Forecasting economic growth by combining local linear and standard approaches

Area of Science:

• Economics
• Econometrics
• Time Series Analysis

Background:

• Developing economies are crucial for global growth.
• Macroeconomic time series forecasting is hindered by data limitations, volatility, and non-linear trends.
• Existing methods often struggle with the complexities of developing economies' economic data.

Purpose of the Study:

• To propose an improved forecasting method for macroeconomic growth data.
• To address challenges in forecasting time series for developing economies.
• To enhance the accuracy and reliability of GDP growth predictions.

Main Methods:

• Data-driven local linear trend estimation with an extended iterative plug-in algorithm for endogenous bandwidth selection.
• Extension of the random walk model to incorporate a local linear, time-varying drift.
• Application to GDP data from six developing and two advanced economies, comparing forecast combinations.

Main Results:

• The proposed local linear trend estimation method provides smooth trend estimations, adapting to temporary changes.
• The extended random walk model with local linear drift improves forecasting.
• Forecast combinations including the local linear approach demonstrated enhanced accuracy and reduced variance compared to traditional methods.

Conclusions:

• The developed forecasting method offers a more sophisticated approach for macroeconomic time series analysis, particularly for developing economies.
• Improved trend estimation and model extensions contribute to more reliable GDP growth forecasts.
• The findings suggest that advanced forecasting techniques are essential for navigating the complexities of developing economies' economic landscapes.