Deconstructing the construction industry: A spatiotemporal clustering approach to profitability modeling

In spite of the strong influence of the construction industry on the national health of the United States' economy, very little research has specifically aimed at evaluating the key performance parameters and trends (KPPT) of the industry. Due to this knowledge gap, concerns have been constantly raised over lack of accurate measures of KPPT. To circumvent these challenges, this study investigates and models the macroeconomic KPPT of the industry through spatiotemporal clustering modeling. This study specifically aims to analyze the industry in 14 of its subsectors and subsequently, by 51 geographic spatial areas at a 15-year temporal scale. KPPT and their interdependence were firstly examined by utilizing the interpolated comprehensive U.S. economic census data. A hierarchical spatiotemporal clustering analysis was then performed to create predictive models that can reliably determine firm's profitability as a function of the key parameters. Lastly, the robustness of the predictive models was tested by a cross-validation technique called the predicted error sum of square. This study yields a notable conclusion that three key performance parameters - labor productivity, gross margin, and labor wages - have steadily improved over the study period from 1992 to 2007. This study also reveals that labor productivity is the most critical factor; the states and subsectors with the highest productivity are the most profitable. This study should be of value to decision-makers when plotting a roadmap for future growth and rendering a strategic business decisions.