It Takes Two to Tango: Economic Theory and Model Uncertainty for Equity Premium Prediction

Abstract

We ask whether prominent theory-based proxies for \textit{the} expected market return fully capture it, or whether agnostic statistical methods recover information they miss. Economic restrictions—from present-value identities, the CAPM, and no-arbitrage—deliver significant out-of-sample predictability, and among statistical methods, complete subset regressions (CSR) perform best, comparably to the theoretically restricted models. Forecast-encompassing tests show that neither approach subsumes the other, suggesting the two carry complementary information about expected market returns. Combining them improves on either alone, and the equal weight works best because the ex-ante optimal weight is unstable and cannot be estimated reliably in real time; fixing it outperforms estimating it, and the equal-weight forecast delivers the best statistical performance and the highest risk-adjusted returns.

Download

• 📄 SSRN Preprint

Presentations

FMA Annual Meeting 2023, 4th Frontiers of Factor Investing Conference, 2023 QuantMinds International, 2024 Meeting of the Brazilian Finance Society, FinML 2024 at USI Lugano, 11th IAAE at the University of Turin.

Citation

Bianchi, Daniele, Alexandre Rubesam, and Andrea Tamoni. It Takes Two to Tango: Economic Theory and Model Uncertainty for Equity Premium Prediction. Working paper.

@article{bianchi2025takes,
  title={It takes two to tango: Economic theory and model uncertainty for equity premium prediction},
  author={Bianchi, Daniele and Rubesam, Alexandre and Tamoni, Andrea},
  journal={Available at SSRN 4513241},
  year={2025}
}