The management objective for the Government Pension Fund Global (GPFG) is the highest possible return after costs measured in international purchasing power, given an acceptable level of risk. Within this overall financial objective, the fund is to be a responsible investor. As the Ministry wrote in Report to the Storting No. 17 (2011-2012), the geographical distribution of the fund’s investments is to support this objective by contributing to the best possible trade-off between expected return and risk. The Ministry also wrote that the geographical distribution of the benchmark index for equities should help diversify risk in the fund in both the short and the long term, and that the fund’s long-term investment horizon indicates that the emphasis should be on long-term considerations.
The benchmark index for equities is based on, but differs somewhat from, a global float-adjusted market-weighted index from the index provider FTSE Russell. The biggest differences are due to the decision to assign equities adjustment factors according to their country of origin. Equities in European developed markets have been assigned a factor of 2.5, North American equities a factor of 1, and equities in other developed markets and in emerging markets a factor of 1.5. As a result, the fund has a much larger ownership share in companies in European developed markets than in North America. Since the sector distribution varies between countries and regions, these adjustment factors impact on the index’s sector composition. For example, the fund is invested more in financials and health care, and less in consumer services, technology and utilities, than a float-adjusted market-weighted index.
In its letter of 6 November 2018, the Ministry asks for Norges Bank’s assessment of the geographical distribution and composition of the benchmark index for equities for the GPFG. We have decided to respond to the Ministry in two separate letters. In this letter, we discuss developments, weighting principles and other factors with a bearing on the choice of geographical distribution. Drawing on this discussion, we then consider the need for changes to the adjustment factors. The Ministry has also asked the Bank to report on its experience of, and the framework for, equity investments in emerging markets. These topics are covered in a separate letter.
Developments relevant to the geographical distribution of the benchmark index
The Ministry has asked the Bank to assess developments with a bearing on the geographical distribution of the benchmark index.
We have looked at how the return and risk characteristics of equity investments have varied between the different regions in the benchmark index. In keeping with the letter from the Ministry, emerging equity markets have been considered separately. We have also analysed the extent to which differences in equity returns can be explained by country and sector affiliation, and the extent to which this has changed over time.
The benchmark index for equities has had a high weight of European developed markets ever since the fund began investing in equities in 1998. Historically, this was justified by the fund’s overall currency risk. Norway imports most from Europe, making it natural to assume that the fund’s purchasing power can be better protected against currency risk by investing more in European markets. In Report to the Storting No. 15 (2010-2011), the Ministry concluded that the fund’s currency risk appeared to be smaller than previously assumed, and there was no longer a basis for such a strong concentration of investments in Europe.
It was therefore decided in 2012 to reduce the weight of European equities in the benchmark index. At that time, the allocation to equities was set at 50 percent Europe, 35 percent North America and 15 percent Asia/Oceania. These fixed regional weights were then replaced with today’s adjustment factors. The changes decided on in 2012 meant that the regional distribution of the benchmark index for equities would no longer be fixed but vary with market developments.
There have been relatively large differences in regional returns since then, and so the current regional distribution is different to that in 2012. Stronger returns in North America than in European developed markets have meant that the weight of the former has risen from 35 to 41 percent, while the weight of the latter has fallen from 38 to 34 percent. The changes for other developed markets and emerging markets have been only minor.
In the enclosure, we show how return and risk characteristics have varied across the four regions and different time periods over the past 25 years. In most of these periods, returns have been higher in developed equity markets than in emerging equity markets. This is due primarily to the performance of North American equities.
Over the past 25 years, North American equities have produced an annualised return of 9.9 percent, against 8.3 percent for European developed markets, 5.0 percent for other developed markets and 6.7 percent for emerging markets. Over the same period, volatility as measured by annualised standard deviation has been 14.8 percent in North America, 17.3 percent in European developed markets, 17.1 percent in other developed markets and 23.3 percent in emerging equity markets.
We have previously shown that a global investor with a 70 percent allocation to equities can achieve a substantial reduction in risk in the long term by diversifying these investments across numerous countries. This applies even though short-term returns in different countries seem to be moving more closely together than in the past. In connection with this letter, we have performed calculations to identify any differences in the correlations between the four regions in the benchmark index.  We find that the correlations have varied over time but have been relatively similar in the period since the financial crisis. However, the returns in the different regions have not been perfectly correlated. Broad diversification of equity investments across these regions can therefore help reduce the expected volatility of equity returns.
Emerging equity markets have historically been associated with higher market risk than developed equity markets. Higher market risk can provide a basis for higher expected returns. Along with the diversification benefits of investing in emerging markets, this can help improve the trade-off between return and risk in the fund. However, the realised return over the past 25 years has been lower than could reasonably have been expected given the higher market risk in emerging equity markets.
Since 2011, the fund has also invested in government bonds from emerging markets. We have looked at the correlation between returns on bonds and equities in emerging markets and found that bond prices have a tendency to rise when equity prices in these markets rise, and fall when equity prices fall. The sign for the correlation between equities and bonds is the opposite to what we have found in developed markets. Bond investments in emerging markets thus increase the implied equity exposure of a portfolio of both equities and bonds. The Ministry decided in Report to the Storting No. 20 (2018-2019) to remove bonds from emerging markets from the benchmark index for bonds. Following this decision, the index’s implied exposure to emerging equity markets would be somewhat reduced.
Diversification benefits and higher expected returns are among the reasons why two of the funds with which the GPFG is often compared aim to have substantial investments in emerging markets. The Canada Pension Plan Investment Board (CPPIB) has set a target of having one-third of the fund invested in emerging markets by 2025. Singapore’s sovereign wealth fund, GIC, aims to have 15-20 percent of its capital invested in emerging equity markets. By way of comparison, only just over 10 percent of the GPFG was invested in emerging markets at the end of 2018.
Countries and sectors
Since the sector composition varies between countries, the decision to depart from market weights along the country dimension also impacts on the sector distribution. In the enclosure, we show that there have been substantial differences in the return and risk characteristics of different sectors over time. Equity returns in specific sectors may rise or fall substantially further than the broad equity market at times.
In connection with this letter, we have looked at the importance of country and sector affiliation in explaining equity returns. We find that both country and sector are important, but that the relative importance has varied over time and between regions. In emerging markets, country has been more important than sector throughout the period. In developed markets, the picture is not as clear-cut: country have historically been more important, but we find two periods where developments in individual sectors have been more important than country effects. These two periods coincide with episodes with major price movements: the dot.com crisis of the early 2000s and the financial crisis towards the end of the same decade. These episodes were both global events that originated largely in a specific sector. In more recent years, the relative importance of country and sector in developed markets has been more balanced.
Regional distribution with different weighting principles
The Ministry has asked the Bank to look at the regional distribution with different weighting principles. We have assessed the consequences of different weighting principles for the regional and sector distribution of the benchmark index, and how the return and risk characteristics vary between these different principles.
Full and float-adjusted market weights
An index based on full market weights reflects the capital that is available in the listed equity market. This capital is broadly distributed through equal percentage ownership of all the companies included in the index. However, not all shares included in the index will be readily available for purchase. The most widely used global stock indices allow for this by removing these shares from the index when a company’s market weight is calculated. This is known as free-float adjustment.
A float-adjusted index is easier for typical users of the index to track. A float-adjusted market-weighted index can be implemented at low cost and is a good starting point for transparent and cost-effective equity management. There will, however, be some costs for tracking such an index. Those tracking the index will need to trade whenever the composition of the index changes. For example, this will happen when markets and companies move in and out of the index, and when the index provider amends its estimate of the percentage of a company’s stock that is readily available.
A float-adjusted index will have a different company, sector, country and regional composition to one based on full market weights. Countries with a limited free float will have a lower weight in the index. This applies to most emerging markets as well as to some developed markets in Europe and Asia. Countries with a high free float, such as the US, the UK, Australia, Canada and Switzerland, will have a higher weight in a float-adjusted index. Sectors with high levels of government ownership and so limited free float, such as telecommunications and utilities, will have a lower weight in such an index. Sectors with a high free float, such as health care and technology, will have a higher weight. A float-adjusted index will also typically have lower weights of value stocks, small companies and infrequently traded stocks.
In the enclosure, we show how free-float adjustment impacts on return and risk characteristics. In the short and medium term, there has been a relatively big difference between the returns on a float-adjusted index and an index based on full market weights. The differences in return and risk characteristics in the short and medium term can largely be explained by North America and emerging markets having different weights in a float-adjusted index and a non-float-adjusted index. Within the four regions, differences in free float between countries seem to have had less of an effect on overall return and risk characteristics.
Alternatives to market weights
The Ministry has also asked the Bank to assess the regional distribution with fundamental weights such as GDP and listed companies’ capital and earnings, and weights based on the distribution of risk between the regions. We present the results in the enclosure. None of the indices constructed on the basis of these alternative weighting principles produces a regional distribution similar to that of the current benchmark index.
These alternative indices can be complex to calculate. They will be adjusted more frequently and will therefore be more expensive to track in the management of the equity portfolio than a market-weighted index. Unlike a market-weighted index, an index constructed on the basis of these alternative principles will not necessarily be based on transparent, verifiable criteria. These alternative indices will also be less investable for the fund and so less suitable as a long-term yardstick for the choices made in its management.
The discussion so far in this letter has been based on the situation of a typical global investor. There may be other factors, such as the fund’s characteristics, which should be taken into account when choosing a geographical distribution.
The fund’s investment horizon
Investors have different time horizons for their investments and different risk-bearing capacities. In Report to the Storting No. 20 (2018-2019), the Ministry writes that the GPFG’s long investment horizon makes it well placed to take on risks that require a long time horizon. A long-term investor does not necessarily need to worry about drops in value driven by factors considered to be temporary.
The relevant risk for long-term investors is the risk of permanent losses. These might arise as a result of changes in expected earnings. Should something happen that affects expected earnings for all companies in a market or sector with a high weight in the index, the permanent loss could be substantial. Prolonged economic crises, major natural disasters and wars are examples of events that can have such an effect, cf. the Ministry’s discussion in Report to the Storting No. 17 (2011-2012). Permanent losses can also be triggered by events that affect individual investors in a market or sector, such as the expropriation of assets or introduction of special levies.
It is hard to quantify the probability of such events occurring or the extent to which this risk is reflected in market prices. If account is to be taken of such considerations, one possible approach would be to put a ceiling on the weight a specific country and/or sector may have in the index. Where such a limit should be set is uncertain and could vary with the countries and sectors that have a high weight in the index. This may change over time. As with other deviations from market weights, this type of deviations would need to be evaluated against its purpose at regular intervals.
The fund’s limited liquidity needs
The probability of large unexpected withdrawals from the fund is small. This means that, in principle, the fund is well placed to accept the risk, and harvest the potential reward, of investing in less liquid assets.
Free-float adjustment is performed to ensure that an index can be tracked closely by investors with ongoing liquidity needs. The fund is not such an investor. Free-float adjustment means that the index does not reflect all listed capital, only the capital that the index provider believes is available to a typical investor at the time. This is reflected in the market value of the float-adjusted index being around 20 percent below that of an index based on full market weights.
Another issue is whether an index based on full market weights best reflects global capital, since a large share of economic activity takes place outside the listed market. There are no precise figures for the percentage of capital listed in different countries, but the ratio between the market value of the shares listed on an exchange in a country and economic output in that country (GDP) can provide an indication. In general, it appears that unlisted entities account for a larger share of output in emerging markets, while listed companies play a greater role in developed markets.
Emerging and frontier markets accounted for 55 percent of global GDP in 2018. Dimson, Marsh and Staunton (2019) discuss possible reasons for the discrepancy between emerging markets’ share of global output and their share of the global equity market. They conclude that this discrepancy can be explained in part by free-float adjustment and by index suppliers’ liquidity requirements for individual stocks. Without restrictions imposed by index providers, the authors estimate that the weighting of emerging markets in a global equity index would be more than 20 percent. By way of comparison, these markets accounted for 10 percent of the fund’s benchmark index for equities at the end of 2018.
Equity investments in emerging markets are less liquid than equity investments in developed markets. The fund may be well placed to invest in less liquid emerging equity markets and so achieve a more diversified total portfolio. In Report to the Storting No. 17 (2011-2012), the Ministry writes that the fund’s special characteristics – such as its long investment horizon and limited need to realise assets quickly – might be considered to give the fund an advantage in emerging markets.
The fund’s position in individual markets
The benchmark index consists of almost 8,000 companies. The fund’s average ownership share in these companies is around 1.5 percent. As a minority shareholder in these companies, we are dependent on good corporate governance, limited discrimination and the protection of the fund’s rights in law and legal systems. In the Bank’s letter of 2 February 2012, differences in governance practices were singled out as a possible argument for a higher weight in Europe than in other regions.
Based on indicators from the World Bank and MSCI, it appears that minority shareholders’ rights are somewhat better protected in Europe than in other developed regions, although there are major variations between European countries. To the extent that differences in governance practices from country to country impact on the return and risk characteristics of equities, it is reasonable to assume that general differences in these practices will be reflected in market prices.
The return on equity investments after costs will be affected by the fund’s tax position in individual markets. Tax costs for equity investments consist mainly of taxes on dividends. In some countries where the fund is invested, however, taxes have also been introduced on capital gains. Capital gains taxes are often less standardised than dividend taxes and may vary with the type of investor and with the size and duration of the investment. In most countries where the fund is invested, other than a few emerging and frontier markets, Norges Bank is exempted from taxes on capital gains in tax agreements between Norway and the countries in question.
The fund’s position in individual markets in areas such as tax and governance could, in principle, be taken into account in the choice of geographical distribution in the benchmark index if the fund’s position in selected markets is permanently significantly different to that of the marginal investor. This is not easy to ascertain. The tax rules in each market can vary with the type of investor and with the value and duration of an investment. In addition, both tax rules and governance practices evolve over time.
The Bank’s advice
The investment strategy for the GPFG means that the fund’s return and risk characteristics largely mirror those of the benchmark index. The benchmark index therefore plays an important role in its management. To serve as a long-term yardstick for the choices made, the equity index needs to be constructed on the basis of transparent, verifiable criteria and be investable for the fund.
A float-adjusted market-weighted index such as the FTSE Global All Cap meets these criteria and is investable for the GPFG. As the Ministry writes in Report to the Storting No. 17 (2011-2012), such an index is a natural starting point for the geographical distribution of the fund. Any departures from a float-adjusted market-weighted index should be justified and have a concrete purpose.
The geographical distribution of the fund’s benchmark index has been adjusted over time towards float-adjusted market weights, but still has a much higher weight of equities in European developed markets and a correspondingly lower weight of North American equities. The weights of equities in other developed markets and equities in emerging markets are approximately the same as float-adjusted market weights. The fund may have characteristics that support a geographical distribution that departs to some extent from float-adjusted market weights. We are of the opinion, however, that the geographical distribution should be adjusted further towards float-adjusted market weights by increasing the weight of equities in North America and reducing the weight of equities in European developed markets. The gap to market weights will then be smaller than today. We assume that the Bank will be given an opportunity to return on the issue of how this adjustment to a new geographical distribution should be implemented.
The Executive Board approved the Bank’s reply at its meeting of 14 August 2019, but one of its members, Kjetil Storesletten, had the following special remark:
“The starting point for the Executive Board’s advice is the fund in isolation. I agree with the Board’s assessment that this perspective indicates that the portfolio weights should be adjusted towards market weights. The Bank’s analyses show that, if the fund is considered in isolation, indices based on FTSE full market weights and FTSE float-adjusted market weights would have given a better trade-off between return and risk than the regional adjustment weights the fund uses today.
“The conclusion changes, however, when viewed from a broader perspective of wealth. The very reason why Norway set up the oil fund was a broad view of national wealth, and it was this perspective that was behind the Mork Committee’s advice on the equity share. This perspective indicates that higher weights should be assigned to countries and markets with a lower correlation with Norwegian government revenue and Norwegian economic output, and lower weights to countries and markets that move more closely in line with Norwegian income.”
Jon Nicolaisen Yngve Slyngstad
 See Tables 1 and 2 and Figure 1 in the enclosure, where the choices made in the design of the benchmark index result in a distribution between regions, countries and sectors that departs from both full and float-adjusted market weights.
 The discussion is based on the fund in isolation rather than Norway’s overall national wealth.
 The weight of European bonds in the benchmark index for bonds was reduced at the same time and for the same reasons.
 See Table 3 in the enclosure. The estimates in the table are based on return data measured in US dollars.
 See Figure 2 in the enclosure.
 Further information on the Bank’s long-term estimates of the return and risk characteristics of equity investments can be found in the Bank’s letter of 1 December 2016.
 See Table 3 and Figures 2 and 3 in the enclosure.
 See Dimson E., Marsh P. and Staunton M. (2019) “Credit Suisse Global Investment Returns Yearbook 2019”. A summary of
the report is publicly available at https://www.credit-suisse.com/about-us/en/reports-research/studies-publications.html.
 See Table 3 in the enclosure.
 The Ministry envisages that the Bank is still able to invest up to 5 percent of the bond portfolio in emerging markets.
 See the Canada Pension Plan Investment Board’s annual report for 2019.
 See Table 4 in the enclosure.
 See NBIM Discussion Note 1/2019, available at www.nbim.no.
 See also Melas D. (2019) “The Future of Emerging Markets”, a report from MSCI, for a discussion of the importance of country-specific factors in emerging equity markets.
 See Table 5 in the enclosure.
 Further information can be found in NBIM Discussion Note 5/2014, available at www.nbim.no.
 See Table 6 and Figure 4 in the enclosure.
 Measured over the entire period in which the broad global stock indices have been adjusted for free float, the difference in returns has been limited. Float-adjusted indices did not become common until the early 2000s.
 See Tables 7-10 in the enclosure.
 Table 11 and Figures 5 and 6 in the enclosure show the effects on return and risk characteristics. The benchmark index for equities is broadly diversified across countries and sectors, and only the largest countries and sectors would be affected by such a ceiling. In the period we have analysed, this applies primarily to two countries (Japan and the US) and two sectors (technology and financials). Little importance should therefore be attached to the empirical results.
 Measured as the difference in market capitalisation between the FTSE Global All Cap (which is float-adjusted) and a version of the same index that is not float-adjusted.
 See Figures 7 and 8 in the enclosure.
 See Dimson E., Marsh P. and Staunton M. (2019) “Credit Suisse Global Investment Returns Yearbook 2019”.
 Measured as the percentage of float-adjusted capital.
 See Figures 9 and 10 in the enclosure.
 One example of an emerging market where the fund currently pays tax on capital gains is India, which in April 2018 began applying this tax to shares held for more than a year. Previously the tax was applied only to shares held for less than a year. The rate of tax is lower for shares held for more than a year than for shares held for less than a year.
 See Tables 12-16 in the enclosure for a comparison of the current benchmark index and alternative indices.