Foreign Content in Institutional Portfolios.

Transcription

1 September 2006 Foreign Content in Institutional Portfolios. How Much is Enough and Should it be Hedged? Historical evidence shows that Canadian investors can expect to achieve higher returns and/or less risk by increasing their holdings of foreign securities; yet since the elimination of the Foreign Property Rule in 2005, there has been no mad rush by institutional investors to increase foreign content in tax-preferred funds. At Phillips, Hager & North, we monitor the trends and issues that influence our clients investment experience and periodically produce research articles and commentaries to help them address investment questions. In this article, we consider optimal foreign content allocations (equities and bonds) in institutional portfolios with defined future cash flows (e.g., pension plans and endowments); and we further explore whether any or all of the foreign content allocation should be hedged back into Canadian dollars. We also provide a checklist to assist trustees and other fiduciaries looking at these issues in relation to their own situation.

2 Longer-term q What is your view on currency management tactical hedging, strategic hedging or alpha transport? q Ideally trustees should set aside some time to discuss this core issue and then review the decision on an annual basis, as market conditions change or as trustees become more comfortable/ knowledgeable about strategies such as alpha transport. q Does your SIP&P accommodate your currency-hedging stance? q Once the discussion on hedging has occurred and a decision has been taken, be sure to check your SIP&P to ensure it can accommodate the agreed upon stance on currency hedging. q Have you taken liabilities into consideration in your plan s investment strategy? q Our research highlights the fact that the results for currency hedging and optimal foreign content allocation can vary significantly depending on whether you consider your situation in an asset-only context or an asset/liability context. We would urge trustees to take liabilities into consideration these are the future obligations of your pension plan, endowment fund or foundation and their inclusion can materially alter the outcome of foreign content and hedging deliberations. q Does your domestic fixed income manager have the capabilities to manage foreign bonds? q As part of the due diligence process, trustees should ensure that their domestic fixed income manager possesses the appropriate skill set to manage foreign bonds, especially if you chose to allow this in your SIP&P and have decided to retain your domestic bond manager. q Do you allow investments in foreign bonds denominated in Canadian dollars and/or in foreign currencies? q The elimination of the foreign property rule has played a role in the growing issuance of foreign bonds denominated in Canadian dollars, otherwise known as Maple bonds. Issuance of Maple bonds has exploded in the last year and provides an opportunity for investors to buy foreign credits with ratings comparable to well-known Canadian entities with a yield advantage and no currency or foreign interest rate risk. Does your SIP&P restrict your fund s opportunity set by not allowing investment in foreign bonds denominated in Canadian dollars and/or in foreign currencies? q Should you talk to a foreign currency manager or a foreign bond manager? q As part of due diligence, it may be worthwhile for trustees to seek input from foreign bond managers or foreign currency managers to gain an understanding of the potential risks and rewards and to assess the skill set required for successful management. An investment consultant may provide helpful guidance in this regard.

3 Trustee Checklist In order to help you review our analysis in the context of your own pension plan or endowment fund, we have developed this checklist of salient questions for trustees and other fiduciaries contemplating foreign content in taxpreferred plans and funds. Short-term q Where does your plan/fund fit on the risk spectrum? q Are you a low risk investor with a smaller strategic allocation to equities or a higher risk investor with a larger allocation? q What restrictions (if any) does your Statement of Investment Policies and Procedures (SIP&P) place on foreign content? q Does your SIP&P include a clause with respect to foreign investment, either in stocks or bonds or both? Have you set a limit on the maximum exposure to foreign investments that is appropriate for your plan s risk tolerance? q Does your SIP&P allow currency hedging or the use of derivatives to manage currency? q Some clients allow their investment managers to utilize foreign securities but prohibit them from using derivatives. The manager may wish to use currency forwards, for example, to hedge a position in foreign stocks or bonds. Be sure you are not giving conflicting signals by saying, yes, you can buy foreign securities but no you can t hedge. Those uncomfortable with the notion of allowing their investment managers carte blanche usage of derivatives may decide to restrict the use of derivatives to foreign currency transactions and/or may require prior permission before entering into these kinds of arrangements. To Hedge or Not To Hedge q What is your time frame? q If it is shorter-term, hedge more. In the short-term, currencies can be extremely volatile and can impact the risk profile of your investment portfolio. If your timeframe is longer, hedge less since history suggests that, in the long-term, currencies are mean reverting. q What is your context? q Do you consider your plan in terms of assets only or in terms of both assets and liabilities? If you include liabilities, you may want to hedge less. * q What is your benchmark allocation to equities? q If you have a higher equity allocation, less hedging may be desirable. q What is your investment belief with regard to currencies? q If you have a strong view that currency returns will be positive or negative, this may affect your preferred hedge ratio. * This seemingly unintuitive result arises from the fact that in an asset-only context, the model generates portfolios containing partially-hedged allocations to foreign bonds; whereas, in an asset/liability context, the foreign bond allocations disappear from the recommended foreign content, leaving only foreign equities all of which our model suggests should be unhedged.

4 Introduction In February 2005, the elimination of the 34-year old restriction on investment in foreign property by taxpreferred investors raised the question of what is the optimal allocation to foreign investments. Historical evidence shows that Canadian investors can expect to achieve higher returns and/or less risk by increasing their holdings of foreign securities. Yet, almost two years later, there has been no mad rush on the part of institutional investors to increase investment policy allocations to foreign bonds or equities. This is possibly due to the recent strong relative performance of Canadian equities and the strength of our Canadian dollar, but these conditions may not persist (as recent volatility in global capital markets can attest) and many pension plans and endowments are contemplating what to do with their foreign content. In January 2006, JPMorgan surveyed 104 Canadian pension plans managing $240 billion in assets and found that over 80% of the respondents expected to increase their allocation to foreign assets to over 30% within the next three years. 1 Fully 28% expected to boost their foreign holdings to over 40% in five years time. Drawing on in-depth research using our proprietary analytical model, this paper explores the issues of optimal foreign content and optimal hedge ratios for tax-preferred institutional investors. The model and key assumptions are explained on the page that follows and additional detail is available upon request. Our research studied a myriad of scenarios for a continuum of investor profiles. This paper focuses on the model s recommended portfolio mixes (using nine asset classes) for four risk profiles under three different economic scenarios (current, inflationary and disinflationary). We considered these findings from two investor perspectives: a traditional asset-only perspective in which risk is viewed as the volatility of the asset portfolio s returns (i.e., to what degree the portfolio returns go up or down over time); and a more holistic asset/liability perspective in which risk is seen as the volatility of the difference between the asset portfolio s returns and the changes in the value of the liabilities 2 that are to be funded by that portfolio (i.e., to what degree the returns can fund the expected obligations over time, accounting for the fact that the value of the obligations is not static). We included a focus on foreign bonds because this asset class is somewhat unfamiliar to many investors, overlooked when foreign content was capped in favour of the greater return potential of foreign equities. Finally, we also used the model to help recommend a hedging policy because investors in foreign securities have a choice to expose their portfolio to currency risk or to hedge their portfolio against it. Our study confirmed that there is no one size fits all for foreign content allocation and hedging policy. It showed that foreign equities can be expected to be helpful (in terms of the uncorrelated return potential and the diversification they offer) and should be roughly equal to the Canadian equity allocation. Adding foreign bonds to the portfolio benchmark on a policy basis, however, was shown to be beneficial only when the focus is on assets and ignores the liabilities that the assets are expected to cover. Amending policy documents to allow for opportunistic investing in foreign bonds, instead, allows the investment manager to pursue value-added opportunities as they arise, while still being benchmarked against a domestic bond index. In terms of managing currency risk, we advise that a non-zero hedge ratio, while not prescribed by our model, may nonetheless be prudent, based on investor-specific criteria. To help trustees and other fiduciaries analyse their own situations, we have included a Trustee Checklist at the end of the paper. It poses questions intended to help you identify salient issues, including your pension plan or fund s willingness/ability to invest abroad, your view on hedging, and implementation considerations. 1 Source: Globe & Mail, January 18, Liability refers to the present value of the plan s or endowment s obligations i.e., the actual future payments of benefits, expenses and disbursements. Foreign Content in Institutional Portfolios: How Much is Enough and Should it be Hedged? 1

5 The Model and Assumptions The model employed is a mean-variance optimization, efficient frontier model, that employs analytical techniques developed by Harry Markowitz in the 1950s. It allows for optimizing asset allocation within two contexts: 1) an asset-only environment and 2) an asset/ liability context. Within these contexts, we examined three different scenarios the base case or neutral scenario, reflecting current bond yields and the full historical data set with regard to volatility and correlation; an inflationary scenario; and a disinflationary scenario. All model simulations were unconstrained with regard to the foreign content, equity/bond allocation and the currency hedge ratio. The key assumptions needed for our study are: the expected returns for each asset class, the expected volatility of these returns, and the expected correlations between the returns of the various asset classes identified in the analysis. Our base case assumptions are outlined in Table 1. In each case, the projected returns were meant to incorporate a long-term view of at least ten years. Note that these are geometric returns and that foreign refers to a global portfolio that includes the United States, the U.K., Japan and Germany. Table 1: Base case assumptions for nine asset classes Expected Return* Volatility Cash 3.0% 1.5% Mid-term Cdn nominal bonds 4.0% 5.2% Long-term Cdn nominal bonds 4.2% 15.3% Real Return Bonds 4.2% 15.3% Foreign nominal bonds ** 3.5% 4.6% Canadian equity 7.0% 16.0% Foreign equity 6.5% 16.0% Currency exposure 0.5% 4.5% Currency hedge 0.5% Hedging cost 0.15% * Geometric return ** Mid-term bonds, in local currency We also modeled the foreign asset classes separately in two parts: hedged in Canadian dollar terms and unhedged in local currency terms (thus subject to the impact of relative currency movements) for reasons outlined in the paper. We did not examine emerging markets equities or alternative asset classes such as private equity, hedge funds or real estate in our analysis for two reasons: The federal foreign property rule was not the limiting factor for whether or not Canadian investors held these types of assets; as well, it is very difficult to determine reasonable input assumptions for these asset classes. In the asset/liability context, we modeled both nominal liabilities (i.e., no inflation indexation before or after retirement) and real liabilities (i.e., full indexation both before and after retirement). By combining the two types of liabilities in different proportions, we were able to model partially indexed liabilities as well. Key Assumptions 1. For cash, we assumed a return equal to cash yields as of December 31, 2005 and assumed volatility and correlations were consistent with history For a nominal bond, in the absence of interest rate changes, its return will be equal to its yield at the time of purchase. We set our expected returns on domestic and foreign bonds equal to their yields as of December 31, When adding in the impact of foreign currency, we assumed the expected return on the hedged and unhedged foreign stocks and bonds was the same as that of domestic stocks or bonds less (in the hedged case) the cost of hedging transactions. We determined bond and stock volatility using historical yield, return and currency movements. 4. We assumed that Canadian real return bonds (RRBs) would return current RRB yields (as of December 31, 2005) plus current market inflation expectations. 5. We assumed equities (in local currency terms for foreign equities) would return 3% more per year over the long run than mid-term bonds (using mid-term Canadian bonds for Canada and mid-term foreign bonds for foreign), but would have a higher volatility as well. In other words, we assumed a 3% equity risk premium. 6. The expected return on unhedged foreign currency exposure was assumed to be the difference between the expected return on Canadian mid-term nominal bonds and their foreign counterparts. 7. The expected return on a currency hedge was set to the same amount, minus a small assumed transaction cost to implement the hedge. The assumed hedging transaction costs were based on the approximate size of the bid/offer spread on currency forwards (i.e., 0.10% per year for a U.S. currency hedge and 0.15% per year for a global currency hedge). 8. We also performed a significant amount of analysis of the historical data, to ensure the reasonableness of our base case assumptions and, where justified, additional model simulations based on alternative assumption sets to confirm that our base case conclusions remained valid. 3 We used data going back to 1949 as far back as it is possible to find complete data. The details of our assumptions are available upon request. 2 Phillips, Hager & North Investment Management Ltd.

6 Analysis of Foreign Content Allocation For each given set of assumptions, the model we developed seeks to generate an efficient frontier, or set of portfolios, that provides maximum reward (i.e., return) for a given level of risk tolerance. For ease of illustration, we have defined this as the level of equities within a portfolio s asset mix policy. Chart 1: Traditional efficient frontier analysis 9% 8% 7% 100% equities 80% equities risk portfolio, or MRP 4, to the most risk tolerant with a 70% policy allocation to equities. The four incrementally riskier portfolios with incrementally larger allocations to equities were considered from an asset-only perspective and then again adding in the liabilities. Table 2: Four investor profiles considered from two perspectives Risk Profile Asset Only Asset/Liability* Minimum Risk Portfolio ü ü 20% Equities ü ü 50% Equities ü ü 70% Equities ü ü Return 6% 5% 4% 20% equities 0% equities 40% equities 3% 4% 6% 8% 10% 12% 14% 16% 18% Volatility 60% equities Most analyses of foreign content approach the issue from a traditional asset-only perspective. Thus, we, too, began our analysis from this perspective, testing it in the three different economic environments. This approach may make sense for investors with few funding concerns. For example, a young pension plan with net annual contributions that are large compared to the accrued liabilities; or a well-funded endowment that can afford to ignore asset/liability matching considerations and focus on assets. Since most pension and endowment investors cannot afford to focus exclusively on assets, however, we then expanded the context to include the liabilities that are to be funded by the asset base. In fact, understanding future streams of cash flows is usually our first step in constructing pension or endowment fund portfolios. Efficient Frontier in an Asset-only Environment The strengths of this approach are that the model will derive the asset mix that produces the highest expected rate of return for a given level of volatility of return (or risk), the inputs are readily available and the outputs (optimal mix for a given level of risk and return) are easily understood. As noted, we examined a continuum of risk tolerances and several investor perspectives. In this report, we present eight different situations, as summarized in Table 2. Investor risk tolerance ranged from the least risk tolerant with a portfolio known as the minimum * Assumes 50% real / 50% nominal liabilities. Model Results Chart 2 shows the efficient frontier results from an assetonly perspective, with no foreign content limit. In this context, reward is defined as the expected return of the asset portfolio. Risk is defined as the expected volatility of that return. The addition to the portfolio of an asset class that has a low return correlation with the portfolio s other assets will tend to reduce the volatility of the return of the total portfolio. Along the horizontal axis, we have increasing levels of risk tolerance, defined as policy allocation to equities. Along the vertical axis, we have the allocation to the various asset classes. The expected rate of return is included at the bottom of the chart. Chart 2: Asset-only Context Efficient frontier portfolios feature increasing allocations to both foreign equities Foreign equities Canadian equities Foreign bonds Canadian bonds and bonds The model optimizes by seeking uncorrelated asset classes in order to diversify and reduce the volatility of returns. In an asset-only context, the key factor is Foreign Content in Institutional Portfolios: How Much is Enough and Should it be Hedged? 3 Allocation 100% 80% 60% 40% 20% 0% Expected return MRP 20% equity 50% equity 70% equity Portfolio 3.1% 4.4% 5.9% 6.6% Increasing risk Cash 4 The MRP in an asset-only context is the asset portfolio with the lowest expected return volatility; the MRP in an asset-liability context is the asset portfolio that most closely replicates the economic behaviour of the liabilities being considered.

7 stability of returns. The minimum risk portfolio (MRP) in this context is primarily allocated to cash, the asset class that produces the least volatile return there is no consideration of the liability behaviour in an asset-only definition of risk. As risk tolerances increase, the model allocates an increasing proportion to foreign equities (grey) and foreign bonds (pink). For a highly risk tolerant pension plan with a 70% policy allocation to stocks, the model allocates roughly equal amounts to foreign and Canadian stocks and a higher relative proportion of the bond portfolio to foreign bonds. In an asset-only context, the optimal ratio of Canadian equities to foreign equities is consistently around a 50/50 split, no matter what level of risk tolerance. Chart 3: Asset-only Context Canadian & foreign equity allocations are roughly equal Percentage of portfolio 40% 35% 30% 25% 20% 15% 10% 5% 0% MRP 20% equity 50% equity 70% equity Portfolio Canada Foreign Asset-only Context Key Findings 1. A low-risk investor (in other words, a portfolio with a low equity benchmark allocation) would allocate very little to foreign stocks or bonds. 2. As risk tolerance increases, the model allocates increasing amounts to foreign equities and bonds because returns are uncorrelated with domestic assets and currency is a good diversifier. 3. Allocations to foreign equities were roughly equivalent to the allocations to Canadian equities, regardless of risk tolerance. Efficient Frontier and Liability-linked Analysis Given that the purpose of a pension fund is to pay its future obligations, the characteristics of the ongoing financial commitments should be taken into consideration in designing the investment policy. A perspective that includes liabilities seeks to ensure that the level of risk that is undertaken through various investment strategies takes into account the nature and economic behaviour of the plan s liabilities. The liabilities of a Canadian pension fund are the present value of a projected stream of future nominal and/or real payments that are denominated in Canadian dollars and whose present value is discounted using Canadian bond yields. This promise or obligation for a defined benefit pension plan often extends for 80 years or more. Similarly, liabilities for other non-pension investors, such as endowments and foundations, are often a series of long-term future payments denominated in Canadian dollars. To the extent this is true, these investors would face a very similar asset/liability environment as do pension plans. In an asset/liability context, the scope of the efficient frontier problem changes. Now reward is the expected excess of the return on the asset portfolio over the change in the value of the liabilities (i.e., the growth in surplus), and risk is the expected volatility of that excess return (i.e., surplus volatility). The model seeks to maximize the expected return of assets over liabilities for a given level of surplus volatility. This is a more comprehensive approach that looks at the whole investment equation rather than just at the assets. The asset/liability risk of a pension plan will be reduced with the addition of foreign securities if it increases the inclination of the plan s assets to move in tandem with the value of the plan s liabilities. Model Results In our analysis, we looked at many different liability structures including the following three: % nominal liabilities this mimics a flat dollar pension plan such as a multi-employer plan with no inflation indexation before or after retirement or a typical endowment % real liabilities this mimics a fully indexed pension plan 3. 50% real/50% nominal a combination of the two liability structures outlined above. This is the structure we present in this research paper. In an asset/liability context, the key factor is how the assets correlate with the liabilities. The model optimizes by choosing assets that are as highly correlated with liabilities as possible. Chart 4 looks at the model results using the combination of 50% real and 50% nominal liabilities. When liabilities are included, the MRP allocates almost the entire portfolio to Canadian bonds. In this context, the MRP is the asset portfolio that produces the least surplus volatility (i.e., matches 4 Phillips, Hager & North Investment Management Ltd.

8 the investment behaviour of the liabilities as closely as possible). High volatility of asset returns is not a concern, and indeed, may be desirable, if it matches a similar high volatility of the liabilities. Chart 4: Asset/Liability Context Efficient frontier portfolios Allocation 100% 80% 60% 40% 20% 0% Expected return MRP 20% equity 50% equity 70% equity Portfolio 4.3% 5.2% 6.2% 6.7% Increasing risk Foreign equities Canadian equities Foreign bonds Canadian bonds Cash feature foreign equities but not foreign bonds As risk tolerance increases, the model allocates increasing proportions to equities both Canadian and foreign but not to foreign bonds. This is because as risk tolerance rises, the model seeks to add risky assets that will deliver as much return as possible for the level of risk (asset/ liability mismatch risk) they produce in other words, equities. Foreign bonds disappear from the efficient frontier allocation because in an asset/liability context, a key determining factor is how the assets correlate with the liabilities. The model optimizes for a given level of return by choosing assets that are as highly correlated with liabilities as possible; since foreign bonds add no additional return relative to domestic bonds, but do increase the asset/ liability mismatch, they are not selected by the model. Asset-Liability Context Key Findings 1. The introduction of liabilities significantly alters the results. 2. Allocations to foreign equities (as a proportion of the equity portfolio) remain fairly constant, with an even split between Canadian and foreign equities, regardless of risk tolerance. 3. Foreign bonds disappear completely from the portfolio. Alternative Scenarios Inflation and Disinflation In the past, there have been prolonged periods of time dominated by a secular trend of rising or falling inflation, and increasing or decreasing bond yields. With bond yields recently touching multi-decade lows, many question whether we are about to return to a period of higher interest rates and higher inflation, or alternatively whether inflation and interest rates will continue their secular downtrend. To broaden the relevance of our findings, therefore, we developed a set of assumptions for an inflationary scenario and another set for a disinflationary scenario (i.e., an environment in which the rate of inflation is positive, but falling). For the inflationary scenario, we assumed that inflation would run at 4% per year as compared to 2.5% in the base case. We assumed that foreign and domestic nominal bond yields would rise by 0.3% per annum and that inflation expectations would also increase. We based the correlation of returns from the asset classes on the period from January 1, 1966 to December 31, 1981 a period in which inflation rose from 3% to over 12%, and Government of Canada 10-year bond yields soared to 15% from 5%. For the disinflationary scenario, we assumed that inflation would run at a 1% rate and that foreign and domestic nominal bond yields would fall by 0.175% per annum alongside a drop in inflation expectations. In this case, we used the period from January 1, 1982 to December 31, 1994 for historical correlations, a period when inflation fell from over 12% to zero and 10-year Government of Canada bond yields slumped from 16% to 9%. With an asset-only context in a world of rising inflation, the model prescribes allocations to cash, domestic equities and foreign equities and not to bonds (domestic or foreign) as compared to the base case. In a disinflationary world, the model allocates more to bonds (domestic and foreign) and less to equities than in the base case. Given an asset/liability framework and rising inflation, the model allocates a significant portion of the portfolio to real return bonds (RRBs). With disinflation, however, the model allocates more to domestic long bonds and more to domestic equities than in the base case. In both cases, the results of the model align with what intuition would suggest. Alternative Scenarios Key Findings 1. Asset-only investor: the model prescribes no bonds in the inflationary scenario and both domestic and foreign bonds in the disinflationary scenario and less equities. 2. Asset/Liability investor: the model prescribes a significant allocation to RRBs in the inflationary scenario and more domestic long bonds in the disinflationary scenario. A Focus on Foreign Bonds 5 Under the constraints of the former federal foreign Foreign Content in Institutional Portfolios: How Much is Enough and Should it be Hedged? 5

9 property rule (FPR), Canadian pension plans tended to allocate the allowable foreign content entirely to foreign equities, which were perceived to offer enhanced returns and the benefits of diversification. With the elimination of the FPR in 2005, investment professionals began to explore whether or not there is a greater role for foreign bonds in institutional accounts and, if so, how to implement investment in this hitherto largely unfamiliar asset class. The focus of our research was on non-canadian dollar bond issues, which we identified first as global bonds (including U.S. and other foreign issues) bonds and, in a separate analysis, as U.S. bonds exclusively. This was done to reflect the greater familiarity of Canadian investors with the U.S. markets and, therefore, the greater likelihood that a foray into foreign bond markets could start with investing in U.S. bonds to the exclusion of other foreign bonds. The arguments for investing in foreign bonds are similar to those used to support investing in foreign stocks the Canadian market is relatively small, accounting for just 3% of the global bond market and investors can access considerable diversity by investing outside of Canada. We examined three potential arguments for holding foreign bonds: Higher expected returns Lower expected risk Higher potential for investment managers to add value ( alpha ) Foreign Bonds Key Findings 1. Neither theory nor historical evidence support the argument that an allocation to foreign bonds (as an asset class and at the expense of domestic bonds) should increase a portfolio s expected returns. 2. Historical evidence appears to support the argument that an allocation to foreign bonds may reduce a portfolio s risk in an asset-only context, but not from an asset/liability perspective. 3. There appears to be some evidence to support the argument that bond managers who have the latitude to operate in larger foreign bond markets may be able to capitalize on the greater breadth of these markets to increase value-added without taking on undue active risk. Implementation as Opportunistic Investment or Policy Allocation? There are two approaches an institutional investor can take to the inclusion of foreign bonds in their portfolio: opportunistic investment or policy allocation. The latter involves a decision by the investor to carve out and allocate a part of the portfolio to foreign bonds on a long-term basis. As the name suggests, opportunistic investment is done on a short-term, security-by-security basis as potential value-added opportunities arise; due to the rapid reaction time necessary to capture such opportunities, opportunistic investment is normally a decision made by the portfolio investment manager (assuming permission has been granted by the institutional investor). We used our model to determine which route institutional investors should take to include foreign bonds in their portfolio. In the asset-only context, we found that foreign bonds should comprise more than 50% of the total bond allocation, as they have the same expected return as domestic bonds, but with lower volatility and a lower correlation to Canadian equities. We also found that the minimum risk portfolio should be mostly in cash as it provides the lowest achievable return volatility. Chart 5: Asset-only Context Efficient frontier portfolios feature a significant allocation to foreign bonds 6 However, in the asset/liability context, we found that Allocation 100% 80% 60% 40% 20% 0% Expected return MRP 20% equity 50% equity 70% equity Portfolio 3.1% 4.4% 5.9% 6.6% Increasing risk Total equities Foreign bonds Canadian real return bonds Canadian long term bonds Canadian mid term bonds Cash foreign bonds should completely disappear from the portfolio because Canadian bonds provide a better match for Canadian-based liabilities. Foreign bonds have no expected additional return but their presence increases the asset/liability mismatch risk. In this context, the minimum risk portfolio comprises Canadian bonds, primarily long nominal bonds and real return bonds. 5 For complete detail of our analysis of foreign bonds, please refer to our upcoming paper entitled Should Canadian Pension Plans Invest in Foreign Bonds?. 6 Chart 5 shows the same efficient frontier portfolios shown in Chart 2, but provides more detail on the composition of the bond allocations in those portfolios. 6 Phillips, Hager & North Investment Management Ltd.

10 Chart 6: Asset/Liability Context Efficient frontier portfolios are void of foreign bonds 7 Allocation 100% 80% 60% 40% 20% 0% Expected return MRP 20% equity 50% equity 70% equity Portfolio 4.3% 5.2% 6.2% 6.7% Increasing risk Total equities Foreign bonds Canadian real return bonds Canadian long term bonds Canadian mid term bonds Cash Implementation Key Findings 1. Most pension plans and many endowment funds are asset/liability investors, thus an allocation to foreign bonds on a policy basis is probably unattractive. 2. Opportunistic investment in foreign bonds is attractive, however. It provides access to additional alpha opportunities, while avoiding systematic exposure to unwanted foreign interest rate and currency risks. 3. We recommend that sponsors retain a Canadian bond benchmark, but allow substantial investment in the Canadian-dollar issues by foreign issuers (called Maple bonds) and allow more limited investments in hedged and unhedged foreign currency issues. Having addressed the approach, size and composition of an allocation to foreign securities, we turned our model to the question of whether or not to leave the portfolio exposed to currency risk. To Hedge or Not to Hedge? Hedging is the decision to neutralize the foreign currency exposure of an investment portfolio. The 40% rise in the value of the Canadian dollar against its U.S. counterpart over the past four years, coupled with the 20% surge in 2005 of our currency against the yen and the Euro, has made the question of hedging a dominant one for Canadian investors in foreign markets. The strength of our dollar has consumed a considerable portion of unhedged gains on international portfolios. The elimination of the foreign property rule enhanced the need to examine the questions of whether to hedge or not, and if so, what the 7 Chart 6 shows the same efficient frontier portfolios shown in Chart 3, but provides more detail on the composition of the bond allocations in those portfolios. 8 The currency hedge ratio is defined as the percentage of the foreign asset portfolio that is hedged. optimal hedging ratios 8 might be. In the JPMorgan survey referenced at the outset, 9 of 104 Canadian pension plans that participated, over half expected the value of the Canadian dollar to rise over the next three to five years; 41% said they employ currency management practices today; and 1/3 plan to review these practices if foreign investments rise as a percentage of their total portfolio. There are strong arguments for and against hedging. In general, those arguing for hedging are of the view that currencies are inherently difficult to predict and do not add expected return, but can add to the risk of a portfolio. Those arguing against hedging believe that over the long term, currencies are mean reverting and hedging costs can quickly add up, so why bother? The two sides of this debate have been presented in two seminal academic papers. In 1988, André Perold and Evan Schulman 10 argued that portfolios should be fully hedged, as foreign currency risk was not offset by a corresponding increase in return. They set forth the view that because currency has a zero long-term expected return, currency risk can be removed from the portfolio without causing a drop in the expected long-term return, something they referred to as a free lunch. Kenneth Froot 11 of Harvard University presented the other side of the debate in a paper published in Froot argued that over long time frames, real exchange rates are mean reverting and as a result, investors should leave their portfolios unhedged, thereby avoiding transactions costs that would negatively impact returns. While this debate has yet to be resolved, it is important for trustees and other fiduciaries to be aware of the issues surrounding currency management. The objectives of currency management are the same as for an investment plan to maximize risk-adjusted returns. If the portfolio is hedged, the return will be equal to the return on the foreign security plus (or minus) the return on the hedging instrument. If it is not, the additional volatility of the foreign currency exposure may increase or decrease the overall volatility of the portfolio, depending on the 9 Source: Globe & Mail, January 18, Perold, A. and Schulman, E. (1988) The Free Lunch in Currency Hedging: Implications for Investment Policy and Performance Standards, Financial Analysts Journal, May-June. 11 Froot, K. (1993) Currency Hedging Over Long Horizons, National Bureau of Economic Research. Foreign Content in Institutional Portfolios: How Much is Enough and Should it be Hedged? 7

11 correlation between the currency movements and the asset class returns. There are three broad approaches to managing currency strategic, tactical, and alpha transport. Passive or strategic hedging programs usually establish a static, uniform hedge ratio across all of the portfolio s foreign currency exposure. For example, a plan sponsor may elect to employ a 50% hedge ratio. A drawback to this approach is that the optimal hedge ratio may vary over time and it reduces the benefits from diversification of foreign currency exposure for Canadian investors. Tactical hedging actively adjusts the hedge ratio over time as economic conditions change. A limiting factor in this approach is that managers tend to be constrained to hedging only benchmark currency exposures. For example, if the benchmark for global equities is MSCI World ex-canada Index, then the benchmark exposure for Australia was 2.52% as of January If the manager wanted to take a large position on the Australian dollar, this would not be possible because of its small weight in the benchmark. Alpha transport involves the creation of a diversified portfolio of currency positions that are not constrained by or necessarily even influenced by the underlying assets. The portfolio contains long and short positions (by definition), usually enacted through the use of derivatives. While this approach may be of interest as a return-enhancement technique, it is not a currency hedging technique; therefore, we will not address this approach further in this research paper. As noted earlier, those arguing against hedging suggest that over the long run, currencies are mean reverting, so why bother hedging? As well, because currencies do not pay dividends or interest and do not have a claim on underlying assets they suggest that the expected returns from currency are zero. 12 However, how long is the long run? As Table 3 shows, U.S. equity returns in Canadian dollars and in U.S. dollar terms from 1989 to 2005 were almost exactly equal over this 16-year period currency was a neutral factor. This changes though when we look at the 4-year period ending in In U.S. dollar terms, U.S. equity returns 12 Froot, K. (1993) Currency Hedging Over Long Horizons, National Bureau were almost 50% but in Canadian dollar terms, they tumble to just 10%. U.S. $ C $ U.S. equity returns 1989 to % 398.3% U.S. equity returns 2002 to % 10.4% Table 3: Are long-term expected returns from currencies = zero? In the short-term, currencies can be extremely volatile and notoriously difficult to predict and can significantly alter the risk profile of an investment portfolio. In the past few years, Canadian investors have been painfully aware of the impact of currency on their U.S. dollar investments. Chart 7 looks at the annual percentage change in the U.S. dollar value of the Canadian dollar over the past 25 years. The last three years have been a very unusual period in which the Canadian dollar has moved from just over 60 cents U.S. to over 90 cents, % year-over-year providing a major headwind to foreign asset returns for Canadian investors. Chart 7: The strength of the C$ versus the US$ has been a headwind for Canadian investors in the past four years * 2006 to June 30 The volatility of the Canadian dollar against non-north American currencies is even greater. Chart 8 examines the performance of the Canadian dollar against a basket of major non-north American currencies including the U.K. pound, the Euro and the yen. Annual changes in the value of the Canadian dollar against this basket have swung from -8% in 1990 to +15% in 1981, a large and unpredictable source of volatility. of Economic Research. 8 Phillips, Hager & North Investment Management Ltd.

12 Chart 8: The C$ versus foreign currencies* has been a large and unpredictable source of volatility for Canadian investors % year-over-year * a basket of currencies comprising: 25% yen, 25% pound, 50% euro (synthetic euro prior to 1999) ** 2006 to June 30 The Pros and Cons of Hedging 100% Unhedged Traditionally, many Canadian investors have approached currency management from the position of 100% unhedged. The benefits associated with not hedging are that both Canadian and foreign equity returns are negatively correlated with currency returns, so leaving the portfolio unhedged offers additional diversification. In addition, it is inexpensive to do nothing that is, no hedging transaction costs are incurred. The cons are that there is a mismatch between foreign assets and Canadian dollar liabilities and a negative or poor correlation to changes in the liabilities. Also, by not hedging, investors are implicitly positioning their portfolios against a rise in the value of the Canadian dollar some argue that not hedging is equivalent to taking a market position. In our view, taking positions based on expected short-term movements in currencies is the lowest quality alpha decision an investor can make. Investors may simply believe the expected return on currencies is zero. In that case, avoidance of hedging costs would provide a higher expected rate of return. Whether the risk-adjusted return of an unhedged portfolio were higher or lower would depend on correlation and volatility assumptions. 100% Hedged The pros of hedging a portfolio fully are that it completely insulates the portfolio from currency movements and from their negative correlation to changes in liabilities. The cons of 100% hedging are two-fold. Hedging the entire portfolio can be costly, thereby eliminating any potential benefit from currency diversification. In addition, despite slightly negative liability correlation, the more pronounced negative correlation of currencies with equities might still cause some currency exposure to be risk reducing in an asset-liability context. 50% Hedged Many institutional investors have opted for (or defaulted to) the so-called path of least regret the 50% hedge ratio. The pros to this approach are that the portfolio has partial downside protection and partial upside benefits from currency movements. The cons are that the transaction costs can be a factor and you will always be half wrong. Model Results for Hedging Each pension plan is unique a more customized approach may be required than either the fully hedged or fully unhedged portfolios. Chart 9 looks at the efficient frontier portfolio allocations in an asset-only context. The model hedges only at low levels of risk tolerance, defined as a low policy allocation to equities. As risk tolerance increases, the hedging allocation gradually disappears. Chart 9: Asset-only Context Efficient frontier portfolios hedge only at low levels of risk 13 Allocation 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% MRP 20% equity 50% equity 70% equity Portfolio Foreign equities (unhedged) Foreign equities (hedged) Canadian equities Foreign bonds (unhedged) Foreign bonds (hedged) Canadian bonds Cash Adding in liabilities significantly changes the results as seen in Chart 10. In an asset/liability context, the model does not advise hedging any portfolios, regardless of risk 13 Chart 9 shows the same efficient frontier portfolios shown in Chart 2, but provides more detail on the level of currency hedging in those portfolios. Foreign Content in Institutional Portfolios: How Much is Enough and Should it be Hedged? 9

13 tolerance. One point to bear in mind when considering this outcome is that none of the asset/liability portfolios contain any allocations to foreign bonds. Therefore, any potential foreign currency exposure comes only from the portfolio s allocation to foreign equities. Foreign equity allocations are unhedged, as currency returns diversify equity returns; because foreign equities are such a poor match to Canadian dollar liabilities in the first place, the volatility-reducing diversification effect of currency exposure overcomes any slight increase in liability mismatch it might create. Chart 10: Asset/Liability Context Efficient frontier portfolios are not hedged 14 Allocation 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% MRP 20% equity 50% equity 70% equity Portfolio Foreign equities (unhedged) Foreign equities (hedged) Canadian equities Foreign bonds (unhedged) Foreign bonds (hedged) Canadian bonds Cash Hedging Key Findings 1. In the asset-only context, hedging is prescribed by our model only at low levels of risk tolerance. 2. In the asset/liability context, hedging is not prescribed by our model. 3. Despite our model results, you may still wish to hedge depending on your risk tolerance and timeframe. Despite Model Results, You May Still Wish to Hedge Our analysis suggests little justification for hedging global stocks or bonds over the long run. However, investors may wish to adopt a different strategy depending on their time frame and their tolerance for risk. Many investors adopt the path of least regret, a static 50% hedge ratio. Others may wish to consider hedging the entire portfolio if their appetite for risk is extremely low. However, low risk tolerance does not necessarily mean 100% hedging the actual percentage depends on the correlation of currencies with other assets and liabilities. The Trustee Checklist at the end of this article includes some questions which may help you in your hedging deliberation. Implementation Issues If you do contemplate hedging some or all of your portfolio, here are a few things to consider. Monitoring costs: Both fees to manage the hedge program (i.e., transaction costs) and the cost of the hedge itself. The cost of the hedge itself is a function of the spread in interest rates between the base country (Canada) and the country of the hedged currency. Sometimes you get paid to hedge (when interest rates in the base country are higher than in foreign countries), and sometimes you would have to pay. At the date of this report, you would be paid to hedge the yen but would have to pay to hedge the U.S. dollar. Checking the Statement of Investment Policies & Procedures: Does it allow your manager to hedge? Some clients have asked us to hedge, but do not allow the use of derivatives. Forward contracts employed in hedging are considered to be derivatives. Monitoring the hedge: Perfect hedging is rarely possible. If your portfolio has a market value of $1 million when you hedge, and then subsequently increases in value to $1.1 million, the incremental $100,000 is not hedged. Thus, changes in portfolio value must be monitored if a hedging policy is pursued. Not all currencies can be hedged easily: While you can cheaply and readily hedge the Euro against the Canadian dollar, the Polish Zloty may be another matter. How to Hedge Currency Risk If you decide to hedge, there are various strategies you can employ using either OTC (over-the-counter) market currency forwards, options, exchange-traded futures or options on futures. A forward contract locks in the price at which an entity can buy or sell currency at a future date. A forward hedge is created by selling the non-domestic currency and buying the domestic currency forward for a specified period up to five years, i.e., at a future date, at a specified price and in a specified quantity. For example, if a person were buying a home in Florida in three months time, they could hedge the transaction by entering into a forward contract that would commit them to buy a set amount of U.S. dollars ($500,000 for example) at a fixed exchange rate for example C$ Chart 10 shows the same efficient frontier portfolios shown in Chart 3, but provides more detail on the level of currency hedging in those portfolios. 10 Phillips, Hager & North Investment Management Ltd.

14 The risk is that the Canadian dollar will appreciate over the 3-month time frame, let s say to $1.05. In that case, the purchaser would have been better off to have left the transaction unhedged. Of course, the spot rate could move the other way as well, say to $1.50, in which case the buyer would be pleased with the purchase of the insurance policy. For cost reasons, hedging is preferred using forwards rather than options. To reiterate a point made earlier, an investor could also elect to hire a currency overlay manager to actively manage currency exposure irrespective of the exposure of the underlying portfolio. This is a type of portable alpha strategy and is beyond the scope of this paper. Having decided that foreign content of one kind or another is desirable, our focus shifted to how to manage the currency risk associated with foreign investments. Our analysis in this regard suggests that in an asset-only context, significant hedging should be employed only by investors with very low tolerance for risk. Factoring in liabilities, the model shuns hedging all together. However, put into practical terms, you may find you want to hedge some or all of your portfolio depending on your risk tolerance, your time frame, and your investment beliefs with regard to currencies. Conclusion Our analysis confirmed that there is no single answer to the question of optimal portfolio allocation to foreign content for defined benefit plans, endowments and foundations as each entity has its own unique characteristics with regard to risk tolerance, funding status and other considerations. We concluded, however, that the addition of global equities can be beneficial in either an asset-only context or when liabilities are incorporated in the analysis, with the relative allocation to Canadian versus global equities fairly constant at around 50%/50%. Global bonds were found to be attractive for investors in an asset-only context but not for investors interested in matching liabilities. The model recommended, for example, that a medium risk investor (defined as a portfolio with a benchmark or policy allocation to equities of 50%), should have an allocation to foreign content of about 55% in an asset-only context. Of this, about 30% would be allocated to foreign bonds. In an asset/liability context, the foreign allocation would fall to just under 25%, all in equities. Thus, while the optimal recommended allocation to foreign bonds is zero on a policy basis i.e., there should be no allocation to foreign bonds within the portfolio s benchmark foreign bonds could be added to the portfolio opportunistically at the discretion of the investment manager. Note that the investment policy statement may have to be amended to permit this. Foreign Content in Institutional Portfolios: How Much is Enough and Should it be Hedged? 11