Oracle of Bids: Preaward Bid Dispersion as an Early Warning Signal of Schedule Delays in Infrastructure Projects

Abstract Schedule delays in public infrastructure projects frequently result in liquidated damages (LD), yet most early warning tools rely on postaward information, limiting opportunities for prevention. This study proposes a preaward early warning framework that uses bid-stage dispersion as a market-generated signal of latent project uncertainty. Drawing on collective-intelligence and procurement-economics perspectives, the study tests whether dispersion among submitted bids is associated with LD outcomes using a manually constructed data set of 642 projects awarded by a US state transportation agency between 2015 and 2020. The analysis is associational and intended for preaward risk screening rather than causal inference. A logistic-regression model using only preaward information demonstrates meaningful discriminatory performance, indicating that projects incurring LD are systematically ranked higher in predicted risk at bid opening. Among dispersion measures, interquartile-range (IQR) dispersion exhibits the most persistent association with LD risk and remains nonzero under highly restrictive specifications that absorb county and award-year effects and apply regularization. Grouped cross-validation shows that predictive performance remains stable across counties and award years, with only modest attenuation under strict spatial and temporal hold-out. Robustness checks confirm that results are not driven by outliers or heavy-tailed LD severity. The findings indicate that bid dispersion‍—particularly central-mass disagreement captured by IQR—provides a low-burden, preaward early warning signal of schedule-delay risk. Because bid distributions are already available in modern e-procurement systems, dispersion-based analytics can be readily integrated into procurement workflows to support early screening, targeted review, and proactive oversight planning.