God is a Capitalist

Tuesday, March 18, 2014

Markov Confirms ABCT

Greg Davies and Arnaud de Servigny offer a different take on diversification in their book Behavioral Investment Management: An Efficient Alternative to Modern Portfolio Theory. Chapter 6, “Representing Asset Return Dynamics in an Uncertain Environment was the most interesting chapter to me, and the one that adds confirmation to using the ABCT as a guide to timing the market. 

Modern portfolio theory tells investors to diversify their portfolios at least between two asset classes, stocks and bonds. A simplistic summary of the method is to use the statistical measure called standard deviation to assess the risks of asset classes and diversify according to risk. But in reality, advisers have found that a fixed ratio, say 70% stocks and 30% bonds, often works better without requiring as much work. 

The point of diversification is that we don’t know what the future will hold so we try to prepare for the most likely contingencies. As Warren Buffet once said, “Diversification is a protection against ignorance. It makes very little sense for those who know what they’re doing.”[1] As the market cycles between escaping the claws of bears to riding the bulls, or running with the bulls, one asset class will lose investors money while the other will make up for the losses, hopefully. 

BTW, diversification is one of the main reasons your portfolio will never make the same returns as an index like the S&P 500. If you are diversified, gains in the index will be offset by the losses in bonds. That’s the tradeoff cost for increased security.

The authors of Behavioral Investment Management describe four approaches to diversification across assets:

  • 1.       Unknown pattern - fixed ratios of say 70% stocks and 30% bonds.
  • 2.       Through-the-cycle stability – using statistical techniques such as regression and correlation analysis to determine past risks and returns and assuming the pattern will remain the same in the future. This is modern portfolio theory.
  • 3.       Regime switching – determining if an asset class is in a bear or bull market and adjusting the allocation accordingly. For example, if it appears stocks are in a bull market the allocation may jump to 90% stocks.
  • 4.       Time adaptive dependency – Don’t worry about the regime. Rebalance often, say every month.

The “Unknown pattern” method above is more robust (produces better results) than modern portfolio theory because it avoids the error of over fitting. That error happens when advisors use a data set of historical returns to fit a statistical model that determines risk and returns from the past, but the portfolio fails to achieve the results predicted by the model because the future didn’t unfold exactly as the past events on which the model was based. 

The authors settle on method #4 above for practical reasons, but for me #3, Regime switching, was the most interesting. Introducing the method, they write

A regime-switching model assumes the existence of different states of the world. These states of the world are discrete, and they are not directly observable through a clear, recognizable variable... Assuming that this framework is a fair representation of reality, the questions we are left with relate to identification of the state of the world we are currently in, as well as to an understanding of the dynamics that will enable us to define the most probable state over the next period...the state dynamics follows a first-order Markov process.
The authors test a two-state theory of past risks/returns of the S&P 500, bonds and cash. They find two clear states of the markets existed historical data over the past decades:

The first regime is typical of a bull market, where the average equity return is positive, its volatility is low, and the correlation with bonds positive, whereas the second regime can be associated with a bear market, where the average equity return is negative, its volatility is higher, and the correlation between equities and bonds is expected to turn negative. 
Then the authors rebalance their portfolios according to the regime that their model predicts will happen next. Their regime-switching model tells them to be fully invested in stocks during bull markets and fully in bonds during bear markets. Imagine that! 

What the authors don’t discuss, because they haven’t read my book Financial Bull Riding is that the market switches between bear and bull regimes because of the business cycle. Fed monetary pumping unleashes the bulls while recessions call out the bears. The authors advocate the fourth method of frequent re-allocation as the practical alternative to the complexity of the regime-switching process. The re-allocation is based on the most recent, short term risk/returns analysis.

But frequent re-allocation will drive up expenses and fees. An alternative would be to learn the Austrian business-cycle theory well and determine to buy most of your stocks during the depths of a recession and stay with them until the recovery gets long in the tooth. Then frequent re-allocation from stocks to bonds will save you from the crash that will come soon but no one can predict exactly when.

[1] Greg Davies and Arnaud De Servigny, Behavioral Investment Management: An Efficient Alternative to Modern Portfolio Theory, New York: McGraw-Hill, 2012, 131, quoting Buffett from investor meeting in 1996.

No comments: