Description
A traffic collision, also known as a traffic accident, motor vehicle collision, motor vehicle accident, car accident, automobile accident, road traffic collision, wreck (USA), car crash, or car smash (Australian) occurs when a vehicle collides with another vehicle, pedestrian, animal, road debris, or other stationary obstruction, such as a tree or utility pole. Traffic collisions may result in injury, death, vehicle damage, and property damage.
Journal of Transport Economics and Policy, Volume 40, Part 3, September 2006, pp. 359–381
Tra?c Risk Mitigation in Highway Concession Projects
The Experience of Chile
´ M. Vassallo Jose
´ M. Vassallo, Associate Professor, Departamento de Address for correspondence: Jose ´ cnica de Madrid, Transportes, ETSI de Caminos, Canales y Puertos, Universidad Polite 28040 Madrid, Spain ( [email protected]). The author would like to thank ´ Antonio Go ´ mez-Iban Professor Jose ˜ ez of the Kennedy School of Government (Harvard University) for proposing the idea for this paper and for reading and correcting with extreme patience all the drafts he gave him. A special debt of gratitude is owed to Mark Fagan, my colleague as a visiting fellow at Harvard, for the countless hours spent helping with the structure of the paper, as well as for the thankless task of ´ lez, former director of concession proediting its ?nal version. Finally, Alvaro Gonza jects at the MOP and the ?rst person to approach the LPVR mechanism, helped obtain the information necessary to write the paper and also made useful comments about highway concessions in Chile.
Abstract Tra?c risk mitigation remains a challenging aspect of highway concessions. This paper evaluates three mechanisms applied in Chile to mitigate tra?c risk: the ‘Minimum Income Guarantee’ (MIG); the ‘Least Present Value of the Revenues’ (LPVR); and the ‘Revenue Distribution Mechanism’ (RDM). Speci?cally, the paper focuses on the performance of LPVR and MIG during the economic recession that took place between 1998 and 2002. In the context of this recession, the paper explains the reasons that led the government to implement the RDM mechanism. The paper gives some guidelines about the applicability of these mechanisms in other countries, highlights the bene?cial features of LPVR in reducing tra?c risk and avoiding concession contract renegotiations, and ?nally provides some recommendations as to how to make LPVR more attractive to private promoters.
Date of receipt of ?nal manuscript: October 2005 359
Journal of Transport Economics and Policy
Volume 40, Part 3
1.0 Introduction
Highway privatisations around the world have been completed using concession contracts, often in the form of Build, Operate, Transfer ´ mez-Iban (BOT) projects (Go ˜ ez, 2003). Successful BOT projects require the proper allocation of risks among the project’s stakeholders. Whereas some risks — such as construction, operational or legal risk — are clearly controllable by some stakeholders, tra?c risk can not be controlled by any of them. As a result, allocating tra?c risk is one of the greatest challenges in designing highway concession contracts. Chile has been a pioneer in the implementation of tra?c risk mitigation mechanisms in highway concessions. The objective of this paper is to study the e?ectiveness of those mechanisms, particularly the ‘Least Present Value of the Revenues’, and to provide recommendations for its implementation in the future. The paper is organised as follows. The ?rst section highlights tra?c forecast challenges and provides a taxonomy of the di?erent mechanisms to mitigate tra?c risk as the context for examining the Chilean experience. The second section presents a case study of the highway concessions in Chile, focusing on three mechanisms: the ‘Minimum Income Guarantee’ (MIG), the ‘Least Present Value of the Revenues’ (LPVR), and the ‘Revenue Distribution Mechanism’ (RDM). The third section provides lessons based on the Chilean experience.
2.0 The Extent of Tra?c Risk and Mitigation Strategies
Tra?c risk depends primarily on the performance of the economy, the reaction of users, and the competition with other means of transport. Thus, none of the concession stakeholders can directly mitigate the risk. Moreover, unlike other BOT projects — such as mining and oil production — in which demand risk is usually hedged through long-term purchase contracts, highway BOT tra?c risk is di?cult to hedge because of the large number of customers and lack of an insurance market. 2.1 Tra?c risk inaccuracy Tra?c demand is very di?cult to predict. Table 1 shows the results from several studies that measure the inaccuracy of tra?c demand estimation. Standard and Poor’s (S&P) compared ?rst-year forecast tra?c levels with actual tra?c volumes for 82 projects. The results of the study showed that, during the ?rst year, tra?c volumes averaged about 76 per cent of 360
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
Table 1 Tra?c Forecast Deviation Comparison in Several Studies
Main geographical areas studied North America, North Europe, Asia, South Europe, Latin America Denmark, European Union South Europe, Latin America Mean Sample (real/ size forecast) 87 0.76 Standard deviation 0.26
Study
Projects
Standard & Poor’s Toll roads (2004) Flyvbjerg et al. (2004) Vassallo (2002) Free roads Toll roads Toll roads Shadow toll roads
183 18
1.09 1.03
0.44 0.24
their predicted values, and the error — measured through the standard deviation — was 0.26 (Bain and Wilkins, 2002; Bain and Plantagie, 2004). Some data provided by Spanish concessionaires to the author report similar results in term of standard deviation although, unlike S&P data, they do not prove any bias towards underestimation. A large and recent study by Flyvbjerg et al. (2005) — focused primarily on free roads in Northern Europe — also reports substantial forecasting errors; the standard deviation estimated is 0.44. Unlike the S&P study, this study shows a slight bias towards underestimation. According to these studies, tra?c forecast errors for toll highways seems to be consistently in a range between 0.20 and 0.30 in terms of standard deviation. Assuming that tra?c deviations are distributed according to a normal function with mean equal to 1 and standard deviation equal to 0.25, which seems reasonable according to the empirical results, 42 per cent of the toll highway projects will experience ?rst-year tra?c misestimation of more than 20 per cent. This situation is accentuated by the fact that, as mentioned by Zhao and Kockelman (2002), the further in the future the tra?c forecast, the greater the inaccuracy. 2.2 Tra?c risk as a cause of renegotiation Tra?c risk in highway concessions has generally been assigned to concessionaires. That fact has often prompted asymmetrical behaviour. If ultimately the tra?c is higher than expected, the concessionaire will reap excess pro?ts, whereas if the tra?c is lower, the concessionaire will incur losses, and may attempt to force a renegotiation with the government. This asymmetric behaviour has occurred for instance in Spain (Izquierdo and Vassallo, 2004). 361
Journal of Transport Economics and Policy
Volume 40, Part 3
As Guasch (2004) demonstrates, concession contract renegotiations are quite common across a range of industries. This study shows that 54.7 per cent of the transport projects analysed were renegotiated, and the average time to renegotiation after award was only 3.12 years. Further, 57 per cent of the renegotiations in the transport sector were initiated by operators, 27 per cent by governments, and only 16 per cent by a common agreement between both parties. Renegotiations are generally problematic because recontracting undermines both the credibility of the concession scheme and, more broadly, trust in the government, particularly with foreign investors. Also, renegotiations encourage concessionaires to bid based on in?ated tra?c forecasts. 2.3 Tra?c risk mitigation approaches A range of tra?c-risk mitigation approaches has been tested. Table 2 shows a taxonomy of tra?c risk-sharing mechanisms. These mechanisms can be classi?ed according to three criteria: (1) (2) (3) The variable used to trigger the measures to mitigate tra?c risk (trigger variable); The extent to which risk is shared; and The compensation mechanism adopted.
The trigger variable is the variable used as a reference point for initiating either the implementation of a guarantee, the modi?cation of contract Table 2 Classi?cation of Tra?c-Risk Mitigation Mechanisms
Trigger Variable Annual tra?c or revenues Risk Sharing Approach Compensation Subsidy/ payment Min Point Max Accumulative tra?c or revenues Min Point Max
Pro?ts/IRR Min Point Max
Approach 2: Minimum Income Guarantee Korea, Chile, Colombia Approach 1: Highway concessions in France and Spain Approach 3: (1) Severn Bridge (2) LPVR–RDM Chile (3) Portugal
Toll
Contract length
362
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
conditions, or a system to share gains and losses. The trigger variable can, for example, be: tra?c, revenues, pro?ts, or internal rate of return (IRR). The second criterion in the taxonomy is the degree to which the risk is shared. One possibility is to limit tra?c risk through the establishment of a ?xed target goal. A second possibility is reducing the downside by establishing a minimum target goal, so that the concessionaire is compensated only when the actual target variable falls below the minimum target goal. A third possibility is to share losses and gains through the establishment of maximum and minimum targets, outside of which, at either end, sharing mechanism is initiated. The third criterion shown in Table 2 for classifying tra?c risk-sharing mechanisms is the way in which compensation is established. Three di?erent ways are identi?ed in this classi?cation: a subsidy from the government, a change in the level of tolls, or a modi?cation of contract length. In practice, concession contracts have primarily occupied three regions of the taxonomy. The ?rst approach, which was developed in France ´ mez-Iban (Go ˜ ez and Meyer, 1993; Shugart, 1998) and recently adopted by Spain with some di?erences (Vassallo and Gallego, 2005), consist of guaranteeing the ‘economic balance’ of the concession, which is generally interpreted as the expected project IRR. Generally, the compensation measures to re-establish the economic balance of the contract are not pre-established, but rather negotiated when the IRR falls above or below the target levels. This compensation can include the variation of tolls, the change of the contract length, or the provision of public subsidies. The second approach, used in Chile and in some other developing countries such as Korea and Colombia (Irwin, 2003), guarantees either tra?c or revenues, and usually has both lower and upper bands to share tra?c risk between the concessionaire and the government. In the case that in one year the revenues fall below the bottom band, the government will have to pay the concessionaire the di?erence between the revenues guaranteed and the revenues collected. If the revenues fall above the upper band, the concessionaire has to share a percentage of the extra revenues collected with the government. The third approach, which has been adopted in several contracts recently, is to match the duration of the concession to the precise moment when the concession achieves a target variable, generally tra?c or revenues. This approach was ?rst applied in 1990 in the concession of the Second Severn Crossing in the United Kingdom. The length of the concession was pegged to a ?xed target of ‘Required Cumulative Real Revenue’ (Foice, 1998). Thus a ?gure was established in 1989 prices, which, once collected from toll income, would end the concession. Another similar experience is the Lusoponte concession in Portugal, which was 363
Journal of Transport Economics and Policy
Volume 40, Part 3
awarded at the end of the 1990s. The concession agreement was designed so that the concession would expire no later than March 2028 or once a total cumulative tra?c ?ow of 2,250 million vehicles had been reached (Lemos et al., 2004). A variant of this approach is the ‘Least Present Value of the Revenues’ mechanism (LPVR). LPVR was developed by Engel, Fischer and Galetovic (1997, 2001) in response to a proposal from an o?cial in the Chilean Ministry of Public Works (MOP). The concession is awarded to the bidder who o?ers the least present value of the accumulated revenues — discounted according to a discount rate pre-?xed in the contract — and the concession ends when that LPVR had been reached. Consequently if real tra?c is ultimately higher than expected, the concession will ?nish earlier, whereas if it is lower the concession will ?nish later. This mechanism has been implemented mainly in Chile. Based on the LPVR approach, there have been some proposals, inspired by this mechanism, from academics (Nombela and De Rus, 2003; Vassallo, 2004). The LPVR approach adds two interesting points to the Severn’s and Lusoponte’s experiences. First, the revenues are discounted according to a discount rate that should re?ect the weighted average cost of capital (WACC) of the project. Second, the LPVR requested by the bidders becomes the key economic variable in selecting the bidder that will win the tender. Beyond the risk mitigation e?ect, LPVR provides the government with a price to buy out the concession, a feature thought to discourage the concessionaire from renegotiation since the government can opt to buy out the concession at the established price.
3.0 The Chilean Experience
The case of highway concessions in Chile is interesting for several reasons. First, Chile has tendered many highway concessions in the last ?fteen years. Second, the government during this period designed and tested several di?erent mechanisms. Finally, an economic recession caused a slowdown in tra?c growth just a few years after the beginning of the operation of the concessions, which put to the test the performance of those mechanisms. 3.1 Highway concessions in Chile In the early 1990s Chile introduced private capital into the transport infrastructure sector by o?ering BOT concessions. The concessionaire was to recover its investment from tari?s paid by users, in some cases 364
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
supplemented with a subsidy from the government. In 1991, this process was reinforced with the approval of a new Public Works Concession Law, which was subsequently extended and improved in 1996. Table 3 shows the 26 highway concessions that have been successfully procured to date in Chile. Most of the interurban highway concessions were awarded between 1994 and 1998. Two features distinguish highway Table 3 Main Characteristics of the Highways Concessions granted in Chile by Year of Award
Investment million US$ 38 26 196 180 12 10 169 250 152 246 241 200 249 255 575 340 384 442 104 280 240 19 19 180 165 70 Total investment million US$ 38 222 371 Main economic tender variable Mixà Mixà Tari? Tari? Tari? Tari? Tari? Tari? Tari? Tari? Duration Upfront fee Upfront fee Upfront fee Upfront fee LPVR Upfront fee Upfront fee Subsidy Upfront fee Upfront fee Subsidy N/A N/A LPVR Upfront fee
Year of award 1992 1994 1995
Concession ´n Tu ´ nel del Melo Camino de la Madera ´n A. Norte Concepcio Santiago S. Antonio Nogales Puchancav?´ Acceso a AMB ´n Talca Chilla Los Vilos Santiago Santiago Los Andes La Serena Los Vilos ´ n Collipulli Chilla Temuco R?´ o Bueno R?´ o Bueno Pto. Montt Collipulli Temuco Santiago Talca Santiago Valpara?´ so Costanera Norte Norte–Sur Red Vial Litoral Central Vespucio Sur Vespucio Norte Talcahuano–Penco Variante de Melipilla Camino Internacional Ruta 60 Acceso Nororiente a Santiago El Salto–Kennedy
Highway Interurban tunnel Interurban Interurban Interurban Interurban Suburban Interurban Interurban Interurban Interurban Interurban Interurban Interurban Interurban Interurban Interurban Urban Urban Interurban Urban Urban Suburban Suburban Interurban Suburban Urban tunnel
1996
648
1997
945
1998 1999 2000 2001
915 384 546 558
2002 2003 2004
Ã
180 165 70
Mix: Several economic variables are employed. N/A: Information not available.
365
Journal of Transport Economics and Policy
Volume 40, Part 3
concessions in Chile from those in other countries. First, most of the highway projects are substantial upgrades to existing roads. The typical concession contract gives the concessionaire the responsibility for widening and improving the existing road into a high performance highway, and also for the maintenance and operation of this highway during the concession term. Second, most toll highways in Chile do not have free parallel roads competing with them, since Chile is roughly 5,000 kilometres long but only 100 kilometres wide, and has a low population density. Chile is using concessions to develop both interurban and urban roads. The majority of the interurban concessions, in terms of investment volumes and kilometres of road, have been the di?erent segments of the main north–south Pan American Highway, named Route 5. The total amount of investment in interurban highways totals more than US$ 3,400 million. The duration of both interurban and urban concessions was generally between twenty and thirty years. Beginning in 2000, the Chilean government began to grant six new urban highway concessions in the city of Santiago. The investment in those urban projects during the last ?ve years totalled more than US$ 1,500 million. Concession procurement in Chile takes place in a three-stage process. The ?rst stage is pre-quali?cation, to assure that bidders have both the knowledge and the resources to carry out the project. The second stage is technical evaluation of the proposals submitted by the pre-quali?ed bidders. The ?nal stage is selection based on economic variables established in the bidding terms, such as tari? level, subsidy (if any) required from the State, concession term, income guarantee requested by the State, revenue o?ered to the State for existing infrastructure, and LPVR (Ru?an, 1999). Starting in 1998, coincident with the beginning of the operation of most of the intercity concessions, the Chilean economy faltered. GDP per capita in Chile, which had grown during the last ten years at about 7.5 per cent per year on average, declined sharply. Table 4 shows the decline in tra?c on two of Chile’s most heavily travelled highway concessions during this economic downturn. When tra?c forecasts were carried out for most of Table 4 Three-year Chile’s GDP Growth Compared to Santiago–Talca and Santiago– Valparaiso Tra?c Growth
1989–1992 1992–1995 1995–1998 1998–2001 2001–2004 (%) (%) (%) (%) (%) GDP Growth in real terms AADT Growth Santiago–Talca AADT Growth Santiago–Valparaiso 7.94 7.46 7.20 7.75 7.87 4.54 5.72 6.48 7.64 2.34 ÿ0.84 ÿ3.24 3.56 2.76 5.74
366
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
the interurban highway concessions (between 1992 and 1996) planners did not anticipate that, after more than a decade with annual tra?c increments over 7 per cent, tra?c would decline. Consequently, concessions had to cope with lower tra?c levels than expected. Chile has implemented three tra?c-risk mitigation mechanisms: Minimum Income Guarantee (MIG), Least Present Value of the Revenues (LPVR), and Revenue Distribution Mechanisms (RDM). Although all three mechanisms mitigate tra?c risk, there are di?erences between them in terms of both the extent to which they have been applied and the objective the government had with their implementation. MIG was designed only as a way of mitigating tra?c risk so that lenders would perceive less risk and as a consequence the ?nancial cost of the project would become lower. The extent of MIG’s application to the Chilean situation has been widespread, since almost all the highway concession projects have incorporated this guarantee. LPVR was not only designed as a mechanism to mitigate tra?c risk, but also as a procurement mechanism detailing the terms of the regulation of the concession contracts. Unlike MIG, the application of LPVR in Chile to date has been limited; only two concessions have currently been awarded using this mechanism. RDM presents a way to mitigate tra?c risk ex-post by means of a modi?cation of contract terms after the 1998–2002 economic recession. At the end of 2004, six concessions had already implemented this mechanism, and many others were under renegotiation with the government. These mechanisms are not mutually exclusive. For example, the Santiago–Valparaiso project was procured under the LPVR mechanism, and it incorporated the MIG guarantee as well. In this case, the government considered that, despite LPVR, the implementation of MIG was necessary to reduce lenders’ perception of risk. Similarly, RDM is being implemented in projects that have already implemented MIG. 3.2 The minimum income guarantee The primary mitigation mechanism used in Chile was the ‘minimum income guarantee’ (MIG). The total guaranteed income in present value is the same for all the bidders, and it is equal, in present value, to 70 per cent of the investment cost plus the total maintenance and operation costs estimated by the MOP. The 70 per cent ?gure was chosen because, as Esty (2003) shows, on average the percentage of debt in a project ?nance structure is 70 per cent. The mechanism was designed to allow bidders enough ?exibility to establish the minimum band as long as the present value of the guaranteed revenues was no more than 70 per cent of the investment cost plus the total maintenance and operation costs. To that end, the MOP de?ned a range to set up the bands in the bidding terms. If the real revenues fall 367
Journal of Transport Economics and Policy
Volume 40, Part 3
below the lower band in any year, the MOP will have to compensate the concessionaire for the di?erence between the MIG Band revenues and the real revenues at the end of that year. If the concessionaire decided to take the MIG guarantee, it has the obligation to share part of the revenues obtained whenever real tra?c is higher than expected. Two di?erent methods have been implemented for triggering these revenue-sharing mechanisms. The ?rst method triggers revenue-sharing when the rate of return on investment is above 15 per cent in a given year. This rate of return is estimated by the MOP according to the actual revenues and the investment and operation costs estimated initially. The second method established a symmetrical band called a ‘mirror band’, in such a way that if real revenues surpass this band in any year, the concessionaire has to share 50 per cent of the di?erence between the real revenues and the mirror band revenues with the govern´ mez-Lobo and Hinojosa, 1999). ment (Go The MIG has some important advantages. First, the tra?c risk is shared, in that both unjusti?ed pro?ts and losses derived from uncertain tra?c are avoided. Second, the debt holders will feel more comfortable because of the guarantee of part of the revenues, and consequently the ?nancing cost of the project will be lower. This in turn will lower users’ tolls. In spite of the advantages, the MIG could be dangerous for the public sector if many concessions fail simultaneously, as might occur during a substantial economic downturn. In view of this concern, the MOP commissioned a study to address the ?scal implications derived from the implementation of the MIG in order to determine whether budgetary provisions were necessary to cover expected future payments. The study was based on a Monte Carlo simulation and concluded that the 70 per cent guarantee ´ mez-Lobo and level was su?ciently low to avoid this problem (Go Hinojosa, 1999). This study has recurred periodically, incorporating actual information about the concessions. The last version of this study concluded that the value of all the expected cash-?ows payable by the government in the form of minimum income guarantees for highway concessions, from 2004 on, totalled US$ 194 million at present value (Government of Chile, 2003). This amount represents only 3.84 per cent of the total investment involved in the highway concession programme. The performance of the highway concessions in Chile is consistent with the MOP study ?ndings. Almost all the road concessions procured since 1995 requested the MIG, and only two of them had performed under the bottom band at the end of 2003. The government’s payout for these contracts totalled US$ 5 million (US$ 4.8 million and US$ 0.2 million for the Santiago–Los Vilos and Nogales–Puchuncav?´ concessions, respectively). 368
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
Despite the tra?c decrease at the end of the 1990s, the total contribution from the government to the concessionaires due to MIG during the last ten years totalled less than 0.15 per cent of the toll road investment in Chile during the same period. 3.3 The LPVR mechanism The Chilean Public Works Concession Law de?ned the possibility of using the sum of total revenues to be obtained by the project as the main economic variable for tendering concessions. LPVR was primarily designed as a procurement mechanism, although one of the most important consequences of LPVR was the reduction of tra?c risk. Before describing the way the LPVR approach has been applied in Chile, it is important to show the relationship between the present value of the revenues and the ?nancial balance of a concession (equation (1)). Io ÿ S ¼ where: Io : S: a: n: pi : qi ð pi Þ: ci : ti : Initial investment Initial subsidy provided by the government Weighed average cost of capital (WACC) on the project Concession term Price for year i Actual tra?c in year i depending on pi Operation and maintenance costs in year i Taxes or concession fees in year i.
i¼n X ð pi Á qi ð pi Þ ÿ ci ÿ ti Þ ; ð1 þ aÞi i¼1
ð1Þ
Assuming that the public authority in charge of tendering the project does not provide any subsidy at the beginning, equation (1) can be restated as: Io þ
n i¼n X X ci þ ti ð pi Á qi ð pi ÞÞ ¼ : i ð1 þ aÞi i ¼ 1 ð1 þ aÞ i¼1
ð2Þ
The left-hand-side of the equation shows the discounted costs that the concessionaire has to bear in operating and constructing the concession, and the right-hand-side shows the revenues that the concessionaire will obtain along the contract duration. The right-hand-side of the equation re?ects actual LPVR during the life of the contract. At the beginning of the concession, the left-hand-side of the equation will be much higher than the right-hand-side. However, when the operating and maintenance costs are lower than the revenues, the right-hand-side of the equation will 369
Journal of Transport Economics and Policy
Volume 40, Part 3
Figure 1 Economic Balance and Variable Duration Models of a Concession under LPVR Mechanism
ECONOMIC BALANCE OF THE CONCESSION $
Construction Operation Discounted Revenues Discounted Costs
Io
Equilibrium VARIABLE DURATION OF THE CONTRACT DUE TO LPVR $
Construction Discounted Revenues Expected Traffic Low Traffic High Traffic
A
T1
T0
T2
Years
grow more quickly that the left-hand-side. This is the reason why the project needs to borrow money from lenders and promoters at the beginning. The point at which the two terms of the equation are equal means that the concession has covered all its costs — according to a cost of capital equal to a. Figure 1 illustrates graphically how this condition is reached. The auction mechanism based on the LPVR consists of granting the concession to the bidder that requires the lowest present value of the revenues to recover its costs. Supposing that the most competitive bidder — and therefore the winner — o?ered a net present value of the revenues equal to A, the concession will end when the real discounted ?ow of 370
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
revenues reaches A. This system substantially reduces the likelihood of renegotiations in the future. If tra?c is as expected, the duration of the concession will be T0 . However if the tra?c is ultimately lower or higher than expected, the concession duration will be either extended or reduced. In Chile, however, the Concessions Law establishes a maximum duration of 50 years so the concession term can not be extended more than this. This fact poses the problem that if, at the end of the maximum duration, the LPVR reached is lower than the LPVR requested, the concessionaire will not be able to recover its costs. An interesting feature of LPVR is that the discount rate can be either ?xed or variable. For example, in the Santiago–Valparaiso concession, the MOP gave the bidders the choice between a ?xed rate set as a riskfree rate of 6.5 per cent plus a risk premium of 4 per cent, totalling 10.5 per cent; or a variable rate established as the monthly average real riskfree rate of the ?nancial system plus a 4 per cent risk premium. Despite the high level of interest in the LPVR mechanism, it has been used only on a few occasions in Chile (see Table 5). The ?rst concession using LPVR, and also the most successful, was the Santiago–Valparaiso highway (Route 68). An analysis of the procurement process of this ´ mez-Lobo and Hinojosa (2000). Four consortia highway can be found in Go passed the technical evaluation stage. The competition for the award was ?erce, and the economic o?ers were within a very tight range. Table 5 Experiences in which the LPVR has been used as a Procurement Variable to Tender Highway Concession Projects in Chile
Year Investment1 PVR2 Maximum Number of US$ US$ term of tender million million (years) bidders3 1998 2000 340 384 381 – 25 30 4 0
Project Santiago–Valparaiso Costanera Norte
Situation Successfully awarded In operation Not awarded Tendered again a year later under other economic variable Awarded No bidder arrived to the LPVR step Successfully awarded In construction
Talcahuano–Penco
2001
19
–
31.5
2
Acceso Nororiente
1 2
2003
165
346
40
1
Investment predicted by the government. Present value of the revenues o?ered by the granted bidder. 3 Bidders in the last stage of the project.
371
Journal of Transport Economics and Policy
Volume 40, Part 3
The second attempt to tender a highway concession in Chile under the LPVR mechanism took place at the beginning of 1999. The highway selected was the Costanera-Norte, an urban expressway in Santiago, which joins the Northeast and the Northwest of the city passing under the downtown in a tunnel. Three factors contributed to a level of risk for this project that turned out to be much higher than previous interurban experiences. First, the construction was challenging because of both the need to construct a large tunnel below the Mapocho River so as to cross the city centre, and some unresolved environmental problems. Second, tra?c risk was substantial, owing to uncertainty about the competition from other roads or means of transport in the future in the urban area. And third, the project was planned to be the ?rst use in Chile of an electronic free-?ow toll collection system scheme. The substantial risks of the project encouraged MOP to tender this concession under the LPVR mechanism. The results of the tender, however, were disappointing. Only one consortium presented an o?er and it was ultimately disquali?ed because the guarantee bond o?ered was below the level established in the bidding documents. This experience proved that the LPVR was not a magic remedy to get very risky projects o? the ground without public support. In fact the project, which was re-tendered at the end of 1999 — this time not under the LPVR mechanism — was awarded without problem. This time the government committed itself to construct and ?nance part of the works of the project, reducing considerably the concessionaire’s investment. In addition, tra?c risk was reduced through establishing a minimum income guarantee band that was 25 per cent higher than usual. The Chilean government tried LPVR again in 2001 for the Talcahuano– Penco road (Ruta Interportuaria) near the city of Concepcion. This project was much smaller in terms of investment than the two previous ones. The procurement terms of this project provided bidders the option of competing in terms of the LPVR as long as no subsidy was required from the State. However, the three bidders requested a subsidy, therefore the concession was awarded ultimately with a subsidy and LPVR was not used. The last concession procured in Chile under LPVR was the ‘Acceso Nor Oriente a Santiago’ in 2003. This project is a suburban highway that provides easier access from the highways coming from the north of the country to the wealthy neighbourhoods in East Santiago. Although the concession was awarded, only one bidder responded. Like Costanera Norte, this project was extremely risky. The uncertainty was re?ected in a high (more than twice the investment of the project) LPVR requested. 372
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
3.4 The RDM mechanism The problems of many concessionaires as a consequence of the economic recession in Chile combined with the need for additional investment in many of the highway projects moved the government to introduce a new mechanism in order to modify the original contracts. This mechanism was called ‘Revenue Distribution Mechanism’ (RDM). RDM guarantees that a pre-?xed amount of revenues — in present value — will be received by the concessionaire. Thus the duration of the renegotiated concession turns from being ?xed to being variable. The government expected that the average annual tra?c growth would remain at around 3.5 per cent for most of the concession contracts at the time the RDM was considered. The government gave three alternatives to the concessionaires in terms of the present value of the revenues to be guaranteed. Those three alternatives were established in terms of an average annual tra?c growth of 4, 4.5, and 5 per cent respectively. Three reasons explain why the government adopted those values. First, the government decided to guarantee an amount of revenues in present value higher than the revenues forecast so as to take advantage of RDM to require additional investments upfront from the concessionaires. Second, the government decided not to guarantee an average annual tra?c growth much higher than the one predicted so as to avoid a long extension of the length of the contract. Third, the government wanted to give ?exibility to the concessionaires by o?ering them several di?erent alternatives. In exchange for the guarantee of extra revenues, the government required the concessionaire to carry out initial investments, which were calculated by the government as the di?erence between the present value of the revenues guaranteed and the present value of the revenues expected. Thus, the higher the tra?c growth guaranteed, the higher will be the amount of upfront investment to be implemented by the concessionaire. The main innovation of this mechanism is that the term of the concession contract changed from ?xed to variable and is therefore similar to LPVR, since the contract will expire as soon as the concessionaire has collected the revenues guaranteed in present value. RDM causes the concession to ?nish earlier if average tra?c growth is ultimately higher than the guaranteed level. If that happens, the concessionaire will ultimately fare worse than it would have done if it had not obtained the RDM guarantee. If the average tra?c growth is ultimately the same as the guaranteed level, the concession will end in the year initially ?xed. If the average tra?c growth turns out ultimately to be as forecast, the concession contract will expire some years later than the term originally established in the contract. 373
Journal of Transport Economics and Policy
Volume 40, Part 3
Figure 2 RDM Mechanism
Revenues Revenues guaranteed Real revenues
C F E
G
A Revenues expected in year k Additional Investment k
Operation costs H n I m
Year
Figure 2 sketches how RDM works. Year k is the year in which the rede?nition of the contract terms takes place. Tra?c forecasts from this year on are represented by the long-stroke broken line AE. The shortstroke broken line AC represents the average tra?c guaranteed — 4, 4.5, or 5 per cent according to the selection of the concessionaire. In compensation for the extra-revenues guaranteed, the concessionaire has to carry out additional investments in year k. These investments will be equal to the area ACE (dotted area) in present value. If, for instance, the real tra?c level follows the way of the AF line, the duration of the concession contract will be extended until year m, in which the present value of the revenues agreed would be reached. According to the characteristics of the RDM the area FGHI in present value should be equal to the area ECF in present value. There are three main di?erences between LPVR and RDM. First, under LPVR there is a competition to o?er the lowest LPVR, whereas under RDM there is no competition. In this case the government o?ers several alternatives that can be taken or rejected by the concessionaire. Second, unlike LPVR, under RDM the public sector has a choice of obligation: whether to extend the concession contract until the present value is reached or to compensate the concessionaire for the remaining LPVR. In other words, the concessionaire will be certain to receive the whole revenues guaranteed in present value. Third, under RDM once the initial duration of the contract has been reached, the revenues calculated in order to determine the end of the concession are net revenues (revenues minus costs). The last two di?erences between LPVR and RDM imply that, unlike LPVR, RDM does not transfer any downside risk to the concessionaire. 374
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
The additional investments required by RDM have to be approved by the MOP. The amount of these investments in the year envisaged for their realisation — year k — is ?xed by the MOP according to equation (3): AIk ¼ where: AIk : y: REi : RGi : Additional investment required in year k Discount rate equal to the cost of capital (9.5 per cent annually) Revenues expected in year i calculated by the government Revenues guaranteed in year i.
i¼n X RGi ÿ REi i¼k
ð1 þ yÞi ÿ k
;
ð3Þ
The discount rate adopted was ?xed by the MOP as 9.5 per cent annually, according to the MOP’s estimations of the weighted average cost of capital based on previous tenders. In principle, if the guarantee was correctly priced, concessionaires should be indi?erent as to whether to buy this guarantee, or to choose any of the three guarantee alternatives o?ered. However, almost all the concessionaires so far have shown an interest in purchasing the guarantee, and all of them have chosen the highest one (average annual growth equal to 5 per cent). That issue seems to show that actually the discount rate adopted by the government to calculate the present value of the revenues (equal to 9.5 per cent) was higher than the real discount rate as perceived by the companies. An explanation for that di?erence could be that the Chilean Government did not take into account the fact that the discount rate (weighted average cost of capital) to be applied to the RDM should be lower than the discount rate before the project has been constructed. Two reasons justify that fact. First, as the project has already been constructed, the uncertainty regarding construction costs and tra?c generation is much reduced. Second, the modi?cation of the contract from ?xed to variable length substantially reduces the risk of the project and consequently the discount rate to be adopted. The latter reason can be regarded in another way. The RDM guarantees a certain amount of revenue, which includes both the expected revenues and some additional revenues. The additional investment in compensation for the additional revenues is estimated as the present value of the additional revenues discounted at a rate of 9.5 per cent. However, that does not take into account that the risk of the whole project has substantially diminished since the expected revenues are in fact guaranteed. This approach is undoubtedly one of the main reasons why the RDM mechanism became so popular among concessionaires and why all of them chose the highest guarantee. 375
Journal of Transport Economics and Policy
Volume 40, Part 3
RDM is not a substitute for the ‘Minimum Income Guarantee’ (MIG). MIG has the role of providing security to lenders and reducing the ?nancial cost of the project, while RDM has the role of mitigating shareholders’ tra?c risk. The revenues adopted by RDM in order to calculate the present value that will eventually determine the end of the concession contract are the revenues that the concessionaire actually collects no matter where they come from. For instance, it could happen that, after the RDM has been implemented, the tra?c growth greatly increased and, according to the MIG approach, the concessionaire should share with the government part of those extra-revenues collected above the upper band. In that case, the concessionaire should pay to the government the amount of money that the MIG approach establishes for the corresponding year. However, the revenues paid to the government are not taken into account in estimating the present value of the revenues that will eventually determine the end of the concession contract. The revenues that are to be calculated for that purpose are the revenues actually received by the company. Table 6 shows the highway concessions where the concessionaire has already purchased the RDM. Between 1994 and 1998 fourteen highway concessions were awarded in Chile. By the beginning of 2005 six highway concessionaires had purchased the RDM mechanism and many other concessionaires were negotiating with the MOP in order to acquire the guarantee. The RDM guarantee has been used by some of the concessionaires as collateral for new bonds to ?nance the additional highways investment required by the RDM. For instance, the Santiago–Talca highway — the largest project using the RDM guarantee — issued a new bond in 2004 to ?nance both a second phase of the project and the additional works required by the RDM. Table 6 Characteristics of the Highway Concessions which have already purchased the RDM in Chile
Year concession awarded 1994 1995 1995 1997 1997 1998 First year of operation 1996 1997 1998 2001 2002 2002 Initial duration (years) 25 22 10 22 25 25 Premium (US$ million) 3.06 1.47 29.00 19.89 25.23 73.73 Initial investment (US$ million) 26 12 169 241 255 575
Project Camino de la Madera Camino Nogales Puchuncav?´ ´n Ruta 5 Tramo Talca–Chilla ´ n–Collipulli Ruta 5 Tramo Chilla Ruta 5 Tramo Collipulli–Temuco Ruta 5 Tramo Santiago–Talca
376
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
3.5 The consequences of the 1998–2002 recession Chile’s economic recession from 1998 to 2002 provides an opportunity to assess the performance of tra?c and income guarantees under lowerthan-projected tra?c volumes. Although the recession caused economic growth to be lower than predicted, Chile’s recession is not comparable to other huge recessions experienced in Latin America, which reduced GDP by more than 10 per cent in a year (that is, Argentina 2002, Venezuela 2003, Mexico 1995). The recession a?ected the shareholders more than the lenders, since the payback of the loans was largely covered by MIG. The concession shareholders attempted to renegotiate the contract terms based on an article of the Chilean Concession Law that permits the economic terms of the concession to be modi?ed in case of unpredictable circumstances. The Ministry of Public Works rejected this approach on the grounds that the concessionaires’ tra?c risk was mitigated through the MIG, and because the Chilean concession model clearly de?ned tra?c risk as a risk to be held by the concessionaire, regardless of any guarantees that could be applicable according to the law. At the same time that the concessionaires were arguing for renegotiation, the MOP realised that additional investments would be necessary in most of the highway concession projects. Some designs of earlier concessions were now thought to be inadequate (poor connections to the local network, safety problems, and so on). A common problem was that additional pedestrian and road bridges were needed to provide access all along the highway. This is the reason why the MOP decided to apply the RDM guarantee. The MOP permitted the possibility of obtaining this guarantee to all the highway concessions that were awarded before 2000. The sole exception was the Santiago–Valparaiso concession, which was the only highway project then in operation that had been tendered under the LPVR mechanism. This is an interesting demonstration that LPVR performed very well in terms of tra?c risk mitigation, since it is the only concession that neither the government nor the concessionaire tried to renegotiate.
4.0 Lessons from the Chilean Experience
LPVR is conceptually very attractive for several reasons. First, a variable term is a highly e?ective compensation method that neither commits public resources nor entails tari? increases. Second, LPVR sets up a clear buy-out price. And third, LPVR reduces renegotiation expectations so 377
Journal of Transport Economics and Policy
Volume 40, Part 3
the bidders have less incentive to in?ate their o?ers. However, among the 26 road projects that were granted in Chile during the last decade, only four were tendered, and only two were successfully awarded under this approach. In a set of interviews conducted by the author in Chile, all the stakeholders — the MOP, the concessionaires, university professors, and so on — point to the strong opposition from concessionaires as the main reason limiting the application of LPVR. Concessionaires’ concerns about LPVR are threefold. First, LPVR does not improve the capacity of the project to ful?ll its commitments to the lenders every year. Second, the variable length of the contract makes the concession operation di?cult to organise since resource planning cannot be programmed in advance. Finally, LPVR always limits the upside pro?tability of the concessionaire but not always the downside, so the mechanism is not symmetric. Chile’s experience suggests that the ?rst two criticisms are not a major concern. Regarding the ?rst criticism, the Santiago–Valparaiso highway bond issued in 2002 was the largest and least expensive infrastructure bond issued in Chile until then. Due to the variability of the concession term, the bond was structured with a mechanism of mandatory prepayment. Regarding the operational organisation and resource planning, the Santiago–Valparaiso experience demonstrates that information about actual tra?c during the concession facilitates the prediction of when the contract will end. Thus there is ample time for resource planning. Moreover, the organisation of the concession operation under a variable term approach was ultimately not as complicated as expected. The third criticism — an asymmetrical risk pro?le — is of concern. As Brealey, Cooper, and Habib (1996) assert, private shareholders are interested in the upside of the project. This is easy to understand since, as concession projects are usually highly leveraged, the sponsors expect a large upside that compensates for the possibility of losing all their capital. Concessionaires are also interested in avoiding the downside because this way the lenders perceive lower risk, permitting the concessionaires to enjoy the bene?ts stemming from higher leverage and lower interest rates. The risk pro?le drawn by LPVR is the opposite of the concessionaire desires described above. On the one hand, the upside is almost non-existent. On the other hand, as a maximum duration is established in the concession contract, the concessionaire bears the risk that the project will not reach the LPVR requested at the maximum term. This is the reason why promoters regard this mechanism as asymmetrical. This asymmetry was already ´ mez-Iban pointed out by Go ˜ ez (2003) who mentions that whereas the government has a call option on the project for the remaining present 378
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
value of the contract, the concessionaire does not have a put option whereby it can sell the project to the government at the end of the contract in exchange for the present value remaining. The duration constraint problem is di?cult to solve with an extension of the concession contract duration. First, the maximum duration is limited by the Concession Law so it is not legally possible to establish concession durations longer than 50 years in Chile. Second, even if there was not a duration limit, it would not seem appropriate to establish very long durations since the longer the duration the more di?cult it is to specify complete contracts. An interesting approach to solving this problem is the one developed by RDM. This mechanism forces the government to pay the LPVR left at the end or extend the concession until the LPVR requested is ultimately reached. In this extension, the government can modify the contract with the concessionaire to adapt the new environmental and technological requirements. Three lessons can be highlighted from Chile’s experience. The ?rst lesson is that establishing a mechanism, such as LPVR, to mitigate tra?c risk based on extending or reducing the concession length has been demonstrated to work quite well with the caveat that the risk pro?le drawn by it was not well-regarded by private promoters. Moreover, LPVR has proved e?ective in reducing renegotiation expectations. In fact, as a result of the 1998–2002 economic recession, the LPVR-based Santiago–Valparaiso concession was the only one where the government did not allow the possibility of reopening renegotiation. The second lesson is that, unlike LPVR, the establishment of a mechanism like MIG does not reduce renegotiation pressures from promoters when the tra?c is lower than expected. Moreover, although this mechanism behaved reasonably well in Chile during the 1998–2002 economic crisis, such a crisis was very small compared to the economic crises su?ered by other developing countries in the last few decades. This fact suggests that, in spite of the good performance of MIG in Chile, it still transfers an important risk because of the important correlation between tra?c and economic growth. This point can be a serious problem for the implementation of this mechanism in unstable economies. The third lesson is that, although the risk pro?le of this mechanism was not initially attractive for private investors, the mechanism seems to become more attractive when the downside risk is limited. This lesson, however, has to be approached carefully. Chilean concessionaires assert that they positively respond to a mechanism, such as RDM, in which the downside risk is eliminated. Yet the success of RDM among concessionaires seems to be motivated not only by this factor, but also by an overestimation of the discount rate. 379
Journal of Transport Economics and Policy
Volume 40, Part 3
These lessons suggest that LPVR is a very attractive mechanism for procuring highway concessions and limiting tra?c risk. However, from experience so far, two measures can be suggested that can help to improve concessionaires’ perception on LPVR. First, there is a need to implement a limit on the downside risk. Second, it is necessary to establish a minimum concession duration in such a way that the concessionaire will enjoy an upside if the tra?c is higher than expected. Since to date LPVR has been implemented only in highway concessions, it would be useful to devote future research to studying the applicability of LPVR to demand-risk mitigation in other infrastructure facilities — such as railroads, ports, airports, water and sewage facilities, and so on — and even to long-term service contracts.
References
Bain, R. and M. Wilkins (2002): ‘Credit Implications of Tra?c Risk in Start-Up Toll Facilities’, Standard & Poor’s Project and Infrastructure Finance, Standard & Poor, London. Bain, R. and M. Plantagie (2004): Tra?c Forecasting Risk: Study Update 2004, Standard & Poor, London. Brealey, R. A., I. A. Cooper, and M. A. Habib (1996): ‘Using Project Finance to Fund Infrastructure Investment’, Journal of Applied Corporate Finance, 9, 25–38. Engel, E., R. Fischer, and A. Galetovic (1997): ‘Highway Franchising: Pitfalls and Opportunities’, American Economic Review, 87, 68–72. Engel, E., R. Fischer, and A. Galetovic (2001): ‘Least Present Value of Revenue Auctions and Highway Franchising’, Journal of Political Economy, 109, 993–1020. Esty, B. C. (2003): Modern Project Finance, a Casebook, John Wiley & Sons. Flyvbjerg, B., M. K. Skamris Holm, and S. L. Buhl (2005): ‘How (In)accurate Are Demand Forecasts in Public Works Projects? The Case of Transportation’, Journal of the American Planning Association, 71, 131–46. Foice, D. (1998): ‘Second Severn Crossing’, Proceedings of the Seminar PPP Risk Management for Big Transport Projects, Ministerio de Fomento, Spain. Government of Chile (2003): Report on Public Finances. Government Budget Bill for 2004, Ministry of Finance, Santiago. ´ mez-Iban Go ˜ ez, J. A. (2003): Regulation of Private Infrastructure: Monopoly, Contracts, and Discretion, Harvard University Press. ´ mez-Iban Go ˜ ez, J. A. and J. R. Meyer (1993): Going Private: the International Experience with Transport Privatization, The Brookings Institution, Washington, DC. ´ mez-Lobo, A. and S. Hinojosa (2000): Broad Roads in a Thin Country: Infrastructure Go Concessions in Chile, Research Paper 2279, The World Bank, Washington DC. Guasch, J. L. (2004): Granting and Renegotiating Infrastructure Concessions Doing it Right, WBI Development Studies, The World Bank, Washington DC. Irwin, T. (2003): Public Money for Private Infrastructure: Deciding When to O?er Guarantees, Output Based Subsidies and other Fiscal Support, World Bank Working Paper 10, Washington DC. Izquierdo, J. M. and J. M. Vassallo (2004): Nuevos sistemas de gestio´n y ?nanciacio´n de ´ n SEINOR no 35, Colegio de Ingenieros de infraestructuras de transporte, Coleccio Caminos, Canales y Puertos, Madrid.
380
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
Lemos, T., D. Eaton, M. Betts, and L. Tadeu de Almeida (2004): ‘Risk Management in the Lusoponte Concession — a Case Study of the Two Bridges in Lisbon, Portugal’, International Journal of Project Management, 22, 63–73. Nombela, G. and G. de Rus (2003): ‘Flexible-term Contracts for Road Franchising’, Transportation Research Part A, 38, 163–79. Ru?an, D. M. (1999): Manual de Concesiones de Obras Pu´blicas, Fondo de Cultura ´ mica, Santiago, Chile. Econo Shugart, C. T. (1998): Regulation-by-Contract and Municipal Services: The Problem of Contractual Incompleteness, PhD Dissertation, Harvard University. Vassallo, J. M. (2004): ‘Short Term Infrastructure Concessions: Conceptual Approach and Recent Applications in Spain’, Public Works Management and Policy, 8, 261–70. Vassallo, J. M. and J. Gallego (2005): ‘Risk-sharing in the New Public Works Concession Law in Spain’, Transportation Research Record 1932, 1–8. Zhao, Y. and K. M. Kockelman (2002): ‘The Propagation of Uncertainty through Travel Demand Models’, The Annals of Regional Science, 36, 145–63.
381
doc_583185278.pdf
A traffic collision, also known as a traffic accident, motor vehicle collision, motor vehicle accident, car accident, automobile accident, road traffic collision, wreck (USA), car crash, or car smash (Australian) occurs when a vehicle collides with another vehicle, pedestrian, animal, road debris, or other stationary obstruction, such as a tree or utility pole. Traffic collisions may result in injury, death, vehicle damage, and property damage.
Journal of Transport Economics and Policy, Volume 40, Part 3, September 2006, pp. 359–381
Tra?c Risk Mitigation in Highway Concession Projects
The Experience of Chile
´ M. Vassallo Jose
´ M. Vassallo, Associate Professor, Departamento de Address for correspondence: Jose ´ cnica de Madrid, Transportes, ETSI de Caminos, Canales y Puertos, Universidad Polite 28040 Madrid, Spain ( [email protected]). The author would like to thank ´ Antonio Go ´ mez-Iban Professor Jose ˜ ez of the Kennedy School of Government (Harvard University) for proposing the idea for this paper and for reading and correcting with extreme patience all the drafts he gave him. A special debt of gratitude is owed to Mark Fagan, my colleague as a visiting fellow at Harvard, for the countless hours spent helping with the structure of the paper, as well as for the thankless task of ´ lez, former director of concession proediting its ?nal version. Finally, Alvaro Gonza jects at the MOP and the ?rst person to approach the LPVR mechanism, helped obtain the information necessary to write the paper and also made useful comments about highway concessions in Chile.
Abstract Tra?c risk mitigation remains a challenging aspect of highway concessions. This paper evaluates three mechanisms applied in Chile to mitigate tra?c risk: the ‘Minimum Income Guarantee’ (MIG); the ‘Least Present Value of the Revenues’ (LPVR); and the ‘Revenue Distribution Mechanism’ (RDM). Speci?cally, the paper focuses on the performance of LPVR and MIG during the economic recession that took place between 1998 and 2002. In the context of this recession, the paper explains the reasons that led the government to implement the RDM mechanism. The paper gives some guidelines about the applicability of these mechanisms in other countries, highlights the bene?cial features of LPVR in reducing tra?c risk and avoiding concession contract renegotiations, and ?nally provides some recommendations as to how to make LPVR more attractive to private promoters.
Date of receipt of ?nal manuscript: October 2005 359
Journal of Transport Economics and Policy
Volume 40, Part 3
1.0 Introduction
Highway privatisations around the world have been completed using concession contracts, often in the form of Build, Operate, Transfer ´ mez-Iban (BOT) projects (Go ˜ ez, 2003). Successful BOT projects require the proper allocation of risks among the project’s stakeholders. Whereas some risks — such as construction, operational or legal risk — are clearly controllable by some stakeholders, tra?c risk can not be controlled by any of them. As a result, allocating tra?c risk is one of the greatest challenges in designing highway concession contracts. Chile has been a pioneer in the implementation of tra?c risk mitigation mechanisms in highway concessions. The objective of this paper is to study the e?ectiveness of those mechanisms, particularly the ‘Least Present Value of the Revenues’, and to provide recommendations for its implementation in the future. The paper is organised as follows. The ?rst section highlights tra?c forecast challenges and provides a taxonomy of the di?erent mechanisms to mitigate tra?c risk as the context for examining the Chilean experience. The second section presents a case study of the highway concessions in Chile, focusing on three mechanisms: the ‘Minimum Income Guarantee’ (MIG), the ‘Least Present Value of the Revenues’ (LPVR), and the ‘Revenue Distribution Mechanism’ (RDM). The third section provides lessons based on the Chilean experience.
2.0 The Extent of Tra?c Risk and Mitigation Strategies
Tra?c risk depends primarily on the performance of the economy, the reaction of users, and the competition with other means of transport. Thus, none of the concession stakeholders can directly mitigate the risk. Moreover, unlike other BOT projects — such as mining and oil production — in which demand risk is usually hedged through long-term purchase contracts, highway BOT tra?c risk is di?cult to hedge because of the large number of customers and lack of an insurance market. 2.1 Tra?c risk inaccuracy Tra?c demand is very di?cult to predict. Table 1 shows the results from several studies that measure the inaccuracy of tra?c demand estimation. Standard and Poor’s (S&P) compared ?rst-year forecast tra?c levels with actual tra?c volumes for 82 projects. The results of the study showed that, during the ?rst year, tra?c volumes averaged about 76 per cent of 360
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
Table 1 Tra?c Forecast Deviation Comparison in Several Studies
Main geographical areas studied North America, North Europe, Asia, South Europe, Latin America Denmark, European Union South Europe, Latin America Mean Sample (real/ size forecast) 87 0.76 Standard deviation 0.26
Study
Projects
Standard & Poor’s Toll roads (2004) Flyvbjerg et al. (2004) Vassallo (2002) Free roads Toll roads Toll roads Shadow toll roads
183 18
1.09 1.03
0.44 0.24
their predicted values, and the error — measured through the standard deviation — was 0.26 (Bain and Wilkins, 2002; Bain and Plantagie, 2004). Some data provided by Spanish concessionaires to the author report similar results in term of standard deviation although, unlike S&P data, they do not prove any bias towards underestimation. A large and recent study by Flyvbjerg et al. (2005) — focused primarily on free roads in Northern Europe — also reports substantial forecasting errors; the standard deviation estimated is 0.44. Unlike the S&P study, this study shows a slight bias towards underestimation. According to these studies, tra?c forecast errors for toll highways seems to be consistently in a range between 0.20 and 0.30 in terms of standard deviation. Assuming that tra?c deviations are distributed according to a normal function with mean equal to 1 and standard deviation equal to 0.25, which seems reasonable according to the empirical results, 42 per cent of the toll highway projects will experience ?rst-year tra?c misestimation of more than 20 per cent. This situation is accentuated by the fact that, as mentioned by Zhao and Kockelman (2002), the further in the future the tra?c forecast, the greater the inaccuracy. 2.2 Tra?c risk as a cause of renegotiation Tra?c risk in highway concessions has generally been assigned to concessionaires. That fact has often prompted asymmetrical behaviour. If ultimately the tra?c is higher than expected, the concessionaire will reap excess pro?ts, whereas if the tra?c is lower, the concessionaire will incur losses, and may attempt to force a renegotiation with the government. This asymmetric behaviour has occurred for instance in Spain (Izquierdo and Vassallo, 2004). 361
Journal of Transport Economics and Policy
Volume 40, Part 3
As Guasch (2004) demonstrates, concession contract renegotiations are quite common across a range of industries. This study shows that 54.7 per cent of the transport projects analysed were renegotiated, and the average time to renegotiation after award was only 3.12 years. Further, 57 per cent of the renegotiations in the transport sector were initiated by operators, 27 per cent by governments, and only 16 per cent by a common agreement between both parties. Renegotiations are generally problematic because recontracting undermines both the credibility of the concession scheme and, more broadly, trust in the government, particularly with foreign investors. Also, renegotiations encourage concessionaires to bid based on in?ated tra?c forecasts. 2.3 Tra?c risk mitigation approaches A range of tra?c-risk mitigation approaches has been tested. Table 2 shows a taxonomy of tra?c risk-sharing mechanisms. These mechanisms can be classi?ed according to three criteria: (1) (2) (3) The variable used to trigger the measures to mitigate tra?c risk (trigger variable); The extent to which risk is shared; and The compensation mechanism adopted.
The trigger variable is the variable used as a reference point for initiating either the implementation of a guarantee, the modi?cation of contract Table 2 Classi?cation of Tra?c-Risk Mitigation Mechanisms
Trigger Variable Annual tra?c or revenues Risk Sharing Approach Compensation Subsidy/ payment Min Point Max Accumulative tra?c or revenues Min Point Max
Pro?ts/IRR Min Point Max
Approach 2: Minimum Income Guarantee Korea, Chile, Colombia Approach 1: Highway concessions in France and Spain Approach 3: (1) Severn Bridge (2) LPVR–RDM Chile (3) Portugal
Toll
Contract length
362
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
conditions, or a system to share gains and losses. The trigger variable can, for example, be: tra?c, revenues, pro?ts, or internal rate of return (IRR). The second criterion in the taxonomy is the degree to which the risk is shared. One possibility is to limit tra?c risk through the establishment of a ?xed target goal. A second possibility is reducing the downside by establishing a minimum target goal, so that the concessionaire is compensated only when the actual target variable falls below the minimum target goal. A third possibility is to share losses and gains through the establishment of maximum and minimum targets, outside of which, at either end, sharing mechanism is initiated. The third criterion shown in Table 2 for classifying tra?c risk-sharing mechanisms is the way in which compensation is established. Three di?erent ways are identi?ed in this classi?cation: a subsidy from the government, a change in the level of tolls, or a modi?cation of contract length. In practice, concession contracts have primarily occupied three regions of the taxonomy. The ?rst approach, which was developed in France ´ mez-Iban (Go ˜ ez and Meyer, 1993; Shugart, 1998) and recently adopted by Spain with some di?erences (Vassallo and Gallego, 2005), consist of guaranteeing the ‘economic balance’ of the concession, which is generally interpreted as the expected project IRR. Generally, the compensation measures to re-establish the economic balance of the contract are not pre-established, but rather negotiated when the IRR falls above or below the target levels. This compensation can include the variation of tolls, the change of the contract length, or the provision of public subsidies. The second approach, used in Chile and in some other developing countries such as Korea and Colombia (Irwin, 2003), guarantees either tra?c or revenues, and usually has both lower and upper bands to share tra?c risk between the concessionaire and the government. In the case that in one year the revenues fall below the bottom band, the government will have to pay the concessionaire the di?erence between the revenues guaranteed and the revenues collected. If the revenues fall above the upper band, the concessionaire has to share a percentage of the extra revenues collected with the government. The third approach, which has been adopted in several contracts recently, is to match the duration of the concession to the precise moment when the concession achieves a target variable, generally tra?c or revenues. This approach was ?rst applied in 1990 in the concession of the Second Severn Crossing in the United Kingdom. The length of the concession was pegged to a ?xed target of ‘Required Cumulative Real Revenue’ (Foice, 1998). Thus a ?gure was established in 1989 prices, which, once collected from toll income, would end the concession. Another similar experience is the Lusoponte concession in Portugal, which was 363
Journal of Transport Economics and Policy
Volume 40, Part 3
awarded at the end of the 1990s. The concession agreement was designed so that the concession would expire no later than March 2028 or once a total cumulative tra?c ?ow of 2,250 million vehicles had been reached (Lemos et al., 2004). A variant of this approach is the ‘Least Present Value of the Revenues’ mechanism (LPVR). LPVR was developed by Engel, Fischer and Galetovic (1997, 2001) in response to a proposal from an o?cial in the Chilean Ministry of Public Works (MOP). The concession is awarded to the bidder who o?ers the least present value of the accumulated revenues — discounted according to a discount rate pre-?xed in the contract — and the concession ends when that LPVR had been reached. Consequently if real tra?c is ultimately higher than expected, the concession will ?nish earlier, whereas if it is lower the concession will ?nish later. This mechanism has been implemented mainly in Chile. Based on the LPVR approach, there have been some proposals, inspired by this mechanism, from academics (Nombela and De Rus, 2003; Vassallo, 2004). The LPVR approach adds two interesting points to the Severn’s and Lusoponte’s experiences. First, the revenues are discounted according to a discount rate that should re?ect the weighted average cost of capital (WACC) of the project. Second, the LPVR requested by the bidders becomes the key economic variable in selecting the bidder that will win the tender. Beyond the risk mitigation e?ect, LPVR provides the government with a price to buy out the concession, a feature thought to discourage the concessionaire from renegotiation since the government can opt to buy out the concession at the established price.
3.0 The Chilean Experience
The case of highway concessions in Chile is interesting for several reasons. First, Chile has tendered many highway concessions in the last ?fteen years. Second, the government during this period designed and tested several di?erent mechanisms. Finally, an economic recession caused a slowdown in tra?c growth just a few years after the beginning of the operation of the concessions, which put to the test the performance of those mechanisms. 3.1 Highway concessions in Chile In the early 1990s Chile introduced private capital into the transport infrastructure sector by o?ering BOT concessions. The concessionaire was to recover its investment from tari?s paid by users, in some cases 364
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
supplemented with a subsidy from the government. In 1991, this process was reinforced with the approval of a new Public Works Concession Law, which was subsequently extended and improved in 1996. Table 3 shows the 26 highway concessions that have been successfully procured to date in Chile. Most of the interurban highway concessions were awarded between 1994 and 1998. Two features distinguish highway Table 3 Main Characteristics of the Highways Concessions granted in Chile by Year of Award
Investment million US$ 38 26 196 180 12 10 169 250 152 246 241 200 249 255 575 340 384 442 104 280 240 19 19 180 165 70 Total investment million US$ 38 222 371 Main economic tender variable Mixà Mixà Tari? Tari? Tari? Tari? Tari? Tari? Tari? Tari? Duration Upfront fee Upfront fee Upfront fee Upfront fee LPVR Upfront fee Upfront fee Subsidy Upfront fee Upfront fee Subsidy N/A N/A LPVR Upfront fee
Year of award 1992 1994 1995
Concession ´n Tu ´ nel del Melo Camino de la Madera ´n A. Norte Concepcio Santiago S. Antonio Nogales Puchancav?´ Acceso a AMB ´n Talca Chilla Los Vilos Santiago Santiago Los Andes La Serena Los Vilos ´ n Collipulli Chilla Temuco R?´ o Bueno R?´ o Bueno Pto. Montt Collipulli Temuco Santiago Talca Santiago Valpara?´ so Costanera Norte Norte–Sur Red Vial Litoral Central Vespucio Sur Vespucio Norte Talcahuano–Penco Variante de Melipilla Camino Internacional Ruta 60 Acceso Nororiente a Santiago El Salto–Kennedy
Highway Interurban tunnel Interurban Interurban Interurban Interurban Suburban Interurban Interurban Interurban Interurban Interurban Interurban Interurban Interurban Interurban Interurban Urban Urban Interurban Urban Urban Suburban Suburban Interurban Suburban Urban tunnel
1996
648
1997
945
1998 1999 2000 2001
915 384 546 558
2002 2003 2004
Ã
180 165 70
Mix: Several economic variables are employed. N/A: Information not available.
365
Journal of Transport Economics and Policy
Volume 40, Part 3
concessions in Chile from those in other countries. First, most of the highway projects are substantial upgrades to existing roads. The typical concession contract gives the concessionaire the responsibility for widening and improving the existing road into a high performance highway, and also for the maintenance and operation of this highway during the concession term. Second, most toll highways in Chile do not have free parallel roads competing with them, since Chile is roughly 5,000 kilometres long but only 100 kilometres wide, and has a low population density. Chile is using concessions to develop both interurban and urban roads. The majority of the interurban concessions, in terms of investment volumes and kilometres of road, have been the di?erent segments of the main north–south Pan American Highway, named Route 5. The total amount of investment in interurban highways totals more than US$ 3,400 million. The duration of both interurban and urban concessions was generally between twenty and thirty years. Beginning in 2000, the Chilean government began to grant six new urban highway concessions in the city of Santiago. The investment in those urban projects during the last ?ve years totalled more than US$ 1,500 million. Concession procurement in Chile takes place in a three-stage process. The ?rst stage is pre-quali?cation, to assure that bidders have both the knowledge and the resources to carry out the project. The second stage is technical evaluation of the proposals submitted by the pre-quali?ed bidders. The ?nal stage is selection based on economic variables established in the bidding terms, such as tari? level, subsidy (if any) required from the State, concession term, income guarantee requested by the State, revenue o?ered to the State for existing infrastructure, and LPVR (Ru?an, 1999). Starting in 1998, coincident with the beginning of the operation of most of the intercity concessions, the Chilean economy faltered. GDP per capita in Chile, which had grown during the last ten years at about 7.5 per cent per year on average, declined sharply. Table 4 shows the decline in tra?c on two of Chile’s most heavily travelled highway concessions during this economic downturn. When tra?c forecasts were carried out for most of Table 4 Three-year Chile’s GDP Growth Compared to Santiago–Talca and Santiago– Valparaiso Tra?c Growth
1989–1992 1992–1995 1995–1998 1998–2001 2001–2004 (%) (%) (%) (%) (%) GDP Growth in real terms AADT Growth Santiago–Talca AADT Growth Santiago–Valparaiso 7.94 7.46 7.20 7.75 7.87 4.54 5.72 6.48 7.64 2.34 ÿ0.84 ÿ3.24 3.56 2.76 5.74
366
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
the interurban highway concessions (between 1992 and 1996) planners did not anticipate that, after more than a decade with annual tra?c increments over 7 per cent, tra?c would decline. Consequently, concessions had to cope with lower tra?c levels than expected. Chile has implemented three tra?c-risk mitigation mechanisms: Minimum Income Guarantee (MIG), Least Present Value of the Revenues (LPVR), and Revenue Distribution Mechanisms (RDM). Although all three mechanisms mitigate tra?c risk, there are di?erences between them in terms of both the extent to which they have been applied and the objective the government had with their implementation. MIG was designed only as a way of mitigating tra?c risk so that lenders would perceive less risk and as a consequence the ?nancial cost of the project would become lower. The extent of MIG’s application to the Chilean situation has been widespread, since almost all the highway concession projects have incorporated this guarantee. LPVR was not only designed as a mechanism to mitigate tra?c risk, but also as a procurement mechanism detailing the terms of the regulation of the concession contracts. Unlike MIG, the application of LPVR in Chile to date has been limited; only two concessions have currently been awarded using this mechanism. RDM presents a way to mitigate tra?c risk ex-post by means of a modi?cation of contract terms after the 1998–2002 economic recession. At the end of 2004, six concessions had already implemented this mechanism, and many others were under renegotiation with the government. These mechanisms are not mutually exclusive. For example, the Santiago–Valparaiso project was procured under the LPVR mechanism, and it incorporated the MIG guarantee as well. In this case, the government considered that, despite LPVR, the implementation of MIG was necessary to reduce lenders’ perception of risk. Similarly, RDM is being implemented in projects that have already implemented MIG. 3.2 The minimum income guarantee The primary mitigation mechanism used in Chile was the ‘minimum income guarantee’ (MIG). The total guaranteed income in present value is the same for all the bidders, and it is equal, in present value, to 70 per cent of the investment cost plus the total maintenance and operation costs estimated by the MOP. The 70 per cent ?gure was chosen because, as Esty (2003) shows, on average the percentage of debt in a project ?nance structure is 70 per cent. The mechanism was designed to allow bidders enough ?exibility to establish the minimum band as long as the present value of the guaranteed revenues was no more than 70 per cent of the investment cost plus the total maintenance and operation costs. To that end, the MOP de?ned a range to set up the bands in the bidding terms. If the real revenues fall 367
Journal of Transport Economics and Policy
Volume 40, Part 3
below the lower band in any year, the MOP will have to compensate the concessionaire for the di?erence between the MIG Band revenues and the real revenues at the end of that year. If the concessionaire decided to take the MIG guarantee, it has the obligation to share part of the revenues obtained whenever real tra?c is higher than expected. Two di?erent methods have been implemented for triggering these revenue-sharing mechanisms. The ?rst method triggers revenue-sharing when the rate of return on investment is above 15 per cent in a given year. This rate of return is estimated by the MOP according to the actual revenues and the investment and operation costs estimated initially. The second method established a symmetrical band called a ‘mirror band’, in such a way that if real revenues surpass this band in any year, the concessionaire has to share 50 per cent of the di?erence between the real revenues and the mirror band revenues with the govern´ mez-Lobo and Hinojosa, 1999). ment (Go The MIG has some important advantages. First, the tra?c risk is shared, in that both unjusti?ed pro?ts and losses derived from uncertain tra?c are avoided. Second, the debt holders will feel more comfortable because of the guarantee of part of the revenues, and consequently the ?nancing cost of the project will be lower. This in turn will lower users’ tolls. In spite of the advantages, the MIG could be dangerous for the public sector if many concessions fail simultaneously, as might occur during a substantial economic downturn. In view of this concern, the MOP commissioned a study to address the ?scal implications derived from the implementation of the MIG in order to determine whether budgetary provisions were necessary to cover expected future payments. The study was based on a Monte Carlo simulation and concluded that the 70 per cent guarantee ´ mez-Lobo and level was su?ciently low to avoid this problem (Go Hinojosa, 1999). This study has recurred periodically, incorporating actual information about the concessions. The last version of this study concluded that the value of all the expected cash-?ows payable by the government in the form of minimum income guarantees for highway concessions, from 2004 on, totalled US$ 194 million at present value (Government of Chile, 2003). This amount represents only 3.84 per cent of the total investment involved in the highway concession programme. The performance of the highway concessions in Chile is consistent with the MOP study ?ndings. Almost all the road concessions procured since 1995 requested the MIG, and only two of them had performed under the bottom band at the end of 2003. The government’s payout for these contracts totalled US$ 5 million (US$ 4.8 million and US$ 0.2 million for the Santiago–Los Vilos and Nogales–Puchuncav?´ concessions, respectively). 368
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
Despite the tra?c decrease at the end of the 1990s, the total contribution from the government to the concessionaires due to MIG during the last ten years totalled less than 0.15 per cent of the toll road investment in Chile during the same period. 3.3 The LPVR mechanism The Chilean Public Works Concession Law de?ned the possibility of using the sum of total revenues to be obtained by the project as the main economic variable for tendering concessions. LPVR was primarily designed as a procurement mechanism, although one of the most important consequences of LPVR was the reduction of tra?c risk. Before describing the way the LPVR approach has been applied in Chile, it is important to show the relationship between the present value of the revenues and the ?nancial balance of a concession (equation (1)). Io ÿ S ¼ where: Io : S: a: n: pi : qi ð pi Þ: ci : ti : Initial investment Initial subsidy provided by the government Weighed average cost of capital (WACC) on the project Concession term Price for year i Actual tra?c in year i depending on pi Operation and maintenance costs in year i Taxes or concession fees in year i.
i¼n X ð pi Á qi ð pi Þ ÿ ci ÿ ti Þ ; ð1 þ aÞi i¼1
ð1Þ
Assuming that the public authority in charge of tendering the project does not provide any subsidy at the beginning, equation (1) can be restated as: Io þ
n i¼n X X ci þ ti ð pi Á qi ð pi ÞÞ ¼ : i ð1 þ aÞi i ¼ 1 ð1 þ aÞ i¼1
ð2Þ
The left-hand-side of the equation shows the discounted costs that the concessionaire has to bear in operating and constructing the concession, and the right-hand-side shows the revenues that the concessionaire will obtain along the contract duration. The right-hand-side of the equation re?ects actual LPVR during the life of the contract. At the beginning of the concession, the left-hand-side of the equation will be much higher than the right-hand-side. However, when the operating and maintenance costs are lower than the revenues, the right-hand-side of the equation will 369
Journal of Transport Economics and Policy
Volume 40, Part 3
Figure 1 Economic Balance and Variable Duration Models of a Concession under LPVR Mechanism
ECONOMIC BALANCE OF THE CONCESSION $
Construction Operation Discounted Revenues Discounted Costs
Io
Equilibrium VARIABLE DURATION OF THE CONTRACT DUE TO LPVR $
Construction Discounted Revenues Expected Traffic Low Traffic High Traffic
A
T1
T0
T2
Years
grow more quickly that the left-hand-side. This is the reason why the project needs to borrow money from lenders and promoters at the beginning. The point at which the two terms of the equation are equal means that the concession has covered all its costs — according to a cost of capital equal to a. Figure 1 illustrates graphically how this condition is reached. The auction mechanism based on the LPVR consists of granting the concession to the bidder that requires the lowest present value of the revenues to recover its costs. Supposing that the most competitive bidder — and therefore the winner — o?ered a net present value of the revenues equal to A, the concession will end when the real discounted ?ow of 370
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
revenues reaches A. This system substantially reduces the likelihood of renegotiations in the future. If tra?c is as expected, the duration of the concession will be T0 . However if the tra?c is ultimately lower or higher than expected, the concession duration will be either extended or reduced. In Chile, however, the Concessions Law establishes a maximum duration of 50 years so the concession term can not be extended more than this. This fact poses the problem that if, at the end of the maximum duration, the LPVR reached is lower than the LPVR requested, the concessionaire will not be able to recover its costs. An interesting feature of LPVR is that the discount rate can be either ?xed or variable. For example, in the Santiago–Valparaiso concession, the MOP gave the bidders the choice between a ?xed rate set as a riskfree rate of 6.5 per cent plus a risk premium of 4 per cent, totalling 10.5 per cent; or a variable rate established as the monthly average real riskfree rate of the ?nancial system plus a 4 per cent risk premium. Despite the high level of interest in the LPVR mechanism, it has been used only on a few occasions in Chile (see Table 5). The ?rst concession using LPVR, and also the most successful, was the Santiago–Valparaiso highway (Route 68). An analysis of the procurement process of this ´ mez-Lobo and Hinojosa (2000). Four consortia highway can be found in Go passed the technical evaluation stage. The competition for the award was ?erce, and the economic o?ers were within a very tight range. Table 5 Experiences in which the LPVR has been used as a Procurement Variable to Tender Highway Concession Projects in Chile
Year Investment1 PVR2 Maximum Number of US$ US$ term of tender million million (years) bidders3 1998 2000 340 384 381 – 25 30 4 0
Project Santiago–Valparaiso Costanera Norte
Situation Successfully awarded In operation Not awarded Tendered again a year later under other economic variable Awarded No bidder arrived to the LPVR step Successfully awarded In construction
Talcahuano–Penco
2001
19
–
31.5
2
Acceso Nororiente
1 2
2003
165
346
40
1
Investment predicted by the government. Present value of the revenues o?ered by the granted bidder. 3 Bidders in the last stage of the project.
371
Journal of Transport Economics and Policy
Volume 40, Part 3
The second attempt to tender a highway concession in Chile under the LPVR mechanism took place at the beginning of 1999. The highway selected was the Costanera-Norte, an urban expressway in Santiago, which joins the Northeast and the Northwest of the city passing under the downtown in a tunnel. Three factors contributed to a level of risk for this project that turned out to be much higher than previous interurban experiences. First, the construction was challenging because of both the need to construct a large tunnel below the Mapocho River so as to cross the city centre, and some unresolved environmental problems. Second, tra?c risk was substantial, owing to uncertainty about the competition from other roads or means of transport in the future in the urban area. And third, the project was planned to be the ?rst use in Chile of an electronic free-?ow toll collection system scheme. The substantial risks of the project encouraged MOP to tender this concession under the LPVR mechanism. The results of the tender, however, were disappointing. Only one consortium presented an o?er and it was ultimately disquali?ed because the guarantee bond o?ered was below the level established in the bidding documents. This experience proved that the LPVR was not a magic remedy to get very risky projects o? the ground without public support. In fact the project, which was re-tendered at the end of 1999 — this time not under the LPVR mechanism — was awarded without problem. This time the government committed itself to construct and ?nance part of the works of the project, reducing considerably the concessionaire’s investment. In addition, tra?c risk was reduced through establishing a minimum income guarantee band that was 25 per cent higher than usual. The Chilean government tried LPVR again in 2001 for the Talcahuano– Penco road (Ruta Interportuaria) near the city of Concepcion. This project was much smaller in terms of investment than the two previous ones. The procurement terms of this project provided bidders the option of competing in terms of the LPVR as long as no subsidy was required from the State. However, the three bidders requested a subsidy, therefore the concession was awarded ultimately with a subsidy and LPVR was not used. The last concession procured in Chile under LPVR was the ‘Acceso Nor Oriente a Santiago’ in 2003. This project is a suburban highway that provides easier access from the highways coming from the north of the country to the wealthy neighbourhoods in East Santiago. Although the concession was awarded, only one bidder responded. Like Costanera Norte, this project was extremely risky. The uncertainty was re?ected in a high (more than twice the investment of the project) LPVR requested. 372
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
3.4 The RDM mechanism The problems of many concessionaires as a consequence of the economic recession in Chile combined with the need for additional investment in many of the highway projects moved the government to introduce a new mechanism in order to modify the original contracts. This mechanism was called ‘Revenue Distribution Mechanism’ (RDM). RDM guarantees that a pre-?xed amount of revenues — in present value — will be received by the concessionaire. Thus the duration of the renegotiated concession turns from being ?xed to being variable. The government expected that the average annual tra?c growth would remain at around 3.5 per cent for most of the concession contracts at the time the RDM was considered. The government gave three alternatives to the concessionaires in terms of the present value of the revenues to be guaranteed. Those three alternatives were established in terms of an average annual tra?c growth of 4, 4.5, and 5 per cent respectively. Three reasons explain why the government adopted those values. First, the government decided to guarantee an amount of revenues in present value higher than the revenues forecast so as to take advantage of RDM to require additional investments upfront from the concessionaires. Second, the government decided not to guarantee an average annual tra?c growth much higher than the one predicted so as to avoid a long extension of the length of the contract. Third, the government wanted to give ?exibility to the concessionaires by o?ering them several di?erent alternatives. In exchange for the guarantee of extra revenues, the government required the concessionaire to carry out initial investments, which were calculated by the government as the di?erence between the present value of the revenues guaranteed and the present value of the revenues expected. Thus, the higher the tra?c growth guaranteed, the higher will be the amount of upfront investment to be implemented by the concessionaire. The main innovation of this mechanism is that the term of the concession contract changed from ?xed to variable and is therefore similar to LPVR, since the contract will expire as soon as the concessionaire has collected the revenues guaranteed in present value. RDM causes the concession to ?nish earlier if average tra?c growth is ultimately higher than the guaranteed level. If that happens, the concessionaire will ultimately fare worse than it would have done if it had not obtained the RDM guarantee. If the average tra?c growth is ultimately the same as the guaranteed level, the concession will end in the year initially ?xed. If the average tra?c growth turns out ultimately to be as forecast, the concession contract will expire some years later than the term originally established in the contract. 373
Journal of Transport Economics and Policy
Volume 40, Part 3
Figure 2 RDM Mechanism
Revenues Revenues guaranteed Real revenues
C F E
G
A Revenues expected in year k Additional Investment k
Operation costs H n I m
Year
Figure 2 sketches how RDM works. Year k is the year in which the rede?nition of the contract terms takes place. Tra?c forecasts from this year on are represented by the long-stroke broken line AE. The shortstroke broken line AC represents the average tra?c guaranteed — 4, 4.5, or 5 per cent according to the selection of the concessionaire. In compensation for the extra-revenues guaranteed, the concessionaire has to carry out additional investments in year k. These investments will be equal to the area ACE (dotted area) in present value. If, for instance, the real tra?c level follows the way of the AF line, the duration of the concession contract will be extended until year m, in which the present value of the revenues agreed would be reached. According to the characteristics of the RDM the area FGHI in present value should be equal to the area ECF in present value. There are three main di?erences between LPVR and RDM. First, under LPVR there is a competition to o?er the lowest LPVR, whereas under RDM there is no competition. In this case the government o?ers several alternatives that can be taken or rejected by the concessionaire. Second, unlike LPVR, under RDM the public sector has a choice of obligation: whether to extend the concession contract until the present value is reached or to compensate the concessionaire for the remaining LPVR. In other words, the concessionaire will be certain to receive the whole revenues guaranteed in present value. Third, under RDM once the initial duration of the contract has been reached, the revenues calculated in order to determine the end of the concession are net revenues (revenues minus costs). The last two di?erences between LPVR and RDM imply that, unlike LPVR, RDM does not transfer any downside risk to the concessionaire. 374
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
The additional investments required by RDM have to be approved by the MOP. The amount of these investments in the year envisaged for their realisation — year k — is ?xed by the MOP according to equation (3): AIk ¼ where: AIk : y: REi : RGi : Additional investment required in year k Discount rate equal to the cost of capital (9.5 per cent annually) Revenues expected in year i calculated by the government Revenues guaranteed in year i.
i¼n X RGi ÿ REi i¼k
ð1 þ yÞi ÿ k
;
ð3Þ
The discount rate adopted was ?xed by the MOP as 9.5 per cent annually, according to the MOP’s estimations of the weighted average cost of capital based on previous tenders. In principle, if the guarantee was correctly priced, concessionaires should be indi?erent as to whether to buy this guarantee, or to choose any of the three guarantee alternatives o?ered. However, almost all the concessionaires so far have shown an interest in purchasing the guarantee, and all of them have chosen the highest one (average annual growth equal to 5 per cent). That issue seems to show that actually the discount rate adopted by the government to calculate the present value of the revenues (equal to 9.5 per cent) was higher than the real discount rate as perceived by the companies. An explanation for that di?erence could be that the Chilean Government did not take into account the fact that the discount rate (weighted average cost of capital) to be applied to the RDM should be lower than the discount rate before the project has been constructed. Two reasons justify that fact. First, as the project has already been constructed, the uncertainty regarding construction costs and tra?c generation is much reduced. Second, the modi?cation of the contract from ?xed to variable length substantially reduces the risk of the project and consequently the discount rate to be adopted. The latter reason can be regarded in another way. The RDM guarantees a certain amount of revenue, which includes both the expected revenues and some additional revenues. The additional investment in compensation for the additional revenues is estimated as the present value of the additional revenues discounted at a rate of 9.5 per cent. However, that does not take into account that the risk of the whole project has substantially diminished since the expected revenues are in fact guaranteed. This approach is undoubtedly one of the main reasons why the RDM mechanism became so popular among concessionaires and why all of them chose the highest guarantee. 375
Journal of Transport Economics and Policy
Volume 40, Part 3
RDM is not a substitute for the ‘Minimum Income Guarantee’ (MIG). MIG has the role of providing security to lenders and reducing the ?nancial cost of the project, while RDM has the role of mitigating shareholders’ tra?c risk. The revenues adopted by RDM in order to calculate the present value that will eventually determine the end of the concession contract are the revenues that the concessionaire actually collects no matter where they come from. For instance, it could happen that, after the RDM has been implemented, the tra?c growth greatly increased and, according to the MIG approach, the concessionaire should share with the government part of those extra-revenues collected above the upper band. In that case, the concessionaire should pay to the government the amount of money that the MIG approach establishes for the corresponding year. However, the revenues paid to the government are not taken into account in estimating the present value of the revenues that will eventually determine the end of the concession contract. The revenues that are to be calculated for that purpose are the revenues actually received by the company. Table 6 shows the highway concessions where the concessionaire has already purchased the RDM. Between 1994 and 1998 fourteen highway concessions were awarded in Chile. By the beginning of 2005 six highway concessionaires had purchased the RDM mechanism and many other concessionaires were negotiating with the MOP in order to acquire the guarantee. The RDM guarantee has been used by some of the concessionaires as collateral for new bonds to ?nance the additional highways investment required by the RDM. For instance, the Santiago–Talca highway — the largest project using the RDM guarantee — issued a new bond in 2004 to ?nance both a second phase of the project and the additional works required by the RDM. Table 6 Characteristics of the Highway Concessions which have already purchased the RDM in Chile
Year concession awarded 1994 1995 1995 1997 1997 1998 First year of operation 1996 1997 1998 2001 2002 2002 Initial duration (years) 25 22 10 22 25 25 Premium (US$ million) 3.06 1.47 29.00 19.89 25.23 73.73 Initial investment (US$ million) 26 12 169 241 255 575
Project Camino de la Madera Camino Nogales Puchuncav?´ ´n Ruta 5 Tramo Talca–Chilla ´ n–Collipulli Ruta 5 Tramo Chilla Ruta 5 Tramo Collipulli–Temuco Ruta 5 Tramo Santiago–Talca
376
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
3.5 The consequences of the 1998–2002 recession Chile’s economic recession from 1998 to 2002 provides an opportunity to assess the performance of tra?c and income guarantees under lowerthan-projected tra?c volumes. Although the recession caused economic growth to be lower than predicted, Chile’s recession is not comparable to other huge recessions experienced in Latin America, which reduced GDP by more than 10 per cent in a year (that is, Argentina 2002, Venezuela 2003, Mexico 1995). The recession a?ected the shareholders more than the lenders, since the payback of the loans was largely covered by MIG. The concession shareholders attempted to renegotiate the contract terms based on an article of the Chilean Concession Law that permits the economic terms of the concession to be modi?ed in case of unpredictable circumstances. The Ministry of Public Works rejected this approach on the grounds that the concessionaires’ tra?c risk was mitigated through the MIG, and because the Chilean concession model clearly de?ned tra?c risk as a risk to be held by the concessionaire, regardless of any guarantees that could be applicable according to the law. At the same time that the concessionaires were arguing for renegotiation, the MOP realised that additional investments would be necessary in most of the highway concession projects. Some designs of earlier concessions were now thought to be inadequate (poor connections to the local network, safety problems, and so on). A common problem was that additional pedestrian and road bridges were needed to provide access all along the highway. This is the reason why the MOP decided to apply the RDM guarantee. The MOP permitted the possibility of obtaining this guarantee to all the highway concessions that were awarded before 2000. The sole exception was the Santiago–Valparaiso concession, which was the only highway project then in operation that had been tendered under the LPVR mechanism. This is an interesting demonstration that LPVR performed very well in terms of tra?c risk mitigation, since it is the only concession that neither the government nor the concessionaire tried to renegotiate.
4.0 Lessons from the Chilean Experience
LPVR is conceptually very attractive for several reasons. First, a variable term is a highly e?ective compensation method that neither commits public resources nor entails tari? increases. Second, LPVR sets up a clear buy-out price. And third, LPVR reduces renegotiation expectations so 377
Journal of Transport Economics and Policy
Volume 40, Part 3
the bidders have less incentive to in?ate their o?ers. However, among the 26 road projects that were granted in Chile during the last decade, only four were tendered, and only two were successfully awarded under this approach. In a set of interviews conducted by the author in Chile, all the stakeholders — the MOP, the concessionaires, university professors, and so on — point to the strong opposition from concessionaires as the main reason limiting the application of LPVR. Concessionaires’ concerns about LPVR are threefold. First, LPVR does not improve the capacity of the project to ful?ll its commitments to the lenders every year. Second, the variable length of the contract makes the concession operation di?cult to organise since resource planning cannot be programmed in advance. Finally, LPVR always limits the upside pro?tability of the concessionaire but not always the downside, so the mechanism is not symmetric. Chile’s experience suggests that the ?rst two criticisms are not a major concern. Regarding the ?rst criticism, the Santiago–Valparaiso highway bond issued in 2002 was the largest and least expensive infrastructure bond issued in Chile until then. Due to the variability of the concession term, the bond was structured with a mechanism of mandatory prepayment. Regarding the operational organisation and resource planning, the Santiago–Valparaiso experience demonstrates that information about actual tra?c during the concession facilitates the prediction of when the contract will end. Thus there is ample time for resource planning. Moreover, the organisation of the concession operation under a variable term approach was ultimately not as complicated as expected. The third criticism — an asymmetrical risk pro?le — is of concern. As Brealey, Cooper, and Habib (1996) assert, private shareholders are interested in the upside of the project. This is easy to understand since, as concession projects are usually highly leveraged, the sponsors expect a large upside that compensates for the possibility of losing all their capital. Concessionaires are also interested in avoiding the downside because this way the lenders perceive lower risk, permitting the concessionaires to enjoy the bene?ts stemming from higher leverage and lower interest rates. The risk pro?le drawn by LPVR is the opposite of the concessionaire desires described above. On the one hand, the upside is almost non-existent. On the other hand, as a maximum duration is established in the concession contract, the concessionaire bears the risk that the project will not reach the LPVR requested at the maximum term. This is the reason why promoters regard this mechanism as asymmetrical. This asymmetry was already ´ mez-Iban pointed out by Go ˜ ez (2003) who mentions that whereas the government has a call option on the project for the remaining present 378
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
value of the contract, the concessionaire does not have a put option whereby it can sell the project to the government at the end of the contract in exchange for the present value remaining. The duration constraint problem is di?cult to solve with an extension of the concession contract duration. First, the maximum duration is limited by the Concession Law so it is not legally possible to establish concession durations longer than 50 years in Chile. Second, even if there was not a duration limit, it would not seem appropriate to establish very long durations since the longer the duration the more di?cult it is to specify complete contracts. An interesting approach to solving this problem is the one developed by RDM. This mechanism forces the government to pay the LPVR left at the end or extend the concession until the LPVR requested is ultimately reached. In this extension, the government can modify the contract with the concessionaire to adapt the new environmental and technological requirements. Three lessons can be highlighted from Chile’s experience. The ?rst lesson is that establishing a mechanism, such as LPVR, to mitigate tra?c risk based on extending or reducing the concession length has been demonstrated to work quite well with the caveat that the risk pro?le drawn by it was not well-regarded by private promoters. Moreover, LPVR has proved e?ective in reducing renegotiation expectations. In fact, as a result of the 1998–2002 economic recession, the LPVR-based Santiago–Valparaiso concession was the only one where the government did not allow the possibility of reopening renegotiation. The second lesson is that, unlike LPVR, the establishment of a mechanism like MIG does not reduce renegotiation pressures from promoters when the tra?c is lower than expected. Moreover, although this mechanism behaved reasonably well in Chile during the 1998–2002 economic crisis, such a crisis was very small compared to the economic crises su?ered by other developing countries in the last few decades. This fact suggests that, in spite of the good performance of MIG in Chile, it still transfers an important risk because of the important correlation between tra?c and economic growth. This point can be a serious problem for the implementation of this mechanism in unstable economies. The third lesson is that, although the risk pro?le of this mechanism was not initially attractive for private investors, the mechanism seems to become more attractive when the downside risk is limited. This lesson, however, has to be approached carefully. Chilean concessionaires assert that they positively respond to a mechanism, such as RDM, in which the downside risk is eliminated. Yet the success of RDM among concessionaires seems to be motivated not only by this factor, but also by an overestimation of the discount rate. 379
Journal of Transport Economics and Policy
Volume 40, Part 3
These lessons suggest that LPVR is a very attractive mechanism for procuring highway concessions and limiting tra?c risk. However, from experience so far, two measures can be suggested that can help to improve concessionaires’ perception on LPVR. First, there is a need to implement a limit on the downside risk. Second, it is necessary to establish a minimum concession duration in such a way that the concessionaire will enjoy an upside if the tra?c is higher than expected. Since to date LPVR has been implemented only in highway concessions, it would be useful to devote future research to studying the applicability of LPVR to demand-risk mitigation in other infrastructure facilities — such as railroads, ports, airports, water and sewage facilities, and so on — and even to long-term service contracts.
References
Bain, R. and M. Wilkins (2002): ‘Credit Implications of Tra?c Risk in Start-Up Toll Facilities’, Standard & Poor’s Project and Infrastructure Finance, Standard & Poor, London. Bain, R. and M. Plantagie (2004): Tra?c Forecasting Risk: Study Update 2004, Standard & Poor, London. Brealey, R. A., I. A. Cooper, and M. A. Habib (1996): ‘Using Project Finance to Fund Infrastructure Investment’, Journal of Applied Corporate Finance, 9, 25–38. Engel, E., R. Fischer, and A. Galetovic (1997): ‘Highway Franchising: Pitfalls and Opportunities’, American Economic Review, 87, 68–72. Engel, E., R. Fischer, and A. Galetovic (2001): ‘Least Present Value of Revenue Auctions and Highway Franchising’, Journal of Political Economy, 109, 993–1020. Esty, B. C. (2003): Modern Project Finance, a Casebook, John Wiley & Sons. Flyvbjerg, B., M. K. Skamris Holm, and S. L. Buhl (2005): ‘How (In)accurate Are Demand Forecasts in Public Works Projects? The Case of Transportation’, Journal of the American Planning Association, 71, 131–46. Foice, D. (1998): ‘Second Severn Crossing’, Proceedings of the Seminar PPP Risk Management for Big Transport Projects, Ministerio de Fomento, Spain. Government of Chile (2003): Report on Public Finances. Government Budget Bill for 2004, Ministry of Finance, Santiago. ´ mez-Iban Go ˜ ez, J. A. (2003): Regulation of Private Infrastructure: Monopoly, Contracts, and Discretion, Harvard University Press. ´ mez-Iban Go ˜ ez, J. A. and J. R. Meyer (1993): Going Private: the International Experience with Transport Privatization, The Brookings Institution, Washington, DC. ´ mez-Lobo, A. and S. Hinojosa (2000): Broad Roads in a Thin Country: Infrastructure Go Concessions in Chile, Research Paper 2279, The World Bank, Washington DC. Guasch, J. L. (2004): Granting and Renegotiating Infrastructure Concessions Doing it Right, WBI Development Studies, The World Bank, Washington DC. Irwin, T. (2003): Public Money for Private Infrastructure: Deciding When to O?er Guarantees, Output Based Subsidies and other Fiscal Support, World Bank Working Paper 10, Washington DC. Izquierdo, J. M. and J. M. Vassallo (2004): Nuevos sistemas de gestio´n y ?nanciacio´n de ´ n SEINOR no 35, Colegio de Ingenieros de infraestructuras de transporte, Coleccio Caminos, Canales y Puertos, Madrid.
380
Tra?c Risk Mitigation in Highway Concession Projects
Vassallo
Lemos, T., D. Eaton, M. Betts, and L. Tadeu de Almeida (2004): ‘Risk Management in the Lusoponte Concession — a Case Study of the Two Bridges in Lisbon, Portugal’, International Journal of Project Management, 22, 63–73. Nombela, G. and G. de Rus (2003): ‘Flexible-term Contracts for Road Franchising’, Transportation Research Part A, 38, 163–79. Ru?an, D. M. (1999): Manual de Concesiones de Obras Pu´blicas, Fondo de Cultura ´ mica, Santiago, Chile. Econo Shugart, C. T. (1998): Regulation-by-Contract and Municipal Services: The Problem of Contractual Incompleteness, PhD Dissertation, Harvard University. Vassallo, J. M. (2004): ‘Short Term Infrastructure Concessions: Conceptual Approach and Recent Applications in Spain’, Public Works Management and Policy, 8, 261–70. Vassallo, J. M. and J. Gallego (2005): ‘Risk-sharing in the New Public Works Concession Law in Spain’, Transportation Research Record 1932, 1–8. Zhao, Y. and K. M. Kockelman (2002): ‘The Propagation of Uncertainty through Travel Demand Models’, The Annals of Regional Science, 36, 145–63.
381
doc_583185278.pdf