A Risky Assumption Lurks in Your Portfolio

Your portfolio may be riskier than you think, and it’s a good time to take note.

Hiding in most portfolios is a big assumption about risk — namely, that assets are uncorrelated (or at least less than highly correlated) to each other. So when one asset is down — emerging-market stocks, for example — a typical portfolio relies on some other asset to pick up the slack — U.S. high-yield bonds, let’s say.

Looking at the long-term data, the assumption that different assets chart different paths appears to be true. The following table shows the correlations since 1988 for a variety of assets commonly found in investors’ portfolios. The data show those assets are not highly correlated, implying they don’t move in sync — and, more importantly, they won’t fall apart simultaneously.

There are two problems with this assumption, however. For one, the long-term numbers belie the fact that correlations among assets have risen over the past two decades.

But the bigger problem is that correlations tend to spike when markets are in turmoil. In other words, assets do fall apart at the same time — they just wait for the most inconvenient time to do it.

The 2008 financial crisis is instructive. Investors no doubt remember the way diversification failed them in that period. Even hedge funds, which were still revered in those days for their investment savvy, suffered declines of nearly 20 percent in 2008, according to the HFRI Fund Weighted Composite Index.

Here’s why: Correlations swelled during the financial crisis, as the following table shows. Assets that were supposed to zig and zag raced each other to the ground instead.

Now correlations are on the rise again. As Bloomberg News recently noted, a measure of asset correlations developed by Credit Suisse Group AG and known as the “cross-market contagion indicator” shows correlations have crept up since 2014 and are higher today than at any point since 2008.

That jibes with my own research. For example, the three-year trailing correlation between the S&P 500 and the MSCI EAFE Index — an index of international stocks — hit a post-financial crisis low of 0.69 in July 2015, but has since spiked to 0.85. Another example is the correlation between the Russell 2000 and the MSCI Emerging Markets Index, which touched a post-financial crisis low of 0.3 in June 2015, but has since risen to 0.57.

It’s not entirely surprising that correlations appear to be flashing warning signals. Eight years have passed since the financial crisis; Thursday marked the anniversary of Lehman Brothers’ collapse. The passage of time by itself doesn’t mean another market upheaval is in the offing. But market panics come along with surprising regularity, and if history is any guide we’re due for another one.

More than a few asset prices have been stretched to (and maybe even beyond) their reasonable limits over the past eight years. European and Japanese government bonds spring immediately to mind, and so do U.S. stocks and Treasuries, and corporate bonds and real estate investment trusts. High valuations have often set the stage for corrections.

Before you run out and bury your money in the backyard, you should note some assets have actually held their value — and even thrived — when most other assets fell apart.

The following table, for example, shows the correlations between U.S. government bonds and the assets shown in the tables above since 1988 and during the financial crisis. There was little or no correlation between government bonds and those other assets in either period. The one exception is that government bonds were highly correlated to corporate bonds in both periods, but the correlation actually diminished during the financial crisis.

So rather than rely on correlations to bail them out during the next meltdown, investors should instead ask themselves one simple question: How much of my portfolio am I prepared to put in harm’s way? Then they should move the rest to more solid ground.

Source: Bloomberg Gadfly, https://bloom.bg/2z5D9g5

Leave a Reply

Your email address will not be published. Required fields are marked *