Transcript of Question & Answer Session Some Innovative Mortgage Data

You just mentioned arrears levels at the moment are quite low. What interest rate level, or what arrears level do you think would need to be to be considered medium into high?

Sorry, can you just repeat that?

Arrears levels at the moment are now considered low. At what levels do they need to be to be considered from your perspective, medium to high?

Well, what I don't have here for housing loans is a long-run chart on non-performing loans. But, even during the height of the GFC, when the unemployment rate in Australia went up, and went up fairly quickly, that's often a thing that leads to arrears. They didn't go much higher than we currently are, in other words, they were reasonably low then. They were much higher, particularly for banks, but not so much in mortgages as much as for commercial property and the like back in the early 90s recession. So, you'd need something like a significant rise in unemployment and a recession to really drive arrears rates much higher.

I think that's really a sort of stark comparison. And the reason why I say arrears rates now are low, and even during the GFC were pretty low, was just because compared to countries offshore, particularly say North America and parts of Europe, you know, they had orders of magnitude higher rates of arrears. Could we ever get there? Possibly. If we had a dramatically weaker economy. I'm not going to speculate though on what sort of level of interest rates would generate such problems and of course, as we've suggested in our recent discussions on this sort of issue, the Board is going to be very attentive to any pressures in the economy that come about at a time when eventually rates start to normalise. And they're going to take that into account and take a cautious approach. That's not any time soon though I don't think.

Given that your base portfolio is dominated by self-securitised assets by the banks. I'm not sure what the percentage is, probably 60 to 80 per cent, I would say. What are you doing to increase the coverage of non-self-securitised assets because they stand to be high quality, specifically because they've got a need to be used as a CLF. And obviously the coverage of more quality mortgages and including all other things like seasoning. Do you think that this is a statistical issue that you can adjust for the quality of data that you get or do you think you need to widen the coverage of the data set to receive other data from the other mortgage providers?

Well we receive this data when the banks want to be able to pledge a security to us as collateral. They want that option. So it's not entirely up to us to determine that mix, and the self-securitisations, as I've suggested, in the paper are there primarily to sort of facilitate the banks participation in the committed liquidity facility. So it's not something we direct. I think you raise a good question that we can look into. I just don't happen to know the answer as to whether there's any significant difference between the marketed and self-securitised, sort of the nature of the pools of those assets. But my suspicion is, if there is a difference, it's not going to be particularly stark. Particularly because we wouldn't want to have the banks self-securitise and pledge a pool of assets that is of significant poorer quality for example. Sorry, the next question.

Sorry about that. I'm going to ask and try to combine what you've shared in the previous session. So the question is, going forward do you see RBA will be exploring other dimension or attributes of data? Like, so far what you've shared with the payment characteristics, LVRs, and you know, whether it's owner occupied or investment type, do you see going forward RBA might be exploring other dimension or other types of data?

Well within this database, yes, I mean that's going to be something that we will use regularly just to better understand developments in the mortgage market. Because as I've suggested, this data is pretty timely and it might not be entirely representative of all mortgages but it does capture a sizable share of the mortgage market and there are, as I've said, those hundred fields that are covered for each loan. So there's a lot of rich data there to be used to better understand the market. So on that dimension, yes. I wasn't sure about your question, whether you were talking of other datasets and marrying this across. Of course, the data in here is, it's a horrible word, but it's de-identified. So we can't actually sort of link this in to sort of other databases. Obviously. But you can do useful things like I showed in the arrears chart. Where you take sort of some of the loan data by region and then you marry that against sort of the data from the ABS on unemployment rates by region. So there's those sort of comparisons that can be made. I'm sure we'll be doing a lot more of that sort of thing.

Well thank you very much, thank you for your time.