Bank of Israel

Foreign Currency Department

Investment of the

Foreign Exchange

Reserves

Annual Report 2006

2

Contents 1. THE LEVEL OF THE RESERVES AND THEIR MANAGEMENT

FRAMEWORK 4

a. The level of the reserves 5

b. The investment policy of the reserves 7

2. THE HOLDING PERIOD RATE OF RETURN AND THE RISK OF THE

RESERVES PORTFOLIO RELATIVE TO THE BENCHMARK 14

a. The holding period rate of return on the reserves portfolio 14

b. Rate of return and risk in terms of the benchmark 16

c. The contribution of ongoing management of the reserves by

component 21

1) Contribution of currency management 21

2) The contribution of management of duration and dispersion

along the yield curve 22

3) Intermarket positions 23

4) Contribution of asset selection 23

d. Control of credit risk in ongoing management 24

3. YIELD ON THE RESERVES PORTFOLIO RELATIVE TO OTHER

MANAGED PORTFOLIOS 26

4. THE LIQUIDITY OF THE RESERVES 29

3

Tables

1. The Level of the Reserves Relative to Other Aggregates, 1997-2006

2. The Performance of the Actual Portfolio vis-à-vis the Benchmark Portfolio,

1997-2006

3. Contribution of Management Decisions to the Yield Spread, vis-à-vis the

Benchmark, 2006

4. The Contribution of Asset Selection, 2006

Figures

1. Gross Foreign Exchange Reserves, 1997–2006

2. Yield and the Total Management Contribution, 1997–2006

3. Distribution of Yields of the Portfolio and the Benchmark, 1997–2006

4. Yield spreads vis-à-vis the Benchmark, 1997–2006

5. Tracking Error of Active Management of the Reserves, 2005-2006

6. VaR of the Positions, June 2002–December 2006

7. Duration Positions in the Total Portfolio, 2005-2006

8. Asset Distribution of the Reserves Portfolio, 2006

9. Exposure to Banks and Governments, 1997–2006

10. Performance Distribution of Managers of Short- and Medium-Term Funds in

the US Market, 1997–2006

11. Yield and Risk: the Dollar Portfolio vis-à-vis Funds in the US Market, 1997–

2006

12. Liquidity of the Reserves Portfolio, 2001–2006

Boxes

1. Major Investment Decisions Implemented in 2006

4

Investment of the foreign exchange reserves in 2006––Main developments1

♦ The average level of the Bank of Israel's foreign exchange reserves in 2006 was

about $28 billion,2 equal to 4.8 months' worth of imports, and constituting about

75 percent of Israel's short-term foreign debt (a fall of 0.2 months and 4

percentage points, respectively, on 2005).

♦ There was no change to the framework of the management of the reserves, which

was based on the possible uses of the reserves and their benefit to the economy.

The management framework was under review in 2006, a process which continues

into 2007.

♦ The holding period rate of return on the reserves in terms of the numeraire in 2006

was 3.8 percent, up from 2.6 percent in 2005.3 The return reflected the rise in

yields to maturity in the capital markets in which the reserves were invested

during the year, where rising interest income was accompanied by capital losses.

♦ In 2006 the holding period rate of return on the reserves was 12 basis points

higher than that of the neutral benchmark, a spread that reflects the contribution of

active management of the portfolio. As in previous years, most of the incremental

yield derived from asset selection.

♦ The exposure of the reserves to the world banking system averaged 32 percent of

the reserves this year. This exposure is managed under a system of quotas and

rules, which plays a central role in credit-risk management of the portfolio.

♦ The reserves have a very high liquidity: about 92 percent of the portfolio is

invested in liquid assets, and the rest in assets with lower liquidity.

1. THE LEVEL OF THE RESERVES AND THEIR MANAGEMENT FRAMEWORK

The management of Israel's foreign exchange reserves is subject to the Bank of Israel

Law, 5714–1954 and its accumulated legal interpretations. These define how the 1 The Foreign Currency Department has published reports on the investment of the foreign currency reserves since 2000. For reports on earlier years, part of which were covered by sections in the Bank of Israel's Annual Reports, see the Bank's website www.bankisrael.gov.il. These earlier reports contain a glossary of terms used in this report, and discussions on various aspects of management of the foreign exchange reserves that do not appear in the current report. 2 The level of reserves is calculated on the basis of the daily balances of the reserves assessed at their market value. 3 All holding period rates of return in this report are in terms of the numeraire, unless stated otherwise.

5

Bank may conduct its foreign-currency activities and set limitations on the types of

assets in which it may invest. In areas in which the Bank is not restricted by the

wording of the Law, it acts within a framework that reflects the spirit of the Law and

the Bank's priorities, and limits the various risks to which the reserves portfolio is

exposed. The main financial risks are credit risk, controlled by a system of rules and

quotas; interest–rate risk, controlled mainly by setting a target average duration for

each currency portfolio; and currency risk, controlled by defining a neutral currency

composition for the portfolio––called the numeraire––that serves as a yardstick for

measuring the performance of the actual portfolio. Other risks are also taken into

account, such as operational and legal risks.

a. The level of the reserves

At the end of 2006 the reserves stood at $29,028 million, a rise of $1,170 million from

their level at the end of 2005. An increase of about $1.8 billion resulted from the

current income from investment of the reserves (interest and capital gains) and from

the strengthening of non-dollar currencies in the numeraire, in which some of the

reserves are invested. Against this, the activities of the government (net borrowing

and operational withdrawals) lowered the level of the reserves by about $0.7 billion.

About one quarter of the net increase in the reserves––$0.3 billion––was a result of

short-term deposits of foreign currency at the Bank of Israel, placed by the local

commercial banking system for its operational needs at the end of 2006.

The average level of the reserves in 2006 was about $28 billion, $1 billion

higher than their level in the previous year (Figure 1). Although the reserves

increased, this rise was small in relation to the rise in the other economic aggregates

which are usually used as a measure with which to compare the level of the reserves,

such as imports, short-term external debt, unindexed local-currency assets held by the

public (M2), and GDP (Table 1). In dollar terms, Israel's foreign exchange reserves

rose by 4.2 percent in 2006, while the total reserves (excluding gold) of the member

countries of the IMF rose by 18.5 percent during the year.

6

Gross Short-term Unindexed GrossImports external external local-currency domestic

(months) debt debt assets (M2) product

1997 16,797 4.66 29 73 52 161998 21,392 6.03 35 86 61 201999 21,718 5.42 33 82 55 202000 21,854 4.58 32 77 46 182001 23,523 5.53 34 77 43 202002 23,943 5.84 34 75 48 222003 24,002 5.64 32 78 46 212004 25,908 5.30 33 79 47 212005 27,020 5.02 34 79 46 212006 27,976 4.81 32 75 45 20

a Based on daily valuation of the reserves.SOURCE: Bank of Israel, The Central Bureau of Statistics, and returns from the banks.

Table 1 The Level of the Reserves Relative to Other Aggregates, 1997-2006

Reserves as percent of aggregate

Average levelof reservesa

($ million)

Figure 1-Gross Foreign Exchange Reserves, 1997–2006 (monthly average)

0

5,000

10,000

15,000

20,000

25,000

30,000

$ million

Annual average

Reserves

20011997 1998 1999 2000 2002 2003 2004 2005 2006

SOURCE: Bank of Israel

The reserves serve as a source of liquidity to be used as and when necessary,

and their purpose is also to yield benefits that derive from the very fact that the

country is holding a certain level of foreign exchange reserves. At the end of 2003, the

7

Bank adopted a definition for the role of the reserves, according to which the uses of

the reserves are divided into the uses of the government––for which the Bank would

be likely to sell to it foreign currency––and the uses of the Bank. The benefits enjoyed

by the economy from maintaining foreign exchange reserves are the reduction in

probability of a crisis in the Israeli foreign exchange market and the enhancement of

Israel's standing in the international financial environment. The framework for

estimating the desired level of reserves in order to achieve these benefits was revised

in 2006, and following that, it can be concluded that the current level of reserves is

sufficient for this purpose. The framework for estimating the desired level of reserves

needed to provide for expected uses, including implications for the reserves'

investment policy, is currently undergoing a review, and it appears that changes will

be implemented in 2007, all of which will be reported separately.

According to the current method, which is described in the Department’s

annual reports for 2003-2005, the desired level of reserves according to the uses

approach was estimated at the end of 2006 at about $36 billion, of which $22 billion

was to cover possible government uses and about $14 billion was to cover the Bank's

possible uses. This compares to estimated levels of about $34 billion at the end of

2004 and about $35 billion at the end of 2005. The desired level needed to achieve the

benefits to the economy from holding reserves is not added to the level derived from

the reserves’ uses, but overlaps with it. There is no direct economic method for

determining the amount required for this purpose, but indirect indications can be

obtained in several ways. These show that the amount of the reserves calculated from

the benefits is smaller than that calculated from the uses. Thus the level derived from

the uses approach will also suffice for the benefits approach.

b. The investment policy of the reserves

The investment policy of the reserves portfolio is set by the Foreign-Currency

Committee headed by the Governor of the Bank. The Foreign Exchange Department

suggests topics to be discussed by the committee, and reports to it on the performance

of the portfolio, on investment decisions it has taken and on current developments in

the international markets.

The investment policy guiding the management of the foreign exchange reserves

is founded on three principles:

8

• Maintaining the value of the reserves in terms of their intended uses, as described

above. This is expressed in the determination of the currency composition of the

reserves, the limitations on their exposure to credit risk and to some extent the

management of their interest-rate risk.

• Managing the reserves with a high degree of liquidity. This is expressed mainly in

the restrictions on the types of asset in which the reserves may be invested. (Some

of these restrictions are specified in the Bank of Israel Law and the relevant legal

interpretations of it, and some are determined by internal Bank of Israel

decisions).

• Achieving a reasonable yield on the reserves portfolio, subject to the above two

principles. This is expressed in decisions on the duration of the portfolio, the

permitted level of exposure to credit risk and the use of active portfolio

management.

Control over most aspects of financial risk of the reserves portfolio is anchored in

the management of the reserves portfolio against a benchmark, a hypothetical

portfolio chosen according to predetermined rules, as part of the investment policy.

These rules determine the currency composition of the benchmark, its duration in each

currency, the types of assets included, and the dispersion of these assets along the

yield curve. The Bank of Israel uses active management in investing the reserves, and

therefore the reserves portfolio differs from the benchmark, generally, in terms of its

exposure to various risk factors, such as currency risk and interest-rate risk. The

investment policy of the reserves limits the scope of deviation from the benchmark

portfolio in a number of ways, expressed principally in five dimensions: 1) a limit on

the Value-at-Risk (VaR) of the gap between the actual portfolio and the benchmark;

2) limits on the differences in currency composition between the actual portfolio and

the benchmark; 3) a limit on the total difference between the duration of the actual

portfolio and the duration of the benchmark in various currencies; 4) limits on the

scope of investment in various types of asset which are not included in the

benchmark; and 5) limits on exposure to credit risk.

The VaR of the gap between the actual portfolio and the benchmark is defined as

the greatest loss the portfolio is expected to suffer from managing all positions over

the course of a week, with a probability of 95 percent. The VaR of the positions

managed in the reserves portfolio is restricted to a fifth of 1 percent (20 basis points)

9

of the value of the portfolio. It should be noted that there is no guarantee that a greater

loss will not be suffered in any given week, as the VaR restriction can only estimate

the largest expected loss 95 percent of the time. Moreover, the VaR calculation is

based on the assumption that the mathematical characterization of the market's

behavior in the current period will be similar to its characterization in a given

historical period. If, contrary to this assumption, the characterization changes, then the

scope of the expected loss could differ from the estimate that the VaR calculation

gives. However, one should also note that in practice only a part of the allowed range

of VaR generally is used, as described in Section 2b.

The currency component of the benchmark––the numeraire––is determined

according to the designated uses of the reserves, the two major components being the

US dollar and the euro. In contrast, the structure of the benchmark’s exposure to

interest rate risk in each currency—its duration and dispersion along the yield curve—

is not derived from the structure of exposure of the reserve uses in that currency to

interest rate risk, but rather from the goal of achieving maximum yield with low risk,

with the risk-return profile derived from the portfolio holder's preferences. The

duration of the benchmark for the dollar portfolio was revised in 2006 (see box

below). The investment policy limits the scope of exposure of the reserves to currency

risk and to the element of interest rate risk measured by duration differences.

Regarding the element of interest rate risk that stems from differences in distribution

along the curve, there are no explicit restrictions, but it is controlled by the VaR

restriction. The investment policy also places quantitative restrictions on exposure to

credit risk and on the investment of the reserves in assets that are not included in the

benchmark (spread assets). It establishes a minimum threshold for the credit quality of

an individual institution and also ensures an appropriate dispersion among institutions

and countries, according to their size and credit quality. There are also quantitative

restrictions on the total exposure of the reserves portfolio to the global banking

system.

The added value of active management is manifested in the differences in yield

and risk between the reserves portfolio and its benchmark, which is analyzed by

component in Section 2c below. The decision of the Bank regarding controlled active

management of the reserves is primarily justified by its contribution to the yield on

10

the portfolio during the last decade, which amounted to 9 basis points4 annually

(Table 2). Active management has additional advantages including strengthening of

the business relationships between the Bank and financial institutions abroad—which

increases the Bank's ability to obtain economic and business information—and

enhancement of the professional level of its reserves management staff.

Box 1

Major Investment Decisions Implemented in 2006

Change in duration of the reserves portfolio benchmark

The target duration in all currencies of the benchmark is set by the shortfall approach,

according to which, the long term duration target is set such that the annual holding

period rate of return on the portfolio will not fall below a given minimum threshold,

subject to a given confidence level. The minimum threshold and the confidence level

are the two parameters that define the desired level of risk according to risk

preferences appropriate for the reserves portfolio. The parameters of the shortfall

approach that were used in 1999 to set the benchmark duration for each currency are:

a confidence level of 95 percent and a minimum threshold of half of the yield on a 3-

month Treasury bill. The target duration as derived from these parameters was 16

months.5 In light of developments in recent years in Israel's economic environment, in

monetary policy and in exchange rate policy, the appropriate risk level for the

reserves portfolio is also perceived to have changed, and hence the Bank's

management decided to increase the expected mean holding period rate of return of

the reserves portfolio, at the price of a moderate increase in the risk level. In the third

quarter of 2006 the parameters of the shortfall approach were changed to a confidence

level of 99 percent and a minimum threshold of zero rate of return, in order to more

accurately reflect the risk-return tradeoff of the Bank's management. The

implementation of the new parameters––which have so far been implemented only in

the dollar portfolio––led to a change in the duration of the dollar's benchmark to a

fixed level of 24 months. Until such time as the new parameters are implemented in

4 A basis point is one-hundredth of a percent, or one in 10,000. At the current level of the reserves, one extra basis point is worth about $2.9 million. 5 From January 2003 to July 2005 the duration of the dollar portfolio (only) was temporarily shortened to 11 months, which was not according to the shortfall approach. A discussion on this can be found in the reports of the years in question.

11

the other currency portfolios, their durations have been left as derived from the

previous parameters, that is at a fixed level of 16 months. It should be noted that the

target duration is set according to long-term considerations, with a multiyear view,

and therefore decisions regarding it should not be judged according to investment

results over a shorter term.

Distribution along the yield curve of the dollar benchmark

Changing the benchmark duration also requires the desired dispersion of the

benchmark along the yield curve to be considered. From 1999 the benchmark has

included bonds of up to 5 years to maturity.6 Based on examining the profitability of

portfolios of various compositions, it was decided that the benchmark will continue to

include bonds of up to 5 years maturity instead of longer periods to maturity. As a

result, due to the increase in the duration of the benchmark and, consequently, that of

the portfolio as well, the short-term share of the portfolio was greatly reduced. The

reduction in the short-term part of the portfolio also reduced the opportunities for

investing in short-term spread assets. These assets, which are traded at a positive

spread compared to the short-term benchmark assets, had been a source of riskless

positive yield spread over the benchmark, since, due to their short terms to maturity,

they tend to be held to maturity in the reserves portfolio, avoiding any risk of capital

losses on them. The loss of this source of riskless income was taken into account in

historical analysis and in an examination of various scenarios, and still the portfolio

that was chosen above was found superior to those alternative portfolios examined.

Adding external managers of mortgage portfolios

At the beginning of the 1990s it was decided to invest in the mortgage market via

external managers.7 The percentage of the reserves invested in GNMA mortgage-

backed securities under external management grew gradually to 4 percent of the

reserve portfolio in the mid-1990s. In recent years, the share of these securities in the

reserves has fallen, reaching 2.5 percent at the end of 2005, mainly as a result of the

increase in the reserves. In light of the historical outperformance of mortgages

6 Except for the period January 2003 to July 2005, the period of shortening the duration of the dollar portfolio, when the benchmark included bonds of up to 3 years to maturity. 7 For a discussion on the motives behind the use of external managers see the 2001 annual report of the Foreign Exchange Department, Box 1.6. Since the early 1990s, GNMA mortgage backed securities have also been managed in-house.

12

compared to US government bonds, it was decided at the beginning of 2005 to raise

the share of the reserves invested in these securities to about 4 percent of the reserves

by raising the allocation to external management. This approach was chosen due to

the expertise of external managers in managing a wide range of assets in mortgage

portfolios, among other reasons. It was also decided to raise the number of external

managers, so as to avoid an excessive increase in the exposure of the reserves'

performance to the decisions of any particular manager. In 2005, a process of

selecting suitable management candidates for additional mortgage portfolios was

conducted. In the process, the Foreign Exchange Department consulted with other

central banks and international institutions. All candidates examined were top tier

investment institutions and met the Bank of Israel's criteria for managing mortgage

portfolios. In the end, two managers were chosen, which, in the opinion of the Bank,

demonstrated superior performance potential. These managers began managing

mortgage portfolios for the Bank of Israel in July 2006.

Extending the list of countries eligible for investment

Until this year, the reserves were invested only in developed countries of the first

rank, those with flexible exchange rate regimes and with high credit ratings from the

three main rating agencies: S&P, Moody's and Fitch. This year, after an in-depth

analysis of the probability of default for countries with various credit ratings against

the additional yield that may be earned from investing in debt instruments of states

with lower credit ratings than those where the reserves traditionally have been

invested, the Bank concluded that it was possible to liberalize the minimum

requirements for inclusion of a country on the list of those eligible for investment. As

a result of this change, a group of about a dozen countries was immediately added to

the list. The countries that were added all have investment-grade credit ratings, and

include states in Asia, Europe, and Latin America, including some considered to be

emerging market economies. An investment in a country may be an investment in the

currency of that country, or in debt instruments issued or backed by the government of

that country, whether in local currency or foreign currency. Adding a country to the

list of those eligible for investment is not an automatic result of the country's credit

rating, but rather is done only after a thorough individual check of the economic and

regulatory conditions of the particular country.

13

Incremental yieldDispersion

Decision to positionsActual Neutral Currency Intermarket invest in Duration Spread Dispersion and other

portfolio benchmark Total management positions GNMAb management effectc effect contributions1997Yield 4.99 5.24 -0.24 -0.26 0.03 -0.17 0.18 -0.01Standard deviation 0.44 0.49 0.16 0.08 0.02 0.09 0.07 0.02

1998Yield 6.00 5.99 0.01 0.00 -0.03 0.01 0.10 -0.07Standard deviation 0.63 0.69 0.08 0.03 0.06 0.02 0.05 0.02

1999Yield 3.26 3.17 0.08 0.02 0.03 -0.06 0.05 0.05Standard deviation 0.66 0.60 0.13 0.02 0.05 0.08

2000Yield 6.79 6.78 0.01 -0.15 -0.04 0.00 0.19 0.01Standard deviation 0.89 0.86 0.11 0.06 0.03 0.02

2001Yield 6.35 6.13 0.22 0.00 -0.05 -0.01 0.23 -0.01Standard deviation 1.44 1.36 0.20 0.03 0.04 0.04 0.09 0.00

2002Yield 5.18 4.98 0.20 0.03 0.00 -0.02 0.20 -0.01Standard deviation 1.32 1.41 0.17 0.04 0.05

2003Yield 2.15 1.94 0.21 0.04 -0.01 -0.02 0.19 0.02 0.00Standard deviation 0.81 0.79 0.09

2004Yield 1.70 1.67 0.03 0.02 -0.01 0.00 -0.05 0.11 -0.02 -0.02Standard deviation 0.73 0.73 0.06

2005Yield 2.64 2.44 0.21 0.00 0.02 0.00 -0.03 0.14 0.05 0.02Standard deviation 0.67 0.71 0.09

2006Yield 3.83 3.70 0.12 -0.02 0.00 0.00 -0.05 0.21 0.01 -0.01Standard deviation 0.77 0.77 0.11

1997-2006Yield 4.28 4.19 0.09 -0.03 -0.01 -0.04 0.16 -0.01

a 5.5 basis points of total incremental yield not attributed to any listed component in this year.b During 1997-2003, certain funds were managed (externally and internally) against a GNMA market index benchmark. Therefore, this benchmark was substituted for part of the long-maturity portion of the US dollar benchmark. The incremental yield from this substitution is given here.c For 1997-2002, the incremental yield from spread includes also the effect of dispersion.SOURCE: Bank of Israel.

(percent, in annual terms)

Asset selection

Table 2The Performance of the Actual Portfolio vis-à-vis the Benchmark Portfolio, 1997-2006

Portfolio performance

a

14

2. THE HOLDING PERIOD RATE OF RETURN AND THE RISK OF THE

RESERVES PORTFOLIO RELATIVE TO THE BENCHMARK

a. The holding period rate of return on the reserves portfolio

The total holding period rate of return on the reserves portfolio in terms of the

numeraire was 3.8 percent this year, compared to 2.6 percent in 2005 and an average

of 4.3 percent in the years from 1997 to 2006 (Table 2 and Figure 2). The volatility

(standard deviation) of the portfolio return was 0.77 percent, compared to 0.67 percent

in 2005. Three main factors affect the holding period rate of return on the foreign

exchange reserves: (i) developments in the financial markets; (ii) long-term

investment decisions, expressed in the composition of the neutral benchmark of the

portfolio; and (iii) day-to-day portfolio management, including active management

(decisions on deviations from the neutral benchmark).

Figure 2-Yield and the Total Management Contribution, 1997–2006 (percent, in annual numeraire terms)

-2.00

-1.00

0.00

1.00

2.00

3.00

4.00

5.00

6.00

7.00

8.00

9.00

1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

Yield

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6Contribution

Total management contribution Actual portfolio Neutral benchmark SOURCE: Bank of Israel.

15

The Bank of Israel normally evaluates the holding period rate of return on the

reserves portfolio in terms of the neutral currency composition––the numeraire––

calculated according to intended uses of the reserves. The return on the reserves can

also be calculated in shekel terms, which this year reached a negative level of 2.3

percent, after a positive return of 6.5 percent in 2005. The negative return in shekel

terms in 2006 is a result of the weakening of the US dollar against the shekel by 8.2

percent during the year. This weakening of the dollar was partly offset by the

strengthening of the other currencies in which the reserves were invested against the

shekel and by the income of the reserves from interest and capital gains. Over the last

ten years the holding period rate of return in shekel terms was 7.2 percent a year, and

after deducting the rise in the domestic price level over that period, 4.6 percent a year.

In terms of the dollar, the reserves made a profit of $1.8 billion. Income from

interest, capital gains and active management reached $1.1 billion and a further $0.7

billion resulted from exchange-rate differentials.8 It should be noted that measuring

post factum the return on a portfolio that was managed ex ante according to a multi-

currency target, in terms of one currency only, is considerably affected by the

currency of measurement. Thus for example, the return this year in dollar terms was

6.5 percent, compared to a negative return of 4.3 percent in euro terms. Therefore the

return on the reserves is generally measured in terms of the numeraire. The dollar and

euro portfolio of the reserves yielded returns of 4.4 percent and 2.2 percent

(respectively in terms of the portfolio currency). These returns are the main

components of the total return on the portfolio in terms of the numeraire. The total

return is also affected by the returns on the other portfolios and by the elements of the

active management contribution which are not attributed to any specific currency

portfolio.

The return on the reserves this year was affected to a great extent by the rising

trends in yields to maturity in the US and euro zone markets, with the inversion of the

US curve and flattening of the European curve. The rise in yields occurred against a

background of a continuing tight monetary policy in the US, which began in 2004,

and which saw the interest rate of the US Federal Reserve rise gradually from 4.25

percent to 5.25 percent during the first half of 2006, as well as the tight monetary

8 These amounts are calculated according to the average level of the reserves, and they include unrealized profits/losses, so that differences may appear between them and the amounts quoted in the Bank of Israel's financial statements.

16

policy of the European Central Bank, which began raising its interest rate at the end of

2005, and which raised its rate gradually from 2.25 percent at the beginning of 2006

to 3.5 percent at the end of the year. The increased level of yields on bonds in the US

and European markets raised the amount of current interest income included in the

total holding period rate of return on the reserves, but the rise in yields to maturity in

these bond markets also was expressed as a fall in prices, which were recorded as

capital losses. As the duration of these portfolios is longer than the period of

measurement, the reduction of the holding period rate of return from the capital losses

is necessarily greater than the additional holding period rate of return from the rise in

current income. Therefore, the holding period rates of return on the dollar and euro

portfolios are lower than the average level (during the year) of the yield to maturity in

the US and European markets, respectively.

b. Rate of return and risk in terms of the benchmark

The holding period rate of return of the benchmark was 3.7 percent this year while its

volatility was 0.77 percent. Given the developments in the markets, the benchmark

was the dominant factor in determining the holding period rate of return on the

portfolio in view of the relatively small magnitude of deviation of the composition of

the portfolio from the composition of the benchmark.

17

Figure 3-Distribution of Yields of the Portfolio and the Benchmark, 1997–2006

(percent, in annual numeraire terms)

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0Standard deviation

Yield

BenchmarkPortfolio

SOURCE: Bank of Israel.

Figure 3 presents holding period rates of return and volatilities for the

benchmark and the reserves portfolio since 1997, showing that in each year the yield

and volatility of the portfolio are very similar to those of the benchmark. The figure

also clearly shows the fluctuations in the annual holding period rates of return on the

benchmark and the reserves portfolio, which were largely determined by

developments in the markets in each year. In contrast the fluctuations from year to

year in the volatility of holding period rates of return were relatively small. The ratio

of the rate of return of the benchmark to its risk in any given year reflects the trade-off

of yield to risk as implied in the market. This ratio is shown in Figure 3 by the slope

of the line that connects the observations for all previous years with the origin. It can

be seen from Figure 3 that this ratio in 2006 was better or similar to the ratios in

earlier years in the past decade, except for 1997, 1998 and 2000, the observations for

which lie above the line.

The spread between the yield on the benchmark and the yield on the actual

reserves portfolio is the contribution of the ongoing management of the reserves.

Ongoing management this year contributed 12 basis points, with a volatility––known

18

as the tracking error––of 11 basis points (see Table 2 and Figure 2), based on monthly

observations. The contribution of ongoing management and its volatility came mainly

from the decisions on use of the degrees of freedom provided by the investment

policy regarding portfolio characteristics which can match those of the benchmark or

deviate from it––in its currency composition, its duration, the assets included in it and

their dispersion along the yield curve. The contribution of ongoing management also

may be influenced by operational factors, which can affect the ability to reach and

carry out portfolio management decisions.

Figure 4-Yield spreads vis-à-vis the Benchmark, 1997–2006

(percent, in annual numeraire terms)

-0.4

-0.2

0

0.2

0.4

0.0 0.1 0.2 0.3 0.4Standard deviation

Yield

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

SOURCE: Bank of Israel.

Analysis of the yield spread relative to the benchmark, and its volatility, can

provide an indication of the level of skill of the portfolio management. Ideally,

management of the portfolio should achieve a positive yield spread with volatility

lower than the yield. Figure 4 (and Table 2) show average yield spreads in terms of

the benchmark and their volatility in each of the years 1997-2006. The ratio of

incremental yield and its volatility (the “information ratio”) is reflected in the figure

by the slope of the line for each year. The slope of the line will ideally be greater than

one, in which case the incremental yield more than compensates for the incremental

19

risk. It can be seen that the information ratio was in fact greater than one in 2006 and

in four out of the five preceding years (2004 was the exception), since the slope of the

lines for those years is steeper than the dotted line which has a slope of one, was

between zero and one in 1999 and 2004, close to zero in 1998 and 2000, and negative

only in one year in the past decade, 1997.

The increase in the information ratio from 2001 onward was apparently the

result of decisions taken then, to increase the scope of asset-selection management

and to reduce the extent of exposure resulting from duration management (from 2000

onward) and currency management (from 2001 onward), two areas of activity that

contributed significantly to the yield on the portfolio during the early 1990s but were

less profitable subsequently. However, the Bank reconsiders from time to time the

ideal extent of active management of the reserves in various fields.

Figure 5-Tracking Error of Active Management of the Reserves, 2005-2006

(based on moving 52-week calculation)

5

6

7

8

9

10

11

12

13

14

15Basis points

2005Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4

2006

SOURCE: Bank of Israel.

Toward the end of 2006 the Bank decided to set a target for active

management's tracking error at 30 basis points. The aim of this was to increase the

contribution of active management by assuming moderate additional risk in those

fields where there is a chance of achieving a high information ratio, while creating an

20

incentive to improve the information ratio in those fields where today it is lower than

ideal. It is appropriate to mention here that in contrast to the measurement in Table 2

and Figure 4, the tracking error target was set on the basis of weekly observations.

Figure 5 shows the development of tracking error on this basis for 2005-2006. It can

be seen from the figure that the tracking error has been rising since the beginning of

the second quarter of 2005, though a slowdown in the rate of increase is also

noticeable in the second half of 2006. It should also be noted that the reserves’

tracking error is measured on a moving annual basis, so it is to be expected that the

effect on it of any change in the Bank’s style of active management will be moderate

and continuous.

Figure 6-VaR of the Positions, June 2002–December 2006

(one-week horizon, parametric method, 95% confidence level)

0

1

2

3

4

5

6

7

8

9

06/2002 12/2002 06/2003 12/2003 06/2004 12/2004 06/2005 12/2005 06/2006 12/2006

GNMATIPSTotalDuration positions, asset dispersion and security selection Currency positions

Basis points

SOURCE: Bank of Israel.

The incremental risk involved in managing positions in a particular period can

be measured after the fact by the volatility of the incremental yield, the approach

which was used in the discussion of the information ratio. Another measure used in

evaluating the risk of holding positions relative to the benchmark, which can be

measured ex ante, is the VaR (Value-at-Risk) of the positions, or in other words the

21

VaR of the gap between the actual portfolio and the benchmark. VaR is an estimate of

the maximum capital loss which can occur during a certain period of time and with a

certain probability. The VaR of all the positions, defined as the largest loss expected

in a week at 95 percent confidence level, plays a role in the investment policy of the

reserves and is restricted to a fifth of a percent (20 basis points) of the value of the

reserves. Figure 6 presents the breakdown of VaR due to various types of positions

that were held in the portfolio from June 2002 to the end of 2006, from which one can

see that this degree of freedom was only partially utilized, so that the average daily

VaR during 2006 was 4.4 basis points.

c. The contribution of ongoing management of the reserves by component

As part of the ongoing management of the portfolio, decisions are made to deviate

from the composition of the benchmark (positions). The additional yield from

management of the various types of positions is presented in Tables 3 and 4. A

position constitutes an addition of risk relative to the benchmark which may be

rewarded by incremental yield. As part of managing a position, a pre-determined

ceiling is generally put on the potential loss; the position is closed if the cumulative

loss reaches the ceiling. The scope of actual positions is much smaller than the

maximum permitted.

Total Spread Dispersion Dispersion Intermarket Currency Othercontribution Duration effect effect positions positions management contributions

Total 12.4 -5.3 20.5 0.8 -0.6 0.0 -2.2 -0.8Currency portfolios

Total 15.0 -5.3 20.5 0.8 -0.6 -0.4Dollar portfolio 18.9 -0.3 19.2 0.9 -0.8 -0.2Euro portfolio 1.9 1.1 0.7 -0.2 0.4 -0.2Other portfolios -5.8 -6.1 0.6 0.1 -0.2 0.0

Intermarket positions 0.0 0.0Currency positions -2.2 -2.2Other contributions -0.4 -0.4

SOURCE: Bank of Israel.

Asset selection

Table 3Contribution of Management Decisions to the Yield Spread, vis-à-vis the Benchmark, 2006(basis points, in annual terms)

1) Contribution of currency management

A number of currency positions were managed on the basis of an analysis of

economic variables, on the basis of tactical considerations or with the aid of models

22

(used for the management of small-scale, short-term currency positions). The

contribution of currency management includes exchange rate and interest rate

differences on the currency positions. In 2006, the management of currency positions

brought about a loss of 2.2 basis points to the reserve portfolio (Table 3).

2) The contribution of management of duration and dispersion along the yield curve

In 2006, duration positions managed in the various currency portfolios made a

negative contribution of 5.3 basis points to the yield spread relative to the benchmark

(Table 3 and Figure 7). The duration positions are manifested in the difference

between the duration of the currency portfolio and the neutral duration determined for

it: 16 or 24 months (see box above). The major component of the negative

contribution of the duration positions managed in 2006 was the loss absorbed by the

short duration positions managed in the other currency portfolios.

The total contribution of dispersion relative to the benchmark was a positive

0.2 basis points, of which a negative 0.6 was due to dispersion positions and a positive

0.8 was due to the influence of dispersion that resulted from the asset selection

positions (Table 3). The positions involving the dispersion of assets are created by

purchasing assets in one area of the yield curve and selling assets in another while

minimizing the impact of an upward or downward shift (without a change in slope) on

the profitability of the positions. Frequently these positions are managed through

futures contracts.

Figure 7-Duration Positions in the Total Portfolio, 2005-2006

-2.00

-1.50

-1.00

-0.50

0.00

0.50

1.00

01-05 04-05 07-05 10-05 01-06 04-06 07-06 10-06

Duration in months

USEuroOthers

`

SOURCE: Bank of Israel.

23

3) Intermarket positions

These positions are created in such a way that their profitability depends on the

development of the yield spread between two debt instruments. The instruments are

denominated in different currencies and the resulting currency exposure is neutralized

using a currency swap or through futures contracts. In 2006, the management of this

type of position did not lead to any additional yield (Table 3).

Total reserve portfolio

Securities lending by the Department9 2.8Eurobonds 6.7Synthetic investment in Japanese bonds10 0.5CD + CP 4.4Inflation-linked securities 0.3GNMA (external management)11 1.8Other assets and residual 4.1Total contribution of asset selection 20.5Dispersion deriving from asset selection 0.8Total 21.3

SOURCE: Bank of Israel.

(basis points, in annual terms)

Table 4The Contribution of Asset Selection, 2006

4) Contribution of asset selection 9,10,11

Decisions regarding asset selection contributed 20.5 basis points to the yield on the

reserves portfolio in 2006 (Table 4 and Figure 8). This increment was due to the

investment in spread assets, assets which are not included in the benchmark. Yield to

9 The lending of securities involves the linking of two transactions––a repo transaction and a reverse repo transaction or bank deposit. In the repo transaction a security is loaned in exchange for cash which is "deposited" in a reverse repo transaction against another security or invested in a bank deposit. The two transactions are for the same time period, and have no effect on the duration of the portfolio. The Bank of Israel's profit from such a pair of transactions results from the fact that the loaned securities in the repo transaction are in demand in the market by parties who are prepared to borrow them and to lend the equivalent value in money to the securities lender at a lower interest rate than that paid to the lender by depositing it elsewhere. 10 A synthetic transaction comprises three transactions that take place simultaneously: In the first, a security denominated in a particular currency is bought. In the second, the currency of investment is exchanged for the currency of the security, and the third is a currency forward transaction, that is the currency will be exchanged at a future date––the date of maturity of the security––from the currency of the security back into the currency of the investment. The sums exchanged are equal to the cost of buying the security and the sum received at time of maturity, respectively. 11 See box above.

24

maturity of such an asset can be split into two factors: i) expected yield to maturity of

government bonds in the same currency with the same maturity, and ii) the yield

spread, which reflects the characteristics of the issuer of the asset, and which usually

changes continuously and moderately. The contribution of the selection of spread

assets is measured by the spread between their holding period rate of return and that

of assets, with identical duration, of the type included in the benchmark while taking

into account their weight in the portfolio. In calculating the overall contribution of

current management, the impact of the differences in dispersion of the assets along the

yield curve––arising from the decisions on the selection of spread assets––are also

taken into account, a factor which is included in the calculation of the contribution of

dispersion as described in paragraph 2) above.

Figure 8-Asset Distribution of the Reserves Portfolio, 2006

(percent, annual average)

-0.4

1.83.5

-14.9

17.6

0.8

30.8

0.3

15.5

44.9

-30

-20

-10

0

10

20

30

40

50

60

70

80

Benchmarkassets

Eurobonds FRN CD + CP InflationLinkedBonds

(Dollar+JPY)

Deposits +r.repo

Repo GNMA (External

Managers)

JPY bills Other

Percent

SOURCE: Bank of Israel.

d. Control of credit risk in ongoing management

As with many other central banks, the Bank of Israel's sensitivity to credit risk is

higher than its sensitivity to other risks, such as interest risk. The credit risk which is

necessary to the ongoing management of the reserves portfolio is managed by the

25

Bank's internal restrictions––quantitative ceilings on various exposures and a system

of investment regulations. Credit risk is also measured by the pricing of risk of

tradable spread assets, which are marked-to-market daily. This year it was decided to

lower the country ranking permitted for investment (see box above).

Exposure of the reserves to the international banking system, which includes

exposure to banks through deposits with them (including tradable certificates of

deposit––CDs) and other transactions connected to them, is an important part of the

reserve portfolio's exposure to credit risk. This exposure is restricted to a ceiling of 35

percent of the portfolio. The actual annual average exposure of the reserves to banks

stood at 32 percent of the reserves in 2006, similar to the rate of exposure in 2005 of

31 percent (Figure 9).

Figure 9-Exposure to Banks and Governments, 1996–2005

(figures in boxes––percent of total capital of banks approved for investment)

0

10

20

30

40

50

60

70

80

90

100

1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

Percent

0

1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

9,000

10,000$ million

Exposure to banks (percent of reserves, left scale)Exposure to governments (percent of reserves, left scale)Exposure to banks ($ million, right scale)Annual average exposure to banks ($ million, right scale)

0.61 0.800.600.780.910.810.540.700.74 0.83

20011997 1998 1999 2000 2002 2003 2004 2005 2006

SOURCE: Bank of Israel.

26

3. YIELD ON THE RESERVES PORTFOLIO RELATIVE TO OTHER MANAGED

PORTFOLIOS

The comparison of performance between portfolios is problematic both because they

are generally managed relative to different benchmarks and because of the differences

in the rules applying to them. Nonetheless, something can be learned from a

comparison of portfolios with similar characteristics.

The performance of 11 mutual funds which operated in the American market

during the period 1997 to 2006 will be used for the comparison. These funds were

primarily invested in US government bonds. Some of the funds are classified as

“investors in short-term government bonds” while the rest are “general investors in

government bonds," which, in practice, invest in the medium term. The funds invested

no more than a small proportion of their portfolios in indexed US government bonds

(TIPS––Treasury Inflation-Protected Securities) or in assets with a low rating (less

than AA).12 These traits make it possible to make a rough comparison between the

performance of the Bank of Israel’s dollar-denominated portfolio and the performance

of these funds in spite of the differences between them. However, it should be

remembered that the performance of the funds is net of commissions which reflect the

cost of ongoing operations while the performance of the dollar-denominated reserves

portfolio is reported without subtracting this cost.

12 It is possible that there was a component of corporate bonds among these assets which the Bank of Israel is not permitted to hold.

27

Figure 10-Performance Distribution of Managers of Short- and Medium-Term

Funds in the US Market, 1997–2006

(percent, in annual terms)

-5

0

5

10

15

20

Yield

Bank of Israel portfolio

1997 Average1997-2006

1998 1999 2000 2001 2002 2003 2004 20062005

- Average of short-term funds - Average of medium-terms funds

Range of distribution SOURCE: Lipper Inc. via the Wall Street Journal internet site, and Bloomberg.

Figure 10 presents the range of annual holding period rates of return on the

funds since 1997 for each type of fund––the "short terms" and the "medium terms".

For each period, the figure shows the lowest and highest yields from among the group

of portfolios, the average yield for the group and the performance of the dollar-

denominated portfolio. An examination of the range of the yields indicates that the

performance of the dollar-denominated portfolio during the period was within, or

above, the range of fund yields. In 2006 the holding period rate of return of the dollar-

denominated portfolio was 45 basis points above the highest rate of return of the

short-term funds and 1.25 percent above the highest of the medium-term funds. The

years during which the dollar-denominated portfolio considerably outperformed the

funds, i.e. 1999 and 2006, were years in which the yield-to-maturity of US Treasury

securities increased. The superior performance of the dollar-denominated portfolio

during these periods implies that its duration was shorter than that of the other funds

which also explains its lower ranking during the years 2000 to 2002 when yields-to-

maturity declined.

28

The data also demonstrate the wide variation of yields among the funds which

implies differing investment allocations and the large magnitude of volatility of the

yields on the medium-term funds relative to the short-term funds. The latter

apparently reflects the longer duration of the medium-term funds. Nonetheless, the

range of the cumulative yields of each of the two groups of funds is very limited and

also similar between the two groups. The return for additional risk associated with the

higher duration of the medium-term funds, is expressed in the range of cumulative

yields of the medium-term funds, which is higher by some 1.2 percent than the short-

term funds.

Figure 11-Yield and Risk: the Dollar Portfolio vis-à-vis Funds in the US Market,

1997–2006

(percent, in annual terms)

0

1

2

3

4

5

6

7

8

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0Standard deviation

Yield

Medium-termfunds

Short-term funds

Dollar portfolio

SOURCE: Bank of Israel

This finding is even more apparent when the average yield of each fund is

compared to its volatility during the period 1997 to 2006. These are presented in

Figure 11. Also presented is the average yield for the dollar-denominated portfolio

and its volatility during that period. The yield-to-risk ratio for the dollar-denominated

portfolio is significantly higher than those of the other funds, as can be seen in the

diagram from the slope of the line joining the observations to the origin. Also evident

is the higher yield-to-risk ratio for the short-term funds relative to the medium-term

29

funds. The differences in the yield-to-risk ratios between the dollar-denominated

portfolio and the various funds is the result of the narrow range of variation among the

yields relative to the wide range of variation in volatility.

4. THE LIQUIDITY OF THE RESERVES13

The level of liquidity of the reserves portfolio is an estimate of that portion of the

portfolio which can be realized quickly and without a loss in value. In order to

monitor the general level of liquidity of the reserves portfolio on an ongoing basis, the

assets in the portfolio were divided into four groups according to the possibility of

selling them quickly without the sale itself causing a loss. The classification of the

securities into different groups was done according to the bid-offer spread for tradable

assets, and for nontradable assets, their time to maturity. The groups are:

1. Very liquid securities––with a spread of 0-2 basis points or 0-2 cents.

2. Less liquid securities––with a spread of 2-5 basis points or 2-6 cents.

3. Securities maturing within a month’s time –– deposits, repo and

reverse repo transactions which are shorter than one month.

4. All remaining assets.

The liquid component of the reserves consists of the first three groups. The

difference in levels of liquidity of these three groups is relatively small, and the

ranking of liquidity level of the third group vis-à-vis that of the first two is a matter of

judgment. The fourth group, however, is clearly less liquid than the first three. It

should be stressed that the bid-offer spread for a particular type of asset could expand

considerably in times of crisis in the international financial markets. Therefore, the

cost of realizing part of the reserves at a time of global crisis (as opposed to only a

local crisis) could be greater than that implied by the level of liquidity of the reserves

measured according to the bid-offer spread as observed in normal times.

The liquid component of the reserves in 2006 accounted on average for

approximately 92 percent of the total reserves as compared to 84 percent in 2005

(Figure 12). This 8-percentage point rise in the liquid component of the reserves is

13 A detailed discussion of the level of liquidity of the reserves and the management of that liquidity can be found in Box 2.1 in the 2002 Annual Report of the Foreign Currency Department.

30

related to two factors: a) an increase in the depth and liquidity of the Eurobond

market, which led to reductions in spreads and the movement of some of these types

of securities into the liquid component, and b) a move from bank deposits with terms

of more than one month to deposits with terms under one month and to tradable assets

such as CDs or CPs. Of the 92 percent liquid component, some 45 percent was

invested in assets with the highest level of liquidity, 39 percent in assets with high

liquidity, though less than that of the first group, and 8 percent in assets whose time to

maturity was shorter than one month. The share of assets in the first ("very liquid")

group rose by some 20 percent on 2005. About half of this increase can be attributed

to a fall in the second group ("liquid"), as a result of the reduction in bid-offer spread

of some types of government bonds during the year. This reduction, though small in

actual size, was sufficient to bring these assets across the threshold from the second

group into the first group. From this it appears that there was no significant change in

the level of liquidity of the liquid component itself.

Figure 12-Liquidity of the Reserves Portfolio, 2001–2006

(annual average)

0

10

20

30

40

50

60

70

80

90

100

2001 2002 2003 2004 2005 2006

Percent

Very liquid assets Liquid securities Assets of less than one month

SOURCE: Bank of Israel.

On average, approximately 8 percent of the reserves portfolio in 2006 was

invested in assets which were not classified as being part of the liquid component of

31

the reserves portfolio since they are characterized by a low level of liquidity. This

group is composed of funds under external management (which are invested primarily

in GNMA securities), bank deposits with a term of more than one month and it also

includes some Eurobonds and CPI-indexed bonds. In previous years, it also included

internally managed GNMA securities.

In light of the objectives of holding foreign exchange reserves and the ability

to sell a large part of them in a short time without thereby depressing their value, it

appears that their liquidity level is high enough, despite the gap between their actual

level and the desired level. This is because the liquidity component of the reserves

would appear sufficient to cover any uses that are likely to be required at short notice.

The high liquidity of the reserves derives from two main factors: one is the

investment policy based on the Bank of Israel Law, that prescribes a conservative

approach to the management of financial risks, and sets liquidity as one of the main

objectives in managing the reserves. The second factor consists of economic

considerations based mainly on the level of financial spreads of the various spread

assets and the paths they are expected to follow relative to their inherent risk. In light

of these considerations, the Bank has made only partial use of the degrees of freedom

to invest in spread assets with low liquidity during the last few years.