Volatility Targeting

The narrative du jour

It always amazes me how quickly things change in financial markets. Every day, without fail, the front page of financial media will feature a new story for why asset prices are doing what they do, often linking some overnight news as a catalyst. Not unlike those chalkboards you used to see in front of restaurants touting daily specials, the market narrative seemingly changes every morning and, undoubtedly, will catch someone off guard. That uncertainty in financial market media broadly represents how complicated the entire system has become. Market participants all behave differently and have competing objectives, making it nearly impossible to predict where financial markets are going. Moreover, the rise of computing power has shortened the time to take in all the information and adjust views accordingly. The behavior of these automated strategies can also be counterintuitive in some cases and lead to suboptimal outcomes, which makes the task of understanding financial markets all the more challenging.

The latter part of what I described above remains largely opaque. On the surface, automation is a good thing. Machines are harnessing technology to improve price discovery in the market, which, in theory, should lead to more efficient markets. But, of course, everyone has their ax to grind, and market participants that understand these dynamics shy away from talking about the trade-offs of their activities. Although we'll never know for certain what these trade-offs are, we can say with conviction that market crashes have become more frequent. For example, the chart below shows years in which market volatility over 1-week has spiked above 3-month volatility, indicating a rapid correction in the market. Moreover, we can see that these crashes have become more frequent over the past decade.

Although I think it's fair to say that automation and systematic investing more broadly bring benefits to the broader market during periods of calm, I have a more challenging time believing that these strategies are a net positive for markets during periods of turmoil. I think this because many of these algorithms have embedded risk management frameworks standardized across the industry. For example, if a market falls 3 percent, an algorithm might place a sell order because it thinks the market is crashing and wants to avoid further losses. That's fine if you have one fund doing this, but when hundreds of billions of dollars follow the same investment rules, a market that should have fallen 3 percent can quickly lose 5 percent or more due to algorithms selling. These risk management frameworks create a form of herding, where algorithms rush to get out of positions like people in a crowded theater running to the door when the fire alarm rings. Now, as with anything in finance, there's plenty of nuances here. Over the coming months, I'm hoping to roll out a few posts that break down a few critical systematic investment strategies that I think are important to track and look at where they are similar and where they differ.

With that said, one common objective in most strategies is to reduce portfolio volatility. At the heart of this is the notion of volatility drag, which lowers overall wealth over time. I do a deeper dive on this in Quit dragging me down for those that are interested. Since portfolio managers are concerned about volatility, many will look at their returns and normalize them by risk (The Sharpe Ratio). The Sharpe ratio is calculated by taking the return divided by the standard deviation. Historically, stable risk-adjusted returns were delivered by adding bonds to the portfolio. However, since the Great Financial Crisis (GFC), bonds have done a poor job of delivering high risk-adjusted returns since interest rates are near secular lows. As such, investors have had to go into riskier assets to meet required/desired rates of return. The above dynamic is something I've written about in Searching For Yield for those looking for more details. With investors taking on more risk, the need to manage that risk increased, which has led to plenty of innovations in risk management. The innovation we will focus on today is the volatility targeting portfolio.

Volatility targeting now accounts for around $US400 bln in assets under management.

A volatility targeting strategy is a systematic strategy that targets a constant level of volatility for the overall portfolio. The strategy maintains a consistent level of volatility by dynamically adjusting its asset allocation in response to changes in volatility. For example, when volatility increases (decreases), a volatility-targeting fund would decrease (increase) exposure to the underlying assets. Said another way, these funds buy/sell assets with a negative relationship to volatility. These strategies are typically found in a portfolio of riskier assets since the main benefit of implementing this type of strategy is that it limits severe losses. In practice, most funds will use these techniques to gain exposure to equity markets, although funds also invest across asset classes using similar rules.

The main benefit of volatility targeting strategies is that they reduce tail risk. They do this by relying on the fact that volatility in financial markets tends to cluster. When I talk about clustering, I mean that if volatility is high today, there is a reasonably high probability that volatility will also be high tomorrow. As such, using a volatility targeting strategy, investors can reduce exposure during periods of high volatility and reduce volatility drag across time. As such, these strategies tend to outperform the broader index on a risk-adjusted basis; however, they will lag on outright performance. In addition, the more astute readers might have caught on that volatility targeting looks to predict the distribution rather than the direction of returns.

Volatility targeting has grown in popularity over the past decade. Part of the increase is due to the macroeconomic environment, which has challenged pension and insurance companies to find ways to earn predictable, stable returns. Analysts estimate that the AUM for these types of strategies ranges from US$250 to US$400 bln. However, their AUM is likely larger since other systematic investing strategies can use Volatility targeting as an overlay. For example, long-short hedge fund strategies often use volatility targeting to dictate how much leverage they use.

Volatility targeting can be destabilizing for financial markets.

The buying/selling from volatility targeting funds tends to be pro-cyclical. From the above, we know that buying/selling of volatility targeting is negatively correlated with volatility. As I've discussed in Masters of the crayons, volatility increases with large price movements. However, volatility tends to accelerate faster with falling prices compared to price increases. That's because, on average, the S&P500 tends to increase more than it decreases. Therefore, when markets are falling, volatility rises, forcing volatility targeting strategies to sell into already falling markets. Technically, the same dynamic happens when markets are rising rapidly; however, in this case, volatility targeting strategies act as a market stabilizer as they sell into rising markets.

Prolonged periods of low volatility also lead to more violent selling from volatility targeting strategies. Unfortunately, the amount that volatility targeting strategies buy/sell in response to changes in volatility is not linear. The chart below shows the allocation for a given level of volatility.

We can see that the allocation increase exponentially as volatility approaches the lower bound. Historically, the allocation for a 10 percent volatility targeting strategy hit as high as 190 percent when volatility was 5 percent in 2017.

The bottom line is that any fund that uses a volatility targeting framework depends on market stability to deploy capital. That means that when times are good, they are active buyers, and when times get rough, they are active sellers. An increase in stock market turbulence can be made worse by volatility targeting funds, which would sell their positions when the market is falling. Understanding these dynamics is important because, under the right conditions, volatility control funds could extend market declines.

Where are we today?

Allocation in volatility targeting funds sits at around the 35th percentile going back to 1960. Exposure was as high as the 80th percentile over a month ago. However, the recent bout in market volatility due to the Evergrande debacle, rising inflation, and slowing growth have dented sentiment. The chart below shows exposure to the S&P500 going back to 2017. We can see that exposure remains very low.

Volatility targeting funds will likely be buying in the coming weeks. The chart below shows a simulation of 1000 different return streams over the next month based on the returns seen over the past year. Based on this analysis, volatility funds should increase allocation to the S&P500 by an additional 10 percent, which equates to around US$25bln in buying over the coming month. Of course, the lower end of these estimates, a drop in 20 percent exposure, would be triggered by a decline of just 3 percent and would equate to roughly US$50bln in selling. Most of the buying from these funds should come next week, as more volatility periods are falling out of the windows used to estimate their volatility.

Thanks for reading,

Tiago Figueiredo