Forecasting Model of Construction Cost Index Based on VECM with Search Query

Abstract

The Construction Cost Index (CCI), published by the Engineering News-Record (ENR), is a cumulative index of construction labor and material costs. Recent CCI forecasting studies, based on multivariate time series, have conducted forecasts using economic and social indices as variables, with better forecast results than conventional univariate time series approaches. However, a problem with these economic and social index variables is that, because of the time required for collection and analysis of data, up-to-date data is not available at the time of forecast. To overcome the limitation of conventional CCI forecasting using multivariate time series analyses, specifically the limitation that the latest data cannot be used owing to the time required till their availability, and to enhance the accuracy of the forecasting model, this study proposes a CCI forecasting model based on the Vector Error Correction Model (VECM) with search query frequencies. A VECM-based forecasting model using the ‘project manager salary’ query was selected as the final model. This CCI forecasting model showed better predictive ability than a cointegrated vector autoregression model using the United States consumer price index, and it was confirmed that the advantages of query frequency over conventional economic indices could prove helpful for forecasting purposes.