How to Account for Tail Risk in Wealth Management
Our software is a breakthrough in financial modelling, applying advanced algorithms, machine learning and data mining to measure our clients’ key performance indicators under any future scenario. It enables a new range of financial analyses that were previously impossible.
portfolio management, risk managment, scenario planning, scenario, financial strategy, investment plan, wealth management
16854
post-template-default,single,single-post,postid-16854,single-format-standard,bridge-core-1.0.4,cookies-not-set,ajax_fade,page_not_loaded,,qode-title-hidden,footer_responsive_adv,qode-child-theme-ver-1.0.0,qode-theme-ver-18.2.1,qode-theme-bridge,qode_header_in_grid,wpb-js-composer js-comp-ver-5.7,vc_responsive

How to Account for Tail Risk in Wealth Management

Traditional measures for assessing portfolio risk are too reliant on historical data and do not properly account for rare, market-altering events. Given the ominous signs of a potential 2020 recession, asset managers should adopt a dynamic, forward-looking, multi-scenario approach to portfolio benchmarking.

Currently, we’re in the longest period of fairly stable markets, but it is still critical to get ready for a potential downturn.

Traditional measures of risk in investment portfolios are always challenged during crises, when paradigm shifts occur. Standard risk measures are based on historical performance and fail to foresee the dramatic changes in correlations that always occur when the market takes a dramatic turn for the worse. Fortunately, modern risk management techniques can help dealing with such regime shifts.

A portfolio manager’s job is to outperform the benchmark. During crises, however, all benchmarks are suddenly down, as we have seen several times in the past 20 years. Under such circumstances, outperforming the benchmark is no longer an appropriate objective for portfolio managers, and they should therefore switch their attention to absolute performance while adjusting their benchmarks.

Traditional vs. Alternative Risk Measures

Modifying the measure of risk is one of potential ways of adapting to changing market dynamics. The traditional measure of risk-return tradeoff is based on classic mean-variance portfolio optimization framework – the cornerstone of which is the assumption of constant portfolio variance-covariance over the investment horizon.

Combined with the static (mostly normal/lognormal) distribution of underlying asset returns, this assumption might produce deceptive confidence intervals for portfolio values over the longer term, when volatilities and correlations dramatically change. Taking this problem into account, the portfolio benchmark should be revised according to the dynamics of market and the economic environment.

While the traditional approach to benchmarking works well enough in stable market conditions, it can be extremely deceptive if the probability of tail events increases. Indeed, since tail risk is often unprecedented, risk measures based on historical performance are not an adequate solution.

The common way of expanding risk analysis beyond available history is to consider a few handcrafted scenarios for stress testing. Since optimal allocation (of assets or capital) is based on the classic mean-variance concepts and normalized distribution assumptions, standard stress testing usually doesn’t affect it.

The real solution lies in constructing much more dynamic, forward-looking distributions of potential outcomes. Of course, proper incorporation of tail risk might affect the entire distribution, dragging down the expected value.

Over a longer-term horizon, such realistic distribution of investment portfolio outcomes can improve multiple risk facets. It can, for example, help a firm align its portfolio strategy with its investment risk appetite and time frame, as well as to determine its tolerance to real “worst-case outcomes.” Furthermore, it can lead to better-informed reallocation decisions, eventually yielding better-than-expected returns.

This dynamic approach also addresses emerging regulations on the best interest of the investorquantitative suitability and other fiduciary rules. Additionally, it prevents panic withdrawals of funds at the bottom of the market by ensuring that investors are better informed about potential outcomes.

The Multi-Scenario Approach

Generation of multiple forward-looking scenarios is the best way to construct realistic distribution of investment outcomes that incorporate dynamic changes in market environment. Scenarios should encompass all the underlying variables that are needed to evaluate portfolio performance, both from cash flow and mark-to-market angles.

It is critical to incorporate potential market shocks that can simultaneously change correlations between all risk drivers and impact equity markets, interest rates and credit spreads. Scenarios should pay particular attention to increases in credit spreads, which can lead to accelerated rating transitions and heightened correlations between the downgrades and defaults. Eventually, these types of problems can have a snowball effect, causing much bigger dent in portfolio values than the standard mean-variance analysis can predict.

Correlations between risk drivers and portfolio values vary in the central part of outcome distribution and in the tails, and the greatest advantage of linking portfolio values to the scenarios is in identifying early warning signals.

Identifying risk drivers that are correlated with the tail portfolio outcomes allows risk managers to establish better-adjusted mitigation strategies, hedges and reallocations. Portfolio adjustments driven by early warning signals help to maintain desirable expected returns while reducing tail risk.

Scenarios that should be considered include yield curves, credit spread curves, equity indices, exchange rates and commodity prices – as well as macroeconomic variables like GDP, unemployment, housing prices and production indices. Specific portfolio assets can either be priced on the scenarios or linked to the scenarios using robust machine learning techniques, like regression with regularization. From the long-term perspective, it’s important to aggregate portfolio instruments with similar risk characteristics (e.g., short-term A-rated financial sector or mid-term BBB-rated industrial fixed-income instruments) into uniform segments.

While security-level analysis is needed for tactical buy-sell decisions, uniform asset segments are essential for rollover assumptions built into portfolio analysis. These assumptions can be subsequently fine-tuned, based on a respective scenario analysis.

For example, if a particular A-rated bond matures when single A credit spreads are two standard deviations higher than average, the matured amount might not be rolled over into a similar rating. (First, it’s outside of the investor’s risk appetite; second, the issuer probably will not go for such expensive funding). So, the investment manager might choose to stay in cash for a while (until credit spreads come back to normal levels) or purchase either a higher rated or government bond.

Since scenarios contain all critical information necessary to make these decisions, contingency plans can be derived, overlaid over the scenarios and analyzed ahead of the time – so that as a particular market environment unfolds, it doesn’t catch anyone by surprise. The investment manager can just pull the trigger and initialize all actions that were laid out by previous analysis of potential scenarios and derived in the best interest of the client’s objectives and risk appetite.

Parting Thoughts

As it can be used by both traditional wealth managers and widely spread, low-fee robo-advisors, this multi-scenario approach to investment portfolios is complementary to more tactical mean-variance optimization of assets. It is also fully transparent, incorporates all types of risk considerations and truly informs investors about not only the potential risks and outcomes but also the timing of potential recoveries.

What’re more, it gives a clear perspective on expected returns over time, enabling identification of opportunities (through consistent modeling of the entire distribution of risks) and dynamic modification of allocation strategies.

Given the signs of a potential downturn, risk management and investment communities should watch out for the tail risk right now

Published on Friday, October 04, 2019, GARP (Global Association of Risk Professionals)