Mise à jour du BSIF. Sylvain St-Georges. ?? = Inaudible/Indecipherable ph = phonetic U-M = Unidentified Male U- F = Unidentified Female

Transcription

1 November 2009 General MEETING Ottawa (PD-17) 1 PD-17: TR-17 : OSFI Update Mise à jour du BSIF MODERATOR / MODÉRATEUR: SPEAKERS / CONFÉRENCIERS: Bernard Dupont Bernard Dupont Sylvain St-Georges?? = Inaudible/Indecipherable ph = phonetic U-M = Unidentified Male U- F = Unidentified Female Moderator Bernard Dupont: This is session PD 17 called OSFI Update. Sylvain and I will be speaking in English during that presentation but if you have any questions at the end as good civil servants we re ready to answer them in French or in English. I ll introduce myself, I m Bernard Dupont. I m the Director of the Insurance Capital Division at OSFI. That s an organization I ve been with for over 20 years now. I m not a real actuary; I m a Fellow of the Casualty Actuarial Society and a Fellow of the Canadian Institute of Actuaries as well. In my role at OSFI I m responsible for the Minimum Continuing Capital and Surplus Requirements (MCCSR) and for the Minimum Capital Test (MCT), so basically updating and maintaining the MCCSR and that s what we re going to be talking about today. Talking about real actuaries, we re lucky to have as a panellist Sylvain St-Georges from the AMF Quebec. Even though the session is called OSFI update. Unfortunately for me, Sylvain is not with OSFI yet and just like me he has been with the AMF or the IGIF for over 20 years as well. He is responsible for the life capital requirement at the AMF and he s also serving on many CIA committees. On this the agenda for today I ll be talking about the MCCSR overall direction, where are we going with the MCCSR and what are the stakeholders asking us for and then we ll talk about more in detail the various work streams of the MCCSR like the IFRS modification. Sylvain will talk after that about the market and credit risk QIS and then I ll come back at the end to talk about seg fund requirements and mortality improvements. We re getting pressure these days from all sorts of stakeholders to modify the MCCSR. Every company is looking at the MCCSR and there is always something in there that is an irritant for them and they don t like it and they phone us on a regular basis. So if you multiply this by 300 companies you can imagine we re fairly busy looking at requests from the industry and also coming from CLHIA or the IBC. Also, there is lots of movement internationally. There is Solvency II being developed in Europe, the U.S. have started what they call a Solvency Modernization initiative and also IFRS will modify the balance sheet of our companies and we have to adapt the capital requirement because of those changes. Also, we need to sometime redesign the MCCSR because certain portions are getting obsolete and we have new and more risk based techniques being developed. Many years ago when we were developing factors for the capital requirement we were using sophisticated actuarial techniques and we were developing our factors this way. Today it doesn t work this way, the stakeholders don t accept this and OSFI doesn t accept that either. When we develop capital requirement nowadays we have to make sure we know what confidence level they are at? what technique is being used? is that technique being consistent with what is on the banking side? is it consistent with what is being done internationally? So there is lots of pressure on us to look at all sorts of level items. PROCEEDINGS OF THE CANADIAN INSTITUTE OF ACTUARIES Vol.41, No.1, November 2009

2 2 Novembre 2009 ASSEMBLÉE GÉNÉRALE OTTAWA (TR-17) Also, we re looking at allowing the use of internal models for determining a regulatory capital. I won t go into detail about this because Jean-Guy Lapointe has a session tomorrow, I think it s called Internal Models or Use of Internal Models. So I encourage you to attend that session if you want to know a little bit more about that. But our session was supposed to talk also about the MCCSR Advisory Committee and to provide an update. That s fairly easy. The MCCSR Advisory Committee. When we started our work we made the assumption that the models being used for seg funds are being well used and they are pretty solid and therefore we said now we want to develop other market risks and credit risks. The last crisis showed us that maybe there were deficiencies with the use of internal models for seg funds so we have taken a step back, put on hold credit risk and other market risk and now we are sort of looking at seg funds again. So the MAC is a little bit on hold for the moment. So I mentioned there is lots of work we are having here and there but the first work stream if we can call it this way is the implementation of IFRS. For January 1, 2011 IFRS will come in place, IFRS 1 and Phase 1 of IFRS 4 and that will modify of course the balance sheet. The general principle we ve been working with to modify the MCCSR was to reduce the MCCSR modification to a minimum until Phase II is in place. We don t want to modify materially the MCCSR and then realize in two years the liabilities are being calculated totally differently and therefore we have to modify again our capital requirements. So we re trying to limit the changes either to limit our work and to limit the work for the industry to update their system. So when companies are given the option to retain the current accounting methodology we are going to retain the current MCCSR treatment. A good example of that is for insurance contracts. OSFI knows that under IFRS Phase I companies will be allowed to retain the CALM methodology to value their liabilities so we re saying if you re allowed to use it then we want to keep it for capital requirement. If companies use something different then we will reverse the difference in between the IFRS valuation and the CALM valuation in the MCCSR. An option is being considered to allow IFRS valuation for non-insurance contracts to use maybe a different methodology than the CALM approach. We re still undecided about that but we re having discussions with the industry. I think it s going to be a surprise for nobody we have learned from the recent financial crisis that securitization was not always transferring the risk the way we believe it is so the assets that are not derecognized and which are not exempted from consolidation should be included in the capital available and capital required calculation in the future. So basically, we won t exclude any new item coming into the balance sheet. Examples of that are the consolidated mutual funds. If you have mutual funds that will be now appearing on your balance sheet and we will apply a capital charge to it. The only exception to this of course will be for segregated funds. If you have a consolidation and you have an asset that will appear on your balance sheet just because it s a seg fund asset and it s being consolidated because these assets are already being taken into account in the segregated fund capital requirement then of course we won t apply a second capital charge so these will be excluded from that capital charge. For real estate own-use property, that s maybe one of the only places where effect on the balance sheet will be reversed. We re saying for companies that are using a fair value option any other unrealized gains, either at implementation of IFRS or in the future, will be disallowed in the capital. That s, I guess, the only thing I had on that. The difference though is if it s real estate investment property and if you have been using fair value to value that asset then that will be recognized in regulatory capital. The net impact from IFRS should be reflected in retained earnings except for own-use property real estate as we mentioned earlier. Of course, certain companies approach us and say Well, the impact might be material to us mainly because of the consolidation now of pension plan deficits and OSFI realized that. We re a nice regulator so we said Okay, no problem. If those amounts are material we re ready to phase in the impact over a certain period of time. So it could be quarterly; phase in could be one year. If it s really material it could be two years, three years. We ll see, it s going to depend on the amount we re collecting from the industry. The second work stream that we are working on these days is the Standardized Approach and for that we have formed a working group called the Standardized Approach Working Group. Participants on this group are Vol.41, N o.1, Novembre 2009 Délibérations de l Institut canadien des actuaires

3 November 2009 General MEETING Ottawa (PD-17) OSFI, the AMF and Assuris. The goal of that group is to look at modifying the MCCSR for the elements that will be impacted by IFRS 4, Phase II. So in 2013, 2014, we don t know when, but the valuation of actuarial liabilities are likely to change. The CALM methodology is likely to disappear, we don t know yet, for the financial statement. So we need to probably modify the capital requirement because of that. So we already started looking at that element. I won t get too much into details here because Sylvain will talk about it pretty soon. So I ll skip and just say briefly that we issued on that work stream a framework paper in October 2008 and we issued as well a market risk paper in October Both papers were issued more than a year ago. We just posted a QIS a month ago and we re expecting the modification following from those work streams to be in place for The important thing for us is to have in place new capital requirement before IFRS Phase II comes in place. It s either at the same time or before but we want to make sure it doesn t come and then there are certain elements of the PfADs that are being lost in the future. Work on other MCCSR elements after Phase I will be following if necessary. I m talking here about any other like policyholder behaviours, mortality. All those other elements will be looked at after these ones. It is always preferable to do all your capital changes at the same time but in this case here, in the case of the MCCSR it s so big that doing all of them at the same time would be impossible in terms of resources either at OSFI or within the industry. So we re focusing now on market and credit risk. So Sylvain will explain a little bit more what we ve been doing on that and I ll be back to talk about seg funds. Speaker Sylvain St-Georges: So good afternoon everybody. Just a brief word in French for my friends here. Bon après-midi tout le monde. Comme Bernard l a dit, je peux parler en anglais mais ceux qui étaient à Ste-Hyacinthe il y a quelques semaines c est la même présentation que j ai faite. Je vais la faire en anglais cette fois-ci. So I m back in English. I just said that this is the same presentation I did to Quebec actuaries a few weeks ago in French so I ll try to do it in English. So I ll talk about the QIS that OSFI posted on their website and we sent to appointed actuaries in Quebec. There are five parts to my presentation: a brief introduction; talk about the market risk QIS; the credit risk part; a word about participating products that are on the same requirement for both market and credit risk; and some closing comments. I wanted to ask: are there a lot of you that have looked at the QIS yet? Not that much? So I ll try not to be too long but still to give some information that will be interesting to everyone. So the main purpose of the QIS is to test the practicality of the methods and to estimate the potential impact. The approach is based on shocks but as you will see for the interest rate risk method, it s a shock on interest rate risk as compared to currently when it s a shock, it s a factor applied to actuarial abilities. So there are some differences. We re going to test if it s doable for companies to do the test. And as has just been said by Bernard there is going to be some calibration required after we receive the results when we know the impact of those proposals. So for the market risk, these are all the components that will be tested with the market risk part. There is interest rate risk, equity risk, real estate risk, pass through products risk, currency risk, liability markets options risk and asset market options risk. So for the interest rate risk as I said it s the part that is the most changed, the most different current method. We ll talk about the cash flows using that method; the risk free interest rates won t be used. What is the method? What does the method consist of? And I will talk about what additional scenarios will we ask with the QIS? So the cash flows, the method, just to have a brief introduction, it is present value of the cash flows of assets and liabilities. It s a kind of test of what is the impact on the balance sheet of a change in interest rate curve. So while we ask companies to do a projection of their cash flows, it s the CALM cash flows. So it s the expected plus margin cash flows that are projected on the asset side and liability side. PROCEEDINGS OF THE CANADIAN INSTITUTE OF ACTUARIES Vol.41, No.1, November 2009

4 Novembre 2009 ASSEMBLÉE GÉNÉRALE OTTAWA (TR-17) However, contrary to the CALM method there is no reinvestment so it s a simple projection to the end of the asset and liabilities term. For the assets, the fixed cash flows are projected. So for example because it s not that clear in the paper that if you have real estate with a contract that guarantees you will receive a payment on the location used by those that occupy the property. If there is a contract those fixed cash flows can be used in the projection. However for the non-fixed cash flows there are no projections; they stay at time zero. So the assumption on this is that even if the interest rate changes there is no impact on those other assets like equities or the real estate except for the fixed cash flows. You also have to project every asset and liability. So CALM cash flows but also the assets and liabilities that are not included in the CALM valuation. So every asset backing surplus and all the other liabilities that you have that are not in the actuarial liabilities. So once you have projected those cash flows you discount them with the risk free interest rates for the base scenario. Those interest rates are the spot rates for Government of Canada bonds and other rates that are equivalent in other parts of the world. We have provided with the QIS the rates for Canada and U.S.A. but you have to determine similar rates for Asia, or other parts of Europe. Those rates are provided for the first 30 years and I ll explain soon what to do with the cash flows that have longer terms than 30 years. So as I said the method is to assess the economic impact of a sudden change in interest rates at time zero. The first step is to discount the cash flows after 30 years to the year 30 by using a flat 6 percent rate. That is the rate that is provided. That is the rate for the Canadian assets. So you discount those cash flows to year 30 then you add them to your standard year 30 cash flows. Then you discount all the cash flows from year 0 to 30 to the year zero by using the prescribed interest rates that I just mentioned. That s for the base scenario without the risk free rates. We also told you how to do shocks to those rates. That s my next slide and so we have base scenario plus 10 scenarios and for each of those scenarios you calculate the net present value which is the difference between the asset present value and the liability present value and the solvency buffer for the interest rate risk is the difference between the net present value of the base scenario and the minimum of the net present value of the test scenarios. So it s kind of what could be the impact on the equity and the capital? So what are the interest rate shocks? They are the tests that are estimated to be at 99.5 percentile of the potential upward or downward change in 30 day T-bill rate over one year and also on the potential upward or downward change in 30 year spot rate over one year. So you shock the short term, you shock the long term and use a linear interpolation of shocks between these two rates. Then you add these shocks to the risk free interest rate base scenario to have the stressed scenarios. Those shocks are based on a simplified Cox-Ingersoll-Ross model which was fitted to historical data, so it s a simple formula to use the current interest rate to obtain the shocks. Still on the interest rate shock method there are a few things about the Universal Life products. When doing the projection the guaranteed credited rate should be considered in the calculation. Cash flows should be consistent with the scenario interest rates, for example the account values, but there should be no adjustments due to anticipated changes in lapse rates and expense charges because they would be considered an insurance risk later in another few years. We also ask for another 10 additional scenarios. Those are only used to assist us in calibrating the metrics; they are not used in the calculation of the whole solvency buffer. Those additional scenarios are based on shocks which are based on a 95 th percent confidence level, or some different confidence level. So that was the part that was very different than the current requirements. The other parts are more similar but there are some differences. The first is equity risk. Equity risk is only for common shares because we decided that it was more appropriate to have a buffer on preferred shares that is included in the credit risk component. So don t be surprised not to see a preferred share in the market risk because it is in the credit risk. It s an immediate shock at time zero so that s the same thing that we have currently except that we estimate that for an equity index a= 20 percent decline is what could be similar to one year 99 percent CTE drop, but since the managed equity portfolios are more risky, have more volatility than an indexed stock we ask for factors over 30 percent, the same as a 30 percent decline in the equity. Vol.41, N o.1, Novembre 2009 Délibérations de l Institut canadien des actuaires

5 November 2009 General MEETING Ottawa (PD-17) For the real estate risk it s also an immediate shock at times zero; this time it s a 20 percent decline. In our paper on QIS there is not much explanation of where the 20 percent comes from but in the paper I ll talk about, the market risk paper that was published earlier, we explained that we talked about a 20 percent decline. With the little information available it seemed to be an appropriate factor. For the pass through product risk, for all those products like the current requirement so there is no credit or market risk applied to those assets. You use as currently the coefficient factor approach or discount coefficient factor calculated between the asset variation, the liability variation as it is in the current requirements except that currently the threshold is 85 percent and in the QIS its 70 percent, and that reflects the 30 percent equity risk compared to the 15 in the current factor. But we clarify that this requirement based on the correlation factor is optional so you can use the standard credit and market risk to those assets. There is a new risk but that s not true; it is included in the MCCSR recently a currency risk. It also is an immediate shock. The shocks are applied to the net mismatch of cash flows in each currency and it s a 20 percent rise or decline in currency s value against the Canadian dollar. That was explained in the paper on market risk; it s related to the 20 percent. 20 percent seems to be the appropriate factor; there is not much calculation behind that. For the liability market options risk we talked about, there were two types of those risks. There is a risk related to the minimum interest rate guarantees reflected in annuities and those guarantees are reflected in the interest rate risk solvency buffer that I talked about earlier. So the only other risk in this category is the risk related to seg fund guarantees and is based on deterministic shocks. That s the difference compared to the current requirement which could be based on an internal model or a kind of program that calculates factors on a lot of dimension matrices. We wanted to keep it simple for this test and you have to remember that in the next phase I would say when we could allow companies to use internal models, even for those companies we expect that the internal models will be used for the target ratio and the simple approach would be used for the minimum ratio. So this calculation will be used by all companies. For this risk too we will ask for an additional seg fund scenario for supplementary information. It is not included in the calculation but it will be used for us when we do our review of the results. So the risk related to segregated fund guarantees is calculated by calculating the cost for two scenarios and we are taking the maximum of those two scenarios. The first one is the immediate 30 percent fall in the equity market and the immediate 20 percent fall in the bond market with no recovery thereafter. In the second scenario, it is assumed that the equity index is increasing 100 percent and the tax is 200 percent. You start with the index of 100, you increase it to 200 up to the 44 th month and then at the 50 th month you drop the index to 75 so it s a 100 percent increase. At the same time that the equity index dropped the fixed income market dropped a lot by 20 percent, and at that point there is no recovery either for the equity index or the fixed income market. The last sub-risk is the asset market option risk. This risk is covered products described in section 3.7 of the MCCSR guideline. Those are the assets replicated synthetically and the derivative transactions and we use the same requirements as we have currently so you just take what you have in the current calculation and use them in the QIS. So that was the market risk. Now then to the credit risk. So those are the sub-risks that are covered by the credit risks. We have a component for short term investments, one for public bonds and private bonds, asset backed securities, mortgages, preferred shares and other items I ll explain later. For the short term investments in the QIS we use the current factors. We made some analysis and were consistent with factors that we established for the public bonds. So let s keep in mind that they are the current requirement and then I ll talk about how we established the public bond factors. Currently the factors are only dependent on the ratings of the assets. In the QIS we have factors that depend on rating but also on the remaining term to maturity and we established those factors by using the Basel Foundation IRB approach. We found that it was best comparable for Basel of the three approaches possible and found that the simple approach, the standard approach for Basel was not a really good comparison mainly because there no banks, or Desjardins in Québec, used the factor approach for Basel; they use the internal model. PROCEEDINGS OF THE CANADIAN INSTITUTE OF ACTUARIES Vol.41, No.1, November 2009

6 Novembre 2009 ASSEMBLÉE GÉNÉRALE OTTAWA (TR-17) So we made some modifications to the calculations: we used the 99.5 percent confidence level instead of the 99.9 percent that was used for Basel and we modified the maturity adjustment to take into account the longer duration of bonds held by Canadian life companies because for Basel the maturity adjustment is limited to five years and it was not appropriate to extrapolate those adjustments. So this is what are the new factors in the QIS. I put beside the QIS factors the standard Basel factors to show you that the factors we are proposing are much lower than the standard Basel factors, but you will see by yourself that the new factors could be much higher than the current factors because the current factors are much in line with the zero-to-one or one-to-two-year term. We did some tests with data from experience real loss and bond data and we tried to develop factors directly from the experience and they are very similar to those factors that were developed based on the foundation approach of Basel. So we are satisfied that they are appropriate factors to reflect 99 percent risk. For private bonds, I d just like to clarify that this category also includes leases and other loans. The factors should be based on the inferred rating from other issues by the same issuer. So I think that s the same as currently in the MCCSR and if not possible, if the issuer has no outside public ratings for other issues then the factors that should be used are an average of the public bond BBB and BB factors for the equivalent term to maturity. For the asset backed securities we use the same factors as for the public bonds or private bonds; there is no similar issue where they are private. They are calculated as a separate category however; that s the same situation as currently but the calculation is the same as public bonds. Mortgage, for the commercial mortgage the factor proposed is a 6 percent factor. Currently we have 4 percent; Basel uses an 8 percent. And the 6 percent factor is based on evidence from the 1990 s real estate downturn. We looked at the experience of that period and we feel that 6 percent accurately reflects the risk. For the single family residential mortgages we used the current factors. We received some data from the industry and it seems to show that the current factors are in line with the experience of the industry. As I said, we apply the credit risk factors to preferred shares. We keep the same categorization as we use currently. The factors however are consistent with public bond factors in this QIS. So for example, if you compare it to public bond, where the current bond factors are 1 percent, for those bonds is the 3 percent factor in the QIS. So there is a 3 percent factor for the preferred share. The last category is 30 percent; this is consistent with the common shares and between there is a fine graduation to reflect the risk. The other items are miscellaneous items of balance sheet exposures and securities lent; for all those we keep the current requirements. So I said that there is something particular to participating products. All the calculations in the QIS should be as if the policies or the assets are backing non-par products. Then we use this methodology to apply reduction for participation. So the first step is to calculate the gross reduction which is based on the maximum reduction in present value of future dividends and since the gross reduction is expected to be applied to all risks, the reduction in the dividends is a reduction that could be applied to all risks and is a reduction in the worst case. Then for this QIS use this gross reduction and reduce it by the reduction for par products included in the other components like mortality, morbidity and the other. So we re left with a net reduction which is applied to the market and credit risk components; we thought about market and used it for credit because we developed a market risk component first. You recall that the worksheet should be returned by December 11 th and I replied that all the submissions will be shared with Assuris and OSFI and AMF so they were either sent to us or I ll receive those from OSFI. At the end of both QIS risk and market risk there is a page called interrogatories and preparatory comments. There are some questions and we will require you to respond to those questions; we ll have lots to understand the results. There are also other questions where we ask comments and we encourage you to give your comments on those questions and any comments you have on those QIS. Those comments will help us to have better requirements and it s better to have the comments now than later when everything will be decided. Vol.41, N o.1, Novembre 2009 Délibérations de l Institut canadien des actuaires

7 November 2009 General MEETING Ottawa (PD-17) 7 Moderator Dupont: Thank you very much Sylvain. As you can see there was lots of work put into the development of that QIS so I guess I should thank Sylvain for helping us on that and there is Daniel Mayost sitting quietly in the back and Jean-Guy Lapointe as well with us. They ve been putting lots of time into this. Thank you very much. The next work stream I want to talk about is segregated funds. So for companies currently using internal models to derive their statutory capital requirement we are using at OSFI the CIA methodology developed in the year In the past we ve been working a lot with the CIA when there are more technical issues where we want to develop capital requirements, such as for the mortality formula for segregated funds. So we ve been working with the CIA and it has been working very well in the past. For seg funds though the market has continued to evolve and companies have been introducing new and very complex products over the years. So because of that OSFI started to have concerns a couple of years ago and we thought it may be the current modeling approach developed by the CIA wasn t always capturing all the risk embedded into those products. So we sent a letter to the CIA in 2006 outlining concerns we had about the methodology, I think it was six or eight of them, and we asked the CIA to review that. The events of the past 18 months have sort of shown to us that yes there were probably deficiencies with the methodologies and that maybe there is something that needs to be done because we believe that the way the capital requirements reacted in the last crisis that probably some of those products are currently underpriced in the market. One of the concerns we re having also about that product is it doesn t necessarily react like a traditional life insurance product. As you know for life insurance products if you sell more and if you are well diversified by population or geography, you can sometimes reduce a little bit your risk. For seg funds it s a little bit hard to reduce it unless you hedge it but it s not by selling more of that product that you will diversify yourself. The stock market unfortunately showed us at the last crisis that when it goes down for one sector it goes down for all the sectors in the stock market. So that s one thing we re sort of worried about now with those products. So what OSFI did is we undertook to develop a comprehensive review of the seg fund guarantee requirement. What we re trying to do is to come up with a list of various options: looking at keeping the current actuarial methodology, updating it and looking also at using market consistent methodologies and looking at various options or mix of both methodologies to see if we can arrive with one methodology that would provide a better valuation. To provide some input into that study we did a comparison of the models and assumptions currently being used by the different companies to determine the capital requirements and we also did a comparison of the amount determined by different companies for identical risk. On that one, we called it at OSFI the benchmarking study. So what we did is we sent to companies using internal models for seg funds, identical products with certain required assumptions. We sent it to them and said please calculate that using your model and we then compared the results of all those companies. We also hired an external consultant. We hired a Canadian firm but having lots of international ramifications and we re hoping that the consultant will provide us an independent view as to what could be used to value guarantees for seg funds. We re not sure yet but we believe that once we will have done our review the capital requirement for seg funds are likely to go up. We re basing that view on the fact that if we look at the cost of hedging the product today usually the cost of hedging it is much higher in many cases than the price being charged by the company. So we have a feeling that probably the products might be underpriced in a couple of places. The preliminary results of our review suggest that first yes there is a wide range of practice with respect to the setting of assumptions. So when you look at the assumption being used by the various companies, either discount rate, lapse rates, equity returns, there is a large range of practice in there. Also, when we did the benchmarking study we realized that companies could be holding materially different capital amounts for the same products. PROCEEDINGS OF THE CANADIAN INSTITUTE OF ACTUARIES Vol.41, No.1, November 2009

8 Novembre 2009 ASSEMBLÉE GÉNÉRALE OTTAWA (TR-17) We have to be careful and we realize that at OSFI that doesn t mean that when you look at the range of practice the company who has the lowest assumption is necessarily underpriced. Maybe the other ones are too conservative, but that s something we re keeping in mind. On the other hand, we don t think that there was any company that was outside of the range of practice of the CIA when we looked at the results. We also believe it would be good to increase the disclosure of the assumption to the public; that might help bring a little bit of discipline and a little bit of convergence to the entity practices. So once we have finalized our views on the methodology that we re going to use for valuing seg fund product, we will do an industry consultation. It s likely to be, I would like to say, early 2010 but every time after saying that I realize it s a little bit later. So it s going to be during 2010 sometime where we re going to come up with our views as to what might be used to value those products. The implementation is expected then to occur as soon as possible after that. It s always a question then depending how different it is from the current methodology. We have to give time to either OSFI or the companies to change their system if there is a need to change that, but I don t think we want to wait too much. If the new methodology is proving to us that there are deficiencies or that certain products are underpriced then we think it s very urgent that we move rapidly on that. The CIA has established a new task force to look at seg fund valuations. We re going to use that as well as an input into our study. So we re looking at what we think might be good then we re going to sit down with that task force from the CIA and look at what might be done to value those products. OSFI is participating as well in the work of that task force. Now, in other work streams the CIA is expecting to have a new standard on mortality improvement. It s expecting to allow mortality improvement to be taken into account in the valuation of the actuarial liabilities. OSFI has concerns with that. We agree that disregarding totally mortality improvement might not be the right approach which is the current approach now, but allowing and recognizing mortality improvement over 25 years might be excessive. Maybe not for valuing the liabilities, we re leaving it to the CIA to decide, but for capital purposes though we think that at the 99.5 percentile we think that the assumption is probably a little bit weak. So what we ll try to do with the help of the CIA is to have next year in 2010 a QIS and look at what methodology could be used for capital purposes recognizing some mortality improvement but not necessarily to the same extent that the CIA would like to recognize it in the valuation of the liabilities. So for 2010, since our methodology is very unlikely to be in place for whenever the standards will come in place, OSFI will reverse the impact of mortality improvement in the MCCSR until we have developed and implemented the new capital requirements. That capital requirement I would assume will be in place in When you re looking at all the work streams we re always talking about 2012, 20 something. It s not that we like to take our time, it is that by the time we sit down with the industry and discuss that with the different stakeholders and then we finally come to a decision, then we need to allow at least eight months internally within OSFI to develop our system. For IFRS we are now developing the capital requirement that will be in place January 1 st, 2011, so we re expecting our guideline to be ready in January. So we always need at least a year in advance to look at the various capital requirements otherwise we re going to miss the deadline in terms of having our system ready. Then the other issues that we re also looking at, we re now assessing the appropriateness of the current capital metric for holding companies including the associated disclosure requirement and leverage ratio. And finally, I kept it for the end but it s the closest in time to us, it s the MCCSR 2010 modification. Only minor changes will be made to the MCCSR for January 1st, We will clarify certain hedging rules in the MCCSR and we will also include definitions of unregistered reinsurance and stuff like that that used to be in our reinsurance guidelines; we re now going to implement that into the MCCSR. So it s going to be fairly minor changes; we re expecting to circulate them for informal consultation with the CLHIA pretty soon and then probably post them later in December. So in conclusion, you can say we have a lot of work stream on the go and that s without really talking about all the work being done on the internal model side. It does require lots of work. It does require lots of resources Vol.41, N o.1, Novembre 2009 Délibérations de l Institut canadien des actuaires

9 November 2009 General MEETING Ottawa (PD-17) 9 but you know I think with good cooperation with the industry and with the CIA, we ll arrive at a capital requirement that makes sense. So on that I ll open the floor to any questions that you might have. U-M: (off mic) Bernard, I wonder if you could confirm with respect to when you say reverse the impact of release in reserves from mortality improvement that you could confirm that what you don t mean is that it would be added to the required capital but more that you re looking at something on available capital perhaps on an aftertax basis? Moderator Dupont: Yeah exactly, when we re saying reversing the impact it s really the impact, it s not necessarily the amount, of the mortality improvement. I really said the impact. We re hoping that once we adjust the MCCSR the impact on the MCCSR will be neutral. U-M: From all the factors that I saw up there, capital requirements are going to increase dramatically. Does that mean OSFI is going to be changing their target ratios on their communication or how is that all going to fit together? Moderator Dupont: That s a very good question. That s something we ve been looking at internally. That s also why the mortality improvement modification can come in place at the same time. So we re hoping that whatever is being gained in one place and lost in other places will sort of counterbalance each other. We need to look at the impact of all this as well. The last thing we want we re not doing this, like it s a capital grant. That s not what we re trying to do here. We don t think the industry is overly under-capitalized or something like that so we re not going after more capital but we re going to have to look at it. Do we need to modify the targets? Do we need to reduce capital requirements somewhere else? Do we need to rebalance? I think it s going to be good for us. I m saying we re overall satisfied with the amount of capital that we have currently in the MCCSR but pretty often we re not sure if we have the right balance between capital requirement for mortality, for market risk, for the different risks within the MCCSR. So by doing that QIS and by doing it hopefully with a certain confidence level in mind we ll be able to say: Oh, if we re putting all this on an equal footing here is the kind of capital requirement, here is the kind of proportions we re coming up with. And maybe the current MCCSR doesn t have the right proportion in it. So by looking at that we re hoping to rebalance maybe the MCCSR. So it s only preliminary work; we re testing those methodologies. We re going to look at what the impact is going to be and then we re going to have to sit down and say: Okay, what do we do? But you know if we arrive and capital requirements are being doubled by that, you know there is no way we can implement that. We have to be realistic. U-M: To what extent are you looking at what s happening with Solvency II, the standard formulas that they have and to what extent do you think it s relevant? Do you think it s totally irrelevant or do you think we can learn from that? Moderator Dupont: I think your last comment is probably the right one: we can learn from that. We re looking at Solvency II. I mean we don t have to, but as I said at the beginning there is lots of pressure from everybody. Everybody is aware of what is going on in Europe in Solvency II so people are saying: Well, why are you doing that? Why aren t you looking at Solvency II? How come Solvency II is doing this? So it s really hard for us to disregard what is being done over there. The same thing it s very hard for us to disregard what is being done on the banking side or P&C side. So more and more everything is more international for capital requirements. PROCEEDINGS OF THE CANADIAN INSTITUTE OF ACTUARIES Vol.41, No.1, November 2009

10 10 Novembre 2009 ASSEMBLÉE GÉNÉRALE OTTAWA (TR-17) We re trying, especially at the IAIS, to have some convergence in the capital requirement. So I would say the official position of OSFI is that we don t have to look at it but we re looking at it because it s an interesting development going on these days. But we re also looking at what is being done in Australia and the U.S. is starting to review its capital requirements. So we re going to look at what we can do by looking at what they are doing and maybe even harmonize a few things with them. U-M: You seem to be looking at Basel II and what is being done in the banking sector as probably a better source of a capital model for the insurance industry than Solvency II. Maybe I got the wrong impression but I personally feel that insurance companies are not just like banks, there seems to be a feeling that Moderator Dupont: No, I fully agree with that and it is sometimes frustrating even for us within OSFI because you cannot modify Basel II really. I mean it s international, it has been agreed on. So when we develop something, if we try to harmonize within OSFI, Basel II is Basel II. There is nothing you can do. You cannot modify it to be a little bit more like the MCCSR even if we re doing a better job on certain risks. Overall we re not trying necessarily to force what is being done in Basel II but as an integrated regulator we have to look at for the same risk do we have different capital treatment? That is something that is very important for us. Is it possible for a financial institution to move risk from one institution to another and have different capital requirements and therefore arbitrage the rules? That s why it s very important for us to make sure that our rules are consistent between the different sectors. Speaker St-Georges: And I will add that we have an adjustment to the Basel as I said. It s not just take the factor for banking and extend it for insurance. We adjust for the particularities of the insurance companies. The difference between Basel II and Solvency II is I think we could say that Basel II is more directly comparable because banks and insurers having some product diversity in the same market compared to two different markets with the insurance in Europe and insurance in Canada. Moderator Dupont: I would add that also the difficulty with Solvency II is that it s moving a lot and it depends who you re talking to because if you re talking to the European regulators they have their own views on how it should be applied and the industry has its own views. Then you have the European Commission who has to submit it to the European parliament. So it s not the regulator over there that has the last say. So when we talk to people about how is it going to work then people that see it are saying: Well, here is what we ve been proposing and other people are saying: Here is what has been proposed and so you never know exactly where it is going. So it s a little bit hard to know exactly what Solvency II is really going to look like in a couple of years. U-M: You described the participating policies. Any idea or any guidance on whether that s going to put par policies in the same relevant capital position as they currently are in non-par or a strengthening or weakening relative to non-par? Speaker St-Georges: I think the biggest difference is now a company could have a reduction for par products in different risks and those are more than what the company could really transfer to par policyholders. That s what we try to get rid of. U-M: But no ideas yet whether these, it looks like a strengthening of the requirement of non-par, is par going to be the same ratio of strengthening or a greater strengthening or a lesser one relative to the non-par? Vol.41, N o.1, Novembre 2009 Délibérations de l Institut canadien des actuaires

11 November 2009 General MEETING Ottawa (PD-17) 11 Speaker St-Georges: It would be the same thing because just reduce the par requirement for the reduction related to the the reduction in the non-par development. It s possible for a company but then if we took the real reduction as currently at 50 percent on par products Moderator Dupont: Correct me if I m wrong Sylvain but I don t think we had any intent to lets say increase or decrease them compared to non-par products. U-M: (off mic) in the same relative position? Moderator Dupont: We ll know U-M: Time will tell. Moderator Dupont: We ll see but there was no specific intent saying that s an issue we have to address. U-M: One of the things we noticed in the last downturn was the cyclical nature of the equity market. What are you guys thinking about in the MCCSR formula to take into account the economic cycle? Moderator Dupont: Again, that s a good question; I mean we have been looking at it. There are certain people that say when your equity market goes down then you re already contra-cyclical because then your capital requirement will go down as well because then your 15 percent, the same factor, will be applied to a lower exposure. So certain people are saying this way will be sort of contra-cyclical from that point of view. But I don t think it s an easy question to answer. I mean we ve been looking at contra-cyclical policies or requirements but so far nobody has come with something that I would say was satisfactory to us. So we re still looking at ways to make our requirements contra-cyclical but I think it s going to be a little bit harder. Speaker St-Georges: And the difference between banks and insurers is the problem with banks is the cycle with the economy so when the economy goes bad, banks lend less and it s bad for the economy and it s going on and on. And then the converse is true that when banks lend a lot and the economy goes well but they lend a lot and then they have bad risks on their books. That s a problem with banks that we mainly don t have with insurers. Mr. Frank Grossman: This question is for Sylvain. The capital regime or concept in Basel has columns and in the market discipline part an important role is allocated or given to credit rating agencies in terms of the correct appreciation of the risk and classification and a consistent approach over time. Over the last few months, I think the consistency and perhaps accuracy has been in question somewhat. It s a broader question but do you have any thoughts about reliance on other agencies when you re trying to do your important work? Moderator Dupont: That s a question we looked at at a certain point at OSFI, reliance on rating agencies for our capital requirement and at a certain point we said: Well that doesn t really make sense and we should probably have less reliance on rating agencies for our capital requirement. Then as soon as we started looking at options and how we could forget about the rating agencies and we re saying: Okay, we re going to remove that rating of various companies or products then people are saying: Oh well! Now your requirements are not risk based because now you don t take into account the rating of the products. So either you do it yourself at OSFI and have your own valuation which would be almost impossible for us or you are sort of stuck to rely on rating agencies. So we had a project at a certain point how we could rely a little bit less on rating agencies and we re trying whenever we have a chance to rely less on rating agencies but to be PROCEEDINGS OF THE CANADIAN INSTITUTE OF ACTUARIES Vol.41, No.1, November 2009

12 12 Novembre 2009 ASSEMBLÉE GÉNÉRALE OTTAWA (TR-17) honest that s another place just like contra-cyclical requirements where we haven t been very successful I would say at finding a solution to that. Speaker St-Georges: May I add that the biggest problem that we re seeing with everything involved on the more complex products and for those we currently ask for more than one rating for some of the more complex products so we re not just letting one agency giving the factors to the asset. U-M: For the equity risk, credit risk test, you re going with the 20 percent decline for equity index stocks and 30 percent for managed equity portfolio and presumably there is a rationale that there is uncertainty around volatility with a managed portfolio. I just wondered are you contemplating if and when you actually introduce the change that if somebody can demonstrate that their volatility of their managed portfolio over time is, let s say similar to an indexed that maybe they could get closer to the 20 percent? Speaker St-Georges: And maybe I said that could be more a part of our internal model. Always we have to take into account that with the standard approach and we try to be more risk based if possible but it s still the standard approach and we have to be manageable and be able to check it and watch for other parties. Moderator Dupont: I guess the more direct answer is no we didn t look at that and I suggest you provide it as a comment to the QIS. Any other questions? Well if there are no more questions then thank you for coming and have a nice evening. (Applause) Vol.41, N o.1, Novembre 2009 Délibérations de l Institut canadien des actuaires