UK: Building Flexibility. New Delivery Models for Public Infrastructure Projects

Infrastructure development remains at the heart of the modernisation of UK public services. From the upgrade of secondary schools to the development of new IT systems, the public sector has rarely been under such pressure to deliver effective and efficient improvements in its infrastructure.

The public sector has always worked with private sector partners to help deliver its infrastructure requirements – typically through so-called ‘conventional procurement’. Recent years have seen the private sector play a new role in the Private Finance Initiative (PFI), and the partnership models of Local Improvement Finance Trusts (LIFT) and Local Education Partnerships (LEP). However, these schemes continue to be subject to considerable discussion and debate.

This report focuses on assessing alternatives to conventional procurement. It attempts to evaluate the PFI and LEP/LIFT models and describe the conditions in which they are likely to work best. It proposes five new delivery models that Deloitte believes have the potential to offer improved outcomes in situations where neither PFI nor LEP/LIFT is suitable.

We believe that the PFI and LEP/LIFT models will go on being good options in many situations, but that they need to be supplemented by alternative approaches if the public sector is to achieve its infrastructure objectives. Our aim is to raise the awareness of the public sector to a broader range of delivery models than those commonly considered and to provide a rough and ready guide to their selection.

Existing models such as the PFI and LEP/LIFT are not sufficient to meet the public sector’s infrastructure needs. While both the PFI and LEP/LIFT models work well in many circumstances there is a range of situations for which they are unsuitable. The PFI works best for large projects in conditions of relative certainty. The LEP/LIFT model can be more flexible than the PFI, but concerns have been expressed regarding its value for money owing to a lack of competitive pressure. Importantly, there are significant risks from using either of these models where uncertainty is serious and cannot easily be reduced.

New policies and likely changes to accounting rules are putting increasing pressure on the use of these models. The introduction of market mechanisms is bringing new uncertainties into the provision of health and education services, while possible changes in the accounting treatment of PFI may mean that some off-balance sheet schemes come back on-balance sheet. In addition, the nature of the infrastructure challenge is evolving. In many sectors there is now a need to deliver upgrades (rather than new build), and infrastructure solutions that are subject to considerable future risks and uncertainties (e.g. in technology or waste). These are areas in which the PFI and LEP/LIFT are either untested or unsuitable.

New models are emerging which can help to meet some of these challenges. In some circumstances the affordability of PFI schemes is being improved through a more selective transfer of risk to the private sector partner. The LEP/LIFT model is being adapted through a stricter separation of the project development and delivery roles. It can also be modified by using a structure that puts private sector partners in competition with each other. These are referred to as the ‘Integrator’ or ‘Competitive Partnership’ models. Where uncertainty over the future needs is serious and unavoidable the ‘Alliancing’ or ‘Incremental Partnership’ models are being used to enable projects to go forward.

The level of certainty the public sector has about its infrastructure and service requirements should be a key determinant of the choice of model. A high level of certainty suggests that the main options are a Private Developer Scheme, PFI or conventional procurement. The Integrator, LEP/LIFT or Competitive Partnership models should be considered where there is more limited certainty. A low level of certainty suggests the use of Alliancing or Incremental Partnership.

By applying a broader range of models in the right circumstances, public sector organisations can improve the likelihood of achieving their infrastructure objectives. The PFI and LEP/LIFT models will continue to be good options in many situations. But if public sector organisations are to meet the challenges posed by ongoing uncertainties amid a changing policy environment, increasingly they will need to utilise a broader range of delivery models. If they do so, they will increase the likelihood of meeting their infrastructure objectives in the future.

The United Kingdom has been at the forefront in developing innovative models for the delivery of public infrastructure projects. Since its emergence in the early 1990s the PFI has evolved and adapted to meet changing needs and requirements. More recently, the LEP/LIFT model was developed to help meet challenges for which PFI was seen to be unsuitable. This section explains why the PFI and LEP/LIFT models will not be sufficient to meet the public sector’s needs, and therefore why this innovation must continue. A description of the PFI and LEP/LIFT models together with a fuller description of their merits and demerits can be found in the Appendix.

The PFI has many advantages. It enables public sector organisations to spread the cost of infrastructure investment over the lifetime of the asset, avoiding some of the uncertainties present in conventional procurement. In addition, because the payment mechanism is aligned with project objectives, PFI offers improved likelihood of projects being on time, to budget, and meeting the original specifications. It also encourages a focus on value for money over the lifetime of the asset. Finally, because of the high level of risk transfer typically involved, PFI projects can be off-balance sheet, which can be desirable from the perspective of departmental budgets and economic indicators.

However it also has some disadvantages that make it unsuitable in some situations. The PFI can be a high cost option owing to the costs of procurement, risk transfer, and private finance. The cost of procurement is high, because of the length of time taken to reach financial close – which is typically two years or more. The cost of risk transfer can be higher than the cost of retaining the risk in the public sector, particularly where uncertainties mean that the private sector partner expects a significant premium to compensate for the possibility of downside risk. Finally, the cost of private finance is typically higher than the cost of public borrowing. For PFI projects to be value for money these costs must be exceeded by the savings.

PFI can also be inflexible due to the length of contracts and difficulty of changing requirements. Although it is possible to build in opportunities to change service requirements during the contract term, this will be cost effective only where the scope of such change is reasonably predictable. This is because, once appointed, the PFI contractor is in a strong position in any future negotiations over contractual changes. Furthermore, in practice the public sector is often reluctant to appoint a new provider because of the consequent complexity and potential for conflict that could be generated.

Due to the cost of procurement, PFI is rarely considered for small projects (less than £20 million in value)¹. It works best where the public sector can have a relatively high degree of certainty about future service requirements. Then, the services can be properly specified, and the public sector can achieve the benefits of whole-life costing and strong performance incentives. The public sector can also avoid the costs of risk transfer in conditions of uncertainty and future contractual change. On the other hand, where certainty is lower – perhaps because of a lack of knowledge of the condition of the assets, or of future service requirements – the PFI’s disadvantages begin to outweigh its advantages. (See Figure 2).

The LEP and LIFT models were developed partly to facilitate infrastructure development in some of the situations where a conventional PFI scheme might be less suitable. In particular, the aim was to enable a combination of new build and upgrade work to be carried out in successive phases without the need for several lengthy procurements. In addition, there was a need for a model that could allow work to begin where there was continuing uncertainty about the exact timing and scale of the work to be carried out over the lifetime of the project.

It is early days in the experience with the LEP/LIFT model; however it is possible to identify its principal advantages and disadvantages.

It has the advantages of lower procurement costs over the project life (because only one EU procurement is typically required), greater flexibility over programme delivery (because work can be commissioned in separate phases) and an ability for the public sector to continue to influence the direction of investment (because of the joint venture structure). The retention of a single strategic partner throughout the project has the potential to enable continuous improvement through successive phases of work. In addition, there are potentially significant advantages from the commercial input of a private sector partner early in the planning phase.

However, it also has some disadvantages. In the case of both LIFT and LEP there is a potential conflict of interest for the private sector partner. This is because the partner has both the role of seeking to ensure maximum value for the public sector but also of frequently carrying out most of the work.

The effectiveness of the LEP/LIFT model depends on the use of benchmarking² 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 that it hopes will provide robust data to ensure the value for money of LEP proposals.³ 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. The first phase of work may not provide a sufficiently clear benchmark for further work if subsequent phases are significantly different from the first. Site specific costs and the existing condition of the infrastructure cannot be benchmarked. Furthermore, there are always numerous possible explanations for why a proposal might depart from the established benchmark.⁴ This means that in practice it may be difficult for the public sector to know whether or not proposals constitute value for money.

Furthermore, although the public sector retains the right to use alternative providers if it is unconvinced of the value for money of the LEP/LIFTco proposals, in practice it is unlikely to do this. This is because if it did so it would immediately lose the primary benefit of the model – the presence of a single strategic partner to manage the programme of work and provide continuous improvement.

In any event, benchmarking is unlikely to offer the same kind of savings that direct competition provides. Data from the Building Cost Information Survey (BCIS) shows that procurements that are carried out without direct competition are on average 11-13 percent more expensive.⁵

Accordingly, the LEP/LIFT model will work best for projects where benchmarking is an effective tool for ensuring value for money. These are projects whose elements are relatively homogeneous in nature – i.e. of the same kind and 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, benchmarking will be less effective.

Because the model offers greater flexibility and lower overall procurement cost, it is also suitable for projects in which the infrastructure needs and objectives are relatively clear, but where there is a degree of uncertainty about the timing and scale of the work that will be carried out.

Both the PFI and LEP/LIFT models require a degree of certainty about the type of infrastructure or services needed. In the case of PFI, the public sector needs a high degree of certainty about the desired output specification; in the case of LEP/LIFT, the public sector needs relative certainty about the kind of infrastructure required, but can be less certain about the scale or timing of the work.

But neither scheme will work well where there are serious and irresolvable uncertainties over the required infrastructure and services or the cost of meeting those requirements. Such uncertainties might be present as a result of: latent defects (flaws in the infrastructure that are not apparent until work commences), policy changes (which imply a change in service requirements), demand risks (resulting from e.g. the introduction of user choice), changes in public needs, or rapid changes in technology.

The potential implications of applying these models where such uncertainties are serious are one or more of the following:

• The private sector partner requires a significant risk premium to compensate for the uncertainty.

• The public sector needs to change its specifications post-contract, leading to extra costs.

• The public sector is stuck with a contract that fails to meet the public’s needs.

• In extreme circumstances contracts have to be terminated at considerable public expense.

As it stands, one of the central issues is therefore how to develop infrastructure solutions in situations where such uncertainties are present.

While these models are least applicable in conditions of serious uncertainty, it is precisely here where many of the remaining infrastructure challenges lie.

For instance, there is a significant need to provide upgrade and refurbishment of existing schools, health care facilities, social housing and road and rail networks. But refurbishment and upgrade projects are less suitable for the PFI because they often involve a risk of latent defects.

In addition, in some areas, the infrastructure requirements are subject to significant technological uncertainties. In the waste sector, there are uncertainties about the optimal technical solution to enable local authorities to meet the EU Landfill Directive. In many areas of defence, it is often unclear what kind of infrastructure solution is technologically possible – let alone whether it would be affordable or value for money if it were possible. Information Technology (IT) projects are subject to well known uncertainties and complexities that make them unsuitable for delivery under a PFI model.

While the limitations of existing models are becoming more apparent, changes in the policy and financial framework are also putting pressure on their use.

For example, in health the government is introducing patient choice and Payment by Results in the secondary care system. This will mean that money will increasingly follow the choices of patients who will seek the best combination of care, convenience, and waiting time. As a result, hospitals may need to be able to adapt their infrastructure and services at relatively short notice in order to meet changing public needs and demands. While this is an issue that could potentially affect all hospitals, regardless of how they have been procured, it may be particularly challenging for hospitals procured under PFI. In the case of PFI hospitals, any future changes to the infrastructure or service requirements would have to be negotiated with the private sector contractor. Given the strong position of the private contractor (see above) this could be difficult and costly. Partly as a consequence of the perceived expense and inflexibility of PFI, the future of several schemes is now in question.⁶

Similar consequences may result for schools from the implementation of the Schools White Paper.⁷ If implemented, this may increase pressure on schools to adapt themselves to the changing demands of parents. Again, this would tend to make the long-term service contracts of the kind used under PFI increasingly unsuitable.

Lastly, there are forthcoming changes in the accounting treatment of PFI transactions that may make them less attractive to the public sector. The accounting treatment of PFI transactions and many elements of the LEP/LIFT schemes can often lead to the relevant assets and liabilities being ‘off-balance sheet’ as regards the public sector. This can be seen as helpful with regard to the impact on the capital control budget and also the Net Debt economic control total. However, over the next few years UK Generally Accepted Accounting Principles (GAAP) are expected to converge with the rather different International Financial Reporting Standards. As a result, in the medium-term it may be challenging to achieve off-balance sheet treatment for such projects^.8

So, for all these reasons there is an increasing need to think beyond the current delivery models to identify hybrid or amended models which can help meet these challenges.

More specifically, it would be desirable to:

• Improve the value for money and flexibility of PFI where possible.

• Develop models that have similar advantages to LEP/LIFT, but with more competitive pressure.

• Identify models that could be suitable in conditions of serious uncertainty – where neither PFI nor LEP/LIFT work well.

This section describes five new models that have the potential to help public sector organisations begin to meet these challenges. The ‘De-risked PFI’ has the potential to improve value for money in some situations. The ‘Integrator’ and ‘Competitive Partnership’ models have many of the advantages of the LEP/LIFT model, but provide more ongoing competitive pressure. Lastly, the ‘Alliancing’, and ‘Incremental Partnership’ models can be used in situations where uncertainty is great.

In a conventional PFI scheme, significant risks are transferred to the private sector partner. These include the risks of cost over-runs and delays in construction as well as meeting the operational requirements. The decision as to whether or not to transfer particular risks depends not only on who is best able to manage that risk but also on the financial implications of doing so. As a result, in some situations, value for money can be improved by reducing the overall risk transferred relative to a conventional PFI scheme.

One model that has been used in the concession for the DLR Woolwich Extension is to reduce the risk to the contractors by underwriting some of the financial risk during the operational phase (see Sidebar). In this case the contractor adds most value in the construction phase – so those risks remain transferred.

Docklands Light Railway Limited (DLRL) is extending its network from near London City Airport to Woolwich Arsenal. The project cost is £240 million and involves a new twin bore tunnel under the Thames as well as creating a new DLR station at Woolwich Arsenal.

The primary delivery vehicle is a PFI (Design, Build, Finance and Maintain), but with ‘de-risking’ during the operational phase. The greatest risks in the project are during the construction phase (particularly the creation of the tunnels) – and it is here where DLRL sees the benefits of significant risk transfer. However, DLRL believe that once the infrastructure exists, the costs of transferring significant operational risk outweigh the benefits (because the actual benefit of risk transfer is dwarfed by the cost of private finance incurred). So they are guaranteeing 75 percent of the unitary payment after the infrastructure has been constructed and in satisfactory operation for two years (with 25 percent still at risk). With public sector agencies typically having good credit ratings (Transport for London currently has an AA credit rating), this is enabling the project to go ahead at substantially lower cost. The project will in any event be on-balance sheet, and therefore this enables DLRL to significantly improve affordability.

The advantages and disadvantages of the model are described in Figure 5.

Accordingly, this ‘de-risking’ approach is likely to be a good option where:

• The risks at the operational stage are perceived to be minor relative to risks during construction, or to decline swiftly after delivery.

• The project would in any event be on-balance sheet.

• The public sector body can afford to meet the extra costs that are incurred if the relevant risks materialise and/or the public sector is sufficiently confident that the risks will not materialise.

As discussed above, the LEP/LIFT model has important benefits but may not offer optimal competitive pressure over the lifetime of the project. One way of improving the overall competitive pressure is to retain the basic structure but to separate the role of the strategic partner (the ‘Integrator’) from that of direct delivery (design, construction and so on). We refer to this as the ‘Integrator’ model.

The distinctive element of the Integrator model is the inclusion of a private sector partner who has responsibility for project development (taking significant project risk) but has a less direct role in service provision. The Integrator is rewarded according to overall project outcomes wherever that is possible (with penalties for lateness, cost over-runs, poor quality etc). The Integrator then undertakes to arrange the necessary delivery functions, potentially using a variety of procurement options, including PFI and conventional procurement, as appropriate.

In some cases the Integrator is barred from being involved in direct delivery at all. In other cases, the Integrator is appointed to carry out the first phase of work, or specified works, but then barred from carrying out subsequent phases (see MoDEL below). The purpose of this prohibition is to remove the potential for conflict of interest between achieving best value for the public sector and maximising returns through the supply chain.

The advantages and disadvantages are set out in Figure 7.

The London Borough of Greenwich has been using an Integrator model to deliver their Building Schools for the Future programme. The Integrator Model has also been used in the recent MoDEL project for the Ministry of Defence (MoD).

The Integrator model works best in similar circumstances to the LEP/LIFT model, but particularly where the work is heterogeneous (composed of different elements whose costs are uncertain), making ongoing competitive pressure over all elements of the work vital.

Project MoDEL involves the consolidation of seven MoD sites to a single location in RAF Northolt in London. The consolidation will relocate up to 3,500 military and civilian personnel into modern fit-for-purpose accommodation. This requires creating new facilities to house the personnel, enabling the transfer of staff to the new accommodation, and selling sites that are surplus to requirements. The consolidation budget is around £200 million and it will take up to five years to complete.

Neither a conventional procurement nor PFI would be suitable because of the uncertainty over the costs, work required, and future values of the sites for disposal. Instead, the approach taken has been to appoint an Integrator (termed the ‘Prime Plus Contractor’ in this case) who will take the principal risk on project delivery. All elements of the work are subject to open competition. The integrator competes for the initial specified works, and all currently unspecified works are then competitively procured by the Integrator. The Integrator is barred from carrying out works that are unspecified at financial close.

The Integrator is primarily remunerated through net disposal receipts and is responsible for ensuring that timescales and quality standards are met. This ensures that there are strong incentives for keeping costs down, consistent with other project objectives.

This model offers a potential approach for rationalising and upgrading estates within a single contractual framework, with a partner who can take significant risks on behalf of the public sector.

The project will reach financial close in the summer of 2006. 

In the conventional LEP/LIFT model, if the public sector is dissatisfied with the perceived value for money of future works, it has the option of having that work carried out by another party (after a specified time). But if it did so it would lose the benefits of having a single agency having responsibility for delivery in that area. On the other hand if it does not have the option of going elsewhere the competitive pressure on the private sector partner can be weak.

To overcome this disadvantage, an alternative approach is the Competitive Partnership model in which several LEPs/LIFTcos are appointed separately, in competition with each other. The contracts allow the public sector to reallocate projects between them at a later date as overall performance becomes clear. The public sector can also use the cost and quality benchmarks of each LEP/LIFTco as a benchmark for the performance of the others.

One natural way of doing this would be to split the responsibilities by area. For instance, in schools, this could mean appointing two or more LEPs, each provisionally responsible for a different area. After the first phase of work, the public sector (for instance, local authority) would be able to compare the timeliness, quality and cost of the work carried out by each LEP. It would then be able to shift responsibilities away from the weaker performers towards the stronger ones as appropriate. A significant advantage here is that the work of each LEP can be used as a benchmark for the others – given that they are likely to be working under very similar conditions.

This model is currently being explored in both the schools sector and in housing and regeneration (Oldham and Rochdale). It has also been used successfully in the private sector by the Bank of America. Bank of America has approximately 4800 branches in 22 states. It procured outsourcing contracts to provide maintenance to their branches and allocated a geographical region to each. The contractors were required to achieve continuous improvement of their service, year-on-year and were incentivised through regular review and reallocation of the size of their respective areas.⁹ 

The advantages and disadvantages of this approach are set out in Figure 9.

This model would work best where the required work was of sufficient scale to make a division of this kind possible, and where the work was reasonably homogeneous and could be divided into sensible and comparable segments (e.g. by area).

All the models discussed so far require at least moderate certainty – about the infrastructure and service specifications; in the case of PFI, a high degree of certainty, in the case of LEP/LIFT, Integrator, and Competitive Partnership models relative certainty about the infrastructure needs, if not the exact timing and scale of the work. 

But what if it is not possible to have even that degree of certainty, perhaps because future developments are extremely uncertain, or because it is unclear what sort of infrastructure is technologically possible? One possible approach is the use of Alliancing. Alliancing is a term used to describe delivery models in which the focus is on encouraging collaboration through the use of payment mechanisms that ensure that the interests of all parties are aligned with the project objectives. The aim is to avoid the adversarial relationships and acrimony that sometimes characterise more conventional delivery models, and instead seek to ensure that all parties work together collaboratively for the good of the project. Common features of Alliancing agreements are:

• A focus on specifying key project outcomes, rather than inputs, or processes.

• The use of integrated project teams (which include representatives from the public sector and all relevant contractors).

• Gain/painshare arrangements, which mean that all parties (public and private) have an interest in ensuring that all elements of the project are a success, and an incentive continuously to strive for out-performance, not simply minimum expectations.

• Simplified arrangements for managing risk, for example a single insurance policy.

Alliancing has been used in the oil and gas sectors. More recently it has been used by the MoD (see below), and by BAA in the management of its Terminal 5 project. It has also been used for several public infrastructure projects in Australia and New Zealand. ¹⁰

The Alliancing model can take a number of forms. Sometimes the output specification and delivery is carried out by a single party. In other cases, it is possible to run separate competitive processes for the output specification and the delivery phase (depending on the extent of competition in the market place, and the synergy between output specification and delivery responsibilities).

The advantages and disadvantages are described in Figure 10.

Alliancing is not a panacea for infrastructure projects, but can be successful where:

• There is irresolvable uncertainty about the nature of the infrastructure or services required to meet project objectives (e.g. technological risks).

• The infrastructure is large, indivisible (cannot be easily separated into discrete elements), and complex.

• The public sector is an experienced manager of infrastructure projects and prepared and able to retain all significant project risks.

The purpose of this project is to replace the current Invincible class of aircraft carriers (which date from the Cold War period) with two larger vessels that could support a more powerful air group. The approximate budget was £3 billion.

The first phase of the project – the Assessment phase – involved the development of several carrier designs. Two contractors – BAE Systems and Thales were appointed.

The decision as to how to procure this resulted in the selection of an Alliancing model in which KBR act as the ally, with BAE Systems and Thales contracted to produce the necessary components. The Alliancing framework aims to ensure that the ally is rewarded for project outcomes, and all parties are incentivised to economise wherever possible.
Source: MoD Major Projects Report 2005, National Audit Office, 2005.

There are other situations in which although there is a high degree of uncertainty, Alliancing would not be suitable because the required infrastructure is smaller and can be built up in successive waves.

Under the Incremental Partnership approach the public sector enters into a framework agreement with a private sector partner who procures the necessary infrastructure and services on behalf of the public sector. The public sector can ‘call off’ specific projects as its requirements become clear. The private sector partner competitively procures the services and infrastructure from sub-contractors but retains overall responsibility for service levels as assessed against clear performance measures. There is no exclusivity for the private sector partner – the public sector retains the right to use alternative providers if it wishes.

In 2002, the London Borough of Greenwich was faced with reletting its ICT facilities management contract and addressing the increasing requirements for e-enabled service provision. The Council considered but rejected the option of a ‘big bang’ approach which would introduce a new package of infrastructure and services with a private sector partner. This was due to considerable uncertainty about the service requirements, the rapidity with which ICT systems can become obsolete, and the potential risk of entering into a major financial commitment with a private sector partner. It was also wary of entering into a relationship with a private sector partner that was not truly independent and would be constrained to offer its own service solutions.

As an alternative the Council contracted with Deloitte to act as the ‘service integrator’, under a framework agreement lasting for five years. The Council is then able to ‘call off’ individual projects as and when it wishes, without any long-term commitment. Deloitte then sub-contracts on behalf of the Council for the provision of new hardware and software as appropriate – using its procurement expertise to negotiate the best possible deals. The Council is not bound to use Deloitte’s services, but it has continued to see value in the relationship – having now worked on over 40 individual project briefs. Should the contract end, Deloitte’s licences and contracts would simply revert to Greenwich.

Source: Deloitte Research, 2006.

This avoids the weaknesses associated with ‘big bang’, large scale contracts that are difficult to reverse and which require a long-term commitment from both parties. This ‘Incremental Partnership’ model has been used successfully with Greenwich Council for the provision of new Information and Communication Technology (ICT) infrastructure (see sidebar).

Note that although this has similarities to the ‘Integrator’ model (discussed above) it differs in some crucial respects. In an Incremental Partnership (but not with an Integrator), the private sector partner has no exclusivity, the term of the agreement is much shorter (the partner is not necessarily responsible for the assets over their lifecycle) and there is typically no need for project finance structures such as an SPV (as the private sector partner does not provide capital).

The advantages and disadvantages are described in Figure 11.

Accordingly, the Incremental Partnership model is potentially beneficial where the infrastructure in question is divisible (can be built up in discrete phases) where it has a short life, and where there are ongoing uncertainties about the requirements. It is also suitable for smaller organisations that lack the internal capacity to manage the infrastructure procurement themselves. In these situations, the high level of flexibility of the model makes it attractive – allowing the public sector to build up the infrastructure it requires, without the need for a long-term agreement.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.