Mixed Frequency Data Dynamic Models in Macroeconomic Forecasting

Journal Title: Zarządzanie i Finanse - Year 2015, Vol 13, Issue 4

Abstract

The aim of the study was to find out whether the use of mixed frequency data dynamic models can improve the accuracy of quarterly forecasts of selected macroeconomic variables. The research tools were the MIDAS, AR-MIDAS, DFM, ARIMA and VAR models with quarterly forecasts as reference points. The study used a simulation of eight (ex ante) forecasting sessions. To that end, it was necessary to use variables from the real-time database. Yet, due to the limited access to such databases, the study focused on U.S. GDP. The results indicate that the AR-MIDAS class model, which directly incorporates the latest available monthly information as well as „classic” VAR model provided significantly more accurate forecasts of GDP then DFM and ARIMA models. Comparing AR‑MIDAS and VAR forecasts (which were equally accurate in terms of Diebold‑Mariano test) only the previous model proved to be able to provide accurate forecasts of all turning points which occurred within the sample. This unique feature enables to recommend AR-MIDAS as a now‑casting or short term forecasting tool.

Authors and Affiliations

Lech Kujawski

Keywords

Related Articles

 Fiscal meaning and execution efficiency of Customs Service in Poland

 Knowledge about fiscal efficiency of Customs Service is very important in a view of significant scale of public levies collected by this institution. Such knowledge can be particularly useful for service efficiency...

Finansowanie kierunków zamawianych ze środków europejskich jako czynnik rozwoju polskich przedsiębiorstw z sektora MSP

Celem artykułu jest analiza znaczenia finansowania kierunków zamawianych ze źródeł zewnętrznych dla pobudzenia rozwoju polskiej gospodarki opartej w znacznej mierze na sektorze MSP. W opracowaniu syntetycznie zostaje prz...

 The analysis of the diversity of financing the long term-care within the health insurance fund in Poland

 The article presents the essence and diversity of financing the long-term care services in Poland along with a detailed financial analysis based on contracts with National Health Fund (NHF). The system structure of...

Bariery szerszego wykorzystania instytucji upadłości konsumenckiej w Polsce 

 W artykule przedstawiono założenia teoretyczne i koncepcję przyjętej w Polsce procedury upadłościowej. Problemem badawczym podjętym w artykule była analiza przyczyn niskiej skali wykorzystywania instytucji upadłośc...

 Speculation on the land market in the context of its structure on the example of Silesia Province in the years 2003–2012

 This paper presents results of a study of the market structure of corporate land acquired in Silesia Province in the years 2003–2012. The study was based on direct and comprehensive market data taken from notary...

Download PDF file
  • EP ID EP95963
  • DOI -
  • Views 81
  • Downloads 0

How To Cite

Lech Kujawski (2015). Mixed Frequency Data Dynamic Models in Macroeconomic Forecasting. Zarządzanie i Finanse, 13(4), 209-228. https://www.europub.co.uk/articles/-A-95963