How to manage systemic risk in asset management

Systemic crises are rare but critical for long-term performance records. When the financial system fails, good trades become bad trades and many sensible investment strategies incur outsized losses due to deleveraging and liquidation pressure. Managers have two principal sets of tools to address systemic risk. The first is estimation and control of tail risk. The second is a contingency plan for systemic events. Estimating tail risk can make use of extreme value models, Bayesian risk forecasting, bubble indicators, and conditional value-at-risk models. Action plans for systemic crises can be based on research of systemic pressure points, guidance for shedding risk early in distress, rules for following market trends, and an adaptation of risk management to liquidity conditions.

The below post is an updated summary on the management of systemic risk on this site.

The importance of managing systemic risk

A critical and neglected factor of investment performance

Systemic risk in finance refers to the probability of the financial system failing at its essential functions, such as providing credit, making markets, or safeguarding securities and deposits. Typically, systemic risks build gradually over many years but materialize abruptly. They are hence a cause of extreme endogenous or setback risk (view summary of setback risk here) with rare incidences. Since years or decades pass between systemic crises, with many bonuses being awarded and vesting in the meantime, systemic risks are easily neglected in the day-to-day dispositions of investment managers.

However, how a manager prepares for and deals with systemic risk often makes or breaks long-term performance. Most investment strategies rely on some combination of directional or alternative risk premia and estimations of absolute and relative value. The basic principles of these are common across institutions. A systemic crisis typically derails all of these at the same time. This is because, in contrast to normal market drawdowns, systemic pressures trigger funding, accounting, or legal constraints that force position liquidation with little freedom of choice for portfolio managers. When a systemic crisis escalates the priority of institutions shifts from seeking returns to short-term capital preservation. As a result, the principles of efficient positioning or flows are often not only suspended but reversed. The best positions become the worst. With forced liquidations in thin liquidity, the trades that offered the most convincing value positions (by conventional standards) suddenly incur the greatest mark-to-market drawdowns. This happens because the expected return in the asset management industry is correlated with positions in normal times; good opportunities are rarely secret for long.

Moreover, standard risk management techniques fail in systemic crises. As volatility is soaring and correlations “all go to one”, positions that were calibrated according to value-at-risk give rise to unsustainable mark-to-market profit-and-loss volatility. Hedges may reduce average directional risk but give rise to large “basis risk” (profit and loss swings due to disparities in the returns of the main positions and the hedges). Also, as hedges typically raise the leverage of positions they make trades even more susceptible to forced deleveraging. Finally, diversification is of little help because the scope of inefficient flows is wide. Empirical research shows that a single global financial cycle floats or sinks most markets at the same time. (view post here).

A critical and neglected factor of financial system stability

Consideration of systemic risk in asset management also reduces the risk and consequences of crises themselves.

  • Investment managers’ attention to specific systemic risks makes it more likely that markets will charge a price for such vulnerabilities. Most crises follow from excesses and economic imbalances. For example, an excess of leverage reduces the capacity of institutions to absorb price drawdowns. The recognition of this risk means that borrowing costs and equity premia rise upon excesses, slowing or reducing them. Another relevant example is the unsustainable use of the environment and climate change (view post here). However, markets can only serve sustainability if managers think beyond annual performance fees. That is because “shorts” and protection strategies are typically negative carry trades with unknown duration.
  • The impact of shocks on the financial system and the economy is less severe if asset managers are prepared for systemic crises. It is often the very lack of protocol for difficult situations that leads to poor decisions and crisis escalation (view post here). Decision-makers rationally pay attention to rare potential crises only if expected losses from unpreparedness are more than proportionate to their rarity. There is ample evidence of regular boom and bust investment cycles, herding, trend-chasing (view post here), inefficient expectation formation, and speculative bubbles. Also, it has been shown that standard cognitive behavior is often inconsistent with efficient markets (view post here).

The standard policy response to private investment managers’ negligence of systemic risk is regulation. Yet, regulatory rules respond slowly to the evolution of the financial system and invite their own inefficiencies. A complementary policy is to increase the transparency of financial-economic information and reduce costs for allowing such information to be used in the investment process, for example through “quantamental systems” (view post here).

Approaches to managing systemic risk

Calibrating tail risk

Standard risk management relies on past volatility of price changes, historical correlation, and assumptions regarding outliers of price changes beyond normal ranges. On this basis, the majority of portfolios of liquid financial instruments is managed based on some form of Value-at-Risk (VaR) model, a statistical estimate of a loss threshold that will only be exceeded with a low probability.

Unfortunately, past volatility is not always a helpful gauge for financial markets risk. Volatility is merely the magnitude of historic price fluctuations, while risk is the probability and scope of future permanent losses (view post here). The two concepts are not only different but may even become opposites.  In particular, reliance on historic volatility can create an illusion of predictability that gives rise to excessive risk-taking. Indeed, low volatility itself is often a cause of high leverage and crowded positioning and hence conducive to subsequent outsized market movements.

Therefore, it is helpful to go beyond conventional risk metrics when assessing and calibrating the risk of large outlier events (“tail risk”):

  • Risk estimation should incorporate expert assessment. Historic data are not the only source of valuable judgment. For example, the risks and consequences of political upheavals, policy regime breaks, or first-time sovereign defaults are not easily quantifiable through price history. However, judgment and experience can be put into numbers, even if they take the form of wide confidence intervals. Logic, a broad perspective, and common sense prevent risk management from being trapped in the close confines of standard models.
  • Portfolio risk estimates can explicitly consider extreme market regimes. Thus, the basic idea behind extreme value theory is to fit an appropriate limiting distribution over returns that exceed a specific threshold. In particular, extreme value mixture models simultaneously estimate the threshold for extreme distributions and the extreme distribution itself (view post here). An alternative statistical approach is Bayesian risk forecasting, which accounts for the considerable distributional parameter uncertainty of Value-at-Risk and Expected shortfall estimates (view post here).
  • There are also quantitative warning signs of increased “tail risk” other than volatility. The simplest are valuation metrics for detecting bubbles (view post here), i.e. asset prices that are unusually high relative to the present value of estimated future cash flows. Academic papers have argued that equity markets with low dividend yields relative to local government bond yields are prone to large corrections (view post here). Similarly, countries with overvalued exchange rates and high short-term interest rates are prone to currency crises (view post here).
  • Standard volatility-based risk management metrics can be adapted for “tail events” and “gap risk”.Historically, diversification and downside risk analyses have assumed normal (“Gaussian”) probability distributions. Those are convenient for calculation but give little weight to large outliers. By now this “normality assumption” has been widely refuted and better gauges of tail risk are available (view post here), such as conditional Value-at-Risk. The distribution assumption is crucial for setting risk management parameters realistically and for assessing the potential upside of long-volatility and short-risk strategies.

Preparing crisis strategies

Having a plan of action in a systemic crisis can greatly diminish its impact on performance. An effective contingency plan can be based on (i) continuous research of systemic vulnerabilities that gives familiarity with pressure points, (ii) a rational protocol for shedding risk early in market distress, (iii) rules for following market trends, and (iv) adaptation of risk management to liquidity conditions.

  • Know the systemic weaknesses: It is very hard to predict the timing and dynamics of systemic crises. However, it is not so hard to understand the nature of vulnerability in specific economies and markets. Systemic risk indicators have been created for larger markets and have proven their value and relevance in the past (view post here). Their pace of change may be sluggish and their lead times ahead of crises long. Yet, through awareness of (i) systemic vulnerability and (ii) the degree of preparedness of the financial sector, asset managers can judge whether a specific shock is likely to be transitory or escalating after it has occurred. The two key criteria for escalation risk are (i) vulnerability of the system and (ii) rarity of the type of shock that hit it.
    • Vulnerability here measures the reliance of parts of the financial system on favorable conditions. Typically, protracted periods of low market volatility lead to build-ups of leverage and risk until the financial system reaches a tipping point (Minsky hypothesis, view post here). Long-term empirical analysis suggests that asset price booms are most dangerous when they are associated with rising financial leverage. Combinations of housing price bubbles and credit expansions have been the most detrimental of them all. (view post here). Also, economies that have grown accustomed to low real interest rates for long periods of time are typically susceptible to stress when rates or credit spreads are rising. Whether this stress is likely to escalate depends on whether the government or central bank have the means and the mandate to intervene. (view post here).
    • Rarity here measures the inverse of the a-priori probability of a specific type of shock. Rarity of shock means that it was not on the radar screen of either investors or policymakers. Theory suggests that senior decision-makers rationally do not prepare for rare events as they can only process a limited quantity of information. Technically speaking, rare events derail markets by raising “ambiguity premia”, whereby ambiguity means uncertainty about the probability distributions of an asset’s returns. Experimental research suggests that ambiguity premia are typically far more powerful drivers of price changes than traditional risk premia (view post here). Therefore, expected losses from unpreparedness are inversely proportionate to an event’s rarity(view post here). Indeed, expected losses from unpreparedness are even higher if managers bear only limited liability because they have particularly low incentives to prepare for low events.
  • Be early in shedding risk: There are two valid bases for adjusting positions quickly as systemic risk is rising: information advantage and market stress.
    • Information advantage depends on understanding and tracking the main mechanisms of systemic crises, contagion, and self-fulfilling dynamics (view post here). A particularly useful metric is the ‘financial stability interest rate’, a threshold above which the real interest rate in an economy triggers financial constraints and systemic instability (view post here). The relationship between the financial stability interest rate and the natural interest rate may be one of the most important predictors of medium-term market direction and future crisis risk.
    • Tracking market stress is simpler and, hence, far more common. A wide range of systemic risk indicators has been developed that rely on market prices, correlations, and volatility (view post here).
      In particular, volatility targeting is a valid default rule for containing mark-to-market risk if one is uncertain as to the severity of systemic pressure. This is because recent volatility is a good predictor of future volatility but not generally for future returns (view post here). Judging from U.S. equity data, volatility targeting strategies have over longer periods of time produced significant increases in return per unit of tail risk (view post here).
      Effective volatility targeting and dynamic hedging can make use of option-implied volatilities. Implied volatility only helps to predict future actual volatilities and future asset price correlations (view post here). High implied volatilities translate into high correlation whenever a single global factor is dominating price moves across all global markets.
  • Running with the herd: Following market trends is often rational and macro research can help to figure out when this is the case. For example, research may tell us that markets face a critical threat (e.g. a major bank may be at risk of default) but may not tell whether or not the risk will manifest (e.g. there may be a government bailout). In “make-or-break” situations it is rational for non-insiders to herd, i.e. to trade in the direction of prices, as private information disseminates in the market through prices (view post here). In this case, inaction would be an irrational and dangerous choice, even if we had no information advantage on the evolution of the crisis.
    Historically, simple trend-following strategies have reduced maximum drawdowns in equity portfolios and provided some hedging against losses in FX carry trades (view post here). Trend following removes psychological and institutional obstacles to exiting positions early in escalating crises. Trend following is a market directional strategy that promises “convex beta” and “good diversification” for outright long and carry portfolios as it normally performs well in protracted good and bad times alike (view post here). Drawdown control strategies, such as volatility targeting and “MinMax” (view post here), use elements of trend-following for risk management.
  • Controlling liquidation: Market liquidity can evaporate in systemic crises. This can give rise to outsized price movements and, at the same time, make position adjustments prohibitively expensive. To the extent that liquidity rather than fundamental change triggers market moves, position liquidations destroy investor value. Indeed, structural and regulatory changes in recent years seem to have made liquidity more precarious than in the past (view post here, here, and here). Hence, calibrating or structuring positions such as to withstand liquidity events can be a major cost saver and performance factor. This typically requires some flexibility in respect to risk management and mark-to-market-based drawdown limits. Researching the nature and potential of systemic risk is critical, both for preparation and for forming judgment whether it is really in the investor’s interest to liquidate positions.

With the right preparations, investment managers can even benefit directly from systemic events, particularly if they have sufficient flexibility and risk limits to exploit price distortions and high-risk premia paid. For systemic value based on price distortions see the related summary here. And for detecting and receiving high-risk premia see the section on “implicit subsidies” here.

Exchanging market risk information

No single investor or institution has all pieces of the puzzle that is systemic risk. Investors typically specialize in markets or geographic regions. However, every financial market depends on all other financial markets to some extent. At times particular market segments, such as asset-backed securities or technology stocks, can have a dominant global influence. Even small and remote markets, such as Iceland or Greece have in the past triggered sizeable global market moves. Links between seemingly separate markets often reflect their communal dependence on global liquidity, which is the ease of financing transmitted by a small number of financial centers (view post here).

Therefore, investment managers must engage in active risk information exchange, trading their insights for the insights of colleagues. Indeed, theoretical and experimental research suggests that portfolio managers will generally share ideas and research if mutual feedback is valuable (view post here). This creates investor value at all times but particularly when systemic risk is rising because investment managers that are part of an information network are better positioned to act early, as they know more and know better what others know. From a social welfare angle, this process of information exchange is essential to disseminate concerns over systemic crises. The dissemination turn may serve to warn market participants, policymakers, and the broader public, smoothing market volatility.

Diversified reward-risk parity

Risk parity is a portfolio construction technique that seeks to equalize risk contributions from the different components of the portfolio. Risk parity with respect to uncorrelated risk sources maximizes diversification. Simple risk parity rules are based on the inverses of market beta, price standard deviation, or price variance. These methods can be combined with common reward risk metrics, such as the Sharpe ratio, Calmar ratio, STAR ratio, or Rachev ratio. The resulting diversified reward-risk parity allocations have not only outperformed equally-weighted risk portfolios and standard factor allocations but also provided enhanced risk management.

(more…)

Market dynamics: belief, risk, and ambiguity effects

To understand financial market dynamics, it is helpful to distinguish beliefs, attitudes towards risk, and attitudes towards ambiguity. Beliefs are subjective evaluations of future cash flows. Risk refers to uncertainty within a model of the asset’s return. And ambiguity means uncertainty about the model and probability distributions. Accordingly, one can separate price dynamics into three effects: changes in beliefs, changes in risk premia and changes in ambiguity premia. Ambiguity premia seem to be dominant, particularly when investors have little information about the nature of a particular risk. Traditional risk premia seem to be much less significant. Belief effects are negligible when ambiguity is high but increase as information accumulates. Often trading opportunities arise from the mean reversion of ambiguity premia and the “under-adjustment” of beliefs.

(more…)

The financial stability interest rate

The financial stability interest rate is a threshold above which the real interest rate in an economy triggers financial constraints and systemic instability. It is different from the natural rate of interest, which balances growth and inflation. Indeed, the relationship between the financial stability interest rate and the natural interest rate may be one of the most important predictors of medium-term market direction and future crisis risk. A low financial stability rate versus the natural rate will create a tendency for real interest rates to rise to levels that disrupt financial relations. Factors that lower the financial stability rate include leverage and asset quality in the financial system. It is possible to build time series of financial conditions and stability rates.

(more…)

Bayesian Risk Forecasting

Portfolio risk forecasting is subject to great parameter uncertainty, particularly for longer forward horizons. This simply reflects that large drawdowns are observed only rarely, making it hard to estimate their ‘structural’ properties. Bayesian forecasting addresses parameter uncertainty directly when estimating risk metrics, such as Value-at-Risk or Expected Shortfall, which depend on highly uncertain tail parameters. Also, the Bayesian risk forecasting method can use ‘importance sampling’ for generating simulations that oversample the high-loss scenarios, increasing computational efficiency. Academic work claims that Bayesian methods also produce more accurate risk forecasts for short- and long-term horizons.

(more…)

Signaling systemic risk

Systemic financial crises arise when vulnerable financial systems meet adverse shocks. A systemic risk indicator tracks the vulnerability rather than the shocks (which are the subject of ‘stress indicators’). A systemic risk indicator is by nature slow-moving and should signal elevated probability of financial system crises long before they manifest. A recent ECB paper proposed a practical approach to building domestic systemic risk indicators across countries. For each relevant categories of financial vulnerability, one representative measure is chosen on the basis of its early warning qualities. The measures are then normalized and aggregated linearly. In the past, aggregate systemic risk indicators would have shown vulnerability years ahead of crises. They would also have indicated the depth of ensuing economic downturns.

(more…)

How to estimate risk in extreme market situations

Estimating portfolio risk in extreme situations means answering two questions: First, has the market entered an extreme state? Second, how are returns likely to be distributed in such an extreme state? There are three different types of models to address these questions statistically. Conventional “extreme value theory” really only answers the second question, by fitting an appropriate limiting distribution over observations that exceed a fixed threshold. “Extreme value mixture models” simultaneously estimate the threshold for extreme distributions and the extreme distribution itself. This method seems appropriate if uncertainty over threshold values is high. Finally, “changepoint extreme value mixture models” even go a step further and estimate the timing and nature of changes in extreme distributions. The assumption of changing extreme distributions across episodes seems realistic but should make it harder to apply the method out-of-sample.

(more…)

Algorithmic strategies: managing the overfitting bias

The business of algorithmic trading strategies creates incentives for model overfitting and backtest embellishment: researchers must pass Sharpe ratio thresholds for their strategies to be considered, while managers lack interest in realistic simulations of ideas. Overfitting leads to bad investment decisions and underestimated risk. Sound ethical principles are the best method for containing this risk (view post here). Where these principles have been compromised one should assume that all ‘tweaks’ to the original strategy idea that are motivated by backtest improvement contribute to overfitting. Their effects can be estimated by looking at the ratio of directional ‘flips’ in the trading signal. Overfitting typically increases with the Sharpe ratio targets of the business and the scope for applying ‘tweaks’ to the strategy. Realistic strategy performance expectations should be based on a range of plausible strategy versions, not on an optimized one.

(more…)

Drawdown control

Containment of drawdowns and optimization of performance ratios for multi-asset portfolios is critical for trading strategies. Alas, short data series or structural changes often render estimates of covariance matrices unreliable. A popular solution is risk-parity with volatility targeting. An alternative is ‘MinMax’ drawdown control, which builds on a broad interpretation of drawdowns as maximum actual or opportunity losses from not adjusting a benchmark portfolio to a specific underlying asset. In the case of one risky and one safe asset, this boils down to managing simultaneously the risks of conventional PnL drawdowns and foregone risk returns. Optimal asset allocation depends only on aversion to different types of drawdowns. Averaging over a plausible range of aversion parameters gives a model portfolio. Empirical evidence for the case of cryptocurrencies suggests that in an environment of uncertain returns MinMax delivers better PnL return-to-drawdown ratios than conventional volatility control.

(more…)

Realistic volatility risk premia

The volatility risk premium compensates investors for taking volatility risk. Conceptually it is based on the difference between options-implied and expected realized volatility. In equity markets this premium should be positive in the long run and fluctuate overtime depending on the market’s willingness to pay for protection against future changes in price volatility. In practice, measuring the premium overtime is challenging, particularly because expected realized volatility is not known. Using recent realized volatility as a proxy can be highly misleading. However, a realistic estimate can be constructed by considering the trade-off between timeliness and noise ratio of recent price changes and the long-term mean reversion of volatility. This “realistic” volatility risk premium has been positively correlated with subsequent daily volatility index future returns.

(more…)