Risk

Correlation Breakdown and Why Hedges Fail Exactly When They Matter

Understand why the diversification and the hedges you bought stop working in the exact conditions they were bought for, so you can find the common driver hiding under a portfolio of "independent" exposures before it finds you.

  • Advanced
  • 13 min total
  • 15 chapters

What decision this helps you make: Which of your protections depend on a relationship that was measured in calm conditions, and what to hold instead for the case where that relationship does not survive the stress.

What this topic is

Correlation breakdown is what happens when a statistical relationship you relied on changes at exactly the moment you needed it. It shows up in two forms. Diversification failure: exposures that looked independent all move together once a common driver is revealed. And hedge failure: the instrument you bought stops tracking the thing it was supposed to offset, leaving you with the loss you thought you had removed and the cost of the hedge on top.

Why it matters

Correlation is estimated from history, and history is mostly calm. Both diversification and hedging are priced and sized against a number produced by ordinary periods, then relied upon in extraordinary ones. That is not an occasional bad break. It is a structural feature of how the number is built, which is why protection that tests well tends to fail on the specific days it was bought for.

Who should learn it

Anyone whose safety comes from having several of something (customers, suppliers, revenue lines, currencies, funding sources), and anyone holding a hedge, a swap, a forward, or a parametric cover whose payout depends on tracking something other than their own loss.

What you will understand

  • The four mechanisms that make correlations rise together under stress
  • Why a correlation measured in calm conditions systematically understates tail dependence
  • What basis risk is, and why it is largest exactly when the hedge matters most
  • What to hold instead when the relationship you were relying on cannot be trusted

Prerequisites

Common misconception

"We are diversified. No single customer is more than eight percent of revenue." That is a statement about the count of exposures, not about what drives them. If nine of your twelve largest customers sell into the same end market, borrow from the same regional lenders, or depend on the same shipping lane, you have one exposure recorded twelve times. The diversification test is never how many names are on the list. It is whether you can identify a single event that moves several of them at once, and how large that combined move would be.