The Government Petroleum Fund - evaluation of potential new markets

1. Introduction

In the Revised National Budget for 1999, the Government announced that the National Budget for 2000 would include an evaluation of emerging markets that could potentially be included in the Petroleum Fund's country list. Any changes in the country list are intended to apply from 1 January 2000. In this submission, Norges Bank evaluates the countries that might qualify for the list.

In Norges Bank's submission of 16 March 1999, the criteria for inclusion of new countries were discussed, and the following approach was recommended:

"An evaluation of relevant countries should be based on the IFC¹ definition and delimitation of emerging markets. These markets must then be evaluated in terms of the requirements that should be applied with respect to settlement systems, size and liquidity. After narrowing the possibilities on the basis of an appraisal of the openness, settlement risk, legislation and liquidity of the various markets, Norges Bank recommends that the remaining countries be evaluated against a requirement of political and economic stability. To ensure consistency with the existing portfolio, remaining country candidates should then be evaluated according to their contribution to the Petroleum Fund's return and risk."

¹International Finance Corporation (IFC) is a subsidiary of the World Bank.

In the Revised National Budget the Government expressed agreement with this weighting of criteria. In this submission, additional countries are appraised according to these criteria in order to determine which countries meet the minimum requirements it is reasonable to make of countries to be included in the Petroleum Fund's country list. The reference point for these evaluations was the objective of the Fund, which is to maximise future purchasing power, given an acceptable risk level. Norges Bank has therefore limited its review to economic factors related to investing in emerging markets. Other factors, such as the political situations in these countries, are not included in this submission.

Settlement risk in emerging markets is considered under point 2. It is stressed that settlement risk is greater than in developed markets and that the legal framework is less reassuring.

It is important that financial markets in which investment is to take place be of a certain size and relatively liquid. Under point 3, criteria relating to size and liquidity are used to distinguish among emerging markets.

Historical return and risk in emerging equity markets are discussed under point 4. The fact that the risk in emerging markets is substantially greater than that in developed markets is a central point. This means that considerable losses will be incurred at regular intervals on investments in individual countries. There is a lesser risk if investment is spread over several emerging markets. Even when investments are made in several emerging markets concurrently, periods with very negative returns will still occur, and the risk will remain substantially greater than that associated with investments in developed markets.

From a portfolio perspective the risk associated with investment in individual emerging markets or groups of emerging markets is not of the greatest significance; rather, it is the risk associated with combinations of the Petroleum Fund's existing portfolio and investments in emerging markets. This submission presents calculations showing that historically it would have been possible to invest a smaller amount in emerging markets, and achieve a reduction in the portfolio's risk without causing a reduction in return. We emphasise that these calculations are based on historical data and that it can not simply be assumed that history will repeat itself. Therefore, as a supplement to analyses of the historical returns data, we also examine certain country risk indicators, since these are based on the current situation in these countries.

In the previous submission it was stated that an increase in the country list could, in principle, apply to both equities and bonds. It was pointed out that most of the discussion in theoretical and empirical literature focuses on emerging equity markets, and that it is easier to establish good benchmark indices for equities. This makes it easier to evaluate the effect on the Petroleum Fund's risk of investing in equities rather than bonds. Consequently, this submission will focus on equity markets in emerging economies. Three other factors support this conclusion. First, even if the equity market of a given country satisfies certain minimum requirements it does not necessarily follow that the country's bond market fulfils similar requirements. Similarly, the bond market in a country may fulfil the minimum requirements while its equity market does not. A possible consequence of this is that the country list for equities might be different from that for bonds. Second, literature on the subject indicates that the bond markets in emerging economies have some of the same return and risk characteristics as equity markets, ie that the returns in the markets show a high degree of covariation. This makes it important to assess the marginal benefit of investing in bonds in addtion to equities. The third reason that we are focusing on equity markets in this submission has to do with the structure of bond markets in emerging economies. A substantial portion of these markets consists of bonds denominated in USD. Local currency bond markets are not particularly well developed as a general rule. In both of these markets the credit risk is considerably greater than that in bond markets in developed economies, and in many cases these countries do not comply with the guidelines stipulated for credit risk in the Petroleum Fund. It might also be mentioned that approval has recently been granted for up to 10 per cent of the fixed income portfolio of the Petroleum Fund to be invested in private bonds in developed markets. This is a decision that presupposes changes in the operational management which have not yet been made. For this reason it may be appropriate to delay the decision to allow investment in emerging bond markets until investments have been made in private bonds.

Norges Bank aims to analyse emerging bond markets at a later time and will submit a recommendation concerning additions to the country list for bonds.

The examination in this submission is a partial one, in the sense that the various risk factors are discussed independently of one another. In reality, settlement, liquidity and price risk are closely interrelated. For example, an exogenous shock in the form of a sharp price reduction on an important export item may led to a cyclical downturn, which is reflected in a sharp fall in prices on the stock exchange and perhaps a depreciating currency as well. This in turn may lead international and national investors to withdraw capital from the country. In addition to exacerbating the fall in equity prices and of the exchange rate, these capital flows may result in a severe tightening of liquidity in the securities markets. A possible outcome is that many investors do not manage to get out of the market until after prices have fallen significantly and major losses have been sustained. Substantial price drops and capital flows out of the country may destabilise the country's financial system, thereby causing a sharp increase in settlement risk. Dynamic relationships of this kind are highly significant in practice, and there are a number of examples of this type of compound crisis in emerging economies. This means that once the decision is made to invest in these markets, the risk may be very high in periods and may also change very quickly. Investments in emerging markets must therefore be monitored very closely, and there will be higher costs associated with them than with investments in developed markets.

2. Settlement systems and legislation

Norges Bank has chosen to base its analyses on the IFC's definition and delimitation of emerging markets. IFC divides the markets into three main categories: developed markets, emerging markets and frontier markets. IFC classifies an equity market as emerging if it fulfills at least one of two general criteria:

1) The World Bank classifies the country as being a low or middle income economy. This criterion reflects the country's economic development.
2) Compared with developed markets, its investable market capitalisation is low relative to GDP. This criterion reflects how well developed a securities market is.

The third category, frontier markets, includes markets that are small, non-liquid and with a flow of information that is poor even by the yardstick applied to emerging markets.

IFC has constructed equity indices for emerging markets. The IFC-I index was established in December 1988 and includes companies and countries that foreigners can actually invest in, in the sense that these countries are on the whole open to foreign portfolio investment, and that the size of their companies and the liquidity of the markets fulfils certain minimum requirements. Table 1 provides an overview of the emerging markets included in the index (the share of total market value in the index is given in brackets)

Table 1: Emerging markets included in IFC-I

² In Malaysia, restrictions were imposed on outgoing capital flows in autumn 1998.
³ In March 1999, Portugal was classified as a developed market and thus is no longer included in the IFC emerging markets indices.

However, it is important to stress that the fact that an investment bank or an international organisation decides to produce indices for the equity market in an emerging economy, or include the market in an index that measures developments in regional or global equity markets, cannot in itself be used as a criterion for the quality of or risk associated with the market. In some cases, investment banks construct indices to satisfy specific inquiries from investors, who want to invest in specific markets and therefore need a benchmark index. This means that a market may be defined as emerging because some investors are interested in investing in it. As a general rule, the criteria used by the producers of indices to evaluate countries of potential interest do not provide an assurance that the liquidity of the markets is satisfactory, that the markets are large enough or that transaction costs do not appreciably exceed costs in developed markets. Nor is the fact that a market is included in the IFC-I or some other index a guarantee that the risk associated with the clearing and settlement systems is acceptable, or that the country's legislation meets the requirements that it is natural for an investor such as the Petroleum Fund to make. And it cannot be taken as given that all countries included in an index automatically complement the Petroleum Fund's existing securities portfolio. These factors indicate the need for an independent evaluation to determine which countries should be included in the Petroleum Fund's country list.

In Norges Bank's previous submission, it was pointed out that the settlement risk associated with emerging markets is greater than that in developed markets, and that the legal framework is less than adequate, and reference was made to the GSCS risk indices.⁴ This section provides a description of settlement risk in emerging markets based on these indices, the evaluation system of Norges Bank's global custodian institution⁵, and recommendations from international institutions.

⁴This is a private British limited company which constructs indices that measure settlement risk in emerging markets, among other things.
⁵The Chase Manhattan Bank

Custodian institutions have a detailed knowledge of the settlement systems of various countries which it is logical to use in assessing the settlement risk of emerging equity markets. For example, global custodian institutions regularly enter into agreements with sub-custodians in the various countries, and are responsible for ensuring that the correct owner is registered as holder of the securities. Custodian institutions are also responsible for ensuring that payment for and delivery of securities takes place on time. Often the global custodian also administers tax affairs and ensures that interest revenues and dividends are paid on time. Moreover, a custodian institution can keep accounts for securities, and calculate the return on the portfolio.

One central objective of settlement systems around the world is to reduce the risk associated with settlement for securities, and a number of international organisations have recommended establishing systems that satisfy certain requirements regarding the organisation of settlement of securities trading. G-30 has defined "best practice" in clearing and settlement of securities trading.⁶ Both custodian institutions and others who evaluate the risk in the settlement systems of various countries often take as their starting point the recommendations of the G-30 or the BIS⁷.

⁶G-30 is a private alliance of international investment banks, other banks and other key private participants in the financial system. See 'Clearance and Settlement Systems in the World's Securities Markets', March 1989, Group of Thirty.
⁷Bank for International Settlements

In simplified terms, the risk associated with clearing and settlement systems can be split into two main types: risk arising from the fact that the market value of a security may change between the trade date and the settlement date ('market exposure') and risk that the counterparty will not make payment on or after settlement date ('counterparty risk').

In the first category, a number of factors need to be considered. These are discussed briefly below:

• Trade matching: If all trades are matched before settlement, the risk of trade fails is reduced. Trade matching involves identifying counterparties and checking contract terms to ensure that misunderstandings are avoided. When a trade "fails", it is not approved by the system, and this effect can be transmitted to other investors. G-30 maintains that a system should be established to ensure that direct market participants (brokers, market-makers etc.) check that transactions have been correctly recorded on the day following the trade day at the latest. Moreover, G-30 believes that indirect market participants such as institutional investors should be included in a reconciliation system in which details of transactions are confirmed by participants. Central to this process is the establishment of a securities identification system such as ISIN (International Securities Identification System).
• Settlement type: Many settlement systems require that market operators have substantial capital and liquidity, to enable them to finance activities in the period until payment is received for securities that have been sold. The larger the outstanding amounts, the greater the risk that the suspension of payments by a bank or broker will have a domino effect and cause a systemic crisis. Real-time gross-settlement systems (RTGS) enable transfer of title and payment to take place electronically and simultaneously. Such systems make it possible to use cash the same day, which reduces the time lag of traditional settlement systems. There are also some types of netting systems which reduce settlement risk. The G-30 recommends that systems be introduced for netting of payments and securities if this contributes to reducing risk or increasing settlement efficiency.
• Security type: In many markets securities still have a physical form, while securities in other markets, such as the Norwegian market, exist only in electronic form in one or more registers. G-30 stresses that settlement risk will be lower in systems in which securities exist only as electronic entries on an account in a depository.
• Depository: G-30 advocates that all countries should have a central depository organised in such a way that as many direct and indirect market operators as possible can participate in the system.
• Failure provisions: To prevent trades failing it may be necessary to impose fines, or in some other way to penalise market operators, to have buy-in provisions and to establish systems for securities lending and repurchase agreements. Such rules and procedures may either prevent a trade from failing, or contribute rapidly and efficiently to remedying failed trades.
• Trade date/settlement date lapse: The shorter the lag between the trade date and settlement date, the less the mark-to-market exposure. G-30 stresses that risk is lowest in a rolling settlement system, and that the settlement period should not exceed three days.
• Regulatory oversight: Market rules should ideally be applied consistently and effectively. It is important that the overseeing authorities have substantial experience in dealing with complex market issues, and at the same time encourage market stability.
• Securities legislation: The establishment of an effective legal framework and an arbitration process that assures a satisfactory resolution of conflicts between participants in the settlement system is regarded a important for reducing risk in the settlement system.

Table 2 shows the average rating for the 8 assessment factors included in the first main category. For the sake of comparison, the classifications of a number of developed markets, all of which are included in the Petroleum Fund's investment universe, are also given. The ratings are based on The Chase Manhattan Bank's evaluations of the various aspects of the settlement systems of the individual countries. If an aspect of a country's settlement system is considered to be on the same level as the world best, it receives the rating 1 ('Best practice'). If the organisation is on a level with the world average, it is awarded a rating of 2 ('Average practice'), while an organisation on a level with the world's poorest practice is assigned a rating of 3. Norges Bank has chosen to give the various categories the same weights when calculating average ratings. This is not necessarily the best solution, but nor is it immediately apparent which risk components should receive the most weight in an assessment of the overall risk associated with the settlement systems.

The table shows that mark-to-market exposure in the settlement systems of the developed economies is substantially lower than that in emerging economies. This underscores the qualitative difference between developed and emerging markets, and is a clear warning that once investment in this type of market is permitted, a number of unforeseen events can be expected to occur over time.

In the second main category (counterparty risk), three components are assessed from a buyer's perspective (buy-side assessment), and the same three components from a seller's perspective (sell-side assessment). A brief description follows:

• Timing of receipt of good title: In some settlement systems, there is a lag between the settlement date and the date that the buyer obtains good title to the securities. If the seller's broker, and in some cases the seller itself, goes bankrupt, the investor may have paid for the securities without having any collateral other than a claim on the counterpart
  Sometimes cheques or irrevocable promises to pay are used as a means of payment. In such systems, the seller of a security bears the risk that in a bankruptcy situation the broker or broker's bank will not honour the cheque or promise to pay. Exposure increases with the length of the period from the settlement date until good funds are received.
• Interim exposure type: In practice, the actual transfer of ownership of securities can take place in several ways. The most satisfactory is when the transfer takes place in an electronic accounting system organised and operated by a central depository. Manual transfer/registration carried out by a third party, such as a stock exchange or custodian bank, is regarded as mid-range practice, while manual transfer/registration carried out by a broker or counterparty is regarded as worst practice. For a seller of securities, the risk exposure may be to a central bank, a local commercial bank or a brokerage firm. Being exposed to a central bank is more secure than being exposed to a brokerage firm.
• Default protection. Good default protection offsets the risk exposure described in the two previous points. Such protection may consist of stock exchange guarantees or collateral, participant guarantee funds, insurance funds, rules for loss-sharing between operators, and central bank guarantees. G-30 regards systems in which payment and delivery take place simultaneously (DVP) as a type that eliminates the risk associated with one of the parties not fulfilling its commitments.

Table 2 provides an overview of the classification of counterparty risk in settlement systems. The assessments are based on The Chase Manhattan Bank's system, but Norges Bank has calculated the average rating itself by giving the various sub-categories the same weight. From the table we see that emerging markets perform more poorly than developed markets, and the differences are particularly large from the buyer's perspective. This underscores the conclusion that settlement risk in emerging markets is substantially higher than that in developed markets.

Table 2: Evaluation of settlement systems

Average ratings for various aspects of settlement risk in selected equity markets are presented in the table. The rating in the various categories indicates how well the various aspects of the settlement systems function compared to best, average and worst practice in the world. A rating of 1 indicates a settlement system on a par with the world best, a rating of 2 a system on a par with the world average, and a rating of 3 a system that functions more poorly than the average. This is thus a fairly rough sorting of emerging markets. In Category 1, the risk of the market value of the security changing between the trade date and the settlement date is assessed. Category 2a is an assessment of how well the system ensures that buyer actually receives good title to securities he has paid for, while Category 2b is an assessment of how well the system ensures that the seller of securities receives payment for the sale.
In Category 1 there are eight sub-categories, and in Category 2 there are six sub-categories: three viewed from buyer's perspective and three from seller's. The last column represents an average of the ratings in categories 1, 2a and 2b

The next to last column in the table shows the average rating for the two main categories. A rating of 2 indicates that the risk exposure in the settlement system is equivalent to the global average. The developed equity markets have a substantially higher standard than the average. The Swiss equity market appears to represent the lowest settlement risk, but Norway also shows up well, and Portugal, which is the newest country in the investment universe, has an average rating of 1.5.

It appears natural to require a minimum standard for the settlement systems in the countries in which the Fund's capital can be invested. The level of a minimum requirement is open to discussion, but for the Petroleum Fund it is probably not advisable to invest in countries which on average are assigned ratings much poorer than 2.0. The following countries will satisfy such a requirement for clearing and settlement systems:

Argentina, Brazil, Mexico, Peru, South Korea, Malaysia, Sri Lanka, Taiwan, Thailand, the Czech Republic, Greece, Hungary, Poland, Slovakia, Turkey, Israel and Morocco.

On the whole, there is a high degree of consistency between the evaluations of risk index producers and global custodians. The table shows that Greece has an average of 2.14, and for that reason it should not have been included on the list above. It has nevertheless been included because Norges Bank's global custodian has used various means to reduce the risk of the Greek settlement system, with the result that the adjusted rating is roughly the same as the average rating of 2.0. China is an example of a country with an average rating that is marginally poorer than the minimum requirement.

3. Size and liquidity

i) Size
It is important that markets be of a certain size in both an absolute sense and in relation to the Petroleum Fund's investments. There are considerable differences in market size. Because market values in USD fluctuate widely as a result of changes in the dollar exchange rate, and because market values in local currencies are also relatively volatile, a requirement regarding size should refer to an average of the market values over a certain period.

Table 3 shows the market value of 30 emerging equity markets that are included in IFC-I, calculated as an average of the market value at the end of the year for the past three years. Also included, to make relevant comparisons possible, are market values for a selection of developed markets that are already included in the Petroleum Fund's investment universe. The table also shows the individual market's share of the aggregate value of all emerging markets and the number of listed companies. These data sets have also been calculated from data for the last three years.

The majority of emerging equity markets have grown substantially during the past decade. Overall, the growth of emerging markets has been slightly greater than that of developed markets.

A natural starting point for a minimum requirement may be the size of one of the smallest markets in which investment can currently take place. Portugal is both one of the smallest markets, and the most recent country to be included on the Petroleum Fund's country list. The average market value of the Portuguese market in the past three years has been slightly over USD 40 billion. By way of comparison, the value of the Norwegian market, also a small market, has been roughly USD 60 billion.

It may seem arbitrary to set an absolute minimum requirement on the size of the market, since there may be small markets that represent interesting investment possibilities. However, small markets are often volatile, and at the same time, for an investor like the Petroleum Fund, they do not offer any particular diversification gains. For operational reasons, it may be necessary to limit the number of new countries in the investment universe. One way of doing so is to range countries according to size.

The value of the equity market in the following countries/regions exceeds USD 40 billion:

Argentina, Brazil, Chile, Mexico, China, India, Indonesia, South Korea, Malaysia, the Philippines, Taiwan, Thailand, Greece, Russia, Turkey, Israel and South Africa.

The aggregate value of these 17 markets represents about 88 per cent of the aggregate market value of all emerging equity markets.

ii) Liquidity
It is important that the liquidity of the markets is good, and not substantially poorer than that in developed markets. Good liquidity is important for a number of reasons. First, investors must be able to buy and sell equities in normal quantities in a company and in a market without the transactions leading to major price changes. Low liquidity may result in a sale depressing prices, and a purchase pushing prices up. In both cases, the investor will lose compared with a situation with better liquidity and more limited market influence. Second, it is important for an investor to be able to 'exit from' the market in crisis situations without incurring major losses. This latter is particularly difficult in emerging markets, in which liquidity typically weakens substantially when volatility increases strongly.

Table 3 shows the average daily turnover (252 trading days) in the period 1996-98, the individual market's share of total turnover in emerging markets, turnover ratio and average turnover per company.

There was substantial growth in the annual turnover in a number of the emerging markets in the period 1989-98. Nevertheless, growth in developed markets has been higher on average. Daily turnover has been significantly higher in the developed equity markets. However, in some emerging markets daily turnover has been higher than in small developed markets such as Portugal, Norway, New Zealand and Finland. An appropriate minimum requirement for liquidity in emerging markets is that daily turnover is not lower than the turnover in the smallest markets in the current investment universe. In Portugal, the average daily turnover during the period 1996-98 was USD 100 million, and in New Zealand USD 111 million.

Table 3:

The table shows market values in USD for selected emerging markets, calculated as the average of market values at the past three year-ends. The table also shows how large a share the individual emerging market represents of the market value of all emerging markets, and the number of listed companies in each market. The next three columns show average daily turnover (252 trade days), the individual market's share of total turnover and the turnover ratios of the various markets. The rate of circulation is defined as the ratio between annual turnover and the average market value for the year. In order to be able to assess how large and liquid the emerging markets are, corresponding information is given for developed markets which, with the exception of Norway, are already included in the Petroleum Fund's portfolio. All figures are calculated as averages for the past 3 years.
The data is obtained from Emerging Stock Markets Factbook 1999, International Finance Corporation.

One ratio used a great deal in comparisons of liquidity in various markets is turnover ratio. This is defined as annual turnover divided by average market value during the year, and is an expression of the share of total market value that is sold during the course of the year. In the developed markets this ratio is lowest in New Zealand, where approximately 40 per cent of the average market value changed hands in the course of one year.

Since both ratios measure liquidity, but in slightly different ways, it may be appropriate to make it a condition that the market fulfil at least one of the two requirements. The following emerging markets satisfy the minimum requirement regarding liquidity:

Brazil, Mexico, China, India, Indonesia, South Korea, Malaysia, Pakistan, Taiwan, Thailand, the Czech Republic, Greece, Hungary, Poland, Slovakia, Turkey and South Africa.

Eight emerging markets satisfy the minimum requirements regarding settlement systems, size and liquidity. These are Brazil, Mexico, South Korea, Malaysia, Taiwan, Thailand, Greece and Turkey. The aggregate value of these markets is 48 per cent of the total market value, while turnover amounts to 73 per cent of the turnover in all emerging markets.

In Malaysia, the crisis led in autumn 1997 to the fixed exchange rate being abandoned, interest rates being raised, and fiscal policy tightened. In autumn 1998 the authorities introduced controls on outgoing capital flows such that foreigners could only repatriate capital after a certain period had elapsed. Since then, interest rates have dropped slightly, and in spring 1999 the waiting period was replaced with a punitive tax on capital taken out of the country after a short period of time. This illustrates the risk of investing in emerging markets. As long as Malaysia maintains restrictions on outgoing capital flows, it should probably not be included on the country list for the Petroleum Fund⁸.

⁸Morgan Stanley Capital International, one of the major suppliers of indices, has decided to include Malaysia in its collective indices from February 2000, provided that the positive development of the country does not reverse. FT/S&P, which is also one of the major suppliers of indices, is awaiting further developments before taking a decision.

4. A better trade-off between risk and return

In this section, an account is provided of historic return and risk associated with investments in specific emerging equity markets, and comparisons are made with developed markets.

Table 4 shows some key variables associated with the monthly return in some major emerging equity markets, including those markets that satisfy requirements relating to settlement systems, size and liquidity. The table also shows return data for the regional indices for emerging markets in Asia, Europe and Latin America, and for emerging equity markets overall. Corresponding information for developed markets is provided to enable a comparison to be made of return and risk in emerging and in developed markets.

Calculations for most of the countries are based on data for the period January 1989 to May 1999. The time series for the other countries are shorter (cf. the information in the table). In this type of comparison, it is important to be aware that the length of the period, including start and end dates, strongly influences the results. It has been pointed out in the literature that several of the so-called emerging equity markets have a substantially longer history than is evident from the figures series of the index producers.

For example, the global indices from IFC only extend back to 1975, while the indices that foreigners actually could have replicated did not start until December 1988. Analyses based on the last decade, which is the period we actually have figures for, are therefore not necessarily representative in a longer historical perspective. The Argentine equity market is one of several that were opened to foreign investment in the early 1900s, but which stopped being an option for foreign investors for several decades. If these periods are included in return calculations for emerging markets, it can be seen that the return has been smaller, and the risk greater than that indicated by calculations based on the last decade, for example. There is an empirical basis for maintaining that historical return rates and risk vary over time, and that it cannot be assumed that the historical average pattern will be repeated in the future.

Table 4 shows that there is a wide spread between the monthly returns of the emerging markets, and that it is wider than that between the developed markets⁹. Returns in Latin American countries have been very high. This is true of Greece and Turkey, too, but the data for Turkey is based on a slightly shorter time horizon. Returns in Asian countries have been lower on average than in Latin American countries, and also lower than a very large number of developed markets. The return in South Africa has been moderate, but is based on few observations. The average return in China was negative during the period, but the calculations cover a relatively short period (51/2 years).

⁹To ensure comparability, all return rates are calculated in USD.

It is only in Latin America that the average return has been higher than that in developed countries. The return in the emerging markets in Europe, Asia and the Middle East and Africa have been lower than those in the developed markets.

The second column of figures in the table shows the standard deviation of the return. This is a measure of the size of fluctuations in the monthly return rates, and describes the risk of investing in an individual market. The fourth column of figures shows the smallest monthly return observed during the period. The figures in both these columns show that the risk in emerging equity markets is substantially greater than in emerging markets. The Norwegian equity market is often described as a risky market but, as the table shows, the risk associated with emerging markets is considerably greater. Whereas the lowest monthly return lies at between minus 40 and minus 30 per cent for a significant proportion of the emerging markets, negative return rates of 20 per cent and worse are far less common in developed markets.

The table shows monthly historical rates of return for key emerging and developed markets and for selected portfolios. Also shown are the standard deviations of returns, the largest and smallest monthly return in the period, and the proportion of months with a negative return. The standard deviation of the annual standard deviations (moving standard deviation) has also been calculated to provide an impression of how variable risk is over time. Both risk and fluctuation in risk are substantially higher in emerging markets than in developed markets.

The calculations for the portfolio '7 emerging' are based on returns in Brazil, Mexico,
South Korea, Taiwan, Thailand, Greece and Turkey. The calculations for the portfolio '4 emerging' are based on returns in Mexico, Korea, Taiwan and Greece. The Petroleum Fund portfolio consists 50 per cent of European equities, 30 per cent of North American equities and 20 per cent of equities in Japan, Australia, Singapore and Hong Kong.
MSCI W is MSCI's world index (developed markets).
Source: empirical data: IFC
The index series for Peru starts in September 1993.
The data for Turkey starts in August 1989.
Data for Taiwan is available from January 1991 and for Korea from January 1992.
The time series for India starts in November 1992 and for China in October 1993.
Data for South Africa is available from January 1994.
All other time series start in December 1988.

In Brazil, for example, the lowest recorded monthly return during the period is minus 67 per cent, which means that in that month an investor could have lost 2/3 of invested capital. A similar extreme observation has been recorded for Argentina. This means that on average an investor can expect less return stability in emerging markets, and that the largest negative return rates may be substantially larger (in absolute value) in emerging markets than in developing markets.

The fifth column of figures in Table 4 shows the number of months, expressed as percentages, in which the return has been lower than or equal to zero. The number of months with a negative return is on average larger for investments in emerging markets, and the spread is also larger (cf. also Chart 1).

The last two columns of Table 4 show the standard deviation of the moving standard deviation¹⁰ and the largest moving standard deviation. The moving standard deviation is calculated for the last 12 months at any time, and the standard deviation of the moving standard deviation is consequently a measure of how the historical risk in a market has changed. A high figure implies major fluctuations over time. The largest standard deviation provides an indication of how risky a market can be in a single year. The table shows that by investing in emerging markets, one can expect substantially larger variations in risk over time than by investing in developed markets. The level of risk associated with emerging markets in risky periods is also substantially higher than that associated with developed markets in risky periods.

¹⁰The standard deviation of the return measures the spread in return around the average return during the period. The moving standard deviation measures this spread each month on the basis of the return for the past 12 months.

Investing in a number of countries and regions will reduce the risk, because the return rates are not perfectly correlated. Chart 1 shows the spread in the historical rates of return for a portfolio consisting of seven emerging equity markets. This portfolio consists of the seven markets that satisfy the minimum requirements regarding settlement systems, legislative regulation, size and liquidity with the exception of Malaysia, which has introduced restrictions on outgoing capital movements. The portfolio of emerging markets is equally weighted, ie each country makes up 1/7th of the portfolio. The chart also shows the spread of a portfolio of developed markets (the `Petroleum Fund') and the spread of individual markets. Some important return figures for the portfolios are presented in Table 4. The spread and maximum monthly loss in the historical return on the portfolio of emerging markets are substantially less than those for any individual emerging market, but at the same time substantially larger than those for the 'Petroleum Fund's' equity portfolio. Whereas the first comparison illustrates the advantages of diversifying over a number of countries and regions, the second shows that the risk associated with portfolios of emerging markets is greater than that associated with portfolios of developed markets.

Although an emerging market may be very risky in isolation, it is not equally certain that marginal investments in such a market will increase the overall risk of the portfolio. If the covariation between the return in an emerging market and the return on the original portfolio is low, it will be possible to reduce the risk in the portfolio by including the emerging market in the investment universe.

If the fluctuations in the country's equity market are large relative to those in the equity markets of developed countries, however, risk may increase even if there is low covariation.

Chart 2 shows historical development in moving 12-month correlation coefficients, which measure precisely the degree of covariation between some key emerging markets and the equity portion of the Petroleum Fund's portfolio. The correlation coefficients may vary between minus one and plus one, and on the chart the numerical value can be read off the left vertical axis. This covariation fluctuates considerably over time, which indicates that diversification gains are smaller in some periods than in others. Covariation appears to have increased in most emerging markets in recent years. The chart suggests that there may also be a positive correlation between developments in the standard deviation for the individual market and developments over time in covariation. This may indicate that diversification gains are smallest when they are needed most.

[Chart 1]

[Chart 2]

Any investments in emerging equity markets will be additional to the Petroleum Fund's investments in developed equity markets. The calculations above are based on the return and risk of portfolios composed exclusively of investments in emerging equity markets. To show the effect on the total equity portfolio of the Petroleum Fund, return and risk have been calculated for portfolios consisting of investments in both emerging and developed equity markets. The portfolio of developed markets has a composition corresponding to that of the Petroleum Fund. The emerging equity markets are represented by the portfolio consisting of those countries that satisfy requirements regarding settlement systems, size and liquidity, with the exception of Malaysia. To illustrate the effect of reducing the number of emerging markets, we have also calculated return and risk for combinations of the Petroleum Fund and a portfolio of four emerging equity markets. Brazil and Turkey are not included in this last portfolio, because during some periods the risk of investing in these countries has been quite substantially greater than that in the other countries, and because the country risk is regarded as relatively high, and higher than in the other countries. Because Asia is represented by three countries, Thailand has also been excluded, since this market is smaller and has poorer liquidity than those of Taiwan and South Korea. Table 5 shows monthly return and standard deviation for various combinations (weights) of investments in emerging and developed markets. The proportion invested in emerging markets varies from 0 to 100 per cent.

It is clear from the table that the risk associated with the Petroleum Fund could be reduced by investing a small proportion of the Fund in emerging markets. However, these gains would not have been particularly large. The greatest risk reduction would be achieved by investing in the portfolio consisting of four emerging markets. This result may seem rather surprising, since one normally expects that the more countries included, the greater the risk reduction. In this case, much of the explanation lies in the fact that the correlation coefficient between the return on the Petroleum Fund and on the "seven-country" portfolio is higher (0.43) than that between the Petroleum Fund and the "four-country" portfolio (0.40), and that the risk in Brazil and Turkey in isolation is substantially higher than that in the other countries.

Table 5:
Return and risk for combinations of the Petroleum Fund
and investments in emerging markets.

The table shows the standard deviation and return on a portfolio consisting of investments in emerging and developed markets. In the first case, the emerging markets portfolio consists of investments in 7 countries, and in the second of investments in 4 countries. If investment had not taken place in emerging markets, the return for this period would have been about 0.8 per cent per month, while the standard deviation of the return would have been 4.6 per cent. If the whole portfolio had been invested in emerging markets, the return would have been 2.25 per cent per month, while the standard deviation would have been 8.83 per cent for the portfolio of 7 countries and 8.15 per cent for the portfolio of 4 countries.
It is clear from the table that total risk could have been reduced by investing a limited portion in emerging markets. We see for the 7-country portfolio that a share of up to 30 per cent would not entail a greater risk than investing the Petroleum Fund only in developed markets.

Source: primary data: IFC

This type of analysis is highly sensitive to changes in return rates and risk parameters, and there is great uncertainty attached to estimated rates of return, standard deviations and correlation coefficients. There are also weaknesses associated with the tool. Decisions to expand the investment universe can therefore not be based on analyses of this kind alone. It appears more appropriate to use this type of analysis as a supplement to the evaluations made of settlement systems, legislation, size and liquidity.

Fluctuations in share prices reflect to a large degree changes in macroeconomic and political stability. Experience shows that these are parameters that can change rapidly in emerging markets. This means that studies of historical return figures for these markets do not necessarily provide a good indication of how risk will be in the future. However, a number of institutions calculate indices for measuring the risk associated with investments in various countries. These country-risk indices are forward-looking, and can provide a useful supplement to risk assessments based on the volatility of historical rates of return. Table 6 shows how five institutions assessed the macroeconomic and political risk in some emerging markets in summer 1999. The table shows that the risk associated with investments in Turkey and Brazil is regarded as high at present, while it is regarded as low in Taiwan, Chile, South Korea and Malaysia. This in turn underscores the importance of spreading investments over a number of countries, in order to reduce the country-specific risk. One disadvantage of such risk indices is that they are often based on expected developments over relatively short time horizons, whereas the investment horizon of the Petroleum Fund is very long-term. The short time horizon means that the index levels of the individual countries may vary substantially over time. A country that appears risky in the short term may represent a satisfactory investment option in the longer term. Both these factors imply that circumspection should be employed in basing a recommendation to expand the country list on such indices and ratings.

Table 6: Country risk indicators

Sources:
The Economist, scale from 0 to 100 with 100 as highest risk
GIEK (Garantee Institute for Export Credits), scale from 1 to 7 with 7 as highest risk
J P Morgan, scale from 0 to 100 with 100 as highest risk
Moodys, alphabetical ratings with scale from Aaa (best) to C (poorest)
For example, Baa represents a lower risk than B
Standard & Poor, alphabetical ratings from AAA (best) to CC (poorest)
Bb represents less risk than B

The Economist Intelligence Unit, GIEK and J P Morgan are three of many institutions that rate the country risk of emerging economies. GIEK's scale runs from 1 to 7, with 1 as the lowest and 7 as the highest risk. The scales of The Economist and J P Morgen run from 0 to 100, and risk rises with the numerical value. Both Taiwan and Chile are regarded as low-risk countries, while Brazil and Turkey are high risk. It could be useful for investors to be able to place an upper limit on country risk. However, this is difficult, both because there is neither a theoretical nor an empirical basis for setting such a limit, and because the ratings of countries are changed relatively frequently. It is also clear from the table that the ratings of the different institutions vary somewhat. Moreover, a high country risk may be reflected in a high expected return in the securities market. In such case, an investor may want to invest in high risk countries.

Moody's and Standard & Poor's are two credit rating agencies used by many investors to set up investment guidelines. Both award alphabetical grades to bonds issued by national states, and it is these that are shown in the table. Moody's scale runs from Aaa (best) to C (poorest), so that for example Baa implies a lower risk than B. Standard & Poors' scale runs from AAA (best) til CC (poorest), so that Bb represents less risk than B. According to the current guidelines for credit risk in the Petroleum Fund, the fund could not invest in bonds with a rating poorer than Baa3 (Moody's) or BBB- (Standard & Poor's). This means that for example Thailand (Ba1) would be disqualified according to Moody's rating system, and South Africa (BB+) according to Standard & Poor's system

5. Summary

In order to be able to invest portions of the Petroleum Fund's capital in new countries, it must be assumed that the new markets have satisfactory clearing and settlement systems, size and liquidity. A further condition is that the legislation must assure orderly conditions and transparency in the securities markets, and that oversight of financial markets is organised in a satisfactory manner. With respect to the portfolio, it is important that the new markets complement the existing country and asset distribution. In this submission, Norges Bank has considered which emerging equity markets fulfil these criteria. The starting point for the evaluations is the objective of the Fund, which is to maximise future capital at an acceptable risk level. Norges Bank has therefore focused only on economic factors in its assessment of the countries, and political factors have consequently not been discussed.

Investing in emerging markets is highly resource-intensive. Operational considerations thus dictate that the country list should not be expanded to include too many new countries initially. Consequently, some countries are not recommended for inclusion on the new country list, even though the markets in these countries do not distinguish themselves markedly, according to relatively objective criteria, from the markets in the countries it is recommended including on the list. Moreover, there has been no discussion as to whether any expansion of the investment universe other than that of including new countries would provide a better means of achieving the objective of the Fund. Because of the need to prioritise resources in operational management, in principle an overall evaluation of this kind should be made before any investment in emerging markets takes place.

The evaluations in the submission are restricted to the equity markets in emerging economies. Norges Bank aims to make a thorough evaluation of bond markets at a later date, and will then submit a proposal with respect to expanding the country list for bonds.

Emerging markets are generally smaller and have poorer liquidity than developed markets. The risk associated with the clearing and settlement systems in emerging equity markets, including those markets that fulfil the minimum requirements, are normally quite substantially higher than that associated with developed markets. It is generally the case that the legislation does not assure orderly conditions in the securities market to the same extent as in developed markets, and that surveillance of the participants in financial markets is not as well developed. As pointed out in this letter, the various elements of the settlement systems have been given the same weight in the final evaluation, and the minimum requirements for settlement systems have been set on a discretionary basis. Norges Bank must therefore make a separate evaluation of settlement risk before investing in a market.

The return in emerging markets varies more than that in developed markets. From time to time, significant losses will therefore occur on investments in both individual emerging markets and emerging markets as a category. The risk to the Petroleum Fund as a whole is limited somewhat by the relatively low covariation between the yields in emerging and developed markets respectively.

In our evaluations of new countries, no assumptions have been made about the future return in the various markets, and the choice of countries is not based on expectations that the return in these countries will be higher than the return in other emerging economies or developed markets.

After evaluating the various countries on the basis of the criteria described above, we are left with a core of emerging markets that fulfil the minimum requirements. In Latin America these are Brazil and Mexico, and in Asia, South Korea, Malaysia, Taiwan and Thailand. In Europe, Greece and Turkey qualify. Malaysia currently has controls on outgoing capital flows which makes it a less attractive market. The macroeconomic risk in Brazil and Turkey is currently regarded as high. However, it does not seem appropriate to place too much weight on the current economic situation when evaluating new countries.

There is considerable work in progress in a number of the emerging markets to make clearing and settlement systems more efficient and to improve securities and company legislation. There is concurrent work in progress to improve the stability of the financial systems, and publicly owned companies are being privatised. These factors may contribute to improving liquidity such that a larger proportion of capital intermediation takes place in securities markets. In the future, more countries may therefore satisfy the minimum requirements.

One key question when it comes to expanding the country list for the Petroleum Fund is whether the new countries ought also to be included in the Fund's benchmark portfolio. If emerging markets are included in the benchmark portfolio, Norges Bank will have a strong incentive to invest in these countries. This is because the share the new countries will constitute of the benchmark portfolio will be the neutral starting point for Norges Bank in its management of the Fund. If the Bank chooses to deviate from this share, the return on the Fund could vary substantially in relation to the return on the benchmark portfolio, as a result of the wide short-term fluctuations in emerging equity markets. Because Norges Bank's management is evaluated relative to the performance of the benchmark portfolio, Norges Bank will therefore be cautious about deviating from the share emerging markets constitute of the benchmark portfolio. In consequence, Norges Bank will very probably not invest in emerging markets before the new countries are included in the Fund's benchmark portfolio.

It is difficult to judge whether emerging markets should be included in the Petroleum Fund's benchmark portfolio. The most usual way of deciding is to use an empirical portfolio model to see how the Fund's return and risk are affected by the inclusion of new countries. In this letter we have presented the results of such a model, which shows that including emerging markets in the Fund's portfolio may result in slightly lower risk. However, this type of study focuses on short-term variations in the return, whereas the management of the Petroleum Fund is based on a long-term investment horizon, and consequently it is the long-term return and risk that is of greatest interest. Since there are not sufficient data to permit a study of return and risk over long-term investment horizons, one alternative is to use more general principles as a basis for evaluating whether emerging markets should be included in the Petroleum Fund's benchmark portfolio. When establishing the country weights applying to the Petroleum Fund, for example, considerable emphasis was placed on the countries' import weights and GDP weights. Emerging economies represent some 20 per cent of world GDP, and around 15 per cent of Norway's imports. In isolation, this may indicate that emerging markets should be included in the Fund's benchmark portfolio. However, caution should be exercised in using such analyses when it comes to emerging markets. There is considerable settlement risk, and unusually large variations in the return on investments in these countries. Far more resources are also required to invest in emerging markets than in developed countries. Since Norges Bank already faces significant challenges in the operational management of the Petroleum Fund, we recommend waiting to include emerging markets in the Fund's benchmark portfolio. However, an expansion of the country list will lead to increased focus on the new markets, and further work will be done on questions associated with settlement systems, market liquidity, future return and risk. It will be natural to review the question of the benchmark portfolio again in the light of the insight that will be achieved in due course regarding settlement risk, among other things, by investing in such markets.

The Regulation on the Management of the Government Petroleum Fund contains provisions regulating the distribution of securities by region. A logical solution may be to include investments in emerging equity markets under the appropriate regions. This would mean, for example, including investments in Mexico in the Americas region, while investments in Greece would be included in the European region. Moreover, an upper limit must be placed on the total share investments in emerging equity markets can constitute of the Petroleum Fund's equity portfolio. In Norges Bank's view, this share should not exceed 5 per cent as long as these markets are not included in the benchmark portfolio.

Jarle Bergo

Harald Bøhn