The Reserve Bank (RBNZ) is planning a major overhaul of the data it collects from banks about residential mortgage lending, which will hopefully overcome some of the shortcomings in its current mortgage data.
The RBNZ currently collects a wide range of residential mortgage data* from banks, including the volume and amount of new mortgage lending, the type of borrowers taking out loans, such as first home buyers, owner-occupiers and investors, and the purpose of the mortgages, such as buying a property, topping up an existing loan or switching banks.
The information gives valuable insights not only into mortgage lending patterns but also into the wider residential real estate market, and is used by a wide range of organisations ranging from economists to the media, including interest.co.nz.
Unfortunately some of the information, in its current format, is not quite what it seems.
Information on the number and value of new mortgage lending is based on the new mortgages approved by the banks each month.
Unfortunately a reasonably common practice among prospective borrowers looking to take out a mortgage is to get quotes from more than one bank, to see which one will give them the best deal.
So a borrower may request a mortgage from two or three different banks, all of which get approved, but only one of them actually proceeds.
But all of these prospective "deals" are included in the mortgage approval data submitted to the RBNZ, even though some of them will not result in a mortgage being drawn down.
This creates phantom mortgages in the RBNZ data, that is mortgages which are approved but never actually eventuate.
This means both the volume and value of actual mortgage lending provided by the banks is less than that suggested in the RBNZ's mortgage approval tables.
The RBNZ acknowledges it doesn't know how many of these phantom mortgages are included in its mortgage approval tables, but their effect may be significant.
One of the key datasets the RBNZ provides is the number and value of new residential mortgages approved for the purposes of "property purchases."
In the 12 months to June this year, the RBNZ reported 89,626 such mortgage approvals.
By way of comparison, the Real Estate Institute of NZ reported just 76,742 property sales becoming unconditional over the same period.
Another complicating factor is that the RBNZ figures of mortgage approvals for "property purchases" do not involve a purchase at all.
A purchase implies that a property changes hands under the terms of a Sale and Purchase Agreement, but many of the properties included in the RBNZ figures are non-sale transfers.
These can include properties that have a change in their ownership structure as part of a marriage settlement or inheritance, or are transferred into a trust by their owners.
Where a mortgage is involved, this will be included in the new mortgage approvals data provided to the RBNZ and will end up being included in its "property purchase" data.
Additionally, some of the non-sale transfers will also involve "phantom mortgages," where quotes from several banks are obtained for a single mortgage.
All of this tends to inflate the RBNZ's mortgage approvals data compared to the actual amount of new mortgage lending being provided by the banks. And the difference between the two doesn't end there.
The higher volume and value of mortgage lending suggested in the RBNZ figures also flows into its figures on who is taking out mortgages and what they are being used for.
In other words, the number of first home buyers, investors and existing owner-occupiers taking out mortgages is almost certainly less than the RBNZ mortgage approval figures suggest, and the same applies to whether mortgages are for purchasing property, a top up to an existing mortgage or the result of moving from one bank to another.
So at least a modicum of caution is needed when interpreting the RBNZ's mortgage data.
Hopefully, some of these issues will be resolved by the RBNZ's new mortgage data gathering system currently at the advanced planning stage.
The RBNZ is keeping details of the new system under wraps, but interest.co.nz has been told the information it gathers will be much more "granular" than the current data series and should address the issues raised above.
No time frame for its likely implementation has been given.
*All of the numbers referred to in this article refers to data contained in tables C31 and C33 on the RBNZ website, unless stated otherwise.
We welcome your comments below. If you are not already registered, please register to comment.
Remember we welcome robust, respectful and insightful debate. We don't welcome abusive or defamatory comments and will de-register those repeatedly making such comments. Our current comment policy is here.