Geng Integrity Review: 'The Impact of Logistics Performance on Trade' (Hausman, Lee & Subramanian, POM, 2012)

Academic fraud report · Geng Detector

Summary

This AI-assisted integrity review applies the six-step Geng framework to the article 'The Impact of Logistics Performance on Trade' (DOI: 10.1111/j.1937-5956.2011.01312.x), published in Production and Operations Management in 2012 by Stanford-affiliated and World Bank authors. The verdict is CLEAN: no image reuse, image splicing, data fabrication, statistical anomalies, mass-production patterns, or citation irregularities were detected. The paper is a macro-econometric empirical study based on World Bank Logistics Performance Index (2005) and trade data, an article type that limits pixel-level image forensics. Key positive indicators include a realistic R² of 0.717, t-statistics ranging from 1.18 (insignificant) to 75.57, honest reporting of insignificant control variables with counterintuitive signs, transparent acknowledgment that 60 of 140 countries' cost data were unverifiable and dropped, a credible seven-year research timeline (2005 data collection to 2012 publication), and consistent data vintages all predating the 2009 submission. Limitations: the framework cannot detect certain economic fraud types such as data source misrepresentation, and conclusions rest on textual and quantitative inspection rather than replication.

Verdict

Clean (✅) — No evidence of data fabrication, image manipulation, or systematic research misconduct was found. The paper is a legitimate macro-econometric empirical study.

Key findings

Image reuse (Step 1): Not applicable. The paper contains no Western blots, microscopy, or comparable images suitable for pixel-level forensic comparison; Appendix B bar charts are standard statistical visualizations.
Data fabrication (Step 2): No indicators detected. Regression output (Table 3) shows R² = 0.717, consistent with gravity models of bilateral trade. t-statistics span 1.18 to 75.57, including genuinely insignificant variables. Correlation matrices (Appendices C, D) retain multi-decimal precision (e.g., 0.4475, -0.1495) with no implausible perfect correlations.
Image splicing (Step 3): Not applicable. No experimental images present.
Statistical anomalies (Step 4): None detected. The authors report insignificant and counterintuitively signed control variables (e.g., Log of importer's average time, coefficient 0.171; Log of importer's MaxTime-AvgTime difference, coefficient 0.090) without suppression or reframing — consistent with honest statistical reporting rather than p-hacking.
Mass-production patterns (Step 5): None detected. Timeline is internally consistent: 2005 World Bank survey → submission January 2009 → acceptance August 2010 (single revision) → publication 2012, a seven-year span appropriate for large-scale macro research.
Citations and methodology (Step 6): No anomalies. All data vintages (2003 trade data, World Bank GDP, Gleditsch & Ward 2001 distances, 2004 corruption index) predate the 2009 submission. The double-log gravity-model specification is standard for international trade elasticity estimation, and the elasticity derivation is mathematically self-consistent.

Evidence highlights

DOI: 10.1111/j.1937-5956.2011.01312.x
Journal: Production and Operations Management (POM), 2012
R²: 0.717 (Table 3)
t-statistic range: 1.18 – 75.57
Sample reduction: 60 of 140 countries' cost data dropped due to unverifiability, explicitly disclosed by authors
Submission–acceptance–publication timeline: January 2009 → August 2010 → 2012
Correlation matrix precision examples: 0.4475, -0.1495 (Appendices C, D)

Notes

The Geng framework, originally designed with biomedical imaging forensics in mind, has limited applicability to pure macro-econometric papers; absence of image-related findings is partly an artifact of article type rather than exoneration.
The review cannot independently verify that the underlying World Bank dataset was used as described; replication of the regression results using the cited 2005 LPI and 2003 trade data would strengthen confidence.
All assessments are AI-assisted and should be treated as preliminary; authoritative conclusions require institutional investigation.