A0964
Title: Advancing forecast accuracy analysis: A partial linear instrumental variable and double machine learning approach
Authors: Christoph Schult - Halle Institute for Economic Research (Germany) [presenting]
Katja Heinisch - Halle Institute for Economic Research (Germany)
Fabio Scaramello - University of Venice (Germany)
Abstract: The relationship between forecast accuracy and forecast assumptions is explored using German data and a novel empirical approach. Partial Linear Instrumental Variable (PLIV) regression models are employed, combined with Double Machine Learning (DML) methods, to address high-dimensional nuisance parameters and endogeneity issues. This innovative PLIV-DML framework enables a more complex understanding of the relationships between forecast assumptions and forecasts accuracy than traditional OLS-based analysis. The evaluation sample includes 1460 annual GDP forecasts and various oil, exchange rate, and world trade assumptions. The PLIV-DML model's inherent flexibility allows us to examine two possible violations of assumptions: the rationality of forecasters and the linearity of the data generation process. This research contributes to the field of forecasting by providing a more robust and flexible analysis of forecast accuracy determinants. For instance, the proposed method contributes to the discussion regarding weak instruments and instrumental variables' validity in macroeconomic models. Evidence of a constant underestimation of OLS estimators of the impacts of squared assumption errors of oil price and world trade on squared forecast errors of GDP is found. The insights gained from this study have potential implications for improving economic forecasts' accuracy and understanding underlying forecasting processes.
