A dearth of data on the risk-return profile of infrastructure is leaving investors with more questions than answers in assessing this diverse investment class, says Georg Inderst

What can investors expect from infrastructure investments? What is the appropriate risk-return profile? Is infrastructure a separate asset class? How can it be integrated into long-term asset-liability models? What are appropriate benchmarks for infrastructure funds?

These are important questions pension funds are asking but the answers generally still look pretty shaky. There are many figures flowing around the financial industry but it is often less clear what their substance is. Surprisingly, there is still hardly any academic research on the subject.

The immediate question for asset allocators is whether infrastructure is a new and separate asset class in the first place. After all, pension funds have been investing in shares and bonds of utility companies for a long time, and may own related real estate. Some consider infrastructure just as a particular sector in the economy and would classify infrastructure stocks via a traditional sector approach.

In contrast, the proponents of the alternative asset class define infrastructure assets differently. They find common economic and financial characteristics, such as excess returns (from natural monopolies, regulated businesses or government concessions) and stable long-term cash flows.

This has led to a situation where a lot of different sub-classes are being included, such as airports, gas distribution networks, satellites and hospitals. It remains to be seen how the supposed commonalties of infrastructure assets will stand up to scientific scrutiny, as they may turn out to be idealisations of a very diverse reality.

New-age infrastructure investment also comes in different vehicles. The investment industry offers an increasing number of private equity-type funds. Some bigger institutional investors additionally seek direct participation in (unlisted) infrastructure companies. So, should infrastructure investments be subsumed under private equity or real estate?

The experts stress the differences to private equity (a longer time horizon, higher and more stable yields) and real estate (investment in companies rather than physical property, less competition, bigger scale).

When the global infrastructure investment boom started, some fund providers set return expectations of 15%-plus per annum, based on early results from pioneering funds in Australia set up in the mid-1990s. Currently, pension funds are being presented with all sorts of stylised risk-return charts: sometimes showing infrastructure with risk and return both higher than equities; sometimes both are lower, and sometimes at higher returns and lower risk (‘equity-type returns with bond-type risk').

Given the heterogeneity of infrastructure, consultants like to give broad ranges for expected returns, such as 10-15% or 8-18%. However, not all investors are aware of the use and impact of leverage in infrastructure investments, typically 30-80%.
What numbers are being used in asset liability-models? Take Morgan Stanley Investment Management as an example for managers: of five main asset classes, infrastructure (volatility 7.9%, return 9.3%) is placed second only to bonds (4.4%) in terms of expected volatility, and second only to private equity (10%) in terms of expected return.

As an example for pension funds, in 2007 the Dutch ABP scheme set a 10% strategic return expectation against a risk of 7%. In comparison, the corresponding figures are 6%/9% for property and 15%/25% for private equity. Such projected risk-return profiles clearly look attractive to investors. However, it remains unclear, what long-term risk premia (eg, for credit, size, illiquidity, politics) should exactly be used for infrastructure investments, even in theory.

Crucial differentiations will have to be made, for example, between early-stage assets (being closer to private equity) and high-yielding mature assets (similar to utility bonds or low-beta stocks). In practice, investors will need to look deep into the specialist funds as they may contain things ranging from pioneering greenfield energy projects to more pedestrian assets like toll bridges.

How should pension funds benchmark infrastructure funds? What could be considered success or failure?
In theory, there are a number of possibilities, including:

Absolute return figure (eg, 9%); Inflation plus margin (eg, consumer price index (CPI) +4%); Bond yield plus margin; (Inflation-linked) bond index return plus margin; Blend of equity, real estate, bond and private equity benchmark; Listed infrastructure index or global equity index or blend of the two; Peer group of unlisted infrastructure funds.

In practice, there seems to be a great deal of variety in the benchmarks set by pension funds, with the first two seemingly the most important. US scheme CalPERS, for example, is looking for an annual return of inflation plus 5-7%. The choice of an appropriate benchmark depends on a number of factors, relating both to the liability profile of the pension fund and the type of infrastructure investment.

What do the actual performance figures tell? Figures presented, in marketing for example, are to be taken with caution. They regularly refer to the performance of listed stocks. In addition, infrastructure sector indices are often heavily biased towards utility stocks that experienced a strong revaluation in the mid 2000s.

For unlisted infrastructure investments, publicly available reliable data are sparse. Independent data collection is still rudimentary and there are no agreed performance reporting standards. History can offer little guidance, as the experience with infrastructure funds is rarely longer than two to three years for most pension funds, except for Australia.

A recent study analyses 19 unlisted Australian infrastructure funds. Their average annual return over the 10 years to the second quarter of 2006 is 14.1%, beating that of bonds (7.2%), stocks (12.9%) and direct property (10.9%). Volatility of unlisted infrastructure (5.8%) is lower than for stocks (11.0%), but higher than for bonds (4.3%) and direct property (1.5%).

This study of Australian data also seems to confirm the diversification opportunity of unlisted infrastructure, as correlations (using quarterly data) with other asset classes appear rather low: 0.06 with equities, 0.17 with bonds, and 0.26 with direct property.

A number of caveats are necessary, some of them also given by the authors, particularly on the appraisal-based valuation of infrastructure and property; definition of risk; indices used; frequency of data, and the period analysed (before the credit crunch of 2007-08). In particular, such statistical analyses tend to underestimate volatility and correlations of unlisted instruments.

Risk analysis must go much further than backward-looking statistics. Pension trustees will be interested in the specific risks of infrastructure investments, not least the regulatory and political risks. The current financial crisis has increased the awareness for other risks, such as leverage, counterparties, valuations, liquidity, opacity and conflicts of interests.

Infrastructure investments have some promising long-term features for pension funds, but the challenges in governance and management remain high. Many trustees are not yet at ease with private equity-type investment vehicles. It can only be healthy that return expectations are becoming more modest and the understanding of risks more realistic.

Georg Inderst is an independent adviser