There have been a lot of articles written recently about a high average correlation level in the stock market.? I want to take a stab at explaining what it means and implies.
A few notes before I start.? First, remember that cash doesn?t enter or leave the market when we buy or sell.? Cash enters the market when new stocks, bonds, etc., get issued in exchange for cash.? Cash exits the market when stocks, bonds, etc., get retired in exchange for cash through buyouts, maturities, etc.? Second, when we buy or sell, the price changes based on whether buyers are sellers are more motivated to buy/sell the asset and sell/buy cash.? In the short run, even the amount of cash doesn?t change, aside from what the brokers and market-makers scrape off.
Note that this applies to ETFs as well.? Even as they grow, they suck in more of the stocks/bonds that they index, but after fees (more scrape) they are just shells, holding vehicles for assets.
Third, there are two reasons why assets can be highly correlated.? The first reason is that the business performance is geared to the same driver, for example, the expansion of credit.? The second reason is that the current and future ownership has similar motives for each asset, and trade each similarly.? The first is Ben Graham?s weighing machine, while the second is the voting machine.? The second reason is more relevant for what we are experiencing today.
Fourth, remember that correlation is not the same as beta.? Stock A always moves half as much as stock B.? The correlation is 1, but the beta versus B is 0.5.? Just because correlations are high does not mean every stock is moving the same amount.? It does mean that they are almost all moving in the same direction at mostly consistent relative amplitudes.
The preliminaries are done.? The most important aspect of my preliminaries is that we are likely dealing with Ben Graham?s voting machine as the causative factor for the high valuations.
Okay, now think of stocks and other assets as dependent on the time horizons of their investors.??? If the time horizons of investors are predominantly long, correlations on assets should be low in the short-run, because investors don?t make decisions to trade off of short-term macro factors.? But when a large part of the investor base is skittish and is always running to or from the latest bit/byte/bite of data ? that leads to high correlations.
ETFs aren?t necessary for high correlations, but they seem to help the process by creating easy ways for people to implement decisions that are a simple idea.?? ?I want financials, I don?t want energy, buy the long bond, sell gold.?
Thus high short-term correlations indicate a momentum mindset in the investor base.? Momentum investors are the ?weak hands? of ownership.? They don?t have much of a balance sheet, and so their decisions are quick and correlated with short term price action.? The strong hands have balance sheets, or are long-term minded, and can ?buy and hold? or ?sell and sit on cash.?? That takes a lot of fortitude, particularly in the present environment.
In an era of high correlations, I have two things to say:
1)????? When the voting machine is running hot, pay more attention to the weighing machine ? the fundamental values that drive long-run investing.? Pretend you are Seth Klarman, Warren Buffett, or if you can?t imagine that, pretend you are me, and aim for the best over the next three years.
2)????? In general, markets are near short-term peaks when the level of momentum investing is high.?? Volatility tends to be high as well.? Volatility is inversely proportional to time horizon. (I.e., the longer you aim for in investing, the less you care about short-term volatility.)
Thus my conclusion is this: now is a time to pay attention to fundamental values.? Ignore the noise and protect your capital.? I know this sounds too simple, but when correlations get too high, act against the direction of the market.
PS — still don’t have power back from Irene.? Pray for us.? I get my work done at a backup site.
Please comment on whether you believe HFT has impacted correlations. Thanks.
I lack competence to answer that question; I would love to figure that out though.