The 2019 EDHECinfra/G20 survey of infrastructure investors is a detailed study of benchmarking practices amongst asset owners and managers and brought to light a significant issue with regard to the investment process in infrastructure: investors do know how much risk they are taking and they are not happy about it. Key findings include:
Investors mostly use absolute return benchmarks (based on the risk-free or inflation rate), but less than 10 per cent think they are good enough. The major concerns include: they are not representative, do not measure risk and do not allow asset-liability management.
Current absolute return infrastructure equity benchmarks are not ambitious and not hard to beat. Most investors use a spread over real or nominal rate of 400 to 500 basis points. In a low rate environment, this is less than annualized stock market returns.
When investors use relative benchmarks, they use ‘fake benchmarks.’ Preferred relative benchmarks are often listed infrastructure indices, which have been shown to have 100 per cent correlation with broad equity indices by academic research. Otherwise, investors use ‘industry peers’ as a relative benchmark, despite the well-known issues encountered with valuation and return smoothing in private markets, as well as the difficulty to make direct comparisons.
With current benchmarking practices, investors in unlisted infrastructure equity cannot understand their risk and define their infrastructure investment strategy. The practices described by investors correspond more to the definition of a hurdle rate than a benchmark. Current benchmarks cannot be used to identify systematic rewarded risks, monitor risk-adjusted performance or set risk budgets.
A representative survey of infrastructure investors worldwide
More than 300 respondents took part in the survey, including representatives of 130 asset owners accounting for $10 trillion in AUM, more than 10 per cent of global AUM.
This is largest survey ever undertaken of asset owners and managers active in the infrastructure space and is representative of the views of large sophisticated investors, with 50 per cent of respondents reporting more than $25 billion AUM and 30 per cent reporting more than $50 billion AUM.
Infrastructure asset allocation requires a benchmark
Policy benchmarks reflect a long-term risk allocation choice with regards to the relevant asset class and may be a combination of sub-indices representing an investor’s preferred opportunity set. For instance, in the case of infrastructure, one might want to gain exposure to a combination of contracted infrastructure investments in project vehicles in the transport and renewable energy sectors, or focus on regulated infrastructure companies exclusively. Thus, strategic allocation to unlisted infrastructure equity or debt can involve multiple tilts defined in terms of business risk, industrial activity, geo-economic exposure, and corporate governance (see The Infrastructure Company Classification Standard, or TICCS, on the EDHECinfra website for more details).
This policy benchmark is the basis for strategic asset allocation exercises because it provides investors with measures of performance but also risk and correlation with other asset classes.
An intuitive manner to highlight the role of the asset allocation benchmark is the so-called core-satellite approach to portfolio management, by which any investment in a given asset class can be divided into two parts:
- the ‘core’ represents the risk-return profile of the average investment in a representative portfolio of the targeted asset class (e.g. an investor might favour a combination of contracted infrastructure projects and merchant power projects in the OECD) and sets the absolute level of risk (and expected returns) chosen by the investor. In the listed equity space, it would be an index fund. In the unlisted infrastructure space, it is likely to be a non-investible benchmark capturing the characteristics of an investors’ infrastructure investment strategy;
- the ‘satellite’ portfolio(s) are invested by active managers or internal investment teams and defined in terms of their tracking error relative to the core. In the listed space, this can be defined as a portable alpha fund, excluding the effect of exposure to the index from the assessment of the active strategy. In the unlisted infrastructure space, if the core portfolio is not investible directly, managers must deliver both core and satellite exposures together, but the contribution of each part is made explicit.
A core-satellite approach to active asset management has multiple benefits:
- allowing active managers to deviate significantly from the benchmark leads to a better use of the manager’s skills;
- in the case of infrastructure, because building portfolios and achieving a degree of diversification takes time, the manager’s tracking error can be set dynamically to reflect the implementation of the infrastructure investment strategy: a younger portfolio can have a larger tracking error relative to the long-term asset allocation benchmark, but the gradual implementation of the strategy should lead to a closer tracking of the policy benchmark;
- allowing a clear distinction between the value added by the design of the strategic asset allocation represented by the benchmark (core portfolio) and the out-performance generated by active portfolio management.
This last point highlights the importance of selecting the correct benchmark, both to deliver the desired risk exposure and to determine the contribution of the manager or investment team.
With unlisted infrastructure investment because of illiquidity and the difficulty to access the next transaction, the manager’s contribution consists of both creating the core portfolio (transaction by transaction) and improving on the core portfolio expected performance.
Investors acknowledge major issues with the allocation benchmarks they use
In this survey, absolute benchmarks are the most popular among unlisted infrastructure equity investors, be they asset owners, managers, or consultants, with 70 per cent of respondents reporting using such benchmarks to make strategic asset-allocation decisions.
This high reliance on absolute-return benchmarks suggests that investors are restricted to making investment and allocation decisions based solely on target returns rather than taking the risks involved in infrastructure investments into account.
Respondents who picked absolute-return benchmarks use mostly risk-free-rate- and inflation-based benchmarks. In almost 55 per cent of cases, required excess return is below 500 basis points. We note that a small proportion of investors, especially asset managers, require quite even lower excess returns. Thus, excess returns required by infrastructure investors at the allocation stage are often lower than the equity-risk premium found in public markets, which can be a surprise given the illiquid nature of the assets.
Meanwhile about 30 per cent of investors surveyed reported relying on relative asset-allocation benchmarks.
Unfortunately, current choices of relative benchmarks are also reportedly inadequate according to survey respondents. Of those respondents who preferred using relative benchmarks for strategic asset allocation to unlisted infrastructure equity, the majority of respondents said they rely on a listed infrastructure index or industry peers.
According to the survey, almost 50 per cent of asset owners use a listed infrastructure index as their infrastructure-allocation benchmark, despite the majority of them not investing in listed infrastructure as also reported in the survey.
Moreover, previous research has shown that listed infrastructure indices make for a poor proxy of the unlisted infrastructure asset class. Blanc-Brude and Whittaker (2017) apply mean-variance spanning tests to all major listed indices and show that they do not add diversification benefits to an investor’s portfolio. Bianchi et al (2018) show that the returns of listed infrastructure indices are also easily explained away for a standard Fama-French factor model. In Amenc et al (2017) listed infrastructure strategies are found to be all easy betas and zero alpha.
Hence, using listed infrastructure indices as benchmarks for unlisted infrastructure is not very different from using the broad equity market as an infrastructure benchmark, perhaps with a couple of factor tilts. As a result, it is unclear how investors make asset-allocation decisions on this basis, since most optimisers would then recommend either no infrastructure allocation or entirely replacing public equity with infrastructure in the portfolio.
The other main type of relative benchmark used for asset allocation is industry peers, in the case of approximately 25 per cent of investors.
Such peer benchmarks are created by aggregating reported infrastructure funds’ IRRs, and they face their own series of methodological issues. First, the classic issues of stale valuations and return smoothing found in private markets precludes any measure of risk using such indices (see Amenc 2008, for a detailed discussion of similar issues with real estate indices). Second, in such contributed indices, constituents are neither representative of the market nor of the strategy of any given investor, making direct comparisons difficult.
In fact only a handful of respondents think that using such benchmarks does not raise serious issues.
Absolute-return benchmarks do not measure or take into account underlying risks (unless the investment is to be considered risk-free and with an alpha of 5 per cent…) and thus partly defeat the point of using an asset-allocation benchmark, which is fundamentally an exercise about return covariance between asset classes or risk factors.
Almost 75 per cent of respondents said that the aforementioned benchmarks are not representative of the overall relevant infrastructure market.
Over 50 per cent said that these benchmarks do not allow for defining a strategy by subcategories such as business model and sector.
Around 50 per cent of respondents acknowledged that these benchmarks do not allow for the measurement of risk or correlations with other asset classes.
In order to make the best strategic allocations to infrastructure, investors need a customised benchmark of unlisted infrastructure investments — be they equity or debt investments — that is representative of their investment strategy and preferences, provides a measure of risk-adjusted returns, and allows the measurement of correlations with other asset classes.
Such indices and benchmarks are being developed by EDHECinfra using a methodology that ensures the representativeness of index constituents in both time and space and the calibration of expected returns to available transaction data in all principal markets in which this information can be observed, ensuring that such indices reflect the fair value and the risks.
The paper was authored by Noel Amenc, Frederic Blanc-Brude, Abhishek Gupta and Jing-Li Yim.
For further reading see
Amenc, N., F. Goltz, and V. Le Sourd (2006). Assessing the quality of stock market indices. EDHEC Business School Publication.
Amenc, N., F. Blanc-Brude, and A. Chreng (2017). The rise of fake infra. EDHEC Infrastructure Institute Publications.
Amenc, N., F. Goltz, V. Le Sourd, and M. Lionel (2008). The EDHEC European Investment Practices Survey 2008. EDHEC-Risk Institute Publication.
Bianchi, R., M. E. Drew, and T. Whittaker (2017). Is ”listed infrastructure” a fake asset class? EDHEC Business School Working Paper.
Blanc-Brude, F., T. Whittaker, and S. Wilde (2017). Searching for a listed infrastructure asset class: Mean-variance spanning tests of 22 listed infrastructure proxies. Financial Markets and Portfolio Management.