Thursday, May 7, 2009 0 comments ++[ CLICK TO COMMENT ]++

Correlation between your investment decisions may be higher than you think

Reader John responds to the Gary Shilling interview I discussed last week—unfortunately he seems to have lost his original message he was typing up :(—and makes an insightful point generally missed by many, including me. His point is in regards to Gary Shilling's statement that he got all 13 his predictions correct last year and this had a probability of 1-in-8192. John argues that the probability is probably more like 1-in-2 since all the investments were likely to be correlated:

John:

After reading Taleb's writings, I'm now very skeptical towards anyone claiming credit of predicting the economy. (Can a climatologist claims credits for predicting long range the weather pattern?) Shilling said getting 13 out of 13 right has a chance of 1 out of 8192. I haven't got a chance to read his original 13 recommendations. But one would imagine all 13 recommendations should be strongly correlated, or even perfectly correlated. Hence, the chance of getting all of them right is closer to 1/2 than 1/8192.

A layman, not being very probability illiterate, can have the excuse of getting it wrong. But Shilling, becoming a [scholar], shouldn't make such mistake. Either he was hypocritic or he is incompetent. Both sound very dangerous.


The criticism is kind of harsh since I don't think Gary Shilling gave much thought to the probability but the point is very important to consider in terms of portfolio diversification. I haven't read Taleb—not my type ;)—so I can't support or rebut in any concrete manner.

The key argument John is making is that the correlation between investment decisions one makes is likely to be really high. So even if you invest in, say, 10 securities, you may actually be betting on one or two outcomes. I suspect John, or those following Taleb's thinking, may argue it is the case with everyone but I'm not sure it applies to all types of investing.


Passive Investors & Stockpickers

It's not clear to me this is the case with passive investors who pursue diversification or stockpickers.

Ignoring the potential for erroneous decisions (i.e. mistakenly treating correlated assets as uncorrelated), I would imagine that passive investors pursuing diversification will genuinely have uncorrelated investments in the long run. For instance, if someone invested in timber and common stocks, I think they would be uncorrelated in the long run (even though this wasn't the case in the last decade.) So these type of investors are likely making independent bets.

Stockpickers are probably also making independent bets for the most part. Most stockpickers do not put much weight into their macro views. In fact, many don't even spend any time on it. The returns may seem correlated in the short run—say during a recession—but I suspect the long-term returns will be uncorrelated. A good example to consider is Warren Buffett. Buffett generally avoids betting on macro views* and he has said many times to avoid being influenced by them. So, even though everything may seem correlated in the short run, say during a recession, I'll bet that his investments are more correlated than they seem. So even though his investment in, say, Constellation Energy and Wells Fargo may both track each other during down markets, I suspect they are largely uncorrelated.


Macro Investors Do Make Single Bets

However, macro-oriented investors such as Gary Shilling and most amateur investors (those that do not purposely try to build a portfolio of uncorrelated assets) are likely making single bets. I think if you follow this style of investing, it is absolutely critical that you realize that, even if you buy 5 stocks, you may just be making one decision. This view supports the point John is making of Gary Shilling.

What typically happens with macro investors is that a lot of our predictions are based on a thesis or world view. Gary Shilling's investments are probably more similar to making a yes/no 2-probability prediction on the whole outcome rather than making multiple individual uncorrelated investments. If one turns out to be right, chances are most of the other calls will be correct; if one is wrong, it likely that everything will be wildly off. You can see this in practice by observing macro-oriented investors such as Jim Rogers. If you ignore the short-selling of stocks by Jim Rogers, he likely lost a fortune last year (at least on paper.) Even though he was supposedly invested in differing commodities, China, and so on, they are essentially a single bet.

By the way, I do not think the high correlation between decisions is necessarily bad. It's simply the nature of the investing style. If you pursue a macro-oriented investing style, just make sure you realize that you are not diversified even if you own a whole bunch of stuff.


One Other Thing To Consider: Psychology

On top of everything said, and regardless of the type of investor you are, one also needs to consider psychology and the human brain. I have never studied pscyhology and I'm just guessing here but it is possible that the correlation between investment decisions may end up high due to psychological reasons you may not be aware of. Your brain may subconsiously pick up information from the media, friends, bloggers :), other investors, and so on. Even if you independently come up with a bunch of investments, they may actually be correlated highly because everyone around you was bullish on the economy and all your picks were dependent on strong economic growth.


Shilling's 13 Predictions From 2008

Gary Shilling's predictions were all based on a certain macroeconomic outcome. Namely, he was forecasting a deflationary credit bust—and, to the surprise of almost everyone, it happened! He was predicting the bust as far back as the late 90's but he was finally proven right. Therefore, it shouldn't be surprising that he nailed all 13 last year. He isn't really making 13 investments; it's more like one investment. For reference, here were his 13 predictions last year:

1. Sell or sell short homebuilder stocks and bonds.
2. If you plan to sell your home, second home or investment houses anytime soon, do so yesterday.
3. Sell short subprime mortgages.
4. Sell or sell short housing-related stocks.
5. Sell or sell short consumer discretionary spending companies.
6. Sell low-grade fixed-income securities.
7. Sell or avoid most commercial real estate.
8. Short commodities.
9. Sell or sell short emerging market equities.
10. Sell emerging country bonds.
11. Buy the dollar before long.
12. Sell or sell short U.S. stocks in general.
13. Buy long Treasury bonds.


As you can see, although there are 13 investment suggestions, they are all correlated closely. The probability of all 13 being right was really high. Basically if the credit bust materialized, all 13 had a high chance of occuring.

Having said all that, one has to be impressed with Shilling's skill in nailing those 13. Yes, he was calling the bust as far back as the late 90's but he nailed it in the end. Furthermore, although the correlation between these picks may close to one, one still had to pick the macro scenario correctly. You could have called, say, an inflationary bust and you would have suffered huge losses (mainly because the US$ rallied.) You had to have called 'a bust', 'deflation', and 'a credit implosion' and for that Gary Shilling deserves credit. If you were able to stay solvent during the time, you made a killing in the end.



FOOTNOTE:

* If Buffett was macro-oriented he clearly wouldn't have invested in USG (I believe he was concerned about a housing bubble a few years ago); possibly sold off Moody's (he was warning about questionable practices in the securitization market, and Moody's was an enabler and a profiteer of that market); and he obviously wouldn't have bet heavily on ConocoPhillips near a high in oil prices (even oil bulls were getting nervous at the rapid price increase.) Buffett does make some macro bets once in a while, often with questionable results, such as his bearish bet on US$, or bullish bet on stock indexes by writing put options, or bullish bet on junk bonds by writing CDS.

Tags: ,

No Response to "Correlation between your investment decisions may be higher than you think"

Post a Comment