Publications
Time to Slice and Dice Your Portfolio
There have never been so many ways to analyse a portfolio from a risk perspective. Whether the approach is regulatory, traditional or driven by data and advanced statistics...
...it is time to harness the power of the tools at hand and redesign the risk framework of portfolio management.
The Death of the Old Ways
It used to be that conservative portfolios would be managed through a mix of fixed income and equity instruments, overweighting the former in conservative profiles and moving the equity cursor higher as risk appetite grew. Over the past ten years, massive central bank interventionism, the advent of mega ETFs and funds, as well as the emergence of algo-trading, led to the questioning of this approach as correlations between asset classes increased as did the asymmetry of risk driven sustained high valuations and low interest rates. It is worth noting that many wealth management solutions still rely on this approach.
How to Measure Risk
What can I stomach?
The traditional measure of risk as a way to implement portfolios, besides rule of thumb approaches briefly described above, is volatility, which measures the dispersion of price movements around a mean value. The higher the volatility, the riskier the asset and, normally the higher the return potential. There are nonetheless many limitations to volatility, the first being that it measures both positive and negative movements where investors typically only worry about the later. Furthermore, central bank intervention has tamed volatility, giving an artificial sense of comfort only occasionally shattered by volatility spikes uncovering a growing market imbalance. Historical data is thus less relevant, while forward looking measures also suffer from their own shortcomings. Interestingly, historical volatility is the approach favoured by regulators in the MiFID and LSFin context, as a measure of investors’ tolerance to variations in their wealth, pushing banks’ risk departments to implement portfolio management guidelines which may have adverse effects. Is it really optimal for a conservative account to hold at least 70% of investment grade bonds in the current context of extremely low rates just because they have historically shown low volatility? Probably not.
How bad can things get?
In light of the above, investors have developed other risk measures focusing on loss of capital. Value at Risk or VaR estimates the maximum loss a portfolio can experience over a certain time frame with a certain degree of confidence (typically 95% or 99%). It can be computed based on historical data, simulations (Monte Carlo) and some innovative mathematical models are making their way with promising results. If a client does not want to lose more than 5% over a given year, a portfolio can be constructed to achieve this VaR with a 99% confidence over 252 days. But VaR can be misleading it does not cover the 1% remaining confidence probability over which the portfolio could lose 10, 15% or more. That could lead to an uncomfortable portfolio review.
If things do get bad, what is our expected loss?
Expected shortfall (or Conditional VaR) goes a step further and gives the expected loss with the same time horizon and confidence level given that the returns are already below the pre-specified worst-case scenario. It represents the average loss among the losses falling below a certain threshold and is a better measure of risk focusing on loss of capital.
What if?
Modern computational capabilities and data science have opened a new approach to risk management targeting what if scenarii. Assets can be broken down by factor (sensitivity to interest rates, spread moves, certain equity characteristics such as size, etc.) and these factors can be “shocked” according to wanted stresses. For instance, one may wish to establish the sensitivity of a portfolio to a shift and steepening of rates curves. Moreover, historical events can also be broken down into their own impact on factors and these shocks can be applied to today’s portfolios. One can thus experience the loss incurred should the same Global Financial Crisis, oil shock, taper tantrum or tech bubble happen today. While no crisis is ever the same, this is a powerful tool to ensure that the risk embarked in the portfolio is in line with the client’s profile and appetite.
No one size fits all
One can guess that since so many risk management approaches co-exist (and there would be many more to cover), there is no magic recipe. No single number or formula can return a perfect measure of risk and each formula is only as good as its underlying assumptions. All these methodologies should be used concomitantly with a focus on the ones most fit to each client’s best suited definition of risk. Some may be less tolerant to variations in capital while others may focus on the risk of permanent loss. The one thing that is certain is that more advanced risk management tools need to make their way to traditional portfolio management and should be part of any wealth management planning.
Mikaël Safrana, CIIA, CWMA
April 30, 2020
Publications
Publications
Letter from the CIO: "Resilience or the start of recovery?" - April 2024Read our latest Letter from the CIO publication titled "Resilience or the start of recovery?" here:
April 11, 2024
Publications
Letter from the CIO: "The bottle doesn’t matter as much as the drunkenness!" - March 2024Read our latest Letter from the CIO publication titled "The bottle doesn’t matter as much as the drunkenness!" here:
March 18, 2024
Publications
Letter from the CIO: "2024's Journey Begins" - February 2024Read our latest Letter from the CIO publication titled "2024's Journey Begins: January Review" here:
February 09, 2024