Tail risk management strategies An overview

Introduction – why focus on tail risks in the current environment? Institutional investors recognise that there is a need to accept investment risk in order to generate the target return outcomes set out in their investment objectives. A natural outworking of this process (that is, investing in assets which are not ‘risk free’) is that there are likely to be periods where risky investments perform undesirably and poor outcomes occur. Risk is often referred to in terms of the standard deviation/volatility of returns, correlation with equity markets or an overall risk score or measure1. But in managing their portfolios, investors are typically concerned with avoiding, or at least mitigating the impact of, events that occur infrequently, but which have a large negative impact on portfolio returns. These events are often referred to as ‘tail events’, by virtue of the fact that they lie within the left ‘tail’ of the potential distribution of returns. With the rise of the derivatives market, financial instruments are now available to investors which provide insurance against an array of financial risks. However, conventional wisdom and investment theory tells us that the fair price for financial insurance usually exceeds the expected loss that is being protected against. This means that, over the long term, a structural allocation in a portfolio to financial insurance will typically result in a drag on returns and, as a result, investor interest in this space tends to be highly cyclical (and often in response to, rather than in advance of, periods of significant market turbulence). Following several years of very low volatility in capital markets, the return of volatility and drawdowns in risky assets in recent months have brought tail risk protection strategies back to the forefront of investors’ attention. Whilst we view the recent episode of market volatility as being relatively moderate from a medium-term perspective we do believe that, as communicated in our Secular Outlook 2015 publication, there are fundamental reasons why downside risks continue to be elevated at present, including: •• The significant public and/or private sector debt burden which still hangs over the major developed economies and represents a headwind to economic growth •• The fact that we sit in an economic expansion that is now ‘long in the tooth’ (compared to post war averages) combined with our view that there are numerous potential catalysts for a downturn •• The extraordinary levels of monetary stimulus which have been deployed since the Global Financial Crisis that has yet to be removed. This both constrains the ability of policymakers to counteract a future economic downturn and has caused valuations of risky assets to rise to ‘stretched’ levels as investors have pushed along the risk curve in search of additional returns •• We believe that financial markets behave like a complex adaptive system, that is, a system that is constantly evolving, with no central control and complex collective behaviour, where the actions of one participant impact the actions of other participants and vice versa. The complexity and interconnectedness of markets increases the degree of uncertainty around future outcomes and combined with the fragile state of economies (themselves complex and adaptive) at present may amplify any resulting downturn.

1. The risk measures commonly used by investors can understate or ignore the impact of tail events – for example the probability of a negative return does not reflect the severity of negative return outcomes, and the standard deviation of returns gives equal weight to upside and downside volatility.

Towers Watson currently assigns a higher than normal probability of downside risks materialising and an adverse economic scenario playing out. A materialisation of some of these scenarios might see a 30%-50% drawdown in equities markets, similar to the 2000 Tech Bubble or 2008 Global Financial Crisis. As well as experiencing poor asset returns during a crisis, a stressed environment could also expose deficiencies in a fund’s abilities to manage counterparty risk, meet cash flow requirements, rebalance or monetise assets and mitigate detrimental stakeholder behaviour. Successfully navigating these risks may incur a significant administrative burden. With institutional funds facing a rising number of risks to the downside, we believe that clients should now consider adding a form of tail risk protection to their portfolios. While this topic has been discussed at length in the past, in general implementing tail risk management strategies remains poorly understood. In this paper we look to further inform this discussion by: •• Articulating the main objective of tail risk management •• Providing an overview of the broad range of tail risk management strategies •• Assessing the strengths and weaknesses of the various strategies available, and •• Setting out a framework for determining which strategies are most appropriate for a particular institutional investor.

What are tail risk management strategies trying to achieve? The common approach adopted by many Australian funds to try to minimise the impacts of left tail events is to look to hold a diverse portfolio of assets. While a lot of traditional diversification strategies can help to reduce portfolio volatility and improve risk-adjusted returns during ‘normal’ market conditions, we believe there are a limited

number of genuinely diverse strategies that are effective during stressed market conditions (in which asset class correlations converge). Many asset classes considered ‘alternative’ or diversifying prior to the Global Financial Crisis, demonstrated a higher correlation to equity markets during the crisis (for example, hedge funds, property). That is, while we continue to believe in diversity as an effective policy lever, there are tail scenarios where diversity will provide less portfolio protection than would be expected based on ‘average’ correlations between asset classes. In this paper we explore how investors could implement targeted strategies which look to mitigate the effects of left tail events on a portfolio. Firstly, we should consider the key drivers of large negative portfolio returns as this provides a lens through which to view the objectives of tail risk management strategies. Looking at a distribution of returns (example in Figure 01 below) the worst results occur when an outcome is drawn from the left hand side of the distribution. The further to the left of the distribution the outcome occurs, the more painful and undesirable the result. Therefore, what we are really trying to achieve is to manage, or limit, the frequency and magnitude of downside outcomes. There are three characteristics of the return distribution which will ultimately drive the degree of tail risk inherent in a portfolio: •• Location – the mean, or expected return, which the distribution is centred around •• Dispersion – the width of the distribution, often expressed as the standard deviation •• Shape – the most commonly understood distribution shape is the ‘normal’ distribution. However, in reality return distributions for most asset classes are not ‘normal’. Asset return distributions tend to have fat tails and/or are skewed.

Figure 01. Example distribution of outcomes

Very good

Very poor

Painful ‘left tail’ scenarios Funds should be most concerned about tail events, as in these scenarios risky asset correlations increase and even a welldiversified portfolio will suffer.

Normal course of events A diversified portfolio is expected to be well positioned to deal with a range of economic scenarios under ‘normal’ market conditions.

Upside events Nice to have upside returns.

Tail risk hedging objective: manage the frequency and magnitude of downside outcomes 2 towerswatson.com

Risk relates to the variability of future outcomes. An investor generally does not know where, within the distribution of possible returns, the outcome over the next period(s) will lie – put a different way, we do not know in advance whether we will receive a return that is at the median of the distribution, at the lower 5th percentile, the upper 25th percentile or somewhere else. This is often managed by an investor defining a maximum tolerance for downside losses and structuring the portfolio so that the modelled downside risk for the portfolio falls within the stated risk budget. An issue with the conventional approach to risk management is that it relies on knowing in advance the width and shape of the return distribution and in reality this is also an unknown and is potentially time-varying – this is what we refer to as uncertainty. That is, even if a level of tolerance can be articulated (for example, 1-in-20, 1-in-200) the distribution of outcomes could be wider, more negatively skewed or have a fatter left tail than expected, which would result in downside outcomes being worse than predicted. As an example, it can be observed that the realised volatility of equity returns has exhibited significant variation over time and that large drawdowns have tended to coincide with elevated levels of volatility, as shown in Figure 02 below. Figure 02. Volatility vs realised returns

20% 0%

Figure 03. Explicit hedges 6 5

3 2 1 0 -50%

-40%

-30%

-20%

-10%

0%

10%

Return Unprotected

20%

30%

40%

50%

Downside Hedge

Source: Towers Watson

2. Attempt to re-shape the distribution (accept risk, attempt to reduce uncertainty) – Reshaping strategies generally aim to condense the return distribution (particularly the left side), reducing the uncertainty of investment outcomes and, with it, the amount of tail risk inherent in the portfolio. These strategies accept that investment risk will always be present and poor outcomes will occur. Instead, reshaping strategies attempt to reduce or at least control the size of the tail of the distribution and reduce the impact of a left tail event (or increase the degree of certainty around the size of such an event). Following the implementation of these types of strategies the distribution of portfolio returns might look as in Figure 04.

-40%

5

Realised volatility (% pa) Source: Bloomberg LP, Towers Watson

Approaches to tail risk management When looking to implement a tail risk management program, there are two approaches investors can take to try to (at least partially) mitigate investment risk or uncertainty: 1. Explicit downside hedges (accept uncertainty, attempt to remove risk/tail of distribution) – Attempt to remove/offset return outcomes below a certain point through utilising strategies which either explicitly cut off downside

Probability density

35%+

30% - 35%

25% - 30%

20% - 25%

15% - 20%

10% - 15%

6

5% - 10%

-60%

4

Figure 04. Distribution reshaping

-20%

0% - 5%

Realised return (% pa)

40%

outcomes and/or have negative correlations with risky assets in a market downturn, as shown in Figure 03. These strategies attempt to remove the risk of return outcomes being drawn from the far left of the distribution. However these strategies will not necessarily influence the shape or dispersion of the rest of the distribution; therefore investors are still subject to the uncertainty that return outcomes can be drawn from anywhere else in the distribution (which has an unknown shape).

Probability density

Minimising a left tail event will necessitate managing the return distribution through trying to control or influence one or more of these attributes. However, investors are faced with two problems when trying to manage and mitigate left tail events – risk and uncertainty.

4 3 2 1 0 -50%

-40%

-30%

-20%

-10%

0%

Return Unprotected

10%

20%

30%

40%

50%

Reshaped

Source: Towers Watson

We identify strategies under both approaches which can be implemented using either derivatives or physical assets; these are summarised in Figure 05.

Tail Risk Management Strategies

3

Figure 05. Summary of tail risk management strategies

Physical

Explicit Hedges

Distribution Reshaping

•• Sovereign bonds

•• Diversity strategies

•• Gold

•• Dynamic asset allocation

•• Commodities Derivative

•• Vanilla option strategies

•• Managed volatility

•• Long volatility strategies

•• Dynamic asset allocation

•• Constant Proportion Portfolio Insurance (CPPI)

•• Put spreads

•• Foreign currency (FX) exposure

•• Fixed budget option strategies

Below we provide a brief overview of each strategy:

Physical assets – strategies that provide an explicit downside hedge Sovereign bonds – An exposure to long-duration developed market government bonds, particularly in a ‘safe haven’ currency (for example, US 30 year Treasuries) has historically provided investors with protection in most ‘flight to quality’ scenarios. Such strategies may also involve leverage where suitable long-dated bonds are not available (for example, intermediate bonds could be leveraged) or to reduce the capital intensity of the strategy. This strategy may be particularly suited for investors with interest rate-sensitive liabilities. However, there may also be adverse economic scenarios (for example, a sovereign debt crisis or a ‘right tail’ inflationary growth slowdown) where bond exposures could be additive to tail risk. In addition, current levels of yields mean that government bonds may not offer the same degree of downside protection as has been the case in the past. Gold – Historically investors have sought precious metals as a ‘safe haven’ asset during times of crisis. Gold can also act as a store of real wealth in the event of an inflation ‘spike’. While gold may not provide investors with the strongest equity market hedge, it may protect against a variety of severe market conditions, such as a (fiat) currency crisis or a sovereign debt crisis. As well as direct investment in the asset, investors could also gain exposure via ETFs or derivatives – the latter options avoid having to arrange for, and fund the cost of, storage which is a common problem with physical investment in commodities. Another problem with investment in gold is the fact that it does not satisfy the usual criteria for an asset – it does not have an associated cash flow stream, nor is it a common input into production of goods and services. As a result it is extremely difficult to assess the attractiveness or otherwise of gold prices. Commodities – Exposure to a broad index of commodities may provide protection from a tail event that is characterised by an inflationary spike. However the drivers of the spike will determine

4 towerswatson.com

the effectiveness of the hedge and also which commodities will provide the highest level of protection for the portfolio. Exposure to a basket of commodities is likely to underperform in a general economic downturn, where demand for major commodities (for example, oil and iron ore) is curtailed. Physical assets that are expected to act as explicit hedges against downside outcomes have common advantages in that they are relatively liquid, straightforward to implement and can be accessed at a reasonable cost. The key disadvantages of these strategies are that their effectiveness is highly dependent on the nature of the tail event that occurs and the payoffs from these strategies are relatively linear. This means that high and/or leveraged allocations are required to have a meaningful impact in the event that equity markets suffer a significant drawdown. Furthermore, the need to purchase and hold physical assets may lead them to cause noticeable impact/disruption on the structure of a portfolio.

Physical assets – strategies that seek to reshape the return distribution Diversity strategies – True diversity involves reallocating away from equities towards other, more niche, assets that have exposures to return drivers other than the equity risk premium. This, in theory, results in a reduced concentration of macro risks within a portfolio. There are essentially four key ways investors could materially look to increase diversity in decreasing order of impact and governance: 1. Increase the allocation to ‘diversifying strategies’ a. Liquid alternatives (for example, alternative risk premia, skill based strategies) b. Real assets (for example, real estate, infrastructure and natural resources) 2. Better diversify the credit portfolio (for example, high yield, loans, ABS and illiquid credit) 3. Better diversify within the equity portfolio (for example, private equity and long-short equity)

4. Increased targeting of the skill premium within the existing asset mix (for example, higher active share equity managers). While we expect diversity to continue to be a valuable policy lever in the current environment, we would note that these strategies have historically underperformed equities in upwards trending markets. In addition, these strategies generally result in increased complexity/governance requirements, and also generally higher direct costs and illiquidity risk. Further, the nature of these ‘uncorrelated’ strategies is that they are unlikely to be well understood by the market and therefore generate part of their excess return from a ‘complexity premium’ which, if re-priced, could result in these strategies underperforming unexpectedly. Finally, a number of these strategies essentially exploit different forms of an insurance risk premium and therefore have ‘left-tailed’ characteristics (that is, incremental positive returns punctuated by sharp losses). Dynamic asset allocation – By dynamically adjusting allocations to the underlying asset classes/ strategies in the portfolio, an investor can take into account elevated valuations and/or downside risks and allocate away from strategies that offer unattractive risk adjusted returns at a given point in time. Dynamic positions can be implemented using either judgement or a rules based systematic approach (or a combination of the two). While Towers Watson does not currently (as at October 2015) identify any areas where we observe extreme mispricing, we would expect mispricing opportunities to appear in any market characterised by either excess or fear. A dynamic strategy should encourage investors to purchase assets which have been eschewed in an environment of excess (generally, where the market has less exposure to these assets they will likely outperform in a crisis). Conversely, in a market ‘shock’ or downturn, panic is likely to lead to risky assets being oversold and market dislocations/mispricing opportunities to occur. Furthermore, we expect divergence of economic (and therefore capital market) performance and heightened volatility to be key themes over the medium term which should also be favourable for a dynamic asset allocation process. Whilst not an explicit downside hedge, simply holding a higher allocation to cash could be considered as complementary to a downside protection program. During a market downturn, a higher cash allocation would reduce the impact of any losses, provide precious liquidity and could be used to take advantage of any mispricing opportunities which emerge. This takes advantage of the so-called ‘option value’ of holding cash but is

contingent on the skill of an investor to be able to identify an appropriate time to redeploy capital in the market (otherwise this approach reduces longterm expected returns). A well implemented dynamic asset allocation process will require significant internal and/or externally sourced governance and resource in order to ensure that decisions are well informed and can be implemented efficiently.

Derivatives – strategies that provide an explicit downside hedge Vanilla option strategies – Options can be used to provide a specified level of downside protection. This could involve put options, collars and cash plus call option strategies, and can be implemented across various asset classes including equity, interest rates, volatility and credit default swap markets. We explain these terms below. The simplest way to implement downside protection via options is to buy (‘go long’) an out of the money put option (that is, strike price lower than current price) on a specific index (or security). It is analogous to purchasing insurance in that the buyer pays a premium which will profit if a market falls below a specified level. The strike price of the option can be chosen to set a specific ‘floor’ on returns for the asset class being protected (at varying costs). A long equity collar strategy involves the purchase of an out of the money put option in conjunction with the sale of an out of the money call option. This is similar to the strategy above except that any upside in excess of the call option strike level is foregone, thereby somewhat (or in some cases completely) offsetting the cost of the put option. For investors adopting this strategy, implementing low cost protection may take precedence over maintaining upside potential, or alternatively an investor’s preference for additional returns may not be linear beyond a certain level (for example, a liability-driven investor may receive diminishing benefit from returns beyond a certain threshold/required level). A cash plus call strategy involves obtaining exposure to an asset class through holding cash plus a long (typically at the money) call option as an alternative to physical exposure to the asset. This strategy provides exposure to appreciation in asset prices while limiting downside to the cost of the call option (less the return on the cash holding). Implementing an option strategy can be complex and involves a number of decisions other than just the type of strategy to be implemented. Key decisions include the level of protection sought (for example, the strike levels of the options) and whether any ‘upside’ will be given up in order to reduce the cost of protection. In addition, the

Tail Risk Management Strategies

5

choice in which markets to buy protection is not trivial either as there is a natural trade-off between complexity and cost (smaller markets may exhibit lower liquidity/higher derivative trading costs), and basis risk (whilst equity markets would be expected to be globally correlated, idiosyncratic events can impact individual markets in isolation). All of these decisions will impact the cost of an option strategy. The relative attractiveness of option prices may vary across markets, strikes and tenors which means that dynamic management of these factors can add material value. In addition, as options have a fixed maturity, these strategies are not ‘set and forget’ and decisions will need to be made including: •• How to manage counterparty risks if over-thecounter/non-collateralised instruments are used •• Whether to ‘monetise’ options in the event that a tail event does occur and gains accumulate and, if so, what to do with the proceeds (for example, reinvest in the asset class on the basis that valuations are more favourable or purchase cheaper protection instruments) •• How to structure the expiry dates (for example, laddered expiry vs single expiry), and •• What to do when the options expire (for example, reinvest the same protection strategy, adopt a different strategy or invest in physicals). The price of options can vary significantly over time as market conditions change and can be significant in periods of high market volatility. Long volatility strategies – These are more complex option-based strategies which involve purchasing a portfolio of derivatives designed to deliver a ‘convex’ payoff in the event of a market drawdown associated with a sharp rise in volatility. As noted in the first section of this paper, periods of market stress are often associated with a spike in volatility. As a result, derivatives with a high ‘vega’ (that is, high sensitivity to the volatility of the underlying asset – typically long-dated and/or deep out of the money options) will deliver strong returns in these types of environments. Due to the convex nature of the payoff profile, a long volatility strategy may generate a return which is several multiples greater than the underlying drawdown in equities (for example, a 25% drawdown in equities could result in a payoff for a long volatility strategy in excess of 100%). There are a number of these types of strategies offered by fund managers aiming to deliver a strong payoff if a tail event occurs. In addition, these managers generally look to generate ‘alpha’ relative to holding a naïve portfolio of options and dynamically manage the portfolio (for example, by looking across geographies, markets and tenors to identify where volatility can be purchased more cheaply) to mitigate the cost of premium ‘bleed’ in periods where no event occurs.

6 towerswatson.com

Constant Proportion Portfolio Insurance (CPPI) – CPPI utilises quantitative models to allocate dynamically between risk free and risky assets in order to replicate an option-type payoff. A CPPI strategy seeks to cap an investor’s downside risk while maintaining exposure to risky assets in sideways and upwards trending markets. The rebalancing of risky asset exposures is typically done using listed futures to limit trading costs and to improve liquidity. The advantage of CPPI strategies is that they are not subject to the premium drag associated with traditional options. The downside is that because these approaches are reliant on rebalancing between risk free and risky assets in response to market movements (which incurs transaction costs and cannot be done in continuous time), CPPI strategies do not protect against scenarios where markets fall sharply or ‘gap’. CPPI strategies can also be exposed to losses in sideways markets as they are effectively trying to anticipate the beginning of a trend which may not materialise. In addition, as these strategies increase the allocation to cash (as the price of risky assets fall), they effectively become ‘cash locked’ when the protection floor is reached, preventing participation in any future recovery. Foreign currency (FX) exposure – In financial market crises, ‘risk-on’ currencies (such as the Canadian Dollar and Australian Dollar) have tended to depreciate against ‘risk-off’ developed market currencies (such as the US Dollar, Japanese Yen and Swiss Franc). Investors could hold an exposure to a basket of risk-off vs risk-on currency pairs, if they have conviction that the correlation between equities and risk-on currencies will hold in future market downturns. This strategy may incur a cost of carry (depending on relative interest rate differentials of the underlying countries for the currency exposures held), but a specific currency basket could be constructed and accessed relatively easily and inexpensively using derivatives.

Derivatives – strategies that reshape the return distribution Managed volatility – These approaches recognise that the volatility of asset classes varies over time and utilise quantitative models to estimate the prevailing level of asset class volatility using market data (which may include historical return data, implied volatility or a combination of both and other factors) and use this to make predictions about the level of volatility in the near future. These strategies then use listed futures to adjust the exposures to the underlying asset classes to target a level of either asset class or total portfolio volatility that is specified by the asset owner. Simplistically, if measured volatility is higher than

the target level then total portfolio risk is reduced and vice versa – minimum/maximum limits around asset class exposures (for example, to ensure the portfolio remains long-only) can also be applied. In principle, these approaches can be applied across a wide range of asset classes, including equities, fixed income, credit, commodities and currencies. They can also be implemented as an overlay to the portfolio or as a managed fund solution. If the model being used to estimate volatility is effective then the realised volatility of portfolio returns should be controlled within a relatively narrow range rather than being allowed to vary with market conditions. This in turn addresses the problem of uncertainty addressed earlier, providing an investor with greater confidence in the estimate of the magnitude of potential downside outcomes for their portfolio. As an example, a managed volatility overlay could be implemented that targets a level of portfolio volatility that is equal to the volatility assumption used by an investor in its asset allocation model. In addition, if the historical observation that expected returns are higher in a low volatility environment and vice versa (as shown in Figure 02) holds in future, then strategies should both reduce the magnitude of downside outcomes and also be accretive to expected returns. However, it should be noted that these strategies do not explicitly limit downside risk. As these strategies use listed futures they are highly liquid and can be implemented at low cost. A systematic approach is also transparent and avoids risks associated with using human judgement. On the other hand, managed volatility strategies rely on the modelling of prevailing volatility at a point in time. The strategies are therefore susceptible to model risk and will underperform in situations where markets and/or the models used misprice or misestimate volatility. The performance of these strategies is also somewhat reliant on the assumption that the historical negative correlation between volatility and returns on risky assets holds in the future. Dynamic asset allocation – A dynamic asset allocation process could also be implemented using derivative instruments (for example, using futures, options and/or other instruments) to implement over/underweight exposures to asset classes. This would be particularly appropriate for a process that uses a relatively short time horizon (12 months or less) or where only a moderate level of conviction is required to implement a dynamic position, which would in turn imply a higher frequency of position changes. For some asset classes, using derivatives is likely to be more efficient and cost effective than purchasing the physical assets.

Put spreads – A put spread consists of purchasing and selling put options covering the same notional asset class value at differing strike prices. In the context of managing tail risks this will involve purchasing an out of the money put option and selling a put option that is further out of the money – essentially a put spread involves purchasing protection at a certain level and selling away protection at a lower level. This in turn results in the portfolio being fully protected for losses that fall between the two strike levels of equivalently the portion of the return distribution falling between the two strikes being ‘shifted’ to the initial protection level. These strategies can also involve selling out of the money call options to further reduce the overall cost if an investor is willing to forego upside beyond a certain level. Option prices exhibit a volatility ‘skew’ which means that, per unit of protection provided, put options become relatively more expensive as they become further out of the money (that is, as the strike price decreases) – this reflects both risk aversion to significant losses and also market demand/ supply dynamics. This means that the range of returns protected by a put spread can be set to cover the most likely range of losses (for example, those expected to occur with say 90% to 95% likelihood) whilst avoiding paying for protection from more extreme losses that are a) very unlikely and b) expensive to protect against. These types of strategies may also be appropriate for investors who are highly peer sensitive as they are lower cost than simply buying put option protection and generate outperformance relative to long-only investors without an option overlay in the event of moderate market losses. Fixed budget option strategies – As a variation of vanilla put option strategies, an investor could allocate a fixed dollar budget towards purchasing put options. The amount of protection offered by this strategy will vary over time as option prices change. This strategy would be appropriate for a cost sensitive investor who has a fixed amount they are willing to spend on purchasing portfolio protection. In addition, it has often been observed that implied volatility (and therefore option prices) tend to lag market events. In particular, implied volatility tends to rise sharply after a large equity market fall or increase in realised volatility and also that such falls tend to be preceded by a period of unusually low realised and implied volatility. This means that, in many cases, option protection was ‘cheap’ at exactly the time it would have been beneficial to be holding such protection and very expensive after the event – a fixed option budget strategy would therefore be consistent with a belief that this phenomenon is likely to continue to repeat in the future.

Tail Risk Management Strategies

7

Assessing tail risk management strategies

can be made separately from the decision regarding which tail risk management strategies to employ.

In this section we look to measure each strategy against the following criteria for tail risk management strategies.

Liquidity – In a stressed market environment liquidity is often scarce, therefore it would be desirable if the strategy(ies) implemented can be monetised quickly and efficiently so that payoffs can be realised if a tail event occurs.

Strategic and cyclical attractiveness – The economic cycle can be considered to have four broad phases – recession, trough, recovery and peak. As discussed earlier in this paper, we believe that we are currently in the ‘peak’ phase of the global economic cycle and that there is an elevated risk of a recession occurring over the medium term. As the key drivers of the economy and therefore capital markets differ through each phase of the cycle, so too does the relative attractiveness of various risk management strategies. We set out in Figure 06 below the typical attractiveness of different risk levers through the economic cycle (absent of valuation considerations). Figure 06. Cyclical attractiveness of various risk levers Lever

Recession

Diversity



Bonds



Option Strategies Extreme Hedges



Trough



Recovery



Peak

 





(sell)

(buy)



Cost – Strategies which look to mitigate downside risks tend to involve costs that exceed the expected loss being protected against – ideally the strategy(ies) implemented by an investor will be selected and managed in such a way as to minimise these costs. We consider both explicit costs (incurred when adopting/moving to the strategy or maintaining the position) and implicit costs (such as the impact of negative carry on some derivative positions or the opportunity cost of having to hold/ purchase ‘risk off’ assets which may create a drag on performance in normal market conditions). We note that a number of protection strategies can be implemented on a relatively cost effective basis at the moment. One benefit of the monetary tightening which has occurred across the developed world across the past several years is that numerous protection strategies can be accessed relatively cheaply (particularly strategies which incur a cost of borrowing) due to the low interest rate environment. Portfolio disruption – We generally prefer strategies that limit, as far as is practicable, the disruptive impact on the rest of the portfolio – in this way the broader strategic asset allocation decision

8 towerswatson.com

Complexity – Strategies that can be both easily understood and implemented are preferable to more complex ones. Operational risk – The risk of loss resulting from inadequate or failed internal procedures, people and systems. The nature of a number of the tail risk management strategies discussed in this paper is that they involve a greater degree of potential operational risk compared to more conventional investment strategies, whether due to human error or systems failure. Strategies with greater operational risk will require advanced (fund) governance structures in place. Model assumption risk – The risk of the strategy not capturing/mitigating the risks it was designed to due to errors in the models or assumptions utilised. This is predominantly applicable for quantitatively driven strategies and processes. This risk is also relevant for strategies that rely on specific historical relationships applying in the future in order to be fully effective.

We set out in Figure 07 below an assessment of the various tail risk management strategies identified in this paper against the criteria set out on the previous page, as well as assign an overall rating (between one and three). Figure 07. High level evaluation of tail risk management strategies

Strategic and cyclical attractiveness

Cost

Portfolio disruption

Liquidity

Complexity

Operational risk

Model/ assumption risk

Overall rating

  

  

  

  

 

 

 

 

   

   

   

   

   

   

   

   

   

   

Physical assets – strategies that provide an explicit downside hedge Sovereign bonds Gold Commodities

  

  

  

  

Physical assets – strategies that seek to reshape the return distribution Diversity strategies Dynamic asset allocation

 

 

 

 

Derivatives – strategies that provide an explicit downside hedge Vanilla option strategies Long volatility strategies CPPI FX exposure

   

   

   

Derivatives – strategies that reshape the return distribution Managed volatility Dynamic asset allocation Put spreads Fixed budget options

   

   

   

Positive



Neutral



Negative



We summarise the key advantages and disadvantages of each strategy in the Appendix to this paper. Notes •• We assume that all derivative contracts are exchange traded or collateralised and therefore have not factored credit/counterparty risk into the ratings. •• Vanilla option ratings assume a long put strategy is employed.

Tail Risk Management Strategies

9

Aligning a tail risk management strategy with Fund beliefs, objectives and circumstances In addition to the metrics outlined on the previous page, the suitability of the various strategies is critically dependent on an investor’s beliefs, circumstances and objectives. Before adopting a tail risk management strategy we would encourage investors to consider the following steps: 1. Consider the strategies in the context of risk objectives or risk statement – The starting point for all investment strategy decisions is an investor’s risk objectives and/or risk appetite statement as particular approaches are likely to be more aligned with certain objectives. For investors who aim to limit the ‘probability of negative return’, reshaping strategies may perform better as narrowing the return distribution can serve to reduce/control the shortfall probability. Explicit downside hedging strategies that look to remove negative outcomes beyond a certain point will tend to result in similar (or potentially slightly increased) shortfall probability outcomes and are therefore likely to be more suitable for investors who aim to limit the ‘5th percentile return’, or ‘expected tail loss’. For investors who aim to perform well relative to peers in poor markets, strategies which attempt to moderate negative returns (for example, put spreads or [a low] fixed budget option strategy) will perform well. These strategies will materially outperform the ‘average fund’ in a downturn, while only incurring a moderate cost to maintain the position(s) across calm markets (where arguably sensitivity to peer risk is lower in any case). 2. Identify any specific risk(s) to the portfolio – Once key tail risks or ‘pain’ thresholds have been identified, specific strategies can be identified to mitigate these risks. One approach is to determine which particular scenarios and/or subset of the portfolio poses the biggest threat to the portfolio achieving its investment objectives. Having completed this process, investors could decide to target either specific or idiosyncratic risks (for example, long sovereign bonds if liabilities are interest-rate sensitive) or a broader approach encompassing a variety of risk scenarios. However most investors will conclude that equity market risk poses the biggest threat to their mission impairment and it is the risk of a sharp fall in equity markets which needs to be cushioned. In addition, the nature of the downturn will materially impact which lever is most attractive: •• In a sharp downturn, options and other strategies with a convex payoff profile are likely to perform best as this will likely coincide with a sharp rise in 10 towerswatson.com

volatility and the payoff will be monetised before too much premium drag is incurred •• In the event of a continued period of calm in markets or a gradual downturn strategies that look to reshape the return distribution (in particular diversity and other strategies that do not incur cost of carry) will likely perform better. It is important to note that whilst we have evaluated each strategy in isolation in this paper, utilising multiple downside protection tools could be the most effective approach towards protecting a portfolio across a spectrum of potential adverse scenarios. 3. Gauge how much tail risk exposure exists in the portfolio – Quantitative modelling, such as calculating portfolio Value at Risk (VaR), can be useful for indicating the potential magnitude/amount (of the tail loss) a fund could be expected to incur. Even at the portfolio level, left tail events will tend to occur more frequently than standard models predict because in times of stress: •• Correlations among risk factors typically rise •• Different economies and factors may behave like one another, and •• Volatilities across most markets will generally rise. We would encourage investors to look at ways in which their existing modelling frameworks can be enhanced/augmented to compensate for these limitations, including: •• Recognition of the impact of uncertainty by incorporating multiple volatility regimes and/or the possibility of market ‘gaps’/jumps •• Application of ‘left tail multipliers’ to modelled tail risk outcomes and/or carrying out modelling with increased correlations, in particular when considering more complex strategies with limited track records, and •• Stress testing portfolios using deterministic scenarios that represent events that are expected to materially impair the portfolio.

Quantitative measures of downside risk should be treated only as indicative. These are likely to only capture a select number of possible ‘paths’ or adverse scenarios which could play out. However, quantifying downside risk can provide a fund with valuable perspective for how a downturn may affect the portfolio. 4. Have a (broad) cost budget for implementing the tail hedge – Estimating costs for implementing tail risk management strategies can be difficult. However investors should recognise/accept that it is likely that there will be some ongoing cost for sustaining a tail risk management program and then ascertain the level of cost they are willing to outlay for protection.

It is generally acknowledged that implementing tail risk hedges is expensive, as investors end up paying away a risk premium in order to remove these risks from their portfolio. This is because the tail risk that an investor wants to remove has to be passed on to someone else –almost all investors want to remove this risk, and virtually no investors structurally benefit from a market downturn. However, there are two possible approaches to minimising the cost of a tail risk protection program. Opportunistic tail risk protection: The simplest way to implement a position is to purchase the instrument/asset and maintain the protection indefinitely (rolling it over if using derivatives). This is called a static hedge and it can be expensive to roll-over and maintain. A cheaper alternative to a static hedge is to ‘turn on’ the tail risk hedge when there is a signal that there is an elevated likelihood of a left tail event occurring in the near future. The key issue with an opportunistic approach to hedging is timing – both being able to foresee when market corrections may occur and having the appropriate governance structure in place to implement the tail-risk hedge expediently and also to determine when the hedge should be removed. An additional risk associated with an opportunistic approach is that after identifying an appropriate time to add protection the price of the instruments being used to manage tail risks may rise sharply before the relevant strategies can be implemented. Derivatives based strategies are likely to be more compatible with an opportunistic approach where positions need to be adjusted quickly and in a cost efficient manner. In contrast, where protection is to be held on a strategic basis strategies which use physical assets may be more appropriate. Indirect tail-risk hedging: Depending on timing and market dynamics, often the most ‘direct’ hedging solution will be the same protection that other investors are looking to buy and therefore the most expensive. However, there may be ‘indirect’ hedging alternatives that provide sufficient lefttail protection at a significantly cheaper cost. For example, the cost to hedge equity risk using options can vary greatly across global markets for a similar type of protection. Different asset classes or derivatives may also provide suitable hedging proxies (for example, it is often observed that Asian implied volatility tends to sell off in the event of a crisis that impacts developed market risk assets but derivative pricing tends to be cheaper in Asian markets than in the major developed markets due to the demand/supply dynamics of the regional structured products markets). The protection coverage offered by indirect hedging alternatives

can be very strong, especially during times of market stress when asset correlations tend to ‘go to one’. Of course, these patterns will not hold in all circumstances. 5. Other considerations – These may include investor-specific requirements and restrictions. Two common restrictions in the tail risk protection space include the use of derivatives and leverage. Many investors currently do not directly trade derivatives in their portfolio, either due to explicit guidelines and/or laws. While investment into funds which use derivatives, such as ‘real return’ funds or hedge funds is generally acceptable, pure derivatives based strategies are still generally avoided. Similarly there may be a limit on a fund’s ability to use leverage within the portfolio. Investors should confirm and clarify applicable restrictions with their investment guidelines and legal counsel before looking to implement specific tail risk management strategies. 6. Articulate the key parameters for the tail risk management program – Once an investor has decided to implement a tail risk protection program, it is beneficial to articulate the key parameters for any proposed tail risk management strategy(ies), including the objectives of the program and the timeframe over which these are to be achieved. When implementing a risk mitigation strategy, success measures can be difficult to define. For example, the best outcome would be that no negative tail event occurs, in which case most tail risk strategies would be expected to detract value. However, stipulating the goals and objectives should assist in giving a clear purpose for any downside protection program and is also useful for reconciling any strategy implemented against the underlying objectives of the program. This is crucial if the program is to be implemented over an extended timeframe. The types of parameters which could be established include: •• The range of returns that are to be protected against at the portfolio and/or asset class level •• The expected performance of the program (in the event that returns in the ranges set out above materialised) •• Actions to be taken if an event occurs (for example, take profit vs reinvestment rules) •• Time horizon for the program (including criteria for winding-up the program) •• The cost budget for the program •• Liquidity requirements •• Monitoring requirements and responsibilities.

Tail Risk Management Strategies

11

Conclusion As set out in this paper and also in our Secular Outlook 2015, in the current environment there are a range of potential catalysts for a further market downturn. As always, how or when a downturn will occur is unclear. We therefore continue to believe that diversification across a diverse range of risk premia and investment strategies is the primary lever investors should use to manage risk in the current environment. This will result in a portfolio that has less concentration to any one specific macro risk. This said, we believe that investors should consider adding tail risk management strategies to their portfolios to reflect growing downside risks in the current environment. We believe that with a wide array of downside protection tools available (the majority of which can be implemented cost effectively) investors should be able to identify a strategy aligned with their investment objectives. To this end, the aim of this paper is to go beyond the diversity lever and to familiarise investors with various strategies which could add an additional layer of protection to portfolios, as we move to a market environment where risks to the downside are steadily increasing. There is no single perfect solution, and the suitability of each strategy is largely going to be fund specific. Towers Watson’s role is to assist our clients in determining which investment ideas may be suitable for their portfolios. Figure 08 below provides a very high level summary of the types of investors for whom different strategies might be appropriate. Figure 08. Aligning a tail risk management strategy with your fund Explicit Tail Hedges

Reshaping Distribution

Physical Assets

•• Investors defining risk as VaR/expected tail loss

•• Investors defining risk as shortfall probability

•• Low governance investors

•• Investors looking to manage (limit) the dispersion of returns

Derivatives

•• Investors defining risk as VaR/expected tail loss

•• Investors defining risk as shortfall probability

•• Investors looking to protect against a sharp drawdown

•• Investors aiming to outperform peers

•• Investors with a moderate to high cost budget

•• Investors intending to implement a highly dynamic tail risk management program

•• Low/medium governance investors

•• Investors with a low cost budget

•• Investors intending to implement a highly dynamic tail risk management program

•• High governance investors

•• High governance investors

The main actions that we suggest investors take at the current time are as follows: 1. Make a high-level decision as to whether consideration should be given to implementing tail risk management strategies in the portfolio, taking account of investment beliefs, mission and governance 2. Consider existing risk objectives to determine what categories of tail risk management strategies are likely to be appropriate 3. Identify the key drivers of tail risks in the portfolio and a ‘pain threshold’ 4. Assess the level of downside risk in the portfolio using quantitative models, enhanced where possible to reflect the shortcomings of traditional models

12 towerswatson.com

5. Determine an acceptable total budget/cost for purchasing tail risk protection 6. Determine the range of strategies that are in scope, taking into account any investor-specific requirements or restrictions 7. Articulate key parameters for the tail risk management program, and 8. Using internal resource or external advice (or a combination), design and implement an appropriate basket of tail risk management strategies.

Appendix – Advantages and disadvantages of key tail risk management strategies Advantages

Disadvantages

Physical Assets – strategies that provide an explicit downside hedge Sovereign bonds

•• Highly liquid, low cost

•• May not offer sufficient upside at current yields

•• Low governance, simple to implement and understand

•• Capital intensive, requires significant allocation and/or leverage to achieve meaningful protection due to linear payoff

•• Should protect against most likely (disinflationary) downside scenarios Gold

Commodities

•• Store of wealth, should protect against a range of downside scenarios

•• Likely negative expected long-term carry

•• Low governance, simple to implement and understand •• Can be implemented via ETFs

•• Capital intensive, requires significant allocation and/or leverage to achieve meaningful protection due to linear payoff

•• Liquid, relatively low cost

•• Will only hedge against an inflationary/right-tail scenario

•• Low governance, simple to implement and understand

•• Assessing valuation extremely difficult

•• Assessing valuation extremely difficult

•• Capital intensive, requires significant allocation and/or leverage to achieve meaningful protection due to linear payoff Physical Assets – strategies that seek to reshape the return distribution Diversity strategies

•• Reduces concentration of macro risks

•• Typically involves greater illiquidity and higher costs

•• Strategically attractive in all parts of the cycle

•• A number of ‘uncorrelated’ strategies are untested across different environments

•• Positive carry

•• Tends to underperform in strong equity markets •• May still exhibit correlation to equities in a crisis Dynamic asset allocation

•• Addresses elevated valuations

•• Significant governance/resources required

•• Should perform well in an environment characterised by divergence

•• Wide asset allocation ranges required to achieve meaningful protection

Derivatives – strategies that provide an explicit downside hedge Vanilla option strategies

•• Offers absolute downside protection

•• Relatively expensive if held on a rolling basis

•• Low capital intensity/impact on portfolio

•• Ongoing management not straightforward

•• Relatively straightforward to understand Long volatility strategies

CPPI

•• Provides returns that are a multiple of equity market drawdowns in the event of a significant market event

•• Complex, not transparent

•• Relatively small allocation required to achieve material level of protection

•• Significant judgement required/reliance on manager skill

•• Liquid, relatively low cost

•• Exposed to market ‘gaps’/jumps

•• Transparent, systematic approach

•• Complex to implement, material operational risk

•• Potentially less liquid than other strategies •• Negative carry (but this can be managed to some extent)

•• Can become ‘cash locked’ when protection floor is reached preventing participation in any subsequent recovery FX exposure

•• Highly liquid, low cost •• Low governance, simple to implement and understand •• Less capital intensive than other physical solutions as implemented as an overlay

•• Exchange rates (and foreign currency exposure) can be highly volatile •• May incur a carry cost (particularly for unhedged AUD exposures) •• Relies on historical relationship between ‘risk-off’ currencies and risky assets holding in the future

Derivatives – strategies that reshape the return distribution Managed volatility

•• Directly addresses uncertainty associated with measuring downside risk

•• Reliant on effectiveness of model used to estimate prevailing market volatility

•• Liquid, relatively low cost

•• Some exposure to operational risk via execution capability of manager

•• Transparent, systematic approach •• In theory should be accretive to risk-adjusted returns •• Can be run as an overlay, minimal portfolio impact Dynamic asset allocation

•• Addresses elevated valuations

•• Reasonable level of governance/resources required

•• Should perform well in an environment characterised by divergence

•• Wide asset allocation ranges required to achieve meaningful protection

Put Spreads

•• Lower cost than a strategy of holding out-of-the-money put options

•• Somewhat more complex than a vanilla option strategy

•• Protects against most frequently experienced losses and avoids paying for insurance against very rare losses •• Will improve peer relative outcomes in weak markets Fixed budget options

•• Cost is known in advance •• Consistent with empirical observation that significant market downturns often follow periods of low volatility

•• Material governance/resources required to manage on an ongoing basis •• Exposure to operational risk via internal or external execution capability •• More complex to understand and manage than a program with fixed notional exposure •• Level of protection is variable – may not have protection when it is required Tail Risk Management Strategies

13

Contact If you would like to discuss any of the areas covered in more detail, please contact your usual Towers Watson consultant, or:

Jeffrey Chee Senior Investment Consultant, Head of Investment Strategy, Australia and Asia-Pacific +61 3 9655 5126 [email protected]

About Towers Watson Towers Watson is a leading global professional services company that helps organisations improve performance through effective people, risk and financial management. With 16,000 associates around the world, we offer consulting, technology and solutions in the areas of benefits, talent management, rewards, and risk and capital management. Learn more at towerswatson.com

The information in this publication is general information only and does not take into account your particular objectives, financial circumstances or needs. It is not personal advice. You should consider obtaining professional advice about your particular circumstances before making any financial or investment decisions based on the information contained in this document. Towers Watson Australia Pty Ltd ABN 45 002 415 349, AFSL 229921 TW-AP-15-45582. October 2015 Copyright © 2015 Towers Watson. All rights reserved.

towerswatson.com /company/towerswatson @towerswatson /towerswatson