Christian-Marc Panneton. ?? = Inaudible/Indecipherable

Transcription

1 JUNE 2009 ANNUAL MEETING HALIFAX PD-30: TR-30 : Liability-Driven Investing for Pension Plans: Moving from Theory to Practice Les placements fondés sur le passif pour les régimes de retraite : de la théorie à la pratique MODERATOR / MODERATEUR: SPEAKERS / CONFÉRENCIERS Christian-Marc Panneton William Solomon Christopher G. Brisebois?? = Inaudible/Indecipherable Moderator William Solomon: I think we should get started. Good morning my name is Bill Solomon and this is a session on LDI solutions. So if you want to do something else now is your time to find a different room, otherwise you re with us for the next hour and fifteen minutes. I am the moderator and the role of the moderator is to find two people who are smarter than him so that they can actually do the heavy lifting and on the panel today is Christian-Marc Panneton and Chris Brisebois and they will be doing the bulk of the presentation. What I m going to do to start is to give an overview as to how I see liability-driven investing (LDI), and what may differ in what you re seeing this morning is we re actually trying to go from the theoretical to the practical. Liability-driven investing is a customized strategy in which a pension plan s asset risk and returns are directly linked with its liabilities. LDI strategies attempt to optimize risk-adjusted returns. The focus of LDI strategies tends to shift from focusing on the return of the pension plan s assets to the volatility of the funded status of the plan, be it a surplus or deficit. In a nutshell the allocation of the pension fund assets is benchmarked to the plan s liabilities rather than the traditional investment return measures. LDI in the broadest sense is not new. If you re old like me you ve seen it before. I would argue that the long forgotten group annuity contracts that were issued for many years by insurance companies and the Annuities Branch of the Federal Government were probably the earliest forms of LDI. Under such approaches both investment and mortality risks were shifted from the plan sponsor to an insuring entity. However that was done for a price. New terminology that enters the discussion includes such terms as risk analysis and risk budgeting, which lead to value at risk. The message here is clear that the investment risks inherent in the traditional asset classes such as stocks and bonds are only a small portion of the risk that a pension plan sponsor faces when compared with the liability risk. LDI provides an approach to managing those risks. Why has LDI seemed to have taken off, especially in the most recent years? Perhaps the greatest motivating factor for many pension plans to consider implementing some form of LDI strategy are the recent changes to International Accounting Standards. These changes which are soon to be adopted in Canada will cause plan sponsors to focus on the volatility of plan surplus or deficit rather than optimizing the rate of return of the fund. What are likely to be the implications of pension plans adopting LDI strategies? While there may be some shift in asset classes away from equities to other presumably less volatile asset classes, I do not believe that this will be the principal change. Greater emphasis will be placed on the structure of the fixed income investments rather than simply investing in a universe bond fund. Use of alternative investments will also increase as well the use of derivative PROCEEDINGS OF THE CANADIAN INSTITUTE OF ACTUARIES Vol. XXXX, No.2, June 2009

2 2 JUIN 2009 ASSEMBLÉE ANNUELLE HALIFAX instruments and Chris is going to talk about that which Chris? Right, I m not unique, I m not the only Chris on the panel, right. Chris Brisebois will talk about that in his presentation. As not all plan sponsors have the same objectives when seeking out LDI solutions, not all LDI solutions are the same. (By the way, it s interesting that the developers of products for the LDI market tried to call them solutions rather than products.) In order to determine which LDI solution is to be implemented, the plan sponsor needs to assess both the asset and liability risks and then decide which ones to address. Among the liability risks we have interest, inflation, longevity; and on the asset side, among others, we have market values and currency. Furthermore these risks are inter-related and not mutually exclusive. As suggested, a pension plan sponsor must evaluate many risks before deciding firstly whether or not to consider an LDI mandate and then secondly which of the many options that are available to them should they pursue. Among the factors to consider are: the actuarial assumptions being used to fund the valuation and in particular the interest rate assumption, the current economic situation of the plan, the risk tolerance level of the plan sponsor and if applicable, the other stakeholders such as shareholders. LDI is by no means a panacea for the problems that are currently plaguing the defined benefit pension plans. Implementation of LDI strategies is not likely to improve a plan s overall rate of return, either in the short-term or the long-term. To the contrary, it may in fact increase the total cost of running a pension plan. What LDI does strive to do is reduce the plan s funding volatility and to improve the risk-adjusted rate of return of the plan. So with that I ll turn it over to Christian-Marc. Christian-Marc is an actuary with Industrial Alliance and he has spent quite a considerable part of his career there and has developed many of the products for Industrial Alliance. Christian. Speaker Christian-Marc Panneton: Thank you Bill. Good morning everyone, bonjour tout le monde. To be reassured that no one rush out to get French translation, knowing that most of the attendants speak English, I will do this presentation in English. I m pleased to be with you this morning, just a few secs to switch the presentation. So as Bill mentioned I will be covering more the basis of LDI from a Life Insurance company s point of view. As he mentioned Life Insurance companies have been interested in what we called previously matching immunization which are the base concepts behind LDI. So I built this presentation to give you not just theory but also some practical implications so we ll start with covering the liabilities, speaking of interest rate risk which is the most significant risk that can be addressed with LDI currently. I will speak also about credit spread risk, speak of other risks that affect pension plans, cover quickly the management process and I will conclude. So to start, the first risk that we want to consider is the interest rate risk and the objective we have with LDI is to mitigate the interest rate risk. I choose the word mitigate because it s not necessarily to eliminate this risk. Maybe the pension plan has some leeway, some surplus and is able to take some risk to enhance the return of a better return. I ve heard it can go to the extreme of pure immunization where all interest rate risk is removed. So the first step is we want to assess what is the risk exposure, we need some metric to assess this risk and this is the theoretical part of the presentation so we ll start with a simple metric, the duration and what is the duration? It s a single number which indicates the sensitivity to interest rate change and our interest will be in the duration gap between assets and liabilities, based on the theory if the gap is zero, we have a duration match and the interest rate risk shall be under control. So we need to calculate the liability duration. We need some information about liabilities and in practice, for the case we have done, it s the first hurdle in the process we have seen that I will say consultants for pension plans are not used to providing monthly liability cash flows so that s an extra step, it needs some development, it needs some time. So typically we ll work with monthly cash flows, expected cash flows from the pension plan. They must be updated regularly to adjust between the actuarial experience compared to the expected which is built in the cash flow. Here in this presentation I used a retiree group, why did I choose a retiree? Because they are retired, their cash flows are more predictable, they are maybe not a closed group but there s no turnover, an employee when he gets his pension cannot cash it or withdraw from the pension, so it s probably the group where we can Vol. XXXX, Nº.2, juin 2009 DÉLIBÉRATIONS DE L INSTITUT CANADIEN DES ACTUAIRES

3 JUNE 2009 ANNUAL MEETING HALIFAX 3 get the best expected cash flow or more stable expected cash flow. Using 4.5% discount rate, the present value of this cash flow is a little above 100 million and the duration calculation is 8.5 years. Using that information, the rule of thumb with duration is with a given duration number, a variation in yield will change the value, the dollar value of the portfolio by minus the duration times the change in yield. For example, having a duration of 8.5, if the rate increases by 1%, we can expect a decrease in value of 8.5%, that s a sizable change or sizable exposure to interest rate risk. In fact the present value as a function of yield is not a linear function, so the duration is only an estimation, and if you look at the number at the bottom, the real change for an increase of rate of 1% is a change of value of 7.88%, a decrease of Why there s a difference between the 8.5 and 7.88? It s due to the curvature of the price function, it s not a straight line so the real value is little above the straight line interpolation. So we can address that with other measure like convexity and I won t go through all the mathematics but if you add and work with convexity, I should not move away from the mike, now you see that we are very close with duration and convexity the change is -7.83% compared to an actuarial change of -7.88%. Using that information we can now look at our pool of retirees our liability. At 4.5% I said the liability duration was at 8.5 and assuming that we have a nice funded ratio exactly at 100% (maybe too optimistic given the current market situation but that s a case study), starting with the traditional asset mix 60% equity 40% bonds. The duration of equities: there s no link between variation of interest rates and variation in the stock market, at least not a direct link that can translate into specific measures so the convention is to assume that the equities have a zero duration and typically the bond part is invested in a portfolio which tracks the DEX universe index. And as of December 2007, the duration of the DEX universe index was So using that mix of having only 40% in bonds, it s simply an average, 40% of 6.57 gives a portfolio duration of So if the rates decrease by 50 basis points (a basis point is 100th of a percent), so it s.5%, the asset value will increase by 1.32, the liability will increase by 4.25 and the funded ratio will decrease by almost 3% to 97.07%, so we have some interest rate exposure here. And it s only a decrease of 50 basis points so we can consider that with this type of asset mix we have a material interest rate risk in the portfolio. How can we improve the situation by using long-term bonds? So here it s a case study so the numbers I choose to get to the nice result using DEX long-term duration ( that s a sub-part of the DEX universe, it only includes the longer term bonds but it s a published index which is readily available), so that duration of the index is and using raw numbers if I choose an asset mix of 33% equity, 67% in that index, I get a duration of almost 8.5 at So now if the interest rate decreases by 50 basis points I have almost the same change in the assets and liabilities, my funded ratio is almost unchanged. Now I ve immunized my portfolio against interest rate risk change. Is it that easy? Probably not, in fact, it s not so easy. So I look at two recent quarters, the first quarter of 2009, look at different performance of the market, the Canadian Stock index represented by the TSX total return index has made -2%, the DEX universe +1.5, the DEX long-term +.03, the solvency liability which is defined in the Standards of Practice is at, for that group of retirees, will be at -1.5%. If I value the liability using the Canada yield curve, the change of the yield curve between December 31st and March 31st, I get -.03% and if I use the cash flow matching liability, the liability valued with the AA yield curve, I get a +4.4%, so I ve numbers all over the place. If I look at 2008 Q4, we have the market crash so -22.7% on the total return index and Q4 does not include September so it s only October, September was not great also, we have positive numbers with the DEX, the solvency and the liability value with the Canada yield curve and a negative value with the accounting value of the liability. So all that translates in our funded ratio. So starting with the 60/40% traditional asset mix, we start with a 100% ratio. Moving from September to December the stock market didn t do well, the DEX universe we gained some value but we end up with a funded ratio of 85.4; moving to March it goes up a little at 86.5%, that s a loss of 13.5% based on the funded ratio. Using the alternative asset mix, using the long bond index, the results are more positive; we now lost only almost 6%, part of that is due to the reduced investment or reduced exposure to the stock market which has very poor returns, another part is due to the good performance of the bonds in the same time. But what we see here is that even in real time if I choose that long duration bond portfolio which matches PROCEEDINGS OF THE CANADIAN INSTITUTE OF ACTUARIES Vol. XXXX, No.2, June 2009

4 4 JUIN 2009 ASSEMBLÉE ANNUELLE HALIFAX the duration and I believe that I was immunized (didn t have interest rate risks), it didn t behave perfectly, yes I had some equity exposure but it was not perfect. So yes it did reduce exposure to interest rate change. Going back to the solvency liability, it s from my point of view I work on the investment side and I need to be prudent because it s the Standards of Practice but from an investment point of view, for me it s an artificial construction, it s based on a single rate which is the Canada rate published by the Bank of Canada, it s the Canada over-ten-year bond plus a spread, it doesn t capture the full yield curve dynamic so the behaviour or the market value of your asset portfolio will never match that benchmark because it s a single rate which is built for statistics. So if we really want to manage the interest rate risk, it s better to value from a risk perspective, risk management perspective, to value the liability according to the market so we need a yield curve to value the liability. A way to do that is to start with the Canada Bond yield curve but bootstrap that yield curve to get the spot curve and the spot curve is a series of zero-coupon rates for different maturities so with these rates we can discount each cash flow of the liability and get in a sense a market value of that liability based on the Canada Bond yield curve. Having that yield curve we can take a step further instead of looking at a single number, the duration, we can also look at partial duration. Partial duration measures the sensitivity of the change in yield for a given maturity. So here we have different maturity which defines the yield curve, we ll look at the one month, three months, six months, one year, two years, different points. In practice, bond traders don t have specific yields for each possible maturity. They will simply interpolate rates between different maturities so that s why we have for example at 180 the 15 years maturity and we move to the 20 years and the 30 years, but we don t have interim maturities. So calculating the partial duration, what we see here that for short maturities we have very low numbers so what tells us is that our liabilities have a low sensitivity to short-term rates. I ve heard the numbers are higher for the longer durations, so it tells us that we have a high sensitivity to long-term rates. Trying to see what that means in market movement, so if I look at the yield curve shift from December 07 to September 08, so the yield curve at December 07 was almost flat, the blue line here, and from December 07 to September 08, the yield curve steepens, gets more sloped, short-term rate decreases and the long-term rate increases slightly. In this case even if on the previous slide we had low sensitivity on the short-term, because the decrease of rates was so large compared to the change in long-term rates, that impact on the short-term rates dominates and we see that the value of the liabilities based on the Canada Bond yield curve increased by.3%. From December 08 to March 09, the yield curve remains almost unchanged, almost both lines are the same. However in this case because the change was negligible both in short-term and long-term, we see an increase in rate in the long-term. That increase because of the partial duration value we have with the long maturities, that effect dominates and now we see a decrease in the long-term rates. So what is interesting and you can argue that in both of these cases +.3%, -.3% it s not a huge change in the value of liabilities but in one case we had almost no change in the yield curve, in the other case we had significant change in the yield curve and for both of these, because of the structure of the liabilities, the impact on the liability was quite low. So that s for the Canada yield curve. Probably you will tell me that you don t invest only in Canada Bonds for pension plans, you also consider Provincial, so that s where I get into the credit spread risk and here what I see with the credit spread. The graph shows the evolution of credit spread on ten year bonds for Québec, the blue line, and the Ontario ten year bonds the red line, and the Ontario spread increased by 47 basis points in Q4 2008, that s a huge increase, it s related to the financial crisis. That may seem huge but if you look at a typical pension plan fund, probably you won t have only Provincial or Government of Canada securities in the portfolio, you may have corporate credit and during last year in 2008, Ontario s spread increased by almost 1%, 93 basis points, but single A-rated corporate spread increased by 305 basis points, 3% that s huge. Now we have seen recently that spread has started to decrease but all that impact on the spread has specific impact on the pension liability. And what I tried to illustrate here on these two next slides is that the credit spread risk can be more material in some situations than the interest rate risk. So in Q4 2008, if I value the liability with the Canada yield curve, what I see. It s the blue line so from September to December the Government of Canada yield has decreased. So what happened with the liability? The liability increased by 8.9%. If I use the provincial curve based on the Ontario, there was also a decrease, Vol. XXXX, Nº.2, juin 2009 DÉLIBÉRATIONS DE L INSTITUT CANADIEN DES ACTUAIRES

5 JUNE 2009 ANNUAL MEETING HALIFAX 5 the red line, the dotted red line, the decrease was smaller and remember that for this liability the long-term or long maturity has more weight value in the impact on the market value or the maybe not the market value but the value of the liability, so here we have a smaller decrease in the Ontario curve so the increase we see is not at 8.9 it s only 4.2. And if I move with the corporate, in that case the increase of spread overwhelmed the decrease in rate so what we see here, the purple line, the A curve increases, leading to a decrease in the value of the liability by - 5.5%. Moving to Q1 2009, I see that the Canada and Ontario curves remain almost unchanged. However as I said we have seen spread return more toward normal levels, so in that case in Q1, the A-rated corporate curve decreases, and I see an increase in the value of the liability. So what we have in our portfolio of assets to match these liabilities will have a huge impact on the performance. I mentioned that I will speak about risk. So I covered the interest rate risk, the credit spread. Other risks that need to be considered are liquidity and the current financial crisis has been in large part a liquidity crisis in the sense it s nice to have corporate bonds in the portfolio but the liquidity on these bonds has been almost nonexistent at some point in the fall of 2008 or early So maybe you don t want to have cash flow matching in your portfolio because it s too expensive but on the other hand, you need to make sure that if you go with corporate securities you can have the liquidity. I didn t talk about inflation. A lot of pension plans are indexed; there are some bonds which are indexed that s another risk. It s possible also to mitigate that risk but these securities are less diverse than for the interest rate risk. So it s harder to manage the inflation risk than it is to simply manage the interest rate risk. Market risk, it s a decision of the pension plan if you want equity exposure that s a risk that you will have to take. Non-financial risks, mortality, Bill alluded to that in his presentation; longevity risk I understand is currently a hot topic for pensions. New entrants, turnover: as I said, I chose a block of retiree people so I excluded that part and I recognize that in practice applying LDI to an active group of participants will be a tougher challenge. And the last part of the presentation, I want to go more in the day to day process of LDI or ALM process, so what I want to emphasize is the importance to start from the client s needs or the client s risk tolerance. The development of an investment policy is a key step, that s where you have to have to define what is the risk tolerance of the client. I referred to duration matching or partial duration that s a metric, but it s really the client that will decide what is his risk appetite toward equities or how much he is willing to be exposed to diminishing funding ratios. From that, we can construct a portfolio and it is not a static process. Once the portfolio is constructed (I illustrated this as a loop), we will have reports showing how well the portfolio matched the liability, what are the residual risks, are they within the tolerance defined by the investment policy, so there will be some following-up regarding that. If there are issues, even if there s no issue, we ll look at optimization, can we generate additional returns which will lead to portfolio re-balancing and it will be a loop going on constantly. So it s not LDI is not necessarily an active portfolio management in the traditional sense where the manager tries to beat the DEX universe bond index, but it s not a buy and hold strategy either. There s a decision making process which I ve illustrated on the previous slide which is a continuous process which involves many people in the team, the optimization requires testing assumptions, there s portfolio decisions and validation. The yield curve analysis I illustrated previously between different types of yield curves, Canada, Provincial, Corporate, we need sources for these prices to derive the yield curve. We use internal and external sources back to the investment policy that will drive the construction of the yield curve. What are the securities allowed in the portfolio: detached coupons, asset-backed securities, asset-backed mortgages, synthetic bonds, corporations public and private, derivative products (Chris will cover that more), forwards, real return bonds which can be used to manage the inflation risk? So we end up on a regular basis having to take a buy and sell decision. These buy and sell decisions can be driven by three reasons: portfolio re-balancing, opportunities that show up on the market (even if we believe, like to believe, that the market is efficient it s not perfectly efficient and from time to time opportunities arise) and there are market events that may force a sale of a security. PROCEEDINGS OF THE CANADIAN INSTITUTE OF ACTUARIES Vol. XXXX, No.2, June 2009

6 6 JUIN 2009 ASSEMBLÉE ANNUELLE HALIFAX The process to construct the portfolio. The objective is to ensure matching of liabilities according to risk matrix, define investment policies, and some of the steps that are taken are mentioned or illustrated here. So in conclusion, managing the financial risks associated with a pension liability requires a very good understanding of the liability financial characteristics, and I can emphasize that from a Life Insurance company, the LDI clients we have and what we have done, it was the first hurdle in the process to get liability cash flow, expected cash flow, financial information on the liability. Next what I mentioned about the investment policy, define the risk tolerance limits which reflect the risk appetite of the pension plan. We need tools to measure and assess the current risk position, and finally portfolio managers are a key element because they can generate added value within the pension plan risk appetite. Thank you and we will switch to Chris. Moderator Solomon: Thank you Christian. Building on that will be comments by Chris Brisebois. Chris works for BonaVista Investment Management which is an affiliated firm of Phillips, Hager & North which is now part of the Royal Bank and Chris has extensive background in asset consulting and money management, so Chris. Speaker Christopher G. Brisebois: Okay, so thanks Bill. So first of all I guess just to talking about where we ve come from so far, I think Christian-Marc has talked about how we can manage one of the key risks that plan sponsors face which is interest rate risk, largely by investing in a large portfolio that s largely dominated by fixed income securities. And that s good if that s the only risk that you want to manage but a lot of plan sponsors of course are under pressure to also find other ways to add return in their investment portfolio. So what I m going to talk about is some of the non-traditional tools that are available now where you can manage interest risk but still get some excess return from the portfolio. So the first thing I wanted to cover off early was just again the importance of liability interest rate risks and with this slide here is kind of a proxy we put together of changes in the market value of the liability s annualized three year periods looking at the past 40 years. And I think what you can see in this chart pretty quickly is that the numbers are pretty volatile. In fact, you know, when you look at it you can see they ve gone as high as 45% for example going back in the early 80 s as well as down as much as 20%. And of course a lot of the talk that we ve been hearing about the perfect storm and so on has been dominated by equities but also by changes in interest rates, which have gone against pension plans over the past little while. So I don t think there s much argument that interest rate risk is something that plan sponsors need to be very concerned about. And that has implications for the investment strategy of a pension fund. If you look at a sort of how a plan sponsor that s deciding on an investment strategy, they only have so much risk capital and risk capital can be defined as really how much pain are they willing to pay for, either in the terms of increasing their contribution rates or being willing to cut benefits. So whatever that line is in the sand, that s how much risk they re willing or able to take. The amount of risk that they actually take is called the risk budget. So they don t actually have to spend all of their risk capital but they do have to decide how much of their risk capital they want to spend and of course many plan sponsors do spend some risk capital when they invest, you know, in a traditional portfolio like 60/40 for example. But let s say you have a plan that also has a mismatch between its assets and liabilities, you know, in terms of the interest rate sensitivities What you can see here is that given the risk budget, that mismatch can actually spend some of the risk within the context of the risk budget. And the question is whether that s where you want to spend the risk as a plan sponsor because ultimately, given the risk that you can take is kind of a scarce resource, you really want to optimize the risk that you spend and make sure that you re spending it in ways where you actually get rewarded. And the question is, if you have that mismatch is that the best way to spend the risk? And so to answer that question, is it a risk that s worth accepting, I think you kind of have to look at it at a couple of different levels. First of all, you have to look at it at the policy level so this is a committee level type decision and here the committee has to decide whether they even have a view on the long-term direction of interest rates, and the important thing to realize is that when you re making this call, you don t have to necessarily be right Vol. XXXX, Nº.2, juin 2009 DÉLIBÉRATIONS DE L INSTITUT CANADIEN DES ACTUAIRES

7 JUNE 2009 ANNUAL MEETING HALIFAX 7 over the short-term but you definitely have to be right over the long-term and probably over the medium-term in order for it to pay off for you. The other way to take advantage of changes in a liability s interest rates and interest rate risks or whatever, is to look at active interest rates strategies and this is handled really at the manger level where you go out and hire a manager to take advantage of their views on interest rates. Regardless of which way you do it however, you do have to have confidence in your ability to forecast interest rates and I think even looking at professionals who do this and look at this on a daily basis, I think it s easy to see that it s not something that s very easy to do, and in fact in general betting on interest rates and the direction of interest rates is generally viewed as a poor quality bet when compared to other sources of return that you can get out there. So the key thing on this is, given the significance of interest rate risk within the portfolio, not to say that you can t decide to take interest rate risk, but you really have to evaluate it I think relative to other strategies to see if that s where you want to spend the risk within the portfolio. And the good news by the way now is that there s some tools that are available to you so you can still manage this interest rate risk but not necessarily give up returns, which was sort of the lead-in to our discussion and that s pretty much what I ll be talking about is you re not forced to take interest rate risk just because you want to have equity exposure. So how do you do that? Well you can increase the asset/liability hedge ratio that s in effect managing the interest rate risk within a pension fund. And some of the things we ve already covered which are kind of obvious are: increasing the fixed income allocation, this of course though reduces the expected return on assets and to the extent that you want to use the investments in the pension fund to also help pay for benefits obviously this becomes more difficult the higher allocation of fixed income you have within the portfolio and particularly if you re underfunded that means if you go all the way you have an immunized portfolio then all your deficits pretty much have to be funded through contributions as well as the current service costs etc The other way to do it is to look at duration extension and I guess a simple example is building off of what Christian-Marc said about zero-coupon bonds where you look at sort of hyper-long fixed-income securities in terms of duration like zero-coupon bonds. So for example if you have a liability that has a duration of 15 and you look at a 30 year zero-coupon bond that has a duration of 30, if you have 50% allocation to that zero-coupon bond that in effect gets you a 15 year duration. The problem with that though is that it introduces significant yield curve risk because you re focusing really on just one part of the yield curve and as we saw, you know, it doesn t take into account convexity and other things which can introduce some big deviations and risks within the portfolio. So it s something that you can do but there are some risks embedded in that. The other thing you can do is look at overlay programs and that s really what I m going to be focusing my comments on, which essentially use derivatives and other instruments to manage interest rate risks. So when I talk about derivative overlay strategies, what am I really talking about? Well first of all the focus is on the fixed income market here but these strategies use derivative instruments either to obtain or to offset desired portfolio exposures beyond those provided by the underlying physical investment portfolio. So what I ve done on this slide is I ve listed some of the tools that are available to Canadian investors which include interest rate swaps, total return swaps, bond forwards and Government of Canada bond futures, otherwise known as delayed settlement bonds. And I m not going to go into details because I only have about 20 minutes and in fact I can talk a lot more about these things if we had more time but one of the key things with these types of strategies is that they do not involve an exchange of cash flows up front, unlike if you went out to go and buy a bond in the physical market where you actually have to buy a bond. In effect with the derivatives strategy you re actually renting the exposures by paying what s called the financing cost. So in effect you re getting the exposure that you want from the physical underlying investment, say a long-term bond, but you re not paying the whole you re not tying up capital in order to do that. So this is accomplished using leverage and of course this is kind of a dirty word these days but I think in the context of what we re doing here, which is managing risk, I think you can look at it as we re using leverage in one PROCEEDINGS OF THE CANADIAN INSTITUTE OF ACTUARIES Vol. XXXX, No.2, June 2009

8 8 JUIN 2009 ASSEMBLÉE ANNUELLE HALIFAX sense to offset some of the other leverage on the other side of the plan which is really the pension plan maybe borrowing for example from the plan members to invest in non-hedging investment strategies so it s kind of a risk management use of leverage as opposed to a speculative use of leverage. Now the implication of this of course is because when you enter into an overlay program, there s no cash flow exchange up front, so that actually frees up capital which can then be re-deployed in other areas where you might think you d get a better reward and, you know, obviously equities can be one of those things or alternative asset classes, hedge funds, whatever the case may be, it just gives you the flexibility to look at different types of strategies. So I listed a few of the instruments that are available to you on the overlay side. How do you decide which ones are most appropriate to use? Well there s actually a number of factors to consider and in fact a lot of these factors vary in terms of their importance depending on what s going on in the markets. So generally what we would recommend is if you re looking at going into an overlay program, don t restrict yourself to one particular type of tool, allow yourself some flexibility so you can move in or out and there have been recent examples where for example swaps have become, you know, where actually the swap spreads went negative in the latter part of December last year whereas if you have the flexibility, if you were in an interest swap at the time and you had the flexibility to move into other strategies, you could have actually capitalized on that. That s just one recent example of why you wouldn t have wanted to limit yourself to just one kind of strategy. But the things that you should also be focusing on are things like market exposures because even though I listed a few of the tools that are available, they don t all have the same market exposures. For example, delayed settlement bonds are really only available for Government of Canada bonds and Province of Ontario bonds so if you have want any exposure to credit for example you have to use other instruments like total return bond swaps for example or total return swaps. The other thing you have to take into account is the credit within the underlying physical portfolio. And so for example, looking at the late settlement bonds with Province of Ontario bonds you have to take into account the credit analysis of the underlying securities of the Province of Ontario bond as well as any other credit analysis like analysis on counterparties etc You have to take into account market and liquidity conditions which again can vary over time and for example interest rate swaps are generally only really liquid on ten-year and 30-year points of the yield curve. If you look at bond forwards which is one of the other tools that I mentioned, they tend to be more available across the spectrum of the yield curve. And other types of tools I guess are available on a contract basis which have sort of a short time horizon like say a three months time horizon, the swaps are an example of that, for example. And even though the underlying security that is being invested in, like say a ten-year or 30-year bond, has a longer maturity to maintain that exposure because the contracts are only three months in duration, you have to keep rolling those contracts to get to maintain that exposure. So that leads to some of the operational risks in terms of what you have to have at the plan level or at the manager level to manage these things. You have to obviously be able in the case of those contracts to be able to roll over the contracts on a regular basis and also because we re using derivatives strategies you also have to be able to deal with any margin requirements or, you know, cash flows, that you may get a call for cash depending on where the value of the trade is going, so you have to be able to manage that. And as I said because these things can change over time, you really don t do this, this isn t a one time analysis, it s a continuous analysis of the various factors to determine the suitable mix. Now I know there s a number of issues to kind of consider when investing in overlay products but one thing I thought I d step back on was just talk about some of the differences between investing and overlay versus or highlight some of the differences between overlay strategies in the physical market. In the physical market let s say there s a change in the liability interest rates that can actually cause a huge change in the value of the liabilities as we saw on the first chart. And I think that s really what s caused a lot of concern for plan sponsors in terms of managing or recognizing the interest rate risk particularly on a solvency or accounting basis where we ve seen the numbers being very volatile. With overlays however the exposure is really limited more to the variability of financing costs which again are based off of short-term rates which kind of interest on the exposure, if you will. So in fact when you re looking at going into an overlay strategy, you re kind Vol. XXXX, Nº.2, juin 2009 DÉLIBÉRATIONS DE L INSTITUT CANADIEN DES ACTUAIRES

9 JUNE 2009 ANNUAL MEETING HALIFAX 9 of going into a trade-off between trading off the long-term capital gains or losses that you can experience in the physical market with essentially variability and financing costs which for a lot of plan sponsors is something that they re trying to avoid, that is, the big movements in the market rates. But that also means that the financing costs are variable because as we all know cash or short term rates can move around quite a bit and for example if your financing cost go one year from 3% to 4% well that represents a one-third increase in your financing cost. The other thing though that s interesting about this is that kind of your benchmark with overlay becomes beating cash because that s the rent that you pay on the assets that are freed up to invest in other strategies, and so that s different. For example say if you re investing say the traditional 60/40 mix where you re trying to add value relative to say the liabilities, you kind of have to beat the liability rate of return, not necessarily the cash rate of return. And so it kind of brings other types of asset classes with maybe a little bit lower yield into play in terms of what you can look at, particularly if they have good diversification benefits and so on. So there are some differences in terms of how you can perceive the investment problem by going into overlay strategies. One of the things that I also wanted to do to help to illustrate the concept was provide a couple of examples, it s always helpful, and so I m going to just highlight a couple here. One is let s say you re a traditional, you ve got a traditional type of asset mix, 60% equities, 40% bonds and you want to maintain the current level of or current expected rate of return but you also want to manage your interest rate risks. Well you can do that using an overlay program and this is an example of how you can do it. First what you do is you have your 60/40 mix and you enter into an interest rate swap which in effect when combined with the bond portfolio of your portfolio will actually match off the return in risk characteristic of the liabilities. And you can kind of see that graphically if you look on the right hand side of the chart where the right is the blue part of the assets it kind of matches up the liabilities. Now because there s no exchange of cash flow like I said up front that means you don t have to sell equities, you don t have to sell your bond portfolio in order to enter into the interest rate swaps, so even in effect maintain your equity position through this process which leaves your structure kind of, you know, untouched in terms of getting returns. Now there is a caveat to this which of course is there are costs associated with this, the financing and other costs I ve kind of ignored in this chart and I ll give you an example of how they work in a second, but this is kind of on a high level how these things work. Another example that s been getting lots of attention is portable alpha. And portable alpha really is the process of separating market returns or beta from skill-based returns which is alpha. Now there s a variety of different ways and nuances of this that aren t restricted necessarily to LDI but one of the applications to LDI is the beta source can actually be a bond portfolio that matches off the characteristics of your liabilities and the alpha source can be basically anything but a lot of the marketing and products that are available have been focused on kind of hedge funds, so that s the example that I m going to focus in on here, but these all use the fixed income overlay type of tools that I was talking about. One caveat before I kind of illustrate this example though, is that alpha doesn t grow on trees, as it says here, so the structure itself doesn t actually create alpha, you still have to identify a manager that can do this for you and the things that you should be looking for to do this include stability of alpha, in other words, most of the time you win (obviously that s what we d all like to do) and also that your alpha is uncorrelated with other alpha sources or the market. And this again feeds into the idea that you want stability of alpha because if you have less correlated sources then that actually helps diversify from a diversification point of view. And then the very important point too is even if you find this alpha you have to be able to do it in a cost-effective way so that you can benefit from it, so it has to exceed fees and other transaction costs as part of the process. So as the example of this, let s say you re a plan sponsor that starts off with a bond portfolio that matches your liabilities and you d like to beef up the returns through a hedge fund. How do you do that? Again a similar approach, you take some of your bonds that you have in your portfolio, you sell those bonds, you ll get proceeds from that and then you ll enter into an interest rate swap which in effect replaces that interest rate exposure that you just sold to gain the proceeds and then you use the proceeds of those bonds that you sold to buy a hedge fund essentially. And when you add it all together again, because you ve entered into the interest rate swap and you ve PROCEEDINGS OF THE CANADIAN INSTITUTE OF ACTUARIES Vol. XXXX, No.2, June 2009

10 10 JUIN 2009 ASSEMBLÉE ANNUELLE HALIFAX got the remaining bond portfolio, that combination matches the characteristics of your liabilities and then those proceeds that you got from selling the bonds, you buy the hedge fund with it and get the hedge fund exposure. So again in the same kind of context you can see that we ve managed to match off the interest exposure using swaps in the remaining bond portfolio but still been able to add return on top and again a key element of this is that you didn t have to sell anything to enter into the interest rate swap, there was no exchange of cash flows which is why you could use the proceeds to get into the hedge fund. So what have we learned? Some of the key take-aways here are, first of all, that there are some tools that are available to allow you to essentially hedge that asset/liability hedge ratio to whatever it is that you want. I ve given a couple of examples that hedged essentially all of the interest rate risk but you can do it along the spectrum, whichever you prefer to do. Also the LDI framework is a framework but it does not require reducing expected return and as Bill said it doesn t necessarily create return either, it s just a different framework for analyzing really the risks that are embedded in the portfolio more explicitly and making sure that you don t have any unintended or undesired bets within the portfolio and this allows you really to take risk where you think you ll be rewarded as a result. Some caveats which I think have already been mentioned. We ve been talking about this more from an investment perspective but LDI on its own doesn t really manage what we call underwriting risks (which I think was covered by Christian-Marc) which include longevity, you know, terminations, that kind of thing, so that s something to keep in mind. And also while I think that a lot of the interest in these products has been driven by what s been happening in solvency and accounting lately, I think that it s important to note that you can t necessarily hedge, you know, solvency and accounting kind of at the same time because they re valued on different bases but definitely investment strategies can be tailored to help manage these risks and I think that s why they re getting more attention as we ve seen. So that kind of gives you a high level view of how you can use I think overlay strategies and how they can help, you know, maintain structures within a portfolio and manage interest rate risk but I just thought it s important to talk about some of the implementation considerations of doing this in addition to some of the ones I already talked about. And so one of the key ones of course is cost, that s a consideration, and I ve used a portable alpha example which is maybe a little bit more complicated an example but I can just show you how some of the cost can sometimes add up and why you can t just assume that you re going to add value by going into an LDI structure on its own without taking the due diligence. So let s say, you know, you have a long bond portfolio and you d like to overlay a hedge fund manager on top of it, similar to the example that I just demonstrated, and the reason why you want to do that is because you found this miraculous hedge fund manager that you think can generate a return of 10% per year over the long-term which is actually a very, you know, difficult target. If you look at the fees the hedge fund manager is going to charge you for it though, it s going to be 2% flat plus performance-based fee which is typical in the hedge fund space of 20% of the return over and above T-Bills and the T-Bills yield I ve used in this example is 4% (it s actually less than that now but for this example 4%), and transaction costs which include legal cost, other kind of cost that are embedded within some of these derivative structures of say.5%. So let s say your hedge fund manager does exactly what he was supposed to do, gets his return of 10%, what does that mean for the fund? Well if you look at the cost that comes off of that, you can see first of all there s the 2% investment management fee, which is the flat fee plus the performance-based fee of.8%, so that s the 20% over and above the 6% essentially over and above T-Bills plus the financing cost which is 4% plus.5%, so it s kind of plus plus plus, what that leads you to though is an after-fee alpha of about 2.7%. So you had a hedge fund manager that returned 10%, what you get is an alpha of 2.7% over and above your long bond portfolio. So it just highlights the importance I think of doing your due diligence on these things. Now the financing cost is something that would you don t have to, you know, use a hedge fund manager that charges two and 20 for example, but largely the financing costs and the transaction costs are things that would apply to sort of any strategy, fixed income overlay strategy, that you would come across. Vol. XXXX, Nº.2, juin 2009 DÉLIBÉRATIONS DE L INSTITUT CANADIEN DES ACTUAIRES

11 JUNE 2009 ANNUAL MEETING HALIFAX 11 Just thought I d highlight a couple of other implementation considerations as well. I kind of listed them off here and again just in the interest of time, you know, I can t spend too much time on them, but some of the key ones that are specific to these strategies are the use of leverage and that s something that actually is restricted by a lot of SIP&Ps (Statement of Investment Policies & Procedures) currently, you know, it has to be addressed in a lot of SIP&P s at the very least. Counterparty risk is something that s gotten a lot of attention lately because a lot of the key counterparties like for example Lehman Brothers was a counterparty in a lot of derivative transactions, and so on. There s been a lot of issues around counterparty risk and in fact a lot of the margin requirements and other things have been strengthened to help address this and it s also to a certain degree impacted the liquidity of some of the derivative transactions that we ve seen particularly in the latter part of last year and the early part of this year. It seems to be getting a little bit better but it s created some issues. I m sure there s going to be a lot of case studies done on the last few months once this is all over. Also as Christian-Marc talked about, while I ve talked about managing interest rate risk, you know, I sort of the assumption was you can manage all of the interest risks within a portfolio, the reality is that you can t really do it perfectly and do it really in an efficient basis. So there is some tracking there with the liabilities but often this is quite manageable. Other operational things like the margin maintenance and so on also there s additional contracts that you have to enter into, like an agreement in futures trading account and so on which have to be papered up when you go into these types of tools, which can be time consuming, they can have legal costs associated with them etc so you have to be able to handle that. And of course you have to have the knowledge level and comfort level to use these before you go into them. I just think it s always a good idea for any type of product so. So that might lead you to the question, well is this really only a solution for larger plans that have the resources to actually do this? And you can understand that there are some sort of limitations for smaller plans given the complexity, given just even the papering up, the additional work that has to be done to set up these strategies, but that sort of highlights I think the benefits of pooled fund approaches which are becoming more and more available I think in this space, and the advantages of pooled fund approaches is that they require sort of more modest oversight requirements and fewer demands on internal resources obviously which leads to lower minimum asset sizes. Also if there is an agreement in futures trading accounts and so on that I talked about being set up or needed to be set up on the previous slide, those are all done by the pool not by the actual client and so that just makes it a little bit easier to handle. Of course the drawback of pooled funds is that you don t necessarily get the same level of flexibility or customization so that s something to consider. You want to make sure that whatever you go into meets your investment objectives. So I know it s been a little bit technical, a little bit heavy topic, but I just want to leave you with kind of four key comments. One is that LDI is a framework that really is about managing risk relative to the liabilities and by using this framework I think it allows you to manage where the risks are being taken and take them where you want to take them as opposed to being forced into certain risks. And interest rate risk generally is not as I said considered a high quality source of alpha and now there are tools available that allow you to manage that as I said given its significance in the overall scheme of things. With overlay programs, it brings the problem back to beating cash which may be, you know, in terms of adding value perhaps can be viewed as a little bit easier hurdle potentially, which positions the plan for introducing alternative sources of return. And finally, which is probably sort of an implication for everybody in this room, is if LDI really kind of takes off and I think it already sort of has, it really does require a lot more coordination between the various parties including actuaries, investment consultants, investment managers. So anyways that s it. How much time left? Moderator Solomon: We have a few minutes left for questions but before I open the floor up for questions I just wanted to make the comment that for pension clients before they decide whether or not they re interested in implementing an LDI strategy, what they have to determine is what risks are they concerned about? What s the best way to deal with those risks? Do they want to eliminate them? Do they want to mitigate them? Do they want to control them? And then what are the options available? Christian and Chris have talked about a number of them, PROCEEDINGS OF THE CANADIAN INSTITUTE OF ACTUARIES Vol. XXXX, No.2, June 2009