Projects

Bootstrap Based Empirical Joint Prediction Regions for Path Forecasts

Ing. Marián Vávra, PhD.

Members of project
doc. Ing. Tomáš Domonkos, PhD., Ing. Filip Ostrihoň, PhD.

• Basic informations
• Characteristics

• Duration of the project: 01/2021 - 12/2023
  ID number: VEGA 1/0476/21

   

• Due to uncertainty surrounding point forecasts, many central banks regularly calculate and officially publish prediction intervals for key economic variables in order to express and communicate perceived forecast risks with professionals and the general public. The Gaussian-like prediction bands have become a workhorse in the banking industry. However, empirical evidence suggests that their accuracy is very poor. The project aims to modify the recently proposed joint prediction bands using a block bootstrap method in order to achieve empirically reliable results.