Attribution

Why attribution matters

A return number shows what happened but not why. Attribution connects outcomes to investment decisions, allowing clients and managers to test whether results match the stated process. It helps identify repeatable strengths, unintended exposures, implementation drag, and responsibility across teams. It also distinguishes a correct thesis from a favorable market move and supports better decisions about manager retention, risk budgets, and portfolio construction.

How it is applied

Brinson-style attribution compares portfolio and benchmark weights and returns to estimate allocation, selection, and interaction effects. Factor attribution explains active return through modeled sensitivities, while transaction-based analysis can isolate trading and currency. The method must match the mandate and reconcile with reported performance. Analysts define classification hierarchy, benchmark, timing, cash treatment, derivatives exposure, fees, and arithmetic or geometric linking before interpreting contributions across periods.

Formula

Portfolio example

An equity manager overweights technology, which outperforms the benchmark, and selects technology shares that lag their sector. Attribution may show positive allocation from the overweight and negative selection within technology. Total active return can still be positive. A factor analysis might further show that the allocation result largely reflects momentum exposure, giving the investor a different view of how distinctive the decision really was.

How to interpret it

Positive contribution means a defined decision or exposure added return under the chosen framework; it does not necessarily mean the decision was skillful. Interaction terms and classification choices can shift credit between allocation and selection. Results should reconcile to total active return and be examined over a horizon suited to the strategy. Persistent contributions consistent with documented decisions are more informative than isolated quarterly effects.

Limitations and common misconceptions

Attribution is a model, not a unique truth. Benchmarks, sector mapping, factor definitions, price timing, derivative treatment, and linking methods change the answer. Holdings snapshots can miss intra-period trading, while transaction detail can be expensive and noisy. Contributions do not show risk taken or losses avoided. Qualitative review, risk attribution, costs, and confidence in data are needed before using the output to judge people or processes.