Performance attribution: decomposing portfolio returns into allocation and selection effects
A portfolio can beat its benchmark for many reasons. It may have been overweight in the asset classes that performed best. It may have selected superior securities within each class. Or it may have benefited from both simultaneously. Performance attribution disentangles these sources, making it possible to evaluate whether outperformance came from decisions that are repeatable or from exposure to market movements that happened to align with the portfolio. This discipline was formalised by Gary Brinson, Randolph Hood, and Gilbert Beebower in their 1986 paper, which found that asset allocation decisions explained more than 93 per cent of the variation in portfolio returns—a result that remains influential and contested in equal measure.
What performance attribution measures
Attribution analysis compares a portfolio's return against a benchmark by decomposing the difference into three components. The allocation effect measures whether the portfolio was overweight in categories that outperformed the benchmark and underweight in those that underperformed. The selection effect measures whether the securities chosen within each category outperformed the corresponding benchmark segment. The interaction effect captures the combined contribution of being overweight in a category and selecting well within it—or the penalty for being overweight in a category where selection was poor.
The Brinson-Hood-Beebower framework
Allocation effect = (wp − wb) × (Rb − R)
Selection effect = wb × (Rp − Rb)
Interaction effect = (wp − wb) × (Rp − Rb)
Where wp is the portfolio weight in a category, wb is the benchmark weight, Rp is the portfolio return within the category, Rb is the benchmark return within the category, and R is the total benchmark return. Summing the three effects across all categories reconciles to the total active return.
How to interpret attribution results
A positive allocation effect in a category means the portfolio was overweight there and that category beat the overall benchmark. A positive selection effect means the securities chosen within that category outperformed the category benchmark return. Attribution results only become meaningful across multiple periods and market regimes. A single period's attribution is largely noise; a persistent pattern—consistent positive selection in one category, for example—is a signal worth investigating. Attribution that shows positive allocation but negative selection, or vice versa, suggests the manager's skill is concentrated in one dimension. The interaction term, often omitted in simplified attribution reports, can be significant in portfolios with large active tilts.
Rolling performance attribution
Rolling attribution, computed over 12- or 36-month windows, reveals whether value-add is regime-dependent. A strategy that generates positive selection in rising markets but negative selection in falling markets is demonstrating a hidden beta, not pure skill. Stable attribution across regimes is a stronger indicator of genuine edge. In practice, most attribution software computes this on a monthly basis and aggregates geometrically to avoid errors from compounding arithmetic effects.
Limitations
Attribution analysis requires a clear benchmark and a consistent category structure. Portfolios that blend asset classes in unconventional ways—mixing equities with alternatives, for example—can be difficult to attribute cleanly. The choice of benchmark significantly affects the results: a manager benchmarked against the MSCI World will show different allocation effects than if benchmarked against the MSCI ACWI. Attribution also says nothing about whether the active decisions were deliberate or accidental; a lucky overweight in a winning category scores the same as a considered one. Finally, the Brinson model is designed for single-period, long-only portfolios. Extending it to leveraged, short-selling, or derivatives-based strategies requires more complex frameworks.
Performance attribution in pfolio
Formal performance attribution is not currently available in pfolio Insights. Asset class-level return contribution is visible in the portfolio breakdown, which provides a partial view of how allocation decisions have affected performance relative to the benchmark.
Related metrics
Disclaimer
Get started now
