Unlike liquid markets and real estate, a lack of publicly available data has made assessing risk more challenging for infrastructure equity assets. Laurence Monnier considers the options for investors looking to complement qualitative assessments with quantitative rigour.
Modern portfolio theory’s influence on public markets is still widely felt
It is nearly 70 years since Harry Markowitz unveiled his modern portfolio theory (MPT) and, while active investing advocates have since credibly challenged the notion of perfectly efficient markets, MPT’s influence on public markets is still widely felt.1,2
Building on Markowitz’s legacy, new portfolio construction tools and theories have emerged to assist investors in measuring risk and return within public markets. Such efforts have been supported by rich and long datasets. Indeed, quantitative analysis is so prevalent in liquid markets that 70-80 per cent of US stock trading is algorithmic.3
Private markets, on the other hand, have no such data feast to frenzy on. A lack of access to transparent, comprehensive and long-term pricing history means infrastructure equity risk calculations are largely qualitative, as shown in Figure 1.
Figure 1: Infrastructure equity calculations
More quantitative market analysis is being developed by academic research institutes and index providers, such as EDHEC infrastructure and MSCI. This is welcome but translating it into active investment decisions is not easy. Furthermore, the lack of transactions with publicly available prices means analysis has to either rely on valuations reported by fund managers or on internal models. This absence of a single price reference for each asset results in inconsistent output.
The prospects for computer-driven algorithms replacing human investors seem quite far off
The opacity of private markets, and high transaction costs, means the prospects for computer-driven algorithms replacing human investors seem quite far off. However, this does not mean investors should abandon more quantitative, risk-based measures altogether; it just requires a more thoughtful and nuanced approach.
Debt rating equivalents
In Aviva Investors’ 2020 Real Assets Study4, which took the pulse of over 1,000 institutional investors representing over €2 trillion of assets under management, more than 70 per cent of respondents said they allocate to real assets for cashflow-matching reasons.
These investors may invest in equity as well as debt. Income-focused equity investors often ask what the “rating equivalent” of their investment is. Although equity holdings do not have credit ratings, investors need to understand the intrinsic income risk of their investment (ignoring changes in capital value).
Ratings are not quantitative measures: they are opinions on the credit quality of an asset. Yet, they can be translated into “probability of default” or “expected loss” figures based on historic data, which provides quantitative insight to investors and regulators. Figure 2 shows the link between ratings and probability of default on debt over 20 years, based on Moody’s data.5 Twenty years has been chosen as a typical maturity for buy-and-hold investors in infrastructure equity.
Figure 2: 20-year cumulative probability of default (per cent)
There are, of course, fundamental differences between equity and debt income. Debt risk is downside only, with a binary outcome (performance or default), while income risk for equity assets is generally greater6, with multiple sources of upside and downside variations.
Infrastructure investment decisions are usually based on the expected return for an asset and scenario analysis around the base case. If we assume the expected return is the median of the potential return distribution, it has a 50 per cent chance of being missed and, conversely, a 50 per cent chance of being exceeded, unlike debt.
There is a demonstrable and systematic tendency for investors to be overly optimistic
This means the probability of achieving or exceeding the median return would correspond, based on Figure 2, to a B rating with a 50 per cent probability of default. In fact, there is a demonstrable and systematic tendency for investors to be overly optimistic, a phenomenon referred to as optimism bias.7 Factoring this would imply the expected return may have less than a 50 per cent chance of being reached.
There is also at least a 90 per cent chance the same asset will produce (or exceed) a lower return. For investors focusing on downside risk, this would be consistent with a BBB8 scenario in the debt universe, which is a return that has only a ten per cent chance of being missed over 20 years. Figure 3 shows an asset with a mean return of eight per cent but a “BBB probability return” (i.e. with at least a 90 per cent chance of being reached) of 1.6 per cent. The same analysis can be used to assess the probability of capital loss.
For income-focused investors, understanding what part of the income has an investment-grade probability of exceedance is invaluable, and may potentially open the door to cashflow-driven investment strategies.
Figure 3: Return distribution illustration (per cent)
Of course, equity investors do not generally know the future return distribution of their investments to support this analysis. It might not follow a normal distribution around the expected return either. One reason is optimism bias. Another is that some risks inherent to infrastructure, such as regulatory risk, are quite binary.
To help overcome this, we have developed our own methodology to model the return distribution of various equity sectors and asset classes. This approach is based on cashflow forecasting appropriate for each sector and measures the expected dispersion of future income over the life of the asset. It has been developed to inform asset allocation decisions at a sector level,9 which enables us to assess the risk (i.e. dispersion of future returns) for particular asset classes and, in turn, the likelihood of achieving or exceeding certain return levels.
Taxonomy confusion
Infrastructure assets encompass a wide range of activities. Assets are generally classified according to certain characteristics: geography; sector; type of revenues (e.g. regulated/contracted/merchant); and stage of development (greenfield or brownfield).
Often, this taxonomy informs the initial view of the likely risk of an investment. For example, an infrastructure fund with a low-risk target may cap the level of greenfield investment (i.e. new-build assets) or merchant risk, both of which are perceived as carrying higher risk.
Our methodology indicates this taxonomy, although useful, does not tell the full risk story. Figure 4 compares three types of renewable sectors in the UK: brownfield solar with feed-in-tariffs (FIT), onshore wind with renewable obligation certificates (ROC), and a greenfield energy-from-waste plant with no subsidies.
Figure 4: Comparing return distribution of three renewable assets (per cent)
In Figure 4, we annotated as “BBB” the return with a 90 per cent probability of being exceeded.
While ROCs and FITs are no longer available for new projects, operational projects can still benefit from these subsidies for part of their remaining life (assumed to be 16 years in this example). The first two sectors would be classified as subsidised renewable projects. By contrast, energy-from-waste represents unsubsidised assets, but these plants derive around two thirds of revenues from treating waste. Such revenues benefit from supportive regulation and can be contracted with a waste aggregator for the medium term.
The three sectors have different levels of complexity
The three sectors also have different levels of complexity. While the first two projects fall in the same category (operational renewable assets with subsidies) and country, they have a markedly different risk and return profile than most would expect.
The dispersion of returns for ROC onshore wind is much closer to energy-from-waste than the solar asset – although energy-from-waste has a higher tail risk. This is due to all the solar revenue being derived from subsidies, while a high proportion of ROC wind revenue is derived from wholesale power sales. The third asset, the energy-from-waste project, additionally has construction risk and more complex operations, but these risks are offset by contracted waste treatment revenues and supportive regulation.
Cashflow matching
For investors seeking low income volatility, the solar asset with feed-in-tariffs appears more attractive. While it has the lowest median return, the return with a 90 per cent probability of exceeding is higher than onshore wind. The return dispersion and left-tail risk is also much lower.
Energy-from-waste has a much higher volatility
Energy-from-waste has a much higher volatility and more left tail risk but would seem attractive given it has a higher “BBB probability” return and significant upside.
However, if the portfolio is required to provide specific cashflows, not only does the investment income need to exceed liability costs over time, it also needs to be highly predictable each year – a dimension not shown in Figure 4.
For assets with construction risk, complex operations and some exposure to market risk, such as greenfield energy-from-waste, the potential for cashflow variations from year-to-year is elevated, particularly if construction is delayed. Therefore, while the asset class could be attractive for investors due to the high probability of achieving an attractive return over the life of the asset, it is less attractive for cashflow-driven investment.
Investors should therefore favour assets with a high proportion of revenues derived from subsidies or contracts
Investors targeting renewables looking for predictable income should therefore favour assets with a high proportion of revenues derived from subsidies or contracts. If the income is to be used for liability-matching purposes, another way of thinking of this is which part of the income may be considered, or even structured, as a “debt tranche”, whereby cashflows are synthetically securitised.
Looking beyond renewables, fully contracted infrastructure assets with similarly low volatility can be found in other sectors, including data centres and accommodation. Outside of infrastructure, income-focused investors may find similar risk profiles in other real assets, such as long-term amortising leases (income strips), or long-lease real estate, as shown in Figure 5. While such assets may not have the same appeal as renewables from a net-zero emissions perspective, they can generate other positive ESG effects.
Figure 5: Return distribution: infrastructure and long lease examples
Look beyond expected returns
Infrastructure equity has been gaining favour with investors seeking attractive risk-adjusted returns. However, the analysis of risk and return for this complex asset class, although improving, lags behind public markets and private real estate due to a lack of price transparency.
Investors should look beyond the expected return to understand the risk profile
Segmenting the wider infrastructure market is helpful to better inform investors on the diversity of infrastructure but is not a substitute for quantitative risk analysis. Investors should look beyond the expected return to understand the risk profile, or return distribution, of their portfolio.
With increased transparency and more sophisticated ways to analyse returns, fund managers should be better placed to build portfolios that meet client objectives while improving risk-adjusted returns. Such analysis may attract more liability-driven investors into the asset class, particularly if they can isolate what part of the return has investment-grade-type characteristics.
