By Robert Barwick
27 March 2019
The Australian Prudential Regulation Authority (APRA) has massively understated average Australian mortgage debt, which seriously impacts its ability to assess financial system stability. APRA's track record shows that this cannot be explained away as an oversight—it's yet another example of APRA being a see-no-evil regulator which ignores information that would inconvenience the banks and their gambling in mortgages.
APRA's data reports that the average balance of an Australian mortgage is $276,000. Observers have been curious about this figure for a long time, as it doesn't compute with the borrowing required to afford the sky-high house prices of the past decade or so.
Last week, analysts at Deutsche Bank revealed they have identified a major flaw in APRA data, which they said has underestimated mortgage debt by up to 40 per cent, "potentially giving regulators, economists and the broader analyst community a false sense of comfort about the indebtedness of Australian households", according to the Australian Financial Review's Jonathan Shapiro.
Given that the average includes mortgages that are almost paid out, with a very low balance, as well as mortgages in the millions of dollars, a 40 per cent discrepancy is huge, indicating a lot more heavily indebted households. Matthew Wilson and Anthony Hoo examined APRA's figure of the number of mortgage loans to households, because, they said, it "makes no sense". They observed that since 2008, the number of mortgages has increased by 75 per cent, while population only increased by 18 per cent.
They concluded that APRA's data doesn't account for loan splitting, by which borrowers split the same mortgage into multiple formats, such as a fixed interest component and variable interest component. AFR cites the example of a $1 million mortgage split 50:50 into fixed interest and variable interest loans, which would appear as two loans with a balance of $500,000.
The Deutsche Bank analysts estimated that half of all mortgages are split, and usually three ways, which formed the basis for their recalculation of average mortgage balances being 40 per cent higher. They also questioned the Reserve Bank's figure of $300,000 for average mortgage balances, which may also not account for split loans.
APRA's flawed data means that its mortgage debt statistics do not reflect true household debt, and hence the vulnerability of Australian borrowers to economic shocks. It also adds to the suspicion about the rest of APRA's mortgage data, such as loan-to-value (LVR) ratios. APRA and the RBA claim that most mortgages at origin have LVRs of 60 to 80 per cent, but the banks are known to cheat on these figures. For instance, loans for investment properties will be valued against both the investment property and the borrower's principal home. Understated mortgage debt and LVRs increase the vulnerability of both borrowers and banks to the falling house prices that are now under way.
In the 27 September 2017 issue of the Australian Alert Service, a former APRA employee pulled the curtain back on APRA's internal practices to expose that the regulator isn't interested in accurate data. The whistleblower charged that "APRA failed to require banks to report accurate data on key features of housing loans in spite of government need for more accurate residential mortgage lending data".
The data at issue was very basic, related to investment properties, and whether the banks recorded a borrower's address as their home address, or the address of their investment property. Other government agencies had asked APRA to accurately record addresses of investment properties, so it could be assessed whether loans were for owneroccupiers or investors. The difference is significant, as investor loans are more risky than owner-occupier loans; but if banks record a borrower's address for an investor loan as their home address, and not the investment property address, such loans can be falsely passed off as owner-occupier loans. In 2017, APRA supposedly cracked down on interest-only loans, which were predominantly to investors and were running at almost 50 per cent of all lending, but as this type of lending fell there was a corresponding spike in owner-occupier loans, raising suspicion that the banks had falsely reclassified their lending.
This should be detectable in APRA's data, but according to the whistleblower, APRA had refused to implement the data collection system that would have collected this information, even though APRA's IT staff had designed the system to do it. "Despite the pressure from other government agencies, APRA refused to insist banks use an accurate record of the address of the investment property purchased for regulatory reporting purposes", he wrote. "Instead, APRA is comfortable with banks using the collateral property [principal residence] as a 'proxy' or guesstimate for the address of the property purchased. This is key because many owner-occupiers use their home as security to buy an investment property, potentially in a different state or territory from where the borrower lives. Therefore the collateral property may not always be the same as the purchased property. To what extent? We don't know, and APRA does not seem to want to know."
This is consistent behaviour by APRA. In 2007, the regulator refused to publish a report by its research department, which showed that lowered lending standards by banks had produced a bubble in mortgage lending on which there were rising defaults that could lead to a recession. This report coincided with soaring subprime mortgage defaults in the USA that culminated in the 2008 banking crash, and therefore should have been taken seriously by APRA; the only reason to suppress it was that it would have been uncomfortable for the banks, which already then were over-exposed to mortgage lending and the property bubble. Not long after APRA's management suppressed this report, they disbanded their research department, confirming their see-no-evil, speak-no-evil approach to bank regulation.