Building benchmarks for infrastructure investors
EDHECinfra was created to address the profound knowledge gap faced by infrastructure investors by collecting and standardising private investment and cash flow data and running state-of-the-art asset pricing and risk models to create the performance benchmarks that are needed for asset allocation, prudential regulation and the design of infrastructure investment solutions.
Majid Hasan
Senior Reserarch Engineer
majid.hasan@edhec.edu
#asset pricing
Sharyn Chang
Infrastructure Data Analyst
sharyn.chang@edhec.edu
Qi Wang, PhD
Senior Reserarch Engineer
qi.wang@edhec.edu
"statistics alive!"
"the infrastructure adventure"
Frederic Blanc-Brude, PhD
Director
frederic.blanc-brude@edhec.edu
Christy Tran
Infrastructure Data Analyst
christy.tran@edhec.edu
"Mr Data"
Tim Whittaker, PhD
Associate Research Director
tim.whittaker@edhec.edu
"I build bridges too."
Grace Chen
Senior Relationship Manager
grace.chen@edhec.edu
Aurelie Chreng
Senior Reserarch Engineer
aurelie.chreng@edhec.edu
#portfolio design
Jing-Li Yim
Infrastructure Data Analyst
jing-li.yim@edhec.edu
Terrance Teoh
Information Systems Manager
terrance.teoh@edhec.edu
Silvia Garcia
Infrastructure Data Analyst
silvia.garcia@edhec.edu

What we do

Collecting and analysing data

We collect, clean and analyse the private infrastructure investment data of the project’s data contributors as well as from other sources, and input it into EDHECinfra’s unique database of infrastructure equity and debt investments and cash flows.We also develop data collection and reporting standards that can be used to make data collection more efficient and reporting more transparent. This database already covers 15 years of data and hundreds of investments and, as such, is already the largest dedicated database of infrastructure investment information available.

Designing cash flow and discount rate models

Using this extensive and growing database, we implement and continue to develop the technology developed at EDHEC-Risk Institute to model the cash flow and discount rate dynamics of private infrastructure equity and debt investments and derive a series of risk and performance measures that can actually help answer the questions that matter for investors.

Building reference portfolios of infrastructure investments

Using the performance results from our asset pricing and risk models, we can report the portfolio-level performance of groups of infrastructure equity or debt investments using categorisations (e.g. greenfield vs brownfield) that are most relevant for investors’ investment decisions.

International Advisory Board

As is the case for all research at EDHEC, the work of the Infrastructure Institute is the object of a double validation process: academic validation (internal and external) and industry validation. The second leg of this process is driven by an Advisory Board of high-level individuals who are convinced about the role of applied research in financial practice and committed to providing feedback and to steer the research of the Institute.

  1. AIA, Mark Konyn (Chief Investment Officer) & Mr Tan Soo Thiam (Regional Director of Investment Management – Fixed Income Group Investment)
  2. Abu Dhabi Investment Council, Adriaan Ryder (Chief Investment Strategist)
  3. AP2, Tomas Franzen (Chief Strategist)
  4. Aviva, Ian Berry (Head of Infrastructure) & Laurence Monnier, Head of Strategy and Research
  5. Caisse de Depots du Quebec, Dave Brochet (Head of Infrastructure Risk and Managing Director, Asia)
  6. Clifford Capital, Premod Thomas (Head of Strategy) & Richard Desai (Chief Risk Officer)
  7. FWD Life, Paul Carrett (Group Chief Investment Officer)
  8. Government Investment Corporation of Singapore, Chia Tai Tee (Chief Risk Officer) & Tan Hsiao Mein (Senior VP, Risk & Performance)
  9. Kernmantle Advisors, Ajay Sawhney (CEO)
  10. NTUC Income, Mark Wang (Senior Vice President & Chief Investment Officer)
  11. OPTrust, James Davis (Chief Investment Officer)
  12. Prudential (Eastspring), Tony Adams (Head of Infrastructure)
  13. QSuper, Brad Holzberger (Chief Investment Officer)
  14. Sun Life Financial Asia, Sancho Chan (Chief Investment Officer and Head of ALM)
  15. Swiss Life, Christoph Manser, Head of Infrastruture
  1. Noel Amenc, Professor, EDHEC Business School & CEO of Scientific Beta
  2. Robert Bianchi, Associate Professor, Griffith University
  3. Antonio Estache, Professor, Université Libre de Bruxelles
  4. Stefano Gatti, Professor, Bocconi University
  5. Timo Valila, Visiting Professor, UCL
  1. IE Singapore, Taik Him Chua (Deputy-CEO)
  2. Global Infrastructure Hub, Chris Heathcote (CEO) & Brer Adams (Director)
  3. OECD, Andre Laboul (Deputy Director, Directorate for Financial and Enterprise Affairs) & Raffaelle Della Croce (Lead Manager, Long-term Investment Project)
  4. Monetary Authority of Singapore, Bernard Wee (Executive Director, Financial Markets Development Department)
  5. World Bank, Jordan Schwartz (Director, World Bank, Singapore) & Cledan Mandri-Perrot (Head of Infrastructure Finance and PPPs)
  1. Campbell-Lutyens, John Campbell (Chairman) & Conrad Yan (Partner)
  2. Long-Term Infrastructure Investors Association, Thierry Deau (Chairman)
  3. Meridiam, Julia Prescott (Chief Strategy Officer)
  4. NATIXIS, Anne-Christine Champion (Global Head Portfolio Management)

EDHECinfra Research Associates


The benchmarks: Frequently Asked Questions

Badly. In its 2016 survey of infrastructure asset owners and managers , we found that the immense majority of investors think that current options for benchmarks are utterly lacking.

Not really.

On the private debt side – and infrastructure is mostly about debt –very little usable benchmarks exist as the recent debates around the calibration of Solvency-II have shown. Rating agencies have documented default frequencies in certain classes of private project debt but this is insufficient to arrive a full-scale valuation results of the kind that are required to build investment benchmark.

One the equity side, investors have been using combinations of public equity indices – so-called “listed infrastructure” – and indices created by average the internal rates of return of private assets. But neither are good solutions.

In the first case (listed infrastructure indices), it fails to capture the characteristics of private infrastructure projects. In general, these indices have proven to be extremely concentrated in a few stocks and to be more volatile than the market as a whole.

In a recent paper , we show that for a large, well-diversified investor exposed to capital market instruments available globally, in the US or in the UK, adding “infrastructure” to the portfolio does not shift the efficient mean-variance frontier. In other words, it does not improve diversification.

From a factor investing perspective, an “infrastructure sector” filter applied to public equities is unlikely to be the most efficient way to gain exposure to remunerated risk factors. The question is not “what is the infrastructure investor factor?” but “what combination of remunerated factor exposures can be accessed through infrastructure investments?”

The second option, which consists of averaging the internal rate of return of a group of private projects is even more problematic. First, these rates of return spring from the net asset value reported for individual assets. As is well documented in the academic literature, appraisal-based net asset values tend to be very “stale” – they seldom change because they embody a simple cash flow forecast and an ad hoc choice of discount rate which does not necessarily reflect the riskiness of the investment. So return volatility is “smoothed” and you end up with promises of high returns with little or no risk, which is dubious.

Second, as every corporate finance textbook can attest, you cannot simply average IRRs!

More generally, the benchmark construction part of the question is completely ignored with this approach. For all we know the reported performance could be that of a single large asset in the basket.

EDHECinfra is developing explicit portfolio construction rules for benchmarks of highly illiquid assets.

The methodology is very simple.

To create investment benchmarks in private infrastructure debt or equity, we first build a statistical model of the cashflows of each instrument. The model derives the expected value but also the volatility of cash flows.

Next, we build discount rate term structures as a function of by the conditional volatility of cash flows to either equity or debt investors. Hence, we can price and measure the periodic returns of any of these assets.

Finally, these assets are aggregated into reference portfolios that capture the systematic dimensions of infrastructure investments: the type of infrastructure “business model” (whether income is contracted, merchant or regulated), the relevant moment in their lifecycle (greenfield vs brownfield infrastructure) and different geographic segments (OECD vs EM) corresponding to a broader level of political and regulatory risk.

We will publish several benchmarks. Initially they will embody different aspects of what it means to be investing in infrastructure: different moments in the investment lifecycle (Greenfield vs. Brownfield), different type of infrastructure business models (Contracted, Merchant or Regulated), or different jurisdictions, either as a debt investor or an equity investor.

In due course, such benchmarks can also be designed to help address broader investment solutions e.g. aim to achieve a combination of duration and risk-adjusted performance.

Only in part. The (term structures of) discount rates applied to each expected cash flow are “filtered” from the investment data (the initial price) observed in each actual investment and the conditional volatility of these cash flows, which is itself calibrated using a combination of model prior and observable data.

By design, measuring the performance of assets that are very seldom traded requires a degree of marking to model. The approach retained by EDHECinfra makes the best possible of available data by integrating sequentially and “blending” it with any assumptions made ex ante about the nature of the cash flow process. In computer science this is referred to as Machine Learning.

We strongly believe that this is the most adapted approach to measuring performance in long-term, illiquid investments.

Infrastructure projects are unique, but so are most firms. This does not preclude the objective to invest across a group of firms for investors nor does invalidate the possibility to measure the contribution of that group to the overall portfolio.

The question at hand is: does focusing on infrastructure investments isolate a risk-adjusted profile that can improve the outcome of the investment process for asset owners?

In terms of measuring risk across of range of arguably heterogenous infrastructure firms, we first measure the volatility and correlations of cash flows streams to debtors or investors. This drives the estimation of discount factors, which in turn determines the measures of financial performance. These metrics are measured by groups or families of projects, which have similar profiles both ex ante (same financial structure) and ex post (same realised volatility).

But through the implementation of advanced statistical techniques (e.g. particle filters), we can also track individual assets and how they evolve between different “states” or “regimes” of their cash flow process and what the implications are for valuation and performance measurement in time. For instance, individual projects may experience a drop in expected cash flows and in cash flow volatility at the same time.

The existence of reasonably homogenous groups of investments by risk profile allows building expectations ex ante. The possibility to track their behaviour on an individual basis is what makes the updating benchmarks of illiquid assets possible.

Our approach improves on the typical valuation exercise in several ways. Typically, the valuation of private investment projects is done using a cash flow forecast (or base case) derived from the financial model of the project. This forecast is not the necessarily the most likely (the expected value) series of future cash flows. It is also the result of a static model usually created before the investment starts and relying on numerous inputs about development and operating costs.

Such models are hard to update. They would require updating values for all realised inputs and revisiting the output of the model, including how the project might have changed in nature or structure since the model was made. This is typically not possible.

Hence valuation rely on ad hoc cash flow projections, which may become more and more obsolete as the investment lives its life.

Next, these ad hoc forecasts are discounted using flat (constant) rates of return. The choice of the discount factor is necessarily subjective and research has shown that investment managers can have incentives to use discount rate opportunistically. Even if they are not manipulated, the choice of discount rate is the reflection of a belief or expectation but is not grounded in actually measures of the riskiness of such investments (i.e. the volatility of cash flow streams).

In the case of infrastructure investment, we expect ex ante (before the fact) that the volatility of cash flows will evolve from higher to lower risk, due to the sequential resolution of uncertainty that characterises most ring-fenced or relationship-specific investment. Hence, we should be using a term structure of discount factors, not a flat rate.

Our approach improves current methods on both counts:
– We have developed techniques designed to transform ad hoc cash flow projections into full-fledged statistical forecasts, including a conditional (before the fact) measure of cash flow volatility;
– Using this measure of the volatility of cash flows, we can derive full term structures of of discount factors to be applied to these cash flows, thus taking into account the expected change in risk profile of infrastructure investments

Einstein once remarked that compounding interest is the eighth wonder of the world.

Indeed, improve our estimation of the discount factors of cash flow streams that extend far into the future can have a significant on valuations and therefore on reported performance and risk.

Such performance metrics can be measured annually. This is long Term Investment indeed and there is no need to update performance measures quarterly when the underlying data (the accounts) is mostly available on an annual basis only.

Data collection: Frequently Asked Questions

EDHEC infra collects cash flow data and the characteristics of underlying infrastructure projects and ”single-asset” firms and utilities.

We do not collect data on investment funds or cash flows in and out of investment funds.

The relevant data falls into three categories:

  1. Data corresponding to investments’ characteristics and the systematic determinants of their financial performance;
  2. Forecast equity and debt cash flow data;
  3. Realised equity and debt cash flow data.

Further details about the EDHECinfra data collection template can be here.

EDHEC infra collects several types of cash flow data from individual infrastructure investments

  1. Realised and forecast cash flows from long-term investors (as well as project characteristics, timeline and events)
  2. Realised and forecast debt service and ratios from creditors and bond holders
  3. Historic data from statements of accounts and annual reports
  4. Business case/ nancial close models through freedom-of-information requests
  5. Project descriptions from open-source and commercial sources

EDHEC infra saves contributors data using a dedicated, high-security system:

  1. All data is saved on stand-alone servers, separate from any webserver. These servers are back-up on several external, asynchronous drives at regular intervals.
  2. Off-line data entry is performed in EDHECinfra’s offices in Singapore by data analysts who never have access to the database itself.
  3. On-line data entry can be performed by contributors to update historical records or create a new project entry in the database through a secure online interface accessing the EDHECinfra database through a 128-bit encrypted SSL ”tunnel”, guaranteeing maximum security to save contributed data into the remote database. The data entered by contributors does not stay on the webserver. Historical data from the database is never uploaded to the webserver
  4. Only validated and publicly available benchmarking reports in PDF format are stored on the EDHEC infra webserver.
EDHEC is committed to preserving the full confidentiality of the data contributed by infrastructure investors and creditors and enters into a non-disclosure agreement with each of them. Raw contributed data is never shared or distributed to anyone.
Any private data contributed by infrastructure investors and creditors remains their entire property. EDHECinfra is the stewart of this information.

Any analysis performed using the database (such as calibrating volatility models) is not the property of data contributors but that of EDHECinfra.