# Archived - Evaluation of the Asset Allocation Framework 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.

Department of Finance

Final Report
Submitted by
Jonathan Jacob
Rainer Kaufmann

Forethought Risk
www.forethoughtrisk.com

Approved by the Deputy Minister of Finance on the recommendation of the Audit and Evaluation Committee on March 2, 2012

Executive Summary

Introduction

Description of the Asset Allocation Framework

Evaluation Context and Design

Evaluation Findings

Conclusions

Recommendations

Appendix A: Matrix of Results

Appendix B: Investment Management Process

Appendix C: Foreign Central Bank Comparison

Appendix D: Case Studies

Appendix E: Literature Review

Appendix F: Acronyms

Appendix G: Departmental Response and Action Plan

## Executive Summary

The Asset Allocation Framework (AAF) is the framework of governance and process that guides the portfolio management and decision-making activities involved in managing the Exchange Fund Account (EFA). The EFA is the portfolio that represents the largest component of Canada’s official foreign reserves and is largely comprised of liquid financial instruments. The purpose of the reserves is to provide a mechanism of support to the external value of the Canadian currency. It is also intended to act as a reserve source of liquidity for federal government operations. Investments within the EFA are governed by the Statement of Investment Policy (SIP) established by the Minister of Finance.

The AAF evaluation was done as a part of the Treasury Evaluation Program. The objective of the AAF evaluation is to assess the relevance and performance, including the efficiency and economy, of the AAF, with particular emphasis on governance, control mechanisms and decision making, as well as internal and external accountability and reporting. The evaluation is intended to be useful for officials within the Department of Finance and the Bank of Canada, parliamentarians and the public. The scope of this evaluation covers the activities of the AAF from 2000 to 2010, and was performed during the period from October 2010 to June 2011.

The evaluation found that the objectives of the AAF were consistent with priorities of the Government of Canada and were aligned with the government’s roles and responsibilities.

The evaluation found that the AAF achieved its expected outcomes of liquidity and capital preservation. The evaluation found no methodical investment process in place to target portfolio return and thus it cannot be stated that returns were optimal.

Due to the absence of a methodology for optimization of returns, it could not be stated that the AAF achieved its objectives efficiently. Aside from this deficiency, the AAF has a well developed governance structure guiding its activities, yet some processes within the AAF were found to need improvement. In particular, the portfolio management process was found to lack clear definition of investment preferences and accountabilities. Communication processes were also found to lack sufficient definition.

The evaluation found that there are alternatives to the current AAF, but they would result in changes that may conflict with the three primary objectives of the EFA: liquidity, capital preservation and return optimization.

The AAF should be strengthened by adding and refining operational processes concerning both portfolio management and group communications. The most pressing need is to enhance and formalize the investment process within the AAF. Twelve specific recommendations are made in this report to address improvements in these various areas.

## Introduction

This report discusses the evaluation of the AAF performed by Forethought Risk in 2010 and 2011 as part of the Treasury Evaluation Program of the Department of Finance. The objective of the evaluation is to assess the efficiency, accountability, performance (effectiveness[1]) and appropriateness of the AAF used to manage the EFA of Canada. Forethought Risk is an external, third-party firm, and the report and evaluation have been objectively performed.

## Description of the Asset Allocation Framework

The AAF is the framework guiding the portfolio management activities and decision-making roles involved in managing the EFA. The EFA is a balanced portfolio of high-quality financial instruments that are held as assets of the Government of Canada. EFA investments are funded by liabilities, in the form of debt instruments that are issued by the Government of Canada.

The EFA portfolio is made up of Special Drawing Rights (SDR), gold and fixed-income debt investments. This portfolio is held in the name of the Minister of Finance and managed jointly by the Department of Finance and the Bank of Canada. The objective of the EFA, as specified in Part II of the Currency Act, is to aid in the management of the external value of the Canadian dollar, with emphasis on the control and protection of the currency. The portfolio is expected to provide access to foreign currency reserves, which will be used as a source of liquidity in times of need.

The EFA has three strategic objectives. The two primary strategic objectives are to provide ready access to liquid foreign assets and to preserve the capital value of the portfolio. The third strategic objective is to optimize return, defined as achieving the highest possible rate of return on the portfolio, subject to meeting the liquidity and capital-preservation objectives. The EFA is structured as a portfolio of matching assets and liabilities, so that funds raised by the Government of Canada are invested in foreign securities in a manner that very closely matches their currency and maturity timing. This is done to minimize the financial risks to the government of the foreign-currency and interest-rate markets.

The liabilities used to fund the EFA are typically fixed-income debt securities, issued in the name of the Government of Canada. Some debt securities are issued directly in the target foreign currency and the funds raised are then invested in foreign assets of the same currency. Currently the largest portion of securities is issued as Government of Canada bonds in Canadian dollars and the funds are converted into foreign currency through cross currency interest-rate swaps. Once the currency conversion has occurred, the funds are then invested in appropriate foreign assets of the swapped currency.

The majority of the EFA assets are contained in an actively managed portfolio of liquid, high quality, debt securities and term deposits, denominated in foreign currencies such as U.S. dollars, Euros and Yen. The composition of EFA assets is governed by the SIP. The SIP defines the policy guiding the overall management of the assets of the EFA, including their purchase, management and sale. Asset allocation decisions are limited to investing funds in the assets approved by the SIP. Currency and interest-rate risk exposures are excluded as a result of the currency and maturity-matching requirements of the SIP.

The AAF is concerned with the decision-making processes and activities related to asset allocation and performance measurement within the EFA. Investment activities within the EFA can be categorized as strategic asset allocation, tactical asset allocation and active portfolio-manager trading decisions.

• Strategic asset allocation decisions are concerned with the longer term investment of funds in the different fixed-income asset classes allowed by the SIP. The objective of these decisions is to optimize the return of the portfolio for the amount of risk taken, given the constraints of the SIP on the EFA portfolio.
• Tactical asset allocation decisions are concerned with shorter-term investment actions, intending to either take advantage of relative-value changes in asset sectors or to reduce the risk exposure of the EFA to specific asset sectors.
• Active portfolio-manager decisions are very short-term investment actions, intending to take advantage of opportunities in the relative value of specific securities within asset classes.

The only definitive guide for asset allocation for the EFA is the broad-limit structure of the SIP, which results in a strategic asset allocation process that lacks clear definition. The current asset-allocation investment process can be summarized simply as determining the cost of any new investable funds, and buying securities that provide a greater return while respecting the EFA asset-class limits set by the SIP. Agreement, or not, by senior management in the asset-class and security decision is subsequently communicated, resulting in an investment process of revealed preferences.

The management structure of the EFA, within which the AAF operates, is specified in the document Funds Management Governance Framework. This document describes the roles and responsibilities of officials within the Department of Finance and the Bank of Canada, and of the various committees created to govern activities in managing funds, including the EFA.

The most-senior management group is the Funds Management Committee (FMC), which oversees the EFA as well as all other activities involving wholesale debt and cash management. This committee reports to the Deputy Minister and is directly supported, in matters concerning the EFA, by four management groups.

The Risk Committee (RC) is responsible for providing the FMC with opinions on EFA risk limitations, and forwarding recommendations to the FMC from other operating committees.

The Asset Liability Management Committee (ALMC) provides the FMC with recommendations concerning strategic and policy matters affecting the EFA. This is the committee that most actively manages EFA activities.

The Funds Management Coordinating Committee (FMCC) provides the FMC with recommendations concerning strategic and policy matters affecting the EFA as well as domestic debt- and cash-management decisions concerning the EFA.

The Treasury Evaluation Committee (TEC) is an advisory group that assesses policies and operational, as well as debt and cash, management.

In addition to the Funds Management Governance Framework, the AAF operates within the overview of a number of other governance documents and reports.

The Statement of Investment Policy (SIP) specifies the policy, approved by the Minister of Finance under the Currency Act, governing the acquisition, management and divestiture of assets for the EFA.

The Memorandum of Understanding on Treasury Risk Management between the Bank of Canada and the Department of Finance defines the treasury risk management policy framework of the government, and the roles and responsibilities of the Bank of Canada and Department of Finance in the implementation of the framework.

The Government of Canada Treasury Risk Management Framework document presents the risk management framework for the government’s liquid financial assets and market debt.

The Swap Management Policy for the Government of Canada defines the policy for swaps used to manage the liability structure of the government’s marketable debt.

The Report on the Management of Canada’s Official International Reserves is an annual report that provides details on the operations of reserves.

The Exchange Fund Account Monthly Performance and Risk Report is an internal summary report that presents the performance and risk exposure of the fund each month.

### Evolution of the AAF since 2000

The processes, performance measures and objectives of the EFA have evolved since the year 2000 and affected the AAF. Canada’s official reserves and the EFA grew quickly in the second half of the 1990’s. This was a time of historic weakness in the currency and also a period of frequent currency interventions by all major nations. From 1995 to 2000, the EFA roughly tripled in size from $10 billion to$31 billion. As a result, the risks in the portfolio grew and various changes were made over time in response to evolving market conditions in order to better define and strengthen the AAF supporting the EFA.

In 2000, the EFA had three objectives[2]:

• To promote orderly conditions in the foreign exchange market for the Canadian dollar.
• To provide general liquidity for the government.
• To minimize the cost of carry of the government’s foreign currency liabilities used to fund the EFA’s assets, with due attention to risk management.

Governance of the EFA began with the Minister of Finance, who approved the general policies related to the management of the EFA, established the target level of reserves, and provided an annual report to Parliament on the operations of the Account. Responsibility for the management of the EFA was shared by the Department of Finance and the Bank of Canada.

The Director, Financial Markets Division (Department of Finance), and the Chief, Financial Markets Department (Bank of Canada) were responsible for the on-going management of the EFA. A policy committee, comprised of senior management of the Department of Finance and the Bank of Canada, met semi-annually to review developments, approve major policy initiatives, and provide guidance and accountability on the management of the EFA.

The Risk Management Unit (RMU), within the Bank of Canada oversaw the risk position of the Government of Canada, and monitored and advised on the risk position of the EFA, including market, credit and liquidity risk. The governance structure for risk management was defined by the Government of Canada Treasury Risk Management Framework which was first released in the year 2000.

The management of the EFA was governed by seven principles and three key management guidelines[3]:

Principles

1. There should be a sufficient amount of high-quality, highly liquid, foreign-exchange reserve assets available for intervention and general liquidity purposes.
2. The difference between the interest paid on the foreign currency liabilities the government used to fund EFA assets, and the interest earned on those assets should be minimized (to the extent that the former is higher).
3. Foreign reserves should be managed to ensure, as much as possible, that the assets match the liabilities in currency and duration.
4. A prudent maturity structure and maturity profile should be maintained to limit refinancing needs.
5. Best practices, with regards to risk management, should be applied in the overall management of the EFA.
6. Credit risk should be managed prudently through diversification of the EFA asset portfolio, with appropriate use of credit ratings and counterparty limits.
7. Foreign currency borrowing activities, to fund EFA assets, should be conducted so as to maintain Canada's reputation as a successful borrower in international capital markets.

Guidelines

• The EFA portfolio is structured into a Liquidity Tier and an Investment Tier.
• At least 50% of the EFA's assets are denominated in U.S. dollars. The remainder of the portfolio is split among euro- and yen-denominated assets.
• The EFA holds debt in the designated currencies of highly rated sovereign governments and their agencies, and by supranational organizations having an AA rating or better.

In 2001, collateral agreements were introduced to reduce credit risk to trading counterparties, by allowing cash lending based on the security backing in the repo market. Collateral agreements were also developed for the cross-currency swap program, allowing the Bank of Canada to enter into transactions with single A-rated counterparties. This increased the potential number of trading counterparties, while providing a degree of risk diversification. The process of using collateralized cash deposits in the repo markets was further refined, leading to the introduction of tri-party repo agreements in 2003. All of these changes had the effect of increasing the set of trading counterparties for the Bank of Canada, thereby increasing liquidity while reducing risk through both the collateralization effect and the counterparty risk-diversification effect.

The effects of the introduction of the collateral framework can be seen in the funding method of the EFA. The EFA is funded through a variety of sources, some directly denominated in foreign currency such as global bonds, Canadian bills and euro medium-term notes. Another important source of funding is the use of cross-currency swaps to convert the proceeds of Canadian-dollar bond issues into foreign currency. The economics of this source of funding is generally highly beneficial to Canada but comes at the cost of counterparty credit risk, as well as basis spread risks. The introduction of collateralization allowed Canada to mitigate counterparty risk and, as this limiting factor was removed, the percentage of funding done through swap funding grew from 25% of the EFA portfolio in 2000 to 55% by 2004.

In 2004, the Currency Act was modernized and changes were enacted reflecting the increasing size of the EFA and the evolution in both governance practices and the financial markets. The EFA investment principles and objectives were reviewed, with emphasis on the range of eligible investment assets and on the underlying policies regarding the operation of the EFA, given the best practices of comparable sovereign reserve managers. The result was a change in the principles and objectives governing the management of the EFA, in order to enhance decision making.

The Currency Act was amended so that, instead of providing an inflexible list of approved instruments, a range of investment asset classes was allowed, offering the potential of greater investment returns at lower risk.

The management framework of the EFA was also changed. The purpose of the EFA was then redefined as providing a source of funds needed to help promote orderly conditions for the Canadian dollar in the foreign exchange market and to provide foreign currency liquidity for the Government of Canada.

The key objectives became[4]:

• Maintain a high standard of liquidity: Hold reserves in assets that mature or can be sold on very short notice with minimal market impact and, therefore, minimal loss of value.
• Preserve capital value: Minimize risk of loss of market value by holding a diversified portfolio of high-quality assets (in terms of credit rating and type of issuer), managing assets and liabilities on a matched basis (in terms of currency and duration) and using appropriate practices to mitigate risks.
• Optimize return: Achieve the highest possible level of return, while respecting the liquidity and capital preservation objectives.

The list of guiding principles was reduced from seven principles to three core and two additional ones[5]:

• Governance: The governance framework should ensure that debt and asset management activities are conducted in line with clearly established operational and risk guidelines, and that risk-monitoring and oversight are independent of treasury management operations.
• Transparency: Information on the management and performance should be disclosed in a timely manner so as to ensure public understanding and accountability.
• Leading practices: Operational frameworks and practices should be in line with leading practices of other comparable sovereigns. Regular evaluations should be conducted on the effectiveness of the governance framework, policy initiatives and operations.

• Prudence: Borrowing and investment activities should be managed to limit exposure to risk and protect the value of the government’s assets.
• Cost effectiveness: The carrying cost of reserves, taking into account the return on assets and the cost of liabilities, should be kept as low as possible.

Ultimate responsibility for governance rests with the Minister of Finance, with operational oversight now done through the FMC, which is composed of senior management from the Department of Finance and from the Bank of Canada, similar to the earlier Policy Committee it replaced.

The Memorandum of Understanding on Treasury Risk Management between the Bank of Canada and the Department of Finance was also released in 2004. The RC was formed to oversee and advise on risk management policy and report on risk positions and exposure to the FMC. The Financial Risk Office was formed at the Bank of Canada to provide support to the RC and report on performance and risk measures of the EFA, which include market, credit, operational, liquidity and legal risks.

Immediate changes that occurred as a result of this modernization step included the increased performance and risk reporting through monthly and quarterly reports. The calculation of value-at-risk (VaR) measures for the EFA was introduced along with the results of stress tests meant to mimic EFA performance during periods of extreme market behavior. The VaR calculation was intended to estimate the financial risk inherent in the EFA, consistent with the major policy principle that all material exposures to financial risk should be identified, measured, monitored, controlled and reported. Two VaR measures were introduced as a result. The first VaR attempted to measure the risk in the EFA assets only, and the measure was most highly influenced by the effects of the EFA duration and credit risks. The second VaR measure attempted to measure the risk of the net EFA, including both assets and liabilities. The value of these measures differed greatly – by factors of 50 or more. As is found in this evaluation, the net VaR calculation ignored many spread risks inherent in the net EFA portfolio and was, therefore, a severely limited measure by its design. In subsequent years, the net VaR measure became known as the market VaR, within the Department of Finance and the Bank of Canada.

Another immediate change was that the set of allowable EFA assets was increased and the portfolio was invested in a wider range of securities within those asset classes, as portfolio managers attempted to reduce risks through diversification of assets.

On the EFA liability side, the use of swap funding continued to increase as the relatively small market VaR number, combined with the return optimization objective, led to the increased use of this profitable funding mechanism. The percentage of EFA funding through swap funding expanded from 55% in 2005 to 81% by 2008.

The changes in 2004 also included the introduction of Total Return as a performance measure. In the past, the performance of the EFA had been evaluated on the basis of carry. Carry can be viewed simply as the difference between the rate of interest earned on the assets and the rate of interest paid on liabilities. Carry, as a return measure, was adequate when the portfolio was relatively small and the assets and liabilities were well matched in both qualitative and quantitative aspects. As the portfolio grew and EFA assets and liabilities became less tightly related as a result of swap funding intermediation, carry no longer provided adequate performance measures. Total return added the effects of the changes in the value of the securities within the EFA to their carry income, to produce a more accurate measure of how the EFA performed. This introduced a measure of mark-to-market accounting standards to the EFA, consistent with the best practices of the private sector asset managers.

With all of the changes in the Currency Act and the restatement of objectives for the EFA, there was a need to better define the governance structure surrounding the EFA, and thereby strengthen the AAF. In 2005, the SIP was introduced to provide added structure to the new investment flexibility resulting from the reworked Currency Act. This resulted in increased opportunity to diversify the investments of the EFA. The Government of Canada Treasury Risk Management Framework document was also reviewed and an updated version was published in January 2006.

In 2007, the Funds Management Governance Framework was released. This document described the governance framework for managing the Government of Canada’s financial assets and liabilities. The framework was intended to[6]

• Facilitate the discussion of key policy matters and set out decision-making authority for policy and operational changes.
• Support accountability through clear and transparent assignment of roles and responsibilities.

For the AAF and the EFA, this latest governance document once again defined the role of the FMC as overseeing all activities covering reserves and risk management. The governance role of the ALMC was split between an advisory role in support of the FMC and as a decision-making body, within limits, as defined by the following paragraphs from the Funds Management Governance Framework:[7]

The Asset Liability Management Committee (ALMC) meets regularly (typically every six weeks or as needed) and provides recommendations to the FMC, in its advisory role, on strategic and policy matters affecting the management of foreign reserves, including changes to the limits and guidelines pertaining to the foreign reserves established by the Minister and the FMC.

Within limits delegated by the FMC, the ALMC is also a decision-making body, whose decisions are executed by officials at the Bank of Canada and Department of Finance. In particular, the ALMC oversees the development of the annual funding and investment plans for the foreign reserves as well as the performance metrics regarding these plans, within the limits set by the FMC. Determining the annual funding plan involves setting targets and/or ranges for funding and investment activities.

What was not covered in this definition was the day-to-day decision-making role of the ALMC in regards to the management of the EFA. The definition stated that the ALMC will make limited decisions on an annual basis in overseeing the development of investment plans. It also stated that the ALMC will advise the FMC on strategic matters concerning the foreign reserves. Clarity of decision making over operational changes, a stated goal of the document, was not fully addressed and therefore some aspects were left open to future decisions by the FMC. For the EFA, the ambiguity of roles means that the second objective of the document, accountability through clearly defined roles and responsibilities, was unlikely to be met.

The credit risk policy was also reviewed in 2007, in light of the increasing credit risk potential arising from the evolving SIP and market conditions.

In 2008, the start of the global financial liquidity crisis, the credit classes within the SIP were expanded as governments, particularly the U.S., began to offer explicit credit guarantees to sub-sovereign and super-national debt. The allowable assets were expanded to include government-guaranteed or government-supported securities rated less than AAA, and a limit placed upon this sub-category of 20% of EFA assets. This removed the onus on the EFA to sell investments in those classes, should their rating fall below the AAA level. The credit crisis also changed the nature of liability funding for the EFA due to the nature of cross-currency swaps, which have large counterparty credit requirements, but are collateralized. Since that time, the EFA has been changing its funding composition to include more global bond issues, in an attempt to remove the stress on bank counterparty credit, reduce the mark-to-market volatility created by the swap market and add diversification.

## Evaluation Context and Design

### Evaluation Objective and Scope

The AAF evaluation was done as a part of the Treasury Evaluation Program. This program manages third-party reviews of policies and operational programs in treasury management. The objective of AAF evaluation is to assess the relevance and performance, including the efficiency and economy of the AAF, with particular emphasis on coordination and control mechanisms, decision making, as well as internal and external accountability and reporting. The evaluation is intended to be useful for officials within the Department of Finance and the Bank of Canada, parliamentarians and the public. The results of the evaluation will be used to aid in decision making. The scope of this evaluation covers the activities of the AAF from 2000 to 2010 and was performed during the period from October 2010 to June 2011.

### Evaluation Methodology

The evaluation addressed five issues, as follows.

Table 1
Evaluation Issues
Evaluation Issue Questions to Address Issue
Relevance
Alignment with Government of Canada priorities Are the objectives of the AAF consistent with the priorities of the Government of Canada?
Alignment with Government of Canada roles and responsibilities To what extent is the AAF aligned with the roles and responsibilities of the Government of Canada?
Performance
Achievement of Expected Outcomes To what extent has the AAF been able to achieve its objectives and expected outcomes? Are adjustments needed to improve management performance?
Demonstration of Efficiency and Economy Has the AAF achieved its objectives efficiently? How can its efficiency be improved?
Exploration of Alternatives Are there more cost effective ways to achieve the same results or objectives?

### Data Collection

Data and information used in answering the evaluation questions were collected from independent sources and include results from internal and external interviews, reports, documents and literature reviews.

### Interviews

Internal interviews were held with personnel from the Department of Finance (four) and the Bank of Canada (seven) who are directly involved in management of the EFA. The interviews were structured around a set of 18 questions dealing with the evaluation issues vetted by the Internal Audit and Evaluation team of the Department of Finance.

External interviews were held with personnel at seven foreign central banks who manage foreign reserves on behalf of their nations. The interviews were structured around a set of 12 questions dealing with the structure of their reserves management and asset allocation process. Case studies were prepared for a few of these banks, based on information from these interviews and from previously published documents. The case studies can be found in Appendix D. Less structured interviews were held with three private sector asset managers to deal with specific topics within their areas of expertise.

### Document Review

Department of Finance governance documents related to the activity of asset allocation for the EFA were reviewed. Also reviewed were the monthly and annual performance reports, minutes of meetings of management groups and internal memoranda.

### Data Review

Performance data were obtained from a combination of monthly and annual reports, in order to perform data analysis, which consists mainly of the presentation of time-series charts. Data values were compared with limits for those values as determined by the EFA governance documents. Further comparisons of risk measures were performed, especially comparing actual historical performance results with VaR predicted measures.

### Literature Review

A literature review was undertaken to discover articles concerning methods and best practices used in foreign reserves management and asset allocation for central banks, including articles published by the BIS, IMF and other central banks. Results from this review were used in the preparation of the foreign-bank case studies.

### Evaluation Limitations

Internal interviews were held in January 2011 with only 11 persons – a small sample of the many who are involved in performing AAF-related activities; thus the evaluation suffers from potential sampling-type errors. Many of these 11 persons had not been involved in the AAF from the year 2000, which limited the amount of historical information available. Similarly, external interviews were limited to a single person from each external institution, again creating the potential for sampling errors and incomplete information. The foreign central banks each have models of foreign-reserves management that differ from those in Canada, which thereby limits the applicability of the other central banks’ experiences. Private sector asset managers likewise have business models that vary from those of the government.

The data used for analysis came from annual performance reports for the period 2000 to 2010, or from monthly performance reports from 2004 to 2010. Since some data measures such as value at risk and total return were not available prior to 2004, there are some incomplete data series within the full 2000-to-2010 evaluation period. This limits the historical completeness of the evaluation.

The minutes of the FMC and ALMC meetings were reviewed for the period 2004 to 2010. Therefore, no data arising from the meetings of these groups are presented for the period of 2000 to 2004 of the evaluation period. This limits the historical completeness of the evaluation.

The literature review was limited to researching articles concerned with central bank foreign reserves and asset allocation methods since 2000.

## Evaluation Findings

Evaluation Issue 1: Are the objectives of the AAF consistent with Government of Canada priorities?

As indicated above, the AAF is the framework guiding the portfolio management activities and decision-making roles involved in managing the EFA. In the Department of Finance’s Program Activity Architecture, the AAF is placed under the Reserves Management Sub-Activity and the Treasury and Financial Affairs Activity. This activity is meant to contribute to the achievement of the Government of Canada’s fiscal and economic priorities in general, and its “strong economic growth” outcome in particular. As one of the four main lines of activities in the Department of Finance, it is also meant to contribute to achieving the Department’s strategic objective of building “a strong and sustainable economy, resulting in increasing standards of living and improved quality of life for Canadians”, and its associated priorities of Sound Fiscal Management, Sustainable Economic Growth, Sound Social Policy Framework and Effective International Influence.[8]

The EFA contributes to the achievement of these priorities and outcomes by supporting the Government of Canada’s financial situation. For example, the high standard of liquidity ensures that the government has ready access to foreign funds it can use to support the external value of the Canadian dollar and as a source of liquidity for government needs. For a nation with its own currency, sound fiscal management includes the need to manage the stability of that currency in order to protect both the domestic and international economies from exogenous currency shocks. The liquidity objective of the EFA is consistent with both of these needs, and with the priorities of sound fiscal management, sustainable economic growth and effective international influence. Sound social policy is an indirect consequence of economic growth and sound fiscal management and thus also consistent with this objective.

The capital preservation objective ensures that the capital invested in the EFA is at minimal risk of loss. The minimization of loss is part of any sound fiscal management process. Not only does it minimize negative influences on sustainable economic growth and act to limit negative influences on reputations, it also increases the government’s ability to take action. Therefore, the capital preservation objective is consistent with the priorities of sound fiscal management, sustainable economic growth and effective international influence.

The optimized return objective attempts to ensure the highest possible rate of return on the portfolio, subject to meeting the liquidity and capital preservation objectives. Sound fiscal management includes making investment decisions based upon the risk and return potential of the investment. Sustainable economic growth is based upon achieving consistent growth in earnings. Social policy is predicated upon the ability to fund such policy. International influence is based upon domestic positions of economic strength.

These conclusions were also confirmed by each of the four Department of Finance and seven Bank of Canada personnel who were interviewed as part of this evaluation. As a result, all evidence suggests that the objectives of the AAF are consistent with the Government of Canada’s fiscal and economic priorities and contribute positively to the achievement of its strong economic growth outcome.

Evaluation Issue 2: Is the AAF aligned with the roles and responsibilities of the Government of Canada?

This evaluation issue concerns the proper jurisdiction for AAF activities.

Canada’s Constitution Act of 1867 determines the distribution of legislative powers between the various levels of government in Canada. A review of the 1982 Consolidated Constitution Act indicated that all matters concerning currency and coinage, and the issue of paper money would fall under the exclusive legislative authority of the Parliament of Canada. Thus, the federal government has the exclusive responsibility for managing all matters concerning the Canadian currency including, by definition, the activities of the AAF and EFA.

The Currency Act was passed to provide the federal government with legislative authority to manage the responsibilities assigned to it by the Constitution. Part II of the Currency Act deals with the EFA and activities related to it, and assigns management responsibilities for the EFA to the Minister of Finance and his delegates. Accordingly, the Minister of Finance has the authority to acquire, borrow, sell or lend assets held in the EFA in accordance with the SIP[9]. Therefore, the AAF and EFA are aligned with the federal government’s roles and responsibilities.

This evaluation question was also asked of the same four Department of Finance and seven Bank of Canada personnel who were interviewed as part of this evaluation. Each affirmed that the AAF was aligned with the roles and responsibilities of the Government of Canada.

Evaluation Issue 3: Has the AAF achieved its expected outcomes?

The expected outcome of the AAF is the performance of the EFA, consistent with its objectives of liquidity, preservation of capital and optimization of returns. These objectives are defined within the SIP (2008) as the following:

• Maintain a high standard of liquidity: Hold reserves in assets that mature or can be sold on very short notice with minimal market impact and, therefore, minimal loss of value.
• Preserve capital value: Minimize risk of loss of market value by holding a diversified portfolio of high-quality assets (in terms of credit rating and type of issuer), managing assets and liabilities on a matched basis (in terms of currency and duration), and using appropriate practices to mitigate risks.
• Optimize return: Achieve the highest possible level of return, while respecting the liquidity and capital preservation objectives.

These three objectives are appropriate, given that the purpose of EFA assets is to provide foreign currency liquidity to the government and to promote orderly conditions for the Canadian dollar in the foreign exchange markets, if required.

The liquidity of investment assets is not directly observable as a metric in the market place, but rather must be inferred from the pricing behaviors of market participants, through their trading activities. Methods for measuring liquidity include polling dealers about their ability to make markets, transacting in the marketplace to observe actual liquidity, and attempting to infer liquidity from published market-volume data for those financial instruments that publish such data on a timely basis, such as futures contracts. All of these methods are subject to high levels of error. Polling and transacting suffer from the inability to obtain an ongoing large and unbiased data sample. Published volume data are available for only a few specific investment assets, and these provide only a broad measure of market conditions. Furthermore, volume data differ from liquidity, since the latter is concerned with both price and volume metrics. Liquidity is not a constant, either across or within asset classes, nor is it constant through time. That liquidity is so changeable and difficult to quantify often forces market participants to define liquidity in qualitative terms, based on the credit quality of the investment assets as well as their maturity. For the EFA, the definition of the liquidity objective is reformulated into a statement concerning the composition of EFA assets and their assumed liquidity characteristics. Thus the intangible quality of liquidity is transformed for the EFA into a set of specific and measurable portfolio-composition metrics that can be managed. The global liquidity crisis of 2008 provided extreme market stresses that tested this construct, and it was exactly those Tier 1 assets that were the most liquid of all asset classes, due to the collective actions of the world’s central banks. The 2008 credit crisis further demonstrated that asset liquidity was heavily dependent upon the ability of financial market participants to make trading markets. The conclusion drawn from this experience is that it is necessary to maintain the broadest possible set of dealing lines with private sector dealers and with central banks in order to ensure that able trading counterparties are available when needed.

The capital preservation objective is similarly reformulated into a statement concerning the composition of EFA assets. Thus a forward-looking statement concerning the objective to minimize the potential for future loss is transformed into a set of specific and measurable portfolio metrics concerning portfolio credit and asset/liability matching that can be managed. The inclusion of the sentence portion concerning, “…appropriate practices to mitigate risks…” may appear to be vague, but it reflects the reality that a brief objective statement cannot reasonably be expected to include a full definition of all modern methods of risk management.

The return optimization objective states that the portfolio is to achieve the highest possible level of return, subject to liquidity and capital-preservation constraints. This objective has not been reformulated into a workable definition. It could be restated to target the highest expected level of return on an ex ante basis, with the recognition that ex post results will differ, depending on financial market developments during the investment period. This restatement of the return optimization objective would not only result in specific and measurable metrics but, when fully acted on, would result in the implementation of enhanced portfolio methods, such as benchmarking, asset-class performance, forecasting and return optimization. These topics are discussed in further detail below.

Recommendation

1. The return optimization objective of the EFA should be rewritten to incorporate the concept of targeting expected returns.

The EFA is analogous in many ways to money market investment accounts for individuals. These accounts are intended to serve the function of acting as a store of liquidity, while not placing capital at risk. The portfolio guidelines for these types of accounts typically limit investments to short-dated assets of high credit quality such as Treasury Bills. Portfolios will include limits and provide metrics on the maturity profile and credit composition of the assets. In doing so, portfolio managers target their objectives of liquidity and capital preservation by relying upon the maturing flow of high quality sovereign debt. The performance of these portfolios will typically be measured against a benchmark portfolio of Treasury Bills with maturities out to 365 days. This investment approach results in low but stable returns that meet the financial security needs of individuals. Such measurements would not be appropriate for the EFA given its explicit asset liability framework, the larger range of asset classes available for investment and its longer maturity profile.

The specifics of how the EFA performed relative to its objectives will now be considered in light of the above. To determine if the EFA portfolio performance has been consistent with its objectives, it is necessary to review the details of its operation.

The EFA has many performance dimensions including economic return, risk profiles, net duration and asset composition. Some of these dimensions can be viewed from only the asset side of the EFA portfolio, but most lose meaning unless viewed in keeping with the asset and liability portfolio construct of the EFA. Therefore, rather than limit this discussion purely to elements of the asset side, we will include the liability side as well when appropriate.

Analysis of economic return is useful in order to determine capital preservation behavior of the EFA as well as return maximization. Historical returns have shown that most years the portfolio has had positive total return earnings, but that there have been periods of high losses caused by negative mark-to-market swings within the swap-funding elements of the liability side. Figure 1, the chart of performance results seen below, illustrates these swings. The EFA has historically been able to wait out adverse market moves and regain profitability, in part because of the lack of need to divest the portfolio to raise funds for currency intervention or liquidity purposes. However, had funds been necessary during those adverse periods, the portfolio would have locked-in high losses in the process of divestment. Indeed, in 2007–2008, losses in the EFA portfolio, on a mark-to-market basis, were $394 million. Clearly the EFA has suffered loss of capital at times, but not to a large degree relative to the size of the portfolio (1.05%). Figure 1: This chart shows that monthly performance has fluctuated between positive and negative returns with periods of exceptionally high volatility leading into, during and following the global financial crisis marked by the September 2008 bankruptcy of Lehman Brothers. The average annual performance of the EFA from 2004 to 2010 is about$68,000,000, and the chart shows that individual monthly returns frequently exceeded this in absolute value. Data used in this chart were taken from the Exchange Fund Account Monthly Performance and Risk Report issues for the period.

The risk profile of the EFA is intended to be limited by construct through the currency and matched timing constraints of the SIP, as well as by the various asset-class limitations included in the SIP. The use of asset-liability matching and asset class limits, however, does not fully address risk taking. VaR is a more fulsome reflection of the risk of a portfolio, and two measures of this type are estimated for the EFA. The “market VaR” metric of the EFA focuses solely on currency and maturity mismatches, and shows that these risks are relatively small. Unfortunately, as shown in Figure 2, below, the variance of historical past earnings is much larger, and so the EFA market VaR metric vastly underestimates the total risk of the EFA. The market VaR measure does not include either credit risk or the spread risks inherent in swap funding, and therefore does not measure the full risk of the portfolio.

The credit VaR estimate of the portfolio is calculated, and while the value of this is more in keeping with the historical variance of total returns, it still does not adequately reflect the total risk of the EFA. The market risk effect of swap funding is presently not included in any VaR calculation, and in order to better understand the risks of the EFA we recommend the inclusion of these risks in VaR calculations.

Recommendations

2. Total market VaR for the EFA portfolio assets and liabilities should be calculated daily. This VaR measure should encompass all risk elements including swap funding, and credit, currency and duration risks.

3. The use of the risk measure called “Market VaR” should be severely curtailed and a full description of its limitations included whenever it is referred to in reports.

4. All VaR-calculated measures of expected risk should be accompanied by appropriate historical measures of similar EFA risk in order to judge their appropriateness.

5. Portfolio risk limits should be set for the total market VaR of the EFA portfolio. A process of timely review of this measure, and action responses for risk-limit contraventions should be created.

Figure 2: This chart overlays the monthly performance results with the calculated anticipated Market VaR and also with the historical three standard deviation confidence bounds. The Market VaR was scaled to make its 10-day calculation base consistent with the monthly series of performance returns. Market VaR is a 99% confidence calculation, and so should be comparable to the historical level of three standard deviations of monthly returns. That the values of the two measures are not remotely similar means the Market VaR values should not be considered an adequate estimate of the total risk contained in the EFA. Data used in this chart were taken from the Exchange Fund Account Monthly Performance and Risk Report issues for the period.

Net duration of the EFA is a measure of the timing differences of the assets and liabilities of the portfolio. The SIP specifies duration matching as an EFA investment objective and this implies a net duration of near zero. The time series of duration for the EFA, shown in Figure 3, reveals that the portfolio has a bias to greater asset duration than liability duration, thus benefiting from an environment of falling interest rates. The duration mismatch could be a result of swap funding or credit-spread duration effects that result in a net duration in spite of matching notional amounts of assets and liabilities. However, future contracts or swaps can be used to eliminate this duration bias, regardless of the source.

Recommendation

6. Portfolio risk limits should be set for the duration risk of the EFA portfolio. A process of timely review of this measure and action responses for risk-limit contraventions should be created.

Figure 3: This chart shows that the duration of the EFA portfolio has often been significantly different than zero. The average duration result for the period suggests a bias to a net-negative duration, reflecting greater duration of assets than liabilities. Such a portfolio would tend to positive performance results in an environment of falling interest rates. Data used in this chart were taken from the Exchange Fund Account Monthly Performance and Risk Report issues for the period.

The asset mix of the portfolio is constrained by the limit structure of the SIP to target liquidity and safety of assets. The monthly and annual performance reports show the asset mix always met the many limits placed upon it. However, the mix of assets does generally vary over time, and on occasion reaches the boundary limits for a particular asset class. Examples of this are shown in the chart in Figure 4 for United States Treasury (UST) holdings and in Figure 5 for Tier 1 and Tier 2 liquidity assets.

Figure 4: This chart of UST security holdings as a percentage of EFA holdings shows that the holdings have always exceeded the 10% minimum set for the portfolio by the SIP. The decision process causing the large increase in holdings during 2009 and 2010 was not recorded in the ALMC or FMC minutes of 2009 and 2010. Data used in this chart were taken from the Exchange Fund Account Monthly Performance and Risk Report issues for the period.

Figure 5: This chart of Tier 1 and Tier 2 liquidity asset holdings shows they always met the limits set by the SIP. The total amount of liquid assets varied modestly over the period, as shown by the Total line at top. The breakdown of assets varied more greatly however, with Tier 1 assets ranging from a low of about 15% to a high of almost 30%. The records of ALMC and FMC minutes do not show any decisions made concerning shifts in asset balance. Data used in this chart were taken from the Exchange Fund Account Monthly Performance and Risk Report issues for the period.

The use of a broad limit structure as the only published guide for EFA asset allocation, combined with limited record keeping of strategic portfolio decisions, results in a portfolio with composition that wanders over time and in no ability to attribute responsibility or accountability for portfolio performance to strategic actions. The asset investment process was previously summarized as determining the cost of any new investable funds and buying securities that provide a greater return, while respecting the EFA asset-class limits. Agreement or disagreement by senior management in the asset class and security decision is subsequently communicated, resulting in an investment process of revealed preferences. This process does not include any explicit optimization algorithms to guide strategic asset allocation decisions. As no optimization process is followed, it cannot be stated that there has been any optimization of returns as required by the third objective of the EFA.

While the governance structure surrounding the AAF is well developed, the actual process to be followed for asset investment activities within the AAF needs to be formally laid out. The investment objective would be to target the maximum expected return for the EFA, given the maximum acceptable amount of risk. This addresses the SIP objective of “optimizing return.” The process would define the roles and responsibilities of all groups involved and should be built around modern portfolio management concepts such as benchmarking and risk-versus-return measurement.

All external central banks interviewed for this evaluation use some form of a benchmark to measure performance of the foreign exchange reserves. The Bank of England’s Centre for Central Bank Studies states:

The most commonly used method by which decisions from one level are passed down to the level below, and by which results of decisions are calculated and passed back up the management chain, is the Benchmark process. A benchmark is a notional or imaginary portfolio constructed to provide a yardstick or baseline against which the return on an actively managed portfolio can be measured.[10]

Furthermore, the Bank of International Settlements in its FX Reserve Management: Trends and Challenges, states:

This Strategic Asset Allocation (SAA) is articulated by selecting a benchmark portfolio and by defining the tolerance ranges within which the actual allocation is allowed to vary. These decisions are usually based on advice provided by risk analysts and other specialists involved in the reserve management process. The trend represents a major shift in practices from a world in which the portfolio was more commonly constructed bottom-up, and in which the role of “drift” in asset allocations was greater.[11]

Also, most large-fund managers make extensive use of benchmarking as a means of guiding investments within the many component portfolios they comprise, and to serve as an objective measure with which to assess performance. Appendix B details the suggested approach for the EFA, based on accepted portfolio management practice, including the concepts of risk/return frontiers and benchmarking.

Most private sector managers and many foreign central banks have taken a more structured approach and added benchmarks to the limit structure defining their portfolios. Benchmarks are constructs that serve to explicitly communicate and measure the asset mix of the desired portfolio. This asset mix is often determined by the most senior levels of the portfolio management team and approved by the owners of the portfolio. Benchmarks act to provide a measurable and investable asset mix for the portfolio traders to target. The AAF would benefit from formal benchmarking for the EFA as a means to more strongly define the responsibility, authority and accountability of the roles of the various groups involved in managing the EFA.

Recommendations

7. A portfolio benchmark should be set for the EFA as part of an annual strategy-setting process to aid in the communication of investment preferences.

8. An enhanced portfolio process that involves targeting optimum expected return for risk taken should be created. This is discussed further in Appendix B.

The evidence collected shows that the AAF has not achieved all of its expected outcomes. The liquidity and capital-preservation objectives have been achieved. There is no explicit process to address optimization-of-return of the EFA, thus it cannot be stated that return optimization has been achieved. Clarity of responsibility and authority and accountability of roles should be improved. To manage a process, it must first be measured. The addition of more defined portfolio objectives in the form of benchmarking would act to address many of these concerns.

Evaluation Issue 4: Has the AAF achieved its objectives efficiently?

The many different ways in which efficiency can be viewed within the AAF can be broadly divided into EFA performance-related topics or into organizational topics.

The commonly accepted metrics for portfolio performance include benchmark comparisons and Sharpe ratio level calculations. The EFA does not have an explicitly set benchmark. That said, the EFA is a relative-value portfolio for which the funding costs of the liabilities serve as a base for the assets to outperform. The EFA has had positive earnings in most years, averaging about $68,000,000 per year from 2004 to 2010, though it experienced losses in 2008, the year of the global liquidity crisis. A measure of efficiency of economic return for the amount of risk taken is the Sharpe ratio. This ratio divides the excess return of a portfolio by the level of risk taken to achieve that return. The Sharpe ratio of the EFA, shown in Figure 6, has usually been below 1, indicating a poor return for the amount of risk taken. However, it should be noted that the foreign reserves portfolios of most countries have negative returns, due to high funding costs. Canada benefits from an advantageous swap market that allows the country to achieve relatively cheap foreign funding. Therefore, a more applicable Sharpe ratio is the measure of excess return produced by a swap-funded EFA over a non-swap-funded EFA, divided by the incremental risk added by the use of the swap-funding mechanism. This measure would provide a clear indication of the added value of funding through the swap market versus the incremental risk of such a strategy. As these measures are not presently estimated, we recommend performing these calculations in order to improve the strategic asset allocation decision-making process. Appendix B addresses this topic more fully. Figure 6: The chart of Sharpe ratios for the annual returns of the EFA from 2004 to 2010 shows that the ratio never exceeded 1 and was most often below 0.2. The Sharpe ratio is a measure of return achieved versus the risk taken and a level of 1 is generally considered adequate portfolio performance. Levels well below 1 suggest that returns have not been commensurate with the level of risk taken. Data used to calculate the Sharpe ratios in this chart were taken from the Exchange Fund Account Monthly Performance and Risk Report issues for the period. The conclusion from a review of the annual return series and the Sharpe ratio of these returns is that the AAF is modestly efficient most of the time. Efficiency for portfolio management organizations is often measured as the costs of managerial operations relative to the funds under administration, expressed as a percentage of assets. This ratio is referred to as the management expense ratio (MER). The table below shows common MER ratios for a variety of asset classes and organization types. True comparison of these figures would be unfair, as some organizations require marketing staff to attract assets, whereas others are managing captive assets. Furthermore, some managers require administrative staff to oversee individual accounts or pension values, while others deal with a single pooled account. The estimated cost of running the EFA, as provided by Department of Finance and Bank of Canada personnel, is approximately$16.5 million per year (including all related operating costs, commissions and fees). Given that the EFA portfolio is approximately $45 billion to$50 billion, the MER for the EFA is exceedingly small by all comparisons, a result of the efficiency of the asset classes in the EFA, the economies of scale inherent in the EFA, the modest size and costs of the Bank of Canada and Department of Finance groups involved in EFA management, the fact that the assets are in a sense captive to the Bank of Canada and Department of Finance, and the management of the assets as a single pooled fund. The true MER for the EFA is unclear, as many groups and personnel involved in managing the EFA have multiple roles outside of the EFA. But assuming the estimate for MER is reasonably accurate, the low operating costs suggest that the AAF may be suffering from a lack of resources relative to its peer group, which may prevent it from achieving higher performance results.

Table 2
MER Ratios for a Variety of Assets and Organizations
Asset Manager MER
Mutual Fund – Equity 2% - 3%
Mutual Fund – Fixed Income 0.5% - 1.5%
Index Fund – Equity 0.35%-0.50%
Index Fund – Fixed Income 0.15%-0.35%
Private Pension Plan –- >$50B 0.35%-0.45% Public Pension Plan – CPP 0.30% Estimated EFA 0.035% The interviews of personnel also revealed an unintended inefficiency concerning the work expectations of the portfolio managers. The use of portfolio managers as researchers and report writers limits their ability to perform their roles of asset managers and relative-value traders. It was recognized that some report generation and research was a necessary part of their job and indeed added to the traders’ knowledge and skill sets. However, there was a consensus that there are too few research-oriented personnel resources available and thus many such tasks were given to traders, to the detriment of return-optimization of the EFA. Recommendation 9. A review of appropriate personnel resource levels should be done with a view to increasing resources, as required, to perform portfolio-optimization development work and to undertake other research and report-generation tasks. During the interview process with Bank of Canada and Department of Finance staff, it was reported that there were occasions when reputational and communications considerations had an influence on the asset holdings of the EFA. These considerations are concerned with market perceptions and reactions resulting from the investing actions of the Government of Canada. It is easy to imagine there could be occasions in which apparent sub-optimal investment decisions are taken as a result of such considerations. This may appear to be a source of inefficiency for the EFA when viewed from the limited perspective of the portfolio manager. Ultimately, however, the responsibility for the EFA portfolio and its performance lies with the Minister of Finance. The Minister has the broadest perspective of the economic well-being of Canada and its currency, and the role of the EFA within that worldview. The implementation of a benchmark process for the EFA would allow the FMC, as delegate of the Minister, to set the strategic focus of the portfolio. The benchmark would include those asset classes that the Minister and his delegates deem appropriate, given all considerations. In exceptional circumstances, where reputational considerations appear to override one or more of the three priorities of the EFA, the strategic benchmark can be modified by the FMC to guide the investing actions of the ALMC and portfolio managers, thus removing the onus from those groups to perform investment actions that appear to be in conflict with EFA priorities. While this may result in lower short-term returns for the EFA, it would be done in the hope that longer-term returns would benefit. Efficiency of the investment process is an important aspect of fund management organizations. The lack of an optimization process to guide the asset allocation process for the EFA as described earlier is potentially the greatest inefficiency in the AAF. Incremental return improvements on the$45 billion to \$50 billion EFA can easily represent millions of dollars of added revenue.

The groups involved in the AAF include the Minister of Finance, FMC, ALMC, RC, and portfolio managers. A review of the minutes of the FMC, ALMC and RC reveals that these groups did act in accordance with their roles, as defined by the various governance documents. The interviews of personnel of all levels, however, revealed a perceived tendency for more junior levels to seek pre-approval or non-disapproval of decisions by authority levels more senior than required, including the FMC. This was done either to seek explicit approval, or the weaker “non-disapproval” for decisions and actions. While this is not in contravention of the governance structure, it is a source of inefficiency as the decision process becomes delayed and encumbered. This also has the effect of diluting responsibility and accountability for decisions taken at all levels. Establishment of a rigorous investment process would lessen the prevalence of this negative behavior. Since many decisions involve the asset mix of the EFA, the use of benchmarks would greatly reduce the negative effects of this process.

Record-keeping of group meetings appears to lack efficiency. Four interview respondents said the ALMC minutes were often incomplete and often circulated just before the following ALMC meeting. Further evidence of incompleteness of minutes was discovered when the minutes were reviewed to determine what decisions were driving the strategic-asset shifts within the EFA. The lack of recorded direction found in this instance once again served to dilute responsibility and accountability for decision making. A similar review of the minutes of the FMC and RC showed a lack of recorded direction in their records of decisions and actions.

Recommendations

10. Minutes of all group meetings should be prepared and circulated in a timely fashion. This typically means within two days following the meeting, so participants’ memories are still fresh concerning the events of the meeting.

11. Strategic decisions concerning changes in the asset mix of the EFA should be recorded in the minutes of the ALMC and FMC meetings as appropriate, and responsibilities for those decisions assigned to appropriate group members.

The asset/liability structure of the EFA is highly influenced by the use of swap funding. The primary profit vector of the EFA is to raise funds at advantageous levels through swap funding in the sovereign name of Canada and to reinvest the funds in higher yielding securities, which may be of lower liquidity or credit quality. The performance-attribution discussions presented in the monthly and annual performance reports focus on items such as yield curve and coupon effect, but don’t touch on the large benefit gained from the use of swap funding. To improve clarity of performance communication, the performance results should include a fulsome performance attribution of the asset/liability mix, including the extensive use of swap funding.

Recommendation

12. EFA performance reports should contain a performance attribution discussion centered on the use of swap funding and credit investing. Included in the discussion should be a quantitative analysis of the risks taken and the returns provided.

The evidence collected suggests that the governance structure of the AAF is efficient, but that some of the processes inside the AAF need to be improved and revitalized. The most pressing need is to enhance and formalize the investment process for the AAF to fully achieve the SIP objective of optimized return for the EFA.

Evaluation Issue 5: Are there alternative methods for achieving the same results?

There are alternative methods for achieving similar, but not identical, results.

The AAF could achieve greater efficiency through adoption of enhanced portfolio management methods, including the use of benchmarking and return/risk ratio targeting. This was discussed above and a process for such is presented in Appendix B.

The AAF could be modelled after those of other central banks. No other country manages their foreign reserves with currency and maturity-matching the way Canada does. Most foreign countries have either a pure-currency portfolio or a mixture of matched and pure portfolios. A table comparing the various central banks interviewed for this evaluation can be found in Appendix C, and case studies of three central banks are presented in Appendix D. In particular, we consider the process of foreign reserves management in the U.K. an interesting model, which could be replicated by Canada. In the U.K., the central bank is responsible for managing the reserves, and reporting regularly to Her Majesty’s Treasury. Treasury is not directly involved in the operational aspects of foreign reserves management.

A final alternative is to employ external portfolio managers to invest a portion of the EFA assets. A number of central banks do this in order to gain access to investment expertise that they lack for certain asset classes. The costs of this are high, as most of the better asset-managers operate as hedge fund managers, who typically charge 2% of assets plus a 20% performance fee. Institutional fund managers perform this service for lower fees, but don’t typically offer the same degree of specialized asset-class skills. Investing assets with external managers would require separating EFA assets and liabilities. It would also raise issues concerning liquidity of funds and confidentiality of divestment activities.

There are alternatives to the current AAF, but most would result in changes that could conflict with the three primary objectives of the EFA and which are likely more expensive than our estimates of the current costs of operating the EFA. The most promising alternative is to update the current AAF to include enhanced portfolio management methods.

## Conclusions

The evaluation, based on multiple lines of evidence, found that the AAF aligned with federal government priorities as well as with its roles and responsibilities, during the period 2000 to 2010.

The evaluation found that the AAF achieved most of its expected outcomes. The expected outcome of the AAF is the performance of the EFA, consistent with its objectives of liquidity, preservation of capital and optimization of returns.

The EFA has always operated within its liquidity limits.

Performance data evidence shows that the EFA has sometimes performed outside of its objective of capital preservation on an absolute basis, but mark-to-market capital losses, when they did occur, were modest, relative to the size of the EFA. It is reasonable to expect that any diversified investment portfolio will occasionally suffer losses. It is highly unlikely that the performance of the EFA was ever “optimal,” as there is no method being used to perform return optimization. An investment optimization process, as discussed in Appendix B, entails forecasting risks and returns for the set of allowable investments, then determining the portfolio asset mix that results in the maximum expected performance.

The evidence collected found that the AAF did not achieve its objectives efficiently. The governance structure of the AAF was found to be efficient, but some of the processes inside the AAF were inefficient and need improvement. The most pressing need is to enhance and formalize the investment process, in order for the AAF to achieve the SIP objective of optimized return for the EFA.

There are alternatives to the current AAF, but most would result in changes that would be in conflict with the three primary objectives of the EFA. The most promising alternative is to update the current AAF to include enhanced portfolio management methods.

## Recommendations

1. The return optimization objective of the EFA should be rewritten to incorporate the concept of targeting expected returns.
2. Total market VaR for the EFA portfolio should be calculated daily. This VaR measure should include all risk elements such as swap funding and credit, currency and duration risks.
3. The use of the risk measure called “Market VaR” should be severely curtailed and a full description of its limitations should be included whenever it is referred to in reports.
4. All VaR-calculated measures of expected risk should be accompanied by appropriate historical measures of the same risk, in order to judge their reasonableness.
5. Portfolio risk limits should be set for the total market VaR for the EFA portfolio. A process of timely reviews of these measures and action responses should be created.
6. Portfolio risk limits should be set for net duration of the EFA portfolio. A process of timely reviews of these measures and action responses should be created.
7. A portfolio benchmark should be set for the EFA as part of an annual strategy setting process to aid in the communication of investment preferences.
8. An enhanced portfolio process that involves targeting optimum return for risk taken should be created. This is discussed further in Appendix B, which also addresses much of the recommendations in 1 and 5 above.
9. A review of appropriate personnel-resource levels should be conducted, with a view to increasing resources, as required, to enable personnel to perform portfolio optimization-development work and undertake other research and report-generation tasks.
10. Minutes of all group meetings should be prepared and circulated in a timely fashion. This typically means within two days following the meeting, so participants’ memories are still fresh concerning the events of the meeting.
11. Strategic decisions concerning the asset mix of the EFA should be recorded in the minutes of the ALMC and FMC meetings, as appropriate, and responsibilities for those decisions assigned to appropriate group members.
12. EFA performance reports should contain a performance-attribution discussion centered on the use of swap funding and credit investing. Included in the discussion should be a quantitative analysis of the risks taken and the returns provided.

## Appendix A Matrix of Results

Summary of Results

Evaluation Issue

Central Bank Interviews

Literature / Data

Are the objectives of the AAF consistent with the priorities of the Government of Canada?

• Yes – based on evidence from the Currency Act and the fact that all respondents answered in the affirmative
• Changes to the governance have been consistent with government priorities
• Decisions are being transmitted in a method consistent with government priorities
• No evidence
• Review of ALMC, RC and FMC minutes indicate a focus on prudent management of the EFA
• Review of internal governance documents indicate a focus on management practices consistent with prudence

Is the AAF aligned with the roles and responsibilities of the Government of Canada?

• Yes – All respondents answered in the affirmative
• The AAF is consistent with the Government of Canada’s roles and responsibilities for fiscal prudence and appropriate decision making
• No evidence
• Review of ALMC, RC and FMC minutes indicate a focus on maintaining liquid reserves and minimizing capital risk
• Review of internal governance documents indicate a focus on maintaining liquid reserves and minimizing capital risk

Has the AAF achieved its expected outcomes?

• Positive developments since 2000 include enhancements in the governance policy and creation of the various committees which oversee the EFA
• Risk oversight has been positive
• Market has changed – sovereign credit no longer regarded as safe – likely requires some changes to governance structure / decision-making processes
• No evidence
• Liquidity ratios (Tier1, Tier 2, UST %, Encumbered ratio) remained consistent throughout crisis
• Encumbered ratio did reach close to 30%, though could result from a large maturity
• Fluctuations in P&L and duration measures resulted from extreme moves in BA/Libor basis and Canadian swap spreads
• Fluctuations were outsized relative to all historical measures, which highlights a weakness of current ALM structure
Has the AAF achieved its objectives efficiently?

Formal decision and action delegation within the governance framework

• Seven respondents indicated that there were levels of uncertainty concerning authority and delegation
• All central banks indicated benchmarks provide clarity of decision making as the benchmark determination process requires preferences to be revealed (further discussion below under the benchmarking topic)
• BIS and IMF reports indicate the need for not only a structure of governance, but also for specific decision-making role-delegation to various levels of management (IMF Report 2004, p. 12: “The internal governance structure of the reserve management entity should be guided by and reflect the principles of clear allocation and separation of responsibilities.”)

Efficiency of communications processes within the management structure

• Four respondents indicated that meeting minutes were not timely and recommendations for senior management were often watered down
• No evidence
• No evidence

The issue of benchmarking the portfolio and its use as a management tool

• Seven respondents indicated that benchmarks would be useful in managing the investment process
• Benchmark use is recommended for credit-risk component of EFA, as current asset-liability framework has embedded benchmark for interest rate and foreign exchange risk
• All central banks interviewed use benchmarks, ranging from public-market benchmarks (New Zealand, Brazil, Colombia) to customized benchmarks (U.K., Denmark, Australia, Switzerland) created internally, depending on nature of reserves
• Several central banks indicated benchmarks were a crucial component of managing the reserves, in terms of communication and delineation of responsibilities
• Colombia indicated that benchmarks create a dialogue with the Board
• IMF and BIS reports, as well as interviews with private sector organizations, also indicate that benchmarking is used as a best-practices method of portfolio management. (IMF Report 2004, p. 17:“Investment benchmarks are an important tool for assessing performance and enforcing the accountability of reserve managers.”)

Risk measurement deficiencies of the EFA portfolio with consideration to swap-funding activities

• No evidence
• All central banks, aside from New Zealand and Colombia, use VaR reports for risk determination and budgeting
• Switzerland uses VaR-based methodology to determine asset mix
• Brazil uses daily benchmark-relative VaR for risk measurement
• Denmark indicated VaR cannot be relied upon for non-linear payoffs
• Credit-VaR methodology used by Bank of Canada should include all instruments and risks
• BIS and IMF reports speak of the need to have timely and meaningful VaR reports in order to manage reserve portfolios. (IMF Report 2004, p. 5:“Risk exposures should be monitored continuously to determine whether exposures have been extended beyond acceptable limits.”)
• VaR-based risk measures produce extremely low values that do not adequately explain the mark-to-market swings observed, and that are inconsistent with the size and asset/liability mix of the portfolio

Profit-attribution reporting, with particular focus on the importance of swap-funding as a primary source of portfolio profitability and risk

• Performance reports do not provide a detailed attribution analysis of profit-and-loss results
• Central banks such as Colombia indicated value to bank staff of performance attribution
• Attribution analysis would aid in the transparency of strategic portfolio decisions, including the profits effects of swap-funding. (BIS Report No. 38, p. 12 “The measurement and evaluation of performance consists of a number of steps: the valuation of the portfolio and calculation of the returns, the attribution of those returns to different decisions and the evaluation proper.”)

The effect of differing risks of the FIN and the BoC, namely reputational and communications risks versus credit risk, on the performance of the EFA portfolio

• Eight respondents indicated conflicting portfolio constraints that limit the actions of portfolio managers and that subject the portfolio to risk of capital loss (for example, the lengthy timeline for determination of the ability to sell certain bonds)
• The U.K.’s foreign exchange reserves: no constraints are imposed by the government (nor is the principal represented on committees), though the process used to manage FX reserves is extremely methodical, and requires regular reporting to the principal
• Brazil is similar to U.K.
• Denmark, Australia, Switzerland: – central bank owns reserves
• No evidence

Liquidity and its implications for mark-to-market accounting for the assets and liabilities of the EFA portfolio

• The primary portfolio objective of liquidity and the purpose of assets being available for intervention should the need arise, implies a mark-to-market accounting standard and a VaR calculation that includes the effects of swap-funding activities
• Several respondents indicated that there is no clear framework for managing resulting asset-liability gaps in the case of intervention. By necessity, both foreign exchange and duration gaps will result from intervention, and it would be helpful for management to have an operating framework to manage these gaps
• Australia indicated that it does not believe in asset-liability management for reserves, since it hampers the ability to intervene; intervening creates an open position, rather than closing a position
• New Zealand runs simulations to determine optimal ratio of hedged and unhedged; in 2007, began using unhedged reserves due to analysis based on modelling of a crisis, its duration and impact, and the resulting losses caused if reserves fully hedged
• U.K. indicated recent increase in unhedged reserves due to swap counterparty credit limits and rationale for unhedged reserves
• Portfolio is currently managed on a matched asset/liability framework that implies a hold-to-maturity investment strategy. (BIS Paper 38, p. 12, “Measurement relies on proper valuation procedures. Regardless of whether portfolios are managed in-house or externally, processes need to be in place to value portfolios. As FX reserve management has become more sophisticated, marking to market for management, as opposed to external reporting, purposes, has naturally become the rule.”)

Reserves are managed in an efficient manner, but perhaps there are opportunity costs to the management

• Current structure lacks incentive and staff to perform relative-value trades
• Private-sector bonus structure would provide incentive, but could also encourage greater risk-taking, which would not be consistent with EFA objectives
• We believe the use of benchmarks will provide greater clarity to portfolio managers and remove some of the disincentive to engage in relative-value trades
• Australia has limited active management of reserves
• All other central banks use active management to increase profitability and market intelligence, and for operational preparedness in the event of intervention
• ECB Risk Management paper discusses the values of active management (p. 50) and the limited additive risk of relative-value trades (p. 41).

Are there alternative methods to achieve the same results more cost-effectively?

• U.K.
• HMT provides input into liquidity model
• Bank of England manages reserves based on liquidity model
• Allows greater efficiency in decision-making process

## Appendix B Investment Management Process

The ideal investment management process involves defining portfolio objectives, setting management group structure and outlining the steps of asset allocation. The AAF already has many defining governance documents, such as the Currency Act, SIP and Funds Management Governance Framework. What remains to be defined is the detailed process used to actually allocate assets within the limit bounds of the SIP using the management structure of the Funds Management Governance Framework.

### Investment Management Process

Figure 7: A diagrammatic illustration of the investment management process in a given year starting with a forecast of risks and returns by asset class. The diagram shows the roles of the key players – FMC, ALMC and Portfolio Managers – in the investment management process throughout the year, as described in this Appendix.

The objective of the investment process is to achieve the optimum portfolio for the SIP, where optimum is defined as the target portfolio that obtains the maximum expected return for the targeted amount of risk to be taken.

The process should be overseen by the FMC at a policy level and managed by the ALMC at a strategic level.

Sometime prior to the start of a new fiscal year, the ALMC should institute an economic forecasting process in which the expected returns and risks for all investable asset classes are prepared. Using these forecasts, a detailed risk map, consisting of the total market VaR of all potential portfolio asset mixes, should be prepared. The set of portfolio asset mixes should be constructed to stay within the limit bounds set by the SIP for the EFA. Expected returns should be prepared for these portfolios as well. The optimum portfolio should then be defined as the one portfolio of all these that obtains the highest return, given the targeted amount of risk.

Once the asset mix of the optimum portfolio has been determined, an index should be set reflecting that asset mix. That index should be reviewed by the FMC and, if found appropriate, set as the strategic benchmark for the EFA portfolio for the coming fiscal year. It may be necessary to reset the index more frequently than annually, should the policies or strategy shift substantially. If so, the annual portfolio review process should be redone in light of these changes, in order to redefine the new optimized portfolio and the benchmark index for it.

Investment decisions to move away from the index in absence of policy changes are strategic and should be reserved for the ALMC or FMC, and recorded as such in minutes of their respective meetings. In the base-case scenario, portfolio managers would make security-specific value decisions, but would not execute trades which shift the portfolio away from asset-class benchmark allocations set by the ALMC. Thus, a single strategic benchmark would be employed in the management of the EFA.

There exists a second option for a further level of accountability and risk taking by introducing the concept of a tactical benchmark. The ALMC would set the asset allocation on a tactical basis and these decisions would be translated into a tactical benchmark. The portfolio managers would be given limits against the tactical benchmark and would be allowed to make security allocations that include deviations from the asset allocation specified by the ALMC. Decisions by portfolio managers would be measured against the tactical benchmark, while the asset allocations set by the ALMC would be measured against the strategic benchmark. In this scenario, a review of the strategic benchmark would occur annually and such decisions would likely determine, in the case of the EFA, allocations to various credit tiers, such as sovereign, supranational and agency debt allocations. The tactical benchmark could be changed by the ALMC as frequently as monthly, based on perceived risk-and-return for the various asset allocations, and would likely involve decisions at a more granular level including targeted allocations on an issuer-specific basis.

The decision to manage the portfolio on an active versus passive basis may affect the choice of whether to use a single strategic benchmark or both a strategic and tactical benchmark. A passive portfolio will not require a tactical benchmark to measure portfolio-manager performance relative to the allocation recommended by the ALMC. On the other hand, an active portfolio does not necessarily require a second benchmark, but that will depend on the level of portfolio activity and the degree of mismatch between the ALMC-recommended portfolio and the actual portfolio.

It is our opinion that a single strategic benchmark should be employed; details of how we reached this decision appear below. Although the current governance process does not require any changes to accommodate this approach to managing the EFA, we believe there are potential alternatives that should be explored, as detailed in Evaluation Issue 5.

If funds are required for liquidity or foreign exchange intervention, the open-currency and duration positions created, including liabilities and assets, should be moved to another tracking portfolio. The core EFA portfolio should be left currency- and duration-matched. The performance and risk of the open-tracking portfolio should be kept separate, so that management results for the core EFA do not get lost in the performance results of the open positions. The net results of the core EFA portfolio and the tracking portfolio will need to be added together to show the total performance of the EFA for the annual Report on the Management of Canada’s Official International Reserves.

### Benchmark Determination

Figure 8: A diagrammatic exposition of the benchmark optimization process that is described in the following paragraph. The diagram demonstrates how the optimal portfolio (strategic benchmark) is derived as an outcome of an optimization process subject to a number of factors – forecast asset-class risks and returns; minimum allocation to sovereigns; cost of swap funding / maximum allocation to swap funding; and liquidity, maturity, currency distribution and risk constraints.

It should be noted as well, that a single benchmark remains appropriate within the confines of the asset-liability matching constraints of the AAF. Once a study has determined the risk/return parameters for swap funding (Recommendation #12), a constraint will be developed which limits swap funding based on results of the study. Further constraints will exist in terms of liquidity, distribution of liabilities among the primary currencies (US dollar, Euro and Yen) and minimum allocations to sovereign credits. Once the constraints can be articulated mathematically, an optimization process can be used which will maximize the net return of the fund subject to the constraints. The optimization process will be used to determine the term, currency and method (swap versus direct issuance) of funding, as well as the ideal credit mix for the fixed income assets.[12]Thus, the optimization will focus on relative credit spreads between potential assets and potential liabilities across term and currencies subject to the above constraints.

Single Benchmark Versus Multiple Benchmarks

In Borio (2008), reference is made to vertical tiering; that is, the establishment of two or three levels of responsibility for management of foreign exchange reserves. In a three-tiered structure there is a natural desire to establish benchmarks to divide the tiers. This would entail the creation of a strategic benchmark and a tactical benchmark. In the case of the EFA, for example, the strategic benchmark would be set by the Funds Management Committee, and the Asset Liability Management Committee would set the tactical benchmark. Thus, the ALMC would be evaluated on its decision to deviate from the strategic benchmark and the portfolio managers would be measured against the tactical benchmark.

We have chosen to recommend the establishment of a single benchmark, based on the experience and investment structure of other central banks. Switzerland uses both a strategic and tactical benchmark for purposes of asset allocation. Within asset classes, public benchmarks are used to measure relative performance of the portfolio. It is our belief that this structure works well for the Swiss National Bank due to the many asset classes available for investment and the fact that these investments do not require customization. Denmark, on the other hand, initially established both a strategic benchmark and a tactical benchmark, but recently abandoned the tactical benchmark due to the amount of time required to maintain such a benchmark. The benchmarks that Denmark established were customized for the constraints imposed on their foreign exchange reserves portfolio and little additional information was gained by maintaining a second benchmark.

Canada’s EFA has a very narrow scope for potential investments and requires significant customization of investment due to the asset-liability management framework. Although, in our opinion, there is much to be gained in terms of structure and decision-making discipline by setting two benchmarks for the three-tiers of responsibility for the EFA, based on Denmark’s experience it may not be an efficient use of EFA resources.

## Appendix C Foreign Central Bank Comparison

Foreign Central Bank Comparison: Foreign Exchange Reserve Management

Central Bank

Owner of Foreign Exchange Reserves

Use of Asset-Liability Management

Use of External Fund Managers

Canada Government – Decisions made by joint committees and ultimately Minister of Finance For Foreign Exchange and Interest Rate risk None
United Kingdom Government – HMT is responsible for reserves policy but it delegates management to BoE. HMT and BoE meet quarterly, but HMT not represented on operational committees. BoE manages reserves, provides policy advice to HMT, and arranges financing in foreign currencies. Reserves policy is assessed under an ALM framework but decision taken to hedge FX and IR risk for a portion of portfolio only; this portion has grown since the recent financial crisis. None
New Zealand Government FX and IR risk – amount hedged is based on quantitative model (was fully hedged until 2007) None
Denmark Denmark’s National Bank No – assets are used to keep currency within a range U.S. mortgage bond portfolio
Australia Reserve Bank of Australia No Asian bond funds
Brazil Government, but not represented on BCB Board or Investment Committee FX only – recently loosened this policy Fixed Income, FX, Corporate Bonds, Equity Indices, Commodities
Colombia No explicit ownership – managed by central bank and registered in its balance sheet FX constraint on benchmark construction Fixed Income – high percentage
Switzerland Swiss National Bank No None
Foreign Central Bank Comparison: Allowable Currencies and Assets

Central Bank

Allowable Currencies

Allowable Assets

• USD, EUR, JPY
• Bonds – sovereign, supranational, agencies
• Commercial paper, CDs
• Swaps (IR & FX), FRAs, FX Forwards
• SDRs

United Kingdom

• Primarily USD, EUR, JPY
• Gold
• Bonds – sovereign, supranational, agencies
• Swaps (IR & FX), FRAs, FX Forwards, futures
• CDs and bank CP (being phased out)
• SDRs

New Zealand

• was limited to USD, EUR, JPY
• Bonds – sovereign and supranational
• Deposits, CDs

Denmark

• mostly EUR
• Gold (predefined)
• Foreign Bonds
• Bank deposits (secured and unsecured)
• SDRs

Australia

• USD, EUR, JPY
• Gold
• Bonds – sovereign, supranational and agencies
• FX Swaps, FX Forwards

Brazil

• USD, EUR, JPY, GBP, AUD, CAD, DKK, SEK
• Gold
• Sovereign debt and deposits

Colombia

• CAD, AUD, NZD, JPY, EUR, GBP, CHF, SEK, DKK, NOK
• Sovereign, supranational, agencies (incl. MBS)
• FX Forwards
• Deposits, CDs and bank CP
• Corporate debt

Switzerland

• EUR, CAD, GBP, USD, JPY, AUD, SEK, DKK, SGD
• Gold
• Sovereign and corporate bonds
• Equities

Foreign Central Bank Comparison: Risk Management and Benchmarks

Central Bank

Risk Management

Benchmarks

• Use of 10-day VaR for credit
• ALM sets limits on FX, IR
• None

United Kingdom

• Use 10-day 99% VaR
• Customized
• Unhedged currency benchmark is 40% USD, 40% EUR and 20% JPY
• Issuer allocation based on liquidity model

New Zealand

• Limitations on liquidity, duration, credit
• Minimal active management
• Public benchmarks with FX determined by model

Denmark

• Use VaR (analytical and historical) with 95% probability of loss not exceeding 3-4B DKK annually
• Interest rate limits set by limit on dollar duration and key-rate duration
• Had 2 benchmarks – strategic and tactical
• Recently stopped using tactical benchmark – took up valuable resources, unclear additional benefit from two benchmarks
• Customized benchmark based on required term to maturity

Australia

• VaR based on historical model and correlations
• Dollar duration, asset allocation, currency and credit limits
• Customized benchmark

Brazil

• Benchmark-relative VaR
• Limit maturity of fixed income portfolio to 3 years, to limit potential for loss
• Public benchmarks – liquid, replicable, standard

Colombia

• Duration and FX limits
• Tracking error for active management
• Credit limits (sector and issuer)
• Composition of public benchmarks
• Currency composition determined by Foreign Reserves Committee
• Duration and asset class constraint determined by model

Switzerland

• Use of VaR and cVaR with risk simulations
• Official benchmark with customized maturity weightings

Foreign Central Bank Comparison: Responsibility

Central Bank

Committees/Departments Responsible

• Department of Finance
• Funds Management Committee
• Asset/Liability Management Committee
• Risk Management Committee

United Kingdom

• HMT
• Executive Director of Markets at BoE
• FED – Foreign Exchange Division – execution and issuance
• RMD – Risk Management Division – risk analysis, methodologies and limits
• MSROD – Markets Strategy and Risk Operations Division – operational risk, static data and risk reporting

New Zealand

• Governor – ultimate strategy and policy with advice from ALCO
• Reserve Bank Board
• ALCO
• Financial Markets Department
• Risk Management Department

Denmark

• Market operations – day to day PM, portfolio composition
• Financial Markets – risk mgmt, outline strategies for Board, policies that implement Board’s decisions
• Accounting – back office, controls
• Board of Governors – determine policies and make strategic decisions

Australia

• Governor – benchmark risk profile and risk delegation
• International Department (front office)
• Credit Committee
• Risk Management Committee
• Internal audit and External audit

Brazil

• BCB Board of Governors
• Investment Policy Committee
• Depin (operations – front, middle, back office)

Colombia

• Foreign Reserves Committee (policies)
• Bank’s Internal Foreign Reserves Committee (operational framework)
• Foreign Reserves Department (implementation)

Switzerland

• Governing Board - determine investment strategy
• Investment Committee – set tactical asset allocation
• Implementation by Portfolio Managers

## Appendix D Case Studies

The following case studies have been chosen because they provide a diverse selection of how other central banks manage foreign exchange reserves. It should be noted that these central banks may have different constraints from those employed in the management of the Exchange Fund Account (EFA); in the individual case studies we list these differences.

### Swiss National Bank

The Swiss National Bank (SNB) has seen its reserves quadruple since the onset of the financial crisis, from 50B to 200B CHF. The increase in reserves resulted from an attempt to slow the increase of the Swiss franc against the euro and to provide support to Swiss banks requiring foreign currency. Portions of the intervention funds have been diversified out of euro into U.S. dollars, pound sterling, Canadian dollars, Japanese yen, Swedish kroner and Danish kroner; these currency exposures are left unhedged. Prior to the financial crisis, investment limits were determined by currency, but as the European credit crisis has progressed, additional limits have been established by country of issuance. Reserves are managed with three priorities in mind: liquidity, safety of funds and return on investment. Although return on investment is the lowest priority, changes were made to portfolio governance in 2005 to allow allocations to equities and corporate bonds, since they provide diversification to portfolio risk. The SNB uses a risk model that runs simulations, to determine the portfolio with the lowest risk. The model begins with a portfolio of U.S. and Euro sovereign bonds then gradually relaxes asset constraints by allowing investment in additional currencies and asset classes. The model optimizes the asset allocation by finding the portfolio with the lowest acceptable risk exposure. Equities and corporate bonds, while risky as individual asset classes, provide sufficient diversification to reduce portfolio risk according to the SNB model. In fact, the predominant source of portfolio risk is found in the currency exposure. The model’s output serves as a benchmark for the asset allocation of the fund. Individual asset classes are benchmarked against publicly available benchmarks. The Governing Board approves the investment strategy, which includes the strategic asset allocation as well as benchmarks for the portfolios. The Investment Committee establishes the tactical benchmark by deviating, within specified limits, from the strategic benchmark. Portfolio manager’s asset allocation can differ from that of the tactical benchmark, although the limits for doing so are substantially smaller than those afforded to the Investment Committee.

The SNB management differs from that of Canada in several key areas:

• SNB does not employ an asset liability management framework.
• Allocations to corporate asset classes are not permitted by Canada.
• SNB employs a structured framework of risk versus return to determine strategic asset allocation.
• SNB uses two tiers of benchmarks for the portfolios.

Note that the SNB owns its reserves and does not adhere to a strict ALM framework; instead, it manages with a strong foundation of maximizing return over risk.

### Reserve Bank of New Zealand

The Reserve Bank of New Zealand (RBNZ) utilizes reserves to provide stability to the New Zealand dollar (NZD) FX-rate. Prior to 2007, the RBNZ managed reserves by using an asset-liability management framework, similar to that currently underlying the Bank of Canada’s management of the EFA. In 2004, the RBNZ initiated a study of the framework, specifically focused on the impact of intervention and its interference in the asset-liability management framework. The study tested the reserves in times of crisis, including scenarios such as a long-term crisis in which the NZD continues to depreciate, and an inability of the bank to refinance foreign debt. The study raised concerns regarding the efficiency of reserves when hedges must be unwound early, and the significant losses precipitated by a long-term decline in the NZD. Foreign exchange reserves had been compared to insurance, with the cost of carrying reserves similar to insurance premiums, yet the outsized losses created by these stress tests were not consistent with this model. The RBNZ re-evaluated the entire ALM framework, and after assigning probabilities to the occurrence of a crisis, the duration of the crisis and the cost of refinancing in a crisis, created an optimization process to determine the allocation to unhedged reserves. While differences between Canada and New Zealand are many, we would encourage the Bank of Canada and Department of Finance to undertake a similar study to that of the RBNZ[13].

RBNZ’s management of foreign exchange reserves uses publicly available benchmarks, which provide a basis for analyzing strategic asset allocation versus actual asset allocation. The performance is measured against the benchmark and the liabilities, while the unhedged portion is managed with use of an overlay strategy. Assets are restricted to government and supranational issued bonds, with an assigned minimum allocation to government bonds. Portfolio managers are always aware of the risk-allocation available. The Governor of the RBNZ is ultimately responsible for the reserves, which are owned by the central bank. The Governor delegates responsibility, within acceptable ranges, for reserve management and is advised by the Asset Liability Committee, which determines risk management policies and monitors policy adherence and portfolio performance. The head of the Financial Stability Department is responsible for benchmarks, compliance with risk management and portfolio management.

The management of foreign exchange reserves in New Zealand differs from EFA management in the following ways:

• RBNZ does not utilize an asset-liability management framework for its entire reserve portfolio.
• RBNZ does not use cross-currency swaps in managing the portfolio.
• RBNZ uses publicly available benchmarks.

### Bank of England

The U.K. management of reserves is most similar to that of Canada in terms of the ownership of reserves and the ALM framework used. In both countries the government owns the foreign exchange reserves, though they are managed by the central bank. In Canada, based on the ALM framework, all assets are hedged relative to liabilities, whereas in the U.K., the percentage of hedged assets has fluctuated. The Exchange Equalisation Account (EEA), the U.K.’s foreign exchange reserves, owns gold as well as Special Drawing Rights (SDRs) issued by the IMF on behalf of member countries. The majority of the remaining assets were hedged prior to the financial crisis, primarily through the use of cross-currency swaps. Following the onset of the crisis, there was a slow-down in the use of cross-currency swaps, due to its credit intensity, and a larger percentage of the assets were held unhedged (below 50%)[14]. Since that time the U.K. has increased swapping out of sterling to increase the percentage of hedged assets, although strategically the U.K. will always hold a portion of assets unhedged. The EFA experience through the credit crisis is very similar to the EEA in regard to its liability financing, with the exception that the EFA never ran a portion of its assets unhedged.

There are, however, many differences between the EEA and EFA in terms of governance. The Bank of England (BoE) manages reserves on behalf of Her Majesty’s Treasury (HMT) under a Service Level Agreement (SLA), which specifies how the benchmark should be determined. A key part of that is the use of a liquidity model. The model’s input includes holdings by issuer, issuer returns, liquidation costs for each issuer by amount, liquidation events (size, duration to liquidate) and probability of liquidation. The investment universe consists of sovereign, supranational and government-backed agency bonds, with minimum allocations for some assets and maximum allocation based on dollar amounts, not percentages, since liquidation costs are a function of liquidation size. The optimization of portfolio liquidity determines the allocation, which minimizes the probability-adjusted cost of divesting portfolio assets during a liquidity event, versus the asset return if liquidation is unnecessary. The output allocates investments by issuer-country, along with a predetermined maturity profile for investments in those countries. In order to minimize portfolio turnover, the model allocates investments for coupon and principal payments, but does not reallocate current investments. The model’s output forms the benchmark for the portfolio. In implementing the benchmark, BoE staff must purchase the cheapest bonds that meet the benchmark criteria. The actual management of reserves includes active management, based on benchmark-relative risk budgets. Credit risk is monitored for both actual investments and for counterparties with whom the BoE conducts transactions.

HMT oversees management of the reserves by determining the amount of reserves to be held; as well as the investment strategy, including oversight, of the liquidity model. HMT will also be actively involved in any important risk-management decisions. The BoE and HMT maintain regular informal contact at all levels, as well as meeting quarterly to discuss strategic issues. However, the BoE manages the funds and HMT is not represented on the operational committees overseeing EEA management, as follows:

The Bank reports to the Treasury on size, composition and liquidity of the reserves and their continued consistency with policy objectives, investment performance, returns made and risk exposures, including those relating to active management, every month and at a quarterly meeting chaired by the head of the Treasury’s Debt and Reserves Management team. Every six months there is a meeting at which the EEA Accounting Officer and the BoE’s Executive Director for Markets, or delegated senior officials, review investment performance and discuss strategic issues relating to the reserves.[15]

There remains an ongoing dialogue between the BoE and HMT, as HMT must approve any changes to the strategic management of reserves. The cited advantages of the U.K.’s management of reserves are a clear delineation of responsibilities, where the BoE manages the operations of the EEA and HMT is responsible for policy. One interviewee posited that the current structure is “better than a private sector SLA (Service Level Agreement).”

Key differences between the EEA and the EFA are the following:

• Although the EEA employs an asset-liability framework, only a portion of assets are hedged relative to liabilities.
• The BoE uses a liquidity model, which optimizes allocation to assets based on the ability to liquidate and the probability of such an event.
• The BoE uses a customized benchmark for measuring EEA performance.
• The EEA’s operational management of reserves lacks government representation.

## Appendix E Literature Review

Literature Related to Management of Reserves by the Bank of Canada

• Treasury Risk Management Framework for the Government of Canada, Department of Finance, Bank of Canada, January 2008
• Draft: Funds Management Governance Framework, Department of Finance, Bank of Canada, December 2010
• Romanyuk, Yuliya. Asset-Liability Management: An Overview, Bank of Canada discussion paper, August 2010
• Statement of Investment Policy for the Government of Canada, January 2008
• Swap Management Policy for the Government of Canada
• Report on the Management of Canada’s Official International Reserves, April 1, 2009 - March 31, 2010
• Report on the Management of Canada’s Official International Reserves, April 1, 2008 - March 31, 2009

Literature Related to Central Banks for Interviews and Case Studies

• New Zealand: Research of all available documents online including:
• Hayward, Tore, McKenzie, Paxton and Potter, Warren. Managing New Zealand’s Foreign Reserves, Financial Markets Department.
• Foreign Reserves Management in Briefing on the Reserve Bank of New Zealand, August 2002.
• Archer, David and Halliday, Jerse. The Rational for Holding Foreign Currency Reserves, Financial Markets Department in Reserve Bank of New Zealand: Bulletin Vol. 61 No.4.
• Reserve Bank of New Zealand Risk Management Document, General Policies, updated March 8, 2010.
• Background information on the Reserve Bank’s proposal to extend the purpose for which it holds foreign exchange reserves. Dated March 17, 2004.
• Governor’s mandate for the management of the foreign reserves portfolio
• Letter to Minister of Finance to Increase Foreign Exchange Reserves, dated February 9, 2004.
• Foreign Exchange Intervention Options, Ref # 147138
• 2007 Treasury Report: Reserve Bank Foreign Reserves
• Letter to Minister of Finance describing new operating regime for Foreign Exchange Reserves, dated June 19, 2007.
• Balance Sheet Review Q&A
• England: research of following available documents
• Outline of the U.K.’s Official Reserves of Foreign Currency and Gold dated June 2008.
• Debt and Reserves Management Report 2008-09, HM Treasury.
• Exchange Equalisation Account: Report and Accounts 2009-10, HM Treasury
• Halme, Liisa, Hawkesby, Christian, Healey, Juliette, Saapar, Indrek, and Soussa, Farouk Financial Stability and Central Banks: Selected Issues for Financial Safety Nets and Market Discipline, Centre for Central Banking Studies, Bank of England, 2000.
• Switzerland
• Swiss National Bank International Investment Position in 2009.
• IMF Country Report No. 10/140, May 2010.
• SNB Investment of Assets.
• Pierdzioch, Christian and Stadtmann, Georg. Effectiveness of the Intervention of the Swiss National Bank: Kiel Working Paper No. 1160 , April 2003.
• Economic Situation and Monetary Conditions in Switzerland, speech given by Philipp Hildebrand, June 17, 2010.
• Macroprudential Instruments and Balance Sheet Risks, speech by Thomas Jordan, December 16, 2010.
• The Challenge of the Global Economic Crisis, speech given by Thomas Jordan, February 1, 2011.
• Swiss Monetary Policy and Target2-Securities, speech given by Jean-Pierre Danthine, December 16, 2010.
• Australia
• Brazil
• Colombia
• Foreign Reserve Management, Central Bank of Colombia, March 2009.
• Denmark
• Financial Management at Danmark’s Nationalbank, Danmarks Nationalbank, 2004.
• Danmarks Nationalbank Report and Accounts, 2009.
• Kjoer Jensen, Peter. The Foreign Exchange Reserve, 1999.

Literature Related to Best Practices for Exchange Reserve Management

• Risk Management for Central Bank Reserves, edited by Carlos Bernadell, Pierre Cardon, Joachim Coche, Francis X Diebold, and Simone Manganelli, European Central Bank, May 2004.
• Nugee, John. Foreign Exchange Reserves Management, Handbooks in Central Banking No. 19, Centre for Central Banking Studies, Bank of England
• Gray, Simon T. Liquidity Forecasting. Centre for Central Bank Studies, Handbook No. 27
• Borio, Claudio, Ebbesen, Jannecke, Galati, Gabriele and Heath, Alexandra. BIS Papers No 38 FX Reserve Management: Elements of a Framework, Monetary and Economic Department, March 2008.
• Borio, Claudio, Galati, Gabriele and Heath, Alexandra. BIS Papers No 40 FX Reserve Management: trends and challenges. Monetary and Economic Department, March 2008.
• Guidelines for Foreign Exchange Reserve Management, and accompanying documents. Prepared by the Staff of the International Monetary Fund

Private Sector Interviews

• Interview with Senior Vice-President, Head of Fixed Income at Canadian Fund Manager on management of segregated funds and governance thereof
• Interview with former manager of Balance Sheet risk at major Canadian Bank
• Review discussions of governance at previous employers
• investment bank;
• hedge fund;
• pension fund.

## Appendix F Acronyms

AAF = Asset Allocation Framework

ALCO = Asset Liability Committee

ALM = Asset Liability Management

ALMC = Asset Liability Management Committee

BIS = Bank of International Settlements

BoE = Bank of England

CD = Certificate of Deposit

CP = Commercial Paper

cVaR = Credit Value-at-Risk

FIN = Department of Finance (Canada)

EEA = Exchange Equalisation Account (UK’s foreign exchange reserves)

EFA= Exchange Fund Account (Canada’s foreign exchange reserves)

FMC = Funds Management Committee

FMCC = Funds Management Coordinating Committee

FRA = Forward Rate Agreement

FX = Foreign Exchange

HMT = Her Majesty’s Treasury

IMF = International Monetary Fund

IR = Interest Rate

PM = Portfolio Manager / Portfolio Management

RBNZ = Reserve Bank of New Zealand

RC = Risk Committee

SDR = Special Drawing Right (notes created by the IMF based on major world currencies)

SIP = Statement on Investment Policy

SNB = Swiss National Bank

VaR = Value at risk

## Appendix G Departmental Response and Action Plan

The Department is pleased to note that the evaluation report submitted by Forethought Risk concludes that the current objectives of the asset allocation framework for the Exchange Fund Account are consistent with Government of Canada priorities and are aligned with the roles and responsibilities of the government in its funds management function. The evaluation concluded that the asset allocation framework achieved its expected outcomes of liquidity and capital preservation and could be enhanced to better target return.

The evaluators noted that the governance structure for making asset allocation decisions is well developed; however, they suggest that decision making could be improved by clarifying investment preferences and accountabilities, and through more timely and detailed documentation of decisions taken.

The report finds that the asset allocation practices of other sovereign and central bank reserve managers differs from that of Canada, owing to the asset-liability matching nature of Canada’s reserves portfolio. Nevertheless, the report concluded that Canada’s asset allocation practices could be improved by adopting several portfolio management methods employed by large institutional investors, such as evaluating risk and returns of investable asset classes, defining risk tolerances, and establishing and using benchmarks in a more systematic manner.

Forethought Risk has provided constructive recommendations to improve the decision-making framework for Exchange Fund Account investing and funding activities. The Department of Finance agrees or agrees in principle with all of Forethought Risk’s recommendations and has begun, in collaboration with the Bank of Canada, to address these.

In the accompanying table of recommendations, management responses, and planned actions, the twelve recommendations by Forethought Risk are grouped into two categories: Category 1, with emphasis on the framework of asset allocation, its governance, investment process, internal and external accountability; and Category 2, with emphasis on its risk reporting, risk models and control mechanism.

Category I: Governance, Investment Process, and Accountability
Recommendation Management Response Action Steps Lead Date
1.[16]The return optimization objective of the EFA should be rewritten to incorporate the concept of targeting expected returns. The Department agrees with this recommendation.
Comment: The EFA portfolio is managed with a high priority placed on maintaining appropriate levels of liquidity and preserving portfolio capital. As such, investment returns of the EFA have not been targeted ex ante; in addition, the tools to optimize and measure return subject to liquidity and capital preservation are currently not employed.
As part of work undertaken to assess return-based benchmarks and how they could be used in the investment process to influence asset allocation decisions, incorporate targeting expected returns.
1) governance documents to clarify objectives; and
2) the investment process
Department of Finance and Bank of Canada The work to update the governance documents to clarify objectives will be completed by the end of 2012-13.
The work to update the investment process will be completed by the end of 2013-14.
7. A portfolio benchmark should be set for the EFA as part of an annual strategy-setting process to aid in the communication of investment preferences. The Department agrees with this recommendation.   Work will be undertaken to develop a benchmarking approach in the asset allocation framework, and to assess the appropriate use of benchmarks for the purpose of the EFA.   Department of Finance and Bank of Canada Work will be completed by the end of 2013-14
8. An enhanced investment management process that involves targeting optimum return for risk taken should be created, as discussed in Appendix B. The Department agrees with this recommendation.     Work will be undertaken to design and develop an enhanced investment management process.     Department of Finance and Bank of Canada Work will be completed by the end of 2013-14
10. Minutes of all group meetings should be prepared and circulated in a timely fashion. This typically means within two days following the meeting, so participants’ memories are still fresh concerning the events of the meeting. The Department agrees with this recommendation.   Decision items of meetings will be prepared and circulated within two business days. Process for integrating comments on the minutes will be shortened. Department of Finance and Bank of Canada Work was completed in January 2012
11. Strategic decisions concerning the asset mix of the EFA should be recorded in the minutes of the ALMC and FMC meetings as appropriate, and responsibilities for those decisions assigned to appropriate group members. The Department agrees with this recommendation.   1) Minutes will record decisions and non-decisions taken on the investment and funding activities for the EFA.
2) In addition, an investment plan that is prepared on an annual basis will also document strategic decisions.
Department of Finance and Bank of Canada Work was completed in January 2012
9. A review of appropriate personnel-resource levels should be conducted, with a view to increasing resources, as required, to enable personnel to perform portfolio optimization-development work and undertake other research and report-generation tasks. The Department agrees with this recommendation.     A review of personnel resources, training requirements, and analytical tools will be conducted in the context of the design and implementation of the enhanced investment management process of Recommendation 8. Department of Finance and Bank of Canada Work will be completed by the end of 2013-14

Category II: Risk Reporting, Risk Models and Control Mechanism
Recommendation Management Response Action Steps Lead Date
2. Total market VaR for the EFA portfolio should be calculated daily. This VaR measure should include all risk elements such as swap-funding and credit, currency and duration risks. The Department agrees in principle with this recommendation. 1) A market VaR for the EFA portfolio assets and liabilities will be calculated daily.
2) Work will be undertaken to investigate why the current market VaR measures do not reflect actual (or historical) profit/loss measures of the portfolio and to assess missing risk factors. Costs and benefits of implementing a total market VaR model or an alternative metric that encompasses all risk elements of the portfolio will be assessed
3) Subject to the outcome of the above assessment, a total market VaR model that captures all market risk elements, including swap-funding, credit, currency and duration risks, would be developed.
These actions steps also address Recommendation 3.
Bank of Canada Work will be completed by the end of 2013-14
3. The use of the risk measure called “Market VaR” should be severely curtailed and a full description of its limitations should be included whenever it is referred to in reports. The Department agrees with this recommendation. Steps have been taken to note the limitations of the Market VaR measure in reports. Bank of Canada Work was completed in October 2011
4. All VaR calculated measures of expected risk should be accompanied by appropriate historical measures of the same risk, in order to judge their reasonableness. The Department agrees with this recommendation.   Actual profit and loss (historical measures) will be added to all VaR-based measures in reporting. Bank of Canada   Work will be completed by the end of 2012-13
12. EFA performance reports should contain a performance-attribution discussion centered on the use of swap funding and credit-investing. Included in the discussion should be a quantitative analysis of the risks taken and the returns provided. The Department agrees with this recommendation.   1) Work will be undertaken to include performance attribution numbers for swap funding in the Monthly Performance and Risk Report.
2) Work will be undertaken to develop quantitative analysis of risks and returns of the portfolio.
Bank of Canada   The work on performance attribution will be completed by the end of 2012-13
The work to develop quantitative analysis of risks and returns of the portfolio will be completed by the end of 2013-14
5. Portfolio risk limits should be set for the total market VaR for the EFA portfolio. A process of timely reviews of these measures and action responses should be created. The Department agrees in principle with this recommendation.   Subject to the outcome of the total market VaR model assessment, work would be undertaken to assess the appropriate use of risk limits (or thresholds) on the total market VaR.   Bank of Canada   Work will be completed by the end of 2013-14
6. Portfolio risk limits should be set for net duration of the EFA portfolio. A process of timely reviews of these measures and action responses should be created. The Department agrees in principle with this recommendation.   Net duration will continue to be reported in the Monthly Performance and Risk Report. Work will also be undertaken to assess the use of risk limits (or thresholds) for managing market risk. Bank of Canada   Work will be completed by the end of 2013-14

1.The extent to which the AAF achieved its objectives/expected outcomes (its performance) is a measure of its effectiveness.

2. Exchange Fund Account Annual Report 2000, p. 1

3. Exchange Fund Account Annual Report 2000, p. 2

4. Exchange Fund Account Annual Report 2004 (web-based, no page number available)

5. Report on the Management of Canada's Official International Reserves: 1, web-based, no page number available.

6. Funds Management Governance Framework 2008, p. 2

7.  Funds Management Governance Framework 2008, p. 21

8. Finance Canada. Report on Plans and Priorities 2011-2012, pp. 9–12 and 25.

10. Nugee, John Foreign Exchange Reserves Management, p. 17.

11. FX Reserve Management: Trends and Challenges: BIS Papers No. 40, May 2008, p. 7.

12. Note that the term and currency for the assets will be the same as that of the funding (liabilities).

13.  Note that such a suggestion falls outside the scope of the current evaluation.

14. HM Treasury, Exchange Equalisation Account: Report and Accounts 2009-10, July 2010, p 5.

15. HM Treasury, Exchange Equalisation Account: Report and Accounts 2009-10, July 2010, p 3.

16. Numbers in the recommendation column correspond to Forethought Risks’s recommendations in the evaluation report.