A computer firm once proposed a novel way of charging the NHS for a maternity IT system. It would install the computers free, but receive a fee for every baby born at the hospital.
That was in an early flush of enthusiasm over private finance initiatives (PFI). Fortunately, wiser counsel has prevailed. The company soon realised that the birthrate was completely outside its control. Even worse, payment by results would create a clear incentive for the health authority to save money by shifting births to hospitals not on the system. The project's main goal, to capture better information about births, would soon be thwarted.
An extreme example, but it illustrates the difficulty of paying for government IT projects through PFI deals. This week, after 10 years of trying to make the process work, the Treasury announced that no new IT projects would be funded through PFI. (Existing projects, and those "on the point of signature" will still go ahead.)
Yet the government seems as committed as ever to PFI funding for schools, hospitals and other infrastructure. What's different about IT? Quite a lot, experience has shown. IT projects are difficult and expensive to fund privately because they are highly risky and lack an obvious revenue stream.
The whole point of the PFI is to transfer risk. The scheme was originally designed to pay for projects such as toll bridges, at which the government was notoriously slow. It made a lot of sense: if contractors make money only after the bridge or road is open, they have a strong incentive to do it on time.
With IT projects, the risk is less transferrable and the revenues, in consequence, become harder to identify. Revenue streams are usually based on organisational savings, which are often out of the IT contractor's ability to achieve: a change in the government's policy can make the whole business plan obsolete overnight.
These difficulties mean that PFI IT contracts are hugely complex. On average, PFI procurements take 18 months from the initial advertisement to the contract signature - in the NHS, they have frequently taken longer. Many procurements have broken down after months of negotiations because the customer and the contractor cannot agree on the transfer of risk.
Finally, if the whole thing does go pear shaped, the contractor does not even have the residual value of the technology as an asset. A three-quarters-built hospital is worth at least the value of its site: a three-quarters-built IT system is almost worthless.
All this means that banks charge a premium when funding IT projects, which means they are then always more expensive than conventionally funded ones.
For years, the Treasury (and the IT industry) has said that this price is worth paying if systems are built on time. But, palpably, they are not. The story of delays is as old as the first big IT contract under the PFI, the national insurance recording system deal signed in 1995. The main surprise about this week's announcement is that it has taken so long to come.
