Mr. Kieran Culhane:

We tried to add any possible additional information we could by linking to administrative datasets. We looked at a lot of administrative datasets. In the end we came up with four datasets, I think, that were most relevant for the private rented sector and for the types of dwellings we were looking at. They included the local property tax, the HAP scheme, rent supplement and a couple of others. Then we linked the properties on the RTB register to those datasets. Much like the Deputy, who started with the legislation, we could only start with the data. The dataset that we worked off was the RTB register, which we matched to those individual administrative datasets. From that, we could get a profile of what a rented property that is on the register looked like when we compared it with these register datasets. We found four real indicators there. I will not go through everything. For example, when we looked at the local property tax, and we looked at the principal private residence indicator, we found that in the case of the majority of properties on the RTB register it was not their principal private residence, which is what would be expected from a rental property. That, on its own, was not an indicator of a private rented property but when we put it together with the other indicators we had, it built a statistical model for us that helped us to identify what we thought should be on the register. Next, we compared our different set - the 73,000 - to those same four administrative datasets and we looked for the same indicators for the different set. Of the ones that fit our model that look like the RTB-registered properties, we said there were 25,000 that we thought were "formal" rental arrangements, and we called the others "informal" arrangements. It might have been better to call them "not formal" in that it was more by omission that they did not look like the properties on the RTB register. That was how we identified those. Then to get the extra little bit about semi-detached-----