Estimating traffic congestion cost uncertainty using a bootstrap scheme

Abstract

This study introduces a bootstrap-based approach to estimate the uncertainty in the total economic cost of congestion (TCC) due to traffic delays. Focusing on Medellín, Colombia, we employed a stratified random sampling plan of road segments to capture real-time traffic data via Google Maps. By integrating a Linear Mixed-Effects model with a nonparametric Bootstrap method, we produced robust hourly delay distributions, which were then used to estimate TCC. Our findings estimate Medellín’s annual congestion cost at approximately USD 375.7 million, with a 95 % confidence interval ranging from USD 348.2 to USD 405.2 million. This range not only quantifies the uncertainty in congestion costs but also provides a benchmark for future comparisons, enabling policymakers to distinguish significant changes from random fluctuations. The results offer critical insights for urban planning, highlighting key road characteristics that could reduce congestion and addressing the variability often overlooked in cost estimates.