Augmenting macro panels with synthetic factor-model copies creates a factor-structured kernel ridge regressor that outperforms the Stock-Watson benchmark in point forecasts, with gains rising at longer horizons.
Title resolution pending
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
econ.EM 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
A global AR(1) plus block-specific local residual models raises full-panel out-of-sample R² by 0.047 on heterogeneous investment data, with gains replicated across US, UK/EU, and combined panels.
citing papers explorer
-
Double Descent and Benign Overfitting in Macroeconomic Forecasting
Augmenting macro panels with synthetic factor-model copies creates a factor-structured kernel ridge regressor that outperforms the Stock-Watson benchmark in point forecasts, with gains rising at longer horizons.
-
Global Persistence, Local Residual Structure: Forecasting Heterogeneous Investment Panels
A global AR(1) plus block-specific local residual models raises full-panel out-of-sample R² by 0.047 on heterogeneous investment data, with gains replicated across US, UK/EU, and combined panels.