What is the allocation effect?
In the Brinson-Hood-Beebower attribution framework, the allocation effect isolates the contribution to active return from the manager's decision to deviate from benchmark weights across segments -- sectors, asset classes, countries, or any other grouping used in the attribution.
The allocation effect for a segment is calculated as: (portfolio weight - benchmark weight) multiplied by (benchmark segment return - total benchmark return). The subtraction of total benchmark return is the key element: it measures whether the overweighted segment outperformed the overall benchmark, not just whether it had a positive return.
A manager who overweights technology by 5 percentage points when technology outperforms the benchmark generates a positive allocation effect from that overweight. The same overweight generates a negative allocation effect if technology underperforms the benchmark even if technology had a positive absolute return.
The allocation effect is zero for any segment where the manager holds the exact benchmark weight. It is positive when the manager overweights outperforming segments or underweights underperforming segments. It is negative when the manager overweights underperforming segments or underweights outperforming segments.
How the allocation effect is calculated in practice
The standard Brinson formula for allocation effect in sector i is: (w_p,i - w_b,i) * (r_b,i - r_b), where w_p,i is the portfolio weight in sector i, w_b,i is the benchmark weight in sector i, r_b,i is the benchmark return in sector i, and r_b is the total benchmark return.
The formula applies to each segment independently, then the segment-level allocation effects are summed to produce the total portfolio allocation effect. In a multi-sector portfolio with k sectors, the total allocation effect is the sum of (w_p,i - w_b,i) * (r_b,i - r_b) for i = 1 to k.
The formula uses the benchmark return for each segment (not the portfolio return) in the benchmark column. This separates the allocation decision -- the weight -- from the security selection decision -- which specific securities within the segment to hold. A pure allocation model asks: given that you took a position in this segment, what would have happened if you had held the benchmark's securities within it?
Multi-period allocation effects are not additive. The allocation effect in period one becomes embedded in the portfolio weights for period two. Geometric linking of attribution returns is required to make multi-period attribution sum correctly to the total active return.
Allocation effect versus selection effect
The Brinson framework decomposes total active return into three components: allocation effect (from weighting decisions), selection effect (from security selection within each segment), and interaction effect (the cross-product of deviating on both weight and selection simultaneously).
A manager can have a positive total active return while having a negative allocation effect and a strongly positive selection effect. This is common for concentrated bottom-up stock pickers who generate most of their alpha from security selection rather than from top-down sector weighting.
The interaction effect is often small and is sometimes eliminated by using a Brinson-Fachler or geometric attribution method that reformulates the three terms to avoid the cross-product. Different attribution methods will produce different allocation and selection numbers for the same portfolio -- consistent methodology within an evaluation period is what matters.
Evaluating a manager's skill requires understanding which decisions they actually made. A fundamentals-driven bottom-up manager who shows a large allocation effect is not necessarily being rewarded for intentional tactical tilts: sector weights are often a byproduct of security selection in a benchmark-unconstrained strategy.
Frequently asked questions
What is the allocation effect in Brinson performance attribution?
The allocation effect measures the return contribution from the manager's decision to overweight or underweight segments relative to the benchmark. In the Brinson formula, it is calculated as (portfolio weight - benchmark weight) multiplied by (benchmark segment return - total benchmark return). It captures only the weighting decision, using the benchmark's segment return rather than the portfolio's to isolate weighting from stock selection.
How is the allocation effect calculated in practice?
For each segment (sector, country, asset class), the allocation effect equals (w_p - w_b) * (r_b,segment - r_b,total). The portfolio weight minus benchmark weight is the active weight; the benchmark segment return minus total benchmark return is the excess return of that segment relative to the overall benchmark. Sum this across all segments to get the total allocation effect. The formula uses benchmark returns for both the segment and the total to isolate weighting decisions from security selection decisions.
What does a positive allocation effect mean?
A positive allocation effect means the manager's weighting decisions added value. Specifically, the manager overweighted segments that outperformed the overall benchmark, underweighted segments that underperformed, or both. A positive allocation effect does not require overweighted segments to have positive absolute returns -- only that overweighted segments outperformed the benchmark average, and underweighted segments underperformed it.
How does allocation effect relate to selection effect in performance attribution?
In the Brinson framework, allocation effect and selection effect are orthogonal decompositions of total active return. Allocation effect captures the value from sector/country/asset-class weighting decisions; selection effect captures the value from choosing different securities within each segment than the benchmark holds. A manager can produce a negative allocation effect and still add total value if selection effect is large and positive -- common for bottom-up stock pickers who generate most alpha from security selection rather than top-down sector calls.