Detail práce

GDPNow for the Czech Republic

Autor: Mgr. Jan Kutman
Rok: 2022 - zimní
Vedoucí: prof. PhDr. Tomáš Havránek Ph.D.
Typ práce: Diplomová
Finance, finanční trhy a bankovnictví
Jazyk: Anglicky
Stránky: 93
Ocenění: Ocenění DOT Award za vynikající magisterskou práci.
Abstrakt: The gross domestic product (GDP) is an essential measure of the state of economic activity
and serves as a crucial tool for policymakers, investors, or businesses. However, the official
GDP estimate in the Czech Republic is only available with a lag of approximately 60 days, and
the Czech National Bank (CNB) announces its GDP forecast once in each quarter. This thesis
focuses on predicting GDP growth in the current quarter, referred to as nowcasting. I employ
several methods to nowcast the real GDP growth in the Czech Republic in a pseudo-real-time
setting and compare their performance. Additionally, I investigate the possibility of creating
an ensemble model by using a weighted average of several nowcasting models. The results
suggest that the Dynamic Factor Model (DFM) performs best in the GDP nowcasting task, and
its predictive accuracy is comparable with the official CNB nowcast. Furthermore, the model
averaging process yields accuracy close to the best individual model while addressing model
uncertainty. The GDP nowcast of the DFM will be made available to the public in real-time on
a website and updated with a daily frequency.




Patria Finance
Česká Spořitelna