Bank Reporting & Housing Update
In this video, Scott Rundell, Chief Investment Officer at Mutual Limited, looks at what a softer Australian housing market means for residential mortgage-backed securities, or RMBS. Using recent bank reporting season data and observed market evidence, Scott explains why falling property values do not automatically translate into mortgage credit losses, and why borrower repayment capacity, arrears and unemployment remain the key indicators to watch. Watch the full video below (16:40).
Transcipt
Scott Rundell (0:00:07)
Today we're covering two connected stories. First, a short look at the August bank reporting season: what the major and regional banks told us about profitability, capital, credit quality, funding, and their outlook. Second, and this is where we'll spend most of our time, the Australian housing market and what it means for residential Mac mortgage-backed securities or RMBS. House prices have started to fall, and it's a fair question to ask what mean what that means for investments backed by mortgages. We'll work through the question step by step using observed data wherever we can. Let's start with the banks.
If I had to sum up this reporting season in a sentence, it would be this: the profit lines were fine, the balance sheets are strong, but the forward-looking credit indicators have started to drift. Starting with profitability. Earnings generally beat expectations, but those expectations were low. You can see the profit growth figures at the bottom left, CBA up 7%, Westpac and NAB up 2%, ANZ up 5%, Bendigo up 3%, and Judo the standout, with profit before tax up 34%. The more interesting point is where the margin came from. At every major, the pricing on loans actually went backwards. What helped margin was treasury income, hedging, and deposit pricing. That matters because some of these supports are more durable than others. On capital, every bank is above its target. The CET rate, CET one ratios on the chart range from about 11.3% at Benigo up to 12.5% at ANZ. The direction of travel is slightly low at CBA, Westpac, and Bendigo, but that's because capital is being used to fund growth, not absorb losses, which is important.
ANZ is the one building capital. NAB has the thinnest headroom at roughly 50 basis points on top of pro forma basis.
Finally, asset quality. Realised losses remain very low by historical comparisons. Mortgage arrears are edging up at CBA and ANZ, but falling at Westpac. On the business side, watch lists are growing, but non-performing loans have barely moved. Provision buffers remain well above the bear case, the bank's bear case scenarios. So, this story is about loan potentially migrating to weaker grades over time. It's not a solvency question. Turning to liquidity and funding. The banks are comfortable here. Liquidity ratios are between 130 to 140% compared to a regulatory requirement of 100%. And the net stable funding ratios between 100 and 111%, 125%, again compared to a regulatory requirement of 100%, all well above regulatory minimums. But there's a subtle shift underway. Deposits are no longer keeping pace with lending, so these stable funding ratios are grinding lower. The banks are issuing a lot more term debt. For example, Westpac last year issued $38bn this financial year, against $28bn the previous year. The competition for deposits is now where margin's being won or lost.
Now for the outlook statements, which is our bridge into housing. The banks agree on the direction: economic growth is slowing, inflation remains high, and housing is turned down. Mortgage applications fell between 15% - 20% following the May Federal Budget and three rate rises this year. Where they disagree is in the magnitude. Westpac made the standard revision of the season, cutting its calendar 2026 house price forecast from a rise of 2.5% to a fall of 1%. And on the chart, you can see the spread of forecast for housing credit growth, next financial year from 2.5% at NAB to between 4% and 6% of CBA. All three grew roughly 6.5% - 8.5% last year. So, everybody's expecting a slowdown. The debate is how sharp. The share market's verdict was quite clear. Banks with heavy housing exposure were marked down, while ANZ and Judo with stronger business lending were rewarded.
So, what are the key takeaways? First, credit is sound. Capital, liquidity, and provisions are comfortable across the sector. Nothing this season changes the credit picture for the major banks. The realistic risk is gradual deterioration from a historically strong base, not solvency concerns. Second, bank debt supply will stay heavy through the next financial year. Westpac has 31 billion of maturities ahead, and NAB has raised over 30 billion, including 6 billion of tier two capital. Expect an active primary market.
Third, deposit pricing is the live battleground. It drove margins at CBA, Judo, and Bendigo, and it's where competitive presses show up first. Fourth, growth is rotated from housing to business growth. That's where above system growth margin and demand will set in. And fifth, bank specific factors increasingly separate one bank from another. Bendigo's regulatory conditions, a control issue identified at Judo, and BOQ’s limited disclosure until October are all examples that sector-wide numbers won't show you. That's the banking picture. Now let's turn to housing and RBS. Here are the six numbers we think matter most. I'll introduce them now and we'll unpack each one over the rest of the presentation. On values, national dwelling prices have fallen 3.1% over the quarter to the end of August. That's a meaningful quarterly fall, though prices are still higher than a year ago. On market sensitivity, during the risk period or risk off period, I should say, in March, triggered by the conflict in Iran, our sample of mezzanine RMBS holdings moved by less than 0.5% in capital price terms. And in the 30 days after the May budget, those same holdings actually rose by just over 50 basis points or 50.5%. On credit: weight average 30 day plus day arrears on RMBS mortgages in the pools we hold is 2.59%. That compares with around 3.5% for the broader market as measured by Standard Poor's SPIN data.
On structural protection: house prices would need to fall 25%, and nearly 4% of the mortgage in a pool would need to default before a single B rated tranche was at risk of a capital loss. And on history, the realised loss rate on Australian rated RMBS, both prime and non-conforming, is zero, no losses ever. If you take one idea away from this section, make it this one: falling valuations are not the same thing as borrowers not being able to pay. Mortgage pools are damaged when borrowers lose the capacity to meet their repayments, not when property values drift lower. Let's look where prices have actually moved because the headlines don't tell the full story. The dark blue lines in this chart show the change over the quarter. Almost every capital city fell, and nationally prices were down 3.1%. Sydney and Melbourne led the decline down 4.7 and 3.9%, respectively. But now look at the lighter coloured bars, which show the change over the full year. Sydney and Melbourne are the only two major cities that are meaningfully negative over that period, both around 4.5% lower. Canberra’s roughly flat.
Everywhere else is still up, and in some cases strongly. For example, Perth is up over 15% for the year, Darwin also nearly 15%, Brisbane 11%, Adelaide, Hobart around 8%, and regional markets are up 8%. Nationally, house prices are still 3% higher than they were a year ago. So, the picture is one of momentum moderating sharply in two markets rather than a broad-based collapse. The quarterly numbers that make the headlines are coming off a very strong base, and in most of the country prices still above where they were twelve months ago.
That raises the obvious question. Does the weakness in Melbourne and Sydney matter for mortgage pools? Let's look at that next. Our view is that the composition of this decline matters much more than its size. And two features make a valuation story rather than a credit story. First, nobody's being forced to sell. Transaction volumes have fallen because buyers are pausing, expecting lower prices, while sellers are refusing to accept prices. We're not seeing any material for increase in foresales. That's important because losses in a mortgage pool come from distress sales, where a lender has to realize the property at a discount. They don't come from valuations drifting lower while the owner keeps paying.
Second, the weak just largely sits outside the pools. The falls in Sydney and Melbourne are concentrated in higher value property, the top quartile. The average mortgage in the deals we hold is $750,000, and concentration limits cap exposure loans above two and a half million. The segment showing the most price stress is substantially absent from these pools. Now, we don't dismiss the headwinds. Three RBA rate hikes and likely another possible two, cost of living pressure, weak consumer sentiment, and the budgets of tax changes are all real. But there are real structural offsets too: constrained housing supply, continued population growth, and most importantly, a resilient labour market. On housing supply, the government is likely to fall two hundred thousand dwellings short of its election target of 1.2 million.
Let's put the forecast into context. Consensus is now for a peak the trough fall of anywhere between six to ten percent fall over the next twelve to fifteen months. In the 2019-2020 COVID period, prices fell around six to eight percent. In 2022 to 2023, when the RBA started to hike rates aggressively, prices fell eight to ten percent. RMBS losses in both episodes were zero. It was very much the same story in the GFC in the 2008-2010 period.
And our stress-testing bear case already assumes a 10% fall in the first year of a deal we look at, plus a 15% haircut on the value of the property at the point of default. That's effectively a 25% decline. So, the fall now being forecast is of the same order as the last two downturns, and it sits comfortably inside what we already test for. Common concern is, even if there are no credit losses, won't RMBS prices fall when markets get nervous? We don't need to model the answer to that. In March, the conflict involving Iran produced a broad multi-asset risk off event. So, the market ran the test for us, and this chart shows the result. Over that period, the ASX 200 fell nearly eight percent, the S&P 500, about five percent, even government bonds fell around 1.4% and fixed rate credit 1.2%. Mutual High Yield Fund, which is a big investor in RMBS, was able to maintain its capital position and gained point zero four percent. Slow, but it's still positive.
Floating rate credit also in this chart you can see up 0.2%. And our sample of RMBS holdings moved less than 0.5% in the capital price. And the coupon from that series meant that the fund delivered a positive return. The reason is structural. These are floating rate notes because the coupons reset the bank will swap above the bank will swap rate. There's no interest rate duration to reprice. The only thing that can move the price is credit spread. And in a period like March, only spread widening is partly offset by higher BBSW and the ongoing coupon income. It's also worth remembering that these figures are price movements only. Coupon income continued to accrue throughout, And as we saw in the earlier slide, in the month of after the May budget, those same holdings recovered and moved higher despite the new structural risk to housing. This is the most detailed slide in the presentation, so let me guide you through it rather than read every number. Each row is a house price decline from down 5% to the top of 60% at the bottom. Each column is a tranche rating from single B through to AAA. And each cell shows the percentage of mortgages in the RMBS pool that would need to default before the tranche loses any capital. So, start at the top. At a 5% or 10% price decline in housing.
There's no plausible default rate that produces a capital loss at any rating. Remember, consensus is a 6% to 10% national fall. Now move down to 25% row, which is the effective decline our bear case already tests. Here, a single-A tranche needs nearly 10% of the pool to default before capital is lost. Even a single B trance, the lowest rating, needs nearly 4%. Compare that with what we've actually seen, arrears of 2.59%, and arrears are not defaults. The vast majority of loans in arrears catch up on their repayments. Historically, cumulative loss rates on mortgages have been in the order of two to ten basis points or 0.02% to 0.10%. Put differently, that's $200 lost for every $1 million lent, which is very, very low by historical and global standards.
It's also worth noting that Australia has never recorded a 25% national price decline in any given 12-month period. Through the early 1990s recession, the GFC, COVID, peak declines were around 10%. One honest point that at this about this table, it isn't linear. As price falls deepened, each default causes a bigger loss. So, the required default rate drops away quickly. But even at a 60% decline, a scenario well beyond anything in Australian history, a single B trend still needs 1% of mortgages to default. And triple-A tranche is nearly fourteen percent.
So how do we apply this in practice? Every deal is stress tested before you can buy it. It's a three-step process. First, we analyse the pool loan by loan, forecasting arrears and foreclosure rates on every individual mortgage. Second, we run those foreclosures through the deal's cash flow structure to test that every layer of the capital stack is repaid in full over the life of the deal. And third, it's a binary decision. If the model forecasts a capital loss, the deal isn't approved, and our portfolio managers are prohibited from investing in that particular deal.
The bear case is deliberately conservative. The cash rate peaks at 5% by the end of 2027, increasingly more likely than not, given the current inflation scenario, but still. Unemployment rises to 5%. And the over the remaining life of the exposure and house prices fall that 10% plus the 15% haircut I mentioned earlier. So, 25% house price decline. Timing matters as much as magnitude too. A deal that collects excess spread for several years before defaults emerge is far better protected than one that defaults in day one, and that's extremely unlikely. And importantly, the tests generally exclude deals. We only approve about two-thirds of what we look at. So, four in ten deals fail because of the forecast on cash flow loss. Those rejected deals are available to buy. This is a reflection of process, not a shortage of supply. We just don't invest in everything. And you can see the outcome in our numbers. Our arrears numbers are 2.59%, sitting against 3.5% for the broader market.
Stress tests are useful, but history is more persuasive. Australian dwelling prices have fallen twice since 2014, and this table compares the two episodes. In 2019 to 2020, house prices fell 6 to 8% while the RBA was cutting rates, with the three-month bank bill falling from 2.9 to just under 1%, and that was the COVID impact. Unemployment was around 5% and rising. In 2022 to 23, prices fell 8 to 10% alongside a really aggressive RBA rate height cycle, and then from 0.1% to 4.35%. Fastest policy tightening in a generation. The bank bill went from 0.3 to 4%, which is what our the bank's price in mortgages off. And non-conforming arrears nearly doubled from 2% to 4%. Spreads moved in both episodes. They widened further down the capital stack but stayed contained higher up. And in both cases, RMBS losses were zero.
So, we've seen downturns of the same order as the one we now forecast under both falling and rising interest rates without a single loss. And going further back, non-conforming arrears exceeded 23% in 2000 in the early 2000s, an average in the mid-teens during the GFC. In both scenarios, no rated RMBS trench ever incurred a loss or a default.
So, what do we watch? You'll notice house prices and not on the list. Historically, they haven't been a meaningful driver of RMBS performance. We watch arrears first, and specifically non-conforming arrears, because they move first and the furthest. Unemployment matters, but perhaps less than you expect. The chart on the left here shows you that.
During COVID, unemployment spike to around seven percent, yet arrears peaked at only five percent. On the right-hand side, interest rates have historically been the stronger influence, which is why inflation and the rate path it drives is the indicator most likely to change our view. So, to bring it together, the housing market is softening, and that's a valuation story. The stress required to impair capital in rated RMBS is has no local precedent. And most importantly, what matters is whether borrowers can keep paying, not where valuations sit.
Thank you for watching. If you'd like to discuss any of this in more detail, please get in touch with the Mutual team.
