How to develop a more
suitable risk management
framework
I often hear comments about how many times we’ve experienced a 1-in-20 risk recently. Over the last 20 years we’ve seen a global financial crisis, a global pandemic, a housing crisis, double digit inflation, a gilt crisis, political populism, the Russia-Ukraine war… the list goes on.
So, what do I say to this? Well, the reality is, if you’re measuring 20 different risks, you’d expect a 1-in-20 risk to hit roughly every year.
SECTION 1
Understanding and
embracing risk
I spend a lot of time talking to people about risk, and this unfortunately risks people not wanting to talk to me! However, we live with a huge amount of risk each day and we need to give it the attention it deserves.
We can’t escape risk, and if we can eliminate an individual risk, we’re likely to create or increase a different one in the process. A former colleague of mine used to describe this as packing a suitcase for holiday - you sit on one corner to get the zip closed, only for another to pop open.
Also, not taking risks can pose a risk in itself. I’ve had this discussion with my parents – by their own admission, they’re pretty risk averse (is it surprising I became an actuary?!) – and they invest in cash-like assets to ensure “they don’t lose money”. This suits their risk tolerance but, for many investors, the problem with cash is that you can guarantee you will lose money. In real terms, inflation will erode the value of that money.
We need to embrace risk (and its opposite cousin: opportunity) but this requires understanding it better than we have in the past.
SECTION 2
Value at Risk -
a measure for the past
In the pensions world, risk management has been simplified to improve the efficiency of decision making, often wrapped up into a single risk metric: Value at Risk (VaR).
This simplification was a noble aim to avoid decision paralysis, but it is no longer fit for purpose.
VaR was broadly effective when pension scheme outcomes were historically dominated by two key risks: equity markets and gilt yields movements.
Both these risks are captured reasonably well by VaR because the distribution of daily changes in their value is a bell curve (see the charts below), and VaR relies on risks being shaped like a bell curve.
Daily movements in gilt years (15-year maturity)
Number of days (y-axis) per quarter percentage point size move (x-axis) since 1979.
Source: Bank of England
You can read more about gilt yields, inflation, and the risks and opportunities associated with them here.
Daily movement in equity markets (FTSE All World)
Number of days (y-axis) per half percentage point size move (x-axis) since 1993.
Source: FTSE
a statistical measure of financial risk/loss based on probabilities, typically 1 in 20 (or greater) is used for pension schemes.
struggling so much to choose between different options given 'information overload' that we eventually do something completely different or nothing at all.
SECTION 3
Why Value at Risk is
no longer fit for purpose
As schemes have de-risked (see chart below showing how the average asset allocation for pension schemes has changed over time), equity risk and interest rate risk (which used to dominate) have reduced materially. The problem is that the industry has continued to use the same risk metrics.
Average pension scheme asset allocation over time
From 2006 to 2022. Source: PPF
We have been managing down VaR over time as our target, but it is simply not fit for purpose for dealing with the risks that remain.
For many schemes, two of the most significant risks now are credit and longevity risk and you cannot measure either with VaR effectively – these risks do not fit the ‘bell curve’ shape (despite some VaR statistics including them).
Longevity risk and credit risk are ‘tail risks’ - the extreme items are more likely and impactful.
Typical corporate bond expected payment profile
Probability of different outcomes for corporate bond.
Source: Barnett Waddingham
SECTION 4
Further issues
Another key problem with our existing risk metrics (VaR) is that they ignore a whole series of risks that cannot be easily measured by historical data. Examples include:
- Operational risk – the gilt crisis exposed this.
- Regulatory risk, such as Retail Price Index (RPI) reform.
- Benefit/data risk – not least the administration of this.
- Impact of GMP equalisation – adding to liabilities.
- Liquidity risk – see the gilt crisis again.
- Curve risk - this remains even with liability-driven investment (LDI).
- Investment manager risk, for example fund failure/winding down.
- Inflation mismatch risk – upset funding levels during the first half of 2022.
- Risk of insurer failure – we could easily delve much deeper into this topic.
where the forecast of future interest rates fluctuates but not in the same way across all years. For example, the 5-year interest rate forecast may rise by 1% per annum (pa) but the twenty-year interest rate forecast may fall by 0.5% pa. Without a perfectly matched hedge, these differences can lead to mismatches between asset and liability movements.
All the above impact pension scheme funding but are not in existing risk metrics.
With VaR we’ve experienced ‘what gets measured gets managed’. However, this leads to schemes making perverse decisions that do not take account of the wider risk position. It’s what’s known as Goodharts Law - when a measure becomes a target, it ceases to be a good measure.
Such decisions can ultimately lead to systemic risk – the risk of collapse of an entire financial system or market, as opposed to the risk associated with any one component.
The gilt crisis last year was a prime example of a risk that wasn’t measured in traditional risk metrics - it was a tail risk that came to back to bite the pensions industry.
SECTION 5
So, how can we measure
risk more effectively?
We need to move beyond traditional risk metrics (like VaR) and develop a more appropriate risk management framework.
Here, we can take lessons from the insurance industry by looking, not at VaR-like risk measures, but at stress and scenario testing our risk metrics instead. This means:
- Stressing our portfolios for acute credit risk and liquidity tail risks, including how this might occur and how a portfolio could be rebalanced if a stress occurs. For example, what happens if the counterparty investment bank to the derivatives portfolio closes, like Lehman Brothers? We would also need to consider the replacement risk and costs to these derivatives.
- Measuring risks that have traditionally had less focus, such as curve risk and inflation mismatch in our schemes.
- Scenario testing decision making in acute risk events such as operational risk events or regulatory changes.
- Understanding which risks are directly manageable and which aren’t, and developing ways that the manageable ones can be mitigated.
What this will likely lead to is:
Essentially if you have two risks, one you’re measuring (investment risk) and one you’re not (non-investment risks) it doesn’t mean the latter risk isn’t there. Allowing only for investment risk would lead to greater de-risking of the investment portfolio than allowing for both investment and non-investment risks, because the two types of risk would counterbalance each other.
Therefore, having a broader risk management framework avoids excessively de-risking the investment portfolio, thus maintaining higher returns without impacting overall risk (once you have allowed for the non-investment risks) which improves funding over time.
A framework for monitoring and documenting key risk metrics that doesn't over-simplify and cause perverse risk management decisions, for example the over de-risking your investment portfolio described above.
SECTION 6
In conclusion
Returning to the original point, we can’t eliminate risk but we can try to understand it better.
With greater understanding we can stop ignoring risks that don’t fit into existing frameworks and instead, move to a risk measurement approach that is fit for purpose for de-risked schemes. We will also be able to take advantage of more opportunities if we can assess them more effectively.
Do you want to achieve a better risk framework for managing and monitoring your scheme? Please visit our risk management page to see how our experts can help. You can also contact David Barnett directly.
Risk management services
DB Navigator is our decision-making framework. We'll give you all the tools you need to help you sail smoothly towards your defined benefit (DB) pension endgame.
Find out more
BW Risk Portal
Unlock over 1,000 hours of topical business risk management content, guidance and training in our BW Risk Portal.
Explore more