What is your toll actually worth?

Four governments just did four different things with the same asset — a future stream of user revenue. One sold it forward. One auctioned it away by design. One wrote down that it might not arrive. In the fourth it already didn’t, and 31% of the financing landed on the budget.

Four decisions and one anchor. Everyone is trading the toll; the forecasts underneath it run a quarter high.

Week of 1 September 20261 September 20265 cases · 7 slides
India · selling the toll

India is selling fifteen years of toll revenue to build the next road.

NHAI has finalised its list of national-highway stretches for monetisation in FY 2026–27 under Toll-Operate-Transfer and infrastructure investment trusts: 17 assets, 1,692.5 km, across nine states. Ownership stays with the State — what is sold is the operating right and the toll stream that comes with it, and the proceeds fund fresh construction.

1,692 km
of operating highway whose future toll revenue is being converted into cash today — 17 assets, nine states

This is good treasury management and a real answer to a capital constraint. It is also, in substance, a financing transaction: cash now against revenue the State was already collecting and will not collect again.

Source · India · NHAI / MoRTH, PIB release, 28 May 2026
Brazil · bidding the toll away

In Brazil, the winning bid is the toll you agree not to collect.

On 21 May 2026 the ANTT board homologated the concession of BR-116/251/MG, “Rota Gerais”, to Ecorodovias — the winning bid expressed, in the Diário Oficial, as a 19.00% discount on the basic toll tariff. Weeks earlier EPR took the São Paulo–Curitiba Régis Bittencourt corridor at a reported 22.53% discount, against R$7.2bn of works.

19.00%
off the basic toll — not a side effect of the auction but the award criterion itself: the concession goes to whoever promises to charge users least

This is competitive tension working as intended, and users get the benefit on day one. It also means the revenue stream is deliberately thinned at the moment of award — before a single vehicle has been counted.

Source · Brazil · Deliberação ANTT nº 143, 21 May 2026 · D.O.U. 22/05/2026
Kenya · counting the toll

Kenya’s Treasury wrote down that the toll might not come.

The 2026 Budget Policy Statement puts the PPP portfolio at 36 projects and, on the flagship Rironi–Nakuru–Mau Summit road (KSh 150 billion), names the exposure in the government’s own budget document: “traffic demand uncertainty, revenue shortfalls and potential renegotiation risks.”

KSh 150bn
on one road inside a 36-project portfolio — with demand risk named by the Treasury, in the budget statement, before anything has gone wrong

This is the case on this page that is going right. Most finance ministries disclose PPP exposure as a stock of commitments. This one names the mechanism — demand — and the remedy the mechanism triggers: renegotiation.

Source · Kenya · National Treasury, 2026 Budget Policy Statement §5.3.2.2
The Bahamas · when it doesn’t come

Thirty-one per cent of the financing has already become debt.

The Fiscal Strategy Report 2026, tabled with the budget, records that of US$140m in PPP financing from the Africa Export-Import Bank for road upgrades on Eleuthera and Cat Island, US$43.1m has crystallised as an obligation the Public Treasury must repay. The report rates PPP exposure as of “moderate potential fiscal impact and a possible likelihood of realisation.”

31%
of the PPP financing has moved from contingent to actual — US$43.1m of US$140m, now a Treasury repayment

Note what the rating and the outcome do together. The exposure was assessed as moderate and possible — and roughly a third of it realised anyway. “Possible” is not a number, and it cannot be budgeted for.

Source · The Bahamas · Fiscal Strategy Report 2026, tabled with the 2026/27 Budget
Global · the anchor

The forecasts under all four run about a quarter high.

The largest study of toll-road forecasting performance — over 100 internationally, privately financed toll roads — found the mean ratio of actual to forecast traffic at 0.77 in the first year of operation. By year three it had moved to 0.79. The optimism does not wear off with operating experience.

0.77
actual traffic as a share of forecast, first year — a systematic 23% optimism bias across geographies, road types and decades

Read the four pages before this one against that number. The monetisation price, the Treasury’s named demand risk and the crystallised third are not four unrelated events. They are the same forecast error at four stages of one contract’s life.

Source · Bain (2009) · Error and optimism bias in toll road traffic forecasts · Transportation 36(5)
The through-line

You can only sell the toll once.

Sold forward in India, bid away in Brazil, named as a risk in Kenya, already crystallised in the Bahamas — and underpinned everywhere by a forecast that runs 23% high and stays there. The toll is not spare capacity. It is the only thing standing between the contract and the budget:

VfMvalue for money
Re-run the case at 0.77 of the forecast, not at the forecast.
FAROPPP fiscal risk
Price demand shortfall as a distribution, with a budget line attached.
GFSaccrual accounting
Show revenue sold forward as what it is — financing, not income.

Four governments traded the same asset four ways this year. Only one of them wrote down what it might not be worth.

Papers, models & FARO

Every fix above points at a paper in the Austral series, and at the instrument that implements it.

ResearchPlatform