This section discusses revenue risk sharing mechanisms that could potentially be applied in the U.S. As summarized in Table 4, these mechanisms can create a range of impacts. For example, the extent to which a project is made financeable by a Minimum Revenue Guarantee depends on the level of protection provided by the guarantee.
| Criterion | Present Value of Revenues | Minimum Revenue Guarantee | Contingent Finance Support | Availability Payment & Revenue Sharing | Shadow Tolls | Regulated Returns | Innovative Finance Programs |
|---|---|---|---|---|---|---|---|
| Value for Money | ●●● | ●●● | ●● | ●●● | ● | ● | ●● |
| Fiscal Impact | ●●● | ●● | ●● | ● | ● | ● | ●● |
| Financeability | ●● | ●●● | ●●●● | ●● | ●● | ●●● | ●●● |
| Ease of Implementation | ●●● | ●●●● | ●●● | ●● | ●●● | ● | ●●● |
Key: Most value or benefits = ●●●●
Least value or benefits = ●
* Benefits are in terms of maximizing value for money, reducing fiscal
impact, enhancing financeability, and increasing ease of implementation.
Enhancing financeability benefits the Developer, who usually is responsible
for the financing, yet it also benefits the Agency, whose interest is
also in a successful financing.
Examples of these mechanisms are provided in Appendix I. Most of the examples are international with the exception being Contingent Finance Support. Some mechanisms are not included in this section (e.g., Revenue Distribution Mechanism, Rate of Return, and Price Cap), since they are not sufficiently developed for consideration in the U.S. and/or are difficult to implement for standalone projects.
Under a PVR approach, Developers bid a minimum gross revenue discounted at a pre-determined discount rate. The P3 contract ends when the NPV of gross revenue is reached. The concession term varies as a function of realized gross revenues, but the contract provides for a base case and minimum and maximum terms as discussed in Appendix 4I.1.
Figure 3 illustrates PVR's impact on debt service, with debt prepaid if PVR revenues are higher than expected (Scenario 1) and maturity of debt extended if revenues are lower than expected (Scenario 2).
Figure 3: Illustration of PVR Mechanism
View larger version of Figure 3
Under a PVR mechanism, the revenue risk transfer to the Developer is limited as the contract can be extended if revenues fall short of expectations. However, in extreme downside cases, Developers will still be exposed to revenue risk as the maximum contract duration will typically be capped. Furthermore, for the PVR mechanism to effectively reduce revenue risk, the base concession term should not be too long (e.g., no more than 20 to 30 years) so that an extension of the concession can indeed achieve the desired gross revenue NPV bid value. If the base concession term were to be 50 years, an additional 20 years (or even 50 years) may not have a significant effect on the gross revenue NPV due to the effect of discounting.
Assuming that the selected PVR structure addresses these concerns, the Agency effectively accepts the revenue risk through the flexible contract end date (i.e., the Agency will only start receiving toll revenues after that date). Since a contract extension could result in additional O&M costs that are non-linear, negatively impacting the Developer's returns, the Developer is incentivized to maximize early revenues.
Compared to a full revenue risk transfer, inefficient risk pricing is less likely as the PVR provides downside revenue risk protection for Developers. Given the potential for excessive pricing without PVR in a less-than-robust market, the mechanism is expected to create more VfM than a full revenue transfer to the private sector.
The PVR mechanism does not have any immediate direct or contingent fiscal impacts for the Agency. However, as explained, the concession term variation does impact when the Agency will start receiving toll revenues, which obviously has future fiscal impact.
Subject to the limitations listed in the VfM section above, the PVR mechanism can significantly reduce the revenue risk for the Developer, thereby enhancing financeability. However, not only the revenues, but also the Developer's costs vary with the contract term. To the extent the costs are more or less linear, the Developer will be able to account for them in determining the minimum gross revenue. To the extent the costs are non-linear, such as major maintenance costs or expansions, the variable term contract creates a risk.
Just as in other P3s, Lenders would have step-in rights in case revenues are too low to meet debt service obligations. In that case, a contract extension would improve the possibilities for a successful restructuring, compared to a situation without the possibility of such an extension.
While U.S. terms for toll concessions are now around 50 years, most Lenders will currently not lend for a period of more than 40 years, creating a refinancing risk in downside cases. Private Activity Bonds, the typical choice of senior debt in current U.S. P3s, have provisions limiting early repayment (especially in the first decade) and generally have a maturity of no longer than 30 years, making it difficult to adjust the senior debt tenor. To overcome this, Agencies could consider 1) basing the minimum and maximum contract terms stated in the P3 agreement upon what is achievable in the financing market, 2) obligating the Agency to partially or fully repay the bond if it is not repaid after 30 years (effectively shifting more of the revenue risk to the Agency), as is the case in some termination provisions of U.S. P3s, and/or 3) encourage Developers to use zero coupon instruments in their financial structure, which has been used in some non-recourse toll roads such as the San Joaquin Hills Transportation Corridor Agency 73 toll road financing in California. 2
Some Developer respondents said they might not like the fact there is no or little "upside." That may be the case, but the flip side is that the mechanism offers significant protection. A combination of significant downside protection and significant upside potential should be attractive to Developers, but unrealistic and unreasonable from an Agency perspective.
The proposed mechanism is based on gross revenues, which can be easily monitored. A mechanism based on net revenues, including O&M, has proven difficult to monitor in some P3s, since Developers included what are essentially profits in certain cost categories. Therefore, a gross PVR mechanism scores high on ease of implementation.
Some Developer Respondents stated that PVRs add complexity to an already complex market made up of fifty or more DOTs with a variety of legislative and political requirements, because the mechanism is new and requires a strong understanding of finance. However, calculating the PVR is relatively straight forward and is not much more complicated than current P3 or revenue bond financings currently in the market.
If the discount rate specified in the PVR mechanism reflects the Developer's weighted average cost of capital (WACC) based on previous P3s, Lenders are relatively indifferent to when they receive their payments. If the discount rate is different from the Developer's WACC, this may create problems. For instance, if the Developer's discount rate is lower than the Agencies' prescribed WACC, then the Developer may have an incentive to prolong the concession. Getting the discount rate right is therefore the main challenge. Obviously, information from recent transactions should get close to the "right" discount rate. To minimize negative impact for the Developer, the actual WACC at financial close could be included in the P3 agreement.
PVRs may fail to incentivize performance related to maintenance and quality of service, and the uncertain duration of the concession could present challenges relating to handback of the facility. These issues can be dealt with by structuring and enforcing performance standards through penalties, as would be the case in AP P3s.
Under a MRG, the Agency sets a base case revenue line and guarantees to cover revenues in any year in which revenues fall below this line. Appendix I discusses how Agencies have guaranteed revenue ranging from 60% of base forecast in Canada and Brazil to 80%-85% in Chile. Initially, South Korea guaranteed 80%-90% of the revenue forecast, but it has become less "generous" over time. The MRG can be sized in different ways. For example, it could cover debt service - partially or fully - possibly in combination with expected O&M expenses, or even providing some protection to equity. Its duration can extend to the entire concession period or can be limited, for example, to only the debt tenor. The MRG level can also change over time.
MRGs often have a corresponding upside revenue sharing mechanism, in which the Agency shares in upside revenues above a certain level, for example 120% of forecasted revenues. Such provisions are common in U.S. revenue risk toll roads, such as in Texas and Virginia, but they have rarely been tested due to lower than expected realized revenues. Figure 4 illustrates an example of a MRG mechanism where the Agency partially covers the gap between actual and forecasted revenues in combination with an upside revenue sharing mechanism.
Figure 4: Illustration of Minimum Revenue Guarantee Mechanism
View larger version of Figure 3
From a flow of funds perspective, MRG payments are defined as project revenue, entering into the priority payment structure at the top of the cash waterfall.
A MRG mechanism could also include a claw back clause, where the Agency is repaid its earlier contributions if the project's revenues exceed forecasted revenues at a later date.
MRG mechanisms help protect Developers' downside risk as the Agency guarantees revenues up to a certain level. A MRG provides better revenue risk protection to Developers in extreme downside revenue cases than a PVR due to PVR's contract duration limits. However, the overall revenue risk protection will ultimately depend on the revenue level that the Agency guarantees.
Compared to a traditional toll concession with full revenue risk transfer, MRGs should generate more efficient risk pricing and likely additional private sector interest leading to more competition as it can provide significant downside protection. As a result, MRGs are expected to create more VfM than a full revenue risk transfer.
MRGs are contingent liabilities as the Agency's contributions will ultimately depend on realized traffic. As mentioned, accurately forecasting traffic can be challenging, which is why valuing contingent liabilities is difficult, although academics continue to develop new techniques that hold promise (Chiara and Garvin, 2007). In any case, Agencies need to have adequate funds to cover downsides, most likely in gas or sales tax revenues. A portion or all of these liabilities would be included in the Agency's budget, based on how each jurisdiction "scores" and how credit rating agencies view the obligation, which can vary.
As one credit rating agency Respondent pointed out, if an economic downturn reduces project revenues, it is likely that Agency's funding sources, such as gas taxes, will also decrease, as experienced in the global financial crisis. Therefore, the Agency may struggle to meet its obligations on a timely basis. This is true for any Agency contribution, but particularly contingent liabilities.
Compared to an upfront subsidy or AP mechanism, MRGs can help reduce Agencies' direct liabilities. Depending on how contingent liabilities are accounted for, MRGs may allow Agencies to support a larger number of projects than direct subsidies or AP mechanisms. This assumes, of course, that the Agency has accurately calculated its contingent liabilities and made requisite budgetary provisions.
MRG mechanisms help reduce revenue volatility and eliminate extreme downside revenue risk for Developers and Lenders. Through this improved risk profile, Lenders be may be more able and willing to finance a MRG-supported project.
Depending on how the MRG is structured, it may cover the majority of debt service costs, thus addressing a key concern of Lenders in current revenue risk sharing P3s. As the price of debt reflects the nature of the risks to which it is exposed, MRG should lead to lower interest rates. For certain projects with high revenue risk, Lenders may be unwilling to lend money, even at high interest rates. In that case, MRGs - like any other revenue risk sharing mechanism that reduces risks for Developers and Lenders - can help make such projects financeable. Similarly, in order to obtain a TIFIA loan, Developers usually need to attract senior debt, often in the form of private activity bonds (PABs) which are rated "investment grade" by major credit rating agencies. An MRG can be a key credit factor in obtaining such a rating.
A revenue risk sharing mechanisms like a MRG may also affect the financial structure of a P3 transaction. Indeed, besides attracting lower cost debt, a MRG could result in higher debt-to-equity ratios, thereby lowering the overall cost of capital, as debt is cheaper than equity. Depending on the revenue guarantee level, a MRG can create different risk allocations between Lenders and Developers. They may also reduce Lenders' monitoring costs, especially if the MRG covers all or close to all debt service. However, this could have a negative impact on P3 efficiencies, as the discipline imposed by Lenders is an important value driver in P3s.
A MRG could also cover equity payments, thereby guaranteeing an equity return that could be equivalent to a U.S. Treasury or similar instrument. Most MRG mechanisms used today provide minimal or no equity protection. Many Developer and Agency Respondents felt that the MRG should not directly benefit Equity Investors. Respondents shared anecdotal evidence that some Developers, dominated by Strategic Investors, do not focus on equity returns because most profits of their combined companies are made in construction activities. Furthermore, there may also be "agency issues" as economists define them, in that Developer staff are highly incentivized to "win and close the deal" with less long-term interest in equity returns. Furthermore, providing too much revenue risk protection may reduce Developer incentives.
Several credit rating agency Respondents also focused on the need for the mechanics of MRGs and CFS (the latter discussed below) to be crystal clear: do they result in timely payment of obligations, i.e., within 30 days? These Respondents have had experience with Agencies who failed to make timely payments, even though they had strong financials. This issue can likely be addressed with clearly stated legal obligations and periodic funding of a reserve.
MRGs have been used extensively throughout the world, including in Europe, Latin America, and Asia and many Developers have experience with them. As MRGs are gross revenue guarantees, they do not pose any particular practical problem in terms of monitoring.
Some U.S. states may not be able to legally provide revenue guarantees. For instance, TxDOT cannot provide revenue guarantees to toll road projects. In order to address this, TxDOT provided a subordinated loan to the North Texas Transportation Authority in the SH-161 financing, a public toll road project subject to revenue risk. In that transaction, if lower than expected revenues forced debt service coverage to dip below specified levels, then this would require TxDOT to take out senior and TIFIA debt, making this subordinated loan essentially a financing support mechanism, although it was not legally considered a revenue guarantee.
A MRG is first and foremost a revenue risk sharing mechanism. However, it can also be used as a procurement bidding parameter. In that case, bids could be evaluated based on 1) the total amount of revenue covered by the MRG in present value terms or 2) the required (upfront) subsidy for a given MRG level. Alternatively, the MRG can be treated as a form of insurance, as was the case in Chile, where the Developer paid a fee of 0.75% of the guaranteed revenue.
Contingent Finance Support (CFS) is a risk sharing mechanism that addresses financing as opposed to revenue risk. Under a CFS, the Agency guarantees that the project will be able to repay debt, even under downside scenarios. While CFS is technically not an MRG, it has many similarities and could potentially be viewed as a MRG subset. However, downside scenarios could be caused by lower than expected revenues (as under a MRG) or higher operating costs.
An example of a CFS would be the Developer Ratio Adjustment Mechanism (DRAM) used in the I-77 project in North Carolina. Under this mechanism, NCDOT guaranteed a debt service coverage ratio of 1.00x, effectively covering the Lenders but not equity investors. However, to limit its exposure, NCDOT limited the DRAM payments to a maximum of $12M per year and $75M in total. To the extent that the DRAM had not been exhausted, it would available for the entire concession period. Towards the end of that period, the DRAM would be released from NCDOT's budget and available for other projects. As the I-77 transaction closed in 2015, no DRAM has been requested or is anticipated to be requested at this time. To be prudent, however, NCDOT has programmed the expected cost of the DRAM in its long-term state transportation improvement program based on a severe downside case. This serves as an example of how Agencies can account for CFSs and MRGs. An example of a CFS is illustrated in Figure 5.
Figure 5: Illustration of CFS Mechanism
View larger version of Figure 5
Similar to a MRG, a CFS can provide significant revenue risk protection to Developers and Lenders. Depending on the guarantee's limits, in annual or overall value, a CFS can be particularly valuable in the project's early years when liquidity is tight.
However, as a CFS covers all risks, not just revenue risk, it means that the O&M cost risk is no longer being fully transferred to the Developer, which is less efficient, since the Developer is better placed to manage such costs. Retaining this risk would partially remove the incentive for the Developers to minimize life cycle costs, which is considered one of the key value drivers of a P3 structure. Therefore, this aspect of CFS is not conducive to VfM.
Similar to a MRG, a CFS is a contingent liability. It can be difficult for Agencies to establish the value and budget implications of such liabilities. As Agencies become more familiar with MRGs and CFSs, they may develop better ways to assess the fiscal impacts of contingent liabilities.
Depending on how the CFS is structured, the primary beneficiaries are most likely Lenders, for whose protection there may be a broader public understanding. Most Respondents believed that this was a public relations positive, since the general public continues to object to Agencies providing financial support that appears to "subsidize" or "guarantee" private company investments. Furthermore, several Respondents believed that Developers are readily able to obtain equity for P3s, from Strategic and/or Financial Investors, so that this mechanism correctly focuses on the problem of obtaining debt. Several Respondents referred to the sale of the Indiana Toll Road in 2015 as an example of a large supply of equity.
As was the case for MRG and other revenue risk sharing mechanisms, reducing a project's risk profile makes financing easier and cheaper. In the I-77 case, the CFS focuses on the worst case, when annual debt service coverage ratios (DSCR) fall below 1.00x, when Lenders absolutely need it. As a result, the DRAM was essential to that transaction reaching financial close. Compared to an MRG, the CFS will further improve financeability as it considers both revenues and operating costs.
Although a CFS is very similar to a MRG, its implementation and monitoring may be more difficult as it requires a detailed analysis of a project's cash waterfall and the mechanics of it may be problematic since it is not common to the U.S. municipal finance market. For example, in the case of I-77, NCDOT will need to determine both gross revenues and expenses ahead of debt service in order to determine the DSCR, which in turn triggers the DRAM payment. In that context, a MRG may be easier to implement while also more directly addressing revenue risk.
An Agency could combine an AP and revenue sharing mechanism, similar to the payment mechanism in the A25 project in Canada. In that project, the Developer compensation consisted of:
Depending on structuring, it can be similar to a MRG, although its role in a procurement may be different, as discussed below. Figure 6 illustrates an AP/Revenue Sharing combination.
Figure 6: Illustration of AP in combination with Revenue Sharing
As with other mechanisms discussed earlier, revenue risk sharing mechanisms that limit Developers' exposure to revenue risk should create VfM when competition for full revenue risk concessions is weak and/or inefficient risk pricing is likely. In this case, the extent to which the Developer is shielded from revenue risk depends directly on the level of AP in relation to the level of toll revenues to be received. If the AP is relatively high, it means that the transaction will be similar to a normal AP transaction with limited revenue risk. As a result, inefficient risk pricing is unlikely, which in turn should result in VfM. If the AP is relatively small, the Developer will still be exposed to significant revenue risk, which could lead to inefficient risk pricing if there is insufficient competition among Developers for high levels of revenue risk.
As the Agency is paying an AP to the Developer, the AP is a direct liability. However, as toll revenues are used to pay the AP, the shortfall - AP minus the realized toll revenues - becomes an Agency's contingent liability. On the upside, the Agency can expect to earn a share of the revenues. Furthermore, as a portion of the project will still be financed through toll revenues, this mechanism allows for some off-balance sheet financing.
To the extent that the mechanism reduces revenue risk for Developers, it should make obtaining financing easier. However, there is a risk associated with combining multiple risk sharing mechanisms. Bringing together an AP structure with revenue risk sharing could confuse Lenders on the nature of the credit. Is this primarily an AP or a revenue risk credit? In discussions with credit rating agencies, Developers and Lenders, there were differences of opinion on whether such an instrument would be rated as 1) a hybrid credit with melding of AP and revenue risks or 2) a credit based on its weakest element, revenue, even when the AP covered the majority of the debt. One Respondent commented, however, that credit analysis also depends on how well the mechanism is defined in the legal documentation.
Implementing a combined AP/revenue sharing mechanism necessitates a coherent procurement strategy. As this mechanism has two distinct variables - required AP and the revenue sharing level - the Agency needs to decide on which variable it wants Developers to bid. Comparing bids that have different APs and revenue sharing levels is extremely difficult due to the challenges in determining the value of uncertain revenues.
One option would be for the Agency to set the AP and let the Developers bid on the level of revenues they are willing to share. In this case, the AP has to be set at a level well below a normal AP which would cover all investments and operational cash flows.
Alternatively, the Agency could specify a revenue sharing profile, using revenue bands, asking Developers to bid on the required AP. This is similar to bidding the lowest subsidy as in many U.S. P3s, yet the subsidy payments are performance-based and made over the concession period (potentially in combination with a milestone payment) as opposed to upfront payments, as in current practice.
The involved procurement process combined with potential difficulties in assessing this mechanism's credit may make it more difficult to implement.
In a shadow toll approach, an Agency pays a Developer on a per vehicle basis, with the Agency retaining any toll revenues, if it is a toll road. As discussed in Appendix I.6 shadow tolls eliminate risks to the Developer related to perceived or real economic barriers that tolls and tolling technology create. However, under most shadow toll mechanisms implemented in the U.K. and contemplated in the U.S., Developers are still subject to all traffic risk. Furthermore, Agencies are subject to 1) payment liabilities (which depend on traffic) and 2) revenue risk (only if the facility is tolled). As Figure 7 shows, the Agency pays the Developer shadow tolls that are subject to traffic volatility, so that the Developer still faces revenue risk.
Figure 7: Illustration of Shadow Toll Mechanism
View larger version of Figure 7
Under shadow tolls, the Developer's exposure to traffic risk means that it will face similar challenges as in a full revenue risk transfer. In particular, if there is inefficient risk pricing (e.g., due to insufficient competition among Developers) it may result in a loss of VfM compared to a more balanced approach to revenue risk sharing.
If shadow tolls are used as a partial subsidy to ensure a more efficient use of managed lanes, the economic benefits for society in the form of additional time savings could be significant since the Developer will attempt to maximize vehicle throughput on the managed lanes in order to maximize its revenue. Furthermore, the negative impact on risk pricing may be limited if the Developer believes there is sufficient demand to fill the managed lanes' capacity.
Under a shadow toll, the Agency bears the traffic and tolling risks (if there is tolling), as under an AP. Unlike an AP, it also faces a payment risk, as payments to the Developer are not fixed or indexed to inflation, as in an AP, but vary according to traffic. These risks depend on the facility's nature and a large Agency may be able to absorb these risks, especially if the expected payment profile of a shadow toll - lower payments in early years and higher in later years, as traffic grows - benefits their constrained budget in the short-term and they are able to adequately manage the longer-term liabilities.
In the I-595 case (see Appendix I.6), the shadow toll option was judged to have higher financing costs compared to an AP based on Developer's perceptions of the higher risk in the payment structure. During the period that the I-595 procurement was conducted, 2007-2009, Developers' perception of magnitude of traffic and revenue risk increased, and hence so did the premium required for shadow toll or full revenue risk transfer mechanisms. Some Developers and Lenders may perceive the traffic risk as high as the traffic and revenue risk of a full risk sharing toll road, which would mean that the financeability of a shadow toll concession is similar to that of a toll concession.
Implementing a shadow toll scheme should be no more difficult to implement than a P3 toll concession. A key component of such agreements is reliable traffic counting and electronic tolling technology, both of which are well-tested in the U.S. Inexperience with shadow tolling may pose minor challenges to U.S. Agencies and their advisors in drafting contracts. More importantly, as with MRG and CFS, Agencies may have challenges in accurately predicting their payment obligations and incorporating these obligations in their budgets.
Under a regulated return mechanism (RRM), the Developer benefits from different possible remedies to ensure that its originally proposed IRR will be realized. These include:
Figure 8 presents a case in which the Developer's IRR is rebalanced by increasing the toll tariffs above the regulated price or by receiving an ongoing government subsidy.
Figure 8: Illustration of Regulated Return Mechanism
View larger version of Figure 8
As discussed, VfM is about efficient risk allocation. In principle, a RRM could cover any risk whenever a Developer's target returns are under pressure. For example, the Developer could be compensated if O&M costs are higher than expected or if the facility's completion is delayed. However, this would contradict the logic of efficient risk transfer and VfM, as there are many risks that are better managed by the Developer than the Agency. If, on the other hand, RRM is used specifically to transfer revenue risk to the Agency this could create VfM.
The RRM fiscal impact depends on the rebalancing mechanism. In the case of a contract extension, the effects are the same as under PVR. If higher tolls are imposed on users, and assuming that the demand elasticity is sufficiently low, there is no Agency impact.
A subsidy, which can be one-off or continuous, will have a direct impact on the Agency's budget. Due to the uncertain nature of the subsidy, both in terms of timing and level, the Agency may find it difficult to value this contingent liability. Furthermore, this liability depends not only on actual revenues, but also on the performance of the Developer.
Furthermore, if a one-off subsidy compensates a Developer, the impact would be of such magnitude that it may create payment difficulties for the Agency. In Colombia and Brazil, provisions in the concession agreement provided for such "true up" payments - payments that result in the Developer's returns equaling those that they originally bid - after five or more years. As a result, the payment would either be 1) too late to cover short-term liquidity issues or 2) could be such a large amount that the Agency would have a problem paying it from constrained budgets, both of which manifested themselves in Latin America. To avoid this, Agencies could fund a reserve fund to reflect the expected liabilities. However, this would require significant budget discipline and an accurate accounting of contingent liabilities.
As for fiscal impacts, financeability will depend on the adjustment mechanism used to achieve the target IRR. If the concession is extended, this provides significant protection for Lenders and Developers. However, Lenders would have to be comfortable with the refinancing risk.
If toll rates can be raised or the project benefits from a government subsidy, this again provides protection for Lenders and Developers. However, if this compensation happens only after a number of years, the project may already be facing short-term liquidity issues, as happened in Latin America. As long as Developers and Lenders are comfortable with restructuring the debt when required, RRM should provide sufficient protection against revenue risk.
RRMs can be difficult to implement due to monitoring issues, as acknowledged by academic, Agency, and Lender Respondents. In order to regulate a Developer's return, Agencies must monitor capital costs and timing of expenses. However, if no adjustments for costs, financing, and timing (including delayed completion) are allowed, these challenges are reduced. Nevertheless, RRMs still require regular updating of the project's financial model to calculate Developer's returns in order to determine compensation. Furthermore, depending on the compensation mechanism, elasticity of demand may limit the potential for additional revenues.
The Latin American experience with RRMs has been problematic because of Agency-Developer disputes over IRR calculations, such as the Mexico Fumisa Airport P3 (Moody's Investor Services, 2012). In some examples, when the realized IRR is very low, the Agency may need to extend the contract for many more years than policymakers are comfortable with, creating a "never ending concession" risk if no cap on the concession term is defined.
The mechanisms discussed so far have mainly focused on revenues. However, there are also other ways to share risks. For example, if Lenders were willing to offer flexible financing terms, this could help alleviate some of the financial impacts of revenue uncertainty. Outside the U.S., publicly-owned lending institutions or development banks often provide this type of debt. In the U.S., the USDOT's TIFIA program plays such a role as do state infrastructure banks on a smaller scale. These lending instruments can work in conjunction with the other risk sharing mechanisms. The following TIFIA provisions help reduce cash flow pressures in early years:
The TIFIA statute requires that the senior debt in the transaction be rated investment grade, and some rating agencies treat TIFIA debt as if it were investment grade. Therefore, Developers perceive TIFIA as closer to senior debt than deeply subordinated debt that takes equity-like risks. This means that there are limits to TIFIA's ability to absorb revenue risk.
The USDOT's TIFIA program only receives debt service payments and does not participate in any "upside" of the project other than benefitting from early repayments in the event that revenues are higher than expected. However, TIFIA's scheduled/mandatory debt service approach, which allows for interest capitalization in the first years of the project after substantial completion and flexible repayment minimizes cash flow requirements in the early years.
In general, post MAP-21 requirements have become stricter reflecting: 1) the overall credit market's stricter lending following the global financial crisis, and 2) specifically TIFIA's experience with the financial difficulties of P3 projects to which it has lent, such as the South Bay Expressway and the Pocahontas Parkway.
Developer Respondents also suggested that TIFIA or Agencies offer revenue risk insurance that would cover some or all of expected toll risks. Developers would pay a fee based on the amount of the coverage, similar to bond insurance or bank letters of credit. Generally, these products are not available today or are available for a much smaller set of projects than before the global financial crisis. It would appear that the MRG as practiced in Chile comes close to this product. The TIFIA program does offer a similar product in the form of a credit guarantee. However, Agencies and Developers have overwhelmingly preferred TIFIA's direct loans, since they are simpler to administer and less expensive than a credit guarantee, the latter necessitating an outside lending bank with additional fees.
Furthermore, before the global financial crisis some commercial banks operating in the U.S. provided early period relief in an "A Loan/B Loan structure," which may make some debt service payments flexible and subject to "cash sweeps." Since then, commercial banks have adopted more conservative lending practices. As they do not offer long tenors like PABs or TIFIA, they are not competitive in financing most revenue risk projects in the current market.
Internationally, there are more examples of innovative finance being used to directly address revenue risk. For example, in Australia, the Agency providing a subordinated loan to a P3 project has the right to receive "promissory notes" once certain ROR targets are met, resulting in future cash repayments on those notes (PricewaterhouseCoopers, 2011, p.13). In other words, subordinate debt providers share in the project's upside. These notes appear to be somewhat similar in purpose to revenue sharing mechanisms of current U.S. toll roads, in which Developers and Agencies share revenues at pre-established payment bands.
Spain has a subordinated program in which interest rates may change based on revenue levels, with higher-than-anticipated revenues resulting in higher interest costs (Sanchez-Solino, A. and J. M. Vasallo, 2006). This has a similar intent, but differs from the approach in the Australian promissory note program. Furthermore, the Spanish approach effectively addresses downside risk as interest rates are reduced when traffic is low, hence offsetting some of the revenue reductions. Figure 9 shows the Spanish example and the program's impact on debt service.
Figure 9: Illustration of Alternative Subordinate Debt Mechanism
View larger version of Figure 9
Given the wide variety of innovative financing solutions, the evaluation below will focus on mechanisms that absorb revenue uncertainty similar to TIFIA's mandatory/scheduled debt service and Spain's variable interest approach.
As in the case of the other mechanisms, a transfer of revenue risk from Developer to Agency or public debt provider should create VfM if the Developer is less well-positioned to manage revenue risk than the public Agency. However, as financing solutions can only absorb a small portion of the revenue uncertainty, it is likely that innovative finance by itself will not result in a very significant revenue risk transfer. Those innovative financing mechanisms that do not focus on revenue risk, yet provide protection for all project risks, contradict the P3 philosophy of efficient risk transfer and VfM.
In the case of TIFIA's mandatory/scheduled debt service approach, there is no direct cost to TIFIA for accepting later repayment as the Developer will be expected to pay interest on the outstanding balance. However, with more debt still outstanding, TIFIA's exposure will be higher for a longer time period. Furthermore, TIFIA's interest rates are significantly below market interest rates, which means that TIFIA is subsidizing the project. The value of this annual subsidy can be estimated by taking the difference between the market interest rate and the TIFIA interest rate and multiplying it by the outstanding balance. As this amount can be estimated at financial close, this subsidy can be seen as a direct liability to TIFIA.
In the case of Spain's variable interest rate approach, the lending agency effectively accepts waiving a part of the interest payments if revenues are lower than expected, which in itself is a subsidy. As the interest rate is now tied to traffic, this means that the lending agencies face a contingent liability, making it difficult to value.
TIFIA loans play a key role in U.S. P3 projects. Without this source of subsidized financing, projects would be more expensive or may simply not be feasible. The flexible TIFIA repayment conditions also help absorb revenue uncertainty. In the I-77 managed lanes project, scheduled interest payments can be delayed if traffic conditions are lower than expected. Given the uncertain nature of managed lanes revenues, this project would probably have struggled to obtain financing without TIFIA's participation.
Variable interest rates could also help absorb revenue risk. In the most extreme case, interest rates on the tranche provided by the Agency could fall to zero if revenues are much lower than expected. If such conditions were to be acceptable, it would provide significant protection to other Lenders as more cash flows would be available to service their debt, hence improving the overall financeability. The Agency could of course include a claw back clause to ensure that the Developer doesn't unfairly benefit from its lenient lending conditions.
Although P3 financing is highly complex and requires significantly more legal work than a conventionally procured project, the above innovative finance solutions do not pose any particular difficulty for implementation. As P3 transaction documents are drafted and negotiated on a case-by-case basis, some Agencies may include these or other financial innovations, which may involve minimal additional costs and complexity. However, some innovations may require legislative changes.
2 In a zero coupon bond or capital appreciation bond, in the municipal market, interest payments are capitalized until the final year of maturity, providing interest debt relief for the early years in which revenues are uncertain. Such instruments are not always available in the market and their amounts may be limited.
