Correlation Swap: Trade Average Pairwise Correlation Directly

Key Takeaways

• A correlation swap pays notional times (realized average pairwise correlation minus strike); for a 50-name basket, realized correlation is the equally weighted average of 1,225 distinct pairwise correlation terms.
• Correlation spikes fast and mean-reverts slowly, a global risk-off event can push realized correlation from 0.45 to 0.62 in weeks, while short-correlation positions suffer mark-to-market losses that may force exits before expiry.
• The correlation swap and a variance dispersion trade both give correlation exposure, but their P&L profiles diverge during vol spikes because the dispersion trade embeds a volatility-of-volatility component.
• Structured-product desks are structurally short correlation because selling autocallables and worst-of products embeds a short-correlation position; they buy correlation swaps to hedge this accumulated exposure.

What It Is

The buyer of a correlation swap receives the difference between the realized average pairwise correlation of the basket and the strike correlation written into the contract, multiplied by a notional amount. The seller takes the other side.

Reference baskets are typically equity index constituents, such as the top 50 names in the S&P 500 or the full Euro Stoxx 50 roster. Dealers use correlation swaps to hedge the correlation risk they accumulate when they sell structured products linked to index volatility or worst-of baskets.

Correlation swaps are OTC-only. They do not trade on exchanges. Documentation follows an ISDA master agreement with a bespoke confirmation. Liquidity is modest and concentrated among a handful of equity derivatives desks.

The Intuition

Index volatility is not the average volatility of the constituents. It is lower, because individual moves partially offset at the basket level. The gap between index vol and average single-name vol is driven by correlation. When correlation rises, the index behaves more like one big stock and diversification inside the basket collapses.

A dealer who sells a volatility product on an index ends up short index vol and implicitly short correlation. If something scares the market and every name sells off together, that short correlation position loses money fast. The dealer can offset this risk by buying a correlation swap. The counterparty, often a hedge fund running a dispersion book, takes the short side because they get paid for carrying it.

The correlation swap is therefore a hedging tool first and a speculative vehicle second.

How It Works

The payoff to the correlation buyer at maturity is:

Payoff = Notional * (RealizedCorr - StrikeCorr)

Realized correlation is the equally weighted average of all distinct pairwise realized correlations among the n basket members, measured daily over the life of the swap:

RealizedCorr = (2 / (n * (n - 1))) * sum over i<j of rho(i, j)

Each rho(i, j) is the historical correlation of daily log returns between stocks i and j over the observation window. Pairs with n names generate n(n-1)/2 pairwise measurements. For a 50-name basket, that is 1,225 distinct correlation terms averaged into a single number.

Pricing is not a closed-form exercise. Dealers typically quote strikes at a few points below the implied correlation extracted from index and single-name variance swaps. That gap is the dealer's expected profit for running the residual risk that cannot be perfectly replicated.

Worked Example

A structured-products desk at a large bank has sold $500 million of autocallable notes linked to the Euro Stoxx 50. The hedging book is short index vol and short correlation. The desk decides to hedge part of the correlation exposure.

It buys a 1-year correlation swap on the top 50 Euro Stoxx names, notional $10 million per correlation point, at a strike of 0.45. That means for every 0.01 increase in realized average pairwise correlation above 0.45, the desk receives $100,000.

Twelve months later, a global risk-off event pushes realized correlation across European equities to 0.62. The payoff is:

$10M * (0.62 - 0.45) = $10M * 0.17 = $1.7M per correlation point

At $10M per point, that resolves to a $1.7M payoff. If correlation had instead come in at 0.35 (below the strike), the desk would owe $1M to the counterparty. The hedge fund on the short side built the position into a wider dispersion book and treats the loss as the cost of carry on the short-correlation leg.

Common Mistakes

1. Thinking the swap replaces a dispersion trade. A correlation swap and a variance dispersion trade both give correlation exposure, but their P&L profiles are not identical. Jacquier and Slaoui show that the implied correlation from a dispersion trade includes a volatility-of-volatility component, so the two products can diverge materially during vol spikes.

2. Ignoring basket composition drift. Index membership changes over the life of the swap. Document whether the basket is fixed at trade date or adjusted at each reconstitution. A mismatch here can cost several correlation points.

3. Underestimating one-sided liquidity. Very few market participants are natural buyers of correlation. Dealers are structurally short and hedge funds are the main sellers. If you need to unwind early, expect wide bid-ask spreads, often 5 correlation points or more.

4. Assuming symmetry in a stress event. Correlation tends to spike much faster than it mean-reverts. Short-correlation positions suffer path-dependent drawdowns even if the terminal realized number lands near the strike, because mark-to-market losses can trigger margin calls that force an exit before expiry.

Frequently Asked Questions

Q: What is a correlation swap in simple terms? A correlation swap pays the difference between the average realized pairwise correlation of a basket of stocks and a fixed strike, multiplied by a notional. It is the cleanest way to take a direct bet on whether stocks move more or less in lockstep, without the complexity of managing a basket of options.

Q: How does a correlation swap affect investment decisions? Structured-product desks use correlation swaps to neutralize the correlation risk they accumulate when they sell products linked to worst-of baskets or index volatility. Hedge funds running dispersion trades are natural sellers of correlation swaps, pocketing the risk premium as carry.

Q: What is a real-world example of a correlation swap? A structured-products desk buys a 1-year correlation swap on the top 50 Euro Stoxx names at a strike of 0.45, notional $10 million per correlation point. A global risk-off event drives realized average correlation to 0.62. The payoff is $10M times (0.62 minus 0.45) = $1.7 million, offsetting losses on the desk's short-correlation structured product book.

Q: How can investors use correlation swaps in a portfolio? Correlation swaps are institutional instruments suited for hedge funds running dispersion books. As a seller of correlation (short the swap), a fund receives carry because implied correlation typically prices above realized. The risk is a sudden market-wide selloff that spikes realized correlation far above the strike.

Q: How is a correlation swap different from a dispersion trade? A dispersion trade achieves similar correlation exposure by going long single-name variance and short index variance, but its payoff depends on the ratio of index to single-name volatility, not directly on correlation. The two instruments can diverge materially when volatility-of-volatility is high, making the correlation swap a more direct but less liquid instrument.

Sources

1. Jacquier, A. and Slaoui, S. (2010). Variance Dispersion and Correlation Swaps. Imperial College London. https://www.ma.imperial.ac.uk/~ajacquie/index_files/Jacquier,%20Slaoui%20-%20Dispersion.pdf
2. Bossu, S. (2016). A Primer on Correlation Trading via Equity Derivatives. http://docs.sbossu.com/bossu-correl_primer-20160408.pdf
3. Foresi, S. and Vesval, A. Equity Correlation Trading. Goldman Sachs Asset Management / NYU Stern. https://web-docs.stern.nyu.edu/salomon/docs/derivatives/GSAM%20-%20NYU%20conference%20042106%20-%20Correlation%20trading.pdf

Disclaimer

This article is educational content only and is not financial advice. Nothing here is a recommendation to buy, sell, or hold any security. Consult a licensed advisor before making investment decisions.