What Are The Key Risks Associated With Private Investment In Start Up Toll Road Projects In Developing East Asian Economies

1,108 views
1,052 views

Published on

MBA Dissertation, 2006

Published in: Technology, Economy & Finance
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
1,108
On SlideShare
0
From Embeds
0
Number of Embeds
11
Actions
Shares
0
Downloads
17
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

What Are The Key Risks Associated With Private Investment In Start Up Toll Road Projects In Developing East Asian Economies

  1. 1. Henley Management College What are the key risks associatedwith private investment in start-uptoll road projects in Developing East Asian Economies? Richard F. Di Bona ID No.: 1005661 Dissertation submitted in partial fulfilment of therequirements of Master of Business Administration 2006
  2. 2. Dissertation Richard F. DI BONAHenley Management College (1005661) ACKNOWLEDGEMENTSI am indebted to many for assistance and advice given during the preparation of thisDissertation. Firstly, to my supervisor, David Parker; also to all the staff of the HenleyHong Kong office, and to Ken Bull in Henley.Within transport planning and associated professions, there are simply too many peopleto thank individually. I believe I have learnt something from almost everyone I haveworked with over the last 14 years, who afforded me the opportunity to work across afascinating mix of countries. Over the last couple of years I have picked the brains ofmany colleagues and clients, past and present; and due to frequent commercialsensitivity, many comments and discussions have been on an anonymous basis. Manyalso acted as disseminators of my questionnaire and as “sounding boards” to discussideas and informally corroborate “ball park” figures used in the Monte Carlo risksimulations.I should also like to thank Consolidated Consultants in Amman, for their assistance withprinting the Dissertation.Finally and most importantly, I must thank my wife Mariles for her moral supportthroughout the course of my MBA studies and our daughter Vanessa (for helping metake my mind off of my studies for essential relaxation).DissFinal i December 2006
  3. 3. Dissertation Richard F. DI BONAHenley Management College (1005661) DECLARATIONI confirm that this Dissertation is my own original work. It is submitted in partialfulfilment of the requirements of Master of Business Administration in the Faculty ofBusiness Administration of Henley Management College. The work has not beensubmitted before for any other degree or examination in any other university.DissFinal ii December 2006
  4. 4. Dissertation Richard F. DI BONAHenley Management College (1005661) ABSTRACTSince the 1980’s there has been a resurgence in private sector involvement ininfrastructure, especially in tolled highways, including in developing economies(Malaysia, Mexico and Thailand were early adopters). Activity expanded during the1990’s across much of Latin America and East Asia, the latter region being where theauthor has worked extensively. Following a slowdown in the aftermath of the 1997Asian Financial Crisis, activity has recently picked-up again.The 1980’s and 1990’s were characterised by generally declining price inflation andinterest rates; whereas now there is evidence of them increasing. Based on theKondratieff Wave (long-term business cycle; a.k.a. “K-Wave”), price inflation andinterest rates could be expected to trend upwards significantly over the coming 10-15years. This Dissertation seeks to determine whether this will significantly change thenature of project risk. Thus the specific hypothesis is: “There is a significant change in the nature and extent of project finance risks for private stakeholders in East Asian toll roads during a period of increasing price inflation and interest rates”The focus is on inter-urban toll roads in Cambodia, Mainland China, Indonesia, Laos,Malaysia, Myanmar, the Philippines, Thailand and Vietnam.The Literature Review begins with basic taxonomy and a review of infrastructureprivatisation trends (globally and in East Asia), illustrating likely future demand.Financial valuation methods are reviewed, suggesting that whilst FIRR and NPV can beused, the upfront capital-intensity of toll roads makes annual ratios such as Return onCapital Employed less relevant to ex ante project evaluation. Generic project risks areDissFinal iii December 2006
  5. 5. Dissertation Richard F. DI BONAHenley Management College (1005661)then investigated, showing that most project-risks are “front-loaded” on toll roads. TheKondratieff Wave is then introduced and its potential applicability discussed, followedby Kuznets’ work on both infrastructure development cycles and developmenteconomics. The implications of cycles on over-investment are then discussed, withspecific emphasis on the genesis and aftermath of the 1997 Asian Financial Crisis.Transport modelling theory is presented, followed by discussion of traffic risks andforecasting issues, resulting variously from uncertainty, institutional risks andmethodological weaknesses, but also demonstrating the primacy of economic growth onoutturn performance. Construction risks are also considered, followed by a briefdiscussion of other issues (primarily related to governance and business norms).Forecasts of toll road demand and construction cost have often been unreliable, withserial underestimation of cost and overestimation of demand.Environmental analyses of the East Asian countries studied are then presented, usingPESTLE and stakeholder analysis. Focussing on Thailand (for consistency with theLiterature Review’s analysis of the Asian Financial Crisis), recent economicperformance is assessed, suggesting that recovery is underway. Potential growth invehicle ownership and the demand for roadspace is then considered, benchmarking thestudied countries against more developed economies; this shows substantial up-sidepotential. The performance of a number of Chinese expressways is then examined. Theopportunities and threats facing the studied countries are discussed, grouping thecountries into three categories corresponding to risk-versus-potential characteristics.Finally, analysis of gold price and treasury bill rates are used to postulate the currentglobal economy’s position on the K-Wave, showing that it is likely in the early stages ofan upswing.DissFinal iv December 2006
  6. 6. Dissertation Richard F. DI BONAHenley Management College (1005661)Next, practitioner perceptions, expectations and experience were tested using aquestionnaire survey (which generated over 160 responses; respondents having a meanof 20.6 years’ working experience). These showed that legal and political factors weredeemed most significant; but once detailed evaluation (i.e. modelling) commences,economic factors predominate. As expected, data were perceived as less available andreliable in developing economies. However, no strong preferences regarding the choiceof modelling method were shown; rather that the approach should be tailored to eachproject in turn. Under-forecasting demand seemed rare and over-forecasting it relativelycommon, in line with Literature Review findings. There was evidence of transportmodellers being pressured by clients to adjust forecasts. There was also evidence thatmany forecasters do not appreciate differences between equity- and debt-side evaluationrequirements. NPV and FIRR are both widely used in evaluation. Based on perceptionsof individual countries’ prospective toll road markets, the country categorisationsproposed in the environmental analysis were broadly supported (with the exception ofIndonesia being seen more bearishly by respondents). Interestingly, respondents seemedto generally expect many symptoms of the K-Wave upswing, in terms of rising interestrates and price inflation. However, they were not that convinced of the impacts of theseparameters on forecast performance.Consequently, Monte Carlo risk simulation modelling was employed to quantitativelytest likely impacts of different risk elements. The model comprised traffic/ revenueforecasts and financial analysis for a notional inter-urban start-up toll road facility.10,000 model runs were undertaken, with each run tested over three economicscenarios, namely:DissFinal v December 2006
  7. 7. Dissertation Richard F. DI BONAHenley Management College (1005661) “Conventional Case” based on recent previous forecast modelling assumptions (e.g. interest rates, price inflation and economic growth at levels similar to recent years); “Respondents’ Case” based on expectations gauged from the questionnaire survey (with slightly higher economic growth, interest rates and price inflation, but markedly higher fuel cost inflation); and, “Kondratieff Case” based on K-Wave upswing conditions (higher economic growth, interest rates and price inflation; though fuel price inflation at the same level as the Respondents’ Case).The Respondents’ Case tended to give the most optimistic results, but results were morevariable than in the Conventional Case. Meanwhile, results from the Kondratieff Caseappeared quite volatile, tending to support theory. Furthermore, interest rates wereshown to become substantially more important to overall risk as they rise; and priceinflation may also increase in importance. Under Kondratieff Case conditions, ifeconomic growth outstrips the impacts of rising price inflation and interest rates, thenprojected returns can be quite significant.What the above implies is that the nature and extent of project finance risks for privatestakeholders are indeed likely to change as price inflation and interest rates increase.However, if investors can lock-in fixed-rate debt (e.g. issuing bonds) before interestrates increase significantly, these risks can be mitigated. Price inflation subsequent tothe issuing of bonds would also serve to decrease the real value of debt outstanding. Butdownstream refinancing is likely to prove increasingly costly (versus experience duringthe 1980s and 1990s when cheaper refinancing was often available as a consequence ofDissFinal vi December 2006
  8. 8. Dissertation Richard F. DI BONAHenley Management College (1005661)declining interest rates). In summary, therefore the hypothesis is broadly supported byevidence. Approximate word count of main text is 16,900 words. KEYWORDS Infrastructure project finance Demand forecasting Developing countries Risk analysis Long wave business cycle (Kondratieff wave) Economic growth Price inflation Interest rates Transport planning Start-up toll road facilitiesDissFinal vii December 2006
  9. 9. Dissertation Richard F. DI BONAHenley Management College (1005661) TABLE OF CONTENTS1. Introduction ............................................................................................................. 11.1 Terms of Reference/ Personal Development............................................................ 11.2 Applicability and Hypothesis ................................................................................... 21.3 Geographic Scope .................................................................................................... 31.4 Research Approach and Dissertation Structure ...................................................... 52. Literature Review.................................................................................................... 62.1 Historical Perspective and Basic Taxonomy ........................................................... 62.2 Economic Benefits of Transport Infrastructure Development ................................. 72.3 East Asian Transport Infrastructure Privatisation Trends ...................................... 82.4 Financial Valuation ................................................................................................. 92.5 Project Risk Analysis ............................................................................................. 142.6 The Kondratieff Wave ............................................................................................ 162.7 Kuznets Cycle, Kuznets Curve and S-Curves ........................................................ 182.8 Infrastructure Development, Cycles and Crises .................................................... 192.9 Transport Modelling .............................................................................................. 232.10 Traffic Risks and Forecasting Issues ..................................................................... 252.11 Construction, Operations and Maintenance.......................................................... 332.12 Other Considerations ............................................................................................ 352.13 Summary of Key Issues .......................................................................................... 373. Environmental Analysis ....................................................................................... 393.1 Introduction and PESTLE Analysis ....................................................................... 393.2 Political, Legal and Stakeholder Issues................................................................. 403.3 Economic Recovery ............................................................................................... 423.4 Vehicle Ownership ................................................................................................. 463.5 Traffic Performance of Existing Toll Roads .......................................................... 483.6 Opportunities and Threats ..................................................................................... 513.7 Postulated Position on K-Wave ............................................................................. 534. Questionnaire Survey ........................................................................................... 554.1 Purpose .................................................................................................................. 554.2 Design Concept and Sample Selection .................................................................. 564.3 Questionnaire Design and Survey Execution ........................................................ 574.4 The Survey Sample ................................................................................................. 584.5 Tollway Appraisal.................................................................................................. 624.6 Transport Modelling Issues ................................................................................... 654.7 Forecast Performance and Evaluation Criteria .................................................... 674.8 Countries’ Outlooks ............................................................................................... 704.9 Economic Outlook ................................................................................................. 734.10 Other Comments .................................................................................................... 754.11 Key Conclusions from the Questionnaire Survey .................................................. 755. Risk Simulation Modelling ................................................................................... 775.1 Introduction ........................................................................................................... 775.2 The Case Study and Its Parameterisation ............................................................. 785.3 Methodology .......................................................................................................... 825.4 Comparison of Cases under “Base Run” .............................................................. 845.5 Comparison of Simulation Results between Cases ................................................ 85DissFinal viii December 2006
  10. 10. Dissertation Richard F. DI BONAHenley Management College (1005661)5.6 Analysis of Individual Risks ................................................................................... 885.7 Discussion of Results ............................................................................................. 916. Discussion and Conclusions.................................................................................. 926.1 Introduction ........................................................................................................... 926.2 Evaluation Criteria and Implications of the Time-Nature of Risk ........................ 936.3 Macro-Level Risks and Opportunities ................................................................... 946.4 Market Risks .......................................................................................................... 966.5 Forecasting Risks................................................................................................... 986.6 Is the Market Anticipating a Change in the Rules-of-the-Game? ....................... 1006.7 What Lessons for Practitioners? ......................................................................... 1016.8 Conclusions: Evaluation of Hypothesis ............................................................... 103References: Literature ................................................................................................ 105References: Internet Resources ................................................................................. 117Appendices ................................................................................................................... 119 LIST OF TABLESTable 2.1: Investment and Maintenance Needs in East Asia, 2006-2010 ......................... 8Table 2.2: Bain and Polakovic Forecast Performance Statistics ..................................... 26Table 2.3: Bain and Wilkins Ramp-Up Revenue-Adjustment Profiles .......................... 30Table 2.4: Estimated Expressway Construction Costs .................................................... 34Table 2.5: Operations and Maintenance Costs ................................................................ 34Table 2.6: Summary of Key Risks and Issues ................................................................ 38Table 3.1: Highlights of PESTLE Analysis .................................................................... 39Table 3.2: Vehicle, Trip and Expressway Patronage Income Elasticities....................... 48Table 4.1: Aggregated Respondent Experience Categories ............................................ 58Table 4.2: Respondents’ Mean Years’ Experience in Various Fields ............................ 60Table 4.3: Respondents with Experience in Study Area ................................................. 61Table 4.4: Rankings of Macro-Level Risks by Respondent Category ............................ 63Table 4.5: Rankings of Project-Level Risks by Respondent Category ........................... 64Table 5.1: Basic Link Characteristics of Case Study Network ....................................... 79Table 5.2: Assumed Trip Distribution (% by O-D Pair) ................................................. 79Table 5.3: Comparison of “Base” Runs between Cases ................................................. 85Table 5.4: Summary Results from Simulation Runs ....................................................... 86Table 5.5: Rankings of Risk Categories’ Importance by Case ....................................... 89 LIST OF FIGURESFigure 1.A: Map of East Asia ........................................................................................... 4Figure 1.B: Research Approach ........................................................................................ 5Figure 2.A: Standard & Poor’s Risk Pyramid ................................................................. 14Figure 2.B: Transport Concession Risks ......................................................................... 15DissFinal ix December 2006
  11. 11. Dissertation Richard F. DI BONAHenley Management College (1005661)Figure 2.C: Kuznets Curve and S-Curve......................................................................... 18Figure 2.D: Indexed Thai Real GDP and M2, 1991-1999 .............................................. 19Figure 2.E: Baht-US$ Exchange Rate 1994-2001 .......................................................... 20Figure 2.F: Dollarised Thai GFCF 1994-2001 ................................................................ 21Figure 2.G: Demand, Revenue and Price Elasticity of Demand ..................................... 27Figure 3.A: Typical Concession Stakeholder Map ......................................................... 40Figure 3.B: Thai GFCF 1993-2006 (Rolling Annual Average by Quarter) .................... 43Figure 3.C: Thai GFCF 1993-2006 (Rolling Annual Average by Quarter) in US$ ....... 43Figure 3.D: Thai GFCF, GDP and M2 in Baht, Indexed to 1995 ................................... 44Figure 3.E: Thai GFCF, GDP and M2 in US$, Indexed to 1995 .................................... 44Figure 3.F: Thai GFCF, GDP and M2 in US$, Indexed to 2000 .................................... 44Figure 3.G: Currency Performance since 1994 ............................................................... 45Figure 3.H: Currency Performance since 2001 ............................................................... 45Figure 3.I: Relationship between Wealth and Roads Per Capita .................................... 47Figure 3.J: Relationship between Wealth and Road Density .......................................... 47Figure 3.K: Traffic Growth on Shanghai-Nanjing Expressway...................................... 50Figure 3.L: Traffic Growth on Shanghai-Hangzhou-Ningbo Expressway ..................... 50Figure 3.M: Interest Rates, Nominal Gold Price and Kondratieff Wave ........................ 54Figure 4.A: Respondents by Experience Type ................................................................ 59Figure 4.B: Respondents by Years of Experience .......................................................... 59Figure 4.C: Respondents’ Global Experience ................................................................. 60Figure 4.D: Respondents with Experience in East Asia ................................................. 61Figure 4.E: Attitudes to Macro-Level Risks ................................................................... 63Figure 4.F: Attitudes to Project-Level Risks .................................................................. 64Figure 4.G: Data Availability and Reliability ................................................................. 65Figure 4.H: Attitudes to Transport Model Types ............................................................ 66Figure 4.I: Perceptions of Forecast Performance ............................................................ 68Figure 4.J: Which Forecast Outputs are Considered? ..................................................... 69Figure 4.K: How Often Are Which Criteria Considered?............................................... 69Figure 4.L: Perceived Tollway Market Opportunities by Country ................................. 71Figure 4.M: Impact of Experience on Country Perceptions ........................................... 71Figure 4.N: Country Perceptions by Respondent Category ............................................ 72Figure 4.O: Economic Expectations ............................................................................... 73Figure 4.P: Economic Expectations by Respondent Group ............................................ 74Figure 5.A: Case Study Notional Network ..................................................................... 79Figure 5.B: Volume/Capacity-to-Speed Relationships ................................................... 83Figure 5.C: Cumulative Probability Distribution of FIRR (excluding FIRR<0%) ......... 87Figure 5.D: Cumulative Probability Distribution of Payback Period (years) ................. 87Figure 5.E: Cumulative Probability Distribution of NPV at 16% ($m) .......................... 87DissFinal x December 2006
  12. 12. Dissertation Richard F. DI BONAHenley Management College (1005661) GLOSSARY OF TERMS AND ABBREVIATIONS ADB Asian Development Bank, Manila ASEAN Association of South East Asian Nations BOO Build-Own-Operate (concession form) BOOT, BOT Build-Own &/or Operate-Transfer (concession form) Billion One thousand million, being the international financial standard (as opposed to the strict/ traditional British definition of a million million) China For the purposes of this Dissertation, China is analogous to Mainland China, being the People’s Republic of China, excluding the Special Administrative Regions of Hong Kong and Macau and also excluding Taiwan. CIA Central Intelligence Agency, United States of America DBFO Design-Build-Finance-Operate (concession form) EIRR Economic Internal Rate of Return comprising FIRR plus social impacts Factory Gate Referring to prices of goods once manufactured but not transported, either to port or end user. FCO Foreign and Commonwealth Office, United Kingdom FDI Foreign Direct Investment FIRR Financial Internal Rate of Return FOB Free On Board: being the price of cargo loaded onto a maritime vessel GMS Greater Mekong Subregion, comprising Cambodia, Laos, Myanmar, Thailand, Vietnam plus Guangxi and Yunnan Provinces of China Guanxi meaning connections, a term covering business networks, political connections and a broad sense of developing and maintaining goodwill; see Appendix 6 for full definition HHI Hopewell Highway Infrastructure Limited IBRD International Bank for Reconstruction and Development, analogous with WB IPFA The International Project Finance Association IRR Internal Rate of Return, taken to be analogous to FIRR JBIC Japan Bank for International Cooperation and Development, Tokyo JICA Japan International Cooperation Agency K-Wave Kondratieff Wave or Cycle KOICA Korea International Cooperation Agency Kondratieff Spelling adopted for Kondratieff; alternative Latin spellings include Kondratyev, Kondratiev (original Russian: Кондратьев) NESDB National Economic and Social Development Board, Thailand NPV Net Present Value PBA Parsons Brinckerhoff (Asia) Ltd. PPP Public Private Partnership (when discussing project financing models) PPP Purchasing Power Parity (when discussing national income accounting concepts, such as GDP and GDP per capita), this in contrast to figures derived based on official exchange rates ROT Rehabilitate-Own/Operate-Transfer (concession form) SWHK Scott Wilson (Hong Kong) Ltd/ Scott Wilson Kirkpatrick (Hong Kong) Ltd (including joint-consultant reports with Scott Wilson as one of the authors) UNESCAP United Nations Economic and Social Commission for Asia and the Pacific, Bangkok, Thailand US$ United States Dollars VOT Value of Time: equivalencing time and money in behavioural models. WACC Weighted Average Cost of Capital WB The World Bank, Washington, D.C.DissFinal xi December 2006
  13. 13. Dissertation Richard F. DI BONAHenley Management College (1005661)1. Introduction1.1 Terms of Reference/ Personal DevelopmentFor 14 years, I have worked in transport planning, economics and demand forecastingacross 20 countries/territories, mostly on transport infrastructure scheme appraisal, oftenfor privatisation, and usually in East Asia (covering rich, “tiger” and poor economies).One reason for pursuing the MBA, the Business Finance Elective and this Dissertationtopic was to gain a more comprehensive understanding of projects’ financial risks.Hopefully to make me a “better” demand forecaster and broader project appraiser.During the course of my MBA I rekindled interest in aspects of economics, mostnotably business cycles, leading me to the Kondratieff Wave. This postulates a cycle of48-60 years duration; comprising inter alia phases of increasing interest rates andcommodity prices followed by decreases in same. Given recent increases in FederalReserve interest rates and commodity prices, Kondratieff theorists posit acommencement of an “upswing” phase, qualitatively different from the “downswing” ofthe 1980’s and 1990’s; potentially changing the relative importance of different aspectsof investment risk. Given most transport privatisation and associated literature andexperience are based on “downswing” conditions, reviewing these based on “upswing”conditions could be timely.Though focussed on profit maximisation (through risk management), betterunderstanding of changing risks should result in more efficient use of capital by private,public and aid agency sectors alike.DissFinal Page 1 December 2006
  14. 14. Dissertation Richard F. DI BONAHenley Management College (1005661)1.2 Applicability and HypothesisThe Dissertation focuses on East Asia which is again emerging as a “powerhouse” ofeconomic growth, with commensurately strong demand for transport anticipated. TheWorld Bank (2003a) notes resurgent private sector involvement in infrastructureprovision since the 1980’s, with substantial tollway activity in East Asia (US$34 billionduring 1990-2001 into 149 projects). Although activity slowed following the AsianFinancial Crisis (AFC), by 2001 it returned to 1995 levels. Yepes (2004) expectshighways to be the second biggest infrastructure investment sector in East Asia during2006-2010. In addition to providing profit opportunities, there is evidence that projectscould facilitate substantial economic growth in poorer economies, as well as “tiger”economies (Corbett et al, 2006).However, besides a potential legacy of over-investment prior to the AFC (Di Bona,2002) suppressing the attractiveness of certain new projects, following 20 years ofdeclining interest rates and price inflation, it appears that they are now rising (Faber,2002). Arguably this is connected with an upturn in the long-wave business cycle(Kondratieff, 1926). Thus, the specific hypothesis is: “There is a significant change in the nature and extent of project finance risks for private stakeholders in East Asian toll roads during a period of increasing price inflation and interest rates”DissFinal Page 2 December 2006
  15. 15. Dissertation Richard F. DI BONAHenley Management College (1005661)1.3 Geographic ScopeEast Asia is a large, diverse region, including some of the World’s richest and poorestsocieties, with differing political and legal systems and levels of economic openness.This Dissertation is concerned with its developing economies, which are likely tobenefit as: manufacturing hubs for the world; markets in their own right; and/or, naturalresource providers. It is in such economies that transport infrastructure demand growthmay be most marked.Whilst the literature review is deliberately broad, and the questionnaire survey relativelyso, the main focus is on inter-urban toll roads. Countries are included based on being: Sufficiently large (geographically) to accommodate inter-urban tolled highways; Developing economies; and, Countries where the author has at least some project experience.The countries thus considered are: Cambodia, China1, Indonesia, Laos, Malaysia,Myanmar, Philippines, Thailand and Vietnam; highlighted in Figure 1.A.Appendix 1 gives key demographic and economic data on these countries and a fewothers for benchmarking purposes. Appendix 2 gives headline transport statistics.Whilst countries such as China are anticipated to continue requiring and attractinginvestment in roads, increased scope for PPP is expected in other countries also.1 Being Mainland China, i.e. excluding Hong Kong SAR, Macau SAR and TaiwanDissFinal Page 3 December 2006
  16. 16. Dissertation Richard F. DI BONAHenley Management College (1005661) Mongolia N.Korea S.Korea Japan CHINA Hong Kong LAOSMYANMAR PHILIPPINES VIETNAMTHAILAND Brunei CAMBODIA MALAYSIA Singapore INDONESIA Timor-LesteSource of base map: Google EarthTM 2Figure 1.A: Map of East Asia2 Study Area countries in red on yellow text. Other countries/ territories in black on grey text.DissFinal Page 4 December 2006
  17. 17. Dissertation Richard F. DI BONAHenley Management College (1005661)1.4 Research Approach and Dissertation StructureThe outline research approach is presented in Figure 1.B; also giving relevant Chapternumbers. 1. Introduction and Hypothesis Including definition of geographic scope 2. Literature Review 3. Environmental Analysis Including a priori evaluation Including country economics and and analysis thereof tollway market potential 4. Questionnaire Survey Analysis of respondent perceptions against findings of Literature Review and Environmental Analysis 5. Risk Simulation Modelling Quantitative testing of impacts of different economic assumptions and evaluation of relative importance of different risks, incorporating findings of Chapters 2, 3 & 4 6. Discussion and Conclusions Collating, comparing and summarising findings from Chapters 2, 3, 4 & 5. Evaluation of initial hypothesis and identifying areas for possible future investigation.Figure 1.B: Research ApproachDissFinal Page 5 December 2006
  18. 18. Dissertation Richard F. DI BONAHenley Management College (1005661)2. Literature Review2.1 Historical Perspective and Basic TaxonomyPrivate transport infrastructure financing and operation dates back to at least the 19thCentury, including railways (e.g. UK and USA) and the Suez Canal. IPFA (2006) notesfollowing the First World War government resumed most infrastructure provision,financing projects from public debt; subsequently developing countries followed thispractice, borrowing from development agencies (e.g. WB, ADB).By the 1980’s, government debt constrained public financing of schemes, especiallygiven high interest rates; yet economic and demographic forces continued to demandinfrastructure. Thus was private involvement reborn.There is much overlapping taxonomy regarding types of project privatisation. Guislainand Kerf (1995) note a continuum of options for private sector involvement, fromsupply and service contracts through leasing (wherein management of a built project islet to the private sector in exchange for a revenue-share and/or up-front payment) toBuild-Own/Operate-Transfer (BOT, BOOT) and Build-Own-Operate (BOO); wherein,the project is constructed then operated by the private concessionaire either in perpetuity(BOO) or for a fixed period (BOT). Other forms include Design-Build-Finance-Operate(DBFO) wherein the prospective concessionaire undertakes the design as well as buildof the project, often being wholly responsible for financing.DissFinal Page 6 December 2006
  19. 19. Dissertation Richard F. DI BONAHenley Management College (1005661)2.2 Economic Benefits of Transport Infrastructure DevelopmentWhilst SACTRA (1994) questioned the benefits of additional trunk roads in developedeconomies with built-out highway networks, in developing economies new highwaysoften facilitate economic development. Christensen and Mertner (2004) showedCambodia’s factory gate price advantage over China for garments negated by transportcosts: China FOB prices are lower than Cambodia’s. Di Bona (2005) notedrehabilitation of Cambodia’s road networks transformed traffic levels and patterns;subsequent quantification estimated nationwide road traffic levels increased 83.6%above trend following the rehabilitation-to-date of roughly half of the trunk roadnetwork3 (Corbett et al, 2006, p.A2-99). The benefits of transport infrastructure indeveloping countries can be attested by increasing development aid for same (Luu,2006).In economic terms, rehabilitation greatly reduces generalised costs of travel (e.g. time,fuel, vehicular wear-and-tear and hence fares/ tariffs). Buchanan (1999) recommendsgovernments only approve projects yielding a given socio-economic return, beforedetermining likely profitability.Klein et al (1996) note privatisation appears to increase implementation costs, partiallydue to private sector participation bringing true costs to light. It also increases fundsavailable for development.3 83.6% estimated statistically, with traffic growth attributable directly to economic growth excluded.DissFinal Page 7 December 2006
  20. 20. Dissertation Richard F. DI BONAHenley Management College (1005661)2.3 East Asian Transport Infrastructure Privatisation TrendsDeveloping countries’ transport infrastructure privatisation began in earnest in the1980’s, primarily with Malaysian, Mexican and Thai toll roads (WB, 2003a, p.126).During 1990-2001, East Asia was the second largest market, attracting US$56 billionprivate investment (41% of global total) into 229 projects (Ibid., p.135), particularly tollroads: US$34 billion into 149 projects (Ibid., pp25-26 & p.143). By 2001, China hadattracted more private investment than any other country (US$23.6 billion), andMalaysia the most per capita (US$582) (Ibid., p.136). Whilst activity slowed after the1997 Asian Financial Crisis (AFC), by 2001 it returned to 1995 levels (Ibid., p.2). Table2.1 illustrates substantial anticipated future expenditure (from Yepes, 2004); highwaysare anticipated to require the second most investment of any infrastructure category.Table 2.1: Investment and Maintenance Needs in East Asia, 2006-2010 (US$ million) (percent of GDP) Investment Maintenance Total Investment Maintenance TotalElectricity 63,446 25,744 89,190 2.4 1.0 3.4Telecoms 13,800 10,371 24,171 0.5 0.4 0.9Highways 23,175 10,926 34,102 0.9 0.4 1.3Railways 1,170 1,598 2,768 0.0 0.1 0.1 Water 2,571 5,228 7,799 0.1 0.2 0.3Sanitation 2,887 4,131 7,017 0.1 0.2 0.3 Total 107,049 57,998 165,047 4.0 2.3 6.3Buchanan (1999) notes the Malaysian boom in BOT highways followed the perceivedsuccess of the North-South Highway (PLUS) concession in 1988, through which privatefinance overcame public sector constraints and took-on risk, bringing private sectorskills and incentives to infrastructure operation. However, he believes PLUS appearedprofitable only because Government handed over 225km of existing expressway withtolling rights.DissFinal Page 8 December 2006
  21. 21. Dissertation Richard F. DI BONAHenley Management College (1005661)In China several Provincial Governments established corporations for expresswaydevelopment. Soon after completing a flagship expressway, the company would belisted with revenues raised used to acquire or develop additional highways4. Thisrelatively rapid listing contrasts with experience elsewhere (see Willumsen and Russell,1998). Meanwhile, most foreign-invested BOT or leasing projects were Joint Ventures(JV) with government retaining equity in the operating company.Elsewhere in Asia, BOT concessions were the norm, though often undertaken by listedfirms. Operators occasionally issue bonds, although this practice is more widespread inthe Americas.2.4 Financial Valuation2.4.1 NPV and IRRThe decision to pursue a project and on what terms are primarily questions of projectvaluation and risk. Higson (1995, pp.60-61) notes project value may be defined via NetPresent Value (NPV) or Internal Rate of Return (IRR). NPV values future cashflows as: n Ct NPV   1  r t (1) t 0Where: Ct is net cashflow in period t r is the discount rate (equivalent to opportunity cost of capital) n is the number of periods covering the concession periodIRR expresses scheme value in terms of a percentage return on capital invested, beingthe discount rate at which NPV is exactly nought:4 See Appendix 3 for examples.DissFinal Page 9 December 2006
  22. 22. Dissertation Richard F. DI BONAHenley Management College (1005661) n Ct NPV   0 t 0 1  R t (2)Ct can include social benefits of the scheme (see Section 2.2), as well as social costs(e.g. displacement, environmental degradation etc; not covered in this Dissertation)when used for social analysis.The Fisher-Hirshleifer theorem (ibid, pp.66-67) states firms should undertake projects ifreturn is greater than investors’ required return. Highways require substantial up-frontinvestment and traffic flows often take a few years to build-up to “break even” levels;attractiveness is greatly affected by timing of revenue receipts and the discount rate, aswell as by initial investment size.Investors treat own target FIRR as strictly confidential; so no directly citeable values areavailable. However, from the Author’s experience corroborated by off-the-recordconversations with fellow practitioners, a target FIRR of 16% p.a. is the usual thresholdrequired. This includes a modest risk premium (see 2.4.2); for particularly high riskprojects, or when capital is more expensive, FIRR would increase accordingly.2.4.2 CAPM and WACCThe above assumes certainty regarding all project aspects, including: demand, priceinflation for inputs, selling price, construction cost and time, operating period, implicitassumption of no sovereignty risks etc; yet uncertainty bedevils these parameters. TheCapital Asset Pricing Model (CAPM; ibid., p.123) suggests the return on a risky projectrj is: rj  ri   j (rm  ri ) (3)where: ri is the return on riskless borrowing/ lendingDissFinal Page 10 December 2006
  23. 23. Dissertation Richard F. DI BONAHenley Management College (1005661) rm is the return on the money market as a wholeThe risk premium for j is a proportion βj of overall market risk-premium, as follows:  jm j  (4) m 2Required return can also be calculated as Weighted Average Cost of Capital (WACC;ibid, p.279): E MV  K e   DMV  K d  WACC  E MV  DMV  (5)Where: EMV is total market value of equity employed DMV is total market value of debt employed Ke is cost of equity, given by (6) Kd is cost of debt, given by (7)  Dividend  Ke     ExpectedDi videndGrowth (6)  Share Pr ice     Debenture Pr ice (%ofFaceValu e)   1  TaxRate  InterestRa te Kd    (7)  From (3) and (7) the Fisher-Hirshleifer theorem can be restated as pursue projects if: ri   j (rm  ri )  EMV  K e   DMV  K d  EMV  DMV  (8)2.4.3 Treatment of Price InflationOften (especially in transport scheme appraisal) a constant inflation rate is assumed withcalculations based in real prices (akin to zero price inflation throughout). Such priceneutrality simplifies calculations; however, it does preclude analysis of price-risksassociated with individual project inputs and outputs.DissFinal Page 11 December 2006
  24. 24. Dissertation Richard F. DI BONAHenley Management College (1005661)2.4.4 Problems with CAPM and WACCβj might theoretically be known for existing highways, but is unknown for new projects.There may be insufficient local data to determine  m . β is intended for fully diversified 2investors, rather than appraising a scheme in isolation. Higson (ibid., p.136) notesCAPM assumes: (i) perfect markets, without taxes and transaction costs, full, freely available information and no-one with price-making power; (ii) investors are rational, risk-averse, wealth-maximising, with homogenous expectations of the future; (iii) assets are marketable and infinitely divisible, with normally distributed returns; and, (iv) there is a risk-free asset for comparison.Yet transaction costs can be substantial (professional fees, cross-border know-how, etc);information is imperfect and expectations are heterogeneous. Given skill-sets required,infrastructure investors are unlikely to be highly diversified. Highway projects’ sizemakes them relatively illiquid. There may be no risk-free asset: money is only risk-freeif possible depreciation/ price inflation is ignored.Lumby (1983) notes unless a project is financed with the same capital structure as thefirm itself (unlikely), WACC changes once the project is undertaken. Furthermore,WACC assumes constant cashflows and that project systematic risk to equal that of thecompany’s existing projects; both highly unlikely.DissFinal Page 12 December 2006
  25. 25. Dissertation Richard F. DI BONAHenley Management College (1005661)Ormerod (2005, p.173) notes whilst CAPM requires a normal probability distribution inderivative markets, they exhibit power-law behaviour; this discrepancy caused the 1998collapse of Long Term Capital Management. Whilst CAPM supports currencydiversification (e.g. in borrowing), Beaverstock and Doel (2001) note such borrowingcollapsed Steady Safe (an Indonesian taxi and bus firm) and in turn PeregrineInvestment Bank.2.4.5 Financial RatiosA number of financial ratios may be used to evaluate likely project performance andrisk. Given the capital-intensity of highway construction, coupled with typically longlead-times for demand build-up (see 2.10.4), financial ratios may not always be asrelevant to ex ante project valuation.Return on Capital Employed5 is likely to be poor for early years of a concession (unlessthe project is highly geared). Likewise, Gross Profit Margin, Profit On Sales, Expensesas Percent of Turnover, Sales to Capital Employed, Sales to Fixed Assets and AssetTurnover all typically take many years to build-up to levels normally deemed acceptablein many other businesses5.Some of the above ratios might be improved by heavy borrowing, but such borrowingand resultant debt-servicing increases the importance of Working Capital Requirements,the Current Ratio and the Debt Service Coverage Ratio5. Standard & Poor’s relies onInterest Cover (debt-service coverage) as the primary quantitative measure of a project’sfinancial strength (Rigby and Penrose, 2001, p.28).5 See Appendix 4 for definitions of these financial ratios.DissFinal Page 13 December 2006
  26. 26. Dissertation Richard F. DI BONAHenley Management College (1005661)2.5 Project Risk AnalysisRigby and Penrose (2001) identify a pyramidal five-level framework for credit rating,which can be taken as a proxy for overall project investor risk, shown in Figure 2.A. Credit Credit Enhancement Enhancement Force Majeure Risk Institutional Risk Sovereign Risk Project-Level RisksFigure 2.A: Standard & Poor’s Risk PyramidProject-level risks comprise six broad elements, namely: Contractual foundations Technology, construction and operations: both pre-construction (e.g. construction delay/ quality issues) and post-construction (e.g. Operations and Maintenance) Competitive position of project within its market: including industry fundamentals, project’s competitive advantage/ likely market share, threats of new entrants, etc Legal structure, including choice of legal jurisdictionDissFinal Page 14 December 2006
  27. 27. Dissertation Richard F. DI BONAHenley Management College (1005661) Counterparty risks: e.g. extent to which JV partners can contribute equity if/when debt funding exhausted, reliability of suppliers, political risk guarantees, etc Cashflow and financial risks: in addition to expected cashflow, ability to cope with interest rate, inflation, foreign exchange, liquidity and funding risksGeorge et al (2004) note the uncertainty inherent in start-up tollways requires flexiblefinancing approaches. Willumsen and Russell (1998) illustrate project-level risks asshown in Figure 2.B. Predominating traffic/ revenue risks are discussed in Section 2.10. Construction Delay Change Orders Risk (nominal) Construction Costs Ramp Up Traffic & Revenue O&M er -2 -1 10 0 1 2 3 4 5 ov d an H YearFigure 2.B: Transport Concession RisksSovereign and institutional risks are concerned primarily with the project’s country:ratings usually constrained by government’s debt servicing/ foreign currency record,reflecting risks of currency conversion and overseas transfer. Institutional factorsDissFinal Page 15 December 2006
  28. 28. Dissertation Richard F. DI BONAHenley Management College (1005661)include business and legal institutions, which are often weak/ nascent in developingcountries, with concepts of property rights and commercial law not fully developed,potentially leaving creditors/ investors exposed. La Porta et al (1997) found investorrights in developing countries though limited, are generally better under common lawthan civil law (especially French civil law, which often has weak enforcement).Force Majeure risks includes “Acts of God” (floods, earthquakes, etc) as well as civildisturbances, strikes, changes of law. Rigby and Penrose (2001) note toll roads aretypically less affected/ can return to normal service more quickly.Credit Enhancement refers to insuring/ re-insuring specific risks. However, litigationintrinsic in such claims can delay payment by years, so mitigation may be limited.2.6 The Kondratieff WaveOrthodox economics assumes given policies produce similar results at all times;Ormerod (1999, pp.96-102) notes experience contradicts this, due to periodic exogenousshocks. Others postulate cycles responding to exogenous shocks. But to some cycleadherents, such “exogenous” shocks are mostly endogenous. Schumpeter (1939)consolidated others’ preceding work, specifying three inter-related cycles: Kitchin (1923): based on fluctuations in business inventories (39+/– months) Juglar (1863): based on business investment in plant and equipment (7-11 years) Kondratieff (1926): based on development of new technologies/ sectors and impact of their adoption on socio-economic conditions (48-60 years; a.k.a. “K-Wave”)The K-Wave postulates periodic “Creative Destruction” (Schumpeter, 1950, Chap.VII)intrinsic to industrial-capitalism. Not all cycle proponents accept the K-Wave:DissFinal Page 16 December 2006
  29. 29. Dissertation Richard F. DI BONAHenley Management College (1005661)Kindleberger (1996, p.13) calls it “possibly… dubious and elusive.” There is alsodebate on periodicity. Whilst Schumpeter believed one K-Wave contained three JuglarCycles, each comprising in turn three Kitchin Cycles, Faber (2002, p.110) notesKondratieff never postulated precise periodicity.Kondratieff’s empirical work identified a number of patterns within each cycle. Furtheranalysis by Schumpeter (1939), summarised by Faber (2002, pp.116-138) notes: Before and during the beginning of Upswings there are profound changes in industrial techniques (based on new technologies) and/or involvement of new countries in the global economy and/or development of new transport technologies. Social upheavals and international conflict are more likely during Upswings. Agricultural prices decrease during downswings; industrial prices hold steady or fall slightly. During upswings, commodity price increases can create broader price inflation. Interest rates also follow this cycle. As appears to have been the case in recent years (see Section 3.7). Upswings are characterised by brevity of depressions and intensity of booms; the opposite being true during downswings.There are separate transitional phases at peaks and troughs, usually brief in relation toUpswing and Downswing phases and largely ignored in the context of this Dissertation.Appendix 5 presents K-Waves since 1787. Maddison (1995) estimated real global GDPper capita rose 2.90% p.a. from the 1950s-1970s (K-Wave upswing); but declined to1.11% p.a. until the 1990’s (K-Wave downswing).DissFinal Page 17 December 2006
  30. 30. Dissertation Richard F. DI BONAHenley Management College (1005661)2.7 Kuznets Cycle, Kuznets Curve and S-CurvesKuznets (1930) identified a 15-25 year building construction cycle, concurring withSchumpeter that innovation drives growth endogenously to the economic cycle. He alsopostulated the Kuznets Curve (1955), plotting economic development against incomeinequality: inequality increasing in the early stages of economic development,plateauing then diminishing. Inequality can be measured using the Gini Coefficient(Gini, 1912): 0 denoting perfect equality and 100 perfect inequality (one person has allwealth).This implies few might afford cars or tolls in the early phases of growth, but aseconomies develop, tolls become substantially more affordable. Coupled with demandsaturation, this suggests an “S-Curve”, akin to the innovation/ adoption curve (Rogers,1962). Figure 2.C shows this inter-relationship between a Kuznets Curve and S-Curve,based on normal distribution. Norm al Density/ Kuznets Curve Cum ulative Norm al/ S-CurveFigure 2.C: Kuznets Curve and S-CurveDissFinal Page 18 December 2006
  31. 31. Dissertation Richard F. DI BONAHenley Management College (1005661)2.8 Infrastructure Development, Cycles and CrisesInfrastructure may facilitate Upswings, but its short-term impact may triggerDownswings, fostering “Creative Destruction” (Schumpeter, 1950): purging oldmethods/ technologies for improved methods/ infrastructure to drive Upswings.Lawrence (1999) argues major skyscraper completions are cyclical, precedingrecessions. But do build-out peaks precipitate recessions, or are they “peaks” due tosubsequent demand failure, uncorrelated with preceding build-out (as espoused byKrugman, 2000)?Di Bona (2002) analyses Thailand6, where the Baht’s flotation triggered the AFC.Figure 2.D7 shows impressive real GDP growth until 1996, when close correlation withM2 broke. Continued M2 growth refutes Krugman’s attribution of the AFC to demandfailure, which ignored structural causes. 200 300 Real GDP (1991=100) 180 260 M2 (1991=100) 160 220 140 180 120 140 100 100 1991 1992 1993 1994 1995 1996 1997 1998 1999 Real GDP M2Figure 2.D: Indexed Thai Real GDP and M2, 1991-19996 Much of these Thai analyses originally presented in Di Bona, R.F. (2002) Surviving Bahtulism7 Raw data from APEC (www.apec.org); analysis my own.DissFinal Page 19 December 2006
  32. 32. Dissertation Richard F. DI BONAHenley Management College (1005661)Before the AFC, Thailand enjoyed a virtuous economic development cycle: increasedwealth boosted investment returns, attracting further investment. Keynesian multiplier-accelerator effects boosted growth, encouraging further development. Adaptiveexpectations of investment returns fuelled excessive capital works and otherinvestments. Bangkok planned several new residential and business hubs, which couldnot all be viable simultaneously: eventually supply outpaced demand.The Baht’s July 1997 flotation coincided with doubts regarding the sustainability ofThailand’s growth. Its depreciation (Figure 2.E8) ballooned offshore-financed corporatedebt. Ensuing capital flight intensified the crisis. Long infrastructure lead-times meantthere was still supply-in-waiting; many projects were stalled or abandoned. Figure 2.F9shows GFCF collapsing with no noticeable rebound by 2001. 0.05 0.045 0.04 USD per THB 0.035 0.03 0.025 0.02 0.015 0.01 0.005 0 4 4 5 5 6 6 7 7 8 8 9 9 0 0 1 1 99 99 99 99 99 99 99 99 99 99 99 99 00 00 00 00 n -1 l-1 n-1 l-1 n-1 l-1 n-1 l-1 n-1 l-1 n-1 l-1 n-2 l-2 n-2 l-2 Ja Ju Ja J u Ja J u Ja J u Ja J u Ja Ju Ja Ju Ja J uFigure 2.E: Baht-US$ Exchange Rate 1994-20018 Source data: www.fx.sauder.ubc.ca9 Source data: www.nesdb.go.th and www.fx.sauder.ubc.caDissFinal Page 20 December 2006
  33. 33. Dissertation Richard F. DI BONAHenley Management College (1005661)Hayek (1933) argues artificially low interest rates breed over-investment, precipitatingcrises with debt- and investment-overhangs delaying recovery. Faber (2002, pp.192-193) argues global liquidity injections following the 1995 Mexican crisis fuelled furtherAsian speculative growth, delaying but ultimately amplifying and prolonging the AFC. 14,000 12,000 10,000 million USD (1988 prices) 8,000 6,000 4,000 2,000 0 1994 1995 1996 1997 1998 1999 2000 2001 2002 Gross Fixed Capital Formation Private Construction Government Construction Land Development Construction And Land DevelopmentFigure 2.F: Dollarised Thai GFCF 1994-2001Faber (2002, p.69) notes cycles are “particularly violent in the case of emergingeconomies, emerging industries and emerging companies, which grow and evolverapidly and are, therefore, capital-hungry.” Transport infrastructure construction isespecially capital-intensive.DissFinal Page 21 December 2006
  34. 34. Dissertation Richard F. DI BONAHenley Management College (1005661)Although infrastructure and utilities are often seen as defensive investments, Forsgren etal (1999) argue toll road performance is cyclical, noting with reference to China (notgenerally regarded as badly hit): Challenging business climate with (official) economic growth down to 7% p.a. Delayed construction of connector roads and reduced commerce reducing traffic growth (and occasionally traffic declines) Debt service coverage (operating revenues) short of base projections Growing doubts as to willingness and ability of local partners to pay minimum income guarantees to toll companies (note: these were abolished by decree in 2002) Increased refinancing and foreign exchange risks Periodic toll increases required to meet projections, yet approval process is opaque Problems with toll collection/ leakage Credit ratings deteriorating due to reduced credit quality of counterpartiesIn Indonesia, the rapid devaluation of the Rupiah in 1997, compounded by rapidlyincreasing fuel prices, massive economic and political uncertainty and civil unrest,substantially reduced Jakarta Intra Urban Tollroad traffic volumes (Ibid.).Such patterns are not new. Despite railways driving America’s economic developmentin the 19th Century, Faber (2002, pp.55-63) notes they exhibited cyclical booms andcrises. Moreover, historically overseas investors are often latecomers, repeatedly buyingpeaks to sell-out in the immediate aftermath of crisis.DissFinal Page 22 December 2006
  35. 35. Dissertation Richard F. DI BONAHenley Management College (1005661)2.9 Transport ModellingCorbett and Di Bona (2006) note transport models provide inter alia: assessment ofdemand-side project risks; evaluation of alternative projects against one another; and,forecasts of economic and financial returns, for use in project valuation. Traditional“Four Stage” models (elucidated in Ortúzar and Willumsen, 1994) are outlined inAppendix 7; but such models are data hungry so simplifications are common. Theirapplicability to tollways has been questioned (Willumsen and Russell, 1998).Usually the modelled area is divided into spatial zones. Traditionally, traffic to/ fromeach zone is estimated based on land-use and corresponding trip generation rates.However, given sparseness of robust land use data in developing countries, econometricmodels of traffic levels are often used. Whilst Khan and Willumsen (1986) fitted S-curve models to vehicle ownership and usage, often historical traffic counts areregressed on corresponding income data to estimate income elasticities of trafficdemand, defined as:   t1  t 0      t t   2    T   0 1   y  T  Y   y1  y0  (9)       y0  y1   2    Where: to,t1 are traffic levels in periods 0 and 1 y1,y0 are income (GDP) levels in periods 0 and 1As elasticities might not hold over time forecast values are adjusted, based either on S-curves or a conservative assumption of gradually declining elasticities, taking implicitaccount of longer-term demand saturation or improved logistic efficiency (decreasedlorry empty-running). Though these ignore vehicle ownership/ usage costs, Pindyck andRubinfeld (1981, pp.396-398) note Hymans’s (1970) model of USA vehicle ownershipDissFinal Page 23 December 2006
  36. 36. Dissertation Richard F. DI BONAHenley Management College (1005661)shows such factors have short-term impacts, income-ownership relationshipspredominating thereafter.In developing countries tollway appraisals, driver interview surveys scaled using trafficcounts are often used to obtain trip patterns. Effects of other modes (e.g. rail) arecommonly omitted; impacts might be insignificant, or data unavailable.In order to determine vehicle routeing, a variety of approaches are possible, including:Network Assignment Modelling: Where the network is complex (roads parallel andperpendicular to the toll-road significantly affecting patronage), network assignmentmodels should be used. In addition to interzonal trip matrices, the road network is coded(e.g. length, capacity, tolls and relationships between speed and congestion). Aniterative assignment process is used, with link speeds recalculated to reflect congestion.Typically forecasts are prepared for a base year, opening year and at 5 or 10-yearintervals thereafter, with intermediate years interpolated. Such models are calibrated byadjusting network coding and often using maximum entropy matrix estimation (see VanZuylen and Willumsen, 1980) to better match traffic counts.Logit-Based Corridor Modelling: A spreadsheet-based approach to model a corridor,typically with one competing route (e.g. with no/ lower tolls and lower speeds). Trafficis allocated between routes based on a logit function; (10) shows an absolute logit curvefor forecasting a new road’s traffic. For existing toll-roads incremental logit modelsmay be preferred, shown in (11). Commonly κ and λ would be estimated based onprevious studies (ideally existing toll-roads). Richardson (2004) notes a general biasagainst using toll roads (κ<0). Forecasts may be prepared for selected years(intermediate years interpolated) or for all years. Whilst congestion levels do notfeedback, increasing incomes make tolls more affordable.DissFinal Page 24 December 2006
  37. 37. Dissertation Richard F. DI BONAHenley Management College (1005661) 1 PijXt  ,       GC ij ,t  GC ijX,t L  (10) 1 eWhere: PijXt is the share of trips i→j in period t using the expressway, PijXt  PijLt  1 , , , r GCij ,t is the generalised cost for trip ij by route r (X=expressway, L=local road), comprising equivalenced time and monetary elements in year t κ, λ are calibrated parameters   1        PijXt    GCij ,t GCijX,t    L IPij ,t  Obij ,t 0   X   Obij ,t 0   X X , X 1  e    Pij ,t 0    (11)    1      1  e  GCij ,t 0 GCij ,t 0    L X  Where: ObijX,t 0 is the base year observed expressway market share for trips ij PijXt , is forecast expressway share in year t (absolute logit); t=0 is base year2.10 Traffic Risks and Forecasting IssuesBain and Wilkins (2002) analyse toll-traffic uncertainty and traffic forecast error,showing strong inter-correlation. Average initial year traffic was 70% of forecastoverall, 82% in lender-commissioned projections and 66% when commissioned byothers, suggesting commissioning party influence on forecasts: debt-financiersrelatively more concerned with down-side risk than equity-holders. Their Traffic RiskIndex (shown in Appendix 8) compares low and high risk factors for toll roads andtraffic forecasts in general.Whilst initial year errors might be due to ramp-up (see 2.10.4), which Streeter andMcManus (1999) reckon can last 3-5 years, Bain and Polakovic (2005) note optimismbias is “constant through Years 2 to 5” as shown in Table 2.2, signalling other errorsDissFinal Page 25 December 2006
  38. 38. Dissertation Richard F. DI BONAHenley Management College (1005661)(discussed below). They also note drastic differences in forecasts by different parties forthe same projects, based in part on very different assumptions.Table 2.2: Bain and Polakovic Forecast Performance Statistics Operating Year Mean Actual/Forecast Traffic Standard Deviation 1 0.77 0.26 2 0.78 0.23 3 0.79 0.22 4 0.80 0.24 5 0.79 0.252.10.1 Toll Sensitivity and the Value of TimeExcepting “shadow tolling” (operator reimbursed based on patronage instead of user-tolling), willingness-to-pay tolls is critical. Typically choice is between a slow, cheaproad and a fast toll-road; time and money equivalenced using the behavioural Value ofTime (VOT) to give “generalised cost.” Whilst higher tolls are usually preferred (see2.10.4) sometimes they are too high (Wong and Moy, 2004). The price elasticity oftollway demand is:   q1  q0     Q   q0 q1   2   D  p    P  p1  p0  (14)       p0  p1   2    Where: ΔQ is change in traffic ΔP is change in price (toll) q1,q0 are traffic after and before toll change respectively p1,p0 are new and old tolls respectivelyDissFinal Page 26 December 2006
  39. 39. Dissertation Richard F. DI BONAHenley Management College (1005661)Figure 2.G shows the relationship between demand, revenue and η. When tolls arebeneath the revenue maximising level (i.e. p<Prm) 0   D  1 , toll increases boost prevenue; when p>Prm  D  1 (toll increases decrease revenue).  D  1 when p=Prm. p pWillumsen and Russell (1998) note in developing countries Stated Preference surveys toestimate  D and VOT are scarce and of uncertain quality. Reference is often made to pprevious studies, factored for income levels. But the income elasticity of VOT, VOT is ycomplicated: as income increases, VOT rises (“income effect”), as does expenditure onother products/ services (“substitution effect”) and possibly savings too (“savingseffect”), implying VOT  1 . In developed economies, Wardman (1998) suggests yVOT  0.49 ; Gunn and Sheldon (2001) advocate 0.35  VOT  0.7 . Cross-sectional y yanalysis between developing countries suggests VOT  1 yet time-series analysis within ya country VOT  1 to growth VOT thereafter10. y Revenue Maximisation -η Demand Total Revenue η= −1 Prm Price→Figure 2.G: Demand, Revenue and Price Elasticity of Demand10 Confidential source used in absence of public source.DissFinal Page 27 December 2006
  40. 40. Dissertation Richard F. DI BONAHenley Management College (1005661)Goods vehicles are of particular concern. Bain and Wilkins (2002) note in developingcountries long-distance tolls often exceed drivers’ wages, giving incentive to useuntolled routes (pocketing bosses’ toll money). Some studies (e.g. ADB, 2003) havefailed to establish any VOT for goods vehicles.2.10.2 Competing Routes and Link RoadsContractual guarantees theoretically limit competing routes’ development, presupposingthe contracting branch of government is willing and able to enforce such guaranteesacross multiple government layers.Jiangsu Expressway circumvented this risk by acquiring rights to highways parallel totheir flagship Shanghai-Nanjing Expressway and so manage (and toll) traffic on bothroutes. However, when GZI Transport listed in 1997, it was assumed that the ferryparallel to the (then) soon-to-open Humen Bridge would cease operation. But beingoperated by a different local government, operation continued with fares undercuttingbridge tolls, attracting substantial goods vehicle volumes from the Humen Bridge.Even when concessionaires gets first refusal at planned parallel routes, overinvestmentmay result in excess infrastructure relative to traffic levels. Buchanan (1999) notes inMalaysia those identifying schemes can often proceed (subject to financing) withoutdue diligence of impacts on existing BOT’s.Though more important for urban projects, provision of adequate link roads is alsoimportant. Congested approaches/ exits can result in “hurry up and wait” (Bain andWilkins, 2002), reducing tollways’ attractiveness.DissFinal Page 28 December 2006
  41. 41. Dissertation Richard F. DI BONAHenley Management College (1005661)2.10.3 Toll Increases and Revenue GuaranteesContracts typically allow periodic price-indexed toll increases, or at a percentage ofprice inflation. However, Forsgren et al (1999) note toll increase approval processes areoften opaque and beset with delay. Bain and Wilkins (2002) note tariff escalation isoften politicised, especially where there is little previous “tolling culture.” Sometimessocial unrest follows tolls’ imposition (Orosz, 1998) or toll increases, especially duringeconomic downturns (Dizon, 2002).Some contracts give revenue guarantees to operators, underwritten by government.However, China’s 2002 State Council directive scrapped such revenue guaranteesoverriding contract provisions, leading to New World Development divesting from 13toll roads and bridges (Chan, 2003).Whilst non-toll revenues may be generated (e.g. service stations, advertising), Streeteret al (2004) note their contribution is usually dwarfed by toll revenues.2.10.4 Ramp-UpBain and Wilkins (2002) define ramp-up as information lag for users unfamiliar with anew highway and general reluctance to pay tolls (see Richardson, 2004 for experimentalevidence). Streeter and McManus (1999) reckon on 3-5 years’ ramp-up and note this isoften underestimated in traffic forecasts.Bain and Wilkins (2002) note ramp-up experience tends to cluster to extremes: either oflimited duration (even exceeding forecast traffic levels) or lagging for a long duration,maybe never “catching up”, particularly for projects with a high Traffic Risk Index (seeAppendix 8). They derived revenue-adjustment factors as per Table 2.3 for use infinancial stress-tests.DissFinal Page 29 December 2006
  42. 42. Dissertation Richard F. DI BONAHenley Management College (1005661)Table 2.3: Bain and Wilkins Ramp-Up Revenue-Adjustment Profiles Forecasts Lenders Others commissioned by Traffic Risk Low Average High Low Average High Year 1 revenue -10% -20% -30% -20% -35% -55% adjustment Ramp-up duration 2 5 8 2 5 8 (years) Eventual catch-up 100% 95% 90% 100% 90% 80%2.10.5 Operating CostsIn addition to tolls, many models also apply distance-based monetary Vehicle OperatingCosts (VOC) reflecting fuel, maintenance, depreciation, etc. Whilst economic values forthese parameters are derivable, accurate behavioural values are often elusive. In practicethey may be used to reflect certain advantages of higher quality roads, whereon wear-and-tear may be less and where smoother flow may yield fuel savings. However, theseare typically applied as fixed values with respect to distance and road-type, rather thanfeeding-back modelled forecast speeds. Where there are larger VOC savings from anexpressway ceteris paribus there is more scope for higher tolls. However, there is anissue as to who pays these costs (driver or employer).2.10.6 Toll LeakageSome vehicles use a facility without paying, either legitimately (e.g. certain governmentor military vehicles) or illegitimately. There may be theft by toll-collectors and fraud byadministrators. Forsgren et al (1999) note toll leakage can be as high as 20% ofrevenues. Sometimes computerised toll collection and auditing can restrain losses, buton lower volume routes the cost of such measures might outweigh savings.DissFinal Page 30 December 2006
  43. 43. Dissertation Richard F. DI BONAHenley Management College (1005661)2.10.7 Induced TrafficWhen a new highway significantly reduces transport costs or relieves congestion, it mayresult in additional (induced) traffic. Corbett et al (2006, p.A2-99) report substantial,rapid induction on Cambodia’s roads following rehabilitation. On green-field sites, itmay also over time enable expanded development, generating further traffic demand.However, Willumsen and Russell (1998) note the difficulty of reliably forecasting sucheffects; Bain and Polakovic (2005) report the prevalence of significant errors in inducedtraffic forecasts.2.10.8 AnnualisationBain and Wilkins’ (2002) Traffic Risk Index shows projects with seasonal flow patternstend to be riskier. For inter-urban highways a “typical” day is usually modelled, withresults factored-up to annual forecasts. Thus seasonal changes might not be captured:forecasts represent an expansion of one part of the annual pattern. Even when AnnualAverage Daily Total (AADT) traffic is modelled, larger seasonal variations equate tolarger total variance between modelled day and actual day across the year.For those projects where modelled hours are considered, mathematically the problemincreases, given further factoring from a “typical” hour (or perhaps AM peak and PMpeak) to a “typical” day. Conversely, when modelling a day, future congestion in peakperiods and its impact on effective daily capacities may be under-estimated.DissFinal Page 31 December 2006
  44. 44. Dissertation Richard F. DI BONAHenley Management College (1005661)2.10.9 Economic EffectsEconomic risks feed through many elements of traffic forecasts: Overall travel demand (e.g. car ownership and usage, freight volumes, extent of traffic induction) Willingness-to-pay tolls and try tollways (affordability; ramp-up extent and duration) Toll leakage (incentive for malfeasance) Over-investment increasing likelihood of competing routes being built/ upgradedEconomic cycles affect most aspects of the economy and decision-making, includingevaluation assumptions adopted. Transport consultants define economic growthscenarios either under guidance or instruction of commissioning parties. Whenexpectations are high more projects are evaluated, so proportionally more projects arelikely to founder on downturn (and be blamed on transport forecasts). This may createcynicism regarding tollway investments extending into the early economic recovery,resulting in under-investment in some areas, thence over-investment as returns onoperating (and newly opened) highways exceed expectations, thus creating a new“error of optimism” (Pigou, 1920).Luu (2006) and Gomez and Jomo (1999) cite governments in Vietnam and Malaysiapotentially over-expanding transport infrastructure development.DissFinal Page 32 December 2006
  45. 45. Dissertation Richard F. DI BONAHenley Management College (1005661)2.11 Construction, Operations and MaintenanceConstruction cost overruns and delay (deferred/ lost revenue) may imperil initial debtrepayments. Rigby (1999) notes using engineering, procurement and construction (EPC)contractors’ reputations to proxy technical risk is both commonplace and erroneous:construction risks are often inadequately assessed. Based on UK experience, Flyvbjergand COWI (2004) recommend highway construction cost estimates be uplifted 15% if a50% chance of overrun/ delay is acceptable, or by 32% if 20% chance acceptable.Ruster (1996) notes construction cost overruns, delays and defects can be largelymitigated by liquidated damages, performance bonds, warranties, contingency fundsand insurance. As revenue losses are rarely disputed during delay/ overrun arbitrations,the focus of this Dissertation remains on demand-side risks. However, when thecontractor is the concessionaire, such risks should be analysed. Similarly, operationsand maintenance (O&M) risks should also be considered.Table 2.4 shows estimated costs for new expressways in China and Vietnam. Whilstcosts are dependent on terrain, design standards and local labour and material costs,there is significant difference between HHI costs and others (ADB potential projects),unlikely wholly attributable to differences in local prices, or the difference betweenDual-2 and Dual-3 standard. A distance-weighted average of US$4.633m per km ofDual-2 was derived, to be used in Chapter 5’s simulation model.There is a trade-off between construction and subsequent operations and maintenancecosts. The latter also affected by periodic major maintenance (e.g. immediately beforeconcession handback). Literature review found little agreement as to how to gauge suchcosts, and whether they should be related to construction or traffic flow/ revenue. Table2.5 shows some public domain values; some confidential sources suggested using 6% ofDissFinal Page 33 December 2006

×