A public-private partnership (PPP) is a project undertaken by a private-sector entity to build and maintain an asset for a public-sector entity in return for a stream of payments over a significant period. Several PPPs have recently been undertaken in New Zealand, involving motorways, prisons, and schools, with a collective construction cost of almost $4b.1 The current government seems enthusiastic about continuing with PPPs and has announced in-principle support for several more involving motorways in Northland. These have a collective construction cost of $10b.
PPPs have superior incentives
PPPs have many possible advantages compared to the “traditional” model, in which the public sector instead contracts for the construction of the asset and its maintenance.
One advantage is that the superior incentives faced by the private sector lead to lower construction and operating costs.
Another is that combining responsibility for both construction and maintenance should ensure a better trade-off between these two facets of the job — avoiding cost cutting in construction that gives rise to higher maintenance costs that are more than offsetting, and avoiding gold plating in construction that does not provide more than offsetting benefits in lower maintenance costs.
A further possible advantage is risk transfer: large construction projects typically run over budget. Well-written and well-enforced PPP contracts can transfer blowout risk from the public to the private sector.
PPPs face some disadvantages
PPPs also have some disadvantages, of which the principal one seems to be higher financing costs.
One aspect of this is the inferior liquidity of private-sector borrowing compared to government borrowing. As of 30 May 2024, the Government could borrow for five years at 4.71%. By contrast, I understand that a private-sector entity (a “special purpose vehicle” or SPV) borrowing for a PPP for the same term would pay about 6.75%. This is a 2% differential. Part of this differential will be due to the higher default risk of the SPV compared to the Government (with BBB and AAA credit ratings respectively). The empirical evidence from traded bonds is that a credit rating differential of this type for five-year debt would raise the interest rate by about 0.50%. So, inferior liquidity on the SPV debt raises the cost of borrowing by about 1.5% (2% - 0.5%). This occurs because the debt of an individual SPV (up to $1b for the PPPs to date) is a tiny fraction of the aggregate government debt (about $200b). This “liquidity cost” does not arise under the traditional model.
The second aspect of higher financing costs is the default risk premium of about 0.50%. Part of this is the “pure” default premium, arising when there are no bankruptcy costs from defaults.2 This is not a higher cost of private-sector financing per se, because the higher rate paid if default does not occur is simply compensation for the lower rate effectively paid in the event of default. In effect, this is an arrangement to share the downside between the equity and debt holders. However, in setting the PPP payments required by the SPV, standard practice is to act as if the higher borrowing rate always applies because it is too difficult to estimate and then deduct the part due to this pure default risk premium. The result is that the PPP payments, which will be made to the SPV even if it defaults on its debts, overcompensate the SPV for its debt costs. This problem does not arise under the traditional model.
The third aspect of the higher financing costs is the part of the default premium on the SPV debt that is due to bankruptcy costs. Again, this cost does not arise under the traditional model.
The fourth aspect of the higher financing costs is that part of the cost of equity capital that is due to the low liquidity of the equity of an SPV. Again, this cost does not arise under the traditional model.
What is the WACC?
A proper evaluation of a PPP should consider all of these pros and cons. In view of the possibility that different government entities considering PPPs might evaluate them in different ways, The Treasury promulgated PPP Guidelines to ensure consistent evaluation in 2015. Inter alia, these Guidelines state that government entities considering PPPs should present value the proposed PPP payments of a private provider by using the weighted-average cost of capital (WACC) of the private provider, to compare with the present value of the costs of the traditional model, with the latter being the construction cost plus the operating costs present valued using the private provider’s WACC.3 However this requirement has the effect of completely eliminating the higher financing costs of the SPV, and therefore biasing the decision strongly in favour of PPPs.
To illustrate this point, suppose that a government entity (“the Ministry”) requires an asset to be constructed, which can be done now for $1000m, and with annual operating costs of $15m for 30 years, for certain. One option would be for the Ministry to select a construction company, who would be paid $1000m, and to select another company who would be paid $15m per year to deal with the operating costs (the traditional model). The alternative (PPP) would be to select an SPV to arrange both of these, who would then charge the Ministry a fixed sum per year for 30 years, after which full responsibility for the asset would revert to the Ministry. So as to focus upon the implications of the Guidelines, I assume that the SPV would face the same construction and operating costs as the Ministry. The SPV would then have to raise $1000m in capital. I also assume that the SPV would pay no tax on its income and therefore competitive neutrality with the public sector (which does not pay income tax) would prevail. In this situation the SPV’s cost of capital would presumably be almost entirely debt, and I therefore assume 100% default-free debt.
Suppose the SPV (correctly) estimates its default-free cost of capital (all five-year debt) at 6.25% (6.75% less 0.50% for the default risk premium). To cover its costs, the annual payment A required from the Ministry would have to be such that A net of the operating costs, and present valued at 6.25%, would have to equal the amount borrowed:
The solution is A = $90m, i.e., the SPV would require PPP payments from the Ministry of $90m per year over 30 years.
The Ministry would then have to evaluate this proposal. For the purposes of comparison, suppose that the $1000m paid by the Ministry under the Traditional Model was financed by borrowing that amount and repaying it in equal amounts per year over 30 years. Since the Ministry would pay the government borrowing rate of 4.71%, that amount per year would be B = $63m,4 to which is added the operating cost of $15m to yield $78m, i.e., the Ministry would pay $78m per year for 30 years under the traditional model. So, the traditional model would be superior to the PPP by $12m per year, and this arises because the government borrowing rate is lower than the SPV’s borrowing rate, and this is because of the illiquidity premium on the SPV bonds, i.e., SPVs are far less efficient at raising debt capital.
Each of the payment streams ($78m for the traditional model and $90m for the PPP) can be expressed as an equivalent amount that must be borrowed today in order to cover the payments (an “equivalent loan”). The equivalent loan for PPP payments is an amount borrowed now by the Ministry (at its borrowing rate of 4.71%) that would incur costs of $90m per year for 30 years, and this is $1430m as follows:5
By contrast the equivalent loan for traditional model payments of $78m per year would be $1238m.6
The equivalent loan for the traditional model is also the present value of the operating costs, discounted using the government’s borrowing rate of 4.71%, plus the construction cost of $1000m, which is also $1238m as follows:
The equivalent loan for the traditional model is then $192m less than the PPP. So, the PPP is inferior by $192m, which is almost 20% of the construction cost. So, whether one looks at the annual payments or the equivalent loans, the traditional model is superior.
By contrast, the Guidelines would require the Ministry to present value the proposed PPP payments using the Weighted Average Cost of Capital of the SPV, which is 6.25%. The result would be a present value of $1200m as follows:
The Guidelines would also require the same discount rate to be used in evaluating the traditional model, which would also be $1200m as follows:
The PPP proposal would then seem to be as good as the Ministry undertaking the project itself, when it is actually worse by $192m. If the Ministry adopted the PPP, it would then have overpaid by $192m and would be completely unaware of it. No matter how large the SPV’s cost of capital is, the two options will still seem equally good. For example, if the SPV’s cost of capital rose to 10% then, following equation (1), the required annual PPP payment would rise to $121m. Following equation (2), the present value of these payments discounted using the government borrowing rate of 4.71% would rise to $1923m whilst the present value of the costs under the traditional model would remain $1238m as in equation (3). So, the advantage of the traditional model would grow. However, if the new PPP payments of $121m and the costs under the traditional model were both present valued using the SPV’s cost of capital of 10%, as mandated in the Guidelines, then both would be $1140m following equations (4) and (5). So, the increased cost of capital for the SPV makes the traditional model even better whilst the approach mandated by the Guidelines would still wrongly suggest that the PPP was as good as the traditional model.
These examples assume that the SPV would face the same construction and operating costs as the government entity. Suppose now that the construction costs for the SPV are 10% lower at $900m. With substitution of $900m for $1000m in equation (1), the annual PPP payment required by the SPV falls from $90m to $82m. Substitution of this $82m into equation (2) yields a present value on the PPP payments of $1303m, which is still larger than the $1238m for the traditional model in equation (3). So, The traditional model is still superior, because the lower construction costs of an SPV are insufficient to counteract its higher financing costs. However, the results arising from following the Guidelines in equations (4) and (5) are that the present value of the PPP payment of $82m per year now seems to be $1100m whilst that for the traditional model still seems to be $1200m. So, the PPP seems to be superior, because the lower construction costs of the SPV are recognized but not the higher financing costs. So, again, the wrong conclusion is drawn.
Summary: PPPs face significantly higher financing costs
In summary, PPPs have many potential advantages, but they also have the very significant disadvantage of higher financing costs. A proper evaluation of a potential PPP must consider all costs and benefits. However the Guidelines promulgated by The Treasury would completely mask the higher financing costs and therefore strongly bias the decision in favour of a PPP. More extensive analysis that I’ve conducted suggests that the disadvantage from the higher financing costs is about 30% of construction cost, with the majority of this coming from the inferior liquidity of SPV debt compared to government bonds.7 Across the $4b in PPPs to date, if the advantages of these PPPs are less than 30% of construction cost, the net disadvantage of these PPPs will be up to $1200m. In respect of the proposed PPPs with construction costs of about $10b, if their advantages are less than 30%, the net disadvantage will be up to $3b.
The good: Well-designed and executed PPPs have superior incentives, which can encourage lower construction and maintenance costs.
The bad: PPPs face higher financing costs.
The ugly: Treasury guidelines bias financial evaluations in favour of PPPs.
By Martin Lally, Capital Financial Consultants Ltd.
These are the Transmission Gully motorway ($1.25b), the Puhoi to Warkworth motorway ($1b), Waikeria Prison ($900m), the Auckland South Corrections Facility ($300m), and school projects PPP1 ($60m), PPP2 ($300m) and PPP3 ($70m).
Bankruptcy costs take many forms, including the costs associated with lawyers, accountants and liquidators in the event of default, and the costs arising from writing and enforcing debt covenants.
The relevant document is Public Private Partnership Program: A Guide for Public Sector Entities, particularly pp. 27-31. Until a few weeks ago the document appeared at https://www.treasury.govt.nz/sites/default/files/2015-10/ppp-public-sector-comparator-sep15.pdf. My efforts at contacting The Treasury to determine what happened to that webpage have not so far been successful. However, you can view the document here, courtesy of the Internet Archive.
Responsibility for the PPP guidelines has been passed to the NZ Infrastructure Commission, who have produced a draft new version. The wording at points relevant to this post is essentially identical.
Equation for B:
PPPs are very similar to financing leases, and there is an extensive financial economics literature on the evaluation of financing leases, summarized in books such as Principles of Corporate Finance, 10th edition, R. Brealey, S. Myers and F. Allen (Irwin, 2011, Chapter 25.4). This literature shows that the appropriate discount rate on a fixed stream of payments by any entity (akin to debt payments) is that entity’s borrowing rate. In addition, the mathematics used here is exactly the same as that used by any bank or mortgage broker to determine how much one could borrow at a given annual payment for 30 years.
Equation for EL:
See Lally, Martin. (2024). Public-private Partnerships and the New Zealand Guidelines (June 20, 2024). Available at SSRN: https://ssrn.com/abstract=4871326 or http://dx.doi.org/10.2139/ssrn.4871326.
John, thanks again for your comments and I read Ian's submission with great interest. Ian's most compelling point seems to be that, with an Emissions Trading Scheme, WCC's actions to reduce its emissions through this project will likely not reduce NZ's emissions and therefore the project is not only pointless in that sense but extraordinarily expensive relative to the current carbon price. We see these kinds of alleged emissions reduction projects in numerous other places within the NZ public sector. It is tempting to attribute them to ignorance amongst their proponents about how the Emissions Trading Scheme works. Perhaps the proponents of these schemes do understand how the Emissions Trading Scheme works but extract utility from being seen to be saving the planet by people who themselves do not understand how the Scheme works? Cycle ways seem to be the classic example.
Re: "Well-written and well-enforced PPP contracts can transfer blowout risk from the public to the private sector". Have we had well-written and well-enforced PPP contracts to date? Is it not the case that the public sector had to pick up a big chunk of the cost overruns for Transmission Gully and the Puhoi-Warkworth extension? And don't private sector companies have incentives to submit unrealistically low bids to secure public sector contracts and then bully the public sector once the point of no return is reached to secure additional financial concessions from the government? And does government have the staff expertise to negotiate and manage PPP contracts adequately?