Submission to the Australian Treasury
SUPPORTING RESEARCH PAPER

In a short accompanying paper provided alongside this submission, we document and back up with data four key points and two recommendations, which are summarised below:

1. Superannuation investments in infrastructure focus on those countries and sectors that make up the unlisted infrastructure universe.

The majority of their infrastructure equity investments are made in unlisted assets, via asset specialist asset managers, in Europe and Australia and in the Transport and Utilities sectors. We collect data for 424 investments made by Superannuation funds and still held at the end of 2019, 324 of which are in the 25 countries that make the bulk of the unlisted infrastructure universe, that is more than 70% of the unlisted infrastructure investments currently held by these funds.

The unlisted infrastructure investments made by Superannuation funds span the entire breadth of the unlisted universe including a number of sectors such as social infrastructure or renewable energy companies which are seldom found in listed markets. In terms of both geographic and sector distribution, the infrastructure portfolios of Superannuation funds are in line, albeit not similar to, with the unlisted universe. We report the following stylised facts:

1. There is a home bias: Australia represents close to 12% of the global unlisted infrastructure universe, but a quarter of the AUMs Superannuation funds dedicate to unlisted infrastructure.
2. Superannuation funds make 70-80% of their unlisted infrastructure equity investments in Europe and Australia.
3. Superannuation funds also favour investments in transport and utilities. These two sectors represent around 70% of their unlisted infrastructure investments by size.
4. 80% of investments in unlisted infrastructure by Superannuation funds, by count or by size, are made indirectly via managers. This is relevant when considering the role of fees.
5. The benchmark currently used in the APRA Heatmap test for infrastructure is not representative of the way Superannuations invest in infrastructure.
6. The FTSE Developed Core index corresponds to twice as few firms but 49 times as much market capitalisation than the unlisted infrastructure equity held by superannuation funds. The type of firms found in the FTSE index differs greatly from those unlisted infrastructure companies Superannuation funds invest in. The top 10 largest firms found in the FTSE index are very large telcos and energy firms that are closer to industrial conglomerates than infrastructure projects.
7. The FTSE index is extremely concentrated in Network Utilities which represent about half of the index weight. The FTSE Index is also underweight in key sectors of the unlisted infrastructure sector notably transport and social infrastructure, which are sectors in which Superannuation funds hold a significant part of their unlisted infrastructure investments. The FTSE index fails to represent entire segments of the unlisted infrastructure universe. Moreover, 16% of the index by market capitalisation (18% by size) cannot be considered infrastructure under the TICCS taxonomy.

Thus, the FTSE Developed Core Infrastructure Index is not at all representative of the Superannuation fund portfolios, which are focused on the transport sector and a spread of investments over all other segments of the unlisted universe.

The FTSE Core Index also exhibits strong return correlation with listed markets with very high and significant levels of correlation in the 70-80% range with equities in and outside Australia and Real estate. Indeed, we find that more than half the number of constituents in the FTSE Developed Core Infrastructure Index are found in other APRA equity benchmarks, translating to a 91% overlap in terms of market capitalisation for the FTSE Developed Core Infrastructure Index. We also show that the FTSE Core index does not in fact represent a unique asset class when compared with the other benchmarks of the APRA Heatmap: it is already ‘spanned’ by the risk-returns characteristics of the other asset classes used in the Heatmap.

Looking at the financial characteristics of the FTSE Core Index and the infra300 index of unlisted infrastructure produced by EDHEC, we see that:

1. Unlisted infrastructure is more defensive than the listed infrastructure would suggest: this is illustrated by the Value-at-Risk and Maximum Drawdown of the infra300 (hedged to AUD) and the FTSE Core Infrastructure (hedged to AUD). While the listed infrastructure index exhibits a 99.5% one-year VaR in the 25-30% range depending on the horizon, the VaR of the infra300 ranges between 15-18%. Likewise, the maximum drawdown of the FTSE Core index is in the 16-28% range while the infra300 never exhibits drawdown greater than 10-12%.

• This confirms that unlisted infrastructure does exhibit drawdown protection characteristics. In the accompanying paper, we show that this finding is not driven by any ’smoothing’ of the infra300 returns (which exhibit no serial correlation and significant volatility) but is instead the result of unlisted infrastructure companies being exposed to fundamentally different risks than the constituents of listed indices, including during the Covid-10 lock-downs. This result suggests that unlisted infrastructure could be treated different in the ’simple reference portfolio’ test conducted by APRA and categorised as more defensive than it currently is.

2. Unlisted infrastructure exhibits higher risk-adjusted returns than listed proxies: the Sharpe ratio i.e. the return per unit of risk, is 0.7 in Q2 2020 for the infra300 but 0.5 for the FTSE Core Index. A similar difference exists for all investment horizons. This higher risk-adjusted return is what attracts Superannuation funds to the unlisted infrastructure asset class.

• It must be noted that this is not the result of producing ‘alpha’ over a listed equivalent. First, we established that the listed index is not representative of the unlisted universe. Second, the higher risk adjusted returns of unlisted infrastructure investments are the result of systematic risk exposures and rewards. Proxying these risks in the context of the APRA SAA test requires using the right benchmark corresponding to the risks and rewards of the unlisted infrastructure asset class.

3. Unlisted infrastructure has higher dividend yields compared to listed infrastructure: the infra300 index has a 9.41% dividend yield compared to the FTSE Core mean dividend yield of 3.28%. Such a very significant difference in dividend yield, which is also one of the main reasons why investors are attracted to unlisted infrastructure, shows that the underlying firms and risks of the FTSE Core and infra300 indices are completely different.

Thus, the FTSE Core Infrastructure Index is completely inadequate as a proxy for the unlisted infrastructure portfolios or strategies of Superannuation funds:

• It is not capturing the same universe or the same type of firms;
• It is highly concentrated in a few firms that are not representative of the unlisted infrastructure universe in which the Superannuation funds invest;
• It is highly correlated with other listed indices and in fact cannot be statistically distinguished from them as demonstrated by mean-variance spanning tests;
• As expected, it exhibits risk and return characteristics that are very close to listed equity indices and is quite different from a global index of unlisted infrastructure equity (the infra300) which exhibits more defensive characteristics and different risk dynamics.

3. Better benchmarks exist that captures the characteristics of the unlisted infrastructure asset class in which Superannuation invest.

A better benchmark than the listed infrastructure index put forward to test the performance of MySuper products can be designed using a representative dataset and mark-to-market valuations that adequately capture the risks and returns of the unlisted infrastructure asset class. The infra300 is an index of the international market for unlisted infrastructure equity produced by EDHECinfra each quarter along with several hundreds of indices of the segments of the unlisted infrastructure universe and used by several dozen investors including in Australia.

While a listed index is not adequate when it comes to capturing the characteristics of the unlisted infrastructure asset class, until recently the only alternative was an index based on private appraisals (like the ones mentioned and rejected in the 2018 Productivity Commission report). Indeed, as the 2018 PC report highlighted, this type of data suffers from multiple issues and biases including a lack of representativeness (selection and survivorship biases) and no robust measure of risk due to the ‘smoothing’ of appraisals and returns.

These issues have now been addressed by recent advances in research: a representative dataset of the investible universe, and measures of the mark-to-market performance of the unlisted assets in this representative sample i.e. applying IFRS 13 guidelines and using the latest transaction data to update the estimate of the risk premia that applies to each investment, are possible.

With this approach there is no more smoothing in the returns and a proper measurement of the variance of returns is possible. Representative and realistic risk and risk-adjusted characteristics are produced.

The Covid-19 crisis provides a test of the inadequacy of the FTSE Core of unlisted infrastructure as represented by the infra300 index of unlisted infrastructure companies:

1. In Q1 2020, with the first wave of Covid-19 lock-downs all listed equities, including the FTSE Core index, experienced very negative quarterly returns due to their significant exposure to the market beta. Note that while the FTSE Core is mostly dominated by energy and telecom companies, which were not immediately impacted by Covid-19 lock-downs, the FTSE Core index had a -16% returns in that quarter (table 1 below). In comparison, the infra300 had smaller negative returns. Indeed, while the index includes numerous transport companies that were affected by the lock-downs, it also includes many more ’contracted’ infrastructure businesses which did not see their cash flows impacted by the Covid-19 lock-downs. The infra300 Q1 2020 returns are negative because the unlisted infrastructure equity market risk premia increased for almost all assets but in aggregate the impact on the unlisted infrastructure sector was less dramatic than for listed equities, despite subsegments of the infra300 like unlisted airports having strong negative returns. The FTSE Core Index could not have captured this effect, despite the fact that exposure to unlisted transport assets is at the heart of the Superannuation unlisted infrastructure investment strategies.
2. In Q2 2020, with the spread of the pandemic and the economic impact of the lock-downs, more infrastructure sectors began to be affected negatively such as utilities and roads. The infra300 continued to exhibit negative returns but listed benchmarks including the FTSE Core Index returned to positive quarterly returns as a result of the strong rebound in capital markets. Once again, we see that the unlisted and listed indices follow very different dynamics and that the FTSE Core does not represent what happened to the unlisted infrastructure asset class.
3. In Q3 2020, the infra300 returned to positive territory as the unlisted infrastructure market risk premia stabilised and some infrastructure sectors exhibited a strong rebound in revenues such as toll roads. The FTSE Core index was also positive but not for the same reasons since it is exposed to different risks and does not include significant exposure to transportation assets, one of the main types of unlisted infrastructure held in superannuation products.
4. On a YTD basis in 2020, we see that the FTSE Core exhibits -7% returns in AUD hedged terms whereas unlisted infrastructure has proven more resilient and is down -4.7% globally (AUD hedged), despite the larger losses experienced in the most exposed merchant transport sectors. Clearly, as a benchmark of how the unlisted infrastructure performed during the Covid-19 pandemic, the FTSE Core is a poor proxy of the unlisted infrastructure sector. The infra300, which is build directly from measures of the fair market value of a representative set of unlisted infrastructure companies, shows the actual impact of higher risk premia, lower cash flows and lower interest rates on unlisted infrastructure NAVs.

Thus, while the FTSE Core Index is shown to be wholly inadequate as a proxy for unlisted infrastructure, in particular in terms of coverage and representativeness of the investments made by Superannuation funds, it is possible to build and produce a fair benchmark of the unlisted infrastructure asset class: the infra300 is designed to be a bias-free, representative view of the ’principal’ market i.e. the main markets in which buyers and sellers of unlisted infrastructure companies are the most active, including Australian Superannuation funds. It is based on a mark-to-market asset pricing technology that captures the risks inherent to the asset class and produces robust, realistic results that can serve as the basis for benchmarking the investments made in MySuper products, as the example of the Covid-19 crisis demonstrates.

Table 1: Impact of Covid-19 on Infrastructure: Gross Quarterly Total Returns of Public Equities, FTSE Core Infrastructure Index and the infra300 Index

4. Using the infra300 as the unlisted infrastructure benchmark in the APRA tests is supportive of the regulator’s objectives

We look at the two performance tests that APRA currently produces for the Heatmap: the Simple Reference Portfolio (SRP) test and the Strategic Asset Allocation test (SAA) and consider how they may be impacted by switching the benchmark of the unlisted infrastructure asset class from the listed FTSE Core Index to the infra300 index of unlisted investments in infrastructure equity described in detail in the previous section.

We examine two potential evolutions of the treatment of unlisted infrastructure in the APA performance tests:

1. The impact on the SRP of classifying unlisted infrastructure as more defensive that it currently is, given the evidence provided by the infra300 on the defensiveness of the asset class, which is not captured by the FTSE Core benchmark currently used.
2. The impact on the SAA test of using the infra300 instead of the FTSE Core Index, in particular, whether it would support the regulator’s objective to ‘punish under-performance’ in MySuper products.

We argue that unlisted infrastructure should be considered more defensive since it possess properties that help protect portfolios in downside markets like lower VaR and lower maximum drawdown as shown by the infra300. This index also exhibits significant correlations with both international fixed income and Australian fixed income, asset classes that are considered defensive in the APRA classification, and lower correlations to listed equities.

To conduct this analysis, we use the MySuper asset allocations for the 138 products obtained from MySuper statistics. We make the same SRP calculation than APRA but using a 50-50 split between growth and defensive styles for unlisted infrastructure. We find that increasing the defensive classification of unlisted infrastructure from 25% to 50%, decreases the SRP for products invested in infrastructure from 7.227% to 7.165%. This is normal since the defensive style can be expected to have lower returns but the change is marginal, thus making the test equally robust from the point of view of the regulator but better reflecting the defensive characteristics of the unlisted infrastructure asset class in individual cases.

Next, in the context of the SAA test, the use of the wrong proxy results in making incorrect conclusions as to how much value is added by managers. Implementing the SAA test with listed benchmark assigned to proxy unlisted infrastructure would lead to random, unscientific and fundamentally unfair outcomes:

• In some periods the listed market exhibits much stronger returns than unlisted infrastructure, which is characterised by its defensive characteristics and attractive risk-adjusted returns. In this case, investors in unlisted infrastructure would be unjustly punished by the SAA test.
• In other periods, listed markets may have lower returns than private assets and investors would benefit from an ’apparent’ alpha just by allocating finds to unlisted infrastructure but without exhibiting any skills while doing so.

Using a representative index like the infra300 (here, net hedged-AUD) would solve this problem. The infra300 is much closer to the investment strategy of Superannuation funds in unlisted infrastructure and also designed to be representative of the unlisted infrastructure equity universe.

To determine what the impact of using the infra300 in the SAA test instead of the FTSE Core Index would be, we make a similar comparative analysis between single strategy 63 MySuper products and examine the impact on the SAA test of changing the infrastructure benchmark for the infra300 index instead of the FTSE Core. As shown in table 2, we find that:

1. 14 out of 63 products, score better than the SAA benchmark when using the infra300 instead of the FTSE Core index
2. 39 products score less well than they would using the FTSE index
3. 10 products fare the same irrespective of the choice of infrastructure benchmark, mostly because they invest very little or not at all in this asset class.
4. A single product actually switches from outperforming its SAA benchmark to under-performing it because of the change of infrastructure benchmark from the FTSE Core to the infra300.

Thus, using the infra300 index as the proxy of unlisted infrastructure does not overturn the results of existing SAA tests and preserves the regulator’s objective to apply a robust market test to Superannuation products. Because it uses the correct benchmark, as we argued above, instead of punishing managers randomly and unfairly, using the infra300 would reward those managers who invested skillfully in the relevant unlisted infrastructure market.

Looking at APRA’s Q2 2020 Heatmap, we see that the products that underperform in the SAA test do so on average by -0.63% (median -0.65%). This puts the small size of the mean impact on returns of switching infrastructure index in perspective: while the average effect is unlikely to change the test result, the more infrastructure plays a role in the product, the more relevant using the correct benchmark becomes and the more it makes a difference in the SAA test. More details are provided in the supporting paper.

Table 2: Impact of Using the infra300 (net, hedged-AUD) instead of the FTSE Core in the APRA SAA Test in 63 Single Strategy MySuper Products

 

Our Recommendations

These finding are in line the intent of the ‘Your Future, Your Super’ legislation and demonstrates the importance of including a relevant, representative benchmark for unlisted infrastructure investment.

Using the infra300 in the APRA Heatmap and subsequent performance tests would not only be much more representative of the underlying investments made and risks taken by investors in unlisted infrastructure but also help identify those managers that actually create value through these investments.

Our two recommendations to the Treasury and to APRA in the context of the reform of MySuper are to:

1. recategorise unlisted infrastructure as 50% defensive in the Simple Reference Portfolio (SRP) test of the APRA Heatmap, and to
2. use the infra300 index (Hegded-AUD) to proxy the unlisted infrastructure asset class in the Strategic Asset Allocation (SAA) test of the APRA Heatmap. This index is described on the EDHECinfra website (edhec.infrastructure.institute/infra300-index) and available on the EDHECinfra platform (indices.edhecinfra.com) and via Bloomberg.