Jon Moshier / Notes / Diversification vs Hedging budding
Note · From the Notebook

Diversification vs Hedging

Why spreading risk and offsetting risk are different operations: one lowers variance and is free but fails in a crisis, the other cancels a specific exposure and costs a premium but holds when you need it.

These two words get used as if they were the same risk-management move. They are not. Diversification lowers the variance of an outcome by spreading exposure across things that do not move perfectly together. Hedging cancels a specific exposure by taking a position that pays when the thing you fear happens. The difference is the sign of a correlation, and it decides whether your protection shows up in the exact year you need it.

The definitions turn on one correlation

The cleanest formalization comes from the academic literature on gold. Baur and Lucey’s 2010 paper Is Gold a Hedge or a Safe Haven? (Financial Review 45, 217-229) defines three roles by correlation, and the boundaries are exact:

Their finding: gold was a hedge against stocks on average and a safe haven in extreme stock-market crashes, but the safe-haven effect was short-lived, fading within weeks. Gold was neither for bonds. The point that matters here is the taxonomy. A diversifier and a hedge are separated by whether the correlation is merely below one or actually below zero.

The variance math makes it concrete

Take two assets with equal weight and equal volatility σ. Portfolio variance is w₁²σ₁² + w₂²σ₂² + 2·w₁w₂·ρ·σ₁σ₂, so the whole story lives in ρ, the correlation.

Diversification lives in the region between +1 and 0 and removes only idiosyncratic risk, the part specific to one holding. It cannot touch systematic risk, the part shared by everything. That split is why adding names stops helping fast: Meir Statman’s 1987 analysis put the useful threshold around 30 to 40 stocks, after which each new name buys almost nothing. Hedging lives at ρ < 0 and is the only one of the two that offsets a loss rather than averaging it away.

The asymmetry: free but unreliable vs costly but dependable

Diversification is close to free. You are not buying anything, just declining to concentrate, and you keep the upside of every holding. Its weakness is reliability. Correlations are not constant, and in a crash they tend to converge toward one. In the 2008 collapse, assets that looked uncorrelated in calm markets fell together as forced selling and a flight from risk hit everything at once. Diversification evaporated at the exact moment it was supposed to pay off. This is the same mechanism as the ski case in Ski Industry Risk Management: a broad warm winter is the systematic shock that makes “diversified” mountains dry together.

Hedging is the mirror image. It costs a premium, an option price, an insurance payment, the give-up of upside in a futures lock. But it works in the tail by construction, because it is a contractual offset rather than a statistical tendency. Insurance is the purest hedge: you pay a known small cost every year to be made whole in the rare bad one, and the payout is defined by your loss, not by an average. Hedging is not unconditionally clean, though. Its cost is itself state-dependent, because option premiums and insurance reprice upward in exactly the stress that makes you want them, so the protection is dearest when demand for it peaks. And an imperfect hedge leaks through basis, the gap between the thing you own and the thing that actually pays out. Liability Matching is the same logic applied to known future bills, buying the asset whose payoff is timed to the obligation rather than the one with the highest expected return.

So the choice is not which is better. It is what you are protecting. Diversification is the right tool for the unknowable, scattered, idiosyncratic risks where you have no specific exposure to cancel. Hedging is the right tool for a specific, identified exposure you cannot afford to be wrong about, and where you will pay to be certain. The error is reaching for diversification when the risk is systematic, because that is the one case where spreading out does nothing.

Try it

Sweep the correlation and watch the regimes separate (1-2 hours, Python or a spreadsheet). Build two synthetic assets, each with annual volatility 20% and equal weight. Compute portfolio volatility with the two-asset formula while sweeping ρ from −1 to +1 in steps of 0.1, and plot it. You are looking for three things: the curve is flat-high at ρ = +1 (no benefit), passes through σ/√2 ≈ 14.1% at ρ = 0 (diversification), and collapses toward zero only as ρ goes negative (hedging). Then add the failure mode: simulate a “crisis” by drawing returns where ρ jumps from 0.1 to 0.9 in the worst 5% of months, and confirm that realized drawdown is far worse than the calm-period correlation predicted. That gap is why a diversified book still blows up in a crash and a hedge does not.

See also

Sources

← All notes Read recent essays →