Abstract
The collapse of buildings under construction represents a critical infrastructure failure with severe human and economic consequences across developing economies. Despite growing public concern in Africa — where over 200 documented collapse incidents and several hundred associated fatalities have been recorded in the past decade — there is limited systematic empirical evidence on the institutional determinants of such structural failures. Grounded in institutional theory and principal–agent theory, this study constructs a novel country–year panel dataset of building collapse incidents across 54 African countries and examines whether institutional quality, corruption, and economic development influence the incidence of collapses. Using count data models with year fixed effects over the period 2016–2023 (396 complete-case country–year observations), we estimate the impact of corruption perceptions (CPI score), regulatory quality, GDP per capita, total population, and urban population growth on collapse frequency. Likelihood-ratio tests decisively favour the Negative Binomial (NB2) specification over Poisson, reflecting substantial overdispersion in the data. The main finding is that national population size is the strongest and most robust predictor of both collapse counts (IRR = 2.05, p<0.001) and fatalities (IRR = 3.63, p<0.001), consistent with a scale-of-exposure mechanism. The CPI score exhibits a marginally significant negative association with fatalities (IRR = 0.98, p=0.052), suggesting that less corrupt countries tend to experience fewer collapse-related deaths. Regulatory quality, GDP per capita, and urban population growth are not statistically significant under NB2 once population is controlled. These results indicate that the previously documented positive association between regulatory quality and collapse counts was primarily an artefact of omitting population as an exposure control. Once the scale of the built environment is accounted for, institutional variables exhibit theoretically expected signs but limited statistical power in this sample. The findings highlight the importance of accounting for population-driven exposure in cross-country analyses of infrastructure safety, and contribute to the broader literature linking governance to disaster risk reduction.
•Building collapses across 54 African nations are analysed over 2016–2023.•Population size dictates both building collapse frequency and fatalities.•Corruption correlates with collapse counts but shows no link to fatalities.•Regulatory quality effects reflect population exposure, not governance failure.•Fixed-effects models confirm institutional traits do not explain data trends.