# Archived - Evaluation of the Exchange Fund Account

Archived information

Archived information is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please contact us to request a format other than those available.

July 5, 2006

Many of the graphics provided by the consultant in this report do not work in HTML, and in some cases they actually distort the document. It is highly recommended that the PDF version provided here be used for both viewing and printing.

The French-language version of this document is posted as provided by the contractor, and may contain graphs and charts that have not been translated from the English.

(The French-language version of this document is posted as provided by the contractor, and may contain graphs and charts that have not been translated from the English.)

### Proposal to Evaluate the Exchange Fund Account

Executive Summary

Objectives and Scope of the Project

Current Structure & Management of Canada's International Reserves Programme

Evaluation and Recommendations

## Executive Summary

Fischer Francis Trees and Watts (FFTW) has been engaged as a consultant by the Department of Finance of the Government of Canada (Canada) to evaluate the management of the country's foreign exchange reserves. With over thirty years of experience managing foreign exchange reserves and advising Central Banks on reserve management, FFTW has the skills and experience to assess the current practices and objectives established for the management of the Exchange Fund Account (EFA). Our assessment confirms that the EFA is being managed prudently, effectively and with due regard for the three key objectives, all of which are common and key parameters for most of the world's Central Banks:

• Liquidity
• Capital preservation
• Return enhancement

While having similar objectives to many of the world's Central Banks, Canada differs in so far as the EFA is funded through borrowing, and the management of the assets is matched against the cost of funding with the objective of immunizing the primary market risks - interest rate risk and currency risk. Both the risk and the return of the EFA are evaluated relative to the cost and structure of the liabilities. By realising a moderate positive return over the cost of funding the reserves, Canada has met all stated key objectives, and despite a conservative approach to managing the EFA, has also managed to achieve a contribution to the public purse. The recommendations outlined in this evaluation focus on ways in which Canada could enhance return, should it wish in the future to re-consider its current low-risk tolerance policy. These recommendations focus on two areas: seeking return by reducing liquidity and seeking return by adding market risk, subject to (i) efforts to ensure that there will be no drain on the public purse, (ii) an adherence to the stated key objectives as listed above and (iii) maintaining the ability to explicitly hedge unwanted risks.  Areas identified that Canada may wish to consider and that merit further evaluation are:

1. Re-evaluating the currency allocation to ensure that the designated neutral allocation is indeed a neutral position
2. Increase the number of eligible currencies to allow for  greater diversification and a reduction in the overall risk of the EFA
3. Broaden the list of eligible countries to, again, allow for greater diversification which serves to reduce the overall risk of the EFA
4. Consider allowing / increasing unmatched duration (interest rate) exposure to increase the potential for increased return.
5. Evaluate the impact of yield curve exposure that would result from unmatched duration exposure.
6. Consider adding additional sectors of the fixed income market that provide incremental yield relative to existing eligible investments.

Given the conservative yet professional approach to managing the EFA, all recommendations outlined in this evaluation should be regarded as suggestions with the purpose of enhancing return, subject to an increase in risk, and should not be construed as criticisms of current policies and practices currently undertaken.

### Objectives

The objective of this project is for Fischer Francis Trees and Watts, Inc. (FFTW) to evaluate the existing structure of the Exchange Fund Account (EFA) which comprises the foreign exchange reserves owned by Her Majesty in Right of Canada (the Government of Canada). In the course of the evaluation, FFTW will assess the investment practices to ensure that they are compatible with the conservative nature of the Fund and in line with practices adopted by others in the Central Bank community. FFTW will also make recommendations regarding both the investment structure of the EFA and the practices surrounding the management of the EFA given our understanding of the need to limit risk within the reserve assets in an appropriate way for a country of Canada's standing in terms of the size of foreign exchange reserves and the likely role of, and use of, these reserves. We pre-suppose a desire to optimize the use of such risk but understand that enhancing return, while important, is not a significant objective of the EFA and the appetite to increase risk to achieve higher returns is therefore quite limited. The primary factors behind the management of the EFA are clearly the need to preserve capital and to ensure liquidity, both of which are key and common parameters throughout the Central Bank community worldwide.

### Scope of the Project

The project covers all aspects of the management of the asset side of the balance sheet and the EFA will be assessed in a portfolio context, as a portfolio of assets. The funded nature of the EFA in an asset/liability matched framework, while noted and an important factor in the structure of the reserves, does not fall within the scope of this evaluation. This framework does, however, allow the Bank of Canada and the Government of Canada to accurately assess the true cost of holding reserves which is a valuable metric. Other areas that fall outside the scope of the project are (i) the size or targeted size of the foreign exchange reserves (ii) the rationale behind holding foreign exchange reserves, both of which are policy issues for Canada, and (iii) the governance structure surrounding the management of the EFA, which is a structural but not an investment issue. Following the initial evaluation, during which detailed information was provided by representatives from the Bank of Canada and the Department of Finance, the following areas came to light as requiring comment:

• The portfolio structure and strata definitions
• Limited nature of current sector allocation
• Potential for unmatched duration exposure
• Future management of yield curve exposure
• Neutral currency allocation

The issues raised and suggestions on each of the above will be detailed in this evaluation.

### The Current Structure and Management of the Exchange Fund Account

An initial meeting in Ottawa with members of the Exchange Fund Account (EFA) management team allowed FFTW to better understand the construction and objectives of the EFA, and also provided some insight into the history behind the current structure. Information gathered during this meeting and subsequent discussions with members of the EFA team in conjunction with publicly available data have served as the reference sources for this evaluation. FFTW's goal is to evaluate the appropriateness of the current reserve management structure given the fund's objectives and in comparison to others with similar objectives within the Central Bank community.

Within this evaluation, FFTW has taken into consideration the unique relationship between the Bank of Canada and the Department of Finance to the extent that it is relevant to the management of the assets in the EFA. The current structure of the reserve management program is unusual in that the Bank of Canada and Department of Finance share responsibility for the strategic planning and operational management of the EFA, a structure that is not common within the Central Bank community. Given that most, but not all, countries separate the management of the assets and the management of the liabilities, the ALM approach adopted by the Bank of Canada and the Department of Finance lends itself to a combined approach, overseen and managed jointly by the two entities. While the liabilities are not part of the evaluation, it should be noted that changes to the liability structure or to the method of hedging liabilities will likely impact the asset side of the balance sheet and hence the management of the EFA.

Oversight of the EFA is assumed by the Fund Management Committee (FMC), comprised of senior representatives of both the Bank of Canada and the Department of Finance. Risk within the EFA is overseen by the Risk Committee, which is supported by the Financial Risk Office of the Bank of Canada. The Risk Committee is responsible for monitoring and reporting on performance and positions within the EFA. It will be the responsibility of FFTW's project team to review and raise issues surrounding the management of the EFA and to provide recommendations on investment-related aspects of the current reserve management structure that we believe would benefit from further analysis and consideration for possible changes.

The Department of Finance is responsible for issuing debt on behalf of the Government of Canada, the proceeds of which comprise the major part of the Government's foreign exchange reserves. The other components of the reserves are Gold and a small SDR component, both of which are outside the scope of this evaluation. The small allocation to SDR, a legacy from the Bretton Woods days and from IMF payments, is considered operational reserves and is managed separately from the rest of the reserves. The SDR component of the portfolio is funded with US Dollar liabilities so, as the US Dollar is a sizeable component of the SDR basket, the US Dollar element of the SDR is fully currency hedged. It is therefore only the non-US Dollar component of the SDR that is unhedged and therefore adds some volatility to the aggregate reserves.

As mentioned above, the relationship between the Bank of Canada and the Department of Finance is somewhat different from the working relationship between most Central Banks and Ministries of Finance. Whereas the Department of Finance directly funds the Exchange Fund Account and the assets of the EFA are structured to match the liabilities, most countries completely separate the liability and asset functions. Assets are generally managed by the Central Bank reserve management team whereas the liabilities are managed either by the Ministry of Finance or by a Debt Management Agency specifically established for the purpose of funding. There are exceptions however, and in some countries the Central Bank and not the Government manages both the management of the reserve assets and also the country's debt issuance. A prominent exception, to the general experience is the Bank of England which evaluates the assets in the context of the liabilities and, somewhat similar to Canada, manages the foreign exchange liabilities and assets together.

The Bank of Canada and the Government of Canada look at the outstanding debt and the foreign exchange reserves in a 'whole portfolio' context which has implications to both funding and investments. As a result, the objectives of the EFA in certain areas may be different from those of other more traditional Central Banks who regard, or at the very least evaluate, the investment of reserves as unrelated to the structure of the debt. However, there are many areas of similarity between Canada and the management of the majority of Central Bank foreign exchange reserves, the most notable of which is the conservative and risk averse nature of the EFA and the focus on liquidity. An area of difference is the relatively low appetite for increasing risk in order to seek incremental returns and this is an area that will be explored in this evaluation. The size of reserves and the infrequent use of reserves for intervention purposes is a factor that influences reserve management strategies and practices across many Central Banks. Lowering the probability of use of funds for intervention effectively leads to a lengthening of the investment time horizon and allows greater scope to potentially increase the risk parameters while still retaining a prudent approach to reserve management. One of the outcomes of this evaluation is to question the level of risk taken and to explore whether risk may be increased, and if so, by how much, in order to add to returns while maintaining an acceptable level of downside protection.

The EFA has set currency target levels around which the Fund's assets will fluctuate very little, so the segregation of assets by denominated currency with assigned weightings should not be considered an impediment to the management of the program. The majority of the non-US exposure is in Euro with a small allocation to Japanese Yen. Although all intervention in the past has been executed using the US Dollar, the US Dollar exposure is less than 100% to allow for the Euro's rise as a reserve currency - a move that has occurred with a number of Central Banks, and not just those linked to the Euro. Many Asian Central Banks have gradually increased their Euro holdings over the last 5 years, as have Central Banks in non-Euro denominated European countries. More recently, some Middle Eastern Central Banks, despite having currencies pegged to the US Dollar, have gradually increased the allocation to the Euro versus the US Dollar. The above increased allocations to the Euro have occurred largely for two reasons (i) to increase allocations to the Euro in recognition of its potential role as a major reserve currency, (ii) for investment and return reasons recognizing the poor fiscal and current account fundamentals within the US. The former has led to more strategic, longer term (benchmark) allocation shifts towards the Euro whereas the latter can be categorized as tactical allocations and do not impact the neutral benchmark allocations to the US Dollar and Euro. The Japanese Yen remains a minor reserve currency, largely given the opportunity cost of holding Yen in income terms relative to the higher yielding US Dollar and Euro. The majority of the assets of the EFA are denominated in these three reserve currencies: the Euro, US Dollar and with a small allocation to Japanese Yen.

Generally, the currency allocation within a Central Bank's reserves is determined as a core strategic decision and is taken at the senior management level. In some instances, the decision is taken at the very highest level within the Central Bank, i.e. the Governor's office, and in many cases, if not decided at the highest level, it is approved at the highest level. However, the management of the currency exposures around the strategic allocation remains within the reserves management department and the tolerances and degree of active management is strictly regarded as an investment related decision. The currency allocation within the EFA, is determined by the Fund Management Committee and there is currently very little tolerance for currency risk and currency exposure outside these currency targets. Given the importance of the neutral currency allocation decision and the impact on returns within the EFA, we suggest that the process behind this decision would merit a review in the context of a more analytical framework. Further, the parameters determining the degree of deviation around the core strategic decision should also be subject to review.

The assets held in the EFA are invested primarily in money market instruments and government bonds denominated in the currencies of the predetermined currency allocation. Although the current investment universe appears to satisfy many of Canada's investment requirements, we will be suggesting further analysis into the possibility of the inclusion of non-G3 countries into the investment program to broaden the investment universe, increase the opportunity set, introduce an element of diversification and thereby potentially lower the aggregate risk of the EFA.

The purpose of the EFA is to provide foreign-currency liquidity and sufficient liquidity to allow for effective intervention in the foreign exchange market should the need arise, while respecting the need to preserve capital and subject to these two parameters, to obtain a return above the funding cost. This last factor is currently only a minor focus and return enhancement is not a prominent objective for the EFA.

The EFA assets are classified into two tiers: a Liquidity tier and an Investment tier. The liquidity tier is mostly invested in highly rated US Dollar denominated assets while the investment tier is comprised of a combination of high quality instruments denominated in US Dollars, Euro and Yen. The liquidity tier is invested mostly in US Treasuries as they are generally acknowledged to be the most liquid securities in the market. Although we agree that US Treasuries are indeed considered the most liquid securities available, we would recommend further review as to whether there is a real need to maintain the highest level of liquidity and this will be discussed later in this evaluation.

On another note, it is our understanding that Canada defines liquidity primarily as asset class driven, and to a lesser extent as term driven, whereas other central banks typically define liquidity as a combination of the ability to convert invested reserves into cash at short notice and the potential exposure to loss in the course of converting those assets into cash in the event that the assets are required for immediate intervention purposes. The liquidity component of reserve assets is generally in the base currency of the reserves and defined by duration, not by asset class or credit parameters. Accordingly, as interest rate risk is a concern for most reserve managers, any portfolio defined as a liquidity portfolio should have a short duration, generally no greater than three months, therefore protecting this component of the reserves from mark-to-market losses on an ongoing basis, losses which could become realized should an unexpected need to liquidate assets arise.

A second aspect of liquidity, separate from the inherent risk characteristics of the instruments (duration, credit quality, currency) is the ability to liquidate. Clearly the government asset classes identified by the Canada and in which the EFA is primarily invested are significantly more liquid than non-government 'spread' sectors in times of stress, which supports the prudence of Canada's sector selection decision. However there is an opportunity cost and the question should be raised as to whether there is a need for all of the EFA to be invested in these highly liquid but lower yielding asset classes that will provide the liquidity benefit in times of stress.

The investment tier of the EFA represents approximately 50% of the fund's assets and is invested in assets denominated in US Dollars, Euro and Yen.. This is the result of (i) an explicit target for the US Dollar allocation and (ii) attractive yield opportunities in the Euro market in recent years. However, this allocation should be closely reviewed as the Federal Reserve approaches the end of the current rate hiking cycle and the European Central Bank embarks on a rate hiking campaign of its own. This scenario has the potential to significantly alter the current investment landscape. As such, the target allocation for US Dollars should also be reviewed as already stated earlier in the evaluation.

Following the review of the composition and liquidity of the EFA, we would like to address the return of the EFA in relation to the stated guidelines. The net return over funding earned on the EFA in 2005 was approximately 9 basis points. Achieving an excess return over funding cost, while rigorously adhering to the liquidity and capital preservation requirements, is considered a success for the program, particularly taking into account the minimal unmatched duration and currency risk parameters of the EFA. The team managing the EFA has both prudently and successfully achieved all of their objectives and this evaluation is able to categorically confirm that success. However, we believe that the current practice of matching assets to liabilities and passively managing the duration and currency exposures within the fund should be open to review given the potential for additional excess returns while still remaining true to the first two stated objectives of ensuring liquidity and preservation of capital.

The rationale for not aggressively seeking excess return in the portfolio is clear given the objectives of the EFA and Canada's ability to achieve excess return with very little risk given the favorable funding cost afforded to the Government of Canada. It has historically been a typical approach by Central Banks to limit their return demands and many have maintained a very conservative approach to managing assets as a result, albeit, in many cases, independent from any liability management. However, there has been a steady trend away from this practice over the past ten years as Central Banks attempt to find the balance between the need for liquidity and the potential for greater return. We will discuss this topic in greater detail in the benchmark analysis section of this evaluation. Although most Central Banks do not look at the liability side of the balance sheet, in other respects, their objectives are the same as those of Canada.

Timeframe is a determining parameter for risk, and it is our understanding that Canada monitors performance closely on a monthly, quarterly and yearly basis. While the objective of adding risk is to enhance returns, sometimes the additional risk detracts from returns and we determined through discussion that, similar to other Central Banks, Canada is likely to be uncomfortable with any loss of capital over a one year period, though losses over a quarter or over six months may be tolerable. Subject to these conditions, the issues open for consideration are (i) whether returns can be enhanced while maintaining the current level of risk; (ii) can the amount of risk in the portfolio be increased to enhance annual returns and (iii)if so, to what level can risk be increased in order to generate these incremental returns.

Risk, for the purposes of prudent reserve management, may be increased through the addition of unmatched interest risk (duration), currency or non-government sector exposure. We understand that Canada has researched the latter and has investigated diversifying away from the current eligible instruments, which in terms of longer terms assets are limited to government, sovereign and supranational credits. The analysis evaluated the benefits of including credit securities such as Asset Backed Securities (ABS) and Corporate Bonds, but concluded against broadening the investment parameters to make an allocation to these fixed income sectors for the time being. The decision against investing in US ABS securities was part due to the perception that the legal risks associated with these securities and in particular, the structure of their pools, is too high relative to the incremental return earned. In contrast, however, a decision has been made to broaden the eligible short term investment universe to include Commercial Paper and Certificates of Deposits in addition to repurchase agreements. It is interesting to note that this decision is in contrast to the allocation decisions of many Central Banks, which opt to include AAA-rated ABS, which represent fully collateralized credit exposure, and exclude Commercial Paper as Commercial Paper introduces the reserves to uncollateralized corporate credit exposure which exposes the reserves to both unsecured credit risk and to headline risk.

In summary, the goal of the Exchange Fund Account, which represents the majority of the international foreign exchange reserves, is to provide foreign currency liquidity for Canada and to aid in the management of the domestic currency in the foreign exchange market - a common objective for most international reserve managers. The current structure and management of the EFA consists of liquid foreign government, sovereign and supranational assets managed with emphasis on liquidity, capital preservation and, subject to the first two and to a lesser extent, incremental returns above funding. The operations of the fund are supported, on an asset and liability matched basis, by obligations of the Government of Canada. The purpose of matching the assets and liabilities of the Government of Canada is to reduce exposure to excess currency and duration risk while attempting to earn a net positive return over funding cost. Interest rate risk is managed through the ALM approach and is both monitored and measured using a Value at Risk methodology and by stress testing.

The purpose of the reserves is to be a pool of assets available for intervention should the need arise and the role of Canada is to manage those reserves in a prudent manner taking into account:

1. the nature of the assets as foreign exchange reserves
2. the source and cost of funding
3. the public role of the Bank and the Government of Canada
4. the conservative nature of official institutions
5. the need for return

### Evaluation and Recommendations

After reviewing the current structure of the EFA and the Fund's ability to accomplish its stated goals, this paper will now focus on evaluating the funds comparability to other Central Banks with similar objectives. For this review, we will not venture to stipulate the level of risk that we believe that Canada should be willing to take, but will instead provide a comparison to other Central Banks and will also offer some tools and ideas for Canada to analyze.

In conducting this evaluation, FFTW's first step was to identify the level of risk to which the EFA is currently exposed. As stated earlier, the investment objectives of the Government of Canada are similar to those that are typically stated by many Central Banks:

1. Liquidity
2. Capital preservation
3. Return enhancement subject to (i) and (ii)

This means that we can draw on our knowledge of the reserve management practices of other Central Banks to determine appropriate parameters, how much risk may be acceptable and where that risk should be allocated.

### Liquidity

As referred to earlier, there are two ways of looking at liquidity. Firstly, liquidity is that component of the reserves that can be used as the first source of cash in the event of the need to liquidate investments. In the current structure of the EFA, this would be defined as the Liquidity tranche of the reserves. The second definition of liquidity is marketability, or the ability to sell assets in size and within a short period of time without compromising the price. This would be reflected in the composition of the Investment tranche of the EFA.

#### Liquidity Tranche

It is very typical for reserves to be structured in layers that reflect the likely call on assets. The diagram on the following page reflects the typical structure of reserves for a Central Bank with sizeable reserves. The liquidity component is effectively cash on call and the investments should have a short maturity and be available for sale and settlement on same day or with one day's notice. In many cases the liquidity component does not have a specific benchmark as the purpose is not to match any term or investment structure but to have the reserve currency, in this case US Dollar cash, readily available. Where there is a benchmark, it will likely have a 3 month maturity or less and investments will be purchased purely for safety and availability. This means that liquidity reserves are in practice comprised of money market instruments such as Treasury bills, Agency discount notes, reverse repurchase agreements, time deposits and certificates of deposit. Commercial paper is generally excluded from this definition and only a few Central Banks invest in commercial paper, given a general aversion to taking unsecured credit risk. We note that the EFA guidelines have recently been broadened to include Commercial Paper as an eligible investment. This is something that we would discourage unless there are dedicated credit analysts reviewing investments, as rating is not sufficient by itself as a measure of credit worthiness. It is important to be able to evaluate the direction of the rating, or more specifically, the potential for deterioration in the rating and therefore in the credit so as to avoid anticipatory price deterioration. In the case of commercial paper, the risks are almost entirely default risks as opposed to price deterioration, given the short term nature of this market. The role of the internal analysts is to determine that the borrower is of superior credit quality (as also indicated by the quality ratings provided by the rating agencies) and has sufficient liquidity in the event of a credit event to be able to pay their maturing debt if access to re-funding resources is compromised. A liquidity squeeze caused by limited access to refunding is a risk of purchasing unsecured commercial paper, and is a risk that should be explicitly evaluated by internal credit analysts in the event that commercial paper is permitted in the investment universe.

With regard to other money market instruments traditionally eligible for inclusion in Liquidity tranches, time deposits should also come under some scrutiny. Paradoxically, although time deposits are generally regarded as Liquidity instruments and although they have little risk of loss (other than bank credit risk), deposits technically are not convertible into cash prior to maturity so, strictly speaking, have amongst the least liquidity of many fixed income instruments. Liquidity is therefore very limited in a portfolio of time deposits with an average maturity of anything longer than a week. Ironically this is typically the investment of first choice for the liquidity portfolios of many Central Banks. A further issue that is common within the liquidity component of reserves is the investment of cash on deposit with local banks. It is not unheard of for crises to occur within the domestic banking system that have an impact on the currency, thereby leading to a drawdown on reserves. We therefore, as a policy, would not advocate the investment of reserves in local banks as this subjects the reserves to the same type of risks that they are there to protect against. In the case of Canada, we note that this is provided for in the EFA policy guidelines which, prudently, explicitly prohibit investing any part of the EFA in the local Canadian banking system.

#### Marketability

Another way of looking at liquidity is to evaluate the ability to liquidate and, most critically, to liquidate in times of stress. An argument could be made that all traded securities are liquid under normal market conditions. However, market conditions are not always normal, and it is during those times of abnormal behavior that a security's true liquidity can be gauged. The most appropriate way to evaluate a security's liquidity is to determine the impact on the size of transactions that can be executed and evaluate any change in spread, relative to normal market conditions, that an unexpected market event can cause. From an economic perspective, most investors gauge liquidity in terms of how much money will have to be forgone should they be required to sell a security during an extraordinary event.

High quality government fixed income markets, which comprise the majority of the EFA, offer the highest level of liquidity in terms of marketability. Portfolios can be liquidated within a short period of time (the whole EFA can likely be liquidated in less than 3 days) and losses due to wide bid-offer spreads or price mark-downs for size or immediate settlement will be minimal. US Treasury and agency securities can normally settle T+1, but same day settlement is not unusual. Euro denominated Government bonds are slightly less liquid and although they can be executed in slightly smaller size in normal market conditions, it is our estimate that they can be liquidated in the same size as US Treasuries in conditions of market stress. One key difference, however, is that non-US government securities typically settle T+3, which means that in terms of accessibility of cash, they are less liquid. Finally, sovereign and supranational euro- and global bonds which are eligible instruments in the EFA and in the reserve assets of many Central Banks, while having safety in terms of their credit rating, are among the less liquid (or least liquid) of investment grade instruments. This should be a cautionary note against building sizeable positions in supranationals, although the practical experience has been that it is often harder to build positions than liquidate in sovereigns and supranationals as there is such great demand from Central Banks for this paper, that it often exceeds the available supply.

FFTW has conducted an internal survey (Table 1 below) amongst members of our investment team in order to evaluate the impact on spreads that an unexpected negative event is likely to have on different asset classes.

Table 1

It should be noted that the eligible securities within the EFA are clearly the most liquid in terms of size and bid-offer spreads, both in normal market conditions but particularly in extreme situations. It is evident that Canada has adopted a conservative but also a very prudent approach to market liquidity. A question that Canada may wish to address, however, is whether the highest level of liquidity is required for the whole of the EFA or whether there can be some compromise on a part of the reserves. Liquidity has a cost so it is definitely appropriate to maintain the highest level of liquidity in terms of marketability and maturity in the Liquidity Tranche. However, it may worth consideration to divide the Investment Tranche into two categories a most liquid and a slightly less liquid category, in order to generate a higher return on a portion of the EFA (see diagram above). The less liquid investments would still be very liquid, in the context of world markets, but would mean that the liquidity premium would not be forfeited on 100% of the EFA, when the likelihood of 100% drawdown over a short period of time is relatively small.

#### Capital Preservation

Capital preservation is a key parameter for every Central Bank. The responsibility of the reserve management department is to oversee the reserves and to ensure that they are available for use and that the amount does not shrink and can potentially grow. This requires prudence, conservatism, an understanding of the likely use of assets and an evaluation of the trade-off between protecting the value of the reserve assets and growing the size of the reserves through prudent investment. The assessment of what capital preservation means is very much dependent on both the time frame in which the investment of the reserves are being evaluated and whether the reserves are being viewed on an aggregate basis or whether each strategy/portfolio within the reserves is being reviewed on an individual basis. The following analysis explains the difference in these parameters, the appropriate risk level in each event and the type of benchmark that may be the most suited to each circumstance. There are two types of risk uniformly taken by Central Bank reserve managers, duration risk and non-base currency risk. Currency risk will always be a component by the very nature of foreign currency reserves, duration, however, may or may not.

One of the key purposes of a benchmark is to reflect the neutral risk profile of the pool of assets, in this case, of the EFA. The neutral risk profile, in turn should reflect the risk tolerance of the reserve manager. While "neutral risk profile" can have many interpretations, we generally illustrate this by stating that if the portfolio were to be indexed to the benchmark, the client should feel comfortable with the portfolio returns over the chosen investment horizon, irrespective of market conditions over that period. By portfolio we refer to those assets that may comprise the EFA and are neither hedging nor hedged by liabilities. Unlike the EFA, most Central Banks have a designated benchmark against which the assets are managed but, also unlike the EFA, as discussed before, most Central Banks separate the assets from the liabilities and take little or no note of the cost of the reserves. This means that, in order to have some sort of 'risk compass', most Central Banks choose a benchmark as a measure of neutral risk, against which to manage the reserves. However, taking into account our earlier definition of "neutral risk profile" and given the EFA's risk parameters, the Fund's benchmark in this instance could be considered to be zero and therefore risk free.

Despite the generally conservative nature of investments within the Central Bank community, the EFA's risk profile should be considered conservative by comparison. As will be discussed throughout the next couple of pages, we have witnessed a change in the way that many Central Banks allocate risk within international reserves portfolios as we see more Central Banks allocating duration and credit risk within their portfolios.

It is not uncommon for Central Banks to be hesitant about making changes to an established reserve management program, particularly when the objective is to increase the level of risk in order to potentially increment return within the portfolio. We would anticipate that Canada has similar concerns should the EFA's current risk level be subject to review in the future.

For the purpose of this evaluation, we have outlined the process undertaken by FFTW to determine the level of risk that we would consider appropriate for a Central Bank with Canada's objectives and will compare this with what we understand to be the norm within the Central Bank community. We have focused on historic extremes as a determinant of the range of potential outcomes. Generally we recommend evaluating market moves falling within two standard deviations of the historic mean, with a focus on the downside returns. We do acknowledge, however, that the last decade has presented a number of four (and greater) standard deviation events so chosen risk levels should always be stress tested for the extreme events. While no-one likes returns in down market periods, clients need to evaluate the impact of the different return outcomes to assess which are tolerable and, most importantly, which are intolerable.

In order to gravitate towards the optimal balance between risk and potential return given the individual requirements of most Central Banks, as outlined earlier, we have analyzed the history of various fixed income asset classes in a number of different ways. Chart 1 below shows US Yield Movements going back to 1980. The chart highlights the change in environment between the 1980's, where yields averaged 10.36% and volatility was more extreme, and the period from 1990 onwards, in which the average yield was 5.11% and there has been considerably less volatility.

Chart 1 US Yield Movements, January 1980 to December 2005

To assist in identifying the risks attached to duration, the table below provides the risk/return characteristics of US Treasury securities over the 26 year period ending December, 2005.

The table shows the risk and return history of the US Treasury market, broken down into maturity buckets. The Period End Duration column is the duration as of December 2005. It should be noted, however, that the duration of each index may have changed over the period under analysis as a result of changes in market issuance patterns over the course of the period under review. The mean return represents the average of the monthly rolling annual periods under review for 26 years. The volatility is the standard deviation of these rolling annual periods. The worst return represents the worst rolling annual return and it should be noted that it may not necessarily have taken place in a calendar year. Finally, the Sharpe ratio is calculated using 3 month LIBID as the risk free rate, as this is the most common investment parameter and benchmark for the liquidity component of Central Bank reserves.

Table 2

In reviewing the above data, we can make the following observations. First, if positive return is the major consideration (i.e. capital preservation), then history shows us that in this time frame the 1-3 year sector of the US Treasury market has had negative returns during only two annual periods (the annual periods ending Feb 2005 and March 2005, when starting yields were at record lows). This is important for two reasons:

1. The period under review includes three rising rate cycles (1994, 1999 and 2004/5) so for US Treasury returns, particularly in the short end of the market, it is a highly representative period of analysis.
2. For many Central Banks, as well as Canada, the need for capital preservation over a one year period is a key requirement of an investment strategy and the 1-3 year area of the yield curve is a very commonly chosen neutral position (benchmark) among the more conservative Central Banks throughout the world.

Table 3 below shows similar data to Table 2 above, but over the more representative (lower volatility) period since 1990. It should be noted that the volatility is lower, the mean returns are lower, but despite this, in the short end of the yield curve, the worst annual returns are unchanged.

Table 3

Periods of analysis: January 1990 to December 2005

The 1-3 year area of the curve, with a duration of around 1.75 years, therefore appears to be an appropriate maximum duration level for a risk-averse investor seeking a low probability of negative returns. It is one of the key duration targets within the Central Bank community and not only provides a high probability of capital preservation, but also from an investment perspective includes one of the most advantageous areas for beneficial yield curve roll down in a normal market environment.

The current asset-liability management policy in Canada requires that the duration of the assets in the EFA and the liabilities funding those assets be matched as close as possible. The objective is to reduce the amount of interest rate risk to which the reserves are exposed at any given time. Any deviation from a duration neutral position is, in most cases, the result of timing differences in the reinvestment of maturities although modest and specific deviations from a neutral position are permitted, based on the maturity of the asset being reinvested. This practice significantly deviates from that of most Central Banks in so far as most don't have the objective of completely eliminating duration risk and, in fact, actually aim to take and actively manage the duration to which the reserve program is exposed.

We would suggest that Canada consider reviewing the duration exposure within the current fund structure. If Canada had an appetite to increase duration risk as a means of incrementing return, the duration gap could be increased up to a level representative of the duration exposure suitable for many other Central Bank with similar stated objectives. As discussed above in our analysis of interest rate exposures, a typical duration gap could be up to 1.75yrs, which is the duration of a typical 1-3 year Central Bank benchmark. As can be seen in Table 3, a 1-3 year US Treasury benchmark only had 2 negative rolling one year return periods during the 16 years analyzed and the worst return during that period was negative 35 basis points at a point in time when aggregate yields were close to 1%. Two years of duration exposure has been a common target duration in the Central Bank community and is often the first step for Central Banks looking to extend the duration of reserve portfolios beyond money market parameters. In fact, FFTW is aware of at least 25 Central Banks around the world who have chosen a 1-3 year benchmark as an internal and/or external benchmark.

The concept of a benchmark will be different for the EFA compared to typical Central Bank reserve managers given that the assets are liability matched and therefore already have their own benchmark in a sense. A discussion point will center around the implementation issues inherent in assuming duration risk while operating within the liability matching framework, and will address the potential for, and the need to avoid, residual yield curve risk. Increasing duration in an otherwise liability-matched structure will necessitate some yield curve exposure in order to avoid leverage. We have debated several options for establishing a benchmark for the EFA. The first is to establish a benchmark structure to optimize the asset side while the hedging of liabilities in terms of duration and currency will be managed as an overlay, a structure which would necessitate the use of futures as eligible instruments in the guideline parameters. The second is to recognize the liabilities themselves as a type of benchmark and limit taking off-benchmark duration exposures to non-systematic positions or opportunistic, tactical duration exposures. The third approach is to assume that the benchmark duration is zero thereby viewing returns in a total return concept, allowing both overweight and underweight (long and short) positions using futures to manage the risk. This as a concept, can of course be extended to other risks, such as currency through an overlay and even, credit through swaps. A final alternative is to maintain the current duration matched structure but to introduce different risks, such as the addition of new asset classes to enhance the yield spread between the assets and the cost of funding (liabilities).

Each of the above structures has its own pros and cons. The first will require the use of derivatives, and specifically the use of interest rate futures and interest rate swaps to manage the assets as an overlay, a potential impediment to Canada as it is to many Central Banks. The issue with the second approach is that one of the key requirements of a benchmark is its long term strategic nature, so non-systematic duration risk assumes an alpha and not a beta approach to returns. We believe that in an unconstrained world, a combination of these two approaches is appropriate. The third option is one towards which the industry as a whole is turning as the concept of total return management becomes both more recognized and more popular. The separation of beta (benchmark or market returns) from alpha (excess returns) in a total return context allows the industry to assume a cash-like benchmark but still generate excess returns while specifically isolating and targeting the risks that they wish to assume. In the case of the EFA, the objective is to generate excess returns (alpha) over the funding cost subject to the ability to preserve capital and maintain liquidity needs.

Another way in which duration risk may be more effectively used, is to diversify the risks across multiple markets. Currently the duration risks offset liabilities in three currencies: US Dollars, Euro and Yen. The chart below illustrates the degree of correlation between the major bond markets over the last 25 years.

In addition, appendix 1 shows a correlation table of the major investment grade fixed income markets and although correlations are relatively high between many markets, they are still less than 1.0 which indicates that a higher degree of diversification is achievable by adopting a broader set of markets. An important point to note is that the decision to diversify market exposure can and should be a separate decision from the currency allocation decision, which will be addressed later.

The eligibility of an increasing number of countries in global benchmarks has been steadily evidenced as the smaller markets have developed and borrowing, and therefore liquidity within markets, has increased. This would be an area for Canada to analyse further. By way of comparison, many Central Banks that adopt a multi-currency approach in their reserves do so by adopting a published multi-currency benchmark which includes 11 or more countries. The Citigroup World Government Bond Index, the JP Morgan Global Bond Index and the Lehman Brothers Global Aggregate Index are all multi-country indices currently used by Central Banks, and are often custom-weighted to target specific country allocations or sub-components are used to target preferred duration risk, such as a WGBI 1-3 year benchmark. The risk lowering benefit of country diversification can be seen in Table 4 below.

Table 4

Finally, we will assess the currency breakdown of the EFA. The currency allocation is currently a combination of US Dollars, Euro and to a small extent Japanese. The currency exposure is determined as a neutral position and the allocation around that neutral position remain very limited. In assessing the decision process behind the currency breakdown of reserves, there appeared to be scope for further evaluation of the currency mix. The allocation of the non-base currencies represents the single largest contribution to risk in a reserve portfolio and is a decision that lends itself to a quantitative determination that we suggest includes:

1. An assessment of the correlation of each currency and combination of currencies relative to the home currency (the Canadian Dollar). This will intrinsically include the impact on the currency due to trade patterns as well as non-trade flows and influences.
2. An estimate of the likely use of a given currency for intervention purposes.
3. An evaluation of the expected long term returns within each currency.

We would suggest conducting this analysis using long term (10 and 20+ year) data histories to fully assess the long term historic relationships between the optimal currency mix for the reserves and the Canadian Dollar as the neutral currency decision is a long term strategic decision and one which we would expect and recommend be changed only infrequently. While the US Dollar, Euro and Japanese Yen are the main reserve currencies, we suggest that this analysis be broadened to include other currencies to allow a degree of diversification which can be expected to reduce the overall risk attributable to the currency mix. It should be noted, however, that the three eligible currencies are the most liquid in the world and that through introducing other less liquid currencies, depending upon the size of the intended neutral allocation, the liquidity requirement of the EFA may be compromised.

Finally, having evaluated other risks, we can turn to an evaluation of different asset classes within the fixed income universe as a means of generating incremental return

Return Enhancement

The challenge of managing a pool of assets against a zero or 'cash' neutral position is to identify strategies which outperform cash while not exposing the portfolio to capital depreciation during periods of rising yields. In our experience, the objectives of low volatility or enhanced cash investment returns may be achieved by seeking incremental yield in a diversified portfolio of low volatility assets, thereby achieving higher returns over cash while limiting additional risk. Historically, returns achieved have typically paid the investor well for the additional risk taken, often in the short term (1 year) and almost always in the long term (over a cycle). The chart below summarises the strategies that are generally adopted to achieve different objectives.

### Objective

 Maintain Liquidity ê Maximise Returns ê Liquidity: High Liquidity Lower Liquidity Risk Level: Low Risk Higher Risk Duration: 3 months 1.7 years 5.0 years Credit: Government Agency, sovereign Corporates Credit: AAA rating BBB Rating Structure: Bullets ABS, CMO's Mortgages Country: US Global Hedged Global Unhedged Portfolio: Money Market Short-Intermediate Broad Market Benchmark: LIBOR 1-3 Yr US Treasury Global Aggregate

This section will focus upon the left side of the chart and will examine where and how far to move to the right to achieve excess returns while not losing sight of the short term (annual) capital preservation needs.

Return enhancement for conservative Central Banks can be divided into two categories: duration and sector. Currency may also be used for return enhancement purposes, but within reserve management, that tends to be for the more aggressive Central Banks. The more aggressive Central Bank investors, with some exceptions, are either those who

1. have little expectation of cash needs for intervention purposes such as the national reserves of those Central Banks which form the European Monetary Union; or
2. Central Banks which have sizeable reserves and therefore the Investment and Excess Wealth component may be invested more aggressively as the drawdown for intervention needs is less likely given the sizeable cushion invested in lower volatility and more liquid components. These would be categorized in the top two tiers in the diagram below.

While return enhancement is not a key objective of the EFA, the potential to increment return at low levels of risk should not be ignored and is an area that we recommend Canada should consider evaluating. The opportunity cost of holding the most liquid government securities translates into forfeited income and while the purpose of reserves is not to make money, it is becoming increasingly important to many Central Banks to enhance the return on reserves. Driven by the search to find yield, Central Banks have continued to expand the universe of acceptable asset classes included in their reserve management programs. Various surveys show that there is a steady move away from the traditionally "safe" asset classes towards highly rated, liquid but higher yielding investment opportunities. The RBS Reserve Management Trends - 2006 survey found that over the past year Central Banks are increasing their allocation to both "riskier" and "new" assets. In response to the question "Which of the following 'new' asset classes is your Central Bank currently invested in?", over a third of survey respondents reported investing in 'A' rated government securities. Almost half of the Central Banks surveyed are introducing "new" asset classes and one quarter of those surveyed are introducing structured securities, namely ABS and MBS.

Table 5

 Asset Class Number of Central Banks Governments bonds (AA) 36 Governments bonds (A) 18 Governments bonds (BBB) 6 Governments (below BBB) 4 Corporate bonds (above BBB) 10 Corporate bonds (below BBB) 2 Agency paper 32 Asset-backed bonds 12 Mortgage-backed bonds 12 Index-linked bonds 12 Equities 4 Hedge Funds 0 Property 0 Alternative investment 4 Source: Central Bank Publications

The outcome, above, shows that moving down the credit curve in terms of rating has been the most popular option and moving into non-government sectors has been the second option. Alternatives, equities and corporate bonds are also included in the asset mix for some reserve managers, but this is a much smaller component and limited to the more aggressive investors.

A second source also supports this finding. The UBS Central Bank Survey that includes the results of more than 50 respondents shows a steady increase in the percentage of Central Banks that approve Mortgage Backed Security and Asset Backed Security (MBS/ABS) allocations. The survey is somewhat similar but shows the progression of moves into new asset classes over the past eight years, which tells a very important story and indicates the trend among reserve managers.

Relevant points to note from the above survey are that the proportion of correspondents investing in non-Treasury sectors has increased steadily over the years suggesting that return is becoming increasingly important. Furthermore allocations to structured securities such as Mortgage Backed Securities (MBS) and Asset Backed Securities (ABS) have increased dramatically, as the percentage approving these structured securities has grown from 2% to 39%, surpassed only by allocations to bank debt. MBS and ABS are clearly identified as sectors of interest and relevance to Central Bank reserve managers, in terms of their risk and return characteristics as well as in qualitative terms such as their rating, the collateralized nature of the asset classes and the relative protection from headline risk. Another point to note from the above is the reduction in the number of Central Banks allocating to sovereigns. Part of this is likely due to availability, but the lower liquidity of this sector is also a relevant factor behind this statistic.

The objective of most yield enhancement strategies undertaken by Central Banks is to introduce credit spreads into what would otherwise be a government portfolio by investing in either structured securities or unsecured corporate securities. Taking into account the objectives and risk tolerance both of Canada, and of most Central Banks, we would not recommend the inclusion of unsecured corporate debt. This introduces headline risk, company preference issues and at the extreme, the potential for accusations of conflict of interest between the corporate world and the Central Bank.

An area we would strongly favor for inclusion in the EFA, and one that is supported by the surveys above in terms of eligibility and appropriateness, is the introduction of spread through investments in structured (collateralized) securities. We recommend considering diversifying some government risk into high quality Asset Backed Securities and low volatility, short duration, prepayment stable US Mortgage Backed Security structures such as Collateralised Mortgage Obligation Floating Rate Notes ('CMO floaters'), Planned Amortisation Class (PAC) CMOs and Adjustable Rate Mortgages (ARMs). Both these sectors generally offer high credit ratings, fully collateralized US agency debt in the case of US Mortgage Backed Securities, and superior yields to government securities while offering low volatility of relative returns. Our recommendation focuses on generating incremental yield above a stated benchmark through investing in Floating Rate and Fixed Rate Asset Backed Securities. In this case the benchmark may be considered to be the current government asset structure of the EFA. Value would be derived from the higher yields generally available in these instruments relative to government debt, as well as potential additional return through the active management of security selection and sector allocation across and within the different ABS sub-sectors. The high quality (AAA and AA) sectors and those in which many Central Banks invest are: home equity loans (HEL), credit card receivables, auto loans, and student loans. The following provides details of the ABS market in terms of size, liquidity and incremental yield as this is an area in which a number of Central Banks invest, Central Banks whose only other investments are those similar to the eligible instruments currently approved in the EFA.

The US ABS market offers broad diversification in terms of sub-sectors, issuers, rating and geography. As issuance has grown, liquidity in the asset class continues to improve. Chart 2 illustrates the increased issuance and shows the growth of different sectors of the ABS market.

Asset backed securities provide a means of adding incremental yield relative to US Treasuries and deposit rates (LIBID), while at the same time maintaining a high credit quality and a high degree of liquidity. The yield spread is largely to compensate the invester for structural risk, or the risk of cash flow volatility, and to a less extent to compensate for the credit of the servicer. All of these risks can be carefully monitored, controlled and managed within the context of an overall portfolio.

For duration-sensitive investors, investing in high quality floating rate ABS benchmarked to one month or three month LIBOR would be an ideal option. However, the Bank of Canada should also consider taking advantage of investment opportunities in fixed rate ABS. Fixed rate instruments are an investment option for the EFA as the fund is allowed to maintain absolute duration as long as it is offset by the liability exposure, thus yielding little net duration exposure to the reserves.

In the following table, we have compared the 1-3 year sector of the ABS market over the 12 year period ending December 2005 to US Treasuries, US Agencies and US corporate bonds (1994 is the first year in which accurate index data became available for ABS). All non-government asset classes represented in the table have generated substantial excess returns at volatility levels only marginally above those of US Treasuries. Corporate bonds offer superior risk/return characteristics but asset-backed securities, which are largely AAA-rated (we would in any event recommend restricting a Central Bank portfolio to AAA-rated issues) have earned a substantial incremental yield relative to US Treasuries, a comparable return to corporates yet are significantly more liquid. Finally, as noted earlier, this asset class is much more acceptable to the Central Bank community given the general aversion to the headline risk associated with corporate bonds.

Table 6

The impact on the EFA of adding ABS into the asset allocation would be to lower the risk of the aggregate portfolio, achieved by introducing diversification but with a similarly low volatility asset class as US Treasuries, and to increase the return potential primarily through yield enhancement. The table above shows that the historical annualized return increment of Asset Backed Securities over US Treasuries over the past 12 years has been 73 basis points for comparable volatility. 73 basis points translates into US$7,300,000 additional income for every US$ 1 billion allocated to the asset class. The forfeited return relative to US agencies is 44bps, or US$4,400,000 in additional income for every US$ 1 billion allocated to ABS and not agencies. These are sizeable forfeitures for Canada and deserve additional consideration as to whether the current conservative allocation to G3 governments, agencies, sovereigns and supranationals could be broadened for the purposes of increasing return potential.

Liquidity in Asset Backed Security Market

ABS are a highly liquid and growing sector of the fixed income universe. The chart to the right depicts the breakdown of the total US Bond Market by sector. The US ABS market has now grown to approximately US \$1.8 trillion, close to 50% of the size of the US Treasury market and just a little less than the size of the US Federal Agency market. The chart below shows the annual growth of the Asset Backed Securities sector and clearly illustrates the steady growth in new issuance of Asset Backed Securities.

In terms of bid/offer spreads, high quality ABS issues trade, as stated in our spread survey earlier, on a spread of less than one thirty second. This allows for a high degree of liquidity in terms of liquidation or sales, with the main impediment being that of sourcing suitable new issues for reinvestment or purchases. Asset Backed Securities generally trade T+3 settlement so have the same settlement characteristics as the Euro denominated government securities already permitted in the portfolio. Although current spreads provide little indication of incremental return potential, the table below provides an indication of the incremental yield currently offered by this asset class.

Table 7

Spreads have tightened substantially in recent years as the ABS market has grown, become more liquid and also more mainstream, This mean that the incremental return above LIBOR is now quite limited. However, the asset class still offers incremental return relative to Government debt and should be evaluated for inclusion in a longer term context and particularly to potentially prepare for future investment should spreads become more attractive.

Active management of Asset Backed Securities within the context of the EFA should generally revolve around the following three fundamentals:

• ABS exposure is duration and yield curve neutral. The sector decision is a separate decision from duration and yield curve.
• Relative value opportunities exist across a diversified portfolio of ABS.
• Sophisticated analytics are essential to assessing value and controlling risk.

There are two main sources of added value in managing Asset Backed Securities. The first is the incremental yield derived from these securities. The second is the value derived from managing sub-sectors within the ABS universe and then selecting specific issues within those sub-sectors. The different sub-sector opportunities within the ABS market include fixed and floating rate notes across home equity loans, credit card receivables, auto loans and student loans. The security selection process should seek to determine whether the investor will receive adequate compensation for the identified risks of a particular security.

### Conclusion

• We would identify the management of Canada's reserves as being among the more conservatively managed within the Central Bank community.
• The EFA is managed prudently and with full regard for the key objectives to ensure liquidity and capital preservation. Subject to the achievement of these goals, return becomes a tertiary consideration. However, Canada have historically managed to contribute to the public purse through the excess return earned above the funding costs of the EFA.
• The reserve management structure is unusual in so far as liabilities are matched against assets, whereas more usually (though not necessarily a superior structure) assets are managed against a benchmark and are evaluated in the context of the benchmark and not against a cost of funding.
• The excess return earned over funding in 2005 was 9bps while simultaneously ensuring liquidity and capital preservation relative to the liabilities - the key objectives of the EFA.
• Liquidity is very high in all investments within the EFA. We suggest reviewing the need for the highest level of liquidity for the aggregate reserves and suggest that the cost of the liquidity premium may be lowered by reducing liquidity marginally within a component of the reserves.
• A trend we have observed in recent years is the more aggressive investment of reserves by many Central Banks as the need for return becomes an objective of increasing importance.
• We suggest a point for consideration might to be to move the management of the EFA from the Liquidity strata closer to the top of the Intermediate Liquidity layer on the triangle of reserves, an increase in risk with the objective of increasing return.
• Other suggestions which lead to an increase in risk to potentially increase the additional return earned over funding include evaluating adding duration to the assets, focusing around the two year duration area, an area that is commonly chosen as a duration target / benchmark for Central Banks moving away from the Liquidity tranche for the first time.
• Another recommendation is to consider evaluating non-government sectors and specifically to consider introducing ABS and low volatility MBS structures to introduce incremental yield at a comparable level of volatility to the current Treasury and Agency allocation.
• The currency allocation is a large component of the volatility of the reserves and we recommend a quantitative analysis of the current allocation to ensure that the breakdown is appropriate.
• We further suggest consideration be given to increasing the number of currencies held in the EFA to provide a more diverse and lower risk portfolio of assets