Contingent Liabilities: Issues and Practice; Aliona Cebotari; imf working Paper 08/245; October 1, 2008

Several other instruments for sharing risks have been used. These include: (i) setting a 
time limit on the operation of the contingent liability instruments (Australia); (ii) including 
termination clauses that allow the government the option to terminate the arrangement when 
it is determined there is no longer a need for the instrument (Australia); (iii) requiring 
beneficiaries to post collateral; (iv) requiring an ownership stake, for example through stock 
warrants (U.S.); and (v) sharing the upside potential, in addition to the downside risks, as is 
done in the case of minimum revenue guarantees under PPP arrangements in Chile, 
Colombia, Korea (see OECD/ITF, 2008)
21
or as provided in the contingent liability 
framework for Canada (Currie, 2002).
Charging for Guarantees 
When the government takes on a contingent obligation in order to correct a market 
failure rather than to subsidize, it is good practice to shift the cost of this obligation to 
the beneficiary. This can be done by charging the recipient of the guarantee a fee reflecting 
the market cost of the guarantee, forcing him to compensate the government for the 
committed resources and bear the cost of the guarantee. The rationale for charging for 
guarantees is that, like a traditional insurance policy, the guarantee has value to the insured 
and imposes a cost on the insurer, hence a firm would normally be expected to pay for the 
guarantee an amount at least equal to its actuarial cost. If the government charges the 
recipient only on an expected cost basis, its guarantee operations would break even in the 
long term. However, this fee would be lower than what private market participants would 
normally charge, because they would also demand a risk premium to compensate them for 
the risk that the cost of a guarantee may be higher than expected. Therefore, the market cost 
of the guarantee would equal the expected cost plus a risk premium.
22
Studies of the 
corporate bond market, for instance, suggest that the risk premium included in the corporate 
bond yield may be large and may significantly exceed the value of the expected loss.
23
In 
ued) 
21
For example, under Chile’s minimum income guarantees, all bidders are guaranteed income equal to
70 percent of the investment costs plus total maintenance and operations costs, and accept an obligation to share 
part of the revenues obtained if traffic is higher than expected. In this way, traffic risk is shared and high losses 
and windfall gains are avoided. This mechanism has generally performed well; from 1995–2003, the 
government had to pay out only US$ 5 million to cover revenue shortfalls from investments worth close to
US$ 5 billion (OECD/ITF, 2008).
22
The expected cost measure reflects the cost the government could be expected to bear on average over the 
lifetime of the guarantee, calculated as the expected annual cash payments times the probabilities of having to 
make them. 
23
Elton, Gruber, Agrawal, and Mann (2001), for example, show that the spread between the rates for a ten-year 
A-rated corporate bond and a ten-year treasury bond in the U.S. is accounted for largely by the risk premium
(40 percent), less so by the differential tax treatment of the bonds (36 percent), and only to a small extent by the 
expected default loss (17.8 percent). Similar results were obtained in Weber (2007). There are several reason for 
(contin

---

16
sum, it is important to avoid implicit subsidization, by charging a fee equivalent to the 
market cost of the guarantee. Nevertheless, if it is difficult to gauge the market cost, chargin
fees on the basis of expected cost or other value is clearly still preferable to not charging
g 
nything. 
r 
d 
, Chile 
or the 
or these guarantees (see Box 3 and Annex II for 
discussion of how guarantees are valued).
y 
ted 
a