Putting APRA’s Stress Test to the Test

An excellent piece from LF Economics, which chimes with DFA data too by the way (more on that later). APRA’s “stress tests” are merely window dressing.

So yesterday APRA came out and said that if unemployment rose to 11%, House prices fell by 35%, and the Chinese economy tanked, that the Australian banking system would be able to withstand the economic stresses associated with this type of economic destruction.

So let’s use a bit of history as a guidance and put it to the test using the most abnormal of leniencies to assess whether APRA is full of cow-pat.

What history tells us from housing crashes in the past in other jurisdictions is that when house prices crash by 35% that the riskiest borrowers (those with the highest outstanding loans to income ratio’s and/or the lowest buffers) are totally wiped out alongside those who lose their jobs long enough to run out of savings and default on their mortgage.

As an example, those borrowers in the US just prior to the GFC who had a total liabilities to income ratio higher than 6x income were at a very high risk of default (and many did) when the GFC arrived. Those with a debt to income ratio of  8x or higher were all but guaranteed to go into foreclosure and lose everything they had.

Furthermore, it was all but guaranteed that those who were living beyond their means on the slimmest of income buffers (income – debt repayments- cost of livings) would too be foreclosed upon as the cost of risk rose.

Before even going there and factoring in job losses, the above two cohorts (the highly leveraged and those living beyond their means) alone represented a dangerous fringe of borrowers in the US that cost its economy, and its banking system very dearly when house prices fell.

The Oz banking stress test.

With limited data available here in Australia on banks mortgage books, it has been incredibly hard to find a sample of a banks mortgage book over the years to be able to conduct a stress test of some sort. But several weeks back, The Royal Commission released Westpac’s mortgage book sample that was used in APRA’s highly secretive ‘Targeted Reviews’. These reviews were never meant to see the light of day. But with good fortune, the Royal Commission ascertained these reviews and released Westpac’s, including the mortgage book sample. This mortgage book sample consisted of 420 loans issued over the 2015/16 period.

If this mortgage book sample has any resemblance to the greater mortgage market in general then it would be fair to say that stress testing this sample in a way that gives more benefit of the doubt than we should be giving…should be able to give insight whether a bank like Westpac….indeed the banking system in general would be able to survive in real life the elements that APRA used to conduct its stress test.

Now before we get to the figures I think it’s important to note that in relation to the data in the mortgage book sample, we assume that the data is correct. Yes correct! So correct, that for the sake of this stress test on this mortgage book sample we assume that the borrowers actually earn as much in income as the data suggests they do. Furthermore, we assume that the borrowers monthly costs of living data is accurate…despite some of this data pretty much implying that a fair cohort of borrowers will neither purchase a car, go on a nice holiday or buy any family members Christmas presents over the life of the loan.

We also assume that the total and existing liabilities of borrowers were not modified (reduced) to make borrowers appear more creditworthy than what they really are.

For the purpose of this stress test, and with history telling us that borrowers with the highest leverage ratios and lowest buffers are those who get wiped out, we snippet out the absolute fringes from the mortgage book sample to illustrate the collateral damage that coincides with an all-out economic catastrophe as APRA used in its stress test.

And just to give more than any reasonable absolute benefit of the doubt; instead of calculating credit write-offs of borrowers leveraged 8x or more, we assume that only borrowers who are leveraged 11x or more when the loan was issued are wiped out. Furthermore, we assume that borrowers whom only have a monthly uncommitted income of $70 or less also go into receivership. The findings do not double dip if a borrower has both 11x leverage to income and a uncommitted monthly income of $70 or less. We also assume that borrowers outside of the scope of the selected fringe thresholds ‘do not’ lose their jobs, or for any other reason default on their liabilities.

The findings.

  • Westpac mortgage book sample value – $397,364,308.15
  • Number of borrowers in mortgage book sample – 420
  • Number of borrowers who fail in the stress test according to our assumptions – 44
  • Sum of debt held by borrowers with total liabilities 11x income or greater – $41,180,593
  • Sum of debt held by borrowers with an uncommitted monthly income of $70 or less, but leveraged less than 11x – $14,345,483.91
  • Percentage of borrowers who fail the stress test – 10.48%
  • Proportion of mortgage book sample value that fails stress test – 13.97%
  • Losses to the banking system if scaled: $298 Billion

Conclusion

Despite giving more than the benefit of the doubt on highly questionable data, if Westpac’s mortgage book sample has any broader resemblance or correlation with the broader profile of the Australian household debt market, there is simply no way Australian banks would ever survive APRA’s implied elements used in its stress test once you factor in the further losses outside retail banking in real life….. In other words, there would simply be too much distressed debt, not even the funds from the Committed Liquidity Facility will have enough to cut the mustard to cover the shortfalls. And this is just based on the assumption that fringe borrowers leveraged 11x income or greater, and/or have less than $70 a month buffer wont repay will default in an economic disaster.

In ending, the results of APRA’s stress test further provide evidence that they are a captured regulator..

Reproduced with permission.

 

Author: Martin North

Martin North is the Principal of Digital Finance Analytics

Leave a Reply