Mechanical condition of vehicles and the durability of road safety systems: A 17-year time-series analysis in Benin (2007–2023)

Abstract

Objective. This study analyses the relationship between vehicle mechanical condition and the frequency of road accidents in Benin over seventeen years, testing whether vehicles in poor condition are disproportionately involved in accidents and whether this relationship remains stable over time, to inform sustainable road safety policies in a resource-constrained context. Methodology. Using national monthly accident data from 2007 to 2023 (204 observations, covering 88,605 accidents, 153,531 vehicles in good condition and 3,957 vehicles in poor condition), the study employs descriptive statistics, paired Wilcoxon tests, Pearson’s correlation analysis, multiple linear regression and Poisson regression modelling (GLM), supplemented by interaction and structural break analysis to assess temporal stability. We conducted robustness analyses using a quasi-Poisson model with Newey-West robust standard errors (lag = 12) to correct for autocorrelation and underdispersion. We performed a Granger causality test to assess the direction of causality. Results. The Poisson regression shows that each additional vehicle in poor condition increases the number of monthly accidents by 0.15 per cent (IRR = 1.0015; 95% CI [1.0005 – 1.0025]; p = 0.002), a modest but statistically significant and temporally stable effect (interaction p = 0.649). Robustness analyses using Newey-West standard errors confirm this effect (z = 5.933; p < 0.001). A negative aggregate correlation between the proportion of vehicles in poor condition and the total number of accidents (r = −0.308) suggests confounding by inspection cycles or economic conditions rather than a genuine protective effect. The Granger causality test indicates that vehicles in poor condition statistically predict future changes in the number of accidents (F = 3.27; p = 0.022), alleviating concerns about reverse causality. A policy impact simulation estimates that a 10 per cent reduction in vehicles in poor condition would prevent approximately 15 accidents per year. Theoretical contribution. This paper offers one of the few long-term, monthly-resolution analyses of the relationship between vehicle condition and accident frequency from a sub-Saharan African country. It is among the first to formally test the temporal stability of this relationship. Practical implications. The results support sustained investment in periodic vehicle inspections, incentives for maintenance and targeted roadside checks as part of Benin’s contribution to the United Nations Decade of Action for Road Safety and SDG 3.6, whilst emphasising that vehicle condition accounts for only a modest proportion of the variance in accidents compared with exposure and infrastructure factors. Sustainable Development Goals (SDGs): SDG 3: Good health and well-being; SDG 9: Industry, innovation and infrastructure; SDG 11: Sustainable cities and communities