Empirical Investigation of Temporal Association between Architecture Billings Index and Construction Spending Using Time-Series Methods

Abstract

Construction spending (CS) represents the level of construction activities in the United States. It has been widely used for evaluating and tracking the construction market. Evaluation of the construction market provides valuable information for construction firms to plan resources and make strategic business decisions. Architecture Billings Index (ABI) has been introduced as a potential leading indicator of construction spending. Correlation analysis and regression modeling have been used in the literature to conclude that ABI leads nonresidential construction spending up to 1 year. This paper argues that these methods are not the right methods for studying the temporal relationship between these two time series because they ignore autocorrelation among the values of each time series. In other words, the autocorrelation may result in spurious correlation between variables simply by coincidence. This is further confirmed in this paper showing that neither ABI nor construction spending are stationary, and therefore, correlation analysis and regression modeling could lead to wrong interpretations. The research objective is to empirically analyze the temporal association between ABI and construction spending using time-series methods. Unlike the findings indicated in the literature, the results of this research show that institutional ABI leads nonresidential construction spending only in higher lag lengths (12 months and longer). There was no evidence that commercial/industrial ABI leads nonresidential construction spending. The results also show that residential ABI leads residential construction spending up to approximately 3 years. Moreover, it is found that institutional and residential ABIs have long-term temporal relationships with nonresidential and residential construction spending variables, respectively. These results are validated through the creation of appropriate time-series models that are capable of identifying future trends in construction spending. The findings of this paper contribute to the state of knowledge by identifying and quantifying the temporal relationships between ABI and construction spending. It is expected that these findings will help construction firms identify future trends in construction activities. The identified trend in construction spending provides useful information to construction firms in making better strategic business decisions in the volatile construction market.