Market-led proposals

Tabled: 27 November 2019

3 West Gate Tunnel project: value for money

The 2015 interim MLP guideline required VFM assessments for MLPs to include:

  • a quantitative assessment involving a comparison of the final proposal's cost to a state benchmark, either a public sector comparator or a realistic alternative, and
  • a qualitative assessment of the value of the proposal as a whole and the individual value drivers underpinning the proposal.

In December 2017, DTF advised the government, based on the Stage 4 assessment, that the Transurban WGT proposal offered VFM based on a comparison with state benchmarks for the same project funding sources.

The total project cost is $6 689 million (nominal) including the Monash Freeway upgrade and the Webb Dock access improvement. Works will be funded by:

  • new tolls on users of the project, including heavy commercial vehicle tolls on the upgraded West Gate Freeway and car, light commercial vehicles and motorcycle tolls on WGT
  • adjustments to various CityLink tolls during the remaining term of Transurban's existing CityLink concession, including:
    • a fixed 4.25 per cent annual toll escalation rate for 10 years from July 2019
    • a 10-year extension of the CityLink concession, from 2035 to 2045
  • a state funding contribution of $2 661 million (nominal).

More than 93 per cent of Transurban's design and construction-related costs were tendered out to the market in a competitive process. Transurban will transfer the WGT back to the state in January 2045.

3.1 Conclusion

DTF's VFM assessment and advice to the government met the 2015 interim MLP guideline requirement for a quantitative assessment of the Transurban WGT proposal.

DTF advised the government that Transurban's proposal met the requirements for VFM. However, DTF's advice to the government was not sufficiently comprehensive due to:

  • limited analysis of alternative funding and delivery options
  • state VFM benchmarks used to compare against the Transurban proposal being expressed as a wide range without inclusion of a best estimate
  • lack of sensitivity analyses to challenge discount rate assumptions in the state's VFM assessment.

In addition, DTF's presentation to the government of the VFM assessment results on Transurban's proposal did not clearly disclose the sensitivity of key assumptions and calculations which underpinned the reliability of the assessment.

3.2 Adequacy of the VFM assessment

Analysis of funding and delivery options

Transurban's proposal offered an apparent benefit to the state because Transurban intended to finance most of the project delivery itself, primarily through CityLink tolls—the unique characteristic the proposal offered. This would remove the need for the state alone to fund the build. Assessing the value of Transurban's proposal therefore required a comparison with an estimate of what it would cost if the state delivered the project or used other possible funding and delivery approaches.

Assessment and analysis process

DTF committed to the government to comprehensively assess whether Transurban's proposed approach to funding and financing the project was the best option available to the state, or whether an alternate funding and contracting approach would deliver better VFM. This analysis was to include approaches other than granting a concession extension to Transurban such as the state:

  • borrowing through securitisation of future CityLink tolls
  • monetising future CityLink tolls through a competitive process.

However, DTF did not meet this commitment. Neither the Stage 3 or Stage 4 reports or advice to the government provide this broader assessment. DTF's VFM assessments only compared the Transurban proposal to state benchmarks using the same funding sources. This narrow assessment approach deprived the government of critical information to support its decisions on whether, and how, to progress the proposal.

Exploration of alternative funding sources and delivery options in the business case and VFM assessments

DTF's Investment Lifecycle and HVHR 'Prove' guidelines on the development of business cases state that alternative funding sources should be considered when assessing commercial and financial issues for projects.

Given this, we expected that DEDJTR's business case for the project would include assessment of a wider set of alternative project funding options and issues such as the difference between the costs and benefits of raising funds from tolls on CityLink compared to other state sources of funding such as taxes and charges, borrowing or other monopoly concessions.

This was not the case. DEDJTR's business case included a DTF report on the approach to tolling. However, this analysis was restricted to a narrow set of toll pricing alternatives, not broader options associated with alternative mechanisms for funding the project.

We expected the VFM assessments to include assessments:

  • of the risk associated with entrenching Transurban's incumbency advantage in bidding for future toll roads
  • combining qualitative and quantitative analysis of the incremental costs and benefits to the state of accepting Transurban's proposal.

At Stage 2, DTF noted that the government should consider whether extending an existing concession would create an unintended monopoly advantage for Transurban in the construction and operation of toll roads. However, DTF did not, in Stages 3 or 4, advise government on whether extending the CityLink concession would create an unintended monopoly advantage for Transurban.

By accepting Transurban's proposal, the state effectively chose not to exercise the other options available to it. The business case explored whether the state should do nothing or whether a similar package of works would be of value without Transurban involvement. However, it did not examine whether the state would be better off accepting the Transurban proposal, undertaking the works on its own or seeking private sector involvement in the marketplace. This was the intended role of the VFM assessment.

The quantitative VFM assessments at Stages 3 and 4 compared Transurban's proposal to a state VFM benchmark intended to represent the costs to the state of undertaking the project without Transurban, but assuming the state could access all the same funding sources. This approach was reasonable.

However, the uniqueness and VFM assessments and advice to the government did not clearly provide the intended broader qualitative VFM assessment of the benefits and costs, including opportunity costs, of either pursuing or not pursuing the Transurban proposal.

An Availability PPP is where the primary funding source that repays the private sector finance used to build the asset takes the form of a payment from government over the operational phase of the project to ensure the continuing availability of the asset.

Such an assessment would have examined the incremental benefits to the state of pursuing the Transurban proposal against alternative options of state or private sector delivery. The benefits to the state under the Transurban proposal included Transurban's capacity to fund the project by accessing CityLink toll revenues at low cost, reduced costs to manage interface issues between the project and Transurban's CityLink road and avoiding costs to set up a tolling operation. The opportunity costs of accepting Transurban's proposal would include costs associated with potentially entrenching Transurban's incumbency advantage in bidding for future toll roads.

The individual costs and benefits of accepting Transurban's proposal were, for the most part, articulated either qualitatively or quantitatively in the VFM assessments. However, these strands of analysis were not brought together in a meaningful way in the VFM assessments or advice to the government, meaning it would have been difficult to assess the benefits and costs, including opportunity costs, to the state of either pursuing or not pursuing the Transurban proposal.

Reliability of VFM benchmark inputs

The Stages 3 and 4 VFM assessments included a quantitative comparison of the Transurban proposal against state benchmarks for project costs and revenues. The VFM assessment reports refer to these benchmarks as the state-owned enterprise (SOE) or SOE delivery model benchmark. We refer to this as the 'state VFM benchmark'.

State VFM benchmark (SOE)

Assumed that the state …

Project costs for construction and operation

Independently delivered the same project and scope proposed by Transurban under an Availability PPP delivery model.

Project funding solution

Collected and retained the same toll revenue streams assumed under the Transurban proposal, and that those revenues delivered a commercial rate of return.

Source: VAGO.

The state's VFM assessment involved discounting the estimated nominal cashflows for project costs and revenues using a range of discount rates intended to reflect a market-based WACC valuation range. This gave a range of state VFM benchmark results.

Discount rates used to reflect market-based WACC valuation range

Determined by an …

Intended to reflect …

Resulted in …

Used to compare with …

Investment bank engaged by DTF.

the state's requirement for a commercial rate of return given the commercial risk position assumed under the SOE delivery model.

a range of values for the state VFM benchmarks for each cost and revenue element in NPV terms.

corresponding elements in Transurban's proposal and presented as overall results in the VFM assessment reports.

Source: VAGO.

WACC is the rate that a company is expected to pay on average to all its security holders to finance its assets. WACC is commonly referred to as the cost of capital.

Estimated toll revenue and discount rate assumptions were the most important drivers of the overall determinants of the VFM assessment results presented to government.

Toll revenue estimates

The state's VFM assessment used benchmarks which included the state's estimates of the toll revenues it could raise from the WGT, the West Gate Freeway, and the CityLink escalation and extension sources.

A large proportion of the reported VFM benefits to the state in Transurban's proposal arose from the differences between Transurban's and the state's valuation of future tolling revenue. At Stage 4, the state's lower CityLink toll escalation revenue estimate was the major contributor to the overall difference of $272 million (NPV) between the Transurban proposal and the midpoint of the state's VFM benchmark. This was because the low estimate had the effect of dropping the bottom end of the benchmark range.

This difference arose because Transurban valued tolling revenues, particularly toll escalation revenue, more highly than the state did. The difference between the valuations reduced the size of the state contribution needed under Transurban's proposal. DTF rightly advised us that Transurban was prepared to accept the risk involved in achieving its higher toll revenue estimates. Transurban also faced a real risk that the state would not proceed with the proposal if the state assessed Transurban's forecast tolling revenues as unrealistic and not providing VFM against state benchmarks.

Notwithstanding this, DTF rightly obtained its own estimates of revenues to develop its VFM benchmark. However, there were clear reasons for the state to be sceptical if its own toll revenue estimates were less than Transurban's because:

  • as noted by DTF in its Stage 3 and 4 assessment reports, Transurban had greater knowledge of CityLink revenues and their drivers above all other market participants
  • Transurban's better knowledge means its estimates may have been more accurate than the state's
  • Transurban had commercial incentive to undervalue future CityLink toll revenues in its proposal to make a case for a longer extension of the CityLink concession, higher toll escalation and a larger state contribution.

Given this, DTF should have been wary of significant differences between Transurban's and the state's tolling revenue valuations and the potential impact of those differences on the size of the state contribution and the extension and escalation of the CityLink concession.

Given that Transurban had little incentive to overstate toll revenue, and had the best information, where Transurban's estimates were higher than the state estimates, DTF's assessment and advice would have been more comprehensive had it used the Transurban estimates as a sensitivity test for its VFM benchmarks and shared the results with the government.

Estimation of CityLink escalation revenue in Stage 4 assessment

The state's forecast 'low traffic scenario' benchmark for the CityLink escalation revenue was significantly below Transurban's forecast. DTF's VFM assessments counted Transurban's higher estimate as a material source of VFM for the state. However, Transurban's knowledge about this revenue source raises questions about the reliability of the state's low estimate.

At Stage 4 both the state benchmark and the final Transurban proposal used the same toll price assumptions. Given this, differences in forecast toll revenue were driven by differences in the assumed traffic demand and/or the discount rate applied.

Figure 3A shows that the Stage 4 VFM assessment included a single point estimate for the state's benchmarks for all revenue sources, except CityLink escalation revenue, where DTF reported both a high and low benchmark.

The difference between the state's high and low CityLink escalation revenue benchmarks arose from an assumption that around 6 to 7 per cent of traffic would be diverted under a 'low traffic scenario'.

Figure 3A
Nominal project revenues as reported in the Stage 4 assessment

Project

Transurban proposal ($m)

State benchmark ($m)

WGT

1 854

2 163

West Gate Freeway

3 010

3 109

City Access

137

149

CityLink Extension

14 345

15 061

CityLink Escalation

4 758

2 893–4 788

CityLink Traffic Impact (during operation)

(210)

(300)

Total

23 894

23 075–24 970

Source: VAGO based on information from DTF.

DTF advice to the government on the Stage 4 assessment did not:

  • highlight that the state benchmark's forecast 'low traffic scenario' for the CityLink escalation revenue was 65 per cent lower than Transurban's
  • discuss the possibility that the state benchmarks undervalued the CityLink escalation revenue
  • explore the suitability of the assumption that resulted in 6 to 7 per cent of traffic being diverted under the 'low traffic scenario'.

The differences in revenue assumptions significantly impacted the calculation of the overall project NPV. The state's low CityLink escalation revenue estimate had a material effect as it was used to estimate the Stage 4 'low' state benchmark, dropping the bottom end of the benchmark range.

This was further exacerbated by DTF reversing the discount rates applied to this revenue source, without explanation, in contrast to those applied to all other revenue estimates. In the Stage 4 assessment, for revenue sources other than the CityLink escalation revenue:

  • the 'low' state benchmark was derived by applying a high discount to future revenues and costs
  • the 'high' state benchmark was derived by applying a low discount rate to future revenues and costs.

Conversely, for the CityLink escalation revenue:

  • the 'low' state benchmark was based on a low traffic scenario and a low discount rate
  • the 'high' state benchmark was based on a high traffic scenario and a high discount rate.

Applying the discount rates this way, in the state benchmark, placed the Transurban proposal as offering comparatively better VFM (all else remaining equal). DTF did not explain, in the Stage 4 VFM assessment or related advice to the government, the nature, impact or basis for this approach. This was an omission given how sensitive the state VFM benchmark was to the CityLink escalation revenue 'low' traffic scenario.

DTF advised us that its approach to combining the range for CityLink escalation toll revenues with the range for the state's discount rates was intended to reflect an inherent link between the risk profile of the two revenue forecasts and the discount rate applied to each, suggesting that the high escalation revenue estimate was considered less realistic. If that were the case, it should have been clearly explained in the Stage 4 assessment report, but it was not.

DTF's explanations to us on this issue suggest that the treatment of the discount rates in relation to the CityLink escalation toll revenues was their way of seeking to address concerns about the plausibility of assumptions about traffic diversion that underpinned each of those forecasts. If that was the case, there would have been better approaches for addressing those concerns.

The discount rate should reflect the variability of the cash flows that are being discounted—not the degree of doubt about the plausibility of those cash flows.

If DTF was attempting to account for the high CityLink escalation revenue forecast scenario being less realistic than the low scenario, the appropriate way to address that concern would have been to probability-weight the scenarios, according to the likelihood of each of those scenarios arising. DTF argued that adopting a probability-weighted approach to deriving a best estimate of toll revenues would have introduced unnecessary subjectivity into the forecasts. However, it would have been far better to express those judgments transparently, so that the impact of those assumptions about likelihood could be scrutinised and tested.

Figure 3B shows the combined effect on the estimated range for the state's VFM benchmark of:

  • removing the low traffic scenario CityLink escalation revenue estimate
  • applying discount rates to the CityLink escalation revenue consistent with those applied to other sources of revenue.

Figure 3B
Comparison of state contributions under Transurban proposal, the state benchmark from the Stage 4 assessment and our re-estimated state benchmark ($ million NPV)

Figure 3B shows comparison of state contributions under Transurban proposal, the state benchmark from the Stage 4 assessment and our re-estimated state benchmark ($ million NPV)

Note: The middle bar represents the state VFM benchmark range as given in the Stage 4 VFM assessment. The right bar is our re-estimated benchmark range based on our adjustments.
Source: VAGO based on the DTF Stage 4 assessment report and our re-calculations.

The impact of these adjustments is that the state contribution, represented in Figure 3B by dots, under the Transurban proposal of $1.96 billion (NPV) is no longer near the top of the state benchmark but closer to the mid-point.

Discount rate issues and impacts on state VFM benchmark

The Stage 4 VFM assessment and advice to the government showed the state benchmark as a wide NPV range—positive $55 million to negative $600 million—largely because of the range of discount rates used.

DTF engaged an investment bank with experience in the financing of toll roads, including the original CityLink transaction for Transurban in the mid-1990s, to provide advice on the discount rates in the state's VFM assessment.

Relevant guidance

The MLP guideline does not advise how to determine discount rates. Therefore, it would have been necessary for the state to consider the appropriate methodology for determining discount rates for this MLP.

The state benchmark for the WGT project assumed an availability PPP to deliver the infrastructure and a state-owned entity operating the road and collecting tolls. This involved the state accepting traffic demand risk for the WGT and receiving the extended CityLink tolling concession. The state's commercial advisers used discount rates in the Stage 4 VFM assessment that were intended to reflect commercial rates of return given the commercial risk position taken on by the state under this model.

The National PPP Guidelines on discount rates are relevant. DTF advised that this project is akin to an economic infrastructure project, rather than a social infrastructure project and, therefore, the relevant methodology for determining discount rates is established in Appendix D of Volume 5 of the National PPP Guidelines, which indicates that:

Public Sector Comparator

The National PPP Guidelines define the PSC as an estimate of the hypothetical, risk‑adjusted whole-of-life cost of a public sector project if delivered by government.

The PSC is used as the financial benchmark in the quantitative VFM assessment during the PPP procurement process.

  • there is no valid reason that the state should seek a lower net return on its participation in economic projects than that sought by private sector participants in the same market. This means that when evaluating bids in a PPP process for economic infrastructure the Public Sector Comparator (PSC) should ordinarily be developed as an asset purchase model and the decision structure based on selecting the best value investment
  • the PSC model should be developed using observed returns private sector entities are seeking for similar projects to develop the model. This will naturally include a market premium for systematic risk, since economic infrastructure cashflows are subject to such risks
  • the PPP bids will include either a payment to or from the state, independent of the actual future revenue experience, and as such are devoid of systematic and project risk from the government perspective and so should be discounted by the risk-free rate. There is no need to adjust this rate since the PSC directly values the cost of systematic risk and therefore there is no differential between valuation approaches as there is in social infrastructure projects.

The state's investment bank adviser on discount rates used a 'bottom-up' financial model to derive the set of discount rates used in the VFM analysis. This financial model used inputs including equity internal rates of return that were benchmarked using observed rates of return for actual toll road projects. This approach was consistent with the broad requirements of the relevant part of the National PPP Guidelines.

Reasonableness of the discount rate range

The investment bank engaged by DTF determined a range for the rates of return on capital used in the state VFM benchmark. These rates were intended to reflect the state's need for a commercial rate of return under the SOE model assumed as part of the state's VFM benchmark.

Discount rates as ranges

DTF's commercial adviser discounted a single set of nominal cashflows using a 'high' discount rate and a 'low' discount rate provided by the investment bank DTF engaged. This determined the range for the state's benchmarks.

DTF treated forecast cashflows and discount rates inconsistently in the state's VFM assessment.

The VFM financial model assessed the NPVs of cashflows from June 2017 to March 2045. This required forecasting cashflows 29 years into the future. The level of uncertainty of these cashflow forecasts is at least as significant as the uncertainty of discount rate assumptions.

Despite this uncertainty, DTF's commercial adviser formed a best view of future cashflows to be discounted and used a single scenario for forecast future cashflows. The one exception to this was CityLink escalation revenues, used to estimate the NPV of the project for the purposes of establishing the state's VFM benchmarks.

The investment bank engaged by DTF to advise on the appropriate discount rate did not provide a single point best estimate for the discount rate. Figure 3C presents the nominal discount rates used in the Transurban proposal and in the state's Stage 4 VFM assessment benchmark analysis.

Figure 3C
Discount rates used in the State Benchmark (per cent)

Figure 3C shows discount rates used in the State Benchmark (per cent)

Source: VAGO based on information from DTF.

It is not unusual to express discount rates as ranges because discount rates (estimated as the WACC) are difficult to estimate with a high degree of precision.

Expressing discount rates as ranges recognises the uncertainty involved in deriving reliable estimates of the WACC and DTF advised that the discount rate range reflected uncertainty around the true commercial rate of return appropriate for this project. However, it is unusual to express discount ranges solely as ranges, and adopting this approach produced a very wide band for the state VFM benchmarks.

Figure 3D shows:

  • that even the modest range in the state's discount rate resulted in a very wide range for the state VFM benchmarks of $655 million NPV
  • the extent to which the state VFM benchmarks shift if the discount rate range is varied by just +/- 0.25 per cent.

Figure 3D
Sensitivity of VFM state benchmarks to changes in discount rates ($ million NPV)

Figure 3D shows sensitivity of VFM state benchmarks to changes in discount rates ($ million NPV)

Source: VAGO based on data from DTF.

The reason the VFM benchmarks are so sensitive to the discount rate relates to the timing of the project cashflows. In projects such as the WGT, where most of the costs are incurred early and most of the revenues are spread over the lifetime of the project, the longer the life of the project, the more sensitive the state's VFM benchmark range will be to the discount rate range.

Further, the upper bounds of the discount rate ranges used by the state were materially higher than the discount rate used by Transurban, suggesting the state estimated higher required commercial rates of return than that adopted by Transurban.

Transurban had better knowledge of the commercial rate of return in relation to its existing CityLink asset and the project. Transurban also had an incentive to propose the highest possible commercial rate of return, to maximise the state contribution towards the project. DTF advised that Transurban's incentives would have been dampened by the fact that the state could have chosen at any stage in the MLP process to deliver the project itself. That is, the state's ability to 'walk away' would have constrained Transurban's incentives to overstate the commercial rate of return.

Aside from Transurban's incentives, it should have been clear to the state that Transurban had adopted a rate of return that was close to the bottom end of the discount rate range that the state had adopted. There are two possible reasons why the state's estimate of the discount rate was higher than Transurban's:

  • Transurban genuinely had a lower cost of capital than the SOE, in which case adopting the Transurban proposal would have provided the state with VFM, or
  • the state overestimated the required rate of return.

The state should have undertaken further work to rule out the second possible explanation, because if the state had overestimated the discount rate, that would have had a material impact on the VFM assessment. DTF did not demonstrate that it had undertaken any such work. In addition, DTF advised us that it had no information to determine whether the top of the state's discount rate range was more or less likely than the bottom end of the range.

It would have been reasonable for DTF to examine and advise the government:

  • whether there was any reason Transurban may have in fact had a lower required rate of return than the SOE, and
  • the impact of adopting Transurban's rate of return estimate on the outcome of the VFM assessment, as a sensitivity analysis.

There may have been legitimate reasons why an SOE would need a required rate of return higher than Transurban, including assumptions about the type of finance that would be used for this investment. Given the importance of the discount rates to the outcome of the state's VFM assessment, we expected clear evidence that this question was explicitly addressed. However, the:

  • VFM assessment report offered no discussion or evidence about whether the SOE faced a higher cost of capital than Transurban
  • assumptions in the investment bank discount rate analysis advice and models lacked transparency in this area and while evidence provided by DTF shows ongoing interaction with the investment bank as it developed its analysis the evidence does not demonstrate substantive testing and challenge by DTF.

In an investment of this scale, where the discount rate was so influential to the VFM outcome, the reasoning behind these key inputs should have been set out explicitly and unambiguously.

The use of discount rates in the state's VFM assessment that materially exceeded Transurban's proposed rate of return was significant because the top end of the state's discount range produced the lower bound (negative $600 million) of the VFM range, while the bottom end of the state's discount rate range indicated that it would have been better VFM for the state to undertake the project (i.e. the VFM benchmark in that scenario was positive $55 million).

All points in the discount rate range are not necessarily equal

Further to the issue of using discount rate ranges, the state's VFM assessment assumed that discount rate and NPV outcome points within the state's benchmark range were equally likely to occur. However, this may not be true.

The upper and lower bounds of the state's WACC ranges were derived using a set of input assumptions. The most important of those assumptions were benchmarked using actual toll road and infrastructure projects. The projects used in the benchmarking exercise may not have shared close risk characteristics with this particular project and DTF has not provided evidence the toll road and infrastructure projects used in the benchmarking exercise were in fact appropriate comparators for that task. Therefore, the appropriate rate of return for this project may have been located closer to one end of the state's WACC range than the other.

In these circumstances, there is no certainty that the midpoint discount rate estimate is the best or most likely outcome, or that estimates above the midpoint are equally likely as estimates below the midpoint. Given this, and the fact that the VFM state benchmarks were very sensitive to the discount rates used, it would have been preferable to show sensitivity analysis using point estimates for the discount rate that represented advisers' best view of a commercial rate of return for the project. The Transurban proposal could then have been compared to the VFM benchmark derived using the state's best estimate of the required rate of return.

The discount rate ranges could have still been used to demonstrate the range of uncertainty around a point estimate. However, the ranges are better used as a cross-check or for the purposes of conducting sensitivity analysis on the VFM conclusion, rather than to drive the central result, as was the case in the state's VFM assessment of Transurban's proposal.

Determining a point estimate that represents a likely discount rate outcome is not straightforward given uncertainties and challenges in discount rate estimation. However, this should have been confronted as part of the state's VFM assessment. Alternatively, DTF should have provided the government with more transparent advice on how the VFM assessment was undertaken and its sensitivity to different assumptions.

The discount rate used for the state contribution in the VFM assessment

The VFM assessment assumed that the SOE would receive the same state contributions that would be received by Transurban. The size of the state contribution in the Transurban proposal was calculated as the nominal amount required to ensure that the project achieved an NPV of zero, while delivering Transurban an expected commercial rate of return.

The VFM assessment adopted the nominal state contributions proposed in the Transurban final offer and discounted all of those nominal state contributions using the state's estimates of a commercial rate of return for the SOE.

It was reasonable for the state contributions towards state costs and state works to be discounted using a commercial rate of return, as those cash flows had commercial risk attached to them. However, a commercial discount rate should not have been applied to the state contributions towards design and construction costs. These contributions totalled around $1.4 billion in nominal terms in Transurban's proposal.

Standard finance theory advises discounting cashflows using a rate that reflects the risk associated with those cashflows.

The state contribution towards the WGT design and construction costs had low risk attached because those payments were negotiated and agreed with Transurban and were to be reflected in the project agreement and other contractual arrangements for the project. The state was effectively agreeing to make a schedule of payments to Transurban, or the SOE in the VFM assessment, to share a portion of the design and construction costs related to WGT.

Therefore, the nominal state contribution cashflows included in the VFM assessment that related to sharing the WGT design and construction costs should have been discounted using the state's borrowing rate as a proxy for the risk-free rate, not a commercial rate of return. From the perspective of the SOE, the commitment by the state to make the scheduled payments to share these design and construction costs should have been viewed—and valued in the VFM—as a risk-free commitment by the state.

Given this, the discount rate applied in the VFM assessment to the nominal state contribution that related to sharing WGT design and construction costs should have reflected the risk of the state defaulting on its funding commitment to the project, that is, the state's borrowing rate. In early December 2017, the state's five-year borrowing rate was approximately 2.17 per cent, which is considerably lower than the commercial rate of return used in the VFM assessment.

Appendix D of Volume 5 of the National PPP Guidelines provides relevant guidance on this point:

'The PPP bids will include either a payment to, or from the state independent of the actual future revenue experience and are thus devoid of systematic and project risk, from the government perspective and hence should discounted by the risk-free rate.'

The guideline is clear that where cashflows are devoid of risk from the government's perspective (that is, the government will make a set of fixed cash flows), those cash flows should, for the purposes of the VFM assessment, be discounted at the risk-free rate. Where cash flows are subject to commercial risk from the government's perspective (that is, the cash flows to be made by government are uncertain, depending on whether or not particular commercial risks materialise), those cash flows should be discounted at a commercial rate that reflects the risk of those cash flows.

In this case, there was commercial risk attached to the WGT design and construction costs. Therefore, it was reasonable to discount those costs using a commercial rate of return in the VFM assessment. However, the portion of the state contribution towards WGT design and construction costs was essentially a risk-free commitment by the state and should have been discounted at a risk‑free rate. This means that the VFM assessment undervalued (in NPV terms) the state contributions.

The state's VFM assessment sought to determine whether Transurban's delivery of the project represented better VFM than state delivery of the project. To do that, the VFM assessment should have compared:

  • the net cost of the state undertaking the project itself (that is, the commercial revenues that would be generated by the project less the costs of the state building and operating the project), against
  • the savings the state would realise by avoiding state contribution payments to Transurban.

If the net cost of the state undertaking the project exceeds the state contribution costs the state would avoid by delivering the project itself, then Transurban should be allowed to undertake the project. If, however, the state contribution costs that the state would avoid by delivering the project itself exceed the net costs of the state undertaking the project, then it would be better VFM for the state to self-deliver rather than allow Transurban to undertake the project.

The incorrect use of a commercial rate of return to value all of the state contributions in the VFM assessment did not affect the net cost of the state undertaking the project (that is, the revenues generated by the project less the costs of the state building and operating the project). However, as explained above, by applying a commercial rate of return rather than a risk-free rate to discount a significant portion of the nominal state contributions, the state contributions towards design and construction costs were undervalued in the VFM assessment.

Correcting this error in the VFM assessment causes the VFM of the Transurban proposal to decline, since the savings to the state from self-delivery of the project (that is, the avoided state contributions to the Transurban project) would be larger than was estimated in the original VFM assessment. This would effectively mean an upward shift in both the high and low VFM benchmarks.

DTF submitted to us that even if there had been an error in valuing the state contributions, that should have no impact on the overall VFM assessment. That is because applying a risk-free rate, rather than a commercial rate of return, to the state contribution to design and construction costs causes the value of the project to Transurban to increase. If the relevant risk-free rate were taken to be 2.17 per cent, then the NPV of the Transurban project would rise to positive $191 million.

DTF advised us that the VFM assessment should recognise that the NPV of the Transurban project has increased in line with any upward shift in the VFM benchmarks.

This is incorrect. The VFM assessment should have no regard to how Transurban values the project. All that should matter in the VFM assessment is:

  • the net cost to the state of building and operating the project, and
  • the value of the state contributions the state could save if it were to deliver the project instead of Transurban.
Lack of evidence of review

It is unclear whether DTF substantively tested the approach used in the WACC model that the investment bank used to derive the state's discount rate range.

The Stage 4 VFM assessment report stated that 'no review has been undertaken by the State or its advisers in reviewing the underlying assumptions and calculations in [investment bank's] WACC model'. This is despite the outcome of the state's VFM assessment depending heavily on the reasonableness of the discount rate estimates. However, the state's commercial adviser advised that the statement in the VFM assessment report is incorrect and should have stated that the commercial adviser specifically did not undertake a review, without reference to the state.

DTF, in response, advised us that:

  • the engagement scope for its primary commercial adviser did not extend to reviewing the inputs to the state's VFM benchmark, such as discount rates
  • it, as well as DEDJTR, engaged separate advisers to provide expert input to each benchmark element—relying on their respective specialist skills and experience—and therefore did not consider it necessary to also engage an external consultant to further review these inputs
  • its general approach in relation to advice received from advisers was to workshop and challenge such advice to ensure it was understood and appropriate, including for advice received from the investment bank in relation to discount rates.

DTF has provided evidence showing that it had ongoing interaction with the investment bank as it developed the discount rate analysis and advice. However, this evidence does not demonstrate substantive testing and challenge by DTF.

Lack of clarity and specific evidence to support key assumptions

The discount rates used in the state's VFM assessment were highly sensitive to certain input assumptions made by the investment bank when deriving those rates.

The key input assumptions were about the assumed equity internal rate of return, the cost of debt, and the debt structure. However, the basis for these assumptions is not clear and it was not possible to determine, based on the very limited material provided by the investment bank, including its reports to DTF's advisers and its financial models and calculations, whether those input assumptions are reasonable.

The investment bank derived the discount rates used in the VFM assessment by benchmarking against the returns required by investors in other toll road projects—which may or may not have been directly comparable to the project proposed by Transurban. It is possible that the limitations of the data available (for example, a large enough sample of sufficiently comparable projects) meant that there was significant uncertainty over the rate of return for this project. Under these circumstances, the true rate of return could have been located near the bottom end of the discount rate range estimated by the investment bank, close to the rate of return proposed by Transurban. This should have prompted the VFM assessment to explicitly examine this possibility.

For example, a key input assumption in the discount rate calculation was the rate of return equity investors require in commercial toll roads. It appears this input assumption was derived by benchmarking the rates of return equity investors require in a sample of toll road investments. However, the investment bank's work does not disclose details of these toll roads, meaning we cannot assess whether they are comparable to the project proposed by Transurban. Given the significance of this input, the inability to check this is an issue.

We note that DTF's Stage 2 Assessment report stated that 'the CityLink toll revenues are very mature and relatively low risk compared to international toll road comparisons.'

DTF advised that due to the confidentiality of some of this information, it did not have access to the details related to some of the comparator toll roads either. It is difficult to see how DTF could have scrutinised the relevance and comparability of certain toll roads used in the benchmarking exercise if it did not know the identity of those toll roads. This is not necessarily a criticism of DTF. It is understandable that DTF may not have had full access to confidential information. However, this demonstrates the uncertainty that DTF faced around the discount rate used in the VFM.

In these circumstances, significant caution should have been exercised when interpreting the VFM results—particularly since some of the state's own discount rate scenarios suggested that acceptance of the Transurban proposal would not have provided VFM to the state.

Presentation of VFM results to the government

In December 2017, DTF presented the Stage 4 VFM assessment results to the government. The DTF advice included the chart shown in Figure 3E, designed to give the government a comparison of Transurban's proposal and the alternative SOE. The advice noted that 'where the Transurban Proposal sits in the top half of the range, prima facie this suggests it represents VFM'.

This chart shows:

  • the Transurban proposal with an NPV of $0, based on the inclusion of a state contribution
  • a wide state VFM benchmark range of positive $55 million to negative $600 million NPV.

Figure 3E
Presentation of Stage 4 VFM assessment results to the government

Figure 3E shows presentation of Stage 4 VFM assessment results to the government

Source: DTF.

The VFM assessment methodology DTF used required a comparison of the net project funding in the Transurban proposal to a state benchmark range. It did not specify or explain where in this range the Transurban proposal would need to sit to represent VFM to the state.

The presentation of the VFM assessment results in Figure 3E is not transparent because it wrongly implies that the mid-point of the state VFM benchmark is the threshold for achieving VFM. Given that the state faced significant uncertainty in developing key elements of its benchmark, such as future toll revenue estimates, the VFM total benchmark range was important and the midpoint does not necessarily represent the state's best or most likely estimate of VFM.

DTF's presentation approach showed the Transurban proposal falling within the range provided by the state's VFM benchmarks and meeting the VFM assessment criterion that it needed to be above the midpoint of these benchmarks.

We note that:

  • the state's commercial adviser responsible for developing the VFM assessment provided no indication of the likelihoods attached to the 'high' and 'low' state VFM benchmark scenarios and did not express any opinion on the most likely benchmark outcome
  • it was probably not the case that any point within the range was equally likely
  • the upper and lower bounds of the range were driven very materially by the state's discount rate range, and we have raised questions about the lack of review and transparency over the basis for and reasonableness of assumptions underpinning the discount rate estimates used in the state's VFM assessment.
VFM benchmark range versus a single point estimate

The VFM assessment analysis would have been more useful and transparent if DTF had provided the benchmark range, together with a single point estimate. That is, a single point VFM benchmark to represent DTF's advisers' best view of the cost to the government of delivering the project itself.

The government could have used this estimate as the key comparison point for the Transurban proposal. The closer the most likely outcome—the single point estimate—for the state benchmark to the top of the state's VFM benchmark range, the less likely that the Transurban proposal would have met the VFM criterion.

A more transparent presentation of VFM assessment results

DTF's advice to the government on the VFM assessment results showed the Transurban proposal as having an NPV of zero. In this particular case, the NPV of the Transurban proposal was zero, given the way Transurban valued the state contributions it proposed.

Figure 3F shows how the high, low and midpoints were derived for the presentation to the government. It also shows the differences in estimates between the state's Stage 4 midpoint benchmarks for costs and revenues and the final Transurban proposal.

Figure 3F
Comparison of state VFM benchmarks at Stage 4 and the final Transurban proposal ($ million NPV)

 

Transurban Proposal

SOE Low

SOE Mid

SOE High

Difference b/w SOE Mid & Transurban

Revenues

Tolls

4 056

3 339

3 754

4 169

-302

State contribution

1 960

1 926

1 944

1 961

-16

Cost savings for settlement of CityLink historical claims

31

31

31

31

0

Major Maintenance Reserve account

24

24

24

24

0

Total Revenue

6 071

5 320

5 753

6 185

-318

Costs

Design and construction costs contestably procured

4 221

4 148

4 188

4 227

33

Design and construction costs not contestably procured

388

311

314

317

74

State costs

854

848

851

854

3

Operations and maintenance costs not contestably procured

608

613

673

732

-65

Total Costs

6 071

5 920

6 025

6 130

46

Net state benchmark as presented to government

0

-600

-272

55

-272

Net state contribution

-1 960

-2 526

-2 216

-1 906

-256

Source: VAGO based on information from DTF.

DTF's VFM assessment compared the VFM benchmarks (as computed above) to the value of the Transurban proposal. However, the VFM assessment should be independent of how any other party, including Transurban, values the project. The VFM assessment should consider how the state values the project, and what savings it would make (by way of avoided contributions to Transurban) if it were to deliver the project.

DTF's advice to the government on the results of the state's VFM assessment would have been more informative if it had compared the present value of:

  • costs incurred by the state if it proceeded with the Transurban proposal, that is the state contribution component, against
  • the NPV of the estimated cost and revenue cashflows if the state delivered the project itself using the SOE delivery model.

Presenting the information this way would have provided clearer advice to the government to inform decisions on the WGT MLP.

Design and construction costs

The Stages 3 and 4 assessments include non-contestable design and construction costs to the state. These are Transurban's estimates of these costs, and they were not competitively tendered. While only around 6.4 per cent of the overall project costs, these costs were an area of potential VFM risk for the state given the need to rely on Transurban's representations.

The MLP interim guideline required DTF to support the Stage 4 VFM assessment by a cost reasonableness assessment including independent review and open book review of Transurban's proposed costs. DTF and the West Gate Tunnel Authority took steps to gain assurance on the reasonableness of these costs including engaging expert advisers to assess Transurban assertions on its costs and create a state benchmark for the same costs but did not gain open book access to Transurban's costings.

At the end of the Stage 4 assessment of the Transurban proposal, DTF and the West Gate Tunnel Authority were not satisfied that the Transurban proposal for non-contestable design and construction costs provided VFM for the state. However, the summary advice DTF provided to the government when seeking approval to proceed to contract close with Transurban did not raise this issue. The summary advice referred to testing Transurban's non-contestable costs but did not inform the government about the results of that testing.

Instead, the summary advice to the government communicated an overall VFM assessment rating of 'satisfied' for Transurban's final offer and stated that 'all non-contestable costs, including Transurban's rate of return, have been rigorously tested against a robust state benchmark'.

Attachments to the summary advice revealed the actual extent of testing, but this could have been more transparently noted in the summary advice.

Back to Top