Infra Insight Blog

Posted by guest blogger Ryan J. Orr

A recently published paper analyzes returns on infrastructure investments and produces some rather surprising findings. The paper, Risk, Return and Cash Flow Characteristics of Infrastructure Fund Investments by Florian Bitsch, Axel Buchner, and Christoph Kaserer, examines an extensive dataset of infrastructure and non-infrastructure deals and finds that the data does not back up the conventional wisdom that infrastructure investments offer long-term, stable and predictable, inflation-linked returns with low correlation to other assets.  Unfortunately, while the study was an ambitious effort by the Center for Private Equity Research (CEPRES), the authors’ dataset of infrastructure deals fails to consider the distinction between asset classes and therefore, we believe, one can hardly argue that it is an accurate representation of the global infrastructure asset class as it has been defined by most institutional investors.

The authors draw from what is perhaps the most complete dataset of infrastructure investment performance ever assembled—that of CEPRES, a private consulting spin-out of the University of Frankfurt that is also supported by Technische Universität München and Deutsche Bank Group.  The 363 infrastructure deals and 11,223 non-infrastructure deals studied are all unlisted pure equity deals done by private equity firms with an initial investment and ultimate exit between 1971 and 2009.  Uniquely, the authors had access to the full history of cash flows for each. Through a series of regressions, they compare the infrastructure and non-infrastructure deals. Among their most interesting findings: infrastructure deals do not have a longer time horizon, do not provide stable cash flows, do not have inflation-linked returns, do correlate with other kinds of assets, and are also somehow low risk and high return.

What’s going on here? If these findings are true, they turn the entire world of infrastructure investing on its head.

Looking closer at the data, drawn primarily from North America (43%) and Europe (43%), the dataset is largely telecommunications (58.7%) and natural resources (24.8%). This is a heavy skewing. Typically, both natural resources and telecom investments have full demand risk and lack contractual and regulatory protections that give infrastructure its hallmark downside protection. Social infrastructure is absent from the sample, and it is unclear what is included in natural resources. (Do oil and gas exploration and production count as infrastructure?) Additionally, 52.9% of the deals are labeled as venture capital, which clearly diverges from the risk/return profile of mature operating infrastructure companies. On the whole, we conclude that this dataset of infrastructure deals is poorly circumscribed.

Admittedly, the industry’s lack of a universal definition of infrastructure is partially at fault for this. Still—to paraphrase Justice Potter Stewart’s 1964 concurring opinion in Jacobellis vs. Idaho—I know infrastructure when I see it, and it is not this. A more focused cross-section of investment performance might suggest a conclusion more in line with the traditional investment profile of infrastructure assets.

Ultimately, the most significant legacy of this paper could be broader awareness of the dataset behind it. Future researchers should be able to carry out more segmented and nuanced investigations of performance of subsets of the infrastructure asset class, which should yield more precise conclusions.

Dr. Ryan Orr is Executive Director of the Collaboratory for Research on Global Projects at Stanford University. He teaches Global Project Finance and Infrastructure Investment to students in the university's Engineering School and Graduate School of Business.