INFRASTRUCTURE VALUATION

Transcription

1 INFRASTRUCTURE VALUATION A guide to the valuation of privately held infrastructure equity and debt By Frédéric Blanc-Brude and Majid Hasan EDHEC-Risk Institute

2 Published in March 2015 by PEI 6th Floor 140 London Wall London EC2Y 5DN United Kingdom Telephone: +44 (0) PEI ISBN This publication is not included in the CLA Licence so you must not copy any portion of it without the permission of the publisher. All rights reserved. No parts of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means including electronic, mechanical, photocopy, recording or otherwise, without written permission of the publisher. Disclaimer: This publication contains general information only and the contributors are not, by means of this publication, rendering accounting, business, financial, investment, legal, tax, or other professional advice or services. This publication is not a substitute for such professional advice or services, nor should it be used as a basis for any decision or action that may affect your business. Before making any decision or taking any action that may affect your business, you should consult a qualified professional adviser. Neither the contributors, their firms, its affiliates, nor related entities shall be responsible for any loss sustained by any person who relies on this publication. The views and opinions expressed in the book are solely those of the authors and need not reflect those of their employing institutions. Although every reasonable effort has been made to ensure the accuracy of this publication, the publisher accepts no responsibility for any errors or omissions within this publication or for any expense or other loss alleged to have arisen in any way in connection with a reader s use of this publication. PEI editors: Wanching Leong and Helen Lewer Production editor: Julie Foster Printed in the UK by: Hobbs the Printers (

3 Contents Figures and tables About the authors vii xi 1 Introduction Why infrastructure? The demand for benchmarks A roadmap This book 6 PART I: UNDERSTANDING PRIVATELY HELD INFRASTRUCTURE INVESTMENTS 2 The nature of investable infrastructure Existence of investable infrastructure The role of long-term contracts The role of commitment The contractual nature of underlying infrastructure investments Creating and regulating investable infrastructure Monopoly privatisation and regulation Project finance and corporate governance Conclusion: Infrastructure assets are not real assets 19 3 Infrastructure project financing Investment delegation and project financing Definition Agency and monitoring in project finance A response to the long-term investment problem Project financing as risk optimisation Conclusion: Infrastructure project finance as a reference asset 35 4 Risk and the cost of capital in infrastructure projects Construction risk Leverage Revenue risk Contract renegotiation Political risk Conclusion: Systematic sources of risk in project finance 47 iii

4 Contents PART II: ISSUES WITH LONG-TERM INVESTMENT PERFORMANCE MEASURES 5 Benchmarking infrastructure investments: Objectives and challenges Why benchmark infrastructure investments? Two challenges 58 6 Addressing the challenges of long-term investment benchmarking Bayesian inference for long-term investment benchmarking Towards intersubjective pricing measures Families of generic infrastructure projects 65 PART III: MEASURING THE PERFORMANCE OF INFRASTRUCTURE PROJECT DEBT 7 Objectives and characteristics of infrastructure project debt Objectives Defining infrastructure debt Observable asset values Covenants and embedded options Identifying default triggers Reorganisations Illiquidity and lumpiness Documented performance Conclusion 82 8 Approaching the valuation of infrastructure debt Asset-pricing models Corporate debt valuation models Proposed approach Conclusion 93 9 Valuation framework Intuition Cash flow dynamics Structural framework Debt restructuring model Restructuring upon a technical default Restructuring upon a hard default Model implementation Cash flow model Risk-neutral measure Debt rescheduling upon default 117 iv

5 Contents 11.4 Algorithm Results 122 PART IV: VALUING EQUITY INVESTMENTS IN INFRASTRUCTURE PROJECTS 12 Objectives and approaches Objectives Existing approaches to private equity valuation Proposed approach Expected dividend model A Bayesian framework Cash flow metrics and risk measures Dividend state transitions (Conditional) dividend payment distribution Prior elicitation Valuation framework Modern asset-pricing theory A state-space framework Observation (pricing) equation Pricing equation Matrix formulation Observation equation State (discounting) equation Term structure model Matrix formulation State equation Filtering of the term structure 181 State-space model Filtering and updating Parameter estimation Model outputs Conclusion Model implementation Prior elicitation Equity cash flow projections Results Return measures Risk measures Market dynamics Sensitivity to conditional dividend volatility estimates 205 v

6 Contents PART V: CONCLUSIONS 19 Conclusions Key points Data collection and standardisation Next steps 215 PART VI: TECHNICAL APPENDICES A Debt valuation 219 A.1 Risk-neutral measure 219 A.2 Risk measures 224 A.3 Return measures 226 B Equity valuation 227 B.1 Cash-flow metrics 227 B.2 Bayesian estimates 229 B.3 Pricing equation 234 Bibliography 243 About PEI 256 vi

7 Figures and tables Figures Figure 3.1: Simplified project contractual claims structure 24 Figure 3.2: Figure 4.1: Figure 6.1: Figure 6.2: Figure 7.1: Example of base case cash flows and debt service cover ratio for a 35-year old project company with contracted revenues 26 Difference between ex ante and ex post construction costs in traditional infrastructure procurement and project finance 39 Box plots of the average DSCR in projects with contracted, partly contracted and fully merchant revenues 68 Box plots of the tail size in projects with contracted, partly contracted and fully merchant revenues 68 Regional shares of the cumulative project finance deal flow and project bond issuance, Figure 7.2: Marginal probability of default in a sample of project finance loans 82 Figure 9.1: Black-Cox decomposition 101 Figure 10.1: Renegotiation and exit values immediately following a hard default 107 Figure 10.2: Outcome of renegotiation post hard default 109 Figure 11.1: DSCR models for merchant and contracted infrastructure 112 Figure 11.2: Risk-neutral distance to default for rising and flat DSCR 116 Figure 11.3: Risk-neutralised DSCR distributions for merchant and contracted infrastructure projects 116 Figure 11.4: Debt rescheduling upon a technical default 118 Figure 11.5: Debt rescheduling upon a hard default 119 Figure 11.6: Flow chart of the determination of cash flows to debt holders 120 Figure 11.7: Comparison of mean debt payments and CFADS 123 vii

8 Figures and tables Figure 11.8: Comparison of the probability of technical or hard default, hard default only and probability of death for rising and flat DSCR 123 Figure 11.9: Comparison of EL, VaR and cvar for rising and flat DSCR 124 Figure 11.10: Exit value of lenders and total firm and debt values 126 Figure 11.11: Probability of lender and equity haircuts (of any size) 127 Figure 11.12: Loss given default as a percentage of the value of debt 128 Figure 11.13: Loss given default in absolute terms for a $1,000 investment 128 Figure 11.14: Recovery rates 129 Figure 11.15: Time evolution of duration 129 Figure 11.16: Trade-off between credit and interest rate risk 130 Figure 11.17: Comparison of yield and z-spread for rising and flat DSCR 131 Figure 11.18: Range of yields for rising and flat DSCR 132 Figure 13.1: Prior and posterior densities of the beta distribution of π 11 over 12 iterations using 50 data points 152 Figure 13.2: Prior and posterior densities of the beta distribution of π 01 over 12 iterations using 50 data points 153 Figure 13.3: Prior and posterior estimates of the probability of observing a positive dividend at time t 154 Figure 13.4: Prior and posterior densities of the log-normal distribution of ESCR t over 12 iterations using 50 data points 157 Figure 13.5: Prior and posterior densities of the log-normal distribution of ESCR t over 12 iterations using 50 data points 158 Figure 13.6: Prior and posterior densities of the m and p parameters of the log-normal distribution of ESCRt 159 Figure 15.1: Investment and cash flow timeline 170 Figure 17.1: State-space predicting and updating procedure with Kalman filtering 184 viii

9 Figures and tables Figure 18.1: How prior knowledge of infrastructure project finance can be combined with observable data to derive posterior parameters of the distribution of dividends 190 Figure 18.2: Assumed DSCR dynamics for merchant and contracted infrastructure 193 Figure 18.3: Base case and expected free cash flow (CFADS) and dividend payments 194 Figure 18.4: Expected dividends as a multiple of the first non-zero dividend 194 Figure 18.5: Probability of default, lock-up and bankruptcy 195 Figure 18.6: Probability of non-zero dividend payment 195 Figure 18.7: Mean and volatility of ESCR t 196 Figure 18.8: Unconditional density of ESCR t at different times in the project s life cycle 196 Figure 18.9: Range of forward period risk premia of expected cash flows at t = Figure 18.10: Range of term structures of discount rates of expected cash flows at t = Figure 18.11: Range of stochastic discount factors of expected cash flows at t = Figure 18.12: Range of prices during project s life cycle at t = Figure 18.13: Forward curves of period equity risk premiums over the project s life 201 Figure 18.14: Term structure of discount rates over the project s life 202 Figure 18.15: Cash yield or dividend/price ratio in each period, conditional on the base case being realised in the previous periods 203 Figure 18.16: Evolution of equity IRR across the project life cycle 203 Figure 18.17: Expected and extreme loss measures at t Figure 18.18: Evolution of effective duration 204 Figure 18.19: Filtered estimates of the IRR and transaction prices 205 Figure 18.20: Filtered prices and term structure with varying expectations of dividend volatility 206 Figure 19.1: Building blocks of infrastructure project finance debt cash flow dynamics 213 ix

10 Figures and tables Tables Table 4.1: Table 10.1: Differences between ex ante and ex post construction costs in traditional infrastructure procurement and project finance 40 Descriptions and symbols for the variables used in the renegotiation model 106 Table 11.1: Merchant and contracted infrastructure characteristics 112 Table 11.2: DSCR models for merchant and contracted infrastructure 112 Table 13.1: True, prior and posterior values of example transition probabilities at time t 151 Table 13.2: True, prior and posterior values of example dividend distribution parameters at time t 151 Table 18.1: Merchant and contracted infrastructure project financing characteristics 193 Table 18.2: DSCR models for merchant and contracted infrastructure 193 Table 19.1: Data collection requirements 214 Table B.1: Table B.2: Table B.3: Posterior mean and variance of example transition probabilities at time t in each observation round 231 Posterior mean and variance of example ESCR distribution parameters (log-scale) at time t in each observation round 233 Comparison of filtered prices for different choices of initial values of and ø 241 (γ t+i ) T i=1 x

11 About the authors Frédéric Blanc-Brude is research director at EDHEC-Risk Institute and heads its thematic research programme on infrastructure investment. Prior to joining EDHEC, he worked for ten years in the infrastructure finance sector and was actively involved in transactions representing a cumulative value of more than $6 billion in Europe, Asia and the Middle East. Frédéric also teaches the infrastructure investment seminar of the Yale School of Management, EDHEC-Risk Institute Certificate in Risk and Investment Management. Frédéric holds a PhD in Finance from King s College London. frederic.blanc-brude@edhec.edu Majid Hasan is a research assistant at EDHEC-Risk Institute and a PhD in Finance candidate at EDHEC Business School. His research interests include asset pricing in imperfect markets and designing solutions to channel private capital towards public infrastructure. Majid holds a Masters in Applied Mathematics from Western University, Canada, and a Bachelors in Physics from the University of Punjab, Pakistan. Before joining EDHEC, Majid worked as a research and teaching assistant at Western University, and the University of BritishColumbia, Canada. The authors would like to thank the following sponsors for their support of this research: Meridiam Infrastructure, Campbell Lutyens, Natixis and the Monetary Authority of Singapore. n xi

12

13 1 Introduction 1.1 Why infrastructure? Matching the huge demand for capital in infrastructure projects around the world with the available supply of long-term funds by institutional investors whether pension funds, insurers or sovereign wealth funds has never been so high on the international policy agenda. This momentum, illustrated by the G20 s focus on long-term investment in infrastructure, coincides with the steadily growing investment appetite from investors for unlisted and illiquid assets. However, full-fledged investment solutions demonstrating the benefits of long infrastructure investment for institutional investors remain elusive, and documenting the investment characteristics of long-term investment in infrastructure has become a pressing issue. Long-term investment in infrastructure is part of a broader trend among institutional investors to improve portfolio diversification or seek higher returns through alternative investments, to invest increasingly outside of public capital markets, to find sufficiently long-dated instruments with a more attractive performance than government bonds, and to invest in inflation-linked securities other than low-yielding Treasury Inflation-Protected Securities (TIPS). One of the salient features of these emerging investment choices is the decision to buy assets that are infrequently traded and hold them until maturity. The growing interest of investors for infrastructure investment is motivated by what we call the infrastructure investment narrative (see Blanc-Brude, 2013c). This is the notion that infrastructure investments uniquely combine the following characteristics: The provision of essential services, hence low correlation with the business cycle. Low price elasticity of demand for service and a degree of monopoly power, hence pricing power that provides an inflation hedge. Predictable and substantial free cash flow. Attractive risk-adjusted cash flows, available over long periods of time. Access to unlisted, illiquid financial assets. Therefore, investing in infrastructure implies: Improved diversification. Better liability hedging, including inflation protection. Less volatility than capital market instruments. Today, the infrastructure investment narrative is the only available benchmark. However, this narrative is defined a priori and has not been subject to a rigorous and comprehensive 1

14 Introduction performance measurement exercise. Investors considering investing in infrastructure need to form return and risk expectations to make allocation decisions. In this book, we develop asset valuation methodologies and data collection templates that aim to contribute to the effective benchmarking of privately held infrastructure project debt and equity. 1.2 The demand for benchmarks The need for benchmarks of privately held investments in infrastructure may seem incongruous at first. After all, infrastructure projects are lumpy and highly idiosyncratic endeavours. If every infrastructure project is different, what can investors learn from a benchmark? In modern finance, asset allocation is not about picking individual investments, but instead focuses on investing in groups of reasonably homogenous assets giving access to remunerated risk factors. The performance of each of these groups can be evidenced by a benchmark. In effect, the narrative described above is also a model that describes the expected characteristics of the average investment in infrastructure. In turn, individual investments in specific jurisdictions, relying on one form or another of contractual or regulatory arrangement, may only have some or none of these characteristics. However, assessing the opportunity to increase allocations to privately held infrastructure requires that investors understand the risks and performance to expect over time and in different economic environments, and that regulators understand the risks investors are taking. Benchmarking the expected behaviour of privately held infrastructure investments has become necessary to allow investors to fully integrate such financial assets into their asset-liability management exercises, as well as to allow regulators to calibrate the riskbased regulatory frameworks that make these investments possible (or not) for large institutional investors. The objective of building long-term investment benchmarks is simply to create the performance measures that investors and regulators need. These measures include the expected return that reflects the dynamic risk profile of infrastructure projects and adequate risk measures for portfolio management (such as conditional value at risk), as well as correlation measures between infrastructure investments and between baskets of infrastructure investments and other types of assets. The information created by such benchmarks will be instrumental in matching the supply and demand of long-term capital. None of these measures are currently available to long-term investors in infrastructure. Indeed, creating relevant performance measures is not without difficulties. Issues include the absence of readily available and comparable data, and the limitations of using listed proxies to benchmark lumpy and privately held financial assets like infrastructure equity or debt. 2

15 Introduction To achieve this objective, we have designed a roadmap describing a series of applied research and data collection steps to arrive at the desired measures of infrastructure investment performance. 1.3 A roadmap Benchmarking long-term investments in infrastructure requires a two-level approach, starting with underlying instruments and then documenting the behaviour of different portfolios built with such instruments (see Blanc-Brude, 2014a, for a discussion). Documenting underlying instruments At the underlying level, five steps are necessary to clarify and document the performance of infrastructure financing instruments, both equity and debt: 1. Define the relevant financial assets Improving the benchmarking and regulation of any type of investment first requires well-defined underlying instruments. As infrastructure investment is currently illdefined, the first step of our roadmap is the creation of unambiguous definitions of financial instruments for long-term investment in infrastructure. Indeed, as we show in Chapter 2 infrastructure assets are not real assets but financial contracts. From an asset allocation perspective, industrial classifications such as roads or power are close to useless. A first solution to the absence of a widely agreed on definition of infrastructure, which we develop in Chapter 3, is to focus on project finance debt and equity as defined in the Basel II Accord. Other approaches to infrastructure investment at the underlying level must also be developed, as long as they refer to well-identified financial instruments (for example, the equity capital of certain types of regulated network operators). 2. Design adequate valuation and risk measurement methodologies Once a clear and broadly accepted definition of underlying instruments is in place, adequate valuation and risk measurement methodologies that take into account the infrequent trading of most underlying infrastructure equity and debt can be developed. By adequate we mean that such methodologies should rely on the rigorous use of asset pricing theory and statistical techniques to derive the necessary input data, while aiming for parsimony and realism in data collection. The proposed methodologies should lead to the definition of the minimum data requirement (MDR), which is necessary to derive robust return and risk estimates. 3. Determine the data collection requirements While ensuring theoretical robustness is paramount to the reliability of performance measurement, a trade-off exists with the requirement to collect real world data from market participants. In particular, proposed methodologies should aim to minimise the number of inputs in order to limit the number of parameter estimation errors. Adequate models should also focus on using data points that are known to exist and are already collected/monitored, or could be collected reasonably easily. In all cases, 3

16 Introduction data requirements should be derived from the theoretical framework, not the other way around. In reality, the amount of available data initially will be limited both in scope, since not all types of infrastructure projects exist in large numbers, and by time frame, because infrastructure investments may have multi-decade lives and available records are unlikely to span such periods. Such data paucity can also be addressed if models are designed to allow for learning. Whether the necessary data already exists or not, the determination of a parsimonious data set for asset pricing will also inform the standardisation of a new investment data collection and reporting framework. 4. Standardise performance reporting Standardising infrastructure investment data collection allows for the emergence of an industry-wide reporting standard, which can be recognised by investors and regulators alike as best practice. This reporting standard can increase transparency between investors and managers, who can be mandated to invest in a well-defined type of instrument and commit to report the relevant data. Adequate reporting will also maximise industry participation and reduce the cost of compliance. 5. Create a database of infrastructure equity and debt cash flows Once the required data and a standardised reporting/data collection template is identified, a database of infrastructure project cash flows can be built to apply the methodologies mentioned above. Initially, historical data can allow for documenting the past performance of welldefined infrastructure debt and equity instruments. Later, the ongoing collection of project cash flows can permit the production of regular updates of the known performance of such instruments over time. Once the adequate valuation and risk measurement methodologies have been determined for a given type of financial instrument, and data collection and reporting have been standardised, the benchmarking of long-term infrastructure investments can effectively take place by focusing on the relevant performance measures at the portfolio level. While a portfolio consisting of a representative basket of assets is the most intuitive benchmark, this approach is virtually impossible for investment in unlisted infrastructure debt and equity. Given currently available infrastructure investment vehicles, an investor cannot instantaneously buy a basket of assets that is representative of investable infrastructure projects in existence at that point in time. It may be possible to invest in such a representative basket over time, but this may take several years, by which time what constitutes a representative basket of infrastructure investments is likely to have changed with the evolution of public procurement policies. 4

17 Introduction Therefore, the most useful long-term investment benchmarks are likely to be a combination of well-documented building blocks capturing systematic risk factors found in portfolios of infrastructure instruments. Such benchmarks can correspond to well-identified investment strategies and combine the performance of the different groups of financial instruments available in infrastructure finance to optimally achieve this explicit target strategy. Building relevant portfolios At the portfolio level, three more steps are necessary to arrive at useful long-term investment benchmarks in infrastructure: 1. Identify building blocks A number of risk factors can be expected to systematically explain investment performance in infrastructure projects. For example, in Chapter 4, we show that the most important of such factors are the contractual features of individual investment projects, in particular to what extent they are exposed to commercial risks, as well as the development of the typical project life cycle, which, in infrastructure project finance, tends to correspond to the continuous deleveraging of the firm s balance sheet. Therefore, the risk/return profiles of most infrastructure project financing instruments can be grouped by revenue risk profile on the one hand and life cycle stage on the other. In other words, at any given point in time, infrastructure debt or equity can correspond to either what is typically known as greenfield or newly built assets, or brownfield or already existing and operating ones. The former are typically riskier and yield higher returns, the inverse is often true for the latter. The same debt and equity instruments can also correspond to, for example, toll roads and merchant power projects in the higher commercial risk/higher return category, and projects (also roads) that receive an income from the government based on the availability of service or those that have a take-or-pay offtake contract (possibly a power plant) in which their future income is independent of market conditions. Such projects exhibit less volatile cash flows and lower returns than those exposed to commercial risks. Therefore, any equity or debt instrument used to finance an infrastructure project can be categorised according to such a two-dimensional matrix of revenue risk profile and life cycle stage. While these are not the only systematic risk factors found in infrastructure projects, we review recent research in Chapter 4, which shows that such risk categories explain most of the variance in credit spread levels in project finance loans. Once the most homogenous subgroups of individual infrastructure finance equity and debt instruments have been identified, relevant investment strategies using these building blocks can be designed. 5

18 Introduction The statistical validation of these insights is a key step in the roadmap towards building an infrastructure investment benchmark, including ensuring that individual building blocks exhibit low levels of correlation between themselves. 2. Define relevant investment strategies As they comprise long-term illiquid assets, a basket of infrastructure projects is not easily or instantly investable. However, the building blocks discussed above can be combined to meet investors individual long-term objectives. For example, Blanc-Brude and Ismail (2013b) show that combining a block of greenfield debt with one of brownfield debt can create substantial diversification benefits (that is, increase returns and lower portfolio risk). Therefore, along the greenfield/brownfield spectrum there is a continuum of efficient portfolios that can serve as a reference point for investors looking to build a portfolio of infrastructure project debt using available instruments over a given period of time. For each strategy, and using the asset pricing and risk measurement methodologies discussed above, various return measures (period return, yield to maturity, return in excess of the investment base case), risk measures (expected loss, value at risk and expected shortfall), and (effective) duration can be computed and inform both the asset allocation process and calibration of prudential regulatory frameworks or internal risk models. 3. Investment benchmarks Finally, the strategies identified above can be used as long-term infrastructure investment benchmarks. Using historical data, the correlation of each strategy s performance with other investment opportunities, such as corporate debt or public equity, can be measured and estimated on an ongoing basis. This last step answers the all-important question of the diversification potential of individual strategies using infrastructure instruments as underlying assets. Implementing this roadmap requires an extensive data collection and modelling effort. Initially, historical data needs to be collected to calibrate valuation and risk models, and to provide a comparison with other asset types. Such benchmarks can also be computed on an ongoing basis not only to continuously inform investors asset allocation choices, but also provide them with performance assessment tools vis-à-vis infrastructure managers or their own direct investment programmes. The adoption of standardised performance reporting by market participants will be instrumental in this regard. 1.4 This book Objectives This book addresses the first three steps of our roadmap: it focuses on defining, pricing and documenting the behaviour of certain types of financial instruments widely used in infrastructure finance infrastructure project finance debt and equity. 6

19 Introduction It also provides a comprehensive review of existing academic research on the nature and performance of such instruments. It also develops valuation models and data collection templates that can be applied by the financial industry and lead to the creation of global investment benchmarks in privately held infrastructure. Each valuation framework is designed with data availability and collection costs in mind while focusing on applying state-of-the-art valuation approaches that allow computing the required investment metrics, (that is, not only expected returns but also risk measures that are essential from an asset allocation, asset liability or prudential perspective). Organisation This book is organised into five parts. Part I provides a comprehensive discussion of the theoretical and empirical literature on infrastructure investments, and highlights the nature of underlying assets and the determinants of systematic risks found in such investments. Part II reviews the rationale for building long-term investment benchmarks, outlines the empirical and theoretical challenges of valuing privately held infrastructure investments, and discusses several ways forward to building applicable yet theoretically robust valuation frameworks. Part III details the design and implementation of an infrastructure project debt valuation model using a structural credit risk approach that can easily be documented with existing creditor data, as well as a debt renegotiation model that allows valuing the embedded options found in infrastructure project loans, in particular, the value of their tail. Part IV proposes an application of the classic dividend discount model to the case of infrastructure project equity when complete time series of dividends are unavailable and market prices are not uniquely determined by an equivalent (replicating) portfolio of listed assets. Finally, Part V summarises our findings and presents the data collection requirements to implement the proposed valuation frameworks. n 7