What are the key assumptions of classical immunization and when does it fail?

Question

AcadiFi · Accepted Answer

Classical immunization (Redington, 1952; Fisher & Weil, 1971) rests on assumptions that rarely hold perfectly in practice. Understanding the assumptions and their breakdowns is a favorite CFA Level III topic.

Core assumptions:

1. PARALLEL YIELD CURVE SHIFTS: The yield curve moves up or down uniformly — all rates shift by the same amount. Reality: short and long rates move differently (twists, butterflies).

2. SINGLE SHIFT: The curve shifts once immediately after portfolio construction, then stays. Reality: continuous changes require rebalancing.

3. INSTANTANEOUS SHIFT: No time passes between purchase and shift. Reality: if time passes, duration and horizon drift differently.

4. RISK-FREE BONDS: No default risk. Reality: even Treasuries have duration considerations with TIPS vs. nominal.

5. NO EMBEDDED OPTIONS: Bonds don't have calls, puts, or prepayments. Reality: callable bonds have negative convexity.

Failure modes:

FAILURE 1 — NON-PARALLEL SHIFTS: Yield curve flattens (short rates up, long rates down). A duration-matched portfolio with long/short barbell can suffer badly. Hedge via KEY-RATE DURATION matching rather than single Macaulay duration.

FAILURE 2 — STOCHASTIC PATH: Rates move randomly over the horizon. Classical immunization only perfectly works for a single immediate shift. Need to REBALANCE periodically (monthly or quarterly).

FAILURE 3 — CREDIT SPREAD CHANGES: If using corporate bonds, widening credit spreads break the rate-matching logic.

Case of failure: Bluepointe Pension bought a barbell to immunize a 10-year liability (duration 10). Mix: 50% 2-year notes, 50% 25-year bonds gives combined duration 10. The yield curve then flattened — short rates rose 150 bps, long rates fell 50 bps. Parallel-shift math says near-zero change, but actual portfolio lost 4% of value relative to liability because of the high convexity mismatch on the barbell vs. the bullet liability.

Solution: M² (M-squared) measure penalizes portfolio/liability dispersion, and DURATION VECTOR matching handles non-parallel shifts.

What are the key assumptions of classical immunization and when does it fail?

Master Level III with our CFA Course

Related Questions

Practice Questions