Inventory stress as financial risk: Trade shocks, macro-financial conditions and explainable early warning for U.S. retail and manufacturing

Abstract Inventories are among the largest working-capital investments of retailers and manufacturers,and inventory imbalances destroy value: overhang ties up capital and forces markdowns and productioncuts, while shortage costs sales and margins. Lenders, investors and managers would benefit from treat-ing inventory stress like credit or financial-market stress, as a risk event that can be priced and forecast.This study develops an explainable early-warning framework for inventory stress in U.S. retail and man-ufacturing. Stress is defined from the Hamilton regression-filter gap of the Census inventories-to-salesratios, separating overhang (excess stocks) from shortage (lean stocks) episodes between 1994 and 2026.Predictors cover trade conditions (imports, exports, the trade balance, import prices, the real dollar andwholesale inventories) and macro-financial conditions (activity, orders, sales, prices, unemployment,sentiment, the policy rate, the term spread, the credit spread and oil), each aligned to its publication lag.A persistence benchmark (AR-logit), a logit with moving-block bootstrap inference, an elastic-net logit,a random forest, XGBoost and an equal-weight ensemble are compared in a pseudo-real-time evaluation(April 2005–July 2026) with information-set ablation and SHAP attributions. In sample, weak industrialproduction, a flat yield curve, wide credit spreads and wholesale stock accumulation precede manufac-turing overhang, the financing channel of inventory investment, but bootstrap standard errors are up to14.93 times the conventional ones. Out of sample, stress is predictable (mean AUC 0.89 one month and0.81 three months ahead), yet no model significantly beats the AR-logit, and the best-fitting logit fore-casts worst. Trade and macro-financial information helps mainly for shortage stress, especially in retail(three-month AUC 0.83 for the random forest versus 0.80), and its SHAP share rises with the horizon.Persistence benchmarks and dependence-robust evaluation should therefore be standard in financial andoperational stress prediction. JEL Classification G17 · G32 · E22 · E32 · E44 · F14 · C53

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-12
DOI
https://doi.org/10.21203/rs.3.rs-11438198/v1
Primary Topic
Working Capital and Financial Performance
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Inventory stress as financial risk: Trade shocks, macro-financial conditions and explainable early warning for U.S. retail and manufacturing

Dr. Md Mizanur Rahman, Fatema Tuj Zohra, Md Farhad Hossain, Abdullah Al Shohan et al.
Research Square
Working Capital and Financial Performance
preprint

Inventory stress as financial risk: Trade shocks, macro-financial conditions and explainable early warning for U.S. retail and manufacturing

Dr. Md Mizanur Rahman, Fatema Tuj Zohra, Md Farhad Hossain, Abdullah Al Shohan, Md Mazharul Islam
preprint en

Abstract

Abstract Inventories are among the largest working-capital investments of retailers and manufacturers,and inventory imbalances destroy value: overhang ties up capital and forces markdowns and productioncuts, while shortage costs sales and margins. Lenders, investors and managers would benefit from treat-ing inventory stress like credit or financial-market stress, as a risk event that can be priced and forecast.This study develops an explainable early-warning framework for inventory stress in U.S. retail and man-ufacturing. Stress is defined from the Hamilton regression-filter gap of the Census inventories-to-salesratios, separating overhang (excess stocks) from shortage (lean stocks) episodes between 1994 and 2026.Predictors cover trade conditions (imports, exports, the trade balance, import prices, the real dollar andwholesale inventories) and macro-financial conditions (activity, orders, sales, prices, unemployment,sentiment, the policy rate, the term spread, the credit spread and oil), each aligned to its publication lag.A persistence benchmark (AR-logit), a logit with moving-block bootstrap inference, an elastic-net logit,a random forest, XGBoost and an equal-weight ensemble are compared in a pseudo-real-time evaluation(April 2005–July 2026) with information-set ablation and SHAP attributions. In sample, weak industrialproduction, a flat yield curve, wide credit spreads and wholesale stock accumulation precede manufac-turing overhang, the financing channel of inventory investment, but bootstrap standard errors are up to14.93 times the conventional ones. Out of sample, stress is predictable (mean AUC 0.89 one month and0.81 three months ahead), yet no model significantly beats the AR-logit, and the best-fitting logit fore-casts worst. Trade and macro-financial information helps mainly for shortage stress, especially in retail(three-month AUC 0.83 for the random forest versus 0.80), and its SHAP share rises with the horizon.Persistence benchmarks and dependence-robust evaluation should therefore be standard in financial andoperational stress prediction. JEL Classification G17 · G32 · E22 · E32 · E44 · F14 · C53

Research Square
The University of Texas at Dallas (US), Comilla University (BD)
Working Capital and Financial Performance
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