Systematic stock selection with macro factors

Motivation

Conventional equity factor models rely primarily on firm-level data: price momentum, book-to-market ratios, earnings quality, and related metrics. While these factors capture important cross-sectional variation in expected returns, they ignore the macroeconomic context in which firms operate. Two companies with identical firm-level characteristics may have very different expected returns depending on the macro environment of their home country.

We propose a framework that augments standard equity factors with country-level macroeconomic indicators, using the JPMaQS point-in-time dataset to ensure all information was available to investors at the time of portfolio formation.

Framework design

Our approach operates in two stages. First, we estimate country-level macro scores using growth expectations, inflation trends, external balance positions, and labor market indicators. Second, we interact these macro scores with firm-level factors to create conditional stock selection signals.

The key insight is that certain equity factors work better in specific macro regimes. Value stocks tend to outperform when growth expectations are rising from depressed levels, while quality factors dominate during periods of macro deterioration. By conditioning on the macro environment, we can dynamically allocate across factor exposures.

Empirical evidence

Backtesting across 23 developed and emerging equity markets over a 15-year period, the macro-conditioned model delivers a long-short Sharpe ratio of 0.85, compared to 0.52 for the unconditional factor model. Turnover is moderate, and the improvement is robust to different rebalancing frequencies and transaction cost assumptions.

Importantly, the macro conditioning reduces the severity of factor drawdowns. The maximum drawdown of the conditional model is 12%, versus 22% for the unconditional benchmark. This improvement arises because the model reduces exposure to factors that are vulnerable in the prevailing macro regime.

Implementation notes

We discuss practical considerations for implementation, including the treatment of missing macro data, the choice of estimation windows, and the appropriate level of position concentration. The framework is designed to be modular, allowing portfolio managers to incorporate macro conditioning into their existing factor models without a wholesale redesign of their investment process.