Institutional investors are increasingly considering unlisted infrastructure for its potential total returns, diversification benefits, and potential for more consistent returns compared to other illiquid private asset classes. Since 2015 the unlisted infrastructure market has grown at a CAGR of 19.7% for a total of $1.2t by March 2023, almost three times the pace of overall global AUM growth. To optimize infrastructure's integration into asset allocation it is important to quantify and model the components of infrastructure asset returns – both price and income returns.
Unlisted infrastructure equity asset-level income and price returns exhibit distinct dynamic behavior in terms of systematic factors such as sensitivity to both public market performance and the asset's characteristics (e.g., age, sector, etc.), as well as idiosyncratic behavior. As a result, in 2022, we introduced an approach to model infrastructure equity asset-level income and price returns separately to better capture these dynamics.
Modeling Unlisted Infrastructure Equity Income and Price Returns
Since then, we have acquired a new infrastructure equity asset-level time series "dataset" from EDHECinfra. The dataset allows us to test and refine our modelling of income returns and examine our parameter estimation process in more detail. Our asset-level income return model aims to assist investors in integrating infrastructure into their portfolios, seeking to enable a more precise evaluation of infrastructure's potential role in contributing to income returns, diversification, and portfolio liquidity.
The model involves two “transition matrices” that describe an asset’s income return behavior every period, based on the asset’s income return behavior in the preceding period. The first transition matrix (Step 1) determines if an asset will pay income in the current period whereas the second transition matrix (Step 2) determines the magnitude of the asset’s income return for that period.
We examine in detail two approaches to estimate these two transition matrices – Bayesian and frequentist – and compare their key differences in interpreting final estimation. One main advantage of the Bayesian approach is its ability to integrate prior knowledge with available evidence when data are limited. On the other hand, the frequentist approach relies exclusively on data for inference, avoiding the potential subjectivity introduced by Bayesian priors. When analyzing infrastructure asset performance derived from either a frequentist or Bayesian approach, investors should be mindful of the impact from potential data limitations and Bayesian prior assumptions on the reliability of estimates.
We suggest there has been persistence in the income return for an infrastructure asset over its lifetime. Once an infrastructure asset has generated income, it has tended to continue generating positive income. In addition, an asset's income return in a given year has often remained in the same income return quartile as the preceding year.
