Correlation And Meaning

We’re all familiar with the old saw: “correlation is not causation.” Correlation is merely a statistical measure of the strength of the linear relationship between two variables. Correlations can change over time. The fancy stats word for this instability is “nonstationarity.”

Anyway, what I want to suggest in this post is that correlations can often be interpreted as markers of meaning.

For example, stocks and Treasury bonds have been negatively correlated since the financial crisis. The reason is that the meaning of Treasuries to investors, broadly speaking, is “safe haven asset.” A Treasury allocation is an allocation that will perform well in a deflationary environment. And deflation, broadly construed, has topped the list of investor fears for many years now.

A big mistake many investors (particularly younger investors) may make is assuming Treasuries will always be negatively correlated with stocks. This has not always been true historically and will not necessarily be the case in the future. Why? In a highly inflationary environment, both stocks and Treasuries will perform poorly. The two assets classes will become positively correlated.

Another example of this is gold. Traditionally, gold has been viewed as an inflation hedge and has been positively correlated with inflation expectations. These days, however, gold is liable to trade up on deflation fears as well as inflation fears. Why the change in correlation? The meaning of gold has changed. Gold has shifted from a pure inflation hedge to an insurance policy against economic chaos more generally. Gold is now a hedge against policy mistakes by our economic elites (our Ever Wise and Benevolent Central Bankers in particular).

What I’m driving at here is that if you want to better understand the nonstationarity of correlations, you ought to spend some time thinking about narrative.

A stable correlation is a correlation where objective meaning dominates. Objective statements can be proven true or false in a straightforward way. Unstable correlations are correlations where subjective meaning dominates. Subjective statements cannot be proven true or false in a straightforward way.  Subjective statements are reflexive.

George Soros described it this way:

Consider the statement, “it is raining.” That statement is true or false depending on whether it is, in fact, raining. Now consider the statement, “This is a revolutionary moment.” That statement is reflexive, and its truth value depends on the impact it makes.

There’s not much subjective judgement required to evaluate a Treasury bond as an investment. It’s a mostly objective process that more or less boils down to your views on the future path of inflation and interest rates.

Now, your views on inflation and interest rates may make Treasury bonds seem relatively more or less attractive to you at any given point in time. But there’s never any real question in anyone’s mind as to how Treasury bonds will perform in a deflationary environment versus an inflationary environment. This is what I’m driving at when I say the meaning of a Treasury bond for your portfolio is going to remain pretty stable over time. A Treasury bond is protection from deflation.

Credit is a bit more subjective than Treasury bonds because now you’ve got defaults and recoveries in the mix. And equity valuation is far more subjective than credit valuation because the timing and amounts of the cash flows associated with equities are highly variable.

The value of gold is an order of magnitude more subjective than even equities because there aren’t any cash flows associated with gold. Gold is a purely speculative asset. Gold has value because, for whatever reason, human beings have arrived at the collective consensus that it’s a store of value over tens of thousands of years.

At the extreme end of this spectrum you have something like Bitcoin. Bitcoin, of course, has no cash flows. On top of that, there’s no broad consensus regarding what Bitcoin means. Sometimes it’s a currency. Sometimes it’s a speculative risk asset. Sometimes it’s a store of value or even a safe haven asset.

You ought to be extremely skeptical of any MPT-style analysis of Bitcoin’s role in a portfolio at this point. You simply can’t know if, when or how its correlation with other portfolio assets is going to stabilize over time. Just because Bitcoin is uncorrelated today doesn’t mean it will continue to be uncorrelated in the future.

Another practical application of all this relates to factor investing.

Patterns of correlations are the building blocks for factor-based investment strategies (they are literally what the math going on under the hood is measuring). It’s well-known that factor strategies go through extended periods of outperformance and underperformance that are difficult, if not impossible, to time. Factor performance comes and goes in irregular regimes. Regimes are driven by a mixture of objective and subjective factors that influence one another in feedback loops.

If you’re trying to figure out when the relative underperformance of value stocks will end, you need to be thinking about what in the zeitgeist and market regime needs to change so that investors will want to buy stocks with “value” characteristics (how you choose to define “value” is important here). For example, in late 2016 the election of Donald Trump triggered a massive rally in cyclical industrial and financial services stocks. If you’re a long-suffering, old-school value investor who owns a lot of these stocks, what you want at a high level is higher growth, (modestly) higher inflation and (modestly) higher long-term interest rates.

If you’re a growth-oriented investor, such as a VC, who owns unprofitable, high-growth businesses that will not generate free cash flows for many years, what you want is a regime with solid growth but even more importantly with low inflation. More specifically, low interest rates. The value of your equity ownership is extremely sensitive to the cost of capital because your investments are very long duration. Much like a zero coupon bond, their cash flows lie far out in the future.

So anyway, when you’re considering factors such as value and growth what you want to be thinking about when evaluating their potential persistence over time are the drivers of the underlying patterns of correlation. And if you go through this exercise enough, I think you’ll find you keep coming back to investor preferences for different cash flow profiles.

As my friend Rusty Guinn once wrote:

Investment returns are always and everywhere a behavioral phenomenon.

Because, in the words of Marty Whitman, we’re pretty much always looking for a “cash bailout” when it comes to our investments. And our ability to exit an investment almost always ends up depending on a sale to another party. Marty wrote a wonderful explanation of this in an old investor letter (I’ll end on this note because I think it’s a fitting conclusion for this post):

From the point of view of any security holder, that holder is seeking a “cash bailout,” not a “cash flow.” One really cannot understand securities’ values unless one is also aware of the three sources of cash bailouts.

A security (with the minor exception of hybrids such as convertibles) has to represent either a promise by the issuer to pay a holder cash, sooner or later; or ownership. A legally enforceable promise to pay is a credit instrument. Ownership is mostly represented by common stock.

There are three sources from which a security holder can get a cash bailout. The first mostly involves holding performing loans. The second and third mostly involve owners as well as holders of distressed credits. They are:

Payments by the company in the form of interest or dividends, repayment of principal (or share repurchases), or payment of a premium. Insofar as TAVF seeks income exclusively, it restricts its investments to corporate AAA’s, or U.S. Treasuries and other U.S. government guaranteed debt issues.

Sale to a market. There are myriad markets, not just the New York Stock Exchange or NASDAQ. There are take-over markets, Merger and Acquisition (M&A) markets, Leveraged Buyout (LBO) markets and reorganization of distressed companies markets. Historically, most of TAVF’s exits from investments have been to these other markets, especially LBO, takeover and M&A markets.

Control. TAVF is an outside passive minority investor that does not seek control of companies, even though we try to be highly influential in the reorganization process when dealing with the credit instruments of troubled companies. It is likely that a majority of funds involved in value investing are in the hands of control investors such as Warren Buffett at Berkshire Hathaway, the various LBO firms and many venture capitalists. Unlike TAVF, many control investors do not need a market out because they obtain cash bailouts, at least in part, from home office charges, tax treaties, salaries, fees and perks.

I am continually amazed by how little appreciation there is by government authorities in both the U.S. and Japan that non-control ownership of securities which do not pay cash dividends is of little or no value to an owner unless that owner obtains opportunities to sell to a market