We base our portfolios on strategic asset allocations – the most critical factor in determining returns is the baseline mix of equities, bonds, cash and other assets. So the returns and risks of different asset classes and the correlations between them are extremely important for determining the parameters of an investment strategy.

There are many resources we draw on to determine our assumptions for traditional assets, including economic theory, historical data sets and commentary from investment researchers and asset managers. But for unlisted assets this is much harder, because data is hard to come by and definitions of what’s in a particular asset class or subclass vary. There are no universally accepted benchmarks for unlisted assets. So we’re always on the lookout for information that can shed light on risks and returns of unlisted assets.

EDHECinfra has published a lengthy document in the Journal of Portfolio Management describing how (and why) they developed their unlisted infrastructure index. The key point is that the internal valuation models that these funds use smooth returns and dramatically underestimate the volatility. This is a common complaint against private assets. Their work supports the view that most investors have intuitively – that infrastructure should sit between bonds and equities on the risk spectrum and has some equity and some interest rate risk.

More details are below. The full report is available at https://eprints.pm-research.com/17511/78974/index.html?34404#

Details

• They define infrastructure as assets with a single use and a large irreversible capital investment that is paid off over many years. The cost structure is inflexible, and variable costs are low, but the asset provides a valuable service. But the asset has no value apart from that service – unlike a building, it can’t be repurposed for a different use.
• Listed infrastructure indices contain many companies that are not true infrastructure, according to their definition, so listed indices are not a helpful guide. And listed infrastructure companies are biased toward “merchant” assets that don’t have contracted revenues. (Prices might be regulated, but volumes can vary, such as on most toll roads.) These are the riskier, higher-growth end of the infrastructure spectrum. They have more equity risk and are more correlated with the business cycle. Unlisted infrastructure has more assets with contracted revenues where total payments are agreed in advance over long periods. These have less correlation with equity markets but more interest rate risk.
• Most unlisted infrastructure funds do not update their assumptions for their valuation models in a timely way, which means that asset prices are smoothed over time and volatility estimates are unrealistically low. (In other words, the reported Sharpe ratios of unlisted infrastructure assets – often over 3 – are rubbish.)
• It’s hard to use comparable sales for infrastructure valuation because assets are very different from each other and rarely traded, unlike property.
• But five risk factors (plus some control variables for different sectors) can explain almost all of the observed variation in sales prices of infrastructure assets.
• Using these for model valuations across assets can give much more realistic risk figures.
• Building an index with their model puts unlisted infrastructure as an asset class roughly between government bonds and equities in terms of risk, and somewhat riskier than corporate bonds.
• There appear to be a small number of managers employing much more risky strategies that produce much higher returns than the simulations of investments into straight infrastructure assets would suggest are possible. These have a material impact on the reported average return of funds.
• We also note that the dispersion of returns between funds is high (quartile 1 to quartile 3 spans 5% to 14% IRR over 2005 to 2013), so the volatility of a particular fund is likely to be much higher than that of the asset class. This is not usually the case for listed assets.

Their index (Infra300 below) has a much higher volatility than indices based on funds’ own valuations.

Their model uses a selection of economically intuitive risk factors.

And predicts actual transaction prices well.

This method (exhibit 5 below) makes a big difference in the risk estimates when compared to fund internal valuation methods (exhibit 4 below) for individual assets.