Recommendations concerning the investment strategy for the Government Pension Gund Global
Norges Bank has reviewed the long-term investment strategy for the Government Pension Fund – Global. The analysis is documented in separate reports which will be made available to the Ministry. The point of departure for the review is that the Fund has become very large and will in all probability continue to grow rapidly. The Fund is intended as a permanent Fund where only the average real return is to be used. This makes for an unusually long investment horizon. For a fund of this kind, it is important that its investments as a whole are safe in the long term, while the liquidity of investments is less important. In reasonably efficient markets, low liquidity will be associated with higher expected returns than can be achieved with more liquid investments. The stated objective of the Fund suggests that we should endeavour to take advantage of this.
The Fund is currently invested exclusively in markets with relatively good liquidity. The asset mix should gradually be shifted in favour of greater emphasis on illiquid investments with a liquidity premium. The size of the markets imposes limits on how quickly it would be appropriate to proceed with this. However, it is reasonable to start moving away from investing exclusively in liquid markets.
The benchmark portfolio currently consists of listed equities in 27 countries and bonds with a high credit rating in 11 currencies. In the following, we look first at the question of whether the investment strategy should include more asset classes, which would mean a greater emphasis on illiquid investments:
- investments in real estate/infrastructure
- investments in private equity
We then look at possible changes in the two asset classes in which the Fund is already invested. The following topics are covered:
- the equity portion in the benchmark portfolio
- the regional weights in the equity and fixed income benchmarks
- investments in small cap equities
- investments in high-yield bonds
The issue of adding new countries to the equity benchmark or new currencies to the fixed income benchmark is not discussed in this letter.
- New asset classes
The Fund’s asset allocation differs considerably from that of large pension funds and other reserve funds, especially in two areas:
- the Fund has a much higher proportion of bonds
- the Fund does not invest in unlisted markets such as real estate, infrastructure and private equity, or in other alternative asset classes
The split between equities and bonds is currently under review by the Ministry. Investments in unlisted markets were discussed most recently in the National Budget for 2004, where the Ministry concluded that the Fund’s investment universe should be limited to listed equities and bonds for the time being.
The markets for unlisted investments have seen clear growth in recent years. As a long-term investor, the Fund is in a better position than many other institutional investors to accept higher levels of illiquidity in parts of its investments, where this results in higher expected returns or broader diversification of the overall portfolio.
Real estate and infrastructure
As a rule, real estate is included as a separate asset class in the portfolios of large pension funds and reserve funds. Each year, the company CEM Benchmarking Inc. performs an analysis of the Pension Fund’s management costs. Its most recent analysis reveals that the average allocation to real estate, excluding listed real estate stocks, in 2005 was 4.5 per cent for the big pension funds in the USA, Canada and the Netherlands – the peer group with which the Fund’s management costs can best be compared. Including listed property stocks, the average allocation was just under 7 per cent for the same group. Europe’s largest pension fund, ABP in the Netherlands, had 11 per cent of its portfolio invested in real estate at the end of 2005, and the largest US pension fund, CalPERS, had 7 per cent.
In principle, real estate exposure can be achieved through both equity and debt instruments. However, when a separate allocation to real estate is being considered, this should be limited to equity instruments. It is in this way that the Fund can realise diversification gains in the portfolio, which are an important argument for a separate allocation.
Although the Fund does not currently have a separate allocation to real estate, the Fund is exposed to the real estate market through investments in listed property stocks included in the Fund’s benchmark portfolio. Equities classed by the index supplier FTSE as property stocks account for just under 2 per cent of the market value of the Fund’s benchmark portfolio for equities. Furthermore, many of the companies in the Fund’s equity portfolio own their production and office facilities, but the significance of this for the equity portfolio’s risk profile is difficult to gauge. The market for listed property stocks is growing, not least now that more and more countries are permitting tax-transparent real estate investment trusts (REITs) aimed at both institutional and private investors. But the unlisted market is far larger than the listed market, and will probably remain so for a long time to come. It would be unrealistic for a fund as large as the Government Pension Fund – Global to build up a substantial real estate allocation over time without investing in the unlisted market. Such investments could be made through real estate funds, through discretionary management mandates, or through joint ventures or other forms of strategic alliance and co-ownership with local participants in the individual markets.
Real estate can be expected to generate an average return somewhere between that on ordinary equities and bonds in the long run. In the three markets with the longest return history for institutional investments in real estate (the USA, the UK and the Netherlands), the real estate market has generated an excess return relative to bonds of between 0.5 and 1.5 percentage points since the mid-1970s. This has been a period of historically high real returns on bonds.
Indices which measure historical returns on unlisted real estate investments show that risk measured as annual volatility is only slightly higher than for bond investments. When these calculations are adjusted for known factors such as artificial stability in the underlying valuations, and for increased risk due to the funds’ high average debt levels, it would be more reasonable to assume that volatility in real estate will also be roughly midway between equities and bonds.
The diversification gains for the Fund from investing in real estate derive from real estate returns being affected by market-specific factors, or from factors that affect returns in all of the capital markets having a different impact on the property market than on the other financial markets. This will result in a low correlation between investments in the real estate markets and those in the other capital markets. If return series based on unlisted real estate indices are used, the historical correlation between real estate and equities/bonds is very low. However, objections can be raised against a comparison of returns in time series based on different valuation principles. If shorter time series for listed property stocks are used, real estate has still offered clear diversification gains relative to ordinary equities and bonds, but not to the same degree as might be suggested by indices for unlisted real estate investments.
Another important reason why funds generally invest in real estate is that such investments are expected to generate a stable real return over time, i.e. provide good protection against unexpected movements in inflation. For funds which have explicit obligations which rise in line with inflation, the combination of inflation protection and higher returns than on inflation-protected bonds will be valuable. This is not as important for the Fund. The degree of inflation protection will vary anyway between markets and projects, and supply/demand patterns peculiar to the property markets could also limit such an effect. No great importance should therefore be attached to inflation protection when considering a separate allocation to real estate.
Management costs will be substantially higher in the real estate market than in the listed markets. CEM’s cost report for 2005 shows that, on average, the Funds in the peer group recorded management costs for unlisted, externally managed real estate investments of just under 100 basis points.
Investments in infrastructure projects, such as electricity and water supply, toll roads, airports and telecommunications, have traditionally been a limited market for institutional investors. However, in recent years, growing private-sector involvement and financing has made these investments interesting for long-term financial investors too. As with investments in traditional real estate, investors will expect a real return somewhere between that on equities and bonds, with much lower volatility than for equities, and with clear diversification gains relative to these asset classes. One clear difference to real estate investments is that the public sector will generally be involved in infrastructure projects as principal, contractual party and/or regulator.
For the foreseeable future, it will not be practicable to achieve exposure to real estate and infrastructure beyond the equivalent of about 10 per cent of the Fund’s overall portfolio. Model analyses based on the assumptions discussed above – and in more detail in the full market analysis – confirm that such an allocation will result in clear diversification gains. If the allocation is matched by a reduced allocation to fixed income investments, the expected return on the overall portfolio will rise, even after taking account of the higher management costs. However, it will take time to build up this level of exposure. This is discussed in the full analysis report.
Norges Bank recommends that, in the light of the probable liquidity premium and the probable diversification gains from investing in real estate and infrastructure, it be made a long-term strategic target for up to 10 per cent of the Government Pension Fund – Global to be invested in this asset class. This portfolio will have to be built up gradually over a period of several years.
Before activities in this area can be started up, a detailed investment mandate specifying required rates of return, risk limits and reporting requirements needs to be formulated.
Private equity is also a common asset class among large institutional investors. Private equity includes investments in both pure equity instruments and hybrid equity instruments such as subordinated, convertible debt. The Fund’s peer group in the CEM report for 2005 had, on average, almost 6 per cent of their assets invested in private equity at the end of the year.
As a rule, investments in private equity are motivated by the goal of achieving higher returns in the longer term than offered by the listed equity market. By actively developing companies in the start-up phase (venture funds) or acquiring and restructuring existing companies (buyout funds), the managing partners in funds which invest in private equity aim to generate a return which is substantially higher than the listed market can be expected to offer in the longer term. This excess return is then divided between the partners and other investors in the funds which are raised for investment purposes. Typically, investors in such funds will demand or expect these investments to yield a return 3-5 percentage points higher than offered by the listed market (net to the investor, i.e. after fees paid to the managing partners in the funds).
However, it is uncertain whether the average investor in this market has actually achieved an excess return relative to listed equities at all. Studies of historical returns in both the USA and Europe have reached differing conclusions on this point. There does seem to be a consensus that the average investor has in any case not achieved a return after costs as high as the return on listed equities plus a margin of 3-5 percentage points. At the same time, the spread of returns has been extremely high. In the USA, return data from analysis firm Cambridge Associates show that, of all the funds that specialised in buyout strategies in the period 1986-2000, managers with returns in the top quartile generated returns 6-9 percentage points higher than that did the median manager. The differences are even larger when it comes to venture funds. There are also clear signs of persistence in these data, i.e. managers who have performed well have a tendency to achieve good returns in subsequent investments or funds.
As is the case with unlisted investments in general, short-term return data for private equity are heavily influenced by the valuations of the underlying investments. Such valuations are always uncertain, and they can be very different from a true market price estimate. This means that calculations of volatility and correlations with other asset classes are also very uncertain. It can in general be assumed that the listed and unlisted equity markets are highly correlated. Diversification gains cannot therefore be a strong argument in favour of investing in private equity.
The main instrument for investing in private equity is investments in funds which then invest directly in the individual companies, and which are actively involved in each individual company in the portfolio. Such a fund will, on average, have an expected life of around ten years, but many of the fund’s investments will be realised long before the fund is closed. Primary investments are by far the most common, i.e. investments which are made with money that the individual fund raises in the market. There is also a smaller market for secondary investments, i.e. purchases of units from other investors in existing funds. Direct investments do happen, but then in the form of co-investments with a private equity fund or an institutional investor with specialist expertise in direct investments.
Management costs can be expected to be much higher for private equity than for listed equities. CEM’s report on management costs in 2005 at the funds defined as the Fund’s peer group shows an average cost of just over 120 basis points of committed capital and just under 240 basis points of invested capital.
In the National Budget for 2004, the Ministry noted that it was uncertain whether investments in private equity would generate an excess return relative to the listed segment. Return data from large investors and from broad return analyses show that investors who manage to identify good managers in advance can achieve an excess return which would be very difficult to realise to the same degree through active management in the listed market. At the same time, these analyses also suggest that the average investor has not achieved this excess return.
Norges Bank recommends that it be made a strategic target for up to 5 per cent of the Government Pension Fund – Global to be invested in private equity, including investments in funds which hold hybrid equity instruments. This portfolio should be built up gradually over a period of several years, and this strategic target should be reviewed after a few years on the basis of an evaluation of the management organisation by then built up.
The main argument in favour of this proposal is that investments in this asset class made by a competent investment organisation can generate a significant excess return. In the continued development of the Fund’s management strategy, importance should be attached to finding new and independent opportunities for generating excess return. Investing in private equity is one such opportunity.
Before such investments can be started up on any scale, an investment mandate needs to be formulated which also addresses the issues of performance expectations, risk management and reporting. However, it may be a good idea to give permission for some instruments bordering on the listed markets relatively quickly, in order both to increase returns and to develop expertise in the management organisation. It is already permitted to retain positions in equities for a while after they have been delisted from an approved marketplace. One relevant extension of this would be to permit the acquisition of stakes in companies which can reasonably be expected to become listed in the foreseeable future.
As a short-term solution, Norges Bank recommends that the investment universe for the Government Pension Fund – Global be extended to permit the ownership of equities which are expected to be listed on an approved exchange in the next 12-24 months.
Way forward for investments in unlisted markets
Norges Bank would like to point out that investments in unlisted markets raise issues of both a technical and organisational nature. In the above, we have recommended on the basis of asset allocation considerations that investments be made in unlisted markets, and we have suggested the long-term allocation which it would be reasonable to aim for.
Investment in private equity will require some adjustment of the Fund’s management structure. It will not be possible or appropriate to define a mandate for investments in unlisted markets in the form of a benchmark portfolio and a limit for tracking error. Nor will it be possible to measure returns continuously based on the market prices of underlying assets, as is the case for bonds and listed equities. It will be possible to estimate returns regularly on the basis of valuations of underlying assets, but these returns will be associated with uncertainty. This uncertainty is something which all investors in this market have to accept, and for which practical adjustments can be made. As a rule, only after several years will there be a good enough basis, in the form of actual cash flows from a concrete investment in unlisted markets, to allow comparison of returns with alternative investments in the same period.
This does not prevent large pension funds, for example, from normally investing a proportion of their portfolio in unlisted markets. Nor should this be a barrier to the Fund investing in these markets. However, as in other funds, there will be a need to establish separate investment mandates specifying what form risk limits, reporting requirements and management follow-up are to take, given the limitations described above.
Norges Bank’s Executive Board is in the process of assessing whether the bank could expand its management activities to include new asset classes. The idea is to be able to give the Ministry a detailed report on how Norges Bank could manage the Fund’s investments in unlisted asset classes.
- Existing asset classes
In its letter of 10 February 2006, Norges Bank recommended that the Ministry increase the equity portion in the benchmark portfolio for the Fund. The updated analysis we have now performed confirms the results underlying that recommendation: a higher equity portion will increase both expected return and return volatility. The trade-off between expected return and volatility appears attractive in the long term. The risk of a negative accumulated real return rises slightly, but it is highly likely that an increased equity portion will be profitable. If we look at a conditional probability distribution, we find that the expected loss, if an increased equity portion turns out not to be profitable, is relatively modest.
Regional weights in the equity and fixed income benchmarks
In its letter of 22 August 2005, Norges Bank recommended that the regional weight for Asia/Oceania be reduced in the fixed income benchmark and increased in the equity benchmark. These changes have now been made.
Based on these new regional weights, we have looked once again at the effects of changes in the regional weights in both the equity and fixed income benchmarks. Our view of the correlation patterns in the markets is not significantly different to before, and we can see no potential changes which would result in clear improvements in the portfolio’s properties.
Shares in companies with a small market capitalisation (small-cap equities) are part of the Fund’s investment universe, but are not included in the Fund’s benchmark portfolio. In its letter of 1 April 2003, Norges Bank recommended including small-cap equities in the benchmark portfolio. The main justification was that this would result in broader representation of the investment universe, and that the Fund’s size and growth warranted the Fund being broadly invested. In the National Budget for 2004, the Ministry decided to stick to a benchmark portfolio consisting of large and medium-sized companies (large/mid-cap equities). The Ministry noted that it was uncertain whether including small-cap equities in the benchmark portfolio would have a measurable positive impact on expected return and risk. The Ministry also deemed it best to wait until the issue of ethical guidelines had been considered before contemplating an increase in the number of stocks in the benchmark portfolio.
The market value of small-cap companies is equivalent to between 11 and 15 per cent of the market value of the large/mid-cap companies currently included in the benchmark portfolio in the three regions of the Americas, Europe and Asia/Oceania. This is the largest single segment of the listed markets not included in the current benchmark portfolio. Table 1 shows the number of companies and market value for the large/mid-cap and small-cap segments in the regional FTSE indices at the end of August 2006 for the countries included in the current benchmark portfolio for the Fund.
The average small-cap company in the Americas and Europe has a market value of USD 1-1.1 billion (approx. NOK 7 billion), while the average in the Asia-Pacific region is much lower at USD 300-400 million. By way of comparison, we can look at the smallest companies already included in the benchmark portfolio: the average value of large/mid-cap companies in New Zealand is USD 1.2 billion. We can also draw comparisons with the most liquid companies traded on the Oslo Stock Exchange (OBX), which had an average market value of around NOK 47 billion (USD 7.1 billion) at the end of August 2006, and with companies with average liquidity on the same exchange (OB Match), which had an average market value of around NOK 2.5 billion (USD 0.4 billion). Thus the companies included in the small-cap segment in the FTSE index are not necessarily small by Norwegian standards.
Table 1: Key figures for FTSE Global Equity Index Series, August 2006.
Time series from the US and UK stock markets show that investments in small-cap equities have generated a slightly higher return than those in large-cap equities, measured over the entire period for which data are available (since 1926 in the USA and since 1955 in the UK). However, the small-cap premium has varied widely and has also been negative for long periods, as can be seen in Table 2.
Table 2: Small-cap premium (annualised) during various periods in the USA, the UK and Japan.
(from 4Q85) 1.6%
Sources: S&P (USA), FTSE (UK), Barra/Nikko (Japan)
In the multi-factor models developed by Eugene Fama and Kenneth French, the size of companies is one of several risk factors used to explain the return on individual equities. According to this theory, shares in small companies are more risky than shares in larger companies, and this is reflected in a higher expected return. A common explanation of why shares in small companies should be more risky than other equities is that, on average, small companies are more likely to run into financial difficulties during periods of slow economic growth. Investors could therefore lose their capital in periods when they have the greatest need for a return (when the markets as a whole are weak). For investors to be willing to hold such equities, they need to be compensated with an extra risk premium. Investors with a lesser need for liquidity in such situations than the average investor, such as the Fund, should therefore be the natural holders of small-cap equities. This explanation is disputed in the academic literature. However, the large (but varying) long-term return differentials between small-cap equities and other equities support the view that the small-cap premium is an independent and priced risk factor in the market.
The return on small-cap equities is positively correlated with the return on other segments of the equity market. The correlation between the small- and large-cap indices underlying the data in Table 2 is in the interval 0.75-0.85 when quarterly data are used, and is still high using non-overlapping annual data (in the interval 0.66-0.78). This suggests that the diversification gains from including small-cap equities in the benchmark index are limited, but still positive. The gains are probably greater with longer time horizons.
Norges Bank believes that the same considerations as were behind its recommendation in 2003 still favour the inclusion of small-cap equities in the benchmark portfolio. Small-cap equities make up a substantial segment of the market. It is difficult to see why the Fund, as a large and long-term investor, should have an exposure to this segment which is substantially lower than that of the market in general. There are also moderate diversification gains to be had from including these equities. If small-cap equities are added to the benchmark portfolio, higher returns can be expected without a significant increase in volatility in the portfolio.
However, a number of objections can be raised against the inclusion of small-cap equities in the benchmark portfolio:
Firstly, transaction costs will be higher in this segment than for other equities, both when establishing the portfolio and when maintaining it. If the entire small-cap portfolio is to be built up over a short period and funded through the sale of large-cap equities, the set-up costs are estimated at USD 153 million. These calculations have been performed using the StockFactsPRO cost model, with the same methodology as Norges Bank uses to calculate the cost of phasing in new capital into the Fund. Table 3 breaks down this estimate between buying and selling costs and between commissions/taxes and estimated market impact.
Table 3: Estimated cost of short-term adjustment of portfolio to new benchmark index (USD million)
Commissions & taxes
Sale of large-cap equities
Purchase of small/mid-cap equities
If the phasing-in period is extended to ten months, the market impact will be significantly less, and the total transaction costs are then estimated at USD 62 million. Further cost reductions can be achieved by using the existing influx of capital allocated to equities by the rebalancing regime, or coordinating the phasing-in of a new benchmark portfolio with an increased equity portion. The costs associated with the sale of large-cap equities will then be reduced. However, the phasing-in period for a new benchmark portfolio will ask a great deal in terms of coordinating the more than 70 sub-portfolios which together make up the Fund’s equity portfolio, and so there may be operational reasons for limiting the length of the phasing-in period.
The cost of indexing the small-cap portfolio will be higher than the indexing cost for the current equity benchmark. This is partly because the cost of each transaction is higher, but mainly because changes in the composition of the index are more frequent. There will be a transaction requirement as a result of new small-cap companies joining the FTSE index, companies leaving the small-cap indices without joining the mid/large-cap segment, and small-cap companies issuing shares and paying dividends.
FTSE’s small-cap index has been produced for only three years, and the transaction requirement associated with changes in the composition of the index have varied considerably over time. The basis for estimating the future transaction requirement is therefore limited. The composition of the small-cap index is reviewed every quarter. During the last four quarters for which we have figures (June 2005 to March 2006), the total transaction requirement as a result of index changes in the small-cap segment of the FTSE index amounted to 23 per cent of market value. There is also the transaction requirement resulting from the issue of shares or payment of dividends outside the quarterly reviews. This transaction requirement can, on an uncertain basis, be estimated at up to 60% of the total transaction requirement.
Based on these assumptions, indexing costs for the equity portfolio will rise from around 3-4 basis points a year with the current equity benchmark to around 8-9 basis points if the small-cap segment is included. Investors normally require a higher gross return to compensate for high transaction costs. How realistic this is in the small-cap market depends on how efficiently the market actually functions.
Another potential challenge is the Fund’s ownership limit of 5 per cent of outstanding share capital. In Europe, the average stake in each company in the third quarter of 2006 is around 0.6 per cent. In the current management of the portfolio, virtually all shares in the benchmark portfolio are purchased in the internally managed index portfolios. When it comes to active management, the 5 per cent ownership limit is a particular problem for positions in small and medium-sized companies. One key strategy for excess return is to identify companies with considerable potential for profitability at an early stage in their development, which typically means small and medium-sized companies. If small companies are included in the benchmark portfolio, the managers will probably want to increase their holdings in order to maintain the size of their active positions. The 5 per cent ownership limit may make this more difficult.
A third challenge relates to the exercise of ownership rights. Norges Bank already has access to the necessary information on all companies in the FTSE index. But the number of companies in the portfolio will rise substantially, and, depending on the level of ambition for voting at small companies, there will be a need for increased resources internally at Norges Bank Investment Management (NBIM) to follow up this increased activity. The rise in the number of companies may also increase the Ministry’s work on reviewing and potentially excluding companies from the portfolio on the basis of the ethical guidelines.
After weighing up the arguments for enlarging the benchmark portfolio against the increased operational challenges, Norges Bank recommends that the benchmark portfolio for equity investments be enlarged to include the small-cap segment in the FTSE index.
This will increase the number of companies in the benchmark portfolio from around 2,500 today to around 7,000. Many of these companies are small, especially in Asia and emerging markets. One alternative might therefore be to include small-cap equities only in the developed markets of Europe and North America. The total number of companies would then be around 5,400.
The main justification for this recommendation is that small-cap equities constitute a large market segment which is omitted from the current benchmark portfolio, and that the expansion of the benchmark will give exposure to this segment more in line with the average for the market. If this recommendation is adopted, coverage of the equity universe represented in the FTSE index will rise from around 85 to 96 per cent. The gap to 100 per cent is due to a number of large markets in Asia and Europe not being included in the benchmark portfolio.
The value of bonds with a high credit risk included in the Lehman Global High Yield Index (LGHY) at the end of August 2006 was equivalent to around 4 per cent of all bonds included in the Lehman Global Aggregate (LGA). More than 60 per cent of bonds in the LGHY were issued in USD. While the average size of each bond in the LGA was just over USD 2 billion, the average size of bonds in the LGHY was just over USD 200 million. This is also substantially lower than the average size of corporate bonds in Lehman’s global index for this segment, which was around USD 600 million.
Return data for high-yield bonds in the USA are available from the mid-1980s onwards. To date, the return on this segment has been around 6.5 percentage points (annualised) above the return on government and corporate bonds. However, this is not a reasonable estimate of the expected return premium in the longer term. During the same period, the stock market as measured by growth in the S&P 500 has generated an excess return relative to government and corporate bonds of more than 10 percentage points, which is much higher than can reasonably be expected in the future. The risk premium for high-yield bonds, i.e. after expected bankruptcy costs are deducted, should be slightly lower than the risk premium for equities. A reasonable estimate might be in the region of 1-2 percentage points above the return on ordinary corporate bonds.
Over the last decade, the return on high-yield bonds has been positively correlated with the return on equities in the US market, and this correlation has been particularly high in the case of small-cap equities. The correlation with equity returns has been higher than with the return on ordinary corporate bonds with a low credit risk. Figure 1 shows these correlations calculated using monthly data in two-year windows.
It is usual for large funds to have a specific allocation to high-yield bonds. Both the expected risk premium and the segment’s diversification properties favour such an allocation.
On the other hand, it will be almost impossible to replicate a market-weighted benchmark index, as is done for both the equity portfolio and the fixed income portfolio in the Fund. The bond indices used to measure returns in the high-yield segment include many bonds which are not traded in the market, or in any case not at prices anywhere near those included in the bond indices. There is also reason to believe that pricing in this market may be inefficient, especially for bonds with the very highest levels of credit risk. Another problem with an indexing strategy is that risk measured as historical tracking error does not give an accurate picture of the real risk, which relates primarily to the probability of bankruptcy. A portfolio manager whose risk is measured as tracking error relative to an index may therefore be given the wrong incentives in terms of controlling bankruptcy risk in the management of the portfolio.
All things considered, Norges Bank does not therefore recommend including high-yield bonds in the benchmark portfolio for fixed income investments.
However, the analysis does show that it may be a good idea to have a separate allocation to this segment. This could take the form of a separate mandate in line with our recommendations for new asset classes. However, as high-yield bonds are a relatively small market segment, and the Fund is already allowed to invest in it, we will not be putting forward such a proposal on this occasion.
The Fund is intended as a permanent fund where only the real return is used. The Fund is well established as part of Norwegian economic policy, and Norges Bank believes that greater emphasis can now be given to the Fund’s long-term objective when choosing an investment strategy. On this basis, Norges Bank recommends that a larger part of the Fund’s investments be made in markets where it is possible to obtain a liquidity premium. The Fund is in a better position than most other investors to accept low liquidity in parts of its portfolio.
Norges Bank has made the following recommendations above:
- It should be made a strategic target for up to 10 per cent of the portfolio to be invested in real estate and infrastructure. Besides an expected liquidity premium for unlisted instruments, such investments will result in diversification gains that improve the trade-off between expected return and risk in the overall portfolio.
- It should be made a strategic target for up to 5 per cent of the portfolio to be invested in private equity. These are relatively non-transparent and illiquid markets, and it is possible to achieve relatively high returns with good management.
- As a short-term solution, the investment universe for equities should be extended to permit the ownership of unlisted equity which can be expected to be listed on an approved exchange in the next 12-24 months.
- The small-cap segment should be included in the benchmark portfolio for equities. This is the largest remaining segment of listed markets not yet included in the Fund’s benchmark.
The risk profile of the overall portfolio will best be preserved if private equity is included as part of the equity portfolio, while real estate and infrastructure are included at the expense of the bond portion. However, as the management structure will be different, it may be more appropriate to define investments in private equity and property/infrastructure as separate mandates, and so allow the rest of the portfolio to retain its current mandate with a benchmark portfolio of equities and bonds.
Sigbjørn Atle Berg
 This ignores the ban on investments in the fund’s homeland, as it is not relevant to draw comparisons with funds with very different objectives to the Fund.
 Real estate investment trusts pay dividends to their shareholders which are tax-free to the company, but must also undertake to invest the bulk of their capital in property and pay out the bulk of their net rental income to shareholders as dividends.
 Sources for property returns: NCREIF for the USA, IPD for the UK and Netherlands, Hordijk (2004) for the Netherlands 1977-94. The data for property returns denote the direct return in the property market, i.e. an assumed equity share of 100 per cent.
 The two main groups of fund which invest in private equity are those which invest in companies in the start-up phase (venture funds) and those which invest in established companies which are bought out, refinanced and restructured (buyout funds).
 Dimson, March and Staunton (2002): “Triumph of the optimists”, Princeton University.
 Fama/French (1996): “Multifactor Explanations of Asset Pricing Anomalies”, Journal of Finance.
Fama/French (1996): “Size and Book-to-Market Factors in Earnings and Returns”, Journal of Finance.
Fama/French (1993): “Common Risk Factors in the Returns on Equities and Bonds”, Journal of Financial Economics.