Prediction method of real discount rate to improve accuracy of life-cycle cost analysis

This study proposes a method to predict the real discount rate i for use in life-cycle cost analysis (LCCA) of buildings because i is compounded to adjust future value to present value. The method combines a multivariate time series and stochastic variability analysis. The value predicted using this method achieved higher prediction accuracy than when the past average value of i was used. The average value of past data has been generally applied to i in previous studies and actual work. Especially, the accuracy was improved when old data that differ greatly from recent data were excluded, and the stochastic variability analysis improved the prediction accuracy when the variability of the past data was large. The proposed method was applied to a case project to select exterior glass on the basis of LCCA. Initial investment over 10 years predicted by the presented method was about 5% higher than the average value of past data. The proposed method can contribute to accurate decision-making because the cost that corresponds to 5% of initial investment increases as the initial investment increases. The method may also be useful to estimate other factors such as electric and gas charges for LCCA.