Evaluating the decision-making on a Public-Private Partnership to fi nance a road project in Vietnam

Public-private partnership (PPP) has become a signifi cant alternative to traditional procurement in improving public projects in many countries since the 1990s. Along with the development of PPP projects, value for money (VFM) assessment is considered as an eff ective tool to support PPP decision-making process, which is widely used by many governments. Like several countries, Vietnam has applied PPP to develop road transport infrastructure. However, the government of Vietnam has never conducted VFM assessment to determine whether using PPP instead of a conventional delivery is better choice to fi nance a proposed project. Th is paper proposes application of the value for money assessment to evaluate the suitability of PPP model for a given project. A case study of My Loi project in Vietnam is used to examine the viability of the method. Th e results of the research using Monte Carlo simulation methodology demonstrate that there is a 23.4 percent chance that PPP could be a good candidate to implement the My Loi project. Sensitivity analysis reveals that the outcome of VFM is the most sensitive to the toll of Public sector comparator (PSC) and the least sensitive to infl ation.


INTRODUCTION
One of the most serious problems Vietnam is facing today is the road system.It is suff ering from increasing demand while the quality of roads is decreasing.According to the General Statistic Offi ce of Vietnam (2016), the volume of road transport passengers in 2001 was more than 500 million, this fi gure has increased rapidly during the past decade and almost peaked 3 bln passengers in 2013.On the other hand, according to the Global Competitiveness Report (2015), the quality of roads in Vietnam ranks 104 th out of 144 countries in the world."Given high and rapidly increasing demand for infrastructure services, and expectations of lower levels of ODA in the near future, mobilizing private sector investment is necessary to fi ll the fi nancing gap.In this sense, PPPs are recognized as one of the most eff ective ways of fi nancing infrastructure development" (World Bank, 2013)."Using a public-private partnership could increase the money available for highway projects and complete the work more quickly or at a lower cost than is possible through the traditional method.Specifi cally, such a partnership could secure fi nancing for a project through private sources that might require more accountability and could assign greater responsibility to private fi rms for carrying out the work" (Kile, 2014).
Th e issue is posed whether decision-making to pursue PPP instead of a conventional delivery is a good choice for fi nancing this project.How to evaluate this decision-making?
According to D. Morallos and Amekudzi (2008), "value for money is one of the leading tools available to public managers to assess the value of pursuing a project through a public-private partnership versus traditional procurement".In a similar way, C.O. Cruz and R.C. Marques (2013) advocated that "PPPs or traditional procurement arrangements are possible alternatives to provide that same output.Because decision makers have diff erent models to ensure the provisions of the infrastructure and/or service, it is necessary to compare both models and select the best one".Th e National Council for Public-Private Partnerships (2012) claimed that "one way of assessing the potential benefi ts associated with a PPP is through VFM analysis, which compares the cost of PPP-based provision to that of traditional projects delivery, providing decision makers with a quantitative tool and data to help them make the case for selecting the most appropriate mode of project delivery".
Th ere has been more than 23 years since Vietnamese government fi rstly stimulated private participation to invest in road projects.Since then, Vietnam has witnessed an increase in the number of PPP projects, which contribute to reducing state budget burden as well as develop economics in the country.For example, 62 road projects have been implemented under PPP model with the total investment of VND186,600 billion.However, according to Deloitte (2015), such the increase in PPP projects does not mean that PPP is always the most suitable choice.In fact, some PPP road projects in Vietnam are facing failures and thus there are debates on the selection of the best option.One of such projects is My Loi Bridge.Some experts in this fi eld believed that PPP model is suitable for this project.On the other hand, others confi rmed that conventional delivery is a better choice.Unfortunately, until now there has been no evidence to clarify this problem.Th is research applies VFM assessment to examine whether PPP or conventional delivery is better to fi nance this particular project.
Th e paper comprises of four parts, it begins with a review of VFM.In addition, the paper also concentrates on the research methodology.Th en, this paper describes the characteristics and the existing problems related to the selection of the best option for the case study applied.Next, in the fi ndings of the research, this paper discusses whether PPP could bring better value as compared with traditional delivery for this case study.Finally, limitation of this research is mentioned, the major shortcoming of this research is the lack of analysis from the qualitative viewpoint.In order to deal with this limitation, suggestions for future work is provided.

LITERATURE REVIEW
Defi nitions of VFM Contreras (2014) claimed that "VFM is the optimum combination of whole-of-life costs and quality to meet requirement, which does not always mean choosing the lowest cost bid.VFM not only measures the cost, but also takes into account the quality and fi tness for purposes, in determining whether goods and services represent good value".Zhen Hu and Shu Chen (2014) found that "Value for money (VFM) is a good measure of the performance of PPP projects, which represents the philosophy of the public sector in partnering with the private sector in delivering public works and services through contract-based PPPs".Once again, Takim, R., et al (2009) confi rmed that "VFM is not the lowest cost option but an understanding of the whole life benefi ts and appropriate risk allocation between public and private sectors".Generally, although there are many diff erent defi nitions of the term "value for money", most scholars agree that value for money is a combination of both quantitative and qualitative assessment in order to determine whether a PPP will save money and bring more effi ciency for public sector compared to a conventional procurement to fi nance a particular project.Sarmanto, (2010) advocated that "there are usually two components of value for money: a quantitative one (including all factors can be measured by the public sector comparator), and a qualitative one (aspects that cannot be quantifi ed)".It is clearly now, theoretically, in order to have a comprehensive assessment, one should conduct both quantitative and qualitative approaches.However, in fact, quantitative VFM has been more popular and widely adopted in PPP decision-making process in many countries compared to the qualitative one.Th is is because that this approach is more transparent and confi dent.In particular, quantitative assessment is based on economic indicators, which can be quantifi ed.Th us, it allows us to know exactly the estimated costs of a proposed PPP project with those of conventional delivery.Meanwhile, qualitative analysis is usually based on the subjective opinions of stakeholders, which could not measure the value of advantages of PPP model compared with the traditional procurement.
In terms of quantitative VFM, World Bank (2013) defi ned quantitative VFM as "involves comparing the value for money of a proposed PPP (or actual bids received) with a "Public Sector Comparator" (PSC)-that is, a model of the project if implemented through traditional public procurement".Likewise, according to D. Morallos and Amekudzi ( 2008) "the quantitative VFM component includes all project factors that can be valued in monetary terms.It features a methodology that compares the PPP bid with a hypothetical scenario called the public-sector comparator (PSC)".Additionally, Tsukada (2015) point out that quantitative VFM is "used to determine in advance how much money the government would eventually pay to the private sector as compared with a publicly provided project".Moreover, D. Tsamboulas et al. (2013) confi rmed that quantitative VFM "is used to compare the required public funds for two cases: the fi rst case refers to the implementation of the project with PPP, while the second case to the delivery of the project by public sector".As can be seen in Figure 1, PSC represents the cost of a project carried out by the public sector, which includes raw PSC, competitive neutrality, transferable risk and retained risk.Meanwhile, PPP bid represents the cost of a project if conducted under a PPP, which combines retained risk and service payments.If VFM is larger than zero, a project should be administrated by private sectors.Otherwise, the project should be incurred by public sector.

METHODOLOGY
Th e research develops Monte Carlo simulation (MCs) to summarize the simulation of the VFM output.In addition, an application of sensitivity analysis helps to identify and compute eff ects of the uncertain input variables on decision-making to pursue PPP model.Th ere are four stages conducted in the simulation model of this research: (1) Determination of input variables (2) Identifi cation of distributions for the input variables and running the simulation for them (3) Estimation of cash fl ow of the PSC and PPP bid (4) Generating distribution of the VFM output (see Figure 2).

Transferred risk variables
Transferred risks means "risks are changed from public sector to the selected private investors" (Infrastructure Ontario, 2007).In other words, if the project is conducted under PPP, the value of risks cost that is expected to be transferred to the private sector.In this research, transferred risk variables are associated with the construction and traffi c volume risk.

Endogenous variables
Th e construction cost, maintenance cost, management cost and toll rates are decided during negotiation process before contract is awarded.Th ey are decision variables or called endogenous variables.Th e general management cost comprises of three components, including the periodic maintenance cost, special maintenance cost and upgrade cost.

Exogenous variables
Th e interest rate of government bond, and discount rate for cash fl ow of PSC mentioned in the case study is defi ned as exogenous variables for the simulation model.Th is is because these variables depend on the country ' s economy.Regarding the interest rate of government bond, this research uses actual data of interest rate of fi fteen-year government bond in Vietnam during the period from 2005 to 2015.With respect the discount rate, this research uses risk-free rate.
On the other hand, according to Malini, (1997), "developed economies strive to contain their infl ation rates to modest levels of <5% per annum, while other countries suff er severe infl ation of a high magnitude, refl ecting a higher level of risk associated with the investment".Th erefore, besides discount rate and interest rate, infl ation is considered as uncertain input variable in the model.Based on the historical infl ation data from 1994 to 2015 in Vietnam, we forecast time series of infl ation in the future using @ Risk software version 7.5, as followings chart: Th e Y-axis characterizes the infl ation rate, and the X-axis represents the past time and future time compared to the base year (the year 2016).As shown in the fi gure, the green path illustrated the infl ation rate in Vietnam during the historical period (from 1994 to 2015), and the red line represents the forecasted infl ation rate in the future.

Step 2: Defi nition of probability distribution for all input variables
Identifying the probability distribution of input variables is very important in the application of Monte Carlo simulation.According to Park K. A., (2007), "if there are proper historical observations, the probabilistic distribution can be defi ned from these data, while if these are absent then this distribution can be assumed by personal or organizational intuitions and decisions".Likewise, C.O. Cruz and R.C. Marques (2012) emphasized that "defi nition of distributions for cost components should be based on historical data or experts' judgment."Furthermore, according to F. Belaid and D. De Wolf (2009), standard statistical distribution such as normal, the log-normal, the uniform and triangular distributions are often used to describe the input parameters.
To run MCs simulation, each variable is assigned the best probability distribution.However, for the transferred risk and exogenous variables, the available data is not suffi cient to determine the distribution.Hence, Bootstrap method is applied to respond to this issue.Alternatively, regarding the endogenous variables, due to lack of relevant data, the probability distribution for these variables is determined subjectively from the previous studies.

Step 3: Estimation of cash fl ows of PSC and PPP bid
For the transferred risk and is determined based on the historical version 7.5.Besides toll, infl ow of SBP could achieve from subsidy from the government.Additionally, cash outfl ow of the PSC as well as the PPP is calculated based on the construction cost during the years of construction, maintenance cost, management cost during the operation phase, and fi nancing cost during the concession period.
Step 4: Simulation Output of VFM Th e simulation output of the model is net present value of the VFM, which is in the form of stochastic variables related to probability distributions.Th e result of the simulations provides the likelihood of the VFM being positive or negative.Apart from this, advanced sensitivity analysis facilities to evaluate how input variables have impact on the VFM when there is interaction of other variables in the model.

CASE STUDY
My Loi Bridge is a project that was initially designed to be carried out under public sector procurement.Th e building of the project started in 2009 by the government direct investment.Th e original investment capital of VND 1,438 billion was funded by issuing government bond.According to the original plan, the project would have been fi nished in 2012.However, since traditional delivery was suggested to implement the project, the construction of the project was not yet completed until 2013.At that time, the Vietnamese government decided to encourage private sector to fi nance the My Loi Bridge.However, according to some PPP experts of Vietnam, the project could be better if implemented by public sector.Th e question is raised that whether the project should be conducted by the government direct investment as an original suggestion or procured by application of PPP model.

FINDINGS AND DISCUSSION
Basic assumption for assessment

* Raw cost in PSC
In order to compute the cost of the project implemented by public sector, the capital cost items of the PSC is extracted from the feasible report of the My Loi project (including design and construction contract price, administration cost, land acquisition and contingency cost for construction).Other costs including operation cost and maintenance cost, which are not available from historical data, are estimated applying the cost estimate standard of the Vietnamese government.For example, regarding the operation cost of the PSC, annual costs are estimated by multiplying the 15 percent (regulated on the Circular 90/2004/TT-BTC of Ministry of Finance dated 2004) and the revenue of the project in the fi rst year of operation.

* Discount rates
Th is research assumes that the discount rate for cash fl ow of PSC is risk-free rate.According to Arrow & Lind1 (1970) "the discount rate to be used in public investments should be the "risk-free" rate refl ecting the risk-neutrality of the public sector", since "government can spread the risks among all taxpayers when it undertakes an investment" (Checherita, 2009).Otherwise, for private projects, investors mobilize capital from equity and borrowing as so the discount rate for cash fl ow of SBP should be weighted average cost of capital (WACC).As shown in Table 1, under PPP model, the total life cycle cost of the project is expected to be VND 6,597.81 billion, while under traditional government procurement the cost of the project is predicted to be VND 3,060.40 billion.Th e comparison of the two net present costs proves a diff erential VFM of VND 273.35 billion in favor of the PPP scheme.Th us, PPP model is expected to be more benefi cial to provide the My Loi project.

Monte Carlo simulation
Th e variation of the VFM of the project is estimated by applying the MCs.Th e frequency distribution of the VFM by stochastic change of input variables (see Figure 4) is utilized to generate the project variation created as an output of the risk simulations.After running 10,000 trials, the Monte Carlo simulation results reveal that mean of VFM is VND -567.44 billion and standard deviation is VND 700 million.Further, the probability value of quantitative VFM distributes between VND -1,939 billion and VND 479 billion is 90%.Additionally, as represented in fi gure 5, the probability of VFM being positive is 23.4%.In the other words, there is a 23.4% chance that application of PPP model to fi nance the project could produce better value for the government than traditional procurement.It should be mentioned that there is a diff erence on the result of VFM between single value estimation and the MCs.Given the result of single value calculation, the value of VFM is VND 273.35 billion (see Table 1), while based on the results of MCs the expected value of VFM is VND -567.44 billion (see Figure 4).In other words, according to the result of single value estimation, PPP scheme is more appropriate than public sector procurement.Meanwhile, according to the results of MCs, public sector procurement is more suitable than PPP model.Th e reason leading to this diff erence may be due to that the input variables estimated by single value estimation method which is diff erent from the mean value by MCs.Especially, the toll of vehicle in the single-value estimations is VND 15,000 per PCU, while MCs creates VND 32,5000 per PCU as an expected value.Likewise, forecasted infl ation rate based on the single-value estimation is 6%, while the historical data creates around 7% as forecast trend.Th ese can make quantitative VFM indicator by MCs lower, compared to the result of single value estimation method.Figure 6 summarizes the distributions generated for the value of VFM range, which helps to know how the value of VFM of the My Loi project is changed during the period from 2014 to 2045.Th e X-axis describes the concession term of the project from 2014 to 2045.Th e Y-axis represents the net present value of the VFM.As can be seen, positive VFM achieved in the fi rst few years, however, in the subsequent years, cash fl ow of the VFM becomes negative.In other words, PPP model can save larger cost to the government than public sector procurement in the construction phase of the project, but public sector can do better than private sector in the operation phase.

Sensitivity analysis
Figure 7 represents the sensitivity tornado graph of the VFM.Th e X-axis characterizes the variation in the outcome value.Th e Y-axis characterizes each input variable.Th e longer the bar, the more sensitivity VFM is to input variables.Th e critical variable has an important eff ect on the evaluation outcome of VFM is PSC toll, while variables with the least eff ect on VFM are infl ation in the year 2020, 2024 and 2025.Especially, when standard deviation of toll increases +1, the standard deviation of VFM will decrease 0.69, equivalent to VND 521.89 billion (see Figure 8).In contrast, a one standard deviation increase in infl ation in the year 2020 makes VFM of the project reduce by 0.10 standard deviation, which respondents to VND 76.470 billion.It implies that in order to improve VFM, one should mostly concentrate on and adjust the toll of the PSC.
Once again, the sensitivity tornado graph confi rms that the positive elements have a infl uence on the VFM of the project are design and construction cost in year 2014, interest rate of government bond, cost overrun risk and discount rate.Th is may be explained that the increases in one of these indicators leads to the increase in value of PSC, therefore, VFM raises.Meanwhile, the inputs have a negative infl uence in the project ' s VFM are infl ation rate, toll and revenue risk.Th is is due to the rise of infl ation produces the increase in PSC infl ow is bigger than the PSC outfl ow, so it leads to the fall down in VFM.Likewise, the increases in the toll of PSC makes the cash fl ow of PSC reduce, leading to the decrease in VFM.It could be concluded that additional construction cost of PSC, interest rate of government bond, cost overrun risk or discount rate is likely to increase the potential viability of PPP model.Otherwise, with a lower revenue risk, infl ation rate or toll is, the higher the viability of PPP model is.

CONCLUSION
PPP model seems not to be a good candidate to implement the My Loi project.Especially, the results of MTCs reveal that the probability that PPP scheme is better than traditional public delivery is 23.4%.It implies that the Vietnamese government's decision-making to stimulate private participation in this project may not be appropriate.
Th eoretically, a PPP project plan is approved after evaluating feasibility and testing value for money.However, in the current context of Vietnam, VFM assessment of a PPP project has not considered and conducted.Th us, the paper proposed the VFM assessment for evaluating the suitability of a specifi c PPP project in Vietnam.Th is method is expected to be a path for further application in Vietnam when choosing PPP model to fi nance a proposed project.In addition, My Loi project is a good practiced case to explore application of the VFM assessment to select the best option between PPP model and traditional government procurement.
However, it should be mentioned that the potential limitation of this paper is lack of comprehensive evaluation of a PPP project in general.Th is research ignores consideration of the standpoint of qualitative PPP decision-making to procure a particular project.On the other hand, insuffi cient available data regarding case studies does not allow us to evaluate whether that PPP model could provide higher VFM in general road projects in this paper.However, in respond to the fi rst issue, future works should employ Analytic Hierarchy Process.Regarding the latter, the method of Bootstrap is recommended to apply in the future research.

Figure 3 :
Figure 3: Time series of infl ation rateSource: Analyzed by author.

Figure 4 :
Figure 4: Distribution for the project ' s value for money Source: Analyzed by author.

Figure 5 :Figure 6 :
Figure 5: Accumulative probability of positive VFM Source: Analyzed by author

Figure 7 :
Figure 7: Sensitivity tornado graph for the project ' s value for money Source: Analyzed by author.

Figure 8 :
Figure 8: Sensitivity tornado diagram for the project ' s value for money Source: Analyzed by author.

Table 1
Comparison of the PSC and PPP bidUnit: billion VND*