The $192m Draw That Explained Ghana’s Real Payment Security

Strategic importance, sovereign support and a signed PPP contract do not by themselves make public payments financeable. From Ghana to Nigeria and Argentina, the projects that held together had tested fallback routes for cash, lender control and termination recovery.

On 7 April 2020, Eni, the Italian energy company, and Vitol, the Netherlands-based commodities trader, drew $192 million under a bank letter of credit supporting Ghana’s Sankofa gas project.

That date matters because Sankofa was not supposed to need the letter of credit. The roughly $4.65 billion gas project had been designed with layers of protection: gas-sale receipts in a segregated account, a 4.5-month escrow reserve, then a $500 million bank letter of credit, with sovereign and World Bank support behind it.

It sounds like overkill. It was not enough to prevent trouble.

The government disbursement account was never made operational. GNPC, Ghana’s national oil company and a project partner, funded only $100 million of a planned $210 million escrow. The reserve was depleted. Then the letter of credit did what letters of credit are meant to do: it paid when the earlier promises did not.

Sankofa showed that a payment reserve matters only when it is funded and its fallback can actually be drawn. Photo: Dimas Rachmadan / Pexels, Pexels licence (free commercial use).

The usual retelling of this kind of project is that international backing made it safe. The more useful reading is harsher. Sankofa survived a failure in its payment machinery because the next part of the machinery was already there.

That is the distinction that gets lost in public-private partnerships, or PPPs. A government can value a road, railway, power plant or gas project intensely. It can sign a contract saying it will pay. It can even stand behind the contract with a sovereign guarantee. Yet none of that answers the question a lender quietly asks before committing money:

What happens when this month’s invoice is approved but the money does not arrive?

The myth of the strategic project

A PPP is simply a long-term arrangement in which a private company builds or runs public infrastructure and is paid under agreed rules. In an availability-payment project, government pays when an asset is ready and performing. In a public-offtake project, a public buyer purchases the power, gas or other output.

Project finance means the lenders expect the project’s own future cash flows to repay them. That does not mean lenders ignore governments. Quite the reverse. If government is the buyer, its payment process becomes the project’s bloodstream.

Most people think the crucial issue is whether government will honour its obligations. That is too broad to be useful. Pakistan shows why.

The country’s independent power producers received broad presidential guarantees in the 1990s, covering the obligations of the utility and fuel suppliers. On paper, this was formidable support. In practice, collections, tariff decisions, subsidies and loss allocation continued to jam up the system.

By June 2024, Pakistan’s power-sector circular debt had reached PRs2.393 trillion, according to NEPRA, the national power regulator. PRs1.710 trillion was owed to independent power producers and public generators.

Saif Power, a Pakistani power producer, offered a particularly uncomfortable example in its 2024 filing. Its receivables from CPPA-G, the government-backed power purchaser, were sovereign-guaranteed. Yet PRs3.83 billion was 61 to 120 days overdue, while another PRs2.93 billion was 121 to 180 days overdue. During contract renegotiation, it agreed to waive PRs1.36 billion of delayed-payment receivables.

A guarantee can make a debt legally good and economically late. Banks cannot use a legal argument to make this month’s debt-service payment.

This is where most people stop looking. They see the guarantee, recognise the sovereign name and assume the risk has been solved. The invoice still has to be certified, budgeted, released, transferred and protected from diversion before lenders are paid.

That is the PPP payment security mechanism in plain English: not one promise, but a chain of promises and accounts that must work in sequence.

Maryland Purple Line. Photo: Elvert Barnes / Wikimedia Commons, CC BY-SA 2.0.

The stack that saved Sankofa

Think of payment security as a stack of five questions.

First, where does the money come from? A payment obligation that depends on a fresh annual political decision is a different creature from one supported by multi-year statutory authority or debt-service-equivalent budgeting. Brazil, for example, treats federal PPP subsidies like automatically appropriated debt service, according to the World Bank’s PPP Knowledge Lab.

Second, does the cash enter a protected account? Sankofa’s segregated receipts and escrow were designed to stop payment money evaporating into other claims.

Third, what happens between invoice and payment? A contract needs fixed certification and invoice deadlines, a route for paying undisputed amounts, rules on late payment and limits on deductions. Otherwise an argument about a small performance issue can become an excuse to delay the whole bill.

Fourth, who controls the waterfall, meaning the order in which cash is used? Lenders care that receipts and insurance proceeds pass through controlled accounts before money can be distributed elsewhere. This is not greed. It is the difference between a project having cash on Tuesday and missing debt payments on Wednesday.

Finally, if the relationship breaks, can lenders cure the problem, step in or recover an agreed termination amount? Step-in rights allow lenders or a replacement operator to intervene before a project collapses. Termination compensation sets the amount payable if the contract ends.

Sankofa’s lesson is not that every PPP needs the same expensive collection of instruments. It is that every layer must be real. An escrow written into a document but not opened, funded or replenished is a decorative escrow.

GI Network's view: Payment security is not a badge a government attaches to a PPP. It is a live operating system. If one account is unfunded or one remedy cannot be used, lenders will price the weakness of the whole system.

Two Nigerian plants, one missing bridge

Nigeria gave the market a near-perfect comparison.

Azura-Edo IPP, a 450 MW power plant, reached financial close in December 2015. It had $877 million of project investment, including $622 million of senior debt, the bank loans paid ahead of equity. Its take-or-pay power purchase agreement meant the buyer had to pay for contracted power under the agreed terms. But the PPA was only the beginning.

Azura-Edo also had a $120 million standby letter of credit for PPA payments, World Bank and MIGA support, a federal Put/Call Option Agreement for termination payments, and direct agreements. Those direct agreements gave lenders routes to cure failures, step in, substitute parties and ensure termination compensation was paid to the security agent acting for the lenders.

The interesting part is not that Azura had protection. It is that its protection joined up. The payment promise, liquidity support, termination route and lender rights were interlocking documents rather than separate gestures of confidence.

Now put it beside Qua Iboe IPP, another Nigerian power project. It was planned at 533 MW and had a proposed $150 million payment guarantee. It was also strategically important. Yet it never reached financial close. The plant and its transmission line were not built, and the guarantee allocation was cancelled.

A proposed guarantee is not cash. Nor is it a complete financing structure.

Put the two cases side by side and a pattern appears that none of the headline announcements says outright: the real dividing line was not the size of Nigeria’s electricity need, or even the availability of international support. It was whether the documents made every stage of payment and enforcement executable before the money was committed.

Founders often encounter the same mistake in a different costume. A signed commercial commitment can look fundable until an investor asks who pays, when, from which account and what happens if the counterparty misses the date. Our examination of why lenders discount celebrated subsidies comes back to the same problem: an entitlement is not spendable cash until its path is reliable.

Argentina sold the exit, not just the electricity

Argentina’s RenovAr renewable-energy auctions offered 20-year power purchase agreements with CAMMESA, the administrator of the country’s wholesale electricity market. For many investors, that was still not enough.

The concern was not hard to understand. A long contract does not remove the possibility of non-payment, currency problems or a counterparty whose finances are weaker than the project’s needs.

RenovAr’s answer was unusual because it accepted that concern rather than arguing it away. If CAMMESA failed to pay, a project could be sold to FODER, the renewable-energy fund. That termination put was supported in sequence by the energy ministry, the finance ministry, earmarked Treasury notes and an optional World Bank guarantee.

The World Bank approved $730 million of guarantees across rounds beginning in 2017. Demand for the guarantee fell from 52% of bidders in Round 1 to 19% in Round 2. The programme supported 147 projects representing about 4,400 MW.

Here is the twist: investors may sometimes value a calculable exit more than a grand promise of uninterrupted operations. RenovAr did not pretend operating-payment risk had vanished. It gave investors a defined route out if it did not work.

That is why termination compensation deserves more attention than it gets. If a public contract ends after a payment failure, lenders need to know whether the amount covers the outstanding senior debt and whether there is a route to collect it. Vague language about fair compensation is not the same thing as a recoverable number.

It is a familiar capital-markets truth. A decent exit can still be worth zero to founders if the waterfall sends money elsewhere first. In PPPs, a respectable termination clause can also be worth very little if the payment source, priority and enforcement route are unclear.

When protection goes too far

The instinct after Pakistan or Ghana is to protect lenders from everything. Britain’s Metronet is the warning against that reflex.

Metronet, the company responsible for upgrading parts of the London Underground under a PPP, entered administration in July 2007 after governance and supply-chain failures. Transport for London guaranteed 95% of its borrowing. Of £627 million in bank loans, only £31 million was genuinely exposed to loss.

The National Audit Office concluded that lenders therefore had weak incentives to monitor. Estimated taxpayer losses were £170 million to £410 million.

Metronet showed the cost of protecting lenders so completely that their incentive to monitor weakened. Photo: Sam Wilson (taken with Samsung Galaxy S21 FE 5G) / Wikimedia Commons, CC BY-SA 4.0.

Maximum protection did not create maximum discipline. It weakened it.

This matters because a payment-security stack is not a licence to shift every risk onto the public. Lenders should have strong rights to cure and step in when a project is salvageable. But they should also retain enough exposure to care whether the project is being run well. Otherwise the state pays for bad oversight twice: once through the guarantee, then again through the rescue.

Maryland’s Purple Line, an availability-payment rail project in the United States, reveals another limit. It closed in 2016, but construction disputes and delays led to the replacement of its design-builder and a second financial close in April 2022. The revised cost reached $5.116 billion, against an original estimate of $2.407 billion. Maryland committed to create a separate trust account one year before revenue service for debt- and equity-related payments.

A credible government payer could not make availability payments appear before the railway was available. Debt continued accruing while construction risk was being fought over. Delay insurance, delay damages, relief events and compensation for government-caused delay have to fit together.

So the stack has a boundary. It secures payment risk. It cannot repair a project whose construction risks were allocated ambiguously in the first place.

What project sponsors should ask before they raise money

If you are developing a pre-revenue infrastructure project, financing before operations is possible. Azura-Edo and the Purple Line both demonstrate that capital can be committed before the asset earns operating revenue. But lenders will not lend against optimism. They lend against documents that explain how future cash reaches them and what happens when it does not.

Start with the payment map, not the pitch deck. Trace one invoice from certification through appropriation, budget release, payment account, reserve and debt repayment. Mark every point where a person, ministry or public body can delay it.

Then make the weak points objective. Is the reserve funded at close? Is there a minimum balance? Who must replenish it, by when, and what happens if they do not? Can an account bank prevent diversion or set-off? Are undisputed amounts paid even while a wider dispute continues?

Do not leave lender rights until the end. Direct agreements, cure rights, step-in rights and the termination-payment route are not legal decoration. They tell a credit committee whether it has a way to preserve value when the original arrangement fails.

The first documents to prepare are therefore the contracts that create cash flow and control it: the public payment agreement, the account and reserve arrangements, the termination provisions and the direct agreements. A guarantee offer without these is closer to a negotiation than a financing.

What experienced investors see

First-time sponsors often hear lenders asking for security and assume the bank is questioning the government’s integrity. Usually, the bank is managing a more mundane fear: timing.

A lender can survive an argument about ultimate liability. It cannot ignore a missed debt-service date.

That is why experienced investors separate three questions. Is the obligation legally valid? Is there liquidity when the invoice falls due? And if the relationship ends, is there a recovery route with a defined amount and priority?

These are not interchangeable. Pakistan had legal support but an invoice chain clogged by underlying system problems. Sankofa had early layers that failed, but a usable bank letter of credit behind them. Argentina made investors more comfortable by pricing the exit route. Metronet showed that the state can absorb too much lender risk and accidentally remove vigilance.

For investors, the smartest diligence question may be the least glamorous: show me the last step before debt service. Not the policy speech. Not the strategic plan. The account, the payment instruction, the priority and the remedy.

GI Network would begin this work by mapping the project’s invoice-to-cash route and testing it against the capital provider’s actual underwriting concerns. That means identifying appropriation gaps, checking whether reserves and accounts are genuinely controlled, stress-testing termination recovery, aligning the payment and direct-agreement documents, and rehearsing the objections an investment committee will raise before investor outreach begins.

The five-link payment test

Use the Five-Link Payment Test before calling a public-counterparty PPP bankable.

1. 1.Source: Is there a credible, continuing source of public funds, not merely an annual intention to pay?
2. 2.Release: Can an approved invoice move through certification and budget release on fixed, enforceable terms?
3. 3.Reserve: Is liquidity already funded, controlled and replenished by rules rather than goodwill?
4. 4.Priority: Do payment receipts flow through a controlled waterfall before cash can be diverted?
5. 5.Recovery: Can lenders cure, step in or collect defined termination compensation if the chain fails?

If one link depends on a future political favour, call the structure what it is: unfinished.

That is not cynicism about government. It is the lesson hidden in Sankofa’s $192 million draw. The first promise may fail. Finance becomes possible when the next promise is already funded, enforceable and ready to work.