Part 2
Topic articles

5 The financial crisis and its impact on the benchmark ind ex of the GPFG

5.1 Introduction

The investment strategy for the GPFG is based on assessments of expected long-term return and risk. The strategy is expressed in a benchmark index consisting of 60 per cent equity and 40 per cent bonds (fixed-income securities). At times, the financial markets will be subject to wide fluctuations, and the returns achieved can deviate significantly from the long-term expectations. Knowledge about turbulent periods is important for the work on the Fund, because it gives insight into the diversification properties of the portfolio and its sensitivity to developments in the economy and other risk factors.

The financial crisis sparked the biggest fall in the stock market in the Fund’s history. The global stock market fell 54 per cent from October 2007 to March 2009, and the benchmark for the GPFG fell 24.9 per cent in the same period. Below we analyse the developments in the benchmark index during the financial crisis. The analysis of the benchmark for equities focuses on developments in the different sectors which the stock market can be divided in to. This approach provides good insight into several of the key characteristics of the financial crisis.

5.2 Developments in the stock market

Figure 5.1 shows the developments in the stock market since the end of 1997. The problems in Asia and Russia in 1998 triggered a brief, less dramatic fall of 14 per cent. The areas that were affected by the crisis constituted a relatively small portion of total market value of the stock market, thus limiting the impact. In the period 2000–2002, the stock market fell 45 per cent. This period is often referred to as the «IT bubble» as the fall followed a period of very high valuation of the sectors information technology, telecommunications and media. Investors were also uncertain of the impact the falling stock market would have on economic growth. There was worries that a marked reduction in households wealth – caused by the fall in the stock market – might lead to increased savings, thereby reducing private consumption sharply. This was avoided, partly because the Federal Reserve reduced interest rates significantly.

Source Ministry of Finance and Norges Bank

The financial crisis triggered expectations of a deep, global recession. The crisis illustrated just how tightly integrated the financial system is and how vulnerable the economy is to problems in this system. The financial crisis was both a credit crisis and a liquidity crisis. The lack of liquidity led to different markets being more closely linked than normal. This was reflected in the returns in the various sectors. The crisis affected all the sectors, and most sectors fell 40 per cent or more (see figure 5.2). The financial sectors in particular suffered massive losses. Banking, insurance and financial services (including investment banks) fell more than 70 per cent.

Source FTSE and Ministry of Finance

At the beginning of the financial crisis, the banking sector was by far the largest single sector in the stock market with a 13.7 per cent share of the total market value. Insurance, banks and financial services accounted for 23 per cent of the total market value of the stock market. The sharp price falls in these sectors thus had a large direct impact on the stock market.

Figure 5.3 shows each sectors contibution to the fall in the stock market so that the total adds up to 100 per cent. The drop in share prices in the financial sectors explains a total 29.3 per cent of the overall decline in the stock market, while the banking sector alone was responsible for 17.7 per cent. In the other sectors, the contributions are far more evenly distributed, even for sectors with high market value such as healthcare and energy. In addition, the financial sectors are major players in the financial markets as investors, issuers of debt securities and by virtue of their central role in the credit market. The problems in these sectors therefore also had a significant, negative, indirect impact on the stock market.

Source FTSE and Ministry of Finance

The developments in the sectors during the financial crisis can be assessed in light of their sensitivity to the stock market during «normal» times. This provides a basis for assessing how the sectors were influenced by the financial crisis beyond the general decline in the stock market. Beta is a measure of a sector’s sensitivity to changes in the overall stockmarket. Sectors with historically small fluctuations and low correlation with the stock market have a low beta value and are expected to fall less than the stock market during periods with falling stock market.

Jensen’s alpha is used to assess a portfolio’s return in relation to its beta. A positive alpha means that the sector had lower depreciation during the financial crisis than expected in view of its beta. Jensen’s alpha shows how large this excess return is – after adjustment for risk-free interest.

According to the Capital Asset Pricing Model (CAPM), beta is also a measure of the overall risk that investors will require compensation for. Jensen’s alpha is therefore a measure of risk-adjusted excess return. In financial literature, however, there is general consensus that also other factors should be included in an analysis of risk-adjusted excess return (see the discussion of risk factors in chapter 7). Figure 5.4 shows Jensen’s alpha for the sectors based on beta values estimated from 2004 until 2007.

Figure 5.4 shows that the financial sectors had a negative return adjusted for beta. The banking sector fell 72 per cent during the financial crisis – 22 per cent more than the historical beta values indicate. The corresponding figure for life insurance was 21 per cent and 9 percent for financial services. Sectors in which developments in emerging economies are important – such as oil and gas, mining, industrial metals and chemicals – made a significant, positive contribution. In recent years, these sectors have had significantly higher returns than the overall stock market.

Source FTSE and Ministry of Finance

In turbulent markets, we often see an increasing correlation between risky assets. This means that the effect of diversification is reduced in these periods. As the financial crisis progressed, growing correlation could be observed among all the sectors – especially in those sectors that had initially had low correlation and thus a low beta. Sectors that historically were not especially sensitive to the stock market therefore achieved a weaker beta-adjusted return in figure 5.4 than sectors that already had high sensitivity. Figure 5.5 shows the return in the sectors during the financial crisis and their respective beta values. On average, sectors with a high beta fell more than sectors with a low beta, but the difference was less than suggested by the historical data. This indicates that other factors may have been more important for the sectors’ performance during the financial crisis than those that formed the basis for the calculation of the beta value.

Source FTSE and Ministry of Finance

Many investors also had similar experiences with different asset classes and investment strategies – which were much more closely correlated with the stock market during the financial crisis than they had been in the past. Many sources cite the massive reduction in access to liquidity as an important explanation. During the financial crisis, large positions in the financial markets had to be liquidated as investors no longer had access to financing. Falling prices and lower liguidity led to increased demands for provision of collateral for leveraged positions. This in turn caused more securities being sold off. The banks had also major problems borrowing, and any excess liquidity was placed in liquid government bonds, rather than loans to investors. During the financial crisis, the investors were also very uncertain about the potential outcomes of the crisis. It was difficult to assess the consequences of the extensive problems in the financial system. In a situation with sharply falling growth, a banking crisis and extreme market volatility, investors were looking to reduce their exposure to shares in all sectors. This had the effect of increasing the correlation among the sectors. In the financial markets, there was growing demand for low-risk securities – primarily government bonds – while risky assets became more closely correlated.

From 2003 to 2008, the market went from having very good access to liquidity to a severe lack of liquidity. The outcome of this was a sharp increase in the volatility of risky assets, and the stock market plummeted from October 2007. In turbulent markets, investors demand safe and liquid securities, as they offer predictability and flexibility. In markets with good access to liquidity, investors are not willing to pay as much for these properties. Figure 5.6 shows the relationship between the developments of a liquidity premium in the government bond market (which is described in more detail in chapter 7) and volatility in the stock market.

Source FTSE and Ministry of Finance

The stock market is driven by expectations of future earnings and valuation. Valuation of the stock market is often procyclical and reflects increased optimism and willingness to take risks among investors. Factors such as good access to liquidity often contribute to higher valuation of the stock market. However, the rise in the stock market prices prior to the financial crisis was driven less by high valuation than often seen previously. This probably helped moderate the fluctuations in the stock market during the financial crisis. A higher valuation at the onset of the crisis could have led to an even larger fall.

Figure 5.7 shows the developments in the stock market and valuation. Valuation is measured using the value of the US stock market compared with ten-year rolling earnings. Prior to 2000, increased valuation was a very important factor behind the upturn. There were extremely high valuations in the technology, telecommunications and media areas in particular. When these sectors did not deliver results consistent with the expectations, there was a significant drop in valuation. The stock market fell 45 per cent – only 9 percent less than during the financial crisis – despite the fact that financial crisis had far greater negative consequences for the real economy.

Source FTSE and Ministry of Finance

5.3 Developments in the fixed-income market

The level of risk in the GPFG is determined primarily through strategic asset allocation. The interaction between the equity and fixed-income indices is the main source of risk diversification in the fund.

During the decline in the stock market in the period 2000–2002, the fixed-income benchmark rose by 19.5 per cent, whereas during the financial crisis it rose by 6.3 per cent. The fixed-income benchmark thus helped mitigate the drop in the Fund’s total benchmark index in both periods. In particular, government and government-guaranteed securities helped dampen the impact on the Fund’s total return in these periods. This is due not only to the fact that sharp falls in the stock market have been followed by a more expansionary monetary policy, but also that government bonds are very liquid and are generally considered a «safe haven» in turbulent markets. Corporate bonds will also benefit from falling interest rates, but usually the yield differential between government bonds and corporate bonds increases as the stock market falls. Falling stock markets reflect expectations of lower cash flows from the corporate sector and thus reduced debt-servicing capacity.

Figure 5.8 shows the returns in the different issuer segments in the fixed-income benchmark. With the exception of corporate bonds, all the segments developed positively during the financial crisis. Roughly two-thirds of the fixed-income benchmark has an explicit or an implicit government guarantee. This was central to the return on the fixed-income benchmark during the financial crisis.

Source Ministry of Finance and Norges Bank

Corporate bonds had a negative return during the financial crisis. The benchmark index for the GPFG consists exclusively of corporate bonds with high credit ratings – i.e. investment grade bonds. Increased credit risk was an important cause of the fall, but a sharp reduction in liquidity also served to lower the return on these securities. The fixed-income markets were hit much harder by lower liquidity during the financial crisis than the stock market.

Figure 5.9 shows the value development of the fixed-income benchmark and the 12-month rolling correlation between the return on the equity and fixed-income benchmarks. With few exceptions, the fixed-income benchmark has risen throughout the period, meaning that developments in the stock market have been important for the measured correlation. The correlation between equities and interest rates has varied over time and has been positive for prolonged periods. Since 1997, however, inflation has been low and stable, giving central banks considerable scope to lower interest rates in response to problems in the financial markets or the economy.

Source Ministry of Finance and Norges Bank

6 The academic research on efficient markets and active management

6.1 Introduction

Chapter 2 of this Report provides a detailed account of the evaluation of the active management of the GPFG. It states, among others, that the Ministry of Finance ordered an external report from professors Andrew Ang (Columbia Business School), William N. Goetzmann (Yale School of Management) and Stephen Schaefer (London Business School); the purpose of which was to account for the theoretical and empirical basis for active management, analyse the results achieved by active management of the fund in the past, and discuss the foundation for active management in the future. This article presents a summary of the results of the expert group"s review of the academic research on efficient markets and active management.

6.2 The Efficient Market Hypothesis

The report from Ang, Schaefer and Goetzmann is based on the «Efficient Market Hypothesis» (EMH), which was developed in the 1960s and 1970s. The efficient market hypothesis states that the price of a financial asset, like a share or bond, always reflects all available information about the asset"s fundamental value. If this hypothesis is correct, it will be impossible for a manager to consistently «beat the market». Active management will thus not have much of a role to play regarding adding value.

The theoretical starting-point for the efficient market hypothesis is based on a large active marketplace for listed securities, characterised by tough competition between investors, leading to the lack of any extra return beyond normal compensation for market risk. An implication of this is that investors who seek to achieve a higher return through active management will accrue a loss. A pure indexing strategy will consequently outstrip a strategy that is based on active management, where the purpose is to use erroneously-priced securities.

The efficient market hypothesis is based on a number of strict assumptions. During the past two decades, the academic research on efficient markets has attempted to further develop the hypothesis by making less strict assumptions. The result of this research is a more modern description of the EMH, which takes consideration of the existence of market frictions, the costs of information gathering, principal–agent problems, and restrictions related to capital structure, among others.

The report by Ang, Goetzmann and Schaefer shows that this development does not change the basic insight from the efficient market hypothesis; namely that it is very difficult to achieve excess return in a market characterised by a high level of competition. However, the academic research does not preclude the existence of cases of erroneous pricing. Recent research has focused in particular on the institutional framework conditions that must be present to be able to use such erroneous pricing, and the extent to which an active role yields benefits that can be transferred to tasks other than «beating the market».

Academic studies are described below that attempt to test the extent to which the efficient market hypothesis can be rejected within different markets and asset classes.

6.3 Empirical studies of the Efficient Market Hypothesis

In the report, Ang, Goetzmann and Schaefer point out that much of the academic research on the efficient market hypothesis and active management is based on data from the US stock market, and US equity funds. This limit is due to the lack of access to data for other markets and investors.

Empirical studies of the efficient market hypothesis are either based on tests of prices or of managers. Price tests are typically based on investigating whether specific types of active strategies could have been exploited to create a positive risk-adjusted return when they are tested on historical data. If it is possible to prove profitability in such strategies, this is considered a breach of the efficient market hypothesis. Active strategies or trading rules that contain this quality are often considered «anomalies».

In general there are a number of methodical issues associated with this type of test. So-called "data mining" is one issue, and is related to the fact that when many researchers actively analyse specific data, it is almost unavoidable that one or more trading rules will incidentally appear to be profitable later. Another issue is related to price tests in practice being a shared test of the model for the expected return and level of market efficiency.

Price tests are generally based on a specific pricing model that is presumed to represent the "correct" price of the financial asset (for example a share) in the form of exposure to a set of common risk factors. The simplest model is the «Capital Asset Pricing Model» (CAPM), while the one used most lately is the multi-factor model developed by Ross (1976): «Arbitrage Pricing Theory» (APT). This model is based on investors being compensated in the form of a higher expected rate of return to assume the risk of exposure to a set of risk factors. Both the CAPM and APT model place emphasis on the risk factor"s importance regarding setting the expected future return on an investment.

A number of empirical studies in the 1970s and 1980s showed a breach of these models in that accounting issues, different goals for company earnings and more company-specific issues could be exploited to predict future movements in share prices. These findings were called anomalies because they could not be explained by the theoretical models. For example, the capital asset pricing model states that the expected return only depends on a single factor: market risk (beta). Later studies also revealed breaches of models that include several risk factors other than market risk; among others, size, value and momentum (cf. table 6.1). This is described in further detail in article 7 in this report.

As tests of the efficient market hypothesis based on prices are a shared test of the specific pricing model and the level of market efficiency, the specification of the pricing model will be of great importance to the result. Many of the early studies that detected a breach of market efficiency (anomalies) were based on a model where market risk is the only risk factor. Specification of this model was further typically based on a reference index that consisted of a market-weighted stock portfolio for the US market.

The debate regarding the efficient market hypothesis has lately focused particularly on the extent to which such breaches must be interpreted as inefficiency, or whether they reflect that other relevant risk factors have been inadequately identified and specified. The report from Ang, Goetzmann and Schaefer points out that independently of whether the efficient market hypothesis is due to inefficiency or no consideration has been taken of other relevant risk factors, studies of price tests may be of potential relevance to the GPFG, if they can state possible sources of return.

In their report, Ang, Goetzmann and Schaefer argue that if the benchmark is exclusively a market-weighted portfolio made of up all securities traded, active management (defined as a deviation from these market weights) may be appropriate to achieve exposure to risk factors that are not covered by the exposure to market risk. Academic research supports investors over time receiving compensation for taking systematic risk, for example by investing in value shares, compared with growth shares. If more systematic factors are included in addition to the market risk factor in a price test, empirical studies show that a hypothesis that the market portfolio is efficient can be rejected.

Testing active strategies that are based on creating risk-adjusted excess return (pure alpha), based on historical data, indicates that there are a number of potentially profitable investment strategies. However, Ang, Goetzmann and Schaefer point to several issues that mean that these strategies have limited transfer value to the fund. First, the profitability of these strategies represents a calculated, and not actually achieved excess return. No consideration has further been taken of transaction costs, management costs and market influence. Second, the results may be influenced by the potential skewedness that follows from the methodical problems associated with data mining (cf. above). Changed market conditions, including a variation of the profitability of active strategies over time, mean that it is generally difficult to extrapolate the results. Third, Ang, Goetzmann and Schaefer make reference to several of the proven profitable deviations from the market portfolio not being scalable, and that they therefore cannot be executed on a scope relevant to the GPFG.

The other type of empirical studies of the efficient market hypothesis is tests based on managers and institutional investors (cf. table 6.2). These tests are more relevant to the fund than tests based on prices.

Many of the empirical studies of research results among managers are based on equity funds, but typically will have a cost structure and a starting-point for value creation that significantly deviates from the Fund. The results of these studies therefore have limited relevance to the management of the GPFG.

More recent academic research indicates that certain managers have special skills that allow them to achieve a better return than the return in the market, calculated before management costs. However, the research shows that there is no statistical basis for determining that the special skills are reflected in excess return for the client. The empirical studies generally show that for equity funds, the yield after costs and risk-adjusted excess return (alpha) is on average zero or negative (cf. table 6.2). At the same time, there is support for some managers being better than others, even though an average equity fund typically performs more poorly than a passive portfolio on a risk-adjusted basis after management costs.

The empirical studies also show that for management through hedge funds there is no statistical support for the creation of a positive risk-adjusted return after deducting management costs. The data series at the heart of these analyses are significantly shorter than the studies of equity funds, and the quality of the data is also less certain. At the same time, the market for hedge fund management is changing, which makes it difficult to extrapolate the results of these studies. However, Ang, Goetzmann and Schaefer state that there is little support in statistics for private equity and venture capital to create a positive risk-adjusted excess return. Some studies indicate a certain level of persistence in skills, but it is not possible to draw clear conclusions, based on available data. For property management, there is not enough data to be able to assess the extent to which managers have created value on a risk-adjusted basis.

There is an even poorer foundation within other types of institutional management, like large university funds (endowments", pension funds and sovereign wealth funds, to determine whether active management can yield a positive risk-adjusted return. Certain US university funds achieved very good results before the financial crisis through an investment strategy based on alternative asset classes. It is asserted that a long time horizon allowed these funds to focus on alternative asset classes: Ang, Goetzmann and Schaefer make reference to most studies indicating that pension fund managers are unable to identify the best managers in advance, at the same time as the managers have little ability to create a risk-adjusted excess return. The few studies of sovereign wealth funds that are based on investments in listed securities show that share prices react positively when a sovereign wealth fund makes an investment, at the same time that the long-term results of these investments are not particularly good.

6.4 Summary

The modern description of the efficient market hypothesis allows for the existence of profitable active management strategies, based on exploitation of comparative advantages. Such advantages can be specialised knowledge, lower transaction costs, lower management costs or lower costs linked to principal-agent problems, and a financing structure that allows trade in securities with a long verification horizon. Such strategies mean that liquid securities markets are generally very efficient, even though there may be certain deviations (anomalies). Ang, Goetzmann and Schaefer argue in favour of being able to view the consideration between passive indexing and active management as a decision-making variable. The best adaptation will generally depend on perceptions of the existence of and potential linked to the manager"s skill, the pricing of the securities on the market in question, the investor"s time preferences and risk aversion, and the manager"s competence and reward system.

Ang, Goetzmann and Schaefer believe that even modest skill on the part of a manger speaks in favour of at least part of the GPFG being managed actively.

7 Risk factors in the stock and bond market

Report to the Storting no. 20 (2008–2009) contained a discussion of priced systematic risk factors in the stock, bond and property market (cf. box 2.3, page 54). It showed that multi-factor models have taken over for the simple capital asset pricing model as an explanation of stock return. Credit risk, term risk and liquidity risk are recognised priced risk factors in the bond markets. Also possible priced risk factors in the property market were discussed.

Systematic risk is the risk one is left with in a well-diversified portfolio, while the diversifiable risk goes under other names like security-specific and idiosyncratic risk. The systematic risk reflects the inherent uncertainty of the economy. Investors cannot diversify away from economic downturns, credit crunches, lack of liquidity and market collapses, etc. However, an investor can refrain from investing (or only invest a smaller portion) in securities, such as shares, which tend to fall sharply in value in bad times.

As most investors are risk averse, and because systematic risk is inextricably associated with investments in securities, investors demand compensation for this risk in the form of a higher expected return. An important insight from finance theory is that the required return on a share or bond is linked to the investment’s contribution to the portfolio’s systematic risk, and not the risk of the security in isolation.

It was assumed in the finance theory developed in the mid-1960s (the capital asset pricing model) that the systematic risk of a security depends on the covariation between the return on the security and the return on the market portfolio, measured by the beta. Securities with a beta greater than one are more risky than the market in general, and will have a correspondingly higher required return than the required return on the market portfolio.

The capital asset pricing model is of great importance to the understanding of return and risk in the securities markets. A security"s beta is still considered an important priced risk factor. However, empirical research has shown that the relationship between return and risk is more complex than assumed by the model. For example, the model cannot explain a number of phenomena observed in the securities markets. Small companies have, for example, proven to outperform large companies with regard to their average rate of return, and this cannot be attributed to differences in the companies’ betas. Nor can the model explain why companies with a low market value compared to recorded equity (so-called value companies) have later had a higher average return than companies with a high market value compared to recorded equity (growth companies). Empirical studies also show that companies that have had high return over the past 3–12 months tend to have high return also during the subsequent 3–12 months (cf. chapter 6). This so-called momentum effect cannot be explained by different market betas.

This indicates that the capital asset pricing model is too simple. Moreover, it is based on a set of unrealistic assumptions.

In light of this, Fama and French 1 expanded the capital asset pricing model to include two more systematic risk factors: company size (given by market value) and value (market value compared with recorded equity). According to this model, small companies will have a higher expected return than large companies, corrected for the market beta. The same applies to value companies. Carhart 2 expanded this so-called three-factor model to include a momentum factor («four-factor model»).

A possible theoretical explanation of the size and value effects is that small companies and value companies are more vulnerable during crises. During economic downturns it is assumed that it is easier for this type of company to go bankrupt or suffer greater financial problems than other companies. The stock return will thus be especially poor under such market conditions, and poorer than the companies’ market beta exposure would suggest. Instead of securing stability, these shares contribute to reinforcing the fluctuations in the portfolio"s return. The average investor will then demand a higher return on such shares.

A corresponding «rational» explanation of the momentum effect has not been developed.

Even though these three and four-factor models provide a better explanation of the return differences in the stock market, and some of the factors have a somewhat intuitive explanation, financial theorists do not fully agree that all of these factors are actually systematic risk factors, priced in the market. Some prefer a behavioural explanation which involves temporary mispricing, which may reflect less rational investor behaviour.

Liquidity is another known risk factor. As liquidity varies over time, there is uncertainty about future transaction costs. Liquidity also affects price levels, and liquidity fluctuations may therefore also affect the price volatility. Liquidity variations therefore represent an extra risk in addition to the ordinary market risk. Securities that are relatively illiquid when the market return is generally low are particularly risky for investors, because they reinforce the return variation. In theory, this is compensated through a higher expected return.

Empirical surveys support the idea that liquidity risk is priced, and may explain some of the difference in the average rate of return between different investments. However, it has turned out to be difficult to calculate reliable time series for the liquidity premium.

A globally diversified portfolio like the GPFG is probably exposed to many of these systematic risk factors, both in the strategic benchmark and in the actual portfolio. The report from professors Ang, Goetzmann and Schaefer about active management of the GPFG points out that the fund"s purpose and uniqueness may call for another exposure to some of these risk factors than what is embedded in the market portfolio – see the discussion in paragraph 2.3. While the average investor keeps the market portfolio per definition, investors with different risk tolerances will seek different levels of exposure to the priced risk factors, thereby optimising the trade-off between expected return and risk. Cochrane 3 provides a good description of such portfolio adaptation to several risk factors.

A key issue is how different risk factors may affect return and risk in the GPFG. Historical data shows that the return distributions to risk factors can be very different. An investor looking for another factor exposure than given by the market portfolio must therefore expect different return and risk. The return distributions of some of the most important risk factors in the stock and bond market are described below. The historical data used is based on the report by Ang, Goetzmann and Schaefer. This data set contains monthly factor returns for the period of December 1997 – September 2009.

The monthly excess return in a globally diversified portfolio of shares in value companies , relative to growth companies, is shown in figure 7.1. Value and growth companies are defined here by the index provider MSCI. Value companies have had higher return than growth companies during the period (2.1 per cent annual average excess return). This is approximately the return an investor would have achieved (before deductions for transaction costs) by selling the growth companies in the MSCI index and buying the value companies. The annual standard deviation (volatility) of this return has been 9.1 per cent. Figure 7.2 shows the value over time of a portfolio that has been invested according to such a strategy. The volatility is relatively low, and the fall in value during the financial crisis has been moderate, compared with the strong fall in the global stock market in general (shown by the FTSE All-World total return index). This reflects the fact that the return on this value portfolio has been weakly correlated with the return in the global stock market.

Source Ang, Goetzmann and Schaefer (2009)

Source Ang, Goetzmann and Schaefer (2009), and Datastream

The monthly excess return in a globally diversified portfolio of shares in small companies , relative to large companies, has also been positive (3.5 per cent annual average excess return since 1997). The volatility of the excess return has been 7.7 per cent. Large and small companies have again been defined by MSCI. This is the return an investor would have received, before transaction costs, by selling the large companies in the MSCI index, and buying the small ones. Also this portfolio would have been less volatile than the general stock market. The correlation with the stock market has been low.

The monthly excess return on a momentum strategy is shown in figure 7.3. This is the difference in return between US shares that have had, respectively, a high and a low return during the past 12 months. The «winner shares» have on average yielded almost two per cent higher annual return than the «loser shares» during the period. However, the volatility of the excess return has been very high, fully 23 per cent. The return distribution also has so-called fat tails, or greater weight in the tails than does the normal distribution. It is also skewed towards low return (negative skewness), which reflects increased downside risk. A momentum strategy based on buying US winner shares and selling loser shares would have given a return that is weakly correlated with the US stock market, except during the financial crisis when the return on the strategy fell dramatically, shortly after the broad stock market had fallen. This is illustrated in figure 7.4. Dimson, Marsh and Staunton 4 have studied the momentum effect in 17 developed stock markets using return data dating back to 1900. They find that a momentum strategy as described above would yield a very high return (before transaction costs) in most stock markets. However, it would be costly to implement the strategy, due to high transaction costs. Their result therefore says little about future return or costs associated with this strategy.

Source Ang, Goetzmann and Schaefer (2009)

Source Ang, Goetzmann and Schaefer (2009), and Datastream

In the bond markets, term risk and credit risk are known priced risk factors, giving rise to term and credit premia. US government bonds with long time to maturity (over 20 years) have had about a 4 per cent higher annual return than government securities with short time to maturity (US Treasury bills, 1–3 months) since 1997. The volatility of the excess return has been about 12 per cent. However, the high excess return is not representative of the long run expected term premium on US bonds, as it was generated by falling interest rates during the period. This fall led to high return on bonds with a long duration. A more reasonable estimate of the long run term premium is 0.5 – 1.5 per cent, depending on the duration (cf. chapter 8).

Credit risk gives rise to an expected credit premium, over and above the yield on corresponding bonds without credit risk. The size of the premium depends on the creditworthiness of the issuer. However, the realized (ex post) credit premium can vary greatly over time, and over extended periods it has been negative. Examples are US corporate bonds with a medium to high credit rating (Aa), which since 1997 have yielded an average annual negative excess return of 0.6 per cent, relative to US government bonds. Over the same period, US corporate bonds with a very low credit rating («high yield») have had annual negative excess return of 3.8 per cent, relative to corporate bonds with a somewhat higher credit rating (Baa). These losses are mainly due to the financial crisis (and to a lesser extent the technology bubble that burst in 2000–2002), which pushed up the expected credit premium and thereby drove down the value of corporate bonds. The negative return since 1997 is therefore not representative of the expected long run credit premium. Dimson, Marsh and Staunton 5 have estimated the historical credit premium of US corporate bonds dating back to 1900. They find a premium, relative to government bonds, of about half a percentage point for bonds with a medium to high credit rating (Aaa and Aa).

As already mentioned, it is difficult to estimate the liquidity premium . However, the data set from Ang, Goetzmann and Schaefer contains information that can be used to estimate the realized liquidity premium. This information is the difference in yield on 10-year US government bonds which are, respectively, «off-the-run» (older securities that are fairly illiquid) and «on-the-run» (newer securities that are relatively liquid). Empirical studies indicate that this difference, multiplied by the duration, can provide an approximate estimate of the realized liquidity premium, at least in the government bond market (cf. Ang, Goetzmann and Schaefer"s report). The result of such a cal­culation is that the average annual return on liquidity risk has been marginally negative since 1997 (-0.1 per cent), and that the return distribution has very fat tails and is skewed towards low returns. A liquidity strategy based on "harvesting" the liquidity premium in the US government bond market by selling the liquid government bonds and buying the illiquid ones would thus have lost value in 1997–2009, as shown in figure 7.5. The loss probably reflects rising liquidity premia during the financial crisis and during the bear markets in 2000–2002. This led to negative return on liquidity risk. However, as mentioned earlier, it is reasonable to assume that the liquidity premium is positive in the long run.

Source Ang, Goetzmann and Schaefer (2009)

Many researchers believe that there are even more priced risk factors, and that their importance can vary over time.

Ang, Goetzmann and Schaefer estimate the contribution to excess return from two additional risk factors: volatility and foreign currency investments financed by foreign currency loans («foreign exchange carry»). The volatility factor captures the difference between the expected and realized volatility. Investors who can tolerate high market volatility can earn a risk premium over time by selling insurance against unexpectedly high volatility to investors who want protection against such volatility shocks. Figure 7.6 shows the return since 1997 associated with such a volatility strategy in the US stock market, over and above the risk free interest rate. The return was consistently high until the financial crisis hit. Then heavy losses effectively wiped out all accumulated return earned since the end of 1997. The return improved in 2009, so that the average annual return was marginally positive at the end of September 2009 (1.2 per cent above the risk free rate). The low return reflects the unexpectedly high stock market volatility during the financial crisis, which entailed great losses for the issuers of volatility insurance. The extent to which this strategy will yield a positive risk premium over the long run depends on the pricing and assessment of future market volatility. As with any type of insurance, the strategy can be profitable if the price of risk is set correctly. Ang, Goetzmann and Schaefer argue that volality belongs to the set of priced risk factors in the stock market.

Source Ang, Goetzmann and Schaefer (2009)

8 Expected long-term real return and risk

In Reports no. 16 (2007-2008) and no. 20 (2008-2009) to the Storting, the Ministry explained the market expectations underlying its calculations of the GPFG"s long-term return and risk. More specifically, expectations were formulated for long-term real return and risk (volatility) associated with the global equity and bond indices in the Fund"s strategic benchmark, and for a globally diversified real estate portfolio. In addition, long-term correlation coefficients between the real returns on these asset classes were estimated.

It is necessary to periodically assess these estimates in light of new information. The financial crisis and its effects on world capital markets over the past few years represent important new information in this respect. This chapter reviews the estimates and considers the need for changes. It is shown that only some minor adjustments are needed.

The Ministry"s earlier estimates of expected long-term annual real return, risk and correlations are shown in tables 8.1 and 8.2. Of course, considerable uncertainty is associated with such point estimates.

The basis for these market expectations is an analysis of historical real returns, and a review of recent research literature.

The estimates are used to calculate the expected real return and risk on the Fund"s strategic benchmark. This is done by using a stochastic simulation model, which simulates the portfolio value over time.

The expectations are intended to apply over very long investment horizons. They represent estimates for average annual real return and volatility over a period of many decades, or over a period that is long enough to encompass many economic and financial cycles. Such long-term expectations are often called unconditional, since they are little affected by time varying factors. In the medium term, for example the next 10 to 20 years, the expected average real return and volatility may differ from the long-term estimates, for example due to time varying risk premia or imbalances in market prices, as pointed out in the Report no. 20 (2008 – 2009) to the Storting. Estimates for such time varying expectations are called conditional. The final part of this chapter contains a brief discussion about whether the return expectations for the medium term deviate from the long-term expectations.

Since the unconditional expectations are for very long time horizons, the threshold is relatively high for making major changes solely based on market developments over the past two years.

In the following, the market expectations in Tables 1 and 2 are discussed, with an emphasis on equities and bonds.

Expected real return and volatility in the long run

Government bonds (in the form of a globally diversified government bond portfolio with a duration of around 5 years, and country weights approximately as in the GPFG"s strategic benchmark) is expected to provide an annual real return of 2.5 per cent 6 and volatility of 6 per cent, according to Table 8.1. The expected real return is higher than the historical average for the period 1900-2009, which is 1.1 per cent, according to annual data from Dimson, Marsh and Staunton (2010), see Table 8.3 which shows the historical returns and volatility. At the end of 2007, before the financial crisis had fully impacted the financial markets, the historical average real return was marginally lower (1 per cent). The low historical real return shows that government bonds were a bad investment during several periods in the last century, due to high inflation (or hyperinflation) and wars.

However, the Ministry"s estimate is considerably lower than the real return on government bonds after 1975 (after the collapse of the Bretton Woods fixed exchange rate system), which is a period that may be more relevant to compare with (5.3 per cent real return at the end of 2009, and about the same at the end of 2007). The high real return during this period reflects the battle against inflation and the decline in inflation expectations through the 1980s and 90s, which more than offset losses in the bond markets during the period of high inflation in the 1970s. Given the current situation, where central banks largely aim for low and stable inflation, a recurrence of a similar disinflationary path is not considered the most likely scenario for the coming decades.

The Ministry"s estimate of future long-term real return on government bonds (2.5 per cent) therefore seems reasonable, even after the experiences of 2008 and 2009. The estimate lies between the very low average for the period 1900-2009 and the very high average for the period 1975 to 2009. The estimate is consistent with an expected real return on short duration government bills of 1 to 2 per cent and a term premium of 0.5 to 1.5 per cent. Historically (1900-2009), this term premium has been about 1.4 per cent according to the data set from Dimson, Marsh and Staunton (for government bonds with long duration).

The expected volatility of bonds (6 per cent, see Table 8.1) is lower than the historical volatility, which has been 8.9 per cent in the period 1900 to 2009, see Table 8.3. In the period 1975-2009 volatility was about the same (8.3 per cent). However, the average duration of the GPFG"s government bond portfolio is significantly lower than the duration of the government bonds that are part of the historical return series of Dimson, Marsh and Staunton. It is therefore reasonable to assume a lower volatility for the GPFG"s bond portfolio.

The Ministry therefore keeps unchanged its estimates of the expected real return and risk on the bond portfolio.

Equities (in the form of a globally diversified equity portfolio with country weights approximately as in the GPFG"s strategic benchmark) are expected to yield an average annual real return of 5 per cent (geometric), according to Table 8.1. The historical data set of Dimson, Marsh and Staunton shows that the corresponding historical average for the period 1900-2009 is about 6.3 per cent, as shown in Table 8.3. At the end of 2007 the historical average was 6.7 per cent.

This shows that the experience of the past two years, with the dramatic fall in equity markets in 2008 followed by strong recovery in 2009, pulls down the historical average real return by only 0.4 percentage point.

If the historical stock market analysis is limited to the period after the Second World War (from 1946), thereby giving modern times greater weight, the real return is somewhat higher (7.1 per cent at the end of 2009, and 7.8 per cent at the end of 2007).

As can be seen from these statistical measures, the Ministry"s estimates for future average real return on equities is lower than the historical average since 1900, although the difference is reduced somewhat when one includes the return figures from 2008 and 2009 in the historical average. As financial research has shown, however, historical returns are not necessarily a good estimate of future returns. Nor can future equity return be estimated independently of the return on alternative, less risky investments (or independent of the pricing of the stock market and expected future dividend streams). The estimate will also depend on the assumed size of the equity risk premium («equity premium»).

The expected equity premium can be defined as the difference between the expected return on a globally diversified equity and government bond portfolio. According to the Ministry"s estimates the long-term equity premium is 2.5 per cent (5 per cent real return on stocks minus 2.5 per cent real return on government bonds). This estimate can be compared with the historical premium. The average historical equity premium may be determined by taking the difference between the (geometric) average real returns on global portfolios of stocks and government bonds, as described above (country weights about the same as in the GPFG"s strategic benchmark). One then finds that the average equity premium was significantly greater than 2.5 per cent over the period 1900 to 2009 (5.2 per cent, see Table 8.3). However it has fallen in recent times, to 2.6 per cent for the period 1975 to 2009. The low equity premium in the recent period is partly connected with the high real return on government bonds during this period. Over the last decade (1999-2009) the realized equity premium has actually been negative (minus 6.5 per cent), reflecting the large drops in world stock markets during the bursting of the dotcom bubble and during the financial crisis. In Japan, the equity premium has been negative over an even longer period. This is an effect of the sharp decline in Japanese asset prices that began in 1990.

In light of these historical figures for the equity premium the Ministry"s long-term expectation of 2.5 per cent seems realistic. The expectation is in line with the equity premium realized in recent times. The estimate is also consistent with some important research results. Dimson, Marsh, Staunton and Wilmot (2009) 7 estimate an expected long-term equity premium in the global equity market of 3 to 3.5 per cent against short duration government securities, which corresponds to approximately 2 to 3 per cent relative to long duration government bonds (geometric mean). According to these researchers, there are three main reasons why the historic equity premium has been higher than this: dividends have been historically high, there has been unexpectedly strong growth in the world economy after the Second World War, and the required return has fallen. A repeat of this course of events cannot be expected. Ibbotson and Chen (2003) 8 estimate a future equity premium 1.25 percentage points lower than the historical average. Arnott and Bernstein (2002) 9 find that a «normal» equity premium is below 2.5 per cent. However, some forecasts are more optimistic. A late 2009 survey of the Chief Financial Officers of a large number of U.S. companies shows that on average, they expect a long-term (10-year) equity premium in the U.S. stock market of about 4.4 per cent, when measured against 10-year U.S. government bonds. 10

In this light, the Ministry"s estimates of the expected real return on government bonds (2.5 per cent) and the equity premium (2.5 per cent) still seem realistic.

Expected stock market volatility has until now been 15 per cent, see Table 8.1. This expectation value was largely intended to reflect historical volatility in the stock market, as this is considered a good reference point for estimating future volatility. Historical volatility was also 15.0 per cent at the end of 2007, when the estimate was prepared, see Table 8.3. However, the financial crisis, which resulted in wide fluctuations in equity returns in 2008 and 2009, pushed the historical volatility up to 15.6 per cent at the end of 2009. In light of this, the Ministry is adjusting the estimate of future stock market volatility upwards, to 16 per cent.

Another consequence of the financial crisis is that the historical distribution of annual stock market returns has become slightly more skewed in the direction of lower (real) returns (a little more «negative skewness»). Although this effect does not have very high statistical significance due to few observations, the Ministry now chooses to model a slight degree of negative skewness when the real return on equities is simulated in the stochastic portfolio model used to calculate the strategic benchmark’s real return and risk. This procedure is expected to give a more realistic description of the risk of losses, which is often underestimated when a normal distribution, without skewness, is used. The modelling of negative skewness is done using stochastic shifts of log-normal return distributions.

Real estate investments are usually expected to provide a long-term return that lies between government bonds and equities, and this also applies to expected volatility. Given that the Ministry assumes an equity premium of just 2.5 per cent, the possible range for the «real estate premium» (the difference in the real return on a global real estate and government bond portfolio) is narrow. The Ministry has estimated this premium to about one percentage point, which gives an expected real return of 3.5 per cent, see Table 8.1. The estimate of volatility (12 per cent) is in the upper part of the interval between government bonds and equity volatility, and thus takes into account that real estate investments are leveraged to a certain extent. Reliable historical return data for comparison are hard to find. Available data go only a few decades back in time, and do not reflect the true time variation in the return. Estimates of future real return and volatility on global real estate investments are therefore associated with particularly high uncertainty. The Ministry finds no reason to change its estimates.

Expected correlations in the long term

The correlation between the annual real return on equities and government bonds is estimated at 0.4, see Table 8.2. This is a measure of covariation, where a correlation of 1 means that the annual real returns show synchronous variation. The estimate of 0.36 is close to the historical correlation for the period 1900-2009, for globally diversified portfolios of equities and government bonds with country weights about the same as in the GPFG’s strategic benchmark (again according to the data set from Dimson, Marsh and Staunton). However, the historic correlation has varied considerably over time. More recently (1975-2009), the average was 0.08, meaning that the real returns have varied almost independently of each other. This low correlation is usually attributed to the low and stable inflation in the industrialised world since the 1980s. When inflation is stable, variations in economic growth become more important for the correlation, which is pushed down. The correlation also goes down when investors «flee» to secure government bonds in turbulent times. By assuming a correlation of 0.4, the Ministry has taken into account that the correlation may pick up again for various reasons (for example, because of greater variation in inflation), and fluctuate around the historical average since 1900. The estimate is conservative, in the sense that it does not underestimate the risk in the GPFG"s strategic benchmark (a higher correlation means higher portfolio volatility).

The estimates of the correlations between real estate and, respectively, government bonds and equities are explained in Report no. 16 (2007-2008) to the Storting. The Ministry finds no reason to change them.

The conclusion of this review of long-term expectations is that the earlier estimates in Table 8.1 will be carried forward with the following two exceptions. First, the expected volatility of the strategic benchmark for equities is adjusted up, from 15 to 16 per cent. Secondly, a slight degree of negative skewness is assumed in the distribution of annual stock market real returns. The skewness is incorporated into the simulation model used to calculate the expected real return and risk of the GPFG"s strategic benchmark.

The revised expectations are summarised in Table 8.4.

Expected real return in the medium term

As pointed out earlier, the unconditional expectations are intended to describe the long-term trend rate of return that is consistent with long-term fundamental economic variables. In the shorter term, there may be deviations from the trends, for example because of time varying risk premia or imbalances in the financial markets. Such fluctuations around a more fundamental trend return are often called mean reversion. Market expectations for a medium investment horizon (next 10-20 years) are therefore discussed in the following.

The real return on the bond portfolio will depend on the nominal interest rate today, the future interest rate path, and inflation. Current interest rates on U.S., European and Japanese government bonds are low, which is partly due to the financial crisis. The real interest rate on a government bond portfolio with 3 to 5 years duration (as in the GPFG) is significantly lower than 2.5 per cent, which is the Ministry"s long-term projection. If one assumes that the real interest rate remains as low for the next 10 to 15 years, the expected real return on government bonds will be substantially lower in the medium term than the long-term estimate of 2.5 per cent. However, it seems more likely that the real interest rate will gradually move back to a higher level. Such a course could push up the expected real return in the medium term, since bonds can be re-invested at higher real interest rates. By how much the expected real return will increase depends on the assumptions for the interest rate path.

In the medium term the expected real return on the equity portfolio may also differ from the long-term expectation of 5 per cent. Both the bond return and equity premium vary over time. Many also believe that the stock market is at times mispriced, which may affect future returns. However, there is still considerable disagreement among researchers about whether stock market returns can be predicted. Analyses of predictability are usually based on an assumption of a correlation between certain valuation indicators and future equity return. Examples of such indicators are the cyclically adjusted price-earnings ratio (CAPE), proposed by Robert Shiller, and Tobin"s Q. The first measures the level of the stock market relative to average earnings per share over the last decade (inflation adjusted). The second measures the level of the stock market in relation to the replacement cost. Historically, a high indicator value relative to the historical average has signaled low returns in the medium term, and vice versa. Smithers shows that the signal from Tobin"s Q has been somewhat stronger than the signal from CAPE. 11 This relationship can be understood theoretically using models for pricing future dividend flows. In these models, a high CAPE value means low expected equity return and/or high expected growth in dividends, and vice versa. Alternatively, the signal can indicate mispricing of the stock market (in relation to the equilibrium earnings).

According to Shiller"s own calculation, the CAPE for the U.S. stock market was 20.12 at the end of 2009. The historical average since 1881 is 16.35 (again according to Shiller). This is interpreted by many as an indication of the overpricing of the U.S. stock market (by some 20 per cent), and a (weak) signal of abnormally low real return in the medium term.

Values for Tobin"s Q can be found in the reports «Flow of Funds Accounts of the United States» that the Federal Reserve publishes quarterly. The latest update, for the third quarter of 2009, gives a Tobin"s Q value of 0.91. During the fourth quarter the U.S. stock market rose more than 5 per cent. Given this rise, it is reasonable to assume that Tobin"s Q was very close to 1 at the end of 2009. The historical average, however, has remained well under 1. This may reflect a tendency to overestimate replacement costs in the official data. According to Smithers the historical average is 0.63. Using this average as a basis, Tobin"s Q also signals overpricing of the U.S. stock market.

However, such analyses are associated with considerable uncertainty, since the predictive power of the various valuation indicators is debatable. This uncertainty is reinforced by the fact that current signals for the U.S. stock market only indicate moderate overpricing. Similar analysis of other stock markets would be even more uncertain, because of limited access to historical data. There are therefore no strong indications that the expected equity return in the medium term is significantly different from the Ministry"s long-term estimates.

9 Expectations documents as tools in exercising ownership rights

9.1 Introduction

Norges Bank is the formal owner of the securities in the GPFG"s portfolio and is responsible for exercising the ownership rights that come with the role of shareholder in the companies in the portfolio. The overall purpose of the active ownership work is to safeguard the Fund"s financial values by contributing to good corporate governance and by striving to achieve higher ethical, social and environmental standards in the companies. Norges Bank"s principles for the execution of ownership are based on internationally recognised guidelines: the UN Global Compact, OECD"s principles for corporate governance and OECD"s principles for multinational companies.

Norges Bank has selected six strategic priority areas within its ownership work:

• Equal treatment of shareholders

• Shareholder influence and board accountability

• Well-functioning, legitimate and efficient markets

• Climate change

• Water management

• Children"s rights

In selecting priority areas, Norges Bank, among other things, emphasised that the areas shall be relevant to investors generally and the GPFG"s portfolio in particular; they shall be suitable for dialogue with the companies and/or regulatory authorities, and provide an opportunity for real impact. They should be justifiable financially, since Norges Bank acts in the capacity of investor.

Norges Bank uses several instruments as part of its ownership activities, including voting at annual general meetings, shareholder proposals, dialogue with companies, legal action, contact with regulatory authorities and collaboration with other investors. In recent years Norges Bank has prepared and published what are called expectation documents that are used as tools in the ownership work.

9.2 Specifics of expectation documents

Within the focus areas children"s rights, climate change and water management, Norges Bank has formulated its expectations of the companies in dedicated documents. In these documents the Bank sets out its expectations of how companies should manage the risks associated with these factors. Norges Bank"s objective is to ensure that the companies" business operations create long-term financial gains with acceptable social and environmental effects.

The documents are used both as a starting point for identifying and analysing various sectors and as a basis for feedback and dialogue with individual companies. Follow-up of the expectations documents is discussed further below.

The public can provide input to the formulation of expectations documents through a consultation process, which was adopted during the evaluation of the ethical guidelines in 2009.

The first expectations document Norges Bank published was in the field of children"s rights. The expectations cover, inter alia, prevention of the worst forms of child labour, sustaining of the minimum age standard, monitoring systems, measures to ensure children"s health and welfare, and the companies" governing structure and reporting systems with regard to these issues. The document requires the companies to have clear guidelines regarding child labour, perform risk analysis, have plans and programmes to prevent child labour and have corrective measures and systems pertaining to the supply chain. Moreover, companies are expected to have monitoring systems and report on their results. They should integrate the potential economic impacts of social-related factors in the company"s strategic planning and have a transparent and well-functioning governance structure.

In August 2009, Norges Bank published an ­expectation document on the companies" management of risk factors related to climate change. Climate change is expected to have major economic consequences, both as a result of the physical impact of the changes and as a result of regulatory measures that could lead to a different pricing of carbon than is the case today. As a long-term investor with a broad portfolio the GPFG will be exposed to risks as a result of such change. Companies are expected to have strategies for managing both physical and economic climate effects, to measure its emissions and set targets for reducing them, to explore and exploit opportunities to develop new products and services that will help the transition to a low-carbon economy, and to develop a strategy for dealing with climate change risk in the supply chain. Companies are also expected to be constructive when engaging with policy makers regarding regulatory response to climate change and to be transparent regarding their strategies, objectives and progress on climate change management, both in terms of their own operations and in the supply chain.

In February 2010 Norges Bank published an expectations document related to companies" water management. Water is a limited resource, and water shortage may have major environmental, social and economic consequences. Water shortage and poor management of water are risk factors that may have an increasing significance for companies" long-term financial results. For investors such as the GPFG, which is broadly invested in sectors exposed to high water-related risks, companies" management of such risks can have great significance.

The Fund is broadly invested in sectors exposed to high water-related risks. Norges Bank has identified seven sectors that are particularly vulnerable: the food industy, agriculture, pulp and paper, pharmaceuticals, mining, industrial and power generation and water supply.

Expectations in the document relate to three main areas: the companies shall have a clear strategy for water management, a sustainable water management and the governance structure must facilitate realistic strategies and responses. The document requires a clear strategy regarding water management, «water footprint» and risk analysis, preventive and corrective plans for identified risks, supply chain management systems and monitoring systems for environmental and social impacts of activities with regards to water, including sustainable water measures. The company is expected to conduct consultations and collaborate with groups of stakeholders, have a clear policy for water management, transparent perfomance reporting and a transparent and well-functioning governance structure.

9.3 Follow-up of the expectations documents

The expectation documents provide the basis for a comprehensive follow-up process. On the basis of each document, the problems and risk factors of vulnerable sectors are identified.

Norges Bank uses the expectations as a starting point for a systematic analysis of the companies" management of climate change and children"s rights, and from 2010 water management. The analysis of the companies is based on publicly available information. The results are published in an annual sector compliance report.

The results are also used as a basis for giving companies individual feedback, in addition to Norges Bank selecting some companies for further dialogue with the aim of contributing to positive changes in the companies.

The expectations document on children"s rights was the first to be adopted, and in this area, Norges Bank has conducted follow-up as described above since 2008. In the first public report 430 companies in the cocoa, apparel, mining and steel industries were analysed against the expectations document on children"s rights. Based on the findings, Norges Bank contacted 135 of these companies and suggested specific improvements. When the Bank undertook a review of the same companies in 2009, 33 percent of the companies had improved their performance with regard to child labour and safeguarding of children"s rights and reporting on such issues. Norges Bank will continue follow-up work in 2010.

The cocoa and textile industries accounted for the major improvements with regard to the number of companies that had publicly available guidelines on child labour, and other measures such as having plans and programmes to prevent child labour and corrective measures. Improvement was also evident in the mining and steel industries, including making guidelines for child labour public. Furthermore, there was increased transparency when it comes to relations with other stakeholders. Generally, the level of transparency on child labour and children"s rights increased in all sectors.

Norges Bank conducted dialogue on water management with 14 companies in the car, construction, mining, oil, gas, retail and media industries. Norges Bank had dialogue on climate change management with 29 companies in the energy and cement sectors.

In 2010 Norges Bank will publish its second annual sector compliance report on children"s rights and an initial report based on the expectation document relating to climate change. The first report based on the expectation document on water management will be published in 2011.

10 International development in the area of responsible investment

10.1 Introduction

The term «responsible investment practice» has begun to take hold as a recognised and applied concept in the global investment community. It springs historically speaking from the idea that business has an ethical and social responsibility that extends beyond directives to comply with laws and regulations.

At the same time the debate about what constitutes responsible investment practice has gradually moved back to the core of investment management: managing capital with the aim of achieving the highest possible financial return within an acceptable risk, in line with shareholders" interests. There has been a move away from a purely philanthropic or ethical point of view to greater awareness of self-interest. From a perspective of ensuring a long-term return on capital values, many investors consider the following questions relevant: What ensures the companies" assets in the long run? What risk factors must a broadly diversified, long-term investor consider? Are there sufficient converging interests between owners and managers of the capital? Should companies demonstrate that they take due account of environmental and social factors, so as to convince investors that they create value over time?

There is reason to believe that the financial crisis has pushed such issues higher up on the agenda. Financial markets that do not function satisfactorily lead to costs and create uncertainty about the pricing of securities. Questions about good corporate governance, including core matters such as equal treatment of shareholders and transparency in compensation systems and incentive structures contributing to greater correlation with owners" long-term interests, are now considered more important than ever. Institutional investors were the subject of criticism in the wake of the financial crisis. «The Walker Review» 12 in the UK highlighted shortcomings in active ownership on the part of UK banks and other financial institutions and the failure of owners to follow up their long-term financial interests. Among other things, the report points out the reciprocal obligations between major institutional shareholders and the boards of the companies in which they invest:

«The potentially highly influential position of significant holders of stock in listed companies is a major ingredient in the market-based capitalist system which needs to earn and to be accorded an at least implicit social legitimacy. As counterpart to the obligation of the board to the shareholders, this implicit legitimacy can be acquired by at least the larger fund manager through assumption of a reciprocal obligation involving attentiveness to the performance of investee companies over a long as well as a short-term horizon. On this view, those who have significant rights of ownership and enjoy the very material advantage of limited liability should see these as complemented by a duty of stewardship.»

Many investors are also showing growing interest in issues related to the environment and handling of the effects of climate change. Whether and to what degree it will be possible to limit these changes will have great significance not only for the global economy, but potentially also for the financial markets and the pricing of companies. It is essential to have national and international regulation in place to limit greenhouse gas emissions by means of pricing or quota mechanisms and technology, to reduce or eliminate the climate threat. Investors are therefore pressuring governments to put in place clear targets for emissions reductions.

For two years in a row a large group of investors has called on government leaders and climate negotiators to put in place a future climate agreement that is strong and binding. The last petition, promoted in September 2009, was signed by 181 investors, who together manage more than USD 13,000 billion. Investors are also exerting pressure nationally. Recently, the U.S. Securities and Exchange Commission (SEC) introduced new guidelines for environmental reporting from listed companies, following pressure from a U.S. group of institutional investors including the Environmental Defense Fund and investors led by the Ceres investor coalition. The guidelines aim to ensure more uniform reporting of potential effects climate change might have on a company"s performance.

10.2 What do recent studies show about development in the area?

Many investors have over time considered ESG factors (environmental, social and governance issues) in their investment management. Some of these operate according to a goal of financial returns, at the same time as they have clear ethical, social and environmental demands from the owners of the capital. There is also a growing number of investors who consider ESG factors as directly relevant to financial returns in a more narrow sense. This is often referred to as incorporating or integrating ESG factors in investment activities. The growing support for the UN Principles for Responsible Investment (PRI) – where safeguarding ESG considerations is central – is an indication of such a trend.

This trend was given a closer look in a report from the consulting and research organisation «Business for Social Responsibility» (BSR). 13 The idea of incorporating ESG considerations in investment strategy and investment decisions suggests that these types of considerations are viewed as part of investment management. This means in turn that it is very difficult to measure results directly ensuing from an emphasis on these types of considerations. Some investors, however, cite qualitative reasons why they emphasise ESG considerations in their management. Strong ESG performance in a company may be seen as a proxy for strong and effective management. Moreover, attention to ESG factors is considered to indicate a high awareness of the company’s external operating environment and the ability to identify new risk factors, a generally increased ability to manage risk, and an ability to turn risk into opportunity.

The BSR report shows that while interest in ESG integration is increasing, only a handful of investors actually integrate good corporate governance, environmental and social factors in all parts of their asset management. Most investors use specialised products and services by, for example, investing in specific funds or indeces that use ESG data as a supplement to traditional financial analysis.

The BSR report points out several changes that are necessary for obtaining a higher degree of integration of ESG considerations. Companies need to communicate actively and regularly about factors that may have particular significance for the company"s long-term value creation, including ESG issues. Investors must in turn request ESG data and develop analyses and methods to take these types of factors into greater consideration. Focus on returns in the short term is deemed an obstacle to the appreciation of ESG factors, which normally will have effects on value performance in the longer term. Even institutional investors, who constitute an important group among the investors who have requested information on ESG factors, by and large measure and reward their managers based on short time horizons. This can be an obstacle to real integration of ESG considerations.

It is still challenging to obtain information on ESG factors from the companies. Nevertheless, positive change is evident through initiatives that contribute to more uniform reporting on these topics. Examples of this are the Global Reporting Initiative (GRI) and Carbon Disclosure Project (CDP). 14 Several of the findings from the BSR are also supported by the consulting firm Mercer in a report from November 2009. 15 The report compares 16 academic studies done after 2007 that look at the relationship between ESG factors and company and portfolio performance. 10 of the 16 studies in the Mercer report shows a positive relationship between ESG factors and financial performance, two studies show a negative-neutral relationship, and four studies reported a neutral association.

Responsible investment practice is a broad topic that covers a fairly diverse set of tools: voting, engagement with companies, investor collaboration, negative and positive screening, as well as ESG integration into valuation metrics. The breadth of instruments shows that it will be too easy to claim that incorporating ESG factors in investment activity will automatically limit the investment universe, and for that reason will always lead to lower returns. What can rather be assumed, as Mercer does in a report to the Ministry of Finance of December 2009, 16 is that expectations of a clear ownership strategy on these issues will only increase. This will apply regardless of whether the fund assets are managed actively or passively.

«This is a dynamic area with increasing pressure on investors to play a more active role in capital markets. Such an environment is likely to continue to influence investor behaviour and practice for both funds and fund managers alike. If pressures from fund members and beneficiaries, regulators and industry norms continue to be put on funds to take on the role of active owners, we believe that – eventually – it will become harder for investors to justify the absence of a thoughtful and well-defined active ownership strategy, backed up by the resources to ensure its effective implementation, and with the transparency to communicate its results.»

As Mercer points out in its report of November 2009, it is not obvious that taking ESG factors into account in investment management will have a specific or uniform impact on portfolio performance. Various elements such as manager skill, investment style (active or passive management) and time period will affect the relationship between ESG factors and financial returns. According to Mercer, this means that an attempt should be made to assess this relationship on more disaggregated levels, where distinctions are made between the sectors and asset classes. A majority of the studies Mercer compared showed that strong corporate governance has a positive impact on financial returns, and that active ownership can increase shareholders" values.

Comparable and robust standards for reporting are considered essential for ensuring that ESG integration becomes widespread in general investment management. A greater degree of transparency and information about ESG data will also make their possible effect on financial returns more visible and verifiable.

The UN Environment Programme – Finance Initiative (UNEP FI) has published two reports on the relationship between responsible investment practices, including integration of ESG considerations, and what Anglo-American law tradition refers to as the «fiduciary duty». 17 The final report was published in July 2009 («Fiduciary II»). The issue looked at in these reports is whether there can be said to be any conflict between safeguarding ESG considerations and sound investment management, or whether fund managers can, should or must incorporate ESG considerations in order to safeguard shareholders" interests in a satisfactory manner. The first report gave the following answer:

«…integrating ESG considerations into an investment analysis so as to more reliably predict financial performance is clearly permissible and is arguably required in all jurisdictions.»

The Fiduciary II report is based on the premises of the first report. Fiduciary II gives advice on how the obligation to integrate ESG considerations should be met. Among other things it claims:

«It is an obligation on pension fund trustees not simply a right or option to state in their Statement of Investment Principles what the fund"s guidelines are on responsible investment and to what extent social, environmental or ethical considerations are taken into account.»

Furthermore:

«…it is necessary for investment management agreements or the equivalent contract between pension funds and asset managers to use ESG language in order to clarify the expectations of the parties to the contract. In particular, it is important that it is made absolutely clear to beneficiaries, pension fund trustees and asset managers that ESG is regarded as a mainstream investment consideration.»

10.3 Contribution to the development of best practice in this area

The Ministry of Finance believes it is important to support initiatives that can help develop a clearer common understanding of what constitutes best practice for responsible asset management. At the same time it is necessary to provide a certain amount of flexibility depending, for example, on ownership, size and structure. The UN PRI (Principles for Responsible Investment) is an important initiative that in the Ministry"s view combines the need for a common platform and understanding of the issues and the need for a certain amount of flexibility in execution on the part of the individual investor. As manager of the GPFG Norges Bank participated together with other leading investors in the preparatory work and design of the UN PRI. The Ministry of Finance, Norges Bank and Folketrygdfondet have signed the principles on behalf of the GPFG and GPFN. The PRI is discussed further in section 3.3 of this report.

Good access to information about actual conditions and the companies" strategies for corporate governance and environmental and social issues is necessary in order to be able to take such factors into account in investment decisions and in the exercise of ownership rights. Transparency can help counteract harmful practices, such as corruption. As manager of the GPFG Norges Bank supports initiatives that contribute to a greater degree of transparency and reporting of factors assumed to affect long-term returns. This applies to initiatives such as the Extractive Industries Transparency Initiative (EITI), Carbon Disclosure Project (CDP) and Forest Footprint Disclosure Project. Norges Bank has a leading role in the development of CDP Water Disclosure, which was launched in November 2009.

As much as investors expect companies to be open about their strategies and their value-creation to the investor, investors must also be open about their strategies and their management – both to their owners, but also to companies and society at large. The Ministry of Finance makes an annual report to the Storting on the management of the Government Pension Fund, which includes reporting on the Ministry"s responsible investment practices. In March 2010 the Ministry published a strategy document showing the breadth of its work on responsible investment, the means it uses and its future goals. It is the Ministry"s ambition to continue to develop this strategy in order to incorporate good corporate governance and environmental and social factors into more aspects of the investment strategy and management. This is in line with our obligations as a member of the UN PRI. Norges Bank publishes quarterly as well as annual reports on the management of the Fund, including active ownership. There is also transparency in investments in the Fund and on all voting at annual general meetings. Recommendations from the Council on Ethics to exclude companies from the GPFG are made public. A list of the companies that have been excluded or are under observation is available on the Ministry of Finance"s website.

In managing the GPFG, given the Fund"s size and long-term horizon, it is important to help ensure that markets are efficient and well-functioning. This is a special area of focus in Norges Bank"s ownership strategy. Norges Bank in 2009 led an initiative to strengthen parts of the European covered bonds market, which was hit hard by the financial crisis. Together with a number of other major investors, Norges Bank established the Covered Bond Investor Council (CBIC) to improve liquidity and information in the market for covered bonds, after the financial crisis had undermined investor confidence in these kinds of securities. A well-functioning covered bond market is crucial for banks" long-term financing ability, in addition to financing of mortgages and the public sector in Europe. 18

The Ministry of Finance considers it particularly important in a long-term perspective to support the building of standards that can help to clarify and raise corporate standards in issues that affect our values and our long-term financial returns. This is also connected with our role as a universal owner, with a need to improve the quality of the investment universe and contribute to sustainable development. It will be in our interest as a long-term and diversified owner that as many companies as possible are well managed and safeguard environmental and social factors in a responsible, sustainable manner. The most appropriate way to ensure such a development will in many cases be public regulation. In light of this the Ministry gives priority to contributing to initiatives that set standards and rules for good corporate practice in different areas.

The Ministry of Finance and the Council on Ethics for the GPFG take part in a project coordinated by the UN Global Compact, where the goal is to develop a set of guidelines that provide guidance for responsible corporate and investment practice in conflict areas. Such guidelines are hoped to provide investors and companies a greater degree of insight into each other"s experiences and perspectives, contribute to better and more efficient use of suitable tools when companies operate in such areas, as well as raise awareness and clarity about what is acceptable, responsible behaviour. The guidelines will emphasise that it is important for a company"s profitability and implicitly an investor"s long-term return how the company handles the type of risk that operations in war and conflict areas often entail. The Ministry takes a positive view of the fact that the guidelines are being prepared in consultation with international representatives of companies, investors and organisations. This increases the likelihood of the guidelines being accorded weight and adhered to.

The guidelines are scheduled for release in summer 2010.

11 The Fund as a universal owner

11.1 Introduction to universal ownership

A universal owner (UO) is defined as an owner with investments spread across a large number of companies in many industries and countries. In this way a UO indirectly owns a share of the world"s production capacity. A UO will also often have a very long investment horizon. Because of its size, long-term horizon and broad social basis the GPFG falls under the UO category.

For a UO, overall economic performance over the long term determines the value of the assets to a greater extent than factors related to individual companies or sectors. This is in contrast to less diversified owners, who may have large holdings in a few companies, or who have specialised in one sector. Gjessing and Syse (2007) 19 look specifically at the GPFG and the UO hypothesis and write that «[..] the Fund"s long-term financial interest lies in the global markets" ability to produce economic growth, and in functioning securities markets.» A universal owner therefore has a special interest in working for well-functioning, legitimate markets and reduced negative effects on the environment and society.

Universal owners have for this reason developed increasingly comprehensive strategies relating to the practical implementation of sustainable and responsible investments, and integration of so-called ESG factors 20 in the asset management.

11.2 What are externalities and how can they affect a universal owner?

The externality concept is a good starting point for better understanding why and how a broad and universal owner should focus on the long-term sustainability of the economy. In the economic literature externalities are defined as unpriced social costs or benefits from production or consumption. That a social cost is not priced means that the cost of production imposed on society is higher than the cost the producer pays. The externalities thus lead to market failure, and different consumption of a resource than if the full social cost was priced.

An example of a negative externality is child labour. The producer pays only what it costs to employ the children, not the cost to society when the health of children suffers or they have less time for education. Externalities lead to efficiency losses and lower economic growth than if they did not exist. For a UO the effects of the externalities, however, are often internalised in both the short and long term. Lower production costs in a company that uses child labour can reduce the profitability of other companies that do not use this type of labour. A lower education level today may lead to lower economic growth in the future.

A UO can benefit from reduced externalities in several ways. 21 One effect of more efficient use of resources is increased productivity. This provides increased value creation that a UO can take part in. Lower risk of future uncertainties in the economy as a result of externalities (such as major climate change) may reduce the risk associated with different investments. This provides lower capital costs and more viable investments.

11.3 Investment horizon and other aspects of universal ownership

Challenges associated with externalities may also have a time dimension. Completely efficient markets will immediately price both the negative and the positive effects of an externality. It is the market"s ability to correctly predict the future income stream that is crucial for their ability to price externalities properly. Mispricing can occur as a result of markets failing to account for the probability of various future events (especially extreme ones). Unexpected, negative events, especially recurring unexpected negative events such as climate change, may also increase uncertainty and reduce the value of a financial investment. The Ministry of Finance participates in a project that looks at how climate change may affect different asset classes, see discussion in Box 2.1.

The UO hypothesis"s implicit long-term perspective suggests that there is a limit to how widely the hypothesis will actually apply to the players in the financial markets. It has been claimed that many companies are short-sighted because there are strong forces and dynamics (for example bonus programs and other incentives) that reward short-term objectives at the expense of long-term return the way the markets are organised today. 22 Since owners often will benefit from a long-term horizon, this is an example of the principal-agent problem in economic theory. A UO should endeavour to reward companies on the basis of what creates the highest possible long-term profitability.

11.4 The Fund"s strategy as a universal owner

The GPFG is managed through investments in bond and equity markets, within the limits specified for risk and the ethical guidelines. The return of the Fund is heavily influenced by the performance of the global equity markets. The factors discussed so far provide a good reason for the Fund to engage in active ownership. The possibilities for mispricing and a different investment horizon mean that a UO may have different interests to safeguard than other investors.

Gjessing and Syse (2007) write that a UO has a vested interest in trying to «lift» the quality of the whole market and reduce future welfare losses. Since lifting the entire market will often be practically impossible, a UO may also have to make specific investment strategy choices in order to «adapt» to the market"s inadequate pricing of the externalities or because of a different investment horizon.

It is important to achieve an efficient division of labour between companies, authorities and owners in the approach to these challenges. While local problems 23 may best be resolved through active ownership and cooperation with other major owners, the global problems are more difficult to reduce.

An important prerequisite for influencing companies to change their behaviour is that such a change is also in the companies" interest, if not the results may soon become arbitrary. Where it is difficult to find a solution in isolation at the company level, a broader industry approach may be relevant. An example of successful ownership work in this context is the GPFG"s initiative in India which contributed to a new industry standard for combating child labour.

The most appropriate form of sustainable, long-term and predictable solutions to global problems will often be through regulation. In this case, the GPFG will primarily be interested in influencing global authorities in the direction of integrating the external effects with the economy, either directly or in partnership with portfolio companies and other investors. Work on climate change or regulation of the financial markets so that risk-taking is more in line with long-term interests are good examples of issues where global solutions are most appropriate.

Also when it comes to the need for «adapting» management strategy and asset allocation it is the major, global externalities that can be expected to have a measurable effect. Adaptation could be achieved through greater use of screening criteria, for example on carbon emissions if one has good data. The integration of ESG factors in the asset management, to which the GPFG has committed itself by signing the UN Principles for Responsible Investment, is a relevant approach. There are also management strategies that actively analyse ESG factors to achieve excess return. 24 These active managers seek to exploit the markets’ potential mispricing of externalities to achieve excess returns.

To provide further practical content to the UO-hypothesis is a topic that concerns many funds the GPFG can be compared with. Scenario thinking and a focus on the major global externalities is a useful starting point in the analysis of how universal ownership can be implemented in practice. The Ministry of Finance is, as mentioned above,participating in a project that looks at hese issues in terms of climate change. Several other large investors also look at how various externalities will affect investment management in various scenarios.

11.5 Some limitations to the hypothesis of universal ownership

There are some limitations associated with the UO hypothesis, especially how the hypothesis should be approached in practice. Even a diversified global owner is not invested in the entire world economy. Some local externalities will therefore be less important 25 in a UO context. It might also be sensible to consider the following issues:

Long-term horizon

A universal owner will normally have a long investment horizon. 26 If the rest of the market does not have the same long-term horizon, a UO that adapts to a long-term horizon may perhaps live through a long period of mispricing before the effect of an externality is reflected in the market. It is therefore difficult to say when a UO best adapts to a future problem area. Relative performance during an adjustment will depend on whether it takes a long time before an externality is priced correctly.

Measurement and practical approach

Lack of measurability 27 can also make it problematic in practice to set good goals for how to «lift» the quality or «adapt» the portfolio, especially for a substantial passive manager. Good company or sector-specific data an owner can take advantage of does not exist for many externalities. An owner can try to influence this. The Fund"s support of CDP in terms of good data on CO₂ emissions is one example.

Nor is it necessarily conducive to encourage company management to follow targets with a high degree of inaccuracy, or that an investor requires companies to maximise shareholder value while simultaneously refraining from making use of profit opportunities. However, it may be important for an owner to seek to help a portfolio company to credibly build short-term strategies that support a long-term goal of maximum return if the company is not able on its own to make short-term (perhaps costly) choices that support the long-term goal. 28 A good example of how this can be done is NBIM"s various expectations documents.

Cost efficiency

It is important that thorough cost-benefit assessments be carried out on various actions related to a broad UO approach. The fact that an owner has an economic interest in reducing global market failure does not mean that it necessarily has a good business rationale for investing in this happening (Gjessing and Syse, 2007). Selective ownership initiatives and cooperation on limited issues through international meeting places like the UN PRI is probably a cost-efficient and good way to approach many challenges.

Ethics beyond the financially founded

The UO hypothesis does not address all ethical considerations an owner may wish to take. The UO hypothesis will in practice be based on forecasts and partly on qualitative assessments and therefore does not lead to any direct or indisputable financially rooted ethics. There will be situations where an owner may wish to go further in the ethical direction than the UO hypothesis might provide a basis for – either by conviction or because the analysis and basic data are missing.

11.6 Conclusion

The UO hypothesis is relatively new and although it has been the subject of academic interest, it is, at least in practice, quite unexplored. However, many of the issues are not new, but familiar from, for example environmental and resource management, the debate about the short-term horizon in business and stakeholder theory. What the UO hypothesis does is to anchor the issues theoretically and intuitively in the context of a universal owner in a financial market context. The hypothesis shows that the issues are relevant to the goal of maximum long-term risk-adjusted returns.

The Fund is a universal owner by definition and should therefore have a concrete approach to what this means in practice. Such an approach should look at the need and possibilities for reducing the short- and long-term welfare losses by «lifting» the quality of the investment universe. It should also look at the dynamic need to «adapt» to the issues through changes in the investment strategy.

The Ministry of Finance will continue the analysis into what the role of universal owner practically implies. It is important to point out that many of the challenges and opportunities the GPFG meets in this context are addressed through the Fund"s work as a responsible investor. For example, the GPFG already has a broad approach to exercising ownership and produces expectations documents aimed at portfolio companies. The new observation list for exclusion may also provide an important contribution to improving the «quality» of individual companies in the portfolio. As such, the theory of universal ownership can be seen to form a superstructure for much of the Fund"s work on responsible investments, and help focus and clarify priorities both in this work and in the main objective of maximum, long-term risk-adjusted returns.

12 International reporting standards

12.1 Global Investment Performance Standards (GIPS)

GIPS is an international standard for calculating and presenting historical investment results for companies that manage funds on behalf of a third party. Complying with the standard is voluntary, and it is based on the principle of self-regulation.

The standard was developed by the Chartered Financial Analyst (CFA) Institute in collaboration with many other organisations, and was launched in 1999. At the moment institutions in 32 different countries are administering and continuing to develop the standard. The standard is designed to instil greater confidence in the completeness and correctness of performance information. Common use of methods and principles also makes it easier to compare figures from different institutions. This benefits both managers and investors.

The standard requires the basic data to be consistent, both across portfolios and over time. This is essential if the return figures are to be complete and give an accurate picture. Valuation must be based on market values (fair value). The portfolios must be valued at least monthly.

In calculating returns, GIPS requires the use of total return, including realised and unrealised gains plus other income. Furthermore, time-weighted rates of return that adjust for cash flows must be used. Periodic returns must be geometrically linked, i.e. the return figures for the interim periods must be linked together to find the return for the entire period.

All returns shall be calculated after deduction for actual transaction expenses. The return shall not be adjusted for estimated transaction costs. GIPS covers both gross return and net return, and recommends in presentations the use of gross return (in addition to relevant management fees for current clients), as shown in Figure 12.1.

Source GIPS

Reporting requires, among other things, disclosure of annual return, benchmark return where such exists, market value, overview and description of portfolio groups, the currency the returns are presented in and whether the net or gross return is presented. At least five years of GIPS-compliant annual return data, or since firm inception, are required.

The total return for the benchmark must be presented for each annual period. In addition, it is recommended that the cumulative return for both the actual portfolio and the benchmark is stated for all periods for which the returns are presented.

GIPS has special requirements for calculating returns on and reporting of investments in real estate and unlisted equities.

The standard requires that a manager must meet all requirements listed in GIPS to be able to claim that it is GIPS compliant. Both Norges Bank and Folketrygdfondet follow the GIPS standard.

12.2 International Financial Reporting Standards (IFRS)

IFRS includes principle-based accounting standards, interpretations and frameworks for financial reporting prepared by the International Accounting Standards Board (IASB).

Some of the standards contained in IFRS are known under the old term International Accounting Standards (IAS). IAS were issued between 1973 and 2001 by the Board of the International Accounting Standards Committee (IASC). The responsibility for issuing international accounting standards was transferred from the IASC to the newly established IASB on 1 April 2001. At its first meeting the IASB confirmed that previously released IAS would be part of IFRS.

IASB is an independent standard-setting body overseen by the IASC Foundation. The members (currently 15 full-time members) are responsible for developing and issuing IFRS and approving interpretations prepared by IFRIC (International Financial Reporting Interpretations Committee).

IFRS is under continuous development and significant changes are expected in coming years. Among other things, a major change regarding financial instruments is under preparation. The framework is also being revised.

In addition to the standards, interpretations are issued to provide guidelines for problems not covered in the standards and in areas where dissimilar or unintentional practices have evolved. These are issued by IFRIC, the International Financial Reporting Interpretations Committee.

IFRS is required or permitted in more than 100 countries and has helped harmonise the globalisation of accounting standards, including through converging projects with FASB, the American standard-setting body and the Japanese standard-setting body. Pursuant to Section 3-9 of the Accounting Act, listed companies in Norway shall prepare financial statements according to international accounting standards (IFRS), while others with a statutory obligation to keep accounts can choose to follow IFRS or the other rules of the Accounting Act. For Norwegian banks, financing and insurance companies there are separate accounting rules in the regulations pursuant to the Accounting Act which in essence mean that these companies must develop financial statements according to the recognition and measurement provisions of IFRS.

Reference is otherwise made to the coverage of new rules for the management of the GPFG in Chapter 4.

13 GPFG compared with other large funds

13.1 Introduction

Globally, there are many large capital owners, funds and managers. In the 12 years since its foundation, the GPFG has grown to become one of the world"s largest sovereign wealth funds, and is also among the world"s largest funds in total. Figure 13.1 shows the size of the GPFG compared with a selection of other large pension and investment funds and investment managers.

¹ Different periods of financial reporting.

Source Fund and manager websites

The Government has high ambitions for the management of the Government Pension Fund. This makes it natural to compare the management of the GPFG with other large funds internationally. The Ministry of Finance therefore receives annual reports prepared by CEM Benchmarking Inc. (CEM) and WM Performance Services (WM). The CEM report compares the GPFG"s asset allocation, return, excess return and management costs with large pension funds, while the WM report compares excess return of pension funds, life insurance companies and sovereign wealth funds.

The reference group consists of the largest funds in the CEM survey (10 US, 3 Canadian and 3 European funds). In 2008, the median size of the reference group was NOK 660 billion, while the GPFG"s average market value in 2008 was NOK 2,063 billion. The data are based on self-reporting by the funds that purchase services from CEM. The WM report shows the results of all the funds in their data and for the 15 and 50 largest funds. In 2009 the WM funds had an average market capitalisation of NOK 18 billion while the 15 largest funds" average market capitalisation was NOK 410 billion. For comparison, the GPFG"s market value, calculated as the monthly average, was NOK 2,365 billion in 2009.

WM"s data are different from CEM"s. CEM has a larger database of funds from the entire world, but does not cover funds in the United Kingdom. The United Kingdom makes up a large portion of the funds in WM"s data.

The CEM report is completed in the second half of each year. The last published report therefore contains only data up to and including 2008. The WM report is completed in the first quarter, so that the final report contains data for 2009. The reports are published on the Ministry of Finance"s website (www.regjeringen.no/spf).

13.2 Strategic asset allocation

One difference between the GPFG and other institutional investors such as pension funds and life insurance companies is that the GPFG does not have specific obligations or short-term liquidity requirements that directly influence strategic asset allocation. The GPFG can on this basis be said to have a very long investment horizon and a higher risk-bearing capacity than other funds it would otherwise be natural to compare the fund with.

It may therefore also be relevant to compare the GPFG with funds without clearly defined liabilities, such as North American university endowment funds.

The choice of equity allocation is the decision that has the greatest impact on the funds" returns and risk. The comparison with other funds in Table 13.1 shows that the GPFG"s division between fixed-income and equity instruments is close to the average of the reference group – large pension funds. The university funds stand out with a much lower bond allocation.

The GPFG is different from other funds in that the strategic benchmark does not include smaller asset classes like private equity, infrastructure and natural resources. University funds have significant investments in such alternative asset classes. Such a strategy is feasible amongst other things because they are much smaller than the GPFG and the world"s largest pension funds. However, some of the university funds are large. The size of the two largest university funds, Harvard and Yale, were USD 37 and 23 billion, respectively, by the end of the first half of 2008. Both of these funds are often cited as examples of university funds that have developed a distinctive investment strategy with a high percentage of illiquid equity investments.

13.3 The benchmark"s rate of return up until 2008

The CEM analyses are conducted for a five year period. The investment strategy underpinning the composition of the benchmark is based on trade-offs between long-term return expectations and risk in the capital markets. In such a perspective a five-year period is short. If the comparisons had been made for a different five-year period, the findings might have been different.

The analyses by CEM show that the average annual rate of return of the GPFG"s benchmark for the five-year period until 2008 was 1.5 per cent, measured by the currency basket of the benchmark. Similarly, the reference group"s median return was 2.7 per cent measured in each fund"s respective currency.

Comparison of the aggregate rate of return between funds is difficult because different funds have different currency compositions and benchmark currencies. Exchange rate fluctuations mean that the rate of return will depend on the benchmark currency that is used. During the five-year period until 2008, for example, the rate of return on the Fund"s benchmark measured in euro was 0 per cent, while it was 2.0 per cent measured in USD.

The difference in the benchmark"s rate of return will also be the result of differences in terms of asset classes and regional composition. Up until now the GPFG has distinguished itself from other large pension funds by a higher fixed-income allocation and the fact that the Fund is not invested in real estate and private equity. CEM has calculated that if the other funds had held the same asset class composition as the GPFG over the five-year period until 2008, their annual indexed rate of return would have been reduced by 0.5 percentage point for the reference group. This difference is attributable primarily to the fact that other funds have had higher allocations to private equity and real estate. The GPFG is also different from other funds because of its diversification over many markets and currencies. Most of the pension funds in the reference groups hold the majority of their investments in their domestic markets.

13.4 Excess return up until 2008

A comparison over time of the actual return of the GPFG with the return of the benchmark shows the gross excess return Norges Bank has created. Figure 13.2 displays the average gross excess return and the realised relative volatility for the five-year period 2004-2008 for the GPFG and the various reference groups. The figure shows that the GPFG had a negative excess return over this period, while the average of other funds is close to zero. One can also see that the GPFG has had a somewhat higher variation in excess returns than the other funds.

Source CEM Benchmarking Inc.

CEM has estimated that in the five-year period to 2008 the GPFG achieved an average annual gross negative excess return of 0.6 percentage point. By comparison, the most typical gross excess return (median) was 0.2 percentage point in the reference group. The analyses also show that the GPFG has had a somewhat higher realised relative volatility than the other funds.

The reference group has created excess returns in asset classes in which the GPFG is not invested. The comparison of the excess return from fixed-income and equity management with other funds provides, therefore, a better illustration as to how Norges Bank has succeeded compared to other managers. The average gross excess return from equity management over the five-year period was 0.5 percentage point, while the median for the reference group was 0.4 percentage point. The average gross negative excess return from fixed-income management over the period was -1.4 percentage points, while the median rate of return for the reference group was -0.9 percentage point. The analyses also show that variation in excess returns for both equity and fixed income management in the GPFG in the five-year period until 2008 is higher than what was typical of other funds.

13.5 Excess return up until 2009

Figure 13.3 shows the average gross excess return and the realised relative volatility of the last three years to 2009 for the GPFG compared with the WM sample. The figure shows that the GPFG, like the majority of the other managers, had a negative excess return during this period. The GPFG has had a somewhat smaller variation in excess returns, i.e. lower relative volatility than the other funds.

Source WM Performance Services.

The WM analyses show that the high excess returns in 2009 placed the GPFG among the 25 per cent best funds. The significant negative excess return in 2008 means, however, that the excess return in the three-and five-year period up to 2009 approaches the most typical excess return (the median) of the other funds in the survey.

The report from WM also contains a comparison of the excess return of the equity and fixed-income portfolios with other funds. The analyses show that although equity management had significantly lower negative excess return in 2008, the equity portfolio created excess return for both the three-and five-year period to the end of 2009. In both these periods, equity management created higher excess return than the average of other funds. Excess return in the five-year period placed the GPFG among the 25 per cent best funds. In fixed income management, as in equity management, significant excess return were created in 2009. The results were also good compared with other funds. However, for the three-and five-year period to the end of 2009, fixed-income management had a negative excess return, and did somewhat worse than the other funds in WM"s source data.

13.6 Management costs up until 2008

The CEM report shows that Norges Bank"s total management costs for the GPFG in 2008 were 0.106 per cent (i.e. about 11 basis points) of average assets under management. With the exception of the GPFN, these represent the lowest management costs of all the funds from which CEM collects data. Since the total management costs largely reflect the funds" asset composition, such comparisons do not provide an adequate picture of whether Norges Bank"s management is cost-efficient. For the GPFG the asset class composition is a result of the Ministry of Finance"s investment strategy.

CEM has, therefore, prepared a cost benchmark based on the asset structure of the GPFG. The cost benchmark indicates what costs the reference group – the world"s largest pension funds – would have incurred with the same asset structure as the Fund. The analysis shows that actual management costs in the GPFG in 2008 were 0.03 percentage point lower than the cost benchmark. This is primarily due to Norges Bank having chosen more internal management than the reference group. Internal management is considerably less expensive than external management. In addition, Norges Bank has paid less for external management than comparable funds.

13.7 Transparency of management

The Ministry of Finance emphasises transparency and public access in its work with the Government Pension Fund. This is important for ensuring the credibility of the Fund and the Fund"s general approval.

Internationally, there has been considerable attention related to the fact that some sovereign wealth funds have little transparency about their activities and may have non-financial objectives for their investments. In particular, the recipient countries of foreign, state investments have in different forums pointed out the importance of transparency with respect to such investments.

Both the U.S. and the EU Commission outlined principles for adequate transparency in 2008. The European Commission called for annual disclosure of investments and asset allocation, disclosure of level of leverage, currency distribution of assets, the size of the fund and the source of the fund"s assets and the home country"s regulation and control of the fund (European Commission, 2008). 29 Reporting on the management of the GPFG is in line with the principles outlined by the U.S. and the EU.

Several independent sources refer to the GPFG as one of the world"s most transparent funds. One example is the article «A Blueprint for Sovereign Wealth Funds Best Practices» by Truman (2008) 30 which gives the Fund 100 of 100 possible points for transparency. The report grades the degree of transparency in the investment strategy, investment activities, reporting and auditing. Of 34 state-owned investment funds, only the Alaska Permanent Fund in the United States and the GPFG achieve the highest score. The average score in the group is 44.

The Sovereign Wealth Fund Institute (SWFI) also gives the GPFG a score of 10 out of 10 on transparency. SWFI believes that a fund should have a score of at least 8 in order to call itself transparent. Figure 13.4 shows the funds that get over 6 points from the SWFI. There are only 8 global sovereign wealth funds that have 10 points in the SWFI ranking. The items included in the review are listed in Box 13.1.

Source www.swfinstitute.org

13.8 The organisation of large government investment and pension funds

There are various models for how sovereign wealth funds and pension funds are organised, and the degree of independence of the state owners varies.

IMF (2008) 31 writes that there is no clear evidence that the investments of any sovereign wealth funds have been clearly politically motivated. A review of sovereign wealth funds by Balding (2009) 32 shows that it looks like they are governed primarily by financial objectives. Another study by Truman (2008) finds, however, that the governance model is poorly defined in many state-owned investment funds and that the manager role is also unclear or undefined in 12 of 34 cases. Truman"s review also shows that investment decisions in half the funds are not independent of state or state-controlled boards. Such a practice is in conflict with the Santiago principles in general, and the principle of operational independence in investment decisions in particular. In its report on the Santiago principles the International Working Group of Sovereign Wealth Funds (IWF) describes a governance structure that divides the ownership, governance and management functions, and where the management function is politically independent. The Santiago principles express good principles of management for sovereign wealth funds. The principles are minimum standards that the current GPFG framework already satisfies.

14 Systematic risk factors in the active management of the GPFG 1998–2009

14.1 Introduction

The GPFG is managed by Norges Bank in accordance with guidelines set by the Ministry of Finance. The Ministry of Finance has provided Norges Bank with a benchmark consisting of 60 per cent shares and 40 per cent bonds. In addition, the benchmark is divided into regions, countries, sectors and companies. The Ministry has also set limits for how much Norges Bank can deviate from the benchmark in the actual investments in the Fund. Divergence of this nature is called active management.

A limit has been set for how much the difference in return between the actual portfolio and the benchmark is expected to vary (so-called expected trackig error). Under certain statistical assumptions, and provided Norges Bank fully utilises the limit, the current limit for expected relative volatility of 1.5 per cent means that the difference in the return between the actual portfolio and the benchmark is expected to be less than 1.5 percentage points in two out of three years.

The three professors Andrew Ang from Columbia Business School, William N. Goetzmann from Yale School of Management and Stephen M. Schaefer from London Business School have analysed the results achieved by the active management of the GPFG in the period January 1998 to September 2009. 33 This special topic article presents the main findings of this study.

14.2 Degree of active management

The analysis by professors Ang, Goetzmann and Schaefer reveals that only 0.9 per cent of the fluctuations in the return on the actual portfolio, measured in variance, were caused by divergences made in the active management. The remaining 99.1 per cent are the result of fluctuations in the return on the benchmark. In the period prior to 2008, 0.3 per cent of the fluctuations were the result of active management. This means that it is mainly the choice of benchmark that determines the level of risk in the actual portfolio.

Only 0.3 per cent of the fluctuations in the equity portfolio were the result of active management. This applies to the period up until the end of 2007 as well as the entire period from January 1998 to September 2009. For the fixed-income portfolio, 0.2 per cent of the fluctuations up until 2007 were due to active management. Looking at the entire period up until September 2009, 2.9 per cent of the fluctuations can be attributed to active management. This increase in the importance of active management in the fixed-income portfolio suggests that the active management of the fixed-income portfolio has entailed exposure to systematic factors that generally made only a minor contribution to the fluctuations, but entailed major fluctuations for brief intervals.

14.3 Risk factors

Studies on historical returns have shown that it is possible to improve the results achieved by a portfolio relative to a benchmark over time through exposure to systematic risk factors. Sometimes exposure to these factors will yield gains; at other times it will entail losses. Over time, however, the gains will be greater than the losses.

Ang, Goetzmann and Schaefer have analysed the excess return of the GPFG using a set of the most studied and accepted risk factors identified by research. The factors that have been investigated can be categorised as factors that affect the return on fixed-income securities and factors that affect the return on shares. The fixed-income factors are term (TERM), credit (CREDITAa, CREDITBaa, CREDITHY), currency carry trades (FXCARRY) and liquidity (LIQUIDITY), while the share factors are market value compared with book value (VALGRTH), size (SMLG), momentum (MOM) and volatility (VOL). See chapter 7 on risk factors for a more detailed description of the factors.

Source Ang, Goetzmann og Schaefer (2009)

Figure 14.1 shows accumulated return of fixed-income factors in the period January 1998 to September 2009. Correspondingly, figure 14.2 shows accumulated return of share factors, and figure 14.3 shows the developments in the liquidity factor. 34 The volatility factor is used in both the analysis of the equity portfolio and the analysis of the fixed-income portfolio.

Source Ang, Goetzmann og Schaefer (2009)

The analysis of how exposure to the different risk factors has affected the Fund’s excess return was performed using regression analysis. This is described in more details in the report by Ang, Goetzmann and Schaefer.

Source Ang, Goetzmann og Schaefer (2009)

14.4 Total portfolio

Figure 3.14 C in chapter 3 shows developments in accumulated excess return from January 1998 to December 2009. Up until autumn 2007, there was relatively steady positive growth. Autumn 2007 saw the beginning of a negative development that resulted in a steep decline in autumn 2008. The return on the Fund in 2008 was 3.4 percentage points poorer than the return on the benchmark. From the second quarter of 2009, the trend has been reversed, with the Fund performing much better than the benchmark.

For the period as a whole, Ang, Goetzmann and Schaefer find that active management has yielded a return that was on average 0.02 percentage points better than the return on the benchmark per month. Statistically, this difference is too small and the period too short to be able to conclude whether the excess return is due to skilful management or merely random. Ang, Goetzmann and Schaefer therefore conclude that the overall excess return is marginal, but positive.

Ang, Goetzmann and Schaefer point out that the active strategies that Norges Bank has adopted basically entail exposure to systematic factors. The analysis shows that the Fund has overweighted fixed-income securities with a credit rating of Aa and underweighted fixed-income securities with a credit rating of Baa. The Fund has also been overweighted in securities with low liquidity, overweighted in shares in companies with a high market value compared with book value (growth companies) and overweighted in shares in small-cap companies. In addition, the Fund has sold volatility, which is equivalent to having sold insurance against greater fluctuations in the market.

By studying the developments in the active management results since 1998, Ang, Goetzmann and Schaefer have calculated that more than two-thirds of the variations in the Fund’s overall excess return can be explained statistically by the risk factors. The equity and fixed-income portfolios were analysed using the factors considered most relevant for each asset class. According to this analysis exposure to systematic risk factors explains scarcely a third of the developments in the results achieved by the equity management and about half of the developments in the results achieved by the fixed-income management.

Ang, Goetzmann and Schaefer also point out that exposure to these systematic factors was not adequately communicated prior to 2008. If the factor exposures had been communicated and the sample space of factor returns had been known the losses in 2008 would not have come as a surprise, given the developments in these factors. In the same way that one would expect the Fund to lose money when the stock markets fall, it would also be natural to expect the Funds to incur losses relative to the benchmark when the fluctuations in the market increased and liquidity was reduced.

14.5 Fixed-income portfolio

In the period January 1998 to September 2009, the average return on the fixed-income portfolio was very similar to the return on the benchmark. In the period prior to 2008, the return on the actual fixed-income portfolio was 0.01 percentage point higher per month than the benchmark, but this figure is too small to statistically conclude that the performance is a result of skilful management.

Using the period from January 1998 to September 2009 to analyse the fixed-income portfolio, Ang, Goetzmann and Schaefer find that 54 per cent of the fluctuations in the excess return are due to developments in systematic factors. Retrospectively it has therefore been possible to identify the exposure to the different factors. Analysing the period prior to 2008, a significantly smaller proportion of the fluctuations in the excess return could be ascribed to developments in the systematic factors. At the beginning of 2008, it would therefore have been difficult to estimate the exposure to the factors. Nevertheless, the analyses by Ang, Goetzmann and Schaefer show that throughout 2008 and 2009 it would have been possible to estimate the various factor exposures by making a new estimate every month, based on developments to date. Assuming this kind of approach and assuming that the sample space for the return attributable to the factors was known, the losses incurred by active management of the fixed-income portfolio in 2008 would not have been completely unexpected according to Ang, Goetzmann and Schaefer.

The analyses indicate that it is exposure to securities with low liquidity and exposure to the volatility factor that were the main causes of the poor results from active management of the fixed-income portfolio in 2008, and the ensuing good results in 2009. The fixed-income portfolio has also had an underweight of bonds with a long term to maturity and an overweight of investment-grade securities relative to securities with a lower credit rating.

14.6 Equity portfolio

In the equity portfolio, active management contributed 0.05 percentage to the return per month in the period as a whole and 0.06 percentage points in the period prior to 2008. The analyses show that statistically the latter figure is significantly different from zero, while the first figure is almost statistically significant.

Some 29 per cent of the fluctuations in the excess return can be ascribed to developments in systematic risk factors. According to Ang, Goetzmann and Schaefer, this shows that the active management of the equity portfolio has been based on exposure to risk factors to a lesser degree than the management of the fixed-income portfolio, but that the exposure to systematic factors is still statistically significant.

The equity portfolio has been overweighted in growth companies and small-cap companies. In addition, the portfolio has sold volatility. In the second half of 2008, this strategy performed very poorly and accounted for much of the losses incurred in the active management of the equity portfolio.

14.7 Internally managed assets

The management of the GPFG is divided into internal and external management. Internal management is carried out by Norges Bank, whereas external management is done by other managers on behalf of Norges Bank. At the end of 2009, 17 per cent of the equity portfolio and 3 per cent of the fixed-income portfolio were being managed externally. The financial crisis entailed a substantial reduction in the proportion of the fixed-income portfolio that is being managed externally.

Internal management consists primarily of strategies which are meant to exploit weaknesses in the benchmark. Company-specific strategies are also used (see section 2.3).

Since a large proportion of the fixed-income portfolio is managed internally, the results achieved by the internal fixed-income management will largely reflect the results for the total management of fixed-income assets. The main difference is that the internally managed fixed-income portfolio has less exposure to the TERM factor.

Ang, Goetzmann and Schaefer find that the results achieved by the internal active management of the equity portfolio are positive both for the period as a whole and for the period prior to 2008, with an excess return of 0.05 percentage points and 0.09 percentage points respectively. The latter figure is statistically significant. The analyses show that the excess return from the internal equity management has covaried significantly with all the share factors. The portfolio has been overweighted in growth companies and small-cap companies, underweighted in companies that have had strong price developments, and has sold volatility. Much of the covariance with the size and momentum factors observed in the internally managed equity portfolio is caused by the company-specific strategies. The excess return attributable to Norges Bank’s strategies for exploiting the weaknesses in the equity benchmark has only had significant covariance with the growth / value factor and volatility.

14.8 Externally managed assets

External managers are mainly used to gain access to expertise that Norges Bank does not possess internally. Norges Bank selects these managers and gives them a mandate with a benchmark. The objective for the managers is to deliver returns that are better than the return on the benchmark within defined risk limits.

The average excess return in the externally managed fixed-income portfolio has been -0.45 percentage points per month. The number of large negative impacts from individual funds is larger than the number of large positive impacts. Typically, the large negative impacts occurred in 2008. Less than a third of the external fixed-income managers have delivered higher returns than the benchmark. The factor analysis shows that the external fixed-income mandates have been underweighted in fixed-income securities with a long term to maturity and have sold volatility.

Active management of the external equity mandates has been more successful and, according to Ang, Goetzmann and Schaefer, has increased the return on the Fund by an average of 0.09 percentage points per month compared with the benchmark. Almost two-thirds of the external equity managers have delivered a positive excess return. Nevertheless, for most of the equity mandates, the monthly fluctuations in excess return are so great that it is difficult to say statistically whether the excess return is due to skilful management or whether it is merely random. According to the analysis, the external equity management has been overweighted in growth companies and small-cap companies, and has pursued a strategy of buying shares whose prices have been rising (momentum), and has sold volatility.

14.9 Summary

The main conclusion of the report by Ang, Goetzmann and Schaefer is that active management has had very little impact on the total return on the Fund over time. Only a small part of the risk associated with the Fund is due to active management, indicating that it is the choice of benchmark that determines the overall level of risk in the Fund.

There are major differences in the results achieved by the fixed-income management and the equity management. The active management of the fixed-income portfolio has yielded close to zero excess return, taken as a whole. Overall, the external fixed-income management has yielded a negative excess return. The management of equity assets has yielded better results with a statistically significant positive excess return in the period up until the end of 2007, and an almost statistically significant positive excess return for the period as a whole.

Since 1998, the return attributable to active management shows exposure to several systematic factors. This applies especially to the fixed-income management, but also to the equity management to a lesser degree. Most of the negative excess return within the active management post-2007 was from the fixed-income management, and in the analysis by Ang, Goetzmann and Schaefer this is explained as an outcome of the high degree of exposure to systematic factors, especially liquidity and volatility.

Ang, Goetzmann and Schaefer believe that if both the level of exposure to the various factors and the entire sample space of factor returns had been known, then the results achieved in 2008 would also have been within the predicted opportunity set of returns.