Extending Parametric Portfolio Policies: Improving Performance through Macroeconomic Conditioning

Abstract

This thesis investigates whether incorporating macroeconomic information into parametric portfolio policies improves portfolio performance. Building on the framework of Brandt, Santa-Clara, and Valkanov (2009), the model allows firm characteristics to interact with macroeconomic indicators, generating time-varying portfolio weights that respond to shifts in the economic environment. Using monthly U.S. stock data from 1970 to 2022, the results show that conditioning on macroeconomic variables improves in-sample Sharpe ratios across four model specifications, each based on distinct macroeconomic themes. However, out-of-sample performance is more volatile, especially early in the sample when estimation risk is high. To address this, the thesis introduces an ℓ2-norm constraint on the characteristic-macro interaction coefficients (θ). This regularization technique limits the magnitude of parameter estimates, improving robustness and reducing turnover without substantially sacrificing in-sample performance. Applying regularization on the parameters in a parametric portfolio policy framework represents a key contribution of the thesis, as previous literature has applied the regularization on the final portfolio weights (DeMiguel et al., 2009).The constrained models often outperform their unconstrained counterparts out-of-sample, suggesting that regularization enhances the practical applicability of macro-conditioned strategies. The study also benchmarks the four model specifications against 495 alternative macro-variable combinations, finding that the chosen specifications are representative and not sample-specific outliers. Finally, the analysis shows that macroeconomic interactions help uncover conditional effects in traditional factors, such as size and value, that are not observable in static models. The findings support the role of macro conditioning in asset allocation and highlight the importance of balancing flexibility with robustness in dynamic portfolio design.