Boosting Non-linear Predictability of Macroeconomic Time Series

Arbeitspapier

We apply the boosting estimation method to investigate to what ex-tent and at what horizons macroeconomic time series have nonlinearpredictability coming from their own history. Our results indicate thatthe U.S. macroeconomic time series have more exploitable nonlinearpredictability than previous studies have found. On average, the mostfavorable out-of-sample performance is obtained by a two-stage proce-dure, where a conventional linear prediction model is fine-tuned by theboosting technique.

Object Details

• Language: Englisch

References and Relationships

• Bibliographic citation: Series: Discussion paper ; No. 124

Classification and Topics

• Classification: Wirtschaft ; Single Equation Models; Single Variables: Time-Series Models; Dynamic Quantile Regressions; Dynamic Treatment Effect Models; Diffusion Processes ; Forecasting Models; Simulation Methods ; Macroeconomics: Consumption, Saving, Production, Employment, and Investment: Forecasting and Simulation: Models and Applications ; Prices, Business Fluctuations, and Cycles: Forecasting and Simulation: Models and Applications ; Money and Interest Rates: Forecasting and Simulation: Models and Applications
• Subject: boosting ; forecasting ; linear autoregression ; mean squarederror ; non-linearity

Contributors, Places and Time

• Event: Geistige Schöpfung
• (who): Kauppi, Heikki ; Virtanen, Timo
• Event: Veröffentlichung
• (who): Aboa Centre for Economics (ACE)
• (where): Turku
• (when): 2018

Further information

• Handle: http://hdl.handle.net/10419/233339
• Last update: 10.03.2025, 11:42 AM CET