Vessel Profiles in the Shadows: A Binary Logit Analysis of Technical and Operational Divergence in Global Tanker Fleets

This article examines whether sanction-designated crude tankers can be distinguished from the wider global fleet through vessel-level characteristics and a cross-validated screening model. Using 2,277 crude oil tankers from Clarksons SIN, of which 382 are positive on a composite designation indicator, the study separates structural inference from operational prediction. Logit and probit models with a spline age term show that designation risk is concentrated in older Aframax tonnage. Relative to Aframax vessels, Suezmaxes and VLCCs exhibit substantially lower odds of designation, while shuttle-tanker status is not robust. The age effect is distinctly non-linear, peaking in the early-to-mid twenties. A screening model that adds pooled flag information attains ROC-AUC 0.9306, PR-AUC 0.7288, and a Brier score of 0.0750. The results support vessel-centred triage, but the model is best interpreted as a designation-risk classifier rather than direct proof of latent shadow-fleet membership.