Let us endulge in a little gedanken experiment: It's July 2006, you have worked hard over the last years, had a little luck here and there and you have now saved up $250k. You wonder whether that might suffice to retire early and live a modest but very happy life on Bali, where the cost of living is still fairly affordable. You do some back-of-the-envelope calculation and find out that you need about $1,750 per month to live comfortably on Bali, or $21k per year. That means that as long as you can invest your savings at an annual rate of 8.4% you won't lose any money while the interest pays for your cost of living. This number is at least close to what the S&P 500 index has returned in the longterm, but you are worried that it may not suffice going forward. However, you have recently heard of leveraged ETFs that return a multiple of the S&P 500 index returns. That sounds perfect, as it will give a bit a leeway to your investment while still covering your expenses. But is it really that easy?
In this section, we will explore the return profile of leveraged ETFs and show that longterm returns depend heavily on how you want to make use of your profits. In particular, we will simulate and discuss that difference between just buying and holding for decades and paying yourself a monthly salary from the profits of your investments. We will find that regular withfrawals require us to be much more risk-averse compared to a pure longterm investor! Finally, we will take a sneak peak at an active investment strategy that uses leveraged ETFs to see that leveraged ETFs can actually allow for regular withdrawals if risk is properly managed.
As a first example of a leveraged ETF, we take a closer look at SSO, the ProShares Ultra S&P 500 ETF that seeks 2-times the daily performance of the S&P 500 index (before fees and expenses). Below, we load the data for SSO and SPY (the unleveraged ETF that tracks the S&P 500 index) to directly compare the two:
Ok, so yes, the 2x leveraged ETF indeed outperforms the unleveraged ETF, but not really by a factor of two. So where does the factor two come in? Well, leveraged ETFs like the ProShares Ultra S&P 500 say nothing about the long-term performance; instead, they seek 2 times the daily performance of the index. If the S&P 500 index goes up by 1% on any day, the leveraged ETF will gain 2%. If the S&P 500 falls by 2%, the leveraged ETF will fall by 4%. We can easily visualize this relation by dividing the daily percentage returns of SSO by the daily percentage returns of SPY, and then showing a histogram of these ratio values:
As we can see, the ratio values concentrate tightly around a value of two, the expected daily leverage factor. However, there are outliers, that means days when the leveraged ETF failed to realize the exact multiple of the S&P 500 returns. This mismatch occurs because the ETF does not actually borrow money and invests it in an unleveraged S&P 500 index. Instead, leveraged ETFs use derivative instruments such as futures and swaps to obtain a leveraged position at low trading cost (c.f. previous section on the nightly SPY ETF). But because these derivatives are traded on their own markets and have their own dynamics (especially during times of high market volatility), the realized returns may deviate quite a bit from the target returns.
Still, even with these tracking errors, seeing the leveraged ETF outperform SPY makes us reasonably confident that we will be able to live happily from the profits that it will generate. As a first test, we simulate buying $250,000 worth of SSO in July 2006 and then simply holding it:
Watching these long-term gains in portfolio value, we are very optimistic regarding our early retirement. But in fact, we don't plan to just buy and hold SSO, we want to live off of the profits. This means that returns do not fully compound anymore as we spend part of the profits for living costs. Even more importantly, rent is still due in times when our investments lose money, so losses will always be larger than the pure investment losses. How this affects our investment decisions can be best (or at least most easily) described by a simulation.
To simulate the effect of recurring withdrawals, we write a short Python function that simulates how the value of our portfolio evolves over time. This simple simulation uses a for loop to iterate over all trading days of our investment period (starting in July 2006) and handles two basic processes:
After defining the simulate function, we set the two parameters that everyone who has thought about returing early has pondered about:
We run this simulation twice, once investing in the leveraged SSO, and once investing in the more conservative, unleveraged SPY. Finally, we plot the value of our portfolio (with withdrawals already subtracted) over time:
So how have we done living on Bali so far? Well, the results are quite surprising: Both portfolios actually lost money, so taking out $1,750 per month was overstretching the returns of the portfolios. Probably even more surprising is that the leveraged SSO portfolio went bankrupt around 2017, while the unleveraged SPY portfolio survive and seems somewhat stable in value since 2009, albeit at a lower remaining value of a bit more than $100,000. How can we reconcile this result with the fact that the leveraged SSO ETF clearly beat the unleveraged SPY ETF in the longterm?
The take-away message here is that the average return is not the most important metric for investors who wish to withdraw regularly from their account. Rather, these investors should look at metrics such as maximum drawdown that describes the worst loss in the history of the investment. If you want to live off of your profits, you will also need to withdraw money after the investment has suffered a significant loss, and that additional "loss" due to the withdrawal will prolong the way back up significantly. If further withdrawals need to happen before recovery, the portfolio value decreases in an unsustainable manner. In our case, the Great Financial Crisis of 2007/2008 decided that fate of the SSO: as the loss was larger than for the SPY, it left the SSO portfolio with too little capital to sustain further withdrawals in the longterm.
While this example surely is oversimplified (we did not account for inflation, and for the fact that one could also adjust one's spending to the state of the portfolio, etc.), it illustrates an important distinction between pure longterm investors, and investors that want to make use of their profits outside of reinvesting them: if you want to to live off of your profits, you need much more stable returns and avoiding steep drawdowns becomes the top priority. In our example here, the returns of SPY are just stable enough to sustain the withdrawals, but at the cost of having no longterm capital growth at all anymore.
So far, it looks like leveraged ETFs are not the right investment vehicle for investors who regularly withdraw money from their accounts. However, we have learned that it's all about avoiding drawdowns, so if we can actively manage our investments is such a way to avoid very steep, unsustainable drawdowns, then we will likely benefit from leveraged ETFs. The following portfolio equity curve was created by an active investment strategy that dynamically allocates capital to three leveraged ETFs: SPUU (2x SPY; stocks), UBT (2x TLT; bonds), and UGL (2x GLD; gold). The strategy changes its allocation weights to these three ETFs dynamically in response to changes in market dynamics. We will discuss this model in detail in Chapter 3-5 of this course, but for now we simply load the result:
As you can see, this active strategy vastly outperforms investments in SPY or its leveraged ETF SSO (it requires trading weekly, and assumes trading costs + slippage of 0.1% per trade; more details in Chapters 3-5). Let’s see how much we could have withdrawn from this portfolio on a monthly basis while retaining its value over time:
Equipped with this active trading strategy, you can afford to withdraw up to $5,000 per monthwhile retaining (and even slowly growing) your initial capital. This example nicely shows the power of a diversified, but actively managed portfolio (sometimes also called "smart beta" investment). For the details behind this portfolio, check out Chapters 3-5, in which we introduce classical as well as more modern ways of portfolio optimization, regime-switch detection methods, and more.