Predictive Density Aggregation: A Model for Global GDP Growth
IMF Working Papers, May 29, 2020
Source details
- Canonical URL
- Predictive Density Aggregation: A Model for Global GDP Growth
Other formats
Bibliographic details
- Authors: Francesca Caselli, Francesco Grigoli, Romain Lafarguette, Changchun Wang
- Published: May 29, 2020
- Series: IMF Working Papers
- DOI: https://doi.org/10.5089/9781513545653.001
Methodological approach
- Proposes a bottom-up probabilistic model to obtain the predictive density of global GDP growth.
- Estimates and combines single countries’ predictive GDP growth densities while taking into account cross-country interdependencies.
- Each economy’s predictive density depends on a set of observable country-specific factors.
- Uses sampling techniques to aggregate individual countries’ densities into a world aggregate while preserving the non-i.i.d. nature of the global GDP growth distribution.
Modeling contemporaneous interdependencies
- Models non-parametrically the contemporaneous interdependencies across the United States, the euro area, and China via a conditional kernel density estimation of a joint distribution.
- Characterizes potential amplification effects stemming from other large economies in each region using kernel density estimations.
- Models the reaction of all other economies with parametric assumptions.
Estimation and aggregation
- Combines non-parametric kernel density estimations for key large economies with parametric assumptions for other economies.
- Aggregation is performed via sampling techniques to preserve cross-country dependence and the non-i.i.d. global distribution.
Evaluation and performance
- Out-of-sample metrics confirm the accuracy of the approach.
Content in this bundle
- Working Paper