Risk Management Lessons Learned in 2020 Covid Crisis
Every crisis is an opportunity to learn and grow. Although history does not necessarily repeat itself, risk management flaws and weaknesses often do. To avoid making the same mistakes again, it is useful to reflect on lessons learned and in many cases re-learned.
To that end, we have conducted a survey of a select group of risk management professionals to tease out their reactions and conclusions to the events of 2020.
It is an appropriate moment to reassess risk management, both because we just went through a small market crisis and because it has been years since the big crisis of the Great Recession. Risk management hit its peaks in the aftermath of the 2008-2009 crisis. Risk management at that time was in many cases revised, new people were hired, new procedures developed, and more money invested. Risk manager compensation anecdotally may have peaked a few years later and stayed flat or even declined marginally as market conditions remained placid. The procedures that were put in place in the years after 2009 can stagnate when nothing much happens. Management understandably pays less attention to low probability, high impact events when they just don ́t happen. Now we had a real mini-crisis, with the potential for more to come and so now there is a reasonable reason to pay more attention to risk once again.
The lessons we have heard are explored below. The list is undoubtedly not comprehensive.
1. Hypothetical stress scenarios need to be used more. Many risk managers use hypothetical stress scenarios as well as historical ones, but the historical stresses are frequently easier to understand and to pay attention to. It is easy and wrong to learn the lessons of 2008-2009, put in place hedges or strategies to deal with those and then assume you are done with risk management. The future will not be like the past, even if it has some similarities. In particular, the liquidity stress of 2008-2009 lasted much longer than previous episodes of liquidity stress and possibly longer than future ones.
Hypothetical scenarios should be a key part of good risk analysis. They need to be plausible and credible. They need to resonate with risk managers and with investment managers. They need at times to employ stresses that are larger than historical in at least one dimension (for example high volatility without simultaneous new extremes in rates) without simply stressing every possible risk beyond historical extremes. And they mustn‘t be put in place and then forgotten.
Boards, with their diverse experiences and perspectives should also add scenarios that are their worst nightmares.
2. Stress testing needs to be revised periodically. Once a package of stress tests have been put in place, they should be revisited and revised from time to time. That process is partly science and partly informed judgment. Market conditions evolve. Perceived risks change. Market liquidity changes. Yield spreads, credit conditions and new issuances all change.
The Spring of 2020 now gives lots of examples of potential revisions. Were negative oil prices in your stress scenarios? Did the actual stress just experienced go beyond your stress scenarios? What about CDS spreads, implied volatilities, commodity prices? We give examples here of just three quick and extreme moves.
If stress tests have stayed in place unchanged for a while, then take a look at them and see if they can be improved based on recent market experience.
3. Stress test results should not merely be admired. One consequence of market stability is that the impacts of stress tests tend to be ignored. Stress tests that have been in place for years showing a low probability of very serious losses serve little purpose. But even new, plausible and creative risk analysis and stress testing are useless if the results are admired but ignored. Coming up with interesting stress tests that demonstrate exposures that are both serious and reasonably hedgeable need to be actually met with hedges or portofolio allocation decisions. If the reaction to the results is “h’mm, interesting,” then the whole exercise has little or no value. The responsibility lies with both risk management and portfolio management. The scenario needs to be serious enough and likely enough and the hedge cost reasonable enough to make the remediation valuable.
4. Good risk management needs to pay attention to gut feelings. Many of us saw Covid-19 expanding in China, acknowledged the risk of it spreading globally, and then ignored it because the spread was viewed as unlikely. Of course, it is easy now to say we should have anticipated the spread, but why was the potential risk often ignored?
From our interviews we find that one reason that the anticipation of COVID 19 was not always absorbed into risk decisions is because it was seen as unlikely and hard to quantify. Those are bad reasons.
If risk management only proposes hedging likely risks, then risk management becomes another part of investment management. The whole point of risk management is to specialize in the unlikely and high-impact risks, not the likely ones. Investment management and trading, in contrast, specializes in high-probability, low-impact situations.
The risk of Covid-19 was tough to quantify. But risk management must not rely exclusively on science-based, mathematical loss calculations. It must also rely on gut instincts and judgments.
5. Risk models and investment models may need to be validated daily during a crisis rather than quarterly or annually. Model risk validation has turned into a regulatory exercise rather than a risk exercise in some cases. The pain of quarterly, annual or even bi-annual model validation or of model risk analysis can be a useful check to continue to ensure the plumbing is working correctly, but it is not a substitute to ensure that risk and investment models are actually serving their intended purposes during times of market volatility. For those models that depend on observed historical relationships and correlations, a time of market stress is the time to do a model risk update or calibration. Model risk validation should be a daily occurrence during market stress, not an annual one.
6. Liquidity premia usually revert more quickly than fundamental risk premia. When fundamentals change, for example when credit risks rise and credit spreads widen, it is reasonable to expect those risk premia will stay high until more is learned about defaults and recoveries. In contrast, liquidity problems follow market volatility and a withdrawal of financing. The factors driving liquidity are usually resolved more quickly than the factors that drive credit.
The contrary lesson was learned in 2008 and 2009 and perhaps learned too well. The liquidity stresses of 2008 extended into 2009 and resulted in numerous failures. That element of the Great Recession was shocking at the time and may be one stress dimension that is less likely to be repeated in the future.
If your assets have dropped in price to the point that it would be wise to consider a stop loss sale, distinguish between the loss that is the result of a change in fundamentals and the loss that results from a worsening of liquidity.
Alliance Bernstein – Daily Fixed Income Trading Liquidity Update, March 24, 2020
“Bid-offer spreads in the most liquid on-the-run benchmark cash US Treasuries have widened by 2-3x vs normal market conditions.”
Bond Market Liquidity A Problem, Says Morningstar, March 24, 2020
“The cost of exiting global bond funds has shot up from between 3 and 13 basis points to between 12 and 124 basis points.
Corporate bond funds have also upped their exit fees, lifting them from between 10 and 55 basis points to as high as 2 per cent.”
We have heard of dealers that specialize in providing liquidity having a very profitable Spring in 2020 as risk mitigation actions forced investors to reduce risk positions.
JPMorgan Trading Bonanza Creates Bonus Dilemma, April 15, 2020
“As the VIX measure of S&P500 equity volatility hit an all-time high of 82 in March, bid/offer spreads were at times as wide as 7% of its price, for example, and bid/offers for variance swaps (which are derived from strips of option prices) went as high as 20%.”
“Fixed income revenue growth in the first quarter was driven by increased client activity and wider spreads in rates, currencies and emerging markets trading.”
Risk reduction can be critically important in an adverse market, but it is worth using judgment to distinguish between the part of the price decline that is due to fundamental risk and the part due to market liquidity. It may advisable to wait for a few weeks on the market liquidity risk component to see if it reverses and only cut risk insofar as fundamental risks have gone up. Distinguishing between the two is of course a matter of judgment, but risk managers are supposed to exercise that judgment.