The challenges of benchmarking infrastructure in a pension portfolio
BY Yaelle Gang | January 7, 2020
Canadian pension funds’ median allocations to infrastructure are higher than their global counterparts, with these allocations increasing significantly since 2010.
In fact, according to data provided to the Canadian Investment Review by Preqin Ltd., the median allocation to infrastructure for a global public plan was 1.7 per cent in 2010, while a Canadian public plan’s allocation was four per cent. In 2019, both numbers increased to 2.5 per cent and 7.8 per cent respectively.
For private plans, the median allocation to infrastructure for a global plan was 2.1 per cent in 2010, compared to three per cent for a Canadian private plan. In 2019, the median allocation for a global private plan dropped to two per cent, and for a Canadian private plan it rose to 6.2 per cent.
As the asset class grows in popularity, pension plan sponsors continue to face challenges when it comes to benchmarking infrastructure.
The importance of the benchmark
Benchmarking is important because it’s the only way to understand infrastructure investment on a multi-asset class basis, says Frederic Blanc-Brude, chief executive officer at Scientific Infra and director of the EDHEC Infrastructure Institute. “Before knowing how well your asset managers perform or anything like this, the first other questions for a [chief investment officer] are, ‘Should I invest in infrastructure in the first place? What role does it play in my portfolio? What does it contribute? How different is it from bonds and stocks? Does it have a risk profile that is interesting for me?’”
Plan sponsors will all have different answers to these questions, but until they have something that is mark-to-market and measures risk, they’re stuck in limbo, he says. “You have your investment beliefs about infrastructure, but do you actually know if they’re true or not? So that’s what these benchmarks do. They help you understand and validate your investment beliefs about unlisted infrastructure.”
Benchmarks are also important because they influence behaviour, says Andrew Claerhout, the former senior managing director of infrastructure and natural resources at the Ontario Teachers’ Pension Plan and a partner at Searchlight Capital. He notes benchmarks are directly linked to compensation, which influences behaviour.
“If you adopt a very low benchmark and you say our benchmark for infrastructure is five per cent, you’re going to have people probably piling into a bunch of lower risk infrastructure assets . . . because they can get comfort that they will comfortably beat their benchmark to get paid,” he says. “If you have a higher infrastructure benchmark, people are going to need to be more creative and people are going to need to selectively move up that risk-return spectrum. So you just need to be aware of that.”
Challenges in benchmark selection
Infrastructure is generally considered half debt, half equity, says Janet Rabovsky, a senior investment professional. “So there is no real benchmark.”
As well, infrastructure isn’t as homogenous as some of the other asset classes and it’s a fairly new asset class without a long history, she adds.
Further, it’s hard to benchmark infrastructure because assets are illiquid and seldom traded, says Blanc-Brude. “If you look at the entire population of infrastructure secondary market transactions, you find that, on average, these companies trade once in their life. But actually many of them never trade.”
This means very little pricing data is available, he adds.
And institutional investors often don’t have a lot of assets in their infrastructure portfolios. “It’s quite difficult for investors to get an understanding of what it is they’re looking at,” Blanc-Brude says. He adds investors can also try to find a public proxy to use as a benchmark by looking for listed infrastructure, but notes this approach has issues.
Firstly, most infrastructure is unlisted and, when it is listed, it’s only in some sectors and some countries. “If you wanted to find some listed proxy or listed equivalent of, say, renewable energy projects or social infrastructure or P3s or even toll roads, it’s very hard to find companies that are listed and that actually are like that,” says Blanc-Brude.
When looking at listed infrastructure indexes, most of the constituents aren’t actually infrastructure, he adds. “The common story is Amazon is infrastructure, because they have data centres, but really all you’re doing is kind of hugging the S&P 500 and not really investing in infrastructure.”
Using listed infrastructure as a proxy for unlisted infrastructure is also very volatile and perfectly correlated with the stock market, he says.
Among the challenges of using unlisted infrastructure as a proxy is the fact that it relies on data that’s been provided by certain managers, so the data doesn’t represent all countries and sectors. “It’s not so representative and it also suffers from the problem of what you call survivorship bias, which is that the only thing managers report is the assets they hold,” says Blanc-Brude. “And so the assets they hold are the ones that are performing because the ones that are not performing, they’re either bankrupt, gone or it sold them. So you sort of overrepresent the winners, if you use that kind of information.”
The data also includes smoothed valuations, which doesn’t really represent the state of the market, he adds. “And so if you look at this data you see very little volatility and the implied risk-adjusted return is extremely high. And so the Sharpe ratio of these contributed indices is usually around three. If you’re in financial markets, you know that there’s no such thing as an investment with a Sharpe ratio of three, especially over a long period of time. It makes no sense. There’s nothing that cheap. It’s not possible.
“So these are the issues that investors face. They have a problem of finding the right proxy in the listed space or finding a representative and a fair proxy in the private space.”
Solutions on the table
With these various challenges in mind, most investors use an absolute-return benchmark for lack of a better alternative, Blanc-Brude says. “In fact, it’s not really a benchmark, it’s more like a hurdle rate.”
At the Ontario Teachers,’ the benchmark is the local consumer price index, plus a local credit default spread, plus four, notes Claerhout. “That’s their benchmark. And it doesn’t really change whether they’re talking about a super core asset with an expected return of six or seven per cent or they’re talking about a value-add asset with expected return between 12 and 15 per cent.”
This can be problematic because plans try to pick one benchmark in an asset class that has many alternatives in terms of the risk and reward trade-offs, he says.
However, using one benchmark provides investment teams with a simple way to think about their required rate of return and a simple way to model in terms of asset allocation, he adds. It also gives investment teams discretion.
The approach isn’t wrong, notes Claerhout. “I don’t think I would necessarily change the way they’re doing it. I would probably continue to have one benchmark and then say to the team, ‘Exercise your discretion in terms of how you create your own portfolio of infrastructure assets to outperform that benchmark.’”
Rabovsky agrees that a lot of pension funds will choose the approach of CPI plus four or five for asset allocation. Another approach, she notes, is using a blended equity and bond benchmark plus a performance premium. “When you get down to the manager, it is things like purchase price multiples, [internal rate of returns], multiple of capital, etc. Core is going to have a different approach than core-plus and value-add.”
Blanc-Brude’s team has created infrastructure indices by building a representative set of the investable universe, he says. “We spent four years documenting the state of the universe, so finding the unique identifier of all the infrastructure companies that we could invest in in 25 countries.”
About 7,000 of these companies exist, he notes. “And so that tells us what the universe is like and so what we should aim to capture in the index.”
His team has also created a taxonomy of infrastructure companies to categorize infrastructure assets. Once the universe was mapped and categorized, he worked to price the companies.
The index has three levels. “There’s the global index with everything in it and, because it’s made of a sufficiently large sample, we can slice and dice it according to the taxonomy to make sub-indices.”
This can help investors create customized benchmarks to their unique portfolios, he adds.
Ultimately, the discussion around infrastructure benchmarking is ongoing, Rabovsky notes. “You can measure a deal, but it’s very hard to measure funds right now.”