Abstract:
Lowest bid has often been favoured for award of construction contracts in most competitive bidding due to its perceived advantage to the client. However, this practice has also been found to eclipse with anomalies. For example, some bid figures are unrealistically low because desperate contractors cut down the figure to enhance the chances of winning contracts with the hope to recover the losses during project
implementation. As a result, researchers recommends lowest bid that is responsive in place of just any lowest bid for the award. Bid evaluation has been used to identify the most responsive lowest bid where clients go through all bid documents in a process called bid evaluation. This process of bid evaluation could be very tedious if bidders are many. This research develops a model that can identify the lowest responsive bid very
early among competitors without the need to go through tedious bid analysis. It is a further research consideration after Carr (2005)”s model. A set of 36engineering projects of diverse magnitude that went through competitive bidding process in Nigeria were obtained and reports on bid analyses collated. Extracted from the reports are the Consultant’s Estimate, The Bid Prices, Error Analysis and the Number of Bidders. Literature documents that these four factors influence significantly the lowest responsive bid in competitive
bidding. Using the four factors as independent variables and the lowest responsive bid as dependent
variable, four simple and three multiple regression models were generated and compared along Carr (2005)’s model. Findings show that the number of bidders and consultant’s estimate are best variables to predict the lowest bid if combined in a regression model. Apart from eliminating unrealistically low bids, the model abstracts the need for tedious bid analysis and reduce the time taken in bidding process. Furthermore, error was found not dependent on the magnitude of a project. Bidders should stick to ethics of estimating to reduce error in bids. Researchers should consider combining three and the four variables in future models to determine comparatively, the one that offers the best predictive power
