary nonnegative setup costs and associated retailer demands, a consolidation policy holds allocated stock until a dispatch threshold is met and gives a factor below $4.9$. On 864 instances built from Walmart store sales, the simulated cost of an integer-rounded implementation of the FIFO policy averages 1.45 times the lower bound and reaches 1.75, and our analytical upper bound on its cost is below 1.94 times the lower bound. With no simulation search, the policy comes within 8.2 and 12.2 percent, on average over 216 instances, of an echelon batch policy and an integer-ratio policy, both tuned by simulation for each instance.

Authors

Publication Details

Journal
arXiv (Cornell University)
Published
2026-11-17
DOI
https://doi.org/10.48550/arxiv.2611.22468
Primary Topic
Supply Chain and Inventory Management
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

A 5/2 Guarantee for Echelon Stock (s,S) Policies in One-Warehouse Multi-Retailer Systems

Ming Hu
arXiv (Cornell University)
Supply Chain and Inventory Management
preprint

A 5/2 Guarantee for Echelon Stock (s,S) Policies in One-Warehouse Multi-Retailer Systems

Ming Hu
preprint en

Abstract

We study a periodic-review warehouse serving multiple heterogeneous retailers with stochastic demand and fixed costs and lead times at both echelons. We construct a deterministic echelon stock $(s,S)$ policy whose long-run expected average cost is at most $5/2$ times the optimum over all nonanticipative policies. The guarantee holds when the common backlog penalty and retailer holding rates are at least the warehouse holding rate and retailer setup costs are at most $3/2$ times the supplier's. The retailer-specific setup-cost condition extends beyond supplier setup-cost dominance: a retailer dispatch may cost up to 50 percent more than a supplier order. Demand vectors are i.i.d. over periods with finite positive means, and demands of different retailers may be correlated. Each retailer selects the full cycle of its single-location plan or a shifted cycle of half its width, and the warehouse serves requests in arrival order, allowing partial shipments. The upper bound on the policy's cost includes each retailer's mirror cost, which prices the warehouse imbalance created by its batching. The upper and lower bounds share a pooled supplier problem and add one problem per retailer, giving a factor independent of all lead times and the number of retailers. For arbitrary nonnegative setup costs and associated retailer demands, a consolidation policy holds allocated stock until a dispatch threshold is met and gives a factor below $4.9$. On 864 instances built from Walmart store sales, the simulated cost of an integer-rounded implementation of the FIFO policy averages 1.45 times the lower bound and reaches 1.75, and our analytical upper bound on its cost is below 1.94 times the lower bound. With no simulation search, the policy comes within 8.2 and 12.2 percent, on average over 216 instances, of an echelon batch policy and an integer-ratio policy, both tuned by simulation for each instance.

arXiv (Cornell University)
Supply Chain and Inventory Management
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon