Delays and doubts about Solvency II implementation have created fresh challenges for insurers in managing their capital. In part one of this InsuranceERM/Towers Watson roundtable, participants discussed how it can be an opportunity to do things better
Graham Fulcher, practice leader, P&C UK & Ireland, Towers Watson
Ian Farr, global ERM product leader, Towers Watson
Julian Ross, chief risk officer (CRO), Talbot
Heloise Rossouw, head of Solvency II, Old Mutual
Mark Chaplin, group enterprise risk manager, Aviva
Simon Gadd, group CRO, Legal & General
Justin Skinner, enterprise risk management director, QBE European Operations
Penny Shaw, CRO, Hiscox
David Weymouth, group CRO, RSA
Andrew Pryde, CRO, Beazley
Chaired by Peter Field, InsuranceERM
Solvency II becomes less of a distraction
Field: What are the general effects of the delay in Solvency II implementation?
Gadd: I think it has to be considered now if, not necessarily when, it comes in. I think that gives a slightly different emphasis in terms of how much priority you give to it. It has probably had a disproportionate amount of emphasis in all of our firms over the last couple of years, and distracted some of us from our core role of developing risk management capability. I think now there will be less of an emphasis around necessarily Solvency II as such, but more an emphasis on risk-based capital generally on a more economic basis.
Field: How would that differ from what you would have done had you been preparing for Solvency II for next year, say?
Gadd: Probably less time for arguing about the rules and more time thinking about what we think is the right basis for our risk-based capital.
Ross: For us, a Lloyd's entity with Bermudian parent, our capital management will be no different over the next few years because it's not driven by Solvency II. It is driven by our desire to maintain our rating from the rating agencies in the US, which is no different now from how it was pre-delay, obviously.
Chaplin: In some ways there is no change, just the benefits we were hoping to get to – where there was a greater alignment between the regulatory regime and our internal economic capital-based measures – have been pushed further out, and perhaps even called into question.
That can create tensions, simply because you are trying to serve two masters or operate to two regimes. Those regimes can conflict, particularly in the life business in some of the continental European countries, where the local regime and the focus on book value accounting and smoothing is quite different and incentivises quite different risk management actions to those incentivised under an economic capital-based regime.
David Weymouth, RSA: "We see the opportunity to ensure that what we do adds real value to the business, as opposed to following a quasi-political timetable"
Skinner: On the non-life commercial side, regulatory capital is less relevant, so any delay to Solvency II doesn't make a difference. It is all about rating agencies managing the economic capital, which does have some alignment with the regulatory capital, but unfortunately not a huge alignment.
A number of these insurers would breathe a sigh of relief, because it means they can take longer over projects and they can do model development over a long timescale. You do not need advance permission from the FSA [Financial Services Authority] for restructuring, and then a six-month model-change approval process. It makes insurance companies focus on what is important, which is managing risk, not doing submission after submission and reviewing consultation papers.
Weymouth: We see the opportunity to ensure that what we do adds real value to the business, as opposed to following a quasi-political timetable. In particular we believe that Eiopa [European Insurance and Occupational Pensions Authority] should reassess the burden of reporting, which is particularly onerous for those businesses with a multi-national corporate structure. This is also the area where we have slowed down our systems investment.
Dealing with the doubt
Field: How do you make a business plan that deals with the doubt about when Solvency II comes in?
Ross: I think it is somewhat academic. Our regulatory capital regime is already quite like Solvency II. So much has been invested in this by the FSA, and in our case, Lloyd's, that it will carry on looking something like Solvency II.
Rossouw: A large part of our business is in South Africa, so we are in a different position, given that SAM, which is the South African version of Solvency II, is going ahead with dual reporting during 2014 and a go-live date of 1 January 2015. Our South African business is preparing for this date, and well placed given the work that has been done in preparation for Solvency II. We perform economic capital calculations twice a year, and are focussing on further embedding and integrating our economic capital model within our business.
Chaplin: There are clearly some very technical debates around particular issues in Solvency II that will possibly have material effects on how we see our business. But provided we get a sensible outcome on Solvency II, it does not really affect us that much.
Simon Gadd, L&G: "The change of regulatory regime and the harsh calibration of Solvency II is just another stress that you test your plan against"
We are not taking for granted the ability to remove some of the additional regulatory constraints that exist under Solvency I, and are managing the two measures until we see potential progress on some of the sticking points. It is more a case of hoping for the best, but planning for the worst, with a continuation of the two regimes at the moment.
Fulcher: Are you finding that the bad way in which Solvency II is being implemented acting as a stress on your business?
Gadd: Yes, it is a stress. You have your plans, and then you stress your plans for a variety of different outcomes, and the change of regulatory regime and the harsh calibration of Solvency II is just another stress that you test your plan against.
Weymouth: Faced with regulatory change you always need a set of good planning assumptions. In our case we have a plan that says if Solvency II were to be implemented on 1 January 2016 we could meet the deadline. You work back from that assumption, have a series of checkpoints and you say at this point I've really got to press the button and do more stuff, depending on the state of knowledge at the time. In a way you are running two parallel universes. If it did happen and it happened in a hurry you have to be able to deliver it. The important thing is to manage investment with real care that every dollar you spend is worth spending in the event that Solvency II does not happen. That is easy to say but harder to manage.
Chaplin: Reflecting our view of the importance of our economic capital models, we have renamed our Solvency II programme the Economic Capital Infrastructure programme.
Effect of Solvency II on managing capital
Field: But are you all saying that Solvency II did not have any effect at all on the way you manage your capital, or that you are doing it anyway but maybe tweaked it at the edges?
Chaplin: There are various elements. Certainly one of the things we think about is how we manage our diversification benefits across the group and how we optimise that. Clearly, that will be different if we get to a position where a regulatory regime recognises that in a way that the Insurance Groups Directive and Solvency I does not. We still optimise it within those constraints under our economic capital regime, but we will be able to do more when we get to a regime that recognises diversification benefits.
Pryde: Like most people in this room who operate at Lloyd's, it is business as usual, in that the 2013 underwriting year was capitalised using a Solvency II model. I think that was a good experience both for managing agencies and for Lloyd's. It probably involved more teams than the ICA [individual capital assessment] model – not just the capital modelling team but the actuarial team, who for us sit outside the risk management function, and the finance team – and so required much more collaboration to get to the capital estimate. It has improved understanding across the organisation.
Fulcher: Do you think the Lloyd's businesses have gained from the fact that Lloyd's has accelerated the timescale, or lost?
Skinner: It has probably dragged the market up to where it needed to be, so it is a positive in that regard. If I look at our company operation, I think Solvency II is 80-90% relevant for Lloyd's. It was a necessary evil to kick-start the market in the right direction.
Shaw: When Lloyd's of London saw the first QIS [quantitative impact study] results, it must have caused considerable alarm, as it could have meant a significant capital increase for Lloyd's. Lloyd's well-run Solvency II programme was a reaction to a key risk. Okay, it did not actually materialise in the end or in the expected timescales, but Lloyd's just had to approach Solvency II in that definite manner.
Andrew Pryde, Beazley: "Solvency II triggered the thought process, and created a target to work towards"
Pryde: I am not sure things would have happened automatically without Solvency II. If I look at Pillar II, I think there is much more focus around evidencing, and it has really driven consistency for us. We probably would have got there anyway, but Solvency II triggered the thought process, and created a target to work towards.
I think most people sitting round the table would agree that the governance around the Solvency II model is far superior to an ICA model. The biggest benefit of that for us is bringing the business into the calibration and understanding how the capital model works. We had gone quite far with our ICA approach, but it helps ensure that committees and boards remain interested.
More time for troubleshooting internal models
Field: How do the delays affect the way you have to run your internal models?
Skinner: We are a long way through the IMAP [internal model approval process] with the FSA, so we are now dealing with a number of difficult-to-deal-with issues that were left at the bottom of the pile. We can now take slightly longer to resolve them. I don't think we will change what we are doing in the internal model other than we can extend our final development push slightly.
I think the FSA has acknowledged that there are things in Solvency II that are just not practical. You unsuccessfully try to implement it as per the regulation, and do not get an answer when you ask how others are implementing it. Partly because there is no right answer. There are three or four things that I can put in that category, and the delay will give us time to develop as an industry.
Field: Can you give an example of a nagging issue?
Skinner: There are two that have always jumped out at me. One is the realisation of reserve risk. Solvency II has one of the oddest risk measures from a non-life perspective, in that it makes no sense to say how much risk will you see or realise during the first year. If you take something like asbestos, it took 30 years to manifest itself, so you would not have capitalised against it. From a non-life risk and capital perspective, Solvency II just does not work. As an industry, I do not think we are very far down the line trying to do this assessment meaningfully
The other one is my pet hate, which is diversification, something we have no data for, and yet drives a huge component of capital requirements.
Fulcher: I remember many years ago in some forum where correlations were being debated suggesting to Lloyd's that they should just tell people what correlations they wanted managing agents to assume as otherwise people would largely be forced to make up the numbers. At the time Lloyd's did not want to go down that route but I think, with the benefit of hindsight, Lloyd's should have adopted that strategy.
Penny Shaw, Hiscox: "The main issue with the models is spurious accuracy and detail masking big assumptions, which is possibly a systemic risk"
Ross: In effect, that is what Lloyd's does anyway through its benchmarking.
Fulcher: Yes, but they do not do it explicitly. They do not just give their assumed correlations to the market. I think for diversification nobody, not even the largest managing agents, has enough data to parameterise correlations, particularly in the tail.
Pryde: We take a slightly different approach, which is a 'drivers of risk' approach of modelling the market cycle rather than using correlation matrices. This is a key design concept of our internal model due to our medium-tailed casualty business risk profile. The market cycle is indeed the first thing that Lloyd's itself talks about when they set the ICA guidance.
Our approach of modelling the market cycle correlates classes and years of account naturally, based on how the business sees the risk. So when we come to parameterise our dependencies we have quite a lot of data because we have got 50 to 60 years of the US combined ratios, and we overlay our understanding how our portfolio has changed. This exposure-based approach provides a prospective view of our risk. Of course, the result has implied correlations, which we use to validate our approach. We find our drivers of risk approach easier to explain because that is how the business thinks about risk.
Shaw: The problem with Solvency II is there is so much focus on the 1-in-200 or tail for all risk types and then the diversification between them, when we lack credible data to actually measure it. A benchmarking-type approach that is simpler to understand and compare across risks and companies would have won a lot of support in the industry board rooms. The main issue with the models is spurious accuracy and detail masking big assumptions, which is possibly a systemic risk.
The real value for business in all these models is that they measure volatility around the mean. That is something that we did not as an industry routinely have before. Everybody had their business plan and a prospective assessment of capital to support the plan, but now we have a tool to look at volatility around the business plan, i.e. the probability of making a loss. That is actually much more interesting information.
Julian Ross, Talbot: "This delay is very much the way that Pillar I should have been implemented from the start"
Pryde: We have set risk appetite at a 1-in-10 likelihood which focuses the board on the risks they have under their watch and what earnings volatility they are prepared to accept. Traditional risk management techniques have a two-by-two matrix of frequency and impact. But I have always struggled with trying to compare risks on a frequency dimension. How do you financially compare a 1-in-50 likelihood with a 1-in-100 likelihood?
So when we set risk appetite we focus on the earnings volatility at one point of the distribution, the 1-in-10 earnings, for each of our 57 risk buckets, and we manage against that. That in itself takes us straight back into your point; what risks are we facing now rather than what might happen in a 1-in-200 event. Our view of the 1-in-10 and 1-in-200 gives us a good range on the distribution, which we probably did not have under the ICA regime.
Getting used to the use test
Field: Do any of you feel you need to fundamentally change your approach?
Ross: This delay gives us an opportunity to enhance how we comply with the use test. This delay is very much the way that Pillar I should have been implemented from the start, in a phased way. One of the disservices done in Solvency II to my mind is that by being increasingly theoretical and prescriptive, it ignores how things are done in practice. For example, part of the validation process is for many people in the business fully to understand the internal model, what goes in and what comes out, and to offer feedback. That positive feedback causes you to change some of the inputs, and to decide you can use it here and you cannot use it there. Trying to rush to a day-one 'big bang' implementation to my mind just does not work. You are not really going to get much use and value out of a model that way.
Weymouth: I would agree. You are more likely to get business more engaged in working on risk-adjusted capital measures with a greater time period than banging it through and saying, 'This is Solvency II, here are the use-test criteria,' because they will get to understand it and see the value in it.
Skinner: Australia's regulator, APRA, has phased-in their Solvency II equivalent more naturally. They said, 'Here is the regulation, here is what you have got to do to have an approved internal model,' and companies took three or four years to digest what those requirements were, and it probably has achieved a better outcome.
Part two of this roundtable can be read here