What are the key differences between survey-based and historical equity risk premium estimates?

Question

AcadiFi · Accepted Answer

The equity risk premium (ERP) is the excess return investors require for holding equities over risk-free assets. The two most common estimation approaches -- historical and survey-based -- often produce meaningfully different results.

**Historical ERP:**
Calculate the average excess of equity returns over government bond returns over a long sample period.

**Survey ERP:**
Ask institutional investors, CFOs, or academics what ERP they expect going forward.

**Worked Example -- Divergence in Practice (fictional compilation)**

| Method | Period/Source | ERP Estimate |
|---|---|---|
| Historical (US, geometric) | 1926-2025 | ~5.5% |
| Historical (US, arithmetic) | 1926-2025 | ~7.2% |
| Historical (Global ex-US) | 1900-2025 | ~3.8% |
| Survey (CFO panel, 2026) | Forward-looking | ~4.2% |
| Survey (institutional, 2026) | Forward-looking | ~5.0% |

**Key Differences:**

| Issue | Historical | Survey |
|---|---|---|
| Backward vs forward | Past realized returns | Expected future returns |
| Survivorship bias | US data overstates (winner's market) | Not affected |
| Arithmetic vs geometric | Choice matters enormously | Not applicable |
| Time period sensitivity | Very sensitive to start/end dates | Reflects current conditions |
| Structural breaks | Includes regime changes | Adapts to current regime |
| Anchoring bias | N/A | Respondents anchor on recent memory |
| Precision | Statistically imprecise (wide CI) | Subjective but reflects consensus |

**Which to Use?**

The CFA curriculum recommends using multiple approaches and triangulating. The Dimson-Marsh-Staunton global dataset suggests historical ERP of roughly 3.5-5.5% (geometric, various markets). Surveys tend to cluster around 4-5%. Forward-looking supply-side models (discussed separately) provide yet another perspective.

For valuation work, many practitioners use a blended estimate of 4.5-5.5% for developed markets, recognizing that the US historical premium likely overstates the true expected premium due to survivorship bias and the extraordinary performance of the US equity market in the 20th century.

What are the key differences between survey-based and historical equity risk premium estimates?

Master Level II with our CFA Course

Related Questions

Practice Questions