With the European Commission due to release the Solvency II Directive in 2007, some of the industry's leading risk,ating, actuarial and broking experts offer their views on the growing significance of enterprise risk management.

Dr Frank Achtert
Instrat, Guy Carpenter (chair)

Oliver Peterken
Chief Risk Officer, Aspen Re

Mark Nicholson
Director, Fitch Ratings

Andrew Hitchcox
Chief Actuary & Risk Officer, Kiln

Steve Taylor-Gooby
Managing Director, Tillinghast

David Paul
Senior Manager of Insurance and Actuarial Advisory Services, Ernst & Young

Professor Philippe Foulquier
Director of EDHEC Financial Analysis and Accounting Research Centre

Charlie Shamieh
Chief Risk Officer, Munich Re

Andrzej Czernuszewicz
Partner, EMB

Frank Achtert: Welcome to this roundtable discussion on enterprise risk management and Solvency II. It gives me great pleasure to see all of you today. From a quantitative perspective, Solvency II is now gaining momentum, as shown by the CEIOPS QIS2 [The Committee of European Insurance and Occupational Pensions Supervisors' Quantitative Impact Studies 2] summary report in which more than 500 undertakings participated from 23 countries. The focus of this roundtable is on where companies are positioned in respect of enterprise risk management both inside and outside the Solvency II environment. Can we start with some general observations? Do you think that Solvency II should be viewed by the European market as an opportunity or a burden?

Steve Taylor-Gooby: We have done a lot of research among insurance companies and our survey, which covers just over 300 companies, suggests that the industry is leading the regulators on this subject. Improving risk identification, measurement and management is something that companies are doing whether or not the regulators force them to. I think companies are seeing the benefits, so I would say that it is more of an opportunity. Obviously, there will be some burdensome aspects as well, but I think that companies are going there anyway.

Andrzej Czernuszewicz : If we look at what has happened in the UK over the past few years, when the FSA [Financial Services Authority] introduced its ICA [Individual Capital Assessment] regime - which could be regarded as a precursor for what is going to happen in the rest of Europe - we saw that people went through the process of the new regime and as a whole they have not been complaining about it. They have seen the benefits of it, which is contrary to what we have seen in other capital setting regimes. In the Lloyd's market, for example, people complained a lot about it. Some have found the ICA quite painful but they have accepted it without the degree of complaints that you might have expected beforehand.

David Paul: The answer will depend on the size of the company. In the UK context, larger companies might have started doing economic capital models two years ahead of finishing their ICA whereas smaller and medium players will have done no more or no less than what they have needed to do to comply. I would imagine that pattern will be repeated in other European countries, as some will use their economic capital models that they have already built in the context of Solvency II. I thought it was interesting to look at the QIS2 returns in terms of the number of companies. Something like 95 German insurers subscribe to QIS2, and then there were 60 or so in France, including the mutuals. There are relatively few in comparison doing it in the UK. It makes me think that France, Germany and Spain will have a lot of smaller insurers that will find this a burden.

Mark Nicholson: It is certainly a much greater leap for them to get from where they were previously to the present, which is possibly why there was greater interest in their being involved. There is certainly a division between the smaller and the larger companies. This could be an opportunity for the smaller companies, whether they believe it is worth the pain or not. I suspect that many will see it as a burden initially, but ultimately it could be an opportunity for them. I suppose the only difference is for companies that are very small in niche sectors. The review that they are looking to carry out might be slightly more burdensome than the benefit that they might derive from it, but generally it is probably an opportunity.

Philippe Foulquier: I think Solvency II demands more systematic and proactive risk management and asset management, which is an opportunity. However, it will be a burden if the formula is not adequate. For example, it will become a burden on the efficiency of asset-liability management, if the equity holding is too demanding in terms of capital requirement. It would be dangerous if the company had to change its asset optimisation founded on a long-term view, just to satisfy the Solvency II rules, which by their nature have more of a short-term view. Naturally, the target will be an opportunity for the company but the calibration of the formula could be a burden if it is not adequate or in accordance with internal practice.

Andrew Hitchcox: David mentioned economic capital models and some firms are creating their own economic capital models. We have settled on having an economic capital model to run the business, which uses the same process as the regulatory capital model but has different outputs because we have different inputs to it. Our optimisation is done on an economic basis, subject to the impact on the regulatory capital.

Steve Taylor-Gooby: It would be easier for you if you could just move to a single economic capital model.

Andrew Hitchcox: There are features where we see the regulatory decisions being different from ours. For example, we would applaud group diversification benefits but regulators tend to be restrictive. We feel that we are going to have to live with two in parallel. I do not mean that we will have a double bill of them, but we are doing both at once and then weaving them together. Is that a flavour that other people have picked up?

Andrzej Czernuszewicz: I would agree with that. That is what we have seen as well. People are using the same models and processes but are looking at different outputs specific to the different requirements. The core process is the same, which is a benefit, but using different models makes it much more difficult to manage the business. I think that companies that initially rely on the formula will not see as many of the benefits. As they rely on that formula, it becomes more of a burden. Once they move across, they might see the benefits but that could take a long time.

Frank Achtert: Therefore there will be an incentive even for small and medium-sized companies to implement an internal model, even partially, to quantify the underwriting risk and the market risk in a probabilistic manner. That could at least get rid of the issue of inappropriate calibration in the standard formula which does not always reflect the company specific risk profile adequately.

Charlie Shamieh: I would challenge the inaccurate calibration on equities. Following the capital markets crises in 2000 and 2002, European reinsurers saw the reality of inadequate calibration of equities. There were not any explicit challenges in regulatory models at that time. I believe that when people apply the economics to the topic, they may be shocked by the answer. However, I think it is a good reality test; I do not think it is a theoretical issue. It is far more connected with the philosophy of whether good insurers and reinsurers should be dabbling in leveraged investment fund activity as opposed to their core business of underwriting. With the frictional costs associated with the former, I certainly believe that most European insurers should have learned the lessons from 2000 and 2002. I think that the key exception is the primary life business, where you are not taking on pure insurance risk but there is risk sharing with the policyholder. The new business is in fact dependent on the extent to which you are taking on the investment risk.

Oliver Peterken: When you look at the influences on an insurer or a reinsurer, there is no doubt that the primary driver of capital is the rating model, closely followed by whatever valuation model your shareholder market is using. Regulatory models would come third on that list. What worries me is that all of our effort is going into regulatory modelling, which may or may not bear much relation to the other constraints we face. I see a big disadvantage in accounting on one basis, with regulatory modelling on another, rating agencies on another and shareholder valuation on another. You almost end up with four systems to find your way through. The only thing you can say is that you probably end up with suboptimal decisions at the end of that.

Mark Nicholson: In terms of ratings models, what we have seen of the capital requirements in Solvency II, including the equity component, is closer to that of the regulators than has hitherto been the case. The challenge for ourselves as rating agencies is to look at some of the internal models and compare them with our own calculations. There are some areas where they are probably not going to be identical, such as in the area of diversification and the capital credit given for that. Overall, I see those two aspects coming together, although there are other constraints, not least the shareholders. I do not think there will be such a great divergence as there has been in the past because we are using one model that will cover the whole of Europe. In the past there have been many different solvency models, but now there will be one comparator. I think it should be closer together on that score.

Steve Taylor-Gooby: A lot of the companies that I work with see Solvency II as an opportunity to unify their models, building an economic capital framework that is risk-based and a shareholder value framework based on the same principles. There will always be compromises in a regulatory framework and I think companies do see this as an opportunity to bring the way they run their business into line with that.

Oliver Peterken: I think that is the intention of a lot of people. The paradox is that the most complex, most economically satisfying work is going into the Solvency II economic risk models. The problem is then how quickly the different players will move and abandon positions. The rating agencies have invested years of time in many cases on numerical, predictive models. The stock markets have their own ways of valuing companies. We have found that even when you disclose a lot of your ICA work it is surprising how far people are away in other constituencies from being able to use that. You can put up the risk capital by business line and show them the returns on economic risk capital, but you will still be valued on a very simple statistical basis. There are probably five to ten years of education required before they get really comfortable with what we are all doing.

Andrzej Czernuszewicz: I think what you are seeing is a change in the way that people use models. In the not too distant past, the essential view was that the regulator or the rating agency knew best. They told you what capital you should have and what to hold. That view has been changing first with the regulator, but now the rating agencies are following. The regulator's view is now that you should know your risks and that you should be able to model the risks. There has been a switch in philosophy from the regulator saying that it knows best to the regulator saying that the company knows best. The rating agencies developed their models based on the initial view, but they are now moving across to the view that each individual company should know its own model and risk. I think we are now in a transition phase, switching philosophies, and Solvency II is consistent with that.

Philippe Foulquier: While the convergence of the systems is a good thing I hope that we will not focus on it to the detriment of the different strands of the company. More precisely, a problem could occur if companies were to focus only on the management of volatility and value-at-risk with a short-term view, regarding Solvency II rules and IFRS standards. This could generate conflict between, on the one hand, efficient risk management, asset management and ALM, and, on the other, simply opportunistic arbitrages, making strategic decisions to offer outsiders a good perception of risk exposure.

Mark Nicholson: In terms of the way that we have been developing how we look at capital, I suppose it is a sort of triangulation approach. We are developing our own capital model on perhaps a more stochastic basis than hitherto has been the case, and we are also looking at the regulatory regime and the internal view of companies and their capital. Our primary mechanism will probably be our own capital model, although we will look at the others to see if there are great differences. Obviously if there are differences, there could be valid reasons for that. There has been a growing sophistication of companies' internal modelling, so we can have greater comfort in that in the future. In terms of the Solvency II regime, if there could be some internal consistency in the way internal models are assessed, that again could give it greater credence for informing our capital view. Of course, we will have our own view as well and we will use that as our primary mechanism. In terms of overall risk management, I suppose the ideal from our point of view is if the way in which capital is measured in the internal model and the way in which a company would use it for its own purposes are as clear as possible to show how it is affecting the precise risks within the business. Obviously there will be a slight difference, but the closer the better from our point of view. I suppose there are many different ways to look at capital, but I think it is converging much closer than it was a few years ago.

Frank Achtert: How does that fit into the overall picture? The Ernst & Young survey stated that 80% of the major insurance companies have started to implement Solvency II criteria, but only 20% believe that their own capital model is already up-to-speed and compliant with Solvency II.

Charlie Shamieh: We have been focusing a lot on the numbers and the technical models. At Munich Re, we took a very early decision to use the economic view because, especially for reinsurance portfolios, the factor-based approaches were not really built for them. Steering your business on that basis in terms of allocating capital and technical pricing was easily flawed. We made constant comparisons with the factor based approaches and it is more of a coincidence that you get a similar answer. For me, the most important thing is what is called under Solvency II, "Satisfying the use test". That means the extent to which the model is embedded in day-to-day management. There is a survey of the top 200 global insurers which shows that nearly two thirds of those have economic capital models so it is no longer just the domain of the sophisticated. However, only 20% use this to incentivise management so that it affects people's pay. We have certainly found internally that this is the way to satisfy the use test. It is quite simple; you allocate capital from this model to the divisional board members responsible. They must allocate it down to a meaningful level and they have a return base on that to ensure that we meet our promised returns to financial analysts. That for me is far more important than the comparisons to factor-based approaches or indeed the discussion about the convergence of these models. There will be convergence, but I do not think that you will ever get full alignment or full agreement. You can never have one model that is so tailored to everyone's business. All of the comparisons are of interest, but they are usually more of academic interest than in indicating something that is over or under capitalised. I think that would be the wrong conclusion.

Frank Achtert: Once Solvency II is in place and assuming that the regulator has approved your internal model, would you expect that it will always be necessary to make the transition calculation from internal modelling for management purposes to regulatory compliance?

Andrew Hitchcox: Having heard that only 20% believe their model will comply, my model does not comply with Solvency II today. However, if a regulator turned up I could give him my internal view of the firm. I am doing QIS3 over the summer and I believe that when Solvency II goes live I will have adapted that module. Although I am not there yet, I do not feel that I will miss the deadline.

David Paul: I think you have to remember that the statistic was from a survey of European insurers. I think the readiness of firms that have not been subscribing to the UK's ICA is probably quite different.

Charlie Shamieh: In Germany, our regulator has an arm dealing with cross sectoral risk management. These guys have been auditing the internal models of the banks for the past ten years under the Basel II concept in a very detailed modular form. For a group like Munich Re, it would entail ten people for ten weeks running through the organisation and looking at data, systems and risk governance. The model is more used as the Trojan horse for looking at absolutely everything. They will not just look at whether the maths is good, although that is part of it, it is much more about what you did based on the analysis. I heard a few weeks ago in Munich an example from Gerhard Stahl of BaFin [the German regulator] about the banking industry of a very sophisticated approach for operational risk. It was clear to him that it was purely incentivised by an attempt to reduce the capital charge and nothing to do with an attempt to manage the operational risk better. Those guys are there to see whether you are actually using the model, which is the key test. To be honest, if we get that far more practical approach to Solvency II I think it will be great for the entire industry. It will help people to manage their risks better.

David Paul: Within the use test, they will also analyse what you do once you have that. I think the other part is that they will also analyse what you did before you set up the model. Did you do a review of the risks of the business? Did you iterate that with your management? If you have not done that at the start, your management team is probably not going to engage. The Germans have been doing this for ten years in their own context, but I think that the whole science of scrutinising and validating models - either by external auditors or by the regulators - is not very advanced yet. I think it might be the IAA [International Actuarial Association] that is in the process of doing a working paper to set out the guidelines for that.

Charlie Shamieh: On the banking side, I would say that it is extremely advanced. If you see these guys in action, they are extremely impressive. They have sound practical experience of the banking industry and they are very strong technically. For example, they took our copula and recoded it themselves to test the capital impact. This is the way that we allow for correlation between different lines of the business, just to stress the extent to which we are being conservative. I think it is more about how risk management is using it and how sound decisions are being made in terms of risk mitigation and technical pricing.

Steve Taylor-Gooby: It is a mistake to focus too much on models. Perhaps the bigger change for a lot of companies will be the governance, process and control part of Solvency II rather than simply calculating a capital number, which in the UK and in Switzerland companies have found pretty easy to do without massive investment in systems.

Andrzej Czernuszewicz: I think that the use test is key. If you do not believe in it, why should anybody else?

Oliver Peterken: I agree with that but I would say that you have to understand how the usability of models and the modelling errors vary by line. There is a very specialist reinsurance operation and we spend a lot of time trying to understand the modelling error because we are dealing with distributions of liabilities where there is not reliable past data to price or model. What we saw in the reinsurance market in 2005 was a systemic failure to understand modelling errors just around hurricane modelling, let alone anything else. What emerged out of 2005 was quite worrying. A lot of the reinsurers who suffered unexpected losses in 2005 were probably what we would call advanced economic risk capital modellers. For a number of years they had taken a load of inputs for granted without trying to understand the variability of the results coming out. We are then faced with the situation where losses they thought they suffered once every 200 years were actually once every 20 to 50 years. That is enormous modelling error by anybody's standards. One of the things that the Solvency II project has not fully appreciated is the difference between modelling errors for large retail operations with fairly stable results and people at the more specialist end. We are all going into this same approach, but we are actually coming out with models with quite different parameters and usability.

Andrzej Czernuszewicz: You are almost describing the misuse test of models. The model is just meant to be a tool; you should not always follow it blindly. It is there to assist in decisions rather than make decisions for you.

Oliver Peterken: I would like to come back to a point about the usability and the use test. I think you cannot underestimate the influence of different shareholder groups. As a US quoted company, earnings per share and return on equity are the only things our stakeholders concern themselves with. A European quoted entity will have different performance metrics. When you go down the US route and combine it with quarterly reporting, what you are focusing on is delivering a quarterly financial result. Failure to do so has very quick and very severe repercussions compared to other markets. When you are in that situation, your economic capital modelling is an optimisation tool but what you are focusing on is that quarterly result. That is what your stakeholders are judging you on. To say to them that you are not doing well on RoE quarter by quarter, but that you have the most fantastic economic optimisation, does not really interest them. The short term is what counts. That, I would say, is an important limiting factor in how much you can invest in economic capital modelling. You have to get this connection between modelling the economics of your business by line and getting what your stakeholders are interested in. The regulators are not paying you.

Charlie Shamieh: In very simplistic terms, the convergence between economic measures and what investors are looking at has all got to do with the extent to which one needs to carry burdensome capital, or capital that you do not need to hold for the target level of security that you want to give to all stakeholders. If you have built the business from the economic perspective to be what is the best representation of your own portfolio but someone says that you still need an extra £5m, that is what is driving the RoE drag. I do think that it is entirely relevant for investors. The more that one can be very explicit about this and say that it has a cost to our investors, lowers returns and raises prices and potentially also leads to wrong decisions in management, then it is relevant to external stakeholders. I believe that this has led many rating agencies to say that they do have a point. If you compare the discussion this year with the discussion two years ago, I think rating agencies have very much moved to recognise the importance of what they call strategic risk management, not just looking at the absolute amount of capital that you hold but much more the sustainability of the earnings. I think that is a very positive development.

Philippe Foulquier: I have a more pessimistic view. For me, the problem is how, on the one hand, regulators currently consider companies to be well-capitalised and, on the other, regarding calibration, companies require two to four times more capital than their internal model allows for with regard to some particular risks. Secondly, if your internal model concludes that thanks to your optimal risks and capital management, you need three or four times less capital than the standard formula, what will happen? I understand why the companies that have an internal model are fighting over the definition of the standard formula. The gap appears too high and there is probably something wrong in the calibration.

Mark Nicholson: Whatever the type of capital test, I think there is always going to be some pressure between having what the regulators and rating agencies want to see in the form of capital and what a company's internal model would like to have given the pressures they have for return and so on. A lot of it is based around the incentives to individuals, and as a regulator or a rating agency you are naturally more risk averse. Coming to the company without the detailed knowledge that a company has of its own capital requirements, there will be an element of caution built in as an external viewer. There may be some moving together of the different capital requirements, but I do not think that it will ever be the case that a company gets precisely what it wants.

Oliver Peterken: Do you think that regulators know how much of a margin they want insurers to hold? It seems to me that rating agencies have a very clear idea and you can have quite precise conversations about what you need to be single A, double A or whatever. I was very surprised when the FSA came out with almost a triple B standard because that would be completely unacceptable to every single client we have. Why should that be acceptable to a regulator?

Steve Taylor-Gooby: It was not really a triple B standard.

Oliver Peterken: They tried to imply that it was that low.

Andrew Hitchcox: It is the standard that you do not want to get to, so you have to operate with your own view of your risk above that.

Oliver Peterken: Nevertheless, it was an interesting benchmark to go to.

Mark Nicholson: Historically, pretty much all national regulators have had lower absolute standards than we would want to see for investment grade companies, whatever market we are looking at. Again, it is uncertain. It could well be the case that for some companies the SCR [solvency capital requirement] they hold is less than the capital they are holding at present. Conversely, it may for some of the smaller entities be a higher level, especially if an internal model is not being run by them.

Charlie Shamieh: I would like to take issue with the fact that you can never really trust whether these things are working. In the context of having ten people running around your organisation, one would have to say that it is far better than a one day session with a rating agency and then some add-ons to that. I do think that the rating agencies will have to consider the extent to which and how they are going to cooperate with the lead regulator concept under Solvency II. We recently did a review of our internal model externally to get an independent auditor and I tried to do exactly the same with accounting firms. It was so clear that that view would be far inferior to a regulatory test. I say that quite provocatively because it was so apparent to me that given the scale of our organisation, there will not be anyone out there to give me a better view. They themselves are proactively testing proprietary models so surely that is a far more rigorous process than simply tagging along with one proprietor and saying, "Well, this is what I hold". If you think about Solvency II and the effort that regulators are going to spend under that, it is going to be very important for all other stakeholders who are looking at internal models to question the justification for holding additional buffers over and above this auditor model. I would fully agree if the model had not been properly subject to some independent testing, but not under Solvency II.

Mark Nicholson: The points I made were not regarding a regulatory auditor of an internal model; they were regarding internal models per se. Obviously, as with any piece of evidence, it is a combination of who created it and what it actually says. We will certainly look at the Solvency II regime and look at that in the context of our own models as well, trying to bring those together. I think that there will be a convergence in terms of the measures that we and other agencies are taking to assess capital and what actually comes out of Solvency II at the end. It will be interesting for us to get some greater certainty about what the parameters actually are for all of us.

Frank Achtert: On the insurance side, I think there are still some years to go before supervisors are as experienced as the supervisors on the banking side.

Charlie Shamieh: Realistically, this will only be happening with the largest groups in most of the countries. I do not think that BaFin intends to staff up to do this for 120 insurance groups.

Oliver Peterken: One of the things the insurance industry is suffering from is a lack of acceptance of this concept of a lead regulator, which I think banking has managed to establish. Effectively, we are all being regulated in multiple jurisdictions. We can agree one thing in Europe but then we can end up with dozens of different discussions around the world. What we are all trying to do is manage a single global balance sheet around the world, regardless of legally where the capital is. I get the perception that the banks dealt with these issues a number of years ago, but the insurance industry has not really progressed. Solvency II is almost making it harder to progress, because, certainly in the US, you do not see any movement on this issue.

David Paul: The signs are not good on that front. The last thing from CEIOPS on that seemed to be very lacking in courage about trying to get our own supervisor. Maybe part of the difference between insurers and bankers is that the raison d'être of insurance supervisors is customer protection rather than a protection of the banking system from a systemic risk. That leaves the national supervisors' heads on the block for consumer protection so you can almost understand why they seem to be fighting for their autonomy. It does clearly lead to something that is contradictory compared with the way that large multinational groups are trying to run a holistic capital approach in their organisations. The FSA and Treasury proposal is far more ambitious than what has come out of CEIOPS.

Steve Taylor-Gooby: I have been quite close to the political process in Brussels around Solvency II and there is a lot of political resistance to the Treasury's proposal.

Oliver Peterken: That is ironic because improved consumer protection comes from greater diversification and greater financial strength, not from local silos of capital.

Mark Nicholson: I think it is questionable whether the objectives of Solvency II have been met in terms of mirroring how groups manage their own capital, unless you have a lead regulator system. I think the diversification question has not been reconciled but it is probably the most challenging one. How do you assess that when different regulators have different views on how much credit to give or otherwise? It is hard to scale a system that will achieve its objectives.

Steve Taylor-Gooby: I think there will come a time because it is so obviously a sensible idea. It is just political opposition that has to be overcome, but that will take several years.

Frank Achtert: The implementation of a full scale lead supervisory system will take several years, even in the EU. If we go across the Atlantic, I do not believe it will be possible to have a more global view of lead supervisors in the medium term. I think it is important for international groups that a lead supervisor concept is established as soon as possible to reduce the regulatory burden and to benefit adequately from diversification effects at a group level. But, having a look at the discussions going on in the US with respect to the collateral issue on the reinsurance side, this is a sign that it will be very difficult to agree even on the more simple things.

Maybe let us take one step further and come to the question on current IT systems and the data warehouse facilities of the companies. What is the current state in respect of Solvency II? I am talking about smaller companies as well as multinational groups.

Steve Taylor-Gooby: I think that most companies will be able to be compliant on day one of Solvency II, assuming it sticks to the current timetable, but in a suboptimal way. To get to the ideal position where you have enterprise systems end-to-end from where the data is created through to producing your economic capital models, most companies have a long way to go. However, I think most companies will be able to do some stress test models pretty soon.

Philippe Foulquier: I think it is also linked to current national regulator models. The UK is in a good position because the ICA regime already covers a large part of the Solvency II requirements. Other countries have fewer additional rules or stress tests beyond Solvency I. The gap is very high. I am afraid that until now, smaller companies have not really cared much about Solvency II, firstly because the rules are not yet clearly defined and the 2010 minimum deadline is far away, and secondly, because they do not have the means. QIS2 exercise required several man months.

Frank Achtert: I fully agree. It is a little different on continental Europe from the UK. As the QIS2 exercise has shown, the gap between the data available in theory and the data that was forthcoming from business units was in many cases significant. That is obvious in Germany and France, and also in the Nordic area. Small and mid-sized companies have really suffered from the lack and quality of data. In the end, of course they have to be compliant but currently they have only just started to think about the process in more detail. That is clearly a difference between the two sides of the Channel.

Steve Taylor-Gooby: It is not just in the UK, though. I was impressed with the speed and the efficiency with which the Swiss companies took the solvency test on board.

Philippe Foulquier: The gap is still very high in Germany, France and, in fact, most continental European countries.

David Paul: I think that even small companies are getting to the level of data required to do their pricing and reserves, but there is quite a lot of reassembly involved in moving that on to calibrate the distributions that you use to build a model. I do not think that a revolution of the information systems is necessarily needed, but a lot of actuarial resources are needed to reassemble and build these models, especially if you have not started yet.

Oliver Peterken: There is a difference between building and operating them. It seems to me that there is plenty of consulting expertise and resources available to help get the models built. The real challenge is internalising the process, maintaining the model and getting people to use the results. The build stage is actually not that technically challenging anymore because a lot of people have done it. However, what a lot of people have underestimated - and I would say that it has taken us a good two years to get to it - is the change in the way underwriters and executive managers think about a business. The concept of economic capital is a much bigger jump for people than many of us really thought. The technical building process is not too much of a challenge, but when you have the model and you have to internalise the process to get updated results, it is a much bigger challenge than many people thought it would be.

Steve Taylor-Gooby: Even the companies that are furthest ahead, such as ING, would say that they are really at the earliest stages of a very long journey of having a common language in which all the senior managers across the company can talk and understand each other. It is a long way to go.

Oliver Peterken: You can start to get benefits very quickly. In our experience, just being able to tell people how much risk capital is being consumed by underwriting or by reserving risk, for the first time you are giving them a much more enterprise wide view of the company. Previously, they probably understood that more subjectively and qualitatively. As soon as you start quantifying it, people can be surprised at how much capital certain products require. However, it is not something that you can achieve overnight. It takes a long time and then, of course, you have to start to explain to people how that allocation of risk capital is changing and why it is changing. I have been very interested in the whole operational risk area, where I think insurers have generally completely underestimated how much operational risk they carry. The banks have had a much better handle on it for a number of years. In the banking sector, because the amount of capital for operational risks can be enormous, they have really paid a lot of attention to process and operational risk management. The amounts of capital for insurers' operational risk are really quite low, but if you look back at some of the issues, you could say that 2005 for insurers was as much about operational risk as about underwriting risk. People generally underestimated how much risk capital they needed for operational risk.

Charlie Shamieh: I am not so sure that the banks are ahead in operational risk. BaFin's advanced measurement approach under Basel II was providing incentives more on the side of people having very sophisticated databases, using extreme value theory and esoteric mathematical techniques to estimate operational risk capital. It is actually a good thing that Solvency II is more principles-based than rules-based. The key thing for us on the operational risk side is investing in more automated systems with audit trails that are banking standard to limit the tracking systems, like the algorithms that the Fitch group uses. A few years ago, we might not have considered that because we were not thinking in terms of the use test or making sure that the limit we are thinking of for value-at-risk in capital is exactly the same as at the portfolio manager level.

Frank Achtert: That is a clear statement that your organisation is at the top end in terms of preparation for Solvency II. For most others, operational risk is not yet on their radar screen.

Charlie Shamieh: Shaun Wang from the Enterprise Risk Management International Institute did a great study recently that would challenge this. He took casualty lines from small and regional players in the US and firstly looked at the numbers and then showed that they actually had a far better experience than the national players. Both the average loss ratio as well as the deviation loss ratio were far better the smaller they were. He then conducted a series of interviews to find out what was behind this, and it was actually because the portfolio was much simpler so you could run the business with very simple models and systems. You have far less turnover, so the underwriters had ownership for the portfolio that they were writing. As a small player, I am not sure that you are so far disadvantaged relative to the global players that can invest in the systems. I think there are some advantages as a smaller player using a simpler portfolio. You do not need elaborate systems, so you can do well.

Mark Nicholson: I suppose that is an area where companies might see it more as a burden. It is not that they do not have their operations under control; it is more about showing what they have done and quantifying it. Once you are doing it already, having to meet the requirements of the regulators in that regard could be more problematic.

Frank Achtert: How important do you think it is to have disciplined underwriting now and in the future to prepare for Solvency II?

Steve Taylor-Gooby: Is it not important anyway?

Andrew Hitchcox: Even without thinking of Solvency II, suppose that you are trying to change your model from a one-year model to a multi-year model. The cycle overrides everything for bread and butter reasons of validity, but it is good to make the managers wrestle with that as hard as they can.

Frank Achtert: Do you not see any additional impact from Solvency II? In the past, companies could absorb underwriting losses by investment gains but now the target capital will be higher under the regime of Solvency II. Therefore, disciplined underwriting will in some ways become even more important but I fully agree that in general, underwriting discipline has to be key to sustainable profitable business anyway.

Mark Nicholson: From a regulatory point of view I suppose it could be proved that previous rates have not been adequate in either direction, but the fundamental point has always been important.

Frank Achtert: How do you think Solvency II is driving enterprise risk management approaches in the industry?

Oliver Peterken: I think it has had a huge impact. Without Solvency II, I suspect that risk capital modelling would have gone down a very different underwriting and investment path. Solvency II, because it has come at it from an enterprise risk point of view, has made the risk capital models cover the whole enterprise. You have not had the option; you have to build your models across all the areas identified almost simultaneously.

Andrzej Czernuszewicz: I was going to say essentially the same thing. Solvency II has made these things much more certain than they were before and it can mitigate the silo mentality.

David Paul: Are you thinking of the standard formula approach when you say "silo"?

Andrzej Czernuszewicz: Yes, switching the transition from the standard formula to the full model. It assists people in getting to the final goal.

Oliver Peterken: The pot of gold at the end of the rainbow has been diversification benefit. I suspect that as an industry we are at a very early stage of understanding that. It seems to me that the real value comes from being able to get a very good handle on that, because that is going to add a lot of value to your value creation process. There has probably been less progress by the industry as a whole on that because everyone has been so focused on building the component models.

Andrzej Czernuszewicz: Related to what Charlie was saying about allocating capital to drive people through their bonuses, we have seen that it is helping the local behaviour but it is not optimal from a global perspective. We are essentially seeing systems being set up within companies to transfer the risk around. Rather than purchasing as much reinsurance as they would externally, they are using internal risk mechanisms to pass that around. That is a direct consequence of building their internal models, which is a switch in the way risk has been managed.

Steve Taylor-Gooby: I think a relatively small number of companies are getting quite sophisticated in the way they allocate capital, but not in the traditional way by product type or business unit but actually by risk type and managing diversification to optimise the portfolio.

Philippe Foulquier: Solvency II has an important role to play. As you mentioned, its target is to improve risk management. However, if things like hedging and better asset management are not all integrated into the Solvency II approach, I think we will lose its positive aspects. It is up to everyone to extract the maximum positive effect from risk management; this cannot be underestimated.

David Paul: Even in the larger organisations that have some capability in terms of modelling their capital management, although the general direction over the years may be positive for enterprise risk management, you have to fear that there will be a one year hiatus when you get to 2010 and the company has to comply. Even when the bigger organisations are trying to comply with a regime, there is bound to be a management overhead that distracts you from the more positive internal uses of your economic capital model. You can probably identify that having happened in the UK in 2004 and 2005. Bigger organisations have stalled trying to get the compliance right before going back to the added value.

Charlie Shamieh: I believe that is the danger of the regulatory regime being compliance orientated. I remember that the FSA did a survey two or three years ago on the ICA. It found that most insurers were doing things internally to keep the FSA happy, which was at the time when 50,000 consultation papers were coming out from the FSA every month. I remember thinking that the BaFin was being a lot smarter in working quietly behind the scenes and talking about principles. When we did economic capital modelling in 2000, we were not doing it for Solvency II, which was the last thing on our minds. We were doing it more because we thought that the way that capital was thought of traditionally in the whole insurance industry was not sending the right incentives. People did not consider it a scarce resource and it depends on what the motivator is. That is the danger of Solvency II becoming far too prescriptive and compliance-driven, because then people do get distracted. Right now, I would say that I am not at all worried about preparations. What I am worried about is making sure that we properly communicate it when capital numbers change so that people understand what that really means for their pricing or capital allocation. There are an infinite number of ways that you can allocate capital and also not allocate capital. There are good academic studies on this from guys like [ERM experts] Don Mango and Gary Venter. I think it is good to keep thinking about what would happen if we did change the policy on capital allocation. Frankly, that is nothing to do with Solvency II but it is because we do not feel that we are getting this prescriptive list of 50,000 things that someone will come and check.

Oliver Peterken: I agree with Charlie. Depending on where Solvency II lands on certain technical issues, the people who have been running risk capital models for a number of years could well stick to those risk modelling policies and just treat Solvency II as another regulatory return. Like Munich, we started building the risk capital models without an eye on Solvency II but for our own purposes. You get more and more comfortable with the risk modelling policies that you are adopting internally. If someone wants you to change direction, you probably will not because you are starting to orientate the whole business. In terms of performance measurement, the time it takes to get the underwriting management comfortable with how much risk capital they are being allocated means that you would not want to change that just because of Solvency II. That is what concerns me with how much is outstanding on Solvency II. It could still take some quite sharp turns to the left or the right.

David Paul: I think the balance of emphasis between the standard formulas and the willingness will come down to national supervisors and local implementation of the framework.

Andrzej Czernuszewicz: A few years ago the FSA said that the whole regime would only produce what companies should be producing anyway. As to whether that happens, we will have to wait for the final document.

Frank Achtert: How stringently will they define the use test? Is it satisfactory testing the numbers from and across different business units or will the test make you change your internal modelling approach? That then becomes a burden for the company.

Charlie Shamieh: BaFin very much likes the holistic framework of risk culture for use test purposes, looking at controls and strategic management. However, the key difference between BaFin as a regulator and rating agencies is that it would act as a consultant if it thought that a risk control you were adopting had a weakness in it. I think most rating agencies would say that they were not consultants and would not tell you how to fix things that they did not like. BaFin does feel that is its responsibility because it is trying to protect policyholders. In that respect, I am not so frightened of the regulators. If you are getting them to play the role of consultant when they see a weakness, they will provide an education to help you lift it up to a better standard. That is nothing to be frightened of.

Frank Achtert: What is considered strategic enterprise risk management in this context? What are the key elements of appropriate risk management in addition to solid capital models? One side is what companies do internally, but another is what regulators will ask them to have in force under Solvency II.

Andrew Hitchcox: If you go back 20 years, the main information flow was loss ratios and perhaps insurance results. That information would flow up to the top of the organisation, but now you are adding on the third thing that is something to do with risk. At the risk of oversimplifying it, as long as that something is roughly right it is a good risk culture. Each time the underwriter speaks to the director of underwriting, he should tell him about the loss ratio, the technical result, the risk that he took on the way and how that might have been different. That, to me, is evidence of the risk culture. We have to admit that information did not flow regularly in the past, whereas now all of our underwriters and all of our departments expect on a quarterly basis to have some management discussion around your results, your plans and your risks. They might be defined in different ways, but they are on the quarterly agenda and they are built into people's job descriptions. People who are good at it will rise up through the organisation faster.

Oliver Peterken: I think you also have to be very clear about who the risk management is for. In a way, Solvency II is taking the whole ERM debate down a particular consumer protection route. Regulators seem to have confused the different goals. Banking is about systemic failure, whereas the consumer protection agenda of insurance regulators is quite a different one. There is a very good argument that you actually do not need it like you do in other sectors. With ERM, I have noticed that you have to ask who you are doing it for and that leads you as a business to consider the risk that you are prepared to accept in managing ERM. You then run into the big driver, which is the credit rating you need to trade in those products and those clients that you want. We have found that you have to be very explicit with ERM on your criteria that you are managing against. The regulators and rating agencies are in different places and I find that clients actually have a far more sophisticated understanding of ERM than perhaps the regulators do in terms of maintaining your franchise and so on. I think Solvency II is becoming more and more numerical, which is trying to bring us out on that wider ERM piece.

Andrzej Czernuszewicz: On a day-to-day management basis, you are thinking about other things as well.

Frank Achtert: Can ERM improve the competitiveness of insurers and reinsurers?

Steve Taylor-Gooby: I think it is becoming something that you cannot do without today, particularly if you have to deal with capital markets or with rating agencies. ERM is effectively mandatory.

Mark Nicholson: Anything that allows you to target your capital more effectively and explicitly would give you a competitive advantage.

Frank Achtert: In the end, do companies that do not have proper enterprise risk management fail? Everyone talked about a consolidation phase, but it never really happened intensively up until now.

Steve Taylor-Gooby: If they are unlucky, they will fail. In the meantime, though, they will find that they have a higher cost of capital.

Mark Nicholson: I think it would be going too far to suggest that major companies not embracing ERM will fail. If anything, as we have seen looking through the Solvency II tests, companies that have more effective risk management will hold lower capital.

Philippe Foulquier: Yes, and if you improve your risk management, you can free up some capital to invest elsewhere.

Oliver Peterken: I wish we could do that. In practice we tend to find that you convince the regulator and your management, but then the rating agencies say that they still want you to have this much.

Mark Nicholson: It is not always a case of the more capital you pile up, the better. It is perhaps in the nature of rating agencies to be cautious in this regard, but I think it is interesting when companies develop their ERM systems for us to focus on as well. At the moment, when we are looking at ERM systems, most of the component parts are already covered by different sectors of the review that we already carry out. I think it is about whether or not it is actually used. If it is used, it will be positive for the industry. If not, it will be a burden.

Steve Taylor-Gooby: One area where insurance companies are still quite a long way behind the banks is in the transparency with which they talk about their ERM systems and communicate with their various stakeholders. What we are seeing now is much more space in the Annual Report devoted to how good a company's ERM is. It may be more about figures on potential loss distributions etc, but I think it is a trend that will continue because companies are seeing an advantage in doing that.

Oliver Peterken: I think we are seeing people release more base data and that greater disclosure is very interesting.

Andrew Hitchcox: Financial analysts can ask you tough questions, though, when they just run their eyes down the columns.

Frank Achtert: Once companies start to disclose more about their risk management processes, I suppose that third parties and stakeholders will include it more in their long-term decisions. I hope that this will reduce the short term view.

The last aspect that I would like to touch on is the relationship between the primary insurers and the reinsurers. Do you think that their relationship will be changed by Solvency II? We are currently seeing an increasing desire from our clients to better understand the risk selection process and therefore the pricing of reinsurance. In the end, we end up with greater transparency in risk and volatility transfers and therefore pricing. The first potential outcome is that there will be risk adjusted product pricing, which some companies already do. However, a lot of undertakings are not there yet. Once the price of reinsurance is more transparent, other techniques such as capital market techniques will be there as a supplement to traditional reinsurance. With transparency, capital market techniques will be used more commonly as an instrument in addition to traditional reinsurance. A remaining obstacle is that investors are still not very familiar with insurance-linked securities, but that will change once transparency has been enhanced.

On the other hand, in terms of the relationship on the product side we might see non proportional reinsurance becoming more common. Especially in Continental Europe, proportional reinsurance is mainly bought to create capacity and improve the Solvency I ratio; but Solvency II is a different story.

Andrew Hitchcox: There are circumstances in which proportional reinsurance offers better protection.

Charlie Shamieh: I think that the best starting point for this is actually Steve's ERM survey [Tillinghast's most recent survey of risk and capital management practices among insurers]. The average revenue of the firms surveyed was $6bn, so you can safely say that we are talking about the larger players. Roughly 15% thought that Solvency II would increase capital requirements, but 15% thought that Solvency II would decrease capital requirements. They offset each other, and then the biggest group - about 36% - reckoned that it would have no impact or a very slight impact. The last group had no idea, and that one third was equivalent to the number that had no economic capital model. My conclusion is that the larger players that have been running economic capital models will not change their view of their reinsurance requirements dramatically. Many of them have in place, not necessarily extremely sophisticated, but certainly adequate means of assessing their reinsurance needs.

For smaller players, I do think that Solvency II will invariably put a spotlight on a need to better protect all the risks, especially when you are adding on a new element of risk - ALM [asset liability management] risk. I genuinely believe that it has been done fairly. The French seem to have this penchant for saying that we should not discourage equity, but smaller players will require an uplift of capital requirements and hence reinsurance purchasing. I think the top line impact will be materially positive, and I think profitability will be better. Smaller players can control their portfolios better and use reinsurance more as a business plan enabler because they have advantages over the larger players in terms of distribution costs. The key disadvantage is simply the lack of diversification, which they can buy via reinsurance. That may still allow them to be far more competitive than their peers. If you think that Solvency II will encourage better behaviour, that will naturally flow through to reinsurance portfolios. I do think it is too simplistic to say that proportional will suffer more. You are right that the formula now in Solvency I is anti non-proportional, but I do not think that too many people buying proportional reinsurance are just buying it to meet Solvency I requirements.

Frank Achtert: I would not say that it is solely for Solvency I, but it is one additional consideration. The Solvency II formula will most likely give credit to a mixture of proportional and non proportional reinsurance, but getting the right balance will be a lot more challenging. It will also give the companies more leeway to use reinsurance. Thus, even smaller companies now think of reinsurance more as an instrument of capital management rather than considering it as pure risk transfer mechanism.

Charlie Shamieh: I chair the German Insurance Association's reinsurance subcommittee on Solvency II. We will publish the paper early in the new year, but we are very much advocating that Solvency II will impose no restrictions on the level or form of reinsurance contracts. The lifting of this arbitrary 50% ceiling on P&C and 15% on life and health will also be beneficial. In return for that, we are also proposing more requirements. If you are using reinsurance, for example, it must be firmly integrated as a part of your risk and capital management, and there must be a risk management statement, as there is in Australia. The credit risk will also have to be assessed on an economic basis. It is not just capital relief; it is net capital relief after you have taken into account the counterparty risk with the reinsurer. As a package, if this is successful in driving how Solvency II recognises reinsurance and all forms of risk mitigation - such as on the life side, interest rate risk and ALM risk - then I think this will be positive for the industry.

Oliver Peterken: I think it will force people to manage their counterparty risk in a far more sophisticated way. When you look at a lot of the economic risk capital models, their great weakness is credit risk. That is often the missing link; economically you can establish the benefits or lack of benefits associated with risk transfer, but the counter party risk is just missing. This is reflected in pricing, where there is no real pricing benefit from having a different credit rating at the moment, which is utterly inefficient but seems to have been a permanent feature of the market.

I think that the capital market influence is usually overstated. There are some real limits because when you look at the capital market approach to ERM, they do not have such a difficult task as us. They have much better data on credit and market risk, and it is real time data. We are trying to value some very unknown possible outcomes and very complex over-the-counter contracts. On top of the whole modelling challenge, once you get away from motor and consumer products, our products are characterised by asymmetrical information flows and some real inefficiencies in the market to do with distribution and local effects. It is very hard to see us ever having an arbitrage free market. It will always be a market characterised by the ability to build a short-term advantage just through superior knowledge or distribution, which other people will find very hard to attack. I think there will be a limit to how far we move towards the way capital markets trade risk. We saw the problems that capital markets had with cat bonds - there are always people long and short of the risk. When you get into cat risk, no one in the world is short of that risk, everyone is long of it. So how do you ever get a trade?

Frank Achtert: It is a question of price and understanding and managing basis risk. If we become more sophisticated and have competitive pricing, more and more investors are willing to take the risk. Thus, I believe that securitisation will become more important. I am convinced there will certainly be increasing demand from the investor side, so we have to make sure that the tools utilised are accurate and the outputs understandable to communicate adequately with the investor community. Counterparty risk is not managed well in many cases at the moment, so do you see any movement in the relationship between primary insurer and reinsurer in terms of getting rid of concentrated risk and broadening the panel or focusing more on higher rated reinsurers?

Charlie Shamieh: We did some work on credit risk. You may recall that the FSA introduced an evidentiary provision three years ago where they said if you are going to use a reinsurer for more than 20% of your programme, you are going to have to justify why. The implication was that having five reinsurers was much better than having one. In our study, one of my people from the banking industry looked at credit risk in a completely consistent way under Basel II. We developed an approach for how you reflect the regulatory rating, not the counterparty rating. Remember that under Solvency II we are going to have a measure for capital adequacy that is consistent for everyone and is published, so market discipline will apply to that. We looked at the extent to which the regulatory rating was far more important than the number of reinsurers, and of course we found that from a security perspective the instrument is far more sensitive to the regulatory regime - in other words, the financial strength of the reinsurer - than to the number. Having 10 triple B reinsurers will not really improve the capital. I have confidence that under Solvency II, the financial strength of a reinsurer will finally get recognised.

Andrew Hitchcox: The capital markets are also very used to packaging and segmenting the risk, so the idea of having one rating for one reinsurance contract could disappear.

Oliver Peterken: As a reinsurer, I think that our clients have been far more sophisticated at managing their counterparty risk than regulators have assumed. They are managing it in terms of an overall exposure and also collateralisation. In the past two or three years we are now seeing collateralisation almost as a part of the contract negotiation because they are thinking ahead. You find that credit risk is best managed at the client level, so regulators have to distinguish much more clearly between the risk of systemic failure in banking and insurance. In reinsurance, we have no evidence of systemic as a problem. When you look at all the major losses we have had - the World Trade Center or Hurricane Katrina - the industry does appear at the moment to be incredibly resilient in terms of dealing with that. You then have to ask why you need Solvency II, because it is being handled perfectly well at the moment in client reinsurer interaction. I think you can get some perverse effects. You can get some very unintended consequences once regulators and solvency rules come in at a level where they are not really needed. I do not think credit risk is an area that is an issue at a regulatory level.

Frank Achtert: Is it an issue from an economic capital standpoint?

Oliver Peterken: I think people are dealing with it. We announced last week that we have put a credit wrap around our own reinsurance recoverables, which is a surprisingly difficult undertaking. Having achieved it, all of a sudden you are able to talk from a different position. How you feed that through into an economic capital model will be an interesting challenge for us. The reason for putting in place protection like that is because you feel it is an unquantifiable risk. I think we are quite a long way from having the sort of sophisticated answers you would like us to have.

Frank Achtert: Thank you very much for this lively discussion.