Volatility Matters: More an investor’s friend than foe

Volatility Matters: More an investor’s friend than foe

Market volatility has risen significantly over the last few years, with many factors driving the extremity of events that can very quickly shift market dynamics and the sensitivity of markets. It is critical to manage the risks associated with heightened volatility while still achieving the returns that investors desire. 

Volatility Matters
Muntu Mdwara

Muntu Mdwara

Quantitative Analyst, STANLIB Multi-Strategy

Key takeouts
  • Volatility targeted strategies are defensive risk strategies that vary the level of equity exposure in a structured way in response to market volatility.
  • We find tactical volatility (risk) management is superior to conventional /systematic volatility (risk) management, with improved risk-adjusted returns and decreased tail risk. We take a tactical approach to this type of risk managed strategy as a team.
  • Managing Volatility Risk is not a “one-size-fits-all” approach, but it is just as important as any other part of investing.
Executive Summary
  • Market volatility has risen significantly over the last few years, with many factors driving the extremity of events that can very quickly shift market dynamics and the sensitivity of markets. It is critical to manage the risks associated with heightened volatility while still achieving the returns that investors desire. 
  • Investors know there is ‘no free lunch’ and the best way to achieve more return is to take more risk (normally measured as volatility). The characteristics of volatility mean some times can be better than others to scale up risk. For this reason, Volatility is one of the key lenses in our Tactical Asset Allocation process to capitalise on market opportunities based on volatility dislocation or boosting risk management (e.g. hedging).
  • Volatility-controlled/targeted strategies are a good example of an investment approach designed to meet the demand (especially after the Global Financial Crisis (GFC) and COVID pandemic) for better defensive risk strategies to deliver smoother return outcomes for investors. These strategies vary the level of asset or factor risk in a capital-efficient way to generate superior risk-adjusted returns through a smoother return profile. This provides lower asset volatility and less severe tail loss, without incurring the extra costs associated with pure tail-risk protection.
  • To illustrate the benefits of this tactical approach, we model a tactical volatility-controlled/targeted strategy for SA Equities which takes advantage of hedging based on the clustering of volatility within regimes. We find such tactical approaches are better than conventional/systematic strategies. This is because they are not continuously de-risking /re-risking, which lowers transaction costs, while remaining fully invested in conventional times.
Introduction

Uncertainty in asset behaviour is a concern for most investors, especially those more loss-averse or who have a liability-driven mindset in that they seek higher returns without the related elevated degrees of volatility, drawdowns or cyclical swings. As expected, the impact of the COVID pandemic has amplified this concern. Institutional investors and pension funds are experiencing pressure from their stakeholders to find better methods of limiting risk. This means downside protection strategies have again assumed importance.

 

Volatility targeted strategies are a potential solution for investors seeking smoother return outcomes (but not a lower return over time), particularly when markets are uncertain and unstable. These strategies are designed to scale down riskier assets in times of high market volatility, but ramp up exposure when volatility is low. Volatility targeting techniques can reduce the risks, while retaining much of the gains to be had from “risky” assets like equities. We believe that, although the results will unavoidably be novel, the key is that risk-adjusted returns ought to be significantly better, as volatility targeting will change the portfolio’s risk and return characteristics. Importantly, volatility targeting is a broad risk management strategy, and does not replace traditional tail-risk strategies (such as hedging with options).[1]

 

The STANLIB Multi-Strategy team is very risk-conscious and we believe volatility regimes matter, as does the path dependency taken to generate returns. We have observed the impact of investors responding too slowly to volatility regime changes, resulting in their portfolios being over-risked for too long, or being too concentrated. The challenge of responding a lot quicker to volatility is that this process of scaling risk positions can add trading costs and have undesirable portfolio turnover.

 

To manage this tension, we discuss tactical volatility strategies or conditional volatility strategies based on a volatility regimes-switch model to give us a better signal to opportunistically scale-up risk in low volatility regimes and to dial down risk when the regime switches to high volatility (Dion Bongaerts, Xiaowei Kang & Mathijs van Dijk (2020).[2]

 

Our contribution to research literature on the impact of volatility strategies across various asset classes is in two ways. Firstly, we test the strategy implementation of tactical and conventional volatility targeting in the South African equities space. Secondly, we document the improvement of the risk return characteristics of existing volatility targeting techniques with the tactical regime timing method for South African investors. The results displayed in this paper are from a hypothetical back-test for illustrative purposes only, and do not currently represent a STANLIB Multi-Strategy product or strategy. We found that the tactical volatility strategy produced better risk-adjusted returns than conventional methods, although both methods lower asset volatility.

 

How do they work?

Volatility strategies vary the level of underlying asset exposure in response to a volatility signal to manage portfolio risk. In practice, this can be done with derivatives, like futures or options, to manage costs and avoid the need to trade the underlying shares.

 

Conventional/traditional volatility strategies are straightforward volatility management strategies that forecast future asset volatility (by using historic realized volatility) and then dynamically adjust the asset exposure to target a set level of volatility. The strategy involves a rules-based implementation, which we call a systematic approach. Using a daily measure of volatility to predict the portfolio’s volatility against the set target level, the portfolio manager then increases or decreases the asset  exposure of the portfolio. This ensures that the exposure does not go above the risk target. For our purposes, the strategy does this by holding equities and a risk-reducing asset, normally cash, so that the predicted volatility is in line with the target volatility (Harvey et al. (2018).

 

Dion Bongaerts, Xiaowei Kang & Mathijs van Dijk (2020) extend this by introducing a concept called “conditional volatility targeting strategy”, also known as a tactical volatility targeting strategy where they reduce/increase equity risk exposure in high/low -volatility states and maintain an unscaled exposure in medium-volatility states. They forecast the volatility states (regimes) by sorting the historical realized volatility into quintiles. This conditional approach reduces transaction costs, and avoids missing out on equity market returns in normal volatility states (so as not to constrain the portfolio).

 

To illustrate this strategy for SA Equities, we take a similar approach to Dion Bongaerts, Xiaowei Kang & Mathijs van Dijk (2020), but with a slight caveat. We estimate the volatility regimes (states) differently. We apply a Markov Regime -Switching Model (Hamilton, J.D. (1989)) with the estimated volatility using the GARCH- Model (Tim Bollerslev, 1986) to determine the various volatility states. Predicting volatility states can be an exceptionally complex matter. In this article, we do not provide in-depth articulation of the Markov Regime -Switching Model, but focus on the overall overview and implementation of the model.

 

The Markov Switching Dynamic Regression model is a type of Hidden Markov Model that can be used to represent phenomena in which a portion of the phenomenon is directly observed while the remainder is ‘hidden’. The hidden part is modelled using a Markov model, while the visible portion is modelled using a suitable time series regression model in such a way that the mean and variance of the time series regression model changes, depending on which state the hidden Markov model is in.

S-state Markov process

Figure 1 depicts the various predicted volatility states for the South African Equity market (ALSI) using the Markov Switching Dynamic Regression with the predicted volatility from the GARCH Model.

The purple area shows the probability at which you could be in a high volatility state in the next month. In early 2012, the market softened after the GFC aftershocks, and the model predicted it was time to re-leverage. In this analysis we used a high probability threshold to identify a switch to a regime. In practice, a less mechanical threshold could be used, as an investment manager may be using other signals to assess the risk regime.

Figure 1: ALSI Volatility Regimes

What volatility targeting can achieve

To demonstrate the potential benefits of volatility targeting in the South African context, we use an example based on the FTSE/JSE All Share Index (ALSI) from 1999 to 2022, using daily data converted to monthly data.  This index has realized volatility of 17.3% (annualized) over the period, so we are using the long-run volatility of 15% as our risk target in our back-test. The equity exposure is capped at 100%, as many investors are restricted from leverage, e.g. those invested via CISCA or unit trust funds. We account for a trading cost of 8 bps, which is conservative in our back-test. This allows for slippage and other transaction costs when de-risking out of equity by moving a portion to cash. Our cash return is based on three-month JIBAR. In practice we would deploy derivative instruments to manage equity positioning to reduce costs instead of selling assets outright.

Figure 2:  Equity Exposure for various volatility strategies

Figure 2 shows the variation in equity exposure between the two approaches. In both cases the invested equity allocation over the back-test period ranges between 22% and 100%. There is a much higher variability in equity levels than is seen in traditional SA equity funds, as the effective equity exposure cap of 100% in a pure equity portfolio limits the benefits of scaling in a low regime. A multi-asset setting would be more effective , as you can scale up risks due to the leverage effect (e.g. move from 60% equity to 75% equity, depending on the balanced fund’s risk profile). The lows are around the time of the GFC (2008) & COVID-19 pandemic, when volatility spiked and equity exposure was ratcheted down to its lowest point, precisely to limit the damaging effects of increased market volatility. The regime model is anticipating volatility clustering, when times of high volatility will generally be trailed by periods when risk stayed high. In Figure 2, another observation is the difference in the average exposure, where the traditional approach averages 88%, but the tactical approach is higher, averaging 92%. In recent months, the conventional volatility targeting is de-gearing, where tactically it has not.

 

There is an implementation contrast between conventional volatility target strategy versus tactical strategy equity allocation, as the tactical volatility strategy reduces/increases equity risk exposure in high/low -volatility states and maintains an unscaled exposure in medium-volatility states. In this respect, the key characteristic of the tactical volatility targeting strategy is that it is not a strategy that continually holds a risk-off position. Only a pick-up in volatility triggers de-risking the portfolio. These strategies are better than a buy and hold approach, providing a dramatic improvement in the tail-risk measured by maximum drawdowns, and a better risk-adjusted return over the whole period. This characteristic is evident in Figure 3, where the tactical strategies improve upside participation, outperforming the conventional strategy by 1.2% on an annualised basis. The transaction cost is a big drag on performance for the conventional strategy, due to continuous up/down scaling of equity position.

Figure 3: Risk-Return analysis

Past performance is not a guarantee of future results. Data is from 31 Jul 1999 to 30 April 2022. The data displayed is a hypothetical example of back-tested performance for illustrative purposes only and is not indicative of the past or future performance of any STANLIB Multi-Strategy product. Back-tested performance does not represent the results of actual trading but is achieved by means of the retroactive application of a model designed with the benefit of hindsight. Actual performance results could differ substantially, and there is the potential for loss as well as profit.

 

In both cases, it is important to note that returns will differ because the allocation to risk assets varies over time. As a result of managing the volatility using the two approaches stated in this article, there is an improvement in the risk-adjusted performance as measured by the Sharpe Ratios of each strategy compared with pure equity holding. Volatility targeting also provides a dramatic improvement in the tail-risk measured by maximum drawdowns. The tactical volatility strategy delivers a better risk-adjusted return than the conventional strategy (see Table 1), without compromising risk measures such as better skew and comparable volatility and conditional VaRs. We show the full time period analysis, instead of discrete periods, to cover the various market cycles over the last 20 years .

Table 1: Strategy Performance

[Source: STANLIB Multi-Strategy, Bloomberg. Data is from 31 Jul 1999 to 30 April 2022. The data displayed is a hypothetical example of back-tested performance for illustrative purposes only and is not indicative of the past or future performance of any STANLIB Multi-Strategy product]

 

In Figure 4, we plot peak-to-trough drawdowns of the volatility targeting strategies. Most notably, the volatility-controlled/targeted strategies significantly reduced larger drawdowns during the 2008 GFC and COVID-19 pandemic. In more moderate drawdowns, however, the strategies did not materially improve outcomes. Due to this, volatility targeting strategies may be insufficient for investors seeking explicit downside risk protection. That is why we reiterate in this article that the volatility targeting strategies should complement traditional tail-risk strategies, e.g. options overlays, for more risk-averse investors.

Figure 4: Drawdown Analysis of the volatility targeting strategies

Source: STANLIB Multi-Strategy, Bloomberg. Data is from 31 Jul 1999 to 30 April 2022. The data displayed is a hypothetical example of back-tested performance for illustrative purposes only and is not indicative of the past or future performance of any STANLIB Multi-Strategy product]

 

Our approach to the lens of volatility is multi-faceted. This quantitative research paper is focusing only on the risk management properties of volatility, and some of its potential uses within a portfolio. We are illustrating that a tactical approach to risk management is better than continual/systematic hedging. We believe in taking a tactical approach towards risk management. For example, we track various volatility signals like implied and realised volatility, volatility-term structure, skew, volatility risk premia, cross-asset volatility and cross-asset correlation (not limited to these volatility signals only) across various asset classes to gives us an idea the type of volatility cycle we are in. These volatility signals/indicators assist us in our tactical approach towards tail risk management and opportunistic trades. For example, at market extremes (where we observe extreme volatility spikes and stresses) it can be advantageous to monetize hedges to take advantage of elevated volatility which can lead to changes to risk exposure. We are not paying up to hedge when people are desperate for hedges.

 

Conclusion

Investors looking for smoother yet optimal return outcomes, even during times of market stress, could benefit from tactical volatility targeting strategies. Volatility tends to cluster in different regimes, so tactically thinking about these regimes can guide investors’ risk management. Volatility targeting strategies will alter the breadth of investment outcomes and can therefore help to lower risk and improve portfolio characteristics. As a result, we find these techniques do deliver better risk-adjusted returns while cushioning SA Equity investors from significant market drawdowns. Tail risk is lessened, but not negated. More risk-averse investors can therefore complement this approach with additional tail risk strategies.

 

Tactical volatility targeting outperforms conventional volatility targeting by not continuously de-risking or re-risking. This means incurring lower transaction costs while remaining fully on-risk in conventional times so returns are not compromised. Our analysis confirms these findings locally.

References
  1. Moreira, A., and T. Muir. 2017. “Volatility-Managed Portfolios.” Journal of Finance 72 (4): 1611–44.
  2. Harvey, C. R., E. Hoyle, R. Korgaonkar, S. Rattray, M. Sargaison, and O. Van Hemert. 2018. “The Impact of Volatility Targeting.” Journal of Portfolio Management 45 (1): 14–33.
  3. Liu, F., X. Tang, and G. Zhou. 2019. “Volatility-Managed Portfolio: Does It Really Work?” Journal of Portfolio Management 46 (1): 38–51.
  4. Dion Bongaerts, Xiaowei Kang & Mathijs van Dijk (2020): Conditional Volatility Targeting, Financial Analysts Journal, 76(4): 54–71. https://doi.org/10.1080/0015198X.2020.1790853
  5. Hamilton, J. D. (1989). “A New Approach to the Economic Analysis of Nonstationary Time Series and the Business Cycle,” Econometrica, 57, 357– 384.
  6. https://towardsdatascience.com/a-worms-eye-view-of-the-markov-switching-dynamic-regression-model-2fb706ba69f3

 

 

[1] For approaches to manage tail risk see the Mdwara & Streatfield paper at https://stanlib.com/2021/01/12/diversifications-failure-and-the-need-for-non-linearity-full/

[2] There is extensive research regarding the impact of volatility strategies across various assets classes. For an introduction to this research, see Bongaerts, Kang and van Dijk (2020).

More insights