As I said, we're in this feasibility stage. Early next year we'll make some decisions on which way we're going forward.
Depending on which way we're going, the cost implications could be quite different. I mentioned the possibility of linking to existing census data. That might be a very different proposition from a full-scale data acquisition. To give you an example, it might be the case that, to supply some variables, a large number of institutions will have to make large, significant changes to their IT systems. That could be a significant cost, depending on the way we go.
It is important that we finish our feasibility work and decide which way we are going. That will give us a sense of the cost.