20 February 2024

Stress testing and optimising the business plan

Risk experts discuss the practical realities of stress-testing the business plan, and what's missing from the strategic planning process, in the second of a two-part roundtable held by InsuranceERM and QRM


Charles Richard III, co-founder, QRM
Vinaya Sharma, managing director, QRM
Giles Fairhead, chief risk officer, Pension Insurance Corporation
Ruth Middleton, chief risk officer, AIG Life
Socrates Coudounaris, head of risk management, RGA UK
Finn Clawson, head of financial risk, Aviva
Robert Moorehead-Lane, group CRO, Aventum Group

Christopher Cundy, InsuranceERM (moderator)

Read part one of this roundtable here

Vinaya: When I talk with companies, time is certainly a constraint, as are financial reporting metrics. For example, in America, companies are having to deal with GAAP, statutory, management, tax... How do you juggle multiple reports when coming up with a plan that you feel is appropriate?

Giles: Ours is made difficult because each of those metrics can act in a slightly different way depending on the market environment. So when you're doing stress and scenario testing, I'd say one of our biggest limitations is around our ability to effectively do multivariate stresses on our business planning process and metrics.

If you want to ask what happens if an economic shock hits halfway through next year, that's very complicated to do in a business planning model. I think most organisations probably end up saying: 'Is it worth it?'

"One of our biggest limitations is around our ability to effectively do multivariate stresses"

Ruth: The model has expert judgements that go into it and your output could have a wide uncertainty band around it. Sometimes we just look at a proxy model and say, directionally, this is where it's going to go, and what management actions would be taken? We're interested in the 'so what?' discussion that comes out. Asking 'does this take us to three percentage points below our risk appetite?' is kind of irrelevant at that point because there's so much uncertainty around it.

Giles: I'm in the camp that it's pointless to turn this into too much of a science. But you need to be able to do something, and trying to work out a multivariate economic shock is very difficult because we're all using proxy models that have a lot of limitations associated with them. You don't want to give false comfort to your board either.

"When there are big shocks, it's more about your internal processes and reporting, rather than your forecasting and planning"

Finn: With some stresses – for example, equity shocks – you tend to assume some mean reversion over time – markets fall but tend to recover.

Some stresses – such as yield curve shifts – might be a reflection of a new world: we might go back, we might not. So, understanding what your sensitivities are for, and whether it's telling you something about short-term volatility or long-term structural change, is important.

Robert: When you've got lumpy business, or when there are big shocks, it's more about your internal processes and reporting, rather than your forecasting and planning. Are you able to look at what's going on in real time on the ground, as opposed to what did the plan say should happen?

Vinaya SharmaVinaya: When it comes to stress testing, do you have control over that process or are you sending it to the first line to do the work?

Ruth: If there are stresses that have been done by the first line, we will review and challenge those. If there are additional stresses that we would want to have run, then we have the capacity to do that too.

Finn: The first and the second lines use broadly the same system. While there are core sensitivities that the first line run, they don't necessarily want to pander to or indeed have the capacity to run every stress that risk conceive. Risk needs the capability to be able to run the stress tests themselves.

Robert: We would look at what the first line think is reasonable and opine on some of that. We collate views from different organisations and the general view of the market on the shocks that might occur, and present the stress and scenario plan to the board so we can get their view too.

That 'authorisation process' is quite critical because you want to get buy-in from the board that you have stressed the plan enough and that you have been rigorous enough in defining and determining what stresses you want to run.

"You want the first line to appropriately consider risk and to a large extent they do"

Finn: The first line are pretty risk aware, so it's not like we're telling them: 'These are all the risks to your plan'. You want the first line to appropriately consider risk and to a large extent they do. Given this, you shouldn't expect a formal plan risk opinion that's dramatically different to the risks highlighted by the first line. It would concern the Board if risk and the first line were highlighting completely different risks.

Giles: Most organisations have different models for different purposes owned in different places. One problem is balancing speed versus accuracy from our modelling suite: we have a lot of proxy models that are capable of getting us results very quickly. However, those are designed for day-to-day use in the business: they have been calibrated to small movements and there is a risk that those proxies break down at larger movements. The only way you can really do big stresses or multivariate stresses in to put them through our heavy models.

Vinaya: Are you seeing the planning process as a race to the finish line? Are people so busy trying to get the work done that picking a right strategy is lost in the shuffle?

Finn: There are always dates you have to meet because it's hard to get the Board or ExCo in a room, but planning is an ongoing cycle. You might have a strategy away day relatively early on after year-end disclosures where the strategy is discussed and debated before moving into months of formal planning and (re)planning. Ultimately we have customers, and the strategy helps ensure we meet their needs. The plan provides the structure around the strategy.

Charles: How do you optimise your strategy? Given all these opportunities, what is the best way to solve for that, giving us the least amount of risk?

Robert MooreheadRobert: The strategy and plan are points in time. But you live risk management as you go through that year and you know if you're doing it well if, when big strategic opportunities come up, it's not a mad dash. People aren't asking: 'How do we take advantage of this?'. Instead they are saying: 'We have everything lined up. As we know what our downside is, we can minimise our downside.'

This is the value-add of risk management: when the upsides appear and opportunities present themselves, we are structured in such a way that we can take advantage of that without adding to our risk profile. And when big opportunities come to the table, they can throw your strategy out the window anyway.

Socrates: That's exactly our business model. The first line, second line and to a certain extent the third line are saying: 'What deals are you doing? How are you assessing those deals?'. And we focus on being able and agile to take on those deals.

We've always got the ORSA half open. For big transactions we will ask: 'Do you need to open the ORSA? Do we need to restate the ORSA?'. There is never a start and a stop. It just carries on.

Charles: Do you use those metrics that you calculate in an acquisition model, to try and figure out where to deploy capital?

Giles: Within the business plan, we get to high-level numbers, but then within different areas of the business, we have optimisation engines that can optimise. But we tend to not to bake that into our business plan. Our business plan is very much about new business.

Vinaya: From a risk perspective, what areas you would like to add that are currently missing to the strategic planning process?

Ruth: The multivariate models that you can stress at any time in the year and which you can base management actions on. As we were saying, you can apply instantaneous stresses, but that's not necessarily how things work in reality. They can come on slowly, and during that slow period you can take management actions. Building that into a model is very difficult.

Giles FairheadGiles: We would all like models where you could run the stress over a longer period of time and the regulator is increasingly asking for this too.

That's probably best done in a business planning type model, rather than in our heavy models. I think we all know what the outcome would be - it would look a lot better than an overnight stress as it can more effectively take account of management actions.

The other challenge for us is trying to reduce the number of metrics we have to monitor. At the moment the model has to do EV, own funds, IFRS, solvency and liquidity; that's quite a lot of parameters to have in a business planning model.

Robert: The Holy Grail is if you end up with a process whereby you get a gradual scenario coming through into the model halfway through the year, but then you can also hit it with a sudden shock.

"Stresses can come on slowly, and during that slow period you can take management actions. Building that into a model is very difficult"

Finn: I would like to see more outside-in estimates used in planning to provide better context. For example, if your plan is stating that a product line is going to grow 5%, how does it compare to the market? If the market is growing at 2%, that's much harder than if the market is growing at 10%.

Charles: That consumer behavioural type of analysis is really tough to do.

Ruth: We have that discussion as part of the quarterly business review. It's part of the ongoing process to look at what data has come out, but it can be backward looking. There are other approaches: for example, if you're selling mortgage protection products, you can look at macroeconomic forecasts and infer what that might mean for mortgage demand.

Robert: You can think about growth versus market share, but you've also got new entrants, new products, retractions from the market and products being withdrawn. There are so many moving parts to how to understand where growth could come from.

Charles: A lot of work has been done over the years on how you define the new business. Many of our clients will literally do their new volume; it could be a target growth; or it could be average growth or could be a formula based on an index.

But I should say that everything you mentioned on the wish-lists is something that we actually are doing.

Part one of this roundtable is available here.