Much of today’s investment advice is built on a concept of Modern Portfolio Theory (MPT), one pioneered by Harry Markowitz. Markowitz was an economist trained at the University of Chicago with the likes of Milton Friedman and won a Nobel Prize in Economics for his work.
MPT focuses on the fact that investors will either maximize their expected return given a desired risk level, or minimize the expected risk level given the desired return. These two are the same concept, just different sides of the coin. The idea of MPT is to create a series of portfolios and then use an efficient frontier to pick your portfolio based on your required rate of return or desired risk. If you go looking for information on MPT, you’re likely to come across some sort of graphic like this:
The general theory was to add asset classes to a portfolio that have low correlation to each other and still have expected positive returns. Nirvana would be a negatively correlated asset class with a positive expected return. Volatility (e.g. the VIX) and equities (e.g. the S&P 500) have a strong negative correlation by definition. Sounds like a great setup for a portfolio of VIX and S&P 500, right? Unfortunately, VIX has a long term negative expected return, which means it’s a pretty substantial drag on the portfolio, even though it performs exactly when you need it to. Imagine an environment where both VIX and the S&P 500 have positive expected returns. That would truly be the ultimate portfolio!
MPT did lead to a vast improvement in the design and implementation of asset allocation and portfolio management for investors. Unfortunately, MPT was only one step in the right direction. And in fact, in key assumptions, MPT is flat out wrong.
What Modern Portfolio Theory Gets Wrong
Modern portfolio theory bases its entire analysis on three inputs:
- Expected Return of each individual asset
- Volatility (as measured by standard deviation) of each asset
- The correlation of each asset to the others
I won’t get into the idea of expected returns. If you can predict the returns of any given asset (but in particular “risky” assets) within a few hundred basis points, even over a 10 year period, then you are a better investor than I and should be very wealthy! For example, the S&P 500 returned a 5.9% annualized return in the 1970s, a 17.5% return in the 1980s, an 18% return in the 1990s, and a dreadful -1% per year in the 2000s. And by the way, those are nominal returns. How are you at predicting inflation?
Regardless, here’s the biggest error in modern portfolio theory. Your risk is not volatility! Your risk is drawdown risk. What’s the difference? All volatility measures is the dispersion of return. However, from an investor’s perspective, volatility only matters in one direction. Imagine a choppy but upward moving market. Volatility could be quite high, but from an investor who owns the upward moving asset, the volatility could in fact represent opportunities. Drawdown, on the other hand, represents a peak to trough decline in the value of an asset. And therein lies your actual risk. Take a look at the below chart of the S&P 500 index (price only, not total return). If you owned the S&P 500, would you rather deal with volatility or drawdown?
Drawdown, and in particular, realized drawdown (there’s a difference), is what represents an actual loss of capital. Drawdown is the destroyer of wealth, not volatility. Volatility, in fact, can be a large creator of wealth if you use it correctly. The goal of any given portfolio should not be to either minimize volatility per unit of return or vice versa; rather, the goal of a portfolio should be to minimize drawdown per unit of return. And there lies another problem related to volatility.
Correlations are not sticky
The motivation behind MPT is having a less volatile portfolio per unit of return. The way this is accomplished is by adding assets with positive expected return and low or negative correlation to one another. The main issue with this is that correlations often do not hold when you need them to the most. And for many asset classes, the correlations can go close to one at the most inopportune times.
For example, let’s use a classic 60/40 portfolio. Most investors would simply use a combination of S&P 500 index for stocks and the Barclays Aggregate Index for bonds. The Barclays Aggregate measures the bulk of the fixed income universe in the US. It includes over 9,300 bonds, and breaks down into roughly (I’m rounding) 40% US government, 30% mortgages, and 30% corporate bonds. All bonds included in the Barclays Aggregate are investment grade; no high yield bonds. Both bonds and stocks have a positive expected return through time. And for the last 15 years, the S&P 500 and the Barclays Aggregate have a negative correlation to each other. A match made in heaven, right?
Unfortunately, once you dig behind the initial findings, things do not look quite as rosy. Due to the greater volatility of stocks, a 60/40 portfolio will be dominated by equity risk (in fact, over 90% of the volatility and almost all the drawdown in this portfolio comes from equities). From 2008 through 2009, the correlation between stocks and bonds went from slightly negative to a positive +.40 correlation! Right when you needed the diversification the most, it wasn’t there. And the Barclays Aggregate is over one-third US Treasuries, bonds that tend to be the destination for a flight to quality.
Heaven forbid you had investment grade corporates or high yield bonds in order to pick up a little extra yield. IG bond went up to a +.60 correlation and high yield bonds to a +.80. Including these bonds instead of treasuries offered you pretty much no diversification at all.
Why Does It Matter?
Drawdown matters more than volatility because we’re human. Personal investing and personal finance is personal, after all. Watching your money evaporate leads to panic and panic leads to bad outcomes. Take a look at the below example. Both return streams have the same annualized volatility and the same annualized return. However, the high drawdown return stream has a max drawdown of 29% (not hard to find), while the second has a max drawdown of 16%.
If you held the high drawdown asset, chances are high that you’d be in a mild state of panic pretty early on. Self-awareness is just another form of discipline. It’s important to try to practice this. If we’re honest with ourselves, I think most investors would struggle to hold onto the high drawdown asset beyond that precipitous drop. Particularly if it was something as central in a portfolio as an S&P 500 index fund. Of course, had you sold, you would have missed a pretty nice recovery.
The Real Danger
If you’re young and do not need your portfolio to live on, chances are higher that you would be able to hold on to assets with higher drawdowns. Although chances are higher, they’re still not high. Again, we’re human.
The greatest danger comes from when you have to make regular withdrawals from your portfolio. I work at an endowment, a theoretical permanent pool of capital. Endowments, to me, are much like a retiree’s portfolio or a pool of family capital. There may or may not be much coming in, but there are definitely withdrawals being made.
Let’s take an example using the same return streams. Imagine that you have a $10,000,000 portfolio and you withdraw $500,000 each year in December. The first thing to notice is that at one point, your portfolio drops below $6.5mm with the high drawdown, before a modest recovery. The second point is much less obvious but extremely important. For the same return and same volatility, the low drawdown portfolio is $300,000 higher at the end of five years. That’s 4.5% higher, or nearly a full percent a year. And this is over only a five year period. Our goal is to focus on extremely long time periods, such as seven, eight, even nine generations. Just imagine what 1% a year can accumulate over 100 years!
What to Do?
What to do about this risk will be the primary discussion in the investing articles in this blog. We’ll get into some specific strategies, but I am still trying to find better and better ways to do this myself. A key strategy will be the use of options in a portfolio, or more broadly, building out positive optionality in the portfolio.
For example, the use of long date call options on equities is an extremely interesting strategy. Because of the inherent downside protection in call options (you can only lose the premium paid), you can feel comfortable taking on more risk. If you would normally keep a 60% or 70% allocation to equities in your portfolio, incorporating long date call options would allow you to go to 90%, 100%, or greater than 100% equity exposure. And because call options allow for the equivalent of term, non-callable leverage, there is no risk in getting stopped out of your positions. Long dated call options, while not cheap, are also not as expensive as many people think (the subject of a future blog post).
Also, the use of different risk premia strategies will also be a key focus. Trend following, for example, tends to have nice levels of convexity on both the left and right tails of the distribution. Particularly on the left tail, this positive convexity leads to increasing negative correlation when you need it the most; a very useful feature indeed.
Modern portfolio theory was a great step forward in the construction of better portfolio management. But like anything else, MPT is based on a series of assumptions that can be, and often are, incorrect. Drawdown risk, not volatility, is the primary concern for investors. Correlations can break down when you need them the most. As we move forward in the low-expected-return world, we must think about and design our portfolios more intelligently.
Keep building my friends.