Prediction of Final Tender Sums of Construction Projects from the Design-Stage Elemental Cost Plans: A Decision Support Tool for New Zealand

Abstract

Reliable prediction of final tender sums (initial contract sums) of building projects from the cost plans is of great importance to construction clients and practitioners to prevent unpleasant consequences of cost overruns, disputes and project abandonment. Various approaches have been adopted in tender price forecasting, ranging from statistical/mathematical to the use of artificial intelligence techniques. Most of these approaches employed various elements of tender price equation to model their forecasts. However, despite these research efforts, a model that could improve the accuracy of final tender sum predictions is yet to be developed for New Zealand (NZ). Despite the aspirations of NZ to have one of the best construction industries in the world, it is still inefficient. This is especially regarding contract management, as characterised by poor labour productivity, health and safety issues and unpredictability in project delivery within budget and on time. Of these challenges, the cost aspect tends to be the most important, owing to its direct economic impact. Armed with this information, an accurate forecast of final tender sum has been a difficult task due to uncertainties and hence risks inherent in construction projects. Knowledge of how these risk factors combine to impact on budget overruns is the concern of this current study. As such, significant deviations are observable between the predicted (ECPs) and actual sums (FTS). This current assessment therefore attempts to develop a trial model on the variation between the design-stage ECP and FTS due to the inherent risks in construction projects. The aim of the model is to provide a decision support tool that could assist NZ construction industry practitioners to have a better and reliable prediction of final tender sums from the cost plans. The theoretical exploration of key risk factors inherent in construction projects is firstly identified through a literature review and then validated with construction practitioners, which resulted in 16 potential risk factors being identified. Based on the identified factors, a quantitative and a qualitative methodology (mixed methods research) was employed, first with 32 consultants (Architects, QS and PMs), and second, with 5 consultant QS participating in the questionnaire- and interview-based surveys respectively. A comparative analysis of survey results emanating from 208 consultants’ perspectives, revealed the top ten risk factors requiring focus for predictive modelling. Risk matrix technique prioritized 9 from the top 10 risk factors obtained by using risk mean analysis, degree of risk measure, coefficient of variation, Spearman’s coefficient of correlation and Kendall’s coefficient of concordance analyses. These risk factors are (1) client’s change/change in an owner’s requirements; (2) complexity of design and construction; (3) quality of information and flow requirements; (4) availability of design information; (5) expertise of consultants; (6) market condition; (7) project team’s experience of the construction type; (8) site investigation; and (9) inadequate tender documentation. The resulting risk impact assessment model was then developed by using multiple linear regression (MLR), based on the archival cost data (ECPs and FTS) from 62 case study projects, the dominant risk factors and their relative impacts of occurrence. The predictive model (trial) developed was validated using 12 case study projects to determine its accuracy and prediction performance, as well as the practical relevance to the NZ construction clients and consultants. The multiple linear regression model developed was promising in that it helped to establish that the phenomenon under consideration could be modelled. It also provided some insights in explaining the observed variability between the design-stage ECPs and FTS on risk impacts. This research study is first to develop an assessment model that could assist the construction industry practitioners in NZ to have a better and reliable final tender sum predictions from cost plans. The primary contribution is to provide quantitative confirmation of the more general statements made in the literature from around the world. This therefore adds to and consolidates existing knowledge. Moreover, as risk elements are inherent in construction project developments globally, these findings have important ramifications for all construction projects in expanding and clarifying existing knowledge on what is needed for reasonable budgetary performance and successful delivery of construction projects.