Brinson Attribution Model: Allocation, Selection & Interaction

Published: March 18, 2026
Ryan O'Connell, CFA, FRM
Article by Ryan O'Connell, CFA, FRM

The Brinson attribution model is the most widely used framework in institutional investment management for answering the question every portfolio review demands: why did a portfolio outperform or underperform its benchmark? While metrics like Jensen’s alpha quantify the total value a manager added, they do not reveal whether that value came from smart sector bets, superior stock picks, or the interaction of the two. Brinson attribution decomposes the portfolio’s active return — the difference between the portfolio return and its benchmark return, typically measured using time-weighted returns — into three distinct sources.

It is important to distinguish performance attribution from return contribution. Attribution explains why active return differs from zero — what decisions drove the deviation from the benchmark. Contribution analysis, by contrast, explains what each holding added to the portfolio’s total return regardless of benchmark. The Brinson model is strictly an attribution framework. Note also that the Brinson attribution framework (decomposing active return into effects) is separate from the famous Brinson, Hood & Beebower (1986) study that found asset allocation explains the majority of return variation across time — related but distinct concepts.

What Is the Brinson Attribution Model?

Key Concept

The Brinson attribution model decomposes a portfolio’s active return (portfolio return minus benchmark return) into three effects: allocation (did the manager overweight sectors that outperformed?), selection (did the manager pick better securities within each sector?), and interaction (did the manager overweight sectors where stock picks also outperformed?). Together, these three effects fully explain the portfolio’s active return for a single evaluation period.

Two primary variants of the model exist. Brinson-Fachler (BF, 1985) measures the allocation effect relative to the overall benchmark return, making it more intuitive for most practitioners. Brinson-Hood-Beebower (BHB, 1986) uses a different allocation baseline. This article presents the BF three-component formulas as the primary framework — consistent with our Performance Attribution Calculator — and explains how BHB differs in the comparison section below.

Prerequisites & Assumptions

The Brinson model requires: beginning-of-period portfolio and benchmark weights, arithmetic single-period returns, the same currency and classification scheme for both portfolio and benchmark, and categories that are mutually exclusive and collectively exhaustive (every security belongs to exactly one category, and all categories together account for 100% of the portfolio and benchmark).

Performance attribution operates at two levels. Macro attribution works at the fund sponsor level — decomposing returns across policy decisions, asset class timing, and manager selection. Micro attribution works at the individual manager level — decomposing returns into sector weighting and stock selection decisions. The Brinson model is a micro attribution framework, answering how a single manager added or detracted value relative to their assigned benchmark.

The standard Brinson model is holdings-based: it uses beginning-of-period holdings and assumes a buy-and-hold posture throughout the period. This means it does not capture the impact of intra-period trades. Transaction-based attribution explicitly accounts for trades during the period and significantly reduces the reconciliation gap between attributed effects and the portfolio’s actual time-weighted return, but requires complete trade-level data. For most institutional reporting, holdings-based Brinson attribution is the standard starting point.

The Brinson-Fachler Attribution Formulas

The Brinson-Fachler model decomposes active return into three additive effects. In each formula, j indexes the sectors (or asset categories), w denotes weight, R denotes return, and subscripts p and B refer to the portfolio and benchmark respectively.

Allocation Effect (BF)
Allocation = Σ(wp,j − wB,j) × (RB,j − RB)
The return impact of overweighting or underweighting sectors that outperformed or underperformed the overall benchmark return
Selection Effect
Selection = Σ wB,j × (Rp,j − RB,j)
The return impact of picking securities that outperformed or underperformed their sector benchmark, isolated at benchmark weights
Interaction Effect
Interaction = Σ(wp,j − wB,j) × (Rp,j − RB,j)
The joint effect of overweighting sectors where the manager also selected outperforming (or underperforming) securities
Active Return Identity
Rp − RB = Allocation + Selection + Interaction
The three effects sum exactly to the portfolio’s total active return in the single-period arithmetic case
Pro Tip

The three effects sum exactly to active return only in the single-period framework. For multi-period analysis (e.g., linking monthly effects to a quarterly total), you need a linking method such as Carino, Menchero, or GRAP to ensure that individual-period effects compound correctly to the cumulative active return.

Brinson Attribution Example

BF Attribution: US Large-Cap Equity Portfolio

Consider a US large-cap equity manager whose portfolio is benchmarked against a broad market index such as the Russell 1000. The manager overweights Technology and Health Care — a common tilt among active managers like T. Rowe Price and Fidelity Contrafund who have historically generated selection alpha in those sectors. The portfolio and benchmark are classified into five GICS sectors (plus an “Other” category to ensure exhaustive coverage) for a one-month evaluation:

Sector wp (%) wB (%) Rp (%) RB,j (%)
Technology 32.00 28.00 3.20 2.50
Health Care 18.00 15.00 1.80 1.00
Financials 10.00 13.00 0.50 0.80
Consumer Staples 15.00 20.00 1.20 1.50
Industrials 25.00 24.00 −0.40 −0.20
Other 0.00 0.00 — —
Total 100.00 100.00

Portfolio return (Rp): 0.32 × 3.20 + 0.18 × 1.80 + 0.10 × 0.50 + 0.15 × 1.20 + 0.25 × (−0.40) = 1.478%

Benchmark return (RB): 0.28 × 2.50 + 0.15 × 1.00 + 0.13 × 0.80 + 0.20 × 1.50 + 0.24 × (−0.20) = 1.206%

Active return: 1.478 − 1.206 = 0.272% (27.2 basis points)

Sector Allocation (%) Selection (%) Interaction (%) Total (%)
Technology +0.052 +0.196 +0.028 +0.276
Health Care −0.006 +0.120 +0.024 +0.138
Financials +0.012 −0.039 +0.009 −0.018
Consumer Staples −0.015 −0.060 +0.015 −0.060
Industrials −0.014 −0.048 −0.002 −0.064
Total +0.029 +0.169 +0.074 +0.272

The portfolio’s 27.2 basis points of active return was driven primarily by security selection (+16.9 bps), particularly in Technology (+19.6 bps) and Health Care (+12.0 bps) where the manager picked outperforming stocks. The interaction effect (+7.4 bps) was positive because the manager overweighted sectors where stock picks also outperformed. The allocation effect was modest (+2.9 bps) — the benefit of overweighting Technology (which outperformed the overall benchmark) was partially offset by underweighting Consumer Staples, which also outperformed.

Notice also that the Financials allocation is positive (+0.012%) even though the manager underweighted the sector. This illustrates the BF same-sign rule: the manager underweighted a sector that underperformed the benchmark (Financials returned 0.80% vs. the 1.206% benchmark), and underweighting an underperformer adds value. This pattern is common in real-world attribution — for example, during the 2008 financial crisis, many active managers who underweighted Financials showed significant positive allocation effects from that single decision.

Try the Performance Attribution Calculator

BHB vs Brinson-Fachler Allocation Formulas

Both variants produce the same total active return — the difference is how the allocation effect is defined. Selection and interaction formulas are identical in both models.

BHB Allocation (1986)

  • Formula: Σ(wp,j − wB,j) × RB,j
  • Uses raw sector benchmark return as the allocation baseline
  • Can show positive allocation from overweighting a sector with positive returns — even if that sector underperformed the overall benchmark
  • Three-component: allocation, selection, interaction shown separately
  • The original Brinson framework (1986)

Brinson-Fachler Allocation (1985)

  • Formula: Σ(wp,j − wB,j) × (RB,j − RB)
  • Uses sector excess return over the overall benchmark as the allocation baseline
  • Allocation is positive when active weight and sector excess return have the same sign — overweighting outperformers or underweighting underperformers
  • Three-component: allocation, selection, interaction shown separately
  • More intuitive allocation interpretation
BHB Allocation Formula
BHB Allocation = Σ(wp,j − wB,j) × RB,j
Uses the raw sector benchmark return rather than the sector’s excess return over the total benchmark

Because the BF allocation baseline subtracts RB, it produces a more intuitive reading: a sector’s BF allocation is positive whenever the active weight and the sector’s excess return have the same sign — overweighting a sector that beat the benchmark, or underweighting one that lagged it. The BHB version can be positive for overweighting any sector with a positive return, even one that lagged the overall benchmark — which can be misleading in reporting. Most commercial attribution systems (FactSet, Bloomberg PORT, Morningstar Direct) offer both variants. Some reporting conventions combine allocation and interaction into a single “allocation” line for a two-component presentation, but this is a display choice rather than the core BHB/BF distinction.

How to Calculate Brinson Attribution

  1. Define the evaluation period. Choose a single period (day, month, quarter) over which to measure attribution. Multi-period attribution requires separate single-period calculations followed by a linking step.
  2. Obtain beginning-of-period weights. Record portfolio and benchmark weights at the start of the period. Using end-of-period weights confounds performance during the period with the initial allocation decision.
  3. Calculate sector-level returns. Compute the return of each sector in both the portfolio and the benchmark, plus the weighted overall portfolio and benchmark returns.
  4. Compute the three effects. Apply the BF allocation, selection, and interaction formulas for each sector, then sum across sectors.
  5. Verify the identity. Confirm that allocation + selection + interaction equals the portfolio’s total active return. Any residual indicates a data or calculation error — or, for holdings-based attribution, a trading/reconciliation effect from intra-period cash flows.
Try the Performance Attribution Calculator
Explore the Portfolio Analytics & Risk Management Course

Common Brinson Attribution Mistakes

1. Confusing BHB and BF allocation effects. The two models use different allocation baselines — BHB uses the raw sector return, BF uses the sector’s excess return over the total benchmark. Comparing BHB allocation numbers from one system against BF numbers from another produces misleading conclusions about where value was added.

2. Applying single-period formulas to multi-period data. Summing monthly allocation effects across a quarter does not produce the correct quarterly allocation effect. Multi-period linking methods (Carino, Menchero, GRAP) are required to ensure that individual-period effects compound correctly to the cumulative active return.

3. Using end-of-period weights instead of beginning-of-period weights. The Brinson model requires weights as of the start of the evaluation period. End-of-period weights already reflect performance during the period, which inflates the allocation effect for sectors that appreciated and deflates it for sectors that declined.

4. Using categories that do not fully cover the portfolio or benchmark. If the sector groupings are not mutually exclusive and collectively exhaustive, the three effects will not sum to total active return. The residual becomes an unexplained “other” bucket that obscures the attribution story. Always verify that portfolio and benchmark weights each sum to 100%.

5. Ignoring the trading/residual reconciliation effect. Holdings-based Brinson attribution produces a buy-and-hold return that may differ from the portfolio’s actual time-weighted return due to intra-period trades and cash flows. For high-turnover portfolios, this residual can be large enough to distort the attribution results. Transaction-based attribution addresses this but requires full trade-level data.

6. Confusing attribution with contribution analysis. Attribution explains why active return differs from zero — what allocation and selection decisions drove the deviation from the benchmark. Contribution analysis explains what each holding added to the portfolio’s total return. Presenting contribution numbers in an attribution context (or vice versa) confuses the audience about the source of value added.

Limitations of Brinson Attribution

Key Limitations

1. Sensitive to classification scheme. Results depend on how securities are grouped into sectors. When a company operates across multiple GICS sectors (e.g., Amazon spanning Technology and Consumer Discretionary), the attribution results change with the classification choice. The quality of any benchmark selection directly affects attribution validity.

2. Holdings-based approach ignores intra-period trading. For actively traded portfolios, the buy-and-hold assumption can miss significant sources of return. Transaction-based attribution captures trading impact but requires complete trade-level data.

3. No risk adjustment. A large positive selection effect concentrated in a single volatile sector may reflect a lucky bet rather than genuine skill. The Brinson model explains where active return came from, not whether the manager was compensated for the risk taken. Risk-adjusted frameworks and metrics like the information ratio are needed to separate skill from risk-taking.

4. Designed for a single-manager, single-benchmark context. Multi-manager fund structures require macro attribution (decomposing returns across managers and asset classes) before applying Brinson at the individual manager level.

5. Multi-period linking introduces methodological choices. Carino, Menchero, and GRAP linking methods can produce slightly different effect magnitudes for the same underlying data. There is no universally accepted “correct” linking method, which creates comparability issues across firms and systems.

The classic equity-sector Brinson framework is also not well-suited for fixed-income portfolios (which require yield curve and spread attribution), derivatives-heavy strategies, long/short portfolios, or absolute-return mandates that lack a meaningful benchmark. For performance reporting standards, see our guide on GIPS standards.

Frequently Asked Questions

The key difference is the allocation effect baseline. BHB (Brinson-Hood-Beebower, 1986) calculates allocation as Σ(wp,j − wB,j) × RB,j, using the raw sector benchmark return. Brinson-Fachler (1985) calculates allocation as Σ(wp,j − wB,j) × (RB,j − RB), using the sector’s excess return over the total benchmark. The selection and interaction formulas are identical in both variants. BF allocation is generally considered more intuitive because it is positive whenever the active weight and the sector’s excess return have the same sign — overweighting outperformers or underweighting underperformers. Both variants produce the same total active return.

The classic equity-style Brinson model can be applied to fixed-income portfolios by grouping bonds into sectors (e.g., by credit quality, maturity, or issuer type), but it is generally not the best approach. Fixed-income returns are driven primarily by yield curve movements, spread changes, and duration positioning — factors that do not map cleanly to a sector-weighting/stock-selection framework. Dedicated fixed-income attribution models such as the Campisi model decompose returns into income, Treasury (duration), and spread components, providing more meaningful insight into fixed-income portfolio decisions. If you use Brinson for bonds, treat the results as a rough first approximation rather than a definitive attribution.

Single-period Brinson effects do not compound across time periods — you cannot simply add January’s allocation effect to February’s to get a two-month total. Multi-period linking methods adjust each period’s effects so they chain correctly to the cumulative active return. Common approaches include the Carino method (uses logarithmic smoothing factors), the Menchero method (optimizes linking coefficients), and GRAP (Geometric Return Attribution Program). Each method can produce slightly different effect magnitudes for the same data. The choice of linking algorithm is primarily a technical implementation decision; the economic interpretation of allocation, selection, and interaction remains the same regardless of which method is used.

The interaction effect captures the joint impact of overweighting a sector where the manager also selected outperforming stocks (or underweighting a sector with poor stock picks). It is positive when overweight sectors have positive selection and negative when overweight sectors have negative selection. In practice, the interaction effect is often small relative to allocation and selection, but it raises an attribution “ownership” question: should it be credited to the portfolio manager (who controls sector weights) or the security analysts (who recommend stocks)? Because security selection decisions often inadvertently affect sector weights, many practitioners attribute interaction to the analysts by combining it with the selection effect in their reporting — a display convention, not a formula change.

The benchmark must be specified in advance, investable, measurable, and appropriate for the portfolio’s mandate. A poorly chosen benchmark invalidates the entire attribution analysis — the allocation and selection effects become artifacts of benchmark mismatch rather than meaningful measures of manager decisions. For example, attributing a US small-cap portfolio against the S&P 500 would show large “allocation” effects that simply reflect the style difference, not active decisions. The benchmark must also use the same sector classification scheme as the portfolio to ensure categories are comparable. For a deeper discussion of benchmark properties, see our guide on benchmark selection.

Disclaimer

This article is for educational and informational purposes only and does not constitute investment advice. Attribution results cited are from hypothetical examples for illustrative purposes. Actual portfolio attribution depends on the specific holdings, benchmark, classification scheme, and methodology used. The Brinson model explains sources of active return but does not by itself prove or disprove manager skill. Always conduct your own analysis and consult a qualified financial advisor before making investment decisions.