India's 108 service bought emergency ambulances by the trip, not the fleet

Before 2005, dialing for an ambulance in most of India meant pleading with private hospitals or hiring any vehicle available; the poor simply did not reach hospital in time. The '108' model changed the shape of the market: one free number feeding a centralized call centre that dispatches GPS-tracked Basic Life Support ambulances, and a public-finance twist — the state pays a private operator per ambulance under contract, with a minimum of about three calls a day, instead of buying and running its own fleet.

The fullest evaluation, in Punjab, analysed 391,369 calls across 20 districts in a single year: 240 BSL-standard ambulances averaged 3.5 patients a day, at a cost of 1,361 rupees ($22.7) per patient transported. Poor patients and those living farther from facilities were more likely to use the service, and the average call-to-hospital time of 48 minutes matched the 54 minutes recorded in states run directly by the operator EMRI — while comparable PPP states paid as little as 565 rupees per patient.

A six-state analysis of maternal transport showed the model's reach: '108' carried 9–20.5% of expected pregnancies depending on the state, disproportionately rural and low-income women, and 89–98% of obstetric emergencies reached hospital within two hours in five of the six states, with ambulances covering half of all emergencies within 10–11 kilometres. The honest caveat from Punjab: after adjustment, the service showed no significant lift in institutional deliveries where baseline coverage was already high.

Buy trips, not ambulances

Paying per trip with a utilization floor kept the operator's incentives aligned with answered calls, not parked fleets.

One number and one call centre made the network legible — a single dispatch brain instead of district-by-district improvisation.

Costs split cleanly: government bore the welfare price per ride while the operator bore depreciation, drivers and idle time.

GPS and call logging made every claim auditable — response minutes, distances, costs — so evaluations could price the model honestly.

Outcomes the state wants but can't build can be bought per unit — if the unit is small enough to log and a payment floor keeps capacity in place. Welfare pricing and private capital can share a ride.

The PPP emergency-service model spread across Indian states; Punjab's emergency referral spend consumed about 6.2% of its National Rural Health Mission budget in 2012, against 3.1–5.3% in comparable states, and per-patient costs varied roughly two-fold — keeping the payment design under review. The evaluations' combined verdict: real access gains for the poor, and a standing question about whether to keep paying for rides or start paying for outcomes.