This paper by Giacomini and White (2003) proposes an alternative framework for testing conditional predictive ability in forecasting, which is more relevant for real-time forecast selection. The authors argue that the current framework, as proposed by West (1996), focuses on unconditional expectations and does not account for important determinants of forecast performance, such as estimation uncertainty and the choice of forecasting method. Their approach evaluates the forecasting method, which includes the model, estimation procedure, and estimation window, rather than just the model itself. This allows for a more realistic assessment of forecast performance, especially in heterogeneous data environments. The paper also discusses the practical advantages of their tests, including their applicability to nested and non-nested models, and their ease of computation using standard regression packages. The authors illustrate the usefulness of their conditional predictive ability tests through an empirical application comparing three leading parameter-reduction methods for macroeconomic forecasting.This paper by Giacomini and White (2003) proposes an alternative framework for testing conditional predictive ability in forecasting, which is more relevant for real-time forecast selection. The authors argue that the current framework, as proposed by West (1996), focuses on unconditional expectations and does not account for important determinants of forecast performance, such as estimation uncertainty and the choice of forecasting method. Their approach evaluates the forecasting method, which includes the model, estimation procedure, and estimation window, rather than just the model itself. This allows for a more realistic assessment of forecast performance, especially in heterogeneous data environments. The paper also discusses the practical advantages of their tests, including their applicability to nested and non-nested models, and their ease of computation using standard regression packages. The authors illustrate the usefulness of their conditional predictive ability tests through an empirical application comparing three leading parameter-reduction methods for macroeconomic forecasting.