Mr. John McKeon:

I have to say that the illness benefit experience was not our proudest moment. I want to acknowledge that. We made some mistakes. At the time it happened, I got in touch with the then Chairman of the Joint Committee on Social Protection, John Curran, and I asked to be allowed to appear before the committee so I could explain it. We appeared before it and we apologised and explained what happened. There is a record of the notes if the Deputy wants to follow up on this. I will summarise the position very quickly. We have more than 80 schemes in the Department and we have been migrating them on a rolling basis to our more up-to-date IT platform. There are about four schemes left. We got through nearly all of them with no trouble whatsoever and then ran into a problem with the illness benefit.

Regarding testing, we did the full range of standard tests, integration testing, regression testing, system testing and end-user testing. We did all the usual tests. In fact, we implemented the system live earlier in 2018 on a subset of illness benefit claims, the partial capacity benefit claims, and it worked fine. When we went live with the big range of claims at the end, it fell over. There are a couple of reasons. One was that we were working on the basis of what we believed was an agreement with the medical profession on the use of a new form of medical certificate. Many doctors followed through on that but some did not and we ended up getting certificates that we could not put into the new IT system. That was the major problem that started it all off. We had people who are probably entitled to the payment but whose doctor sent in the wrong certificate. We could not get the certificates into the system on time so we had to make a judgment call - it was a bit like the PUP - and bias it towards paying the person rather than processing the certificate. That is where the problem arose.

To be honest, there were also other issues. We made some design changes in implementing the system that, in retrospect, we should not have made. We moved from a payment-in-arrears model to a real-time payment model, which we believed was the right thing to do. What happens in practice, however, is that it can take people a week to get an appointment with their doctor after having been ill for a week, and the certificate is received a week later. The payment-in-arrears model gave a week or two of a buffer allowing all that to happen whereas a real-time model gave no buffer. That caused problems.