How do you decompose tracking error into factor-based and stock-specific components?

Question

AcadiFi · Accepted Answer

**Tracking error decomposition** separates total active risk into two components: risk arising from **systematic factor bets** (sector tilts, style exposures) and risk from **idiosyncratic stock selection** (individual security positions deviating from the benchmark).

**Decomposition Framework**

**TE^2 = Factor TE^2 + Specific TE^2**

- **Factor TE** = active risk from differences in factor exposures between portfolio and benchmark (size, value, momentum, sector, country, etc.)
- **Specific TE** = active risk from stock-specific overweights/underweights not explained by factor exposures

**How It Works**

Using a multi-factor risk model (Barra, Axioma, Northfield), each position is decomposed into factor exposures and a residual. The active weights (portfolio weight - benchmark weight) are multiplied by the factor covariance matrix to get factor TE, and the residual variances give specific TE.

**Worked Example**

Caldwell Active Equity Fund vs. S&P 500:

| Component | TE Contribution | % of Total TE^2 |
|---|---|---|
| Sector tilts (overweight tech, underweight energy) | 1.8% | 36% |
| Size tilt (slight small-cap bias) | 0.9% | 9% |
| Value/growth tilt (growth bias) | 1.1% | 13% |
| Momentum loading | 0.5% | 3% |
| **Total Factor TE** | **2.3%** | **~59%** |
| Stock selection (individual picks) | **1.9%** | **~41%** |
| **Total TE (diversified)** | **3.0%** | **100%** |

(Note: components don't add linearly due to diversification; they add in variance terms)

```mermaid
flowchart TD
    A["Total Tracking Error: 3.0%"] --> B["Factor TE: 2.3% (59% of variance)"]
    A --> C["Specific TE: 1.9% (41% of variance)"]
    B --> D["Sector tilts: 1.8%"]
    B --> E["Size tilt: 0.9%"]
    B --> F["Style tilt: 1.1%"]
    C --> G["Stock-specific bets"]
```

**What This Tells You**

| Manager Type | Factor TE | Specific TE | Interpretation |
|---|---|---|---|
| Sector rotator | High | Low | Active risk from top-down bets; could be replicated with ETFs |
| Stock picker | Low | High | Active risk from bottom-up selection; genuine alpha potential |
| Closet indexer | Very low | Very low | Total TE is tiny; not worth active fees |
| Style tilter | High factor-specific | Moderate | Active return driven by value/growth/size bets |

**Implications for Fee Justification**

If Caldwell's factor TE (59% of risk) could be replicated cheaply through factor ETFs, only the 41% from stock selection truly justifies active management fees. The plan sponsor should evaluate whether paying 65 bps for factor tilts that cost 10 bps through ETFs is rational.

**Active Share vs. Tracking Error**

Active share measures the percentage of holdings that differ from the benchmark (a weight-based metric). Tracking error measures the volatility of active returns (a return-based metric). Decomposing TE into factor and specific components bridges the two — high factor TE often corresponds to systematic active share, while high specific TE corresponds to stock-level active share.

For more on portfolio analytics, visit our CFA Level III course.

How do you decompose tracking error into factor-based and stock-specific components?

Master Level III with our CFA Course

Related Questions

Practice Questions