UK: Building Flexibility. New delivery models for public infrastructure projects (Continued)

Given the plethora of delivery models available the question arises: how can the right model be selected from the available alternatives? On what basis should that decision be taken?

Infrastructure development is difficult because typically a decision must be taken now, which will have serious implications over a long period of time. This is why, when deciding on an appropriate model, it is vital to consider how certain the public sector can be about its infrastructure and service requirements. Certainty is crucial because without it, it is difficult to achieve a fair price on contracts, or to ensure that the infrastructure will continue to meet needs in the future.

In the case of existing assets, the key issue is the level of knowledge the public sector has about the condition of those assets and therefore the extent of work needed to meet future asset and service requirements. This issue can partly be remedied by maintaining adequate data about the condition of assets and/or undertaking the necessary surveys before contractual negotiations begin. However the risk of latent defects remains a challenging one, requiring careful consideration.

In the case of new assets, the key issue is how certain the public sector can be about the nature of the infrastructure and services it will need. The public sector need to attempt to assess their own level of confidence about their future asset and service requirements. Important questions to consider are:

• How confident can we be now about the type of infrastructure and services we need over the next 5, 10, 15, 20 years?

• How likely is it that the needs of citizens in this area will change?

• How likely is significant policy change?

• How easy is it to specify what we will need?

• When will advances in technology make these assets obsolete? There will always be uncertainties in these areas, which should not necessarily be a reason for inaction. However, the key thing is to assess how serious the uncertainty is and to what extent it is possible to adequately foresee the changes that are likely to be required.

A high level of certainty would mean that the public sector knows with confidence either the condition of the assets and/or the future asset and service requirements at a detailed level. In this case the main options are a Private Developer Scheme (PDS), a PFI, or a conventional procurement. A PDS will work best where the assets have a high residual value because they have multiple alternative uses such as office accommodation. It will not work well if the assets have little or no residual value, for example a section of motorway. On the other hand, a PFI will work best where residual value is low and the project size is large (greater than £20 million). This is because of PFI’s high procurement costs – which are only acceptable if the project is large enough to absorb them. If risks lie mostly at the construction phase and it is financially feasible to do so (e.g. balance sheet status will be unaffected), it is worth considering a ‘De-Risked PFI’. Where the project is small (less than £20 million), the main option is conventional procurement, or variants of it, because of its lower procurement costs.

A medium level of certainty would mean that the public sector knows the kind of infrastructure it needs, but is less certain about the timing and exact extent of work it wishes to undertake. In such a case, the main options to explore are LEP/LIFT, Integrator or Competitive Partnership. All these models enable the public sector to avoid several lengthy procurement processes, while ensuring that successive waves of work can be delivered quickly and without excessive cost.

Broadly, the Integrator will work best for projects where the overall programme of work is heterogeneous (i.e. different types of infrastructure), whereas the Competitive Partnership and LEP/LIFT models will work best where the elements of work are more homogeneous. The Competitive Partnership model will be best if the work programme can easily be separated into discrete and similar elements, whereas the LEP/LIFT model will work best where the project needs to be tackled as a whole.

A low level of certainty would mean that the public sector is unsure about the infrastructure it needs (or even what is possible), let alone when or how it wishes to have it delivered. In such a case the Alliancing or Incremental Partnership models are worth considering.

The choice is fundamentally driven by the nature of the infrastructure requirement. Where the infrastructure is large, indivisible and complex (e.g. a new, expensive item of defence technology), then Alliancing is worth exploring. However, if the infrastructure is smaller, more divisible (can be acquired in discrete phases) and simpler, then it can be procured using an agreement of the kind involved in an Incremental Partnership.

To reiterate, this is for indicative guidance only, and any final decision should only be taken after a full options appraisal, specific to the particular circumstances has been carried out.

The United Kingdom has been a path-breaker in the development of innovative delivery models for infrastructure projects. Both the PFI and LEP/LIFT models have many merits and will remain useful models in the right circumstances. But they have their limitations and, quite understandably, many sectors are now experimenting with new or hybrid models that are more suitable for smaller projects or where uncertainty is greater.

What this report argues is that this innovation must continue – that there can be no ‘one size fits all’ in infrastructure development. Instead what is needed is to make a principled and informed choice based on an awareness of the full range of delivery models and the conditions in which they are successful.

For central government and policy-makers, this means supporting the exploration and use of alternative models of the kind outlined here. When new programmes of infrastructure development are designed, the model should be selected from the full range of possibilities and on the basis of its likelihood of delivering the optimal mix of government’s objectives.

For local government and public sector delivery organisations, this report presents some new models for consideration in situations in which the conventional or existing models are unsatisfactory. By making the best use of the full-range of models that are available the public sector can maximise the likelihood of achieving its infrastructure objectives in the future.

This appendix provides a fuller explanation and discussion of the PFI and LEP/LIFT models and their relative strengths and weaknesses. It is intended for those readers who are unfamiliar with the details of the models and require further explanation and elaboration of their advantages and disadvantages.

Over the past few years government has used the PFI to deliver a large number of major infrastructure projects. Since 1997, 725 PFI schemes have been signed with a total capital value of £46 billion.¹² The department with the largest number of schemes has been health (excluding devolved administrations), but the department whose schemes have the largest value is Transport due to the London Underground PPP which accounts for £17 billion.

The PFI is now a well established procurement model that has the potential to provide strong incentives for delivery on time and to budget, while enabling the public sector to spread the cost of the investment over a 25-30 year period. Government supports the use of the PFI in local government through the allocation of ‘PFI credits’ which effectively provide additional funds for the capital element of a PFI project.

Typically, although not always, the PFI involves a long-term contract between the public sector and a private sector Special Purpose Vehicle (SPV) to deliver infrastructure and services in exchange for an annual, performance-related payment. Payment for the infrastructure does not begin until it has been commissioned and meets the required specification. The SPV is funded through a combination of bank debt (typically 90 percent) and equity (typically 10 percent). The structure of the PFI contracts and the extent of risk transfer they can involve mean that in many cases they can be offbalance sheet for the public sector.

It is difficult to carry out a systematic review of the performance of PFI relative to other procurement approaches. There have been few comparable schemes that have been carried out under both PFI and conventional procurement for which adequate data is available. Furthermore the PFI is still in its relative infancy, with very few schemes being more than a few years into their operational phase. Despite these difficulties, the experience so far and available evidence suggest that it has a number of merits.

Conventional procurements require the public sector to provide significant up front capital when the benefits are delayed and uncertain. As a consequence many otherwise viable projects are rejected because of the potential mismatch of cost and benefit.

Under the PFI, the public sector is able to spread the cost of investment over time, rather than having to provide large up-front capital investments. This means that the timing of the costs of infrastructure schemes can be better aligned with the timing of the benefits that accrue from those schemes. This has enabled projects that would otherwise not have been approved due to either the uncertainty or immediate cost of investment, to go ahead. In addition this is one of the major reasons why the government makes PFI credits available to local government for the purpose of infrastructure development.

Greater predictability of cost and timeliness. Partly as a result of the fact that payments are better aligned to the delivery of project objectives, projects delivered under the PFI are more likely to be delivered on time and on budget compared to conventional procurements. A 2003 NAO Report found that while 73 percent of non-PFI construction projects were over budget and 70 percent delivered late, the figures for PFI were just 22 percent and 24 percent, respectively. There has been no similar, more recent review of performance.

Focus on value for money over the lifetime of the asset. Under a conventional procurement the focus is on providing value for money in the short-term (often two to three years) i.e. during the design and construction phase. However the consequence is that sometimes short-term savings are delivered that result in higher costs over the lifetime of the asset (e.g. the use of cheaper building techniques which require higher maintenance).

In a PFI scheme this is less likely as the contractor has responsibilities to meet required levels of maintenance and operational requirements over the lifetime of the infrastructure. There are some indications this has led to good quality design and construction: an NAO report included the results of a survey that suggested that over half of managers surveyed considered the design and build quality to be good or very good.¹⁴

Strong performance-related incentives. Under the PFI the performance of the contractors is strongly related to their achievement of key project outcomes: notably the delivery and availability of the infrastructure on time and to budget, and in the operational phase, the achievement of required levels of service. The unitary charge is automatically adjusted to penalise poor performance. This should increase the likelihood of high performance in the operational phase. Although evidence on this is modest, there is some emerging evidence of good feedback from user satisfaction assessments.

Reduced impact on public borrowing. The UK government is subject to both external and self-imposed limits on the amount of borrowing it is prepared to undertake. This includes the condition imposed by the European Union as part of the Stability and Growth Pact that current account deficit should not exceed three percent of GDP. In addition, the current government has sets itself the objective of limiting net debt to a ‘stable and prudent level, below 40 percent of GDP’. This means that government must consider the borrowing implications of all decisions on infrastructure investment.

Under conventional procurement approaches, new infrastructure will almost invariably be on the balance sheet of the public sector, as key risks are retained by the public sector. In a PFI scheme, many risks are transferred to the private sector, often enabling the scheme to be off-balance sheet for the public sector.¹⁵ This means that the scheme will not count against the net debt totals, enabling some projects to go ahead that would otherwise not be viable.

As experience with the PFI has grown its limitations are now becoming relatively well understood.

High cost. PFI schemes can be more costly than other procurements due to three main factors: the cost of procurement, the level of risk transfer, and the cost of private finance.

The PFI procurement process is long and costly. HM Treasury has estimated that the average procurement time is 22 months.¹⁶ The length of the contracts and relative certainty that PFI schemes aim to give the public sector over costs mean that a great deal of pressure is placed on both parties to negotiate a contract that is acceptable in the long-term. It takes a long time to agree the risk transfers, payments and terms that are acceptable to both parties – imposing considerable legal and due diligence costs on both the contractors and client side.

In addition, PFI contracts seek to transfer more risk to the private sector partners than under other models. This is particularly so for PFI schemes that are off-balance sheet. However private sector partners will expect the contracts to cover the financial risks they face. The cost of risk transfer is particularly high where the condition of the assets is uncertain, or where the future asset and service requirements are unclear. In these situations the private sector will expect to receive a ‘risk premium’ to make the project viable. In some cases this exceeds the public sector’s estimate of the cost of the risk materialising.

Lastly, because PFI schemes involve private finance and it is more expensive to borrow money privately than publicly, this constitutes an additional cost under most PFI projects. For PFI projects to be value for money these costs must be exceeded by the savings.

Inflexibility. PFI contracts typically include detailed specification of the outputs required and the penalties for failing to meet them.¹⁷ If the public sector wants to change its service requirements at a later stage, this is usually possible, but it may be costly. This is because for small changes, the strong position of the incumbent partner usually makes the competitive pressure fairly weak.

Consequently there is a range of situations in which the PFI is an unsuitable delivery model. It is particularly unsuitable for small projects because of the high procurement costs, or projects for which the lead time is short, because of the lengthy procurement time. It is also unsuitable where there is a high level of uncertainty over the condition of existing assets, or future asset and service requirements. Such uncertainties may mean that the public sector finds itself with a contract that is unsuitable in the long-term or poor value because the contractor has had to add a significant premium in the price to cover the extra risks.

This explains why government guidance introduced in 2003 recommended that the PFI not be used for small projects (below £20 million), or for IT projects (where uncertainty about future needs is too great).¹⁸ This is also why PFI is generally thought to be less suitable for most upgrades or refurbishments (rather than new builds) – because of the risk of latent defects.¹⁹

The emergence of hybrid models: LEP and LIFT 

For several reasons government has developed an alternative delivery model for investment in the primary care and schools estates.

In the case of the primary care estate, individual projects tend to be small and the estate has required both upgrade as well as new build work. This has meant that neither single nor successive PFI schemes would be optimal because of the procurement costs and difficulties in transferring risk. Simultaneously, there was a desire for a vehicle that could enable the public sector to commission successive phases of work with a single partner who could provide clear lines of responsibility. As a result, the Department of Health developed the LIFT model.

A similar model was subsequently developed for the Building Schools for the Future programme, for similar although slightly different reasons. In the case of schools, PFI has already been used extensively, and largely successfully. However the remaining problem was that PFI remained largely unsuitable for school upgrades or refurbishments due to the issue of latent defects. At the same time, there was a similar desire for a vehicle that could enable a series of waves of investment in the schools estate without the need for multiple separate procurements. Accordingly, the LEP model is being used.

Although the two approaches are not identical they are sufficiently similar to be treated as variants of a single model. The basis for the model is a joint venture company that is majority-owned by a private sector partner. In the case of schools, the Local Education Partnership (LEP) has equity investment from the local authority, Partnerships for Schools, and the successful private sector partner. In the case of primary care, Primary Care Trusts (PCTs), or the Strategic Health Authority, together with Partnerships for Health and the strategic partner invest in the joint venture – the Local Investment Finance Trust Company (LIFTco).

The private sector partner is selected through a competitive process that includes a fixed price for some of the initial work to be carried out. So-called ‘soft FM’ services are excluded in the case of LIFT but not in LEP.²⁰ The contract is for 20 years in case of LIFT and ten years in the case of LEP. Subsequent phases of work are commissioned by the public sector partner, but (typically) carried out by the strategic partner using the first phase of work as a benchmark to the appropriateness of future costs. In the first five years of the contract, the value for money of work is assessed using the first phase of work as a benchmark. Thereafter the public sector has the right to market test proposals if it is unconvinced of value for money.

The joint venture can utilise both conventional procurement and PFI mechanisms according to the work needed.

The model is still in its infancy. In LIFT, 51 projects over four waves have been approved but currently only 31 buildings are open to patients.²¹ In the Building Schools for the Future programme, three waves of investment have been launched, covering 38 local authorities. However no projects have yet reached financial close.

It is therefore too early to evaluate its performance fully. However there are several reasons for concluding that it ought to offer some significant advantages over conventional procurement or PFI in certain circumstances.

Reduced procurement cost and time over the project life. While the initial procurement can be lengthy, over the lifetime of the project the overall procurement costs should be lower than under a number of separate PFI schemes. This is because once the initial procurement has been completed, successive phases of work often do not require an EU procurement process, or if they do, they are likely to take less time. For the same reasons this should enable more rapid delivery of new infrastructure.

Flexibility over programme of delivery. The structure enables phases of work to be commissioned as and when the public sector decides on the type and scale of work required. In this respect it is more flexible than a conventional PFI scheme.

Potential for continuous improvement. The appointment of a single strategic partner (LEP/LIFTco) who commissions all phases of work should enable continuous improvement to occur. This is because the strategic partner can learn lessons from the early phases and incorporate them into subsequent elements of the work.

Ability to retain influence over strategic direction of work. The joint venture arrangements should enable the public sector to retain influence over the strategic direction of development, without having to take responsibility for delivery.

Conflict of interest. There is a potential conflict of interest for the private sector partner who is expected to provide value for money for the public sector, while simultaneously seeking to extract maximum return through the delivery of most or all of the required work.

The effectiveness of the LEP/LIFT model depends on the use of benchmarking as a means of ensuring the value for money of subsequent phases of work. To this end, Partnerships for Schools is investing in the development of a comprehensive benchmarking database which it hopes will provide robust data to ensure the value for money of LEP proposals.²² The intention is that the public sector can use this data to establish helpful parameters of cost and quality for their schemes.

While benchmarking of this kind can be a useful tool, in some circumstances it does not offer sufficient assurance that value for money will be achieved. Sometimes subsequent phases of work are significantly different from the first – making the benchmarks from the first scheme inadequate. There are always numerous possible explanations for why a proposal might depart from the established benchmark,²³ and site specific costs cannot be benchmarked. This means that in practice it may be difficult for the public sector to know whether or not successive proposals constitute value for money.

Strong disincentives to utilise alternative providers. Although the LEP/LIFT model gives the public sector the right to market test and use alternative providers if it is not convinced of the value for money of the LEP/LIFTcos’ proposals, in practice there are very strong disincentives for it to do this. If an alternative provider is appointed, the public sector immediately loses the central benefit of the model, which is to commission work through a single partner, with a single point of responsibility. As a result even if the public sector is doubtful of the value for money of proposals in practice it is unlikely to use alternative providers.

Accordingly, the LEP/LIFT model works best for projects where benchmarking is likely to be an effective tool for ensuring value for money. These are projects whose elements are relatively homogeneous in nature – i.e. of the same kind, subject to the same expectations on cost and quality. Where projects are composed of many different elements, and vary considerably from one place to another, the LEP/LIFT model will be less suitable.