Proposal for guidelines for the Government Petroleum Fund

 Background and summary

Important underlying assumptions for the management of the
Government Petroleum Fund have changed since the existing guidelines were set
out in a regulation issued by the Ministry of Finance in May 1996. The Fund is
being accumulated at a swifter pace than previously anticipated. In addition,
projections for the central government budget balance in the Long-term Programme
1998-2001 indicate that it is not likely to be necessary to draw on the Fund for
a long time. Even though such projections are shrouded with uncertainty, it now
seems appropriate to base management of the Fund on a long investment
horizon.</P>

This has several important implications for the choice of investment strategy for the Petroleum Fund. It was against this background that Norges Bank raised the question of changing certain points in the regulation in a letter to the Ministry of Finance dated 10 April 1997. Among other things, Norges Bank recommended that Fund investments should be expanded to include investment in international equity markets, and that the share invested in European currencies and markets should be reduced in favour of investment in the US and Asia.

In the Revised National Budget 1997, the Ministry of Finance’s preliminary evaluation was that the equity proportion of the Government Petroleum Fund’s investment portfolio should be in the range of 30-50 per cent. The Ministry also conceded that the Fund’s currency and market distribution should be revised. A majority of the Standing Committee on Finance and Economic Affairs supported the view that 30-50 per cent of the Fund’s resources may be invested in equities. Norges Bank has used this as a basis for the guidelines proposed in this letter.

A majority of the Standing Committee on Finance and Economic Affairs agreed with the Government’s view that the Petroleum Fund should act exclusively as a financial investor. A proposal that the Fund’s investment profile should include other requirements did not gain a majority in the Storting. Norges Bank has therefore not discussed any constraints to the Fund’s investment strategy other than those resulting from an evaluation of risk in relation to return.

Against this background, Norges Bank presents a proposal for new guidelines for the management of the Petroleum Fund in this letter. The most important proposals are:

1. Risk measure. In the letter it is proposed that a tolerance level should be set for the expected differential between the return on the Fund’s investments and the return on the benchmark portfolio in the form of a measure for relative volatility. The choice of risk measure has implications for foreign currency and market distribution and the distribution between equities and bonds, which is why this topic is dealt with early in the letter.
2. Foreign currency and market distribution. The proposal entails considerable changes in the Fund’s foreign currency and market distribution. This is a follow-up to the proposals in the letter of 10 April 1997.
3. Interest-rate risk. It is proposed that the duration for the Fund’s bond portfolio shall be between 3 and 7.
4. Instruments. The current guidelines contain a list of the types of instruments permitted in the management of the Fund. It is argued that this is not necessary and that the regulation should be more general regarding this point. Furthermore, it is proposed that investment in private bonds should be permitted.
5. Liquidity. It is proposed that special liquidity requirements should no longer apply.

2. Relationship between Norges Bank and the Ministry of Finance

The Ministry of Finance is responsible for the management of the Petroleum Fund. The operational management of the Fund has been delegated to Norges Bank. It is important that the guidelines for the Fund clearly define the division of responsibility between the Ministry of Finance and Norges Bank. This may be done by setting out in the guidelines the framework for the investment strategy to be used in the management of the Fund. It is usual to assume that this strategy comprises both a long-term (passive) investment strategy and a more short-term (active) investment strategy.

The long-term investment strategy shall reflect the strategic choices made in the management of the Fund regarding foreign currency and market distribution, the distribution between equities and bonds and between securities in different market segments. The strategic choices may be specified by defining a benchmark portfolio. It is proposed that this portfolio may be composed of a selection of certain securities from different markets. The portfolio manager will use the benchmark portfolio as a basis for actual investments.

The more short-term investment strategy represents the variations the manager may wish to achieve in terms of the benchmark portfolio. Such variations may stem from a wish to reduce management costs, eg by not investing in all securities included in the benchmark portfolio. Variations in relation to the benchmark portfolio may also be due to situations arising in financial markets which entail that the manager deems it preferable to change the investment strategy for shorter periods. The guidelines should set out limits for the variations permitted in relation to the benchmark portfolio. The challenge facing the manager will be to use these variations to achieve a higher return in relation to the benchmark portfolio.

As the Ministry of Finance is responsible for the management of the Petroleum Fund, the Ministry must define a benchmark portfolio for the Fund and set limits for the maximum variations permitted in relation to the benchmark portfolio. This will ensure a clearly-defined division of roles between the Ministry of Finance as the delegating authority and Norges Bank as the manager. It is assumed that Norges Bank will be responsible for the management of the Fund in the guidelines adopted by the Ministry of Finance.

Norges Bank will prepare detailed annual reports on the management of the Petroleum Fund. These reports will show how the Fund is managed, including the companies in which the Fund’s capital has been invested. The reports will also show the total return on Fund investments. It is important to emphasise that the Fund will be managed in such a way that the return on the Fund’s capital will primarily be determined by the strategic choices reflected in the benchmark portfolio.

Norges Bank will report regularly to the Ministry of Finance on variations between the benchmark portfolio and actual investments of the Fund’s capital. Any variation may result in a higher or lower return in relation to the benchmark portfolio. Differences in the return and the breakdown of the return into different components will be reported at regular intervals to the Ministry of Finance. This will be an important element in the continuous assessment of management.

In addition to the Ministry of Finance’s guidelines for the Petroleum Fund, an agreement will be drawn up which regulates the relationship between the Ministry and Norges Bank in connection with the management of the Fund. This agreement should set out how Norges Bank’s management costs shall be covered.

3. Benchmark portfolio and risk management

The benchmark portfolio plays a central role in the management of the Petroleum Fund. In addition to reflecting the strategic choices made in connection with the management of the Fund, the benchmark portfolio shall also be used to assess management performance. In order to ensure this, it is important that the construction of the benchmark portfolio is, to the greatest extent possible, based on clear and objective criteria and that it is possible to replicate the portfolio in a cost-effective manner.

Two different methods may be used to define the Petroleum Fund’s benchmark portfolio. The one is to tailor the benchmark portfolio to represent the strategic choices made in connection with the management of the Fund in the best possible way. The second is to use market indices (reference portfolios) developed by large international investment banks. Norges Bank is of the view that it may be appropriate to use a combination of both methods when defining the benchmark portfolio for the Petroleum Fund. The market indices should be the basis for defining benchmark portfolios in individual countries’ equity and bond markets. This provides an objective basis for choosing securities in benchmark portfolios as well as facilitating the evaluation of external managers. However, the overall benchmark portfolio for the Petroleum Fund must be tailor-made as it shall contain a currency and market distribution and distribution between equities and bonds. This distribution must be decided on the basis of the objective for the management of the Petroleum Fund. This will be discussed later in the letter.

A portfolio manager may wish to deviate from the benchmark portfolio when such variations will result in a higher expected return or contribute to more cost-effective management (fewer transactions). The manager’s scope for deviating from the benchmark portfolio is limited in the current guidelines for the Petroleum Fund, in that partial limits are set for foreign currency and market distribution and for interest-rate risk. When equity instruments are also included in the guidelines, it will be necessary to set limits for the spread between equities and bonds. One disadvantage with this is that there may eventually be so many limitations to take into account that it will be difficult to have a clear view of the total risk in the portfolio in relation to the benchmark portfolio. Partial limits do not take into account covariance between the different asset classes, markets and currencies. Consequently, positions may offset one another and overall not entail any more risk than set out in the guidelines, whereas a number of smaller variations (in the same direction) which are individually permitted may in total generate considerable risk in the portfolio.

It may therefore be more appropriate to limit the degree of deviation from the benchmark portfolio using an overall risk measure. This may be done by setting limits for acceptable variations in the return on actual investments and the benchmark portfolio. One measure of risk which is often used in this context is relative volatility, defined as the standard deviation of the differential between the return on actual investments and the return on the benchmark portfolio.1 In simplified terms, if permitted deviations from the benchmark portfolio are limited by setting an upper limit for relative volatility, it is possible to say that the actual return will vary within the range set for the return on the benchmark portfolio. The lower the limit for relative volatility is, the narrower the range will be.

1 Standard deviation is a measure of spread which shows how much a time series varies around its average.

An estimate of relative volatility is required for operational management which shows the extent of expected differences in the return between actual investments and the benchmark portfolio. This estimate depends on two factors. The first is actual variations between the investment portfolio and the benchmark portfolio. The estimate for relative volatility increases if deviations from the benchmark portfolio increase, whereas the estimate will be zero if there are no variations. The second factor which influences the estimate of relative volatility is volatility of the return in each market and the correlation between returns in the different markets. Both the volatility and the correlation can be calculated using historical data. If it is assumed that the historical volatility and correlation will also apply in the future, relative volatility can be used to indicate the degree to which the return on the actual portfolio may be expected to vary from the return on the overall benchmark portfolio.

An example may help to illustrate how deviations from the benchmark portfolio may be managed using relative volatility. Based on a completely defined benchmark portfolio, the principal sets the upper limit for relative volatility at 1 per cent. Such a limit for relative volatility provides the manager with some room for manoeuvre in operational management. The manager may choose how this is used. The limit for relative volatility will, however, entail that permitted deviations from the benchmark portfolio will be small in the most volatile markets, whereas in markets with little fluctuation the possibility of deviation is greater. It is assumed that the manager chooses to construct a portfolio where the expected return is 0.5 per cent (randomly chosen) higher than the benchmark portfolio and a relative volatility of 1 per cent. In this case, the manager has chosen to utilise the limit for relative volatility to the full, which does not necessarily need to be the case. Given certain underlying assumptions regarding the statistical properties of returns, the realised differential in the return between the benchmark portfolio and the actual portfolio may, with a probability of 95 per cent, prove to be between -1.5 per cent and 2.5 per cent.2 If the return on the benchmark portfolio is 10 per cent, the return on the actual portfolio can be expected to be in the range 8.5 to 12.5 per cent. This illustrates how a maximum limit for relative volatility may constrain the manager’s scope for deviating from the benchmark portfolio, as the return on the actual portfolio is not permitted to vary excessively in relation to the return on the benchmark portfolio.

2 The range (confidence interval) for the return differential is calculated by taking the expected higher return (0.5 per cent), then adding and subtracting 2 times the magnitude of relative volatility (1 per cent): 0.5%+ 2x1%. The number 2, which is derived from a statistical distribution, results in an interval within which the return differential will lie with a probability of 95 per cent.

Norges Bank currently uses relative volatility as an operational measure of risk in the management of the foreign exchange reserves. This measure of risk is also used by Norwegian and international management institutions. Furthermore, it is expected that the supervisory authorities in many countries will accept a similar principle for managing market risk in banks and securities firms (see the recommended capital adequacy requirements in the BIS’ and EU’s proposal for the revised CAD). Norges Bank will recommend that the Ministry of Finance stipulate a maximum limit for relative volatility for the management of the Petroleum Fund. The limit for relative volatility set by the Ministry of Finance should apply to variations between the expected return on Petroleum Fund investments and the total benchmark portfolio. Norges Bank should be responsible for setting the variations permitted in the different market segments and the local benchmark portfolios.

It is important to point out that relative volatility is a predicted value and not a realised value. If the correlation between returns in the different markets in the future deviates from historically observed correlations, the realised volatility will also vary from the predicated volatility. The return on Petroleum Fund investments may then deviate by a greater margin from the benchmark portfolio than the measure of relative volatility would indicate. However, it is necessary that the Ministry of Finance, as principal, accepts the underlying assumptions and calculations for relative volatility.

As relative volatility is only a predicted value, it may be appropriate to underpin this measure of risk by including special limits for the most important risk factors (foreign currency and market distribution, the equity proportion and interest-rate risk) in the guidelines for the Petroleum Fund. However, it is important that these limits are not too restrictive as they are primarily deemed to be a ‘security net’ for the operational management of risk through relative volatility. The mid-value of these limits should be the basis for the composition of the Petroleum Fund’s overall benchmark portfolio.

What should the limit for relative volatility be?

When setting the limit for relative volatility in the management of the Petroleum Fund, it may be appropriate to differentiate between the two reasons why a manager may wish to deviate from the benchmark portfolio: the consideration of cost-effective management, and the wish to achieve a higher return in relation to the benchmark portfolio.

As regards the consideration of cost-effective management, experience has shown that it may be difficult to achieve a complete replica of the benchmark portfolio as this often entails that the manager must carry out a substantial number of transactions. Taking transaction costs into account, this could indicate that it should be possible for the manager to construct a portfolio with slightly different features (fewer securities) than the benchmark portfolio. In addition, it may be cost-effective in some markets to use derivatives to achieve exposure to the underlying securities over a shorter or longer period of time. Equity exposure through the use of derivatives may often differ somewhat to exposure in the benchmark portfolio. It is difficult to give any precise estimate of the magnitude of relative volatility required as regards cost-effectiveness. However, it cannot be ruled out that relative volatility between the actual portfolio and the benchmark portfolio would be several per cent in some market segments. As it is likely that deviations from local benchmark portfolios will show little correlation, relative volatility for the Petroleum Fund’s overall benchmark portfolio will be considerably lower. Norges Bank has made calculations which show that a relative volatility for the total benchmark portfolio of around 0.3-0.4 per cent would be sufficient with regard to cost-effective management.

The extent to which a higher limit than this should be set for relative volatility depends on how the Petroleum Fund is to be managed. It is usual to differentiate between passive and active management, but there is, in fact, no exact division between the two types of management. For example, methods have been developed that border on both active and passive management, which provide the opportunity to achieve a higher return with a limited degree of risk.

In Norges Bank’s view, a large part of the Petroleum Fund should be managed on the basis of a long-term strategy with only small deviations from the benchmark portfolio, as substantial amounts of the Fund’s capital will gradually be invested in international financial markets and experience has shown that it may be difficult to engage in active management involving substantial amounts. Furthermore, such a strategy would involve lower management costs.

However, it should be possible to carry out a more active management of parts of the Petroleum Fund. Based on a clearly defined benchmark portfolio, active management can be carried out in four different ways: i) by changing the foreign currency and market distribution, ii) by changing the spread of equities and bonds, iii) by including sectors or companies in the equity portfolio that are expected to perform better than others, and iv) by changing interest-rate risk or credit risk in the bond portfolio. Which active investment strategy will provide the best results depends on where deviations from efficient pricing of securities can be expected and on the manager’s ability to exploit such deviations in order to generate a higher return in relation to the benchmark portfolio. Empirical studies do not provide unequivocal results as regards the possibility of generating a higher (risk-adjusted) return by means of the different types of active management.

Norges Bank will seek to choose active investment strategies which may give a higher expected return without incurring considerably more risk for the Fund. In the light of this, it may be that the two latter forms of active management should be emphasised in the management of the Fund. Norges Bank has solid experience of bond investment and the management of the foreign exchange reserves is currently based on a maximum limit for relative volatility of 0.5 per cent. Active management of equities will primarily be carried out by external managers. Norges Bank is of the view that there is reason to test managers who have developed special investment styles which have proven to generate positive results. The choice of managers and follow-up will be important in terms of whether such a strategy will give a sufficiently high return in relation to the extra risk incurred. According to Norges Bank’s assessment, a limit for relative volatility in the range of 1-1.5 per cent would be compatible with a prudent form of active management of parts of the Petroleum Fund’s equity and bond portfolios.

As regards management of the Petroleum Fund, the limit for relative volatility should be set following an overall assessment on the basis of cost-effective management and active management. Central to this assessment is the Ministry of Finance’s attitude to the risk that the return on the Fund’s capital deviates from that which could be achieved by strictly shadowing the benchmark portfolio. In Norges Bank’s opinion, an upper limit for relative volatility of 1.5 per cent would permit cost-effective management of the Fund and an active management of parts of the Fund’s portfolio. It is unlikely that Norges Bank will fully utilise this limit in the start-up phase, but will rather gradually gain experience of the different types of active management.

4. Long-term investment strategy for the Petroleum Fund

Before discussing the choice of investment strategy for the Petroleum Fund, it is important to define the object of the management of the Fund. In the Revised National Budget 1997, the Ministry of Finance writes: "In terms of the Petroleum Fund, it is natural to apply a long investment horizon and to recognise the importance of preserving the Fund’s international purchasing power". Norges Bank has therefore assumed that the object of the management of the Petroleum Fund is to preserve the Fund’s international purchasing power in the long term. Such an object has two important implications for the management of the Fund. A long investment horizon means that little emphasis should be placed on annual variations in the return on the Fund’s capital. The consideration of international purchasing power means that the focus should not be on the return on Fund investments measured in Norwegian kroner. It would be more meaningful to measure the return on the Fund’s capital in foreign currency. The consequences of this as regards the choice of long-term strategy are discussed below.

In which countries can the Petroleum Fund’s capital be invested?

The Petroleum Fund shall only be invested in countries which have smooth-functioning financial markets and sound legislation relating to securities. In addition, requirements should be set regarding the size and liquidity of the financial markets. It is also important for the Fund that foreign participants do not have restricted access to the markets. As it is not clear which countries fulfill these requirements, it would be advantageous to establish unambiguous and objective criteria for selecting relevant countries. One possibility is to limit investment to countries which participate in international organisations where membership requires that the financial markets function in a certain way (eg the OECD). Another alternative is to base investment on the large investment banks’ market indices (benchmark portfolios) for investors who want a global diversification of their wealth. The advantage of this alternative is that the indices are constructed exclusively according to criteria which are relevant to the management of the Petroleum Fund (eg the size and liquidity of financial markets, legislation and accessibility for foreign investors). Norges Bank is therefore of the view that it may be appropriate to use the investment banks’ market indices when specifying in which countries the Petroleum Fund’s capital may be invested.

Risk management considerations may also limit the countries in which the Fund can be invested, due to the fact that when relative volatility is used as a measure of risk, the Fund should only be invested in countries/markets which make it possible to project the size of this variable.

Thus far, the Petroleum Fund has only invested in developed countries, and there are special market indices for investors who only wish to invest in such countries.3

3 Morgan Stanley has constructed a global share index for developed countries. According to the criteria set, 22 countries satisfy the requirements, comprising the developed countries in the OECD, Hong Kong, Singapore and Malaysia. (According to the World Bank’s definition, Malaysia is not included in the group of developed countries, but Morgan Stanley has included Malaysia in its index as the country has well-developed financial markets.). As regards bonds, many international investors use market indices developed by Salomon Brothers or JP Morgan. These portfolios include fewer countries than Morgan Stanley’s market index for equities (Hong Kong, Singapore, Malaysia and New Zealand are not included in the Salomon Brothers’ market index, and the JP Morgan index, in addition to these countries, does not include Finland and Switzerland).

However, the possibility of investing in emerging markets should also be considered. These markets include developing countries and economies in transition, which are characterised by rapid economic growth and steadily improving financial markets. It is conceivable that these countries will experience a favourable economic development in the years ahead as they will probably progressively close the technological gap which exists between them and developed countries. This may generate expectations of solid returns on investments in emerging markets. If the economic situation in these countries is not synchronised with developments in other countries, investment in emerging markets may also contribute to reducing variations in the Fund’s total return. Norges Bank finds that the guidelines for the Fund should on this basis provide for investment in these markets of up to 5 per cent of the total capital in the Fund. However, it is uncertain whether Norges Bank will invest in emerging markets in the start-up phase of the Fund. The development of systems and skills to handle investments in these markets must be considered in relation to other priority tasks.

It has been proposed that the Ministry of Finance, on the basis of a proposal by Norges Bank, approve a list of countries in which the Fund can be invested. The authorisation to invest in a country entails that Norges Bank is permitted to buy equities in listed companies in the country and bonds issued in the country’s currency. A criterion for inclusion on the country list should be that the country or emerging market is included in internationally recognised market indices. Norges Bank will also emphasise the importance of not investing the Fund in countries which make it difficult to apply the risk measure of relative volatility.

Empirical studies show that very little is lost, particularly for bond investment, in the form of reduced diversification gains by excluding some of the smaller countries from the portfolio. Moreover, this would facilitate the operational management of the portfolio. It should therefore be considered whether the Petroleum Fund’s benchmark portfolio should include somewhat fewer countries.

If investment of the Fund’s capital in both developing countries and emerging markets is permitted, the Fund will have the possibility of investing in virtually the same countries as Norwegian financial institutions which manage international funds.

Currency and market distribution

In principle, the choice of currency distribution and market distribution involve two independent decisions as the distribution can be differentiated through the forward foreign exchange market. In a letter of 10 April 1997, Norges Bank pointed out that it would not be appropriate to operate with such a permanent difference between currency and market distribution. This is partly related to the Fund’s long investment horizon which makes it difficult to reveal the real currency exposure in the Fund as exchange rate changes in the longer term are often offset by interest-rate and inflation differentials. A significant difference between the currency and market distribution would over time necessitate a frequent renewal of short-term contracts (currency swaps) involving substantial amounts, which would give rise to considerable transaction costs and higher counterparty risk. The choice of currency and market distribution should therefore be the same in the guidelines and the benchmark portfolio for the Petroleum Fund.

Import weights have been used as a basis for choosing the currency and market distribution under the existing guidelines for the Petroleum Fund, with about 75 per cent of the Fund invested in Europe, whereas the remainder is invested in the US, Canada and Japan. This distribution should be reconsidered in the light of the changes in the parameters for managing the Fund. The proposal to change the currency and market distribution for the Fund should preferably be based on an empirical portfolio analysis, where the covariance between exchange rate changes, return on equities and bonds in the various markets is taken into account in a systematic fashion. However, the long investment horizon associated with the Fund complicates the use of such a portfolio analysis as it does not comprise sufficiently long time series in order to quantify the covariance between the return and the various investment alternatives in a satisfactory way. Another drawback is that it may prove difficult to measure the real currency and market exposure in the case of equity investment as many of the large companies are active in more than one country. It is therefore difficult to avoid a discretionary approach when choosing the Fund’s currency and market distribution. Norges Bank would argue that the considerations above indicate that the Fund’s investments both in equities and bonds should have a global spread other than that provided by import weights. There are four particular reasons for this:

1. The main argument for using import weights is related to the aim of reducing the exchange rate risk associated with Norway’s imports. Two factors would reduce the importance of this risk when the investment horizon is increased. First, it is reasonable to assume that goods which are traded internationally over time will show about the same price trend in a common currency (purchasing power parity, PPP, holds). Second, the risk associated with any deviation from PPP will be limited as imports can be shifted to those goods which present the most favourable price trend. This illustrates that less emphasis can be placed on import weights in selecting the currency and market distribution for the Petroleum Fund.
2. It is important that the Petroleum Fund’s currency and market distribution provide a sufficient spread of the risk associated with changes in equity and bond prices. This can be achieved by spreading investments in a country where price changes are not closely correlated. This correlation is often stronger between countries located in the same region than between countries located in different regions. This would then favour spreading investments over various regions. Another factor supporting this approach is that the Fund’s long investment horizon entails that greater emphasis should be placed on risk associated with various incidents such as natural disasters, technological breakthroughs, war, social upheaval, and other political events. Since such incidents are seldom recurrent it is difficult to quantify the probability of such events. However, it is important to take into account that such incidents may occur. History shows that problems arising in one country may easily spill over into neighbouring countries. This may indicate that one should spread investments across various regions.
3. Norway’s national wealth primarily consists of real capital and domestic human resources, but also the petroleum reserves in the North Sea and our foreign assets through the Petroleum Fund. In theory, the various components of this wealth should be managed as a whole. In practice, it may be difficult to implement this with a consistent approach. Yet it should still be an objective to manage the Petroleum Fund so that the return on the Fund does not show strong covariance with the return on the other wealth components. According to the existing guidelines for the Fund, the largest share of the Fund’s capital is to be invested in European countries. However, as these countries are located in the same geographical region as our own and have the same basic economic structure as Norway, it is likely that the return on investments in this region will be highly correlated with the return on other components of our national wealth. This may indicate that a larger share of the Fund should be invested outside Europe than is the case today.
4. The existing guidelines entail that close to 25 per cent of the Petroleum Fund is to be invested in Denmark and Sweden. According to the estimates in the Long-Term Programme 1998-2001, the Fund may reach a substantial size in a few years, also on an international scale. With the current currency and market distribution, the Fund will eventually acquire substantial market shares in the Swedish and Danish capital market. Equity investments in these markets may lead to a situation whereby the Fund has large ownership interests in individual companies, which would not be in keeping with the recommendations in the Revised National Budget for 1997 which state that the Fund should exclusively be a financial investor in equity markets. Another drawback associated with large market shares is that market prices are influenced by purchases and sales involving substantial amounts. This illustrates that in selecting the currency and market distribution for equities and bonds, greater emphasis than previously should be placed on the size of the national financial markets in question.

The discussion above underlines that import weights do not provide an appropriate currency and market distribution for the Petroleum Fund and that the Fund’s investments in both equities and bonds should feature a wider global spread. There are two weight bases which are commonly used among international investors in addition to import weights, notably market capitalisation weights and GDP weights.

With market capitalisation weights, the share invested in a market is proportional to the size (market value) of the market. There are several advantages associated with this principle. First, market capitalisation weights entail that the investments are concentrated in the largest and most liquid markets. As a rule, these markets feature small differences between bid and offer prices, and thereby low transaction costs. Moreover, applying the principle of market capitalisation weights will resolve the problem of having to rebalance the portfolio when price trends diverge in various markets. Market capitalisation weights will automatically ensure that the weights in the benchmark portfolio are adjusted in line with price changes, thereby avoiding unnecessary transaction costs. A drawback of market capitalisation weights is that in periods of widely divergent price trends in various markets, the weights may result in extreme market distributions in that the weights for countries with high price rises increase rapidly.

GDP weights will provide a global spread of the Fund, with an associated sound diversification of the risk linked to changes in securities prices. Furthermore, GDP weights are closely correlated with market capitalisation weights, which ensure that the investments are channelled to the most liquid markets. In addition, GDP weights provide an indication of a country’s production capacity, and growing international trade may render this capacity relevant in terms of assessing the future pattern of Norwegian imports.

Table 1 shows the distribution by region resulting from the different weight sets on the basis of the assumption that the Fund is invested in countries which are included in investment banks’ market indices for developed countries (see footnote 1).

* The regional distribution is used as it is common internationally to divide countries into regions by time zone. The import weights are based on national accounts figures for 1996. The GDP weights are based on the IMF’s GDP figures for 1995 translated into USD. Market capitalisation weights for equities are calculated using Morgan Stanley’s share index from 30 April 1997. Market capitalisation weights for bonds are based on Salomon Brothers’ bond index from 30 April 1997.

Norges Bank recommends that the choice of the regional currency and market distribution for the Petroleum Fund should primarily be based on GDP weights, but that some emphasis should also be placed on Norway’s import patterns. The following regional distribution is proposed:

Americas: 20-40%, Europe and Africa: 40-60%, Asia and Oceania: 10-30%

The mid-point of these intervals should be the reference mark for the composition of the Fund’s overall benchmark portfolio. The regional distribution should be the same for equities and bonds. However, Norges Bank is of the view that the currency and market distribution does not have to be the same for equities and bonds in the various regions. As regards equities, the consideration relating to transaction costs and avoiding large stakes in single companies would indicate that market capitalisation weights should be used. However, these weights are not an appropriate alternative in the case of bonds as it may entail that investments are automatically increased in countries with growing debt. When bond investments are to be spread within various regions it would therefore be more appropriate to use GDP weights as a basis.

Equities

The choice of equity proportion will probably constitute the single most important decision in terms of the return on the Petroleum Fund over time. In its letter of 10 April 1997, Norges Bank argued that no strong professional objections could be raised against a minimum equity proportion of 30 per cent, and it was mentioned that other management institutions with responsibility for large portfolios with long-term investment horizons tended to have higher proportions than this. The Revised National Budget for 1997 stipulated a preliminary limit in the range of 30-50 per cent. A majority of the Standing Committee on Finance and Economic Affairs supported this evaluation.

It is important that the equity proportion be permitted to vary over time as changes in the market value of the equities automatically change the equity proportion in the total portfolio. During periods of marked changes in equity prices the equity proportion may shift by a substantial margin. If the equity proportion were to be held constant during such periods, one would have to sell down continuously in equity markets where prices are advancing and buy up in markets where prices are declining. This would involve substantial transaction costs. Therefore, it is important that the guidelines indicate an interval within which the equity proportion is allowed to vary. It would still be appropriate to rebalance the portfolio at regular intervals so that the equity proportion is equal to the mid-point of the interval.

Norges Bank has taken note of the evaluation made by the Ministry of Finance and the majority of the Standing Committee on Finance and Economic Affairs, recommending an equity proportion between 30-50 per cent. As the equity proportion should be allowed to vary over time, the benchmark portfolio should be based on an equity proportion of 40 per cent.

A substantial equity proportion in the Petroleum Fund may lead to wide short-term variations in the return on the Fund. The regular reports to be published on the Fund’s return may therefore show negative returns during periods. However, provided the Fund features a long investment horizon, short-term variations in returns should not have implications for the choice of equity proportion.

The Petroleum Fund may have an equity exposure through investments in unit trusts, direct investment in equities and/or by using derivatives, eg futures or swaps on share indices. Ownership is indirect when investments are made through unit trusts. One drawback of unit trusts is that the manager may have less control over the portfolio’s composition and risk management may be more difficult. In addition, such unit trusts are not very cost-effective. On the other hand, unit trusts may be a relevant alternative in some markets, particularly in emerging markets should these markets be included. Derivatives have proven to be appropriate in terms of operational risk management. In some markets derivatives may also be a cost-effective means of achieving equity exposure. Norges Bank would recommend direct purchases of equities and derivatives in the management of the Fund, provided purchases of derivatives do not exceed the exposure which would have been the case if the same capital had alternatively been directly invested in equities.

Norges Bank emphasises that equity investments in each market must be spread over a large number of companies. This provides a sound diversification of risk and thereby limits fluctuations in returns. With the proposed currency and market distribution, ownership interests will be largest in European equity markets. The projected size of the Fund indicates that in a few years these holdings may be sizeable in these markets. A full replication of the benchmark portfolio would entail that a Fund of NOK 500 billion, for instance, at current share prices and with an equity proportion of 40 per cent would result in ownership interests of around 0.3-0.4 per cent in all the companies included in the benchmark portfolio for European stock markets. However, as full replication may result in substantial transaction costs, it may be appropriate to concentrate investments in a fewer number of companies. This would increase the ownership interests in the companies selected. In addition, external managers may be used to engage in active management of a limited share of the Fund’s total equity portfolio. Active management will entail that investments in some companies will be overweighted in relation to the benchmark portfolio. Against this background, Norges Bank recommends an upper limit of 3 per cent for allowed ownership interests in individual companies. It is reasonable to assume that this limit will only apply to a small number of companies in the coming years. A maximum of 3 per cent should ensure that the Fund does not engage in strategic investments in equity markets.

Interest-rate risk

A common means of managing the bond portfolio’s interest-rate risk is a target figure for modified duration. Modified duration shows the percentage change in the value of an interest-bearing instrument (commercial paper, notes and bonds) in response to a one percentage point change in interest rates. Under the current guidelines for the Petroleum Fund, the maximum limit for duration is 5, whereas the duration for the current benchmark portfolio is 3.5.

The appropriate duration for the Petroleum Fund will have to be determined on the basis of the relationship between return and risk. The average return will increase with duration if the risk premium is rising in pace with duration. Theory does not provide an unambiguous answer to whether there exists a systematic correlation between duration and risk premium. However, empirical studies conducted in the largest international bond markets tend to indicate that the average return rises when the duration rises up to a certain level. Beyond this level it is difficult to substantiate whether return rises when duration increases. Economists are divided as regards the critical duration level and it would appear that the critical level varies from market to market and over time. In the elaboration of the guidelines for the management of foreign exchange reserves, it was pointed out that it is difficult to substantiate whether the return rises by increasing the duration to above 2-3.

Portfolio simulations show that the duration which minimises a portfolio’s interest-rate risk increases with the investment horizon. The revised projections for the Petroleum Fund, which show that it may not be necessary to draw on the Fund for more than 15-20 years, may indicate that the Fund’s duration should be increased. A possible objection to increasing the duration is that it will lead to greater short-term variations in return. It can be argued, however, that such variations should not be decisive for choosing duration in view of the long investment horizon of the Fund.

Norges Bank proposes that the interest-rate risk associated with bond investments should be regulated by confining the duration of the Fund’s total bond portfolio to an interval of 3 to 7. The benchmark portfolio for bond investments in each country should be set at the same level as one of the internationally recognised market indices. The duration for these indices will vary somewhat from country to country but the duration for the Fund’s total bond portfolio will most likely be near the mid-point of the interval proposed.

Credit risk

Credit risk arises because the issuer of securities or the counterparty in a transaction may not honour the associated obligations. Issuers of securities may be the public authorities, banks, investment firms and companies, whereas typical counterparties are banks or investment firms. The Petroleum Fund’s credit risk can be regulated by drawing up requirements as to the type of issuer which is accepted and the creditworthiness of the issuer and counterparty.

The existing guidelines for the Petroleum Fund require issuers of securities to be sovereign states, companies with government guarantees or international organisations. However, if the guidelines are extended to include investment in equities, it should be considered whether the Fund should also have access to invest in bonds issued by private companies. In a bankruptcy situation bond holders normally have higher priority than shareholders. Norges Bank is of the view that private bonds should be an interesting alternative for the Petroleum Fund in that such investments enhance the efficiency of overall credit and interest rate risk management. This is illustrated in empirical studies from US markets which show that a portfolio with private and government bonds feature lower volatility and higher returns than a portfolio confined to government bonds only. Norges Bank recommends the inclusion of private bonds in the Petroleum Fund’s portfolio. However, such investments should not be made until the necessary risk management systems are in place. Requirements relating to creditworthiness are generally based on the evaluations of internationally recognised rating agencies. Under the existing guidelines for the Petroleum Fund it is the central bank governor who draws up the rules relating to credit assessment. This means that the central bank governor has set a limit for the creditworthiness required of a bond issuer if the bond is to be included in the Fund’s portfolio. The central bank governor has also stipulated the banks with which the Fund can place deposits and carry out transactions. Norges Bank recommends that the new regulation for the Petroleum Fund delegate the responsibility for issuing more detailed rules for credit risk management to the central bank governor in accordance with the guidelines drawn up by the Ministry of Finance.

Instruments

The existing guidelines for the Petroleum Fund specify all the instruments which can be used in the management of the Fund. When the guidelines are extended to include investments in equities and private bonds, this list may prove to be too extensive as it is important to provide an exhaustive overview of the instruments authorised. Another problem is that new financial instruments are continuously being developed, which Norges Bank may want to used in order to enhance the cost effectiveness of management or as part of overall risk management. It may be burdensome to change the regulation every time one may wish to use a new instrument. Norges Bank is therefore of the view that it would suffice for the guidelines to stipulate that the Fund shall invest in equity instruments and interest-bearing instruments. Norges Bank should be assigned the task of selecting the most appropriate types of instruments in these asset classes. Norges Bank will ensure that sound risk systems and control routines are established before a new instrument is included in the management of the Fund.

Liquidity risk

The need for liquidity in the Petroleum Fund depends on the likelihood of having to draw on the Fund as a result of central government deficits and the liquidity required for managing the risk in the Fund.

It is hardly likely that drawings on the Fund will be required over the next few years. If this should prove necessary, however, any deficits will be known well in advance of their occurrence. If the central government budget shows a deficit, more than one year will be available for planning the disposal of securities. If the deficit should occur as a result of a deterioration in state finances over the year, a period of several months will be available for selling off paper.

Nor do risk management considerations require highly liquid assets. Risk management can in the vast majority of cases rely on derivatives such as forward contracts or futures.

This shows that the Petroleum Fund will have a limited need for liquidity. Moreover, investments in less liquid paper as a rule yield higher returns than liquid investments. Norges Bank therefore recommends that no separate liquidity requirement be stipulated in the guidelines for the Fund. The Fund’s limited need for liquidity should be satisfied by the broadly constructed portfolio including marketable securities and by setting limits for the Fund’s equity shares in individual companies.

5. Implementation of new guidelines

The proposed changes to the guidelines for the Fund will require substantial changes in the portfolio. First, investments in bonds and money market instruments will have to be shifted to equities. Second, the proposed changes in the currency and market distribution will entail a redistribution of investments among markets, primarily from Europe towards the US, Canada and Asia.

The overall conversion operation will depend on the extent to which transactions can be offset. When rebalancing emphasis will be placed on minimising transaction costs and risk management so that during the transitional phase the exposure desired is achieved as far as possible.

The amount of capital which is to be invested in stock markets is relatively small in relation to the individual markets. However, the amounts are of a size which may influence prices in the event of rapid implementation. In order to minimise the market effects the changeover should be implemented over a period. On the other hand, the timeframe should not be excessively long as new funds will probably be transferred to the Petroleum Fund at the end of next year. The rebalancing of the currency and market distribution can also be effected gradually over the year, although potential market effects are more limited.

When the new guidelines have been established, a conversion plan will drawn up. The new guidelines will result in an overall benchmark portfolio established by the Ministry of Finance. As returns on the Fund will be evaluated on the basis of the benchmark portfolio, Norges Bank recommends that the benchmark portfolio is changed in line with the implementation of the new guidelines. However, as a result of substantial shifts in investments it is likely that the difference in risk and return between the Fund and the benchmark portfolio will be wider during the implementation phase.

The transfer from the central government’s current account in Norges Bank to the Petroleum Fund is made at the beginning of each year once the budget is finalised. Norges Bank builds up the investments in advance of this by purchasing foreign exchange as part of the ordinary foreign exchange reserve operations through the year. To date the guidelines for the Fund have been virtually identical to the guidelines for the foreign exchange reserves. As a result, there have been no practical problems associated with the transfer of part of the invested foreign exchange reserves to the Fund at the beginning of the year. If the asset and market distribution for the two portfolios varies widely, the need for conversion may be substantial at the beginning of each year. Norges Bank will evaluate whether the investment horizon for a share of the foreign exchange reserves should also be extended to a more long-term horizon. If parts of the reserves are managed in line with the guidelines proposed for the Petroleum Fund, it would ease the practical transition problems. Alternatively, the transition problems can be alleviated by increasing the frequency of transfers to the Petroleum Fund.

The new guidelines for the Petroleum Fund will entail important challenges for Norges Bank in its role as manager. As stated in the Revised National Budget for 1997: "Investments in equity instruments will place increased demands on Norges Bank’s risk management and risk control". The central bank governor has appointed a project group with the mandate of improving the management of the Petroleum Fund. The group has assigned priority to working on risk management systems and selecting global custodian institutions and external managers for the Fund’s equity investments.

An important prerequisite for managing the Petroleum Fund’s overall risk is a readily accessible overview of the Fund’s positions. This may be complicated by the fact that both internal and external managers will be involved. Norges Bank is working on organising risk management. The Bank will use a global custodian institution which will be responsible for the settlement of equity transactions. This institution will also be responsible for reporting on the Fund’s overall equity exposure. Norges Bank has evaluated several prospective candidates and will make the final selection later this autumn. In addition, Norges Bank is testing risk management systems which will make it possible to provide a continuous overview of the Fund’s equity and bond exposure. Such a system will also be used to measure the relative volatility of the return on actual investments and the return on the benchmark portfolio.

Norges will make extensive use of external equity managers in the Fund’s start-up phase. The managers will be selected on the basis of an evaluation of skills, risk management systems and management costs. Norges Bank is currently evaluating managers interested in managing parts of the Petroleum Fund.

6. Summary

Reference is made to the proposal for new guidelines for the management of the Petroleum as set out in the annex.

Annex

Changes to §1 and §§4-11 are proposed.

Proposal for regulation on to the management of the Government Petroleum Fund