Is the small-cap premium real or is it just time-period bias? The evidence seems to flip depending on which years you examine.

The small-cap premium is one of the best illustrations of time-period bias in finance. Whether you conclude it exists depends almost entirely on when you start and stop counting.

The Dramatic Sensitivity:

Notice how skipping just the 1926–1932 period (the Great Depression, which devastated small firms) changes the annual spread from 0.43% to 3.49% — an eightfold increase. This is time-period bias in its most dramatic form.

Example — Birchwood Endowment's Allocation Decision:

Birchwood University's endowment committee is debating a 10% tilt toward US small-cap equities. The investment staff presents two analyses:

• Analyst A uses 1932–2010 data: small-cap premium of +3.1%/year with a t-statistic of 2.4. Recommends the tilt.
• Analyst B uses 1990–2023 data: small-cap premium of -0.8%/year with a t-statistic of -0.5. Recommends against the tilt.

Both analysts are technically correct — their statistics are accurately computed. The disagreement is entirely about which period is relevant.

Is the Premium Real?

The honest answer is nuanced:

1. Economic rationale exists: Small firms face higher business risk, less diversified revenue streams, greater information asymmetry, and higher trading costs. Investors should theoretically demand compensation for these risks.
2. The evidence is fragile: The premium is concentrated in specific periods (especially January) and among the smallest, least-liquid stocks that institutional investors can't easily access.
3. Transaction costs erode it: The theoretical premium may exist in gross returns but be largely captured by market makers and brokers in the form of wider bid-ask spreads.
4. It may have been arbitraged away: Once the academic research was published (Banz, 1981; Fama-French, 1993), capital flowed into small-cap strategies, potentially compressing the premium.

What This Means for CME:

The small-cap example teaches a broader lesson: any factor premium or return spread that is highly sensitive to the choice of sample period should be treated with extreme caution in forward-looking CMEs. The analyst should:

1. Report results for multiple sub-periods
2. Test whether the economic rationale still applies in the current market structure
3. Consider whether the premium has been arbitraged since its discovery
4. Use conservative (lower) estimates rather than the most optimistic sample window

Exam Tip: On the CFA exam, time-period bias questions typically present two analyses of the same relationship using different date ranges that produce opposite conclusions. You need to identify this as time-period bias and recommend robustness testing.

Practice factor premium analysis in our CFA Level III question bank.