Norges Bank

Submission

Emerging equity markets and the Petroleum Fund's benchmark portfolio

Norges Bank submitted the following letter to the Ministry of Finance on 30 August 2000

1. Introduction

In a submission of 16 March 1999, Norges Bank discussed the requirements that should be met before new countries are included in the Fund's investment universe. The Bank's considerations were based on the purpose of the Fund, which is to maximise future international purchasing power, given an acceptable risk level.

An evaluation of specific countries was sent to the Ministry of Finance in a submission dated 26 August 1999, and was based mainly on the general criteria described in Norges Bank's first submission on emerging markets. Following an evaluation of the various countries on the basis of general criteria, a core of emerging markets that fulfilled the minimum requirements remained. These were Brazil, Mexico, Greece, Turkey, Taiwan, Thailand and South Korea. These countries were included on the Petroleum Fund's country list on 1 January 2000.

In the National Budget 2000, it was stressed that it was not regarded as appropriate to include the new countries in the Fund's benchmark portfolio from the date that the investment universe was expanded. The Ministry of Finance would need more time to consider this matter, among other things in order to identify indices appropriate for inclusion in the benchmark portfolio. The Ministry aimed to complete the work by the end of 2000.

In this submission, Norges Bank considers whether the seven emerging markets should be included in the Petroleum Fund's benchmark portfolio. Before a new country is included in the benchmark portfolio, its clearing and settlement systems must satisfy certain minimum requirements. In addition legislation and supervisory systems must function satisfactorily, and the markets must be liquid. Chapter 2 commences with some comments regarding the clearing and settlement systems of emerging economies. The same chapter provides an account of the historical performance and market risk associated with the individual markets. Chapter 3 stresses that the most important considerations concern whether the trade-off between expected return and risk can be improved by including new countries in the benchmark portfolio. Chapter 4 provides a conclusion.

2. Characteristics of investments in emerging economies

Norges Bank's submission of 26 August 1999 presented an evaluation of settlement risk, supervisory systems and legislation in the seven emerging markets. It was stated that it would be natural to reconsider the question of the benchmark portfolio in the light of the insight that would be acquired by the Bank on, among other things, the settlement risk associated with investment in emerging equity markets.Over the past few months the Bank has continued its evaluation of aspects of operational management in these markets. In addition to its own investigations, the Bank has made a thorough review of written sources. The conclusions are consistent with the Bank's assessments of August 1999. Supervisory bodies and a legal framework for the securities markets have been established in all these markets. The quality of the legislation and supervisory authorities varies, and there is a lack of consistency between the formal apparatus and real follow-up in some cases. However, the situation is such that, as far as requirements regarding supervisory systems and legislation are concerned, it will be advisable to invest in these countries. Systems for registration and safekeeping of securities are of generally satisfactory quality in all the countries. The quality of the clearing and settlement systems varies. Simultaneous payment and delivery is achieved in only a few of the markets, but on the whole risk exposure is only of a relatively brief duration. As a general rule, Norges Bank's counterparties must fulfil the credit rating requirements stipulated in the regulations governing investment management in Norges Bank. Settlement risk is higher, and supervisory systems and the legal framework are of a lower standard in emerging than in developed economies. In the view of Norges Bank, however, the seven economies satisfy the requirements it is reasonable to make of emerging economies.

Another important factor that distinguishes emerging from developed economies is the returns and market risk associated with investments in these countries. Table 1 shows movements in the equity indices of the seven emerging markets and in selected developed markets in the 1990s. The indices for most of the markets start at 100 in December 1988, and all indices end in June 2000. However, time series for Turkey, Taiwan and South Korea are somewhat shorter, and their commencement dates are given in the table. All time series for returns are converted into USD to make it simpler to compare the different countries.

The US stock market has shown very favourable developments during this period, and therefore provides a useful basis for comparison. The indices for Mexico and Greece have risen more than the US index. The Brazilian index has also risen substantially, but not as much as the US index. Developments have been poorest in the Asian countries, particularly South Korea and Thailand. It is also worth noting that developments in the Japanese market have been negative during this period.

The table also shows the arithmetic mean of monthly returns, the standard deviation of returns and the confidence interval for average returns.

The variation in returns from month to month is generally large, and largest in Brazil and Turkey, while the markets in Taiwan and Mexico appear to be least risky. Since the Petroleum Fund has a long investment horizon, it is also interesting to look at the uncertainty associated with the estimate for the average return in the period being studied. If, as a simplification, it is assumed that the return is normally distributed, a confidence interval can be calculated for the average return. The standard deviation of the average return in the Brazilian stock market is about 1.7 per cent in the period under consideration. Confidence interval around the average return thus extends from 0.05 to 6.8 per cent. Confidence intervals for the other markets are smaller, but still larger than those for the developed markets. Thus there is great uncertainty associated with using historical averages as estimates of expected return. This applies to all emerging markets, and to a lesser, but still substantial, extent to developed markets.

Table 1: Index, monthly return, standard deviation of monthly return

and 95% confidence interval for average return;

December 1988 - June 2000

Index June

Arithmetic average

Standard deviation

Confidence interval for average return

2000

Brazil

709.81

3.43 %

19.88 %

0.05 %

6.82 %

Mexico

993.40

2.23 %

10.28 %

0.48 %

3.98 %

Greece

1209.22

2.47 %

12.16 %

0.40 %

4.54 %

Turkey

550.29

2.98 %

19.31 %

-0.41 %

6.37 %

Taiwan

213.23

1.17 %

10.44 %

-0.80 %

3.13 %

Thailand

74.76

0.57 %

12.56 %

-1.57 %

2.71 %

Korea

105.67

0.96 %

14.15 %

-1.85 %

3.78 %

Equally weighted

685.65

1.74 %

8.06 %

0.37 %

3.12 %

Market-weighted

735.23

1.80 %

8.31 %

0.38 %

3.21 %

US

734.07

1.53 %

3.93 %

0.86 %

2.20 %

Japan

89.94

0.18 %

7.30 %

-1.06 %

1.43 %

Hong Kong

607.62

1.70 %

8.83 %

0.19 %

3.20 %

Europe

484.92

1.24 %

4.35 %

0.50 %

1.98 %

"The Petroleum Fund"

487.95

1.23 %

3.87 %

0.57 %

1.89 %

Source: Datastream The time series for Turkey starts on 31 August 1989.

The starting dates for Taiwan and Thailand are 31 January 1991 and 31 January 1992 respectively.

It should be stressed that the time series with returns data for the emerging equity markets are relatively short. The indices that non-residents can replicate only start in 1988. In a longer historical perspective, it may turn out that the developments of the last 12 years are not representative. The literature shows that in the previous century a number of the emerging stock markets were closed for long periods, and that if these periods were included in calculations, returns would be smaller and risk higher than the figures yielded by calculations based on the past 12 years.

3. Improving the trade-off

In an assessment of whether the new countries should be included in the benchmark portfolio, one important aspect will be whether the new benchmark portfolio has a lower risk and/or higher expected return than the existing benchmark. The risk can be reduced if the covariation between returns in the emerging stock markets and the return on the existing benchmark portfolio is low. The expected return may increase if future returns in emerging markets are higher than those in developed markets. In Norges Bank's submission of 26 August 1999 to the Ministry of Finance, the Bank stated that it would do further work on questions relating to future returns and risk and the composition of the benchmark portfolio.

In practice, it is difficult to arrive at a benchmark portfolio that can be claimed with certainty to be better than other portfolios. This is partly due to the fact that in order to calculate the composition that such a benchmark portfolio should have, the expected return on the various assets class, the variation in return and the covariation of the returns on different assets must be estimated. All these parameters have to be calculated, and the estimates will be uncertain. In addition, the parameters change over time, so that what applied in one period may not necessarily apply in later periods.

The objective of the management of the Petroleum Fund is to maximise returns in the long term. This means that it is long-term expected return, risk and covariation that are relevant. It is more difficult to estimate the parameters when the investment horizon is long than when it is short, partly because there are a limited number of observations for ‘long' rates of return.

Table 1 presents the results of calculations of index development and returns for two portfolios of emerging markets. In the one portfolio, each market counts equally (one seventh - equal-weighted) while the other portfolio index is calculated by weighting the markets within each region with their respective market capitalisation weights (market-weighted). The regional distribution is the same as that for the Petroleum Fund. The index for stock markets in Europe shows the increase in value in the developed markets in this region, while the index for the ‘Petroleum Fund' shows approximately how the equity benchmark for the Petroleum Fund would have developed.

Both the equal-weighted and market-weighted indices for emerging markets rose substantially in the 1990s (cf Chart 1 and Table 1). If USD 100 had been invested in the equal-weighted portfolio, it would have had a value of USD 685 by the end of the period. The value of the market-weighted portfolio rose to USD 735, which is also what an investor in the US market would have ended up with. The index for "the Petroleum Fund" increased to almost USD 500.

The average monthly return on the emerging markets equity portfolios was 1.7-1.8 per cent per month, whereas the return on a portfolio with the same composition as the Petroleum Fund equity benchmark would have been about 1.2 per cent. The risk of investing in emerging markets would also have been higher, however. The standard deviation of the return on the portfolio of emerging markets was just over 8 per cent, while that for the ‘Petroleum Fund' was 3.9 per cent.

As commented under point 2, it is possible to calculate the uncertainty interval for the average return in the long term from historical data. The confidence interval for the emerging markets portfolio is considerably narrower than that for emerging markets considered individually. Expected return for the equal-weighted portfolio varies from 0.4 to 3.1 per cent. This interval is still so broad that there is great uncertainty associated with estimating future returns on the basis of historical data.

Calculations for the Petroleum Fund based on data from the period 1989 to June 2000 show that the return on the Petroleum Fund's equity portfolio would have increased, whereas the risk would have been slightly lower, if a smaller portion had been invested in emerging equity markets (cf. Table 2). This is because the covariation between the return on emerging and developed markets has been relatively low. The correlation coefficient in the 1990s was about 0.4.

Table 2: Combinations of the Petroleum Fund and investments in emerging markets

Share in emerging markets

7 countries - equal-weighted

7 countries - market-weighted

Monthly return

Standard deviation

Monthly return

Standard deviation

0

1.23 %

3.87 %

1.23 %

3.87 %

0.1

1.28 %

3.75 %

1.29 %

3.74 %

0.2

1.33 %

3.80 %

1.34 %

3.81 %

0.3

1.38 %

4.01 %

1.40 %

4.04 %

0.4

1.43 %

4.37 %

1.46 %

4.43 %

0.5

1.49 %

4.84 %

1.51 %

4.94 %

0.6

1.54 %

5.40 %

1.57 %

5.52 %

0.7

1.59 %

6.01 %

1.63 %

6.17 %

0.8

1.64 %

6.66 %

1.68 %

6.85 %

0.9

1.69 %

7.35 %

1.74 %

7.57 %

1

1.74 %

8.06 %

1.80 %

8.31 %

The results in Table 2 may suggest that emerging markets should be included in the Petroleum Fund's benchmark portfolio. However, the great uncertainty associated with estimates of return and risk means that decisions as to what share emerging markets should constitute of the benchmark portfolio cannot be based on this type of analysis. The following paragraphs contain a discussion of some factors that may have a bearing on this type of evaluation.

Returns:

Because historical data are fairly uncertain, it may be useful to attempt to use other methods to estimate future returns. In theory, factors such the risk associated with the investment and the investors' risk tolerance determine expected return. The higher risk in emerging markets may mean that the expected return on investments in emerging markets is higher than that on investments in developed markets.

Another alternative to historical data is to assume that the total value of the stock market is equal to the present value of the dividends that companies will pay their shareholders in the future. The present value of the stock market is known, and expected GDP growth can be taken as an estimate of future growth in the companies' earnings and dividends. The relationship between present price and expected future dividends can then be solved for expected return. Given reasonable estimates of expected GDP growth in developed and emerging economies, and assuming that current prices reflect the underlying fundamental values in the equity market, it can be argued that expected future return are somewhat larger for emerging than for developed markets.

A number of arguments have been put forward in the debate on economic growth in emerging as compared with developed economies. In the long term, ie a period of 10 years or more, average growth will depend mainly on the supply of production factors, and how efficiently they are utilised. In the period since 1960, economic growth in emerging economies has been greater, on average, than that in developed economies. This is primarily because emerging economies have accumulated capital and increased their labour forces at a far faster pace than industrial countries, and may well have faster economic growth in the future too. There are several reasons for this. Their growth starts from a far lower welfare level, and the return on both investments in real capital and education may therefore be greater. There are also indications that investments will grow faster in emerging than in developed economies in the years ahead. There is considerable potential for educating the population of emerging economies. Another important factor that may lead to higher economic growth in emerging economies is that both the population in general and numbers employed grow fast in these countries. There are also substantial opportunities for improving the efficiency of the economy through structural reforms in both the public and the private sector.

Risk:

One important consideration when composing the benchmark portfolio is to spread risk over a variety of countries and instruments. Our calculations are based on historical return rates for a limited time period. They may not adequately reflect the differences between emerging and developed economies. Emerging economies have an economic structure which is very different from that of developed economies, and the former therefore face different challenges from developed countries. In both the short and the long term, emerging economies will be exposed to different types of shock from those experienced by developed economies.

The risk associated with a portfolio of investments in developed and emerging markets depends heavily on the covariation, or correlation, between returns in these two market types. Our calculations show that covariation varies considerably over time, and there are indications that there has been a slight increase in covariation. This may indicate that diversification gains may be smaller in the future. Chart 2 shows developments in the correlation coefficient between the return on the ‘Petroleum Fund's' equity portfolio and the equal-weighted portfolio of emerging markets. In periods with high volatility, it looks as though correlation may also be high. The economic significance of this is that diversification gains are smallest when they are most needed.

Chart 2: Variation of volatility and correlation with time:

‘Petroleum Fund' and 7 emerging equity markets

Emerging markets in the benchmark portfolio:

Within each region, the current country distribution in the Petroleum Fund's benchmark portfolio is based on market capitalisation weights. This means that most is invested in large markets. With one exception, which is commented upon below, Norges Bank recommends that the emerging markets be included in the benchmark portfolio within the respective regions according to market weights. At end-May 2000, the market value of the 7 emerging markets accounted for 2.35 per cent of the aggregate market value of the 21 developed and 7 emerging markets in the Petroleum Fund's investment universe. This is a smaller percentage than appears optimal from the portfolio analysis. Several factors form the basis for this evaluation. The use of market weights is consistent with the current system, and there are no obvious grounds for departing from this principle specifically for emerging markets. Emerging equity markets also constitute a new asset class, and general precautionary principles tend to suggest that investments should be limited. Calculations of optimal portfolios are moreover based on uncertain historical data. The figures also appear to indicate that correlation coefficients are increasing, which, in the context of a portfolio analysis, will contribute to a reduction in the proportion invested in emerging markets.

Table 3 shows the number of companies in the FTSE country indices and the market value of the indices at the end of May 2000. The third column shows how large a share of the equity portion of the Petroleum Fund's benchmark portfolio the individual market will make up. The calculations are based on FTSE data.

Table 3: Number of companies and total market value in indices

 

Number of companies

Market value in millions of USD

Share of the Fund's equity benchmark

Brazil

Mexico

Greece

Turkey

South Korea

Taiwan

Thailand

Total

34

27

59

20

25

35

30

230

88687

89815

78681

18837

92819

132199

8321

509359

0.21 %

0.21 %

0.57 %

0.14 %

0.49 %

0.70 %

0.04 %

2.35 %

From the table it can be seen that Thailand will account for a very small portion of the Petroleum Fund's equity portfolio, while at the same time the index for the stock market in Thailand consists of a relatively large number of companies. In consequences, investments in the individual companies in the country index for Thailand will be substantially smaller than the Fund's average investments in companies from other countries. Equity investments in Thailand will therefore be more costly than investments in other countries. For operational reasons it therefore does not seem appropriate to include Thailand in the benchmark portfolio. Nor would doing so have any effect on the qualitative or quantitative assessments made in this submission. Norges Bank therefore proposes that Thailand should not be included in the Petroleum Fund's benchmark portfolio.

The FTSE country indices are used for the 21 developed markets currently in the benchmark portfolio. FTSE also produces indices for all the seven emerging markets. Before a company is included in one of the FTSE indices for emerging equity markets, three criteria must be satisfied. The first criterion concerns company size. The second is intended to ensure that a sufficient share of a company's equities are available to foreign investors, while the third is a general liquidity criterion. Because the existing Petroleum Fund benchmark portfolio is based on FTSE indices, it will be consistent and natural from an operational point of view to use the same indices for emerging equity markets. Using the same indices ensures both transparency and that roughly the same principles form the basis for the construction of the various country indices.

4. Conclusion

Following an overall evaluation, Norges Bank proposes including the emerging markets, with the exception of Thailand, in the Petroleum Fund benchmark portfolio.

Settlement risk is higher, and supervisory systems and the legal framework have a somewhat lower standard in emerging than in developed markets. The increased settlement risk incurred by the Petroleum Fund is nevertheless regarded as relatively limited.

It may be argued that in the long term returns in emerging markets could be higher than those in developed markets. It is furthermore assumed that the total risk to the Fund will not be altered significantly by including emerging markets in the benchmark portfolio. It is stressed that the risk of investing in emerging markets is greater in the short term than in the long term, and that there may be large variations in returns in the short term.

It is recommended that six emerging markets be included in the benchmark portfolio under their respective regions. Brazil and Mexico will thus be included in the Americas region according to their market capitalisation weights, while Greece and Turkey will be included in region Europe, and Taiwan and South Korea will be included in the Asia and Oceania region. It is furthermore recommended retaining the upper limit of 5 per cent on investments in emerging markets that is laid down in the Petroleum Fund Regulation.

On 1 January 2001 Greece will become a member of the EMU. Some index suppliers have already classified Greece as a developed market, while others, including the MSCI, have given notice of their intention to do so. We therefore propose that Greece be defined as a developed market in the next revision of the Petroleum Fund Regulation.

Yours sincerely

Svein Gjedrem

Harald Bøhn

Published 30 August 2000 18:58