With the rapid growth of factor investing, a key question remains: Can active factor allocation add value over a diversified passive factor allocation?

The authors explore forecast-based timing of factor allocation, distinguishing between factor timing (using time-series forecasts) and factor tilting (using cross-sectional factor information to allocate to factors).

The authors apply two parametric portfolio policy (PPP) frameworks, which produce factor-timing and factor-tilting distributions and are designed to avoid overlap with each other. They construct equity factors from a representative set of 20 investable equity factor portfolios covering Value, Momentum, Quality, and Size.

The authors test whether factor timing can improve the information ratio of a multi-factor portfolio beyond an equally weighted factor portfolio. They identify buy and sell signals for each factor by using fundamental economic variables—such as dividend yields, earnings-to-price ratios, and inflation—and technical indicators, such as momentum and moving averages.

Next, they test factor tilting the same way. Factor tilt is established by assessing cross-sectional factor characteristics, such as relative factor valuation, spread, crowding, momentum, and volatility.

To compare an equally weighted factor strategy with the high turnover and correspondingly high rebalancing costs of factor timing and tilting, the authors account for the associated transaction costs and assess the impact on returns of these costs.

Returns from factor timing are statistically significant compared with those generated by investing in a passive, equally weighted equity factor portfolio.

The factor-timing strategy delivers a gross return of 4.17% a year over the 20-year period. This is 95 bps higher than the passive, equally weighted benchmark, which returned 3.22% a year.

A factor-tilting strategy improves risk-adjusted returns only when short-term factor momentum is used.

However, the high portfolio turnover and the consequent high transaction costs erode much of the potential return of both timing and tilting, putting into question the applicability of these approaches in a real-life investment context. Smoothing the factor rebalancing by imposing constraints on weights, Black–Litterman shrinkage (which reduces the deviation of the portfolio from the market equilibrium), and transaction cost penalties is found to reduce turnover and costs, but only factor timing remains of value net of costs. Factor tilting proves to be of little material value after the increased costs are taken into account.

The findings provide reason for investors to be cautious about dynamic equity factor allocation compared with the simpler strategy of constructing a passive multi-factor portfolio.

The diversification inherent in the passive multi-factor index would appear to offer better value, given that transaction costs are far lower than for the dynamic approach.

Nevertheless, the authors argue that the gross alpha created by factor timing and tilting means their methodology is worth further exploration, given the expectation that the innovative process could be further refined to reduce transaction costs and increase net alpha.