Meld. St. 23 (2015–2016)

The Management of the Government Pension Fund in 2015

To table of content

Part 2
Thematic articles

7 Gross excess return as a management performance measure

7.1 Introduction

The Ministry of Finance has adopted mandates for the management of the Government Pension Fund Global (GPFG) and Government Pension Fund Norway (GPFN) by Norges Bank and Folketrygdfondet, respectively. The mandates express the funds’ investment strategies, including provisions on the composition of the respective benchmark indices and risk limits. The mandates provide that the managers shall seek to achieve the highest possible returns on the GPFG and GPFN – measured in international currency and Norwegian kroner, respectively – after asset management costs.

The mandates permit Norges Bank and Folketrygdfondet to deviate from the applicable benchmark indices subject to given risk limits. The limit on market risk as measured by expected tracking error is key. It allows the managers to exploit the funds’ distinctive characteristics and advantages to achieve excess returns compared to the benchmark indices, and offers an opportunity for cost-effective adjustment to the benchmark indices. Deviations from the benchmark indices may raise or lower the funds’ absolute risk relative to the benchmark indices.

The Ministry of Finance employs gross excess return, defined as the difference in the returns on the Fund and the benchmark index, as the management performance measure. The figure is based on the audited accounts and is independent of model assumptions and estimated parameters. The performance measure is a gross quantity that ignores costs. However, index management also incurs costs. The Ministry has previously referred to calculations prepared by Norge Bank and Folketrygdfondet which support the view that gross excess return may be a reasonable approximation of net value added.

7.2 The division of labour between the asset owner and the manager

The benchmark indices adopted by the Ministry of Finance for the GPFG and GPFN are based on leading, easily accessible and broad market indices for equities and bonds prepared by recognised index providers. The expected returns on the benchmark indices reflect the excess returns compared to risk-free investments achievable by harvesting risk premiums with a robust theoretical and empirical foundation, such as equity and credit premiums. Investments in line with broad market indices also ensure good risk diversification by reducing the significance of incidents in individual companies, sectors and markets.

Market indices are standardised products in which index providers decide on various choices and priorities relating to the representation of markets and securities. The requirement that included securities and markets must be liquid allows investors to buy securities in an index without incurring high transaction costs. In principle, indices can be followed closely at low cost. The benchmark indices for the GPFG and GPFN may therefore provide a suitable starting point for measuring the value added by the managers.

The choices and priorities of index providers are adapted to a theoretical average investor in the market. Accordingly, market indices do not necessarily express the optimal risk-return ratio for investors with distinctive characteristics that deviate significantly from the average, such as large size and a long investment horizon.

The scope for deviations from the benchmark indices stipulated in the mandates adopted by the Ministry of Finance is designed to afford the managers some leeway to make progressive adjustments to the composition of the investments in the two funds. Changes to investment composition to ensure cost-effective adaptation to the benchmark indices, greater risk diversification than in the benchmark indices, exploitation of factor strategies and appropriate security selection may, over time, improve the risk and return properties of the GPFG and GPFN. Such changes in investment composition require specialist expertise, market proximity and the ability to make time-critical decisions, and are therefore delegated to the two managers.

In accordance with this division of labour, the Ministry of Finance uses gross excess return to measure the results of Norges Bank and Folketrygdfondet’s management of the GPFG and GPFN, respectively. Gross excess return encompasses all deviations from the benchmark indices. This means that the results of all changes in the composition of the investments in the GPFG and the GPFN compared to the benchmark indices are attributed to the two managers. This is consistent with the management objective as expressed in the mandates from the Ministry.

7.3 The link between gross excess return and net value added

To assess the value added to the GPFG and GPFN by Norges Bank and Folketrygdfondet, respectively, the returns on the two funds must be compared to the returns, after costs, which could alternatively have been achieved. In principle, the returns on the benchmark indices cannot be achieved at zero cost.

The purchase and sale of securities in order to match investment to the benchmark indices – so-called index management – incur both management and transaction costs. Moreover, in the GPFG and GPFN, some transaction costs will be due to rebalancing of the equity portion to the strategic weighting stipulated by the Ministry of Finance. In the GPFG’s case, costs have historically also been incurred in conjunction with capital inflows to the Fund. Changes in asset allocation have also historically imposed transaction costs on the GPFG and GPFN. Transaction and asset management costs are excluded from the return calculation for the benchmark indices. In principle, therefore, index management will entail a negative excess return compared to the benchmark indices.

Net value added through active management is a measure of the capital value added after costs, and is defined as the difference between the returns on the funds less actual asset management costs and the returns which could have been achieved through index management. The returns on the two funds, the returns on the benchmark indices and actual asset management costs have been audited and verified. Historical index management costs, on the other hand, are based on uncertain estimates of costs linked to factors such as capital inflows, rebalancing, changes in portfolio composition and index adjustments. It is also necessary to estimate historical asset management costs related to index management.

Uncertainty about the size of these costs renders the precise amount of net value added uncertain.

As part of their management, Norges Bank and Folketrygdfondet lend securities to other investors and receive compensation for doing so. Securities lending is associated with risk, including counterparty risk. There is thus a need for reliable risk management systems and a robust legal framework. The Ministry of Finance has previously pointed out that, to some extent, it is reasonable to regard securities lending revenues as additional returns secured through active management; see the report on the Fund for 2009. Such revenues are included in the gross excess return, but are excluded from the return on the benchmark indices.

In connection with reviews of the management of the GPFG and GPFN discussed in the reports on the Fund for 2013 and 2014, respectively, both Folketrygdfondet and Norges Bank presented estimates of historical index management costs.1 These estimates indicate that comparisons of gross amounts (gross excess return) and net amounts (net value added) produce similar results. This is because the estimated historical negative excess return achieved through index management is approximately equal to the actual historical asset management costs. The calculations do not take excess returns due to securities lending revenues into account. Accordingly, the gross excess return has thus far been regarded as a reasonable estimate of net value added.

7.4 Risk adjustment

Managers can achieve excess returns compared to benchmark indices in various ways, including by assuming additional risk. In financial literature, models are commonly used to explain historical results. A distinction is made between returns a manager secures by assuming systematic risk and returns due to other deviations. The latter estimates the risk-adjusted excess return, and is termed “alpha”. In principle, the purpose of such risk adjustments is to provide greater insight, in hindsight, into factors which may explain the excess return. The models assume that the owner can accept higher systematic risk directly, by adjusting the benchmark index. However, no single model or single set of assumptions provides an unambiguous answer to the question of how, in hindsight, risk has impacted performance. Nor is there an unambiguous answer to the question of what adjustments the owner can make to the benchmark index.

Under the Capital Asset Pricing Model, the risk associated with investing in the market portfolio is the only risk compensated for by the expected excess return beyond risk-free investment of the capital. Risk adjustment often employs the manager’s benchmark index as an approximation of the market portfolio. Under this model, the manager’s return is adjusted for the return achieved by accepting more or less systematic risk than the benchmark index. For example, the manager can accept more systematic risk by concentrating investments in assets which increase fluctuations in the return on the benchmark index. In the Capital Asset Pricing Model, the owner of the capital can directly increase or reduce systematic risk by debt financing investments in the benchmark index or making risk-free investments, respectively. Positive alpha can be interpreted as meaning that the manager has achieved a return above the return implied by compensation for risk-taking.

A manager may also employ so-called factor strategies through which investments are systematically tilted towards assets with particular characteristics, such as low market value, low relative price and low liquidity. Harvesting risk premiums has historically improved the risk-return ratio for investors like the GPFG. However, such strategies may entail long periods of higher or lower returns than the benchmark index and have a sizable negative impact during periods of market turbulence.

Although there is no agreement in the financial literature regarding which factors are priced into the capital markets or what risks factor premiums may potentially compensate for, models which adjust for the harvesting of factor premiums are still frequently used. In such models, the manager’s return is adjusted for the return achieved by accepting more or less systematic risk than in the benchmark index and other (assumed) systematic risk expressed by a selected number of factor premiums. However, these factor premiums can reasonably be regarded more as explanatory variables than as compensation for accepting systematic risk (Eckbo and Ødegaard, 2015).

Different risk-adjustment models are based on different assumptions, and may produce different results. As Eckbo and Ødegaard (2015) have pointed out, adjustments may produce a positive, zero or negative alpha depending on which factor strategies are adjusted for and how these are measured.

Alpha is used to measure the manager’s skill in security selection, the exploitation of time variations in risk premiums and the harvesting of risk premiums excluded from the risk adjustment (Hsu, Kalesnik and Wermers, 2011). Alpha is therefore a conditional estimate of risk-adjusted excess return which depends on the risk model and data employed. Uncertainty about the choice of model, data and theoretical basis for different factor premiums suggests that care should be taken when interpreting estimates produced by such models. Using alpha from models which adjust for the harvesting of factor premiums as a management performance measure can also be problematic. As Dahlquist, Polk, Priestley and Ødegaard (2015) have pointed out, it can be argued that the results achieved by harvesting certain factor premiums must be attributed to the manager, and that no adjustment should be made for them.

An important point is that not all risk premiums can be harvested by adjusting the benchmark index. In the report on the Fund for 2010, the Ministry of Finance concluded that giving greater emphasis to factor strategies in the management of the GPFN was ineffectual. Among other things, it was pointed out that it is difficult to identify an unambiguous set of factors in the Norwegian market, and that factor premiums are insufficiently stable over time. Such factors mean that there is little purpose in seeking to adjust the benchmark index for the GPFN to harvest factor premiums.

The report on the Fund for 2012 contained comprehensive analyses of factor strategies for large global equity portfolios like that of the GPFG.2 It was stated that it is difficult to identify the most appropriate adjustments, and that such adjustments may change over time. For a large fund like the GPFG, some strategies will entail large transaction volumes and costs, and have low investability. The Ministry therefore emphasised the necessity of adapting the strategies to the Fund. It was concluded that the GPFG’s distinctive characteristics mean that it is, in principle, well positioned to harvest factor premiums, but that the strategies should not be incorporated into the benchmark index stipulated by the Ministry.

Models which seek to estimate risk-taking by managers are also a useful tool for investigating how risk may, in hindsight, have impacted results. The Ministry of Finance regularly reviews the management of the two funds. This includes thorough analysis and evaluation of risk-taking by the managers. The risk-taking analyses in the most recent review of the GPFN’s management do not indicate that Folketrygdfondet is taking on more systematic risk by concentrating investments in assets which display a higher degree of covariation with the benchmark index.3 In the GPFG’s case, the analysis results are not clear-cut.4 The analyses also show that harvesting factor premiums appears to account for a considerable proportion of the excess return, particularly in the GPFG. Section 3–5(2) of the mandate for the GPFG provides that “[t]he equity and bond portfolio should be composed in such a way that the expected relative return is exposed to several different systematic risk factors”.

7.5 Summary

The Ministry of Finance uses gross excess return to measure the results achieved by Norges Bank and Folketrygdfondet in their management of the GPFG and GPFN, respectively. The performance measure is based on the division of labour between the Ministry and the two managers, and on the audited accounts. Calculations indicate that the gross excess return is also a reasonable measure of net value added through active management. This is because the actual historical asset management costs are approximately equal to the estimated historical negative excess return achieved through index management. The calculations exclude the excess return due to revenue from securities lending.

The gross excess return is not dependent on model assumptions or estimated parameters. The Ministry of Finance emphasises that performance measures should be as unambiguous as possible, and simple to understand and communicate. Moreover, the gross excess return should be supplemented by analyses that adjust for any returns achieved by the manager through systematic risk-taking, and analyses illustrating the harvesting of any factor premiums.

On 1 February 2016, the Ministry of Finance adopted new requirements for more detailed reporting on performance, risks and costs in the management of the GPFG; see section 3.4. Among other things, reports must now be made on the risk-adjusted return and the composition of the gross excess return.

Although gross excess return is used as the primary measure of the results achieved by the two managers, several supplementary measures and analyses are also employed.

The managers give an account of risk-taking in their reports on Fund performance. The Ministry of Finance presents supplementary measures in its annual report to the Storting on the Government Pension Fund, and conducts regular reviews of the management of the two funds. These reviews include evaluation of risk-taking.

7.6 References

Ang A., Brandt M., Denison D. (2014). Review of the active management of the Norwegian Government Pension Fund Global. www.regjeringen.no/spf.

Dahlquist, M., Polk, C., Priestley, R., Ødegaard, B. A. (2015). Norges Bank’s expert group on principles for risk adjustment of performance figures – final report. www.norges-bank.no

Eckbo, E., Ødegaard B. A. (2015). Metoder for evaluering av aktiv fondsforvaltning. Praktisk økonomi & finans, 31, 343–364. (In Norwegian only. English translation of title: Methods for evaluating active fund management.)

Hsu J., Kalesnik V., Wermers, R. (2011). Performance evaluation of active managers: An overview of current practice. Investments and Wealth Monitor, January/February, 37–40.

MSCI (2013). Harvesting risk premia for large scale portfolios: Analysis of risk premia indices for the Ministry of Finance, Norway. www.regjeringen.no/spf.

8 The GPFG’s benchmark index for equities and bonds

8.1 Background

The Ministry of Finance expresses the investment strategy for the GPFG through a benchmark index, risk limits and other provisions in the mandate issued to Norges Bank. The benchmark index plays an important role in the management of the GPFG, as it specifies the desired capital allocation across asset classes, geographical regions and currencies. The index incorporates separate benchmark indices for equities and fixed-income securities, and the real estate portfolio.

The long-term strategy for the GPFG specifies a fixed allocation to equities (60 percent), up to five percent in real estate and the remainder in fixed-income securities. See Figure 8.1. Distribution accross asset classes is the primary determinant of the Fund’s total risk.

Figure 8.1 Benchmark index for the GPFG

Figure 8.1 Benchmark index for the GPFG

Source Ministry of Finance.

The overarching principles underpinning the design of the benchmark indices for the equity and fixed-income portfolios are discussed below. Detailed rules for the fixed-income index and equity index are set out in section 3–2 and section 3–3, respectively, of the mandate for the management of the GPFG.

8.2 Broad market indices

The investment strategy for the GPFG is based on the assumption that total risk in the Fund can be reduced by spreading the investments across a large number of securities, i.e. through diversification. The benchmark indices chosen for the GPFG’s equity and fixed-income investments are based on broad global indices and generally reflect the investment opportunities available in global equity and bond markets. The benchmark indices provide a clear, detailed description of how the Fund should in principle be invested in different countries, sectors, currencies, individual companies and bonds; see Figure 8.2.

Figure 8.2 Distribution of the equity and fixed-income benchmark indices across geographical regions and sectors as of 31 December 2015

Figure 8.2 Distribution of the equity and fixed-income benchmark indices across geographical regions and sectors as of 31 December 2015

Source Norges Bank and Ministry of Finance.

The broadness of a market index also depends on what inclusion criteria the index provider applies to securities. Among other things, providers decide which markets and securities should be included in a given index. Indices are generally constructed with the objective of broad risk diversification. A further aim is that indices should be investable. Highly liquid securities are therefore accorded priority in the composition of such indices, to ensure that broad groups of investors can invest in the selected securities at a low transaction cost.

To some extent, different index providers apply dissimilar criteria when determining which securities are to be included in an index. As a result, different equity indices for a given region will not necessarily contain the same equities, and the selected equities may not have the same relative weightings. Moreover, various indices may incorporate different countries and markets. There may be considerable composition differences between global indices prepared by different index providers, even though the indices all aim to reflect developments in global stock markets. The Ministry of Finance would emphasise that the Fund’s benchmark index is based on index products developed by leading, recognised index providers.

Basing the Fund’s benchmark index on standardised products largely leaves decisions regarding market and company representation to the chosen index provider. The management mandate nevertheless includes a provision stating that Norges Bank must pre-approve all markets before funds may be invested. This approval requirement applies regardless of whether a market is included in the benchmark index stipulated by the Ministry of Finance. Accordingly, the manager is obliged to conduct an independent assessment to establish which markets provide the Fund with satisfactory security.

8.3 The equity index

The benchmark index for the GPFG’s equity investments is based on the FTSE Global All Cap index, and includes all countries classified by the index provider as developed or emerging markets, with the exception of Norway. At the end of 2015, the index incorporated 24 developed markets and 21 emerging markets.5

The index provider classifies stock markets as developed, emerging or frontier by reference to criteria such as data quality, currency restrictions, per capita GDP, the number of limited companies, securities pricing quality, the country’s credit rating and restrictions on foreign ownership. The efficiency of settlement systems, market liquidity and maturity, total market capitalisation and scope for exercising ownership rights also affect classification.6 The inclusion of countries in the equity index is determined by the index provider’s assessments. The country composition of the index will vary over time as individual markets are included or excluded from the index.

An index in which each individual company is given a weighting corresponding to the market value of the shares in the company is referred to as a market-weighted index. A market-weighted index reflects the investment opportunities available to a typical investor, and can be regarded as the portfolio held by the average global investor. Developments in a market-weighted index reflect the value performance of the stock market as a whole, as represented by the index. Basing investment composition on market-weighted indices entails tracking the market’s pricing of the included equities. The geographical distribution of such an index is determined by the included companies’ market value and where they are listed.

The FTSE Global All Cap index is based on global market weights, but is adjusted for so-called free float. Free-float adjustment involves altering the index weightings for individual companies to reflect the ownership shares of large long-term owners and cross-ownership. This adjustment reduces index weights of companies with many long-term owners. The argument for such adjustment is that these ownership shares are not freely tradable. The adjusted weights provide a better measure of the investment opportunities available to international financial investors, and promote lower transaction costs.

Adjusting index weightings to take free float into account also impacts the geographical distribution of such indices. In emerging markets, a material proportion of shares in listed companies are not freely tradable. These markets are therefore given a lower weighting in a market-weighted index adjusted for free float than in an index based on full market capitalisation.

A geographical distribution in line with market weightings provides a natural starting point for the composition of the Fund’s equity index. Market value weighting also implies that countries in which the stock market is largely listed and market value is high are accorded a high weighting in the index. The proportion of the capital market which is listed varies between different countries, regions and market types. In the Ministry’s view, pure market weighting (free-float adjusted) will entail an undesirably high concentration of investments in the US stock market; see the report on the Fund for 2011. The benchmark index for the equity investments is supplemented by country and market adjustment factors to ensure a more balanced geographical distribution.

8.4 The fixed-income index

The purpose of the fixed-income investments is to reduce fluctuations in the total return on the Fund, provide liquidity and exposure to risk factors such as interest and credit risk. The composition of the GPFG’s fixed-income benchmark index is based on assessments of the risk and return properties of different parts of the bond market, in line with the purpose of the fixed-income investments.

While the Fund’s equity benchmark index broadly represents the global listed stock market, the fixed-income index covers a narrower range of investment opportunities in debt instruments. The design of the Ministry’s fixed-income index balances the desire for a simple, transparent and verifiable index with the objectives of broad risk diversification and representation of investment opportunities in the global bond market. Some sub-segments included in broad market indices have been excluded based on, inter alia, evaluations of market structure, concentration risk and whether the sub-markets are suited for passive management; see the report on the Fund for 2011.

The fixed-income benchmark index is based on the currencies and individual securities included in selected sub-indices provided by Barclays. To be included in Barclays’ broad index of fixed-income securities in local currencies, securities must have a high credit rating (investment grade). The securities and the local currency markets must also be sufficiently liquid and investable. Further, there must be a liquid currency market in which international investors can hedge against future fluctuations in the value of the local currency.

70 percent of the fixed-income benchmark index comprises government bonds, inflation-linked bonds and bonds issued by international organisations (the government sub-index). The remaining 30 percent comprises corporate bonds (company sub-index). The allocation between the two sub-indices is fixed, and there is full monthly rebalancing to the specified proportions.

The role of the government sub-index is particularly to reduce fluctuations in the Fund’s aggregate return. The composition of the government bond index is based on the currencies included in Barclays’ broad index7 of nominal government bonds issued in local currencies, with the exception of Norwegian kroner. If the index provider alters the selection of currencies in the underlying indices, the benchmark index for the GPFG’s fixed-income investments will be changed correspondingly. The government sub-index presently comprises nominal government bonds issued in 23 currencies, including 11 emerging-market currencies. Other securities included in the government sub-index are inflation-linked bonds issued in currencies of developed economies and bonds issued by international organisations like the World Bank.8

Market weighting can be a less suitable starting point for investment in government bonds than for equities and corporate bonds. The overall supply of government bonds is significantly impacted by the borrowing needs of individual sovereign states. Market weighting implies high and increasing exposure to countries with high and rising debt levels, and does not necessarily ensure good risk diversification.

The weighting of the GPFG’s sub-index for government bonds is based on the size of each country’s economy as measured by gross domestic product (GDP). GDP weighting entails lower index exposure to countries with high national debt relative to the size of the economy than with a market-weighted index, and vice versa. The distribution across different regions and individual countries depends on developments in GDP and on which markets are included or excluded from the index. In each country, sub-segments and individual bonds are weighted according to market weights. Market weighting ensures equal ownership shares in all bonds in the same country.

Some countries have high GDP compared to the size of the government-bond market. In the interests of investability, individual country weights in the government sub-index are therefore supplemented by adjustment factors. The size of the Fund makes the investability requirement particularly important.

The management mandate also includes a requirement that Norges Bank must take differences in national fiscal strength into account when composing the actual portfolio. This requirement is intended to emphasise that one of the objectives for the Fund’s investments in government bonds is to reduce fluctuations in the Fund’s aggregate return over time.

The benchmark index for the company sub-index contains covered bonds and corporate bonds.9 Expected return on corporate bonds is assumed to exceed that on government bonds, partly due to the compensation expected for accepting the credit risk associated with such bonds, i.e. the credit premium. The composition of the company sub-index in the fixed-income index is based on market weights. The index incorporates seven approved developed-market currencies in Europe and North America.10

Footnotes

1.

The analyses are documented in a letter from Norges Bank to the Ministry of Finance of 12 March 2014 and a corresponding letter from Folketrygdfondet of 10 March 2015. The letters are available on the Ministry’s website. Norges Bank has updated the analyses in its annual report for 2015.

2.

The analyses were prepared by the index provider and consultancy firm MSCI (2013).

3.

See the enclosure to the letter of 10 March 2015 from Folketrygdfondet to the Ministry of Finance. The letter is available on the Ministry’s website.

4.

See the report of the expert group comprising Ang, Brandt and Denison (2014), the enclosure to the letter of 12 March 2014 from Norges Bank to the Ministry of Finance, which is available on the Ministry’s website, and Norges Bank’s annual report for 2015.

5.

FTSE classifies Belgium and Luxembourg as a single stock market. The number of individual countries is thus 46.

6.

FTSE Russell Country Classification Process (September 2015).

7.

Securities are selected in accordance with the Barclay Global Treasury GDP Weighted by Country Index.

8.

Securities are selected in accordance with the Barclays Global Inflation Linked Index and the “Supranational” sub-segment in the Barclays Global Aggregate Index. Bonds issued by international organisations are allocated to countries in accordance with the currency denomination of each security.

9.

Securities selection is based on the sub-segments “Covered Bonds” and “Corporates” in the Barclays Global Aggregate.

10.

USD, CAD, EUR, GBP, SEK, DKK, CHF.

Go to front page