5 March 2012

Solvency II roundtable: embedding processes and live testing are benefits of "extra" year

Priming the board, the use test, the ORSA, "walking" the regulators through insurers' business and processes - these are some of the benefits of the delay in full implementation of Solvency II cited by the CROs and Solvency II experts who debated these issues in an InsuranceERM roundtable, sponsored by SAS.

Note: this is the concluding part of the discussion; the first part was posted on 22 February (see Solvency II roundtable: how to use the "extra" year)

Participants

Greg Gould, Chief Risk Officer, RSA UK
David Gulland, Chief Risk Officer, MGM Advantage
Simon Kirby, Consultant, SAS UK
Andy Sharpe, Financial Risk Director and Lead on pillar 2 for Solvency II, Prudential
Penny Shaw, Head of Risk Management and Capital Planning and Solvency II Lead, ACE European Group
Dave Smith, Head of Actuarial Development, LV=
Peter Field, Partner, InsuranceERM (chairman)

Peter Field: How are you preparing the board for their extended responsibilities under Solvency II?

Penny Shaw: The level of detail the boards are expected to know is very extensive. They are going to be subject to quite intense interviews with the FSA. We have had half a dozen extra board meetings and lots of training sessions.

Dave Smith: The CEO is not going to need to know the fine details of the matching premium calculation. I do not believe the FSA is going to require the whole board to have a detailed knowledge of everything.

Penny Shaw: Should a non-executive director (NED) really get into the detail of how we model credit risk? Is that their role, or is it more the strategic direction?

Simon Kirby: Every supervisor will want to know: does the board understand what this is about at the generic level and why it is important, and are they challenging the business effectively? That is what the FSA want NEDs to do: to challenge.

David Gulland: Do you mean challenge the output of the model or challenge how businesses use them?

Simon Kirby: That challenge is really up to each board. If they do not trust the model, because for some reason the people that run the model have not convinced them enough that it can be trusted, then they should challenge it.

David GullandDavid Gulland: To my mind, the board should not need to get into the detail of the model. It should absolutely understand what the outputs are telling them and how they are being used in the business. They ought to be able to rely on their risk function to explain exactly how the model works and the finer detail, or to just provide an assurance that they can rely on this output.

Simon Kirby: I think it depends on the individuals on the board. Some individuals have got a very technical background.

David Gulland: But there is no requirement to have somebody on the board who has that?

Penny Shaw: Board composition has changed. Certainly when I spoke with FINMA [the Swiss Financial Market Supervisory Authority] and Dr. Philip Keller, who was key in developing the SST [Swiss Solvency Test], he saw a transition of the membership of the board to become more technical.

Andy Sharpe: We do not have any NEDs on the UK board, although the group board does. We have chosen to write our ORSA report as though there were an NED on the UK board, and answer the question, "What would they need to know?" We think that is a way to ensure that the board does get the opportunity to look across the full operation.

Peter Field: What form does board training take?

Penny Shaw: We went for an external person who is already a NED himself and very much up to speed on Solvency II. Then we arranged a series of board meetings where we really went through the various pillars. Then the ACE teams arrived and said, "You've just heard the generic piece, now here is our Solvency II balance sheet for the syndicate and for the company, and let us look at it compared to GAAP".

The same happened with the internal model. We put a specific training pack together that showed how you determine undiversified risk and how that flows through the model in simple and transparent steps for each risk category and class of business. We felt that the board could then easily apply that knowledge to other models, focus on the idea that transparency equals understanding and require that - thereby pushing back on models that are "black boxes". It is an ongoing process but board understanding is of key importance.

Greg Gould: We have learnt some lessons from the ICA reviews and what they asked the board to do from those.

Penny Shaw: What is not clear is the differentiation between the NEDs and the executive directors. How do you get that balance right?

Peter Field: Let's move on to modelling. In order to get your model approved, how much of your Solvency II compliance do you have to show has become "business as usual" even before the IMAP has been completed?

Dave Smith: I have always thought if you are aspirational, you are heading towards the right thing, and you have a firm road map and actions showing you are getting there, compliance should be relatively straightforward. But for IMAP, we have taken the attitude that we are going to demonstrate that we are compliant, and, for the stuff that we are not absolutely compliant on, we have a very firm road map that we are in the process of delivering.

Penny Shaw: I think we have taken a similar approach on the use test. Solvency II is not live for another year essentially, so you have that time to build up that year of evidence with the new Solvency II model.

Dave Smith: If I was the FSA and if firms had not got fully-embedded use tests, I would understand this more than them not being able to do the numbers.

I just do not really get a sense of what the things are that mean a complete red flag if we do not demonstrate them.

Penny Shaw: Part of the issue the FSA is having on model review is because of the changes in the wider IMAP team. At Lloyd's, they had the same project team who very much set the direction.

When we put the list of documents for IMAP together and went through a six-month process of feedback with the FSA, it was so detailed we thought, "How are they going to read this?"

We suggested walk-throughs would enable more effective transfer of understanding of our model than just documentation. We have had a series of walk-throughs with Lloyd's, and the FSA attended all of them, and now the FSA is conducting a number with us.

However, you still do not really know what is coming round the corner in the IMAP process.

Peter Field: Does the regulators' approach make sense, then?

Andy Sharpe: It does worry me that we are building things to satisfy the regulatory requirements rather than demonstrating that we are building a model that is appropriate for our own company.

Dave Smith: And we understand our risks.

Andy Sharpe: Of course, they have to regulate and they have to set standards. Actually, when they come in to have a look, some of it tends to go out the window. There is a bit of a danger that we all get worried about the FSA as if we were preparing for an examination.

David Gulland: What worries me about that is the lessons learnt from the last eight or so years since the ICA. The regulators all start off with proportionality, but then, as they go round the companies, they ratchet up the requirements.

Andy Sharpe: I think it would be very hard for us to say to the FSA, "We do not see the need to do that", or, "We do not see the need to do that to the same level that you do". It is quite a challenge.

Dave Smith, Peter Field and Andy SharpeSimon Kirby: You have to demonstrate why -- you cannot just say that.

Andy Sharpe: It is quite a hard conversation to have.

David Gulland: We should probably insist that we should know what our companies are doing and therefore we design our processes and governance structures for reporting for our purposes.

Greg Gould: You should be able to have a discussion with the FSA and say, "That is the outcome you are looking for, this is how we currently do that", without going down the process you are describing.

David Gulland: Is there actually a danger that it will get ratcheted up on a European level as well? For example, the Spanish authority likes people doing this, therefore, "FSA, why are you not forcing your companies to do that?"

Greg Gould: We find the Swedish regulator is the toughest and we tend to have to almost use the UK regulator against them. We have to say, "This is okay in the UK. Please explain why it is not okay in Sweden."

Peter Field: What problem do you face trying to model for a whole group?

Simon Kirby: The whole point of Solvency II is to try to ensure that the group as a whole is solvent, not just each individual subsidiary within it. The concept of contagion risk from one part of a group dragging down the rest of the group is very real and has happened.

Andy Sharpe: Our group being domiciled in the UK means that we have exactly the same people supervising us. The group is making a submission, of which the UK solo entities are part. We have conversations and at times you have to stop and say, "Are we talking about the group, or are we talking about the UK here?" It is complicated enough for us, and we understand what is going on. For the FSA guys it is quite tricky.

Peter Field: Pushing back full implementation of Solvency II presumably helps on the systems and modelling side?

Dave Smith: For IMAP we are using our ICA process but ensuring we can meet all the six tests.

In parallel, and not for IMAP, but rather for our long-term vision, we are changing our core modelling capabilities. We are going to be building in a quick system for doing the proxy modelling. This is basically a cash-flow model, and a risk and capital model.

We want to move to multivariate market risk modelling. We are running that in parallel but that is not being used for the IMAP, because it is just too risky to try to do that by October. We are doing conforming things for the data side as well. We are a relatively small group, so it is probably a lot easier for us than it is for a Pru or Aviva.

Penny Shaw: It is putting it into the same level that you do your financial accounting, and the same control framework. The ORSA piece has been funnelling things right up to the board in a format that is very easy to digest.

Dave Smith: The pricing guys at the moment use spreadsheets for doing all their pricing. We use VIPitech [Algo Financial Modeler] for our financial reporting. We are building this one system which everyone will use across the business. Prophet will be the pricing and financial reporting core model. It's about getting consistency.

Penny ShawPenny Shaw: I think consistency has been the big underlying driver of what we have done. When we looked at Solvency II, we used it as an opportunity to start with a blank sheet of paper. We brought in some process mappers to document what we currently did. Then we brought the heads of each of the areas together and asked the question, "How do we improve the processes we have in place and also layer in Solvency II."

Simon Kirby: Firms that are doing an internal model will have a capital calculation engine that will essentially be the heart of the internal model. You also have to be able to calculate the standard formula, so that you can actually benchmark your internal model results against it and have that conversation with the FSA.

Equally, you also have to generate your own number as part of the ORSA process. Essentially, you have these three different numbers coming out. Are you using the same engine?

Dave Smith: Yes, the plan is to use the same engine. At the moment you can have peak 1, peak 2, ICA, pillar 1 Solvency II, pillar 2 Solvency II and ORSA -- a range of capital calculations. We are trying to rationalise to limit the sheer number of different balance sheets. IFRS 4 phase 2 is going to throw in another balance sheet. At the moment we are calculating on an engine that basically, if not in one run, in a series of runs, will do all of this.

Penny Shaw: We use Igloo, and Towers Watson has informed us that the intention is to build the standard formula into it which will facilitate more direct comparison. It would also be useful to have the rating agency capital models in the same "capital calculation engine" to aid more detailed comparison.

Andrew Sharpe: We might do a piece of UK annuity business, but that UK annuity business is just one component of a very complex group-wide model. Similarly, there might be something happening in another part of the whole organisation that still requires the whole group-wide model to be re-run, even though it is having no impact on our entity whatsoever because the group diversification changes.

I think this is one of the real challenges that Solvency II does bring. When do you re-run your whole group-wide model, other than when you are doing, say, your quarterly reporting?

David Gulland: Do we not actually need to have that in as part of the ORSA policy? We are drafting that now, trying to specify what events trigger when we do it.

Andrew Sharpe: Absolutely, but until you have had one, how do you know when it is the right time to do it? How do you know that, if you are doing something in the UK, or somebody is doing something in Indonesia, it is going to have that impact at the group level? You do not.

David Gulland: I think the focus of a lot of that is the capital diversification and who, which product line and which company or business unit should get the benefit.

Andrew Sharpe: In one of my past roles I discovered that we gradually became the biggest group risk. We went from being number three on the group's list to number one, partly because the group investment policy changed. As a consequence, the way we then had to deal with our capital on a group-wide basis became totally different, almost through no fault of our own.

Dave SmithDave Smith: It is probably easier for us because we are rather a small group compared to the Pru. Our long-term view is that every month-end should be the same essentially, so December should just be another month-end. In terms of the heavy lifting, running a best-estimate liability is a couple of runs.

Andrew Sharpe: You should have a good idea of how the results should come out, which means, if you have a massive surprise in there when you run it, then you have probably missed what has been happening for the last month.

Dave Smith: It is also not just the estimation at that point. It is also the very key thing of being able on a monthly basis to say, "What do you think the year-end one is going to be?" To forecast is almost actually more important than where you are now. All of the development that we are doing on the models at the moment, which is not part of IMAP, is to try to achieve a big step towards that goal.

Peter Field: How do you plan for the "extra" year when so much is still up in the air?

David Gulland: We are in the annuity market, so clearly the rules around matching premium are pretty important for long-term purposes. Do we invest now in line with our Solvency I requirements? Or do we invest where we think we might end up? We are trying to find an investment strategy which gives us the flexibility to adapt two or three years down the line.

Andy Sharpe: If you are writing a policy today, you expect that to still be there long into the new regime. I think it is incredibly difficult.

Simon Kirby: I think a lot of firms originally thought that they could use their ICA models to calculate the SCR, and then realised that they needed to change their models extensively.

If you have built something brand new, it is an interesting challenge. It would make sense to be up-front with the FSA and ask them what their expectations are: "What exactly are you expecting the use test to be? What do you need from us to demonstrate use of a Solvency II internal model?"

Penny Shaw, Simon Kirby and Greg GouldPenny Shaw: From ACE's perspective, the same assumptions go into both models for our volatility assumptions. The main difference is that the ICA uses UK GAAP and the Solvency II one is based on discounted best estimate.

We have done a lot of comparison of the numbers in the validation. We are still keeping the ICA live because we understand it and it aids comparison and validation.

The problem we have with the SCR calculations is that you have your "SCR one-year" and your "SCR to ultimate". We are using emergence patterns to get to the one-year. That is not really something that the P&C insurance industry has lots of data on, whereas at least for your projection "to ultimate" you have your triangles, you have all that information.

The one-year SCR is not really a very good measure of insurance risk, especially for longer-tail lines of business.

David Gulland: That does seem to me to raise a fundamental difference between what the life approach is and the non-life approach? Dave, how do you actually do that in a composite?

Dave Smith: The GI side has moved from undiscounted to discounted liabilities for the internal models. I think they are moving from the ultimate to a one-year time horizon. I am not that close to what changes it has made in the ICA business. But you are right. We have a big issue in terms of the whole aggregation piece and about diversification benefits -- whether we can recognise them or not.

Penny Shaw: I thought Solvency II was intrinsically designed with life companies in mind, with the one-year focus.

Simon Kirby: I do not think so. Solvency II has been designed to harmonise the insurance industry and try to put in place a set of common standards for all insurers.

Peter Field: Can everybody say what their priority would be for the extra year?


Dave Smith:
Embedding the use test.

Penny Shaw: I think we all agree.

Dave Smith: More time to manage your balance sheet.

Andrew Sharpe: I would vote for the ORSA. I think the more you can get out of your ORSA the better, of which embedding the use test is part.

David Gulland: I agree; it is the ORSA and the risk dashboard -- to be able to give the board what they need to understand what the risks are, and looking at the asset strategy with all the different possible outcomes of matching premium and capital charges. Trying to run a business, while not knowing what the future regime is like, is tricky.

Penny Shaw: I guess actually running a real life situation through it -- getting out of build and out of PowerPoint into results where you actually think: this is the impact.

If you can get that and have a full year of that being escalated up through management, I think you would feel a lot more comfortable about it.

Peter Field: You may have an extra year. Do you have enough resources?

David Gulland: We are under a lot of pressure and I am quite concerned about keeping going and trying to make sure people are proud of what they have achieved in the company. All our staff have done a lot of work, yet we need to keep on going and pushing more things out. From a management perspective, I think we need to think about that as well - about morale and workload.

Penny Shaw: Two years running now we have had a letter from the board saying "thank you" to the Solvency II project team. In my career of 17 years I have not seen much of that. It went down well with the project team.

Dave Smith: If you thought last year was bad, it is going to be worse this year in terms of burn-out and the amount of work.