A System dynamics model to determine concession period of PPP infrastructure projects: the overarching effects of critical success factors

UPCommons Portal del coneixement obert de la UPC http://upcommons.upc.edu/e-prints Aquesta és una còpia de la versió author’s final draft d'un article publicat a la revista Journal of Legal Affairs and Dispute Resolution in Engineering and Construction URL d'aquest document a UPCommons E-prints: http://hdl.handle.net/2117/123255 This material may be downloaded for personal use only. Any other use requires prior permission of the American Society of Civil Engineers. This material may be found at <https://ascelibrary.org/doi/pdf/10.1061/(ASCE)LA.19434170.0000280>. 1 A System Dynamics model to determine concession period of PPP 1 infrastructure projects: The overarching effects of critical success factors 2 3 Fahim Ullah, MS (Corresponding Author) 4 1. PhD Student, Faculty of the Built Environment, University of New South Wales, Sydney,5 Australia. 6 2. Lecturer, Dept. of Construction Engineering & Management (CE&M), NIT-SCEE,7 National University of Sciences & Technology (NUST), Islamabad, Pakistan. 8 Phone: +92-51-90854165 9 Email address: [email protected] 10 Postal Address: Dept. of Construction Engineering & Management (CE&M), NIT-SCEE, 11 National University of Sciences & Technology (NUST), H-12, Islamabad, 44000 Pakistan. 12 13 Muhammad Jamaluddin Thaheem, PhD 14 Assistant Professor, Dept. of Construction Engineering & Management (CE&M), NIT-15 SCEE, National University of Sciences & Technology (NUST), Islamabad, Pakistan 16 Phone: +92-51-90854164 17 Email address: [email protected] 18 Postal Address: Dept. of Construction Engineering & Management (CE&M), NIT-SCEE, 19 National University of Sciences & Technology (NUST), H-12, Islamabad, 44000 Pakistan. 20 21 Samad M.E. Sepasgozar, PhD 22 Lecturer, Faculty of the Built Environment, University of New South Wales, Sydney, 23 Australia 24 Phone: +60 (2) 9385 6528 25 2 Email address: [email protected] 26 Postal Address: Faculty of the Built Environment, University of New South Wales, Sydney, 27 Australia. 28 29 Nuria Forcada, PhD 30 Associate Professor (Group of Construction Research and Innovation – GRIC), 31 Department of Project and Construction Engineering, Universitat Politècnica de Catalunya – 32 UPC, Terrassa (Barcelona), Spain. 33 Phone: + 34 93 7398153. 34 Email address: nuria.fo[email protected] 35 Postal Address: C/Colón, 11 - Edifici TR5 - 08222 Terrassa (Barcelona) – Spain 36 37 Abstract 38 The determination of the concession period (CP) in Public-Private-Partnership (PPP) infrastructure 39 projects has presented complexities to decision makers since various critical success factors (CSF) are 40 involved which may be overlooked. This paper outlines a System Dynamic (SD)-based approach to 41 provide an in-depth understanding of CSFs that determines PPP projects’ CP and models them for 42 localized use. The CSFs are obtained from published literature, duly vetted through a survey of 56 experts 43 and are used to develop a quantitative and qualitative SD model, validated by simulating case studies of 44 five infrastructure projects. A total of 59 concession affecting CSFs are highlighted that are reduced and 45 compared for localized infrastructure projects. The findings, pointing to the need of CP being dynamic 46 instead of traditional fixed and static, warrant an extension in the originally proposed concession in three 47 cases and reduction in the remaining two. The decision-making implications of this study target the three 48 key stakeholders: public body through reduced financial risks, private organization through increased 49 confidence and reassessment of concession during project life, and end user through nominal tolls. 50 3 Keywords: Public-Private-Partnership; concession period; critical success factors; infrastructure 51 projects. 52 Introduction 53 Private sector participation in infrastructure financing and management is growing around the 54 world and this increase is taking special relevance in the case of road projects procured under public55 private-partnership (PPP) (Rahmani et al., 2017). As evident from the case studies in Europe, the growing 56 trend of private sector participation in road projects has resulted into more stringent regulations (Albalate 57 et al., 2013). Privatization of construction projects is an emerging market in the United States of America 58 (USA) as well. There have been significant greenfield projects such as the Dulles Greenway, and 59 brownfield projects including Chicago Skyway, Indiana Toll Road and Pennsylvania Turnpike (MWAA, 60 2015). A general trend of prospects for privatization of roads in New Jersey and build-operate-transfer 61 (BOT) concessions in Texas has surfaced. This wave of privatization is accompanied by a renewed 62 interest in the way the public sector regulates (Cheung et al., 2010; Meng and Lu, 2017). This interest 63 can be explained by the potential redistributive effects due to the right of exploiting a network asset such 64 as a motorway. 65 PPP is evidently getting a lot of attention by governments in different parts of the world. One 66 stated reason behind this trend is the concern over public expenditure (Cheung et al., 2010; Ullah et al., 67 2016). PPPs are often presented as a core part of modernizing public services, since they improve the 68 quality and efficiency of public services (Liu and Wilkinson, 2014). However, from a more skeptical 69 view of their value, some studies point to their complexity, shortand long-term costs and consequences 70 for labor and those reliant on public services (Osei-Kyei and Chan, 2015). Such complexities often lead 71 to the failure of PPP projects. 72 One of the most complex decisions in PPP projects is the determination of the concession period 73 (CP) which is the time allotted to a private entity for operating a constructed facility to recover its 74 4 expenditures and gain some acceptable return on investment (Ng et al., 2007b; Osei-Kyei and Chan, 75 2017). The project is handed over to public authority at the end of this term. CP reconciles risk-reward 76 balance for the private party and end user: the prior in form of speedy financial recovery, and the latter 77 in form of economic tolls and greater public welfare (Akintoye and Chinyio, 2005; Carmona, 2010; Hu 78 and Zhu, 2014; Shi et al., 2016). Thus, if not estimated properly, CP may trigger project failure (Cheung 79 et al., 2010; Tieva and Junnonen, 2009; Trangkanont and Charoenngam, 2014). 80 In terms of legal aspects, PPPs and their CPs are evolving, and the methodology to calculate it 81 has yet to be matured. This is mainly due to the length of CPs (typically 15 to 30 years or longer) making 82 the formulation of a comprehensive contract very difficult. Furthermore, PPP projects are frequently re83 negotiated or bought back, when parties attempt to amend the critical elements, such as tolls and tariffs 84 and subsequent readjustments for financial obligations, making the formulation of a globally accepted 85 PPP legal framework nearly impossible (Mouraviev and Kakabadse, 2017a). Therefore, it is better to 86 look at PPP in terms of the principal features that distinguish it from other types of agreements between 87 the government and private organizations, keeping in view its unique contractual aspects. According to 88 Mouraviev and Kakabadse (2017b), some of the legal challenges faced by PPP CPs are, but not limited 89 to, incomplete and inconsistent legislation surrounding partnerships, weak institutional development, 90 virtually non-existent civil participation, underdeveloped financing institutions, unclear stance from the 91 government regarding approaches to PPP project preparation, lack of PPP-specific governance structures 92 and established procedures for partner interaction, tariff adjustments, dispute resolution, ambiguous 93 government approach towards risk allocation, excessive government regulation where the government 94 focuses on input, and a contradictory perception of a policy paradigm, where the latter serves as an 95 instrument for massive PPP deployment, although the government commitment to partnerships quickly 96 disappears after a PPP is launched. 97 Although it is crucial to precisely determine the CP due to the growing innovation and quality 98 demands (Carbonara et al., 2015; Ullah et al., 2017b), the estimating complexities result into a lack of 99 5 valid and globally applicable model. In response to this research gap, the current study aims to develop 100 a novel model based on systems dynamics (SD) concept to estimate the CP realistically and induce more 101 adoptability by incorporation of localized critical success factors (CSFs). The model, based on the stocks 102 and flows, and positive and negative loops, operates in a systemized way to simulate various possible 103 decisions, and yields the results based upon randomized iterations until a constant trend is observed 104 where the iterations are stopped and results are compiled. This model underlines the dynamic behavior 105 of CPs instead of static traditional approach. Therefore, CPs should be revised after some time and 106 adjusted accordingly depending upon the set terms and conditions, and project forecasted performance. 107 Concession period of PPP projects 108 Various concession-based failure stories have been reported citing the unforeseen costs, improper 109 risk allocation, legal ambiguities, lack of responsibilities and local public resistance as contributing 110 factors (Carbonara et al., 2015; Domingues and Zlatkovic, 2015; Khanzadi et al., 2012). Similar to other 111 project types, value for money (VFM) in PPP projects is carefully assessed due to the associated financial 112 risks (Kumaraswamy et al., 2007; Love et al., 2015). Although before making a PPP procurement 113 decision, the economic justification of VFM is assessed, some other factors including project risks, 114 construction complexities and innovations, traffic congestion problems, lack of proper and legal 115 government policies, and social welfare are ignored (Carmona, 2010; Hu and Zhu, 2014; Jefferies and 116 McGeorge, 2009; Love et al., 2015). For example, the M25 widening scheme in United Kingdom (UK) 117 is criticized for its higher than necessary costs and lack of alternative options of using the hard shoulder 118 as an extra lane during peak hours which led to a potential extra cost of around GBP 1 billion to taxpayers 119 (Marsden, 2005). Similarly, Lagos-Shagamu-Ibadan concession road project is a PPP failure case in 120 which many people had to be relocated and payment of royalties emanated. Several obligations, not 121 captured in the contract package at the pre-contract stage, escalated into serious cases of protracted 122 litigations (Akintoye and Main, 2007; Opawole and Jagboro, 2016; Trangkanont and Charoenngam, 123 2014). 124 6 CP is dependent upon various dynamic factors like project income (PI), toll price, severity of 125 involved risk (SIR) and market situation (MS) (Akintoye and Chinyio, 2005; Wang et al., 2011). Its 126 allocation and optimization decisions have been discussed by various researchers (Karim, 2011; Ke et 127 al., 2010). However, the effects of CSFs on concession estimation are less explored proving a research 128 gap for exploration (Ullah and Thaheem, 2017). This effect is multiplied in case of developing countries. 129 Literature reports a simulation model proposed by Ng et al. (2007b) for determining CP that incorporates 130 the risk of optimum tariff and payback in tendering stage. A regulatory contractual regime has been 131 developed by Cruz and Marques (2012) for evaluating the risk allocation in Portuguese road concessions. 132 They suggest that government must assume risk of production and provide incentives in case of highway 133 development. Similarly, several authors have come up with theoretical frameworks for obtaining VFM 134 from PPP concessions using assessment techniques of public sector comparator, competitive bidding, 135 shadow bid, cost benefits analysis, lease-purchase analysis and public interest test central guidelines 136 (Love et al., 2015; Tieva and Junnonen, 2009). Trebilcock and Rosenstock (2015) propose that 137 governments should anticipate and plan contingencies as well as conduct enforcement and monitoring 138 of long-term PPP contracts. They highlight the capacity as a key determinant of PPP success and 139 reduction of potential concession renegotiations in Latin America. 140 These frameworks not only address the development goals but also attempt at reducing the legal 141 risks of such CPs. In general, there are two types of perspectives on PPPs in terms of the relations between 142 the public and private organizations: the contractual perspective focusing on the legal aspects, and the 143 task specific focus aimed at the project execution and delivery through various stages of its life cycle 144 (Zhang et al., 2015). The legal aspects include, but are not limited to, proper rights, polity, administrative 145 regulations, judiciary and bureaucracy. These aspects, if not properly catered for, may lead to project 146 delays or subsequent failures. For example, the first Polish PPP was delayed by more than a decade due 147 to the lack of legal implications (Albalate et al., 2015). Similarly, Gurgun and Touran (2013) compare 148 various legal frameworks for PPPs and stress their importance. These legal systems must exist but should 149 not be overly complicated as suggested by Edkins and Smyth (2006). The authors argue that an overly 150 7 complicated legal system will not only demand hiring of and extensive payments to legal advisors but 151 also hinder the much aimed ‘partnership’, making the relation an adversarial one instead of collaborative. 152 Although recent studies have shifted their focus towards CP risk, new challenges are brought 153 about by emerging risks associated with lately highlighted critical factors. New influential factors that 154 must be incorporated into the decision models are identified (Carmona, 2010; Yu and Lam, 2013). PPP 155 CSFs are revised from time to time to enable a better execution and management that lead to new 156 challenges. 157 CSF to determine the CP 158 In infrastructure projects, CSF to determine the CP include financial aspects such as equity 159 allocation, toll and service price, economic viability, operation and construction costs to the public, social 160 welfare such as end user purchase subsidy, environmental concerns and elimination of traffic congestion 161 (Liu and Wilkinson, 2014; Shi et al., 2016; Wibowo and Wilhelm Alfen, 2014). These factors, , affect 162 project success and increase its complexity. 163 CSFs for infrastructure projects include, but are not limited to, net present value (NPV), traffic 164 congestion and road usage, effectiveness of public bodies, population in nearby area, adequacy of project 165 funding and annual operational revenue (Alireza et al., 2013; Zhang and AbouRizk, 2006). Since CSFs 166 induce risk in PPP projects, they warrant effective management to achieve successful execution and 167 subsequent closure. 168 Further, closure risk, which refers to a funding agency’s capability to reach financial closure, poses 169 great concern for the public entity (Domingues and Zlatkovic, 2015). However, at the feasibility and 170 planning stages, a poor CP estimate poses even greater challenge to both public and private entities. For 171 the public agency, an administrative and decisive scenario arises that sways the bid process. Having 172 estimated the total project cost, the public agency may decide for the minimum possible offer to accept 173 (Kumaraswamy et al., 2007; Wibowo and Wilhelm Alfen, 2014). It has also implications in establishing 174 8 the toll rate as the public agency is responsible for not only ensuring financial recovery on part of donor 175 agencies but also providing market compatible rates in the light of economic capacity of users (Shi et al., 176 2016). 177 For the private sector, this risk is businessand project-critical since the corporate sector needs to 178 guarantee a profitable return on investment within a stipulated time (Carmona, 2010; Ng et al., 2007a). 179 Any extension in this time may pose serious financial pressures demoralizing the private investors from 180 doing further business in that market (Carbonara et al., 2015). In the case of no extensions, the financier 181 has no other option but to escalate the revenue by increasing usage charges, transferring the financial 182 burden to end user. Therefore, it is crucial to perform thorough risk management for concessions in the 183 early stages of project development. To meet this demand, considerable research has gone into 184 identifying CP duration of BOT based PPP projects (Meng and Lu, 2017; Yu and Lam, 2013). 185 The general trend suggests that risk allocation between the public and private sectors involved in 186 public-purpose transport investments is an onerous and tricky matter owing to the opportunistic behavior 187 of PPP stakeholders (Domingues and Zlatkovic, 2015; Wibowo and Wilhelm Alfen, 2014). Although the 188 allocation of risk between public and private stakeholders varies from country to country, a major 189 consensus is found over the assumption of entire financial risk by the private sector who undertakes the 190 long-term maintenance and operation responsibilities (Ke et al., 2010). An exception to this risk sharing 191 is such projects where promotion and development of some underdeveloped and deprived part of the 192 country is the priority (Wibowo and Wilhelm Alfen, 2013). The effect of risk is further exacerbated due 193 to the time factor as these concessions are usually very lengthy (25 – 30 years), giving rise to the 194 stochasticity of various estimates (Tieva and Junnonen, 2009). 195 Owing to these challenges, concession-based projects based on their context in terms of financial, 196 social and environmental aspects should be investigated (Hu and Zhu, 2014; Jefferies and McGeorge, 197 2009; Wibowo and Wilhelm Alfen, 2013). The current literature mainly covers developed countries such 198 as USA, UK, Canada and China. The challenges of developing and underdeveloped construction 199 15 direct involvement in PPP projects in senior management capacity. The CSFs weights obtained in the 344 previous steps were sought along with comments over using NPV for CP estimation. Projects not using 345 NPV were not considered for this study. The 40% of surveyed experts were from Pakistan while others 346 were from Middle East, USA, UK, Europe, Africa, Australia, China, India and Malaysia. 347 Step 3: Development of a qualitative SD model 348 Based on the CSFs and the quantitative model, a generic qualitative SD model was developed. 349 The complexity of the model suggested developing a reduced SD model including the most significant 350 variables (CSFs), combining variables (clusters) including sub systems (NPV and estimation process) 351 and the main system (CP). The model moves forward starting from the very basic factors to their 352 constituent cluster factor which in turn adds into the two mentioned sub systems finally giving the value 353 of K to be multiplied with the main system. The SD models are developed using Vensim PLE ®, a 354 dedicated software designed for modelling one or more quantities that change over time. 355 Step 4: Model validation 356 A total of 5 PPP infrastructure projects were used to validate the developed models. For the sake 357 of overarching effect, the case projects encompass completed, in execution and under planning projects. 358 This allows examining holistic pros and cons of the proposed models. 359 The basic details of the case projects such as starting dates, type, length, location, key parties 360 involved, original concession and project costs are presented in Table 1. It can be observed that all case 361 studies are mainly headed by National Highway Authority (NHA) and Frontier works organization 362 (FWO), the prior being the public body and latter being a major concessionaire. Four of the projects are 363 in Punjab province whereas one is in Sindh province. The minimum inclusion criteria for the projects is 364 set at having a CP of at least 10 years to study it thoroughly. 365 [Insert Table 1 here] 366 16 To evaluate the weights of the CSFs for each case study, an updated version of the previously 367 used questionnaire survey was utilized with the additional option of an increase or decrease in the value 368 of a CSF over project life cycle. This was done to incorporate the dynamic values of CSFs. Project 369 managers and upper managerial executives were asked to rank and determine the increase or decrease of 370 CSFs for each case study. In case of difference in values, multiple rounds of discussions were performed 371 to reach a consensus and avoid any potential conflict. 372 Results 373 Factors affecting CP 374 From the literature review, a total of 70 factors were identified, some of which were merged. For 375 example, the factor constructability was named as “construction ease”, “buildability” and “erecting” in 376 various papers which were merged. Similarly, traffic congestion was named as “increase in traffic”, 377 “more traffic movement” and “escalated traffic”. Thus, 59 factors were used in the questionnaire survey 378 as shown in Table 2. 379 [Insert Table 2 here] 380 From the questionnaire survey, the top 10 factors affecting CPs and their weights were 381 determined as shown in Table 3. The factors are arranged based on the global (overall) weights. RS is 382 top ranked factor because PPPs involve huge investments that add to serious financial complications, 383 attracting academic research. Industry is also concerned about its finances and assigns high value to this 384 factor. Similarly, risk in PPP arise from many sources including complex relationships, legal aspects, 385 huge stakes and stakeholder satisfaction, making it a significant factor (Osei-Kyei and Chan, 2017). 386 Further, the scores from both AW and IW were harmonized and converted to the same scale (out of 10). 387 Thus, the value 7.438 for RS means that it has a score of 74.38% in AW. The IW score was obtained 388 directly on a scale out of 10 so no conversion was required. 389 17 [Insert Table 3 here] 390 The top 10 factors were reduced to six factors as shown in Table 3 in which the percent effect 391 shows the normalized influence of a factor on the concession decision and formulates the weight given 392 to the factor in the developed model. For normalizing the factors scores, overall score was calculated by 393 adding all values and then finding their percent contribution out of 1 so that the overall factors values 394 will sum up to 1 in the absence of any external factor. The merging is carried out based on the functional 395 similarities and discussion with the experts. This was done to reduce the number of factors and eliminate 396 the confusion of certain factors pointing to the same thing. 397 Table 4 indicates that NPV receives the highest effect. This means that the experts are inclined 398 towards using NPV based assessment of concession which is in line with the existing literature. Keeping 399 this in view, NPV value is fixed in all case projects and remaining are sought while collecting data. Also, 400 changing NPV will mean challenging existing published frameworks which is not the aim of this study, 401 rather it incorporates additional CSFs. Further, the aim is to induce dynamism in CP estimation. 402 [Insert Table 4 here] 403 By substituting the values of percent effect in corresponding variables given in Equation 10, the 404 following equation is obtained. 405 𝐊 = 𝟎. 𝟐𝟗𝐍𝐏𝐕 + 𝟎. 𝟐𝟗𝐏𝐈 + 𝟎. 𝟏𝟒𝐑𝐒 + 𝟎. 𝟏𝟒 𝐒𝐈𝐑 + 𝟎. 𝟎𝟖𝐌𝐒 + 𝟎.𝟎𝟕𝐒𝐈 406 407 408 SD model 409 Based on the previously described methodology, the proposed SD model to estimate modified 410 CP is developed (see Fig 2). The positive or negative influences of each factor are defined with the “+” 411 or “-” signs. For example, the traffic count positively effects the NPV of a project and is marked with a 412 positive sign. Similarly, construction costs negatively affect the NPV as evident from the corresponding 413 18 sign. Thus, all the factors have been marked accordingly to represent their positive or negative influence 414 on CP. These influences result into reinforcing and balancing loops whose combination keeps the system 415 in a natural balance. 416 [Insert Fig.2 here] 417 Adopting the FIDIC price adjustment formula (PEC, 2009), the value of K comprises of fixed 418 as well as adjustable portions. The formula fixes 35% cost items and allows for a variable portion of 419 65%. Since the formula caters for the factors in the same way as intended by the developed model, its 420 logic can be borrowed for the current study. For example, in the formula, 35% items are fixed in any 421 case and should not be disturbed. Following the same lines, the developed model fixes 29% influence of 422 decision for NPV based on the survey responses. The remaining 71% can be varied based on the 423 identified CSFs and their localized percent effect. Thus, the value for K will be between 0.29 to 1.71 424 with 1, as the normal value in expected circumstances, 0.29 meaning a 100% decrease and 1.71 meaning 425 100% increase in the value of CP0. 426 Furthermore, the values for the factors should always be used in portion of 1 and added to or 427 subtracted from 1 depending upon the relations shown in equations. For example, in case of a project 428 being 20% riskier than normal, the CP value will be factored in at 1.2. This value will be multiplied with 429 the corresponding coefficient to get the value of K. Similarly, in case of the expected PI being 20% more, 430 the CP value will be factored in at 0.8. Thus, the range is fixed and cannot be altered due to both 431 mathematical and simulation limitations. Other ranges can be explored in future studies. 432 Based upon these logical constraints and the quantitative model, the SD model is developed. For 433 the synthesis and validation of the proposed model, various iterative simulations are run under best, worst 434 and normal values. Based on the triangular distribution, corresponding input values are provided to 435 achieve graphical results. The input values are in the range of 0 – 2 with 1 being the normal value. In all 436 cases, a test concession of 20 years is simulated. The resulting graphs display the number of years on 437 vertical axis where ranges are shown on left side and exact year on right. Start and end years of 438 19 concession are shown on the horizontal axis. The dark lines refer to the trend for concession whereas 439 lighter lines show the randomized values at any point in the project life cycle. It must be noted that the 440 system was not restricted to a particular value but a range of three-point random values was used as an 441 input, giving a random pattern. To make sense of graphs, focus must be on the trend lines which is the 442 actual CP pattern. Fig. 3 shows that the maximum allowable extension for an original period of 20 years 443 is 34.2, which is quite logical since the system allows for 71% extension in the light of proposed model. 444 In addition, the maximum reducible length of this concession is 5.8 years due to 29% fixed portion. Thus, 445 the system is working in accordance with the developed equation with maximum allowable extension 446 between 0.29 and 1.71. 447 [Insert Fig. 3 here] 448 To make it further understandable, the maximum extendable and reducible lengths of CPs were 449 checked separately. This was achieved with two-point iterations with values 0 for obtaining maximum 450 reduction and 2 for maximum extension. The resulting graphs are shown in Fig. 3. The maximum 451 allowable extension is graphically represented using values of 1 for the 20-year project and 2 for the 452 factors to show the extension. The red trend line and corresponding pattern refers to the 20-year 453 concession whereas the blue trend line and corresponding pattern refers to the allowable extension. 454 Model validation 455 After obtaining the trend lines for maximum possible variations in the CP0, the model is validated 456 using five case studies. The case study projects were simulated using the developed model based on the 457 increase (I) or decrease (D) in the value of a particular factor from inception to its current state, the 458 corresponding K and the CP0 to obtain the MCP. The increase or decrease incorporates the dynamism 459 in values as at any particular time; there can be an increase or decrease in the value of a CSF that can be 460 catered for in the upcoming revision of CP. Thus, CP can be revised accordingly to ensure a win-win 461 situation for both public and private entities. Project managers and upper managerial executives ranked 462 and determined the increase or decrease of each CSFs for each case study. In case of difference in values, 463 multiple rounds of discussions were performed to reach a consensus. Table 5 presents the accorded 464 20 factors for each project. For example, the first project observed an increase of 11% in its RS as reflected 465 in corresponding RS column. The resulting K value indicates a reduction in the original concession. 466 Similarly, the second project shows a decrease in IS and SIR and increase in RS and PI which are 467 reflected in its MCP using obtained K value. Detailed simulation output is shown in Fig. 4. 468 [Insert Table 5 here] 469 Although an initial CP of 10 years was awarded to Project 1, the payback was achieved in almost 470 half of that time. According to the findings of this study, the post-feasibility studies conducted in 2015 471 revealed an interesting pattern of recovery according to which the project concession was overestimated. 472 Then, the CP was revised with mutual consent and handed back to NHA only in 6 years in 2015. The 473 calculations suggest a justified CP of 6.1 years as shown in Fig. 4. The red trend line refers to CP0 and 474 the blue line to MCP. The calculation only strengthens what was already known in this scenario. The key 475 factor leading to the success in this project was its substantial payback in the form of large toll volume 476 due to existing higher traffic. Further, owing to positive socio-political conditions, the market was 477 supportive. In addition, the smaller project size, the less complicated nature, brownfield construction and 478 easy access approach due to non-proximity with any urban area aided to the quick financial recovery. 479 Had a proper procedure been followed in the planning phase or a revision mechanism for CP existed, the 480 project could have been realistically planned to award a justified concession or revise it accordingly. 481 [Insert Fig. 4 here] 482 Project 2 was an extension of an existing route with tolling as the source of major income. The CP0 was 483 25 years based upon local experience. As per the findings, it seems that the project would have been 484 justified even if a CP of 24.5 years was committed, under the existing conditions. This reduction in 485 concession is associated with an average growth of 30% in the number of vehicles or transportation 486 means during years 2003 – 2005. No route promotion was required to attract new traffic as the route was 487 already well established. High tolling is possible during the years of concession. The companies involved 488 in construction have high expertise in field of highway construction. Based on this finding, it can be 489 21 recommended that if the above conditions prevail, the CP can be shortened. In the current form, the 490 concessionaire seems to have been incentivized for committing to the project. It must be appreciated but 491 not at the cost of opportunities for public revenue generation. This warrants dynamism in CP which is at 492 the core of this study. 493 Project 3 output suggests that based upon the current situation, the CP must be increased to 494 almost 20.5 years to incorporate the additional risks and decreasing RS. Keeping in mind that only three 495 years have passed since the start of construction, the situation seems very alarming and demands to keep 496 the concession in check, otherwise the project will need extension for proper financial recovery or may 497 fuel disputes among the key stakeholders. Among other factors, PI and MS balance each other because 498 of soon-to-be-constructed interchanges and, subsequently, more toll production. Therefore, there will be 499 no overall change in expected motorway usage. The main reason for the increase in CP can be linked to 500 the lack of coordination and communication between the stakeholders as quoted by the PM 501 representatives, and the increased risks of rework due to poor quality management and schedule crashing. 502 In a follow-up inspection, non-conformance was issued for a 6 km patch and reconstruction notice was 503 served. This validates the findings of the model and points to early corrective actions or otherwise 504 concession extension will be required. 505 Project 4 originally sought CP of 25 years. Due to risky nature of project because of lack of 506 proper tolling, prequalification technique was used for shortlisting the potential contractors. The PI is 507 supposed to follow the tolling stream mainly along with government subsidies since tolls are not expected 508 to entirely meet the contractor’s expenses. The total project cost was estimated at PKR 52 (USD $ 0.50) 509 billion. An amount of PKR 18 (USD $ 0.17) billion was committed by the Ministry of Planning and 510 Development under Viability Gap Funding (VGF) while PKR 5 (USD $ 0.048) billion will be lent by 511 the Ministry of Finance to NHA. The accumulated sum of PKR 23 (USD $ 0.22) billion will be provided 512 to FWO as financial assistance over the construction phase. The results indicate that the CP should be 513 increased to 27.04 years to incorporate the additional financial risks associated with the project due to 514 lack of tolling and required traffic. Furthermore, since it is a greenfield construction, project risk is high. 515 22 Thus, the originally determined CP is not feasible for the project and is recommended for revision based 516 upon the findings. 517 Project 5 is considered as an ill-fated and highly convicted project due to previous procurement 518 failures, serving as a bad example of project planning. Since the project could not start in last three 519 procurements, its cost has jumped from an originally estimated value of PKR 7 (US $ 0.067) billion in 520 2005 to latest value of PKR 36 (US $ 0.34) billion in 2015. The main reason for unsuccessful 521 procurement is lack of political will of prime stakeholders. The socio-political complexity in the form of 522 land acquisition has been one of the primary issues in this project. As a result, the government awarded 523 the project to FWO, which is a state-owned enterprise with similar success stories to its credit. The input 524 values were obtained from project experts in accordance with the previously mentioned experience 525 guidelines. The simulation results, shown in Fig. 4, point to an increased CP of over 28 years. This 526 increase in concession is directly associated with increased risks, deteriorating MS and lesser revenue 527 availability. The concessionaire confidence is also disturbed by the project history of previous failures 528 to start as reflected by the values assigned to each factor. In case these conditions prevail, the findings 529 advocate a concession extension up to 28.2 years for incorporating the increased risks and poor MS. 530 Discussion 531 The findings of simulated case projects show that considering all risk in concession estimation 532 is a challenging task due to data demands and analytical complexities. The proposed methodology 533 advocates revisiting the originally awarded concessions to accommodate complex emergent risk. This 534 points to a weak PPP system in the country which is primarily due to a recent focus to this delivery 535 mechanism. The system is infested with weak legal, contractual and methodical mechanisms. Since only 536 one PPP project has completed its concession, the possible documented repercussions cannot be 537 discussed with proper reference. This study highlights the upcoming problem of possible concession 538 extension which can cause disputes. Therefore, proactive management is required to get the projects back 539 on track by either developing contingency plans or increasing the concessions as highlighted by the 540 23 findings. The inability to do so may result not only in failure of specific projects but will paint a negative 541 picture of entire PPP procurement capability of the concerned authorities. Such negative image will 542 discourage the investors who may otherwise be agents of running the engine of economy and uplifting 543 the social standards by providing jobs and better infrastructure (Rahmani et al., 2017) . The proposed 544 model can be used for verifying an estimated concession and dynamically revising it at regular pre545 agreed intervals. This will help the local stakeholders in realistic concession assignment and subsequent 546 modification. Other market segments can modify the model according to their contextual needs and 547 priorities. 548 As mentioned, the revision mechanism along with incorporating the strategy offered by this 549 study may help rationalize the concession lengths. This may result into increased investor confidence in 550 two ways: working out holistic concession agreement that considers the locally focused CSFs and 551 ensuring timely recovery of capital as well as interest due to realistic assessment at planning stage and 552 reliable monitoring during project life. The elated confidence will boost the infrastructure procurement, 553 resulting in better and economic constructed facilities. 554 The agencies will also benefit in the form of better estimate of total project cost during the pre555 feasibility phase where the financial evaluation is performed due to country-specific factors that affect 556 private and public stakeholders. With this knowledge, public agencies may better perform the concession 557 negotiation and set the toll rules keeping in view not only the present economic conditions but also the 558 financial projections (Ullah and Thaheem, 2017). End users will benefit in the shape of nominal tariff 559 charges and improved quality of service as the probabilistic analysis, in the light of performance curves, 560 will dictate the maintenance and rehabilitation decisions ensuring top-quality road conditions resulting 561 in reduced travel time and improved ride quality. Thus, the models provide reliable estimation and 562 reassessment of CP at any stage within the project life cycle. 563 24 Conclusions 564 This study identified 59 CSFs that affect CP. The results of the survey of 26 industrial and 30 565 academic experts determined that the NPV, PI, RS, SIR, MS and IS are the most influential aspects with 566 a minimum of 8% influence by MS and IS, and a maximum of 29% by NPV. Further analysis shows that 567 NPV received the highest value which is in accordance with the published literature, and its association 568 with financial concerns and return, which is seemingly the top priority in PPP projects. 569 Additionally, a quantitative model based on the top influential factors was developed and served 570 as an input to develop a SD model to determine the CP under the influence of CSFs. Since the existing 571 models are not capable of meeting the increasing complexities and demands of PPP concessions and a 572 universal model may not be the best solution, the proposed model solves this dilemma through localized 573 CSFs. The proposed model was validated by simulating five local case studies of completed and ongoing 574 infrastructure projects with a minimum concession of 10 years to forecast the required concession under 575 current circumstances. 576 The case projects 3, 4 and 5 indicate that due to non-considerations of the identified CSFs, an 577 extension is necessary. The reasons for this anomaly extend from quality management to risky greenfield 578 construction, lack of promotions and interests of the public bodies in the form of local PPP policies. The 579 projects in initial stages of execution can still be controlled if the identified CSFs are incorporated, 580 otherwise extension will be the only way out as shown by the simulation results. Further, the conclusions 581 are strengthened by the findings of project 1 which clearly highlights an erroneous CP assessment and 582 had it not been the concessionaire presenting the idea of transfer, the project would have been a failure 583 based upon post completion assessments. Project 2 is the only project demanding concession reduction 584 pointing to its better assessment mainly due to the increased traffic and more usage of the route. 585 The key takeaway from this study is the introduction of dynamic CP modification model. Thus, 586 a CP can be estimated properly if localized CSFs used in this model are incorporated that may be changed 587 per local requirements. Even if due to some unforeseen reasons, an unrealistic CP is estimated, the model 588 31 Tieva, A., and Junnonen, J. M., (2009), "Proactive contracting in Finnish PPP projects". 728 International Journal of Strategic Property Management, 13(3), 219-228. 729 Trangkanont, S., and Charoenngam, C., (2014), "Critical failure factors of public-private 730 partnership low-cost housing program in Thailand". Engineering, Construction and 731 Architectural Management, 21(4), 421-443. 732 Trebilcock, M., and Rosenstock, M., (2015), "Infrastructure public–private partnerships in the 733 developing world: Lessons from recent experience". The Journal of Development 734 Studies, 51(4), 335-354. 735 Ullah, F., Ayub, B., Siddiqui, S. Q., and Thaheem, M. J., (2016), "A review of public-private 736 partnership: critical factors of concession period". Journal of Financial Management of 737 Property and Construction, 21(3). 738 Ullah, F., and Thaheem, M. J., (2017), "Concession period of public private partnership projects: 739 industry–academia gap analysis". International Journal of Construction Management, 740 1-12. 741 Ullah, F., Thaheem, M. J., Siddiqui, S. Q., and Khurshid, M. B., (2017a), "Influence of Six 742 Sigma on project success in construction industry of Pakistan". The TQM Journal, 29(2), 743 276-309. 744 Ullah, F., Thaheem, M. J., and Umar, M. (2017b), "PUBLIC-PRIVATE PARTNERSHIPS IN 745 PAKISTAN: A NASCENT EVOLUTION. In N. Mouraviev & N. Kakabadse (Eds.), 746 Public-Private Partnerships in Transitional Nations : Policy, Governance and Praxis 747 (Vol. 1, pp. 127-150), UK: Cambridge Scholars Publishing. 748 Wang, J. Y., Lindsey, R., and Yang, H., (2011), "Nonlinear pricing on private roads with 749 congestion and toll collection costs". Transportation Research Part B: Methodological, 750 45(1), 9-40. 751 32 Wang, Y., (2015), "Evolution of public–private partnership models in American toll road 752 development: Learning based on public institutions' risk management". International 753 Journal of Project Management, 33(3), 684-696. doi: 10.1016/j.ijproman.2014.10.006 754 Wibowo, A., and Wilhelm Alfen, H., (2013), "Fine-tuning the value and cost of capital of risky 755 PPP infrastructure projects". Engineering, Construction and Architectural Management, 756 20(4), 406-419. 757 Wibowo, A., and Wilhelm Alfen, H., (2014), "Identifying macro-environmental critical success 758 factors and key areas for improvement to promote public-private partnerships in 759 infrastructure: Indonesia's perspective". Engineering, Construction and Architectural 760 Management, 21(4), 383-402. 761 Xu, Y., Skibniewski, M. J., Zhang, Y., Chan, A. P., and Yeung, J. F., (2012), "Developing a 762 concession pricing model for PPP highway projects". International Journal of Strategic 763 Property Management, 16(2), 201-217. 764 Yu, C., and Lam, K. C., (2013), "A Decision Support System for the determination of concession 765 period length in transportation project under BOT contract". Automation in construction, 766 31, 114-127. 767 Zhang, X., and AbouRizk, S. M., (2006), "Determining a reasonable concession period for 768 private sector provision of public works and service". Canadian Journal of Civil 769 Engineering, 33(5), 622-631. 770 Zhang, S., Gao, Y., Feng, Z., and Sun, W. (2015), "PPP application in infrastructure 771 development in China: Institutional analysis and implications". International Journal of 772 Project Management, 33(3), 497-509. 773 774