65 – Forecasting and Time Series Estimation
Improved Boosted Two Parameter Breitung Estimator: An Application to the Crude Oil Prices
Talha Omer
Jönköping University
Breitung and Roling (2015) presented a superiority of the nonparametric approach for estimation of mixed-frequency forecast. This approach remained outer performed than the usual parametric approach. In this paper, a boosted two parameter Breitung estimator by using the Lipovetsky and Conklin (2005) is proposed. Monte Carlo Simulation experiment suggests that boosted two parameters with an additional predicting parameter provides more reliable, and efficient approximation to the actual lag distribution than the one parameter Breitung nonparametric estimator. One parameter and two parameter Breitung estimators are applied to judge the predictive power of monthly Brent crude oil prices (low frequency variable) of the world on the bases of various daily stock market index (high frequency variable) indicators. In our real-time forecasting application, we find that the two parameter Breitung estimator performed well.