Chapman's Peak Drive runs along the side of a mountain near Cape Town in South Africa. Described as "impossible" when first proposed, the road was nevertheless hacked and blasted out of the steep cliffs of the mountain between 1915 and 1922. Its location has, however, always made it vulnerable to falling rocks and other debris, and in January 2000 the Western Cape Provincial government closed the road after fire and heavy rain caused major rockslides and the death of a passenger. The provincial government chose to use a public-private partnership, or PPP, to improve the road, an arrangement that was then becoming popular in South Africa. The government called for proposals in August 2001. Two consortia bid, and in May 2003 the government and the Entilini concession company signed a 30-year concession contract. Entilini repaired the road and, using state-of-the-art modeling and engineering, reduced its vulnerability to rock falls. The road reopened in December 2003, at an estimated capital cost of about 150 million rand, split roughly equally between the Province and the concessionaire.1
It was hoped that tolls would cover the concessionaire's costs. But the Province agreed to compensate the concessionaire in certain circumstances if toll revenue was less than a forecast made in 2002. When the road opened, the concessionaire would have to collect tolls from a temporary plaza while it waited for approval from the national Department of Environmental Affairs and Tourism to build permanent toll plazas; the provincial government agreed to bear traffic risk until that approval was granted and the plazas were built. Second, the provincial government agreed to bear traffic risk during certain road closures. Third, the government gave a revenue guarantee that, independently of the provisions relating to toll plazas and road closure, partially protected the concessionaire's lenders from revenue risk.
After lengthy appeals, final approval of the toll plazas was granted in June 2008, and only then could construction of the toll plazas begin. As often happens,2 traffic initially fell short of forecasts. In the absence of permanent toll plazas, the government therefore had to top up the concessionaire's revenue.
Eventually revenue reached forecast levels. But in July 2008 the concessionaire closed the road because of another rockslide. Because the toll plazas are not yet constructed, the government still bears the traffic risk and is having to pay the concessionaire an amount equal to all its forecast revenue. From December 2003 and January 2009, it paid the concessionaire 57 million rand.
This doesn't imply that the government got a bad deal or bore too much of the project's risk. One can reasonably ask whether the best means of compensating the concessionaire for the absence of permanent toll plazas was for the government to pay the difference between actual and forecast traffic. But it is easy to be wise after the fact; decisions about risk bearing are better judged on the basis of the information available at the time. Nevertheless, the case illustrates the kinds of contingent liabilities that arise in PPPs and the need for governments to pay attention to them, both after a contract is signed and, especially, before.
In the last two decades, many other governments have also used PPPs to obtain financing for infrastructure projects. Also known as concessions, these arrangements usually allow the government to get infrastructure built without having to pay for it immediately or, as in the case of Chapman's Peak, without having to pay for all of it immediately. In some cases, the government pays for the service in installments over the term of a contract. In others, users pay. In both cases, the government typically bears some of the risks of the project. As in the case of Chapman's Peak, the government may protect the project company or its lenders from some of the risks of uncertain user-fee revenue. It may also agree to bear the unknown costs of cleaning up possible environmental problems or of acquiring land for the project. It usually agrees to make a compensating payment to the project company if the PPP contract is terminated before its scheduled end. The pressures and rationales for such risk bearing are enduring, but they became stronger during the financial crisis of 2008-09, as lenders and investors grew more cautious.
Although not all these risks create contingent liabilities for accounting purposes, they do create obligations that are often conveniently, if loosely, called contingent liabilities.3 Roughly speaking, contingent liabilities require expenditure only if an unpredictable future event occurs. The probability of that event occurring is typically low. Contingent liabilities differ from the "direct" liabilities that a government incurs when it borrows money or commits itself to paying for a service. In those cases, the government usually plans to pay; when it assumes a contingent liability, it seldom does.
Contingent liabilities create management problems for governments. They have a cost, but the cost is uncertain, so judging whether it is worth incurring is difficult. And a contingent liability seldom requires budgetary approval or recognition in the government's accounts, so a government may prefer contingent liabilities to other obligations. (The uncertainty surrounding contingent liabilities can, however, work the other way. It is well known that PPPs create contingent liabilities, and the IMF, the World Bank, and others often warn of the risks. The initial reaction of a cautious ministry of finance may be to seek to avoid all contingent liabilities.) Management problems also arise once a government has incurred a contingent liability. Projects need to be monitored so that things can be done to reduce risks if possible. Spending must sometimes be forecast, despite the difficulty.
There are many sources of recommendations on managing contingent liabilities created by PPPs. An idea underlying most of the recommendations is that the rules governing PPPs should ensure that the people in charge have incentives, information, and the capability to take account of the costs and risks of contingent liabilities. Specific proposals have included the following (we list them without endorsement):4
Cost-benefit analysis should be used to select projects and value-formoney analysis should be used to choose between PPPs and public finance.
The costs and risks of contingent liabilities should be quantified.
PPPs should be approved by the cabinet, the minister of finance, or some other body with an interest in future spending. The ministry of finance should review proposed PPPs.
Governments should bear only those risks that they can best manage, which are generally those that they can control or at least influence.
Modern accrual-accounting standards should be adopted for financial reporting, to reduce the temptation to use PPPs to disguise fiscal obligations.
PPP contracts should be published, along with other information on the costs and risks of the financial obligations they impose on the government.
Budgetary systems should be modified so that they capture the costs of more contingent liabilities.
A guarantee fund should be used to encourage recognition of the cost of guarantees when they are given, or to help with payments when guarantees are called.
Governments should charge fees for guarantees.
Although there is no shortage of recommendations, it is harder to find out what governments have done to try to improve the management of contingent liabilities associated with PPPs.5 This report aims to help remedy this problem by describing the policies of governments in three countries that are often considered examples of good practice: Australia, Chile, and South Africa. In particular, it considers for each country who must approve a proposed PPP contract and the contingent liabilities it creates, what analysis is undertaken of contingent liabilities, and how PPPs and contingent liabilities are reported to the public in the government's accounts and other documents.
It is difficult to draw conclusions for other countries from just three case studies, and the main aim of this report is simply to describe the relevant practices of the three countries. But, drawing on the experience of these and other countries, the report also discusses which of the three countries' practices appear to be suitable candidates for adoption by other countries, including those with less administrative capacity than Australia, Chile, and South Africa.
_________________________________________________________________________________
1 This account is drawn from the South African National Treasury's PPP Quarterlies 3 (June 2001) and 7 (June 2002); Farlam (2005); Dreyer and others (2005); Gosling (2009); Yeld (2009); and http://www.candor.com/chapmanspeak/ and discussions with officials.
2 On the general issue, see Skamris and Flyvbjerg (1997) and Standard & Poor's (2003).
3 The term 'contingent liability' is problematic, both conceptually and in practice, and the International Accounting Standard Board has proposed eliminating it from accounting standards (IASB 2005). One issue is that the probability of payment under a contractual obligation can vary continuously from 0 to 1, and any division of that interval into two parts, one for contingent liabilities and the other for ordinary liabilities, is arbitrary. For more on definitions of contingent liabilities, see Blair and Jagolinzer (2008), Irwin (2007, ch. 6), and IASB (2008).
4 For recommendations on the management of contingent liabilities associated with PPPs specifically, see Lewis and Mody (1997); Currie (no date); Hemming and others (2006); Irwin (2007); and Schwartz and Corbacho (2008). The guidelines produced by government agencies in charge of PPP policy also contain a great deal of relevant advice, even if they do not refer specifically to the management of contingent liabilities. See for example Government of Australia, Infrastructure Australia (2008a,b) and Government of South Africa (2004b,c). On the management of contingent liabilities generally, see Brixi and Schick (2002) and Cebotari and others (2008).
5 On Colombia, however, see Lewis and Mody (1997) and Echeverry and others (2002).