Home / Wiki / P / Performance Attribution

Performance Attribution

Investment and Capital MarketsUpdated 16 Nov 2026

Performance attribution is a set of methods for explaining why a portfolio's return differed from that of its benchmark. It divides the active return, the portfolio return minus the benchmark return, into parts linked to specific decisions: how much was placed in each sector, asset class or country, which securities were chosen within them, and sometimes currency, duration or factor exposures. Each part is called an attribution effect, and together they add up to the total difference.

Attribution is used by asset managers, pension funds, consultants and wealth platforms to judge the skill behind a track record and to report results to clients. It complements return measures such as alpha and the information ratio, which say how much value was added, by showing where it came from. The best-known framework for equity portfolios is the Brinson model, set out in work by Gary Brinson and co-authors in the mid-1980s.

01

The Brinson model

For each segment ii of the portfolio, such as a sector, let wp,iw_{p,i} and wb,iw_{b,i} be the portfolio and benchmark weights, Rp,iR_{p,i} and Rb,iR_{b,i} the portfolio and benchmark returns in that segment, and RbR_b the total benchmark return. In the widely used Brinson-Fachler form, the allocation effect is:

Ai=(wp,i−wb,i)×(Rb,i−Rb)A_i = \left( w_{p,i} - w_{b,i} \right) \times \left( R_{b,i} - R_b \right)

The selection effect is:

Si=wb,i×(Rp,i−Rb,i)S_i = w_{b,i} \times \left( R_{p,i} - R_{b,i} \right)

The interaction effect, which captures the combined result of overweighting a segment and also outperforming within it, is:

Ii=(wp,i−wb,i)×(Rp,i−Rb,i)I_i = \left( w_{p,i} - w_{b,i} \right) \times \left( R_{p,i} - R_{b,i} \right)

Summed over all segments, A+S+IA + S + I equals the active return Rp−RbR_p - R_b. Allocation rewards overweighting segments that beat the benchmark as a whole, selection rewards picking securities that beat their own segment, and some providers fold interaction into selection to simplify reporting.

02

Worked example

A benchmark holds 60% equities returning 10% and 40% bonds returning 2%, for a total return of 0.6×10+0.4×2=6.8%0.6 \times 10 + 0.4 \times 2 = 6.8\%. The portfolio holds 70% equities returning 12% and 30% bonds returning 3%, for 0.7×12+0.3×3=9.3%0.7 \times 12 + 0.3 \times 3 = 9.3\%. The active return is 2.5 percentage points.

  • Allocation. Equities: 0.10×(10−6.8)=0.320.10 \times (10 - 6.8) = 0.32. Bonds: −0.10×(2−6.8)=0.48-0.10 \times (2 - 6.8) = 0.48. Total 0.80 points.
  • Selection. Equities: 0.6×(12−10)=1.200.6 \times (12 - 10) = 1.20. Bonds: 0.4×(3−2)=0.400.4 \times (3 - 2) = 0.40. Total 1.60 points.
  • Interaction. Equities: 0.10×2=0.200.10 \times 2 = 0.20. Bonds: −0.10×1=−0.10-0.10 \times 1 = -0.10. Total 0.10 points.

The three effects sum to 0.80+1.60+0.10=2.500.80 + 1.60 + 0.10 = 2.50, matching the active return. Most of the outperformance came from choosing better securities, and about a third from the tilt towards equities, an asset allocation decision.

03

Other approaches

Several extensions deal with features the basic model ignores. Fixed income attribution separates returns due to the level and shape of the yield curve, changes in credit spreads, carry and security selection, because sector weights alone describe bond portfolios poorly. Currency attribution isolates the effect of exchange-rate movements and hedging decisions in international portfolios. Factor-based attribution uses a factor model to explain returns by exposures to style factors such as value, size and momentum, which helps show whether apparent stock-picking was really a factor bet.

Results for single periods must be linked to report longer periods, since the effects of successive months do not simply add up when returns compound. Providers use smoothing or geometric methods to make multi-period figures reconcile. Attribution can also be run on risk rather than return, decomposing tracking error into the contributions of each position or factor.

04

Uses and limitations

For a portfolio manager practising active management, attribution shows whether results match the stated process: a stock picker should earn most of the excess through selection, while a top-down allocator should show it in allocation. It helps clients and committees ask whether results reflect skill or luck, and helps managers spot unintended exposures. Consultants use it when selecting and monitoring external managers, and many fund platforms include it in standard client reports. The Global Investment Performance Standards issued by the CFA Institute govern how firms calculate and present their returns, but they do not prescribe any single attribution method, so firms choose and disclose their own.

The results depend on choices made by the analyst, including the benchmark, the way segments are defined, the formula variant and the handling of cash flows and transaction costs. The same portfolio can show quite different effects under a sector breakdown and a country breakdown. Attribution is usually calculated on time-weighted returns, which remove the effect of client deposits and withdrawals, and is not directly comparable to a money-weighted return. Finally, attribution is descriptive: it explains a past outcome and does not show whether the same decisions will add value in future.

Last updated 16 Nov 2026