International Journal of Project Management 20 (2002) 107118

www.elsevier.com/locate/ijproman

Evaluating the risks of public private partnerships for infrastructure projects

Darrin Grimsey a, Mervyn K. Lewis b,*
b a PricewaterhouseCoopers, Spring Street, GPO Box 1331L, Melbourne, Victoria, Australia School of International Business, University of South Australia, North Terrace, GPO Box 2471, Adelaide, Australia

Received 14 October 1999; received in revised form 16 May 2000; accepted 9 July 2000

Abstract In many countries, limitations upon the public funds available for infrastructure have led governments to invite private sector entities to enter into long-term contractual agreements for the nancing, construction and/or operation of capital intensive projects. For the public procurer, there is an obvious need to ensure that value-for-money has been achieved. To the project sponsors, such ventures are characterised by low equity in the project vehicle and a reliance on direct revenues to cover operating and capital costs, and service debt nance provided by banks and other nanciers. Risk evaluation is complex, requiring the analysis of risk from the dierent perspectives of the public and private sector entities. This paper analyses the principles involved, drawing on practical experience of evaluating such projects to present a framework for assessing the risks, and using as illustration a case study of a waste water treatment facility in Scotland which is typical of most PPP projects. # 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Public private partnerships; Infrastructure; Project nance; Risk analysis

1. Introduction For most of the post-war period, government has been the principal provider of infrastructure (at least outside of the United States). Over the last decade, that position has begun to change. Faced with pressure to reduce public sector debt and, at the same time, expand and improve public facilities, governments have looked to private sector nance, and have invited private sector entities to enter into long-term contractual agreements which may take the form of construction or management of public sector infrastructure facilities by the private sector entity, or the provision of services (using infrastructure facilities) by the private sector entity to the community on behalf of a public sector body. These arrangements often take the form of a buildoperate-transfer (BOT) arrangement [1]. The acronym BOT was rst used in the early eighties by Turkey's Prime Minister Targut Ozal [2]. However, the concept itself can be traced back to Hong Kong in the late fties when a privatised vehicle tunnel was rst talked about, and if regarded as a form of concession or franchise
* Corresponding author. Tel.: +61-8-8302-0536; fax: +61-8-83020512. E-mail address: mervyn.lewis@unisa.edu.au (M.K. Lewis). 0263-7863/01/$22.00 # 2001 Elsevier Science Ltd. All rights reserved. PII: S0263-7863(00)00040-5

arrangement has even earlier origins. According to Monod [3], the rst ``concession'' was granted in 1782 to Perrier in France and concerned water distribution, which in the context of this paper seems entirely appropriate. In Australia, public/private sector infrastructure arrangements date back to the bicentennial year 1988. In the UK, the Private Finance Initiative (PFI) was introduced by the Conservative Government in November 1992, widening its privatisation and contracting out policies to incorporate the provision of infrastructure and public services by a hybrid approach of combined public and private sector funding [4]. Procurement was subsequently redened as the provision of services rather than the ownership of assets, with partnerships sought on a range of projects and PFI promoted as a preferred procurement method a position which did not change with the incoming Labour administration which in 1997 built on the initiative with Public Private Partnerships [5]. The European Commission, through its grant mechanism, is encouraging PPPs with projects so far in Portugal, Italy, Netherlands, Greece and Ireland [6]. The techniques have been adapted to promote investment both in local authority services and in infrastructure projects more generally elsewhere [7,8]. This paper analyzes the risks of PPP arrangements from the perspectives of the various parties. The character of

108

D. Grimsey, M.K. Lewis / International Journal of Project Management 20 (2002) 107118

infrastructure investments and the nature of PPPs shape the riskiness of any individual project. Thus we begin by examining the nature of infrastructure and how PPPs are structured. 2. What is infrastructure? Infrastructure is easier to recognize than dene. Investment in infrastructure is thought to provide ``basic services to industry and households'' [9], ``key inputs into the economy'' [10], and ``a crucial input to economic activity and growth'' [7], although what is ``basic'', ``key'' and ``crucial'' varies from country to country and from one time to another (steel production was once regarded as essential infrastructure). Recently, the activities regarded as infrastructure investment include:
. Energy (power generation and supply); . Transport (toll roads, light rail systems, bridges and tunnels); . Water (sewerage, waste water treatment and water supply); . Telecommunications (telephones); . Social infrastructure (hospitals, prisons, courts, museums, schools and Government accommodation).

. Externalities, whereby benets and costs are conferred upon those not a party to the transaction (e.g. spillovers); . Natural monopolies, for which scale economies make it ecient to have only one provider (for example, of an electricity grid).

These share with other types of xed investment (such as property development, oce construction [11]) a number of common characteristics:
. Duration (infrastructure is long-lived, and has a long gestation process); . Illiquid (the lumpiness and indivisibility of infrastructure projects makes for a limited secondary market); . Capital intensive (projects are large scale and highly geared); . Valuation (projects are dicult to value because of taxation and pricing rules and embedded options [12,13] and guarantees).

The trend away from public to private provision of infrastructure has been underpinned by a marked change in thinking and practice on these matters. There has been the perception, for example, that a move from `taxpayer pays' to `user pays' (i.e. from ability-to-pay to the benet principle) in the provision of infrastructure services (water, power) is likely to be associated with a better economic use of the services. Many industries considered to be natural monopolies, e.g. electricity generation and telecommunications, have been broken up geographically into dierent regional rms or, with deregulation, separated into competitive (or potentially -vis those sectors that remain competitive) sectors vis-a natural monopolies (the distinction between power supply and high-voltage transmission, and between railway operation and rail track services). In those activities which have natural monopoly characteristics, substitution of price-cap regulation for rate-of-return regulation (i.e. xing of maximum prices rather than the mark up over costs) has created strong incentives to reduce costs, while third party access to certain facilities that are not economic to duplicate has widened competition in the upstream and downstream markets served by the facilities. All of this has laid the groundwork for PPP arrangements. 3. PPP arrangements Confusion sometimes exists between `infrastructure nancing' and `infrastructure investment'. The former can arise from the privatisation of existing facilities, whereas infrastructure investment involves the development, operation and ownership either by the private sector alone or in a joint venture between government and the private sector entity. The distinction is analogous to buying an existing oce block, already fully let, as opposed to developing a new site the attendant risks are obviously quite dierent in the two cases. Accordingly, PPPs can be dened as agreements where public sector bodies enter into long-term contractual agreements with private sector entities for the construction or management of public sector infrastructure facilities by the private sector entity, or the provision of services (using infrastructure facilities) by the private sector entity to the community on behalf of a public sector entity. They can take many forms and may incorporate some or all of the following features [14]:

The result is that evaluation of the projects is a complex and specialised activity. The litmus test used to be that infrastructure had to be provided by government-owned enterprises (the predominant approach in Europe) or by privately owned utilities subject to rate of return regulation (the approach in much of the United States). This conviction derived from a number of inherent features, such as the existence of:
. Network services, providing integrative activities which bind economic activity together; . Public goods, from which it is dicult (and perhaps not desirable) to exclude non-payers (the non-excludability principle);

D. Grimsey, M.K. Lewis / International Journal of Project Management 20 (2002) 107118

109

. The public sector entity transfers facilities controlled by it to the private sector entity (with or without payment in return) usually for the term of the arrangement; . The private sector entity builds, extends or renovates a facility; . The public sector entity species the operating features of the facility; . Services are provided by the private sector entity using the facility for a dened period of time (usually with restrictions on operations and pricing); and . The private sector entity agrees to transfer the facility to the public sector (with or without payment) at the end of the arrangement.

A classic example of a BOT arrangement is the third Dartford Crossing of the River Thames linking two stretches of the M25 motorway circling London, to be operated (with virtually guaranteed toll income) by the vehicle company for up to 20 years, following which the facility will revert to the UK government. In Australia, projects such as the Sydney Harbour Tunnel and the City Link (linked motorways) project in Melbourne are also BOT arrangements. With a BOO (build-own-operate) project, the private sector entity nances, builds, owns and operates an infrastructure facility eectively in perpetuity. An example comes from the water treatment plants serving parts of South Australia. These facilities are nanced, designed, built and operated by a private sector rm to process raw water, provided by the public sector entity, into ltered water which is then returned to the public sector utility for delivery to consumers. Although governments have been motivated into entering into PPP arrangements by the desire to reduce debt (and contain taxation), another consideration has been the benets of sharing nancial risks and rewards between public and private sector bodies. We now examine these risks. 4. Risks Much of the risk of a PPP project comes from the complexity of the arrangement itself in terms of documentation, nancing, taxation, technical details, subagreements etc involved in a major infrastructure venture, while the nature of the risk alters over the duration of the project. For example, the construction phase of the project will give rise to dierent risks from those during the operating phase. Some idea of the complexity is given in Fig. 1. The project concerned is the Almond Valley and Seaeld (AV&S) project involving the construction and operation of a water treatment facility for East of Scotland Water (ESW), with a services contract over 30 years

between ESW (the public procurer) and Stirling Water. There is also a separate operating agreement between Stirling Water and Thames Water, the private sector operator of the works. The project reached nancial close in March 1999 and the gure sets out the links and key contractual arrangements between the parties involved. From the viewpoint of the public procurer, there is an obvious need to ensure that money has been spent economically, eciently and eectively. At its simplest, the Government seeks to utilise private sector nance in the provision of public sector infrastructure and services and thereby achieve value-for-money. Value-for-money, dened as the eective use of public funds on a capital project, can come from private sector innovation and skills in asset design, construction techniques and operational practices, and also from transferring key risks in design, construction delays, cost overruns and nance and insurance to private sector entities for them to manage. However, in some cases, the emphasis on risk transfer can be misleading as value-for-money requires equitable allocation of risk between the public and private sector partners, and there may be an inherent conict between the public sector's need to demonstrate the value-for-money versus the private sector's need for robust revenue streams to support the nancing arrangements [15]. From the perspective of the project sponsors, PPP (and PFI) is essentially project nancing, characterised by the formation of a highly-geared special purpose company for the project vehicle and consequently a reliance on direct revenues to pay for operating costs and cover debt nancing while giving the desired return on risk capital. Although there are exceptional cases such as the Hong Kong Harbour Tunnel which started making a prot four years after opening, these arrangements typically last for long periods and take a long time to generate a prot. It is therefore important to have a clear understanding of the principles of limitedrecourse nancing. PPP projects are viable only if a reliable, long-term revenue stream can be established. The risk that the predicted revenues do not materialise is the greatest risk to the commercial viability of a project. This risk is borne by those providing nance or nancial guarantees. Straight equity participation is generally low, only 5% of the total funding (4.95 million out of funding of 99 million) in the case of the AV&S project. This situation is more starkly illustrated by the designconstructmanagenance (DCMF) of a PFI private prison currently operating in Bridgend, South Wales which was funded with only 250,000 of equity constituting 0.3% of the total funding of 83.5 million. Subordinated debt is often regarded as the equivalent of equity, but this comprises only a further 14% of funds invested in the AV& S project.

110

D. Grimsey, M.K. Lewis / International Journal of Project Management 20 (2002) 107118

Fig. 1. Contractual arrangements in a PFI/PPP project.

D. Grimsey, M.K. Lewis / International Journal of Project Management 20 (2002) 107118

111

Typically, the providers of nance look to the cashow of the project as the source of funds for repayments. Financial security against the project company itself is not sought because the company usually has minimal assets and because the nancing is without recourse to the sponsor companies. However, performance guarantees are often made available by the sponsor companies in favour of the lenders. Thus the key principle for large PPP projects is to achieve a nancial structure with as little recourse as possible to the sponsors whilst at the same time providing sucient credit support so that the lenders are satised with the credit risks. What are the risks? At least nine risks face any infrastructure project [1619]:
. Technical risk, due to engineering and design failures; . Construction risk, because of faulty construction techniques and cost escalation and delays in construction; . Operating risk, due to higher operating costs and maintenance costs; . Revenue risk, e.g. due to trac shortfall or failure to extract resources, the volatility of prices and demand for products and services sold (e.g. minerals, oce space etc.) leading to revenue deciency; . Financial risks arising from inadequate hedging of revenue streams and nancing costs; . Force majeure risk, involving war and other calamities and acts of God. . Regulatory/political risks, due to legal changes and unsupportive government policies; . Environmental risks, because of adverse environmental impacts and hazards; . Project default, due to failure of the project from a combination of any of the above.

however, there is clearly the risk of losses arising from a changing political climate toward the provision of public services by the private sector, while the prices charged for many public sector services are politically sensitive and may be price-capped in some way. Nevertheless, in principle, the risks of PPP projects seem little dierent from those of other project nancing activities, and can be evaluated using much the same basic techniques. The critical question, as always, is whether revenue streams can cover operating costs, service debt nance and provide returns to risk capital. Consider the case of infrastructure in the form of a power plant. Sponsors of the power project borrow money to build a generation plant. The sponsors contract to supply power to utilities, projecting that the contract revenues will suce to pay debt service and generate prots. But risks abound. Will the plant actually be built on time? Will the plant work? And will the market value of the contracts enable participants to avoid an income shortfall? Can rates be raised to levels that more or less equal the utility's costs for providing electricity, an activity that has historically been regulated by government? None of these questions can sensibly be dodged or ignored in project evaluation. Ultimately, the `bottom line' (i.e. project default risk) is borne by the nanciers, and when considering this scenario the uncertainties concerning future cash ows can be thought of as falling into two categories: 1. Moderate (and perhaps not so moderate) deviations from estimated cash ow projections, due to uctuating prices, costs, timing delays, minor technical problems, etc. 2. Disasters to a project, due to a major cost overrun, downturn in the economy, change in legal rulings, alteration to the political climate, environmental disaster etc, which could lead to project failure and bankruptcy. The dierence here is not simply one of scale. Rather, the distinction we have in mind recalls that made by Frank Knight in his classic 1921 treatise [21]. Knight distinguished sharply between one type of situation, which he called ``risk,'' and another situation altogether, which he named ``uncertainty.'' In both cases, the actual future outcome is not predictable with certainty. But in the case of risk, the probabilities of the various future outcomes are known (either exactly mathematically, or from past experience of similar situations). In the case of uncertainty, the probabilities of the various future outcomes are merely ``wild guesses'' because the `instance' in question is so entirely unique that there are no others or not a sucient number to make it possible to tabulate enough like it to form a basis for any inference of value about any real probability in the case we are interested in. (p. 226).

Successful project design requires expert analysis of all of them and the design of contractual arrangements prior to competitive tendering that allocate risk burdens appropriately. For this purpose the risks can be broadly categorized as global or elemental [20]. Global risks are those risks that are normally allocated through the project agreement and typically include political, legal, commercial and environmental risks. Elemental risks are considered as those associated with the construction, operation, nance and revenue generation components of the project. Most of these risks are common to any project nancing activity, and apply with more or less force depending on the project concerned. With some PPP agreements, revenue risk and market risk might be low, indeed negligible. For example, the revenue from a toll bridge might be more assured than that of an oileld, while a private prison is likely to operate with a higher occupancy rate (e.g. 100%) than a luxury hotel! At the same time,

112

D. Grimsey, M.K. Lewis / International Journal of Project Management 20 (2002) 107118

Knight went on to argue that the main function of the `entrepreneur' is to bear the brunt of this uncertainty of the future. The prots of enterprise (`pure prots' over and above interest payments on debt capital and/or dividend payments to shareholders) is a reward for facing this uncertainty. However, the reality in the nancing of projects is often quite dierent to that envisaged by Knight. For example, developers who put up blocks of oces or ats for future sale commonly set up a new limited company for each new building. This company borrows money and/or sells equity shares. If the building fails to make a prot, this particular company goes bankrupt. But the sponsor may survive. The successful entrepreneur is often one who knows how to shift the burden of uncertainty onto others, in particular investors and/or creditors, in such a way that he himself will survive, waiting for a more propitious time for others to again be persuaded to take chances with their money. As Blatt [22] puts it, the prudent entrepreneur reacts to true uncertainty by attempting to make others bear the consequences. Clearly, an analysis of the nature of the risks [23 25] and who bears them is vital, and the evaluation of projects requires the use of several risk analysis techniques tailored to suit the interests of the various parties to the project. Nevertheless, the Knightian distinction remains. At least in theory, risk can be insured against, diversied, calculated with dierent probabilities, but true uncertainty or disaster scenarios are something else again. How are risks handled in practice? To answer this question we will look at risk analysis from the dierent perspectives of the parties to the AV&S project. 5. Risk analysis in practice The AV&S project constitutes a major initiative being undertaken by ESW to improve the quality of the water within the River Almond and the Firth of Forth and to implement an alternative to the current practice of disposal of sewage sludge at sea. The project is driven by ESW's requirements to comply with its obligations under the Urban Water Treatment (Scotland) Regulations 1994. The project is just one of around ten PFI schemes currently in procurement with the three Scottish water authorities in order to comply with new regulatory standards to be in force by the end of December 2000. ESW alone has indicated that capital investment over 5 years of 850 million is required to achieve the required standard in its region. Following a long procurement process of over 2 years, in March 1999 ESW entered into the Services Contract with Stirling Water pursuant to which Stirling Water agreed to upgrade and improve the treatment of waste

water from the feeding catchments of Almond Valley and Seaeld and to provide new arrangements as an alternative to the disposal of sewage sludge at sea. The main parties to the project and an indication of their connecting relationships are shown in Fig. 1, but in summary:
. ESW is the procuring entity and the authority responsible for providing water, collecting and treating waste water and disposing of sewage sludges within its region; . Stirling Water is the special purpose company set up for the purpose of nancing, constructing and operating the AV&S project; . Thames Water, MJ Gleeson and Montgomery Watson are the project sponsors providing risk capital and acting respectively as the operator, contractor and designer to Stirling Water; and . MBIA is providing the credit enhancement insurance policy so that the project can attain a credit rating of AAA. While institutional investors purchased the bonds, the MBIA policy unconditionally and irrevocably guarantees scheduled repayments of principal and interest under the bonds issued and therefore MBIA eectively assumed the role of senior lender.

Stirling Water funded the 99 million project from:

. 79.217 million of 27.5 year AAA rated credit enhanced bonds paying a xed coupon of 5.822%; . 14.835 million of subordinated debt provided by Thames Water. The subordinated debt has the same term as the bonds and a similar interest rate; and . 4.95 million of equity split between Thames Water (49%), MJ Gleeson (41%) and Montgomery Watson (10%). The internal rate of return on the equity was projected to be 16.56% nominal and 12.5% real at nancial close.

6. Project risks Stirling Water as the project vehicle has taken on certain construction and operating risks over the contract period but has allocated away contractually many of these risks to the construction contractor and the operator, as is usual under such arrangements. Stirling Water retained the risk that projected maintenance costs are exceeded, such as those arising from shorter than anticipated asset life spans, increased ination on specic items of plant and machinery or the requirement to carry out unexpected repairs. Once the plants are commissioned Stirling Water receives payment based on the volume of waste water treated. The payment mechanism is the key to the risk apportionment and is not like a normal PFI deal where

D. Grimsey, M.K. Lewis / International Journal of Project Management 20 (2002) 107118

113

the operator is paid on making the facility available. The contract is more akin to the UK's design, build, nance and operate (DBFO) programme for roads where shadow tolls are paid within payment bands (ie. dened levels of trac volume), a structure in fact devised by PricewaterhouseCoopers. The aim of such a structure is to allocate a sucient element of volume risk to the service company and also to limit the authority's exposure to an increase in payments arising from a greater than anticipated volume of waste water delivered. The AV&S payment bands are as follows:
. The rst band for waste water treatment is for between 0 and 90,000 cu m pa and pays 13.883 p. This band is designed to generate sucient revenue to cover xed operating and maintenance costs and outstanding amounts due under the bond issue; . The second band for waste water treatment is for between 90,000 cu m pa and 101,000 cu m pa and pays 9.2555 p. This band is designed to provide sucient revenue to meet payments under the subordinated debt; and . The third band for waste water treatment is for between 101,000 cu m pa and 107,000 cu m pa and pays 4.628 p. This band is designed to meet the shareholders' target investment returns. . The fourth band for waste water treatment is for ows in excess of 107,000 cu m pa and is free of charge. This band is designed to cap the payments to be made by ESW.

volume throughput or suering larger than expected deductions for poor performance would result in lower than forecast revenues, which would impact on the equity return and may adversely eect Stirling Water's ability to make repayments of subordinated debt and even coupons on the bonds. The proportion of the tari indexed to the retail prices index (RPI) is 65%, applied to all three revenue-generating bands. The remaining 35% of the tari is not subject to indexation but is xed for the contract term. Proportionally this division approximates to the percentage of the nominal cash ow used to service the bond nancing. However, it is possible that there will not be a precise match between Stirling Water's revenues and costs resulting in costs rising faster than revenue arising under the contract and this risk is borne by Stirling Water and its shareholders. The risk of any adverse changes in law is an important element and on the AV&S project the risk is shared between ESW and Stirling Water over the rst ten years of the contract. After this time, the risk is borne by ESW. 7. Risk analysis Having sketched the key risks inherent in the project it is important to look at the nature and quantum of risk from the dierent perspectives of the main parties to the project. As a guide to the methodology employed, Fig. 2 provides a ow chart of the analytical approach. For each of the three major groups of entities, it summarises their risk perspective, the key variables, the major risks they face, and the risk analysis which is appropriate. There are several risk analysis techniques available ranging in complexity from simple expected cost analysis through sensitivity analysis and on to more complex probabilistic techniques involving computer-aided statistical sampling. The technique that should be used, as indicated, is largely dependent on the exposure of the party to the risk and the nature of the return expected by the party. 7.1. Procurer ESW, as the procurer, was only interested in the expected costs in the form of the payments they would have to make under the contract. In accordance with Government guidelines the projected payments to each of the bidders were discounted at 6% real and summed to derive the net present value (NPV) of each of the bids. From ESW's perspective, the risk analysis centred on establishing equality of treatment between bidders for bid evaluation and to facilitate a comparison with the public sector comparator in order to demonstrate value-for-money in NPV terms. The value-for-money criterion should establish the best means for achieving

Payment above the third band eectively is capped thereby limiting ESW's exposure to payments under the contract, while there is a separate rate for the treatment and disposal of imported sludge. The contract provides for payment to be subject to performance-related adjustments geared to any breaches in environmental discharge consent standards set by SEPA. A major concern for Stirling Water is the impact of the quality of the waste water delivered to the works by ESW upon the operator's ability to treat the waste water to a standard which results in the discharged waste water and sludge complying with the required consents. For this reason the contract contains a list of inuent concentration levels. In the event that these are exceeded and as a result Stirling Water is unable to treat the waste water to the required standard, then Stirling Water will nevertheless be entitled to payment in full. Stirling Water is protected through the contract against circumstances inuenced by ESW that would impact on the volume of waste water such as the issuing of trade euent discharge consents, the introduction of water metering or greatly increased capital investment in the sewerage infrastructure leading to a reduction of water inltration to the sewers. However, under the payment mechanism, failure to achieve forecast levels of

114

D. Grimsey, M.K. Lewis / International Journal of Project Management 20 (2002) 107118

Fig. 2. Flow chart of analytical approach.

D. Grimsey, M.K. Lewis / International Journal of Project Management 20 (2002) 107118

115

the required project function for the least cost. Part of the value-for-money analysis involves a comparison of the project against a traditional public sector procurement and operation route known as the public sector comparator. The main risks analysed from this perspective concerned the following issues:
. The contract was conditional on obtaining planning consent and a risk-sharing mechanism was agreed whereby ESW would pay for a proportion of the planning change costs through an adjustment to the taris; . Stirling Water had qualied their bid in relation to passing the risk of costs resulting from nds, such as antiquities, being uncovered during the construction phase of the project back to ESW; and . Bidders were also asked to calculate the impact of delay to the programmed contract signing date in terms of the acceleration costs necessary still to achieve the commencement date for the main period for waste water treatment of January 2001. ESW and its advisers made an assessment of the likely delay and adjusted the bids accordingly.

Table 1 Procurer: interest rate sensitivity Interest rate % increase/decrease in NPV of expected cost stream from base case 3.3 1.7 Base case NPV +1.8 +3.6 +5.4 +7.5

Base case 1% Base case 0.5% Base case Base case +0.5% Base case +1% Base case +1.5% Base case +2%

The risks were analysed by establishing the expected cost to ESW of these risks and adjusting the NPV of the bid. In the case of Stirling Water this resulted in approximately a one percent increase in the NPV of the bid. When compared with the competitors, this led to the conclusion that their bid provided ESW with the most economically advantageous proposal relative to the other bidders. Under the Project Development Agreement, ESW also retained the risk of movements in the underlying interest rates up to nancial close, a practice common to such projects. In this case, the specic risk is that any increase in the level of interest rates, and thus of the nancing costs for the project, prior to nancial close will result in a higher tari being levied, while the discount rate used in the analysis is xed at 6% real. Interest rate risk is generally dicult to quantify with any real precision. To get a feel for the likely impact of this risk, sensitivity analysis was carried out on the nancial model that yielded the results shown in Table 1. The sensitivity analysis indicates that ESW were reasonably well protected against movements in the underlying interest rate with a moderate rate increase of 1% leading to an increase in the NPV of the cost stream of the project of 3.6% (the approximate gradient of the interest rate sensitivity relationship). ESW also carry ination rate risk to the extent that RPI deviates from that projected in the nancial model. Whilst it is generally accepted that there is a link between interest rates and ination the analysis looked at these independently since ination rates may not be reected fully in the ination premia built into interest rates [26]. Like interest rates, ination is dicult to predict and

sensitivity techniques were again used. Table 2 shows the variation in the NPVs at ination assumptions ranging from 2 to 6%. The NPVs are reduced at higher ination rates partly because the debt element of the nancing being serviced by the revenue stream of the project has not been indexed. The eect of discounting an element of the project's revenue stream xed in nominal terms at a xed discount rate of 6% real will lower the NPV of the debt element of the revenue stream at higher ination rates and therefore lower the overall NPVs. The analysis demonstrates that ESW is well protected against the risk that ination increases in the future. It is widely acknowledged that sensitivity analysis is limited to indicating the potential eects upon an outcome, in this case the NPV, if a movement in the variable occurs. Critics of this form of analysis argue that unless some indication of how likely it is that a quantied movement will occur then this form of analysis is not of much use. This argument ignores the fact that some important economic variables cannot easily be quantied in terms of likelihood and extent of change. Who in the UK, for example, during the late seventies could have reasonably predicted that ination would be 3% or less in the late nineties, a period of only 20 years that is, 10 years less than the contract term in this case study? Under such circumstances, sensitivity analysis provides a useful tool for analysing these risks in terms of bid evaluation.
Table 2 Procurer: ination sensitivity RPI assumption (%) % increase/decrease in NPV of expected cost stream from base case +4.79 +1.47 Base Case NPV 1.36 3.79 5.90

2 3 3.5 4 5 6

116

D. Grimsey, M.K. Lewis / International Journal of Project Management 20 (2002) 107118

7.2. Sponsors From the sponsors' perspective, the risk analysis centres around establishing the potential impact on the equity return. For the purpose of this exercise, the distinction between subordinated debt and equity is ignored because the degree of subordination gives subordinated debt virtually the same characteristics as equity. Subordinated debt is favoured in the UK because it enables the sponsors to extract cash from the project vehicle where dividends would be restricted by the prot and loss account and also because the interest is tax deductible. In order to review Stirling Water's nancial proposals from the sponsors' perspective an analysis of the impact of the equity risks on the nancial model had to take account of potential upside as well as downside risk. A simulation exercise was therefore carried out using the following methodology. First, a realistic downside (and upside) case for each risk was dened by ESW's technical adviser Halcrow Crouch to establish a triangular risk distribution for each risk. The relevant risks for this analysis were:
. volume risk; . the risk of mid-life capital expenditure and asset management costs being greater than forecast; . operating cost; and . operating performance.

all valid combinations of the values of input variables to simulate all possible outcomes. A summary of the results of the quantitative risk analysis on the Stirling Water nancial model is shown in Table 3. This analysis suggests that the sponsors are unlikely to achieve the base level rates of return, yet are taking a reasonable amount of nancial risk commensurate with risk capital investments in such projects. Although the relevant risks modelled contained upside potential, the analysis also indicates that there is, overall, no likely nancial upside for the sponsors. The main opportunity for nancial upside comes from increased waste water volumes, but the payments made by ESW are capped as described earlier. Also, the procurer in this case was in a strong bargaining position; England and France have some of the largest water companies in the world and they were vying for a foothold in the relatively small Scottish market. 7.3. Senior lenders For senior lenders the nature of non-recourse or limited recourse funding clearly carries a rather dierent risk or credit assessment than a conventional full recourse loan where the enforcement of security by the lender is additional to its ability to sue the borrower, and the lender can assess the value of the assets used as collateral. With project nancing, the facilities often do not have a capital worth, in terms of a wide market, to which lenders would wish to attribute value. Lenders, of course, insist on having the opportunity to step in and rescue a failing project but they cannot simply sell o the asset to realise value. In contracts such as AV&S, the assets essentially take the form of the contract with the procuring authority. It is therefore understandable that senior lenders tend to take a pessimistic view where risk analysis is concerned. The key dierence between the senior lenders and the sponsors is that for the senior lender holding debt rather than equity there is never any potential upside gain in the project, only downside risk that could reduce the ability of the borrower to make principal and interest payments under the loan agreement.

Whilst construction delay is an important project risk it was not included. This was because the risk analysis looked at the risks from the project sponsors' perspective as equity investors in Stirling Water. For them, delay risk is dealt with contractually through liquidated damages contained in the construction contract and also business interruption insurance. Eectively, therefore, the risk resides with the construction contractor and the insurer and not with the project sponsors. Second, having delineated the relevant risks for examination:
. a simulation exercise was carried out using the @RISK computer software package to determine the distribution of the relevant risks overall. Then, third, . from this analysis an assessment was made of the impact on the blended equity/subordinated debt IRR.

Table 3 Sponsors: equity/subordinated debt risk analysis Simulation Minimum simulation result 5% probability of returns being less than: 25% probability of the returns being less than: 50% probability of the returns being less than: 75% probability of the returns being less than: 95% probability of returns being less than: Maximum simulation result Decrease in blended equity IRR Base case 7.17% Base case 5.48% Base case 3.9% Base case 2.52% Base case 1.47% Base case 0.57% Base case 0.06%

The @RISK software package uses a Monte Carlo simulation process to perform a risk analysis. Simulation in this sense refers to a method whereby the distribution of possible outcomes is generated by letting the computer recalculate the nancial model over and over again, each time using dierent randomly selected sets of values for the probability distributions contained in the spreadsheet model. In eect, the computer is trying

D. Grimsey, M.K. Lewis / International Journal of Project Management 20 (2002) 107118

117

Senior lenders therefore focus on cover to the income stream over the term of the loan and analyse risk to establish robustness by reference to cover ratios. The most important of these ratios are the loan life cover ratio (LLCR) and the annual debt service cover ratio (ADSCR). The LLCR provides a snap shot on a given date of the NPV of the projected cashows from that date until retirement of the loan relative to the loan outstanding on that particular date. The ADSCR is an historic ratio that measures the cashow for the previous year in relation to the amount of loan principal and interest payable for that period. The sensitivities tested from the lenders' perspective are intended to capture the risks left with the service company rather than the project as a whole, taking account of the fact that it will seek to mitigate key risks by allocating them away contractually. The main instruments used for allocating risk will be:
. The Design and Construction (D&C) Contract which mitigates the Service Company's exposure to design risk and construction cost and time overrun risk; and . The Operation and Maintenance (O&M) Contract which mitigates the Service Company's exposure to performance risk and operations and maintenance cost risk.

Table 4 Senior lenders: robustness analysis of cash ows Sensitivity Base case nancial model Construction cost OPEX Downside ow Midlife CAPEX Operational performance Combined downside: downside ow plus midlife CAPEX
a

Changea (%) +3 +12.5 +10 2

ADSCR 1.26 1.26 1.08 1.07 1.24 1.20 1.04

LLCR 1.32 1.32 1.15 1.12 1.29 1.26 1.10

As dened by ESW's technical adviser (see text).

Overall, the combination of risk analysis techniques used on the AV&S project demonstrated that:
. The project delivers value-for-money to the procuring entity ESW and satises the main government criteria for investing in capital projects; . The project has reasonable but not excessive upside potential for the project sponsors and downside potential commensurate with the levels of return anticipated in the base case; and . The downside sensitivity testing suggests that the project is suciently robust and nancially stable from the viewpoint of lenders.

It is therefore crucial to the nancial robustness of bids, given the highly geared nancing structure envisaged, that these risks should be allocated away from the service company under strong contracts to suitable counterparties. However, even when risk is transferred contractually some residual risk will remain with the service company. For instance, if the operating costs turned out to be signicantly greater than originally forecast it is conceivable that the service company could decide that it was in its best interests to share the pain with the O&M Contractor by agreeing to a price increase to absorb some of the increased cost, rather than running the risk of the O&M Contractor abandoning the contract. The operating expenditure (OPEX) and capital expenditure (CAPEX) sensitivities tested are intended to reect this residual risk. Table 4 below displays the results of the robustness analysis on Stirling Water's nancial model. The ADSCR and LLCR (see earlier denitions of these cover ratios) only fall below the minimum requirement (ADSCR of 1.15 and LLCR of 1.15) under the OPEX, Flow and Combined Scenario. However, in no case do the ADSCR or LLCR fall below 1.0, which would indicate a default under the loan agreement. This analysis indicates to the senior lenders that the Stirling Water nancing plan is reasonably robust to support the level of funding anticipated.

8. Conclusion Project nance and PPP/PFI arrangements are founded on the transfer of risk from the public to the private sector under circumstances where the private sector is best placed to manage the risk. The general principles are common to all public sectors insofar as the projects seek to shift risk from the public sector to the supplier and oer a prot incentive to the private sector in return. However, the principal aim for the public sector is to achieve value-for-money in the services provided while ensuring that the private sector entities meet their contractual obligations properly and eciently. Value-for-money and risk transfer principles accepted, fundamentally PPP projects are viable only if a robust, long term revenue stream, over the period of the concession, can be established. A framework for investigating and carrying out an analysis of the risks has been outlined in this paper that systematically views project risk from the perspectives of the procuring entity, the project sponsors and the senior lenders. For a project to be successful, the diering (and conicting) needs of these parties must be satised in the risk allocation process, and this would seem to have been achieved in the waste water treatment project used to illustrate the application of the framework.

118

D. Grimsey, M.K. Lewis / International Journal of Project Management 20 (2002) 107118 [12] Dixit AK, Pindyck R. S., Investment under uncertainty. Princeton (NJ): Princeton University Press, 1994. [13] Adam C. Embedded options in infrastructure projects. Journal of Applied Finance and Investment 1996;1(1):336. [14] Peirson G, McBride P. Public/private sector infrastructure arrangements. CPA Communique 1996;73:14. [15] Grimsey D, Graham R. PFI in the NHS. Engineering, Construction and Architectural Management 1997;4(3):21531. [16] Chapman CB, Ward SC. Project risk management processes, techniques and insights. Chichester: John Wiley and Sons, 1997. [17] Kerzner H. Project management, a systems approach to planning, scheduling and controlling. 3rd ed. New York: Van Nostrand Reinhold, 1989. [18] Smith RC, Walter I. Global nancial services. New York: Harper Business, 1990. [19] Thobani M. Private infrastructure, public risk. Finance and Development 1998;36(1):503. [20] Merna A, Smith NJ. Privately nanced concession contract, Vols. 1 and 2. 2nd ed. Hong Kong: Asia Law and Practice, 1996. [21] Knight F. Risk, uncertainty and prot. Boston: Houghton Mifin, 1921. [22] Blatt JM. Dynamic economic systems. New York: Sharpe, 1983. [23] World Bank. Toolkits for private sector participation in water and sanitation. Washington: World Bank, 1997. [24] Economic Committee APEC. Infrastructure development conference summary and case studies. Singapore: APEC Secretariat, 1996. [25] CEDA. Infrastructure and economic development. CEDA information paper no. 60. Committee for Economic Development in Australia, Melbourne: 1999. [26] Lewis MK, Mizen, PD. Monetary economics. Oxford University Press, Oxford, 2000.

References
[1] McCarthy SC, Perry JG. BOT contracts for water supply. In: World water 89. Thos. Telford London: 1989. [2] Augenblick M., Scott Custer Jr. B. The build, operate and transfer (BOT) approach to infrastructure projects in the developing countries. Working papers. World Bank, Washington: August 1990. [3] Monod J. The private sector and the management of public drinking water supply. Washington: World Bank, 1982. [4] Owen G, Merna A. The private nance initiative. Engineering, Construction and Architectural Management 1997;4(3):21531. [5] Mustafa A. Publicprivate partnership: an alternative institutional model for implementing the private nance initiative in the provision of transport infrastructure. Journal of Project Finance 1999;5(2):6479. [6] Horner R. Public private partnerships, the way forward. Irish Banking Review Summer: 1999; 25-34. [7] East Asia Analytical Unit. Asia's infrastructure in the crisis, harnessing private enterprise. Canberra: Department of Foreign Aairs and Trade, 1998. [8] International Finance Corporation. Project nance in developing countries. Washington:: The World Bank, 1999. [9] Martini CA, Lee DQ. Diculties in infrastructure nancing. Journal of Applied Finance and Investment 1996;1(1):247. [10] Threadgold A. Private nancing of infrastructure and other longterm capital projects. Journal of Applied Finance and Investment 1996;1(1):712. [11] Lewis M.K. Banking on real estate, In: Fair D.E., Raymond R. editors. The competitiveness of nancial institutions and centres in Europe. Dordrecht: Kluwer, 1994.