PUBLISHED BY AUGUST 2016 VOLUME 1, NUMBER 12 A SUPPLEMENT TO PENSIONS & INVESTMENTS. EDHECinfra Infrastructure Special

Transcription

1 PUBLISHED BY AUGUST 216 VOLUME 1, NUMBER 12 A SUPPLEMENT TO PENSIONS & INVESTMENTS Infrasrucure Special

2 Pay for Wha you Ge Because we believe in he qualiy of our research and he robusness of he performance of he associaed smar bea indices, we are proposing ha invesors who wish o do so can replicae our flagship Scienific Bea Muli-Bea Muli-Sraegy indices on he basis of a pure performance fees mandae. Insiuional invesors who choose his opion from June 1, 216, will pay zero fixed fees and will only pay variable fees if he flagship Scienific Bea Muli-Bea Muli-Sraegy index ouperforms he reference cap-weighed index. For more informaion on his new pricing offer, please conac Mélanie Ruiz on or by a melanie.ruiz@scienificbea.com

3 A SUPPLEMENT TO PENSIONS & INVESTMENTS 3 INTRODUCTION CONTENTS Noël Amenc Associae Dean for Business Developmen, EDHEC Business School CEO, ERI Scienific Bea I is my pleasure o inroduce he laes issue of he Research for Insiuional Money Managemen supplemen o P&I, an infrasrucure special produced by he research cenre in Singapore. We presen he resul of he firs in-deph survey of insiuional invesors percepions and expecaions of infrasrucure invesmen. Almos wo hirds of surveyed insiuions declared ha hey waned o increase heir curren holdings of infrasrucure invesmens. The survey reveals some imporan evoluions and also imporan differences of perspecives, amongs invesors and also beween asse owners and managers. One of he key findings is ha invesors have no benchmarks and do no rus repored valuaions. In a shor aricle, we hen look a wheher an asse class of lised infrasrucure exiss. We do no manage o find lised proxies for infrasrucure asses. We conclude ha wha is ypically referred o as lised infrasrucure is no an asse class or a unique combinaion of marke facors. I canno be persisenly disinguished from exising exposures in invesors porfolios. Expecing he emergence of a new or unique infrasrucure asse class by focusing on public equiies seleced on he basis of indusrial secors is unlikely o be very useful for invesors. In a separae aricle, we find ha infrasrucure firms exhibi a unique business model in erms of revenue and profi dynamics compared o a large conrol group of public and privae firms. Infrasrucure firms have significanly lower volailiy of revenues and profis and pay a much higher proporion of heir revenues much more frequenly o heir owners, independen of he business cycle. We show ha a powerful saisical model of credi risk relying on Deb Service Cover Raio (DSCR) dynamics can be developed. This model provides imporan insighs for he valuaion of privae credi insrumens in infrasrucure projec finance. I also miliaes for sandardising he daa collecion and compuaion of iems such as he DSCR in infrasrucure projec finance, and for pooling his informaion in cenral reposiories where i can be used o creae he invesmen merics ha invesors need (and regulaors require) o be able o inves in large, illiquid asses such as privae infrasrucure projec deb. Having progressed owards clear definiions of underlying asses, and buil robus, sae-of-he-ar pricing and risk models ha avoid he pifalls of exising pracices, we feel ha i is now ime o collec he relevan informaion. The daa collecion framework and emplae ha we propose have been designed o correspond o he requiremens of he relevan asse pricing and risk models. Hence, a raionale exiss o collec daa effecively and efficienly o build infrasrucure invesmen benchmarks. We hope ha you find he aricles in he supplemen informaive and insighful. We wish you an enjoyable read and exend our hanks o P&I for heir collaboraion on he supplemen. INDEXES 4 Towards Beer Infrasrucure Invesmen Producs? 8 Looking for a Lised Infrasrucure Asse Class 1 Is Privae Infrasrucure Differen? Tracking Credi Merics in Privae Infrasrucure Deb 2 Daa Collecion for Infrasrucure Invesmen Benchmarking CONTRIBUTORS research@edhec-risk.com Noël Amenc Professor of Finance Frederic Blanc-Brude Direcor Grace Chen Senior Relaionship Manager Raffaelle Delacroce Lead Manager, Long-Term Invesmen Projec, Financial Affairs Division, Organisaion for Economic Cooperaion and Developmen (OECD) Majid Hasan Senior Research Engineer Cledan Mandri-Perro Head of Infrasrucure Finance and PPPs, World Bank Jordan Schwarz Direcor, Infrasrucure & Urban Developmen Hub, World Bank Qi Wang Senior Research Engineer Tim Whiaker Associae Research Direcor Simon Wilde PhD candidae, Universiy of Bah, Senior Managing Direcor Macquarie Capial

4 4 A SUPPLEMENT TO PENSIONS & INVESTMENTS INDEXES Towards Beer Infrasrucure Invesmen Producs? Frederic Blanc-Brude Direcor Grace Chen Senior Relaionship Manager Tim Whiaker Associae Research Direcor In a new paper sponsored by he Global Infrasrucure Hub (a G2 Iniiaive), presens he resul of he firs in-deph survey of insiuional invesors percepions and expecaions of infrasrucure invesmen. I documens he reasons for invesing in infrasrucure and wheher currenly available invesmen producs or sraegies are helping invesors mee hese objecives. The findings provide a firs sep owards inegraing infrasrucure in long-erm invesmen soluions. The survey repors he views of 184 individuals involved in infrasrucure invesmen; half of hem represen insiuional invesors or asse owners (insurers, pension plans and sovereign wealh funds), one-hird are infrasrucure asse managers and he remainder are infrasrucure invesmen specialiss from mulilaeral developmen banks, raing agencies and consulancies. Respondens are mosly senior execuives acives in he op managemen (CEO, board members 14.5 percen), sraegic (CIO, Head of ALM or Asse Mix 25.5 percen), invesmen (Head of Infrasrucure, invesmen direcor 46.2 percen) or oher (14.5 percen) funcions of he organisaions hey represen. Infrasrucure is popular. Almos wo hirds of surveyed insiuions declared waning o increase heir curren holdings of infrasrucure invesmens. Beyond ha unsurprising poin, he survey reveals some imporan evoluions and also imporan differences of perspecives, amongs invesors and also beween asse owners and managers. In wha follows, we summarise he findings of he survey and provide some elemens for discussion and fuure research. EXHIBIT 1 Inenion o inves in infrasrucure in he nex 3 o 5 years, asse owners only In he coming 3 5 years you inend o inves in infrasrucure... (asse owners only) 64.6% More 8.5% No Sure 1.2% No a all 1.2% Less 24.4% Keep allocaion unchanged INVESTMENT BELIEFS Key findings The main findings on asse owners and managers invesmen beliefs are: 1. There is wide disagreemen amongs respondens abou wheher lised infrasrucure equiy or deb qualifies as an asse class. However, unlised infrasrucure is widely considered o be a unique asse class, boh on he privae deb and privaely-held equiy sides; 2. Mos respondens also believe ha focusing on infrasrucure invesmen only makes sense if i can be defined as an asse class, whereas a minoriy repors preferring o approach infrasrucure as an invesable bundle of facor exposures; 3. Mos respondens perceive infrasrucure invesmen s unique feaure o be eiher is poenial for porfolio diversificaion or for harvesing risk premia, whereas i is less frequenly believed ha infrasrucure has unique ineres rae or inflaion hedging properies; 4. Invesors and managers define infrasrucure in erms of long-erm conracual arrangemens and monopoly regulaion and acknowledge ha indusrial secors are a much less informaive way o caegorise such invesmens. In he same spiri, he sabiliy of long-erm conracs and he role of couner-pary risk are perceived o be he mos imporan and unique characerisics of infrasrucure firms (compared o oher firms). Finally, brownfield (exising) and conraced infrasrucure is repored o be he mos aracive o invesors, closely followed by brownfield regulaed uiliies; 5. Expeced reurns follow a clear paern deermined by he business model (conraced, merchan or regulaed) and he lifecycle (greenfield or brownfield) of infrasrucure firms, wih greenfield merchan invesmens requiring higher reurns han brownfield regulaed and conraced infrasrucure; 6. Despie viewing infrasrucure as characerised by long-erm sable conracs and being mos aracive once i has been buil, mos invesors and heir managers expec relaively high reurns. A majoriy considers ha infrasrucure asses should no be expensive and requires equiy reurns ranging from he high single digis o he low eens. Asse managers sysemaically repor higher expeced reurns han asse owners. 7. More han half of paricipaing asse owners declare ha hey currenly inves in emerging markes or wan o, and ha hey are willing o increase heir curren allocaion. SWFs and pensions plans are he mos involved and willing ypes of invesors invesing or aiming o inves emerging marke infrasrucure; 8. The main repored reasons o expand ino emerging marke infrasrucure are higher reurns and counry risk diversificaion, while he main concerns of invesors are public policy reversals and he enforceabiliy of conracual claims. 9. Required reurns in emerging markes are higher han in OECD markes, bu oherwise follow he same paerns. However, he emerging marke premium on reurns varies for differen ypes of infrasrucure projecs: invesmens in he conraced and regulaed caegories command much higher spreads (above he OECD required reurns), paricularly a he brownfield sage, whereas emerging marke merchan risk is perceived o be almos equivalen o OECD merchan risk. From homogenous o heerogenous beliefs These resuls highligh he degree o which invesors agree or disagree abou wha o expec from invesing in infrasrucure equiy or deb. Infrasrucure has long been considered difficul o define as an invesmen proposiion bu a consensus view is emerging amongs marke paricipans abou he naure of infrasrucure business, and wha drives risk and performance in such invesmens. Hence, wha qualifies or no as infrasrucure is now beer undersood, as he recen debae around he definiion of infrasrucure invesmen in he conex of he Solvency-II direcive has shown. As a resul, invesors express views abou expeced reurns which are coheren wih he risk marix proposed in Blanc-Brude, Hasan, and Ismail (214) and Blanc-Brude and Hasan (2) for insance, by which sysemaic risk in infrasrucure invesmen can in par be broken down according o a simple 3 2 marix made up of hree business models (conraced, merchan and regulaed) and wo key momens in he lifecycle of infrasrucure projecs (greenfield and brownfield). A hird dimension of he risk profile of infrasrucure invesmens is counry or jurisdicion risk, which is confirmed by he repored reurns required by invesors for emerging marke infrasrucure. Ineresingly however he 3 2 paern described above is no changed by he addiion of emerging marke risk: relaively speaking, greenfield risk is sill aracing higher reurns han brownfield and conraced infrasrucure bu less han projecs exposed o merchan risk. However, he premium repored for aking emerging marke risk is driven by consideraions ha are specific o hese jurisdicions: he lower end of he risk specrum in

5 A SUPPLEMENT TO PENSIONS & INVESTMENTS 5 OECD infrasrucure (brownfield conraced and regulaed infrasrucure) is wha aracs he highes relaive premium in emerging markes. This validaes he focus on conracs as he deerminans of he risk profile of infrasrucure invesmens: he higher risks found in emerging markes spring from respondens repor he qualiy of he conracing framework and he abiliy o enforce conracual claims. Hence, he more infrasrucure invesmens rely on conracs (when i belongs o he conraced business model and in he long-erm i.e. a he brownfield sage) he more hey arac relaively higher risk premia in emerging markes. While he asse pricing implicaions make sense, hese resuls are also sriking from a public policy perspecive: counries ha have a bad rack record a respecing and enforcing conracual claims pay a significan premium on heir privaely financed infrasrucure; one ha in all likelihood renders uneconomic many poenial privae invesmen projecs in hese jurisdicions. Beyond he homogeneiy of invesors beliefs in erms of he risk and reurn componens of infrasrucure invesmens, survey resuls also highligh he heerogeneiy of views around hese fundamenal building blocks. Differen ypes of asse owners end o repor differen preferences and views are also highly heerogeneous beween individual invesors of he same ype. Tha invesors require a range of reurns for comparable risk profiles (i.e. wihin one family of infrasrucure invesmens) is congruen wih he noion ha in incomplee markes, he law of one price does no apply and large bid/ask spreads remain. In his survey, he repored range of expeced reurns is considerable, wih similar risk profiles aracing reurn requiremens ranging from less han five o more han percen. Finally, he fac ha asse managers sysemaically repor higher expeced reurns han asse owners can also be inerpreed as a reflecion of he agency issues found beween invesors (limied parners or LPs) and general parners (GPs) which we discuss a lengh below. PRODUCTS AND OBJECTIVES Key findings EXHIBIT 2 Limiaions of curren asse valuaion approaches Do you rus he asse valuaions repored by infrasrucure managers? (asse owners only) EXHIBIT 3 Perceived added value of infrasrucure privae-equiy-syle funds 48% Yes 28% No 24% No Sure Is he classic closed ended PE fund wih fund level leverage in addiion o asse level leverage... (asse owners only).18%...a perfec ec jusified way o approach infasrucure invesmen Wih respec o available invesmen opions and he objecives pursued by asse owners invesing in infrasrucure, key findings of his survey include: 1. The immense majoriy of asse owners are raher dissaisfied wih exising infrasrucure invesmen producs; 2. Fee levels is he firs reason for his sae of affairs, and in second place is he absence of well-defined invesmen objecives of he various infrasrucure funds and plaforms; 3. Even co-invesmen alongside managers or banks is considered by almos half of asse owners o be only a second bes opion i.e. hey would raher have access o he invesmen producs hey need and wan. 4. The immense majoriy of asse owners consider he classic closed-ended privae equiy infrasrucure fund model o be oudaed and no adding value ; 5. The majoriy of invesors also declare ha hey are eiher concerned or very concerned abou he accumulaion of dry powder in numerous infrasrucure fund mandaes, because i could lead o a deerioraion of invesmen/underwriing sandards, if no he creaion of Ponzi unis ; 6. Mos respondens concur in saying ha infrasrucure invesmen only really makes sense as a long-erm sraegy (beyond en years), and a majoriy declares iself willing o buy and hold infrasrucure invesmens unil mauriy. Logically, bu perhaps surprisingly, mos invesors repor no being paricularly concerned by he absence of liquidiy of such invesmens. 7. Mos invesors declare hey prefer invesing in privaelyheld infrasrucure deb or equiy as opposed o public socks or bonds bu hey are evenly divided beween hose who prefer direc invesmen and hose who would raher delegae o a manager. 8. Overall, he objecives pursued hrough infrasrucure by he majoriy of invesors are linked o improving diversificaion and achieving higher performance. Oher objecives ha are inuiively associaed wih infrasrucure invesing such as hedging inflaion or ineres risk are less presen in he series of objecives currenly being pursued. However hey are amongs he highes ranked objecives ha invesors would like o be able o achieve hrough infrasrucure invesing (along wih sable cash flows and illiquidiy premia). Limiaions of available benchmarking opions EXHIBIT 4 Curren opions o benchmark privae invesmens in infrasrucure are....82%...oudaed and no adding value for invesors 62% Lacking 6% Sufficien 32% Wholly Inadequae e

6 6 A SUPPLEMENT TO PENSIONS & INVESTMENTS Marke failure? Combined wih he mos recen repors on infrasrucure fund raising which is a hisoric heighs hese resuls reveal somehing like a quandary: a leas half of invesors would like o inves hrough a manager bu he immense majoriy of hem complain ha exising producs are oo expensive and no designed o help hem achieve heir goals. As we repor in he nex secion, more han half of hem do no even rus he performance merics repored by infrasrucure asse managers. The marke o provide access o infrasrucure invesmen hrough invesmen funds is large and growing, and he number of asse managers acive in his space is also significan. I can be surprising ha compeiion beween general parners (GPs) for he aenion of limied parners (LPs) does no lead o a more aggressive levelling of fees or he design of differen ypes of infrasrucure funds. In effec, a small number of asse managers do offer longer, less aggressive and less expensive infrasrucure funds han he mainsream infrasrucure PE funds, bu hey represen a minoriy of he oal fundraising. Why do asse owners coninue o inves in infrasrucure funds ha 8 percen of hem considers o be oudaed and no adding value? When insiuions allowing marke paricipans o rade wihou resricion on prices or volumes are in place and he expeced benefis of compeiion fail o maerialise, marke mechanisms can be considered o be failing. In effec, i can be argued ha he marke for delegaed invesmen managemen in he infrasrucure secor is a leas parly failing o creae he kind of producs ha invesors need, le alone a a fair price. Nex, we discuss why a marke can be suck in a subopimal equilibrium, in which invesors only have access o inadequae and expensive producs. Say a marke for invesmen managemen services is characerised by differen ypes of service providers (in his case, infrasrucure asse managers): hese managers can be more or less capable, ha is, more or less able o selec and manage infrasrucure deb and equiy invesmens o build a porfolio ha has cerain characerisics of ineres o asse owners. The differen ypes of managers can also be disribued more or less evenly: for insance here could be a few capable managers and many less capable ones. Asse owners who need o choose an infrasrucure asse manager are hen faced wih a simple problem: hey do no know which ones are he capable managers and which ones are no. They are said o be facing he problem of adverse selecion. Nex, say ha asse managers also have he opion o make a cerain effor o creae he kind of infrasrucure invesmen produc ha invesors would prefer. This effor is cosly o he manager bu i leads o he creaion of beer producs e.g. beer defined duraion and risk facor exposures. Hence, invesors are also faced wih a case of moral hazard: hey need o creae incenives o induce asse managers o exer a cosly effor o deliver he kind of produc ha bes uilises he characerisics of infrasrucure asses o achieve heir invesmen objecives. If he capable managers do make his effor and propose beer invesmen producs, invesors can choose he producs hey need and maximise heir long-erm uiliy. If he less capable asse managers made he same cosly effor, hey would go ou of business and be forced o exi he marke. Wih perfec informaion abou manager ype and wha invesmen producs can be creaed by invesing in infrasrucure, compeiion would work as expeced: invesors would require he producs ha are bes suied o heir needs and he capable managers would provide hem, and compeiion in he marke would be limied o he capable ypes. The difficuly arises from he absence of informaion (e.g. benchmarking) for asse owners, who do no know exacly wha infrasrucure invesing can do for hem and canno easily discriminae beween differen ypes of managers. Wihou perfec informaion however (he asse owners will have no knowledge of he managers ype), capable managers can simply mimic he less capable ones, make no cosly effor o design beer invesmen producs and provide he same oudaed producs like any oher providers. Wha drives up coss in his case is no he absence of compeiion, bu he endency for all managers o pool ogeher and behave like he leas capable ones. The presence of asymmeric informaion beween buyers and sellers affecs he funcioning of markes and can lead o marke failure: eiher he absence of rade (invesors exi he marke and decide o inernalise infrasrucure invesmen i.e. he so-called Canadian model) or a very subopimal rade characerised by he pooling of manager ypes (all managers provide he same producs). In his las case, asse owners buy invesmen producs ha are no wha hey need and a a high price given he uiliy hey derive from hem, and even he more capable managers end o offer sandardised, relaively inadequae producs, while hey could achieve a greaer marke share by offering advanced invesmen soluions. Nex, we discuss boh cases in more deails. THE COSTS OF REJECTING DELEGATION Faced wih he kind of marke failure described above, a firs group of paricipans chooses o exi and address agency issues under asymmeric informaion by inernalising he invesmen funcion, in his case by building up inernal capabiliy o source and execue infrasrucure ransacions, manage infrasrucure firms hroughou heir lifecycle and receive he benefis of direc conrol, asse selecion and ransacion iming, including as he majoriy of survey respondens declared he opion o hold invesmens o mauriy. Borrowing from he vocabulary of behavioural sudies in he reail pension secor, hese do-i-yourself (DIY) invesors also end o be he mos engaged and sophisicaed ones, whereas ohers, probably smaller invesors, for whom infrasrucure may be a much newer heme, can be described as passive. 1 For engaged invesors o be beer off following he DIY approach han delegaing o a specialis manager, hey mus be able o deliver resuls which are a leas as good as hose producs provided by he bes managers in he marke (ne of coss). The ne benefis from choosing direc invesing are hus deermined by hree facors: 1. Invesmen coss: wih limied effecs of compeiion beween managers on fees, some asse owners have come o he conclusion ha inernalising infrasrucure invesmen can be worhwhile. Neverheless, a fully fledged infrasrucure eam is only available o large invesors. Such eams may also encouner heir lifecycle issues as invesors buy infrasrucure firms (ransacion srucuring and execuion) and operae hem on a buy-ohold basis (asse 2 managemen), he required skill-se mus change over ime. I is also possible ha some agency issues ha exis beween asse owners and managers are simply re-creaed inernally beween he sraegic asse allocaion level and he invesmen level. 2. Diversificaion benefis: Building a direc porfolio of infrasrucure asses is long-erm goal in iself. The recen experience of some Ausralian or Canadian invesors suggess ha i can ake a leas en o years. Even so, he resuling porfolio of 2 o 25 invesmen is unlikely o be well diversified and may even include very concenraed exposures (i.e. a few very large firms). Of course, he main diversificaion benefis of infrasrucure invesmen accrue o he porfolio as a whole, as survey responses sugges, bu less diversificaion of he infrasrucure porfolio iself can be considered a sraigh loss. In principle, invesors should be able o diversify beer by invesing across a range of infrasrucure funds, hemselves exposed o a range infrasrucure business models, lifecycle sages and jurisdicions. The exen of he failure of he marke for delegaed invesmen in infrasrucure is highlighed by his fac: a growing number of large invesors prefer forgoing diversificaion benefis in favour of a more concenraed, inernally managed porfolio. 3. Porfolio consrucion: Agains hese coss (fees and lower diversificaion) invesors expec benefis ha are hemselves dependen on wha porfolio of infrasrucure asses each one of hem can build. Differen invesors have differen objecives and liabiliy profiles which canno be answered ex ane. Full conrol over he invesmen process may allow asse owners o build infrasrucure porfolios ha are more in line wih heir objecives. However, if a well-funcioning marke for invesmen delegaion led o he creaion of beer-defined invesmen producs using infrasrucure deb and equiy o arge a given se of financial merics, he poenial conribuion of such producs may ouweigh he benefis of conrol on asse selecion and infrasrucure porfolio consrucion. Thus, he ne benefis of inernalising long-erm invesmen in infrasrucure are no self-eviden once he possibiliy o improve invesmen producs is aken ino accoun. These issues hinge around he absence of sufficien informaion abou wha can be achieved hrough infrasrucure invesmen and who can commi o achieving such goals. MARKET SOLUTIONS: BENCHMARKING AND SIGNALLING Why aren he more capable infrasrucure asse managers offering differen producs han he classic wo-andweny closed ended PE fund? In he classic adverse selecion model, he more capable ype of manager is simply beeroff in he shor-erm mimicking he less capable ype, and making no cosly effor o deliver beer service. Bu i can also be he case ha he mos compeen managers would be beer off providing more advanced producs (hey would gain marke share) bu canno effecively ariculae and demonsrae he added-value of hey could creae by designing differen forms of infrasrucure invesmen producs. If informaion asymmery is oo srong hen wha migh be achievable hrough new forms of infrasrucure invesmen producs may be very challenging o communicae effecively o asse owners, who remain faced wih he Scylla of DIY invesing and he Charybdis of infrasrucure PE funds. There are however soluions o minimise he effec of informaion asymmery in marke dynamics. To avoid he pooling of managers, marke paricipans can creae soring devices (Spence 1973; Rohschild and Sigliz 1992) or revelaion mechanisms (Laffon and Marimor 22) o faciliae he processing of informaion from uninformed o informed paricipans. 3 The more capable asse managers may also ry o signal heir abiliy o o creae beer producs o asse owners hrough various devices (e.g. cerificaion schemes). In economics, his problem is ypically modelled as a marke wih adverse selecion and compeiive search, where some agens pos erms of rade (conracual erms) and ohers aim o screen he oher side of he rade by agen ype (see for example Guerrieri, Shimer, and Wrigh 21). In such models, he informed side of he rade (here he asse manager) can move firs and signal o he marke wha erms hey can offer, or he uninformed side can move firs and reques a bid for a given menu of conracs. In oher words, eiher asse owners could reques bids in an aucion for a limied number of well-defined invesmen producs, or asse managers could choose o highligh he differen producs ha are available hrough he kind of performance reporing sandard, valuaion approaches and performance benchmarks ha we discuss nex. BENCHMARKING Key findings On he opic of benchmarking he performance of infrasrucure invesmens, he main findings of he survey are: 1. Invesors curren use of benchmarks for heir infrasrucure invesmens are as likely o be relaive or absolue, nominal or real, or relaive o a marke or a macroeconomic index. There is no clear marke pracice; 2. In fac, he immense majoriy of invesors and managers agree ha currenly available benchmarks are inadequae and ha proper infrasrucure invesmen benchmarks jus do no exis; 3. Survey respondens confirm ha risk merics in paricular are no documened and ha valuaions are sufficienly problemaic o cas doub on any measure of reurns as well. More han half of asse owners reckon ha hey eiher do no rus or do no know if hey can rus he valuaions repored by he infrasrucure asse managers. 1 Sill, i is also possible for large direc invesors in infrasrucure o rerea from he DIY approach and o reurn o managed infrasrucure mandaes. The Vicoria Fund Managemen Corporaion is one such recen example. 2 The infrasrucure asse. 3 For example, Hellwig (1987) discusses he role of deducibles in insurance conracs and how he choice of deducible can be used by insurers o infer he probabiliy of acciden of a given individual.

7 A SUPPLEMENT TO PENSIONS & INVESTMENTS 7 Towards beer benchmarks Roadmap and recen progress In June 214, Blanc-Brude (214) pu forward a roadmap for he creaion of infrasrucure invesmen benchmarks. This roadmap inegraes he quesion of daa collecion upfron, including he requiremen o collec informaion known o exis in a reasonably sandardised forma and limied o wha is necessary o implemen robus asse pricing and risk models. I pus forward he following seps: 1. Defining he relevan insrumens; 2. Developing a relevan asse pricing framework; 3. Defining he necessary daa; 4. Building a global daabase of cash flows and invesmen characerisics; 5. Building reference porfolios of infrasrucure equiy and deb. The implemenaion of his roadmap is described in deails in Blanc-Brude (214) and recen progress in Blanc-Brude e al (216) Defining infrasrucure invesmens from a financial perspecive, he only relevan perspecive o build invesmen benchmarks, is a necessary firs sep. As he resuls of his survey and he recenly proposed definiion pu forward by he European regulaor of pension plans and insurance companies sugges, defining infrasrucure invesmen from an invesmen perspecive has progressed considerably. The growing consensus refleced in his survey around he limied role of indusrial secor caegories in explaining and predicing performance, and he much more significan role played by conracs and by differen infrasrucure business models such as merchan or conraced infrasrucure, or various forms of uiliy regulaion, is encouraging. Once he financial insrumens ha correspond o infrasrucure invesmen are usefully defined, he second necessary sep is o design a performance and risk measuremen framework ha can provide robus answers o he quesions idenified above. Again, our survey responses confirm he urgen need o improve he curren mehodologies o evaluae privae asses given he increasingly imporan role hey play in invesors porfolios. Privaely-held, infrasrucure equiy and deb insrumens are no raded frequenly and canno be expeced o be fully spanned by a combinaion of public securiies. Hence, hey are unlikely o have unique prices ha all invesors concur wih a one poin in ime. A wo-sep approach o measuring performance is herefore necessary: 1. Documening cash flow disribuions (deb service and dividends) o address he fundamenal problem of unreliable or insufficienly repored NAVs or losses given defaul (LGDs); 2. Esimaing he relevan (erm srucure of) discoun raes, or required raes of reurns, and heir evoluion in ime. Here oo, progress has been made and recen research provides a framework addressing boh seps, aking ino accoun he availabiliy of daa, while applying bes-in-class models of financial performance measuremen (see for example Blanc-Brude, Hasan, and Ismail 214; Blanc-Brude and Hasan 2 for applicaions o he privae deb and equiy case). Based on his new asse pricing and risk measuremen echnology, a lis of daa iems required o implemen adequae mehodologies can be drawn ha can be used o collec daa and populae he necessary daabase bu also o deermine a minimal reporing framework for invesors o require from infrasrucure managers. These daa collecion requiremens are described in Blanc-Brude e al. (216). The acive collecion of he necessary daa and publicaion of he relevan invesmen benchmarks has begun o be implemened wih he creaion of he EDHEC Infrasrucure Insiue in Singapore in February 216 and is planned o ake place incremenally unil 22 and beyond. Benchmarking as signalling In his survey and in ohers before, asse owners highligh high fees, insufficien performance reporing and inadequae valuaion mehods as some of he main issues found in delegaed privae invesmen. In recen years, however, asse owners have begun o quesion he level of invesmen fees and o achieve subsanial reducions in he overall level of invesmen managemen fees, hrough self-organisaion as well as wih he help of he regulaor. As we argued above, high fees are only he resul of he informaion asymmery ha exiss beween asse owners and managers. The crux of he maer hinges around repored valuaions. The valuaion of privae asses is he primary source of informaion asymmery beween managers and asse owners. Hence, wih he argumen o lower fees gradually being won by LPs, he nex big issue o open for review is asse valuaion. Privae asse valuaion has long suffered from numerous flaws, in paricular he classic sale pricing problem and he corollary smoohing of asse reurns (see Blanc-Brude and Hasan 2 for a review of he lieraure on he subjec applied o infrasrucure). As we suggesed above, a number of echnical improvemens are possible ha allow beer measuremens of risk-adjused performance in privae infrasrucure invesmens. In due course, furher developmen in applied academic research will allow for even more robus and advanced mehods o be implemened. The maer of reporing adequae performance daa and applying sae-of-he-ar valuaion mehodologies is also relevan o he soring mechanisms or signalling ha we discussed above when suggesing soluions o he marke failures found in delegaed invesmen managemen. When informaion asymmeries are so significan ha asse owners canno know which managers are he capable or he less capable ones, hey could require managers o adop a cerain reporing framework and o implemen advanced valuaion mehods o make he more compeen managers reveal heir ype. Likewise, individual managers could offer o adop an equivalen reporing and valuaion framework o make asse owners aware of heir ype. Once, he more capable managers have agreed o reveal heir ype or have been idenified by asse owners, i becomes possible for he laer o require ha hey exer he kind of effor ha should lead o he creaion of beer invesmen producs. Noe ha revealing heir ypes for he beer managers is no free and ha in he sandard soluion o he principal agen problem wih adverse selecion and moral hazard he incenive compaible conrac beween he clien and he service provider requires ha a premium be paid o he agen of he desirable ype. However, he ne (afer fee) benefis o asse owners should now be much higher (if no, hen inernalisaion he DIY opion remains he preferred roue). Beyond ype revelaion or discovery, he las missing elemen in he relaionship beween principal and agen is for asse owners o acually know wha o ask he beer managers o do for hem hrough infrasrucure invesmen. Infrasrucure invesmen benchmarks are a he hear of his issue: wih fully fledged benchmarks, wha is achievable for invesors hrough infrasrucure invesmen can be known (e.g. wha combinaion of facor exposures infrasrucure invesmen can creae) and only hen can asse owners reques heir managers o build infrasrucure porfolios ha are fully inegraed ino a long-erm invesmen soluion for hem. In effec, privae infrasrucure invesmen benchmarks can improve mos issues of informaion asymmery beween invesors and managers since hey can be used boh o deermine wha invesors should require and o signal wha managers can or canno deliver. ESG Key findings Regarding he environmenal, social and governance impac of infrasrucure invesmen, asse owners responses sugges ha: 1. Invesors acknowledge he relevance of ESG consideraions bu a majoriy considers ESG o be a second order problem i.e. one ha does no rump firs order quesions like sraegic asse allocaion; 2. Neverheless, 17 percen of owners consider ESG o be a firs order quesion; 3. Mos respondens also expec ESG o be posiively relaed o invesmen reurns. Does ESG mean more or less risk? Insiuional invesors all have well-defined mandaes o, for example, ensure he delivery of pension benefis, he solvency of insurance schemes or he preservaion of naional wealh. Respecing hese differen mandaes means achieving a series of nominal or real wealh objecives a cerain horizons and preserving he funding level (liabiliies vs. asses) of each insiuion a each poin in ime. In oher words, i means focusing on risk-adjused financial performance, which is, in urn, he resul of sraegic asse allocaion decisions. This is every asse owner s firs order problem. To he exen ha invesors also wan o avoid invesing in cerain ypes of infrasrucure projecs (e.g. coal-fired power plans) or ensure ha he social consequences of new projecs These resuls highligh he degree o which invesors agree or disagree abou wha o expec from invesing in infrasrucure equiy or deb (e.g. hydro-elecric dams) are limied and well managed, he consideraions mus neverheless remain subordinaed o achieving long-erm financial objecives. I does no mean ha invesors do no care abou invesing in less susainable businesses or projecs. Bu simply ha hey have o mee cerain objecives firs, and ha ESG invesing would be self-defeaing if i undermined heir abiliy o achieve hese goals. In fac, being able o pay he pensions and life insurance policies of millions of individuals is nohing shor of a very worhy social goal. Sill, in his survey, 17 percen of asse owners consider ha ESG is neverheless a firs order problem. Moreover, i is likely ha his number has been increasing and ha even more invesors would give his answer in a fuure ieraion of his survey. ESG invesing can be modelled as a form of guil aversion a noion developed in behavioural economics by which invesors could be willing o forgo some level of performance or fuure income o avoid invesing in cerain ypes asses. In his case, here is a mosly negaive rade-off wih performance and ensuring a minimal hreshold of ESG-qualiy in new invesmens can also be undersood as a form of risk managemen: new infrasrucure projecs ha are less likely o creae environmenal or social issues may also be less likely o experience regulaory or policy shocks in he fuure. If his is he case hen, higher ESG crieria should be synonymous wih lower expeced reurns. Sill, he majoriy of respondens believe ha here is a posiive link beween reurns and ESG qualiy, implying higher risk aking in such projecs. For insance, invesing in renewable energy and reducing carbon emissions qualifies as having a posiive environmenal impac bu also ress on publicly-sponsored ariff subsidy schemes ha are prone o change over he decades ha each wind or solar farm invesmen is supposed o las. Recen evidence of changes in wind-farms' feed-in ariff, someimes rero-acively, is pleniful in European markes for insance. Anoher aspec of ESG in he conex of infrasrucure invesmens is job creaion. While his can be considered a posiive in regards o he social and poliical accepance of privae infrasrucure invesmen (he so-called social licence o operae of he privae secor), commiing o employing a cerain workforce may creae long-erm issues regarding operaional efficiency given he impac of echnological change over several decades. The impac of conainerisaion in he por secor is a good example of a secor ha had o le go mos of is workforce over a couple of decades. If invesors expec higher reurns from ESG complian invesmens, i may be inerpreed as an increase in risk aversion vis-a-vis an economic fuure which changing environmenal and social issues make increasingly uncerain or, perhaps more simply, he recen drive owards ESG could be seen as par of a broader increase in invesor risk appeie in a low yield environmen.

8 8 A SUPPLEMENT TO PENSIONS & INVESTMENTS INDEXES Looking for a Lised Infrasrucure Asse Class Frederic Blanc-Brude Direcor Tim Whiaker Associae Research Direcor Simon Wilde PhD candidae Universiy of Bah Senior Managing Direcor Macquarie Capial In his aricle, we ask he quesion: does focusing on lised infrasrucure socks creae diversificaion benefis previously unavailable o large invesors already acive in public markes? This quesion arises from wha we call he infrasrucure invesmen narraive (Blanc- Brude 213), a se of invesmen beliefs commonly held by invesors abou he invesmen characerisics of infrasrucure asses. In his narraive, he infrasrucure asse class is less exposed o he business cycle because of he low price-elasiciy of infrasrucure services. Furhermore, he value of hese invesmens is expeced o be mosly deermined by income sreams exending far ino he fuure, and should hus be less impaced by curren evens. According o his inuiion, lised infrasrucure may provide diversificaion benefis o invesors since hey are expeced o exhibi low reurn covariance wih oher financial asses. In oher words, lised infrasrucure is expeced o exhibi sufficienly unique characerisics o be considered an asse class in is own righ. Empirically, here are a leas hree reasons why his view requires furher examinaion: 1. Mos exising research on infrasrucure has used public equiy markes o infer findings for he whole infrasrucure invesmen universe, bu robus and conclusive evidence is no forhcoming in exising papers; 2. Index providers have creaed dedicaed indices focusing on his heme and a number of acive managers propose o inves in lised infrasrucure arguing ha i does indeed consiue a unique asse class; 3. Lised infrasrucure socks are ofen used by invesors o proxy invesmens in privaely held (unlised) infrasrucure equiy, bu he adequacy of such proxies remains unesed. The exisence of a disincive lised infrasrucure effec in invesors porfolios would suppor hese views. In he negaive, if his effec canno be found, here is lile o expec from lised infrasrucure equiy from an asse allocaion (risk/reward opimisaion) perspecive and maybe even less o learn from public markes abou he expeced performance of unlised infrasrucure invesmens. TESTING 22 PROXIES OF LISTED INFRASTRUCTURE We es he impac of adding 22 differen proxies of lised infrasrucure o he porfolio of a well-diversified invesor using mean-variance spanning ess. We focus on hree definiions of lised infrasrucure as an asse selecion scheme: 1. A naïve, rule-based filering of socks based on indusrial secor classificaions and percenage income generaed from pre-defined infrasrucure secors (nine proxies); 2. Exising lised infrasrucure indices designed and mainained by index providers (welve proxies); 3. A baske of socks offering a pure exposure o several hundred underlying projecs ha correspond o a well-known form of infrasrucure invesmen defined in conras wih he wo previous cases in erms of longerm public-privae conracs, no indusrial secors (one proxy). Employing he mean-variance spanning ess originally described by Huberman and Kandel (1987) and Kan and Zhou (212), we es he diversificaion benefis of hese proxies of he lised infrasrucure effec. THERE IS NO LISTED INFRASTRUCTURE ASSET CLASS Sylised findings include: 1. Our 22 ess of lised infrasrucure reveal lile o no robus evidence of a lised infrasrucure asse class ha was no already spanned by a combinaion of capial marke insrumens and alernaives, or by a facor-based asse allocaion; 2. The majoriy of es porfolios ha improve he meanvariance efficien fronier before he GFC fail o repea his fea pos-gfc. There is no evidence of persisen diversificaion benefis; 3. Of he 22 es porfolios used, only four manage o improve on a ypical asse allocaion defined eiher by radiional asse class or by facor exposure afer he GFC and only one is no already spanned boh pre- and pos- GFC; 4. Building baskes of socks on he basis of heir SIC code and secor-derived income fails o generae a convincing exposure o a new asse class; 5. Hence, benchmarking unlised infrasrucure invesmens wih hemaic (indusry-based) sock indices is unlikely o be very helpful from a pure asse allocaion perspecive i.e. he laer do no exhibi a risk/reurn rade-off or beas ha large invesors did no have access o already. Overall, we do no find persisen evidence o suppor he claims ha lised infrasrucure is an asse class. In oher words, any lised infrasrucure effec was already spanned by a combinaion of capial marke insrumens over he pas years in Global, US and UK markes. Defining infrasrucure invesmens as a series of indusrial secors and/or angible asses is fundamenally misleading. We find ha such asse selecion schemes do no creae diversificaion benefis, wheher reference porfolios are srucured by radiional asse classes or facor exposures. We conclude ha wha is ypically referred o as lised infrasrucure, defined by SIC code and indusrial secor, is no an asse class or a unique combinaion of marke facors, bu insead canno be persisenly disinguished from exising exposures in invesors porfolios, and ha expecing he emergence of a new or unique infrasrucure asse class by focusing on public equiies seleced on he basis of indusrial secors is unlikely o be very useful for invesors. Exhibi 1 provides an illusraion of hese resuls in he case of he FTSE Macquarie Lised Infrasrucure Index for he U.S. marke. Thus, asse owners and managers who use he common lised infrasrucure proxies o benchmark privae infrasrucure invesmens are eiher misrepresening (probably overesimaing) he bea of privae infrasrucure, and usually have o include various add-ons o such approaches, making hem compleely ad hoc and unscienific. DEFINING INFRASTRUCTURE DIFFERENTLY Overall, we do no find persisen evidence o suppor he claims ha lised infrasrucure is an asse class. Our ess also enaively sugges a more promising avenue o find infrasrucure in he public equiy space: focusing on underlying conracual or governance srucures ha end o maximise dividend payou and pay dividends wih grea regulariy, such as he public-privae parnerships (PPPs) or maser limied parnerships (MLPs) models, we find ha he mean-variance fronier of a reference invesor can be improved. The answer o our iniial quesion parly depends on how infrasrucure is defined and undersood as an asse selecion scheme. Under our hird definiion of infrasrucure, which focuses on he relaionship-specific and conracual naure of he infrasrucure business, we find ha lised infrasrucure may help idenify exposures ha have a leas he poenial o persisenly improve porfolio diversificaion on a oal reurn basis, as Exhibi 2 illusraes. This effec is driven by he regulariy and he size of dividend payous compared o oher corporaions, infrasrucure or no. Wha deermines his abiliy o deliver regular and high dividend payous is he conracual and governance srucure of he underlying businesses, no heir belonging o a given indusrial secor. Bundles of PPP projec companies or MLPs behave differenly han regular corporaions i.e. heir abiliy o reain and conrol he free cash flow of he firm is limied and hey end o make large equiy payous. In he case of PPP firms, as Blanc-Brude e al. (216) show, hey also pay dividends wih much greaer probabiliy han oher firms. Going beyond secor exposures and focusing on he underlying business model of he firm is more likely o reveal a unique combinaion of underlying risk facors. However, i mus be noed ha he relaively low aggregae marke capialisaion of lised eniies offering a clean exposure o infrasrucure business models as opposed o infrasrucure corporaes may limi he abiliy of invesors o enjoy hese poenial benefis unless he far larger unlised infrasrucure fund universe has similar characerisics. Fuure work by aims o answer hese quesions in he years o come.

9 A SUPPLEMENT TO PENSIONS & INVESTMENTS 9 EXHIBIT 1 EXHIBIT 2 Illusraion of he difference ess of mean-variance spanning of he FTSE Macquarie USA Infrasrucure Index Mean-Variance Fronier of oal reurns PFI porfolio and reference porfolio, (a) 2-214, asse class and facor-based reference RETURNS.5 High Yield US Equiies Real Esae FTSE Macquane US Unfa Corporae rae bonds Hedge Funds Governmen Bonds World Ex-US Equiies RETURNS.5 Term Defaul Size FTSE Macquane US Unfa Marke RETURNS.5 PFI Porfolio World Ex-US Equiies UK Gils UK Equiies Hedge Funds Real Esae Value Commodiies STANDARD DEVIATION STANDARD DEVIATION ION REFERENCE ENCE ASSETS REFERENCE & INFRASTRUCTURE ASSETS REFERENCE ASSETS REFERENCE & INFRASTRUCTURE ASSETS Commodiies STANDARD DEVIATION REFERENCE ASSETS REFERENCE & INFRASTRUCTURE ASSETS (b) 2-28, asse class and facor-based reference.25 Governmen Bonds Defaul Value Corporae Bonds FTSE Macquane a US Unfa.5.5 Size Term -.5 Hedge Funds - - Europe Socks FTSE Macquane US Unfa -.5 High Yields US Socks Commodiies US Equiies World Ex-US Equiies Real Esae STANDARD DEVIATION STANDARD DEVIATION REFERENCE ENCE ASSETS REFERENCE & INFRASTRUCTURE ASSETS REFERENCE ASSETS REFERENCE ENCE & INFRASTRUCTURE ASSETS RETURNS RETURNS (c) , asse class and facor-based reference.25 RETURNS R.5 High Yields Hedge Funds Corporae Bonds Governmen n Bonds Real Esae US Equiies FTSE Macquane US Unfa Commodiies RETURNS R.5 Size Term Value US Socks Europe Socks FTSE Macquane US Unfa STANDARD DEVIATION REFERENCE ENCE ASSETS REFERENCE & INFRASTRUCTURE ASSETS STANDARD DEVIATION REFERENCE ASSETS REFERENCE & INFRASTRUCTURE ASSETS

10 1 A SUPPLEMENT TO PENSIONS & INVESTMENTS INDEXES Is Privae Infrasrucure Differen? Frederic Blanc-Brude Direcor Majid Hasan Senior Research Engineer Tim Whiaker Associae Research Direcor Qi Wang Senior Research Engineer In a new paper drawn form he Research Chair endowed by Meridiam, Campbell-Luyens and he Long-Term Infrasrucure Invesors Associaion a he EDHEC Infrasrucure Insiue-Singapore, we conduc he firs large scale empirical analysis of he characerisics of cash flows in privae infrasrucure firms from he perspecive of equiy owners. The paper addresses wo main quesions: do infrasrucure firms correspond o a differen business model han he res of he firms acive in he economy? and do infrasrucure firms exhibi a differen equiy payou behaviour han oher firms? ARE INFRASTRUCTURE FIRMS DIFFERENT? Our moivaion springs from wha we have called he infrasrucure invesmen narraive (Blanc-Brude 213), according o which invesors in infrasrucure can look forward o low reurn correlaions wih he business cycle (hence poenially beer diversificaion), as well as lower sensiiviy o economic shocks (implying beer drawdown proecion). Empirical evidence for or agains such hypoheses has so far been very limied. This sudy is a firs ieraion in a series of research papers using a new, global and growing daabase of infrasrucure invesmen daa, and aims o measure he relaive financial performance of such invesmens hrough he creaion of fully-fledged benchmarks or reference porfolios. Here, we are ineresed in he volailiy of revenues in infrasrucure firms as well as heir relaive correlaion wih macro facors such as GDP growh, inflaion or financial markes. We are also ineresed in he equiy payou behaviour of infrasrucure firms, relaive o he business cycle as well as o oher privae and public firms. public secor and privae eniy (he firm), by which he public secor commis o paying a regular income o he firm as long as he relevan infrasrucure services are delivered according o a pre-agreed specificaion. Merchan infrasrucure firms in comparison are exposed o some degree of marke risk. Such infrasrucure projecs can have long-erm conracs supporing heir revenue in he form of a Power Purchase Agreemen (PPA) or ake-or-pay conrac, bu as such conracs may cover only par of he projec s capaciy or lifespan and he couner-pary is ypically no he Treasury. Oher Merchan infrasrucure firms are fully exposed o end user demand and marke prices and include airpors or oll roads. Finally, Regulaed infrasrucure firms are ypically naural monopolies involved in he provision of essenial services, such as sewage reamen, waer disribuion or power ransmission. In he Unied Kingdom, such companies are regulaed by independen agencies such as Ofwa or Ofgem. 5 The daa span informaion from he early 199s o 2, as illusraed on Exhibi 1. We focus on UK daa because hey represen he larges, longes and mos coheren se of infrasrucure cash flow daa available a his ime, wih he added advanage of corresponding o a single currency and regulaory environmen, hus limiing he need o conrol for hese dimensions in he analysis. Saring from UK infrasrucure firms, we can also build several conrol groups of non-infrasrucure firms, wih which o compare he daa. EXHIBIT 1 CONTROLLING FOR THE DIFFERENT ASPECTS OF FIRM BEHAVIOUR Our sample of several hundred infrasrucure firms is compared wih a mached sample of non-infrasrucure, UK firms, boh privae and lised. Indeed, while public marke daa on infrasrucure relaed firms has someimes been used as a proxy of privae infrasrucure firms, recen research has shown ha privae firms exhibi significan differences in erms of size, capial srucure and dividend policy: privae firms end o be smaller han lised firms, and o exhibi higher leverage, making heir profis more sensiive o flucuaions in performance, o have differen dividend payou policies han lised firms and o be less inclined o smooh heir dividends in he presence of profi shocks. Moreover, differences in ownership srucure in privae firms are also shown o explain differences in dividend payou policy beween differen ypes of firms (see Brav 29; Michaely and Robers 212 for a deailed sudy). To isolae he effec of ownership srucure and corporae governance on he behaviour of infrasrucure firms, we build hree conrol groups for each one of our infrasrucure firm ype: privae firms wih concenraed ownership, privae firms wih dispersed ownership and public (lised) firms. Each of hese hree conrol groups is hen mached o each infrasrucure firm-year observaion of a given ype using a neares neighbour mehodology for oal asse size, leverage and profiabiliy and an exac mach for invesmen Number and ime frame of reporing firms in he daabase. Each line represens a ime series of cash flow daa. Scope and breadh of he infrasrucure cash flow daase A UNIQUE NEW DATABASE This sudy makes use of he daabase: a collecion of cashflow, invesmen and balance shee daa colleced from infrasrucure invesors and crediors, as well as manually from individual firm s audied accouns. To dae, he daabase covers more han 5 individual infrasrucure asses over 1 differen counries and a period of -2 years, making i he mos comprehensive daabase of infrasrucure cash flows available for research. For his sudy, we focus on firms siuaed solely in he UK: firms idenified as being eiher special purpose vehicle creaed in he conex of he financing of a specific infrasrucure projec, or a firm conducing specific infrasrucure-relaed aciviies (such as a por or an airpor) or a regulaed uiliy. Thus, deailed accouns for each firm were obained from infrasrucure invesors, lenders and/or from Companies House 4 and analysed individually in order o classify each firm ino one of hree groups: Conraced, Merchan and Regulaed infrasrucure, which we describe nex. Conraced infrasrucure firms are no exposed o end user demand. In he Unied Kingdom, he Privae Finance Iniiaive (PFI) is he prime example of such projecs. Under he PFI scheme, infrasrucure invesors have delivered a broad range of infrasrucure, including schools, hospials and prisons. Such projecs spring from a long-erm conrac for he provision of an infrasrucure asse or service beween he NUMBER OF REPORTING FIRMS availabiliy-based merchan reglulaed uiliies invesmen n sar dae CASH FLOW OBSERVATIONS IN TIME 4 The UK Company Regiser. 5 See Blanc-Brude (213) for a deailed discussion of hese differen infrasrucure business models.

11 A SUPPLEMENT TO PENSIONS & INVESTMENTS 11 EXHIBIT 2 Esimaes of he mean and variance parameers of he uni revenues and profis in calendar ime for Conraced Infrasrucure and mached conrol firms Revenue/Average Asse raio in he Mean/SD Plane for Conraced Infrasrucure, in Calendar Time Profi/Average Asse Raio in he Mean/SD plane for Conraced Infrasrucure, in Calendar Time 2..8 STD DEVIATION OF PROFIT/AVERAGE ASSETS RATIO Infrasrucure (observed) Infrasrucure (fied) Privae concenraed (observed) Privae concenraed (fied) Privae Dispersed sed (observed) Privae Dispersed sed (fied) Public (observed) ed) Public (fied) STD DEVIATION OF PROFIT/AVERAGE ASSETS RATIO Infrasrucure (observed) Infrasrucure (fied) Privae concenraed (observed) Privae concenraed (fied) Privae Dispersed (observed) Privae Dispersed (fied) Public (observed) Public (fied) MEAN REVENUE/AVERAGE ASSETS RATIO MEAN PROFITS/AVERAGE AGE ASSETS RATIO EXHIBIT 3 Esimaes of he mean and variance parameers of he uni revenues and profis in calendar ime for Merchan Infrasrucure and mached conrol firms Revenue/Average Asse raio in he Mean/SD Plane for Merchan Infrasrucure, in Calendar Time Profi/Average Asse Raio in he Mean/SD plane for Merchan Infrasrucure, in Calendar Time STD DEVIATION OF PROFIT/AVERAGE ASSETS RATIO MEAN REVENUE/AVERAGE ASSETS RATIO Infrasrucure (observed) Infrasrucure (fied) Privae concenraed (observed) Privae concenraed (fied) Privae Dispersed (observed) Privae Dispersed (fied) Public (observed) Public (fied) STD DEVIATION OF PROFIT/AVERAGE ASSETS RATIO Infrasrucure (observed) Infrasrucure (fied) Privae concenraed (observed) Privae concenraed (fied) Privae Dispersed (observed) Privae Dispersed (fied) Public (observed) Public (fied) MEAN PROFITS/AVERAGE E ASSETS RATIO EXHIBIT 4 Esimaes of he mean and variance parameers of he uni revenues and profis in calendar ime for Regulaed Infrasrucure and mached conrol firms Revenue/Average Asse raio in he Mean/SD Plane for Regulaed Infrasrucure, in Calendar Time Profi/Average Asse Raio in he Mean/SD plane for Regulaed Infrasrucure, in Calendar Time 2..8 STD DEVIATION OF PROFIT/AVERAGE ASSETS RATIO Infrasrucure (observed) Infrasrucure (fied) Privae concenraed (observed) Privae concenraed (fied) Privae Dispersed (observed) Privae Dispersed (fied) Public (observed) Public (fied) MEAN REVENUE/AVERAGE ASSETS RATIO STD DEVIATION OF PROFIT/AVERAGE ASSETS RATIO Infrasrucure (observed) Infrasrucure (fied) Privae concenraed (observed) Privae concenraed (fied) Privae Dispersed (observed) Privae Dispersed (fied) Public (observed) Public (fied) MEAN PROFITS/AVERAGE ASSETS RATIO

12 12 A SUPPLEMENT TO PENSIONS & INVESTMENTS EXHIBIT 5 Probabiliy of posiive payous and expeced payou raio in Conraced Infrasrucure, invesmen ime P probabiliy of posiive equiy payou in conraced infrasrucure firms Expeced equiy payou raio in conraced infrasrucure firms 1..3 PROBABILITY OF POSITIVE EQUITY PAYOUT EQUITY PAYOUT RATIO (PAYOUT/REVENUES) INVESTMENT LIFE mean 5 1 INVESTMENT LIFE EXHIBIT 6 Esimaed probabiliy of posiive payous and payou raio densiy in privae firms wih concenraed ownership, mached o Conraced Infrasrucure P probabiliy of posiive equiy payou in mached privae firms (concen. ownership) Expeced equiy payou raio in privae firms (concen. ownership) mached o conraced infrasrucure 1. >) mean PROBABILITY OF POSITIVE EQUITY PAYOUT EQUITY PAYOUT RATIO (PAYOUT/REVENUES) EQUITY PAYOUT RATIO (PAYOUT/REVENUES) INVESTMENT LIFE INVESTMENT LIFE EXHIBIT 7 Probabiliy of posiive payous and expeced payou raio in Merchan Infrasrucure, invesmen ime P probabiliy of posiive equiy payou in merchan infrasrucure firms Expeced equiy payou raio in merchan infrasrucure firms PROBABILITY OF POSITIVE EQUITY PAYOUT EQUITY PAYOUT RATIO (PAYOUT/REVENUES) INVESTMENT LIFE mean 5 1 INVESTMENT LIFE

13 A SUPPLEMENT TO PENSIONS & INVESTMENTS 13 EXHIBIT 8 Esimaed probabiliy of posiive payous and payou raio densiy in privae firms wih concenraed ownership, mached o Merchan Infrasrucure P probabiliy of posiive equiy payou in mached privae firms (concen. ownership) Expeced equiy payou raio in privae firms (concen. ownership) mached o merchan infrasrucure 1. >) mean PROBABILITY OF POSITIVE EQUITY PAYOUT EQUITY PAYOUT RATIO (PAYOUT/REVENUES) INVESTMENT LIFE INVESTMENT LIFE EXHIBIT 9 Probabiliy of posiive payous and payou raio densiy in Regulaed Infrasrucure, invesmen ime P probabiliy of posiive equiy payou in regulaed infrasrucure firms Expeced equiy payou raio in regulaed infrasrucure firms 1. PROBABILITY OF POSITIVE EQUITY PAYOUT EQUITY PAYOUT RATIO (PAYOUT/REVENUES) INVESTMENT LIFE mean 5 1 INVESTMENT LIFE EXHIBIT 1 Esimaed probabiliy of posiive payous and payou raio densiy in privae firms wih concenraed ownership, mached o Regulaed Infrasrucure P probabiliy of posiive equiy payou in mached privae firms (concen. ownership) Expeced equiy payou raio in privae firms (concen. ownership) mached o regulaed infrasrucure 1. >) mean PROBABILITY OF POSITIVE EQUITY PAYOUT EQUITY PAYOUT RATIO (PAYOUT/REVENUES) INVESTMENT LIFE INVESTMENT LIFE

14 14 A SUPPLEMENT TO PENSIONS & INVESTMENTS year i.e. he number of years since he creaion of he firm. Hence, we es he difference in revenue and profi volailiy and in dividend payou raio level and volailiy beween infrasrucure and non-infrasrucure firms using nine differen ess: hree ypes of infrasrucure firms (conraced, merchan and uiliies) each compared o hree ypes of corporae governance (privae concenraed, privae dispersed, public), while conrolling for individual firm characerisics (size, leverage, profiabiliy) as bes as available daa allows. Such ess go a long way in addressing he maer of he uniqueness of infrasrucure invesmens. Indeed, if firm characerisics and corporae governance can be expeced o explain in large par he business model and dividend payou behaviour of he firm, hen for infrasrucure o be unique and no easily replicable by combining oher ypes of invesmens, i mus correspond o a unique combinaion of firm characerisics and corporae governance. Likewise, he revenues of infrasrucure firms can only creae a unique form of exposure o economic facors if heir business model is no an easily replicable combinaion of he business models of oher firms. INFRASTRUCTURE IS UNIQUE We find ha, as far as UK daa show over he pas years, infrasrucure firms are indeed ruly unique: ha is, afer conrolling for size, leverage and profiabiliy, as well as he impac of he invesmen lifecycle, infrasrucure firms exhibi lower revenue volailiy and higher payou raios (dividends o revenue) han any oher group of privae or public firms. Compared o heir conrol groups, infrasrucure firms have lower revenues and profis per dollar invesed, highlighing he capial-inensive and long-erm naure of heir business: 1. They are also characerised by significanly lower volailiy of revenues and profis compared o heir mached conrol groups, boh a he aggregae level (all periods) and a each poin in invesmen and calendar ime; 2. Infrasrucure firms also exhibi a very dynamic lifecycle compared o conrol groups, wih uni revenues and profis evolving by an order of magniude over he invesmen cycle; 3. Regression analyses show ha differen proxies of he business cycle have a srong saisical effec on profis and revenues in non-infrasrucure firms, bu ha his effec is absen in he differen infrasrucure firm es groups i.e. infrasrucure firm revenues and profis are less or no linked o he business cycle. Insead, he effec of he invesmen lifecycle is wha ends o explains he change in uni revenues and profis of infrasrucure firms. 4. The probabiliy of posiive equiy payous in infrasrucure firms is also significanly higher han in any of he conrol groups, reaching as high as 8% afer invesmen year 1 in Conraced infrasrucure and he 6-7% range in Merchan and Regulaed infrasrucure. Conrol groups never reach a (condiional) probabiliy of payou higher han 4%. These resuls are illusraed in Exhibi 2 o Exhibi Finally, equiy payou raios in infrasrucure firms are considerably higher han in he relevan conrol groups, reaching expeced values of beween 1% and % of revenues when mached conrols never payou more han 3-5% of revenues. Infrasrucure firms payou more ofen and significanly higher proporions of heir revenues han oher firms once he lifecycle of he firm is aken ino accoun, as shown in Exhibi 5 o Exhibi 1. Which show he equiy payou probabiliy and payou raio for each of our infrasrucure es groups compared o he mached privae concenraed group. Similar resuls for oher conrol groups are presened in he full paper. The combinaion of hese findings suggess ha infrasrucure firms exhibi a unique business model in erms of revenues and profis dynamics compared o a large conrol group of public and privae firms. Infrasrucure firms have significanly lower volailiy of revenues and profis and pay a much higher proporion of heir revenues much more frequenly o heir owners, independen of he business cycle. Anoher significan resul is ha each of he hree ypes of infrasrucure firms ha we define (according o a ypology we firs described in Blanc-Brude (213)) corresponds o a unique infrasrucure business model as well i.e. albei more alike amongs hemselves hen compared wih he res of he We find ha, as far as UK daa show over he pas years, infrasrucure firms are indeed ruly unique... corporae universe, Conraced, Merchan and Regulaed infrasrucure firms have heir own differen cash flow dynamics. IMPLICATIONS AND NEXT STEPS These resuls have implicaions for invesmen managemen and prudenial regulaion: 1. While hese resuls are no abou invesmen reurns per se, a financial asse ha pays more frequenly a larger proporion of he free cash flow of he firm can be expeced o have a differen pricing dynamic compared o he insrumens in he conrol group. Imporanly, i can be expeced o provide diversificaion benefis compared o oher privae and public equiy invesmens because of he low covariance beween revenues and profis of infrasrucure and non-infrasrucure firms. In oher words, from he unique business model of infrasrucure firms may well spring a unique combinaion of facor exposures ha are relevan o invesors. 2. Likewise, he quasi-absence of impac of discree financial or economic shocks, or of he business cycle capured by macro or marke facors, suggess non only low correlaions bu also a form drawdown proecion in bad imes from holding infrasrucure firms. 3. Hence, i is also likely ha a large baske of infrasrucure firm equiy can claim o be reaed differenly from a regulaory perspecive han he res of available public and privae invesmen opporuniies. Of course, he merics presened here are no direcly compaible wih he ones required o calibrae a prudenial framework such as Solvency-II (which requires compuing he 99.5% VaR) bu hey sugges srongly ha infrasrucure firms, wih heir lower revenue volailiy and lesser sensiiviy o he business cycle are a good candidae for a specific reamen. Nex seps include using hese findings o calibrae cash flow models o derive valuaions of individual infrasrucure asses (see for example Blanc-Brude and Hasan 2) and building reference porfolios ha can be used as benchmarks by invesors and regulaors alike.

15 A SUPPLEMENT TO PENSIONS & INVESTMENTS INDEXES Tracking Credi Merics in Privae Infrasrucure Deb Frederic Blanc-Brude Direcor Majid Hasan Senior Research Engineer Tim Whiaker Associae Research Direcor The objecives of his new paper drawn form he EDHEC/NATIXIS Research Chair on privae infrasrucure deb invesmen were o documen he saisical characerisics of deb service cover raios (DSCRs) in infrasrucure projec finance, and o develop and calibrae a model of DSCR dynamics. For his purpose, we collec a large sample of realised DSCR observaions across a range of infrasrucure projecs spanning more han years, represening he larges such sample available for research o dae, and conduc a series of saisical ess and analyses o esablish he mos adequae approach o modelling and predicing fuure DSCR levels and volailiy. Using hese resuls, we build a model of he condiional probabiliy disribuion of DSCRs a each poin in he life of infrasrucure projecs. DSCR DYNAMICS INCORPORATE KEY INFORMATION ABOUT THE CREDIT RISK OF INFRASTRUCTURE DEBT In a previous paper (Blanc-Brude, Hasan, and Ismail 214), we showed ha deb service cover raios can play a pivoal role in he modelling of credi risk in infrasrucure projec finance. This is because he DSCR of an infrasrucure projec company, which measures he amoun of cash available o make he curren period s deb service, provides us wih: 1. An unambiguous definiion of he poin of hard defaul (defaul of paymen), i.e. DDSSCCRR=1; and 2. An equally unambiguous definiion of key echnical defaul covenans i.e. DDSSCCRR=1.xx, while boh ypes of defaul evens creae significan embedded opions for crediors following a credi even; 3. Moreover, knowledge of DSCR dynamics is sufficien o esimae he firm s disance o defaul (DD), which is he workhorse of he so-called Meron or srucural credi risk model; 4. DSCR dynamics can also be combined wih fuure deb service o compue he expeced value and volailiy of he firm s fuure free cash flow, which is insrumenal in measuring enerprise value in he case of infrasrucure projecs, since hey derive heir value almos enirely from fuure operaing cash flows. For hese reasons, documening DSCR dynamics using realised DSCR daa is an imporan par of he objecive o creae invesmen benchmarks of privae infrasrucure deb, as described in he roadmap published by EDHEC-Risk Insiue in 214 (Blanc-Brude 214). A COMBINATION OF EMPIRICAL ANALYSIS AND STATISTICAL MODELLING IS NECESSARY DSCRs in infrasrucure projec finance are mosly undocumened boh in indusry and academic empirical lieraure. While DSCR informaion is rouinely colleced by he crediors of infrasrucure projecs, his ype of daa is ypically confidenial and no available in large daases. From such daa pauciy, especially in ime series, i follows ha empirical observaions alone are no sufficien o documen he expeced behaviour of infrasrucure projec cash flows over heir enire invesmen life, and a combinaion of ex ane modelling and empirical observaions is necessary. Finally, privae infrasrucure invesmen ends o be characerised by very large individual invesmens, almos necessarily leading o poorly diversified porfolios. This suggess ha assuming he mean-reversion of invesors infrasrucure deb porfolios may no be realisic and ha idiosyncraic risk should be aken ino accoun. In paricular, individual infrasrucure invesmens can exhibi significan pah dependency and invesors canno necessarily ake for graned he noion ha hey are exposed o he median infrasrucure projec. For boh ses of reasons (daa limiaions and he imporance of firm-specific risk), an adequae model of he DSCR should be able o capure condiional dynamics and explicily inegrae he differen credi saes ha an infrasrucure projec migh go hrough. This can help boh learning from he daa as and when i becomes available, and aking ino accoun he pah-dependency of each insrumen when esimaing fuure cash flows, insead of assuming a reversion o he populaion mean. Curren academic and indusry lieraure is saic in naure and relies on reduced form credi models, which are likely o be biased given he naure of empirical daa available and, in he curren sae of empirical knowledge, can only address a limied number of dimensions of privae infrasrucure deb invesmen: he hisorical frequency of defaul evens, and o some exen, average recovery raes. For hese reasons, in his paper we endeavour o beer documen he dynamics of DSCRs in infrasrucure projec finance and build a model of DSCR dynamics using Bayesian inference o describe credi sae ransiions and o esimae he mean and variance of he DSCR in each sae and a each poin in an insrumen s life. This allows beer predicion of defauls, condiional on he acual rajecory of individual invesmens or groups of projecs. The abiliy o predic cash flows and heir volailiy is hen insrumenal in he implemenaion of he privae infrasrucure deb valuaion model defined in Blanc-Brude, Hasan, and Ismail (214). DIVIDING INFRASTRUCTURE INVESTMENTS INTO GROUPS DEFINED BY THEIR BUSINESS MODEL In Blanc-Brude, Hasan, and Ismail (214), we described wo generic and inuiive ypes of infrasrucure projec companies and called hem conraced and merchan. This disincion was informed by he casual observaion ha he financial srucure of infrasrucure projec finance vehicles is ofen such ha i requires, a he onse, eiher a rising or a fla base case DSCR profile. A rising base case DSCR profile hen implies an increasing volailiy of DDSSCCRR. Tha is, crediors would demand a higher DSCR in he fuure o proec hemselves agains rising expeced volailiy of he cash flows available for deb service (CFADS). Such projecs would also have longer ails 6 and exhibi beween 7% and 8% of iniial senior leverage. We argued ha such srucuring decisions signalled infrasrucure projecs ha were exposed o commercial risks, such as a power plans selling elecriciy a marke prices, and referred o hese projecs as Merchan infrasrucure. Conversely, we argued ha he decision o srucure a projec while requiring a lower and flaer base case DSCR profile implied he expecaion of a lower and consan condiional volailiy of cash flows. We observed ha projecs wih lile o no marke risk are financed wih such a fla DSCR base case and also have shorer ails and a higher level of senior leverage, usually around 9%. Examples of hese projecs include social infrasrucure projecs, such as schools or hospials ha receive a fixed paymen from he public secor, or energy projecs ha benefi from a longerm ake-or-pay purchase agreemen. We called hese projecs Conraced infrasrucure. In his paper, we endeavour o deermine saisically wheher realised DSCR dynamics fall ino caegories deermined by he disincions made above beween Conraced and Merchan infrasrucure, as well as exogenous condiions a he ime of financing and when he daa is observed. We hen use our resuls o design a model of DSCR dynamics. THE LARGEST SAMPLE OF DSCR DATA AVAILABLE FOR RESEARCH TO DATE Our daase of realised DSCRs is buil using daa manually colleced and verified from he audied saemens of accouns of several hundred projec companies, as well as DSCR daa repored by privae conribuors. We hand colleced years of realised DSCR daa for more han 2 projecs in Europe and he Unied Saes covering our wo broad caegories of projecs (hose receiving a conraced income and hose exposed o merchan or commercial risks), in seven secors, from he early 199s o 2. Our iniial analysis of he daa reveals some imporan poins ha confirm our inuiion: he average credi risk profile of infrasrucure projecs can be usefully capured by caegorising insrumens in broad groups or families of underlying business models. The wo groups correspond o wo disincive DSCR processes, wih saisically differen mean and variance parameers and following differen projec ime dynamics. We also find, as inuiion predics, ha conraced infrasrucure DSCRs in he cross secion are much less affeced by macrovariables or he business cycle han merchan projecs. We confirm our hypoheses ha he DSCR profile of an infrasrucure projec is srongly relaed o he firm s oal business risk, and show ha more highly leveraged projecs achieve lower levels of realised DSCR volailiy i.e. in projec finance high leverage signals low asse risk as iniially argued by Esy (23). Bu while descripive saisics and linear regression models provide some insighs abou he deerminans of he DSCRs, hey fail o capure DSCR dynamics in full. Indeed, we find ha he DSCR profiles of individual projecs and families of projecs are highly non-linear, auo-regressive and heeroskedasic (variance is no consan). Hence, a more advanced model ha can capure hese dynamics is needed. 6 The amoun of ime beween he original loan mauriy and he end of he projec s life, hus allowing higher recovery raes in he even of resrucuring.

16 16 A SUPPLEMENT TO PENSIONS & INVESTMENTS TRACKING THE COORDINATES OF THE DSCR DISTRIBUTION IN THE MEAN- VARIANCE STATE-SPACE Illusraion of he DSCR pah beween saes EXHIBIT 1 If he DDSSCCRR is serially correlaed and can change profile during he invesmen lifecycle of infrasrucure projecs, he ex pos rajecory of individual projecs could in principle correspond o any combinaion of high/low expeced value EE(DDSSCCRR) and high/low volailiy σσ 2 DDSSCCRR. The DSCR of populaions of projecs would equally reflec he weighed rajecory of heir consiuens in a DDSSCCRR mean/variance plane. Numerous models exis ha aim o deermine he posiion of a dynamic sysem and, based on he laes round of observaions, o predic where i will be posiioned in fuure periods. Such sysems are frequenly used in roboics, aerospaial and chemisry applicaions. In his paper, we apply such approaches o esimae he posiion of a given infrasrucure projec in a mean/volailiy DSCR plane a a given poin in ime, and o predic is posiion is DSCR mean and variance coordinaes so o speak in he following periods. In he descripive par of our analysis of he daa, we show ha realised DSCRs can be fied o a lognormal process up o heir 9h and 85h quaniles for conraced and merchan projecs, respecively, a each poin in heir lifecycle, which allows us o develop an easily racable model of parameer inference. Hence, we propose a wo-sep modelling sraegy combining a hree-sae model corresponding o break up poins in he oherwise lognormal dynamics of he DSCR, wih a filering model o infer he values of he lognormal process parameers (is coordinaes ) in he sae in which he DSCR is indeed lognormal. DSCR Safe Sae Upper Limi Risky Sae DSCR Defaul Threshold Defaul Sae T EXHIBIT 2 DSCR densiies for conraced and merchan families. Probabiliies of he Three Saes for Conraced Projecs Probabiliies of he Three Saes for Merchan Projecs Three-Sae Transiion Probabiliies The DSCR process is assumed o occur in any one of hree saes a ime : a risky sae (R) in which i is indeed an auoregressive lognormal process, a defaul sae (D) defined by a hreshold corresponding o DDSSCCRR=1 in which he DSCR process sops unil i emerges from defaul; and a safe sae (S), corresponding o high realised values above he goodlognormal-fi quanile, in which case, as long as he DSCR says in ha sae, he projec deb is considered risk-free. This illusraed by Exhibi 1. Hence, once a projec s DSCR breaches he hard defaul hreshold represened by DDSSCCRR=1, i eners he defaul sae, which i may or may no leave afer a number of periods. In his sae, crediors can ake over he firm and opimise he value of exercising his opion depending on he size of heir exi coss and of resrucuring coss. They may decide o waive he even of defaul or engage in negoiaions wih projec sponsor o resrucure he firm and is deb or indeed ake over he firm and seek anoher sponsor (see Blanc-Brude, Hasan, and Ismail 214 for a formal model). Hence, he firm may ransi ou of he defaul sae (ino he risky sae) wih some probabiliy (say, ππ ddrr ) a he nex period, or say in his sae and again ransi ou of defaul a he nex period, ec. In his sae, he DSCR process effecively sops (in mos cases, here is no deb service), hence esimaing is mean and variance is irrelevan since he projec is already in defaul. In he safe sae, on he conrary, he realised DSCR is so high ha no maer how volaile he process migh be, from a senior credior perspecive, he probabiliy of defaul is no significanly differen from zero. The deb is (condiionally) risk-free. As before, in expecaion a ime, an infrasrucure projec may ransi in and ou of he safe sae a each poin in he fuure, wih some probabiliy (say, ππ ssrr ). In his sae, esimaing he parameers of he DSCR disribuion, in paricular esimaing is variance, is also irrelevan. Finally, in beween he defaul and safe saes, a projec s DSCR may ake values beween 1 and some higher hreshold DDSSCCRR. From his sae, i may eiher say in he risky sae a he nex period, or ransi ouf of i ino he sae of defaul d or he safe sae s, boh described above. In his sae, we know from our empirical resuls ha if he upper hreshold is se a he 85h/9h quanile of our DSCR sample, he daa follows a lognormal process, he paramaers of which (posiion and scale) have o be esimaed (see below). Formally, his seup amouns o a relaively simple model of condiional sae ransiion probabiliies, which can be se in erms of a series of binomial draws and calibraed using Bayesian inference given some prior knowledge (e.g. we DENSIT Y DSCR DENSIT Y D

17 A SUPPLEMENT TO PENSIONS & INVESTMENTS 17 know from credi raing sudies ha projecs end o say in defaul for 2.3 years) and couning he number of projecs crossing he differen sae hresholds, condiional on which sae hey are in a he previous period. The combinaion of he condiional probabiliies of swiching sae a each poin in ime are hen combined ino he probabiliy of being in any given sae a ime. For conraced projecs he probabiliy of being in he risky sae is much higher compared o he probabiliy of being in he oher wo saes i.e. conraced projecs are more likely o say in he normal risky sae. For merchan projecs, he probabiliy of being in he risky sae is lower, while he probabiliies of being in he defaul and safe saes are higher compared o he corresponding probabiliies for conraced projecs. Thus, merchan projecs are found o have more diverse DSCR rajecories in sae space, and each sae is less persisen (sable). This resul confirms ha pah dependency can be an imporan dimension of infrasrucure invesmen insofar as asses are more or less heerogenous and can be difficul o fully diversify very large and bulky asses. For insance, our resuls sugges ha conraced infrasrucure is more homogenous han merchan projecs, which are more likely o follow pahs ha diverge srongly from he mean of he populaion. Group and Individual DSCR Trajecories To deermine he value of he lognormal process parameers in he risky sae discussed above, we propose o use a sraighforward implemenaion of so-called paricle filering models o infer he parameer values of he DSCR s lognormal process in he risky sae i.e. he sae in which documening and racking he volailiy of he DSCR really maers, because i is a direc measure of credi risk. Filering models are a form of signal processing and aim o arrive a some bes-esimae of he value of a sysem, given some limied and possibly noisy measuremens of ha sysem s behaviour. Given our modelling objecives o accommodae small samples, and o avoid assuming saic values for he DDSSCCRR disribuion parameers, we mus be able o revise any exising parameer esimaes once new daa becomes available. This process is bes esimaed ieraively using Bayesian inference echniques described in deail in he paper. We show ha such a framework allows he dynamics of DSCR o be derived in well defined groups of projecs as well as individual projecs, including racking he individual DSCR pah followed by invesmens ha do no necessarily correspond o he median infrasrucure projec. The esimaed dynamics of he DSCR process in conraced and merchan projecs is shown in Exhibi 2, which describes he change in densiy of he DSCR process in invesmen ime, and Exhibi 3, which describes he rajecory of he DSCR sae in he mean/sandard deviaion plane. From such resuls, cerain credi risk conclusions are immediaely available, such as he expeced defaul frequency for hard defauls bu also any level of echnical defaul (DDSSCCRR=1.xx) as shown in Exhibi 4. These resuls allows us o characerise he behaviour of groups of infrasrucure projecs which exhibi reasonably homogeneous dynamics, however, we know ha highly idiosyncraic rajecories and pah dependency should be a poin of ineres in a conex where diversificaion is difficul o achieve in full. Hence, we also show ha he abiliy o infer boh he expeced value and he volailiy of he DSCR process allows us o ake a much more informed view on he credi risk of projecs ha subsanially deviae from heir base case. For insance, consider an infrasrucure projec ha follows an of-observed rajecory: while i remains in he risky sae hroughou is life, i sars off wih a relaively high DSCR, implying a merchan-ype srucure wih relaively high DSCR volailiy, bu laer on undergoes a large downward shif in is realised DSCR level, e.g. as he resul of a negaive demand shock, while is DSCR realised volailiy from ha poin onwards also decreases markedly. A concree case of such a rajecory could be a oll road experiencing significan loss of raffic afer an economic recession, bu for which he residual baseload raffic is much less volaile han before he shock, and sill sufficienly high o keep he DSCR ou of he defaul sae. Such a projec would no be adequaely capured by he mean DSCR process of is original family, even hough his was he bes available saring poin o anicipae is behaviour a. PD(%) EXHIBIT 3 DSCR rajecories in he sae (m,σ) plane, for boh families. SD(%) Conraced Family Merchan Family MEAN EXHIBIT 4 Probabiliies of hard and sof defauls for conraced and merchan families; compued as he probabiliies of DSCR falling below 1. and 1.5, respecively. Probabiliy of Hard Defaul Conraced Merchan 5 1 OPERATION YEAR PD(%) Conraced Merchan 18 Probabiliy of Sof Defaul 5 1 OPERATION YEAR

18 18 A SUPPLEMENT TO PENSIONS & INVESTMENTS In his illusraion, we know he rue underlying DSCR process ha is oherwise unobservable, and how i is impaced by he negaive demand shock. The poin of he example is o show ha as we observe realised informaion for individual invesmens, our esimaes of he rue process can quickly converge o he rue value and hen rack i as i evolves during he life of he invesmen. Exhibi 6 shows he filered DSCR mean and sandard deviaion along wih he realised DSCR values and he rue sandard deviaion of he projec. As soon as he DSCR diverges from is original rajecory he model akes his new informaion ino accoun, and if he divergence persiss, fuure esimaes of he expeced value of DDSSCCRR are updaed accordingly. Likewise, iniial esimae of he volailiy of DDSSCCRR on he righ panel of Exhibi 5 are correced as informaion abou he lower realised volailiy becomes inegraed ino each poserior value. The abiliy o revise he DSCR dynamics of individual projecs direcly leads o he revision of heir risk merics. For example, Exhibi 6 shows he probabiliies of dividend lockup, sof defaul, and hard defaul, respecively, and suggess ha he negaive jump in he DSCR, combined wih lower realised volailiy of DSCR, has no noiceable effec on he projec s probabiliy of hard defaul, a negligible impac on probabiliy of sof defaul, bu a noiceable impac on he probabiliy of a dividend lockup. TOWARDS LARGER SAMPLES OF DSCR DATA This paper shows ha a powerful saisical model of credi risk relying on DSCR dynamics can be developed, and provides imporan insighs for he valuaion of privae credi insrumens in infrasrucure projec finance. I also miliaes for sandardising he daa collecion and compuaion of iems such as he deb service cover raio in infrasrucure projec finance, and for pooling his informaion in cenral reposiories where i can be used o creae he invesmen merics ha invesors need (and regulaors require) o be able o inves in large, illiquid asses such as privae infrasrucure projec deb. Such analyses will be furher developed as new daa is colleced and sandardised o improve our abiliy o rack he DSCR pah of individual and groups of infrasrucure projecs, and increase he number of conrol variables and he robusness of parameer esimaes. EDHEC is commied o he coninued developmen of his research agenda, boh in erms of daa collecion and echnological developmen. DSCR DISTANCE TO DEFAULT EXHIBIT 5 Filered DSCR quaniles and sandard deviaion for a single projec experiencing a negaive shock. Filered DSCR Mean OPERATION YEAR Observed DSCR True Mean Bayesian Mean 95% Conf In SD EXHIBIT 6.5 True Bayesian OPERATION YEAR Projec s disance o lockup, sof, and hard defauls and he corresponding probabiliies of lockup, sof, and hard defauls, esimaed using filered DSCR mean and volailiy. Disance o lockup, sof, and hard defauls DL DS DDD PROBABILIT Y(%) Filered Sandard Deviaion Probabiliies of Lockup, Sof, and Hard Defauls DSCR<1. DSCR<1.1 DSCR< OPERATION YEAR OPERATION YEAR This paper shows ha a powerful saisical model of credi risk relying on DSCR dynamics can be developed, and provides imporan insighs...

19 Defensive When Needed One of he characerisics of radiional defensive sraegies such as Minimum or Low Volailiy is ha hey are concenraed in low volailiy or low bea socks. While over a very long period hese defensive sraegies ouperform cap-weighed indices, over he shor erm, in a bull marke, hey could seriously underperform. Researchers from EDHEC-Risk Insiue have herefore developed a new muli-facor dynamic defensive sraegy approach. Insead of being solely exposed o he low volailiy facor, he Scienific Bea Muli- Bea Muli-Sraegy Relaive Volailiy (9%) index reduces porfolio volailiy by allocaing dynamically beween smar facor indices based on marke volailiy. The defensive profile of he sraegy is ramped up when high marke volailiy makes i necessary. This new approach o defensive smar bea no only produces excess reurns bu significanly reduces volailiy over he long erm compared wih cap-weighed benchmarks, while a he same ime ouperforming in bull markes. For more informaion on his new form of defensive sraegy, please conac Mélanie Ruiz on or by a melanie.ruiz@scienificbea.com.

20 2 A SUPPLEMENT TO PENSIONS & INVESTMENTS INDEXES Daa Collecion for Infrasrucure Invesmen Benchmarking Frederic Blanc-Brude Direcor Raffaelle Delacroce Lead Manager, Long-Term Invesmen Projec Financial Affairs Division, Organisaion for Economic Cooperaion and Developmen (OECD) Cledan Mandri-Perro Head of Infrasrucure Finance and PPPs, World Bank Jordan Schwarz Infrasrucure & Urban Developmen Hub, World Bank Tim Whiaker Associae Research Direcor The growing ineres of invesors for infrasrucure invesmen has been moivaed by wha Blanc-Brude (213) calls he infrasrucure invesmen narraive : he noion ha infrasrucure invesmens uniquely combine he following characerisics: Low price-elasiciy of demand for essenial services, hence low correlaion wih he business cycle; Monopoly power, hence pricing power, hence an inflaion hedge; Predicable and subsanial free cash flow; Aracive risk-adjused cash flows, available over long periods; Access o unlised, illiquid financial asses. Tha is, invesing in infrasrucure implies: Improved porfolio diversificaion; Beer liabiliy-hedging, including inflaion proecion; Less volailiy han capial marke insrumens i.e. a degree of drawdown proecion. Unforunaely, adequae benchmarks ha could assess he validiy of his inuiion do no exis oday, as 94 percen of he respondens of a new EDHEC/Global Infrasrucure Hub Survey of asse owners involved in infrasrucure invesing aess (see Blanc-Brude, Chen, and Whiaker 216). For his reason, in recen years, frequen calls have been made in policy fora for daa collecion effors o be sepped-up wih respec o infrasrucure invesmen, bu i is ofen unclear which daa should be colleced o achieve wha end and how. In his paper, we propose guidelines for collecing and reporing infrasrucure invesmen daa for he purpose of building invesmen benchmarks of privae infrasrucure deb or equiy. To esablish wha daa needs o be colleced, we sar from he reasons why infrasrucure invesmen benchmarks are in demand and lis he key quesions ha such benchmarks should be expeced o answer. WHAT ARE THE RELEVANT QUESTIONS? Asse allocaion: invesors need risk-adjused measures of performance; Prudenial regulaion: regulaors wan measures of exreme risks; Liabiliy-driven invesmen: some invesors also wan o undersand he liabiliy-friendliness of infrasrucure invesing. WHY THESE QUESTIONS CANNOT BE ANSWERED TODAY These quesions are imporan o he fuure of infrasrucure invesmen, especially for invesors wih a need o manage heir liabiliies and subjeced o prudenial rules. The curren sae of invesmen knowledge does no allow answering hem for he following reasons: 1. Marke proxies are ineffecive, as new research on lised infrasrucure confirms (see Blanc-Brude, Whiaker, and Wilde 216); 2. Exising research using privae invesmen daa is oo limied. Even he deailed sudies prepared by raing agencies fall shor of he requiremens of full-scale valuaion and porfolio models; 3. Repored financial merics are inadequae, in paricular he pervasive use of appraisals and inernal raes of reurns which are incompaible wih modern porfolio consrucion echniques. In he new EDHEC/GIH survey of asse owners, more han half declares no rusing he valuaions repored by heir infrasrucure managers. RECENT PROGRESS: FROM DEFINITIONS TO DATA COLLECTION Blanc-Brude (214) pu forward a five-sep roadmap for he creaion of infrasrucure invesmen benchmarks. They are: 1. Achieving clear insrumen definiions; 2. Developing adequae asse pricing mehods; 3. Arriving a simple ye comprehensive daa collecion guidelines; 4. Populaing a global daabase of infrasrucure invesmen daa; 5. Aggregaing individual invesmens ino represenaive reference porfolios of privae infrasrucure deb or equiy. This roadmap inegraes he quesion of daa collecion upfron, including he requiremen o collec informaion known o exis in a reasonably sandardised forma and limied o wha is necessary o implemen robus asse pricing and cash flow models. Since he publicaion of he roadmap, he firs wo seps have been underaken, and wih his paper a framework required o define and launch he daa collecion process (sep 3) now exiss. STEP 1: DEFINITION Defining infrasrucure invesmens from a financial perspecive he only relevan perspecive o build invesmen benchmarks is a necessary firs sep. For he purpose of building invesmen benchmarks, he poin of defining infrasrucure invesmen is no o declare once and for all wha angible infrasrucure is, bu o clearly define wha empirical phenomena are relevan o observe when rying o documen he infrasrucure invesmen narraive. For his purpose, a clear disincion mus be made beween infrasrucure as a maer of public policy, in which case he focus is for he mos par on indusrial funcions (waer supply, ransporaion, ec.) and ha of financial invesors who may be exposed o compleely differen risks hrough invesmens providing exacly he same indusrial funcions (e.g a real oll road and an availabiliy paymen road). As we have argued before, focusing on conracs, no concree, ends o lead o a beer undersanding of he risk profile of infrasrucure invesmens (see Blanc-Brude 213). Moreover, firms ha are acive in cerain indusrial secors and deliver infrasrucure services can branch ou ino new business areas ha are alogeher differen: for insance, a number of uiliies have had a endency o look ino he media business. Likewise, from a business model poin of view, some airpors are more akin o shopping malls han infrasrucure. Hence, defining infrasrucure invesmens as a collecion of firms in cerain indusrial secors may no allow us o observe he infrasrucure business in a clear cu manner. When observing infrasrucure invesmens, we aim o collec daa ha is no oo noisy : corresponding o invesmens ha are as close as possible o he inuiion ha we called he infrasrucure invesmen narraive, he exisence of which we are rying o assess. In he respec, subsanial progress has now been made owards idenifying hose characerisics ha can be expeced o sysemaically explain he financial performance of infrasrucure invesmens. In paricular, he growing consensus around he limied role of indusrial secor caegories in explaining and predicing performance, and he much more significan role played by conracs and by differen infrasrucure business models such as merchan or conraced infrasrucure, or differen forms of uiliy regulaion, is encouraging, as illusraed by he recen conclusions of he European insurance regulaor abou defining qualifying infrasrucure asses in Solvency-II. A number of corporae forms can hus be included in he definiion of infrasrucure invesing as long as, from he perspecive of observing he phenomenon of ineres, we can ensure ha a pure infrasrucure business is being observed and no a combinaion of, say, seapor operaions and a real esae income. STEP 2: VALUATION Once he financial insrumens ha correspond o infrasrucure invesmen are well defined, he second necessary sep is o design a performance and risk measuremen framework ha can compue robus esimaes of he merics needed o undersand infrasrucure invesmen in an asse allocaion and prudenial conex. In he absence of marke valuaions and valid marke proxies, a wo-sep approach o measuring performance is necessary: 1. Documening cash flow disribuions (deb service and dividends) o address he problem of unreliable or insufficienly repored na asse values (NAVs) or losses-given-defaul (LGDs); 2. Esimaing he relevan (erm srucure of) discoun raes, or required raes of reurns of such invesmens, and heir evoluion in ime. Here oo, progress has been made and recen research reviewed in his paper provides a framework addressing boh aspecs, aking ino accoun he availabiliy of daa, while applying bes-in-class models of cash flow forecasing

21 A SUPPLEMENT TO PENSIONS & INVESTMENTS 21 and financial performance measuremen. These advances allow us o define a lis of daa iems required o implemen hese improved mehodologies. EXHIBIT 1 Daa ypes and aribues of privae infrasrucure invesmens GUIDELINES FOR DATA COLLECTION Following our roadmap, we propose a daa collecion framework respecing he following firs principles: 1. The financial insrumens used o inves in infrasrucure mus be well-defined; 2. Benchmarking resuls mus be based on bes-in-class models of financial performance and economic impac measuremen; 3. The required daa mus already exis and be sufficienly sandard o be observable on a large scale; and 4. I mus be limied o a parsimonious lis o keep he collecion process efficien and realisic. We argue ha collecing daa for realised and forecas cash flow and even daa, adequaely caegorised by physical and business model aribues, and corresponding o a clear se of financial insrumens (and heir aribues), is sufficien o measure he performance of porfolios of privae infrasrucure invesmens. PHYSICAL ATTRIBUTES BUSINESS MODEL ATTRIBUTES Secor caegories Conracual and regulaion caegories Capaciy/size Proporion of income conraced or regulaed Technology caegories Conrac erm, regulaory cycle Geographic aribues Financial srucure CASHFLOW & EVENT ENT DATA REALISED OBSERVATION TIME FORECAST To collec his daa, a firs sep consiss in he idenificaion of all invesable infrasrucure in a given counry, and he aribuion of a unique idenifier o each firm corresponding o a poenial invesmen in eiher equiy or differen kinds of deb. For each idenified firm, wo ypes of observable daa poins are of ineres: 1. Cash flows (and cash flow raios, which may or may no be derived from balance shee iems); 2. Evens (or milesones) in he developmen of he firms and, possibly, he evoluion of is risk profile. Nex, cash flow and even daa need o be caegorised according o economically meaningful aribues. These fall ino hree caegories: 1. Physical aribues of he firm: wha and where he firm is as an infrasrucure invesmen; 2. Business model aribues of he firm: sources of revenues and coss of he firm and wheher or no he risk inheren in hese exposures is edged or insured agains, for insance via conracs wih hird paries; 3. Aribues of available financial insrumens: ype of payou srucure, conrol righs and erms applicable o a firms li abiliies and equiy. Exhibi 1 provides an illusraion. All firm and insrumen aribues should also be repored and recorded dynamically. For insance, a loan may change ineres rae over ime (and his may be known in advance), or a firm may see i s ake-or-pay off-ake conrac expire before he end of he invesmen s life. Capuring realised and DEBT & EQUITY INSTRUMENT ATTRIBUTES insrumen caegories, currency & senioriy insrumen covenans and opionaliy insrumen erms and payoffs CASHFLOW & ASSET PRICING MODELS

22 22 A SUPPLEMENT TO PENSIONS & INVESTMENTS forecas changes in ime of he aribues of eiher firms or insrumens is of paricular imporance in he case of infrasrucure invesmens because of he pah-dependency and sequenial resoluion of uncerainy, which characerises hese ype of invesmens. For example projec deb may change is mauriy dae pos-resrucuring, which is insrumenal in he conex of asse pricing and compuing duraion. Applying he framework deailed in his paper, we propose he following daa collecion guidelines: 1. Building invesmen benchmarks of highly illiquid privae asses like privae infrasrucure deb and equiy requires collecing daa repored a he underlying firm level; 2. These firms should be caegorised according o a limied se of aribues which can be expeced o sysemaically explain he risk profile of individual invesmens: no only he variance bu, mos imporanly, he co-variance of cash flows and of reurns; hese include: Physical aribues: invesmen size, echnology, secor, locaion, lifecycle sage; Business model aribues: naure of income and cos sreams, role of conracs and regulaion; 3. Individual financial insrumens used o inves in such firms should also be recorded and documened o be in a posiion o predic he payoffs o differen invesors; Insrumens should be caegorised by ype of payoff profile (fixed, variable); Any condiions (covenans, embedded opions, pre paymen) should be documened o properly model he expeced payoff o invesors; 4. The wo main ypes of daa o collec relaing o he relevan firms and insrumens are sandardised evens and cash flow iems: Firm and insrumen aribues are conrol variables ha explain he dynamics of differen sream of cash flows o differen claimans (invesors); 5. Each daa poin should be repored using a dual ime frame, capuring boh he ime of observaion/reporing and ha of occurrence (pas, presen or fuure). Applying hese guidelines o collec he relevan daa allows us o implemen he ype of asse pricing and risk models ha, in urn, can be used o compue he merics needed o beer benchmark infrasrucure invesmens in privae deb and equiy. This framework for collecion daa abou privae invesmen in infrasrucure is illusraed in furher deails in he companion spreadshee o his paper, which can be downloaded a he address indicaed in he foonoe. 7 POPULATING THE DATABASE Having progressed owards clear definiions of underlying asses, and buil robus, sae-of-he-ar pricing and risk models ha avoid he pifalls of exising pracices, i is now ime o collec he relevan informaion. The daa collecion framework and emplae proposed in his paper have been designed o correspond o he requiremens of he relevan asse pricing and risk models. Hence, a raionale exiss o collec daa effecively and efficienly o build infrasrucure invesmen benchmarks. Collecing his informaion now requires large-scale cooperaion beween invesors, crediors, academic researchers and he regulaors ha can help make such reporing par of a new sandard approach o long-erm invesmen in infrasrucure by insiuional players. A raionale exiss o collec daa effecively and efficienly o build infrasrucure invesmen benchmarks. References Blanc-Brude, Frédéric Towards Efficien Benchmarks for Infrasrucure Equiy Invesmens. EDHEC-Risk Insiue Publicaions, Meridiam, Campbell-Luyens and EDHEC Research Chair on Infrasrucure Equiy Invesmen Managemen and Benchmarking. Singapore: EDHEC-Risk Insiue, 88. hp://edhec.infrasrucure.insiue/wp-conen/uploads/publicaions/blanc-brude_213u.pdf Benchmarking Long-Term Invesmen in Infrasrucure. EDHEC-Risk Insiue Posiion Paper, June. hp://edhec.infrasrucure.insiue/wp-conen/uploads/publicaions/blancbrude_214f.pdf. Blanc-Brude, Frédéric, and Majid Hasan. 2. The Valuaion of Privaely-Held Infrasrucure Equiy Invesmens. EDHEC-Risk Insiue Publicaions, EDHEC, Meridiam and Campbell-Luyens Research Chair on Infrasrucure Equiy Invesmen Managemen and Benchmarking January (January). Singapore: EDHEC-Risk Insiue. hp://edhec.infrasrucure.insiue/wpconen/uploads/publicaions/blanc-brude_2a.pdf. Blanc-Brude, Frédéric, Grace Chen, and Tim Whiaker Towards Beer Producs for Infrasrucure Invesors? A Survey of he Percepions and Expecaions of Insiuional Invesors in Infarsrucure. EDHEC Infrasrucure Insiue Publicaions, Forhcoming. Singapore: EDHEC Infrasrucure Insiue-Singapore. Blanc-Brude, Frédéric, Majid Hasan, and Omneia R H Ismail Unlised Infrasrucure Deb Valuaion & Performance. EDHEC-Risk Insiue Publicaions, EDHEC and NATIXIS Research Chair on Infrasrucure Deb Invesmen Soluions July. Singapore: EDHEC-Risk Insiue. hp://edhec.infrasrucure.insiue/wp-conen/uploads/publicaions/blanc-brude_214g.pdf. Blanc-Brude, Frédéric, Majid Hasan, Qi Wang, and Tim Whiaker Revenue and Dividend Payou in Privaely Held Infrasrucure Invesmens. EDHEC Infrasrucure Insiue Publicaions, EDHEC, Meridiam and Campbell-Luyens Research Chair on Infrasrucure Equiy Invesmen Managemen and Benchmarking March (March). Singapore: EDHEC Infrasrucure Insiue-Singapore. hp://edhec.infrasrucure.insiue/wp-conen/uploads/publicaions/blanc-brude_216a.pdf. Blanc-Brude, Frédéric, Tim Whiaker, and Simon Wilde Searching for a Lised Infrasrucure Asse Class: Mean-Variance Spanning Tess of 22 Lised Infrasrucure Proxies. EDHEC Infrasrucure Insiue Publicaions June (June). Singapore: EDHEC Infrasrucure Insiue-Singapore. hp://edhec.infrasrucure.insiue/wp-conen/uploads/publicaions/blanc-brude_216c.pdf. Brav, Omer. 29. Access o Capial, Capial Srucure, and he Funding of he Firm. The Journal of Finance 64 (1). Blackwell Publishing Inc: doi:1.1111/j x. hp://dx.doi.org/1.1111/j x. Esy, Benjamin C. 23. The Economic Moivaions for Using Projec Finance. Harvard Business School, 28. Guerrieri, Veronica, Rober Shimer, and Randall Wrigh. 21. Adverse Selecion in Compeiive Search Equilibrium. Economerica 78 (6). Wiley Online Library: Hellwig, Marin Some Recen Developmens in he Theory of Compeiion in Markes wih Adverse Selecion. European Economic Review 31 (1). Elsevier: Huberman, Gur and Shmuel Kandel Mean-Variance Spanning. Journal of Finance 42 (4). American Finance Associaion: Kan, Raymond, and Guofu Zhou Tess of Mean-Variance Spanning. Annals of Economics and Finance 13: doi:1.2139/ssrn Laffon, J J, and D Marimor. 22. The Theory of Incenives: The Principal Agen Model. Princeon Universiy Press. Michaely, Roni, and Michael R. Robers Corporae Dividend Policies: Lessons from Privae Firms. Review of Financial Sudies 25 (3): doi:193/rfs/hhr18. hp://rfs.oxfordjournals.org/conen/25/3/711.absrac. Rohschild, Michael, and Joseph Sigliz Equilibrium in Compeiive Insurance Markes: An Essay on he Economics of Imperfec Informaion. Springer. Spence, Michael Job Marke Signaling. The Quarerly Journal of Economics 87 (3). Oxford Universiy Press: hp://edhec.infrasrucure.insiue/wp-conen/uploads/documens/_example_emplae.xlsx

23 Insiue JOIM-Oxford-EDHEC Reiremen Invesing Conference Sepember 216 Oxford Universiy - Oxford, Unied Kingdom Keynoe address will be delivered by Rober Meron, Nobel Prize Winner Dinner Sponsor Endorsing Parner Media Parner New Developmens in Reiremen Invesing On Geing More Reiremen Income Benefis from he Reiree s Asses: Annuiies and he Reverse Morgage New Developmens in Reiremen Invesing Mass Cusomisaion versus Mass Producion in Reiremen Invesmen Managemen: Addressing a Tough Engineering Problem Funding Raio Flaws Ask a Consulan? The Role of Invesmen Consulans in Pension Fund Governance Errors Hacking Reverse Morgages On he Asse Allocaion of a Defaul Pension Fund Does Hiking Damage Your Wealh?

24 EDHEC-Risk Smar Bea Day Norh America 216 In parnership wih Smar Bea Indexaion & Facor Invesing Performance of Value Indices Facor Invesing and Emerging Markes Wha can be Learned from Academic Research on Smar Bea Invesing Robusness and Live Performance of Smar Bea Smar Bea Soluions Smar Bea Defensive Sraegies Facor Exposure Conrol and Risk Allocaion Soluion Official Media Parner Insiue 14 December 216 The Princeon Club New York