Development of a model to predict construction Cost Over-Runs of building projects in Uganda

Abstract

Cost contingencies that range from 5% to 10% in budgets of construction projects in Uganda are routinely exhausted before projects are complete necessitating mobilization of more funds that results in cost over-runs. The contingencies reserves are determined subjectively. This study developed a model to predict cost over-runs of building projects in Uganda for mitigating cost over-runs. Cost over-run data was collected by structured questionnaires from built environment professionals and contractors who were involved variously in managing building projects that had cost over-runs. Both purposive sampling and snow-balling sampling techniques were used to obtain cost over-run data from 37 past building projects. Cost contingencies that average 14% for project planning were found to under-operate at 89% confidence level leading to rampant cost over-runs in projects. The prediction model developed uses bid time, performance bond status, insurance amount, workload and experience to predict cost over-runs with 63% co-efficient of determination with 9% margin of error. Cost contingencies of 35% was found to be the most adequate to cover cost over-runs at 95% confidence interval. Bid time and performance bond were the most significant factors that reduce cost over-runs. Five variants of the model were therefore developed for each of the five procurement methods of PPDA for public works. This necessitates the upward revision of bid times for open and restricted domestic bidding from the current 12 and 20 days to 20 and 30 days respectively to reduce cost over-run substantially. The study was limited to public building projects as opposed to linear sites like road works and water supply lines that extend over vast areas. Research in these areas is therefore recommended. Keywords: contingencies, cost over-run, prediction models, Uganda.