Backtest Analysis

I have laid out the discourse and theoretical framework that led me to the Permanent Portfolio but now I would like to focus on its historical performance. Before exploring the numbers, it is important to note that historical backtesting has its limitations. One cannot rely strictly on backtesting to generate asset allocations for a portfolio. Without a theoretical framework for why you are choosing the assets, it is very easy to get burned with a portfolio that worked brilliantly until it didn't.

However, historical backtesting still has its merits. Although it cannot help us perfectly predict the future, it helps us better understand the possibility of given outcomes.

Our backtesting begins in 1972 because before then it was illegal for US investors to own gold. Long-term Treasury bonds are represented through the Vanguard Long-Term Treasury Fund. Although the average maturity of 22 years is less than recommended for the PP, the historical data for this fund is easiest to access. Stock returns are measured by the return from the Vanguard Total Market Fund. For ease of calculation, the returns assume annual rebalancing back to the 25% weightings. When inflation adjustments are made, it is according to the Urban Consumer Price Index.

First, let’s break down the compound annual growth rates of the asset classes.

  • Stocks 9.92%
  • Gold 7.24%
  • Long bonds 8.06%
  • T-Bills (Cash) 5.5%

Without knowledge of how the asset classes interact with each other, one might assume that you could simply do an average of these four growth rates to find out the Permanent Portfolios growth rate. After all, the Permanent Portfolio equally weights these four asset classes. Running this calculation gives a growth rate of 7.68%. However, the PP did much better than this.

In reality, the Permanent Portfolio returned 9.22% annually. How can this be? The first calculation, simply taking an average of the returns, is treating the asset classes in isolation. That calculation would track the return of a Permanent Portfolio that was never rebalanced. The four asset classes would have been bought and then never touched again. However, the Permanent Portfolio calls for periodic rebalancing events and an investor would have achieved the 9.22% growth if they had rebalanced annually, bringing the assets back to their 25% weighting each year.

Rebalancing is far from magical, and with many portfolios the effect is not as dramatic. Rebalancing works exceptionally well with portfolios that are well diversified by holding asset classes that are relatively uncorrelated. With a properly diversified portfolio, losses are minimized because it is rare for all of the assets to depreciate at the same time. Minimizing losses is crucial because steep losses kill the effects of compounding interest. A portfolio that loses 50% of its value, a common occurrence in the 2008 crisis, would have to appreciate 100% just to break even. By minimizing losses through diversification, the Permanent Portfolio gives investors firepower to buy assets after they have fallen in value. As far as low correlations go, the Permanent Portfolio has been nearly perfect, holding four asset classes that have nearly zero correlation with each other.

Finding assets with zero correlation is the holy grail of investing. Many investors who understand the power of correlations will even buy an investment with dismal return prospects, if it can offer desirable correlations to a portfolio. In fact, most of us do this in real life without even realizing it. Let’s take a look at a fire insurance contract. In isolation, it is a horrible investment. It pays a negative yield and it has a negative expected return for you as buyer, otherwise the insurance company wouldn’t sell it to you. In isolation, it is a poor investment. However, in the context of your life it is a no-brainer because of its negative correlation. You pay a premium for a contract that has a negative correlation with the value of your house and belongings if a fire occurs. That negative correlation protecting you from catastrophe is so important that you will pay a premium each year.

I never know which malaise will afflict my equities, so I have taken out insurance against the major macro economic problems that they will face. I see gold, bonds, and cash as insurance-like instruments that protect my stocks from economic malaise. Many investors acquire portfolio insurance through the use of options but this does not interest me because options are a zero sum game and after commissions a losing proposition. Unlike traditional insurance, cash, gold, and bonds have all offered real returns over the last 40 years while still offering protection.

All of this data is best synthesized graphically…

Despite the volatility of the individual components, the portfolio itself provides consistent inflation adjusted returns. It is important to note that every asset class in the portfolio has had times of strengths and times of weakness. No matter which asset class was doing well or poorly, the PP performed.

Finally, it is important to emphasize that these correlation coefficients are driven by massive fundamental differences in the asset classes. These are not the kind of coefficients that all converge to one during a crisis, a sign of pseudo-diversification reached by slicing and dicing up equities and combining them with credit risk laden bonds. The macroeconomy is the causation that is driving these correlations and the PP acknowledges that from the start with its theoretical framework.


1 comment: