How does surplus optimization differ from traditional mean-variance optimization, and what does the surplus efficient frontier look like?

Question

AcadiFi · Accepted Answer

Surplus optimization reframes the portfolio problem from an asset-only perspective to an asset-liability perspective. Instead of maximizing the expected return of assets for a given level of asset volatility, it maximizes the expected change in **surplus** (assets minus liabilities) for a given level of **surplus volatility**.

**Why Standard MVO Fails for Pensions**

Standard MVO treats liabilities as external. A portfolio of 100% equities might sit on the asset-only efficient frontier, but if liabilities have bond-like characteristics (long duration, sensitive to rates), an equity-heavy portfolio creates enormous surplus volatility. A small rate decline could devastate the funded ratio even while equities rise.

**Surplus Optimization Mechanics**

The surplus return is:

R_surplus = R_assets − (L/A) × R_liabilities

Where L/A is the liability-to-asset ratio. The optimization minimizes surplus variance:

σ²_surplus = σ²_assets + (L/A)² × σ²_liabilities − 2 × (L/A) × cov(R_assets, R_liabilities)

The key insight: assets that are highly **correlated with liabilities** reduce surplus volatility, even if their stand-alone returns are lower.

**Worked Example — Fielding Pension ($5B assets, $5.5B liabilities):**

L/A = 1.10. Three candidate portfolios:

| Portfolio | E(R_assets) | σ_assets | Corr(assets, liab) | E(R_surplus) | σ_surplus |
|---|---|---|---|---|---|
| A: 80% equity / 20% bonds | 7.8% | 12.1% | 0.15 | 2.1% | 14.8% |
| B: 40% equity / 60% long bonds | 5.5% | 7.3% | 0.72 | −0.3% | 5.1% |
| C: 50% equity / 30% long bonds / 20% RE | 6.4% | 8.5% | 0.55 | 0.8% | 7.9% |

Portfolio A dominates on the **asset-only** efficient frontier. But Portfolio B has the lowest **surplus volatility** because its high correlation with liabilities (0.72) makes assets and liabilities move together, dampening surplus swings.

**The Surplus Efficient Frontier:**

The surplus efficient frontier plots E(R_surplus) against σ_surplus. It looks different from the asset-only frontier:

- **Minimum surplus variance portfolio** is dominated by long-duration bonds (high liability correlation), not diversified equity-bond mixes.
- **Moving up the frontier** adds equities and alternatives, increasing both expected surplus return and surplus risk.
- **The tangency portfolio** (maximum surplus Sharpe ratio) balances return-seeking assets with liability-hedging assets.

**Implementation:**

1. Estimate expected returns, volatilities, and correlations for all asset classes **and the liability**.
2. Solve the quadratic optimization with the surplus objective function.
3. Apply constraints: liquidity, regulatory, concentration limits.
4. The output is the optimal LHP/RSP split and within-bucket allocation.

**Limitations:**

- Sensitivity to input estimates (same as MVO — "garbage in, garbage out").
- Liability returns are hard to estimate — they depend on discount rate changes, actuarial assumption updates, and demographic shifts.
- Non-normal return distributions (especially for alternatives) can distort the optimization.

Our CFA Level III fixed income and portfolio management modules cover surplus optimization in depth.

How does surplus optimization differ from traditional mean-variance optimization, and what does the surplus efficient frontier look like?

Master Level III with our CFA Course

Related Questions

Practice Questions