Identification of factors affecting the performance of rural road projects in Colombia

Abstract

Rural roads play an indispensable role in economic and social well-being, especially in developing countries, contributing to achieving the Sustainable Development Goals. For this reason, it is necessary to plan these projects properly to guarantee their success. In this line, the objective of this research is to identify significant variables generating overruns in time and cost using empirical data of 535 rural road projects in Colombia from 2015 to 2018. Bivariate analysis, with statistical tools like Spearman's Rho and Kruskal-Wallis, allowed identifying that higher values of variables like budget and project intensity are related to higher deviations in cost and time. Additionally, it was found that projects with shorter durations are reporting higher time overruns. The worst performers are projects executed in the year that council mayors start their terms, those developed in municipalities with more resources, and those awarded using a competitive bidding process. Multivariate analysis, through Random Forest, assessed the effect of considering all variables interacting simultaneously and ranking them in order of importance. The results demonstrated a relationship between cost and time performance, and that numerical variables are more significant than the categorical ones. This study contributes to a better understanding of the causes of delays and cost overruns on rural roads, providing useful insight for researchers and industry practitioners.