Pricing multi-asset trades – part two

Last month's discussion of multi-asset trades noted that the two most common product types to use multiple assets as underlyings are worst-of autocallables and reverse convertibles, making them sensitive to correlation. Despite the significant differences between the two products, in both cases investors are 'long' correlation and will benefit from the correlation between the assets in the underlying basket turning out to be higher than is expected or priced in. The issuing bank is therefore 'short' correlation and must price and hedge accordingly.

The fact that virtually all retail products that are sensitive to correlation have the same net exposure for the issuing bank means it is impossible for the banks to find offsetting trades that can help reduce the risk. If there was a two-way flow of products split equally between long and short correlation risk, the fact that correlation is almost impossible to hedge elsewhere would matter much less.

With volatility exposure, on the other hand, there are product types that can be used to hedge volatility that are usually issued in high volumes with differing exposures throughout the investment cycle. Products in which the bank is long volatility include capital-protected participation notes, whereas reverse convertibles and autocallables on a single underlying are short volatility. Volatility exposure is eased by the existence of liquid options on exchanges for the bigger stocks in most markets, which are often available out to a period of at least one or two years.

The only trades that help the position in correlation are dispersion trades, where investors stand to benefit from gains in individual stocks and are short the index or basket return. However, these trades are somewhat artificial and complex, and tend to be taken on only by institutional investors, who benefit from the favourable levels of correlation at which the trades may be priced in order to get the offsetting trade in place.

That the vast majority of trades involve the bank being short correlation with no other hedging mechanism available means one would automatically expect the typical traded or implied value of correlation to be consistently higher than the long-run average, which might be thought of as an equilibrium 'fair' value.

This margin of correlation above its fair value can be quite significant and will affect the likely performance of such trades from the investor's perspective. As with any structured product, a balance will need to be struck between the level at which the bank will be prepared to trade and the product terms that investors will find attractive. If no agreement can be found, then the product type will quickly be abandoned.

One compromise that can suit all parties is a multi-asset trade, which has lower correlation sensitivity than alternative structures. This means the bank is still able to mark correlation at a level it deems prudent, but the lower price sensitivity to correlation means the reduction in product terms is less significant. Another way to mitigate high volatility levels for investors wanting capital protection is to provide a call spread or digital as upside instead of uncapped participation, because there will be less of a perceived reduction in terms.

As a final observation on the complexity of correlation risk management, it should be noted that not only is correlation unhedgeable and unstable, but it also tends to exhibit different behaviour depending on market conditions. Typically in a bear market, particularly when there is a sudden fall, all assets tend to decline together, while in calmer or rising markets they move more independently. Correlation can therefore be thought of as higher in falling markets than in flat or rising markets. This effect is sometimes known as correlation skew. Since most products of the type discussed here have most exposure on the downside, it is clear that the level that correlation is priced at needs to reflect the likely level of correlation in the event of markets declining.