# Performance attribution

*From Chapter 8 of The Analyst's Handbook, Sifter's free textbook for student investors · By Matt Lang and Bree Lang · https://sifter.finance/textbook/topics/performance-attribution*

Suppose that a fund returned 3.10% in a period when its benchmark returned 2.30%. The fund is ahead by 0.80 points, and at the next meeting somebody is going to ask where those 0.80 points came from. Performance attribution is the set of calculations that answers the question.

The simplest version takes one holding at a time. The contribution to excess return (CTE) of a holding is its weight in the fund multiplied by how far it beat or trailed the whole benchmark:

```
CTE = w × (R_h − R_b)
```

Here w is the holding's weight, R_h is its return and R_b is the return of the entire benchmark. Imagine that our fund has three holdings. The first is 60% of the fund and returned 5.00%, so its contribution is 0.60 × (5.00% − 2.30%) = +1.62 points. The second is 25% of the fund and returned 1.00%, so its contribution is 0.25 × (1.00% − 2.30%) = −0.33 points. The third is 15% of the fund and returned −1.00%, which makes its contribution 0.15 × (−1.00% − 2.30%) = −0.50 points.

Take a moment with the second holding. It made money, and it still pulled the active return down, because it made less than the benchmark did. This is the distinction that students miss most often. The three contributions add up to 1.62 − 0.33 − 0.50 = 0.80 points, which is the active return we started with. (If they ever fail to add up, something in the calculation is broken.)

A second method, called Brinson-Fachler attribution, sorts the same 0.80 points by decision instead of by holding. For each sector it asks three questions, which are whether the fund held more of a sector that beat the benchmark (allocation), whether its picks inside the sector beat the sector (selection) and what it earned by doing both at once (interaction):

```
Allocation = (w_p − w_b) × (R_b,s − R_b)
Selection = w_b × (R_p,s − R_b,s)
Interaction = (w_p − w_b) × (R_p,s − R_b,s)
```

Here w_p and w_b are the fund's and the benchmark's weights in the sector, and R_p,s and R_b,s are their returns inside it. Suppose that our 60% holding is the fund's only technology stock, and that the benchmark has 50% in technology and earned 4.00% there. The allocation effect is 0.10 × (4.00% − 2.30%) = +0.17 points, the selection effect is 0.50 × (5.00% − 4.00%) = +0.50 points and the interaction is 0.10 × 1.00% = +0.10 points.

Across all three sectors of this fund, allocation comes to 0.35 points, selection to 0.30 and interaction to 0.15, for the same total of 0.80. Chapter 8 of the textbook works through both methods in full.

The full chapter: https://sifter.finance/textbook/ch08-attribution.md
