Doing Well by Doing Good: Alignment of Revenue and Welfare in Robust Product Line Design through Data-Driven Optimization

Product line design (PLD) selects product variants for a firm to maximize expected revenue. Candidate assortments are usually evaluated under random utility maximization (RUM), in which choice probabilities stem from the joint distribution of latent consumer utilities. As this utility distribution cannot be directly observed and is estimated from choice data, distributional uncertainty inevitably arises. This paper studies how to hedge PLD decisions against such ambiguity. We construct a distributionally robust optimization (DRO) model that maximizes worst-case expected revenue over an l1-Wasserstein ball around a nominal utility distribution. We derive a dual reformulation for general nominal distributions and then specialize to two cases. With independent Gumbel utility shocks, the dual simplifies to a closed-form robust multinomial logit. We derive revenue gap bounds for parameters estimated from finite full-assortment choice data. When the true model structure is unknown beforehand, we adopt a data-driven scheme using the empirical distribution of sampled utility vectors and extend optimality-gap analysis to a general RUM oracle. For the NP-hard empirical problem, we propose Adaptive Parametric Search (APS), an algorithm with certified relative optimality guarantees. Numerical experiments show fast computation and better out-of-sample performance.Dual analysis shows that robustness to preference uncertainty endogenously encourages the firm to raise buyer welfare. This incentive takes distinct forms: robustness supports larger assortments in the unconstrained logit setting, whereas under the empirical nominal, it favors products with lower prices and higher utility relative to the outside option. Numerically, we identify a "doing well by doing good" effect: product lines with higher in-sample buyer welfare attain higher out-of-sample revenue and buyer welfare.

Authors

Publication Details

Journal
arXiv (Cornell University)
Published
2026-11-17
DOI
https://doi.org/10.48550/arxiv.2611.22264
Primary Topic
Supply Chain and Inventory Management
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Doing Well by Doing Good: Alignment of Revenue and Welfare in Robust Product Line Design through Data-Driven Optimization

Xiuli Chao, Maoqi Liu, Yun Hui Lin, Jiajie Zhang et al.
arXiv (Cornell University)
Supply Chain and Inventory Management
preprint

Doing Well by Doing Good: Alignment of Revenue and Welfare in Robust Product Line Design through Data-Driven Optimization

Xiuli Chao, Maoqi Liu, Yun Hui Lin, Jiajie Zhang, Liangjie Li
preprint en

Abstract

Product line design (PLD) selects product variants for a firm to maximize expected revenue. Candidate assortments are usually evaluated under random utility maximization (RUM), in which choice probabilities stem from the joint distribution of latent consumer utilities. As this utility distribution cannot be directly observed and is estimated from choice data, distributional uncertainty inevitably arises. This paper studies how to hedge PLD decisions against such ambiguity. We construct a distributionally robust optimization (DRO) model that maximizes worst-case expected revenue over an l1-Wasserstein ball around a nominal utility distribution. We derive a dual reformulation for general nominal distributions and then specialize to two cases. With independent Gumbel utility shocks, the dual simplifies to a closed-form robust multinomial logit. We derive revenue gap bounds for parameters estimated from finite full-assortment choice data. When the true model structure is unknown beforehand, we adopt a data-driven scheme using the empirical distribution of sampled utility vectors and extend optimality-gap analysis to a general RUM oracle. For the NP-hard empirical problem, we propose Adaptive Parametric Search (APS), an algorithm with certified relative optimality guarantees. Numerical experiments show fast computation and better out-of-sample performance.Dual analysis shows that robustness to preference uncertainty endogenously encourages the firm to raise buyer welfare. This incentive takes distinct forms: robustness supports larger assortments in the unconstrained logit setting, whereas under the empirical nominal, it favors products with lower prices and higher utility relative to the outside option. Numerically, we identify a "doing well by doing good" effect: product lines with higher in-sample buyer welfare attain higher out-of-sample revenue and buyer welfare.

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