Stochastic Forecast of Construction Cost Index Using a Cointegrated Vector Autoregression Model

Abstract

The construction cost index (CCI) has been widely used to measure the cost trend in the construction industry. The index is used as an important input when estimating construction budgets and assessing risks in resource planning and cost management. To ensure accurate measurement, the properties of cost indexes should be investigated in their long- and short-run interactions with other variables, such as the consumer price index. This paper presents a cointegrated vector autoregression (VAR) model for forecasting the construction cost trend. This model has several advantages in terms of flexibility and dynamic interaction, and a comparison with existing methods demonstrates that the cointegrated VAR model can provide more accurate forecasts of the CCI. Practitioners can implement the cointegrated VAR forecasting technique using their own historical data. The index forecasts can not only provide more accurate estimation of construction budgets, but also evaluate the risk and uncertainty of cost escalation.