While the plunge in cross-asset correlations, which as recently as last fall was at all time highs, has been duly documented in recent weeks, nowhere is it more obvious than in the following chart from Morgan Stanley which shows the firm's global correlation index.
Morgan Stanley's Andrew Sheets summarizes the stunning move as follows: "Regional correlations, cross-asset correlations and individual stock and FX correlations have fallen simultaneously. That's unusual; we haven't seen a shift this severe in over a decade" and ultimately calls it for what it is: a "crash" in correlations unlike any seen before.
He then concedes that while 'crash' is not a term used lightly adding that "our editors here at Morgan Stanley won't let us use it without a good reason" he "struggles to think of another word to describe just how much, and how sharply, cross-asset correlations have fallen. In just four months, we have gone from a market of unusually close linkages across markets, to one with usually divergent returns."
There are three direct implications from the move:
- Lower correlation should mean a better 'macro' trading environment (since each market isn't the 'same' trade). It is a mixed blessing for diversification, lowering overall portfolio volatility, but also making hedging through 'proxies' harder.
- Lower cross-asset correlation is common in 'late cycle' environments, fitting our preference for equities > credit. Meanwhile, USDKRW remains our favourite 'proxy' hedge in a lower correlation world, given its linkages to several macro risks.
- We'd expected correlations to rise again if growth data disappointed. Since lower correlation has helped to depress volatility, hedges that benefit from both correlation and vol moving benefit from unusually good pricing now. Our favourites are AUDUSD, gold and EURUSD.
Sheets then says that the sharp drop in cross-asset correlation is not a fluke. It represents a decline in correlations at each of the three key 'levels' that we care about – cross-asset, crossmarket and intra-market. Our correlation indices (Exhibit 2) focus on the first two of these, while the last is an important addition.
- Cross-asset correlation (50% of our index): This refers to how closely prices in different assets are moving relative to each other (credit vs. equities, equities vs. rates, equities vs. FX, etc.). Declines here have been driven by a more ambiguous relationship of rates to equity and credit in recent months. Rates-equity correlation has been this poor only during the taper tantrum and then before that in 2006/07. The dollar has decoupled from stocks, commodities and yields, which have rallied in a rising dollar environment.
- Cross-region correlation (50% of our index): This refers to how closely prices in different regions of the same asset class are moving relative to each other (i.e., US vs. EM equities, European vs. Japanese rates, EUR vs. GBP). Declines here have been sharp as well, as monetary policy and political risks have an impact. Intra-fx correlations have also changed – notably, USDJPY correlations to a number of risk currencies have become weaker.
- Intra-market correlation: This refers to how closely prices of different securities within an index are moving (i.e., are the stocks within the S&P 500 all moving together, or not). We use different indices to measure this, which we plot in Exhibit 2.
From an economic perspective, the dramatic move is troubling because such a decline is common in 'late cycle' environments, as 'idiosyncratic' stories (company-specific M&A, tweets) are larger drivers than economic or earnings fears MS adds. The decline is directly linked to the lower realised volatility seen in credit and equity markets; when underlying components are moving in different directions, it is harder to get large shifts overall.
How does the transition come about? Unusually accommodative monetary policy helped to raise linkages between global asset classes. As this policy tightens, it's reasonable that these linkages should decline. But there's another, more 'normal' force at work here: late in the cycle, we've often seen lower correlations.
There are several reasons for this. Deep into an expansion, higher economic confidence reduces the likelihood that many markets will panic at the same time, and means market-specific stories are often bigger drivers than the more binary question of 'recession, or not?'
Sheets then points out that a correlation of this magnitude both helps and hurts investors, and notes that "Any market with diverse drivers has positives and negatives."
On the plus side, it should be a better backdrop for 'macro' trading. The last six years have seen unusually bad performance of macro hedge funds relative to a 60:40 portfolio. There are several explanations, including simply that these funds have made bad calls. But it seems reasonable that it's hard to extract alpha from macro trends when all markets are moving together. With that shifting, the backdrop should be better. Over nearly two decades, hedge fund returns have outperformed a 60/40 portfolio more often in a low correlation environment than a high correlation environment (see Exhibit 6). Lower cross-asset correlations are also good for any globally diversified portfolio, reducing its volatility all else being equal.
And then there is the question of volatility. In an amusing note earlier today, Bernstein strategist Inigo Fraser-Jenkins writes to "prepare for higher idiosyncratic stock-level volatility under a Trump presidency due to “off-the-cuff” tweeting." However, for now the risk of a volatility spike appears remote, as it appears to trail equity correlation.
Ultimately, what happens to cross-asset correlations next will depend on Trump: should his fiscal policy plan disappoint, and the market revert back to the "monetary stimulus" mindset, correlations will spike back up. On the other hand, if the status quo persists, the index may drop even more as all legacy cross-asset linkages are broken. Whether that finally results in some long overdue alpha creation for hedge funds, many of whom recently flipflopped in their outlook of trump, remains to be seen.