A1773
Title: Forecasting with Q: Density forecasts with local quantile projection and quantile vector autoregression
Authors: Johannes Bleher - University of Hohenheim (Germany)
Sophia Koch - University of Hohenheim (Germany) [presenting]
Thomas Dimpfl - University of Hohenheim (Germany)
Abstract: The challenge of estimating and forecasting conditional densities in high-dimensional time series data, particularly in economic and financial contexts, is addressed. Traditional linear regression models often fall short in describing complex conditional distributions. To overcome these limitations, a robust method based on smoothed quantile regression is introduced which avoids restrictive assumptions about the data-generating process. Quantile regression offers a flexible framework that can capture skewed, heteroskedastic, multimodal, and heavy-tailed conditional distributions. The approach is demonstrated through a simulation study and applied to forecast the conditional density of inflation, considering variables such as GDP growth, industrial production growth, and the unemployment rate.
